This page provides sample research projects.
We replicate eight influential papers to provide empirical evidence that, in the presence of heterogeneous treatment effects, OLS with fixed effects (FE) is generally not a consistent estimator of the average treatment effect (ATE). We propose two alternative estimators that recover the ATE in the presence of group-specific heterogeneity. We document that heterogeneous treatment effects are common and the ATE is often statistically and economically different from the FE estimate. In all but one of our replications, there is statistically significant treatment effect heterogeneity and, in six, the ATEs are either economically or statistically different from the FE estimates.
Quantile Regression for Peak Demand Forecasting with Ahmad Faruqui
We demonstrate that annual peak demand days are characterized by both extreme values of predictors (such as weather) and large unpredictable "shocks" to demand. OLS approaches incorporate the former feature, but not the latter, leading OLS to produce downwardly-biased estimates of the annual peak. We develop a new estimation procedure, optimal forecast quantile regression (OFQR), that uses quantile regression to estimate a model of daily peak demand, then uses a loss function framework to estimate a quantile to predict the annual peak. We compare the results of the OLS and OFQR estimation approaches for 32 utility zones. While the OFQR approach is unbiased, OLS under-forecasts by nearly 5% on average. Further, OFQR reduces the average absolute percent error by 43%. A bootstrapping procedure generates forecast intervals with accurate 95% coverage in sample and 87% coverage out of sample.
We have been asked by the Government of Alberta (or “the Province”) to conduct an economic analysis regarding certain aspects of the allegations that it provides the right to harvest provincially-owned standing softwood timber to lumber producers in Alberta at prices below what would be charged by private owners of timberlands and that these timber sales at allegedly "less than adequate remuneration" constitute a countervailable subsidy. These allegations are the latest in the “Softwood Lumber War” that dates back to 1982... The Government of Alberta requested that we determine whether calculations using the TDA transaction data can serve as an in-jurisdiction benchmark for the stumpage fees paid for the harvest of standing timber on Crown lands. We conclude that log prices in Alberta are not depressed as a result of the Crown stumpage system and, if anything, the Crown system results in higher log prices than would arise under a private timber market. We further conclude that the TDA transaction data can be used to calculate an in-jurisdiction benchmark for stumpage dues.
A Review of the National Fire Incident Reporting System and the National Fire Protection Association’s Upholstered Furniture Fire Statistics with Mark Berkman and Stephen Lagos
The FPA asked The Brattle Group to assess the reliability of upholstered furniture fire death data reported by the National Fire Incident Reporting System (NFIRS) and the interpretation of these data by the National Fire Protection Association (NFPA). The NFIRS data have served as the primary basis to estimate fire risks and fire-related costs (deaths, injuries, and property losses attributable to upholstered furniture) since 1980. Specifically, the FPA asked us to evaluate two estimates: 1) the number of deaths attributable to fires where upholstered furniture was identified as the item of first ignition (smolder + small open flame + other ignition sources) and 2) the number of deaths attributable to fires where upholstered furniture was designated as the principal item responsible for fire spread (numerous larger smolder and larger open flame ignition sources). We focus on home structure fires in our analysis. The estimated based on the item responsible for fire spread has only recently been proposed as an additional source of upholstered furniture fire-related deaths. Whether this addition actually improves the accuracy of the fire death statistics is unclear, especially in view of the limitations of the NFIRS data. Addressing this question requires a broader and more sophisticated review of the NFIRS data and its applications. This paper is an attempt to accomplish this.
R package: BrattleExtras (utilities for internal use) [Github]
TeX document class: brattlereport (document formatting for internal use) [Github]