The Unassuming Economist

From a VoxEU post by Markus Eberhardt:

“Recent evidence suggests that a country switching to democracy achieves about 20% higher per capita GDP over subsequent decades. This column demonstrates the sensitivity of these findings to sample selection and presents an implementation which generalises the empirical approach. If we assume that the democracy–growth nexus can differ across countries and may be distorted by common shocks or network effects, the average long-run effect of democracy falls to 10%.

In a recent paper, Acemoglu et al. (2019), henceforth “ANRR”, demonstrated a significant and large causal effect of democracy on long-run growth. By adopting a simple binary indicator for democracy, and accounting for the dynamics of development, these authors found that a shift to democracy leads to a 20% higher level of development in the long run.¹

The findings are remarkable in three ways:

1. Previous research often emphasised that a simple binary measure for democracy was perhaps “too blunt a concept” (Persson and Tabellini 2006) to provide robust empirical evidence.
2.  Positive effects of democracy on growth were typically only a “short-run boost” (Rodrik and Wacziarg 2005).
3. The empirical findings are robust across a host of empirical estimators with different assumptions about the data generating process, including one adopting a novel instrumentation strategy (regional waves of democratisation).

ANRR’s findings are important because, as they highlight in a column on Vox, there is “a belief that democracy is bad for economic growth is common in both academic political economy as well as the popular press.” For example, Posner (2010) wrote that “[d]ictatorship will often be optimal for very poor countries”.

The simplicity of ANRR’s empirical setup, the large sample of countries, the long time horizon (1960 to 2010), and the robust positive – and remarkably stable – results across the many empirical methods they employ send a very powerful message against such doubts that democracy does cause growth.

I agree with their conclusion, but with qualifications. My investigation of democracy and growth (Eberhardt 2019) captures two important aspects that were assumed away in ANRR’s analysis:

• Different countries may experience different relationships between democracy and growth. Existing work (including by ANRR) suggests that there may be thresholds related to democratic legacy, or level of development, or level of human capital, or whether the democratisation process was peaceful or violent. All may lead to differential growth trajectories.²
• The world is a network. It is subject to common shocks that may affect countries differently. The Global Crisis is one example, as are spillovers across countries (Acemoglu et al. 2015, in the case of financial networks).”

Continue reading here.