Andrew T. Ching

Bayesian Estimation of Dynamic Discrete Choice Models

We propose a new methodology for structural estimation of infinite horizon dynamic discrete choice models. We combine the dynamic programming (DP) solution algorithm with the Bayesian Markov chain Monte Carlo algorithm into a single algorithm that solves the DP problem and estimates the parameters simultaneously. As a result, the computational burden of estimating a dynamic model becomes comparable to that of a static model. Another feature of our algorithm is that even though the number of grid points on the state variable is small per solution-estimation iteration, the number of effective grid points increases with the number of estimation iterations. This is how we help ease the “curse of dimensionality.” We simulate and estimate several versions of a simple model of entry and exit to illustrate our methodology. We also prove that under standard conditions, the parameters converge in probability to the true posterior distribution, regardless of the starting values.

A Dynamic Oligopoly Structural Model for the Prescription Drug Market after Patent Expiration

Motivated by the slow diffusion of generic drugs and the increase in prices of brand-name drugs after generic entry, I incorporate consumer learning and consumer heterogeneity into an empirical dynamic oligopoly model. In the model, firms choose prices to maximize their expected total discounted profits. Moreover, generic firms make their entry decisions before patent expiration. The entry time of generics depends on the FDA random approval process. I apply this model to the market of clonidine. The demand side parameters are estimated in a previous paper (Ching (2003)). The supply side parameters are estimated and calibrated here. The model replicates the stylized facts fairly well. I confirm that consumer heterogeneity in price sensitivity plays an important role in explaining the brand-name pricing pattern. I also apply the model to examine the impact of a policy experiment, which shortens the expected approval time for generics. Although this experiment brings generics to the market sooner, it also reduces the number of generic entrants as the likelihood of entering a crowded market in the early periods increases. Given the change in the magnitude of the policy parameter, the experiment improves the rate of learning, and lowers the equilibrium generic prices throughout the period. However, it hardly raises the overall welfare

Measuring the Informative and Persuasive Roles of Detailing on Prescribing Decisions

In the pharmaceutical industry, measuring the importance of informative and persuasive roles of detailing is crucial for both drug manufacturers and policy makers. However, little progress has been made in disentangling these two roles of detailing in empirical research. In this paper, we provide a new identification strategy to address this problem. Our key identification assumptions are that the informative component of detailing is chemical specific and the persuasive component is brand specific. Our strategy is to focus on markets where some drug manufacturers engage in a comarketing agreement, under which two or more companies market the same chemical using their own brand names. With our identification assumptions, the variation in the relative market shares of these two brands, together with their brand specific detailing efforts, would allow us to measure the persuasive component of detailing. The variation in the market shares of chemicals, and the detailing efforts summed across brands made of the same chemical, would allow us to measure the informative component of detailing. Using the data for angiotensin-converting enzyme inhibitor with diuretic in Canada, we find evidence that our identification strategy can help disentangle these two effects. Although both effects are statistically significant, we find that the persuasive function of detailing plays a very minor role in determining the demand at the chemical level---the informative role of detailing is mainly responsible for the diffusion patterns of chemicals. In contrast, the persuasive role of detailing plays a crucial role in determining the demand for brands that comarket the same chemical. This paper was accepted by Pradeep Chintagunta, marketing.

Learning Models: An Assessment of Progress, Challenges and New Developments

Learning models extend the traditional discrete choice framework by postulating that consumers have incomplete information about product attributes, and that they learn about these attributes over time. In this survey we describe the literature on learning models that has developed over the past 20 years, using the model of Erdem and Keane (1996) as a unifying framework. We described how subsequent work has extended their modeling framework, and applied learning models to a wide range of different products and markets. We argue that learning models have contributed greatly to our understanding of consumer behavior, in particular in enhancing our understanding of brand loyalty and long run advertising effects. We also discuss the limitations of existing learning models and discuss potential extensions. One key challenge is to disentangle learning as a source of dynamics from other key mechanisms that may generate choice dynamics (inventories, habit persistence, etc.). Another is to enhance identification of learning models by collecting and utilizing direct measures of signals, perceptions and expectations.

Invited Paper—Learning Models: An Assessment of Progress, Challenges, and New Developments

Learning models extend the traditional discrete choice framework by postulating that consumers have incomplete information about product attributes and that they learn about these attributes over time. In this survey we describe the literature on learning models that has developed over the past 20 years, using the model of Erdem and Keane as a unifying framework [Erdem T, Keane M (1996) Decision-making under uncertainty: Capturing dynamic brand choice processes in turbulent consumer goods markets. Marketing Sci. 15(1):1--20]. We describe how subsequent work has extended their modeling framework and applied learning models to a wide range of different products and markets. We argue that learning models have contributed greatly to our understanding of consumer behavior---in particular, in enhancing our understanding of brand loyalty and long-run advertising effects. We also discuss the limitations of existing learning models and potential extensions. One key challenge is to disentangle learning as a source of dynamics from other key mechanisms that may generate choice dynamics (inventories, habit persistence, etc.). Another is to enhance identification of learning models by collecting and using direct measures of signals, perceptions, and expectations.

Challenges to Evidence-Based Prescribing in Clinical Practice

Although there appears to be widespread support of evidence-based medicine as a basis for rational prescribing, the challenges to it are signilicant and often justified. A multitude of factors other than evidence drive clinical decision-making, including patient preferences and social circumstances, presence of diseasedrug and drug-drug interactions, clinical experience, competing demands from more pressing clinical conditions, marketing and promotional activity, and systemlevel drug policies.

Mapping Power Law Distributions in Digital Health Social Networks: Methods, Interpretations, and Practical Implications

Background Social networks are common in digital health. A new stream of research is beginning to investigate the mechanisms of digital health social networks (DHSNs), how they are structured, how they function, and how their growth can be nurtured and managed. DHSNs increase in value when additional content is added, and the structure of networks may resemble the characteristics of power laws. Power laws are contrary to traditional Gaussian averages in that they demonstrate correlated phenomena. Objectives The objective of this study is to investigate whether the distribution frequency in four DHSNs can be characterized as following a power law. A second objective is to describe the method used to determine the comparison. Methods Data from four DHSNs—Alcohol Help Center (AHC), Depression Center (DC), Panic Center (PC), and Stop Smoking Center (SSC)—were compared to power law distributions. To assist future researchers and managers, the 5-step methodology used to analyze and compare datasets is described. Results All four DHSNs were found to have right-skewed distributions, indicating the data were not normally distributed. When power trend lines were added to each frequency distribution, R 2 values indicated that, to a very high degree, the variance in post frequencies can be explained by actor rank (AHC .962, DC .975, PC .969, SSC .95). Spearman correlations provided further indication of the strength and statistical significance of the relationship (AHC .987. DC .967, PC .983, SSC .993, P<.001). Conclusions This is the first study to investigate power distributions across multiple DHSNs, each addressing a unique condition. Results indicate that despite vast differences in theme, content, and length of existence, DHSNs follow properties of power laws. The structure of DHSNs is important as it gives insight to researchers and managers into the nature and mechanisms of network functionality. The 5-step process undertaken to compare actor contribution patterns can be replicated in networks that are managed by other organizations, and we conjecture that patterns observed in this study could be found in other DHSNs. Future research should analyze network growth over time and examine the characteristics and survival rates of superusers.

A simple method to estimate the roles of learning, inventories and category consideration in consumer choice

Models of consumer learning and inventory behavior have both proven to be valuable for explaining consumer choice dynamics. In their pure form these models assume consumers solve complex dynamic programming (DP) problems to determine optimal choices. For this reason, these models are best viewed as “as if” approximations to consumer behavior. In this paper we present an estimation method, based on Geweke and Keane (2000), which allows us to estimate dynamic models without solving a DP problem and without strong assumptions about how consumers form expectations about the future. The relatively low computational burden of this method allows us to nest the learning and inventory models. We also incorporate the “price consideration” mechanism of Ching, Erdem and Keane (2009), which essentially says that consumers may not pay attention to a category in every period. The resulting model may be viewed as providing a more “realistic” or “descriptive” account of consumer choice behavior.

Behavioral Economics, Wearable Devices, and Cooperative Games: Results From a Population-Based Intervention to Increase Physical Activity

Background Health care literature supports the development of accessible interventions that integrate behavioral economics, wearable devices, principles of evidence-based behavior change, and community support. However, there are limited real-world examples of large scale, population-based, member-driven reward platforms. Subsequently, a paucity of outcome data exists and health economic effects remain largely theoretical. To complicate matters, an emerging area of research is defining the role of Superusers, the small percentage of unusually engaged digital health participants who may influence other members. Objective The objective of this preliminary study is to analyze descriptive data from GOODcoins, a self-guided, free-to-consumer engagement and rewards platform incentivizing walking, running and cycling. Registered members accessed the GOODcoins platform through PCs, tablets or mobile devices, and had the opportunity to sync wearables to track activity. Following registration, members were encouraged to join gamified group challenges and compare their progress with that of others. As members met challenge targets, they were rewarded with GOODcoins, which could be redeemed for planet- or people-friendly products. Methods Outcome data were obtained from the GOODcoins custom SQL database. The reporting period was December 1, 2014 to May 1, 2015. Descriptive self-report data were analyzed using MySQL and MS Excel. Results The study period includes data from 1298 users who were connected to an exercise tracking device. Females consisted of 52.6% (n=683) of the study population, 33.7% (n=438) were between the ages of 20-29, and 24.8% (n=322) were between the ages of 30-39. 77.5% (n=1006) of connected and active members met daily-recommended physical activity guidelines of 30 minutes, with a total daily average activity of 107 minutes (95% CI 90, 124). Of all connected and active users, 96.1% (n=1248) listed walking as their primary activity. For members who exchanged GOODcoins, the mean balance was 4,000 (95% CI 3850, 4150) at time of redemption, and 50.4% (n=61) of exchanges were for fitness or outdoor products, while 4.1% (n=5) were for food-related items. Participants were most likely to complete challenges when rewards were between 201-300 GOODcoins. Conclusions The purpose of this study is to form a baseline for future research. Overall, results indicate that challenges and incentives may be effective for connected and active members, and may play a role in achieving daily-recommended activity guidelines. Registrants were typically younger, walking was the primary activity, and rewards were mainly exchanged for fitness or outdoor products. Remaining to be determined is whether members were already physically active at time of registration and are representative of healthy adherers, or were previously inactive and were incentivized to change their behavior. As challenges are gamified, there is an opportunity to investigate the role of superusers and healthy adherers, impacts on behavioral norms, and how cooperative games and incentives can be leveraged across stratified populations. Study limitations and future research agendas are discussed.

Identification and Estimation of Forward-Looking Behavior: The Case of Consumer Stockpiling

Understanding how forward-looking consumers respond to price promotions in storable goods markets is an important area of research in empirical marketing and industrial organization. In prior work, researchers have assumed that consumers in these markets are very forward-looking, and calibrated their weekly discount factors to levels around 0.9995. This calibration has been used because earlier research has assumed that a consumer’s storage cost is a continuous func- tion of inventory, which rules out exclusion restrictions that can be used to identify the discount factor. We show that by properly modeling storage cost as a step function of inventory (be- cause storage cost depends on the number of packages stored, instead of the actual amount of inventory), natural exclusion restrictions arise that allow for the discount factor to be point identified. In an application to a storable good category, we find that weekly discount factors are very heterogeneous across consumers, and are on average 0.71. We show through a counter- factual exercise that if one used a model which fixed the discount factor to be consistent with the standard calibrated value, one would overpredict the effect of increased promotional depth for a product on its quantity sold by 18% in the short-term, and 15% in the long-term.