USF Libraries
USF Digital Collections

Determinants of female labor force participation in Venezuela

MISSING IMAGE

Material Information

Title:
Determinants of female labor force participation in Venezuela a cross-sectional analysis
Physical Description:
Book
Creator:
Rincon de Munoz, Betilde
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Married women
Single women
And women heads of household
Micro data
Informal sector
Dissertations, Academic -- Economics -- Doctoral -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: The purpose of this study is to fill the gap in research about women in Venezuela by investigating the determinants of their labor force participation between 1995 and 1998. The Central Office of Statistics and Information in Venezuela provides cross-sectional data collected semiannually about individual, demographic, socio-economic and geographical characteristics of individuals living in Venezuela during this period. This study uses binomial and multinomial logit models to test a number of hypotheses. First, the full sample of women between 15 and 60 years old is used to investigate the importance of individual, demographic, socioeconomic, and geographical characteristics in the labor force participation decision, also controlling for a time trend. The same decision is also analyzed for three subsamples: married women, single women, and women heads of household. Comparisons are made between each subsample and the full sample, and also among the different subsamples.Next, multinomial regressions using the same explanatory variables are performed to examine labor market behavior when there is a three-way choice: whether to participate in the formal sector, the informal sector or not to participate in the labor market at all. The multinomial regressions are also performed on the three subsamples as well as on the full sample. Again comparisons are made between each subsample and the full sample and also among the three subsamples. The results of these analyses show considerable differences in motivating factors among the three groups. The conclusion that must be drawn from this research is that one cannot generalize about the women's labor force participation just by studying the behavior of women in the aggregate. The relative importance of motivating factors depends strongly on the specific subsample to which a woman belongs, a fact unrevealed by previous empirical work.The more detailed analyses produced by this dissertation provide deeper understanding of the labor force participation of Venezuelan women. This information will make a valuable contribution to policy-makers who seek to encourage the important economic contribution of women to this previously under-studied labor market.
Thesis:
Dissertation (Ph.D.)--University of South Florida, 2007.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Betilde Rincon de Munoz.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 189 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001988943
oclc - 307532514
usfldc doi - E14-SFE0001985
usfldc handle - e14.1985
System ID:
SFS0026303:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Determinants of Female Labor Force Participation in Venezuela: A Cross-Sectional Analysis by Betilde Rincon de Munoz A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Economics College of Business Administration University of South Florida Major Professor: Carole A. Green, Ph.D. Don Bellante, Ph.D. Gabriel Picone, Ph.D. Maria Crummett, Ph.D. Date of Approval: April 6, 2007 Keywords: Married Women, Single Women, and Women Heads of Household, Micro Data, Informal Sector, Venezuela, Wo rk, Binomial Logit, Multinomial Logit Copyright 2007, Betilde Rincon de Munoz

PAGE 2

Note to Reader: The original of this doc ument contains color that is necessary for understanding the data. The or iginal dissertation is on file with the USF library in Tampa, Florida.

PAGE 3

Dedication I dedicate this project to my mother and father who, from heaven, are probably feeling very happy and proud of me right now. Thanks to them for always believing in me. I also dedicate this to my dear husba nd, Avilio, because he was always my rock, his enthusiasm never failed me. Thank y ou for your confidence in my ability and determination to achieve this.

PAGE 4

Acknowledgements First of all, I wish to thank God for his immense love for me, and for giving me the strength to finish this. I would also like to thank my professors in Venezuela as well as at USF. They all were dedicated teachers whose opportune advi ce and mentorship was crucial for my scholastic success. In particular, I want to thank the members of my Dissertation Committee: Dr. Carole Green, Dr Bellante, Dr. Gabriel Picone and Dr. Maria Crummett. I want to thank my family, for their c onstant love, support an d above all, patience in seeing me through this process. Special tha nks go to my “assistants ,” Betica, Luchita, Nicole, Qing Su, Josefina and Diane, who re viewed many versions of this manuscript, helped with tables, and the dreadful language barrier. Finally, I want to recognize Dr. Carole Green. Nobody showed more interest in seeing me finish this than her. I was amazed and wonderfully blessed to have a mentor so dedicated to her students and to top-notch scholarship. I could not have done this without her mentorship and support. Thank you very much.

PAGE 5

i Table of Contents List of Tables ................................................................................................................ ...iv List of Figures ............................................................................................................... ...vi Abstract ...................................................................................................................... .....vii Chapter One Introduction .................................................................................................1 1.1 Venezuela in the 1990s ...................................................................................2 1.2 The Venezuelan Economic Crisis: 1980-1999 ...............................................3 1.3 The Venezuelan Population in the 1990s........................................................5 1.4 Education ........................................................................................................6 1.5 The Venezuelan Labor Market in the 1990s ..................................................8 1.6 Venezuelan Women and the Labor Market ..................................................10 Chapter Two The Theory of Allocation of Time and Human Capital Investment ........12 2.1 The Neoclassical Model of Allocation of Time ...........................................13 2.1.1 The Household Production Approach ..........................................21 2.1.2 The Tripartite Choice Mode l of Allocation of Time .....................24 2.2 The Human Capital Investment ...................................................................26 2.2.1 The Human Capital Investment Model .........................................26 2.2.2 Women’s Supply of Labor ............................................................30 Chapter Three Lite rature Review ...................................................................................38 3.1 Women’s Labor Force Particip ation in the United States ............................38 3.2 Women’s Labor Force Partic ipation in Other Developed Countries .....................................................................................................51 3.3 Women’s Labor Force Partic ipation in Latin American Countries .....................................................................................................59 3.4 Women’s Labor Force Participation in Venezuela .....................................69 3.5 The Contribution of My Dissertation Research ...........................................71 Chapter Four Research Design .......................................................................................74 4.1 Objectives and Hypotheses ..........................................................................75 4.2 Data Base .....................................................................................................76 4.3 Methodology .................................................................................................77 4.3.1 Binomial Logit Model ...................................................................77 4.3.2 Multinomial Logit Model ..............................................................78 4.3.3 Estimation Method of Maximum Likelihood ................................80 4.4 Specification of the Models ..........................................................................81

PAGE 6

ii 4.4.1 Dependent Variable .......................................................................81 4.4.2 Independent Variables: Description and Motivation .....................82 4.4.3 Testing Hypothesis No. 1 ..............................................................86 4.4.4 Testing Hypothesis No. 2 ..............................................................89 4.4.5 Testing Hypothesis No. 3a ............................................................94 4.4.6 Testing Hypothesis No. 3b ............................................................95 Chapter Five Research Results .......................................................................................96 5.1 Description of the Samples Us ed for the Regression Equations ..................96 5.2 Determinants of Women Labor Fo rce Participation in Venezuela .............100 5.2.1 Results of the Test ing of Hypothesis No. 1..................................100 5.2.2 Results of the Test ing of Hypothesis No. 2..................................105 5.2.3 Results of the Testi ng of Hypothesis No. 3a ...............................112 5.2.3.1 Results for Married Women .........................................112 5.2.3.2 Results for Single Women ............................................117 5.2.3.3 Results for Women Heads of Household ......................120 5.2.3.4 Comparisons Among the Three Subsamples ...............124 5.2.4 Results of the Testi ng of Hypothesis No. 3b ...............................125 5.2.4.1 Results for Married Women .........................................125 5.2.4.2 Results for Single Women ...........................................131 5.2.4.3 Results for Women Heads of Household ....................136 5.2.4.4 Comparisons Among the Three Subsamples ...............142 Chapter Six Conclusions ...............................................................................................146 6.1 Main Findings .............................................................................................146 6.1.1 Age ...............................................................................................146 6.1.2 Education .....................................................................................148 6.1.3 Marital Status ...............................................................................150 6.1.4 Urban Residence ..........................................................................152 6.1.5 Regions ........................................................................................153 6.1.6 Heads of Household ....................................................................155 6.1.7 Socio-economic Status ................................................................156 6.1.8 Nonlabor Income and the Interaction Terms ...............................157 6.1.9 Survey Date .................................................................................158 6.2 Limitations .................................................................................................159 6.2.1 The Presence of Children ............................................................159 6.2.2 Socio-economic Status .................................................................159 6.2.3 Nonlabor Income .........................................................................160 6.3 Future Research ..........................................................................................160 References .................................................................................................................... .161 Appendix A: Definition of Variables ...........................................................................175

PAGE 7

iii Appendix B: Tables ......................................................................................................182 About the Author End Page

PAGE 8

iv List of Tables Table 1 Description of Samples Used for Regression Equations .......................99 Table 2 Binomial Logistic Regr ession Results Coefficients Sample: All Women 15-60 ..................................................................102 Table 3 Binomial Logi stic Regression Results Marginal Effects Sample: All Women 15-60 ..................................................................103 Table 4 Multinomial Logit Regression Results Formal Sector Sample: All Women 15-60 ..................................................................107 Table 5 Multinomial Logit Regression Results Informal Sector Sample: All Women 15-60 .................................................................108 Table 6 Binomial Logistic Regr ession Results Coefficients Subsample: Married Women ................................................................114 Table 7 Binomial Logi stic Regression Results Marginal Effects Subsample: Married Women ................................................................115 Table 8 Binomial Logistic Regr ession Results Coefficients Subsample: Single Women ...................................................................118 Table 9 Binomial Logi stic Regression Results Marginal Effects Subsample: Single Women ..................................................................119 Table 10 Binomial Logistic Re gression Results Coefficients Subsample: Women Heads of Household ............................................121 Table 11 Binomial Logistic Regr ession Results Marginal Effects Subsample: Women Heads of Household ...........................................122 Table 12 Multinomial Logit Regression Results Formal Sector Subsample: Married Women ................................................................128 Table 13 Multinomial Logit Regression Results Informal Sector Subsample: Married Women ................................................................129

PAGE 9

v Table 14 Multinomial Logit Regression Results Formal Sector Subsample: Single Women ...................................................................132 Table 15 Multinomial Logit Regression Results Informal Sector Subsample: Single Women ...................................................................133 Table 16 Multinomial Logit Regression Results Formal Sector Subsample: Women Heads of Household .............................................137 Table 17 Multinomial Logit Regression Results Informal Sector Subsample: Women Heads of Household .............................................138 Table B.1 Binomial Probit Regr ession Results Coefficients Sample: All Women 15-60 ...................................................................182 Table B.2 Binomial Probit Regres sion Results Marginal Effects Sample: All Women 15-60 ..................................................................183 Table B.3 Multinomial Probit Regression Results Formal Sector Sample: All Women 15-60 ..................................................................184 Table B.4 Multinomial Probit Regression Results Informal Sector Sample: All Women 15-60 ..................................................................185 Table B.5 Women’s Labor Force Participation by Geographical Areas of Venezuela ..............................................................................186 Table B.6 Venezuelan Women’s Labor Force Participation by Samples .............187 Table B.7 Multinomial Logit Regre ssion Results Marginal EffectsFormal Sector ........................................................................................188 Table B.8 Multinomial Logit Regre ssion Results Marginal EffectsInformal Sector .....................................................................................189

PAGE 10

vi List of Figures Figure 1 Distribution of Venezuelan Population ....................................................6 Figure 2 The Labor Force Participation Decision ................................................18 Figure 3 The Production of Child Care ................................................................22 Figure 4 Large vs. Small Substitution Eff ect When the Wage Rate Increases .....25 Figure 5 The Optimal Acquisi tion of Human Capital ..........................................28 Figure 6 Alternative Earnings Streams .................................................................30 Figure 7 The Impact of Work Interrupt ions on the Education Investment Decision of Women ................................................................................32 Figure 8 Sharing of Costs and Benef its in Firm-Specific On-The-Job Training ...................................................................................................33 Figure 9 Investment in General On-The -Job Training Over the Life Cycle .........35 Figure 10 Geographic Ar eas in Venezuela .............................................................84

PAGE 11

vii Determinants of Female Labor Force Participation in Venezuela: A Cross-Sectional Analysis Betilde Rincon de Munoz ABSTRACT The purpose of this study is to fill the gap in research about women in Venezuela by investigating the determin ants of their labor force participation between 1995 and 1998. The Central Office of Statistics and Information in Venezuela provides crosssectional data collected semiannually about individual, demographic, socio-economic and geographical characteristics of individuals living in Venezu ela during this period. This study uses binomial and multinomial logit models to test a number of hypotheses. First, the full sample of women between 15 and 60 years old is used to investigate the importance of individual, demographic, soci oeconomic, and geographical characteristics in the labor force participation decision, al so controlling for a time trend. The same decision is also analyzed for three su bsamples: married women, single women, and women heads of household. Comparisons are made between each subsample and the full sample, and also among the different subsamples. Next, multinomial regressions using the same explanatory variables are performe d to examine labor market behavior when there is a three-way choice: wh ether to participate in the formal sector, the informal sector or not to participate in the labor market at all. The multinomial regressions are also performed on the three subsamples as well as on the full sample. Again comparisons are made between each subsample and the full sample and also among the three subsamples.

PAGE 12

viii The results of these analyses show consider able differences in motivating factors among the three groups. The conclusion that must be drawn from this resear ch is that one cannot generalize about the women’s labor force part icipation just by st udying the behavior of women in the aggregate. The relative importa nce of motivating fact ors depends strongly on the specific subsample to which a woman belongs, a fact unrevealed by previous empirical work. The more detailed analyses produced by this dissert ation provide deeper understanding of the labor force participati on of Venezuelan women. This information will make a valuable contribution to policy-ma kers who seek to encourage the important economic contribution of women to this previously under-studied labor market.

PAGE 13

1 Chapter One Introduction The massive influx of Latin American women into the labor market, and their ensuing contribution to the region's economic growth were two of the most important developments in the Western Hemisphere in the twentieth century. These developments took place in the context of structural cha nges that forced women to look for and respond to the demands of the market in the empl oyment sector. Indeed, for any country, the proportion of various population groups in th e labor force both a ffect, and reflect, the overall rate of economic growth, the economic circumstances of those groups, and the role of women in the society. This project examines the determinants of this increased participation of women in the labor force in Venezuela. Although a study of labor supply includes the level of labor force participation, as reflected in annual hours worked, as well as on the number of individua ls participating in the labor force at a poin t in time, this project will deal only with labor fo rce participation. Since labor force participation in the U.S. and other develope d countries has been studied extensively, a brief survey of empirical evid ence of these countries is included in this project as well as that of developing count ries, including those from Latin America. The level of women's participation in th e labor market in Venezuela has increased dramatically in recent decades, from 17.5 percent to 47.2 percent between 1950 and 2000. The rise of the petroleum state and the rapidly growing economy also created

PAGE 14

2 employment for women in the labor force. However, the economic downturn at the end of the twentieth century impacted the overa ll labor market as well as the labor participation rates of women. The first objective of this research proj ect is to investigate the effects of various factors on women’s deci sion to work during the second half of the 1990s in Venezuela. Secondly, given the increased importance of the informal sector in the labor market, the project s eeks to understand the distributi on of the female labor force among the formal and informal sectors. Fina lly, the differences in choosing employment in the formal and informal sector among married, single women, and female heads of household are investigated. The study contribut es to the economic literature devoted to analyzing labor force participation in Latin Am erica, and specifically Venezuela. It also provides an analysis that coul d serve as the basis for the formulation of emerging public policies oriented towards women's advancement. 1.1 Venezuela in the 1990s In order to understand Venezuelan wome n’s labor force participation during the 1995-1998 period, it is important to mention so me relevant characteristics of the Venezuelan population, the labor market, a nd the most important indicators of the country’s economic activity. The Venezuelan economy is mainly based on the exploitation and commercialization of petroleum and its bypr oducts. For many years, the Venezuelan people enjoyed a relatively good standard of living, as well as urbanization and modernization, as byproducts of the extraordinarily high oil revenues. The quadrupling of crude oil prices in 1973 spawned an oil e uphoria and a spree of public and private

PAGE 15

3 consumption unprecedented in Venezuelan history. During the 1970s, the government established hundreds of new state-owned enterp rises and decentralized agencies, as the public sector became the primary engine for economic growth. In addition to establishing new companies in such areas as mining, petrochemicals, and hydroelectricity, the gove rnment also purchased private companies. In 1975, the government nationaliz ed the steel industry and in 1976, the oil industry was nationalized. This tremendous in flux of money allowed the pub lic sector of the economy to embrace an internal industrialization that substituted for previous imports of many finished goods, capital and technologies, and provided for the construction of an important infrastructure of highways, extensive irrigation of agricultural lands, and a huge hydroelectric system, among others. More over, in the last three decades, the economy reached some degree of diversificatio n with the exploitation and exportation of iron ore, petrochemical products, aluminum, cem ent, steel and other industrial products. The Venezuelan government also made importa nt advances in provi ding public health and education services to the Venezuelan people. 1.2 The Venezuelan Economic Crisis: 1980-1999 In 1983 the price of oil fell, and soaring interest rates caused the national debt to multiply. Oil revenues could no longer support the array of government subsidies, price controls, exchange-rates losses, and the ope ration of more than 400 public institutions. Widespread corruption and political patrona ge only exacerbated the situation. By 1989 the economy could no longer support the high rates of subsidies and the increasing

PAGE 16

4 foreign debt burden, particularly in light of the nearly 50 pe rcent reduction of the price of oil during 1986. In 1989 the government launched profound policy reforms with the support of structural adjustment loans from the Inte rnational Monetary Fund (IMF) and the World Bank.1 The purpose was to reduce the role of government in the economy, orienting economic activities toward the free market, a nd stimulating foreign investment. The most important adjustment was the massive devalu ation of the national currency, the bolivar.2 In spite of these efforts, the extraordinar y outflow of monetary resources from the economy created one of the most serious fina ncial banking crises in Venezuela in 1994.3 As a consequence, another structural adjust ment program called the Venezuelan Agenda was implemented. The final years of the 1990s were marked by great economic structural changes that brought about a sharp decline in the standard of living. Those in the middle and working classes faced increasing financia l hardships: the poverty rate increased by over 60 percent by the end of 1997.4 Low employment in the oil sector, lack of sustained growth of non-oil activities, and shrinkage of the public sector, i.e. the main causes of the contraction of the formal sector, are lik ely to remain unchanged for some time. 1 The document with all the economic adjust ment was known as the “Big Turnaround.” 2 Other related policies sought to elim inate budget deficits by 1991 through the sa les of state-owned enterprises, to restructure the financial sector and rest ore positive real interest rates, to li beralize trade through tariff reduction and exchange rate adjustment, and to abolish most subsidies an d price controls. The government also aggressively pursued debt reduction schemes with its commercial creditors in an effort to lower its foreign debt repayments. 3 The 1994 banking crisis was unprecedented in Venezuela. Very few other Latin American countries experienced a similar situation. It was accompanied by a currency exchange control period of 21 months (1994-1996). 4 The World Bank reported that in Venezuela the poverty rate in 1981 was 17.7 percent; it reached 78 percent by the end of 1990s.

PAGE 17

5 1.3 The Venezuelan Population in the 1990s The Venezuelan population was approximate ly 23 million people at the end of 1998.5 Despite a low overall population density (21.4 persons per square kilometer in 1987), the distribution is extrem ely uneven. The most striking phenomenon in the distribution of the Venezuelan population has been the shift from a highly rural to an overwhelmingly urban population in response to the process of economic growth and modernization due to the development of th e oil industry. Most of its population is concentrated in the western Andean region and along the coast. Although nearly half of the land lies south and east of the Orinoco River, that area co ntained only about 4 percent of the population in the late 1980s. About 75 pe rcent of the total popul ation lived in only 20 percent of the national territory, mainly in the northern mountains (Caracas and surrounding areas) and the Maracai bo lowlands. In the 1990s, the north, which is the area of the country’s first colonial cities, agricultu ral estates, and urban settlements, remained the administrative, economic, and social hear tland of the country. Moreover, 40 percent of the people live in the eight most urbanized cities of the country6 whereas the indigenous population (1.5 percent) lives in the southern area s of Venezuela and also in some part of the Zulian region. Finall y, for the period under analysis, 1995-1998, 86 percent of the Venezuelan population lived in ur ban areas, as the following figure shows. 5 Census of the Republic of Venezuela, 2000. 6 Caracas, Maracaibo, Valenc ia, Barquisimeto, Merida Guayana, and Cumana.

PAGE 18

6 Figure 1 Distribution of Venezuelan Population Source: University of Texas Library, http://www.lib.utexas.edu/maps/americas/venezuela_pop_1972.jpg 1.4 Education Although the issue of free public and com pulsory education at the primary level first arose during the independence struggle7 in 1811, the real beginning of free public 7 The ideal of free, universal education has become inextricably joined to the name of the national hero Simon Bolivar. This ideal has since permeated Venezuelan educational policies.

PAGE 19

7 education began in 1870 when the president of Venezuela, Antonio Guzman Blanco, issued a decree in which he recognized compulsory elementary mass education as the responsibility of national and local governments. At least six years of primary school were compulsory until 1980, when the La w of Education was passed. This law established compulsory preschool educati on and nine years of basic education for children six to fourteen year s of age. For those continui ng their education, the system offered two years of diversified academic, tec hnical, and vocational study at a senior high school, which could be followed by various t ypes of higher educa tion—junior college, university, or technical institute, all paid for the government. In addition, adults were encouraged to participate in special night classes conducte d at all education levels. Overall, Venezuela was among the most liter ate of the Latin American countries. The literacy rate among Venezuelans fifteen y ears of age and older increased from 51.2 percent in 1950 to 91.1 percen t in 1995. College education enrollment has also grown significantly. By 1995, approximately 600,000 people were registered in more than 100 private and public colle ges and universities,8 technical schools, and military institutions. It is also important to hi ghlight the success of the “Gr eat Mariscal of Ayacucho,” a scholarship program implemented in 1975. U nder this program, thousands of students have enrolled in American a nd European universities at bo th undergraduate and graduate levels. 8 The most important public universities are the Central University of Venezuela (founded in 1725), Andres Bello Catholic University (founded in 1953), Metropolitan Univ ersity (founded in 1970) and the Simon Bolivar University (founded in 1970) in Caracas; Los Andes University (founded in 1810) in Mrida; Zulia University (founded in 1891) in Maracaibo; Carabobo University (founded in 1892) in Va lencia; Oriente (founded in 1958) with the headquarters in Cumana and branches in different cities of the North-Ea stern region; and, the Centro-Occidental Lisandro Alvarado University (founded in 1962) in Barquisimeto.

PAGE 20

8 1.5 The Venezuelan Labor Market in the 1990s The significant increase of the rate of la bor market activity in the population is an important characteristic of the economy in recent years. The total labor force participation rate rose from 59 percent in 1994, to 69 percent in 1999. During the 1990s, women increased their labor force participati on by 13 percent while that of men increased by only 3.2 percent.9 Labor force participation was highest among women 25 to 44 years of age. Another of the more significant changes th at occurred in the labor market is the increasing importance of the informal sector.10 After remaining fairly constant at around 40 percent for a decade, the percentage of workers employed in the informal sector increased to 49 percent in 1994 and to 52 percen t in 1999. This meant that the majority of the active population was employed outside of the formal sector of the economy,11 with all the implications and costs th at implies, in terms of (a) lacking the benefits of social security and pensions provided through fo rmal employment, (b) lower average salary levels, (c) depreciation of ma rketable skills, and (d) lost income tax revenues for the government. Employment in Venezuela has historically been concentrated in service activities, specifically health, education, pe rsonal services; and trade. Employment decreased in the 9Labor force participation of men decrea sed from 79.5 percent in 1950 to 69.2 percent in 1990; and, increased their participation to 72.4 percent by 1998 (1950 National Census of 1950, OCEI, 1998). 10 Women reported as self-employed (excl uding professional and technicians), em ployers of business with less than 5 employees, or as family workers, are assumed to be engaged in the informal sector of the labor force. This definition follows the methodology used by the Central Office of St atistics and Inform ation (OCEI, 1990). The increase in employment in the informal se ctor was also a byproduct of the globalization process, which was a factor that helped increase self-employment activities. Albeit important, this is beyond the scope this project. 11 The nation's 1990 labor law incorporated provisions for organized labor, collective ba rgaining, generous fringe benefits, and retirement and disability pensions. Ve nezuela passed a national minimum wage law in 1974.

PAGE 21

9 manufacturing industry, the third-largest sect or in terms of employment, in the late 1990s, exacerbated by a notable reduction of indu strial activities. The agriculture sector showed the same tendency to a lesser ex tent. The construction, transportation and communications industries were the only ones that di d not register sign ificant changes in employment. Unemployment fluctuated based largely on the health of the o il industry which in turn greatly impacted the productive activit ies of other industries. In 1978 only 4.3 percent of the labor force was unemployed, co mpared with the peak level of 14.5 percent in 1984 when oil revenues fell. By 1989, th e unemployment rate was 8.7 percent. In 1994, as a consequence of the implementation of the first macroeconomic adjustment, the Big Turnaround, the unemployment rate fluctu ated between 7 and 9 percent. By 1995, it increased to levels of around 11 percent with a peak of 13 percent in 1996 as a result of the banking crises of 1994 (IN E, Annual Report, 1978-1995). During every year of the 1990s, nominal minimum wages increased due to progressive decreases in real wages because of inflation and currency depreciations.12 This decrease in real wages, combined with high rates of unemploym ent, generated losses in household income which helps explains th e increase in the labor force participation rate of women who entered the labor force to compensate for the loss of real household income. Labor policies contained in the Ven ezuelan Agenda were designed and implemented during the period under study (1995-1998) to mitig ate the fall in real salary of those 12 Nominal minimum wages (including tran sportation and food bonus) in urban ar eas increased from 17,794.4 bolivares per month in 1989 to 242,282.7 bolivares per month in 1998 (OIT, Panorama of Labor, 1999), an increase of 1,261.56 percent.

PAGE 22

10 employed in the formal sector, the costs of wh ich were transferred to the employer. This in turn provided disincentives for em ployers to hire additional workers. 1.6 Venezuelan Women and the Labor Market The labor force participation of Venezu elan women increased from 18 percent in 1950 to 31 percent in 1990, and to 43 per cent by 1998 (National Census of 1950, OCEI, 1998). Factors such as accelerated economic growth, democratization of the educational system, and the decrease in fertility rates, on one hand; and the economic and social deterioration after the fall of the oil reve nues and the financial crisis of 1994, on the other, influenced the upward trend in wome n’s labor force participation (Irene Casique, 1994; Orlandina Oliviera, 1997). In fact, im proved educational and job opportunities since the establishment of democracy in 1958 have enabled more women to enter the labor force, thus helping themselves and/or their families attain middle-class status.13 Not surprisingly, those who moved from the lowe r to the middle class in Venezuela often attributed their changed status to their e ducation, and accordingly, many struggled to send their children to private schools so that they could move still further up the social ladder. The social distinction between private and public school, part icularly at the secondary level, has intensified as a result of the expa nsion of public education. This project aims to investigate and inte rpret those changes among these women. The results of this dissertation resear ch, using the individual data of the 13 Most accounts describe the Venezuelan middle-class as the country’s most dy namic and heterogeneous class in terms of social and racial origins, and as th e greatest beneficiary of the process of economic development. Consisting of small businessmen, industrialists, teachers, government workers, pr ofessionals, and managerial a nd technical personnel, this class is almost entirely urban. Some professions, such as in teaching and government se rvices, were traditionally associated with the middle class, whereas newer technical professions have expanded the options and enhanced mobility within this class.

PAGE 23

11 Household Sample Survey data set from 1995.1 to 1998.2, indicate that women’s labor force participation in Venezuela during the pe riod under study is affected by such factors as demographic characteristics, geographical fact ors, and socio-economic conditions. This study goes still further by examining women’s choices between employment in the formal versus the informal sectors, a nd whether these choices are different if they are single, married, or heads of household. My results are unique as there is no other study of women’s labor force participation in Venezuela after the 1980s using micro data. The remainder of this dissertation is organized as follows. The theories of allocation of time and human capital investment are summarized in Chapter Two. Chapter Three provides a brief review of th e empirical evidence on female labor force participation in the United States, othe r developed countries, and Latin American developing countries, with an emphasis on Venezuela. Chapter Four describes the objectives of this project, the methodologi cal approaches, the data used, and the specifications of the models. Chapter Five pr esents and discusses the research results. Finally, Chapter Six summarizes the contri butions this study makes to the existing literature, and provides a brief disc ussion of future research.

PAGE 24

12 Chapter Two The Theory of Allocation of Ti me and Human Capital Investment Two complementary theoretical approach es support the study of labor force participation of women in this project. Both focus on different aspects of the labor supply decisions. On one hand, the neoclassical model of allocation of time deals with whether a woman will enter the labor market by comparing the value of her time in the market to the value she places on her time spent at home and if she decides to do so, how much time will be spent on market work. On th e other hand, the human capital investment theory stresses the relationship between th e return on the investment of acquiring valuable skills and the time the person expects to work during his/her life. In other words, labor participation outcomes are related to general skills acquire d through education and training [Joseph G. Altonji and Rebecca M. Blank, (1999); Francine D. Blau, Marianne A. Ferber and Anne E. Winkler (2002)]. Indeed those who are planning to participate in the labor market as full-time workers are pr ompted to invest more in education and training (Altonji and Blank, 1999). Moreover, the human capital model emphasizes the role of women's preferences and the choices th ey may make to invest less in job-related education and training, as well as to spend a sm aller share of their ad ult years in the labor force (Blau et al., 2002). Othe r factors include premarket discrimination, or societal discrimination, in which various types of so cial pressures influence women's choices adversely. However, such explanations are beyond the scope of this project.

PAGE 25

13 In the following three sections, a brief de scription of these theoretical models and the contribution of each to the study of th e labor force participation of women is presented. 2.1 The Neoclassical Model of Allocation of Time Economists traditionally analyze labor supply through the use of the neoclassical model of allocation of time14 or the model of labor-leisure choice, which is an extension of the utility maximization problem of consumer theory. The model analyzes how individuals make choices in deciding how th ey will spend a fixed amount of time. They must decide how many hours to work, and how many hours to spend consuming a variety of goods, ranging from computers and cars to DVDs and theater. In the simplest model, an individual ha s two uses for his/her time, either working in the labor market at a real wage rate of W per hour, or “leisure”. According to this basic model, individuals wish to maximize their utility15 or satisfaction ( U ) by purchasing goods and services ( C ) in the marketplace and by consuming time in leisure activities ( L ).16 The amount of both consumed will depend on the individual’s market wage ( W ), personal preferences, and the nonlabor income (V) that person enjoys. The individual’s utility function will be: U = f(C, L) (1) 14 The theoretical treatment of the allocation of time was pioneered by Gary S. Becker (1965). 15 Jeremy Bentham (1780) coined this usage. 16 C and L are “composite” goods. We must be aware that utilit y is, in fact, derived by sp ending income and time on the consumption of a wide vari ety of goods and services.

PAGE 26

14 where U is an index that measures the indi vidual’s well being, assuming people are able to rank in order all possibl e situations from the least desirable to the most.17 Thus, a higher index U means more C and/or L and more satisfaction. Moreover, C and L are economic “goods” – that is, whatever economic qua ntities they represent, we assume that more of any particular good is preferred to less. When the individual seeks to maximize his/ her utility with respect to time in the period under analysis, he is bound by two conditi ons: first, he must allocate the day’s discretionary time ( T ) – that is, 16 hours time, either to working for pay ( H ) or to leisure ( L ). The other condition is related to the inco me he needs to buy goods and services in the market place: Labor wages ( W H ) and nonlabor income ( V )18 are the only sources of the individual’s income. These constraints can be written as the following: L + H = T (2) C = (W H) + V (3) The individual’s budget constr aint is represented by equati on (3). It tells us that individual’s consumption expenditures must not exceed the total income. We can rewrite (2) and (3) as follows: C = W (TL) + V (4) 17 Every individual has his own set of indifference curves ( U1, U2 …) reflecting his preferences. For a complete analysis, see Walter Nicholson (1992), pp. 130-132. 18 Nonlabor income includes from property assets, stocks, and dividends. For women, it also is assumed to include the husband’s earnings. A further assumption in this mode l is that the individual does not save or borrow.

PAGE 27

15 Setting up the Lagrangian expression to represent the individual’s utility maximization problem yields = U (C, L) + {[ W (T – L ) + V ] – C } The first order conditions for a maximum are CMU C U C 0 (5) 0 W L U L W MUL (6) Equating (5) and (6), we get W MU MUC L (7) This expression can be interpreted as th e utility-maximizing labor supply decision principle. That is, in order to ma ximize utility, given the real wage, W the individual should choose to work that number of hours fo r which the marginal rate of substitution of leisure for consumption is equal to W (Nicholson, 1992, p. 683).19 The interior solution of the model answers the question of the num ber of work hours to be supplied by the worker. An increase in W, holding income constant, makes leisure more expensive. Therefore, by consuming additional hours of leis ure, the worker gives up more in forgone wages, producing a negative subs titution effect with respect to hours of leisure. On the other hand, since leisure is a normal good, the in come effect will be pos itive. That is, an increase in the wage rate, W will increase the consumption of leisure, L since the person now feels better off. Since work and leisure are mutually exclusive ways to spend one’s 19 For the graphical approach, see George Borjas (2000), p. 33.

PAGE 28

16 time, these two opposite reactions prevent the model from pred icting the direction of the change in the number of hours worked. Th e ambiguity cannot be solved unless one knows the worker’s actual labor supply decision. If the substit ution effect dominates, the result will be an increase in the number of work hours supplied. On the other hand, if the income effect dominates, the number of work hours supplied by the worker will decrease. Empirical studies have shown th at the income effect tends to dominate for men and the substitution effect, for women.20 When nonlabor income, V changes, there is no ambiguity since the income effect operates alone. Thus, an increase in V will cause an increase in leisure time and a decrease in the hours worked, and vice versa. A corner solution of the model will occur when the individual has decided not to participate in the labor force. Economic theory explains this case through th e definition of reservation wage, W* as the measure an individual places on his/her non-market time. The reservation wage is the wage that w ould make a person indifferent between not working and working that first hour. The value of W* is influenced by his/her tastes and preferences, the leve l of nonlabor income V factors influencing the value of one’s time at home such as the number of children, and marital status. This theory has been successfully used to explain women’s labor force participation. Let us consider Figure 2, the graphical depiction of the utility-maximizing labor supply decision of an individual that is the analytical expre ssion of equation (7). The value of the market goods is measured on the y-axis. The numbe r of discretionary 20 See Thomas J. Kniesner (1976); Mary T. Coleman and John Pencavel (1993); Thomas Mroz (1987); Jeffrey Zabel (1993); and Alice and Masao Nakamura (1994).

PAGE 29

17 hours in a day and leisure are measured on th e x-axis from left to right and hours of market work are measured from right to left. Consider the set of i ndifference curves of a woman, U1, U2 and U3. The higher the level of index U the better off she perceives herself to be; the negative of the slope of the indifference curve is the ratio of the marginal rate of substitution between leisure and income ( MUL/MUC ). The budget constraint that she faces is TPM : the negative of its slope repr esents the market wage she faces ( W ). Her utility maximizing point will be the point where the marginal rate of substitution of the highest possible indifference curve e quals her market wage ( W ). In this figure, her market wage is W and her nonlabor income is V At point P in both panels (a) and (b) of the figure, the nega tive of the slope of the indi fference curve at zero hours of market work represents the reservation rate ( W* ). She will choose not to participate if the reservation wage is greater than or equal to the market wage --that is, if W*W as in panel (a). She will be willing to participate in the labor force only if the wage rate that the market offers her is greater than the reservation wage --that is, W>W* as in panel (b). If so, she will maximize he r utility at point Z where the budget constrai nt is tangent to the highest attainable indifference curve (U2), thus achieving the gr aphical equivalent of equation (7) by spending 8 hours on market work and enjoying 8 hours of leisure time. Her total income will be S = 8W + V This analysis suggests that an incr ease in the value of market time ( W ) will produce an increase in the probabi lity that the individual will choose to partic ipate in the labor force. In other words, labor force partic ipation is positively related to the wage or the value of market time. Conve rsely, factors that increase the value of non-market time ( W* ) tend to lower the probability of labor force participation, ceteris paribus i.e. labor

PAGE 30

18 force participation is negatively related to the reservation wage or the value of nonmarket time. Figure 2 The Labor Force Participation Decision Panel (a) Does not Participate in the Labor Market (Corner Solution). U3 U2 U1 P M V Market Goods ($) 0 16 0 16 Hours of Non-market Time Hours of Market Time -W* W T

PAGE 31

19 Panel (b) Participates in the Labor Market (Interior Solution). Market Goods ($) P ZT U3 U2 U1 M S V 0 16 8 8 0 16 Hours of Non-market Time Hours of Market Time W -W

PAGE 32

20 Panel (c) The Impact of Ch anges in Nonlabor Income V on Labor Force Participation. The assumption that leisure is a normal good implies that the reservation wage W* rises (falls) as nonlabor income V increases (decreases). For those individuals out of the labor force, a higher (low er) reservation wage makes it less (more) likely that a person will participate in the labor market. No ambiguity is present since the income effect operates alone. Figure 2 panel (c) depi cts of the effect of a decrease in nonlabor income from V0 to V1 on the woman’s labor force partic ipation, when the wage rate W is held constant. Initially her budget constraint is M0P0T She maximizes her utility at point T Market Goods ($) P1 S Z1P0 U2 U1 M1 0 16 0 16 Hours of Non-market Time Hours of Market Time 11 5 M0 V0 V1 W0* W -W1* U0

PAGE 33

21 P0 since her reservation wage W0* is greater than the wage rate W so she is out of the labor force, as in panel (a). Her income will be V0. If her nonlabor income decreases from V0 to V1, her budget constraint becomes M1P1T She will now enter the labor force because at point P1, zero hours of work, her reservation wage W1* is less than the wage rate W Thus, she will maximize her utility at point P1 and provide 5 hours of work to the market. Her total income will be S = 5W + V1. Note that she is now on indifference curve U0, with a lower level of utility than previously. 2.1.1 The Household Production Approach In the simplest model of labor supply, individuals decide ho w to allocate their time between labor and leisure. Household time was assumed to be leisure time. It was assumed that utility was generated by dire ctly consuming leisure time and purchased goods. A more complex model describes time as being allocated between market work and household production.21 Time not spent working for pay is viewed not as something that is directly consumed but as an input to the production of household commodities. It is these commodities which are ultimately consumed and thus generate utility for household members. Analysis using this model is shown graphi cally in Figure 3. It is assumed that a single mother is the only decision-maker of the household. She derives utility from the commodity “child-rearing.” Her objective is to maximize utility for herself and her 21 The pioneering model of household producti on is by Becker (1965). For a brief summary of the work in this area, see Reuben Gronau (1997).

PAGE 34

22 children. She will be deciding at the same ti me how much to work for pay and how to take care of her children in a way that best satisfies her preferences. Figure 3 The Production of Child Care In the graph, the value of purchased g oods and services is represented on the yaxis and discretionary time in one day is represented on the x-axis. Hours spent on household production are measured from left to right; hours spent working for pay are measured from right to left. Curves S0 and S1, called utility isoquant s, represent the different combinations of pur chased goods and services and household time that generate the same utility. S1 represents greater utility than S0. They have negative slopes because the inputs are substitutes in child-rearing. That is, if household time is reduced, childrearing affording equal satisfaction can be produced by increasing th e purchases of goods High Ratio of Purchases to Household Low Ratio of Purchases to Household TimeJ K S0 S1 0 9 16 Hours Spent in The Household 7 0 Hours Spent Working for Pay 16 Value of Purchased Goods and Services Z T M

PAGE 35

23 or services outside the home The convexity of the isoquant s reflects the assumption that as household time devoted to child-rearing pr ogressively falls, it becomes increasingly difficult to make up for it with purchased goods and services and still hold utility constant. Moreover, along any ray emanating from the origin such as J or K, the ratio of purchased goods and services to household ti me in the production of child-rearing is constant. Finally, to complete the model, the single mother is restricted by her budget constraint MT which reflects the combinati ons of purchases and household time that are possible for her. As in the basic model of la bor supply, the slope of the budget constraint is her wage rate ( W ) which indicates the increased va lue of purchases made possible by an additional hour of paid work. As in the neoclassical model of allocation of time, this individual maximizes her utility at point Z, where she works for pay 7 hours and devotes 9 hour s to taking care of her children. Note that whether household tim e is conceived of as an input into the production of commodities or as leisure time, the resulting theory of labor supply is unchanged. First, let us consider the case wh ere there is nonlabor income (V) holding the wage rate constant. Her budget c onstraint would shift to the no rtheast (and be parallel to the original one). The income effect would tend to reduce labor supply to the market. She would tend to purchase more, or higher-quali ty, goods and services, and she would spend more time at home. If her wage rate ( W ) were to rise, there would be income and substitution effects. As above, the income effect tends to reduce market labor supply. The substitution effect, i.e. the fact that the higher wage increases the cost of spending an extra hour at home, serves to increase hours of market work. As in the neoclassical model, theory cannot tell

PAGE 36

24 us whether, if wages increase, the income or the substitution effect will dominate. The result will depend on the shape of the utility isoquants. 2.1.2 The Tripartite Choice Mode l of Allocation of Time A more complex household production m odel assumes more than one decision maker in the household (husband, wife and any children old enough to work). The threeway allocation of time model of labor supply considers choices among actual leisure time, time spent on household production, and market work. Hence there are two substitution effects when the market wage increases: one between market and household work and the other between market work and leisure time. It is argued that the magnitudes of these two effects are different an d that the weight of the former in one’s overall response to a wage change is rela ted to one’s role in household production. Regarding substitution between market and household work, purchasing more goods or services can easily compensate for fewer hour s of household work. For example, reduced time devoted to such household chores as cook ing, cleaning, and childcare can be easily replaced through the purchase of a microwave, prepared food, an electric dishwasher, or the services of a babysitter. On the other hand, the substitution between market work and leisure is more difficult since leisure activit ies consume time and the possibilities for economizing on time are thus limited.22 However, those with higher wages are more likely to engage in leisure activities that re quire expensive market-p urchased inputs such as skiing or playing golf. Those with lower wages are more likely to engage in more 22 One definition of the difference between household producti on time and leisure time is that one could pay someone else to perform household production tasks, but not to pursu e leisure activities. However, with respect to the leisure activity of travel, for example, a person with a higher mark et wage might fly to a vacation destination rather than driving, thus saving some time but spending more money.

PAGE 37

25 time-intensive, less goods-intensive leisur e activities such as hiking or working crossword puzzles. One can illustrate the difference in ma gnitudes of both substitution effects using two-dimensional graphs such as shown in Figur e 4 panels (a) and (b). On the y-axis we represent the value of goods (dollars); and, on the xaxis are time spent in household work and time spent in leisure in panel (a) and (b) respectively. Figure 4 Large vs. Small Substitution E ffect When the Wage Rate Increases Value of Goods (dollars) Indifference curve New Wage New Wage Indifference curve Old Wage Old Wage Time Spent in Household Work 0 Time Spent in Leisure Value of Goods (dollars) (a) A Relatively Large Substitution Effect between Market and Household Work (b) A Relatively Small Substitution Effect between Market Work and Leisure 0

PAGE 38

26 To isolate the substitution effect asso ciated with a higher market wage, the new budget constraint with a steeper slope is kept tangent to the same indifference curve. Panel (a) shows the tradeoff between the market goods and household work time that keep utility constant, while pa nel (b) shows the goods-leisure tradeoff. The gradual curve in the indifference curve in panel (a) implie s that a reduction in hours of household work can easily be compensated for by purchasi ng more goods. Conversely, panel (b) shows that the sharper curve in the goods-leisure i ndifference curve reflects the greater difficulty of substituting goods for leisure time wit hout loss of utility or satisfaction. 2.2 The Human Capital Investment 2.2.1 The Human Capital Investment Model Modeling the labor s upply decisions requires not only decision factors such as the current wages, preferences regarding house hold production and/or le isure, but also a framework that incorporates la bor market investment behavior into a lifetime perspective. Many labor supply decisions requ ire a substantial investment on the part of the worker. An individual invests resources in himself today in order to increase his or her future productivity and earnings. Economists refer to this behavior as investment in human capital. The most important kind of invest ment in human capital is education and training. The knowledge and skills a worker has, gained from education and training, including the learning that e xperience yields, generate a certain stock of productive

PAGE 39

27 capital. The value of this pr oductive capital depends on how much one with these skills can earn in the labor market.23 As in any study of investment decisions, to determine whether it is worthwhile, one must compare expenditures and receipts in curred at different periods. The investor must be able to calculate the returns to the investment by comparing the current costs24 with the future returns or benefits. In the case of educational and training investment by workers, the expected returns are in the fo rm of higher future earnings, increased job satisfaction over one’s lifetime, and a greater appreciation of non-market activities and interests. Benefits that are received in the future are worth less to us now than an equal amount of benefits received today.25 The basic model of human capital inve stment assumes that people are utility maximizers and take a lifetime perspective when making choices about education and training. The widely used concept of present value allows us to calculate the value of amounts received in different time periods. Present Value = ..... ) 1 ( ) 1 ( ) 1 (3 3 2 21r B r B r B T Tr B ) 1 ( (8) where Bt is a stream of yearly benefits ( B1, B2,….) over time periods (1 to T ), and r is the discount rate. Since r is positive, benefits into the future will be increasingly discounted. In making decisions, workers compare the present value of future benefits with the costs. 23 Investment in job search and migration also increases the value of one’s human cap ital (Ronald G. Ehrenberg and Robert S. Smith, 200, p. 290). However, th ese last two human capital investments are beyond the scope of this project. 24 The cost of an additional year of schooling includes such co sts as tuition, supplies, and forgone earnings, as well as psychic costs. 25 If people plan to consume their benefits they prefer to consume them earlier; if people plan to invest the monetary benefits rather than use them for consumption, they can earn interest on th e investment and increase their funds in the future.

PAGE 40

28 For example, individuals deciding about an additional year of schooling are assumed to compare the near-term investment costs ( C ) with the present value of expected future benefits. So, investment in additional schooling is attractive if the present value of future benefits exceeds costs. Figure 5 yields some interesting insigh ts about the behavi or and earnings of workers. The human capital decision can be illustrated comparing marginal costs ( MC ) and marginal benefits ( MB ). Figure 5 The Optimal Acquisition of Human Capital The marginal costs, MC of each additional unit of human capital are assumed to be constant. The present valu e of the marginal benefits, MB is shown as declining, because each added year of schooling mean s fewer years over which benefits can be recouped. The utility-maximiz ing amount of human capital ( HC* ) for any individual is shown as that amount for which MC = MB Panel (a) of Figure 5 shows a worker who Units of Human Capital (HC) Units of Human Capital (HC ) Marginal Cost (MC) and Marginal Benefits (MB) 0 0 MB MC’ HC’ HC* ( a ) (b) MB MB’’ MC HC’’ HC* MC

PAGE 41

29 finds learning to be especially demanding, to which he attaches a higher marginal cost, MC’ Accordingly, he will acquire a lower level of human capital, HC’ Panel (b), depicts the case of those who expect lower benefits ( MB” ) in the future from additional investment in human capital, who will acquire less human capital, HC” Many insights from this simple theo ry can be discovered by analyzing the decision a young adult faces about whether to i nvest full-time in education or a training program after leaving high school. Figure 6 illust rates, for example, a person considering college. She or he faces a choice between two streams of earnings over her or his lifetime. Stream A shows the earnings stream of a high school graduate. This stream begins immediately but does not increase very much over time. Stream B that of a college graduate, has negative income for the first four years, followed by a period when the wage may be less than the high school gradua te makes, but then it takes off and rises above stream A.

PAGE 42

30 Figure 6 Alternative Earnings Streams Clearly, the earnings of the college graduate would have to rise above those of the high school graduate to induce someone to i nvest in a college education. The gross benefits, i.e. the difference in earnings betw een the two streams, must total much more than the costs because such returns are in the future and are therefore discounted. This graph relates to equation (8) in th at T=43 (65-22) and in each year Bi represents the difference between the earnings of a colle ge graduate and a high-school graduate. 2.2.2 Women’s Supply of Labor Human capital theory suggests several reasons why women might decide to acquire smaller amounts of formal educati on than men. Many scholars have emphasized the traditional roles of women within the fam ily of which childbearing is one of the most important. Women know that bear ing children might force them to leave the labor market Gross BenefitsB A 22 Direct Costs Indirect Costs E a r n i n g s $ Earnings Stream B Earnings Stream A Age of workers C o s t $ 65 0

PAGE 43

31 for a while. Again, the present value equation (8) gives us the insi ght of the potential behavior of women. If a woma n is planning to interrupt he r participation in the labor market, her investment in additional educati on might no longer be profitable since her time out of the labor market results in a reduc tion in benefits since T would be smaller. Moreover, a woman may decide against invest ment in the types of human capital that require sustained, high-level commitment to the labor force because the investment depreciates rapidly during pe riods of work interruptions.26 26 Moreover, the human capital model helps explain gender differences in fields of specializations. For instance, women would prefer to work as teachers of history or languages which have a slower pace of change, and avoid working in those fields in which technological ch ange is fast, such as engineering.

PAGE 44

32 Figure 7 illustrates the impact of thes e factors on the investment in formal education by women. The total time elapse d since completing high school is represented on the x-axis. Figure 7 The Impact of Work Interruptions on the Education Investment Decision of Women The graph depicts the case of a woman w ho plans to be in the labor force for a period of 6 years after college and then to drop out for 10 years, say, for childrearing. Assuming that she will retire at age 65, her expected work li fe is 33 years instead of 43 years. EF represents her earnings profile if she had decided not to go to college. If we assume that her skills depr eciate during the time spent out of the labor force, upon her return to the labor force, her earnings ( e2) will be less in real terms than she was making when she left ( e1). Consequently, after her return to the labor force she will be facing Returns of a Discontinuous Worker D H F $ e1 e2 E 0 A Potential Experience ( Years ) E a r n i n g s C G 4 10 20 B 47 C o s t s

PAGE 45

33 profile GH rather than profile CD.27 The time out of the labor force has cost her a reduction of earnings over the remainder of her working life. In this example, the benefits of the investment in college e ducation, the sum of the two sh aded areas, may not be large enough to make it worthwhile. Thus, a woman w ith an adherence to the traditional roles in the family is less likely to pursue college and graduate study. Anticipating time out of the labor force, she is lik ely to reduce her amount of educational investment. Other kinds of human capital investments are those made after one has started to work, in training received at the workplace. Al l forms of training, whether formal training programs, informal training under the superv ision of a more experienced worker, or general training, are costly. If the training is specific to one firm or employer, workers and the firm share the cost. Figure 8 Sharing of Costs and Benefits in Firm-Specific On-The-Job Training 27 However, the figure shows that GH approaches CD over time, as she retools or becomes less rusty. Employer’s Share of Costs Employer’s Share of Gross Benefits Worker’s Share of Gross Benefits Worker’s Share of Costs M J’ T’ E’ Experience E a r n i E n g T s J ($) 0 B M P Trainin g Perio d Post – Trainin g Perio d

PAGE 46

34 Consider an individual’s decision to invest in firm-s pecific training. Profile JJ’ represents his productivity if he engages in the training. EE’ is the earnings profile available to him at another firm with no tr aining. Since firm-sp ecific training is not transferable, there are incentives for the worker and employer to share the costs. On the the individual’s side, he is not willing to bear all the training costs because if were to lose his job, all his investment would vanish. By the same token, the employer is unwilling to bear all the costs of firm-speci fic training because if the individual were to quit, the firm would lose its investment. Moreover, if th e employer were to bear all the cost and received all the benefits, the individual’s ea rning profile would be EE’. In this case, he would have an incentive to quit his job when a shift in demand resulted in higher wages or even better working conditions elsewhere. The solution for both employer and the indi vidual (employee) is to share the cost of, and returns to, firm-specific training. TT’ would be the employ ee’s earning profile in that case. He would be paid a wage greater than his marginal pr oduct during the training period (from 0 to M) since his productivity is low; the employer accep ts the lower current productivity in exchange for higher output la ter. But after traini ng (from M to P’) the employee’s wage is below his post-training marginal product. Work ers accept the lower wages for the same reason that one decides to obtain formal schooling: in the expectation of improving the present value of their life time earnings (Becker, 1985). In general, for workers and employers, the increases in produc tivity yield higher ea rnings and profits, which will be greater the longer the worker stays with the firm.

PAGE 47

35 In the case of general on-the-job training, in which employees acquire skills usable elsewhere, the workers alone will pay th e training costs. It this case the magnitude of cost versus future benefits is the indi vidual’s principal concer n (Blau et al., 2002). Figure 9 Investment in General On-The -Job Training Over the Life Cycle Figure 9 graphically depicts the life-cycle implications of human capital theory as it applies to general on-the-job training. Let us consider a woman’s investment decision. She will compare the experience-earnings profile she can expect if she takes a job with no training (NN’) to the profile she can expect if she receives general training (TT’). In this case, there are costs28 and the firm will bear the decline in output for the period of training only if she accepts lower wages at that time, a wage below what she could obtain elsewhere. This lower wage corresponds to her productivity to the firm during the 28 Direct costs would be the expenses for instructors or for th e material used in the training. Indirect costs result when her coworkers or the supervisor transfer their attention from daily productio n to training activities. Ex p erience Costs Gross Benefits N T 0 J N’ T’ P E a r n i n g s ($) L M Training Period Post-Training Period

PAGE 48

36 training period. The area TNJ represents th e costs of the general training during the period OM. As she becomes more skilled, her earnings increase along with her productivity. After the training period, they surpass what she could have earned without training. Assuming a total OP years of labor market experience over her work life, her gross benefits will be the area JT’L. As in the case of formal schooling, she is likely to undertake the investment if the present value of the gross benefits exceed the costs. Thus, the human capital model tells us that not only formal education but also onthe-job training leads to incr eases in productivity (Walter Oi, 1962). People who have the ability to learn quickly (usually those who are better-educated) are those most likely to seek out, and be presented with, training oppor tunities. They tend to quickly select the ultimately highest-paying jobs where much learning is required and thus use their abilities to the greatest adva ntage. Consequently, they are most likely to enjoy greater monetary returns on their human capit al investments during work lives. According to the human capital mode l, women who follow traditional gender roles such as child-rearin g and home production activities will tend to acquire less valuable on-the-job training because of thei r weaker attachment to the labor market (Becker, 1985). Two important implications from the analysis of the firm-specific training will help us to understand why wome n earn less than men over the work life, and, why women are less attached to the labor market. First, as discussed above, a relatively permanent attachment is likely to develop between the firm and the specifically trained worker. Such workers are less likely either to quit or to be laid off their jobs than untrained or generally trained workers. Sec ond, because employers pay part of the costs of firm-specific training, they will be con cerned about the expected employment stability

PAGE 49

37 of workers hired into jobs where such tr aining is important. So women expecting to interrupt their work lives are less likely to be offered or to seek out opportunities to engage in on-the-job training. Accordingly, they will be less likely to be attached to the labor market and to enjoy higher earnings si nce they have less to lose by dropping out. Conversely, as more women are employed in jobs with trai ning opportunities, the opportunity cost of work-force in terruptions is increased and their labor force attachment is further reinforced.29 As Figures 8 and 9 suggest, earnings w ill increase with experience for workers who have invested in training because a wo rker’s productivity is augmented by such training. Finally, the human capital model predicts that recent increases in the labor force participation of women, especially of ma rried women of childbearing age, will cause dramatic changes in the ac quisition of schooling and tr aining by women since the expected return on their investments w ill be greater. 29 This situation has been observed during recent decades (Blau et al., 2000).

PAGE 50

38 Chapter Three Literature Review An analysis of trends in labor ec onomics throughout the world reveals that sustained increase in women’s participation in the labor force during the last century, particularly during its second half. This fact has stimulated considerable interest in the economic analysis of a woman’s decision to work. The pioneering studies of Jacob Mincer (1962) and Glen G. Cain (1966) in the United Stat es have served as a theoretical and empirical foundation for nume rous studies of female labor force participation. This chapter provides a brief review of the empirical evidence about female labor force participation in the United States, in some other developed countries, and in developing countries of Latin America, with special emphasis on Venezuela. 3.1 Women’s Labor Force Participation in the United States Women’s labor force participation in the United States showed a tremendous upward trend during the last century. In 1900, only 20 percent of all women worked for pay. Less than 6 percent of all marri ed women older than 15 were employed.30 By 1930, the figure had risen to 26 percent for all women, but before 1940 the labor force participation rate of married women was s till only 14 percent (Dora L. Costa, 2000). In 1945, after the social and economic disrupti ons caused by two world wars and the Great 30 Those who did work came from predominately working-class families (Costa, 2000).

PAGE 51

39 Depression, only 30 percent of all women were in the labor force. Abundant empirical literature documents the rapid increase in the labor force participation rate of married women after World War II. By the century’s end, the labor force participation of all women older than 16 years old had risen to 60 percent, and among married women, to 62 percent. During the 1990s another shift occurre d in the composition of the female labor force: this time the group of single mothers with young children increased substantially while that of married women sl owed down (Blau et al., 2002). Given the enormous social and political significance of this increase in the percentage of wome n working for pay,31 especially among married women, many scholars from various disciplines began to i nvestigate the reasons for such behavior. In the remainder of this section the most pr ominent empirical studies regarding women’s labor force participation in the United Stat es, particularly after World War II, are reviewed.32 Back in 1962, Mincer analyzed the va riation in labor for ce participation of married women in 57 large northern standard metropolitan statistical areas (SMSAs) in 1950. His original study used a single equati on model of lifetime female labor supply. His model assumes that women choose levels of market time on the basis of “permanent” wage rates and income.33 He points out three factors that influence the timing of female 31 During the twentieth century, the rising labor force participation of women increased the aggregate labor force participation rate of 25 to 44 year-o lds by 50 percent (Claudia Goldin, 1989). 32The earliest studies, those in the1950s to the mid-1970s, an d historical studies, used onl y aggregate data or macro data. However, the empirical analyses since the mid-1970s have used micr o data. Many econometric developments of the 1970s were stimulated by the new av ailability of data from household surveys, both cross-section and panel, which contained information on relati vely large numbers of individuals (Robert Moffitt, 1999). 33He introduces the notion of differential labor supply responses to permanent and transitory wages rate and incomes, and uses this notion to reconcile, in part, the discrepancy between time series and cross section estimates of female labor supply functions

PAGE 52

40 labor force participation: lifetime variation in opportunity cost due to the presence of children, unemployment of the spouse, and general business cycle fluctuations.34 Mincer hypothesizes that an incr ease in family income, ceteris paribus has a positive effect on leisure time, but may also indirectly affect the allocation of work time between home and market.35 His results support his a priori expectations: wives’ wages have a strong positive effect on labor force participation while the husbands’ incomes have a negative but weaker effect.36 He also reports that high unemployment tends to discourage labor force participation. From the 1919-29 decade to the 1949-59 decade, Mincer concluded that changes in family income and in the wi fe’s wage account for at least 70 percent of the increase in labor force participation of married women.37 Following Mincer’s lead, researchers began to further identify important characteristics associated with married women’s labor supply. Several studies applied his conceptual framework to cross-section da ta. The most comprehensive statistical economic study was conducted by William G. Bowen and T. Aldrich Finegan (1969), which consisted mainly of cross-section regre ssions to estimate models of the supply of 34 He develops the conceptual framework that the market wage influences not only the allocation of time between market work and leisure, but also between work in the market and work in the home. 35The relative income elasticities of home-produced vs. market-produced goods wo uld determine the strength of this effect. 36 June A. O’Neill (1981), using aggregat e time-series data and li near equations, finds resu lts that support Mincer’s basic model: the positive effect of the women’s wage rate s and the negative effect of family income on women’s allocation of time to the market. 37 Glen G. Cain and Martin D. Dooley (1976) attempted to improve the specification of the model by using a threeequation system in which wives’ labor fo rce participation, fertility, and wages are jointly determined, a formulation based on the presumption that these variables are e ndogenous. The Cain-Dooley results for 1970 do not differ substantially from those of Mincer with respect to the la bor supply function. The wage and income coefficients are mostly significant, and the point estimat es of elasticities are large--around +2 and -1 for wages and husband’s income, respectively. Thus the results support th e prevailing economic hypothesis that the wage effect on labor supply will be positive and that the income effect will be negative.

PAGE 53

41 married women in each of three decennial Census years: 1940, 1950, and 1960. They use ordinary least squares regression technique s on data from 100 Standard Metropolitan Statistical Areas (SMSAs). Their model explai ns the labor force par ticipation of married women in an area in terms of four variables representing its labor market characteristics: the unemployment rate, the wife’s expected mark et wage rate, the relative supply of adult females, and the relative demand for female la bor based on the industrial structure of that area. They also include some variables to control for differences in the socio-economic characteristics of households in the areas: median level of husband’s income, proportion of black wives, proportion of families with young children, median educational level of adult females, and mean level of family nonlabor income. They find that wives are becoming less sensitive to all four of the labor market variables. The reason is that, as the labor force of married women grew over th e years, the proportion of older, more experienced, work-oriented wives became larger These wives exhibit less sensitivity to the market wage level and to the rough measures of competition for available jobs.38 Many important aspects of household be havior involve choices among discrete alternatives. Recognition of this fact in the 1970s led to the development of statistical models appropriate to the anal ysis of such “quantal respon se” problems in cross-section 38 Weaker results are found in Judith M. Fields (1976) who makes a comparison of intercity differences in the labor force participation rates of married women in 1970 with those of 1940, 1950, and 1960. She reports that during the tenyear period from 1960 to 1970, the married female labor force continued to grow, from 30.5 percent to 40.8 percent of the total civilian labor force, contributing nearly half of the total increase in the labor force during the decade. Her empirical work compares the earlier Census regressions for SMSAs reported by Bowen and Finegan (1969) with a similar model, applied to 1970 data. She finds that the overall pattern of result s indicates that by 1970 the model has lost much of its explanatory power. The nine independent variables together explain only from 37 to 58 percent of the variation in wives’ labor force participat ion rates among SMSAs. This result could reflect a real change over time in the labor supply function, if, as Fields suggests, women were significantly changing their work role orientation. Alternatively it could reflect a change in the correlation matr ix of the independent variables or other underlying statistical problems.

PAGE 54

42 data.39 Yoram Ben-Porath (1973) proposes a mo del in which, aside from transitory factors such as children and income variation, the timing of participation over the life cycle is random. Assuming that all consumers work at some time during their life cycle, regressions of participation on mean values of wages and inco me yield coefficients that may be interpreted as estimates of Hick s-Slutsky substitution and income effects (Nicholson, 1992, pp. 136-7). A recurring discordance between estimate s of substitution effects of wage rates on labor supply obtained from cross-section anal ysis of data and annual hours of work data stimulated the important research by H. Gregg Lewis (1968) who, with Ben-Porath (1973), demonstrated that the la bor force participation decision at any age is a discrete decision and that estimates of labor force pa rticipation equations produce parameters that are conceptually distinct from estimated pa rameters of hours of work functions. Both papers ignore the focus on the life cycle that is implicit in Mincer’s work. In the LewisBen-Porath model, participation and hours of work at any age are generated from a concave utility function defined for that age. By changing the nature of the preference function from that implicitly utilized by Min cer, they demonstrate that participation and hours of work equations are not as closely related as they would be if Mincer’s assumption of perfect substitutability between leisure at different ages were accepted. Ben-Porath and Lewis implicitly ignore all co mponents of intertemporal substitution. The Ben-Porath-Lewis papers were a stimulus to later work by James J. Heckman (1974) who formulates a model of annual labor force pa rticipation, annual hours of work, and wage rates that explicitly models the interrelati onship between hours of work and labor force 39 For a survey of developments in this field, see Daniel McFadden (1976)

PAGE 55

43 participation. The probability that a woman works is estimated from a common set of parameters. His statistical proce dure extends the Tobit procedure40 to a simultaneous equations system. His methodol ogy differs from the Tobit m odel in that it allows different parameters to affect the prob ability that a woman works than the ones influencing her hours of work. The method allo ws him to utilize an entire sample of women, whether or not they work, to estimate the functions determining their wage rates, probabilities of working, and hours of work. Th ese parameters allow an estimation of the value of time for non-working women, and the wa ge rates they would face in the market. James Heckman and Robert J. Willis ( 1977), stimulated by Ben-Porath’s (1973) work, apply a sophisticated methodology for the tr eatment of panel data to the labor force participation of married women. Assuming th at response probabilities are governed by a beta distribution, they derive a generalization of the cross-section logit model to enable it to deal with sequences of discrete events in panel data. Using a beta-logistic model,41 they find that the distribution of labor fo rce participation probabilities is U-shaped, indicating that most women ha ve probabilities near zero or near one. Another way of expressing this phenomenon is “persistence,” th at women who particip ate at one age are more likely to participate at future ages, as is found in Kim B. Clark and Lawrence H. Summers (1982). Late in the 1970s, James J. Heckman ( 1978), after reviewing the results of the latest research on the life cycle labor supply of married women, presented the first of a class of dynamic models of labor supply. His work attempts to merge two interpretations 40 William H. Greene, 2003, p. 764. 41 They call their model a beta-logistic model since under a plausible parameterizati on of this distribution, they derive a likelihood function of the conventional logit mo del in the case of cross-sectional data.

PAGE 56

44 of the coefficients of wage rates and unearned income derived from labor force participation regressions. The first interpre tation, which stems from Mincer’s (1962) pioneering study and subsequent studies of Cain (1966), Marvin Kosters (1969), and Orley Ashenfelter and himself (1974), relate s to a life cycle mode l of labor supply and interprets the estimated wage and income coefficients as estimates of substitution and income effects. However, the same interpre tation is given to the coefficients obtained from hours of work regressions. The sec ond approach considers the labor force participation decision at a point in time as a discrete decisi on. In this sense, wage and income coefficients estimated in participati on equations are conceptu ally distinct from those estimated in hours of work functions.42 This view ignores the fact that most consumers have ample opportunity to substitute time and goods over the life cycle, and to invest in human capital.43 Heckman claims that, in gene ral, one cannot use the crosssection mean to estimate the probability of any sequence of labor force participation decisions over the life cycle if there is any unobserved hete rogeneity in the population. Finally, the author presents a dynamic model of labor force participation that attempts to merge these two traditions. His model can be used to interpret the interrelationship among the various dimensions of labor supply an alyzed in the literature and can also shed some light on certain empirical findings that rigorously analyze one-period models of labor supply. James J. Heckman and Thomas E. MaCurdy’s (1980) study is an extension of the Heckman (1974) analysis to a life cycle mode l of married women. The work described in 42 Labor force participation regressions describe “corner phenomena” and do not estimate “interior solution” HicksSlutsky income and substitution effects, although they estimate parameters of the utility function of consumers (Heckman, 1978). 43 Following this tradition are Ben-Porath (1973) and H. Gregg Lewis (1977).

PAGE 57

45 this paper also extends the dynamic analyses of labor force participation by Heckman and Willis (1977) and Heckman (1978) to a stru ctural dynamic model that accommodates both hours of work and labor force participa tion decisions in a uni fied framework. The model is a simple computed fixed-effect Tobi t model suitable for analysis of panel data which they estimate on eight years of panel da ta drawn from the Michigan Panel Survey of Income Dynamics. Their empirical work re futes the implicit assumption maintained in previous work that non-market time at one age is a perfect substitute for non-market time at any other age. The paper considers th e meaning and measurement of labor supply responses to “permanent” and “transitory” in come and wage rates in a model of decision making under perfect certainty without credit constraints. They find empirical evidence consistent with the permanent income hypothesis but no evidence of a labor supply response to “transitory” income variation am ong married women. Their empirical results agree with the prediction of the theory: labor supply is inversely related to lifetime wealth measures; children affect lifetime labor supply decisions; and future values of variables determine current labor supply decisions. Claudia Goldin (1983a) studied married women’s labor force participation from the perspective of their economic roles. She writes that, although change in the labor force participation rates of married women did accelerate after World War II, many of the preconditions for this expansion had been se t decades before. The education, household roles, the occupations of single women, and the fertility behavior of married women had a lasting impact on their later response to economic factors. She uses a life-cycle approach to understanding cha nge in the economic role of married women. She produces a matrix of cross-section and time-series labor force participation rates by marital status,

PAGE 58

46 age, race, and national origin from 18 90 to 1980 and covers cohorts born from 1816-1825 to 1946-1955. Her empirical resu lts indicate that long-term ch anges in the economic role of white married women have been the result of three sets of f actors: cohort-specific effects, primarily education and fertility; po int-in-time factors, wages assumed to be exogenous and the unemployment rate; and a time trend, which pr obably proxies longrun changes in the structure of the economy such as the growth of the service sector. Goldin (1991) again studies married women’ s labor force participation during the 1940s and 1950s, focusing this time on the role of World War II in the rise of women’s employment. She uses two retrospective su rveys conducted in 1944 and in 1951, from a study directed by Gladys L. Palmer (1954) with the assistance of the U.S. Bureau of the Census. She finds that the 1940s were the turning point in married women’s labor force participation, and more importantly, more th an 50 percent of the women working in 1950 had been working in the 1940s. She argues th at various social c onstraints may have inhibited the work activities of married women prior to 1940. If the impediments to economic change were partly ideological, then a major break with the past, such as that affected by war, could have redefined economic roles. She concludes that the war had far less direct influence on female la bor supply than was believed. Robert T. Michael (1985) discusses three independent inquiries into the consequences of the rise of women’s la bor force participation during the period 19501980. He uses cross-sectional differentials in female labor force participation by characteristics including age, educational at tainment, marital status, and (among married women) the proportion with younger children. He finds that the changes in women’s labor force participation over th e three decades were not unifor m in terms of age, marital

PAGE 59

47 status, or educational composition. The new employees in the 1950s were predominantly older, married, and relatively less educated while in the 1970s they were younger, less likely to be married, and far better educated. However, although the compositional shift in the population by age, marita l status, and the presence of young children seem to have had almost no influence on overall female la bor force participation, the increase in educational attainment has contributed about one-quarter of the rise in labor force participation. Finally, the differences in labor force participation among groups as defined by age, marital status, presence of young children, and education are far less pronounced in 1980 than they were in 1950. James P. Smith and Michael P. Ward (1985) investigate the reasons for the growth in the female labor force in Ameri ca during the twentieth century. They share the opinion of Goldin (1983.b) that the remarkab le transformation of American women at work cannot simply be viewed as a result of ch anges in attitudes or in labor markets that have been taking place exclusively since Worl d War II. On the contrary, they find that the rise in market participation by married women occurred throughout the century and not simply after World War II. The temporal correspondence between the rapid rise in labor force participation and in their educati on, does suggest that American schools were important in transforming the role of Amer ican women at work. They found that the period of rapidly increasing re lative female wages predates this increase in women’s education, so other events were clearly stir ring within the labor market. They argue that the clerical sector opened up a whole new set of jobs that presumably lessened the conflict between work and ma rriage. Regarding the effects of World War II on women’s

PAGE 60

48 labor force participation, the authors allo w that its role has been controversial.44 They claim that only marital status matters: being married acts as depressant on labor market activity. They conclude that the longer-term growth in the female labor force reflects something far more fundamental than the demographic composition of the population. Their study also provides evidence of women’s labor behavior after World War II using formal statistical models for the pe riod 1950-81, when improved time-series data were available. They disaggregated time-serie s data across the period. Over this 31-year time span their observations cons ist of mean values of labor supply variables at each year of age (annual hours worked by wome n, annual hours worked by working women and weekly labor force participation rates), e ducation, potential work experience, weekly wages, and fertility rates.45 The data are arranged in the form of a set of life-cycle histories for individual birt h cohorts. They found that alt hough the work effort of the typical woman has risen a great deal si nce 1950, the amount of labor supplied by a randomly selected working woman has scarcel y changed. The discrepancies between the two annual hours series indicate that much of the expansion in female work involved labor force participation decisions. They find w eekly labor force participation exhibits the largest across-cohort increase in labor supply. They also find that rising real wages accounted for 60 percent of the to tal growth in the female labo r force; and, that half of this wage effect in expanding labor supply was the fertility-reducing consequence of a higher wage. 44 For instance, Clark and Summers (1982) and others view th is event as a catalyst that permanently altered women’s view of their appropriate labor market roles. In contrast, as discussed above, Goldin’s (1983a) empirically based study reports only indirect eff ects of the war on female labor force participation. 45 For the subperiod 1967-80, they used Current Population Surv ey (CPS) micro files to calc ulate means at single years of age. Over the subperiod 1950-66, CPS published tabl es on distribution of weeks worked and income.

PAGE 61

49 In similar research, Zvi Eckstein and Kenneth Wolpin (1989), using data from the National Longitudinal Survey mature women’ s cohort, estimated a structural dynamic model of married women’s labo r force participation and fe rtility in which wages are stochastic and work experience or cumulative participation is endogenous.46 The basic feature of their model is that labor market participation affects future wages, which in turn affects future labor force participation. Susan Elster and Mark S. Kamlet (19 90) model labor force participation of married women from a sociological perspe ctive. They examine whether traditional economic variables have a differential influe nce across social groups (defined in their paper by broad occupational a nd age classifications). They also study whether “income aspirations”47 have a differential influence across th ese same social groups. Data for the study were drawn from the U.S. Bureau of the Census, 1980 and public-use microdata sample for the Pittsburg Standard Metropo litan Statistical Area (SMSA), 1983. Results from their logit equations indicate that individuals’ responses to particular influences such as education, age, past marital history, fertility, and income aspirations, differ across social groups. It follows from this that su ch differences influen ce married women’s labor force participation behavior. 46The model is contained in the class of model that describes the life-cycle capital accumulation process with endogenous labor supply such as Yoram Weis (1972) and Heckman (1976). It is closest in spir it to that of Weiss and Gronau (1981). 47 It is a measure of the relative income. People’s well be ing depends not only on the absolute level of income and consumption but also on the individual’ s aspirations. The determinants of inco me aspirations have been empirically studied. The econometric results show that income aspirations increases with personal inco me and they are related to the aggregate income in the community. Particularly, a higher average income in the community increases people’s levels of aspirations; a nd, the estimated effect are larger for people w ho interact with other community members (Alois Stutzer, 2003).

PAGE 62

50 Claudia Goldin (1994) studies the re lationship between women’s labor force participation and economic development. She explores the hypothesis of the U-shaped female labor force function for more than one hundred countries including the United States,48 using data from the work of many other researchers.49 She examines why the change in the dominance of the income or the substitution effects takes place and why the U-shape is traced out. She asserts that the rising portion of the U has dominated the literature on female labor force participati on in the United States and many developed countries.50 The change of the function from the downward portion of the U to the rising portion, she points out, holds the key to why wo men enter the labor force at higher stages of economic development and why their soci al, political and legal status generally improves with economic progress. She explains that the increase in the education of females relative to males as educational resource constraints are relaxed, and women’s increased ability to obtain jobs in the white -collar sector after sc hool completion were the main reasons for this change. She concludes that women’s increased education and their ability to work in more prestigious occupatio ns both increase the substitution effect and decrease the income effect. As the substitution effect begins to swamp the income effect, the upward portion of the U is traced out, and women’s labor force participation enters the modern era.51 48 Other researchers studying this hypothesis are John Du rand (1975), George Psacharopoulos and Zafiris Tzannatos (1989), and T. Paul Shultz (1991). 49 The education data are from Robert Barro and Jong-Wha Lee (1993), the GDP/capita ( 1985) data are from Robert Summers and Alan Heston (1991), and the female labor force participation rates are from the extensive United Nations WISTAT collection (U nited Nations 1992). 50 Richard Layard and Jacob Mincer (1985) confirm the relationship across a variet y of developed economies. 51 Another relevant empirical study with pretty much si milar results is Kristen Mammen and Cristina Paxon’s (2000) study.

PAGE 63

51 3.2 Women’s Labor Force Participation in Other Developed Countries Most developed countries experienced si zable increases in women’s labor force participation rates since 1960, a nd in most countries this chan ge was primarily a result of the changes in the labor market activity of married women. As the body of literature about female labor force participation in th e United States has grown, the same economic models developed for studying the United Stat es have been used to analyze the labor force behavior of women in other countries. Recent trends in women’s labor force participation in the United States have also been observed in other industrialized countries. 52 Ehrenberg and Smith (2000) examine the trends in women’s labor force particip ation among women ages 25 to 54 in Canada, France, Germany, Japan, Sweden and the United States in 1965, 1973, 1983 and 1997. They use data from the Organization for Economic Co-operation and Development, Labor Force Statistics (Paris: OECD, various dates). Th ey find that the fraction of women in the labor market in all of these c ountries, on average, increased from half or less in 1965 to approximately two-th irds or more in thirty years.53 Although they find some differences in trends across countries, it is likely that common factors such as changes in fertility, educational attainment labor market opport unities, and social attitudes are influencing labor supply trends in the industrialized world (Blau et al., 2002). Constance Sorrentino (1990), points out factors explaining cross-country differences, such as the availability and amount of family leave and whether or not it is with pay, the availability of publicly funded day care, th e design of tax policy, and 52 For a survey of trends between 1890 and 1980, see Mark Killingsworth and James Heckman (1986); for the period 1960-1980, see Mincer (1985). 53 Borjas (2000) finds nearly similar results (p. 53).

PAGE 64

52 variations in wage structure.54 In the remainder of this section, I will examine some of the most relevant empirical studi es from other developed c ountries, roughly in alphabetic order. Robert G. Gregory, P. McMahon and B. Whittingham (1985) study women in the Australian labor force. They find that in creases by married women are particularly significant: about 90% of the increase in women’s labor force participation can be attributed to women employed pa rt time. They note that Au stralian time-series equations are subject to structural instab ility, and that the estimated wage coefficients are rarely significant. They suggest that these severe difficulties arise because the female real wage has not played a market-clearing role and th ere has been excess supply of female labor, not adequately measured by the employment ra te as officially defined. Consequently, labor force participation ra tes do not measure points on th e labor supply curve. They conclude that to explain cha nges in labor force participati on, emphasis has to be placed on the demand side of the labor market, par ticularly the mix of full-time and part-time jobs. They also generate cross-sectio nal results based on the 1976 census—a period during which they believe job rationing was part icularly important. These equations give results similar to those of other developed countries, but fail to ad equately predict the time-series variations of labor force particip ation, which are also subject to structural difficulties. Michelle Riboud (1985) presents a comp rehensive study of France. She uses cross-sectional data to study married women’s labor force pa rticipation and attempts to use economic analysis and methods of statistical inference to interpret the phenomenon of 54 For further discussion of cross-country trends, see Costa (2000).

PAGE 65

53 the increase in women’s labor force partic ipation over the 1965-85 and earlier periods. An estimation of the market wage function s hows that level of edu cation, experience and tenure are important variable s for explaining differences in wages and that women’s withdrawal from the labor market depre sses potential wage of fers. Moreover, she estimates a labor force participation equation using a logit model; and also an alternative method of analysis obtaining Ordinary Least Squares estimates, using relative participation (RELP)55 as an endogenous variable. Discrepancies are found in both methods, reflecting that the effect of schooli ng (via wage) on labor fo rce participation has been rising over time. Using times-series anal ysis for the same period of time she shows that changes in male and female earnings and unemployment rates explain much of the trend in labor force participation. Finally, she uses the results of the analysis of women’s labor force participation based on cross-sec tional data to predict changes for 1965-1975 and 1975-1985; these predictions are compared with observed changes. She concludes that the same model of decision-making ba sed on a comparison between the value of home time and earnings potenti al in the labor market e xplains both earlier and recent historical trends. Wolfgang Franz (1985) analyses female la bor force participation in Germany. He uses the Tobit procedure, wh ich allows him to estimate la bor supply functions including both hours worked and labor force participa tion in a cross-section analysis based on individual data. He found that it is nece ssary to distinguish among women by marital status: while labor force participation of young single women decreased substantially, married women have a higher labor force partic ipation rate in the 1980s than in earlier 55 Women’s share of the labor force.

PAGE 66

54 years. The Tobit estimates show that labor supply increases with higher education and with vocational education, a nd also if the husband is self-employed. As expected, the income of the husband has a negative impact on the woman’s supply of labor. The presence of children reduces labor supply: the younger the childre n, the more labor supply decreases. In general, the author finds that labor force part icipation of married women increases slightly until the age of 28, and then it declines monotonically. Finally, he found foreign-born women work more than German women do. Ben-Porath and Gronau (1985) study the tre nds in the labor force participation of women in Israel during the period 1955-1980 us ing data from the Central Bureau of Statistics of Israel.56 The authors report that the labor force participation of Jewish women in Israel increased between 1955 a nd 1980, accelerating in the 1970s; two-fifths of women were in the labor force by 1985.57 The sharpest rise was among mothers aged 25-44. Their main finding is that schooling ac counts for most of the change in the labor force participation rates. Moreover, the differe ntial in participation by marital status has sharply narrowed and the lif e cycle effects have been transformed: The M-shaped age/labor force participation profile has been replaced by an inverted U with delayed labor force entry due to prol onged schooling and more continuous participation in market work throughout the childbearing period. In compatibility between child rearing and market work has been reduced by the increased availability of part-time work and 56 Tapes of some of the Labor Force Surveys and the Consumer Expenditure Survey came from the data archives of the Faculty of Social Sciences. 57This figure was roughly the same as in such diverse countr ies as Germany, France, and Singapore. However, at the time this was lower than labor force participation in the S candinavian and Eastern European countries (55-60 percent), the United States, Japan, (43-47 percent) and, surprisingly, Portugal (46 percent). Labor force participation in Israel was, however, higher than is some of the smaller or Sout h European countries where it was less than 30 percent in Ireland, Spain, Italy, and the Neth erlands) (Ben-Porath, 1973, table 2).

PAGE 67

55 increased reliance on day-care services. They find that the increased employment of women is concentrated in the service industr ies, mostly in the public sectors, and is accompanied by some decline in the relative wage of highly educated women. Children (and particularly preschool children) are the st rongest deterrent to a mother’s labor force participation, which effect does no t seem to weaken over the period. Daniela del Boca (1988) finds that the Italian pattern of the female labor force participation shows a mild U-shape: women’s labor force participation rates fell as the size of the agricultural sector declined and th en rose as women’s educational levels rose and as the service sector and manufacturi ng industries that employed women became more important.58 Ugo Colombino and Bi anca De Stavola (1985) attempt to develop a behavioral model of female labor supply in It aly. The model of labor force participation is estimated with cohort data and takes in to account not only variables changing during each cohort’s life cycle but also invariant factors summarized by cohort fixed effects. In a second stage these effects are regressed over a set of indicators that are meant to reflect variables unchanging over the cohort’s life cycle and conditi oning factors at the early stages of the working life. The results suggest that the flat female participation rate profile (from both a time-series and a cross-section perspect ive) is presumably produced by economic incentive effects th at counterbalance each other.59 However, with the recent 58 Costa (2000) reports that Ital y stands out as having the consistently lowest labor force participation rate for women. 59 The authors found the following main factors: 1) a negative trend captured by the co efficients of cohort and (cohort)2; 2) a very powerful positive female wage effect; 3) an equally (if not more) powerful negative male wage effect; and 4) significant positive inte ractions, between cohort and age and be tween cohort and presence of children < 6.

PAGE 68

56 rise in cohort-specific earni ng power, they identify a decrea se in the work-disincentive effects of aging and of children under 6. Haruo Shimada and Yoshio Higuchi (1985) study Japan. Their work reviews statistical data centering on female labor force participation and household behavior, using a common format for international co mparison, and estimates income and wage elasticities from female labor supply equatio ns. They suggest that analysis of the aggregate female labor force is misleading because it mixes heterogeneous groups with sharply different behavioral pa tterns. They find that the pattern for total female labor force participation is stagnation,60 a distinct contrast to many advanced economies (e.g., the United States, Canada, the United Kingdom, Australia, Sweden, and Germany) where the labor force participation of wo men rose sharply during that period.61 However, the postwar increase of female wage and salary workers as a percentage of the female population has been sharp and exhibits a steadily rising trend, more or less comparable to that in the United States and European countries in recent decades. Although the historical sequence of events in Japan doe s not seem to fit the sequence of changes logically anticipated by the human capital th eory, the compounding in fluences of social and institutional factors that affect the behavior of house holds must be taken into account. Clearly, in the United States and countries with a similar distribution of the labor force by employment status, the labor fo rce participation decision may be treated as the choice “to work or not to work.” However, if individua ls regard the decision to enter the labor force 60 Labor force participation rates for all women went down slightly from around 57 percent in the mid-1960s to a low of around 52 percent in the mid-1970s; it increased somewhat afterward to reach 56 percent in 1981. A similar pattern occurred for married women, although lower by approximately 10 percentage points. 61The Japanese pattern is also curious in that the trend was reversed around th e mid-1970s when the Japanese economy suffered the depressing effect of the first oil crisis.

PAGE 69

57 as an employee as being distinct from the c hoice to enter the labor force as a family worker, then economic models of labor force participation, which treat these choices as identical, will incorporate a specif ication bias (M. Anne Hill, 1983). M. Anne Hill (1989) produces another s uggestive paper that takes into account the “informal sector”62 of Japan. The presence of an info rmal sector of the labor market allows women to engage in economic activities—by producing goods at home for sale in the market, working on a family farm, or working in a small family-run business—while simultaneously caring for children and perf orming other home-related duties. Thus choices of women may be viewed as trichot omous rather than dichotomous: women may choose to work in the formal sector of the la bor market (as an employee), in the informal sector, or they may choose not to work. Accordingly, Hill estimates a trichotomous labor force participation model for a sample of em ployees, family workers, and nonparticipants from the Tokyo Metropolitan Area. A 1975 survey of married women (with husband present) between the ages of 20 and 59 is the database used for the empirical analysis. She found that education and market experi ence were significantly associated with a greater probability of working in the formal sector. In cont rast she found that husband’s income and the number of young children were significantly associated with a greater probability of being out of the labor force. Variables such as experience and the presence of young children increased the probability of being employed in the informal sector.63 Another country with low labor force pa rticipation rates for women is Spain. Feliciano Hernandez Iglesias and Michelle Ri boud (1985) describe tr ends in labor force 62 She assumes that women reported as self -employed or as family workers are engaged in the informal sector of the labor force following Adam Jaffe and K. Azumi (1960). 63My dissertation research will draw heavily on the work of Shimada and Higuchi (1985), and Hill (1989).

PAGE 70

58 participation of Spanish women since 1900. Th ey estimate earnings functions and labor force participation models using a 1979 surv ey of married women. A model including the average experience of married women as an endogenous variable pr oduces estimates that fit times series data. Their findings suggest an increasing effect of education on labor force participation, which confirm Schultz’s (1991) hypothesis of increasing returns to various quality components of the la bor force in developed economies. Finally, Sweden stands out as special case among the developed countries. Costa (2000) reports that Sweden had the highest la bor force participati on rates of women of any country before dipping in the 1990s. Swed en has actively encouraged paid female labor force participation and promoted pronatalism since the 1930s. Moreover, the Swedish system of taxation provides substant ial incentives for dual earner couples, and subsidized child-care in Swed en reduces the negative effect of children on women’s earnings (Siv Gustafsson and Frank Stafford, 1992). Sweden’s recent decline in labor force participation seems to be primarily re lated to its recession of the early 1990s: the same percentage point decline in labor for ce participation since then has been observed among men as well as among women (Costa, 2000). Siv Gustafsson and Roger Jacobson (1985) perform an empirical st udy about trends in female la bor force participation in Sweden. They estimate the parameters of labor force participation equations using individual cross-section data from the th ree standard-of-living surveys done in 1968, 1974, and 1981. Their main findings are that the labor force partic ipation of married women increased from 49.1 percent to 83.5 percent during the decades of the 1960s and the 1970s: and that increases in their own wages, have been by far the most important explanatory factor. Women’s real wages have increased relati ve to their husbands’ after-

PAGE 71

59 tax earnings both as a result of the introduc tion of compulsory individual taxation in 1971, and of dramatically decreased sex diffe rentials in pay partly associated with increased female education. 3.3 Women’s Labor Force Participation in Latin American Countries Unlike in the United States and other de veloped countries, very little empirical work in economics has been done about wo men’s labor force part icipation in Latin America. However, disciplines such as sociology, anthropology, and psychology have contributed to a body of literature about Latin American women. Under the auspices of the World Bank in Washington D.C., George Psacharopoulos and Zafiris Tzannatos (1992) published a collection of studies evalua ting women's employment and pay in this region64 during the decade of the 1980s. They use the empirical results obtained for each of the countries’ studies to draw conclusions about the general charac teristics and trends in women's labor force participati on in the region. Similar techniques65 and comparable specifications are used for all the countries, to allow for an easier comparison of the results. Latin American women overall had a low ra te of participation in the labor market, averaging only 24 percent in the 1950s. Howe ver, it increased to 33 percent by the 1980s. Most of these women were between the ages of 20 and 50 years old. As to the underlying factors that help explain the increase in wo men’s labor participation in the region, the authors conclude that it can be attributed to the economic crisis of the 1950s and 1960s 64 The countries included are Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Ecuador, Guatemala Honduras, Jamaica, Mexico, Panama, Peru, Uruguay, and Venezuela. Workers in these c ountries account for approximately 90 percent of the total labor force in the region both in the 1950s and the 1980s. 65 Most analysis is conducted us ing probit and logit regressions.

PAGE 72

60 which resulted in a more efficient use of female labor than has traditionally been the case, and the expanding employment of women in th e public sector. After analyzing the results of individual studies, the author s found that most agree that the probability of a woman working for pay is greater (1) as they ente r adulthood and up to the age of 40 to 45 years (after controlling for fertility); (2) if they reside in urban areas; (3) the higher their education level; (4) the more general (rather than technical/vocational) their education; (5) the lower their family re sponsibilities (in terms of young children present in the household); (6) if they live in a female-heade d household; and (7) th e lower other income and family wealth. The remainder of this section will summarize empirical findings of individual countries in alphabetical order. Ying Chu Ng (1992) examines the determinants of female labor force participation in Argentina. The author us es data drawn from the 1985 Buenos Aires Household Survey that was conducted by the National Institute of Statistics (INDEC). Aside from personal characteristics, family composition, and educational attainment, economic factors related to the availability of income such as w ealth, household income, and household production demands are also impo rtant. The Argentine labor market is characterized by cyclical peri ods in which labor is either scarce or relatively abundant because, on one hand, there are substantial fluc tuations in terms of domestic and foreign migration, and on the other, the fact th at unemployment and underemployment rates remain relatively low regardless of whether there is an excess or scarcity of labor suggests that there may be a str ong “added worker” effect operating.66 Ng finds that 66 This effect predicts that the labor force of secondary workers, women and/or teenaged children, has a countercyclical trend, that is, it moves in the opposite direction of the busin ess cycle: During recessions th e labor force participation of this group of people will rise, and it w ill fall during expansions. Jairo A. Rive ros and Carlos E. Sanchez (1990) provide

PAGE 73

61 important factors determining women’s prope nsity to work are marital status and presence of children. It is common for marri ed women to withdraw from the labor force during childbearing and when th eir children are young. The highe st probability of female labor force participation is found among wo men ages 25 to 29. Thus, as the youngest cohorts age, she predicts much higher levels of labor force participation in the future. Katherine Scott (1992) studies female la bor force participation in Bolivia. She uses data from the second round of the 1989 Integrated Household Survey (SIH), a biannual survey carried out by the National Sta tistical Institute of Bolivia (INE). The results reveal that 44 percent of the sample of women work for pay.67 However, the definition of “employed women” used in th e study may underestimate the real female work force because unpaid workers in a fam ily business are not counted. In general, women who have lower levels of education th an men are more heavily concentrated in the informal sector (World Bank, 1989).68 Probit estimates of the labor force participation function shows the greatest likelihood of working for pay among women ages 35 to 44 but the probability declines among older women. Unmarried women and heads of household are more likely to work than are married women. Women high school students are less likely to participate in th e labor market than those who are not. In contrast, attending, or having completed a technical school, teacher’s college, or university degree has a highly si gnificant, positive effect on the probability of labor force evidence that this is the case. They re port substantial increases in female labor force participation rates, particularly among women aged 35 to 49 years, during the economic crisis of the early 1980s. 67 Women’s labor force participation wa s estimated at 35 percent in 1987, up fro m the 1976 level of 20 percent (World Bank, 1989). 68 There are some legal regulations that restrict women’s participation in the formal sector of the labor market: they are not allowed to work at night, they ar e not allowed to work more than 40 hour s per week, and the labor code bars women from carrying out jobs considered to be dange rous, unhealthy or hard labor (World Bank, 1989).

PAGE 74

62 participation. Pregnancy has the expected negative im pact: women who were pregnant in a given year had a lower probability of partic ipating in the labor market than women who had not been pregnant. She also reports that language skills also have a significant impact on labor force partic ipation: bilingual69 women participate at a higher rate than women who speak only Spanish. Jill Tiefenthaler (1992) uses a multi-sector model of female labor force participation to study the effects of econom ic and social adjustment programs on the well-being of Brazilian women, compari ng women’s economic opportunities in 1980 with those in 1989. The National Statistical Se rvice collected the data for this study from 70,777 Brazilian households. The author consider s it important to distinguish between the formal and the large informal sector in analyzing the Brazilian labor market.70 Her model considers those who do not work for pay as not participating in the labor market. The formal sector is defined as al l individuals who work for a wage while the informal sector is made up of the self-employed. Results fr om estimating the three-sector labor force participation equation reinforce many of he r hypotheses: the important determinants are those variables that influence the market wage variables that affect the reservation wage, and proxies for the costs of employment across sectors.71 69Aymara and Spanish. 70 She reports that the importance of accounting for the la rge informal sector in many developing countries was recognized over 40 years ago by Jaffe and Azumi (1960). They observed that women engaged in informal or “cottageindustry” work had higher fertility rates than women who worked in the formal sector. Results from several more recent studies, using more rigorous empirical analysis, ha ve supported Jaffe and Azumi’s supposition that women’s costs of labor force participation are not equivalent acro ss sectors. See Hill (1983, 1989) S. K. Smith (1984), David Blau (1984). 71 Morton Stelcner et al. (1992), who also studied the Braz ilian multi-sectoral labor market, show that education is perhaps the most important determinant of labor force status and earnings. More over, it plays an important role in “sorting” individuals among alternative labor force activities.

PAGE 75

63 Indermit A. Gill (1992) studies female labor force participation in Chile using data from the National Socio-economic Survey (CASEN) of Chilean households conducted in 1987. In general, Chile is a re latively developed labor market. However, women constitute only about 28 percent of the labor force. Female labor force participation is less than half that of males. The author inves tigates why, in the face of the rapid equalization of education levels across sexes, female labor force participation rates have not increased to levels observed in i ndustrialized countries. The results of probit estimates for the labor force participation of women aged 14 to 65 years are as follows: higher degrees are positively associated with the probability of labor force participation; the age profile of female labor force particip ation is an inverted U-shape; married and cohabiting women are less likel y to work for pay than ar e those who are single or separated; being head of household is positiv ely correlated with the probability of labor force participation; higher household income (total income of other members of the household) increases the likelihood of working for pay (a somewhat puzzling result). The case of Colombia has been studie d by Eduardo Velez and Carolyn Winter (1992). Women’s labor force participation increased from 19 percent in 1951 to 39 percent in 1985.72 The authors attempt to identify factors that influence a woman's decision to participate in the labor market using data from the 1988 National Household Survey conducted by the Statistics Admini strative Department (DANE) in the largest Colombian cities. They estimate a probit m odel in which the probability that a woman will participate is estimated based on her parent al status, age, education level, the size of the household in which she lives, and her status as head of household or otherwise. The 72 ILO (1990).

PAGE 76

64 probit coefficients show that the probability of participating increases steadily with each additional level of completed education. A larger household has a positive, although small, effect on a woman's decision to work for pay. By contrast, being head of household has a substantial positive impact. As in many other studies, the presence of young children is shown to reduce the probabi lity that a woman will work for pay.73 In Colombia, however, even women with young childr en continue to be heavily represented in the informal sector. Hongyu Yang (1992) studies female labor fo rce participation in Costa Rica. The author finds the major factors that infl uence women’s labor market activity are educational attainment, marital status, fe rtility, other household income, and age. Education has a powerful positive effect on the probability of female labor force participation: more educated women are more likely to participate in the market and are more likely to be employed. Using the result s of probit estimates for female labor force participation, the author predicts the proba bility of labor force participation for each characteristic holding other char acteristics constant at their means. The author found that high school graduates have the highest probability, 54.2 perc ent. Married women are less likely to participate than unmarried women, 17.7 percent versus 40.4 percent. The more children a woman has, the less likely she is to participate in the labor market. A female head of household has a higher likelihood of pa rticipating, 34.1 percent, compared to 22.7 percent probability for a woman who is not head of household. Finally, women who live in rural areas are less likely to participate in market activities. 73 Thierry Magnac (1992), using samples drawn from urban household surveys between 1980 and 1985, finds similar effects.

PAGE 77

65 There are two empirical papers about fema le labor force participation in Ecuador. The earlier analysis was undert aken by Mary Finn and Caro l L. Jusenius (1975) using data for 1966. They find rather low rate s of labor force participation among urban women, around 25 percent, with the highe st rates among women who had completed college (89 percent). On the other hand, single women were more likely to work in the labor market, but earned substa ntially less than working wive s, who also tended to be older and better educated. George Jakubson a nd George Psacharopoulos (1992) use data from the 1987 Ecuador Household Survey that was conducted in urban households in the three largest cities, Quito, Cuenca and Guaya quil, to study the increase in female labor force participation which their estima tes show had increased to 50 percent.74 They also found that more educated women are more likel y to participate in the market and more likely to be employed; marital st atus and being head of household are the most important social determinants of both labor force participation and employment: i.e., wives with working husbands are much less likely to part icipate in the labor market than female heads of household. Women with young children are also less likely to work for pay. Mary Arends (1992) examines female la bor force participation in Guatemala. This country has the lowest ra te of literacy in Latin Am erica, and there is a large schooling gap between men and women. About 40 percent of its population is Amerindians, many of whom do not speak Spanish, and who ha ve little access to social services or to formal labor markets. About ha lf the work force is employed in agriculture, much of it at the subsistence level. The data source is the 1989 National Socio74 The World Bank (1989) reports labor force participation rates among all women in Ecuador have continued to rise, except among females aged 12 to 19 years who are remaining in school longer. Labor forc e participation rates have been particularly high among women between ages 25 and 34 years, and have increased across all marital status categories. Married women increased their labor force part icipation from 16.8 percent to 21.1 percent between 1974 and 1982. The World Bank also reports that 61 percent of fe male heads of household were in the labor force in 1987.

PAGE 78

66 Demographic Survey (ENSD), carried out by the Statistical National Institute. Results from probit equations estimation are that school ing level is an important determinant of labor force participation; that number of ch ildren and marital status have the expected negative and significant impact on participation; that participation peaks between the ages of 30 and 34, dips for women aged 35 to 39 and then rises again; that being head of household increases the probability of particip ation, as does living in Guatemala province and living in an urban area; that a woman from an indige nous group is less likely to participate in the labor market; and that household income has a positive effect on participation, as Gill (1992) reported for Chile. Honduras is one of the poorest countries in Latin America. Carolyn Winter and Thomas H. Gindling (1992) use probit equations to investigate the fact ors that influence a woman’s decision to enter the formal labor market or the informal sector in this country. The data used in their analysis come from the 1989 national survey, the Honduras Household Permanent Survey of Multiple Pur poses (EPHPM). It is assumed that the decision to work and the decision regarding which economic sector to enter (formal or informal) are made simultaneously. Their results are as follows: 1) Holding all other variables at their mean values, the probability of participation increases substantially with each additional level of education completed. However, women with college degrees actually have a lower probability of particip ation than women with completed secondary education.75 2) Women’s labor force participati on rates by age group show the familiar inverse U-shape. Women’s labor force partic ipation peaks between ages 35 and 45 and then declines. 3) Having child ren aged six years or less re duces the probability that a 75 This may reflect, in part, the high levels of unemploym ent among individuals with college degrees since some may become discouraged and leave the labor force.

PAGE 79

67 woman will work. 4) Women are much more likel y to participate in the labor market if they live in urban areas. Diane Steels (1992) analyses women’s labor force participation in Mexico. She focuses on factors that have been shown in pr evious studies to influence the decision to enter the workforce such as age, education level, presen ce of young children, marital status, and household wealth. The probit co efficients show that probability of participation in the labor market decreases as women become older although it remains relatively high even at older ages. Her analys is shows that with increased levels of education, women are more like ly to participate in the la bor market. Steels’ study shows that women in Mexico are actually more likely to participate when there are children in the household, unlike most studies which indicat e that the presence of at least pre-school aged children reduces the proba bility of labor force particip ation. Finally, living in an urban area increases the probability of a wo man's participation in the labor market. Female labor force participation in Pa nama is studied by Mary Arends (1992). Her empirical study uses data taken from the Household-Man Power Survey of August 1989 by the Office of Statistics a nd Census of Panama (DEC). The author comes to the following conclusions from her results: the likelihood of women working for pay increases with higher educati on levels from 10 percent for those with no education to 48 percent for those with over 4 years of unive rsity education; women with children under 6 years of age are less likely to be in the work force: the probability of labor force participation drops from 27 percent for thos e with no children under 6 years of age to 18 percent for women with three children in that age group; la bor force participation peaks between 35 and 39 years of age; those living in an urban ar ea are 12 percent more likely

PAGE 80

68 to work than those living in a rural area; the probability of work ing increases from 20 percent to 57 percent if the woman is head of household; and addi tional workers in the household increases the probability of working. Shahidur Khandker (1992) studies women' s labor market participation in Peru where it increased from 34 to 43 percent be tween 1970 and 1985 in urban areas. He uses the probit estimates to predict the effect of changing certain characteristics holding other characteristics constant at their mean. He finds that women with university rather than secondary or post-secondary diplomas have subs tantially higher labor force participation, and that single women participate in the labor force more than married women (14.9 percent versus 5.52 percent, respectively). Fi nally, the predicted participation rate for women is the highest in Lima (15 percent) followed by other urban areas (9 percent) and rural areas (6 percent). Mary Arends (1992) also studies women’ s labor force participation in Uruguay. She uses data drawn from the 1989 Househol d National Survey conducted by the General Administration of Statistics and the Census ( DGEC), and finds that the female labor force participation rate is 52 percent. Arends presents the results of a simulation testing for each characteristic while holding all other char acteristics at the value of their sample means. She finds that education plays a key role in predicting whether a female works. For example, the likelihood ranges from 28 percent for women with some primary education to 54 percent for wo men with a college degree. Arends finds that labor force participation is higher at all ages than in other Latin Am erican countries. Labor force participation is lower at ages 14 to 19 than at other ages, which is to be expected given Uruguay's high enrollment rates in secondary ed ucation. The number of children also has

PAGE 81

69 a significant negative effect on labor force participation. Being head of household increases the probability that a woman will work for pay from 34 percent to 65 percent. The number of employed persons in the househol d has a significant, positive effect on the probability that a female will be working. 76 The coefficient on household income is negative, as expected, but small. Lastly, liv ing in Montevideo has a small positive effect on the decision to work. 3.4 Women’s Labor Force Participation in Venezuela According to information published by the Central Office for Statistics and Information (OCEI) in Venezuela, women’s la bor force participation has increased since the 1950s, when the rate stayed between 18 percent and 19 percent. By 1971, it had increased to 23 percent. The 1990 Population Censuses show that women’s labor force participation had increased to 30.5 percent. Fi nally, the Household Sample Survey reports that their labor force participati on rate reached 43 percent by 1998. There are only a few studies about women’s labor force participation in Venezuela. In a descriptive study, Mari a Beatriz Orlando and Genny Zuniga (2000) analyze women in the Venezuelan labor market, focusing on their labor force participation and their income. They use aggr egate data from the National Census since 1950. Their results show that Venezuela is similar to other Latin American countries where older women, “cohabitors”77 and those with the lowest level of education 76 The explanation for the sign is not immediately obvious. It may show a kind of family “work ethic” with members preferring to work outside the home (Arends, 1992). 77 “Cohabitors” refers to a specific group of Venezuelan wo men who formerly lived with a partner but who have been abandoned or have decided to separate.

PAGE 82

70 increased their labor force participation, as a strategy to cope with reduced family income. Leonardo Ledezma, Maria B eatriz Orlando and Genny Zuniga (2003) report on the determinants of labor force participati on of women in Venezuela for the period 19802000. Among the factors that influence Venezu elan women’s labor force participation, the authors point out, are income level, educ ation, and the development of political and social institutions. Age is also important: the highest labor force participation, 46 percent, is observed for those between the ages of 30 and 39.78 Two empirical papers inve stigate the labor market behavior of Venezuelan women: Cox and Psacharopoulos (1992) and Winter (1992). Both studies use the same data source, the Household Survey data, for 1987 and 1989, respectively, and the same methodology. Both studies estimate a probit equation for a sample of working and nonworking women. Not surprisingly, results of the papers are quite similar: education has powerful effects on labor force participation as the human capital lit erature suggests. The probit coefficients show that the probability rises steadily with each successive level of education. Cox and Psacharopoulos find that liv ing in a rural area reduces the probability of participating in the labor force by 13 percent, consider able more than Winter, who finds only a 6 percent difference. Other results are related to specific variables used by each of the researchers. Cox and Psacharopoulos find that being a wife or partner reduces the probability of labor force participation by 22 percent, implying th at family responsibilities compete for time 78 Ledesma, Orlando and Zuniga (2003) also claim that the labor force participation of rural women has traditionally been reported at levels much lower than they really are because women consider some of their economic activities part of their domestic chores. However, it is clear that women have been participatin g in the labor force in greater numbers.

PAGE 83

71 spent in the market.79 Being head of household raises th e probability by 23 percent. An increase in the income of other fa mily members equivalent to 15 US$ per month reduces the probability of working by 4 percentage point s. Winter finds that being a mother of young children significantly increases the probab ility that a woman will withdraw from the labor force. And, finally, age is also an important factor. The probability of women working increases steadily starting in the mi d-twenties and peaks be tween the ages of 41 and 45. Low labor market participation rate s among women in their early twenties are consistent with the high enrollment of wome n in this age group in higher education (44 percent). Moreover, many in their ea rly twenties may be having babies. 3.5 The Contribution of My Dissertation Research Psacharopoulos and Tzannatos (1992) collected a series of empirical studies of many Latin American countries. However, none of them attempt to advance the theoretical understanding of issues pertaining to women’s time allocation between home and market work. These studies take analyt ical approaches used during the 1980s to investigate women’s status in the Latin Am erican labor markets, that is —women’s employment and pay. Almost all the studies attempt to explain why Latin American women have lower rates of both labor force participation and pay, compared to men. They assume that there are no innate diffe rences between the sexes that justify the observed gender differences. Thus basically these papers st udy gender discrimination in 79 This is only a partial explanation. Clearly, havi ng a partner’s income raises the reservation wage.

PAGE 84

72 the labor market using the human capital framework.80 The studies of all the countries use similar techniques and specifications in orde r to facilitate a comparison of the results, to find whether there are common patterns and fa ctors at work in the region with respect employment and pay of women. This project is aimed to provide an in-depth analysis of the labor force participation of Venezuelan women duri ng the 1990s, based not only on human capital theory, but also on the theory of allocation of time between home and market work. The more recent data presented in this stu dy will overcome some of the most serious difficulties that Psacharopoulos and Tzannatos faced regarding the quality and coverage of the micro data collected by household surveys in the region during the 1980s. The descriptive study of Ledezma, Or lando and Zuniga (2003), despite the report’s title, is based on previous data from the 1990 census collected by the Central Office of Statistics and Information (O CEI). The empirical works of Cox and Psacharopoulos (1992), as well as Winter ( 1992), use micro data from 1987 and 1989, respectively. Thus, my project will greatly expand the rather small and now-outdated literature by investigat ing factors influencing female la bor force participation in the decade of the 1990s. There will be a signifi cant improvement in both the quality and quantity of data to be used in this study of Venezuelan women. For instance, the two empirical studies of Venezuela use a binary choice model to estimate the impact of the factors that influence women’s labor force pa rticipation. However, in neither study are the unemployed properly identified: Cox a nd Psacharopoulos’ (1992) definition is not strictly comparable with the official one, which, as in most countries, counts both the 80 Gender discrimination is beyond the scope of this study. However, their findings about the determinants of women’s labor force participation are pertinent.

PAGE 85

73 unemployed and the employed as members of th e labor force. Winter (1992), on the other hand, is unable to determine whether indi viduals are unemployed or employed in the informal sector. This is a serious issue, since a large informal sector exists in Venezuela. Consequently, in both studies, th e authors include as participan ts in the labor market only actively working individuals, identified by their positive responses to questions concerning employment status: weekly hour s worked and monthly income. The data available for the last years of the 1990s provide information to properly identify the formal and informal sector, thus a llowing me to apply the methodology used by Hill (1989), to take into account the informal sector.81 Choices for Venezuelan women are: working for pay in the formal sector, being self-employed or working in family businesses in the informal sector, or bei ng out of the labor fo rce entirely. Thus, by applying methods of analysis not previously us ed to new data, this dissertation will offer a greatly-improved, in-depth, and up-to-date investigation into the labor participation decisions of women in Venezuela. In Chapter Four the specific hypotheses to be tested and the estimation procedures are presented more formally. 81 Shimada and Higuchi (1985) also used the same method to study female labor force participation and household behavior in Japan.

PAGE 86

74 Chapter Four Research Design This present analysis represents several distinct advances in the study of the labor force participation of women in Venezuela. Fi rst, improved data will be used to update previous empirical work on this topic and to overcome serious shortcomings of previous studies. The two empirical pape rs about Venezuelan women’ s labor market behavior by Cox and Psacharopoulos (1992) and Winter (1992) described in Chapter Three approached the labor force pa rticipation decision of Venezu elan women as dichotomous: “working” and “not working.” The second advan ce of this study is that a trichotomous decision is considered, adding th e option of participating in th e informal sector. Finally, the labor force behaviors of different de mographic groups of women are analyzed separately. In this chapter th e data and methodology that will be used to analyze women's labor force participation in Venezuela be tween 1995 and 1998 will be discussed. The chapter is organized as follows: Section 4.1 describes in deta il the objectives and hypotheses of this project. Section 4.2 describe s the data sources to be used. Section 4.3 describes the methodology and estimation me thods, and Section 4.4 describes the specification of the models and the descrip tion of variables invol ved in the analysis.

PAGE 87

75 4.1 Objectives and Hypotheses This study will provide an empirical update on labor force participation among Venezuelan women. This effort will help close the existing chronological gap in the literature by examining their labor force pa rticipation throughout the 1995-1998 period. In this chapter the hypotheses are summa rized more formally. An introduction to the methodology used is also included. In Ch apters 2 and 3, we have discussed the economic theory behind women’s labor force part icipation and a review of the literature, in which age, education, family income, and other factors were discussed in terms of how they affected women’s labor force participa tion. The empirical tests in this study are based on the following hypotheses: Hypothesis No. 1: Venezuelan women’s labor for ce participation has increased during the last decades of the last centu ry, particularly during the period 1995-1998. Factors such as age, education, marital st atus, urban residence, geographic location, headship of the household, socio-economic status, nonlabor income, and time influence their decision to work or not to work. Hypothesis No. 2: Venezuelan women have increased their participation in the informal sector during the last decade of th e last century. Factors influencing their decision to work in the informal sector are the same as those impacting their decision to work in the formal sector during this period. Two additional hypotheses will be tested using the subsamples of married women, single women, and women heads of household.

PAGE 88

76 Hypothesis No. 3a: Demographic, geographic, and socio-economic factors influence the labor force participation deci sion of married women, single women, and women heads of household in the same manner. Hypothesis No. 3b: The factors considered in th is study affect Venezuelan women’s decision to work in the formal or in formal sector similarly, whether they are married, single and/or heads of household. 4.2 Data Base This study utilizes micro data from the Household Sampling Survey (EHM) of the Central Bureau for Statistics and Informati on (OCEI), the agency of the Venezuelan government that collects data and generates official statistics. This is a biannual, nationwide survey that measures the characteristics of the Venezuelan labor market as well as other demographic issues such as family composition, housing quality, access to public services and poverty status. The surv ey is conducted using multi-stage sampling; the sample is rotated to avoid refusal while maintaining consistency and representation. Five surveys will combine to produce pooled cr oss-sectional data: th e first semester of 1995, and both semesters of 1997 and 1998. For purposes of this study the sample is restricted to woman between 15 and 60 years of age for whom all the specified variables are available.

PAGE 89

77 4.3 Methodology A description of research methodology follows. 4.3.1 Binomial Logit Model To test Hypothesis No. 1 and Hypothe sis No. 3a, the study models the binary choice that a woman is in the labor force (Y =1) or is not (Y=0) during the period. A set of factors such as age, education, marita l status, head of household, socio-economic status, urban residence, geographic areas, nonlab or income, and survey time gathered in a vector X explain the decision, so that Prob(Y=1| x ) = F ( x ) (1) Prob(Y=0| x )= 1F ( x ) (2) The logit model uses th e logistic distribution (.) Prob(Y=1| x )= ) ( 1' x e ex x (3) The probability model is a regression: ) ( )] ( [ 1 )] ( 1 [ 0 ] / [ x F x F x F x y E (4) Since the parameters of the model are not marginal effects, in the logit model, [/] (')[1(')] Eyx xx x (5) To interpret the estimated model82, it is useful to calculate th ese values at the means of the regressors or other pertinent values.83 The appropriate marginal effect for a binary independent variable would be 82 The method of the maximum likelihood is applied to estim ate this model’s parameters as in the multinomial logit model explained below.

PAGE 90

78 Marginal effect=Prob[Y=1| 1 Pr[ ] 1 ), ( Y d d x | ] 0 ), ( d d x (6) Where x( d ) denotes the means of all the other variables in the model. 4.3.2 Multinomial Logit Model In order to analyze labor force behavi or of Venezuelan women (Hypothesis No. 2), and then to compare their behavior among the three groups (Hypothesis No. 3b), the study follows Hill (1983) in using Daniel McFa dden’s (1974) model for discrete choice: the multinomial logit model.84 The model is the standard one-period st atic labor supply framework in which each woman may select among three mutually exclus ive choices: working in the formal sector as a paid employee (indexed p), working in the informal sector85 (indexed f), and not working for pay (indexed n).86 The preferences are defined by a utility function whose arguments are the Hicksian composite of all goods, non-market time, and vector of exogenous variables that aff ect labor force decisions. Rational decision-making is reflected in the maximization of utility subj ect to time and budget constraints. In other words, the woman compares the maximum u tility attainable given each participation alternative and selects that alternative which yields the highest utility. 83 To compute the marginal effects, one can evaluate the expressions at the sample means of the data or evaluate the marginal effects at every observation and use the sample average of the individual marginal effects. 84 The labor force participation decision is modeled most appropriately w ithin a life-cycle context, especially if there is heterogeneity across individuals with rega rd to the propensity to work in either labor market sector. (See Ben-Porath, 1973 and Heckman, 1978.) Unfortunately, in Latin American co untries including Venezuela the panel data required to estimate a life-cycle model are not available. 85 Both Cox and Psacharopoulos (1992) and Winter (1992) re port the importance of the informal sector in the Venezuelan labor force. 86 This specification does not allow for the possibility of work ing concurrently in more than one sector. However, the data include no information on multiple job holding and each person reports only one current employment status.

PAGE 91

79 Formally, let Vji be the maximum utility attainable for individual i if she chooses participation status j = p, f, n and suppose that this indirect utility function can be decomposed into a non-s tochastic component ( S ) and a stochastic component ( ): Vji = Sji + ji (7) Where Sji is a function of observed variables and ji is a function of unobserved variables. The probability that the ith woman selects the jth labor force participation status is then given by ij = jiV Pr[ > kiV for k] , n f p k j (8) or, substituting in from (4-8), ki ij jiS S P Pr[> ji ki for k] , n f p k j (9) If the stochastic components have independent and identical Weibull distribution,87 then the difference between the errors (ki -ji ) has a logistic distribution and the choice model is multinomial logit (McFadden, 1974).88 To estimate the model, it is necessary to specify a functional form of the nonstochastic component of th e indirect utility functionjiS This component is approximated in linear form (i j jiX S'), yielding an empirical specification of the form ) exp( ) exp( ) exp( ) exp(' ' i n i j i p i j jiX X X X P (10) 87 The Weibull distribution has a unimodal bell sh ape roughly similar to the normal distribution. 88 For a complete description and disc ussions of the multinomial logit mode l, see McFadden ( 1974) and Domencich and McFadden (1975)

PAGE 92

80 where iX is a vector of independent variables explaining labor force participation and j is the parameter vector. 4.3.3 Estimation Method of Maximum Likelihood For n alternatives in the multinomial logit model, only n-1 distinct parameter vectors may be identified. This linear de pendence requires the normalization of the parameters, i.e. that 0j, (11) For a comparison among the empirical results, it is useful to ca lculate the partial derivatives of the dependent variables, the probability of entering the paid labor force ()pP and the probability of engaging in family work or self-employment) (fP with respect to each independent variab le. The partial derivatives are X S P P X S P P X S P P X Pn n j k k j j j j j ) 1 ( j, k = p, f, j k. (12) The econometric model is specified so that jS /0 ln kW if k j Therefore, the effect of the wage in each sector operates through the effect on the conditional indirect utility in that sector. For example, ln lnp p p f p fW S P P W P Estimated standard errors of thes e derivatives may be calculated in a straightforward manner using the variance-cova riance matrix of the estimated parameters.

PAGE 93

81 McFadden (1974) suggests several measures of goodness-of-fit for the multinomial logit model. Among them is a likelihood ratio statistic: ( ) ( [ 20L L)], (13) which, under the null hypothesi s that all parameters equa l zero, is asymptotically distributed as a chi-square variate with k degrees of freedom, where k is the number of estimated parameters. L() is the log likelihood evaluated at, the maximum likelihood estimate of the parameter vector and ,0a vector of zeros. An analog to the R2 in a conventional regression is McFadde n (1974) likelihood ratio index, ) ( ) ( 10 2 L L R = LRI (14) 4.4 Specification of the Models In order to test the thr ee hypotheses, two models for di screte choice are described in this section: 1) the two-way choice model or binary logit model and 2) the three-way choice or multinomial logit model. Both models use the same set of variables but since they are testing different hypot heses, the expected signs may be different. A general description of the variables follows. 4.4.1 Dependent Variable Women’s labor force participation (WLFP) is a dummy variable which takes two values for the two-way choice model: 0 = not in the labor force, and 1 = in the labor

PAGE 94

82 force. It takes on three values for the thre e-way choice model: 0= not in the labor force, 1=participating in the informal sector, a nd 2=participating in the formal sector.89 4.4.2 Independent Variables: De scription and Motivation Three main categories of variables explai n women’s labor force participation: 1) individual and demographic factors:90 age, education,91 and marital status; 2) geographic location factors: urban reside nce and region; and, 3) soci o-economic condition factors: head of household, socio-economic status, non labor income, interaction terms (nonlabor income survey dates), and survey dates. Individual and demographic factors: Age Numerous empirical analyses have po inted out that the probabilities of labor force participation of women differ by age. In this study, age is entered in the labor force participation function as a series of five du mmy variables in 10year ranges: 15 to 20, 21 to 30, 31 to 40, 41 to 50, and 51 to 60. Education The effects of education reflect both nonpecuniary factors, such as “tastes” for market work versus work at home, and pecuni ary ones such as potential market earnings. Changes of cultural values in Latin Ameri ca and women’s attitudes toward working for 89 What truly determines the difference between workers of th e formal and the informal sectors is the compliance with laws regulating market work. See more detailed definitions of the formal and the informal sector in Appendix A. 90 Although important, the presence of chil dren is not used in this project since the individual data do not have indicators linking children and mothers. 91 Age and education level dummy variab les are included in the labor force pa rticipation equations as proxies for offered wages.

PAGE 95

83 pay brought an increase in formal educati on for women, and consequently opportunities for better jobs. This variable is entered as a series of five dummy variables for the highest level of education completed, or if a student is currently enrolled at the level: primary education, secondary education, t echnical education and college. Marital status In general, married women have the burden of the domestic chores which limits their participation in the labor market. More over, the earnings of husbands and partners constitute nonlabor income for these women, which reduce their likelihood of participating in the labor market. A set of du mmy variables represents five categories of marital status: single, married, divorced, widowed and “cohabitors.” Women currently cohabitating, i.e. with a partner present, are included with married women. The divorced group includes women who are separated fr om their husbands. The separate dummy variable “cohabitor” re fers to a specific group of Venezu elan women who formerly lived with a partner but who have been ab andoned or have decided to separate. Geographic location factors: Urban The dummy variable for urban residence reflects a mix of demand side and taste or preference effects, which are likely to wo rk in the same direction. An urban area may provide more jobs opportunities and better e nvironment for a woman to perform market work. The variable takes the value of one if a women lives in an urban area, and zero otherwise.92 92 Definitions of urban and rural areas are found in Appendix A.

PAGE 96

84 Region The division of the Venezuelan territory into regions organizes states according to their prevalent economic activities and availabili ty of resources. There are nine regions: Capital, Central, Central-Western, Zulian, Andean, Plains, NorthEastern, Insular and Guayana. Since there are different mixes of economic activities, population growth and densities and migration patterns by regions the likelihood of women’s labor force participation may also differ by regions. A dummy variable for ever y region in the labor force participation equation is included.93 Figure 10 Geographic Areas in Venezuela Regions of Venezuela: Zulian (red), Guayana (pink), Central (blue), Central-Western (orange), Insular (purple), Andean (green), North-Eastern (light green), Plains (yellow). Source: Perry-Castaneda Library. Map Collect ion. The University of Texas at Austin. http//www.lib.utexas.edu/maps 93 The Guayana region was dropped from this analysis because recent development activities do not allow for reliable data collection. Detailed information about the regions are found in Appendix A.

PAGE 97

85 Socio-economic conditions factors: Heads of household Many women in Venezuela are heads of household with a large number of dependents and with few wage earners. Be ing head of household suggests the woman needs to work to support the household94 and indicates the presence of children. This group includes women in all marital status categories. This dummy variable takes on a value of 1 if a woman is head of household and a value of 0 otherwise. Socio-economic status These dummy variables are derived from an index calculated by the Modified Graffar method (Hernan Mendez Castellano and Maria Cristina de Mendez, 1994).95 This index defines socio-economic classes based on specific living condi tions of families. There are five classes: high, medium high, av erage, relative poverty, and critical poverty. For purposes of this analysis the lowest two classes are combined. Nonlabor income Nonlabor income is a continuous vari able that enters in the labor force participation equations to captu re the effects of wealth on the likelihood of working. The Venezuelan currency is the Bolivar. Howeve r, for purposes of this study, nonlabor income is converted into U.S. dollars per month. Although in many studies the husband’s income is considered nonlabor income for the wife, it is not included here, nor is the income of other family members. 94 As mentioned in footnote 90, one impor tant limitation of this study is the difficulty of identifying women with children. A growing group of single women with children has been identified lately, a trend also observed in developed countries. 95 A detailed description of the Graffar-Mend ez Castellano method is found in Appendix A.

PAGE 98

86 Survey date Five dummy variables are used to contro l for the time when the survey was taken, to capture the influence of short-term variati on in economic activity. Interaction terms These continuous variables capture the im pact of the combined effects of two variables: nonlabor income va riable and survey dates. Four interaction variables are considered: nonlabor income 1997-1, nonlabor income 1997-2, nonlabor income 1998-1, and nonlabor income 1998-2. These variables al low nonlabor income have differing impacts for the different surveys. 4.4.3 Testing Hypothesis No. 1 Using the two-way choice model, the full sample of women will be used to regress the labor force partic ipation equation on individual or demographic factors, geographic factors, and socio-economic conditions. Regressors’ expected signs Individual and demographic factors: Age The youngest group 15 to 20 years of age is the omitted category. Women of these ages are probably students and consequently they are more likely to be out of the labor force. Empirical evidence also suggests that the oldest women are more likely to be out of the labor force. I exp ect that the coefficients on th e dummy variables for other age groups will be positive indicating a greater probability of participation in the labor force than the youngest group. Moreover, according to the National Census (1991), Venezuelan

PAGE 99

87 women ages 30-39 showed the highest labor forc e participation rate. Therefore, I expect the highest probabilities of pa rticipating in the labor market for women between 30 to 50 years of age. Education The omitted category in the labor force participation equation is the group of women with no education. We anticipate that the coefficients on all the other education variables will have positive signs and that there will be larger coefficients for higher levels of education. Marital status In this study, the omitted category is singl e women. We expect that the coefficient on the dummy variable indicati ng a woman is married will be negative since her husband or partner provides the income needed to s upport the household. We al so expect negative coefficients on the dummies variable for divorced and widowed women because usually they receive some kind of legal child support and inheritance, respectively. We expect no significant differences in the coefficient on the dummy variable for “cohabitors” compared to single women. Geographic location factors: Urban Positive coefficients are expected on th is variable, reflecting a greater likelihood of market work for women living in urban areas compared to those living in rural areas. Region The omitted category is the North-Easter n region which is characterized by low population density, agricultural a nd oil production activ ities. Therefore, I expect that the

PAGE 100

88 coefficients on the dummy variables representing residence in areas such as the Capital, Central-Western, Zulian and Andean regions with greater populat ions and with more industrial and service activities will be positive an d significant, indicating higher probabilities of women’s participation in the labor force. A priori it is not clear what signs to expect on the coefficients on the dum my variables representi ng the other regions. Variables indicating socio-economic conditions: Heads of household The study expects the coefficient on the dummy variable indi cating that a woman is head of household to be positive and signi ficant, indicating a hi gher likelihood of labor force participation. Socio-economic status It has been suggested that those in the lowest socioeconomic status are less likely to work than those in higher classes due to negative attitudes toward work and lack of ambition. Since the omitted category is pove rty, we may anticipate positive and successively higher probabilities of particip ating in the labor force as women achieve higher socio-economic status. Nonlabor income According to the neoclassi cal theory of la bor, higher amounts of nonlabor income are associated with higher reservation wage a nd a lower the probability of participation. Thus, a negative sign on the coefficien t for nonlabor income is expected. Survey date This study of women’s labor force partic ipation corresponds to the period after the financial crisis of 1994 and the further a pplication of the macroeconomic adjustment

PAGE 101

89 program of the Venezuelan Agenda. Since the omitted category is the 1995 survey, we anticipate that the coefficients of the othe r dummy variables will be positive indicating a greater need of women to supplement lost income due to the economic crisis. Interaction terms These terms are entered in the labor fo rce participation equation to allow the effect of nonlabor income to differ between surv ey dates. It is not cl ear a priori what the signs the coefficients of these variables will be. 4.4.4 Testing Hypothesis No. 2 This study relies heavily on Hill’s (1984) paper in which she emphasized the necessity of considering a third choice with respect to labor force participation, working in the informal sector, in countries such as Japan. Using data from a survey of married women, she found that education and market experience were signif icantly associated with a greater probability of working in th e formal sector. In contrast, she found that husband’s income was significantly associated with a greater probability of being out of the labor force. The number of small child ren increased the probability of being employed in the informal sector or out of the labor force. I will use the three-way choice model on the sample of women between the ages 15 to 60 to test the second hypothesis about wo men’s decisions to work in the formal or informal sector. The informal sector b ecame more important during the period under analysis because of the deteriorating ec onomic conditions and the shrinkage of the industrial sector, which cause d a loss of employment in the formal sector.

PAGE 102

90 Regressors’ expected signs This trichotomous model attempts to e xplain women’s decision to work in both the formal and the informal market sectors. The predicted signs fo r the same group of regressors are discussed below. Individual and demographic factors: Age As with Hypothesis No. 1, empirical evidence suggests that the youngest and the oldest women are more likely to be out of the labor force. The omitted category is the youngest group, those 15 to 20 years old. I e xpect that the coefficients on the dummy variables indicating women between 20 a nd 50 years of age will be positive and significant indicating a greater pr obability of participating in the labor force than women 15 to 20 years in both the formal and informal sector. However, a priori it is not clear which age groups will be more likely or less likel y to work in either sector, compared to the youngest women. Education According to human capital theory, educa tion is positively related to labor force participation. I believe it also plays an important role in whether working woman will choose the formal or informal sector. Since the omitted category is women with no education, I expect that higher schooling attainments ar e associated with higher probabilities of participating in the formal sector compared to women no education. However, an increased likelihood of being em ployed in the formal sector may mean a negative likelihood of being employe d in the informal sector.

PAGE 103

91 Marital status The omitted category is the single women. As with Hypothesis No.1, we expect that the coefficient on the dummy variable indicating a woman is married will be negative indicating that she will be less likely to participate in either the formal or informal sector. We also expect negative coefficients on the dummy variables for divorced and widowed women because they us ually receive some kind of legal child support and inheritance, respectivel y. However, it is not clear, a priori their preferences to work in the formal or informal sector. With respect to the “cohabitors,” the coefficient on this dummy variable is not expected to be significant from that of single women. Geographic location factors: Urban Guy Standing (1982) report ed that in Latin Ameri ca young single women moved to the towns and cities to take better and higher-paying jobs, or to have access to job training. Consequently, we expect that the coefficients on the dummy variable for urban residence will be positiv e and higher for women’s labor force participation in the formal sector than in the informal sector since th ere are more job opport unities in those areas than in rural areas. Region Orlando (2001) suggests that, in genera l, formal and informal workers in Venezuela are concentrated in different ec onomic activities accord ing to the level of capital, technology and scale requi red. Formal workers are empl oyed mainly in the public sector, in manufacturing, and in intermediate activities (wholesale, financial and insurance services, real esta te, communications and transportation). Informal workers are

PAGE 104

92 mainly employed in retail commerce, servic es for the community and agriculture. In general, there is not a clear-cut separation of activities to be performed by workers of the formal and informal sector. The omitted category is the North-Eastern region which is characterized by low population density, agri cultural and oil pr oduction activities. However, signs and magnitudes of the coeffi cients on the dummy variables of the other seven regions will depend on the capacity of the region to generate jobs and the special characteristics of their economies in the peri od under analysis. I expect that women living in the most industrialized regions of the count ry will be more likely to participate in the formal sector; and, women living in agricultural and rural areas are more likely to participate in the informal sector.96 In the Capital region where the predominant activities are those of the public sector, industry, commerce and finance, I expect that the coefficients will be positive in the formal sector but negative in the informal sector. The Central region is the most important industrial center in the country. Commercial activities have also taken important role as an economic activity in this region. I expect that the coefficient on the dummy variable will show a positive probability of participating in the formal sector compared to the North-Eastern region because the formal sector dominates this region. The Plains region has a long ag ricultural tradition compared to the others. Therefore we expect a positive coefficient on this dummy variable in the informal sector estimation and a negative coefficient for the formal sect or, compared to the North-Eastern region because the informal sector dominates this region. In the Zulian region beside the agricultural and oil pr oduction activities, commercial activities and a large underground 96 Appendix A contains detailed description of each region.

PAGE 105

93 economy associated with drug trafficking, alco hol, cigarettes and other contraband goods play key roles in the economy. Thus, we expect that the coefficients on this dummy variable will be positive in both labor market s although higher for the informal sector. In the Andean region economic activities are concentrated mainly around agriculture and tourism. So I expect positive signs for the co efficients on this dummy variable indicating that women are more likely to participate in th e informal sector. In the Insular region, the predominant activities are fishing, commerce, and tourism. It is expected that the coefficients on this dummy variable will be pos itive in the informal sector compared to the North-Eastern region. Finally in the Central-Western region where agriculture, and oil refining are the principal ec onomic activities, it is anticipa ted that the expected signs of the coefficients on this dummy variable will be positive for both labor markets compared to the North-Eastern region becau se of generally greater economic activity. Socio-economic conditions factors: Heads of household Female heads of household need to work for pay to support their family’s dependents. I expected that the coefficients on this dummy variable will be positive indicating that women are more likely to par ticipate in both labor markets. Given the inability of the Venezuelan economy to supply enough jobs in the period under study, we expect that coefficient will be higher in th e informal sector than in the formal one. Socio-economic status The omitted category is poverty. We expect that the coefficients on the other dummy variables will be positive, indicating a greater likelihood of participating in both labor markets as women achieve higher st atus in the socio-economic structure.

PAGE 106

94 Nonlabor income According to the neocla ssical theory of labor, th e higher the amount of the nonlabor income, the higher the reservation wa ge and the lower the probability of labor force participation. Therefore, I expect nega tive signs for this variable indicating that women are less likely to part icipate in both sectors. Survey date The period under study was characterized by high rates of inflation, successive depreciations of the bolivar with respect to the American dollar, and, in general, employment suffered as a result of the impl ementation of the Venezuelan Agenda. Since the omitted category is 1995, I expect that the coefficients on the dummy variables for the surveys conducted in 1997 and 1998 will be positive, indicating that women are more likely to participate in both labor markets compared to thei r behavior in 1995. Moreover, given the structural disequilibrium of the labor market, I expect that those coefficients will be higher in the informal sector than those for the formal sector for all periods. Interaction terms This interaction variable is designed to capture the combined effects of nonlabor income and survey dates. I expect to fi nd that the higher value of the combined interaction of nonlabor income and the su rvey date, the lower women’s labor force participation will be in both market sectors. 4.4.5 Testing Hypothesis No. 3a The binomial logit model is used to ex amine the decision to participate in the labor force during the 1995-1998 period of th ree subsamples: married women, single

PAGE 107

95 women, and women heads of household. To te st this hypothesis the three subsamples are regressed on the same set of va riables as described in section 4.4.3. It is assumed that factors considered in this study affect each subsample in the same manner regardless of the group to which they belong. Consequently, the predicted signs for the subsamples are the same as those for the sample as a whole. However, in keeping with the predictions for the marital status dummy variables, on the variables where positive coefficients were predicted for the whole sample, we might e xpect to see smaller values for married women than for the other two groups because it is assumed that wives have spousal income. Moreover, we also expect that women heads of household will show larger values for the same variables since th ey need to support their dependents. 4.4.6 Testing Hypothesis No. 3b The multinomial logit model is again used to analyze the decision about whether to work in the formal or informal sector of the labor market or be out of the labor force for subsamples of married women, single women, and women heads of household. For each of the three subsamples, the three labo r force participation decision options are regressed on the same set of variables as described in sec tion 4.4.4. The predicted signs for the coefficients of the regressors of the three subsamples are the same as those for the sample as a whole, since it is assumed thos e factors affected Venezuelan women from these three groups in the same manner.

PAGE 108

96 Chapter Five Research Results The results of the test ing of the three hypotheses about Venezuelan women’s labor force participation are presented in this chapter. The chapter is divided in two sections. The first section describes the samples used for the statistical analys es. The second section presents the empirical results of the binomial and the multinomial logit models. 5.1 Description of the Samples Used for the Regression Equations Table 1 shows the means and proportions of the variables for the main sample, i.e. women aged 15 to 60 years old, and for th e subsamples: married women, single women and women heads of household. All variable s were tested for the significance of differences between the means and pr oportions of married and single women.97 The results indicate significant di fferences at a 1 percent leve l or greater for nearly all variables. The results from the entire sample in dicate that the largest proportion of women (27 percent) is between 21 and 30 years of age. The largest proportion of married women (32 percent) is between 31 and 40 years old, whereas the majority of single women (45 97 Tests of differences in proportions and means between wo men heads of households and the other two subsamples were not performed because of some coincidences of observations among them. For instance, a woman head of household may be either married or single.

PAGE 109

97 percent) are in the youngest age group, 15 to 20 years of age. The largest percentage of women heads of household (36 percent) is between the ages of 41 and 50. When we compare the subsamples, we find that as expected, there are more 15 to 20 year-old women among single women th an among married women or heads of household. There are also significantly more single women aged 21 to 30 (31 percent). Among married women, the larg est percentage is between the ages of 31 to 40 (32 percent), greater than among single women (13 percent), and greater than among heads of household (27 percent). The results also indi cate that there is a significantly greater number of women heads of household between the ages of 41 to 60 than among the other subsamples. Table 1 also shows that less than half of all women have more than a primary education. Twenty-three percen t have completed secondary or high school education and still fewer (11 percent) ha ve graduated from college. Those who have completed a technical education represent only 4 percent of the sample. Two-thirds of the married subsample has no more than a primary e ducation. Single women have the most education: 26 percent have a high school educat ion; 7 percent have a technical education; and 15 percent have a college education. Women heads of household have the highest proportion of women with no e ducation, 11 percent. Half of the total group of women is marri ed, 38 percent are single, and 13 percent are heads of household. Among women heads of household, 36 percen t are divorced and 32 percent are single. Eighty-ei ght percent of women in the sample live in urban areas. With respect to geographic location, the larg est proportion of women lives in the Capital

PAGE 110

98 region (24 percent). The smalle st proportions are found in the Insular and the Plains regions, 1 and 3 percent, respectively.98 For the sample as a whole, 68 percen t of women are of high and medium high socio-economic status, 38 percent and 30 per cent, respectively. Si milar proportions are observed in each of the subsamples. Fina lly, mean nonlabor income (which does not include the income of other family members) is extremely small. Indeed, 92.4 percent of the sample has none at all. Table 1 also summarizes main differences in percentages of women working in the formal or informal sector, and not working in the labor market at all. A larger percentage of single women are out of th e labor force than among any of the other subsamples (45 percent). This group also has the lowest percentage of women working in the informal sector of any of the other subs amples (12 percent), most likely because so many of this group are young and still in sc hool. Heads of household have the largest percentage in the labor force, over 87 percent. Of these, most (56 percent) work in the formal sector. Finally, 67 percent of married women participate in the labor market. In general, more women from all three subsamples work in the formal sector than in the informal sector, twice as many among married women and heads of household, and more than three times as many among single women.99 98 See Appendix A for more details about th e characteristics and locat ions of the administrative regions in Venezuela. 99 These results contradict what the cu rrent literature suggests, i.e. that in Latin American countries, women’s participation in the informal sector is grea ter than in the formal sector (CEPAL, 1999).

PAGE 111

99 Table 1 Description of Samples Used for Regression Equations ( standard deviations in parentheses ) V V a a r r i i a a b b l l e e1 All Women 15-60 Married Women Single Women Women Heads of Household Age groups 15 to 20 .210 .064 .454a .011 21 to 30 .273 .265 .310a .091 31 to 40 .241 .323 .131a .265 41 to 50 .179 .236 .070a .358 51 to 60 .101 .113 .044a .275 Education No education .055 .060 .041a .114 Primary .561 .615 .483a .604 Secondary .227 .204 .264a .158 Technical .044 .031 .065a .024 College .113 .090 .147a .099 Marital status Married .496 .141 Cohabitors .012 .035 Widows .023 .139 Divorced .087 .362 Single .382 .323 Urban residence .877 .869 .883a .889 Regions Andean .115 .115 .118a 119 Capital .239 .233 .241a 263 Central .127 .131 .127a 116 Central-Western .185 .181 .197a 182 Insular .012 .013 .010a .010 North-Eastern .093 .093 .093 .090 Plains .027 .029 .025a .032 Zulian .202 .205 .190a 189 Head of household .132 .037 .112a Socio-economic status High .379 .365 .390a 391 Medium High .303 .297 .307a 295 Average .171 .178 .166a 164 Poverty .147 .160 .137a 149 Nonlabor income (US$/month) 0.06 (0.001) 0.03 (0.001) 0.05a (0.001) 025 (0.004) Interaction terms Nonlabor income 1997-1 0.010 (0.0004) 0.004 (0.0003) 0.008a (0.0005) 0.040 (0.002) Nonlabor income 1997-2 0.010 (0.0004) 0.005 (0.0004) 0.008a (0.0005) 0.046 (0.002) Nonlabor income 1998-1 0.010 (0.0004) 0.005 (0.0004) 0.008a (0.0005) 0.044 (0.002) Nonlabor income 1998-2 0.011 (0.0004) 0.006 (0.0005) 0.008a (0.0005) 0.047 (0.002) Survey date 1995-1st half .236 .242 .239a 211 1997-1st half .221 .216 .227a 217 1997-2nd half .173 .174 .167a 181 1998-1st half .186 .184 .185 .197 1998-2nd half .184 .184 .183 .194 N 86,199 42,791 32,906 11,365d Formal sector (%) 45.3 44.8 42.2 55.7 Informal sector (%) 18.9 22.1 12.4 31.5 1 For exact definitions of variables see Appendix A. a Significantly different from married women at 1% level. b Significantly different from married women at 5% level. c Significantly different from married women at 10% level. d Includes women of all categories of marital status.

PAGE 112

100 5.2 Determinants of Women Labor Force Participation in Venezuela Regression results are reported in four different sub-sections in accordance with the hypothesis being tested. 5.2.1 Results of the Testing of Hypothesis No. 1 In this section, the determinants of women’s labor force participation and their decision whether to work or not to work are discussed. There are 86,199 observations of which 64.2 pe rcent are in th e labor market. Tables 2 and 3 depict the results of th e logistic regression using the sample of women between 15 and 60 years old.100 Due to the nonlinear nature of the model, the discussion of the results focuses on the margin al probabilities of the variables instead of on their coefficients. The most important factors determining Venezuelan women’s propensity to participate in th e labor force are the individu al characteristics of age, education and marital status. However, intere sting insights can also be drawn from the results of the other variables. Examining first the age variables, we fi nd that, as expected, women between 21 to 60 years old are more likely to participat e than are the youngest women; among those, women 31-40 years old are most likely with a marginal probability of 28.1 percent, followed by women 41 to 50 years old with 25.2 percent.101 Consistent with the general theory of human capital investment literatu re, the results show that education has a powerful impact on labor force participation, and that the probability of participation 100 Similar results were found using a binomial probit model. See Tables B-1 and B-2 in Appendix B. 101 The lower labor force particip ation rate among women in their fifties is c onsistent with the em pirical evidence of retirement from the labor force by women of this age group.

PAGE 113

101 increases with a greater educational atta inment. There are very similar marginal probabilities of participation, 21 to 23 percent, for women who have completed secondary, technical and college education, compared to women with no education. Marital status also influences a woman’s deci sion to participate in the labor force since all dummy variables’ coeffici ents are statistically signifi cant. Contrary to what was expected, married and divorced women are more likely to participate in the labor market than single women. Cohabitors’ marginal probability of labor force participation, unexpectedly, is positive and significant. They are 13.3 percent more likely than single women to participate in the labor market. Wi dows are 4 percent less likely to participate in the labor market. Finally, the marginal pr obability of participation for divorced women (16 percent) exhibits the larg est difference from the single gr oup of women, four times as large as that of married women.102 Women living in urban areas are 3 percent more likely to be in the labor market than rura l residents. As for the geographic areas of Venezuela, as expected, marginal effects of these dummy variables are positive and significantly different from th e North-Eastern area, except for the Insular region. The highest marginal probabilities of labor for ce participation are found for women living in the Andean, Central, and Central-Western re gions with 8.1, 6.6, and 6.2 percentages respectively. The Capital region has the smalle st marginal probability with 2.4 percent. 102 The presence of children is a constraint on the participa tion of married women. Unfortuna tely, the data used in this project does not have this variable available.

PAGE 114

102 Table 2 Binomial Logistic Regression Results Coefficients Sample: All Women 15-60 V a a r ri a a b ble Coefficients Standard Deviations z-values Age groupsa 21 to 30 1.218*** 0.021 56.77 31 to 40 1.573*** 0.025 63.52 41 to 50 1.406*** 0.025 50.74 51 to 60 0.773*** 0.033 23.88 Educationb Primary 0.671*** 0.034 19.48 Secondary 1.089*** 0.038 28.70 Technical 1.392*** 0.053 26.07 College 1.234*** 0.042 29.10 Marital status Married 0.198*** 0.019 10.19 Cohabitors 0.698*** 0.085 8.18 Widows -0.184*** 0.060 -3.06 Divorced 0.870*** 0.040 21.80 Urban residence 0.143*** 0.025 5.83 Regionsd Andean 0.387*** 0.034 11.26 Capital 0.111*** 0.030 3.69 Central 0.314*** 0.033 9.37 Central-Western 0.290*** 0.031 9.38 Insular 0.022 0.073 0.30 Plains 0.124** 0.053 2.33 Zulian 0.186*** 0.031 6.02 Head of household 1.381*** 0.036 38.19 Socio-economic status High 0.085*** 0.025 3.38 Medium High -0.017 0.026 -0.66 Average -0.0001 0.028 -0.00 Nonlabor income (US$/ month) -0.144*** 0.045 -3.20 Interaction terms Nonlabor income 1997-1 0.037 0.093 0.40 Nonlabor income 1997-2 0.012 0.091 0.14 Nonlabor income 1998-1 0.119 0.094 1.26 Nonlabor income 1998-2 0.167* 0.094 1.78 Survey datef 1997-1st half 0.094*** 0.023 4.04 1997-2nd half 0.254*** 0.027 9.46 1998-1st half 0.374*** 0.027 14.11 1998-2nd half 0.455*** 0.027 17.00 Constant -2.077*** 0.054 -38.74 N 86,199 -2* log likelihood ratio 97,140*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is the 1995-1 period.

PAGE 115

103 Table 3 Binomial Logistic Regression Results Marginal Effects Sample: All Women 15-60 V V a a r r i i a a b b l l e e Marginal Effects Standard Deviations z-values Age groupsa 21 to 30 0.237*** 0.004 64.64 31 to 40 0.281*** 0.004 78.50 41 to 50 0.252*** 0.004 64.54 51 to 60 0.150*** 0.005 27.89 Educationb Primary 0.149*** 0.008 19.49 Secondary 0.211*** 0.006 33.68 Technical 0.227*** 0.006 39.76 College 0.219*** 0.006 38.58 Marital statusc Married 0.044*** 0.004 10.20 Cohabitors 0.133*** 0.014 9.83 Widows -0.042*** 0.014 -2.99 Divorced 0.164*** 0.006 26.74 Urban residence 0.032*** 0.006 5.73 Regionsd Andean 0.081*** 0.007 12.00 Capital 0.024*** 0.007 3.73 Central 0.066*** 0.007 9.84 Central-Western 0.062*** 0.006 9.73 Insular 0.005 0.016 0.30 Plains 0.027** 0.011 2.38 Zulian 0.040*** 0.007 6.15 Head of household 0.241*** 0.005 52.75 Socio-economic statuse High 0.019*** 0.006 3.38 Medium High -0.004 0.006 -0.66 Average -0.000 0.006 -0.00 Nonlabor income (US$/ month) -0.032*** 0.010 -3.20 Interaction terms Nonlabor income 1997-1 0.008 0.021 0.40 Nonlabor income 1997-2 0.003 0.020 0.14 Nonlabor income 1998-1 0.026 0.021 1.26 Nonlabor income 1998-2 0.037* 0.021 1.78 Survey datef 1997-1st half 0.020*** 0.005 4.08 1997-2nd half 0.054*** 0.006 9.77 1998-1st half 0.079*** 0.005 14.79 1998-2nd half 0.095*** 0.005 18.06 N 86,199 -2* log likelihood ratio 97,140*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is the 1995-1 period. As expected, women heads of household are more likely to partic ipate in the labor market with a marginal probability of 24 per cent. Contrary to what was expected, women

PAGE 116

104 of medium high and average status are not more likely to participate in the labor market than are women of poverty status. Women of hi gh status are only 2 percent more likely to participate in the labor market. The reason for these unexpected outcomes may be either that the impact of educat ion overwhelms the variables103 or that the variables used to construct the socio-economic index ar e not sufficient proxies of wealth.104 As predicted by neoclassical theory, n onlabor income has a nega tive impact on labor force participation. An additional dollar of nonlabor income per month is predicted to reduce labor force participation by 3 percent. However, when this variable is interacted with the survey dates, the net marginal probability of nonlabor income for the second survey of 1998 becomes positive, but very small, 0.5 pe rcent. Finally, as expected, the marginal probabilities of the dummies for the surv ey dates are increasingly positive, and significantly different from 19951, ranging from 2 to 10 percent in the last period. This behavior might be explained by women’s need to preserve the real income of the family during this period, when the economic crisis deepened due to successive periods of inflation and the devalu ation of the bolvar. 103 However, the correlations between education and the so cio-economic variables are not high enough to preclude including both sets of variables. 104 Javier Parra (2004) asserts that most studies of socio-economic status in Venezuela use measures and definitions that are not exhaustive, and do not apply well to the curren t characteristics of the Venezuelan economy. According to the author, it is imperative that more research be conducted in this area to allow for comp arative analyses of the living conditions of people and to contribute to an understanding of ways to overcome the economic difficulties faced by the majority of the population. Better measur es will facilitate decision-making on soci al policies and income redistribution.

PAGE 117

105 5.2.2 Results of the Testing of Hypothesis No. 2 The full sample contains 86,199 observati ons. Forty-five percent of the women work in the formal sector and 19 percent work in the informal sector.105 The dependent variable in the labor force pa rticipation equation is a tricho tomous variable, which takes the value of zero if the woman is out of the la bor market, one if she works in the informal sector, and two if she works in the formal sector. A logit function is used for the estimations. The regressors measure persona l and demographic characteristics of the individual women, as well as geographic factors and socio-economic conditions. As discussed in the previous sect ion, because coefficients are more difficult to interpret given the nonlinear nature of the model, th e discussion in this section will focus on the marginal effects. Tables 4 and 5 illustrate the coefficients and marginal effects from the multinomial logit regression using the en tire sample of women 15 to 60 years of age.106 In terms of individual and demographic fa ctors, the results indicate that women in all age groups are significantly more likely to participate in the labor market than women 15 to 20 years old. However, the pattern is different in the info rmal and formal sectors. In the informal sector, the participation of wome n in the labor force increases with age with the highest marginal effects fo r the two oldest groups, 41 to 50, and 51 to 60 years of age, with 12 and 11 percent, respectiv ely. In the formal sector, however, the peak marginal impact for age is for the two youngest groups of 21 to 30, and 31 to 40 years of age with marginal probabilities of 19 and 18 percent, respectively. As expected, women 51 to 60 years old are only 4 percent more likely to wo rk in the formal sector compared to the 105 See Table B-6 in the Appendix B for details. 106 Similar results using a multinomial pr obit model were found. See Tables B.3 and B.4 in Appendix B for the results.

PAGE 118

106 youngest women. However, contrary to what was expected, women 51 to 60 years old are just about as likely to work in the informal sector as women 31 to 50 years old. This is probably because women of those ages do not receive the retirement benefits which are available to women who have worked in the formal sector. Finally, for all age groups except for the oldest, the marginal impacts of age on labor force par ticipation are greater in the formal sector than in the informal one. With respect to education, as expect ed, all the dummy variables exhibit positive marginal probabilities of labor force partic ipation in the formal sector, but generally negative, much smaller effects in the inform al sector. The excepti on is women with only primary education who are slightly more likely to participate in the informal sector than are women with no education. As the level of education increases, women become more and more likely to participate in the formal sector with marginal effects ranging from 15 to 34 percent. Women with technical and co llege degrees exhibit the highest marginal probabilities of participation, 34 percent and 32 percent, re spectively. These two groups are also the least likely to part icipate in the informal sector. This is probably due to the fact that employment in the form al sector is considered more desirable because of higher wages and fringe benefits.

PAGE 119

107 Table 4 Multinomial Logit Regression Results Formal Sector Sample: All Women 15-60 Variable Coefficients Standard Deviations Marginal Effects Standard Deviations Age groupsa 21 to 30 1.246*** 0.023 0.192 *** 0.005 31 to 40 1.522*** 0.026 0.182*** 0.006 41 to 50 1.307*** 0.030 0.133*** 0.007 51 to 60 0.621*** 0.036 0.036*** 0.008 Educationb Primary 0.792*** 0.040 0.150*** 0.009 Secondary 1.361*** 0.043 0.278*** 0.008 Technical 1.725*** 0.057 0.335*** 0.008 College 1.554*** 0.047 0.315*** 0.008 Marital statusc Married 0.109*** 0.021 -0.012*** 0.005 Cohabitors 0.726*** 0.088 0.111*** 0.017 Widows -0.223*** 0.065 -0.047*** 0.013 Divorced 0.881*** 0.041 0.122*** 0.007 Urban residence 0.176*** 0.026 0.039*** 0.006 Regionsd Andean 0.375*** 0.036 0.053*** 0.008 Capital 0.189*** 0.032 0.058*** 0.007 Central 0.266*** 0.036 0.024*** 0.008 Central-Western 0.327*** 0.033 0.064*** 0.007 Insular 0.115 0.077 0.050*** 0.017 Plains 0.091 0.057 0.003 0.013 Zulian 0.013 0.033 -0.047*** 0.007 Head of household 1.292*** 0.038 0.126*** 0.007 Socio-economic statuse High 0.066*** 0.027 0.005 0.006 Medium High 0.005 0.028 0.008 0.006 Average 0.012 0.030 0.005 0.007 Nonlabor income (US$/month) -0.151*** 0.048 -0.028*** 0.010 Interaction terms Nonlabor income 1997-1 0.029 0.099 0.002 0.020 Nonlabor income 1997-2 0.054 0.094 0.022 0.020 Nonlabor income 1998-1 0.190** 0.098 0.053*** 0.020 Nonlabor income 1998-2 0.235*** 0.098 0.058*** 0.020 Survey datef 1997-1st half 0.073*** 0.025 0.004 0.006 1997-2nd half 0.243*** 0.028 0.033*** 0.006 1998-1st half 0.321*** 0.028 0.031*** 0.006 1998-2nd half 0.357*** 0.028 0.019*** 0.006 Constant -2.506*** 0.059 N 86,199 % of total 39,036 45.3 -2 log likelihood ratio 160,147.98*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is the 1995-1 period.

PAGE 120

108 Table 5 Multinomial Logit Regression Results Informal Sector Sample: All Women 15-60 Variable Coefficients Standard Deviations Marginal Effects Standard Deviations Age groupsa 21 to 30 1.119*** 0.031 0.047*** 0.004 31 to 40 1.701*** 0.033 0.110*** 0.005 41 to 50 1.611*** 0.037 0.122*** 0.006 51 to 60 1.051*** 0.043 0.113*** 0.008 Educationb Primary 0.528*** 0.042 0.011* 0.004 Secondary 0.523*** 0.048 -0.053*** 0.006 Technical 0.496*** 0.075 -0.093*** 0.006 College 0.466*** 0.055 -0.081*** 0.005 Marital statusc Married 0.444*** 0.027 0.057*** 0.004 Cohabitors 0.645*** 0.107 0.025* 0.014 Widows -0.095 0.071 0.005 0.009 Divorced 0.880*** 0.048 0.047*** 0.006 Urban residence 0.051 0.034 -0.008 0.005 Regionsd Andean 0.416*** 0.046 0.029*** 0.007 Capital -0.123*** 0.042 -0.034*** 0.005 Central 0.430*** 0.045 0.043*** 0.007 Central-Western 0.195*** 0.042 -0.0008 0.006 Insular -0.266** 0.109 -0.045*** 0.012 Plains 0.206*** 0.071 0.024** 0.010 Zulian 0.521*** 0.041 0.084*** 0.006 Head of household 1.583*** 0.042 0.119*** 0.006 Socio-economic statuse High 0.129*** 0.330 0.014*** 0.004 Medium High -0.074** 0.350 -0.011*** 0.004 Average -0.025 0.038 -0.005 0.005 Nonlabor income (US$/month) -0.109* 0.058 -0.003 0.007 Interaction terms Nonlabor income 1997-1 0.058 0.116 0.006 0.014 Nonlabor income 1997-2 -0.096 0.116 -0.019 0.014 Nonlabor income 1998-1 -0.067 0.118 -0.027* 0.014 Nonlabor income 1998-2 0.004 0.112 -0.020 0.013 Survey datef 1997-1st half 0.152*** 0.032 0.017*** 0.005 1997-2nd half 0.291*** 0.037 0.022*** 0.005 1998-1st half 0.510*** 0.036 0.050*** 0.005 1998-2nd half 0.693*** 0.036 0.078*** 0.005 Constant -3.296*** 0.072 N 86,199 % of total 16,303 18.91 -2 log likelihood ratio 160,147.98*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is the 1995-1 period.

PAGE 121

109 Marital status affects women’s labor fo rce participation in various ways. As expected, compared to single women, being married has a small negative impact on labor force participation in the formal sector. Ho wever, contrary to what was expected, we found that married women are 6 percent more lik ely to participate in the informal sector than single women. Cohabita ting women also show unexpected behavior. Their marginal probabilities of labor force participation ar e positive and significant in both sectors although the marginal effect is higher in th e formal sector (11 percent) than in the informal sector (3 percent). Widows, as e xpected, are less likely than single women to participate in the formal sector but, unexp ectedly, their labor mark et behavior is not significantly different from that of single women in the informal sector. Moreover, contrary to what was expected, divorced wo men are more likely than single women to participate in both sectors but the marginal probability is considerably higher in the formal sector (12 percent) than in the inform al sector (5 percent). The highest marginal impacts of marital status on labor force pa rticipation are for divorced women in the formal sector and for married women in th e informal sector with 12 and 6 percent, respectively. Finally, in general, all the effects of marital status are considerably larger in the formal sector. In terms of the impact of geographical factors on labor for ce participation, as expected, women living in urban areas have a significantly positive marginal effect of participating in the formal sector. However, there are no significant differences between women living in rural areas and those living in urban areas with respect to being employed in the informal sector. As for th e geographic areas, as expected, in most regions the marginal probabilities of labor fo rce participation are st atistically significantly

PAGE 122

110 different from those of women living in the North-Eastern region. The exceptions are for the Plains region in the formal sector and the Central-Western region in the informal sector. The signs were almost all expected, except for the Insular region in the formal sector. Women from the Central, Zulian, And ean and the Plains regions are more likely to participate in the informal sector than women from the Nort h-Eastern region. The Zulian region exhibits the highest marginal pr obability of 8 percent and the Plains region the lowest, with 2 percent. On the other ha nd, women living in the Capital and Insular regions are less likely to participate in the in formal sector with ma rginal effects of 3 percent and 5 percent, respectively. In the fo rmal sector, women in almost all regions are significantly more likely to participate in th e labor market than women from the NorthEastern region. The Capital and the Central-We stern regions exhibit the highest marginal effects of labor force participation of appr oximately 6 percent. However, women in the Zulian region are 5 percent less likely to pa rticipate in the formal sector. The Central region shows the smallest significant differe nce from the North-Eastern region with a marginal effect of only 2 percent.107 Women heads of household, as expected are significantly more likely to participate in both the informal and formal sectors; the marginal effects are almost identical with 13 percent for the formal sect or and 12 percent for the informal sector. With respect to socio-economic status, c ontrary to what was expected, we find no significant impact of these vari ables on the marginal probabilities of participating in the formal sector. However, in the informal s ector the pattern is different: women of the 107 For detailed information about women’s sector-emp loyment by regions, see Table B.5 in Appendix B.

PAGE 123

111 highest status are slightly more likely to par ticipate, and those of medium high status are slightly less likely to participate than are women of poverty status. The nonlabor income variable shows th e expected negative marginal effect on labor force participation only in the formal s ector. Labor force participation is reduced by 3 percent for every additional dolla r of nonlabor income per month.108 When this variable is interacted with the survey dates, the net marginal probabilities become positive and significantly different from 1995-1 fo r the two survey periods in 1998.109 The only significant marginal effect in th e informal sector is for the first survey of 1998. At that time, an additional dollar of nonlabor income per month is predicted to reduce labor force participation by 3 percent. Expected but interesting results are obse rved for the dummy variables associated with the survey dates. In the informal sector, we see marginal probabilities of participation increasing from 2 percent to 8 percent between the fi rst survey of 1997 to the second survey of 1998. Conversely, in the formal sector, the marginal effects decrease slightly over these periods. All the probabilities of participati on in the informal sector are positive and increasingly greater than the proba bilities of participating in the formal sector. This increasing partic ipation of women in the info rmal labor market through the successive years is related to the level of economic activity during the period of this analysis.110 In addition, the ability of the formal sector to provide jobs decreased due to 108 This result seems like a very st rong effect, but mean nonlabor inco me is only 6 cents per month. 109 The same reasons pointed out for the results in the first hypothesis apply here, with net marginal effects of around 3 percent in the formal sector. 110 During this period the remedial policy entitled the Vene zuelan Agenda was still bei ng implemented. Miguel A. Santos (2003) asserts that its implemen tation had an immediate negative effect on the GDP per capita of –2.3%. The Venezuelan Agenda was accompanied by a small negative effect on the GDP in 1996 (0.2%), and positive effect in 1997 (6.1%) although there was some improvement of the oil ac tivity due to the increase in oil prices and due to

PAGE 124

112 the costs associated with providing these jobs, i. e. contracting costs, health and retirement benefits, etc. 5.2.3 Results of the Testing of Hypothesis No. 3a In this section the determinants of labor force participation of married, single, and women heads of household are discussed. 5.2.3.1 Results for Married Women The subsample of married women includes 42,791 observations of which 67 percent participate in the labor market. Tables 6 and 7 show the coefficients and marginal effects from the binomial logit regression fo r this group. The most important factors affecting Venezuelan married women’s decisi on to participate in the labor market are age, education, being heads of household, a nd the survey date. Many of these results are similar to those of the full sample of wome n but there are some differences. For example Table 7 shows no significant difference between the propensity to participate of women 51 to 60 years old and that of the youngest group. However, in the full sample, the marginal effect of that age group is pos itive and highly significant at 15 percent. The highest marginal probability of partic ipation is among those married women 31 to 40 years old, and the lowest probability is am ong women 21 to 30 years old, prime childsubsequent investment in this sector. Si nce the oil industry is a capital-intensive activity, this recovery was not able to reduce the rates of unemployment in relation to the pre-adju stment period (second half of 1995). On the contrary, the employment rate registered a sustai ned decrease from 1996 until the end of 1999. The unemployment rate of 10.2% before the launching of the Venezuelan Agenda increased still further. During the seven periods of this study (first half of 1996 to the first half of 1999), a total of 742,139 jobs we re eliminated, 667,221 in the fo rmal sector of the economy, and 73,262 in the informal sector. During these three years, there was also a change in the composition of the population employed in the informal sect or, which increased from 48% of total em ployment in the second half of 1995 to 51% in the first half of 1999 (Santos 2003).

PAGE 125

113 bearing years. But even women in this age gr oup are 7 percent more likely to participate in the labor market than are the youngest women. With respect to education, we found that the probabilities of labor force participation for all levels of education are positive, increasing with the level of education, and significantly different from the group of married women with no education. The largest probability is show n for married women with a college degree, with 28 percent, which is 6 percentage point s higher than the marginal probability for same group of women in the full sample. The lowest marginal probability of labor force participation is found to be for married wome n with a primary education, who are still 10 percent more likely to participate in the labor force than thos e with no education.

PAGE 126

114 Table 6 Binomial Logistic Regression Results Coefficients Subsample: Married Women V V a a r r i i a a b b l l e e Coefficients Standard Deviations z-values Age groupsa 21 to 30 0.351*** 0.035 10.14 31 to 40 0.697*** 0.034 20.34 41 to 50 0.515*** 0.038 13.71 51 to 60 -0.031 0.044 -0.71 Educationb Primary 0.472*** 0.044 10.66 Secondary 1.106*** 0.050 21.90 Technical 1.672*** 0.090 18.55 College 1.916*** 0.067 28.67 Urban residence 0.181*** 0.033 5.51 Regionsd Andean 0.412*** 0.048 8.64 Capital -0.037 0.041 -0.90 Central 0.324*** 0.046 7.01 Central-Western 0.303*** 0.043 7.06 Insular 0.141 0.099 1.43 Plains 0.083 0.071 1.17 Zulian 0.048 0.042 1.13 Head of household 1.060*** 0.071 14.97 Socio-economic status High 0.110*** 0.034 3.20 Medium High -0.031 0.035 -0.88 Average 0.018 0.038 0.48 Nonlabor income (US$/ month) -0.136* 0.078 -1.74 Interaction terms Nonlabor income 1997-1 0.859*** 0.211 4.08 Nonlabor income 1997-2 0.485*** 0.183 2.65 Nonlabor income 1998-1 0.547 0.182 3.01 Nonlabor income 1998-2 0.553*** 0.174 3.19 Survey datef 1997-1st half 0.101*** 0.031 3.23 1997-2nd half 0.302*** 0.037 8.26 1998-1st half 0.445*** 0.037 12.18 1998-2nd half 0.551*** 0.037 14.90 Constant -1.030*** 0.072 -14.22 N 42,791 -2* log likelihood ratio 50,386.58*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is the 1995-1 period.

PAGE 127

115 Table 7 Binomial Logistic Regression Results Marginal Effects Subsample: Married Women V a a r ri a a b ble Marginal Effects Standard Deviations z-values Age groupsa 21 to 30 0.073*** 0.007 10.52 31 to 40 0.142*** 0.007 21.70 41 to 50 0.104*** 0.007 14.63 51 to 60 -0.007 0.009 -0.70 Educationb Primary 0.103*** 0.010 10.52 Secondary 0.204*** 0.008 26.34 Technical 0.241*** 0.007 33.27 College 0.276*** 0.005 50.40 Urban residence 0.040*** 0.007 5.39 Regionsd Andean 0.083*** 0.009 9.31 Capital -0.008 0.009 -0.90 Central 0.066*** 0.009 7.41 Central-Western 0.062*** 0.008 7.37 Insular 0.030 0.020 1.47 Plains 0.018 0.015 1.19 Zulian 0.010 0.009 1.14 Head of household 0.180*** 0.009 20.53 Socio-economic statuse High 0.023*** 0.007 3.22 Medium High -0.007 0.007 -0.88 Average -0.004 0.008 0.48 Nonlabor income (US$/ month) -0.029* 0.017 -1.74 Interaction terms Nonlabor income 1997-1 0.184*** 0.045 4.08 Nonlabor income 1997-2 0.104*** 0.039 2.65 Nonlabor income 1998-1 0.117*** 0.039 3.01 Nonlabor income 1998-2 0.119*** 0.037 3.19 Survey datef 1997-1st half 0.021*** 0.007 3.27 1997-2nd half 0.062*** 0.007 8.63 1998-1st half 0.090*** 0.007 13.03 1998-2nd half 0.110*** 0.007 16.25 N 42,791 -2* log likelihood ratio 50,386.58*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is the 1995-1 period. We find that married women living in urba n areas have a sligh tly higher marginal probability of participation th an the full sample of women. Although almost all regional dummy variables marginal probabilities for the full sample of women are significant and positive, we find that marginal effects for married women are significantly different from

PAGE 128

116 the North-Eastern region only for the Central, Central-Western, and the Andean regions, all of which are quite similar, varying only from 6 to 8 percent for the Central-Western and the Andean regions, respectively. Married women heads of household are 18 percent more likely to participate in the labor market than married women who are not. This is 6 percentage points lower than for the full sample of women. Results for th e socio-economic status variables are similar to that of the full sample of women: only married women of the hi ghest status are more likely to participate in the labor market than those of lowest status. Nonlabor income shows a negative marginal probability simila r to that of the whole sample. However, when we interact the latter variable with the survey date (the interaction term variable) for the subsample of married women, the net probabilities of participating in the labor market are all significantly and, (inexplicab ly) strongly positive compared to 1995-1, for all periods. (For the whole sample, the interaction terms were generally not significant.) An additional dollar of nonlabor income pe r month is predicted to increase married women’s labor force participation from 7.5 to 15.5 percentage points during the entire period. Finally, survey date variables for marri ed women exhibit pos itive and increasing marginal probabilities ranging from 2 to 11 pe rcent significantly different from the 19951 period. These marginal effect s are generally stronger than those of the full sample of women. The main reason for these stronger e ffects is the increases in the unemployment rate, which affected the husbands’ income in the period under study, forcing married women to increase their participation in the labor force to compensate for the loss of family income.

PAGE 129

117 5.2.3.2 Results for Single Women The subsample of single women contai ns 32,906 observations of whom 55 percent are in the labor market. Tables 8 and 9 show the coefficients and marginal effects from the binomial logit regression using this subsample. Many of the marginal probabilities are similar to those we found fo r the full sample of women. However, in general, the effects are much stronger. With re spect to the age variable, all age groups are more likely to participate in the labor fo rce than those ages 15-20. The marginal probabilities range from 32 percent for the olde st women, to 42 percent for those between 31 and 40 years old. Women with all levels of education are signifi cantly more likely to participate in the labor force than are those with none. The hi ghest marginal probability is for those with a technical education, 32 per cent; the lowest is among those with a college education, at 26 percent. Single women living in urban areas are 3 percent more likely to participate in the labor market than are those living in rural ar eas, a result similar to that from the full sample. As for the marginal effects of re gional variables, we see positive marginal probabilities for all regions, similar to those found for the full sample, although considerably stronger. For instance, singl e women living in the Andean region are 11 percent more likely to participate in the la bor force than those in the North-Eastern region, versus 8 percent for the whole sample.

PAGE 130

118 Table 8 Binomial Logistic Regression Results Coefficients Subsample: Single Women V a a r ri a a b ble Coefficients Standard Deviations z-values Age groupsa 21 to 30 1.821*** 0.031 59.60 31 to 40 2.480*** 0.049 50.12 41 to 50 2.629*** 0.073 36.16 51 to 60 1.778*** 0.080 22.10 Educationb Primary 1.499*** 0.073 20.63 Secondary 1.607*** 0.076 21.21 Technical 1.694*** 0.088 19.16 College 1.206*** 0.079 15.36 Urban residence 0.117*** 0.042 2.75 Regionsd Andean 0.484*** 0.058 8.40 Capital 0.351*** 0.051 6.84 Central 0.418*** 0.057 7.40 Central-Western 0.347*** 0.052 6.66 Insular 0.063 0.137 0.46 Plains 0.359*** 0.093 3.86 Zulian 0.526*** 0.053 9.92 Head of household 1.523*** 0.070 21.87 Socio-economic status High 0.043 0.043 1.02 Medium High -0.026 0.045 -0.58 Average -0.019 0.049 -0.40 Nonlabor income (US$/ month) -0.257*** 0.081 -3.17 Interaction terms Nonlabor income 1997-1 -0.592*** 0.177 -3.34 Nonlabor income 1997-2 0.079 0.169 0.47 Nonlabor income 1998-1 0.229 0.185 1.24 Nonlabor income 1998-2 0.256 0.187 1.37 Survey datef 1997-1st half 0.091** 0.039 2.33 1997-2nd half 0.145*** 0.045 3.20 1998-1st half 0.254*** 0.044 5.75 1998-2nd half 0.330*** 0.044 7.45 Constant -3.055 *** 0.100 -30.40 N 32,906 -2* log likelihood ratio 35,458.24*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is the 1995-1 period.

PAGE 131

119 Table 9 Binomial Logistic Regression Results Marginal Effects Subsample: Single Women V a a r ri a a b ble Marginal Effects Standard Deviations z-values Age groupsa 21 to 30 0.389*** 0.005 71.34 31 to 40 0.423*** 0.005 86.04 41 to 50 0.406*** 0.005 75.51 51 to 60 0.322*** 0.009 35.59 Educationb Primary 0.349*** 0.015 22.63 Secondary 0.344*** 0.013 26.03 Technical 0.318*** 0.011 29.60 College 0.258*** 0.014 18.75 Urban residence 0.029*** 0.010 2.73 Regionsd Andean 0.113*** 0.013 8.87 Capital 0.084*** 0.012 7.00 Central 0.099*** 0.013 7.73 Central-Western 0.083*** 0.012 6.84 Insular 0.015 0.033 0.46 Plains 0.084** 0.021 4.04 Zulian 0.123*** 0.012 10.42 Head of household 0.304*** 0.010 30.75 Socio-economic statuse High 0.011 0.010 1.02 Medium High -0.006 0.011 -0.58 Average -0.005 0.012 -0.40 Nonlabor income (US$/ month) -0.063*** 0.020 -3.17 Interaction terms Nonlabor income 1997-1 -0.144*** 0.043 -3.34 Nonlabor income 1997-2 0.019 0.041 0.47 Nonlabor income 1998-1 0.056 0.045 1.24 Nonlabor income 1998-2 0.062 0.045 1.37 Survey datef 1997-1st half 0.022** 0.009 2.34 1997-2nd half 0.035*** 0.011 3.23 1998-1st half 0.061*** 0.010 5.85 1998-2nd half 0.079*** 0.010 7.65 N 32,906 -2* log likelihood ratio 35,458.24*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is the 1995-1 period. Socio-economic status has no significant impact on labor force participation for this group. The impact of nonlabor income is twice as large for single women as for the sample as a whole. However in this regr ession the interaction term between nonlabor income and the period 1997-1 is significan t and has the expected negative sign. An

PAGE 132

120 additional dollar of nonlabor income per m onth is predicted to reduce labor force participation by 21 percent relative to first ha lf of 1995. The coeffici ent on the interaction term for the second half of 1998 is not significan tly in this regression, in contrast to the results for the whole sample. Regarding the su rvey date dummy variables, as expected, the marginal probabilities are increasingly posi tive, and significantly different from 19951 ranging from 2 to 8 percent in the last period. These results are similar to those for the whole sample. 5.2.3.3 Results for Women Heads of Household Finally, the subsample of women he ads of household consists of 11,365 observations of which 87 percent are in the labor market. Tables 10 and 11 display the coefficients and marginal effects from the bi nomial logit regression us ing this subsample. As noted earlier, this group has the highest rate of labor force participation of any of the subsamples. We see a number of differen ces in the marginal effects of the marital status variables, urban reside nce, regions, socio-economic stat us, interactions terms, and the survey date, compared to the full sample of women. Age variables again show positive marginal probabilities of labor force participation compared to the youngest wome n. However, the group 41 to 50 years old exhibits the highest marginal probability: th ey are 16 percent more likely to be in the labor force whereas those of 31 to 40 years ol d have the highest marginal probability for the full sample of women.

PAGE 133

121 Table 10 Binomial Logistic Regression Results Coefficients Subsample: Women Heads of Household V a a r ri a a b ble Coefficients Standard Deviations z-values Age groupsa 21 to 30 1.644*** 0.215 7.65 31 to 40 2.295*** 0.205 11.18 41 to 50 2.092*** 0.201 10.42 51 to 60 1.247*** 0.200 6.23 Educationb Primary 0.727*** 0.079 9.17 Secondary 1.554*** 0.128 12.12 Technical 1.948*** 0.328 5.95 College 1.673*** 0.163 10.24 Marital status Married -0.740*** 0.090 -8.25 Cohabitors 0.036 0.171 0.21 Widows -0.833*** 0.083 -10.08 Divorced 0.124 0.085 1.46 Urban residence 0.013 0.094 0.14 Regionsd Andean 0.147 0.132 1.12 Capital -0.089 0.116 -0.77 Central 0.198 0.135 1.47 Central-Western 0.098 0.120 0.82 Insular -0.401 0.279 -1.44 Plains -0.263 0.178 -1.48 Zulian -0.044 0.118 -0.37 Socio-economic status High -0.049 0.099 -0.50 Medium High -0.189* 0.100 -1.89 Average -0.155 0.109 -1.42 Nonlabor income (US$/ month) -0.191** 0.082 -2.32 Interaction terms Nonlabor income 1997-1 -0.542*** 0.149 -3.63 Nonlabor income 1997-2 -0.410*** 0.144 -2.84 Nonlabor income 1998-1 -0.493*** 0.156 -3.17 Nonlabor income 1998-2 -0.347** 0.149 -2.33 Survey datef 1997-1st half 0.018 0.098 0.18 1997-2nd half 0.196* 0.113 1.73 1998-1st half 0.367*** 0.114 3.22 1998-2nd half 0.444*** 0.116 3.83 Constant -0.333 0.258 -1.29 N 11,365 -2* log likelihood ratio 7,634.36*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is the 1995-1 period.

PAGE 134

122 Table 11 Binomial Logistic Regression Results Marginal Effects Subsample: Women Heads of Household V V a a r r i i a a b b l l e e Marginal Effects Standard Deviations z-values Age groupsa 21 to 30 0.087*** 0.007 12.96 31 to 40 0.145*** 0.010 13.85 41 to 50 0.157*** 0.014 11.03 51 to 60 0.089*** 0.012 7.38 Educationb Primary 0.068*** 0.008 8.46 Secondary 0.092*** 0.005 17.80 Technical 0.085*** 0.006 14.47 College 0.089*** 0.005 17.58 Marital statusc Married -0.080*** 0.012 -6.88 Cohabitors 0.031 0.015 0.21 Widows -0.093*** 0.011 -8.17 Divorced 0.011 0.007 1.49 Urban residence 0.001 0.008 0.14 Regionsd Andean 0.012 0.010 1.17 Capital -0.008 0.010 -0.75 Central 0.016 0.010 1.56 Central-Western 0.008 0.010 0.84 Insular -0.041 0.033 -1.24 Plains -0.025 0.018 -1.34 Zulian -0.004 0.011 -0.37 Socio-economic statuse High -0.004 0.009 -0.49 Medium High -0.017* 0.009 -1.83 Average -0.014 0.010 -1.37 Nonlabor income (US$/ month) -0.017** 0.007 -2.32 Interaction terms Nonlabor income 1997-1 -0.047*** 0.013 -3.63 Nonlabor income 1997-2 -0.036*** 0.013 -2.85 Nonlabor income 1998-1 -0.043*** 0.014 -3.17 Nonlabor income 1998-2 -0.030** 0.013 -2.33 Survey datef 1997-1st half 0.002 0.008 0.19 1997-2nd half 0.016* 0.009 1.83 1998-1st half 0.029*** 0.008 3.53 1998-2nd half 0.035*** 0.008 4.28 N 11,365 -2* log likelihood ratio 7,634.36*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is the 1995-1 period. Turning to the education variables, we see considerably smaller effects than the whole sample. The highest marginal effect s are for secondary education, technical education, and college degrees, all approximate ly 9 percent. For the sample as a whole,

PAGE 135

123 the marginal effects for tec hnical education and college e ducation were both around 22 percent. With respect to marital status, married and widowed heads of household are 8 and 9 percent, respectively, less likely to participat e in the labor market than are single heads of household. Cohabitors and divorced heads of household are not st atistically different from single heads of household, in contrast to the results fr om the full sample of women, where they were significantly more likel y to participate in the labor market. Examining next geographical factors, we find that for women heads of household, living in an urban area has no impact on labor fo rce participation, in contrast to the full sample for whom living in an urban area has a positive and significant effect. As for the region variables, none of the marginal probabili ties are significant. This contrast sharply with the numerous regional differences found for the full sample of women. Among the socio-economic status variable s, only the marginal probability for medium high status is significant, the ne gative sign indicates that women in this group are less likely to participat e in the labor force than heads of house hold classified as being in poverty. These results differ considerably from those from the full sample of women where we found a significant margin al probability only for women with high status, indicating that they are 2 percent more likely to participate in the labor market. The nonlabor income variable shows a negative marginal probability of 2 percent, similar to the result obtained from the full sample of women. When we interact the nonlabor income variable with the survey date, we find significant negative marginal probabilities for all the periods under study, with the highest net effect on labor participation for the first half of 1997 and the firs t half of 1998. For these peri ods, an additional dollar of

PAGE 136

124 nonlabor income per month decreases the likelihood of labor participation of women heads of household by 6 percent. In contrast, th e marginal effects of the interaction terms in the full sample are not signi ficantly different from the firs t half of 1995, except for the second half of 1998 when there is a net posi tive effect on labor pa rticipation of 0.5 percent. Finally, with respect to the survey dates, three of the four marginal probabilities are positive and increasing with successive periods, a pattern similar to that of the full sample. 5.2.3.4 Comparisons Among the Three Subsamples Tables 7, 9, and 11 show the results of the binomial regressions for married women, single women, and women heads of household. Regarding age, we find similar results among the three subsamples although the marginal effects are stronger for single wome n. As for education, the relative marginal effects are similar for the three subsamples, but the magnitudes are considerably less for women heads of household. For example, th e marginal effect of having a secondary education is 34 percent for single women, but only 9 percent for women heads of household. With respect to geographic factors, ur ban residence has a positive impact on the labor force participation of married and single women but not of women heads of household. As for the region dummy variable s, we found considerable differences in marginal probabilities of labor force participation for the three subsamples. Only for single women did we find positive and significant probabilities of labor force

PAGE 137

125 participation for all regions, except for the in sular region (which is not significant for any of the three subsamples), compared to the North-Eastern region. N one of the marginal probabilities for the regional dummy variab les are significant for women heads of household. Regarding the socio-economic status variables, we find positive marginal probabilities for married women with high so cio-economic status but negative marginal probabilities for women heads of household co mpared to women living in poverty. The marginal effects of the intera ction term variables are positiv e and significant for married women, but negative and signi ficant for women heads of household. For single women, only one of the interaction terms had a signi ficant negative impact for the period of 1997-1. Finally, turning to the survey date variables, we find all three subsamples of women showing positive, signifi cant, and increasing probabilitie s of participation in the labor force compared to the 1995-1 period. Again, the Venezuelan economic crisis during this period explains these impacts on the labor force participation. 5.2.4 Results of the Testing of Hypothesis No. 3b In this section, regression results fo r married women, single women, and women heads of household are discussed. In this case, the women’s decisions are whether to work in the formal sector, the informal sector or to be out of the labor force entirely. 5.2.4.1 Results for Married Women The subsample of married women includes 42,791 observations of which 45 percent work in the formal sector, and 22 percent work in the informal sector. Tables 12

PAGE 138

126 and 13 show the coefficients and marginal effects from the multinomial logit regression for married women.111 Married women in all age groups are signi ficantly more likely to work in the informal sector than are those 15 to 20 year s old, as are those 21 to 50, in the formal sector. Those 51 to 60 are 10 percent less like ly and those 21-30 are more likely to be employed in the formal sector. Older married women are more likely to work in the informal sector with marginal probabilities fa irly constant at around 10 percent for all age groups. In the formal sector, we see the highe st probabilities of par ticipating for the two youngest groups (at approximately 5 percent). In general, the marginal effects of age are smaller in both regressions than for the full sample. The greatest differences show up in the formal sector where the marginal effect s are less than one third the size as the corresponding effects for the whole sample among women 40 and younger, and become non-significant or negative for the older age groups. As for education, nearly all the marginal probabilities of labor force participation in both markets are significantly different from those with no education. These results are similar to those obtained for the full sample of women, except that women with only a primary education are no longer more likely to work in the informal sector than are women with no education. As in the whole sample, more education makes it more likely that married women will participate in the fo rmal sector and less likely to work in the informal sector. The marginal effects of having a college or a technical degree are somewhat stronger among married women with respect to the formal sector than among the full sample. 111 This group is comprised of cohabitating women (in the Am erican sense of the term) and women whose husbands are living in the same household.

PAGE 139

127 The results show that living in an urba n area has a positive, significant impact on being employed in the formal sector but not in the informal sector, as for the whole sample. Geographic region does have an important impact on the labor market behavior of married women. For instance, those in the Capital and Insular regions are 4 to 5 percent less likely to participate in the informal sector than those in the North-Eastern region. Conversel y, married women in the Zulian and Central regions exhibit the highest marg inal probabilities of participat ing in the informal sector with 9 and 6 percent, respectively. In the formal sector, married women living in the Capital, Andean, and Central-Western regions are more likely to be employed, with marginal probabilities of 3, 5 and 7 percen tages, respectively. Finally, married women from the Zulian region are the most likely to work in the informal sector, and the least likely to be employed in the formal sector. These results are similar to those obtained from the full sample, except that in the formal sector the marginal probabilities of living in the Central and Insular regions are not significant.

PAGE 140

128 Table 12 Multinomial Logit Regression Results Formal Sector Subsample: Married Women Variable Coefficients Standard Deviations Marginal Effects Standard Deviations Age groupsa 21 to 30 0.354*** 0.037 0.052*** 0.008 31 to 40 0.599*** 0.037 0.048*** 0.008 41 to 50 0.365*** 0.041 0.001 0.009 51 to 60 -0.258*** 0.049 -0.100*** 0.011 Educationb Primary 0.551*** 0.052 0.098*** 0.012 Secondary 1.381*** 0.058 0.270*** 0.012 Technical 2.041*** 0.095 0.363*** 0.013 College 2.296*** 0.073 0.388*** 0.010 Urban residence 0.208*** 0.036 0.039*** 0.008 Regionsc Andean 0.405*** 0.052 0.053*** 0.011 Capital 0.042 0.045 0.031*** 0.010 Central 0.242*** 0.050 0.006 0.011 Central-Western 0.354*** 0.046 0.066*** 0.010 Insular 0.263*** 0.105 0.082 0.023 Plains 0.026 0.079 -0.014 0.017 Zulian -0.184*** 0.047 -0.086*** 0.010 Head of household 1.041*** 0.075 0.114*** 0.013 Socio-Economic statusd High 0.065* 0.037 -0.003 0.008 Medium-High -0.012 0.038 0.004 0.008 Average 0.022 0.041 0.005 0.009 Nonlabor income (US$/month) -0.1026 0.085 -0.005 0.020 Interaction terms Nonlabor income 1997-1 0.914*** 0.222 0.148*** 0.044 Nonlabor income 1997-2 0.596*** 0.193 0.127*** 0.041 Nonlabor income 1998-1 0.671*** 0.191 0.139*** 0.039 Nonlabor income 1998-2 0.690*** 0.182 0.143*** 0.037 Survey datee 1997-1st half 0.068** 0.034 -0.001 0.008 1997-2nd half 0.279*** 0.040 0.030*** 0.009 1998-1st half 0.392*** 0.039 0.035*** 0.009 1998-2nd half 0.450*** 0.040 0.026 0.009 Constant -1.469*** 0.081 N % 19,161 44.8 -2 log likelihood 84,714.50*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is the North-Eastern region. d=omitted category is poverty. e=omitted category is the 1995-1 period.

PAGE 141

129 Table 13 Multinomial Logit Regression Results Informal Sector Subsample: Married Women Variable Coefficients Standard Deviations Marginal Effects Standard Deviations Age groupsa 21 to 30 0.334*** 0.046 0.021*** 0.007 31 to 40 0.895*** 0.045 0.096*** 0.007 41 to 50 0.795*** 0.049 0.107*** 0.009 51 to 60 0.350*** 0.057 0.095*** 0.011 Educationb Primary 0.384*** 0.055 0.012 0.009 Secondary 0.590*** 0.064 -0.055*** 0.009 Technical 0.776*** 0.097 -0.111*** 0.010 College 1.053*** 0.083 -0.101*** 0.008 Urban residence 0.122*** 0.044 0.0007 0.007 Regionsc Andean 0.425*** 0.061 0.031*** 0.010 Capital -0.211*** 0.055 -0.040*** 0.008 Central 0.475*** 0.059 0.061*** 0.019 Central-Western 0.204*** 0.056 -0.002 0.009 Insular -0.163 0.140 -0.052*** 0.018 Plains 0.192** 0.091 0.032** 0.015 Zulian 0.385*** 0.054 0.092*** 0.010 Head of household 1.085*** 0.080 0.068*** 0.012 Socio-Economic statusd High 0.185*** 0.043 0.026*** 0.007 Medium-High -0.072 0.045 -0.011* 0.007 Average 0.010 0.048 -0.0005 0.007 Nonlabor income (US$/month) -0.201* 0.114 -0.024 0.019 Interaction terms Nonlabor income 1997-1 0.774*** 0.260 0.041 0.036 Nonlabor income 1997-2 0.205 0.249 -0.025 0.037 Nonlabor income 1998-1 0.263 0.237 -0.023 0.034 Nonlabor income 1998-2 0.280 0.219 -0.022 0.031 Survey datee 1997-1st half 0.171*** 0.042 0.023*** 0.007 1997-2nd half 0.356*** 0.048 0.033*** 0.008 1998-1st half 0.549*** 0.047 0.056*** 0.008 1998-2nd half 0.732*** 0.047 0.084*** 0.008 Constant -2.146*** 0.094 N % 9,474 22.1 -2 log likelihood 84,714.50*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c= omitted category is the North-Eastern region. d=omitted category is poverty. e=omitted category is the 1995-1 period. Married heads of household are si gnificantly more likely to participate in both markets, although the marginal effect is consid erably stronger in the formal sector with 11 percent compared to 7 percent. The magnitude of the marginal effect in this case is

PAGE 142

130 virtually identical to that of the whole samp le in the formal sector but considerably smaller in the informal sector. As for the whole sample, socio-economic status has no significant effect on the probability of married women participating in th e formal sector. Also similar to the whole sample, high status married women are more lik ely to participate in the informal sector but those of medium-high stat us are less likely to do so. A lthough all of these marginal effects are small, the positive effect for high-status women is slightly larger for married women than for the sample as a whole. Unexpectedly, nonlabor income by itself is not statistically significant for married women in either sector. However, for the in teraction terms between the survey date and nonlabor income, contrary to our expectations, we see positive marginal effects on labor force participation for the formal sector. An additional dollar of nonlabor income per month is predicted to increase the likelihood of being employed in the formal sector by 13 to 15 percent. The intera ction variables have no impact on women’s labor force participation in the informal sector. Compared to the whole sample, the marginal effects of nonlabor income for married women have become significant for all 4 variables, compared with only the latest two we saw in Table 4. Moreover, the coefficients are more than twice as large. In any case, the signs of the coeffici ents and marginal effects are counter to the predictions of economic theory and require additional explanation. Examining finally the survey date va riables, we find that the likelihood of participating in the informal sector is positive and increasing throughout. By the second half of 1998, married women ar e 8 percent more likely to work in the informal sector. The magnitudes of these effects are similar to those of the whole sample. However,

PAGE 143

131 probabilities of being employed in the formal sector are pos itive and significant only for the second half of 1997 and for the first half of 1998 at which times they were around 3 percent more likely to participate, marginal effects very similar to those of the entire sample. The main difference we see is that for married women the dummy variable for the last period is no longer significant. 5.2.4.2 Results for Single Women The subsample of single women cont ains 32,906 observations of which 42 percent work in the formal sector, and 12 percent work in the informal sector. Tables 14 and 15 show the results from the multinomial lo git regression for this group. The tables illustrate that the marginal effects are, in general, significantly different from the omitted categories of the age, education, urban resi dence and head of household dummy variables in both markets. The impact of age on labor force participation is stronger in the formal sector where marginal probabi lities ranging from 21 to 33 percent, compared to the informal sector where they range from 7 to 12 percent. However, in the formal sector the lowest marginal effect on labor force part icipation is among those 51 to 60 years old, while in the informal sector the lowest probability is found for the youngest age group. Compared to the whole sample, the margin al probabilities of the age variables on participation in the formal sector are much higher, while for informal sector are almost identical.

PAGE 144

132 Table 14 Multinomial Logit Regression Results Formal Sector Subsample: Single Women Variable Coefficients Standard Deviations Marginal Effects Standard Deviations Age groupsa 21 to 30 1.846*** 0.032 0.326*** 0.007 31 to 40 2.451*** 0.051 0.318*** 0.009 41 to 50 2.570*** 0.075 0.289*** 0.012 51 to 60 1.688*** 0.084 0.211*** 0.012 Educationb Primary 1.649*** 0.065 0.325*** 0.016 Secondary 1.865*** 0.083 0.381*** 0.015 Technical 1.996 *** 0.095 0.389*** 0.013 College 1.585*** 0.070 0.319*** 0.016 Urban residence 0.154*** 0.045 0.040*** 0.010 Regionsc Andean 0.496*** 0.060 0.096*** 0.014 Capital 0.427*** 0.054 0.106*** 0.012 Central 0.409*** 0.059 0.073*** 0.014 Central-Western 0.381*** 0.055 0.082*** 0.012 Insular 0.115 0.143 0.036 0.034 Plains 0.392*** 0.097 0.083*** 0.022 Zulian 0.432*** 0.056 0.054*** 0.013 Head of household 1.475*** 0.086 0.179*** 0.012 Socio-Economic statusd High 0.042 0.045 0.008 0.010 Medium-High -0.001 0.047 0.006 0.011 Average -0.001 0.052 0.004 0.012 Nonlabor income (US$/month) -0.256*** 0.084 -0.051*** 0.018 Interaction terms Nonlabor income 1997-1 -0.646*** 0.189 -0.135*** 0.042 Nonlabor income 1997-2 0.010 0.176 0.015 0.038 Nonlabor income 1998-1 0.274 0.189 0.065 0.040 Nonlabor income 1998-2 0.260 0.192 0.052 0.041 Survey datee 1997-1st half 0.091** 0.041 0.017* 0.009 1997-2nd half 0.160*** 0.047 0.035*** 0.011 1998-1st half 0.209*** 0.046 0.026*** 0.010 1998-2nd half 0.251*** 0.046 0.023** 0.010 Constant -3.524*** 0.108 N % 13,882 42.2 -2 log likelihood 53,647.16*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is the North-Eastern region. d=omitted category is poverty. e=omitted category is the1995-1 period.

PAGE 145

133 Table 15 Multinomial Logit Regression Results Informal Sector Subsample: Single Women Variables Coefficients Standard Deviations Marginal Effects Standard Deviations Age groupsa 21 to 30 1.706*** 0.049 0.065*** 0.005 31 to 40 2.597*** 0.064 0.110*** 0.008 41 to 50 2.800*** 0.088 0.122*** 0.011 51 to 60 1.988*** 0.099 0.113*** 0.014 Educationb Primary 1.263*** 0.048 0.040*** 0.008 Secondary 0.949*** 0.080 -0.019*** 0.008 Technical 0.768*** 0.126 -0.053*** 0.008 College 0.507*** 0.105 -0.039*** 0.008 Urban residence -0.038 0.064 -0.012* 0.007 Regionsc Andean 0.447*** 0.086 0.018** 0.009 Capital 0.012 0.080 -0.022*** 0.007 Central 0.458*** 0.085 0.026*** 0.009 Central-Western 0.226*** 0.079 0.002 0.008 Insular -0.148 0.241 -0.020 0.020 Plains 0.244* 0.144 0.003 0.014 Zulian 0.811*** 0.078 0.069*** 0.010 Head of household 1.816*** 0.079 0.124*** 0.009 Socio-Economic statusd High 0.51 0.062 0.003 0.006 Medium-High -0.110 0.067 -0.011* 0.006 Average -0.77 0.073 -0.008 0.007 Nonlabor income (US$/month) -0.238** 0.107 -0.011 0.010 Interaction terms Nonlabor income 1997-1 -0.457** 0.229 -0.013 0.021 Nonlabor income 1997-2 0.318 0.215 0.033* 0.019 Nonlabor income 1998-1 -0.058 0.244 -0.009 0.021 Nonlabor income 1998-2 0.198 0.226 -0.007 0.019 Survey datee 1997-1st half 0.095 0.061 0.005 0.006 1997-2nd half 0.092 0.072 0.0008 0.007 1998-1st half 0.433*** 0.067 0.036*** 0.007 1998-2nd half 0.621*** 0.066 0.057*** 0.008 Constant –4.092*** 0.140 N % 4,078 12.4 -2 log likelihood 53,647.16*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is the North-Eastern region. d=omitted category is poverty. e=omitted category is the 1995-1 period. Turning to the education variables, we see similar strong positive marginal effects of education in the formal sector, ranging from 32 to 39 percent, compared to those with no education. In the informal sector, as in the two previous regres sions, negative marginal

PAGE 146

134 probabilities were found, except for those with only a primary education, who were 4 percent more likely to participate. These nega tive effects are greater for higher levels of education. For both sectors, the most signi ficant effects are observed amongst single women with a technical education, who are 39 percent more likely to be employed in the formal sector and 5 percent less likely to work in the informal sector. In the formal sector the positive marginal effects are generally str onger than those observed for the sample as a whole whereas in the informal sector th e negative marginal effects are weaker. Single women living in urba n areas are 4 percent more likely to be employed in the formal sector, an almost identical result as that found for the whole sample. They are 1 percent less likely to work in the informal s ector than are those living in rural areas. In most geographic areas of Venezuela, single women are 5 to 11 percent more likely to participate in the formal sector than in th e North-Eastern region. The highest marginal probability of 11 percent is found for thos e from the Capital region; the lowest probability of 5 percent is exhibit by those from the Zulian region. For the whole sample, women from the latter region ar e less likely to participate in the formal sector compared to women living in the NorthEastern region while the margin al effect for single women is positive. Examining in more detail the informal sector, we find that, as expected, single women living in the Capital region are less lik ely to participate than those women living in the North-Eastern region. Po sitive and significant marginal probabilities of labor force participation were observed among those from the Central, the Andean and the Zulian regions with 3, 2, and 7 percen t, respectively. These differences can be attributed to the varying availability of resources and econom ic activities that exist in each region.

PAGE 147

135 Compared to the full sample, in general the marginal effects of region for single women are larger in the formal sector but smaller a nd less significant in th e informal sector. Single women heads of household are signif icantly more likely to participate in both sectors although the likeli hood is higher in the formal sector with 18 percent compared to 12 percent found in the informal s ector, virtually the same pattern as in the full sample. Also as in the full sample, soci o-economic status has virtually no impact on labor force participation in the formal sector.112 However, in the informal sector, single women with medium-high socio-economic status show negative marginal probabilities of participation indicating that th ey are less likely to participate than t hose living in poverty. This unexpected result is also similar to th at found for the full sample. The effect of nonlabor income is negative, as expected, in the formal sect or, but is not statistically significant in the informal sector, as with the full sample. However, the negative effect is nearly twice as large for single women. When we examine the interaction terms, in the formal sector we find that only for the first half of 1997 is the variable significant. The net effect of nonlabor income is particularly strong for this period: an additional dollar of nonlabor income per month is predicted to re duce labor force participation 19 percent. In contrast, for the whole sample, only the intera ction terms for the last two periods (1998-1 and 1998-2) are significant (but positive). In the informal sect or, the interaction terms are generally insignificant as in the full sample. Finally, the survey date variable sh ows greater marginal probabilities of participating in the formal sector for si ngle women after 1995. The highest marginal 112 Free education in Venezuela might explain this behavior. In other words, higher education allows women to have access to employment in the formal sector regardless of socio-economic status. The so -called democratization of education facilitated the entrance of women into the labor force.

PAGE 148

136 effect occurs during the second half of 1997 with 4 percent. This is generally the same pattern observed for the whole sample. In th e informal sector, the effects of time are smaller and less significant than for the full sample with significant positive marginal effects only for the last tw o periods of 1998, at 4 and 6 percent, respectively. 5.2.4.3 Results for Women Heads of Household The final subsample of women heads of household accounts for 11,365 observations of which 55.7 percent work in th e formal sector, and 31.5 percent, in the informal sector.113 Tables 16 and 17 show the coefficients and marginal probabilities from the multinomial logit regression for women heads of household. There are considerable differences between the results for this subsample and the results for the full sample. For example, this study finds that heads of household over 20 are no more likely to participate in either se ctor than are younger women. This lack of significance may be due to the extremely sma ll percentage of heads of household 20 and younger.114 113 Table B.6 in Appendix B provides aggregat e information from the three subsamples. 114 The results would doubtless be different if the omitted categor y were changed, but it was left the same as in the previous regressions to facilitate comparis ons between results fo r the three subsamples.

PAGE 149

137 Table 16 Multinomial Logit Regression Results Formal Sector Subsample: Women Heads of Household Variable Coefficients Standard Deviations Marginal Effects Standard Deviations Age groupsa 21 to 30 1.661*** 0.231 0.057 0.057 31 to 40 2.303*** 0.221 0.086 0.055 41 to 50 2.080*** 0.217 0.081 0.054 51 to 60 1.204*** 0.217 0.021 0.056 Educationb Primary 0.947*** 0.088 0.140*** 0.016 Secondary 1.974*** 0.134 0.263*** 0.015 Technical 2.376*** 0.332 0.262*** 0.023 College 2.148*** 0.168 0.296*** 0.015 Marital Statusc Married -0.809*** 0.073 -0.075*** 0.016 Cohabitors 0.031 0.180 0.009 0.027 Widows -0.944*** 0.092 -0.102*** 0.016 Divorced -0.027 0.084 -0.005 0.012 Urban residence 0.130 0.099 0.060*** 0.016 Regionsd Andean 0.147 0.137 0.008 0.021 Capital 0.048 0.121 0.070*** 0.018 Central 0.220 0.140 0.019 0.021 Central-Western 0.188 0.125 0.051*** 0.019 Insular -0.216 0.291 0.066 0.050 Plains -0.309 0.188 -0.040 0.032 Zulian -0.309*** 0.125 -0.123*** 0.020 Socio-Economic statuse High -0.087 0.103 -0.024 0.016 Medium High -0.164 0.104 0.0002 0.016 Average -0.184 0.114 -0.025 0.018 Nonlabor income (US$/month) -0.213*** 0.087 -0.022 0.015 Interaction terms Nonlabor income 1997-1 -0.581*** 0.160 -0.052* 0.031 Nonlabor income 1997-2 -0.374*** 0.151 -0.002 0.028 Nonlabor income 1998-1 –0.379*** 0.151 0.038 0.030 Nonlabor income 1998-2 -0.312** 0.157 0.002 0.027 Survey datef 1997-1st half -0.005 0.102 -0.009 0.017 1997-2nd half 0.216* 0.117 0.016 0.018 1998-1st half 0.312*** 0.118 -0.014 0.018 1998-2nd half -0.338*** 0.120 -0.037** 0.018 Constant -0.927*** 0.200 N % 6,330 55.7 -2 log likelihood 19,929.12*** ***, (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is the 1995-1 period.

PAGE 150

138 Table 17 Multinomial Logit Regression Results Informal Sector Subsample: Women Heads of Household Variables Coefficients Standard Deviations Marginal Effects Standard Deviations Age groupsa 21 to 30 1.663*** 0.278 0.033 0.058 31 to 40 2.352*** 0.268 0.065 0.055 41 to 50 2.192*** 0.264 0.082 0.054 51 to 60 1.385*** 0.264 0.073 0.056 Educationb Primary 0.500*** 0.087 -0.064*** 0.015 Secondary 0.930*** 0.139 -0.165*** 0.014 Technical 1.249*** 0.349 -0.172 *** 0.022 College 0.836*** 0.178 -0.201*** 0.014 Marital Statusc Married -0.707*** 0.102 -0.011 0.015 Cohabitors -0.009 0.188 -0.008 0.025 Widows -0.748*** 0.096 0.002 0.015 Divorced 0.060** 0.088 0.008 0.011 Urban residence -0.145 0.102 -0.058*** 0.016 Regionsd Andean 0.149 0.143 0.005 0.020 Capital -0.325*** 0.128 -0.077*** 0.017 Central 0.033 0.146 -0.002 0.020 Central-Western -0.033 0.131 -0.041** 0.018 Insular -0.713** 0.332 -0.104*** 0.043 Plains -0.194 0.195 0.014 0.030 Zulian 0.244** 0.127 0.115*** 0.020 Socio-Economic statuse High 0.016 0.107 0.020 0.015 Medium High -0.218** 0.109 -0.017 0.015 Average -0.110 0.118 0.010 0.017 Nonlabor income (US$/month) -0.159* 0.093 0.005 0.015 Interaction terms Nonlabor income 1997-1 -0.482*** 0.171 0.003 0.030 Nonlabor income 1997-2 -0.479*** 0.169 -0.035 0.028 Nonlabor income 1998-1 -0.699*** 0.182 -0.082*** 0.029 Nonlabor income 1998-2 -0.398** 0.165 -0.029 0.026 Survey datef 1997-1st half 0.053 0.107 0.011 0.016 1997-2nd half 0.194 0.124 0.002 0.017 1998-1st half 0.471*** 0.124 0.045*** 0.017 1998-2nd half 0.614*** 0.125 0.073*** 0.018 Constant -1.057*** 0.317 N % 3,579 31.5 -2 log likelihood 19,929.12*** ***, (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is the 1995-1 period.

PAGE 151

139 The education variables exhibit strong positive marginal probabilities in the formal sector, and strong negative impacts in the informal sector.115 The marginal probabilities of participation in crease in magnitude with the level of education in both sectors. For instance, a wo man head of household with a college education has the highest marginal probability of participati ng, 30 percent in the formal sector, and -20 percent in the informal sector. In the formal sector, the marginal impacts of education are slightly larger in this subsample than in the full sample. In the informal sector, the negative impacts are considerab ly larger for this group. Married and widowed heads of household are 8 and 10 percent, respectively, less likely to participate in the formal sector than single heads of household, similar results as for the full sample but with larger magnitudes for this subsample. However, unlike the full sample, cohabitors and divorced heads of household are not significantly more likely to participate in the formal sect or than single heads of household. Also unlike in the full sample, for h eads of household marital status is not significantly related to participation in the informal sector. This result may be due to the extremely small percentage of single heads of household working in the informal sector (3 percent).116 Living in an urban area increases the likelihood that a woman head of household will be employed in the formal sector by 6 percent, and decreases the probability of 115 Thus their incentive to acquire mo re education, as it allows them to earn more competitive wages. This is not the topic of this study but it is important to mention it. 116 Only 117 single heads of household participate in the informal sector from a total of 3,670 single women heads of household working in both sectors.

PAGE 152

140 working in the informal sector by the same amount, larger and mo re significant impacts than for the whole sample. Turning to the regional variables, we find positive marginal probabilities of participation in the formal sector for wome n heads of household livi ng in the Capital and Central-Western regions, indica ting that they are more likel y to participate than those living in the North-Eastern re gion. In contrast, women head s of household living in the Zulian region are 12 percent less likely to wo rk in the formal sector. As we compare these results to those from the full sample of women, we find that fewer regions from the subsample are significant but the signs and ma gnitudes of these coefficients are generally consistent with those of the larger sample. As for the informal sector, on the other hand, women living in the Zulian region are 12 percent more likely to work in the informal sector, and those from the Capital, the Cent ral-Western and Insular regions are less likely to do so, with marginal probabilities of –8, -4, and –10 percent, respectively. Although results from the full sample show statistical significance for more of the region variables, the coefficients for these four regions are larg er and more significant in this subsample. As in the full sample, socio-economic st atus for women heads of household is not significantly related to labor force participati on in the formal sector. However, unlike in the full sample, neither does socio-economic stat us impact participation in the informal sector. Nonlabor income by itself does not impact participation in either sector for this subsample, unlike the full sample where it has the expected negative influence on being employed in the formal sector.

PAGE 153

141 Examining nonlabor income interacted w ith the survey date, negative probability of participation in the formal sector is found for women heads of household only for the first half of 1997, indicating that an additional dollar of nonlabor income per month variable decreases the probability of part icipation by 5 percent. These results differ considerably from those for the full sample where it is interaction terms for 1998 that are significant in the formal sector. Women heads of household show a negative marginal probability of participation in the informal s ector only in the first half of 1998, indicating that an additional dollar of nonlabor income per month decreases the probability of participation by 8 percent compared to 1995. Th is result is similar to that for the full sample except the impact is over twice as la rger for women heads of household. Turning to the survey date variables, we see that women heads of household are 4 percent less likely to particip ate in the formal sector fo r the second half of 1998 than during the first half of 1995. This contrasts with results fr om the full sample that show positive marginal probabilities of participation in the formal sector after the second half of 1997. As for the informal sector, women heads of household are 5 to 7 percent more likely to participate in 1998-1 and 1998-2 than in 1995-1. The coefficients are slightly smaller and generally less significant than those for the same variables for the full sample. As discussed earlier, the reason for these outcomes is likely the increases in the unemployment rate during those time periods an d the inability of the formal sector to absorb labor.

PAGE 154

142 5.2.4.4 Comparisons Among the Three Subsamples Tables 12, 14, and 16 show the results of the multinomial regressions for the labor market behavior of married, single, and wome n heads of household in the formal sector. These results show that among both marri ed and single women age generally has a strong impact on their particip ation in this sector, although the marginal effects of the latter group are considerably stronger than in the former. For instance, single women 3140 are 32 percent more likely to participate in this sector, whereas married women of the same ages are only 5 percent more likely to do so. Among both subsamples those of all ages are more likely to be participate than the youngest women except for married women 51 to 60 who are 10 percent less likely to do so.117 Turning to the education variables, women from all three subsamples are more likely to work in the formal sector than those with no education, with marginal probabilities increasing with the level of education. Married women with a college education and single women w ith a technical education exhibit the highest marginal effects on participation. In general, e ducation has a somewhat smaller impact on participation in the formal sector among wo men heads of household than among the other two groups. Living in an urban area has similar effects for all three subsamples with marginal probabilities varying only from 4 to 6 percent. As for the geographic areas, living in the Insular region has no significant impact on part icipation in the formal sector for any of 117 As discussed in the previous section, the age variables may not be significant for heads of household because the sample size of the omitted category is so small: only 1 percent of heads of household are under 21 years old.

PAGE 155

143 the subsamples. As for the other regions, we see generally the largest marginal effects among single women: those in all regions are 5 to 11 percent more likely to participate than single women living in the North-Easter n region. Fewer of the marginal effects are significant among married wo men and women heads of household but among both of these groups we see a significant negative effect for living in the Zulian region. Ceteris paribus, socio-economic stat us has no impact on the labor force participation in the formal sector for women of any of the three subsamples. To see the full effect of nonlabor income we need to examine the net effect of the nonlabor income variable plus the interaction term s. Nonlabor income has the expected negative impact generally only among single women, one that is particularly strong in the first half of 1997. Une xpectedly, it has a strong positive impact among married women.118 Finally, the survey date variable has positive and significant marginal effects on the participation of single wome n for all periods, indicating gr eater participation than in 1995. The results for the ot her subsamples are mixed and less consistent. Tables 13, 15, and 17 show the results of the multinomial regressions for married, single, and women heads of house hold with respect to their pa rticipation in the informal sector. As in the formal sector, age has a significant impact only for the subsample of single and married women. However, unlike in the formal sector, the marginal effects of age are generally quite similar for the two groups. Education has a strong impact on labor force participation in this sector as well as in the formal sector. However, unlike in the formal sector, the impact is generally negative. These marginal effects are consid erably stronger among h eads of household: a 118 This result may be caused by the atypical nature of households with any nonlabor income, only 7 percent of the whole sample. Recall that spouse’s income is not included in this variable.

PAGE 156

144 woman head of household with a college degree is 20 percent less likely to participate in the informal sector whereas a single woman with the same education is only 4 percent less likely to do so. Single women and women heads of household living in urban areas are significantly less likely to participate in the in formal sector than are those living in rural areas. This effect is considerably stronger for the latter group. As for the geographic areas, we see some similarities in the results for all three subsamples. For example, women living in the Capital and Insular regions are less likely to participate in the informal sector, and t hose living in the Zulian region are significantly more likely to do so. Although there is cons iderable variability in the levels of significance of the results for the other dummy variables, women living in the Central, Andean and Plains regions seem more likely to participate in this sector and those in the Central-Western region, less likely to do so More of the regional dummy variables are significant for married women than for wome n in the other subsamples, probably since this group is more likely to be place-bound. In general the marginal effects are largest among women heads of household a nd smallest among single women. Socio-economic status (as measured by this survey) seems to have no consistent impact on participation in the informal sect or for any of the subsamples. High status married women are slightly more likely to be employed in the inform al sector than those of poverty status. However, both single and married women of medium high status are slightly less likely to do so. Nonlabor income (eithe r alone or interacted with the survey date) also has no consistent impact on participation in the informal sector.

PAGE 157

145 Finally, marginal probabilities associated with dates of the survey are all positive generally significant, and increasing with time for all three subsamples. These effects are particularly pronounced among married women.

PAGE 158

146 Chapter Six Conclusions This chapter summarizes the main findings regarding the determinants of female labor force participation in Venezuela duri ng the second half of 1990s. Limitations and opportunities for future research are also discussed. 6.1 Main Findings This dissertation is the first attempt to investigate the labor behavior of Venezuelan women at the end of the last century using micro data from 1995-1998. The decision to participate or not to participate in the labor market is studied for the whole sample of women 15-60 years old and also for subsamples of married women, single women and women heads of hous ehold. Labor force particip ation in the formal and informal sectors is also analyzed using the same groups. The main findings drawn from this study are summarized below, organized by the explanatory variables: 6.1.1 Age Age is one of the most important determinants of labor force participation of women. With respect to the simplest decision of whether to work or to be out of the labor market, as expected, the results indicate that women of all ages are significantly more

PAGE 159

147 likely to participate in the labor market th an are women 15 to 20 y ears old. Women 31 to 40 years old show the highest marginal probabi lity, 28 percent. In general, married and single women follow the same pattern although th e marginal effects are much stronger in the latter group. For example, the marginal pr obability associated with the 31-40 year age group is three times higher (42 percent vers us 14 percent). In contrast, among women heads of household, those 41 to 50 years old are the ones exhibiting the highest marginal probability of participation. We also note some differences between the full sample of women and women from the three subsamples with respect to the lowest probability of participation. For the whole sample, women 51 to 60 years old are the ones with the lowest marginal probability, only 15 percent more likely to participate than those 15-20 years old. We see the same pattern among single women and women heads of household; for married women the marginal probability is insignificant. As before, the impact is considerably larger for single women, with a marginal probability of 32 percent compared with only 9 percent for women heads of household. When women face a three-way choice, to participate in the formal sector, to participate in the informal sector, or to be out of the labor market, age impacts women’s labor force participation in the formal and info rmal sectors differently. The impact of age on labor force participation is generally greater in the formal sector than in the informal sector. A number of other differences can be seen. For example, for the full sample, participation in the informal sector increas es with age whereas those 21 to 40 are most likely to participate in the formal sector. Turn ing to the subsamples, the results show that among married women, the marginal effects of ag e are smaller in both sectors than in the full sample. The greatest differences are found in the formal sector where the marginal

PAGE 160

148 effects are less than one third the size of the corresponding effects for the whole sample among women 21 to 40 years old, and become insignificant or negative for the older groups. However, in the informal sector married women over 30 years old are significantly more likely to pa rticipate than the youngest wo men, with generally larger marginal effects than in the formal sector. In contrast the impact of age on the participation of single women is stronger in the formal sector than in the informal sector and stronger than for the sample as a whole. Contrary to what was expected, women heads of household over 20 are no more likely to participate in either sector th an are younger women. The extremely small percentage of women heads of household 20 years old and younger doubtless explains this lack of significance. 6.1.2 Education In general, consistent with the theory of human capital investment literature, the results of this study show that education has a strong impact on women’s decision to participate in the labor market. Examining first the two-way choice, we find that the marginal probabilities are pos itive and generally increasi ng with greater educational attainment both for the whole sample and fo r the three subsamples. However, there are some interesting differences. In general, th e highest marginal probabilities are found for women with a technical or a college edu cation (22-23 percent). Examining next the subsamples, we find that among married women the highest marginal effect (28 percent) is clearly for those with a college degree. Among single women, the highest probabilities are for those with only a primary or seconda ry education, with 34-35 percent marginal

PAGE 161

149 probabilities. It is noteworthy that education exerts considerably smaller effects on women heads of household than among women from the other subsamples. Among this group the highest marginal probabilities are only 9 percent for t hose with secondary, technical and college educations. When women face a three-way decision, the education variables exhibit positive marginal probabilities of labor force particip ation in the formal sector ranging from 15 to 34 percent, generally in creasing with level of education. In contrast, for levels of education higher than primary we find negativ e effects in the informal sector ranging from 5 to 9 percent with the largest marginal effects associated with the highest level of education. For the sample as a whole, bot h the largest positive and largest negative marginal probabilities are found for women w ith technical degrees in the formal and informal sector, respectively. Considering next the results of the three subsamples, we find similar patterns in the formal and informal sectors as for the full sample: women from all three subsamples are more (less) like ly to participate in the formal (informal) sector than those with no edu cation, with marginal probabili ties increasing in magnitude with the level of education. For example, married women exhibit positive and increasing significant probabilities of participation in th e formal sector with the highest marginal probability of 39 percent for those with a colle ge degree, higher than the 32 percent for the whole sample. In the informal sector, the highest negative e ffects are found among married women with technical and college educ ations with marginal probabilities of 1011 percent, slightly larger effects than we find for the whole sample. Among single women in the formal sector the marginal e ffects of education are larger than among the full sample. The greatest difference is for those with a primary education where the

PAGE 162

150 marginal effect is 33 percent compared to 15 percent for the whole sample. The general negative impacts of education upon participation in the inform al sector are smaller among single women than among the whole samp le. However, single women with only a primary education are three times as likely to wo rk in the informal sector as was true for the whole sample. Finally, for women head s of household in the formal sector the marginal probabilities associated with educa tion are somewhat smaller than those of the full sample. In the informal sector, magnitude s of the marginal effects are all negative and considerably larger for this subsample than for the whole sample. Comparing the results among the three s ubsamples, single women with technical degrees and married women with college degrees exhibit the highest marginal probabilities of participation in the formal sector with 39 percent each. These two groups are also among those least likely to particip ate in the informal sector. In general, education has a somewhat smaller impact on participation in the formal sector among women heads of household than among women of the other two groups. In the informal sector, on the other hand, the ne gative impacts are considerably larger for this group. For instance, a woman head of household with a co llege degree is 20 percent less likely to participate in the informal sector whereas a single woman with the same education is only 4 percent less likely to do so. 6.1.3 Marital Status Considering first the impact of the marita l status variable on women’s decision to participate in the labor market for the full sample of women, we see results that are contrary to what expected. For instance, cet eris paribus, married and divorced women are

PAGE 163

151 more likely to participate in the labor market than single women with marginal probabilities of 4 and 16 percen t, respectively, and cohabito rs’ marginal probability of labor force participation is positive and significant (13 perc ent). However, some results are as expected: widows are less likely to participate in the labor market (-4 percent). Examining the subsample of heads of household, we see that married and widowed heads of household are less likely to participate in the labor market than are single heads of household with marginal probabilities of 8 a nd 9 percent, respectively. Cohabitors and divorced heads of household ar e not statistically differe nt from single heads of household, in contrast to the results from the full sample of women, where they were significantly more likely to par ticipate in the labor market. Considering next the three-way choice of women to participate in the formal sector, informal sector, or not to participate at all, we fi nd that being married or widowed has a small negative impact on participati on in the formal sector with marginal probabilities of -1 percent a nd -5 percent, respectively. However, married women are more likely to participate in the informal sector than are single women. Cohabitating women are more likely than single women to participate in both sectors, although the marginal effect is higher in the formal sect or (11 percent versus 3 percent). Widows, as expected, are less likely than single women to participate in the formal sector but, unexpectedly, their labor market behavior is not signi ficantly different from that of single women in the informal sector. Divorced wo men are also more likely than single women to participate in both sectors but the marginal probability is consid erably higher in the formal sector (12 percent versus 5 percent) The highest marginal impacts of marital status on labor force participa tion are for divorced women in th e formal sector where they

PAGE 164

152 are 12 percent more likely to participate than single women. The effects of marital status are generally considerably larger in th e formal than in the informal sector. Finally, with respect to the head of household subsample, married and widowed heads of household are less likely to participate in the formal sector than single heads of household, similar results as for the full sample but with larger magnitudes for this subsample. However, unlike in the full sa mple, cohabitors an d divorced heads of household are not significantly more likely to part icipate in the formal sector than single heads of household. Also unlik e in the full sample, for head s of household marital status is not significantly related to participation in the informal sector. This result may be due to the extremely small percentage of single heads of household (the omitted category) working in the informal sector. 6.1.4 Urban Residence Living in an urban area increases the like lihood that women as a whole participate in the labor market by 3 percent. Results for married and single women show similar marginal probabilities, but for women heads of household this variable has no significant effect. Regarding the decision to participate in th e formal or in the informal sector, we find considerable differences between the sect ors. The marginal effect of living in an urban area on participation in the formal sector is positive for the full sample (4 percent), whereas the variable has no significant impact in the informal sector. When we compare these results to those from the subsamples, we find similar positive marginal probabilities of participation in the formal sector with the highest one among women heads of

PAGE 165

153 household at 6 percent. However, there is considerable disparity among the subsamples with respect to participation in the inform al sector. Single women and women heads of household living in urban areas are significantly less likely to participate in the informal sector than are those living in rural areas. Th is effect is considerably stronger for the latter group and definiti vely contrary to what was found in the full sample. However as in the full sample, the variable is not significant for the subsample of married women. 6.1.5 Regions We find considerable differences among re gions that can be attributed to the varying availability of resources and economic activities that exist in each region. In general, for all women, as expected, almost all of the marginal effects of these dummy variables are positive and significantly different from the North-Eastern area. The highest marginal probabilities of labor force part icipation are found for women living in the Andean, Central, and Central-Western regi ons. The Capital region has the smallest marginal probability. As for the three subsam ples, we also find considerable differences. Only among single women did we find positive and significant probabilities of labor force participation for all regions, except for th e Insular region (which is not significant for any of the three subsamples), compared to the North-Eastern region. None of the marginal probabilities for the regional dum my variables are significant for women heads of household. This contrasts sharply with th e numerous regional differences found for the full sample of women. Probably women who are heads of household have little choice about working, regardless of where they live.

PAGE 166

154 For the whole sample facing a three-wa y choice decision, those living in all regions except for the Zulian and Plains regi ons are more likely to be employed in the formal sector than those living in the Nort h-Eastern region. Only the Zulian region are women less likely to be employed in the form al sector. As for the informal sector, positive significant marginal probabilities are found for women living in the Central, Zulian, Andean and the Plains regions. Thos e in the Zulian region are 8 percent more likely to be employed in this sector; those in the Plains region ar e only 2 percent more likely. Negative marginal effects are found for women living in the Capital and the Insular regions. Comparing results of each subsample to those from the full sample, the marginal effects of region on participation in the formal sector are almost all significant, all larger than in the full sample, and all positive, even in the Zulian region. Among married women fewer of the regional variables are significant than in the full sample, but the signs are generally consistent. Only three of the marginal probabilities for women heads of household are significant, but their signs are consistent wi th those of the whole sample. However, heads of household living in the Zulian region are considerably less likely to be employed in the formal sector than for the samp le as a whole, -12 pe rcent compared to -5 percent. As for the marginal probabilities of pa rticipation in the informal sector, for married women we find generally similar but larger marginal effects than the results from the full sample. For women heads of household, only 4 out of the 7 regional marginal probabilities are significant but the signs are consistent with those of the full sample. In

PAGE 167

155 general, the marginal effect s for this subsample are str onger than those from the whole sample. Turning next to a comparison of results from the subsamples, in the formal sector, we see generally the largest marginal effects among single women, all positive effects varying from 5 to 11 percent. Fewer of the marginal effects are significant among married women and women heads of household but among both of these groups we see a significant negative effect associated with living in the Zulian region. As for the impact of region on the decisi on to participate in the informal sector, we also see some similarities in the results for all three subsamples. For example, women living in the Capital and Insular regions are less likely to participate in the informal sector and those living in the Zulia n region are significantly mo re likely to do so. Although there is considerable variability in the levels of significance of th e results for the other variables, women living in the Central, And ean and Plains regions seem more likely to participate in this sector and those in th e Central-Western region, less likely to do so. More of the regional dummy variables are si gnificant for married women than for women in the other subsamples, probably since this group is more likely to be place-bound. In general the marginal effect s are largest among women heads of household and smallest among single women. 6.1.6 Heads of Household Our analysis of the whole sample indi cates that women heads of household are 24 percent more likely to participate in the labor force than are those who are not. As expected, we find significant results for the subs amples as well, but wi th widely disparate

PAGE 168

156 magnitudes: a marginal probability of 30 pe rcent for single women but only 18 percent for married women. With respect to the three-way choice m odel, we find that all women heads of household are significantly more likely to part icipate in both sect ors, with marginal probabilities of 11-12 percent. In the subs amples of both married and single women, being head of household has a similar effect, a lthough the marginal effect is considerably stronger in the formal sector 11 percent for married heads of household and 18 percent for single heads of household. For married heads of household the magnitude of the marginal effect is virtually identical to that of the whole sample in the formal sector but considerably smaller in the informal sector For single heads of household the marginal effect in the formal sector is very close to that of the whole sample, but considerably larger in the formal sector. 6.1.7 Socio-economic Status Considering first the decision of women whether to participate in the labor market or not, we don’t find much impact for so cio-economic status. The only significant marginal effect is for women of high soci o-economic status who are 2 percent more likely to participate than wo men living in poverty. Results for the three subsamples are not consistent: married women exhibit results ne arly identical to those of the full sample, women heads of household of me dium high status are 2 percen t less likely to participate, but the set of variables has no significant impact on the labor force participation of single women.

PAGE 169

157 With respect to the three-way choice, th is variable has no im pact on participation in the formal sector either for the whole samp le or for any of the subsamples. High status married women are slightly more likely to be employed in the inform al sector than those of poverty status. However, there is no consis tent effect on particip ation in the informal sector for any subsample: both single and ma rried women of medium high status have negative marginal probabilities but socioeconomic status has no impact for women heads of household. 6.1.8 Nonlabor Income and the Interaction Terms To see the full effect of nonlabor income we need to examine the net effect of the nonlabor income variable plus the interaction terms. The results for the whole sample of women in their decision to participate in the labor force are generally negative, as expected, indicating that wome n in general are 3 percent less likely to participate for every additional dollar of income per month.119 As for the subsamples, similar patterns are found for single women and women heads of household with negative marginal probabilities for all periods. Unexpectedly, ma rginal probabilities are positive for married women for all periods after 1995. Turning next to the 3-way choice mode l, we find conflicting results. The net marginal probabilities of participation in the formal sector for the whole sample of women are negative for the survey periods of 1995 and 1997, but become positive and significantly different from 1995-1 for the two survey peri ods in 1998. Single women show the expected negative marginal pr obabilities for the entire period including a 119 This sounds like a large impact but recall that 92 pe rcent of the sample has no nonlabor income at all.

PAGE 170

158 particularly strong effect dur ing the first half of 1997. Unexpectedly, married women exhibit positive strong marginal effects on pa rticipation in this sector throughout all periods. Nonlabor income has, in general, no impact on participation in the formal sector by women heads of household. Nonlabor income generally has no impact on participation in the informal sector, either for the whole sample of women or for the subsamples. 6.1.9 Survey Date This variable is used to see the eff ect of time on the labor force participation decision of women in Venezuela. Results for the two-way c hoice model show, as expected, that the marginal probabilities of the dummies for the survey dates are increasingly positive and significantly different from 1995-1 for the whole sample. Similar results are found for the three s ubsamples. Married women show generally stronger marginal probabilities than those of other subsamples. This behavior might be explained by wives’ desire to preserve the real income of the family during this period when the economic crisis deepened due to successive periods of inflation and the devaluation of the bolvar. Considering next the 3-way choice model, for the full sample, in both the formal and informal sectors, the survey date variable generally has positive, significant increasing marginal effects for all periods, i ndicating greater particip ation than in 1995. The effects which are generally stronger, more significant and m onotonically increasing in the informal sector can be attributed to increases in the unemployment rate in the

PAGE 171

159 formal sector during those time periods and the increasing inability of the formal sector to absorb labor. The regressions for the subsamples yi eld mixed and less consistent results. Married and single women are generally more lik ely to participate in the formal sector over time. Women heads of household seem le ss likely to do so, but only one of the marginal probabilities is si gnificant. However, the results for the informal sector generally resemble those of the whole samp le in that the marginal probabilities are significant and increase over time. These effects are stronger, more significant and clearly monotonically incr easing among married women. 6.2 Limitations 6.2.1 The Presence of Children Although, the Venezuelan data contains information about children, there is no means of linking this data to that of indi vidual women. Consequently, this study cannot shed any light on the impact of children on la bor force participation in general or in a particular sector by those who are mothers. 6.2.2 Socio-economic Status This variable was selected as an explan atory variable to capture the effect of socio-economic conditions, generally a proxy for wealth, on women’s labor force participation. The results reported above s how little or no impact of socio-economic status on participation in the labor market eith er as a whole or in i ndividual sectors. These puzzling results may be attribut ed to a problem in building th e index used in this project

PAGE 172

160 (Parra, J., 2003). Thus this study can add little to knowledge about the impact of wealth on labor force participation among women in Venezuela. 6.2.3 Nonlabor Income In many cases we find either unexpectedl y large or counter-int uitive effects of nonlabor income on women’s labor force participation in Venezuela. These results may be due to the high percentage of women (92 percent) who report none at all. Although the data set does contain information about s pouses, since there are no indicators linking women to their spouses, spouses’ income is not included in this variable. Consequently, the present study probably adds little to our knowledge about the impact of nonlabor income on women’s labor force participation. 6.3 Future Research Further insight into Venezuelan wome n’s labor force participation will be provided by performing Times-Seri es analysis for a period of 30 years, something else which has not been done previously. In part icular, this analysis will answer these questions: To what extent is the decision of Ve nezuelan women affected by the business cycle? In other words, what is the relationship between women’s labor force participation and macroeconomic fluctuations? Among Venezuelan women, does the added-worker effect dominate the discouraged-worker effect in times of recessions?

PAGE 173

161 References Altonji, J. G., and Blank, R.M. 1999. “Race and Gender in the Labor Market.” Handbook of Labor Economics 3C. North-Holland, Amsterdam. Arends, M. November, 1992. “Female Labor Force Participation and Earnings in Guatemala.” In Case Studies on Women’s Employment and Pay in Latin America World Bank Report, no. 12175, Wa shington, D.C., pp. 273-298. ---. November, 1992. “Female Labor Force Pa rticipation and Wages: A Case Study of Panama.” In Case Studies on Women’s Employ ment and Pay in Latin America World Bank Report, no. 12175, Washington, D.C., pp. 349-371. ---. November, 1992. “Women’s Labor Force Pa rticipation and Earnings. The Case of Uruguay.” In Case Studies on Women’s Employment and Pay in Latin America World Bank Report, no. 12175, Washington D.C., pp. 431-449. Ashenfelter, O., and Heckman, J. 1974. “The Estimation of Income and Substitution Effects in a Model of Family Labor Supply.” Econometrica 42, pp. 73-85. Barro, R. J., and Lee, J.-W. 1993. “Int ernational Comparis ons of Education Attainment.” National Bureau of Econo mic Research Working Paper, no. 4349. Becker, G. S. September, 1965. “A Th eory of the Allocation of Time.” Economic Journal 75, pp. 493-517. Ben-Porath, Y. May – June, 1973. “Labor-For ce Participation Rates and the Supply of Labor.” The Journal of Political Economy 81:3, pp. 697-704. --and Gronau, R. January, 1985. “Jewish Mother Goes to Work: Trends in the Labor Force Participati on of Women in Israel, 1955-1980.” Journal of Labor Economics 3:1, Part 2: Trends in Women’ s Work, Education, and Family Building, pp. 310-327.

PAGE 174

162 Bentham, J. 1780. Introduction to the Principles of Morals and Legislation. London: Hafner. Blau, D. 1984. “A Model of Child Nutrition, Fertility, and Women’s Time Allocation.” Research in Opulention Economics 5, pp. 113-135. Blau, F. D., Ferber, M. A., and Winkler, A. E. 2002. The Economics of Women, Men, and Work Prentice Hall. 4th ed. Borjas, G. J. 2000. Labor Economics McGraw-Hill, 2nd ed. Bowen, W. G., and Finegan, A. 1969. The Economics of Labor Force Participation. Princeton University Press. Cain, G. G. 1966. “Married Women in the Labo r Force.” University of Chicago Press. --and Dooley, M.D. August, 1976. “Estimation of a Model of Labor Supply, Fertility, and Wages of Married Women.” The Journal of Political Economy 84:4, Part 2: Essays in Labor Economics in Honor of H. Gregg Lewis, pp. 179-200. Casique, I. 1994. “Women and Labor Ma rket. Recent Changes in the Capital Region in Venezuela.” Andres Bello Catholic University. Academic Paper. CEPAL. 1999. Economic Comm ission for Latin America. Economic Report CIDEIBER. 1997. Center of Information a nd Entrepreneurship Documentation for Latin America. Clark, K. B., and Summers, L. H. 1982. “Labor Force Participation: Timing and Persistence.” The Review of Economic Studies 49:5, Special Issue on Unemployment, pp. 825-824.

PAGE 175

163 Coleman, M. T., and Pencavel, J. July, 1993. “Trends in Market Work Behavior of Women Since 1940.” Industrial and Labor Relations Review 46:4, pp. 653676. Colombino, U., and De Stavola, B. January 1985. “A Model of Female Labor Supply in Italy Using Cohort Data.” Journal of Labor Economics 3:1, Part 2: Trends in Women’s Work, Education, and Family Building, pp. 275-292. Costa, D. L. Fall, 2000. “From Mill Town to Board Room: The Rise of Women’s Paid Labor.” Journal of Economic Perspective 14:4, pp. 101-122. Cox, D., and Psacharopoulos, G. November, 1992 “Female Participation and Earnings in Venezuela, 1987.” In Case Studies on Women’s Em ployment and Pay in Latin America World Bank Report, no. 12175, Washington, D.C., pp. 451-461. del Boca, D. 1988. “Women in a Changing Workplace: The Case of Italy.” In Feminization of the Labor Force New York-Oxford: Oxford University Press, pp. 120-136. Domencich, T. A., and McFadden, D. 1975. Urban Travel Demand: A Behavioral Analysis Amsterdam: North-Holland Publishing Company. Durand, J. 1975. The Labor Force in Economic Development: A Comparison of International Census Data, 1946-66. Princeton University Press. Eckstein, Z., and Wolpin, K. July, 1989. “Dynamic Labour Force Participation of Married Women and Endogeneous Work Experience.” The Review of Economic Studies 56:3, pp. 375-390. Ehrenberg, R. G., and Smith, R. S. 2000. Modern Labor Economics. Theory and Public Policy. Addison Wesley Longman, Inc., 7th ed. Elster, S., and Kamlet, M.S. 1990. “R eference Groups, Occupational Class and Married Women’s Labor Force Participation.” Research in Human Capital and Development 6, pp. 89-123.

PAGE 176

164 Fields, J. M. Autumn, 1976. “A Comparison of Intercity Differences in the Labor Force Participation Rates of Married Women in 1970 with 1940, 1950, and 1960.” Journal of Human Resources 11:4, pp. 568-577. Finn, M., and Jusenius, C. 1975. “The Pos ition of Women in the Ecuadorian Labor Force.” Ohio State University: Center for Human Resource Research, Working Paper. Franz, W. January, 1985. “An Economic Analysis of Female Work Participation, Education, and Fertility: Theory and Empirical Evidence for the Federal Republic of Germany.” Journal of Labor Economics 3:1, Part 2: Trends in Women’s Work, Education, and Family Building, pp. 218-234. Gill, I. A. November, 1992. “Is There Sex Discrimination in Chile? Evidence from CASEN Survey.” In Case Studies on Women’s Em ployment and Pay in Latin America World Bank Report, no. 12175, Washington D.C., pp. 119-147. Goldin, C. Spring, 1983a. “The Changing Economic Role of Women: A Quantitative Approach.” Journal of Interdisciplinary History 13:4, The Measure of American History, pp. 707-733. ---. 1983b. “The Female Labor Force and American Economic Growth: 1890 to 1980. In Long-Term Factors in American Economic Growth Conference on Income and Wealth, Chicago: University of Chicago Press, 51, pp. 557-604. ---. January, 1989. “Life-Cycle Labor-Force Participation of Married Women: Historical Evidence and Implications.” Journal of Labor Economics 7:1, pp. 20-47. ---. September, 1991. “The Role of World Wa r II in the Rise of Women’s Employment.” The American Economic Review 81:4, pp. 741-756. ---. April, 1994. “The U-shaped Female Labor Force Function in Economic Development and Economic History. ” Working Paper, no. 4707. NBER.

PAGE 177

165 Granger, C. W. J., and Newbold, P. 1976. “Spurious Regressions in Econometrics.” Journal of Econometrics 2:2, pp. 111-120. Greene, W. H. 2003. Econometrics Analysis New York University Press, 5th ed. Gregory, R. G., McMahon, P., and Whittingham, B. January, 1985. “Women in the Australian Labor Force: Tre nds, Causes and Consequences.” Journal of Labor Economics 3:1, Part 2: Trends in Women’s Work, Education, and Family Building, pp. 293-309. Gronau, R. December, 1997. “Leisure, Home Production, and Work – the Theory of the Allocation of Time Revised.” The Journal of Political Economy 85:6, pp. 1099-1124. Gustafsson, S., and Stafford, F. 1992. “Ch ild Care Subsidies and Labor Supply in Sweden.” Journal of Human Resources 27:1, pp. 204-230. --and Jacobson, R. January, 1985. “Trends in Female Labor Force Participation in Sweden.” Journal of Labor Economics 3:1, Part 2: Trends in Women’s Work, Education, and Family Building, pp. 256-274. Heckman, J. J. July, 1974. “Shadow Pri ces, Market Wages, and Labor Supply.” Econometrica 42:4, pp. 679-694. ---. May, 1978. “A Partial Survey of R ecent Research of the Labor Supply of Women.” The American Economic Review 68:2, Papers and Proceedings of the Ninetieth Annual Meeting of the American Economic Association, pp. 200-207. --and Willis, R.J. February, 1977. “A Beta-logistic Model for the Analysis of Sequential Labor Force Participation by Married Women.” The Journal of Political Economy 85:1, pp. 27-58. --and MaCurdy, T. E. January, 1980. “A Life Cycle Model of Female Labour Supply.” The Review of Economic Studies 47:1, Econometrics Issue, pp. 47-74.

PAGE 178

166 Hernandez, I., Riboud, F. and M. Janu ary, 1985. “Trends in the Labor Force Participation of Spanish Women: An Interpretive Essay.” Journal of Labor Economics 3:1, Part 2: Trends in Women’s Work, Education, and Family Building, pp. 59-90. Hill, M. A. August, 1983. “Female Labor Force Participation in Developing and Developed Countries: Considera tion of the Informal Sector.” The Review of Economics and Statistics 65:3, pp. 459-468. ---. Spring, 1984. “Female Labor Force Partic ipation in Japan: An Aggregate Model.” The Journal of Human Resources 19:2, pp. 280-287. ---. Winter, 1989. “Female Labor Supply in Japan: Implications of the Informal Sector for Labor Force Participa tion and Hours of Work.” The Journal of Human Resources 24:1, pp. 143-161. INE. National Institute of Statististics. Annual Report, 1978-1995 Caracas, Venezuela. ILO. 1990. International Labor Office. Yearbook of Labor Statis tics: Retrospective Edition 1950-1990 Genova: International Labor Office. Jaffe, A., and Azumi, K. October, 1960. “T he Birth Rate and Cottage Industries in Underdeveloped Countries.” Economic Development and Cultural Change 9, pp. 52-63. Jakubson, G., and Psacharopoulos, G. November 1992. “The Effects of Education on Female Labor Force Participati on and Earnings in Ecuador.” In Case Studies on Women’s Employment and Pay in Latin America World Bank Report, no. 12175, Washington, D.C., pp. 255271. Khandker, S. November, 1992. “Women’s Labor Market Participa tion and Male-Female Wage Differences in Peru.” In Case Studies on Women’s Employment and Pay in Latin America. World Bank Report, no. 12175, Wa shington, D.C., pp. 373-395. Killingsworth, M., and Heckman, J. 1986. “Female Labor Supply.” In Handbook of Labor Economics 1, Elsevier Science Publishers.

PAGE 179

167 Kniesner, T. J. October, 1976. “The FullTime Workweek in the United States: 19001970,” Industrial and Labor Relations Review 30, pp. 3-15. Kosters, M. 1969. ”Effects of an Income Tax on Labor Supply. In The Taxation of Income from Capital Washington. Layard, R., and Mincer, J. January, 1985. “T rends in Women’s Work, Education, and Family Building.” Journal of Labor Economics 3, supplement. Ledezma, L. R., Orlando, M. B., and Zuniga, G. March, 2003. “Determinants of the Labor Force Participation of Women in Venezuela, 1980-2000.” Report prepared for the XXIV Latin American Studies Association. International Congress, Dallas, Texas, USA. Lewis, H. G. May – December, 1968. “Labor Force Participation and the Theory of Working Hours.” Economic Science Department Magazine pp. 49-63. Magnac, T. “Female Labor Market Partic ipation and Wages in Colombia.” In Case Studies on Women’s Employme nt and Pay in Latin America World Bank Report, no. 12175, Washington, D.C., pp. 169-195. Mammen, K., and Paxson, C. Fall, 2000. “Women’s Work and Economic Development.” Journal of Economic Perspectives 14:4, pp. 141-164. McFadden, D. 1974. “Analysis of Qualitative Choice Behavior.” Frontiers of Econometrics New York: Academic Press. ---. Fall, 1976. “Quantal Choice Analysis: A Survey.” Annual Economic Social Measurement 5, pp. 363-390. Mendez Castellano, H., and de Mendez, M. C. 1994. “ Society and Social Stratification. Graffar-Mendez Castellano Method.” Fundacresa Caracas, pp. 7-35.

PAGE 180

168 Michael, R. T. January, 1985. “Consequences of the Rise in Female Labor Force Participation Rates: Questions and Probes.” Journal of Labor Economics 3:1, Part 2: Trends in Women Work, Edu cation and Family Building, pp. 117-146. Mincer, J 1962. “Labor Force Particip ation of Married Women.” In Aspects of Labor Economics Universities National Bur eau of Economic Research Conference Studies, 14, Princeton, NJ: Pr inceton University Press, pp. 63-97. Moffitt, R. A. 1999. “New Developments in Econometric Methods for Labor Market Analysis.” Handbook of Labor Economics 3. Elsevier Science Publishers. Mroz, T. July, 1987. “The Sensitivity of An Empirical Model of Ma rried Women’s Hours of Work to Economic and Statistical Assumptions.” Econometrica 55, pp. 765-800. Nakamura, A., and Nakamura, M. Spring, 1994. “Predicting Female Labor Supply.” Journal of Human Resources, 29, pp. 304-327. National Census of Venezuela. 1950, 1991, 2000. Ng, Y. C. November, 1992. “Female Labor Force Participation and Gender Earnings Differentials in Argentina.” In Case Studies on Women’s Employment and Pay in Latin America World Bank Report, no. 12175, Washington, D.C., pp. 1-20. Nicholson, W. 1992. The Microeconomic Theory fifth edition. The Dryden Press. OCEI. 2002. 30 Years of Performance w ith No Interruptions 1967-1997 The Central Office of Information and Statistics. Compact Disc Digital Data. Venezuela. OCEI. The Central Office of Inform ation and Statistics. 1990, 1998. OECD. Organization for Economic Co -operation and Development, Labor Force Statistics Paris.

PAGE 181

169 Ofer, G., and Vinokur, A. January, 1985. “Wor k and Family Roles of Soviet Women: Historical Trends and Cross-Section Analysis.” Journal of Labor Economics 3:1, Part 2: Trends in Women’s Work, Edu cation, and Family Building, pp. 328-354. Oi, W. Y. December, 1962. “Labor as a Quasi-Fixed Factor.” Journal of Political Economy 70:6, pp. 538-555. OIT. 1999. Labor International Office. Panorama of Labor for Latin American and the Caribbean Countries .” Oliveira, O. January, 1997. “Multiple Analytic Perspectives on Women’s Labor in Latin America.” Current Sociology 45. O’Neill, J. A. May, 1981. “A Time Seri es Analysis of Women’s Labor Force Participation.” The American Economic Review 71:2, Papers and Proceedings of the Ninety-Third Annual Meeting of the American Economic Association, pp. 76-80. Orlando, M. B. 2001. “The Informal Sector in Venezuela: Catalyst or Hindrance for Poverty Reduction.” Institute of Economics and Social Investigations. Andres Bello Catholic University, Caracas, Venezuela. --and Zuniga, G. March, 2000. “Women’s Situation in the Labor Market in Venezuela: Analysis of Female Par ticipation and the Income Gap by Gender.” Conference prepared for the XXII Congress of the Association of Latin American Studies Palmer, G. L. 1954. Labor Mobility in Six Cities: A Report on the Survey of Patterns and Factors Mobility, 1940-1950 New York: Social Science Research Council. Parra, J. June, 2003. “Theoretical-Methodological Reflexion about the Justification of a Social Typology for Venezuela.” Cent er of Statistics and Operational Investigations. Economic and Social Scie nces Department. University of Zulia, Maracaibo.

PAGE 182

170 Perry-Castaneda Library. Map Collection. The University of Texas at Austin. http//www.lib.utexas.edu/maps Psacharopoulos, G., and Tzannatos, Z. July, 1989. “Female Labor Force Participation: An International Perspective.” The World Bank Research Observer 4. ---. November, 1992. “Women’s Employment and Pay in Latin America.” World Bank Regional and Statistical Studies Report no. 11360, Washington D.C., pp. 1-250. Riboud, M. January, 1985. “An Analysis of Women’s Labor Force Participation in France: Cross-Section Estimates and Time-Series Evidence.” Journal of Labor Economics 3:1, Part 2: Trends in Women’s Work, Education, and Family Building, pp. 177-200. Riveros, J. A., and Sanchez, C. E. 1990. “Argentina’s Labor Markets in an Era of Adjustments.” World Bank, Working Paper, no. 386, Washington, D.C. Santos, M. A. June, 2003. “1989,1996 and 2002 Th ree Crises and Three Adjustments. How much was the Cost for the Real S ector of the Venezuelan Economy.” IESA, Caracas. Schultz, T. P. July, 1991. “International Diffe rences in Labor Force Participation in Families and Firms.” Yale University Working Paper. Scott, K. November, 1992. “Women in the Labor Force in Bolivia: Participation and Earnings.” In Case Studies on Women’s Employment and Pay in Latin America World Bank Report, no. 12175, Washington, D.C., pp. 21-38. Shimada, H., and Higuchi, Y. January, 1985. “A n Analysis of Trends in Female Labor Force Participation in Japan.” Journal of Labor Economics 3:1, Part 2: Trends in Women’s Work, Education, a nd Family Building, pp. 355-374. Sito, G. U., and Grau, P. C. 1998. Venezuelan Geography 9th grade Santillana Editors. Caracas.

PAGE 183

171 Smith, J. P., and Ward, M. P. January, 1985. “Times-Series Growth in the Female Labor Force.” Journal of Labor Economics 3:1, Part 2: Trends in Women’s Work, Education, and Family Building, pp. 59-90. Smith, S. K. 1984. “Determinants of Female Labor Force Participation and Family Size in Mexico City.” Economic Development and Cultural Change 30:1, pp. 129-152. Sorrentino, C. March, 1990. “The Changing Family in International Perspective.” Monthly Labor Review 113:3, pp. 41-58. Standing, G. 1982. “Female Labor Supply in an Urbanizing Economy” in G. Standing and G. Sheehan (eds.) Labor Force Participation in Low-income Countries Genova: International Labor Organization. Steels, D. November, 1992. “Women’s Partic ipation Decision and Earnings in Mexico.” In Case Studies on Women’s Employment and Pay in Latin America World Bank Report, no. 12175, Washington, D.C., pp. 339-348. Stelcner, M., Smith, J. B., Breslaw, J. A., and Monette, G. November, 1992. “Labor Force Behavior and Earnings of Brazilian Women and Men, 1980.” In Case Studies on Women’s Employme nt and Pay in Latin America World Bank Report, no. 12175, Washington, D.C., pp. 39-88. Stutzer, A. May, 2004. “The Role of In come Aspirations in Individual Happiness.” Journal of Economic Behavior and Organization 54:1, pp. 89-109. Summers, R., and Heston, A. May, 1991. “The Penn World Tables (Mark 5): An Expanded Set of Interna tional Comparisons, 1958-1988.” Quarterly Journal of Economics 106, pp. 327-68. Tiefenthaler, J. November, 1992. “Femal e Labor Force Participation and Wage Determination in Brazil, 1989.” In Case Studies on Wome n’s Employment and Pay in Latin America World Bank Report, no. 12175, Washington, D.C., pp. 89-118.

PAGE 184

172 United Nations. 1992. WISTAT: Women’s I ndicators and Statistics Spreadsheet Database for Microcomputers (Version 2): Users Guide and Reference Manual. New York: United Nations. Vacchino, J. M. 1981. “Economic and Regional Integration.” College of Political and Judicial Sciences. Central University of Venezuela. Caracas, Venezuela. Velez, E., and Winter, C. November, 1992. “Women’s Labor Force Participation in Colombia.” In Case Studies on Women’s Employment and Pay in Latin America World Bank Report, no. 12175, Washington, D.C., pp. 197-206. Weiss, Y. 1972. “On the Optimal Lifetime Pattern of Labor Supply.” Economic Journal 82, pp. 1295-1315. --and Gronau, R. October, 1981. “Exp ected Interruptions in Labour Force Participation and Sex Related Di fferences in Earnings Growth.” The Review of Economic Studies 48:4, pp. 607-619. Winter, C. November, 1992. “Female Earn ings, Labor Force Participation and Discrimination in Venezuela, 1989.” In Case Studies on Women’s Employment and Pay in Latin America World Bank Report, no. 12175, Washington, D.C., pp. 463-475. --and Gindling, T. November, 1992. “Women’s Labor Force Participation and Earnings in Honduras.” In Case Studies on Women’s Employment and Pay in Latin America World Bank Report, no. 12175, Wa shington, D.C., pp. 299-322. ---. 1989. “Country Assessment of Women’ s Role in Development: Proposed Bank Approach and Plan of Acti on.” R.N. 8064-BO. Latin America and Caribbean Region, Country Operations Division 1. Washingt on, D.C.: The World Bank. Yang, H. November, 1992. “Female Labor Forc e Participation and Earnings Differentials in Costa Rica.” In Case Studies on Women’s Em ployment and Pay in Latin America World Bank Report, no. 12175, Washington, D.C., pp. 209-222.

PAGE 185

173 Zabel, J. April, 1993. “The Relationship between Hours of Work and Labor Force Participation in Four Models of Labor Supply Behavior.” Journal of Labor Economics 11, pp. 387-416.

PAGE 186

174 Appendices

PAGE 187

175 Appendix A: Definiti on of Variables A.1 The Formal Sector vers us the Informal Sector Technically, whether a market sector is cl assified as informal or formal depends on whether labor activities are subject to ta xation and other regulat ions. For the purpose of this project, the informal s ector includes self-employed workers,120 owners of micro enterprises, helpers. The formal sector includes all other workers. A.2 Urban Area This variable includes women living in the metropolitan area of Caracas, in the main cities of Venezuela, and in cities with 25,000 residents or less if they are close to a metropolitan area. Rural areas include resident s of cities with less than 25,000 residents not belonging to the metropo litan areas of Venezuela. A.3 Administrative Regions of Venezuela The nine political-administrativ e regions are described below. A.3.1 Andean Region The Andean region includes the states of Tchira, Mrida, Trujillo, Barinas and the Paez municipality in the st ate of Apure. The region has a long agricultural tradition of producing coffee, vegetables, flowers, peaches plantains and Yuca. Cattle raising is 120 This dissertation uses the methodology used by the O ffice of Statistics and Inform ation (OCEI) to define the informal sector and the formal sector. For instance, the self-employed worker category does not include professionals or technicians.

PAGE 188

176 predominant in the south of the Lake of Maracaibo, in Barinas and in the Paez district where about 30 percent of the cattle in Venezuela are located as well as the principal milk powder producing plants. However, reside nts have been leaving the region because of a lack of land for the expansion of agricu lture. There are also other activities such as tourism, mining, craftwork, fish farming, and small and medium sized industries. A.3.2 Capital Region The capital region is composed of the fe deral capital (Caracas) and the states of Vargas and Miranda. It is the center of po litical power as well as of commercial, industrial and financial activities in the country There are also other traditional activities, such as agriculture, in this region. However, agricultural activity has progressively decreased and exists only in the periphery of the district, in the Barlovento sub region, and in the Miranda slopes. The growth of cacao flowers and vegetables and the raising of pigs, represents the survival of agricultural activities which have been able to take advantage of the proxim ity of large markets. Traditionally, the capital region also opera ted as a reception center for inmigrants which contributed to serious urban planning problems characterized by water shortages, air and noise pollution, a shortage of serv ices and a lack of recreational areas. A.3.3 Central Region The central region is composed of the Carabobo, Cojedes and Aragua states. The main characteristic of this region is its indus trial activity, a lthough it is less developed in the Cojedes state. It was historically the basis of the most prosperous agricultural

PAGE 189

177 activities in the country. Howe ver, with the emergence of the oil economy, most of the agricultural land was taken over fo r urban and industrial use. Th e agricultural activity that still exists is linked to the pr oduction of food, drinks, textile s and tobacco. In the state of Cojedes, cattle-raising is important, as we ll as the cultivation of rice, sorghum and ajonjol seed. The presence of Puerto Cabello as the main national port linked the region to the other more industrialized areas. A.3.4 Central-Western Region This region is composed of the states of Lara, Falcn, Yaracuy and Portuguesa. It has great geographical diversity that allows for a variety of economic activities, including agriculture, oil and mining. Barquisimeto an important urban-industrial center is found in this region. Oil and petrochemical activities are found in Falcn a nd Punto Fijo whereas mining activities are concentrated in Lara. An important agricu ltural area in Portuguesa produces sorghum, rice and cotton. Onions, potatoes, sugar cane, maize and poultry are produced in the valleys of Quibor. A.3.5 Guyana Region The Guyana region includes the states of Bolvar, Amazonas, and Delta Amacuro. It represents the greatest fo rest reservoir and water recourses of the country. It is identified as a mining region, producer of hydr oelectric energy and forestal resources. The region contains almost 50 percent of the land in Venezuela but has less than 6 percent of the total populati on of the country. More than 60 percent of the regional

PAGE 190

178 population lives between Ordaz Port and Ciuda d Bolivar. One-thir d of the country’s indigenous people live in this region. Mining is developed around the basic industries of iron and aluminum. There is also agricultural industries, cattle rais ing, logging, fishing and tourism industries. Guayana was dropped fr om the analysis because it has the newest cities of Venezuela for the residents of whic h socio-economic status codes have not been established. A.3.6 Insular Region The insular territories that compose th e region are the state of Nueva Esparta (Margarita Island), the islands located in th e Caribbean Sea, the islands located in the Gulf of Paria. The state of Nueva Esparta is at the heart of the Insular region that has traditionally specialized in fishing, commerci al activities and especially tourism due to the tax free zone on Margarita Island. Mor eover, the many people moving to Margarita Island have caused a boom in the construction activities. A.3.7 North-Eastern Region The North-Eastern region is composed of the states of Anzotegui, Sucre and Monagas. Historically, the main agricultu ral activity was the production of cacao and coffee as well as tobacco and citrus products. The emergence of oil as an activity took prominence over agricultural activities cr eating a new axis of development around hydrocarbons, especially in the south and center of Monagas and Anzotegui. Agricultural activities have survived with th e inclusion of new crops such as peanuts,

PAGE 191

179 sorghum, sunflowers, and the growth of pines for paper pulp. Tourism is also important given the beauty of the beaches in the region. A.3.8 Plains Region The Plains region is compri sed of the states of Guaric o, and Apure except for the Paz Municipality. Around 40 percent of the pop ulation is concentrated in 5 populated centers: San Juan de los Morro s, Altagracia, Calabozo, Zaraza, and San Fernando. The rest of the population is disperse d around this extensive region’s area. This region has a predominantly agricultur al character. Sales of beef, rice, corn, cotton, and tobacco are especi ally important. Mining is a pr omising activity due to the existence of the oil band in the Orinoco Rive r and of the limestone that the construction industry demands. A.3.9 Zulian Region The Zulian region is composed only of the state of Zulia. It is characterized not only by its great oil pote ntial, but also by other economic activities such as agriculture, mining, commerce, craftwork, petro-chemical industries, and coal mining. Zulia is also one of the main producers of agricultura l and cattle products including milk, meat, cheese, sugar cane, and coconut. Trade is also an important activity especially in the large cities such as Maracaibo, Cabimas, a nd Lagunillas. Because it is a border region, trade with neighboring Colombia is active and growing. Ho wever, the region also has problems with contraband, drug trafficking, and kidnappings, which have increased due to the effects of the civil war in Colombia.

PAGE 192

180 A.4 Socio-economic Status These four variables depict the social st ratification of people living in Venezuela. They are based on an index constructed using the Mendez Castellano-Graffar Method which categorizes households according to diffe rent levels of economic and social wellbeing. The index is derived from four vari ables: the profession of head of household, level of education of the spouse or partner of head of household, the main source of income of the household, and th e physical condition of the house.121 Each one of these variables is composed of five items which have been assigned di fferent weights; the summation of the items’ weights determines th e distribution of the households into five socio-economic strata. The gr eater the number of points, the lower the socio-economic strata. The first three strata (with the fewest points) co mpose the group of the non-poor. Stratum IV, “relative poverty” is the next to poorest. It does not imply absolute deprivation but contains the unemployed working class with some education. Stratum V corresponds to critically poor households suffering a very high level of deprivation (Hernan Mndez Castellano and Maria Cristina de Mndez, 1994). For purposes of this study, the later two categories ar e combined as “poverty level.” 121 This categorization includes a combina tion of the sanitary conditions, the degr ee of luxury, and the size of the house.

PAGE 193

181 A.5 Nonlabor Income This variable includes a pension given to the survivor after the death of a family member, financial assistance of a family me mber to another member, a pension received through social security, retirement funds, rent al income, and interests or dividends. Labor income of a spouse or other family members is not included.

PAGE 194

182 Appendix B: Tables Table B.1 Binomial Probit Regression Results Coefficients Sample: All Women 15-60 V a a r ri a a b ble Coefficients Standard Deviations z-values Age groupsa 21 to 30 0.723*** 0.013 56.86 31 to 40 0.938*** 0.014 64.99 41 to 50 0.850*** 0.017 51.53 51 to 60 0.469*** 0.019 24.09 Educationb Primary 0.399*** 0.021 19.26 Secondary 0.654*** 0.023 28.74 Technical 0.843*** 0.032 26.63 College 0.749*** 0.025 29.51 Marital statusc Married 0.121*** 0.012 10.42 Cohabitors 0.401*** 0.048 8.29 Widows -0.090*** 0.035 -2.58 Divorced 0.497*** 0.022 22.29 Urban residence 0.085*** 0.015 5.75 Regionsd Andean 0.228*** 0.021 11.07 Capital 0.065*** 0.018 3.60 Central 0.189*** 0.020 9.40 Central-Western 0.172*** 0.019 9.26 Insular 0.013 0.044 0.29 Plains 0.067** 0.032 2.11 Zulian 0.109*** 0.019 5.87 Head of household 0.793*** 0.020 39.60 Socio-economic Statuse High 0.053*** 0.015 3.51 Medium High -0.008 0.016 -0.53 Average -0.0002 0.017 -0.01 Nonlabor income (US$/ month) -0.085*** 0.026 -3.23 Interaction terms Nonlabor income 1997-1 0.006 0.054 0.12 Nonlabor income 1997-2 -0.017 0.052 -0.32 Nonlabor income 1998-1 0.036 0.053 0.68 Nonlabor income 1998-2 0.067* 0.053 1.27 Survey datef 1997-1st half 0.056*** 0.014 4.00 1997-2nd half 0.152*** 0.016 9.47 1998-1st half 0.225*** 0.016 14.14 1998-2nd half 0.273*** 0.016 17.09 Constant -1.238 *** 0.032 -38.63 N 86,199 -2* log likelihood ratio 97,140*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is 1995-1 period.

PAGE 195

183 Table B.2 Binomial Probit Regression Results Marginal Effects Sample: All Women 15-60 V V a a r r i i a a b b l l e e Marginal Effects Standard Deviations z-values Age groupsa 21 to 30 0.238*** 0.004 63.95 31 to 40 0.293*** 0.004 79.47 41 to 50 0.261*** 0.004 64.52 51 to 60 0.154*** 0.006 27.42 Educationb Primary 0.146*** 0.008 19.29 Secondary 0.215*** 0.007 32.93 Technical 0.239*** 0.006 38.75 College 0.228*** 0.006 37.86 Marital statusc Married 0.044*** 0.004 10.42 Cohabitors 0.131*** 0.014 9.52 Widows -0.034*** 0.013 -2.54 Divorced 0.161*** 0.006 25.99 Urban residence 0.031*** 0.006 5.68 Regionsd Andean 0.079*** 0.007 11.62 Capital 0.024*** 0.006 3.63 Central 0.066*** 0.007 9.75 Central-Western 0.061*** 0.006 9.51 Insular 0.005 0.016 0.29 Plains 0.024** 0.011 2.14 Zulian 0.039*** 0.007 5.97 Head of household 0.241*** 0.005 51.55 Socio-economic Statuse High 0.019*** 0.005 3.51 Medium High -0.003 0.006 -0.53 Average -0.00007 0.006 -0.01 Nonlabor income (US$/ month) -0.031*** 0.010 -3.23 Interaction terms Nonlabor income 1997-1 0.002 0.020 0.12 Nonlabor income 1997-2 -0.006 0.019 -0.32 Nonlabor income 1998-1 0.013 0.019 0.68 Nonlabor income 1998-2 0.024 0.019 0.68 Survey datef 1997-1st half 0.020*** 0.005 4.03 1997-2nd half 0.054*** 0.006 9.70 1998-1st half 0.079*** 0.005 14.71 1998-2nd half 0.095*** 0.005 17.98 -2* log likelihood ratio 97,140*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is 1995-1 period.

PAGE 196

184 Table B.3 Multinomial Probit Re g ression Results Formal Sector Sample: All Women 15-60 Variable Coefficients Standard Deviations Marginal Effects Standard Deviations Age groupsa 21 to 30 0.992*** 0.018 0.193*** 0.005 31 to 40 1.205*** 0.020 0.197*** 0.005 41 to 50 1.048*** 0.024 0.141*** 0.006 51 to 60 0.502*** 0.028 0.043*** 0.008 Educationb Primary 0.612*** 0.031 0.140*** 0.008 Secondary 1.090*** 0.034 0.272*** 0.008 Technical 1.407*** 0.045 0.338*** 0.009 College 1.269*** 0.037 0.315*** 0.008 Marital statusc Married 0.087*** 0.017 -0.011*** 0.004 Cohabitors 0.561*** 0.067 0.112*** 0.017 Widows -0.150*** 0.050 -0.039*** 0.013 Divorced 0.667*** 0.031 0.119*** 0.007 Urban residence 0.141*** 0.021 0.038*** 0.006 Regionsd Andean 0.294*** 0.029 0.052*** 0.008 Capital 0.154*** 0.026 0.057*** 0.007 Central 0.210*** 0.029 0.024*** 0.007 Central-Western 0.262*** 0.027 0.062*** 0.007 Insular 0.094 0.063 0.047*** 0.017 Plains 0.057 0.046 0.0008 0.012 Zulian -0.005 0.027 -0.049*** 0.007 Head of household 0.966*** 0.028 0.121*** 0.006 Socio-Economic Statuse High 0.053*** 0.021 0.005 0.006 Medium High 0.007 0.022 0.008 0.006 Average 0.009 0.024 0.005 0.007 Nonlabor income (US$/ month) -0.113*** 0.038 -0.023*** 0.010 Interaction terms Nonlabor income 1997-1 -0.002 0.076 -0.003 0.020 Nonlabor income 1997-2 0.010 0.073 0.014 0.019 Nonlabor income 1998-1 0.103 0.075 0.041** 0.019 Nonlabor income 1998-2 0.137* 0.073 0.046*** 0.019 Survey datef 1997-1st half 0.059*** 0.020 0.005 0.005 1997-2nd half 0.195*** 0.023 0.033*** 0.006 1998-1st half 0.257*** 0.023 0.032*** 0.006 1998-2nd half 0.281*** 0.023 0.020*** 0.006 Constant -1.995*** 0.047 N 86,199 % of total 39,036 45.3 -2 log likelihood 160,264.84*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is 1995-1 period.

PAGE 197

185 Table B.4 Multinomial Probit Re g ression Results Informal Sector Sample: All Women 15-60 Coefficients Standard Deviations Marginal Effects Standard Deviations Age groupsa 21 to 30 0.830*** 0.022 0.048*** 0.004 31 to 40 1.261*** 0.024 0.111*** 0.005 41 to 50 1.212*** 0.027 0.124*** 0.006 51 to 60 0.785*** 0.031 0.112*** 0.007 Educationb Primary 0.407*** 0.032 0.014** 0.002 Secondary 0.436*** 0.036 -0.046*** 0.006 Technical 0.448*** 0.053 -0.088*** 0.006 College 0.427*** 0.041 -0.074*** 0.006 Marital statusc Married 0.321*** 0.019 0.056*** 0.004 Cohabitors 0.450*** 0.077 0.022 0.014 Widows -0.057 0.053 0.005 0.009 Divorced 0.622*** 0.034 0.046*** 0.006 Urban residence 0.041* 0.025 -0.008 0.005 Regionsd Andean 0.307*** 0.034 0.029*** 0.007 Capital -0.081*** 0.030 -0.034*** 0.005 Central 0.320*** 0.033 0.043*** 0.007 Central-Western 0.152*** 0.039 0.0002 0.006 Insular -0.171** 0.078 -0.043*** 0.012 Plains 0.137*** 0.052 0.023** 0.010 Zulian 0.377*** 0.030 0.085*** 0.006 Head of household 1.154*** 0.030 0.123*** 0.005 Socio-Economic Statuse High 0.097*** 0.024 0.014*** 0.004 Medium High -0.051** 0.025 -0.011*** 0.005 Average -0.021 0.028 -0.005 0.005 Nonlabor income (US$/ month) -0.091** 0.043 -0.006 0.008 Interaction terms Nonlabor income 1997-1 0.021 0.085 0.005 0.015 Nonlabor income 1997-2 -0.093 0.083 -0.020 0.014 Nonlabor income 1998-1 -0.084 0.085 -0.029** 0.015 Nonlabor income 1998-2 -0.032 0.081 -0.023* 0.014 Survey datef 1997-1st half 0.109*** 0.024 0.016*** 0.004 1997-2nd half 0.216*** 0.027 0.022*** 0.005 1998-1st half 0.377*** 0.026 0.050*** 0.005 1998-2nd half 0.514*** 0.026 0.078*** 0.005 Constant -2.482*** 0.053 N 86,199 % 100.0 16,303 18.9 -2 log likelihood 160,264.84*** *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is 1995-1.

PAGE 198

186 Table B.5 Women’s Labor Force Participation by Geographical Areas of Venezuela FORMAL INFORMAL REGION N Out of the Labor Force (%) Public Sector (%) Private Sector (%) Total (%) SelfEmployed (%) Other (%) Total (%) Andean 9,905 33.92 16.03 30.75 46.78 16.36 2.94 19.30 Capital 20,601 35.06 15.10 35.33 50.43 13.78 0.74 14.52 Central 10,988 35.26 11.20 34.06 45.26 18.47 2.00 20.47 CentralWestern 15,293 35.48 13.48 33.79 47.27 15.75 1.97 17.72 Insular 1,063 41.10 15.20 31.10 46.30 10.90 1.70 12.60 NorthEastern 8,045 41.55 14.72 26.91 41.63 14.74 2.08 16.82 Plains 2,332 39.37 17.24 24.10 41.34 17.15 2.14 19.29 Zulian 17,342 35.50 9.30 29. 30 38.60 23.80 2.10 25.90 Source: Household Sample Survey (19951998) and the author’s calculations.

PAGE 199

187 Table B.6 Venezuelan Women’s Labor Force Participation by Samplesa Samples N Out of Labor Force (%) Labor Force (%) Formal Sector Informal Sector All Women 86,199 35.8 64.2 45.3 18.9 Married 42,791 33.1 66.9 44.8 22.1 Single 32,906 45.4 54.6 42.2 12.4 Head of Household 11,365a 12.8 87.2 55.7 31.5 a Includes women of all categories of marital status. Source: Household Survey Samples (19951998) and the author’s calculations.

PAGE 200

188 Table B.7 Multinomial Lo g it Re g ression Results Marginal Effects-Formal Sector ( standard deviations in parentheses ) Variable All women 15-60 Married Women Single Women Women Heads of Household Age groupsa 21 to 30 0.192***(0.005) 0.052***(0.008) 0.326*** (0.007) 0.057 (0.057) 31 to 40 0.182***(0.006) 0.048***(0.008) 0.318*** (0.009) 0.086 (0.055) 41 to 50 0.133***(0.007) 0.001 (0.009) 0.289*** (0.012) 0.081 (0.054) 51 to 60 0.036***(0.007) -0.100***(0.011) 0.211*** (0.016) 0.021 (0.056) Educationb Primary 0.150***(0.009) 0.098***(0.012) 0.325*** (0.016) 0.140***(0.016) Secondary 0.278*** (0.008) 0.270***(0.012) 0.381*** (0.015) 0.263***(0.015) Technical 0.335*** (0.008) 0.363***(0.013) 0.389*** (0.013) 0.262***(0.023) College 0.315*** (0.008) 0.388***(0.010) 0.319*** (0.016) 0.296***(0.015) Marital statusc Married -0.012*** (0.005) -0.075***(0.016) Cohabitors 0.111*** (0.017) 0.009 (0.027) Widows -0.047*** (0.013) -0.102***(0.016) Divorced 0.122***(0.007) -0.005 (0.012) Urban residence 0.039*** (0.006) 0.039***(0.008) 0.040*** (0.010) 0.060***(0.016) Regionsd Andean 0.053***(0.008) 0.053***(0.011) 0.096***(0.014) 0.008 (0.021) Capital 0.058*** (0.007) 0.031*** (0.010) 0.106***(0.012) 0.070***(0.018) Central 0.024*** (0.008) 0.006 (0.011) 0.073***(0.014) 0.019 (0.021) Central-Western 0.064*** (0.007) 0.066*** (0.010) 0.082***(0.012) 0.051***(0.019) Insular 0.050***(0.017) 0.082 (0.023)0.036 (0.034) 0.066 (0.050) Plains 0.003 (0.013) -0.014 (0.017) 0.083***(0.022) -0.040 (0.032) Zulian -0.047***(0.007) -0.086***(0.010) 0.054***(0.013) -0.123*** (0.020) Head of household 0.126*** (0.007) 0.114***(0.013) 0.179*** (0.012) Socio-economic Statuse High 0.005 (0.006) -0.003 (0.008)0.008 (0.010) -0.024 (0.016) Medium High 0.008 (0.006) 0.004 (0.008)0.006 (0.011) 0.0002 (0.016) Average 0.005 (0.007) 0.005 (0.020)0.004 (0.012) -0.025 (0.018) Nonlabor income (US$/ month) -0.028*** (0.010) -0.005 (0.020) -0.051*** (0.018) -0.022 (0.015) Interaction terms Nonlabor income 1997-1 0.002 (0.020) 0.148*** (0.044) -0.135*** (0.042) -0.052* (0.031) Nonlabor income 1997-2 0.022 (0.020) 0.127***(0.041)0.015 (0.038) -0.002 (0.028) Nonlabor income 1998-1 0.053*** (0.020) 0.139***(0.039) 0.065 (0.040) 0.038 (0.030) Nonlabor income 1998-2 0.058*** (0.020) 0.143***(0.037)0.052 (0.041) 0.002 (0.027) Survey datef 1997-1st half 0.004 (0.006) -0.001 (0.008) 0.017* (0.009) -0.009 (0.017) 1997-2nd half 0.033*** (0.006) 0.030*** (0.009) 0.035*** (0.011) 0.016 (0.018) 1998-1st half 0.031*** (0.006) 0.035***(0.009) 0.026*** (0.010) -0.014 (0.018) 1998-2nd half 0.019***(0.006) 0.026 (0.009) 0.023** (0.010) -0.037** (0.018) N 39,036 19,161 13,882 6,330 *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is 1995-1 period.

PAGE 201

189 Table B.8 Multinomial Logit Re g ression Results Marginal Effects-Informal Sector (standard deviations in parentheses) Variable All women 15-60 Married Women Single Women Women Heads of Household Age groupsa 21 to 30 0.047*** (0.004) 0.021***(0.007) 0.065*** (0.005) 0.033 (0.058) 31 to 40 0.110*** (0.005) 0.096*** (0.007) 0.110*** (0.008) 0.065 (0.055) 41 to 50 0.122*** (0.006) 0.107***(0.009) 0.122*** (0.011) 0.082 (0.054) 51 to 60 0.113*** (0.008) 0.095***(0.011) 0.113*** (0.014) 0.073 (0.056) Educationb Primary 0.011* (0.004) 0.012 (0.009) 0.040*** (0.008) -0.064*** (0.015) Secondary -0.053*** (0.006) -0.055***(0.009) -0.019*** (0.008) -0165*** (0.014) Technical -0.093*** (0.006) -0.111***(0.010) -0.053*** (0.008) -0172*** (0.022) College -0.081*** (0.005) -0.101***(0.008) -0.039*** (0.008) -0201*** (0.014) Marital statusc Married 0.057*** (0.004) -0.011 (0.015) Cohabitors 0.025* (0.014) -0.008 (0.025) Widows 0.005 (0.009) 0.002 (0.015) Divorced 0.047*** (0.006) 0.008 (0.011) Urban residence -0.008 (0.005) 0.0007 (0.007) -0.012* (0.007) -0.058*** (0.016) Regionsd Andean 0.029***(0.007) 0.031***(0.010) 0.018** (0.009) 0.005 (0.020) Capital -0.034*** (0.005) -0.040***(0.008) -0.022***(0.007) -0.077*** (0.017) Central 0.043*** (0.007) 0.061***(0.019) 0.026***(0.009) -0.002 (0.020) Central-Western -0.0008 (0.006) -0.002 (0.009)0.002 (0.008) -0.041** (0.018) Insular -0.045***(0.012) -0.052***(0.018)-0.020 (0.020) -0.104*** (0.043) Plains 0.024***(0.010) 0.032** (0.015)0.003 (0.014) 0.014 (0.014) Zulian 0.084***(0.006) 0.092***(0.010) 0.069***(0.010) 0.115*** (0.020) Head of household 0.119*** (0.006) 0.068*** (0.012) 0.124*** (0.009) Socio-economic Statuse High 0.014***(0.004) 0.026***(0.007)0.003 (0.006) -0.024 (0.016) Medium High -0.011***(0.004) -0.011* (0.007) -0.011* (0.006) 0.0002 (0.016) Average -0.005 (0.005) -0.0005 (0.007)-0.008 (0.007) -0.025 (0.018) Nonlabor income (US$/ month) -0.003 (0.007) -0.024 (0.019)-0.011 (0.010) 0.005 (0.015) Interaction terms Nonlabor income 1997-1 0.006 (0.014) 0.041 (0.036)-0.013 (0.021) 0.003 (0.030) Nonlabor income 1997-2 -0.019 (0.014) -0.025 (0.037) 0.033* (0.019) -0.035 (0.028) Nonlabor income 1998-1 -0.027* (0.014) -0.023 (0.034)-0.009 (0.021) -0.082*** (0.029) Nonlabor income 1998-2 -0.020 (0.013) -0.022 (0.031)-0.007 (0.019) -0.029 (0.026) Survey datef 1997-1st half 0.017*** (0.005) 0.023*** (0.007)0.005 (0.006) 0.011 (0.016) 1997-2nd half 0.022*** (0.005) 0.033***(0.008)0.0008 (0.007) 0.002 (0.017) 1998-1st half 0.050*** (0.005) 0.056*** (0.008) 0.036*** (0.007) 0.045*** (0.017) 1998-2nd half 0.078*** (0.005) 0.084*** (0.008) 0.057*** (0.008) 0.073*** (0.018) N 16,303 9,474 4,078 3,579 *** (**,*) = coefficients significant at 1% (5%, 10%) level. a=omitted category is women 15 to 20 years old. b=omitted category is women with no education. c=omitted category is single women. d=omitted category is the North-Eastern region. e=omitted category is poverty. f=omitted category is 1995-1 period.

PAGE 202

About the Author Betilde Rincn de Munoz graduated as an Economist from the University of Zulia in Maracaibo, Venezuela in 1969. She completed her Law Degree at the same university in 1989, graduating Summa Cum Laude. Sh e has a Master’s degree in Business Administration specializing in Finance completed in 1992, and another Master’s in Economics from the University of South Flor ida’s College of Business Administration completed in 1995. She became a Full Professor in the College of Economics at the University of Zulia in 1991. She retired from her work as a Professor teaching both graduate and undergraduate-level courses in 19 99 after more than 25 years of experience. She entered the PhD program in Economics at the University of South Florida’s Department of Economics in 2000 and has served as Adjunct Professor for the Department since 2004. She has been a devot ed wife for 36 years, and is the happy mother of four grown and beautiful children who made her a grandmother of five also beautiful grandchildren. She lives in Tamp a, FL, but her heart is in Venezuela.


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200397Ka 4500
controlfield tag 001 001988943
005 20090216141344.0
007 cr mnu|||uuuu
008 090216s2007 flu s 000 0 eng
datafield ind1 8 ind2 024
subfield code a E14-SFE0001985
035
(OCoLC)307532514
040
FHM
c FHM
049
FHMM
090
HB71 (Online)
1 100
Rincon de Munoz, Betilde.
0 245
Determinants of female labor force participation in Venezuela :
b a cross-sectional analysis
h [electronic resource] /
by Betilde Rincon de Munoz.
260
[Tampa, Fla] :
University of South Florida,
2007.
500
Title from PDF of title page.
Document formatted into pages; contains 189 pages.
Includes vita.
502
Dissertation (Ph.D.)--University of South Florida, 2007.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
520
ABSTRACT: The purpose of this study is to fill the gap in research about women in Venezuela by investigating the determinants of their labor force participation between 1995 and 1998. The Central Office of Statistics and Information in Venezuela provides cross-sectional data collected semiannually about individual, demographic, socio-economic and geographical characteristics of individuals living in Venezuela during this period. This study uses binomial and multinomial logit models to test a number of hypotheses. First, the full sample of women between 15 and 60 years old is used to investigate the importance of individual, demographic, socioeconomic, and geographical characteristics in the labor force participation decision, also controlling for a time trend. The same decision is also analyzed for three subsamples: married women, single women, and women heads of household. Comparisons are made between each subsample and the full sample, and also among the different subsamples.Next, multinomial regressions using the same explanatory variables are performed to examine labor market behavior when there is a three-way choice: whether to participate in the formal sector, the informal sector or not to participate in the labor market at all. The multinomial regressions are also performed on the three subsamples as well as on the full sample. Again comparisons are made between each subsample and the full sample and also among the three subsamples. The results of these analyses show considerable differences in motivating factors among the three groups. The conclusion that must be drawn from this research is that one cannot generalize about the women's labor force participation just by studying the behavior of women in the aggregate. The relative importance of motivating factors depends strongly on the specific subsample to which a woman belongs, a fact unrevealed by previous empirical work.The more detailed analyses produced by this dissertation provide deeper understanding of the labor force participation of Venezuelan women. This information will make a valuable contribution to policy-makers who seek to encourage the important economic contribution of women to this previously under-studied labor market.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Advisor: Carole A. Green, Ph. D.
653
Married women
Single women
And women heads of household
Micro data
Informal sector
690
Dissertations, Academic
z USF
x Economics
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.1985