Håkon G. Grøtta🇳🇴🇺🇦

It's certainly true that cross-country growth regressions suffer from all kinds of identification problems, but I think we've still learned far more about what generates material improvements in living standards from macro-dev than micro-dev. Empirical macro-dev research isn't bad just because it lacks clean causal identification. That's an impossible bar to meet for macro in general (at least if we want to explain a significant fraction of the economically meaningful variation). There are lots of very bad empirical macro-dev papers, of course, but there are also some good ones. Take Dani Rodrik's 2008 paper on undervaluation and growth (brookings.edu/wp-content/upl…). The main observation is that real exchange rate depreciation (relative to where a country's exchange rate "ought" to be based on its initial level of development) triggers persistent growth in developing countries, but not developed ones. Rodrik's hypothesis is that market imperfections (e.g. financial frictions) that are pervasive in developing countries disproportionately affect the tradable sector (manufacturing relies more on external financing than services in the Rajan-Zingales sense), and depreciation alleviates the effects of these frictions by raising the relative price of tradables. This is important because it tells policymakers that even when fixing the underlying institutional causes of poverty is hard, there may be some second-best options that involve distorting relative prices away from the laissez-faire equilibrium. The paper shows convincing evidence that this is indeed the operative channel: undervaluation reallocates resources from services to industry, and the association between growth and undervaluation is largely explained by this reallocation in a two-stage regression. (IVs can be used to learn something about economics, not just to get identification!) Rodrik is also not just cherry-picking Asian tigers: appreciation hurts growth in lots of African countries in his sample through the same channel. This, in and of itself, provides an important (if underappreciated) cautionary tale about the unintended consequences of foreign aid. Rodrik's identification isn't clean (as Mike Woodford explains in a follow-on comment) and the paper wasn't published in a prestigious outlet (it's in the BPEA), but I think it teaches us more about development than the vast majority of micro-dev papers published in QJE & AER. Moreover, macro-dev isn't just empirics, it's also theory and quantitative modeling. Take the 2019 Econometrica by Itskhoki and Moll (onlinelibrary.wiley.com/doi/abs/10.398…).* I love this paper because it illustrates how market imperfections create a role for government to accelerate growth, but also highlights the tradeoffs and distributional tensions inherent in doing so: "The optimal policy intervention involves pro-business policies like suppressed wages in early stages of the transition, resulting in higher entrepreneurial profits and faster wealth accumulation." Finally, even focusing specifically on the empirical part of macro-dev, it isn't just growth regressions. Like many other subfields of macro, it's also increasingly come to emphasize rigorous microdata work, despite the challenges in collecting this data in developing countries. My colleague, Diego Restuccia, exemplifies this trend with his work on African farm-level data. Overall, I think this research program has been far more successful than many people think. Conversely, the elevation of clean identification as the primary goal of research in development economics has been far less successful in yielding useful insights than many people think. As John Cochrane argues (grumpy-economist.com/p/causation-do…), the primary goal of economic research should always be to explain economically-meaningful variation, even if doing so is inherently messy. * This is the paper I tried (and largely failed) to write in grad school when I was thinking about Rodrik's empirical work. My attempt is preserved for posterity here: joesteinberg.com/pdf/rerpaper. It's amazing to look back and see how far I've come as a researcher in 15 years!