Hi there! I am a doctoral candidate in economics at the University of Cambridge. My research interests are in applied econometrics, statistics, public finance, and quantitative finance. My research papers, outlined below, propose new metrics for causal and comparative inference. They address practical problems with some ratios that are widely used for policy analysis. Comments and critiques are welcome. The email addresses gkarapakula@gmail.com and vgk22@cam.ac.uk can be used to contact me.
arXiv: https://arxiv.org/abs/2301.05703
SSRN: https://ssrn.com/abstract=4324986
Abstract: In this paper, I try to tame "Basu's elephants" (data with extreme selection on observables). I propose new practical large-sample and finite-sample methods for estimating and inferring heterogeneous causal effects (under unconfoundedness) in the empirically relevant context of limited overlap. I develop a general principle called "Stable Probability Weighting" (SPW) that can be used as an alternative to the widely used Inverse Probability Weighting (IPW) technique, which relies on strong overlap. I show that IPW (or its augmented version), when valid, is a special case of the more general SPW (or its doubly robust version), which adjusts for the extremeness of the conditional probabilities of the treatment states. The SPW principle can be implemented using several existing large-sample parametric, semiparametric, and nonparametric procedures for conditional moment models. In addition, I provide new finite-sample results that apply when unconfoundedness is plausible within fine strata. Since IPW estimation relies on the problematic reciprocal of the estimated propensity score, I develop a "Finite-Sample Stable Probability Weighting" (FPW) set-estimator that is unbiased in a sense. I also propose new finite-sample inference methods for testing a general class of weak null hypotheses. The associated computationally convenient methods, which can be used to construct valid confidence sets and to bound the finite-sample confidence distribution, are of independent interest. My large-sample and finite-sample frameworks extend to the setting of multivalued treatments.
Current Version: January 2023 | First Version: July 2022
arXiv: https://arxiv.org/abs/2207.13033
SSRN: https://ssrn.com/abstract=4086081
Abstract: In recent years, the Marginal Value of Public Funds (MVPF) has become a popular tool for conducting cost–benefit analysis; the MVPF relies on the ratio of willingness-to-pay for a policy divided by its net fiscal cost. The MVPF gives policymakers important information about the equity–efficiency trade-off that is not necessarily conveyed by absolute welfare measures. However, I show in this paper that the usefulness of MVPF for comparative welfare analysis is limited, because it suffers from several empirically important economic paradoxes and statistical irregularities. There are also several practical issues in using the MVPF to aggregate welfare across policies or across population subgroups. To address these problems, I develop a new axiomatic framework to construct a measure that quantifies the equity–efficiency trade-off in a better way. I do so without compromising on the core features of the MVPF: its unit-free property, and the main preference orderings underlying it. My axiomatic framework delivers a unique (econo)metric that I call the Relative Policy Value (RPV), which can be weighted to conduct both comparative and absolute welfare analyses (or a hybrid combination thereof) and to intuitively aggregate welfare (without encountering the issues in MVPF-based aggregation). I also propose computationally convenient methods to make uniformly valid statistical inferences on welfare measures. After reanalyzing several government policies using my new econometric methods, I conclude that there is substantial economic and statistical uncertainty about welfare of some policies that were previously reported to have very high or even "precisely estimated infinite" MVPF values.
Current Version: April 2022 | First Version: May 2021
Draft available upon request