بازده خصوصی در مقابل بازده اجتماعی برای سرمایه انسانی: آموزش و رشد اقتصادی در هند

This paper investigates whether differences between private and social returns to education of government sector employees can contribute to an explanation of the “micro–macro paradox” in the literature on education and growth. We hypothesize that in India educated people find privately rewarding jobs in a sector in which social returns are low, namely the government sector. This could help explain high returns to education at the micro level and small or negative coefficients on education growth in growth regressions at the macro level. The empirical results, which are consistent with this hypothesis, are based on an analysis of state-level data from India spanning 40 years.

A common prior of economists and policy makers is that education is an important promoter of economic growth. Indeed, it has consistently been shown that the level of education has a positive association with subsequent economic growth (e.g. Barro, 1991 and Benhabib and Spiegel, 1994). But there remains a considerable debate about the effect of education growth on economic growth (e.g. Benhabib and Spiegel, 1994, Krueger and Lindahl, 2001, Pritchett, 2001, Temple, 2001 and Pritchett, 2006). These findings generate a “micro–macro paradox” (Pritchett, 2001) in the empirical literature: studies at the micro level find throughout that more education is economically beneficial for an average individual (e.g. Psacharopoulos, 1994) and educated individuals also receive, on average, other private benefits, for example with respect to health, while at the macro level this is less clear, with some studies even finding negative effects of education growth on economic growth. A potential explanation for the micro–macro puzzle is that private and social returns to education differ. This might in particular be due to the role that the government sector plays as an important employer of educated people in many countries (e.g. Murphy et al., 1991 and Pritchett, 2001). Our goal in this paper is to contribute to this literature in several ways: First, we provide an in-depth study of the importance of this micro–macro puzzle for a large country in which it is particularly striking, namely India. We will study the time period 1961–2001, a period with, on average, a significant expansion of education. Second, we hypothesize and test via cross-state regressions a specific mechanism that may explain a divergence of micro and macro findings: The hypothesis is that educated people find privately rewarding jobs in a sector in which social returns are low relative to the private returns, namely the government sector. Third, because we use within-country variation for our empirical work we keep many important aspects of the economic and social environment constant and thus can be more confident than with cross-country work that unobserved differences across units of observation or parameter heterogeneity are not driving the results. Fourth, by using variation across Indian states and measures of educational attainment from the regularly conducted censuses, we can provide estimates that are not afflicted by the data comparability and data quality problems that typically exist in cross-country work and that may explain some of the existing findings of small growth effects (see Krueger and Lindahl, 2001, Cohen and Soto, 2007 and de la Fuente and Doménech, 2006). Social returns in the government sector may be low for at least two reasons, namely (a) through direct effects, because government sector employees work in unproductive positions, or (b) through indirect effects, because government sector employees exert negative externalities on the productivity of the private sector, e.g. through government regulations, licensing requirements, or individual rent-seeking activities. Particularly if these reasons are more likely to be relevant in countries with high educational growth and large government sectors, like India, this could explain why educational growth does not consistently show a significantly positive relationship with economic growth in cross-country regressions. Indeed, in many developing countries the government sector is large, particularly as an employer for the educated. Gelb et al. (1991) provide a theoretical explanation that is consistent with theories of rent-seeking behavior, in which governments hire in response to unemployment, which leads to surplus labor in the government sector. Jaimovich and Rud (2012) provide a model in which bureaucrats' rent-seeking behavior leads to an equilibrium with low aggregate output and low levels of entrepreneurship. At the same time there is significant anecdotal evidence for the importance of rent-seeking in these large government sectors, which implies both the direct and indirect reasons for low productivity in the public sector suggested above. For example, Murphy et al. (1991) note that in “many African countries in this century, government service, with the attendant ability to solicit bribes and dispose of tax revenue for the benefit of one's friends and family, was the principal career for the ablest people in the society” (Murphy et al., 1991, p. 505). Pritchett (2001, p. 384) cites the case of Egypt where government guarantees of employment led to a government sector that, in 1998, employed seventy percent of university graduates. Gelb et al. (1991) put together data for 14 developing countries to show that government employment not only constitutes a significant share of all non-agricultural employment in these countries but that it was also growing more rapidly than wage-employment in the private sector over the 1960s and 1970s. Regarding India, Banerjee (2006, p. 1021) states that “[…] highly qualified engineers, educated at great public expense […] cooled their heels as minor functionaries in the overfilled bureaucracies of large public companies”. In the appendix we provide additional evidence at the micro level that supports the above-mentioned anecdotal evidence and that suggests a specific mechanism that leads to our findings at the macro level: India's public sector paid substantially above private sector wages and, presumably in part because of this public-sector wage premium, attracted many of India's well-educated individuals.1 In addition, it is often argued that India's bureaucracy created a mass of regulations whose effect was to inhibit India's economic growth. In sum, the attributes of India's public sector have allowed a critical mass of workers in that sector to be the well paid, but relatively unproductive, professionals that are at the core of the hypothesis of this paper. India presents a paradigmatic case of the effect that is described in some of the empirical papers cited above: a country with substantial educational growth (despite still low absolute levels of education) yet relatively paltry economic growth, in the period before the reforms of the 1980s and early 1990s. We estimate, based on data from Barro and Lee (1993), that the worldwide average of the annual growth rate of average years of schooling (as a proxy for educational capital), was 2.6 percent between 1960 and 1985. India's growth rate of the population's average years of schooling during this period was 3.4 percent. Over the whole period, this suggests that India's educational capital increased by a factor of 2.3; whereas, worldwide educational capital increased by a factor of 1.9. At the same time, India's annual average growth rate of per capita real GDP over this period was just over 1.8 percent, compared to a world average of 2.3 percent. Thus, India was an underperformer in economic growth despite its relatively impressive educational capital growth. The paper is not trying to argue against government generally. Indeed, a certain number of government employees is required to provide a sufficient institutional framework; the government also provides other important services and public goods. In a transition process, it might also be beneficial to have a significant number of government employees working in state-owned enterprises. However, we hypothesize that in the particular case studied, a share of India's government sector employment led to a wasteful diversion of human capital and that India's bureaucracy undermined the potential of India's education growth to contribute to economic growth because of its particular attributes. The suggestion that India's bureaucracy has had limited success in promoting economic growth in the time period that we study is neither unique nor particularly controversial. Especially the role of its complicated set of regulations is studied extensively in the literature (e.g. Aghion et al., 2005, Aghion et al., 2008, Besley and Burgess, 2004 and Sivadasan, 2003). Previous research has also shown that the long-run level of income in Indian states is positively affected by levels of education (Trivedi, 2006). What is new is the connection that we suggest and test between the bureaucracy and the role of education as a promoter of economic growth. Our hypothesis has the following testable implications: first, that educational expansion was effective in promoting growth in the absence of the effect of bureaucracy, and second, that the larger the bureaucracy in a given Indian state, the less effective educational expansion was in promoting growth. The findings of our cross-state regression, in which we also take into account concerns about the endogeneity of key regressors, are in line with the hypothesis: After accounting for the role of government we find that state-level economic growth is positively related to education growth in states with small governments. On the other hand, the baseline regressions, those without controlling for government sector effects, resemble the findings of the cross-country literature (e.g. Benhabib and Spiegel, 1994 and Pritchett, 2001): state-level educational expansion has no statistically significant association with state-level economic growth.2 The paper proceeds as follows. We first discuss the importance of government employment in India. We then provide some theoretical considerations to guide the econometric analysis. Next, we test the predictions of this theoretical framework with cross-state regressions using a variety of methods and robustness checks. After a consideration of possible alternative explanations we conclude.

Education frequently takes center stage in discussions about economic development. Indeed, at the micro level, the positive effects of education for human and social development, including effects on health and wages, have been widely documented. However, at the macro level, the empirical evidence on the relation between education growth and economic growth is mixed. This paper investigates this puzzle, which so far has mainly been investigated using cross-country regressions. One contribution of this paper is therefore to employ instead within-country variation to investigate how educational expansion could have failed to promote economic growth in an economy, India, with a substantial wage premium for education. Our key contribution is to suggest and test a hypothesis that may reconcile the findings at the micro and the macro levels. We hypothesize that in India part of the explanation of the micro–macro puzzle can be found in the relationship between the large public sector and the rest of India's economy: the public sector employed the majority of educated workers at high wages, but this allocation of India's human capital failed to promote substantial growth because of the limited productivity of educated workers thus employed. The theoretical framework suggested in this paper generates a testable hypothesis, and we present evidence that is consistent with this hypothesis: the data indicate that the effect of government sector size on education's effectiveness in promoting growth was negative, economically large, and statistically significant. This suggests that government absorbed human capital in low productivity activities. Once we control for this effect of government sector size on educational effectiveness, the results indicate that the effect of education in India was positive and significant for a wide range of Indian states with relatively low levels of government. At the smallest government sector size in the sample the results imply that moving from the 25th percentile of the education growth distribution to the 75th percentile implies an increase in per capita SDP growth of between 0.2 percentage points28 and 1.1 percentage points.29 This positive effect of human capital growth is not only restricted to states with the smallest government sector size: We find that 95% of state/year observations have a government sector small enough such that they would experience a positive SDP growth effect from an increase in human capital growth.30 Can we say something more positive about the contribution of education to growth in India after the significant reforms that India underwent in the 1980s and early 1990s? First, note that in light of the dramatic improvement of economic performance in the years after 1980, one might suspect educational expansion to have begun to have a more positive effect on economic growth. We do not have enough state-level data to reliably perform pre- and post 1990 comparisons that allow us to test for statistical differences between those periods. However, the micro-evidence on public–private wage differentials shows that large positive wage differentials between the public and private sectors persisted through the 1990s, with the public sector wage premium being up to 140%. Thus, on average, it seems that outside opportunities for educated workers were still unattractive relative to the opportunities provided by the state bureaucracy, but according to our findings would have had significant social returns in terms of economic growth. The results suggest that private and social returns to education are vastly different in India and suggest that the effects of the misallocation of human capital in India were significant. Despite many issues that may be unique to India's situation, the analysis may also provide a lesson for other developing nations in which the public sector constitutes a particularly large fraction of the formal economy. This may, in the end, help reconcile economists' and policy makers' priors on the central role of education in the development process with the empirical evidence.