Table 2.
Comparing measures of human capital and economic growth
| Annual growth rate (2000–2010) | ||||||||
|---|---|---|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | |
| Human capital (harmonized learning outcomes) | ||||||||
| Regression coefficient | 0.072 | 0.059 | 0.061 | 0.069 | 0.066 | |||
| Standard error | 0.018 | 0.023 | 0.022 | 0.023 | 0.024 | |||
| P value | 0.000 | 0.011 | 0.007 | 0.003 | 0.006 | |||
| Human capital (Penn world tables) | ||||||||
| Regression coefficient | 0.033 | 0.012 | 0.019 | |||||
| Standard error | 0.011 | 0.013 | 0.035 | |||||
| P value | 0.003 | 0.358 | 0.597 | |||||
| Human capital (schooling32) | ||||||||
| Regression coefficient | 0.016 | 0.006 | 0.014 | |||||
| Standard error | 0.006 | 0.007 | 0.020 | |||||
| P value | 0.004 | 0.382 | 0.494 | |||||
| Human capital (HDI) | ||||||||
| Regression coefficient | 0.020 | 0.002 | −0.028 | |||||
| Standard error | 0.008 | 0.010 | 0.022 | |||||
| P value | 0.013 | 0.844 | 0.914 | |||||
| Control for GDP at the start of the time period | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Observations | 107 | 107 | 107 | 107 | 107 | 107 | 107 | 107 |
| R2 | 0.300 | 0.261 | 0.255 | 0.240 | 0.306 | 0.305 | 0.301 | 0.318 |
Columns 1–8 include the same dependent variable. The difference across columns depends on which independent variables are included as denoted by whether the rows are filled out. The dependent variable is the annual growth rate averaged across 2000–201014. Human capital (harmonized learning outcomes) refers to the measure of human capital in this database averaged from 2000 onwards. Human capital (Penn world tables) refers to the measure of human capital in the Penn world tables v.9.014. Human capital (schooling32) refers to estimates from 2000, which is the start of our sample period32. Human capital (HDI) refers to the measure of education included in the United Nation’s HDI in the year 200015. Results include 107 countries and exclude countries in civil war, inflation crises and with rents from natural resources above 15%. All independent variables are transformed to log units to derive comparable elasticities. We control for the log transformation of initial GDP per capita levels at the beginning of the period (in the year 2000) in all specifications following standard practice in the growth literature3. We report regression coefficients, standard errors of the regression and P values.