Institutions are now widely believed to be important in explaining performance. In this paper, we analyze whether commonly used measures of institutions have any significant, measurable impact on performance, whether of countries or firms. We look at three 'levels' of institutions and associated conjectures. The first concerns whether the political system affects performance. The second concerns whether the business and investment environment affects the performance of countries and the third concerns whether perceived business constraints directly affect the performance of firms. In all instances, we find little evidence of a robust link between widely used measures of institutions and our indicators of performance. We consider why this might be the case and argue that mis-measurement, mis-specification, complexity and non-linearity are all relevant factors.In the conclusion to the paper Commander and Nikoloski write:
Our paper has taken a close look at this proposition by focusing on three, related questions. The first concerned whether the type of political system, and its associated institutions, tends to affect performance. The simple conjecture, drawn from a significant literature, was that democracy in particular has features that should be encouraging for performance, even if that underlying relationship was not linear. This was addressed using several sets of country level measures of political institutions and through use of leading edge GMM estimation. The second concerned the impact of institutions connected to the investment and business environment on the performance of countries, irrespective of their political configuration. In particular, this part of the analysis focused on a widely cited measure of the business environment that covers 175 countries; the World Bank’s Doing Business. The third question was to ask whether the evidence could robustly support the broad proposition that the performance of firms’ could be materially influenced by the business environment. This required, above all, econometric implementation able to address the pervasive problems of endogeneity and unobserved heterogeneity.There is, I would say, a broad consensus that institutions do matter for economic performance, and its hard to see why they wouldn't. May be we just don't have good measures of those parts of the institutional frame work that do matter for performance.
The results reported in the paper are ambiguous, if not hostile, to the default proposition of institutions affecting performance. In the case of political institutions, none of the explanatory variables was significant. For country level analysis we were limited by an absence of an adequate number of observations on time. But the analysis that we were able to implement indicates that no robust conclusions can be drawn. In the case of firm level analysis, using a large two-period dataset on twenty six transition countries – countries whose initial conditions comprised largely similar institutional formats – we were unable to find any strong relationship between revenues and the institutional constraints. Country effects that captured other sources of cross-country heterogeneity were found to matter for performance.
Finally, the paper addressed why these exercises have yielded a relatively meagre harvest, at least when held up against the prevailing orthodoxy. Put simply, it would appear that issues of measurement – including bias arising from subjective evaluation – misspecification, complexity and non-linearity are all relevant.