An international perspective on Malaysia's schooling and cognitive skills, 1960–2020

Dr David Demery, Formerly Research Fellow, University of Bristol, United Kingdom

Economists have long stressed the importance of human capital and schooling in the process of economic growth. In a paper published 30 years ago, Mankiw et al. (1992, p. 407) reported that ‘an augmented Solow model that includes accumulation of human as well as physical capital provides an excellent description of the cross-country data’. Their human capital variable was the secondary school enrolment rate, aimed at capturing the fraction of gross domestic product (GDP) diverted to raising the stock of human capital.

In subsequent research, for example Cohen and Soto (2007), the stock of human capital in a country is more directly linked to the average years of schooling in its population. Yet more recent research by Hanushek and Wößmann (2012), and Altinok et al. (2018), challenges this approach to human capital measurement as it ignores substantial variations in schooling quality across countries.

This article assesses how Malaysia’s record on schooling, cognitive skills, and economic growth compares internationally over the past six decades. The first section analyses Malaysia’s international rank in average schooling level and GDP per capita. The second focuses on the effect of schooling on long-run economic growth. The final section examines the role of schooling quality by analysing the effect of measures of cognitive skills on economic growth.

Schooling and GDP per capita

Analysis of the cross-country relationship between years of schooling and the level of real GDP per capita shows how Malaysia’s position has changed over the past six decades. Cross-country schooling data have been put together by various researchers. Barro and Lee (2013) published schooling data for 146 countries in five-year intervals from 1950 to 2010. Lutz et al. (2018) have assembled schooling data for a larger sample of countries—up to 202 countries in 2015. Cohen and Soto (2007) and Cohen and Leker (2014) have published average years of schooling for individuals in five-year age groups for between 88 and 95 countries between 1960 and 2020 in 10-year intervals.

De la Fuentea and Doménech (2006 and 2015) published schooling data for 12 countries in the Organisation for Economic Co-operation and Development. While not directly relevant for the focus on Malaysia, they find that ‘there are important differences in quality across data sets’ with their own and the Cohen and Soto series outperforming the other data sets. In particular they report that the Barro and Lee (2013) series ‘contains sharp breaks and implausible changes in attainment levels over very short periods’. As a result, the schooling data used here are those published by Cohen and Soto (2007), referred to as CS.1

Cross-country data for real GDP and population are available in the Penn World Table (PWT) dataset version 10.0 (Feenstra et al., 2015). The latest release (June 2021) covers 183 countries for each year between 1950 and 2019. GDP is defined as expenditure-side real GDP at chained purchasing-power parity, in millions of 2017 US$. This allows comparisons to be made of relative living standards across countries and over time.2  The PWT dataset is merged with the CS data on schooling. Because of variations in the coverage of the two datasets, the sample of countries with data for both schooling and GDP in the merged dataset varies: 82 countries in 1960; 91 in 1970, 1980 and 1990; 93 in 2000 and 2010; and 86 in 2019/20.

For four selected years since 1960, Figure 1 has scatter-plots of the logarithm of real GDP per capita and the CS average years of schooling for males and females aged between 25 and 64. This age group covers working-age individuals, most of whom will have completed their formal schooling. GDP for 2020 is not yet available in PWT, so the GDP per capita markers in the last panel are based on observations for 2019. The size of the marker is based on the country’s population size, so that India, China, and the United States are clearly visible. Malaysia is indicated by the red marker.

Three features are apparent in Figure 1. First, countries with higher levels of schooling generally have higher levels of GDP per capita—the scatter points in every graph are upward sloping. This may not mean that more schooling raises GDP per capita; causation may well be the reverse: richer countries can afford to offer more schooling. Applying econometric methods that remove the effect of possible reverse causation, Cohen and Soto (2007) estimate that an additional year of schooling raises GDP per worker by around 12 per cent, a return that is broadly consistent with the microeconomic evidence reviewed by Psacharopoulos and Patrinos (2018).

Figure 1 GDP per capita and CS schooling data: selected years
Source: Cohen and Soto (2007) and Feenstra et al. (2015)

Second, GDP per capita and schooling levels have both risen over time for most, if not all, countries. For example, in 1960 there is quite a cluster of markers with average schooling levels below four years. By 2020, there are very few in this area. And the reverse is true of schooling levels above eight years.

Third, Malaysia’s position relative to other countries shows a marked improvement over time, as seen in the country’s rank or percentile position.

Table 1 reports Malaysia’s position in 75 countries that are sampled in every year. In 1960, Malaysia was at the 51st percentile position for GDP per capita and the 40th percentile for average years of schooling. This means that just under 50 per cent of the countries in the sample had higher GDP per capita and 60 per cent had higher schooling levels than those of Malaysia. Over the following four decades to 2000, Malaysia’s relative position improved substantially for both variables. Only 31 per cent of the countries had higher levels of GDP per capita in 2000 and only 29 per cent enjoyed higher levels of schooling. In the new millennium, however, Malaysia has failed to advance by much in GDP per capita, and while its average years of schooling has continued to rise, other countries were catching up, some overtaking Malaysia to reduce its percentile rank from 71 in 2000 to 63 by 2020 (Table 1).

High education attainment in Malaysia raises GDP per capita
Table 1 Malaysia’s percentile position in GDP per capita and in schooling
Source: Cohen and Soto (2007) and Feenstra et al. (2015)

Schooling and growth

Much cross-country empirical work has focused on the role of schooling in explaining variations in the growth rate of GDP per capita, rather than its level. In the augmented Solow-Swan model of Mankiw et al. (1992), schooling influences the steady-state (or long run) level of GDP per capita but not its long-run growth rate. In endogenous growth models (for example, Aghion and Howitt, 1998) schooling can raise the long-run growth rate through several channels. A better-educated workforce leads to a larger stream of new ideas that produces technological progress at a higher rate. A better-educated workforce will also more easily adopt and adapt new technologies developed by other, more innovative, countries—enhancing technological diffusion. Better-educated societies should enjoy higher long-term growth.

Two periods of growth and schooling are analysed: the last two decades of the 20th century and the opening two decades of the 21st. Countries with higher initial levels of GDP per capita tend to grow more slowly than poorer countries that are ‘catching up’—so-called conditional convergence. A simple two-dimensional plot of growth against schooling may not successfully pick up the separate effect of schooling on growth.

Figure 2 has ‘added-variable plots’. These show the association between economic growth and schooling after the influence of base-year GDP per capita has been removed. Growth and schooling are first regressed on base-year GDP per capita and only the residuals of these regressions are used in the graph. In this way, the graphs plot conditional growth against conditional schooling, having removed the influence of base-year GDP per capita from both.

From this two-dimensional view of the data in Figure 2, it is apparent that while countries with more schooling also experienced higher growth over the last two decades of the 20th century, this appears no longer to be so over the first two decades of the new millennium.

Figure 2 Conditional GDP per capita growth and conditional schooling
Source: Cohen and Soto (2007) and Feenstra et al. (2015)

These features are confirmed by econometric evidence. Simple regressions for the two sub-periods are reported in Table 2, where the dependent variable is the average annual growth rate of real GDP per capita over the appropriate two decades. The terms in parentheses are standard errors. The schooling and base-year coefficients have the correct sign and are statistically significant in 1980–2000. Over the first two decades of the new millennium, the schooling coefficient is not statistically different from zero, confirming the visual features of Figure 2.

Table 2 Economic growth, schooling, and cognitive skills regressions
Dependent variable: average percentage growth rate of GDP per capita

Schooling quality

The use of simple measures of schooling, like average years of schooling, has been questioned as they fail to account for variations across countries in the quality of schooling. Hanushek and Wößmann (2007) explain:
Schooling has not delivered fully on its promise as the driver of economic success. Expanding school attainment, at the center of most development strategies, has not guaranteed better economic conditions. What’s been missing is attention to the quality of education—ensuring that students actually learn. There is strong evidence that the cognitive skills of the population, rather than mere school enrolment, are powerfully related to individual earnings, to the distribution of income, and to economic growth (p. 1)
Hanushek and Wößmann (2012)—HW for short—developed a measure of cognitive skills based on 12 international student achievement tests conducted between 1964 and 2003. Their measure of cognitive skills is the simple average of all observed maths and science scores. They report that cognitive skills significantly influenced the economic growth rates of 50 countries over 1960 to 2000, and that years of schooling made no additional significant contribution. The students participating in the tests covered by the HW measure will have reached working age in the period 2000 to 2019; so the use of their measure to explain growth since 2000 is even more appropriate than over the period they analysed (1960–2000).

More recently Altinok, Angrist, and Patrinos (2018)—AAP—have assembled a cognitive skills measure for 163 countries over 1965–2015.3  They claim that they have assembled ‘the largest globally comparable panel database of cognitive achievement, including 163 countries and regions, 32 of which are from Sub-Saharan Africa, over the last 50 years (1965–2015)’ (p. 4). In so doing, their coverage includes more developing countries than HW, ‘countries that have the most to gain from the potential benefits of a high-quality education’ (p. 2). AAP confirm HW’s finding that cognitive skills have a positive effect on economic growth. Figure 3 plots conditional growth rates of GDP per capita over 2000–2019 against conditional cognitive skills measures published by HW and AAP. In both cases, cognitive skills appear to have a positive effect on economic growth.

High level skills can enhance economic growth
Source: Economic History of Malaya website

Figure 3 Conditional GDP per capita growth and conditional cognitive skills
Source: Hanushek and Wößmann (2012); Altinok et al. (2018) and Feenstra et al. (2015)

The graphical impression is borne out by more formal econometric evidence, presented in Table 3. The base-year GDP per capita coefficients have the correct signs (poorer countries grow faster) and are statistically significant, as are the cognitive skills coefficients. The unit coefficients on cognitive skills suggest a substantial effect on economic growth. Malaysia’s cognitive skills scores place it high in a ranking of these skills over all countries. Its skill measure of 4.8 places it at the 61st percentile of HW’s sample of 76 countries. And in the AAP sample of 131 countries, Malaysia occupies the 73rd percentile position – only a quarter of countries sampled had higher levels of cognitive skills.

Table 3 Economic growth 2000–2019 and cognitive skills regressions
Dependent variable: average percentage growth rate of GDP per capita, 2000–2019


Cognitive skills matter for economic growth. Singapore’s cognitive skills score of 5.8 places it second only to Taiwan in AAP’s coverage, one point higher than Malaysia’s 4.8, which is similar to the scores of Australia and Sweden. The regression coefficients imply that Singapore’s one-point cognitive skills advantage over Malaysia would lead the former to grow one percentage point faster over the first two decades of the new millennium. In fact, Singapore grew at 4.9 percent over the period compared with Malaysia’s 3.6 percent.

For stronger economic growth, therefore, Malaysian policymakers need to place much more attention on measures that that will improve the quality of the population's education, and not just on years of schooling. Such measures should include standards of teaching, with an emphasis on science, mathematics, and English, as well as other subjects that will meet the skills required by the labour market and support innovation and creativity.

Further reading:

Aghion, P. and Howitt, P. 1998. Endogenous Growth Theory. Cambridge, Mass: MIT Press.

Altinok, N., Angrist, N., and Patrinos, H. A. 2018. ‘Global Data Set on Education Quality (1965-2015)' World Bank Policy Research Paper No. 8414. Washington, DC: World Bank.

Barro, R. J. and Lee, J.W. 2013. ‘A new data set of educational attainment in the world, 1950-2010’. Journal of Development Economics, 104, pp. 184–198.

Cohen, D. and Leker, L. 2014. ‘Health and education: another look with the proper data’. Manuscript.

Cohen, D. and Soto, M. 2007. ‘Growth and human capital: good data, good results’. Journal of Economic Growth, Vol. 12(1), pp. 51–76.

de la Fuentea, A. and Doménech, R. 2006. ‘Human capital in growth regressions: how much difference does data quality make?’. Journal of the European Economic Association, 4 (1), pp. 1–36.

______ 2015. ‘Educational attainment in the OECD, 1960–2010. Updated series and a comparison with other sources’. Economics of Education Review, 48, pp. 56–74.

Feenstra, R. C., Inklaar, R., and Timmer, M.P. 2015. 'The Next Generation of the Penn World Table' American Economic Review, 105(10), pp. 3150–3182.

Hanushek, E. A. and Wößmann, L. 2007. Education Quality and Economic Growth. Washington DC: World Bank.

______ 2012. 'Do better schools lead to more growth? Cognitive skills, economic outcomes, and causation'. Journal of Economic Growth, 17, pp. 267–321.

Lutz, W., Goujon, A., Samir, K. C., Stonawski, M. and Stilianakis, N. (Eds.). 2018. Demographic and Human Capital Scenarios for the 21st Century: 2018 assessment for 201 countries., Luxembourg: Publications Office of the European Union.

Mankiw, N. G., Romer, D., and Weil, D. N. 1992. ‘A contribution to the empirics of economic growth’. The Quarterly Journal of Economics, Vol. 107:2, pp. 407–437.

Psacharopoulos, G. and Patrinos, H. A. 2018. ‘Returns to Investment in Education: A Decennial Review of the Global Literature’. World Bank Policy Research Working Paper 8402. Washington, DC: World Bank.



c/o Asia-Europe Institute
University of Malaya,
50603 Kuala Lumpur

Install EHM PWA App

Generic Popup