Semiparametric Identification of the Discount Factor and Payoff Function in Dynamic Discrete Choice Models
summary
The gist
This paper investigates how the discount factor and payoff functions can be identified in stationary infinite-horizon dynamic discrete choice models because common nonparametric assumptions on…
In short
The paper investigates how common nonparametric assumptions about per-period payoffs—like homogeneity and zero cross-differences—can be used to uniquely identify the discount factor in dynamic discrete choice models. It shows these shape restrictions generate specific polynomial equality and inequality constraints that pinpoint the true discount rate, extending identification to dynamic games.
Key concepts
- Nonparametric Assumptions on Payoffs
- These are broad assumptions about how per-period payoffs behave without specifying an exact mathematical formula. Examples include homogeneity (payoffs scaling predictably with state variables) or zero cross-differences (no interaction between certain state variables). These assumptions create mathematical constraints that help narrow down the possible values for the discount factor.
- Identification Restrictions
- These are the specific mathematical equations and inequalities derived from the nonparametric payoff assumptions. They take the form of polynomial equalities (e.g., $r^T_1 U = c_1$) or inequalities (e.g., $r^T_2 U eq c_2$). These restrictions are powerful because they limit the possible solutions for the discount factor, allowing for its precise identification.
- Finite Dependence
- This concept relates to simplifying dynamic models by assuming that the dependence of a decision in one period only extends over a finite number of previous periods ($ ho$). By focusing on this finite dependence, researchers can reduce the complexity of identifying polynomial restrictions from high degrees to lower, more manageable orders.
- Dynamic Game Assumptions
- When applied to games (multiple firms), identification relies on assumptions about how firms interact. Key assumptions include the irrelevance of other firms' lagged actions and exchangeability. These conditions help derive systems of polynomial inequalities that can uniquely identify firm-specific discount factors.
Terminology used across episodes
This episode discusses
- Semiparametric Identification of the Discount Factor and Payoff Function in Dynamic Discrete Choice Models · Paper Radio
- Identifying Present-Biased Discount Functions in Dynamic Discrete Choice Models
The paper
Semiparametric Identification of the Discount Factor and Payoff Function in Dynamic Discrete Choice Models · Read on arXiv
Yu Hao, Hiroyuki Kasahara, Katsumi Shimotsu
University of Hong Kong · Vancouver School of Economics · University of Tokyo
This paper studies the identification of the discount factor and payoff function in stationary infinite-horizon dynamic discrete choice models. In single-agent models, we show that nonparametric assumptions on per-period payoffs implied by economic theory, such as homogeneity, monotonicity, concavity, zero cross-differences, and complementarity, yield polynomial equalities and inequalities in the discount factor. These restrictions can identify the discount factor and the payoff function without normalizing the payoff of one action. Even when the payoff function is not point identified, they may still identify the discount factor. In dynamic games, we identify firm-specific discount factors by combining the irrelevance of other firms' lagged actions with restrictions such as exchangeability, even when the scrap value is unknown. We illustrate the approach with numerical examples and an empirical application to farmers' crop choices.
Transcript
Introduction to the show: ident: Economics. The day in economics: theory, mechanisms and the evidence beneath them.
Ruth: We're looking at the paper: "Semiparametric Identification of the Discount Factor and Payoff Function in Dynamic Discrete Choice Models".
Warren: Here's the gist of the paper. This paper investigates how the discount factor and payoff functions can be identified in stationary infinite-horizon dynamic discrete choice models because common nonparametric assumptions on per-period payoffs provide substantial identification power for the discount factor.
Ruth: Identification in Single-Agent Models The authors demonstrate that common nonparametric assumptions on per-period payoffs—such as homogeneity of degree one, monotonicity, concavity, and zero cross-differences or complementarity—generate equality and inequality restrictions with substantial identifying power.
Warren: These restrictions take the form of polynomial equalities and inequalities with degrees bounded by the cardinality of the state space.
Ruth: These restrictions also identify payoff functions under standard normalization at one action.
Paper summary: Ruth: To recap this paper, "Semiparametric Identification of the Discount Factor and Payoff Function in Dynamic Discrete Choice Models," it investigates how to identify the discount factor and payoff functions within stationary infinite-horizon dynamic discrete choice models. The thesis is that standard non-parametric assumptions about per-period payoffs—things like homogeneity, monotonicity, concavity, and zero cross-differences—provide identifying restrictions on the discount factor <ref:2507.19814#pg1>.
Warren: They claim these restrictions translate into polynomial equalities and inequalities with degrees bounded by the size of the state space <ref:2507.19814#pg3>. This is a way to limit the possibilities for that discount factor, which is usually what we want when estimating parameters.
Iris: And they also show that these restrictions can identify the payoff functions under a standard normalization applied to just one action <ref:2507.19814#pg1>. That means you don't even have to guess the shape of the utility function if you have these payoff restrictions in place.
Ruth: And they extend this analysis into dynamic game models, using assumptions such as irrelevance of other firms’ lagged actions and exchangeability <ref:2507.19814#pg3>. This addresses a gap because most prior work has focused only on single-agent settings <ref:2507.19814#pg3>.
Warren: So, the main takeaway is that leveraging these theoretically grounded shape restrictions, when paired with finite-set identification, allows researchers to achieve point identification or at least identify a small set of values for the discount factor <ref:2507.19814#pg3>.
Conclusion: Ruth: So wrapping up this paper on "Semiparametric Identification of the Discount Factor and Payoff Function in Dynamic Discrete Choice Models," what it really means is that we don't need to guess the exact functional form of agents' payoffs to figure out how much they value future things.
Warren: They show that just by assuming standard properties like homogeneity or zero cross-differences, we get concrete mathematical constraints—polynomial equalities and inequalities—that nail down the discount factor <ref:2507.19814#pg3>.
Iris: And they extend this to dynamic games too, showing that firm-specific discount factors can be identified using assumptions about other players' behavior and exchangeability <ref:2507.19814#pg3>. This is important because it broadens the scope beyond just single agents.
Ruth: In simple terms, the paper shows that common economic assumptions about how people behave over time provide a solid mathematical framework for identifying key parameters like the discount factor without needing to specify complicated payoff functions <ref:2507.19814#pg3>.
More episodes
- Daily Summary for 2026-11-19
- 2608.21274-Recommendation Quality and the Concentration of Consumption: Experimental Evidence from Netflix
- 2408.02757-A nonparametric test for diurnal variation in spot correlation processes
- 2306.16393-High-Dimensional Canonical Correlation Analysis
- 2605.17299-Geometric Brownian motion with intermittent entries and exits
- 2607.04753-Fooling Yourself: how narratives shape beliefs
- 2408.14671-Inference on Treatment Effects with Repeatedly Mismeasured High-Dimensional Controls
- 2201.06898-Difference-in-Differences Estimators for Treatments Continuously Distributed at Every Period
- 2501.15307-Influence Functions: A Unified Differentiation-Based Framework
- The day in conversation for 2026-11-19