In-sti-tu-tions

When sitting in seminars, I now count the number of minutes until the speaker first mentions the word ‘institutions’. Usually this doesn’t take long.

It has become fashionable for economists to emphasize the pivotal role of institutions as drivers of change, or as potential obstacles to development – reflecting that institutional quality is now perceived as one of the main drivers of investments in various forms of capital (and, hence, one of the main drivers of economic growth). In fact, in both the policy arena and in academic discourse, there seems to be an obsession with institutions. Harvard economist Dani Rodrik recently noted that, after a phase of ‘market fundamentalism’ (‘get prices right!’), the development crowd is now in the middle of a phase best described as ‘institutions fundamentalism’ (‘get institutions right!’).

Where does the obsession with institutions come from? I believe a combination of three factors explains much of the hysteria. First, the deep disappointment with the dismal outcomes of most structural adjustment efforts triggered a search for explanations as to why textbook logic might fail in practice. Second, in an infamous study, Craig Burnside and David Dollar of the World Bank suggested that development aid only ‘works’ when provided to countries with good institutions and policies in place (a result that does not hold up to close scrutiny). Finally, future Nobel Prize winner Daron Acemoglu and colleagues unravelled the ‘chicken-and-egg’ problem that has obfuscated work on institutions and income. The problem was that while institutions may drive income change, income is also a determinant of institutional quality. Acemoglu et al. applied a clever econometric trick (for insiders: they used settler mortality in the 16th century as an instrumental variable for current institutions in a series of income regressions) to demonstrate a strong causal relation running from institutions to income. Follow-up work suggests that institutions trump all other explanations when it comes to explaining the long-term growth of country incomes.

While I don’t wish to deny that institutions matter for income and growth – surely they do! – I also believe that the economics profession may have gone overboard when it embraced institutions as the critical lynchpin of development.

Bones of contention

My first bone of contention is the measurement of institutions. While cross-country datasets are available on the web for all to download, enabling us to search for whatever statistical association tickles our fancy, it is not so clear what is actually captured in such datasets. The most widely used datasets are not based on measurements at all, but on surveys and opinions. For example, one can find monthly data on the ‘rule of law’ or ‘control of corruption’ for most African countries, and interestingly these data indeed display some variations over very short time scales. But this really begs the question of what is captured – real institutional changes, or some other development (income growth?) that affects the opinions of those surveyed. If it is the latter, then relating the ‘institutional variable’ to growth in a follow-up econometric analysis is not sensible, or even perverse.

Second, even if we were to take the institutional data seriously, we don’t know which institutional elements matter most. The catch-all phrase ‘institutions are rules of the game’ is not helpful. There are formal and informal institutions, and political and economic ones, and these categories must be further subdivided to become useful or operational. In practice, various institutional variables seem to ‘hang together’ and there is a highly imperfect understanding of which institutional dimensions matter most for driving development. Probably these will prove to be context-specific, which is not helpful either.

Third, even if we knew which dimensions of institutions matter most, our understanding of how to improve them is, well, laughable. If we can believe the monthly institutional data mentioned above, there has been no sign of progress over the past decade, despite the almost fanatical efforts to link development assistance to proper governance. Just as it has proven difficult to bomb the democratic spirit into people, likewise it seems difficult to change economic institutions.

The complexity of the issue is daunting. Rather than pretending that we can get institutions right (whatever that means), we might want to be a bit less ambitious and instead simply try to ‘fix’ the bottlenecks that hold back economic activity in certain regions. Sometimes bottlenecks are of the institutional kind, at other times they are not. Institutional fundamentalism, like other forms of fundamentalism, is unlikely to illuminate our path.

Footnotes