Jasmine Hao

Ph.D candidate, Vancouver School of Economics


Working Papers

Building up Trust in a Dynamic Game: A study on Collusive Price-fixing in the Chilean Pharmaceutical Retail Industry[Job Market Paper]

Abstract This paper discusses firms’ coordination issues when initiating collusion. By understanding the economics behind the initiation of collusion, the government can tailor policies to prevent collusion from emerging. The paper is the first to model the firms’ initiation problem. The work contributes to understanding the firms’ learning-to-coordination process. From empirical researches, we observe that firms exhibit post-cartel tacit collusion. The observations indicate that once firms build up trust, the market is vulnerable to collusion. Literature in collusion focuses on the implementation but overlooks the initiation of collusion. This paper provides a tractable model that considers firms’ incentive problems and coordination problems separately. We relaxes the rational expectations by estimating firm-specific “belief parameter” that disentangle firms’ information acquisition processes from firms’ strategic interactions. Identifying the belief parameters relies on two exclusion restrictions: (1) one firm’s lagged pricing decision affects his payoff through adjustment costs while other firms’ lagged pricing decisions do not. (2) The profits on a given market are not affected by the market outcomes in other markets. The framework with nonequilibrium belief represents the data observed better than the rational expectation model.

Award: Bank of Canada Graduate Student Paper Award 2020

Presentation: WEAI International Conference 2021(scheduled); Econometric Society DSE Winter School 2020; Bank of Canada Graduate Student Paper Award Workshop 2020; Vancouver School of Economics.

Slides Draft Paper

Using Euler equation to estimate non-finite-dependent dynamic discrete choice model with unobserved heterogeneity(with Hiro Kasahara)


In the dynamic discrete choice analysis, controlling for unobserved heterogeneity is an important issue, and finite mixture models provide flexible ways to account for it. The previous discussion of incorporating finite mixture model in the dynamic discrete choice model focuses on a class of models where the difference in future value terms depends on a few conditional choice probabilities(finite dependence property). In models that do not exhibit finite dependence property, it is computationally costly to estimate finite mixture models with the expectation-maximization(EM) algorithm. Following the previous discussion of the finite mixture in dynamic discrete choice with finite dependence property, this paper adopts the EM algorithm to incorporate unobserved heterogeneity for a broader range of dynamic discrete choice model that does not require the finite dependence property. Following the Euler Equation(EE) representation of dynamic discrete decision problems, we provide an alternative conditional choice probability (CCP) value function representation that relies only on the CCP of one action. Contrasting to the Hotz-Miller CCP representation that relies on all the conditional choice probabilities, this characterization avoids the matrix inversion in each EM iteration. The matrix inversion can be computed outside the EM iterations and therefore is computationally attractive. The characterization provides unbiased estimator for models with and without finite dependence property. We illustrate the computational gains with Monte Carlo simulations.

Presentation: Canadian Economics Association 2019 Annual Meeting; Seattle-Vancouver Econometrics Conference 2020; Bank of Canada; Vancouver School of Economics.

Slides Draft Paper

Testing the number of components in finite mixture model with normal panel regression(with Hiro Kasahara)


This paper develops the likelihood-ratio based test of the null hypothesis of a $m_0$-component model against an alternative of $(m_0+1)$-component model in the normal mixture panel regression. I show that the normal mixture panel regression does not suffer from the Fisher Information matrix degeneracy under the reparameterization proposed in Kasahara and Shimotsu(2012). As a result, the likelihood ratio test statistic can be approximated by a local quadratic expansion of squares and products of the reparameterized parameters. Moreover, I obtain the data-driven penalty function via computational experiments to attend to unbounded likelihood ratio. In addition, I apply the test to random coefficient Cobb-Douglas production function estimation following the framework of Gandhi et al.(2013) and Kasahara and Shimotsu(2015). The empirical findings suggest evidence of heterogeneous production technology beyond Hicks-neutral technology factor.

Presentation: International Association for Applied Econometrics 2019 Annual Meeting; China Meeting of the Econometric Society 2019 Guangzhou China; Vancouver School of Economics.

Slides Draft Paper

Work in Progress

Dynamic Decision Models with Continuous-Discrete Mix Choices


In dynamic decision problems, agents can make both discrete and continuous choices at the same time. The existence of both types of choices is natural under some circumstances. For example, empirical industrial organization literature examines firms’ entry and investment decisions. The decision of entry is discrete, and the decision of investment is continuous. \citet{Blevins2010} provides identification results of the class of dynamic discrete-and-continuous-choice models. We show the discrete-and-continuous model is equivalent to the agents’ making decisions that map every possible state to an outcome simultaneously. With the property, the agent’s future value can be represented as the discounted payoff from repeatedly taking an arbitrary action. The estimation technique is the first to account for the Dynamic decision models with discrete-continuous-mix choices.

Does the EV rebate program raise awareness on the environment: evidence based on China automobile market (with Yiran Hao)


This project uses administrative vehicle registration data from one of China’s major cities to identify consumers’ preference over household vehicles’ gas-efficient attributes over time. We propose to evaluate the long-run effect of electric vehicles(EV) adoption policy on the consumer’s preference using administrative data from one major city in China. The data contains registration, transfer and disposal record from January 2010 to the present. The administrative data include the Vehicle Identification Number(VIN) of the registered vehicle, the household district information, the gender, and the consumer’s date of birth. The identification relies on the relative preference of high displacement vehicles and low displacement vehicles. The Chinese tax structure creates a discontinuity in demand for the displacement attribute. The Chinese government imposes a 7.5 % consumption tax for a vehicle with engine displacement below 1.6 litres and a 10 % tax for those above 1.6 litres. The level of the difference between vehicle above 1.6-litre displacement compared to those below 1.6 litres conditional on rebate program for electric cars over time can explain whether the consumers’ preference for environmentally friendly cars has changed.