Theoretical Econometrics

Decision-theoretic research on statistical treatment rules and optimality.

All research areas

My recent work in theoretical econometrics focuses on the derivation of optimal statistical treatment assignment rules and empirical process theory.

  1. Finite-Sample Minimax-Regret Treatment Choice with Interval-Observed Outcomes (2026)
    Nikita Pavlov
    Working paper
    Abstract

    Consider a setup where a decision maker has to choose whether to apply treatment to a population after observing a finite sample. Stoye (2009) characterises the solution to the minimax regret problem in this setup when a finite sample consists of actual outcome realizations. However, in many applied problems, instead of the realized outcomes, the decision maker observes only the intervals containing them. We study a finite-sample minimax regret problem in this setting. With no restriction on interval width, assigning treatment with probability $1/2$ yields minimax regret. With widths bounded by $h_0$ and $h_1$ in the two arms, we first show that minimax regret converges to $\frac{h_0+h_1}{4}$ in the limiting experiment and derive a corresponding asymptotically optimal plug-in treatment rule. Then, for $h_0+h_1 \geq 1$, we derive an explicit minimax regret rule which attains this value in finite samples. For $h_0+h_1 < 1$, we establish the existence of an optimal treatment rule and characterize it implicitly through a well-behaved finite-dimensional optimization problem. A simulation experiment then shows that the derived analytic rule produces superior finite-sample minimax regret relative to several heuristic alternatives, with the plug-in rule coming very close when the sample size is sufficiently large.

  2. Minimax regret treatment rules with finite samples when a quantile is the object of interest (2026)
    Patrik Guggenberger, Nihal Mehta, and Nikita Pavlov
    Accepted in Econometric Theory
    Abstract

    Consider the setup in which a policymaker is informed about the population by a finite sample and based on that sample has to decide whether or not to apply a certain treatment to the population. We work out finite sample minimax regret treatment rules under various sampling schemes when outcomes are restricted onto the unit interval. In contrast to Stoye (2009) where the focus is on maximization of expected utility the focus here is instead on a particular quantile of the outcome distribution. We find that in the case where the sample consists of a fixed number of untreated and a fixed number of treated units, any treatment rule is minimax regret optimal. The same is true in the case of random treatment assignment in the sample with any assignment probability and in the case of testing an innovation when the known quantile of the untreated population equals 1/2. However if the known quantile exceeds 1/2 then never treating is the unique optimal rule and if it is smaller than 1/2 always treating is optimal. We also consider the case where a covariate is included.

  3. Asymptotically optimal statistical treatment rules when a quantile is the object of interest (2024)
    Nikita Pavlov
    Draft coming soon! Preview of the results in parametric setting:
    Abstract

    Researchers are often presented with sufficient information to construct a class of consistent treatment assignment rules, all of which drive the loss function to zero in the limit. The seminal paper by Hirano and Porter (2009) develops large sample results to compare rules within such a class and derives treatment rules that are asymptotically optimal in the sense of minimizing mean sample loss. In this paper, I follow Hirano and Porter’s limiting experiment framework, but instead of the mean sample loss, I focus on an arbitrary quantile of the sample loss distribution. I derive asymptotically optimal Bayes and minimax sample quantile rules for some common classes of loss functions, both in parametric and semiparametric settings.