Many information structures generate correlated rather than mutually independent signals, the news media being a prime example. This paper shows experimentally that in such contexts many people neglect these correlations in the updating process and treat correlated information as independent. In consequence, people's beliefs are excessively sensitive to well-connected information sources, implying a pattern of "overshooting" beliefs. Additionally, in an experimental asset market, correlation neglect not only drives overoptimism and overpessimism at the individual level, but also affects aggregate outcomes in a systematic manner. In particular, the excessive confidence swings caused by correlated signals give rise to predictable price bubbles and crashes. These findings are reminiscent of popular narratives according to which aggregate booms and busts might be driven by the spread of "stories". Our results also lend direct support to recent models of boundedly rational social learning.
The conceptual framework of neoclassical economics posits that individual decision-making processes can be represented as maximization of some objective function. In this framework, people's goals and desires are expressed through the means of preferences over outcomes; in addition, in choosing according to these objectives, people employ subjective beliefs about the likelihood of unknown states of the world. For instance, in the subjective expected utility paradigm, people linearly combine their probabilistic beliefs and preferences over outcomes to form an expected utility function. Much of the parsimony and power of theoretical economic analysis stems from the striking generality and simplicity of this framework. At the same time, the crucial importance of preferences and beliefs in our conceptual apparatus in combination with the heterogeneity in choice behavior that is observed across many economic contexts raises a number of empirical questions. For example, how much heterogeneity do we observe in core preference or belief dimensions that are relevant for a broad range of economic behaviors? If such preferences and beliefs exhibit heterogeneity, then what are the origins of this heterogeneity? How do beliefs and preferences form to begin with? And how does variation in beliefs and preferences translate into economically important heterogeneity in choice behavior? This thesis is organized around these broad questions and hence seeks to contribute to the goal of providing an improved empirical understanding of the foundations and economic implications of individual decision-making processes. The content of this work reflects the deep belief that understanding and conceptualizing decision-making requires economists to embrace ideas from a broad range of fields. Accordingly, this thesis draws insights and techniques from the literatures on behavioral and experimental economics, cultural economics, household finance, comparative development, cognitive psychology, and anthropology. Chapters 1 through 3 combine methods from experimental economics, household finance, and cognitive psychology to investigate the effects of bounded rationality on the formation and explanatory power of subjective beliefs. Chapters 4 through 6 use tools from cultural economics, anthropology, and comparative development to study the cross-country variation in economic preferences as well as its origins and implications. The formation of beliefs about payoff-relevant states of the world crucially hinges on an adequate processing of incoming information. However, oftentimes, the information people receive is rather complex in nature. Chapters 1 and 2 investigate how boundedly rational people form beliefs when their information is subject to sampling biases, i.e., when the information pieces people receive are either not mutually independent or systematically selected. Chapter 1 is motivated by Akerlof and Shiller's popular narrative that from time to time some individuals or even entire markets undergo excessive belief swings, which refers to the idea that sometimes people are overly optimistic and sometimes overly pessimistic over, say, the future development of the stock market. In particular, Akerlof and Shiller argue that such "exuberance" or excessive pessimism might be driven by the pervasive "telling and re-telling of stories". In fact, many real information structures such as the news media generate correlated rather than mutually independent signals, and hence give rise to severe double-counting problems. However, clean evidence on how people form beliefs in correlated information environments is missing. Chapter 1, which is joint work with Florian Zimmermann, provides clean experimental evidence that many people neglect such double-counting problems in the updating process, so that beliefs are excessively sensitive to well-connected information sources and follow an overshooting pattern. In addition, in an experimental asset market, correlation neglect not only drives overoptimism and overpessimism at the individual level, but also gives rise to a predictable pattern of over- and underpricing. Finally, investigating the mechanisms underlying the strong heterogeneity in the presence of the bias, a series of treatment manipulations reveals that many people struggle with identifying double-counting problems in the first place, so that exogenous shifts in subjects' focus have large effects on beliefs. Chapter 2 takes as starting point the big public debate about increased political polarization in the United States, which refers to the fact that political beliefs tend to drift apart over time across social and political groups. Popular narratives by, e.g., Sunstein, Bishop, and Pariser posit that such polarization is driven by people selecting into environments in which they are predominantly exposed to information that confirms their prior beliefs. This pattern introduces a selection problem into the belief formation process, which may result in polarization if people failed to take the non-representativeness among their signals into account. However, again, we do not have meaningful evidence on how people actually form beliefs in such "homophilous" environments. Thus, Chapter 2 shows experimentally that many people do not take into account how their own prior decisions shape their informational environment, but rather largely base their views on their local information sample. In consequence, beliefs excessively depend on people's priors and tend to be too extreme, akin to the concerns about "echo chambers" driving irrational belief polarization across social groups. Strikingly, the distribution of individuals' naivete follows a pronounced bimodal structure - people either fully account for the selection problem or do not adjust for it at all. Allowing for interaction between these heterogeneous updating types induces little learning: neither the endogenous acquisition of advice nor exogenously induced dissent lead to a convergence of beliefs across types, suggesting that the belief heterogeneity induced by selected information may persist over time. Finally, the paper provides evidence that selection neglect is conceptually closely related to correlation neglect in that both cognitive biases appear to be driven by selective attentional patterns. Taken together, chapters 1 and 2 show that many people struggle with processing information that is subject to sampling issues. What is more, the chapters also show that these biases might share common cognitive foundations, hence providing hope for a unified attention-based theory of boundedly rational belief formation. While laboratory experimental techniques are a great tool to study the formation of beliefs, they cannot shed light on the relationship between beliefs and economically important choices. In essentially all economic models, beliefs mechanically map into choice behavior. However, it is not evident that people's beliefs play the same role in generating observed behavior across heterogeneous individuals: while some people's decision process might be well-approximated by the belief and preference-driven choice rules envisioned by economic models, other people might use, e.g., simple rules of thumb instead, implying that their beliefs should be largely irrelevant for their choices. That is, bounded rationality might not only affect the formation of beliefs, but also the mapping from beliefs to choices. In Chapter 3, Tilman Drerup, Hans-Martin von Gaudecker, and I take up this conjecture in the context of measurement error problems in household finance: while subjective expectations are important primitives in models of portfolio choice, their direct measurement often yields imprecise and inconsistent measures, which is typically treated as a pure measurement error problem. In contrast to this perspective, we argue that individual-level variation in the precision of subjective expectations measures can actually be productively exploited to gain insights into whether economic models of portfolio choice provide an adequate representation of individual decision processes. Using a novel dataset on experimentally measured subjective stock market expectations and real stock market decisions collected from a large probability sample of the Dutch population, we estimate a semiparametric double index model to explore this conjecture. Our results show that investment decisions exhibit little variation in economic model primitives when individuals provide error-ridden belief statements. In contrast, they predict strong variation in investment decisions for individuals who report precise expectation measures. These findings indicate that the degree of precision in expectations data provides useful information to uncover heterogeneity in choice behavior, and that boundedly rational beliefs need not necessarily map into irrational choices. In the standard neoclassical framework, people's beliefs only serve the purpose of achieving a given set of goals. In many applications of economic interest, these goals are well-characterized by a small set of preferences, i.e., risk aversion, patience, and social preferences. Prior research has shown that these preferences vary systematically in the population, and that they are broadly predictive of those behaviors economic theory supposes them to. At the same time, this empirical evidence stems from often fairly special samples in a given country, hence precluding an analysis of how general the variation and predictive power in preferences is across cultural, economic, and institutional backgrounds. In addition, it is conceivable that preferences vary not just at an individual level, but also across entire populations - if so, what are the deep historical or cultural origins of this variation, and what are its (aggregate) economic implications? Chapters 4 through 6 take up these questions by presenting and analyzing the Global Preference Survey (GPS), a novel globally representative dataset on risk and time preferences, positive and negative reciprocity, altruism, and trust for 80,000 individuals, drawn as representative samples from 76 countries around the world, representing 90 percent of both the world's population and global income. In joint work with Armin Falk, Anke Becker, Thomas Dohmen, David Huffman, and Uwe Sunde, Chapter 4 presents the GPS data and shows that the global distribution of preferences exhibits substantial variation across countries, which is partly systematic: certain preferences appear in combination, and follow distinct economic, institutional, and geographic patterns. The heterogeneity in preferences across individuals is even more pronounced and varies systematically with age, gender, and cognitive ability. Around the world, the preference measures are predictive of a wide range of individual-level behaviors including savings and schooling decisions, labor market and health choices, prosocial behaviors, and family structure. We also shed light on the cultural origins of preference variation around the globe using data on language structure. The magnitude of the cross-country variation in preferences is striking and raises the immediate question of what brought it about. Chapter 5 presents joint work with Anke Becker and Armin Falk in which we use the GPS to show that the migratory movements of our early ancestors thousands of years ago have left a footprint in the contemporary cross-country distributions of preferences over risk and social interactions. Across a wide range of regression specifications, differences in preferences between populations are significantly increasing in the length of time elapsed since the respective groups shared common ancestors. This result obtains for risk aversion, altruism, positive reciprocity, and trust, and holds for various proxies for the structure and timing of historical population breakups, including genetic and linguistic data or predicted measures of migratory distance. In addition, country-level preference endowments are non-linearly associated with migratory distance from East Africa, i.e., genetic diversity. In combination with the relationships between language structure and preferences established in Chapter 4, these results point to the importance of very long-run events for understanding the global distribution of some of the key economic traits. Given these findings on the very deep roots of the cross-country variation in preferences, an interesting - and conceptually different - question is whether such country-level preference profiles might have systematic aggregate economic implications. Indeed, according to standard dynamic choice theories, patience is a key driving factor behind the accumulation of productive resources and hence ultimately of income not just at an individual, but also at a macroeconomic level. Using the GPS data on patience, Chapter 6 (joint work with Thomas Dohmen, Armin Falk, David Huffman, and Uwe Sunde) investigates the empirical relevance of this hypothesis in the context of a micro-founded development framework. Around the world, patient people invest more into human and physical capital and have higher incomes. At the macroeconomic level, we establish a significant reduced-form relationship between patience and contemporary income as well as medium- and long-run growth rates, with patience explaining a substantial fraction of development differences across countries and subnational regions. In line with a conceptual framework in which patience drives income through the accumulation of productive resources, average patience also strongly correlates with aggregate human and physical capital accumulation as well as investments into productivity. Taken together, this thesis has a number of unifying themes and insights. First, consistent with the vast heterogeneity in observed choices, people exhibit a large amount of variation in beliefs and preferences, and in how they combine these into choice rules. Second, at least part of this heterogeneity is systematic and has identifyable sources: preferences over risk, time, and social interactions appear to have very deep historical or cultural origins, but also systematically vary with individual characteristics; belief heterogeneity, on the other hand, is partly driven by bounded rationality and its systematic, predictable effects on information-processing. Third, and finally, this heterogeneity in beliefs and preferences is likely to have real economic implications: across cultural and institutional backgrounds, preferences correlate with the types of behaviors that economic models envision them to, not just across individuals, but also at the macroeconomic level; subjective beliefs are predictive of behavior, too, albeit with the twist that certain subgroups of the population do not appear to entertain stable belief distributions to begin with. In sum, (I believe that) much insight is to be gained from further exploring these fascinating topics.
