Benjamin D. Williams
  • Home
  • Research
  • Teaching
  • cv
  • Home
  • Research
  • Teaching
  • cv

Publications

Quantifying the failure of bootstrap likelihood ratio tests (with Mathias Drton)
2011, 
Biometrika, 98, 4, pp. 919-934 (pdf) (published version)
Identification of a Nonseparable Model under Endogeneity using Binary Proxies for Unobserved Heterogeneity.
2019, Quantitative Economics, 10, 2, pp. 527-563 (pdf) (supplementary appendix) (published version)
Controlling for ability using test scores
2019, Journal of Applied Econometrics, 34, 4, pp. 547-565 (pdf) (published version)
The Repeat Time-on-the-Market Index (with Paul Carrillo)
2019, Journal of Urban Economics, 119, pp. 33-49 (link)
Identification of the linear factor model
2019, Econometric Reviews (pdf) (supplementary appendix) (published version)
Nonparametric Identification of Discrete Choice Models with Lagged Dependent Variables
2019, Journal of Econometrics (pdf) (published version)
Minimum wage and women's decision making power within households: Evidence from Indonesia (with Jin Kim)
forthcoming at Economic Development and Cultural Change (pdf)

Working papers

A many item analysis of a linear regression with sum scores as regressors
revision requested (pdf)

Works in progress

Large T Nonparametric Identification of Binary Choice Panel Data Model
abstract. 
This paper studies nonparametric identification in binary panel data models with fixed effects. Point identification of the distribution of treatment effects can be obtained as the number of time periods grows. These results do not require stationarity or time homogeneity. Treatment effects can be identified for the subpopulation of “nonmovers” – those individuals who maintain the same regressor values over time. Apart from imposing that regularity conditions on the conditional distribution of fixed effects given the regressors hold uniformly over the support of the regressors, the dependence between the fixed effect and the regressors is unrestricted. Instead monotonicity in the fixed effect and restrictions on the tails of the distribution of fixed effects are required. A brief discussion shows how the identification arguments can be used constructively to consistently estimate average treatment effects.
A nonparametric analysis of roll call voting
abstract. This paper develops nonparametric estimates of the relative ideological positions of legislators and changes over time in the distribution of legislator ideological position from roll call votes. These estimates are compared to results obtained from the prevailing DW-NOMINATE model. Results corroborate the finding of increased polarization in the US House of Representatives.
 Estimation of the interactive fixed effects model with a short panel (with Bob Phillips)
Powered by Create your own unique website with customizable templates.