WPS8402 Policy Research Working Paper 8402 Returns to Investment in Education A Decennial Review of the Global Literature George Psacharopoulos Harry Antony Patrinos Education Global Practice April 2018 Policy Research Working Paper 8402 Abstract Returns to investment in education based on human higher education have increased over time, raising issues capital theory have been estimated systematically since of financing and equity. Social returns to schooling remain the 1950s. In the 60-plus year history of such estimates, high, above 10 percent at the secondary and higher educa- there have been several compilations in the literature. This tion levels. Women continue to experience higher average paper reviews and highlights the latest trends and patterns rates of return to schooling, showing that girls’ education based on a database of 1,120 estimates in 139 countries. remains a priority. Returns are higher in low-income The review shows that the private average global rate of countries. Those employed in the private sector of the return to one extra year of schooling is about 9 percent economy enjoy higher returns than those in the public a year and very stable over decades. Private returns to sector, lending support to the productive value of education. This paper is a product of the Education Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at hpatrinos@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Returns to Investment in Education: A Decennial Review of the Global Literature George Psacharopoulos* Harry Antony Patrinos** JEL codes: C13, J31 Keywords: Returns to schooling; Investments in education ____________________________ * Georgetown University, gpsach@rcn.com. ** Corresponding author, The World Bank, Washington, DC 20433, USA, hpatrinos@worldbank.org We are grateful to all those who over the years have shared with us their work on the returns to education and to Sean Kelly for efficient research assistance. Introduction With roots in the writings of classical economists (see, for example, Adam Smith 1776; Marshall 1890) the link between education and earnings only recently emerged. Formal modeling did not take place until much more recently (Schultz 1960, 1961; Becker 1964; Mincer 1974; Chiswick 2003). The study of earnings by schooling has led to several empirical works testing hypotheses on a great variety of social issues. These include, for example, racial and ethnic discrimination, gender discrimination, income distribution, and the determinants of the demand for education. But the dominant application that has used earnings by level of education is the estimation of the rate of return to investment in schooling. The concept of the rate of return on investment in education is very similar to that for any other investment. It is a summary of the costs and benefits of the investment incurred at different points in time, and it is expressed in an annual (percentage) yield, like that quoted for savings accounts or government bonds. Returns on investment in education based on human capital theory have been estimated since the late 1950s. Human capital theory puts forward the concept that investments in education increase future productivity. Estimation of the returns to education has been a popular subject in the literature (Ashenfelter and Krueger 1994; Becker 1964; Becker and Chiswick 1966; Card and Krueger 1992; Card 2001; Duflo 2001; Heckman, Lochner and Todd 2006; Oreopoulos 2006; Rosenzweig 1995; Schultz 1961). Since our last review of the literature, contributions on the subject have grown exponentially, to the point of being difficult to track (for previous compilations see Harmon, Oosterbeek and Walker 2003; Psacharopoulos 1972, 1973, 1981, 1985, 1993, 1994; Psacharopoulos and Patrinos 2002, 2004a). There have been several compilations that we have undertaken, including a few encyclopedia articles (Patrinos and Psacharopoulos 2010a, 2010b, 2002; Psacharopoulos and Patrinos 2008, 2004b). There are also a few attempts to create databases of comparable estimates of the return to schooling (Hendricks 2004; Montenegro and Patrinos 2014; Peet, Fink and Fawzi 2015). The popularity of estimating returns to education stems from the resulting efficiency, equity and financing implications. The rank order of returns to a level or type of education, and a comparison with the returns of alternative investments can assist education policy makers to make informed investment decisions. Previous compilations have shown that private returns to primary education decline over time, but slightly (Psacharopoulos 1981). Previous work also shows that returns are highest for primary education, the general curricula, the education of women, and countries with the lowest per capita income (Psacharopoulos 1985). Also, primary education continues to exhibit the highest social profitability in all world regions. Social and private returns at all levels generally decline by the level of a country's per capita income. Overall, the returns to female education are higher than those to male education. The returns to the academic secondary school track are higher than the vocational track – since the unit cost of vocational education is much higher; and the returns for those who work in the private (competitive) sector of the economy are higher than in the public (noncompetitive) sector (Psacharopoulos 1994). The classic pattern of falling returns to education by level of economic development and level of 2    education is maintained. The private returns to higher education are increasing, highest returns are recorded for low-income and middle-income countries, average returns to schooling are highest in Latin America and the Caribbean (Psacharopoulos and Patrinos 2004a). In this update we follow the tradition (see, for example, Psacharopoulos and Patrinos 2004) and present the latest estimates and patterns. We review estimates from 139 countries over several years resulting in a panel of 1,120 country-year cases. Methods Returns to education in the literature have been estimated using two main methods – the full- discounting method and the Mincerian earnings function. For an explanation of these methods see Psacharopoulos (1995) and Psacharopoulos and Mattson (1998). Over the years, researchers have given preference to the Mincerian method because of its convenience (Mincer, 1974). The concept of rate of return to education The rate of return to schooling equates the value of lifetime earnings of the individual to the net present value of costs of education. For an investment to be economically justified, the rate of return should be positive, and should be higher than the alternative rate of return. For the individual, weighing costs and benefits means investing if the rate of return exceeds the private discount rate (the cost of borrowing and an allowance for risk). The costs incurred by the individual are the foregone earnings while studying, plus any schooling fees or incidental expenses incurred. The private benefits amount to how much extra an educated individual earns (after taxes) compared with an individual with less education. More and less in this case refer to adjacent levels of education – e.g., university graduates compared to secondary school leavers. The social rate of return includes the society’s spending on education – for example, money spent on renting buildings and professorial salaries. The social attribute of the estimated rate of return refers to the inclusion of the full resource cost of the investment – the direct costs by government and the foregone earnings of students as they invest in their education. Ideally, the social benefits should include non-monetary benefits of education, such as the number of lives saved because of improved sanitation conditions followed by a woman because she has received more education. Given the scant empirical evidence on the social benefits of education, the social rate of return estimates are usually based on directly observable monetary costs and benefits of education. Since the costs are higher in a social rate of return calculation relative to the one from the private point of view, social returns are typically lower than a private rate of return. The difference between the private and the social rates of return reflects the degree of public subsidization of education – since practically the only difference is the addition of social costs. 3   Estimation issues The Mincerian earnings function has been the subject of controversy in the literature (Psacharopoulos and Layard, 1979; Heckman et al. 2006). One issue with the Mincerian method of estimating returns to education is missing variables, e.g., ability bias. Griliches (1977) analyzed the issue many years ago. He found that the bias is small or negative. Adding more variables to the equation will not solve the problem and might add other biases (Patrinos 2016). The earnings premium associated with the level of education suggests that productivity increases as people acquire additional qualifications. An alternative view is that earnings increase with education due to credential effects. This refers to the idea that higher levels of schooling are associated with higher earnings, not because they directly raise productivity, but because they certify that the worker is likely to be productive. In this sense, education merely sorts workers according to their unobserved attributes; it does not necessarily augment their intrinsic productivity. For public policy reasons it is important to distinguish between the human capital (productivity) and screening hypotheses about returns to education. In very basic terms, these two hypotheses mean, respectively: schooling imparts skills that enhance productivity; hence, increases in earnings are due to the increased productivity brought about by investments in schooling (human capital); while the screening hypothesis maintains that employers select workers with higher qualifications to reduce their risk of hiring someone with a lower capacity to learn; in this case, higher earnings may not be due to productivity alone (screening). With these concepts in mind, if the only purpose of schooling is to sort prospective employees, then questions arise as to the appropriateness of public investment in the expansion or improvement of schooling. Layard and Psacharopoulos (1974) found no support for the screening hypothesis. Psacharopoulos (1979) made a distinction between the weak and the strong versions of the screening hypothesis. It is true that, at the hiring point, employers do not have enough information on the prospective productivity of a job applicant, so they might offer a premium to those who have a higher level of schooling relative to the rest. This is the weak version of screening. But after years on the job, such premium should diminish if the worker did not live up to the expectations. Such finding contradicts empirical data showing the earnings gap between more and less educated workers increases over time. Card (2001) provides a rigorous test of the screening hypothesis by taking advantage of natural experiments such as changes in the school leaving age or college openings. By and large, while some evidence of weak screening is revealed, education is generally associated with higher earnings due to productivity rather than to screening. A recent analysis that uses rigorous evaluation techniques to compare the earnings of workers who barely passed versus those who barely failed high school exit exams finds little evidence of diploma screening effects (Clark and Martorell, 2014). More recent analyses, which exploit data that permit the disaggregation of earnings by years of completed schooling, have questioned the linear nature of the earnings function approach. Furthermore, due to ongoing and rapid technological progress, cross-sectional data based on observations from many years in the past may produce biased estimates of the returns to schooling. Some have even questioned whether it is still possible to interpret the coefficient on schooling as 4   a rate of return (Heckman, Lochner and Todd 2006). For instance, some researchers argue that the literature following Mincer’s estimated returns to schooling using cross-sectional data assumes that younger workers base their earnings expectations on the current experiences of older workers. However, if prices (the cost of schooling, for example, tuition associated with university) change over time and workers can at least partially anticipate these changes, then estimates of the returns to different schooling levels based on cross-sectional data may not represent the ex-ante rates of return governing human capital investment decisions. Relying on past cohorts to assess current investment decisions requires several strong assumptions, such as stability of the economic environment and perfect certainty of future earnings streams. This is difficult in the current context, given that emerging evidence suggests that wage patterns have changed substantially over time, making it difficult to use cross sections to approximate lifecycle earnings. One solution could be to follow actual cohorts over their entire educational and employment lifecycles to measure their earnings patterns to estimate the returns to education. The debate is ongoing about the external validity and causality of rate of return estimates. More conclusive evidence will emerge once data on the lifetime earnings profiles of beneficiaries of voucher programs can be obtained, as many of these programs use lotteries to assign places, thus giving researchers access to randomized data. In 1981, Chile introduced nationwide school choice by providing vouchers to any student wishing to attend a “voucher school” (essentially, a private school participating in the program, whereby the funding from the voucher would be used to pay the fees and tuition). Beneficiaries of the vouchers obtained more schooling and subsequently earned more than non-voucher students. Also, it is estimated that formal sector earnings are higher for lottery winners in Colombia’s large-scale government program which used a lottery to distribute scholarships for private secondary school to socially disadvantaged students. Institutional factors have also been used to more precisely estimate the returns to schooling, including the effect of birth date. Those who are forced to remain in school because of their birth date and the school attendance law receive the same rate of return to education as those who voluntarily continue schooling. The Mincerian method gives private returns, whereas the full discounting method can give private and social returns. In interpreting the evidence presented below, the reader should bear in mind that Mincerian and full-discount returns are not directly comparable because they differ in scale. This is because, by construction, the Mincerian method tacitly assumes foregone earnings even for children aged 6 to 12 years, thus underestimating the true size of the returns (Psacharopoulos and Mattson 1998; Psacharopoulos 1995). In our exposition below we start with private returns based on the Mincerian method, then move on to the full-discounting method that includes social returns. Results Mincerian private returns The coefficient on years of schooling of the basic Mincerian function gives an overall picture of the returns to education. Based on 705 estimates, over the years 1950 to 2014, the private rate of return to an additional year of schooling is 8.8 percent (see Annex 1; given the size of the database, 5   only summary statistics are reported here, while the rest can be found online in Annex 21). This is lower by about one percentage point relative to our (Psacharopoulos and Patrinos 2004a) compilation. It should be noted that the decline over time of the returns to education is very gradual and statistically insignificant (see Figure 1). This could be interpreted as a “race between education and technology” as the price of education fails to decline proportionately in the face of rapid supply increases (Tinbergen 1975). Thus, demonstrating that the demand for skill is outpacing the growth in supply of skills. Figure 1: Rate of Return to Schooling over Time (%) 40 35 30 25 20 15 10 5 0 1950 1960 1970 1980 1990 2000 2010 Note: Regressing the overall Mincerian rate of return on the year of the estimate, gives: Return = 49.611 – 0.020Year; R2=0.003 (t = 1.4) Regressing the Mincerian overall return on the country’s years of schooling (S) in the survey year, yields a very small but statistically significant decline of the returns of about 0.2 percentage points for every additional year of schooling (see Figure 2).                                                              1  http://datatopics.worldbank.org/education/files/GlobalAchievement/ReturnsEdAnnex2.xlsx.  6   Figure 2: Return to Schooling by Years of Schooling 40 35 30 Rate of return (%) 25 20 15 10 5 0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 Years of schooling Note: Regressing the overall Mincerian rate of return on the year of the estimate, gives: Return = 10.749 - 0.246 S, R2 = 0.027 (t = 4.4) Nevertheless, there has been an increase in the returns to schooling since 2000. Parsing the sample of estimates into pre- and post-2000 estimates, the returns have increased, albeit at a slower rate relative to the increase in the years of schooling (see Table 1). The schooling revolution of the 20th century continues, with an average increase in schooling years of more than 10 percent in the first decade and a half of the 21st century, as compared with the latter half of the 20th century. The returns to schooling increased in the 21st century by 4 percent relative to the latter half of the 20th century. Table 1: Years of schooling and returns over time Mean years of Overall Mincerian Period schooling rate of return (%) Number of studies Pre 2000 7.8 8.7 511 Post 2000 8.6 9.1 194 Restricting the sample to the most recent estimate for each country, we get an average rate of return of 9.5 percent, just slightly lower than the 9.7 percent we obtained in our Psacharopoulos and Patrinos (2004a) update. Again, this could refer to Tinbergen’s (1975) race between education and technology. The higher returns since 2000 suggest that the price of education is increasing even though supply is going up. This points to the interpretation that education is no longer winning the race. Focusing on a single country is illustrative. For Colombia, we have a record of 35 return estimates over time (Psacharopoulos 1985, 1994; Patrinos 1995; Kaboski 2003; Sanchez Torres and Nunez Mendez 2003; Psacharopoulos, Arriagada and Velez 1992; Psacharopoulos and Velez 1992; Tenjo et al. 2015) since 1965. Taken together, we note a convergence to a 10 percent overall rate of return to a year of schooling. At the same time, average schooling had gone up considerably, from 7   less than 5 years in the 1960s to more than 10 years by 2013 (see Figure 3). Figure 3: Returns and years of schooling over time: Colombia 16 14 12 10 8 6 4 2 0 1974 1979 1984 1989 1994 1999 2004 2009 2014 Years of schooling Rate of return (%) Focusing on Argentina, for which we have Mincerian returns (Kaboski 2003; Psacharopoulos 1994; Kugler and Psacharopoulos 1989; Lopez Boo 2010; Fiszbein, Giovagnoli and Patrinos 2007) over a 25-year period, we notice a remarkable stability of such returns even during the country’s economic crisis (Figure 4). Note that the wide variation of the returns over time is due to different samples and methodologies. Figure 4: Returns to Schooling and Economic Growth: Argentina 15 10 5 0 -5 -10 -15 1980 1985 1990 1995 2000 2005 Returns Growth Note: Source of change in GDP per capita is World Bank World Development Indicators This finding can be interpreted using Tinbergen’s (1975) race between education and technology, in the sense that over time the education supply curve has been shifting more to the right relative 8   to the demand curve. This is most evident in the returns to higher education (see Figure 5). While enrollment in higher education has gone up three-fold since 1970, the returns have not changed overall. The increased share of the labor force with higher education should have reduced the rate of return on the investment. Yet, the rates of return over time do not fluctuate much because as the supply of educated labor increases, so does the demand for higher skills, hence not depressing the returns to education. As discussed by Goldin and Katz (2010), the race reflects, on the one hand, the skill-biasedness of technological progress with its consequences for income inequality and, on the other hand, to the pivotal role of education in mediating this relation. While higher education increased substantially, the premium on high skills continued to increase. This suggests that educational advancements were insufficient to countervail demand due to technological progress. The rising inequality implies that technology is winning this race. Figure 5: The Race between Higher Education and Technology 35 30 25 20 15 10 5 0 1970 1980 1990 2000 2010 Returns Enrollment Income level and regional differences Private returns to schooling are higher in low-income countries by about one percentage point relative to high-income countries (see Table 2). This is the case even though the mean years of schooling differs by a factor of almost two between country groups. In fact, years of schooling will peak, to about 10 years by 2050. Table 2: Private Returns to Schooling by Income Group Country income level Overall rate of return (%) Mean years of schooling Low 9.3 5.0 Middle 9.2 7.0 High 8.2 9.2 World average 8.8 8.0 Notes: Country per capita income levels based on World Bank (2016) classifications in 2015 US$: low = $1045 or less; middle = $1046-$12,735; high = $12,736 or more 9   Disaggregating further by world region, private returns to schooling are highest in Latin America and Sub-Saharan Africa and lowest in the Middle East and North Africa (see Table 3). The Middle East and North Africa is an outlier, given the relatively low average level of schooling. Kingsbury (2018) provides several hypotheses for the low returns citing such factors as corruption, natural resources and poor academic performance. While South Asia exhibits relatively low returns, within the region India has seen increasing returns since the economic liberalization program of the 1990s. Table 3: Private Returns to schooling by region Region Overall rate Mean years of return (%) of schooling Latin America and Caribbean 11.0 7.3 Sub-Saharan Africa 10.5 5.2 East Asia and Pacific 8.7 6.9 South Asia 8.1 4.9 Advanced Economies 8.0 9.5 Europe and Central Asia 7.3 9.1 Middle East and North Africa 5.7 7.5 World average 8.8 8.0 Gender As in previous reviews, the private returns to female education exceed that of males by about two percentage points (see Figure 6). The gap has increased. The female advantage was just over one percentage point in previous updates (Psacharopoulos and Patrinos 2004a; Psacharopoulos 1994). This does not imply that earnings are higher for females, but only that education is a good investment for women and girls, and a development priority. Figure 6: Private Mincerian Returns to Education by Gender 12 10 8 6 4 2 0 Males Females 10   Private sector of employment The returns for those working in the private sector of the economy are higher than for those working in the public sector. The finding lends credibility that, where productivity matters, education is recognized (Psacharopoulos 1983; Harmon, Oosterbeek and Walker 2003). There is a clear earnings advantage for workers in the private sector (see Figure 7). Figure 7: Returns to Schooling by Sector of Employment 12 10 8 6 4 2 0 Public sector Private sector Full discounting method and social returns Based on 166 estimates using the full discounting method, the returns to schooling fall by level of economic development (see Table 4). The private returns to primary education are still high, but there are fewer countries where this calculation can be made given the drive for universal primary education. The private returns to higher education are substantial and have remained high since the last update. They have increased since the last update in both low- and high-income countries, falling only in middle-income countries. The returns to secondary education have declined, but not in high-income countries. Table 4: Returns by income and educational level (%) Private Social Per capita income level Primary Secondary Higher Primary Secondary Higher Low 25.4 18.7 26.8 22.1 18.1 13.2 Middle 24.5 17.7 20.2 17.1 12.8 11.4 High 28.4 13.2 12.8 15.8 10.3 9.7 Average 25.4 15.1 15.8 17.5 11.8 10.5 Note: The “high” private return to primary education in high-income countries is due to an outlier 1959 estimate of 65% for Puerto Rico, a country classified as high-income under our current-per-capita income classification system. 11   The social returns follow the well-known pattern of falling by level of development and level of education. Social returns to education are universally lower than private returns because of the public subsidization of education. Social returns are lower than private because researchers can account for full social costs, but they do not include social benefits. The costs of higher levels of schooling are much higher, hence the lower returns (see Figure 8). Figure 8: The Pattern of Private and Social Returns 30 25 20 15 10 5 0 Primary Secondary Tertiary Private Social As a rule, returns are higher in lower-income countries relative to higher-income countries (see Figures 9 and 10). This can be attributed to the relative scarcity of human capital in the two types of countries. For example, the mean years of schooling of the labor force in Sub-Saharan Africa is 5.1, compared with 10.2 in advanced economies. Figure 9: The Structure of Private Returns 35 30 25 20 15 10 5 0 Primary Secondary Tertiary Low income Middle income High income 12   Figure 10: The Structure of Social Returns 30 25 20 15 10 5 0 Primary Secondary Tertiary Low income Middle income High income Private and social returns by region In our previous analysis (Psacharopoulos and Patrinos 2004a) we presented estimates from 83 countries and showed that the returns to education were highest in Sub-Saharan Africa, the region with the lowest levels of schooling. With near universal primary education in most regions, it is becoming difficult to estimate returns to primary schooling using recent surveys and the full discounting method. Our update – using estimates since 2000 – is dominated by such returns from high-income countries, which exclude returns to primary education (since the comparison group, that is, workers with no education, is absent). Overall, private returns to secondary and higher education in high-income countries are high, at 13 and 12 percent (see Table 5). Table 5: Private and Social Returns to Schooling by Region, post-2000 (%) Private Social Secondary Higher Secondary Higher Average 13.2 12.4 10.2 10.6 Developing countries -- -- 10.2 16.4 Number of studies 50 54 59 61 Note: For private returns, sample consists of high income countries plus Turkey; for social returns, two are from Africa, two are from Latin America, one is from Europe, one is from East Asia, and the rest are from high income countries. In terms of social returns, these are higher than any plausible social discount rate though lower than private, across all income groups. The social returns to higher education are particularly high, but these are driven by returns in Africa, where the social returns to higher education are 35 percent in Malawi (Chirwa and Matita 2009) and 22 percent in South Africa (Salisbury 2015) – this of course implies that private returns to higher education are even higher. 13   Discussion The rate of return patterns found in previous updates are upheld and reinforced. Regarding efficiency in the use of resources, spending on human capital is a good investment. For example, in the United States the long-term 1966 to 2015 average return on stocks and bonds is 2.4 percent (Damodaran 2016) versus a 10.5 percent overall private return to investment in education in our database for education. Recent research indicates that the social rates of return reported above are under-estimates of the true returns because of the omission of externalities (Munich and Psacharopoulos 2017). Taking just one externality into account, the social rate of return to investment in education could be 50% higher than the one traditionally estimated. Monetizing the value of just one externality of education - reduced mortality - Pradhan et al. (2018) found that the social rate of return to investment in one extra year of schooling in low-income countries is 16%, relative to 11% based solely on earnings differentials. In allocating an education budget among different levels of schooling, priority should be given to the lower levels of education in countries that have not yet achieved universal primary. In countries with disparities between male and female enrollment, priority should be given to the education of girls. The size of the private returns to education and difference between private and social rates calls for selective cost recovery in higher education. The overall 10 percent private return to one year of schooling has marginally declined since first estimated over half a century ago (see, for example, Becker 1964). When making education decisions, it is very tempting to use estimates as those presented above for countries that lack such information. We emphasize that there is no substitute for a country- specific study. In such case, we recommend using the full discounting method as it is more appropriate relative to the Mincerian method. Future updates will attempt to include estimates of the returns to school quality. Although there are plenty of macro studies in this respect, micro studies on the returns to investing in specific school quality inputs are emerging. 14   Text references (References to the database used in this compilation are found on-line in Annex 3.2) Ashenfelter, O. and A. Krueger. 1994. “Estimates of the economic return to schooling from a new sample of twins.” American Economic Review 84(5): 1157-1173. Becker, G.S. 1964. Human Capital: A Theoretical and Empirical Analysis with Special Reference to Education. New York: National Bureau of Economic Research. Becker, G.S. and B.R. Chiswick. 1966. “Education and the distribution of earnings.” American Economic Review 56(2): 358−369. Card, D. and A.B. Krueger. 1992. “Does School Quality Matter? Returns to Education and the Characteristics of Public Schools in the United States.” Journal of Political Economy 100(1): -40. Card, D. 2001. “Estimating the return to schooling: Progress on some persistent econometric problems.” Econometrica 69(5): 1127−1160. Chirwa, E. W., and M.M. Matita. 2009. “The Rate of Return of Education in Malawi.” University of Malawi Working Paper 2009/01. Chiswick, B.R. 2003. “Jacob Mincer, Experience and the Distribution of Earnings.” Review of Economics of the Household 1: 343-361. Clark, D. and P. Martorell. 2014. “The signaling value of a high school diploma.” Journal of Political Economy 122:2: 282-318. Damodaran, A. 2016. “Annual Returns on Stock, T.Bonds and T.Bills: 1928 – Current”. http://pages.stern.nyu.edu/~adamodar/New_Home_Page/datafile/histretSP.html Duflo, E. 2001. “Schooling and labor market consequences of school construction in Indonesia: Evidence from an unusual policy experiment.” American Economic Review 91(4): 795−813. Fiszbein, A., P.I. Giovagnoli H.A. Patrinos. 2007. “Estimating the returns to education in Argentina using quantile regression analysis: 1992-2002.” Económica 53. Goldin, C. and L.F. Katz. 2010. The Race between Education and Technology. Belknap Press. Griliches, Z., 1977. “Estimating the returns to schooling: Some econometric problems.” Econometrica 1-22. Harmon, C., H. Oosterbeek and I. Walker. “The returns to education: Microeconomics.” Journal                                                              2 http://datatopics.worldbank.org/education/files/GlobalAchievement/ReturnsEdAnnex3.docx. 15   of Economic Surveys 17(2): 115-156. Heckman, J.J., L.J. Lochner and P.E. Todd. 2006. “Earnings functions, rates of return and treatment effects: The Mincer equation and beyond,” in E. Hanushek and F. Welch, eds., Handbook of the Economics of Education (Volume 1). Amsterdam: Elsevier: 307−458. Hendricks, L. 2004. “A database of Mincerian earnings regressions.” Iowa State University. Kaboski, J.P. 2003. “Endogenous Mincerian Returns: Explaining Cross-Country Variation in the Returns to Schooling.” Ohio State University, Department of Economics. Kingsbury, I. 2018. “Making sense of low private returns in MENA: A human capital approach.” International Journal of Educational Development 61: 173-183. Kugler, B. and G. Psacharopoulos. 1989. “Earnings and education in Argentina: an analysis of the 1985 Buenos Aires Household Survey.” Economics of Education Review 8(4): 353-365. Layard, R. and G. Psacharopoulos, 1974. "The Screening Hypothesis and the Returns to Education" Journal of Political Economy 82, no. 5: 985-98. López Bóo, F. 2010. “Returns to Education and Macroeconomic Shocks: Evidence from Argentina.” IZA Discussion Paper No. 4753. Marshall, A., 1890. Principles of Economics. MacMillan. Mincer, Jacob A., 1958. “Investment in Human Capital and Personal Income Distribution”. Journal of Political Economy 66(4): 281-302. Mincer, Jacob A., (1974), Schooling, Experience, and Earnings. National Bureau of Economic Research. Montenegro, C.E. and H.A. Patrinos. 2014. “Comparable estimates of returns to schooling around the world.” World Bank Policy Research Working Paper Series 7020. Munich, Daniel and George Psacharopoulos, 2018. “The external and non-market benefits of education: A review”. EENEE, Analytical report No. 3/2017. Oreopoulos, P. 2006. “Estimating average and local average treatment effects of education when compulsory schooling laws really matter." American Economic Review 96(1): 152−175. Patrinos, H.A. 2016. “Estimating the return to schooling using the Mincer equation.” IZA World of Labor 2016. Patrinos, H.A. 1995. “Education and Earnings Differentials.” World Bank (mimeo). Patrinos, H.A. and G. Psacharopoulos. 2010a. “Returns to Education in Developing Countries,” in P. Peterson, E. Baker and B. McGaw, eds., International Encyclopedia of Education 16   (Volume 2). Oxford: Elsevier. Patrinos, H.A. and G. Psacharopoulos. 2010b. “Returns to education in developing countries,” in International Encyclopedia of Education. Harvard. Patrinos, H.A. and G. Psacharopoulos. 2002. “Economic Benefits of Education Investment, Measurement,” in J. Guthrie, ed., Encyclopedia of Education (Second Edition). Macmillan. Peet, E.D., G. Fink and W. Fawzi. 2015. “Returns to education in developing countries: Evidence from the living standards and measurement study surveys.” Economics of Education Review 49: 69-90. Pradhan, Elina; Elina M. Suzuki, Sebastián Martínez,Marco Schäferhoff, and Dean T. Jamison, 2018. “The Effects of Education Quantity and Quality on Mortality”. Chapter 30 in Optimizing Education Outcomes: High-Return Investments in School Health for Increased Participation and Learning. Donald A. P. Bundy, Nilanthi de Silva, Susan Horton, Dean T. Jamison and George C. Patton, eds. The World Bank.   Psacharopoulos, G. 1972. “Rates of Return to Investment in Education around the World.” Comparative Education Review 16(1): 281-302. Psacharopoulos, G. 1973. Returns to Education: An International Comparison. Amsterdam: Elsevier. Psacharopoulos, G. and R. Layard. 1979. “Human capital and earnings: British evidence and critique.” Review of Economic Studies 46: 485-503. Psacharopoulos, G. 1979. "On the Weak versus the Strong Version of the Screening Hypothesis." Economics Letters 4 (1979): 181-85. Psacharopoulos, G. 1981. “Returns to Education: an updated international comparison.” Comparative Education 17(3): 321-341. Psacharopoulos, G. 1983. “Education and Private Versus Public Sector Pay.” Labour and Society 8(2): 123-33. Psacharopoulos, G. 1985. “Returns to Education: A Further International Update and Implications.” Journal of Human Resources 20(4): 583-604. Psacharopoulos, G. 1993. “Returns to investment in education: a global update.” World Bank Policy Research Working Paper Series 1067. Psacharopoulos, G. 1994. “Returns to Investment in Education: A Global Update.” World Development 22(9): 1325-43. Psacharopoulos, G. 1995. “The Profitability of Investment in Education: Concepts and Methods.” 17   The World Bank, Human Capital Development and Operations Policy, Working Papers, no. 63, 1995. Psacharopoulos, G., A.M. Arriagada and E. Velez. 1992. “Earnings and education among self- employed males in Colombia.” Bulletin of Latin American Research 11(1): 69-89. Psacharopoulos, G., 1995. “The Profitability of Investment in Education: Concepts and Methods.” World Bank, Human Capital Development and Operations Policy, Working Papers No. 63. Psacharopoulos, G. and R. Mattson. 1998. “Estimating the Returns to Education: A Sensitivity Analysis of Methods and Sample Size.” Journal of Educational Development and Administration 12(3): 271-287. Psacharopoulos, G. and H.A. Patrinos. 2004a. “Returns to Investment in Education: A Further Update.” Education Economics 12(2): 111-135. Psacharopoulos, G. and H.A. Patrinos. 2004b. “Human Capital and Rates of Return,” in International Handbook on the Economics of Education (Chapter 1). Edward Elgar Publishing. Psacharopoulos, G. and R. Mattson. 1998. “Estimating the Returns to Education: A Sensitivity Analysis of Concepts and Methods” Journal of Educational Planning and Administration, 12, 3: 271-287. Psacharopoulos, G. and H.A. Patrinos. 2002. “Returns to investment in education: a further update.” World Bank Policy Research Working Paper Series 2881. Psacharopoulos, G. and H.A. Patrinos. 2008. “Education and Human Capital,” in International Handbook of Development Economics (Volumes 1 & 2, chapter 24). Edward Elgar Publishing. Psacharopoulos, G. and E. Velez. 1992. “Schooling, ability, and earnings in Colombia, 1988.” Economic Development and Cultural Change 40(3): 629-643. Rosenzweig, M. R. 1995. “Why are there returns to schooling?” American Economic Review 85(2): 153−158. Salisbury, T. 2016. “Education and Inequality in South Africa: Returns to Schooling in the Post- Apartheid Era.” International Journal of Educational Development 46: 43-52. Sanchez Torres, F. and J. Nunez Mendez. 2003. “A Dynamic Analysis of Human Capital, Female Work Force Participation, Returns to Education and Changes in Household Structure in Urban Colombia, 1976-1998.” Colombian Economic Journal 1: 110-49. Schultz, T.W. (1960), ‘Capital formation by education,’ Journal of Political Economy, 68, 571- 583. 18   Schultz, T. W. 1961. "Investment in human capital." American Economic Review 51(1): 1−17. Smith, A. 1776. The Wealth of Nations. Alfred A. Knopf (1991). Tenjo, J., O. Álvarez, A. Gaviria and M.C. Jiménez. 2015. “Evolution of Returns to Education in Colombia (1976-2014).” Universidad de Bogota Jorge Tadeo Lozano Working Paper N. 2015-9. Tinbergen, J. 1975. Income Distribution: Analysis and Policies. North- Holland. World Bank, 2016. http://data.worldbank.org/about/country-and-lending-groups 19   Annex 1: Returns by educational level and type, most recent survey year (%) Full discounting Economy Overall Private Social Year Mincerian Source3 Prim Sec Higher Prim Sec Higher Albania 2012 8.0 Miluka (2015) Algeria 2001 2.2 Arbak (2012) Argentina 2003 11.0 Lopez Boo (2010) Australia 2010 8.3 Mariotti and Meinecke (2011) Austria 2011 10.4 9.9 8.8 7.1 OECD (2015) Azerbaijan 1995 3.7 Peet et al (2015) Bangladesh 2006 10.0 Taposh and Lin (2009) Belarus 2002 6.9 Yemtsov et al (2006) Belgium 2009 10.7 14.5 OECD (2013) Belize 2005 10.0 Hausman and Klinger (2007) Bolivia 2002 10.3 Patrinos et al 2009 Bosnia & 2002 8.1 Peet et al (2015) Herzegovina Brazil 2001 15.7 Patrinos et al 2009 Bulgaria 2003 6.7 Flabbi et al (2008) Cambodia 2007 6.7 Lall and Sakellariou (2010) Cameroon 1995 6.0 Banerjee and Duflo (2005) Canada 2011 11.2 12.9 6.9 7.4 OECD (2015) Chile 2011 14.8 8 15.5 OECD (2015) China 2009 10.3 Ding et al (2013) Colombia 2014 11.3 Tenjo et al (2015) Costa Rica 1992 8.5 Banerjee and Duflo (2005) Cyprus 1994 5.2 Banerjee and Duflo (2005) Czech 2011 14.2 19.9 22.6 15.3 OECD (2015) Republic Denmark 2011 14.4 7.9 8.7 3.9 OECD (2015) Djibouti 1996 11.6 Anos-Casero and Seshan (2006) Ecuador 2002 6.5 Peet et al (2015) Egypt, Arab 2011 3.4 Rizk (2016) Rep. El Salvador 1992 7.6 Funkhouser (1996) Eritrea 2002 10.9 Kifle (2007) Estonia 2011 21.9 17.1 6.6 5.6 OECD (2015)                                                              3 References are listed on-line in Annex 3 (http://datatopics.worldbank.org/education/files/GlobalAchievement/ReturnsEdAnnex3.docx). 20   Ethiopia 2011 12.5 Peet et al (2015) Finland 2011 6.1 8.4 20.6 6.1 OECD (2015) France 2009 9.4 9.5 6.6 6.9 OECD (2013) Gambia, The 2003 6.8 Foltz and Gajigo (2012) Georgia 2006 6.9 Botchorishvili (2007) Germany 2011 OECD (2015) Ghana 2007 4.2 Peet et al (2015) Great Britain 2011 11 12.2 45.6 30.3 OECD (2015) Greece 2009 5.9 8.6 5.4 11.6 OECD (2013) Guatemala 2000 11.2 19 13.1 Laguna and Porta (2004) Honduras 1991 9.3 Funkhouser (1996) Hong Kong 1996 6.3 9.5 10 Yeung (1999) SAR, China Hungary 2011 14.2 20.8 26.5 20.5 OECD (2015) Iceland 2003 7.2 5 7.9 5.6 OECD (2006) India 2008 10.8 World Bank (2011) Indonesia 2007 10.7 Sohn (2013) Iraq 2006 0.7 Peet et al (2015) Ireland 2009 20.9 17 6.2 15.4 OECD (2013) Israel 2011 11.1 12.1 5.1 9.5 OECD (2015) Italy 2010 8.1 9.5 6.6 8.6 OECD (2015) Jamaica 1989 28.8 20.4 15.7 17.7 7.9 Psacharopoulos (1994) Japan 2007 7.6 7.3 OECD (2013) Jordan 2001 6.7 Arbak (2012) Kazakhstan 2001 8.0 Arabsheibani and Mussurov (2006) Kenya 1995 Wambugu (2002) Korea, Rep. 2011 3.7 5.9 -0.7 0 OECD (2015) Kuwait 2010 4.8 Alqattan (2012) Latvia 2012 7.7 Vilerts et al (2015) Liberia 1983 99 30.5 17 41 17 8 Psacharopoulos (1994) Luxembourg 2010 10.2 10.1 OECD (2015) Madagascar 1993 15.1 Cogneau (2003) Malaysia 2002 10.0 Ismail (2007) Malawi 2012 8.8 Gondwe (2015) Maldives 2004 8.1 World Bank (2011) Malta 2001 25.7 Arbak (2012) Mexico 2002 13.2 Metzger and Patrinos (2005) Moldova 2003 8.0 Flabbi et al (2008) 21   Mongolia 2002 8.5 Patrinos et al 2009 Morocco 2001 2.8 Arbak (2012) Namibia 2003 28.0 Godana and Ashipala (2006) Nepal 2008 7.9 World Bank (2011) Netherlands 2011 5.5 9.1 10.8 OECD (2015) New Zealand 2011 8.3 7.6 7.7 5.7 OECD (2015) Nicaragua 2001 5.2 9.5 11.4 Belli and Ayadi (1998) Niger 2011 11.1 Peet et al (2015) Nigeria 2011 5.7 Peet et al (2015) Northern 1995 17.4 Trostel et al (2002) Ireland Norway 2011 8.7 7.1 4.5 OECD (2015) Pakistan 2009 6.2 World Bank (2011) Palestine 2011 5.1 Rizk (2016) Panama 2009 10.9 Freire et al (2015) Papua New 1987 11.1 Gibson and Fatai Guinea (2006) Paraguay 1990 11.6 23.7 14.6 13.7 20.3 12.7 10.8 Psacharopoulos (1994) Peru 2001 10.7 Trivelli (2006) Philippines 2000 12.6 Kaboski (2003) Poland 2010 10.2 26.6 6.9 13.9 OECD (2015) Portugal 2011 13.6 19.6 5.7 11.3 OECD (2015) Puerto Rico 1989 15.1 Griffin and Cox Edwards (1993) Romania 2009 11.3 Ion Zgreaban 2013 Russian 2002 9.2 Gorodnichenko Federation and Klara (2005) Senegal 1985 33.7 21.3 23 8.9 Psacharopoulos (1994) Serbia 2013 10.0 Popov (2013) Sierra Leone 1971 20 22 9.5 Psacharopoulos (1994) Singapore 2000 13.2 Low et al 2004 Slovak 2011 29.5 17.7 18.9 12.6 OECD (2015) Republic Slovenia 2011 10.3 16.8 10.1 12.7 OECD (2015) Somalia 1983 59.9 13 33.2 20.6 10.4 19.9 Psacharopoulos (1994) Spain 2011 12.9 9.8 9.6 6.5 OECD (2015) Sri Lanka 2008 8.9 World Bank (2011) Suriname 1992 9.9 10.8 Horowitz and Schenzler (1999) Sweden 2011 18.3 7.8 21 1.7 OECD (2015) 22   Switzerland 2011 7.6 2.8 OECD (2015) Thailand 2010 12.6 3.4 5.7 11.3 Tangtipongkul (2015) Timor-Leste 2007 3.9 Santos (2014) Tunisia 2011 7.0 Rizk (2016) Turkey 2008 11.8 Tansel and Daoud (2014) Uganda 2008 11.9 Peet et al (2015) Ukraine 2002 4.5 Gorodnichenko and Klara (2005) United States 2011 20.5 14 11.9 11.9 OECD (2015) Uruguay 1990 9.1 Kaboski (2003) Venezuela, 2008 6.4 Gonzales and RB Oyelere (2009) Vietnam 2014 5.7 Doan et al (2016) Yemen, Rep. 1985 10 41 56 2 26 24 Psacharopoulos (1994) Zambia 1995 10.7 19.2 5.7 Banerjee and Duflo (2005) Zimbabwe 1994 5.6 16.6 48.5 5.1 11.2 47.6 -4.6 Banerjee and Duflo (2005) 23