This blog post is contributed by Quy-Toan Do and Jishnu Das.

A recent article in the Pakistani newspaper, Dawn, pointed to a potentially disturbing aspect of World Bank lending in the country. The World Bank has undertaken many more projects when the country has had a dictator at the helm rather than a democratically elected representative. But the number of projects doesn't necessarily reflect the volume of lending. What do the data say about multilateral lending and dictatorial regimes?

The graph on the left plots  Multilateral ODA (Official Development Assistance) obtained from  the OECD DAC database and an index of democracy in the country for the years 1960-2002. The index of democracy combines several features, including political rights and the freedom of expression and ranges from -10 (very autocratic) to +10 (very democratic). The graph also plots the increasing trend in ODA to Pakistan over time.

The index performs well. Starting at 0 for the Ayub years (neither particularly autocratic or democratic) it drops to -10 when Yahya Khan comes into office, increases to 8 with Z.A. Bhutto, drops again with Zia, increases with B. Bhutto/Sharif and drops again with Musharraf. As expected, multilateral ODA (measured in millions of dollars) increases over time for Pakistan—this is shown in the trend line (dashed blue). The bottom line is that there is no systematic evidence that lending is above trend when dictators are in power and below trend when an elected representative is at the helm. Lending dropped with Yahya Khan; it was pretty much at trend with Zia; it was above trend for the first years of the Bhutto/Sharif decade and then dropped off dramatically before rising with Musharraf. We don’t have data after 2002, but suspect it’s been above trend since then.

While total lending to Pakistan is higher when dictators are in power, this is largely because they have been in power for more years. In a regression analysis, there is no evidence that more autocratic regimes are correlated with higher annual multilateral ODA. Neither is there any evidence that more democratic regimes are correlated with higher multilateral ODA. This is true not only for Pakistan, but also for all countries where these data exist. It is also a fairly robust regression—controls for GDP, country fixed-effects, population, FDI don’t alter the results qualitatively. Multilateral lending has little to do with autocracy. It also has little to do with democracy.

Should multilaterals have an explicit policy linking lending to democracy? The cynic in us says that lending is easier when you have a dictator who can undertake stroke-of-the-pen reforms (slash the tariffs! down with barriers to trade!) and can make unilateral decisions without building consensus. We also know that the value of lending for the population may be particularly high when dictators are in power; especially if it can cushion people somewhat against the worse excesses. The two most reliable studies on excess child mortality in Iraq during the nineties suggested that close to 350,000 additional children died, primarily as a result of sanctions imposed after the Gulf War.

We don’t know what the right answer is. But we do need a debate on whether the democracy-neutral lending suggested by the data is the best policy…suggestions, anyone?