Research Abstract

The U.S. government imposes a 10% penalty to discourage preretirement leakage—cash withdrawal from 401(k) retirement savings before the age of 59.5 years. In our data set with 162,360 terminating employees covered by 28 retirement plans, 41.4% of employees leaked by cashing out 401(k) savings at job separation, most draining their entire accounts. We investigate the impact of employer matching contributions on leakage at job termination. The “composition” of funds in one’s 401(k) balance matters: leakage increases with employer contribution proportion. Micropatterns in our data align more with behavioral than with economic explanations of this effect. We estimate that a 50% increase in employer/employee match rate increases leakage probability by 6.3% at job termination. However, there could be a 35.3% reduction in leakage probability if employees ignore the perceived incentive generated by the account composition effect. Approximately 60% of accumulated assets from a 50% increase in match rate leak out of the system due to the account composition effect attributable to the percentage of assets contributed by the employer. Employers with more generous matches care about their employees’ well-being in retirement, but unintentionally nudge employees to cash out when they change jobs. We highlight proposals to help employers curb avoidable leakage. 


This article aims to broaden the understanding of quasi-experimental methods among marketing scholars and those who read their work by describing the underlying logic and set of actions that make their work convincing. The purpose of quasi-experimental methods is, in the absence of experimental variation, to determine the presence of a causal relationship. First, the authors explore how to identify settings and data where it is interesting to understand whether an action causally affects a marketing outcome. Second, they outline how to structure an empirical strategy to identify a causal empirical relationship. The article details the application of various methods to identify how an action affects an outcome in marketing, including difference-in-differences, regression discontinuity, instrumental variables, propensity score matching, synthetic control, and selection bias correction. The authors emphasize the importance of clearly communicating the identifying assumptions underlying the assertion of causality. Last, they explain how exploring the behavioral mechanism—whether individual, organizational, or market level—can actually reinforce arguments of causality. 


The prevalence of strong brands such as Coca-Cola, McDonald’s, Budweiser, and Marlboro in “vice” categories has important implications for regulators and consumers. While researchers in multiple disciplines have studied the effectiveness of anti-tobacco counter-marketing strategies, little attention has been given to how brand strength may moderate the efficacy of tactics such as excise taxes, usage restrictions, and educational advertising campaigns. In this research, we use a multiple discrete-continuous model to study the impact of anti-smoking techniques on smokers’ choices of brands and quantities. Our results suggest that while cigarette excise taxes decrease smoking rates, these taxes also result in a shift in market share towards stronger brands. Market leaders may be less affected by tax policies because their market power allows strong brands such as Marlboro to absorb rather than pass through increased taxes. In contrast, smoke-free restrictions cause a shift away from stronger brands. In terms of anti-smoking advertising we find minimal effects on brand choice and consumption. The findings highlight the importance of considering brand asymmetries when designing a policy portfolio cigarette tax hikes, smoke-free restrictions, and anti-smoking advertising campaigns.


Sports franchises derive significant portions of their revenues from season ticket holders. A development that may affect season ticket management is the growth of legal secondary markets. We develop a structural model that integrates both the supply and demand sides of the secondary market into season ticket buyers’ ticket purchase and usage choices. We use a panel dataset that combines season and single ticket purchase records with ticket usage data to investigate the value of secondary markets. We estimate that the secondary market increases the team’s season ticket revenues by about $1 million per season. At the level of the individual season ticket customer, we estimate an increase in CLV ranging from $1,327 in the lowest quality seat tier to $2,553 in the highest. In terms of value to the customer, the average dollar value of having a secondary market is $138 per season ticket. Across segments, the secondary market provides the equivalent of a 4% discount in the premium seat tier versus an 11% discount in the economy seat tier. While the secondary market creates more value in the premium-ticket tier segments, the secondary market has the most impact on behavior in the low price oriented segment.


This paper investigates the impact of mobile hailing technology on taxi driving behaviors. A controversial feature of mobile hailing applications in China is the disclosure of not only pickup locations but also drop-off destinations before drivers accept offers. It provides taxi drivers two different mechanisms to improve their hourly earnings: reducing cruising time and selecting more profitable trips. We examine 3.6-terabyte minute-by-minute geolocation data of 2,106 single-shift drivers in Beijing. A modified change-point model is proposed to infer the adoption decisions and estimate the changes in driving behaviors. We show that mobile hailing technology adoption is associated with an average increase of 6.8% in hourly earnings, equivalent to an extra CNY 750 monthly income. A typical taxi driver greatly improves hourly earnings through trip selection in favor of longer trips rather than aiming for cruising-time reduction. We find that the relative importance of cruising-time reduction and trip selection depends on driver skills and market conditions. We do not find market expansions on the number of trips or working hours, but rather a redistribution of realized trips toward long distances.


Political advertising is controversial, as there is widespread concern about money from political action committees (PACs and super PACs) distorting the democratic process. Studying advertising effectiveness is, however, a challenging topic for several reasons, including the endogenous nature of fundraising and ad spending rates. However, the extensive use of targeting based on designated marketing areas (DMAs) creates a setting in which neighboring counties with comparable demographics receive different levels of advertising exposure. In this paper, we leverage these advertising discontinuities along DMA borders to study the relative effectiveness of political advertising on vote shares and turnout rates in 2010 and 2012 senatorial elections. We find that negative advertising sponsored by PACs is significantly less effective than that sponsored by candidates in affecting two-party vote shares and voter turnout. A 1% increase in negative advertising by the candidate produces a significant 0.015% lift in the candidate’s unconditional vote shares. By contrast, negative advertising from PACs is ineffective in increasing its supported candidate’s unconditional vote share. Further analysis reveals that the competitiveness of races moderates the effectiveness of political advertising, providing implications for those managing candidates’ campaigns, PACs, and super PACs.


This research investigates whether there are enduring effects of goal achievement and failure within customer loyalty promotion programs. We collaborated with a major hotel chain to launch a large scale field experiment involving 95,532 existing loyalty customers. We observed customers’ hotel stays for eight months before the experiment, eight months during the experiment, and eight months after the experiment. Customers in the treatment group were asked to increase their hotel nights during the 8-month promotion by a set percentage relative to their baseline to receive a reward. Overall, the promotion led to increased purchasing in the post-promotion period. However, only 20% of customers successfully reached the goal whereas 80% missed the goal. We use a propensity score analysis to examine the distinct effects of goal achievement versus goal failure. Results show that goal attainment significantly increased post-promotion purchasing whereas goal failure significantly reduced post-promotion purchasing. Additionally, we use econometric methods to empirically test a behavioral theory of relationship-based reciprocity. We find that customers in a high status tier relationship, with the most invested in the firm, are most affected by goal failure whereas customers in a low status tier relationship, with the least invested in the firm, are most affected by goal success. Because the type of loyalty program described in this paper is widely used in a variety of industries the findings suggest that marketers should set reachable goals within loyalty promotion programs. Firms should be particularly cautious about the impact of goal failures for the firm’s most loyal customers.


Countermarketing, or efforts to reduce consumption of certain products, has become common in categories such as tobacco, junk food, fossil fuels, and furs. Countermarketing has a particularly long history in the tobacco industry. Efforts to reduce smoking have included excise taxes that increase the cost of consumption, smoke-free restrictions that make consumption less convenient, and antismoking advertising campaigns that highlight the dangers of tobacco use. This article presents an analysis of the relative effectiveness of these different strategies. We find that cigarette excise taxes are the most effective tool for reducing overall cigarette sales, followed by antismoking advertising. Smoke-free restrictions are not found to have a significant effect on cigarette sales. We also investigate how these various policy tools induce product substitution. This issue is of considerable importance because some countermarketing techniques may potentially shift consumers to more dangerous, higher nicotine and tar cigarettes. Specifically, we find that excise taxes levied on a per pack basis rather than based on nicotine levels often shift consumers to more dangerous products.