Which Youth Get to Make Human Capital Decisions? Results from Recent Surveys
By David J. McKenzie
Decisions made about what and how much to study, when to finish school, the occupation one has, and whether and who to marry will have profound and long-term life consequences. Enhancing the ability of youth to choose wisely among the opportunities to develop human capital is thus one of the three strategic directions for policy reform proposed in the World Development Report 2007. But to what extent do youth feel that they are the ones making these important decisions? This article investigates this question with recently collected survey data.1
The data come from questions the World Development Report team added to nationally representative audience surveys conducted by Intermedia in late 2005 and early 2006 in seven countries: Albania, Bangladesh, Ethiopia, Iraq, Malaysia, Romania, and Tajikistan. In each country, individuals aged 15 and older were asked:
* Thinking of the years of schooling that you received/will receive, who would you say has had most influence on you?
* Thinking of the person you married/will marry, who would you say has had most influence on you?
* Thinking of whether you work/will work and the type of occupation you have/had, who would you say has had the most influence on you?
* For each question they were asked to choose between “myself,” parents and relatives, friends, and the government as the source of most influence. Using data on the 4,447 youth aged 15-24 in the surveys, I examine the extent to which influence over human capital decisions varies by country, and by sex, rural/urban location, household wealth, education level, and religion.
Youth decision-making varies greatly across countries
Figure 1 shows the proportion of male and female youth who say they themselves have the most influence over whom they will marry. This proportion is calculated separately by sex for urban and rural youth, and for youth in richer and poorer families.2 The majority of points are above the 45 degree line, indicating that male youth generally have greater say over marriage decisions than female youth.
A striking pattern in Figure 1 is that the degree of autonomy in decision-making about marriage varies a lot more across countries than it does between urban and rural youth, or between rich and poor youth within a country. Only 3 percent of rural female youth in Bangladesh think they have the most influence over marriage, compared to 9 percent of urban female youth in the same country. There is the same sized urban-rural gap in Romania, where 90 percent of urban female youth have the most influence, compared to 84 percent of rural female youth. However, this urban-rural difference pales in comparison to the large difference across countries.
The large differences across countries suggest that societal and cultural norms play a strong role in decision-making. Within countries, cultural differences also manifest themselves through differences in autonomy by religion. Looking at marriage decisions, we find 82 percent of young Christian women in Ethiopia feel they have the most influence, compared to 65 percent of young Muslim women; in Iraq, 73 percent of young Shia women compared to 49 percent of young Sunni women.
Figure 2 shows a similar figure for the percentage of youth who believe they have most influence in their schooling. The points here are scattered both above and below the 45 degree line, showing less pronounced gender differences in decision-making about schooling than was the case for marriage. There is also more variation within countries. Nevertheless, one still sees large variation across countries, with youth in Bangladesh once again having least say. Young Ethiopians feel they have much less control over schooling than marriage: 83 percent of urban males and females say they have most influence on marriage decisions, but only 40 percent feel this way about schooling.
Which youth have more say in decision-making within a country?
Multivariate analysis which combines country and individual characteristics confirms that the majority of variation in decision-making autonomy occurs across countries. Approximately 90 percent of the variation across individuals in the decision-making power over schooling can be explained by the country in which they live. Likewise 89 percent of the variation in marital decision-making and 77 percent of the variation in who makes decisions about work are explained by the country of residence.3
Nevertheless, policymakers who wish to increase the decision-making capacity of youth will wish to know which youth within a given country have less say in making decisions about their human capital investments. I therefore report the results of econometric analysis designed to determine which characteristics are associated with more or less decision-making power. Figure 3 illustrates the results of this work for estimation of the likelihood that a given individual says they have the most influence over work. It shows that, on average, within a country, youth generally have less say over work decisions if they are female, Muslim, come from a poor family, and live in a rural area. Education level has no significant effect on work decision-making power.
On average, female youth are 10 percentage points less likely to make work decisions for themselves than males, and Muslim youth are 6 percentage points less likely to make work decisions for themselves than non-Muslim youth. While significant, these effects are much less than the effect of being in Bangladesh or Tajikistan compared to Albania, which lowers decision-making power by 19-21 percentage points, or in Romania or Malaysia compared to Albania, which increases decision-making power by 17-27 points.
The results in Figure 3 show the general associations, but do not allow the impact of individual characteristics to vary from one country to another. Country-by-country analysis reveals the following differences in decision-making power:
* Female youth generally have less influence than males in decisions over marriage and work, but equal say over schooling. In no country do female youth have significantly less say over schooling. Female youth in Ethiopia and Romania also have the same high level of autonomy as male youth in marriage and work decisions.
* Youth in rural areas generally have less say in decisions than urban youth. The one exception is Tajikistan, where rural youth have more influence over schooling decisions than urban youth.
* Less-educated youth often have slightly less say in decisions. Youth with relatively low levels of education, such as those who have less than six or nine years of schooling, typically have three to six percentage points less influence in decision-making than youth with more education. The largest effects are seen in Albania and Tajikistan, where less-educated youth have 12 to 13 percentage points less influence over work decisions.5
* Household wealth plays a relatively minor role within countries. Youth from poor and rich households have quite similar degrees of autonomy in decision-making. Youth in poor households on average have three to four percentage points less influence in decision-making about school and work than richer households, and no less influence in decision-making about marriage.
Implications for policy
The survey evidence presented here shows large differences in the agency of youth across countries, and less sizeable, but still significant, differences within countries. This can be useful for targeting policies designed to increase the agency of youth. The results suggest that such policies should be directed to youth in poor and Muslim countries more than in middle-income countries, and at the national level, directed towards female, rural, and less-educated youth.
The results also have implications for the design of policies which aim to raise human capital investments by increasing the decision-making power of youth. An example of a policy of this sort is the Female Secondary School Assistance Project in rural Bangladesh, which provides secondary school stipends to girls on the condition that they attend school and remain unmarried until completing the secondary school certificate exam.6 The survey evidence in Figure 2 shows low influence in schooling decisions for rural youth in Bangladesh, highlighting the importance of policies which allow youth to have more influence. However, the survey also shows that male youth also have very little influence in schooling decisions, suggesting there is not a strong reason to target only females with this policy.
More generally, the survey evidence reveals greater gender differences in decision-making influence when it comes to marriage and work decisions than for schooling decisions. Yet, outside of microfinance7, there are few examples of policies designed to give women more influence over their occupational choices—which would also be likely to give them more influence over marriage, and more autonomy in general.8 Designing imaginative policies which enhance the ability of female youth to make decisions about work therefore seems an attractive area for policy experimentation.
David J. McKenzie is Senior Economist, Development Research Group, The World Bank.
Thanks to Manny Jimenez and Mattias Lundberg for helpful comments on an earlier draft.
1 The existing literature on the agency of young people as decision-makers is largely qualitative, but does suggest that there can be large variation across cultures, economic status, etc. However, in the absence of quantitative data, it is difficult to assess how much variation there is, and along which dimensions decision-making agency varies most. Box 2.5 and pages 54-55 of Chapter 2 of the World Development Report 2007 summarize the existing literature.
2 The classification into rich and poor is done on the basis of household income, or subjective standard of living, and is done so approximately the top third of households in a country are classified as rich, and the bottom third as poor. It is thus a relative measure of household wealth.
3 These numbers come from pooling the seven surveys and using them to compare the R-squared of an OLS model for who makes decisions as a function of individual characteristics (sex, rural/urban location, education level, household wealth, and religion) to an OLS model with individual characteristics and country dummy variables. Similar magnitudes are present comparing pseudo R-squareds from probit models.
4 The coefficients shown are from a pooled probit with country fixed effects, and are shown relative to an urban non-Muslim Albanian male youth with a middle level of education and middle-level household wealth (the omitted categories when constructing dummy variables).
5 These results from Albania and Tajikistan are from country-level probits, and contrast with the insignificant average effect of low education on work from the pooled sample seen in Figure 3. Pooling countries averages over countries where low educated individuals have higher, but statistically insignificant, likelihoods of making work decisions (as in Bangladesh), countries where there is a small, insignificant negative effect (Ethiopia, Malaysia, Romania) and countries with a significant and negative effect (Iraq, Albania, Tajikistan).
6 See Box 6.6 in the World Development Report 2007 for a more detailed description of this program.
7 And while microfinance often targets women, it is not generally targeted towards youth.
8 The correlation between decision-making influence in marriage and in work is 0.39 for female youth, compared to 0.32 for male youth.