Syndicate content

Human development accounting

Youssouf Kiendrebeogo's picture

The rate of change in human development outcomes varies considerably across countries over long periods of time, as reflected in the two histograms below (Figure 1). For 78 countries in the period 1980-2014, the percentage decline in child mortality was 3.39% on average, with a standard deviation of 1.36%, a smallest rate of 0.89% (Central African Republic) and a highest rate of 8.07% (Maldives). The average percentage increase in school enrollment was 3.35%, with a standard deviation 3.54%, a minimum of 0.37% (Georgia) and a maximum of 19.68% (Maldives). Similar patterns of cross-country variation are found when using alternative proxies for health and education outcomes.

Such variations in rates of change in child mortality and school enrollment reflect important cross-country differences in human development achievements. Understanding the sources of these variations is a central issue for economic policy in that it can help us draw lessons from the best-performing countries for lagging countries. Even though exogenous forces such as technological change may play a role, most of these variations are tied to development policies.

In a recent paper, we identify five main explaining factors using a wide variety of empirical specifications: public spending on human capital; economic growth; nutrition; population density and conditional convergence. Holding constant the starting levels of child mortality and school enrollment, countries with higher levels of public spending on health or education, faster growth rates of GDP per capita, and lower levels of undernourishment experienced faster reductions in child mortality and faster increases in school enrollment. Consistent with the neoclassical growth framework, there is a tendency for countries to have faster improvements in health/education outcomes when they start with worse outcomes, relative to their “steady-state” position. The rates of convergence are 1.53% per year and 4.14% per year, respectively, for child mortality and school enrollment.

We then conduct a “human development accounting” in the spirit of the “growth accounting” literature. While the paper focuses on oil-rich countries, this exercise considers a broader perspective by focusing on the highest and lowest deciles of rates of change in human development outcomes. The idea is to look at the extent to which each explanatory factor contributed to the fitted rates of decline in child mortality and of increase in school enrollment (as deviations from sample means). Much of the variation in the actual rates of change in human development outcomes is captured by the fitted values. The correlations between actual and fitted rates of change are 0.54 and 0.80, respectively for child mortality and school enrollment. For child mortality, on average, the difference in the fitted rates of decline between the highest and lowest deciles is 2.74 percentage points per year. The corresponding difference in the fitted rate of increase in school enrollment is 10.20 percentage points per year. What are the main factors influencing the probability that a given country belongs to the top or bottom decile of the distribution of rates of change in human development outcomes?

In Table 1, the fitted rate of decline in child mortality is broken down into the contributions of each statistically significant regressor, for each country group. For the highest decile, the positive value of the fitted rate of decline in child mortality reflects the contributions from strong conditional convergence, high public spending, low undernourishment, faster growth and high population density. For the lowest decile, although the convergence effect is stronger, its positive contribution is cancelled out by the negative contributions of public spending, undernourishment, GDP per capita growth, and population density. With regard to school enrollment, for the bottom decile, the contribution of conditional convergence turns out to be negative, that is, these countries did not manage to catch up (Table 2). This country group experienced a positive contribution from high public spending, which was cancelled out by the negative contributions of conditional convergence, undernourishment and GDP per capita growth. For the top decile, the positive value of the fitted rate of increase in school enrollment is the result of the contributions from strong conditional convergence and faster GDP per capita growth.

To sum up, differences in undernourishment, public spending on human capital and the speed of convergence were the main sources of cross-country variations in the rate of decline in child mortality, as conditional convergence played an important for both country groups. With regards to school enrollment, conditional convergence and, to a lesser extent, GDP per capita growth, were the main driving forces of the probability that a country belongs to the highest/lowest decile. But in both cases, the extent of the contribution of each factor to human development achievements varied from country to country. Although the fitted values explain a substantial part of observed cross-country variations in human development achievements, the residual errors remain important in some cases. More research is therefore needed to further investigate the drivers of cross-country variations in human development achievements.





Add new comment