Konstantin Kashin

Institute for Quantitative Social Science, Harvard University

Konstantin Kashin is a Fellow at the Institute for Quantitative Social Science at Harvard University and will be joining Facebook's Core Data Science group in September 2015. Konstantin develops new statistical methods for diverse applications in the social sciences, with a focus on causal inference, text as data, and Bayesian forecasting. He holds a PhD in Political Science and an AM in Statistics from Harvard University.

YourCast:Time Series Cross-Sectional Forecasting with Your Assumptions

(with Gary King and Federico Girosi)

At its most basic, YourCast runs linear regressions, and estimates the usual quantities of interest, such as forecasts, causal effects, etc. The benefi t of running YourCast over standard linear regression software comes from the improved performance due to estimating sets of regressions together in sophisticated ways.

YourCast avoids the bias that results from stacking datasets from separate cross-sections and assuming constant parameters, and the inefficiency that results from running independent regressions in each cross-section. YourCast instead allows you to tie the different regressions together probabilistically in ways consistent with what you know about the world and your data. The model allows you to have different covariates with different meanings measured in different cross-sections.

divider

qualCI: Causal Inference with Qualitative and Ordinal Information on Outcomes

(with Adam Glynn and Nahomi Ichino)

Exact one-sided p-values and confidence intervals for an outcome variable defined on an interval measurement scale with only qualitative and ordinal information available.

divider

panelAR: Estimation of Linear AR(1) Panel Data Models with Cross-Sectional Heteroskedasticity and/or Correlation

The package estimates linear models on panel data structures in the presence of AR(1)-type autocorrelation as well as panel heteroskedasticity and/or contemporaneous correlation. First, AR(1)-type autocorrelation is addressed via a two-step Prais-Winsten feasible generalized least squares (FGLS) procedure, where the autocorrelation coefficients may be panel-specific. A number of common estimators for the autocorrelation coefficient are supported. In case of panel heteroskedasticty, one can choose to use a ‘sandwich’-type robust standard error estimator with OLS or a panel weighted least squares estimator after the two-step Prais-Winsten estimator. Alternatively, if panels are both heteroskedastic and contemporaneously correlated, the package supports panel-corrected standard errors (PCSEs) as well as the Parks-Kmenta FGLS estimator.

divider