Search

Online ratings play an important role in many markets. We study the often disputed information content of these ratings, by proposing a reduced-form Bayesian model of the typical buyer's rating decision. Our empirical evidence based on eBay raw data is in line with even intricate predictions from it. We thus have good reasons to calibrate the model to moments of the data. Our simulations suggest that the rating record reveals the seller's type after about 100 transactions, or 65-70 ratings.

We develop a new approach to quantify how patients respond to dynamic incentives in health insurance contracts with a deductible. Our approach exploits two sources of variation in a differences-in-regression-discontinuities design: deductible contracts reset at the beginning of the year, and cost-sharing limits change over the years. Using rich claims-level data from a large Dutch health insurer we find that individuals are forward-looking. Changing dynamic incentives by increasing the deductible by €100 leads to a reduction in healthcare spending of around 3% on the first days of the year and 6% at the annual level. The response to dynamic incentives is an important part of the overall effect of cost-sharing schemes on healthcare expendituresâ€"much more so than what the previous literature has suggested.

We use high frequency data on TV and radio advertising together with data on online sales for lottery tickets to measure the short run effects of advertising. We find them to be strong and to last for up to about 4 hours. They are the bigger the less time there is until the draw. We develop the argument that this finding is consistent with the idea that advertisements remind consumers to buy a ticket and that consumers value this. Then, we point out that in terms of timing the interests of the firm and the consumers are aligned: consumers wish to be reminded in a way that makes them most likely to consider buying a lottery ticket. We present direct evidence that this does not only affect the timing of purchases, but leads to market expansion. Then, we develop a tractable dynamic structural model of consumer behavior, estimate the parameters of this model and simulate the effects of a number of counterfactual dynamic advertising strategies. We find that relative to the actual schedule it would be valued by the consumers and profitable for the firm to spread advertising less over time and move it to the last days before the draw.

Many developing countries have recently increased health insurance coverage at a large scale. While it is commonly believed that this has positive effects, to date, it is not well understood through which channels health insurance coverage contributes to the well-being of individuals. More generally, the effects are usually not quantified at the individual level. There are two main reasons for this. First, we lack detailed data on health care utilization and health outcomes, and second, it is not easy to control for selection into insurance. The second problem means that a regression of utilization or outcome measures on insurance coverage will yield biased results and will not estimate the causal effects of health insurance. In this paper, we make progress in both directions. We use rich survey data to evaluate the impact of access to the Peruvian Social Health Insurance called "Seguro Integral de Salud" for individuals outside the formal labor market on a variety of measures for health care utilization, preventive care, health expenditures, and health indicators. We address the second concern by exploiting a fuzzy regression discontinuity design. A household is eligible for the program if a welfare index that is calculated from a number of variables is below a specific threshold. We base our analysis on a natural experiment that is generated by variation in the index around the threshold. We interpret our results through the lens of a simple model. As expected, and in contrast to studies for a number of other countries, we find strong effects of insurance coverage on measures of health care utilization, such as visiting a doctor, receiving medication and medical analysis. The program does not strongly incentivice individuals or health care providers to invest into preventive care. In line with this, in general, we find no effects of insurance coverage on preventive care. The only exceptions to this are our findings that, controlling for selection into insurance coverage, women of fertile age with insurance are more likely to receive pregnancy care and that insured individuals are more likely to be vaccinated. This is in line with the stark decrease in maternal and child mortality that was observed after the program was introduced. As for health care expenditures, we generally find positive effects on the mean and the variability. We complement these findings with quantile treatment effect estimates that show increases at the high end of the distribution. Our interpretation is that insured individuals are encouraged by health care professionals to undertake important treatments and pay for this themselves. At the same time, we find no clear effects on health outcomes at the micro level.

We study the effects of improvements in market transparency on eBay on seller exit and continuing sellers' behavior. An improvement in market transparency by reducing strategic bias in buyer ratings led to a significant increase in buyer valuation especially of sellers rated poorly prior to the change, but not to an increase in seller exit. When sellers had the choice between exiting'€”a reduction in adverse selection'€”and improved behavior'€”a reduction in moral hazard'€”, they preferred the latter because of lower cost. Increasing market transparency improves on market outcomes.

We confront a representative sample of one 1,102 Dutch individuals with a series of incentivized investment decisions and also elicit their time preferences. There are two treatments that differ in the frequency at which individuals decide about the invested amount. The low frequency treatment stimulates decision makers to frame a sequence of risky decisions broadly rather than narrowly. We find that the framing effect is significantly larger for impatient than for patient individuals. This result is robust to controlling for various economic and demographic variables and for cognitive ability.

We provide an explanation for the common finding that the effect of retirement on life satisfaction is negligible. For this we use subjective well-being measures for life and domains of life satisfaction that are available in the German Socio-Economic Panel (GSOEP) and show that the effect of voluntary retirement on satisfaction with current household income is negative, while the effect on satisfaction with leisure is positive. At the same time, the effect on health satisfaction is positive but small. Following the life domain approach we then argue that these effects offset each other for an average individual and that therefore the overall effect is negligible. Furthermore, we show that it is important to distinguish between voluntary and involuntary retirement. The effect of involuntary retirement is negative because the adverse effect on satisfaction with household income is bigger, the favorable effect on satisfaction with leisure is smaller, and the effect on satisfaction with health is not significantly different from zero. These results turn out to be robust to using different identification strategies such as fixed effects and first differences estimation, as well as instrumental variables estimation using eligibility ages and plant closures as instruments for voluntary and involuntary retirement.

In Germany, employees are generally obliged to participate in the public health insurance system, where coverage is universal, co-payments and deductibles are moderate, and premia are based on income. However, they may buy private insurance instead if their income exceeds the compulsory insurance threshold. Here, premia are based on age and health, individuals may choose to what extent they are covered, and deductibles and co-payments are common. In this paper, we estimate the effect of private insurance coverage on the number of doctor visits, the number of nights spent in a hospital and self-assessed health. Variation in income around the compulsory insurance threshold provides a natural experiment that we exploit to control for selection into private insurance. We document that income is measured with error and suggest an approach to take this into account. We find negative effects of private insurance coverage on the number of doctor visits, no effects on the number of nights spent in a hospital, and positive effects on health.

eBay’s feedback mechanism is considered crucial to establishing and maintaining trust on the world’s largest trading platform. The effects of a user’s reputation on the probability of sale and on prices are at the center of a large number of studies. More recent theoretical work considers aspects of the mechanism itself. Yet, there is confusion amongst users about its exact institutional details, which also changed substantially in the last few months. An understanding of these details, and how the mechanism is perceived by users, is crucial for any assessment of the system. We provide a thorough description of the institutional setup of eBay’s feedback mechanism, including recent changes to it. Most importantly, buyers now have the possibility to leave additional, anonymous ratings on sellers on four different criteria. We discuss the implications of these changes and provide first descriptive evidence on their impact on rating behavior.

We propose and implement an estimator for identifiable features of correlated random coefficient models with binary endogenous variables and nonadditive errors in the outcome equation. It is suitable, e.g., for estimation of the average returns to college education when they are heterogeneous across individuals and correlated with the schooling choice. The estimated features are of central interest to economists and are directly linked to the marginal and average treatment effect in policy evaluation. The advantage of the approach that is taken in this paper is that it allows for non-trivial selection patterns. Identification relies on assumptions weaker than typical functional form and exclusion restrictions used in the context of classical instrumental variables analysis. In the empirical application, we relate wage levels, wage gains from a college degree and selection into college to unobserved ability. Our results yield a deepened understanding of individual heterogeneity which is relevant for the design of educational policy.

A fundamental identification problem in program evaluation arises when idiosyncratic gains from participation and the treatment decision depend on each other. Imbens and Angrist (1994) were the first to exploit a monotonicity condition in order to identify an average treatment effect parameter using instrumental variables. More recently, Heckman and Vytlacil (1999) suggested estimation of a variety of treatment effect parameters using a local version of their approach. However, identification hinges on the same monotonicity assumption that is fundamentally untestable. We investigate the sensitivity of respective estimates to reasonable departures from monotonicity that are likely to be encountered in practice and relate it to properties of a structural parameter. One of our results is that the bias vanishes under a testable linearity condition. Our findings are illustrated in a Monte Carlo analysis.

We analyze the joint determination of wage levels, wage growth and firm tenure. Our analysis is built on estimating a reduced form for tenure, a structural wage level equation and a structural wage growth equation. We disentangle returns to a latent type variable from estimates of general returns to tenure and wage gains from job changes. This type is related to unobservable match quality that is allowed to vary over time and to be correlated with the returns to tenure. The obtained results for Germany indicate that the type plays a crucial role in the remuneration of employees. Those types who change jobs more often obtain steeper wage profiles but earn less on average.

Feedback mechanisms that allow partners to rate each other after a transaction are considered crucial for the success of anonymous internet trading platforms. We document an asymmetry in the feedback behavior on eBay, propose an explanation based on the micro structure of the feedback mechanism and the time when feedbacks are given, and support this explanation by findings from a large data set. Our analysis implies that the informational content of feedback records is likely to be low. The reason for this is that agents appear to leave feedbacks strategically. Negative feedbacks are given late, in the "last minute," or not given at all, most likely because of the fear of retaliative negative feedback. Conversely, positive feedbacks are given early in order to encourage reciprocation. Towards refining our insights into the observed pattern, we look separately at buyers and sellers, and relate the magnitude of the effects to the trading partners' experience.

Over the last decades individual field data has become more and more available to researchers. In three self-contained chapters, this thesis is concerned with the analysis of such data and some associated challenges. Chapter 1 contains an introduction. Chapter 2 is concerned with the problem of evaluating the impact of a treatment on an outcome variable, e.g. the effect of on-the-job training on wages. A fundamental identification problem arises in case the treatment decision and the idiosyncratic gains from participation depend on each other. Imbens and Angrist (1994) were the first to exploit monotonicity of the treatment decision in instruments in order to identify an average treatment effect parameter. More recently, Heckman and Vytlacil (1999, 2000a, 2000b, 2005) suggested estimation of a variety of treatment effect parameters using a local version of their approach. However, identification hinges on the same monotonicity assumption that is fundamentally untestable. We investigate the sensitivity of respective estimates to reasonable departures from monotonicity that are likely to be encountered in practice and relate it to properties of a structural parameter. One of our results is that the bias vanishes under a testable linearity condition. Our findings are illustrated in a Monte Carlo analysis. Thereafter, in Chapter 3, we propose and implement an estimator for identifiable features of correlated random coefficient models with binary endogenous variables and nonadditive errors in the outcome equation. The estimator we propose is suitable, e.g., for estimation of the average returns to college education if they are heterogeneous across individuals and correlated with the schooling choice. The estimated features are of central interest to economists and are directly linked to the marginal and average treatment effect in policy evaluation. They are identified under assumptions weaker than typical exclusion restrictions used in the context of classical instrumental variables analysis. In our application for the U.K., we relate levels of expected wages to unobserved ability, measured ability, family background, type of secondary school, and the decision whether to attend college. While Chapters 2 and 3 focus on recovering identifiable features of a structural equation, Chapter 4 follows a different approach. It is concerned with feedback mechanisms in electronic markets that allow partners to rate each other after a transaction. These mechanisms are considered crucial for the success of anonymous internet trading platforms. Rather than estimating a system of structural equations we document an asymmetry in the feedback behavior on eBay, propose an explanation based on the micro structure of the feedback mechanism and the time when feedbacks are given, and support this explanation by findings from a large data set. Our analysis implies that the informational content of feedback records is likely to be low. We argue that the reason for this is that agents leave feedbacks strategically. Negative feedbacks are given late, in the "last minute", or not given at all, most likely because of the fear of retaliative negative feedback. Conversely, positive feedbacks are given early in order to encourage reciprocation. Towards refining our insights into the observed pattern, we look separately at buyers and sellers, and relate the magnitude of the effects to the trading partners' experience.