Zhenling Jiang

Yanhao Wei and Zhenling Jiang (2024), Estimating Parameters of Structural Models Using Neural Networks, Marketing Science. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3496098#

Abstract: We study an alternative use of machine learning. We train neural nets to provide the parameter estimate of a given (structural) econometric model, e.g., discrete choice, consumer search. Training examples consist of datasets generated by the econometric model under a range of parameter values. The neural net takes the moments of a dataset as input and tries to recognize the parameter value underlying that dataset. Besides the point estimate, the neural net can also output statistical accuracy. This neural net estimator (NNE) tends to limited-information Bayesian posterior as the number of training datasets increases. We apply NNE to a consumer search model. It gives more accurate estimates at lighter computational costs than the prevailing approach. NNE is also robust to redundant moment inputs. In general, NNE offers the most benefits in applications where other estimation approaches require very heavy simulation costs. We provide code at: https://nnehome.github.io.

Rachel Gershon and Zhenling Jiang (2024), Referral Contagion: Downstream Benefits of Customer Referrals, Journal of Marketing Research.

Abstract: Companies often invest in referral reward programs to incentivize their current customers to spread word-of-mouth. Previous work has documented that referred customers tend to be more valuable than non-referred customers through their purchases or engagement with the company. We propose a previously overlooked benefit of encouraging referrals – referred customers are also more valuable because they make more referrals. Using a large-scale field dataset, we show that referred customers make 31-57% more referrals than non-referred customers conditional on purchase activities. Using preregistered lab experiments, we replicate the main effect and propose one underlying mechanism: referred customers perceive referring to be more socially appropriate than non-referred customers. In a field experiment, we build on previous work on norm salience and show that reminding referred customers that they joined through a referral further boosts their referral likelihood by 21%. These results advance our understanding of the social motives that contribute to referral decisions and illustrate that promoting referrals is substantially more valuable than previously estimated.

Zhenling Jiang, Yanhao Wei, Tat Chan, Naser Hamdi (2023), Designing Dealer Compensation in the Auto Loan Market: Implications from a Policy Experiment, Marketing Science, 42 (5), pp. 958-983. https://pubsonline.informs.org/doi/10.1287/mksc.2022.1418

Abstract: We study dealer compensation in the indirect auto lending market, where most lenders give dealers the discretion to mark up interest rates and the markup constitutes a dealer’s compensation. To protect consumers from potential discrimination by this dealer discretion, several banks adopted a policy that removes dealer discretion and compensates dealers by a fixed percentage of the loan amount. We document that this policy decreased (increased) the interest rates for low-credit (high-credit) consumers; however, the market share of these banks also decreased (increased) in low-credit (high-credit) segments—a reversal of the usual demand curve. This reversal highlights a significant influence of auto dealers on consumer choices. Accordingly, we develop an empirical model that features dealer-consumer bargaining. Our estimation results show systematically different levels of bargaining power across consumer groups. We use the model to explore alternative compensation schemes that remove dealer discretion. We find that a lump-sum compensation scheme obtains the most market share. In addition, the optimized lump-sum scheme improves consumer welfare compared with the adopted policy. Our study highlights the importance of accounting for the incentives and bargaining power of middlepersons.

Gad Allon, Daniel Chen, Zhenling Jiang, Dennis Zhang (Under Review), Machine Learning and Prediction Errors in Causal Inference.

Abstract: Machine learning is a growing method for causal inference. In machine learning settings, prediction errors are a commonly overlooked problem that can bias results and lead to arbitrarily incorrect parameter estimates. We consider a two-stage model where (1) machine learning is used to predict variables of interest, and (2) these predictions are used in a regression model for causal inference. Even when the model specification is otherwise correct, traditional metrics such as p-values and first-stage model accuracy are not good signals of correct second-stage estimates when prediction error exists. We show that these problems are substantial and persist across simulations and an empirical dataset. We provide consistent corrections for the case where unbiased training data is available for the machine learning dataset.

Tat Chan, Naser Hamdi, Xiang Hui, Zhenling Jiang (2022), The Value of Verified Employment Data for Consumer Lending: Evidence from Equifax, Marketing Science, 41 (4), pp. 795-814. https://pubsonline.informs.org/doi/10.1287/mksc.2021.1335

Abstract: What is the value of verified employment data in consumer lending? We study this question using a data set covering all employment verification inquiries to Equifax. Using a difference-in-differences approach, we analyze the changes in applicants’ loan outcomes after their employers join Equifax’s digital verification system, which provides lenders with an efficient way of accessing the (employer-) verified employment data in auto loan applications. Holding the employment status constant, we find that the availability of digitally verified data significantly expands credit access: the loan origination rate increases by 35.5% on average, and is more significant among deep subprime (146%) and subprime consumers (44%). The interest rates charged on these loans rise only slightly. The expanded credit access also benefits lenders, with an estimated 19.6% increase in profit. This is because the benefit of the market expansion effect dominates the cost of a higher delinquency risk among the expanded loan portfolio. Our results suggest that, besides seeking new data sources, managers and policy makers should also consider ways to extract more value from existing data.

Zhenling Jiang (2022), An Empirical Bargaining Model with Left-digit Bias: A Study on Auto Loan Monthly Payments, Management Science, 68 (1), pp. 442-465. https://pubsonline.informs.org/doi/10.1287/mnsc.2020.3923

Abstract: This paper studies price bargaining when both parties have left-digit bias when processing numbers. The empirical analysis focuses on the auto finance market in the United States, using a large data set of 35 million auto loans. Incorporating left-digit bias in bargaining is motivated by several intriguing observations. The scheduled monthly payments of auto loans bunch at both $9- and $0-ending digits, especially over $100 marks. In addition, $9-ending loans carry a higher interest rate, and $0-ending loans have a lower interest rate. We develop a Nash bargaining model that allows for left-digit bias from both consumers and finance managers of auto dealers. Results suggest that both parties are subject to this basic human bias: the perceived difference between $9- and the next $0-ending payments is larger than $1, especially between $99- and $00-ending payments. The proposed model can explain the phenomena of payments bunching and differential interest rates for loans with different ending digits. We use counterfactuals to show a nuanced impact of left-digit bias, which can both increase and decrease the payments. Overall, bias from both sides leads to a $33 increase in average payment per loan compared with a benchmark case with no bias.

Zhenling Jiang, Dennis J. Zhang, Tat Chan (2021), How Does Bonus Payment Affect Auto Loan Demand and Delinquency?, Journal of Marketing Research, 58 (3), pp. 476-496. https://journals.sagepub.com/doi/10.1177/00222437211009214

Abstract: This article studies how receiving a bonus changes consumers’ demand for auto loans and their risk of future delinquency. Unlike traditional consumer products, auto loans have a long-term impact on consumers’ financial state because of the monthly payment obligation. Using a large consumer panel data set of credit and employment information, the authors find that receiving a bonus increases auto loan demand by 21%. These loans, however, are associated with higher risk, as the delinquency rate increases by 18.5%–31.4% depending on different measures. In contrast, an increase in consumers’ base salary will increase their demand for auto loans but not their delinquency. By comparing consumers with bonuses with those without bonuses, the authors find that bonus payments lead to both demand expansion and demand shifting on auto loans. The empirical findings help shed light on how consumers make financial decisions and have important implications for financial institutions on when demand for auto loans and the associated risk arise.

Arun Gopalakrishnan, Zhenling Jiang, Yulia Nevskaya, Raphael Thomadsen (2021), Can Non-tiered Customer Loyalty Programs Be Profitable?, Marketing Science, 40 (3), pp. 508-526. https://pubsonline.informs.org/doi/abs/10.1287/mksc.2020.1268

Abstract: We study the impact of launching a non-tiered customer loyalty program on consumers’ spending per visit, frequency of visits and attrition rates, and overall customer value. We demonstrate these results both through descriptive difference-in-difference regressions and a duration-dependent hidden Markov model we develop. We find that the program increases customer value by almost 30% over a five-year horizon, which is considerably larger than has been previously found for non-tiered loyalty programs. Most of the impact of the loyalty program comes through attrition: we show that the program’s reduction in attrition accounts for more than 80% of the program’s total lift, whereas increased frequency accounts for less than 20% of the program’s lift. The program’s lift is highest for least and most frequent automatic members, who experience reductions in attrition rates after joining the program. The impact of the loyalty program on spending per visit is negligible.

Zhenling Jiang, Tat Chan, Hai Che, Youwei Wang (2021), Consumer Search and Purchase: An Empirical Investigation of Retargeting Based on Consumer Online Behaviors, Marketing Science, 40 (2), pp. 219-240. https://pubsonline.informs.org/doi/abs/10.1287/mksc.2020.1255

Abstract: This paper empirically investigates how marketers can retarget consumers who have searched online but did not purchase, based on their search behaviors. To infer the relationship between search activities and preferences, we estimate a structural search model that characterizes the consumer search process. We propose an estimator similar to the Geweke-Hajivassiliou-Keane estimator to evaluate the likelihood function. The proposed estimator makes recursive draws from truncated distributions that arise because of the observed search and choice behaviors in an optimal sequential search model. The recovered preferences are used to improve retargeting strategies demonstrated through a series of counterfactuals. Results show a substantial heterogeneity in responses to retargeting among consumers who exhibited different search behaviors. By contrast, the heterogeneity among consumers based on other characteristics (e.g., age, gender) is moderate. We consider two counterfactual marketing strategies: sending out coupons redeemed upon purchasing and sending seller recommendations that reveal the offering of recommended sellers. We find that although both strategies help increase the conversion rate, seller recommendations are more effective than coupons, suggesting the importance of providing consumers with the sellers’ information for retargeting. We also show that a pricing mechanism such as an auction that makes the seller self-select to participate will improve the effectiveness of retargeting. Finally, online retail platforms can benefit both sellers and consumers by providing sellers with the information on consumers’ search behaviors.