Skip to main content

Econometrics Workshop - Guillaume Pouliot

Tue, 05/02/2017 - 11:40am

Guillaume Pouliot

Polytechnique Montreal

498 Uris Hall

"Spatial Econometrics for Misaligned Data"

What is the impact of environmental variables such as rainfall, soil quality, and pollution on economic outcomes such as employment, income, and education? Research on this question is often stymied by the misalignment problem: the locations of the environmental observations do not generally coincide with those of the economic observations. In this article, I study a class of regression problems with spatially correlated variables. This includes regression analysis with misaligned data. I introduce a quasi-maximum likelihood estimator as well as more robust companion methods which do not require specification of the regression error covariance.  For each, limit distribution theory obtains and reliable confidence intervals are made available. I propose computational strategies and investigate their performance. Simulations show that the methods I recommend, along with the asymptotic distribution theory I derive, yield more reliable estimates and confidence intervals than previously recommended approaches. In the reanalysis of two data sets, I find that these methods yield conclusions that differ quantitatively and qualitatively from published results.

Event Categories: Econometrics