THE GENDER GAP AND GROWTH: Measures, Models and the Unexplained

THE GENDER GAP AND GROWTH: Measures, Models and the Unexplained Written by: Erica Siegel University of Copenhagen Department of Economics Productivity Growth: Theory and Empirics Fall 2005

INTRODUCTION Reducing gender inequality in developing countries is one of the few policy initiatives that are almost unanimously seen as win-win objectives because of the positive repercussions at both the intrinsic and functional level. From the sociological perspective gender equality promotes basic human rights. While from an economic perspective increasing female Human Capital, specifically through education, serves as a potential force to decrease population growth, increase productivity, bring forth positive externalities and in turn increase economic growth. Some original findings were counter-intuitive and did not support this idea, but later analysis contradicted the initial results, asserting they were not robust for various reasons and more recent theoretical models and empirical studies have often confirmed that economic gains from increasing female education exist. Now NGOs, Multilateral assistance organizations, the United Nations and other bodies have made gender equality in education a part of their missions. Most notably, the United Nations has defined one of their eight Millennium Development Goals specifically to Promote Gender Equality and Empower Women with their target being to Eliminate gender disparity in primary, secondary and tertiary education, preferably by 2005, and in all levels of education no later than 2015. The statistics however show that it is not likely that this target will be met. Some countries have actually witnessed a worsening of the situation. In making economic Development decisions and implementing new policies it seems that so often, change is not brought about or is brought about slowly due to conflicting social and economic interests, so the question that comes to mind looking at the current status of the gender inequality is: why haven t greater strides been made in the countries where the returns would be greatest? Are there underlying economic mechanisms that are not being addressed? What common characteristics create the greatest persistence of this inequality? Is addressing gender equality in education getting to the root of the problem if not may policy initiatives in this respect fail to lead to growth? The purpose of this paper will be to identify what some of the deeper-seeded gender inequities may be that work into the feedback loop and seem to have created a low-level education trap for women in developing countries. 2

REVIEW OF SOME LITERATURE: The often-cited empirical findings of Barro and Lee (1994) did not shed a positive light on the link between gender equality in human capital accumulation and growth. Their regressions showed a negative relationship between female level of schooling and growth. They suggest that this negative relationship had to due with the level of backwardness of a society with larger gender gaps. Critics of their work have suggested that it was instead a result of the aggregation of data to include the fast growing East Asian countries and could be resolved through a regional dummy. The results may stem from omitted variables or a simultaneity bias. Suggestions made to eliminate their problems include use of panel data and a General Methods of Moments estimator. In the forefront of the theoretical work on gender gaps, focusing on the gender gap with respect to relative wages, Galor and Weil formulated a three-period overlapping-generations model with two kinds of people, men and women, one of each making up the unit of analysis: a couple. The model assumes men and women have equal brains and men have more brawn, with mental ability more complementary to physical capital than physical ability. The point of departure is a production function in per-couple terms taking the following form: y t = a [ αk ρ t + ( 1 - α ) m ρ t ] 1/ρ + b, at time t k= physical capital m= per-couple input of mental capital where a,b>0, α (0,1) and ρ (-,1) The couple makes labor force participation and fertility decisions to maximize their intertemporal utility function. The utility from the number of children the couple decides to have takes the following form: u t =γ ln (n t ) + (1-γ) ln (c t+1 ) with n t being the number of pairs of children per couple. They find the two key effects in [the] model are the positive effect of capital accumulation on women s relative wages and the negative effect of women s relative wages on fertility (Galor and Weil). They suggest that changes of the legal and social type for women are in fact consequences of economic growth rather than tools o promote it. They admit 3

the complexity of fertility patterns but once again suggest that the causal link runs from higher relative wages for women to lower fertility rather than the reverse. In a region-specific empirical analysis, Kingdon uses household data in an urban area of India to determine the significance of unexplainable differences in the gender gap in education. The Blinder-Oaxaca decomposition is used for the data on difference in enrolment and mean years of education. The findings show statistically significant differences in both enrolment and mean years of education. The findings show that discrimination as the unexplained portion is defined in the Blinder-Oaxaca Decomposition, must exist and accounts for a significant portion of the gap. This is just one example of the various country and region-specific articles of this form. MEASURES AND CURRENT STATE OF GENDER GAPS: The problem of a gender gap in education is most prevalent in Southern Asia, sub-saharan Africa and Western Asia for both primary and secondary education for females. The most recent report of the Millennium Development Goals indicates that these regions have made little or no progress over the period 1998-2002 and in general of those countries where complete data is available about half have achieved gender parity in primary education, about 20 percent in secondary and only 8 percent in higher education. Those regions are thus not predicted to reach the 2005 or 2015 target. In order to assess what factors may not have been accounted for it is crucial to answer the question: how is progress for improvement in the gender gap measured? The United Nations has developed several indicators for gender inequality. The GDI, Gender- Related Development Index is one of these tools and is similar to the standard Human Development Index but it also adjusts for the disparity between men and women for life expectancy, literacy and gross enrolment and income as follows: 4

Source: United Nations Development Project Report, 2005 Three steps to calculating the GDI: 1. The Dimension index uses male and female indices in the following formula: Dimension Index= (actual value- minimum value)/(maximum value-minimum value) 2. The Equally distributed index penalizes for inequality: EDI={[female population share (female index -1 )]+[male population share (male index -1 )]} -1 3. The GDI as unweighted average of component indices: GDI=1/3(life expectancy EDI)+1/3(education EDI)+1/3(income EDI) Another tool used is the Gender Empowerment Measure focusing on opportunities for women in 3 main areas as shown below: Source: United Nations Development Project Report, 2005 5

The Gender Empowerment Measure is calculated using the following determinants: 1. Percentage Shares for women s and men s parliamentary seats 2. Percentage shares of positions as legislators, senior officials and managers and percentage shares of professional and technical positions. 3. Women s and Men s earned income (PPP US$) Using these 3 indicators and Equally Distributed Equivalent Percentage is calculated for each: EDEP= {[female population share (female index -1 )]+[male population share (male index -1 )]} -1 For determinant 1 and 2 above, the EDEP is divided by 50, as equality would be indicated by 50% shares of these components. The GEM is then the simple average of the three indices. Recent data for the GDI and GEM are included in the appendices for a selection of countries. The tables include an additional measure that compares these indicators to each country s Human Development Index. The indicator used to specifically measure the gender gap in education is the Gender Parity Index. In general it follows the form: GPI= Indicator for Girls Indicator for Boys, Where GPI<.97 indicates differences favor boys and and.97<gpi<1.03 indicates parity achieved GPI>1.03 indicates differences favor girls The United Nations uses the GPI to compare enrollment rates. It should be noted that this simplification may however lead to an overstatement in the disparity if there is a significant amount of repetition for either girls or boys, embedded in the data. Table 1 shows GPI by region for recent time periods to evaluate the progress towards parity. 6

Enrollment rates are favored in many empirical analyses due to the time period issues encountered with literacy rates. More specifically, using female literacy rates may overstate the gender disparity as older women likely remain unaffected by policy changes and will skew the overall data. Comparing the data across time periods in Table 1 does not provide a general pattern across regions. Some indices have remained stable, some increased and some decreased. When comparing the indices for the most recent time period Northern and Sub-Saharan Africa, Southern and Western Asia and Oceania appear to be in the worst position. The disparity for region defined as Oceania in fact worsened in the three years prior to the 2001/2002 period, while Northern and Sub-Saharan Africa and Southern and Western Asia witnessed slight improvements. Criticism of the Measures: These indicators were developed in 1995 by the UN to supplement the widely recognized measure for human development, the HDI. Ensuring that these indicators are useful is of obvious importance as they could be the driving forces behind policy analysis and evaluation of progress. In a review of the indices Bardhan and Klasen credit the indices with contributing to the empowerment of women but point out the shortcomings of the design of the measures and provide suggested revisions. 7

In addition to the obvious problems related to data reliability, they point out conceptual problems of the GDI as follows: The life-expectancy and population sex ratio should not be used in combination to measure the gap because one has a historical component while the other does not The earned-income component accounts for most of the gender penalty and are not reliable for gender parity at a household level The assumption that objective is to reach a 1:1 ratio in male to female income is not valid for all societies It does not take into account activities of the unremunerated sort Specific calculation problems arise from the use of nonagricultural wages used as a proxy for wages in the full economy The three components of the GDI should not necessarily be given equal weight They point out two main weaknesses of the GEM: the weakness of parliaments in some countries and the focus of women s participation at a national or formal level rather than attempting to evaluate at a more local level. VARIABLES PART OF THE CAUSE AND EFFECT OF GENDER GAPS: After surveying the articles on gender gaps, it becomes evident that there are a wide spectrum of variables considered endogenous and exogenous to the process of human capital accumulation and growth. It would be nearly impossible to address every variable, instead focus will be directed at a few most often cited. As is the case with much of the empirical research in this area, it is difficult to address these issues with completeness due to a lack of sufficient data. FERTILITY: The pattern called the Demographic Transition is characterized by an intermediate stage where a country will observe a rapid acceleration in population growth from technological advances that bring about a fall in the death rate without a commensurate decline in the birth rate. This leads to problems such as a reduction in savings, a dilution of capital per person and a reduction in the 8

marginal productivity of labor. This stage in the Demographic Transition is currently being witnessed as a constraint on growth in developing countries. Reduction of fertility rates to slow population growth in developing countries is one of the growth objectives of increased female education in developing countries. The following reasons given by Thirwall provide a good general summary as to why women s education lowers fertility: Education Improves work opportunities for women, which makes it more costly in time to have children Educated women want their own children to be educated which raises the cost of having children Education and literacy make women more receptive to information about contraception Education and employment delay marriage and the time available to rear children Education improves the status, bargaining power and independence of women, encouraging and enabling them to make their own choices 1 The theory related to fertility was outlined in the work of Gary Becker when he introduced models for topics in household economics such as division of labor in households and families, the demand for children and intergenerational mobility. These models have been the basis for many subsequent studies that combined family decision-making models with research into effects on growth. These models are especially important in the growth patterns of nations because of the effects of a decreased fertility rate such as the following: increases in capital per worker, lowering of the dependency burden and thus increasing savings rates, temporarily increase share of population in the labor market and thus increasing demand for capital investment and social overhead and potential labor force absorption through increased employment leading to per capita economic growth. The literature was extended to life-cycle variations in consumption, earnings and utility with fertility as a function of expenditures on the subsistence and human capital of children (Becker 1993). The model is first outlined as a two-period analysis with parents making decisions based on preferences for the quantity and quality of children and Becker took it further in order to 1 Taken directly from Thirwall (2003), Development Economics, New York: Palgrave MacMillan 9

answer the question of why parents display certain preferences and how children actually affect their utility. Schematic Framework for Women s Education, Labor and Fertility Decisions: Primary Secondary Higher Woman Labor Market Child 1 Child n The factors that affect the women s family s or own decision to attain education include physical distance to school, facilities such as separate lavoratories, opportunity costs with respect to household participation foregone, and direct costs of schooling needed for fees and uniforms. It is most likely that this decision is entirely in the hands of the mother or father when the female is young while it may be the women s own or her husbands when made later in life. Similarly, some of the factors that affect the women s decision to enter the labor force include distance to workplace, opportunity costs with respect to household participation foregone, wages (both real and relative to husband s if applicable), social norms, education attained and health-related issues. Finally the decision of how many children may have to do with some or all of factors such as cost of raising a child, real income of parents, financial and non-financial benefits, 10

contraceptive education and availability for parents, a preference for quantity versus quality, social norms, etc. A demographic tool such as the representation above of educational, labor force participation and fertility decisions is used to represent the flows of an individual through their life and can be extended to add migration and absorbing dimensions for analysis of exit and death. Using a multi-state stochastic modeling tool, demographers and population economists identify the probability of entry and re-entry into various states. These processes, through the studies of the demographic transition, are what much of the fertility/gender gap and growth literature to date has relied on, but what is relevant to this paper is the unidentified heterogeneity. MARRIAGE: Most of the literature regarding economics and marriage seems to come from the micro-level in the literature of household decision-making. Becker s work on this subject includes a comparison of polygamy and monogamy, assortative mating in marriage markets, dowries and bride prices and the decision to divorce. Taking this further, the paper by Edlund and Lagerlof extended earlier work to compare consequences on economic growth of marriage for love versus marriage by arrangement. They attempt to show that love marriage has redistribution effects from young to old and man to woman. This is relevant because gender-profiling studies in Africa have shown that arranged marriage is still a common practice. Marriage rates for 15-19 year-old girls remain high and polygamy, forced early marriage and discriminatory laws which prevent women from inheriting land (UN, 1999) are still prevalent. Women who marry at young ages tend to be less educated and have higher fertility rates, arguably serving as both a cause and effect in the link to economic growth. Edlund and Lagerlof suggest that love-marriage versus arranged marriages may promote economic growth and that monogamy works favorably towards human capital and other investments in daughters. The model is an overlapping generations model with two periods, young and old, and two types, male and female. The groom pays the bride price to either the bride or her father. The differentiation between love-marriage and arranged marriage in the 11

model lies in the identification of two factors: the person receiving the bride-price and the person with agency. Definitions: Love-Marriage- The bride-price goes the bride and the groom and bride decide whether to marry. Arranged-Marriage- The bride-price is paid to the bride s father and parents decide the marriage. The paper finds that love marriage creates long-run benefits for growth by redistributing resources from older men to younger, promoting savings and has ambiguous results in the shortrun because it benefits younger men and does not benefit older. There is also a redistribution of resources from father of bride to bride and this is seen as a mechanism to improve human capital of children. Besides the direct implications of this model, there are indirect consequences on the forces working to promote gender equality as the practice of arranged marriage is linked to young marriage and young marriage is linked to increase in fertility. URBAN-RURAL DIMENSION: Urban-Rural issues are part of both the causes and consequences of issues of educating women. The Todaro Model of Urban-Rural Migration indicates that educational expansions may actually need to be curtailed as they could lead to over-education that will in turn increase urban unemployment. In urban areas, it has been observed however that employers are expecting employees to have higher levels of education, the standard being raised from primary to secondary for positions such as sweepers, messengers and filing clerks and from secondary to higher for positions such as clerks, typists and bookkeepers. An argument is made that the private demand for education will not be congruent with the public demand and will lead to the educated unemployed. Policies to prevent over-education could have both the obvious consequences on expansion of education for women as well as the indirect effect of widening the gap between the skilled and less-skilled and hindering income mobility. A notable point in the theoretical framework of the division of labor in households is the general absence of an inclusion of the ability to outsource household activities. It is clear that in industrialized countries and in fact in many urban areas of developing countries, household 12

chores are not done by a member of a family unit. In some of the regions of the world where growth is highest and women are most predominant in the labor force, childcare and hired help in the home are commonplace. Where women s potential wages in the labor force exceed the cost of hired help and day care this could prove a more economically efficient model. MIGRATION: Migration can have both positive and negative direct consequences on the economy and human capital investments in countries of origin. Brain drain is most predominantly an issue for those educated at a tertiary level because that is the segment of the population that stands to gain the highest returns. It is widely recognized that there are also potential benefits arising from the actual awareness of individuals that there is an opportunity to emigrate. It can make education a more desirable commodity by offering higher rates of return on the investment. In addition, remittances to families in country of origin and foreign direct investment by diasporas can play an important role in poverty reduction and infrastructural development. It is not evident whether a nation or region stands to gain more or less from open migration for females but clearly it is an area that lends itself to further research as the potential will likely become a reality with the implementation of various initiatives to provide women equal rights. Another important point is the issue of a tight labor market. Labor markets are theoretically more efficient when there are relatively many employment opportunities and a low supply of labor relative to demand and/or available opportunities to migrate to other labor markets. Tight Markets result in higher returns and adequate consumption raising working capacity. On a more technical note, the problems with including migration openness for women arise from the dearth of data. Even when data is available it suffers from problems of disaggregation and sample-selection bias. 13

CONCLUSIONS: Clearly, there is a widespread belief that gender gaps related to a wide range of issues hinder development at both an intrinsic and functional level. A comparison of the measures used to analyze the extent of this gap reveals inconsistencies. Empirical research has resulted in conclusions that vary from revealing negative effects, to no effects to positive effects on growth of reduction in gender gaps in human capital. A perhaps more disturbing finding is that the measures being used to evaluate progress and consider policy objectives do not seem to correlate with much of the literature and may be ineffective in their own right. It does seem that enough progress is not being made. Further research in this area is certainly warranted and would benefit from more data and a comprehensive approach to measurement and policy. 14

APPENDIX: 15

16

17

18

References: Bardhan, Kalpana; Klasen, Stephan. UNDP s Gender-Related Indices: A Critical Review. World Development, 1999. Barro, Robert J. Economic Growth in a Cross-Section of Countries. Quarterly Journal of Economics, May 1991. Becker, Gary S. A Treatise on the Family, Cambridge: Harvard University Press, 1993. Cahuc, Pierre; Zylberberg, Andre. Labor Economics. Caombridge: MIT Press, 2004. Edlund, Lena; Lagerlof, Nils-Petter. Implications of Marriage Institutions for Redistributions and Growth. Population and Economic Growth Conference, April 2002. Galor, Oded; Weil, David N. The Gender Gap, Fertility, and Growth. The American Economic Review, June 1996. Kingdon, G.G. The gender gap in educational attainment in India: How much can be explained? The Journal of Development Studies, December 2002. Klasen, Stephan. Low Schooling for Girls, Slower Growth for All? Cross-Country Evidence on the Effect of Gender Inequality in Education on Economic Development. The World Bank Economic Review, 2002. Ray, Debraj. Development Economics, Princeton: Princeton University Press, 1998 Thirwall, A.P. Growth and Development. New York: Palgrave MacMillan, 2003. 19