Jayant Ray and Francisco L. Rivera-Batiz, “An Analysis of Sample Selection Bias in Cross-Country Growth Regressions.”
Sample sizes in cross-country growth regressions vary greatly, depending on data availability. But if the selected samples are not representative of the underlying population of nations in the world, ordinary least squares coefficients (OLS) may be biased. This paper re-examines the determinants of economic growth in cross-sectional samples of countries utilizing econometric techniques that take into account the selective nature of the samples. The regression results of three major contributions to the empirical growth literature by Mankiw-Romer-Weil (1992), Barro (1991) and Mauro (1995), are considered and re-estimated using a bivariate selectivity model. Our analysis suggests that sample selection bias could significantly change the results of empirical growth analysis, depending on the specific sample utilized. In the case of the Mankiw- Romer-Weil paper, the value and statistical significance of some of the estimated coefficients change drastically when adjusted for sample selectivity. But the results obtained by Barro and Mauro are robust to sample selection bias...
In the Mankiw-Romer-Weil (1997) paper, we found that using their 75-country sample leads to the exclusion of a number of low-income and middle-income countries that results in a substantial sample selection bias. The value and statistical significance of the estimated growth equation coefficients reported by Mankiw-Romer-Weil for this sample of countries change drastically when adjusted for sample selectivity. But in re-examining these results using Mankiw-Romer-Weil’s 98-country sample, we found much smaller differences in estimated coefficients. The impact of sample selection bias on the Mankiw-Romer-Weil results is thus dependent on the choice of sample.
Related;
Growth Regressions and Policy Advising
I have long been skeptical about how much one can learn from cross-country growth regressions. In the early 1990s, I wrote one paper in that literature, coauthored with David Romer and David Weil, and to my surprise, it turned out to be my most cited paper by a very large margin. In a subsequent paper, The Growth of Nations, I tried to spell out the reasons for my skepticism. I emphasized three problems, which I called the simultaneity problem (it is hard to disentangle cause and effect), the multicollinearity problem (most of the potential determinants of growth are correlated with each other and imperfectly measured, making it hard to figure out which is the true determinant), and the degrees-of-freedom problem (there are more plausible hypotheses than data points). To some extent, the subsequent literature addresses some of my concerns. For example, there is more attention now to trying to find exogenous differences across countries, but the task is inherently difficult, so one should not expect to find definitive answers about the causes of growth from this literature.
Regressions: Why Are Economists Obessessed with Them?
Why are some countries richer than others? A skeptical view of mankiw-romer-weil's test of the neoclassical growth model
No comments:
Post a Comment