My Research

My CV

Brown et al. (2005) describe the discipline of marketing as “an important applied constituency that demands tools (theories, methods, and measures) to inform marketing, consumer, or public policy actions.” Policymakers regularly implement policies that affect how firms undertake marketing, and sometimes firms’ marketing actions warrant policy interventions. Yet scant research exists in marketing focused on public policy. 

I address this research gap by investigating substantive and timely questions at the interface of marketing and public policy. Herein, my research tackles some of the hotly debated issues of our time, from public health concerns around physician-directed detailing and direct-to-consumer advertising (DTCA), to the unintended consequences of the Affordable Care Act’s (ACA) sunshine laws on competition, the impact of the health insurance mandate on emergency room utilization, the impact of the federal-auto-bailout-induced Chapter 11 bankruptcies on price competition and consumer demand, the marketing of violent movies to underage youth and its corresponding impact on youth crime, the behavioral biases that lead consumers to purchase extended warranties that they rarely utilize, and the precipitous decline in international student enrollments in U.S. universities, etc. 

I draw on theories from marketing, economics, and psychology and employ a diverse set of research methods, including structural models, descriptive analysis, causal inference methods and spatial econometrics. The complexity of my empirical analysis is always driven by my desire to produce actionable insights for policymakers and/or marketing managers in industries that are critical to the U.S. economy. 

My research has been published in Marketing Science, the Journal of Marketing Research, the American Economic Review (Papers and Proceedings) and the Review of Financial Studies – all top-tier marketing, economics and finance journals. 

My research has been referenced by The White House, analysts at the Federal Reserve Board, and in policy circles, lending it additional credibility and relevance. 

My policy-oriented research has also been recognized by overseas governments. In 2016, I was one of the recipients of the Indian Institute of Management - Bangalore (IIMB) Challenge Research Grant administered by the Office of the Prime Minister of India. 

My research contributions have also been recognized by marketing academia. I was inducted into the inaugural class of 2018 MSI Marketing Scholars, which comprises the top 30 marketing scholars worldwide from the 2004~2008 graduating cohorts. 

It is my hope that my research in concert with the emerging interest in the field around marketing and public policy, will give marketers a voice at the table as policies that affect our discipline are considered. 

Surprisingly, the evidence in the prior academic research is mixed. In -

we resolve this controversy by showing that drug characteristics – such as a drug’s effectiveness and side effects – are the missing link. Once we account for drug characteristics, we show that physicians do, in fact, “carefully weigh what they hear” both from firms and patients. Detailing has a higher return on investment for drugs with more side effects. Additionally, we show that limiting physician access or banning sampling by physicians may be the wrong tack. The paper presents clear guidelines for public policy and managerial practice and envisions that the study of the role of drug characteristics may lead to valuable insights in this surging public debate. 

Beyond detailing and sampling, financial relationships between physicians and medical product manufacturers can include everything from free meals to consulting or speaker fees and direct research funding. These relationships have the potential to create conflicts of interest and negatively impact public health. In order to address this public health concern, the U.S. federal government passed the Physician Payment Sunshine Act as part of The Patient Protection and Affordable Care Act (ACA). However, the law’s transparency requirements led hospitals and physician practices to severely restrict pharmaceutical sales representatives’ direct access to physicians. For example, in 2013, 45 percent of subscribers placed significant restrictions on detailing representatives’ access, compared to 35 percent a year prior to the law’s passage. 

How do pharmaceutical firms respond to reduced access to physicians as a result of the Physician Payment Sunshine Act? Who wins? Who loses?  

In - 

we advance a structural model that accounts for the reality that firms compete dynamically when it comes to scheduling detailing visits to physicians. We find that restriction policies benefit firms that have the weakest detailing effects and hurt firms with the strongest detailing effects. This deleterious effect is especially pronounced for new drugs that need to build up goodwill stock and often have the largest detailing effect. We show that this result may be undesirable when the new drugs have significantly better therapeutic qualities relative to their mature counterparts. We also find that an unexpected outcome of this act is to soften competition between firms, thereby enhancing their profits. (Therefore, firms may want to think twice about lobbying against such restrictions.) Furthermore, the share of non-drug treatments or generic drugs expands. Depending on the effectiveness of the outside option, this may help or hurt public health. 

If reduced access to physicians leads firms to reduce their sales force, how might competitors respond if one firm undertakes a large-scale reduction of its detailing sales force? 

Before a firm contemplates a major change to its detailing policy, either because of these laws and/or firm-related financial objectives, it is important to infer ex ante how competitors might respond. Providing tools to do this is particularly timely in marketing and policy circles today, as regulatory concerns and shrinking pipelines of patent-protected drugs have led pharmaceutical firms to question the size and allocation of their sales force. 

In - 

we propose a novel approach that fuses revealed prescription and detailing data with stated competitive response data obtained from experts through a conjoint experiment. From a methodological standpoint, our generalizable approach to fusing observational and stated-choice data affords a richer set of counterfactual analysis than using observational or stated-choice data alone. We find that the consequences of a policy shift depend heavily on who initiates it. All scenarios in which the market leader initiates a policy shift are profitable for all companies. Furthermore, for every policy shift we consider, the initiator’s profits increase. Only when the market leader decreases its sales force by 40 percent do all other firms decrease their detailing efforts. With this study, we extend the methodological approach of data enrichment to study supply-side competitive reactions. 

Beyond marketing activity directed at physicians (detailing, sampling, fees, etc.), marketing directed at patients is also coming into focus for public health regulators. For example, TV ads urging patients to “Ask your doctor if this drug is right for you” are ubiquitous in the U.S., one of the only countries where DTCA is still legal. However, public health concerns that DTCA leads to overmedication have policymakers wondering if DTCA should be regulated or made illegal, such as in other developed nations. 

In - 

we introduce a model that examines the full chain linking DTCA, patients requesting drugs by brand name, and doctors writing prescriptions. Prior to our work, studies on this issue were rare and inconclusive. Perhaps contrary to popular belief, our study finds that the number of patients who actually ask for drugs by name in response to a TV ad is rather small. For patients who do ask, doctors are overwhelmingly inclined to write the prescription, which is in line with public policy concerns. Our most controversial finding is that drug requests by name in regions with higher percentages of African Americans and Hispanics translate into fewer prescriptions of that brand than in regions with higher percentages of Caucasians. Thus, policymakers may also want to monitor whether the right to participate in medical decisions is equally distributed across socio-demographic groups. 

In the first months of the 2008 economic meltdown, automakers saw sales decline 50 percent and slashed 400,000 jobs. Toward the end of 2008, there was full-scale panic over whether automakers and auto suppliers might fail. 

Theory predicts that financial distress can reduce demand for products if consumers worry about a firm’s ability to supply after-sales flows of goods and services, such as warranties, spare parts, and maintenance (Titman 1984). The concern for after-market flows further harms profitability, exacerbating the firm’s financial distress, which in turn further heightens the demand impact. This “bank-run-like” cycle can spiral until the firm falls into bankruptcy (Diamond and Dybvig 1983). 

Despite the lack of empirical evidence, America witnessed several hotly debated policy interventions, broadly characterized as the “auto bailout,” aimed at arresting this death spiral. In March 2009, President Obama introduced the U.S. Treasury Department-backed Warranty Commitment Program that guaranteed warranties of new GM and Chrysler cars if they went bankrupt. Backers said this could break the vicious circle between consumer expectations of automakers’ financial health and demand for their products. However, if the feedback mechanism were weak to begin with, such an intervention could be an ineffective use of taxpayer dollars. 

In -

we quantify the effect of automakers’ financial distress on consumer demand. This quantification is empirically challenging because causality can also operate in the opposite direction: Negative demand shocks affect firms’ cash flows and therefore can induce financial distress. 

We study the effect of financial distress on the prices of used cars because, unlike new cars, used- car prices do not directly affect manufacturers’ cash flows. Therefore, the potential for reverse causality is lower for used cars than new cars. Specifically, we compare shifts in the prices of used vehicles and their manufacturers’ likelihood of bankruptcy measured using credit default swap (CDS) spreads. Consistent with the economic motivations of the program, our research revealed that a 30 percent increase in the public’s perception of the likelihood of GM’s bankruptcy wiped out 25 percent of GM’s margin and 21 percent of GM’s North American book value. These effects are even larger for vehicles with longer lives remaining on their service warranties. As far as we know, this is the first academic study to measure the indirect effect of financial distress on consumer demand, and the first to formally lend empirical support to the Warranty Commitment Program.

In -

we first confirm the theoretical possibility of self-fulfilling expectations highlighted by defenders of the warranty program. However, when we calibrate the equilibrium model with our estimates from our 2011 (AER P&P) study, we empirically show that while financial distress can substantially reduce automakers’ profits, it is not large enough to trigger bank-run-like effects. Thus, while the warranty program benefited consumers and firms, fears that GM and Chrysler might fail entirely without the bailout were likely overstated. 

To provide these insights, we assess the bank-run-like risk faced by automakers that served as the bedrock of the federal auto bailout. We anticipate the empirical framework we propose will have even greater impact as part of the growing body of research in economics and finance investigating bank failures (Egan et al. 2017). 

As part of the auto bailout, the government also forced GM and Chrysler to restructure through Chapter 11 bankruptcy protection. This restructuring allowed GM and Chrysler to terminate roughly a quarter of their dealer networks. Such large-scale distribution-strategy changes were unprecedented in the auto industry, as state laws prevent automakers from closing down dealerships at will outside of bankruptcy protection.

In - 

we study this question. Evaluating the impact of a bankrupt automaker’s dealer exits on incumbent dealers’ car prices is a nontrivial task because the decision of which dealers to terminate is endogenous. We devised a reduced-form, yet causal-inference-friendly, empirical approach to answer this question with robust econometric identification. 

We find prices increase by about 1 percent following an exit and that the increase is lowest at dealerships more proximate to the exiting dealerships because they suffer the biggest loss in agglomeration benefits. Because these findings could impact consumer welfare, policymakers should evaluate not just the competitive effects of these mergers, but also their consequences for consumer search costs. These implications transcend the auto industry and are relevant in other forms of retailing where competitive and agglomeration forces are at play.

We address this question in 

Specifically, we empirically study the effect of Chrysler’s Chapter 11 bankruptcy filing on the quantity sold by its competitors in the U.S. auto industry. The demand for competitors could increase because they may benefit from the distress of the bankrupt firm (competitive effect). By contrast, competitors could experience lower sales if the bankruptcy increases consumer uncertainty about their own viability (contagion effect). Furthermore, a challenge to measuring the impact of bankruptcies is the coincident decline in economic conditions stemming from the Great Recession and the potential effect of the “cash for clunkers” program (among other confounding factors). 

To identify the effect of the bankruptcy filing, we employ a regression-discontinuity-in-time design based on a temporal discontinuity in treatment (i.e., bankruptcy filing), along with an extensive set of control variables. We find that unit sales for an average competitor decrease by 28 percent following Chrysler’s bankruptcy filing. We advance empirical evidence that suggests that this negative demand spillover effect is driven by a heightened consumer uncertainty about the viability of the bankrupt firm’s rivals.

Verhoef et al. (2015) define omnichannel management as the “synergetic management of the numerous available channels and customer touchpoints, in such a way that the customer experience across channels and the performance over channels is optimized.” Herein, firms interact seamlessly with their customers across a multitude of channels (owned, paid and earned) with relevant and timely communications so as to provide a single unified experience across channels. Omnichannel management, therefore, further expands the definition of channels to include channels that are not owned by the firm (paid and earned). Omnichannel management also creates a unified consumer experience rather than just facilitating transactional/relational exchanges as the central function of marketing channels. 

A key element to the success of omnichannel marketing is social media (an example of earned media). Social media has fundamentally changed the way we communicate and collaborate (Aral, Dellarocas and Godes 2013). However, econometric identification and quantification of these word-of-mouth (WOM) effects remain key challenges. Furthermore, the current practice of using metrics such as valence and ratings raises concerns over endogeneity or, even worse, reverse causality (Hartmann et al. 2008, Aral 2011). Even as social media platforms rise in prominence, firms are still continuing to invest in traditional marketing-mix elements like price promotions, advertising, distribution, etc. Surprisingly, scant empirical research exists that contrasts social media effects (WOM) with traditional marketing-mix elements. My work in this area addresses these two aforementioned empirical challenges.

advances a novel econometric framework to causally identify WOM effects using market-level data and, from a substantive standpoint, answers the following research question: How does consumer-generated online word-of-mouth (OWOM) impact offline product performance when products are sequentially released across markets? From pizzas to automobiles to movies, firms make strategic decisions about how to sequentially release products across geographic markets. Surprisingly, the literature is silent on how previous methods to quantify WOM effects apply to these empirical settings. Three unique identification challenges plague industries where products are sequentially rolled out, namely: (i) spatial aggregation across heterogeneous markets, (ii) serial correlation of unobservables, and (iii) serial correlation as a result of endogenous rollout. We propose an instrumental variable-based limited-information approach to address these three identification concerns. 

The resulting findings challenge some long-accepted findings in this area. Where previous studies found that the volume of user reviews is the main driver of firm performance, we find that valence is what matters instead. Our approach using market-level data reveals that what drives the reversal of previous studies is that they suffer aggregation bias by relying on national-level data. After having illustrated the source of the bias, we outline a way to correct for this bias. Because we use market-level data, we are also able to quantify the responsiveness to WOM across geographically dispersed markets. 

The econometric approach proposed in this study has been leveraged by scholars in marketing, economics, and operations management to study other dimensions of consumer and/or firm outcomes and link them to consumer welfare or firm surplus. This paper has also been very well received in the field and is currently the 6th most cited paper published in Marketing Science.

builds on our 2010 Marketing Science paper, and contrasts the effectiveness of online consumer-generated WOM (earned media) with traditional advertising (paid media) and how this effectiveness varies across consumers. Understanding the relative impact of paid versus earned media is critical to the firm’s resource allocation decisions and thereby social surplus. 

Substantively, this study answers the following research question: How do consumers’ responsiveness to consumer-generated WOM and firm-generated advertising vary across markets? Unlike advertising, little is known about how responsiveness to consumer-generated media varies across consumers and geographic markets. The paper advances an econometric framework to answer the aforementioned substantive research question while addressing potential endogeneity issues that arise due to unobservables that might be: (i) serially correlated and (ii) correlated with both WOM measures and advertising decisions. We find that markets most responsive to blogs and advertising, and markets highly responsive to blogs only, are spread more or less evenly across different regions. However, markets highly responsive to advertising only seem to be concentrated in the Midwest, while markets that are least responsive to both blogs and advertising seem to be clustered around the East Coast. With this study, we provide a road map for firms to improve their distribution channel product-rollout strategies by accounting for synergies between earned and paid media. This study, too, has received considerable attention in the field and is the first study to causally estimate the heterogeneous and relative impact of earned and paid media across markets.

Omnichannel management facilitates a seamless and consistent experience for customers across all touchpoints with the firm and its products. This is particularly true in media industries like television broadcasting and newspaper publishing. As consumers take advantage of the seamless integration of content and experience across different media platforms (television, print, online, radio, etc.), media multiplexing (consuming media across multiple devices) presents both huge challenges and opportunities for media planners. 

In - 

we calibrate a structural model that can predict where a target audience may be at any given point in time, and which media or combination of media a target audience is likely to consume. Our study represents a significant advancement over the literature on media choice modeling, much of which assumes that media are consumed one channel (or type) at a time. We highlight opportunities for firms to improve their targeting while leveraging inter-media synergies. 

However, in practice, firms face significant challenges in harnessing the full potential of omnichannel marketing. These challenges arise because firms generally do not have the single, unified approach that is needed for a personalized customer experience. Thus, scattered and disjointed customer information results in poor orchestration and personalization of cross-channel customer interactions and an inconsistent customer experience. 

In 

my fellow MSI Scholars and I synthesize the state of the extant literature in omnichannel marketing and highlight research questions that are worth investigating in future research in this area. In particular, we discuss the role of emerging technologies such as artificial intelligence (AI) and blockchain in shaping the future of omnichannel marketing. In addition, we consider some issues related to privacy that might arise as these technologies become increasingly adopted. 

In summary, my work in omnichannel marketing represents a novel contribution by allowing firms in multiple industries to better allocate their resources across geographic markets and media platforms. Because of the broad appeal of the empirical models I proposed in my published studies, my work has been widely cited across diverse fields, including marketing, economics, IS, finance and operations.