Hide Notes 

Slide 1: Research Statement (July 2007) Ulrike Malmendier, University of California, Berkeley, Department of Economics I. Overview My research explores the implications of non-standard (behavioral) assumptions in two contexts: the interaction of firms and investors (Behavioral Corporate Finance) and the interaction of firms and consumers (Behavioral Industrial Organization). My goal is to identify whether and under what conditions behavioral biases affect market outcomes, such as prices, contracts, or market participation. I use theoretical and empirical methods, including laboratory and field data to address these questions. A key question of my research is whether or not deviations from traditional preferences and beliefs matter in the presence of competition, arbitrage, high-stake incentives and learning. In addressing this question my research provides a strong test of the significance of behavioral approaches for real-world settings in different markets. It shows where we should expect biases to matter and where we should not. Most importantly, it offers organizing explanations for puzzles in Corporate Finance and Applied Microeconomics. My Behavioral Corporate Finance research, the first main area of focus, has demonstrated how managerial overconfidence helps to explain multiple puzzles, including negative acquirer returns in mergers, investment-cash flow sensitivity, and pecking-order financing. This research has added a new dimension to previous work incorporating behavioral approaches in Corporate Finance: Rather than focusing on investor biases, I investigate managerial biases and their impact on corporate decisions. A recent survey article on Behavioral Corporate Finance by Baker, Ruback, and Wurgler (2006) provides an overview of the two perspectives on the field and emphasizes the role of my research in developing the second angle. (Similarly, see the earlier “Survey on Behavioral Finance” by Barberis and Thaler, 2003.) I have also pursued research projects that fall into the field of Empirical Corporate Finance more broadly, including work on the empirical relevance of (Incomplete) Contract Theory for real-world contract design, the impact of media attention on CEO compensation and corporate decision-making, as well as some work related to Law & Finance. In my second main area of research, Behavioral Industrial Organization, I again provided organizing explanations for empirical regularities and puzzles. Examples are the demand for flat-rate contracts among lowfrequency users, overbidding in auctions, and the trade reaction of investors to overoptimistic stock recommendations. This research emphasizes the market response of firms to consumer biases and has pioneered the literature on Behavioral Industrial Organization. This stream of research was judged to be of sufficient importance that it is summarized in the survey article of Ellison (2006) and discussed in later papers in the field, such as Gabaix and Laibson (2006). My research has attracted substantial interest in the academic community, as revealed by its inclusion in top conferences such as the AEA, AFA and WFA annual meetings and the NBER conferences in Corporate Finance, Behavioral Finance, Industrial Organization, Labor Studies, Organizational Economics, and Personnel Economics. I have been appointed as Associate Editor at the Economic Journal and the Journal of Financial Intermediation and to numerous program committees. Additionally, I have been invited to share my research by teaching at many summer schools in the US and in Europe. (Specifics are in the attached CV). I am excited to foster this interest and to contribute with my own research as well as the active mentoring role I have assumed at Berkeley and previously at Stanford, both at the undergraduate and on the graduate level. In the remainder of this Research Statement, I will first summarize my research in Behavioral Corporate Finance (Section II), then additional research in Corporate Finance (Section III), and finally my research in Behavioral Economics, especially Behavioral Industrial Organization (Section IV). Section V concludes. II. Behavioral Corporate Finance Behavioral considerations are relevant to corporate decision-making from two perspectives. First, agents outside the firm (investors, consumers) may make mistakes which managers can exploit. An example is the timing of equity issuances when investors are likely to overvalue the firms (Baker and Wurgler, 2000). Second, managers display biases that induce non-optimal corporate decisions. Previous literature in Corporate Finance has focused mostly on the first perspective. This is surprising since managerial biases–if they exist–are likely to be persistent and to have a large impact on corporate outcomes for two reasons: they cannot easily 1
Slide 2: be arbitraged away, and top executives may be entrenched and hard to get rid of (Bebchuck, Fried, and Walker, 2002). My research in Behavioral Corporate Finance develops this second perspective. I focus on specific biases that are ex-ante likely to affect high-level managers. An overview over plausible managerial distortions and their applications is included my survey paper on the “Behavioral Economics of Organizations” (in: P. Diamond and H. Vartiainen (eds.), Behavioral Economics and Its Applications, Princeton University Press, 2007), joint with Colin Camerer. In a first set of papers, I examine managerial overconfidence. Overconfident managers overestimate the value they can create in their firm and perceive too many internal projects and mergers as worth undertaking. Roll (1986) first formalized the notion of managerial overconfidence, linking over-bidding in a takeover contest to the winner’s curse. However, the implications of overconfidence are more subtle than the simple winner’s curse view of “more investment” and “worse investment” once we account for market interaction. In the context of a market, rational financiers curb executives’ ability to over-invest by not providing financing for overinvestment, at least not at the price the overconfident CEO expects (see Heaton, 2002). Hence, overconfident CEOs may abstain from projects if internal cash flow is not available. This suggests that the investment decisions of overconfident CEOs are sensitive to cash flow and that the resulting investment-cash flow sensitivity is strongest in firms with few internal resources. These hypotheses are developed and tested in the paper “CEO Overconfidence and Corporate Investment” (Journal of Finance, 2005, vol. 60 (6), pp. 2661-2700), joint with Geoff Tate. The biggest empirical challenge in this research is the construction of a plausible measure of overconfidence. Since biased beliefs naturally defy direct and precise measurement, we use a “revealed beliefs” argument. We classify CEOs as overconfident if they personally over-invest in their companies, i.e. buy excessive amounts of stock and hold executive options until expiration, even though a calibrated model of option exercise suggests that they should diversify (Lambert, Larcker, and Verrecchia, 1991; Hall and Murphy, 2002) and even though the options are highly “in the money,” i.e. the current stock price exceeds the exercise price. We show that these CEOs typically earn below-market returns on their stock and option holdings, which allows us to rule out insider trading or rational empire building as explanations. Holding on to options is also hard to explain with a signaling story (i.e., option-holding CEOs signaling high expected returns of their projects), given that their stock performs worse than the market and the industry and also given the persistence of their holding behavior. After addressing a number of additional interpretations (taxes, risk tolerance, board pressure, procrastination), we conclude that excessive stock purchases and option holding likely indicate overestimation of future returns. We then show that the investment decisions of overconfident CEOs are excessively sensitive to the availability of internal funds. Overconfidence emerges as a novel explanation for the investment-cash flow puzzle (see, e.g., Stein, 2003). Moreover, as predicted by the theory, the relation of overconfidence and investment-cash flow sensitivity is largest in financially constrained firms. Our results show a robust relationship between CEOs’ personal overestimation of their firms’ profitability and their corporate investment decision. Our interpretation of the result incorporates the market interaction of biased managers and rational investors via the financing channel: Overconfident CEOs believe that outside investors under-estimate the profitability of their projects and, hence, demand too high an interest rate or pay too low a price for stock issues. In a recent working paper, “Corporate Financial Policies with Overconfident Managers,” joint with Geoff Tate and Jun Yan, we collect additional data in order to test this channel directly and analyze whether overconfident CEOs are more reluctant to tap external capital markets. We provide a theoretical analysis of the financing choices of rational and overconfident managers. We show that overconfidence, embedded in a market with rational investors, can explain long-standing capital structure puzzles such as the “pecking order of financing” (preference for cash over debt and debt over equity) and firms’ reluctance to access external capital markets, including the “low-leverage puzzle” (if the preference for cash over any external finance is strong enough.) We combine our overconfidence data on managers’ personal portfolio from the first paper with data on security issuances, and find a significant relationship between our overconfidence proxies and the financing puzzles. A third paper, “Who Makes Acquisitions? CEO Overconfidence and the Market’s Reaction” (Journal of Financial Economics, forthcoming), joint with Geoff Tate, provides the strongest and cleanest evidence on the empirical importance of managerial overconfidence. In this paper, we relate overconfidence to mergers and acquisitions. As pointed out above, overconfidence does not necessarily predict excessive mergers when embedded into a market setting with rational financiers. It does so only when the (overestimated) benefits of 2
Slide 3: a merger exceed the (also overestimated) costs of raising external financing. Hence, overconfidence induces more mergers only in cash-rich firms. However, if we do observe that overconfident CEOs undertake net more mergers, we can derive the additional prediction that those mergers have, on average, lower returns on average than mergers undertaken by non-overconfident CEOs. In this paper we directly measure market reactions (announcement effects) and provide additional measures of CEO overconfidence. In the empirical analysis, we find that overconfident CEOs do, in fact, undertake more mergers and worse mergers than other CEOs, in particular in cash-rich firm years. The average announcement effect is significantly more negative for mergers of overconfident CEOs (–90 bps) than for those of their nonoverconfident peers (–12 bps). The empirical analysis in this paper represents an advance over the first (JF) paper for several reasons. First, we refine the identification by allowing time variation in overconfidence. Rather than classifying CEOs as overconfident “once and for all,” we allow within-person variation, capturing the CEO’s personal overinvestment over time. Second, we construct an alternative, media-based proxy for overconfidence, which captures how outsiders perceive the CEO. We hand-collect data on CEO coverage in the business press, either as “confident” and “optimistic” or as “reliable,” “cautious,” “conservative,” “practical,” “frugal,” and “steady.” Characterization as confident or optimistic is significantly positively correlated with our portfolio measures of optimistic beliefs. And, all of our main results replicate using a press-based measure of overconfidence. Overall, we find that managerial overconfidence provides a unifying framework for some of the major empirical puzzles in Corporate Finance. Our findings do not imply that traditional explanations such as misaligned incentives or asymmetric information are not valid. Overconfidence is an additional explanation, applicable to the subset of overconfident CEOs. But its broad explanatory power and its large estimated effects on investment, financing, and mergers indicate that of significant empirical relevance. Moreover, the overconfidence explanation has important governance implications: Overconfidence cannot be curbed by providing incentives in the form of stock and option grants. This stream of research has been of sufficient importance to stimulate the demand for separate session at the main finance conferences, including the AFA, or even actual conference themes (NBER University Research Conference on Behavioral Corporate Finance, or the Seventh Maryland Finance Symposium proposing “Does CEO overconfidence affect investment decisions, including mergers?” as one of the main themes). It has also generated a large volume of follow-up work, as indicated by the large number of citations (199 for the two published papers according to Google scholar). III. Other Research in Corporate Finance Although much of my research is behavioral I am also involved in several ‘traditional’ papers and projects in Corporate Finance. Most closely related to the last-mentioned JFE paper is a project on “Winning by Losing: Evidence on Overbidding in Mergers” with Enrico Moretti. We propose a new approach to tackle the main obstacle in the evaluation of mergers, the lack of a return benchmark. Abnormal long-term returns are hard to identify and may be biased due to market timing. As pointed out by Shleifer and Vishny (2003), CEOs may undertake stock-financed mergers when their firms are overvalued. Despite negative short- and long-term returns, such mergers create value and are in the best long-run interest of current shareholders. As a partial solution to this identification issue, we analyze bidding contests. The basic identifying assumption is that the performance of losing bidder(s) provides a valid counterfactual for future stock returns after adjusting for pre-existing trends and the usual firm- and industry-level controls. In fact, traditional methodologies such as matching by book-to-market ratios, industry, or size can be refined by focusing on winners and losers in contested mergers. The first step is to compare the returns of winners and losers prior to the merger. The identifying assumption requires returns not to be significantly different pre-merger. The second step is to compare returns over different horizons after the merger in order to evaluate the merger. In a preliminary analysis, we included all mergers from 1983-2004 involving U.S. targets that the SDC database flags as contested. We find that, prior to the mergers, the normalized prices of winners and losers co-move closely and are not significantly different. This suggests that loser returns are a useful benchmark for the returns of winners. After the merger, the price series diverge, and winners significantly underperform losers. This finding suggests that, in the subset of contested mergers, mergers destroy value for the acquirers’ shareholders on average. We also find that our methodology is most applicable to stock-financed mergers. When repeating the analysis separately for stock-financed and cash-financed mergers, we find that the win3
Slide 4: ner-loser return difference pre-merger is insignificant only for stock-financed mergers but significantly positive for cash-financed mergers. (It is significantly negative post-merger in both subsamples.) We are currently finishing a large data collection, correcting and enlarging the SDC data, which we found to be incomplete and sometimes imprecise about controls such as the merger financing and cash controls. We also hope to further address the leading alternative interpretation of our results, i.e., that, despite the closely matching return paths pre-merger, winners are systematically more overvalued than losers prior to the merger and hence mean-revert more post-merger. One additional test involves the comparison between several losers of a given merger fight. Are they insignificantly different post merger, as our identifying assumption would suggest? If not, their difference in performance may provide an estimate of the amount of bias. Another test involves changes in return variance pre- and post-merger, separately for winners and losers. Another project studies less permanent associations between firms, so-called research alliances. In the paper “Contractibility and the Design of Research Agreements” (revised and resubmitted to the American Economic Review), joint with Josh Lerner, we analyze how contracts respond to varying degrees of “contractibility.” Understanding the determinants and limits of contract design is central to numerous areas in economics, ranging from labor economics and corporate finance to modern macroeconomics. An important distinction, introduced by the literature on incomplete contracts, is the observability and verifiability of actions and outputs (cf. Hart, 1995). If key variables are not verifiable in front of judges, the contracting parties have to find alternative mechanisms to induce the expected behavior, such as reallocating asset ownership. However, much of the prior literature on “real-world contract design” has focused on complete rather than incomplete contracts, maybe because of empirical difficulties in pinning down concepts such as observability, verifiability and incompleteness. (Two leading exceptions are the work by Kaplan and Stromberg, 2003 and 2004, and by Baker and Hubbard, 2003 and 2004.) We analyze, both theoretically and empirically, how the design of contracts varies as underlying variables become harder to verify. Our empirical application is biotechnology research. Biotechnology research agreements have to deal with the non-verifiability of research efforts. We develop a model that allows for researchers to pursue multiple projects. When research activities are contractible, a complete contract achieves the first best. When research activities are not contractible, an option contract is optimal, which assigns the financing firm the right to terminate the agreement and, in case of termination, broadly defined property rights to the project. The threat of termination induces researchers to adhere to the proposed research agenda, while the cost of exercising the option deters the financing firm from opportunistic termination. We test this prediction using a new data set of 580 biotechnology research agreements. Our large, handcollected data set on research agreements allows us to address several concerns plaguing that literature, such as unobserved firm characteristics (via firm fixed-effects and firm-level controls), and to test directly competing explanations. We find that the option contract is more common when research is non-contractible. We also analyze how the contractual design varies with the research firm’s financial constraints and quality. The additional empirical results allow us to distinguish the property-rights explanation from explanations based on uncertainty and asymmetric information about the project quality or research abilities. A third project, “Financial Expertise of Directors” (Journal of Financial Economics, forthcoming), joint with Burak Güner and Geoff Tate, analyzes the impact of corporate directors on corporate decisions. Much of the corporate-governance debate revolves around the composition of boards. Following the recent wave of accounting scandals, regulators have urged for more ‘financial experts’ on boards to ensure more accurate disclosure and better audit committee performance. However, directors spend a significant portion of their time on advising rather than monitoring. This influence can be problematic if directors are affiliated with financial institutions and pursue the interests of those institutions rather than maximizing shareholder value. In this paper, we test whether directors with financial expertise influence corporate policies and whether affiliation hampers their advisory role. A key concern for any such analysis of the impact of directors is the endogeneity of board composition, a point made both theoretically and empirically by Hermalin and Weisbach (1998 and 1988). We construct a novel panel data set on the board composition of 282 companies over 14 years, which helps to better address these concerns. The fourteen-year time series provides sufficient variation to identify commercial banker effects after controlling for company fixed effects. Thus, the estimated effect does not reflect time-invariant firm characteristics. Moreover, we are able to instrument for the board presence of commercial bankers, using pre-sample shocks to the supply of banker directors due to the banking crisis in the late 1970s and early 1980s. 4
Slide 5: We find that financial experts significantly affect corporate decisions, though not necessarily in the interest of shareholders. First, when commercial bankers join boards, external funding increases and investmentcash flow sensitivity diminishes. But, the increased lending affects mostly firms with good credit and poor investment opportunities, i.e., firms that are able to repay loans but do not have value-creating investment projects. Second, investment bankers on the board are associated with larger security issuance but also worse acquisitions. Both activities generate fees for the investment banks but appear to decrease (or not to increase) shareholder value. Third, whenever the interests of the financial institutions are orthogonal to a corporate decision (e.g., compensation) or the financial expert is unaffiliated (e.g., finance professors) we find little evidence of any influence at all. The results suggest a tradeoff between outside incentives (e.g. bank profits) and the incentive to maximize firm value in the selection of directors with financial expertise. The topic of corporate governance and CEO compensation motivates another project on “Superstar CEOs”, joint with Geoff Tate. We explore the ramifications of a “CEO celebrity culture” for managers and shareholders. We show that media stars subsequently underperform, while extracting significantly higher compensation and spending more time on outside activities (public speeches, writing memoirs, board meetings in other companies, golf course). The empirical basis is a unique, hand-collected data set on CEOs who win high-profile awards from the business press or other prominent organizations between 1975 and 2002.1 The key issue is to identify the right counterfactual. A naive comparison of award-winning and non-awardwinning CEOs would suffer from unobserved differences in CEO ability or capture mean reversion. We employ a two-stage matching procedure and introduce the nearest-neighbor matching estimators (and standard error corrections) of Abadie and Imbens (2002) into the corporate finance setting. The estimation procedure identifies CEOs who were likely to win the award at a specific point in time, based on observables, but did not. Those observables include stock prices, which should aggregate all available firm information. We find that actual award winners significantly underperform the matched sample of not-award-winning CEOs (with close propensity scores), by 12-20% over three years beyond mean reversion. At the same time, the average compensation of superstars increases from about $13m to over $18m, far more than that of “hypothetical” winners. These results are driven by badly-governed firms, e.g. firms without an institutional blockholder. Our findings offer a new perspective on the role of tournaments: Tournament winners, and hence ‘superstars’ in the diction of Rosen (1981), underperform subsequently, as predicted by lower ex-post incentives. The types of activities they pursue (book writing, public speeches etc.), however, also lends itself to alternative interpretations such as a role for ego utility (Kőszegi, 2006). At the same time, good governance can prevent such distractions, without lowering the performance of CEOS, as far as we can infer from the awards data. A final area of my research in Corporate Finance relates to the area of Law and Finance. I have worked on the historical origins of the corporation and its interplay with legal and political development. My research agenda is based on my book on the “Societas publicanorum” (Böhlau Verlag, Cologne/Vienna, 2002, in German) A revised and extended English edition is forthcoming at Yale University Press. This book analyzes the often overlooked origins of the “corporation” during the times of the Roman Republic. I provide a historical overview of its rise in the last two centuries B.C. and its gradual demise under the Roman emperors. The decentralized Roman government with its short electoral cycle and lack of any permanent bureaucracy outsourced practically all economic activity, including the construction of large public buildings and streets or the tax collection in the provinces. to the companies of private entrepreneurs grew with the scope of these tasks in the geographically vastly expanding Roman Republic. The Roman law proved to be flexible enough to accommodate the necessary legal requirements for the growing private companies, such as “shares” (partes) to attract external financing or stability beyond the death or departure of “partners”, akin to the modern corporation with its own legal persona. In a related article, “Roman Shares” (In: W. Goetzmann and G. Rouwenhorst (eds.), The Origins of Value. The Financial Innovations that Created Modern Capital Markets. Oxford University Press, August 2005, pp. 31-42, 361-365), I lay out the corporate features of the ancient societas publicanorum in economic terms. In a recent paper, “Law and Finance at the Origin” (revised and resubmitted [third-round] to the Journal of Economic Literature), I relate the historical evidence to the current literature in Law and Finance. Following the seminal papers by La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1997 and 1998), the Law and Finance literature has argued that a country’s legal sysThe sources are Business Week, Financial World Magazine, Chief Executive Magazine, Forbes, Industry Week Magazine, Morningstar.com, Time, Time/CNN, Electronic Business Magazine, and Ernst & Young. 5 1
Slide 6: tem affects financial development and economic growth. Another strand of literature emphasizes the role of political institutions and points out that the effectiveness of legal rules varies considerably with the political support they receive (Rajan and Zingales, 2003; Acemoglu and Johnson, 2005). While those two approaches do not necessarily contradict each other, the relative roles of legal versus political and other institutional determinants are hard to evaluate. Given the scarcity of perfect natural experiments, careful and detailed analyses of individual cases are a valuable part of the literature, even if they stop short of proving causality. Thus, much of the literature revolves around specific examples, mostly referring to the last two centuries. We use the Roman case to expand the current body of evidence to a much earlier period, ancient Rome. The history of the Roman shareholder company adds to the discussion about the nexus between law, finance, and growth since it addresses a basic identification problem: The evolution of law and politics often evolved in the same direction—either fostering or prohibiting financial development. That makes it difficult to attribute financial development to either source (or their interaction). The societas publicanorum provides a rare case where the evolution of law and politics diverged and it is possible to evaluate the relative role of both determinants. IV. Behavioral Economics and Firm Response The last part of my research agenda pursues the question of when and how biases matter in the context of firm-consumer interaction. I ask which types of biases transfer from laboratory evidence into real-world consumer behavior, and I explore firms’ optimal responses to such biases. Whether biases trigger a reaction by other market participants is, in fact, a key test for the relevance of the behavioral deviations. A first building block of this agenda are two papers, joint with Stefano DellaVigna. The first, “Paying Not to Go to the Gym” (American Economic Review, 2006, vol. 96 (3), pp. 694-719), documents large discrepancies between consumers’ choices of contract and their subsequent consumption. We obtained a data set with information on both the membership choice from a given menu of contracts and the day-to-day attendance of 7,978 health-club members over three years. We find that consumers who choose a monthly membership of over $70 per month pay on average 70% more than they would under the pay-as-you-go contract for the same number of visits. Given their behavior, eighty percent of the monthly members would have been better off had they paid per visit for the same number of visits. Second, we found that consumers who chose a monthly contract were 18 percent more likely to stay enrolled beyond one year than users who chose an annual contract. This is particularly surprising, as monthly members pay higher fees for the option to cancel each month. Our results are hard to reconcile with the standard assumption in the Industrial Organization literature that consumers have rational expectations about their future consumption frequency and choose the utility-maximizing contract. Hence, drawing inferences from observed contract choice under the rational expectation hypothesis can bias the estimation of consumer preferences. And, in turn, the contract design of firms is better understood if we account for non-standard consumer behavior. A leading explanation of our results is that consumers have self-control problems and are naïve about their future self-control. In our second paper, “Contract Design and Self-Control: Theory and Evidence,” published as the lead article in the Quarterly Journal of Economics (2004, vol. 119 (2), pp. 353-402), we analyze the optimal firm response to this type of consumer preferences. If certain goods or services tempt consumers to consume more than they would like to (e.g. cell phone or credit card) and if other goods tend to be consumed less than planned (e.g. health clubs), what fee structure maximizes the profits of the producing firms? We derive a two part-tariff with a zero or negative flat fee but above marginal-cost prices per usage in the case of over-consumption, and the reverse structure in the case of under-consumption (high flat fee and below-marginal cost prices per consumption). The optimality of both results persist even under competition. A third finding is that firms offer a back-loaded fee structure and design contracts with automatic renewal and endogenous transaction costs of switching. This is profit-maximizing since naive individuals underestimate the probability of renewal. The predictions of our theory match the observed contract design in numerous industries, including the credit card, gambling, health club, life insurance, mail order, mobile phone, and vacation time-sharing industries. We also show that time inconsistency has adverse effects on consumer welfare only if consumers are naive. To my knowledge, this paper is the first published paper addressing the core question of Behavioral Industrial Organization: How do firms respond to consumer biases? A second set of papers pursues a similar research question applied to firms providing investment advice. In two papers, joint with Devin Shanthikumar, we explore the market relevance of a different bias: naiveté about incentives. Laboratory evidence shows that individuals (buyers) do not sufficiently account for other 6
Slide 7: individuals’ (sellers’) incentives to distort information about the object to be sold, even if they are explicitly informed about the incentive problem (see, e.g., Cain, Loewenstein, and Moore’s (2005) jar-of-coins experiment). In our first paper, “Are Small Investors Naïve about Incentives?” (Journal of Financial Economics, August 2007, vol. 85 (2), pp. 457-489), we examine how real-world investors react to analyst recommendations. We find that large (institutional) investors rationally discount recommendations while small (individual) investors take them literally. For example, only 4.5% of all recommendations issued between the end of 1993 and the end of 2002 were negative. Large investors account for the upward distortion and sell after “hold” recommendations, hold after “buy” recommendations, and buy only after “strong buys.” Small investors take recommendations at face value. They hold after “hold” recommendations and buy both after “buy” and “strong buy” recommendations. These findings provide an organizing explanation for the existence of “affiliated” analysts and persistence of upward distortion in recommendations. In our second paper, “Do Security Analysts Speak in Two Tongues?” (NBER Working Paper 13124, July 2007), we ask whether the overly positive recommendations are a conscious distortion, as the Behavioral IO perspective suggests, or whether analysts are truly overoptimistic about the stocks they cover. True overestimation is not implausible since analysts have some say in the selection of stocks they cover and may pick stocks for which they see positive prospects. In order to distinguish rational reaction of analysts to investor biases – in particular among analysts affiliated with an underwriting investment bank – from sorting of truly overoptimistic analysts, we exploit the concurrent issuance of different types of investment advice by the same analyst and the heterogeneity in investor sophistication between individual and institutional investors. First, we show that recommendations and earnings forecasts reach different audiences. Individual investors respond strongly to simple buy- and sell-recommendations, while the more subtle content of earnings forecasts affects the trade direction only of large, typically institutional investors. As a result, analysts may choose to distort recommendations but signal the quality of their analyses to large investors by giving unbiased forecasts. If analysts are, instead, truly too optimistic recommendation and forecast overoptimism should be positively correlated. We find that, while affiliated analysts issue more optimistic recommendations than unaffiliated analysts, their earnings forecasts are more pessimistic. Moreover, forecast optimism is positively correlated with recommendation optimism for unaffiliated analysts but negatively for affiliated analysts. The positive correlation suggests that unaffiliated analysts might be truly overoptimistic and express this optimism in all types of communication. The negative correlation, instead, is consistent with distortion, once we account for the additional incentives to bias forecasts downwards when approaching earnings announcements (allowing management to meet or beat the forecast). Additional discrepancies between the timing of recommendations and forecasts confirm that active distortion in response to small investors’ naiveté is a major explanation for the recommendation optimism of affiliated analysts. In an ongoing project, entitled “The Bidder’s Curse” (joint with Hanh Lee), we ask whether a widespread market mechanism, auctions, may itself reflect the impact of consumer biases. Traditionally, economists have explained the century-old popularity of auctions with efficiency and revenue maximization. Auctions help to identify those buyers who have the highest valuations of the object to be sold. However, if some buyers tend to overbid in auctions – due to utility from bidding, bidding fever, or limited attention – then auctions also help sellers to identify those buyers. Even if only few bidders fail to limit their bids to their valuation (or the alternative outside price of the object), overbidding will affect prices and allocations since auctions systematically pick as winners those whose willingness to bid is most excessive. The role of overbidding, however, is very hard to test empirically since we typically do not have an objective measure of buyers’ private valuations. We use a novel research design that detects overbidding in eBay auctions. We compare auction prices to fixed prices at which the same goods are simultaneously available for immediate purchase on the same webpage. Our primary data set contains all auctions of a popular board game from February to September 2004, which was also continuously available for sale at a fixed price on the same website. We find that, in 43% of the auctions, the final price is higher than the fixed price at which the same good is simultaneously listed for immediate purchase. The result is not explained by shipping cost, quality differences, reputational differences, or lack of information about the fixed-price option. Moreover, the result replicates in a large cross-section of 1,926 auctions with simultaneous fixed prices. We also address the question of the pervasiveness of this behavior across bidders. Only 17% of the 807 participants bid above the fixed price, but those bidders suffice to generate large-scale overbidding. All other bidders drop out and may buy at the lower fixed price. Hence, while the majority of bidders may purchase at 7
Slide 8: prices below the rational upper bound (including purchases out the auction), auctions appear to help sellers to “fish for fools,” i.e. for those buyers who tend to overbid. We provide a model of different potential explanations for overbidding, including transaction costs of switching from auctions to fixed-price transactions, utility from winning an auction, bidding fever, and limited memory of the fixed price. A rational transaction-cost model cannot easily explain our results given that even the average auction price is higher than the simultaneous fixed price (significantly after accounting for shipping costs). Each of the other (non-standard) preferences is a plausible candidate. We provide a careful calibration of distributions of valuations and preference parameters to further illustrate that few over-bidders suffice to generate a large fraction of overbid auctions. Finally, we also find that experience, as measured by eBay's feedback score, does not eliminate overbidding. If anything, bidders with high feedback scores make more mistakes. This result confirms that overbidding does not reflect search costs of eBay novices or other standard transaction costs and that biases do not necessarily disappear in a market environment with repeated interaction. While our identification strategy and results rely on online auctions, our argument applies to any auction that is suspected to showcase overbidding, from wine, antiques, and car auctions to free agents in baseball (Blecherman and Camerer, 1996), drafts in football (Massey and Thaler, 2006), or real estate auctions. (Ashenfelter and Genesove, 1992). Another set of projects is investigates the question of how experimentally-observed biases transfer to behavior in the field in the context of social preferences. Experiments allow insights into behavior that cannot be studied easily outside the laboratory. However, by design, most experiments try to select a random sample of the population. The participants are locked into the experimental environment and the specific game presented to them. In non-laboratory environments, instead, individuals sort, based on preferences, beliefs, and skills. Those individuals who choose to participate in markets are not a random sample of the population. In a first paper, joint with Eddie Lazear and Roberto Weber, “Sorting in Experiments with Application to Social Preferences” (NBER Working Paper 12041) we analyze the effect of sorting in a series of different experiments involving sharing. We incorporate the type of sorting that one observes in the field (where individuals can choose not to participate in a market) into the laboratory. In addition, we argue that the introduction of sorting may reveal a different interpretation of the preferences underlying the observed sharing behavior. For example, agents who share in a dictator game but who opt out of the game when possible may not share because of pure altruism or fairness, as one could have inferred from the simple dictator game. We first analyze a modified dictator game. The standard result in dictator games is that a significant fraction of subjects give a positive amount to an anonymous receiver, even when their action is not observable by anyone, including the experimenter (see Camerer, 2003; Hoffman et al., 1994). We modify the game and allow subjects to sort between environments that do and do not allow sharing. We find that the sorting opportunity significantly reduces the probability of sharing. When subjects are locked into the sharing environment and forced to choose how much to share, 61% share a positive amount. But when subjects are given the choice to avoid the situation altogether, only 23% share. Second, we observe a negative correlation between the amount shared and the inclination to participate in the sharing environment in the subset of those individuals who share a positive amount when sorting is not possible but who chose to opt out and hence do not share when sorting is possible. When we offer monetary incentives to “return” to the dictator game (but allow sorting), those who share the most whenever they do participate have to be compensated the most to choose participation. The latter results suggest that large sharing may not be exclusively due to altruism or fairness, though it is consistent with the behavior of 20-30% of our subjects. The remaining 40% who share in the dictator game appear to dislike not sharing, i.e. refusing to share when asked, but prefer to avoid being asked to begin with. The stronger the dislike is (or, the impulse to share out of shame or guilt), the higher is the willingness to pay to avoid participation. In the above set of experiments, we did not find significant differences in the amount shared conditional on being in the environment, i.e., between everybody in the standard dictator game and only those who choose to participate in the treatment with sorting. The latter result camouflages, however, that the environment with sorting attracts two ‘extreme’ types of individuals: those who like to share (e.g., due to altruism or fairness) and those who always keep all surplus to themselves, whether or not sharing is possible. In a new set of experiments, we are currently testing whether it is possible to attract ‘high sharers’ by making it costly to enter the dictator game, relative to the outside option. In this game, we increase the outside option while holding constant the surplus in the dictator game. This treatment might create an environment with high shar8
Slide 9: ing among those who opt in, even if the percentage of those opting in is low. Such a result would mirror, for example, the environment of non-profits: few individuals may choose to participate and to forgo higher earnings in alternative jobs, but those who do might be those who are intrinsically willing to share. We are also extending the analysis to different types of experiments involving sharing to evaluate the importance of other motives for sharing, in particular reciprocity, and their interaction with sharing. The main treatment involves a double-dictator game between the same anonymously matched pair of dictator and receiver, but switching their roles in the second game. This experiment explores whether negative reciprocity towards a non-sharer and positive reciprocity towards a sharer are strong motive and robust to sorting. Such a finding would help to explain the environment non-profits or charitable organizations create to induce sharing, e.g., the small gifts charities often provide ex ante, as predicted by the Behavioral IO perspective. A final ongoing project “What Motivates Giving in the Field?”, joint with Stefano DellaVigna and John List, takes the analysis of sorting and social preferences to field. In a field experiment involving door-to-door fund-raising we test for different motivations of giving in practice. As in the lab experiment above, we vary the option to sort, i.e. to avoid the fund-raiser, and evaluate the effect of the sorting option on giving. The experiment is carried out for the Center for Natural Hazards Mitigation Research at East Carolina University. In our key treatment, potential givers are notified one day in advance about the visit of a fund-raiser at their home. We compare giving in this condition to giving in a more standard treatment, where the visit comes as a surprise. If giving is mostly motivated by altruism or fairness, the advance notice should increase giving since it allows givers to time their presence at home. If giving is mostly motivated by social pressure, shame, or guilt, it should decreases giving since it allows potential givers to avoid being at home or opening the door. In either case, the planned field experiment will provide novel insights into the prevalent sharing motives in practice. To our knowledge, our field experiment is the first to account explicitly for sorting. Its results will also be relevant for the interpretation of laboratory experiments. If they confirm the findings of the laboratory experiment above, the results confirm the usefulness of laboratory experiments with sorting option to explore choices similar to the field under controlled conditions. V. Conclusion The core of my research is devoted to the relevance of behavioral approaches in market settings. It relies on insights about non-standard preferences and beliefs from Psychology and Economics and the empirical and theoretical tools of Applied Microeconomics, including Corporate Finance. All areas of my research are closely related to my teaching and mentoring service. Similar to my graduate courses in Corporate Finance (MBA) and Empirical and Behavioral Corporate Finance (Ph.D.) at Stanford, also my graduate courses here at Berkeley, Topics in Psychology & Economics (Ph.D.) and Corporate Finance (Ph.D.), have relied on my own research, but have also informed my research. The mutual enhancement ranges from practical examples and applications of research topics to Ph.D. theses which concretely rely on themes I taught in class (as in the cases of Burak Güner, Xiaoying Xie, Karen Selody, and Erin Syron). My research has also attracted relatively large groups of about 10-15 undergraduate students per semester, who work with me as Undergraduate Research Apprentices. I am fortunate to attract absolutely outstanding students into these groups, many of which have subsequently pursued graduate studies in Economics or plan to do so. Finally, I am currently designing a novel undergraduate class, in which I will introduce students to the field of Behavioral Finance and Advanced Corporate Finance, including many policy-relevant topics such as corporate governance and the evaluation of recent financial regulations (e.g., Sarbanes-Oxley). With my research focus on Corporate Finance, I am supporting Adam Szeidl and Bob Anderson in their efforts to introduce graduate students in Economics to the field of Financial Economics, including the initiation of a new seminar series, and I am glad to witness increasing numbers of students choosing this field and working with me as their advisor. 9
Slide 10: VI. References Abadie, Alberto and Guido Imbens (2002). Simple and Bias-Corrected Matching Estimators. Working Paper. Acemoglu, Daron and Johnson, Simon (2005). “Unbundling Institutions.” Journal of Political Economy 113, 949-995. Ashenfelter, Orley and David Genesove (1992). “Testing for Price Anomalies in Real-Estate Auctions” American Economic Review: Papers and Proceedings, 82, 501-505. Baker, George P., and Hubbard, Thomas (2003). “Make or Buy in Trucking: Asset Ownership, Job Design, and Information,” American Economic Review 93, 551-572. Baker, George P., and Hubbard, Thomas (2004). “Contractibility and Asset Ownership: On-Board Computers and Governance in U.S. Trucking,” Quarterly Journal of Economics 119, 1443-1479. Baker, Malcolm; Ruback, Richard; and Wurgler, Jeffrey (2006). “Behavioral Corporate Finance: A Survey,” in: E. Eckbo (ed.), The Handbook of Corporate Finance: Empirical Corporate Finance, New York: Elsevier/North Holland. Baker, Malcolm and Jeffrey Wurgler (2000). “The equity share in new issues and aggregate stock returns,” Journal of Finance 55, 2219-2257. Barberis, Nicholas, and Richard Thaler (2003), “A Survey of Behavioral Finance,” Handbook of the Economics of Finance, Chapter 18. Bebchuck, Lucian; Fried, Jesse; and David Walker (2002). “Managerial Power and Rent Extraction in the Design of Executive Compensation,” The University of Chicago Law Review, 69, 366. Blecherman, Barry and Colin Camerer (1996). “Is there a winner's curse in baseball free agency? Evidence from the field,” Working Paper. Cain, Daylain, Loewenstein, George, and Don Moore (2005). “The dirt on coming clean: perverse effects of disclosing conflicts of interest,” Journal of Legal Studies 34, 1–25. Camerer, Colin (2003). Behavioral Game Theory: Experiments on Strategic Interaction, Princeton: Princeton University Press. Ellison, Glen (2006). “Bounded Rationality in Industrial Organization,” in: Richard Blundell, Whitney Newey, and Torsten Persson (eds.), Advances in Economics and Econometrics: Theory and Applications, Ninth World Congress, Cambridge University Press, 2006. Gabaix, Xavier and David Laibson (2006). “Shrouded Attributes, Consumer Myopia, and Information Suppression in Competitive Markets,” Quarterly Journal of Economics 121, 505-540. Hall, Brian, and Kevin J. Murphy (2002). Stock Options for Undiversified Executives, Journal of Accounting and Economics 33, 3-42. Hart, Oliver D., 1995, Firms, Contracts, and Financial Structure, Clarendon Press, Oxford. Heaton, J. B. (2002). “Managerial Optimism and Corporate Finance,” Financial Management 31, 33-45. Hermalin, Benjamin, and Michael Weisbach (1988). “The Determinants of Board Composition,” RAND Journal of Economics 19, 589−606. Hermalin, Benjamin, and Michael Weisbach (1998). “Endogenously Chosen Boards of Directors and Their Monitoring of the CEO,” American Economics Review 88, 96−118. Hoffman, Elizabeth; McCabe, Kevin; Shachat, Keith; and Vernon Smith (1994). “Preferences, Property Rights, and Anonymity in Bargaining Games,” Games and Economic Behavior 7, 346-380. Kaplan, Steven, and Stromberg, Per (2003). “Financial Contracting Theory Meets the Real World: An Empirical Analysis of Venture Capital Contracts,” Review of Economic Studies 70, 281-315. Kaplan, Steven, and Stromberg, Per (2004). “Characteristics, Contracts, and Actions: Evidence from Venture Capitalist Analyses,” Journal of Finance 59, 2173-2206. Kőszegi, Botond (2006). “Ego Utility, Overconfidence, and Task Choice,” Journal of the European Economic Association 4, 673-707. La Porta, Rafael; Lopez-de-Silanes, Florencio; Shleifer, Andrei and Vishny, Robert (1997). “The Legal Determinants of External Finance,” Journal of Finance 53, 1131-50. La Porta, Rafael; Lopez-de-Silanes, Florencio; Shleifer, Andrei and Vishny, Robert (1998). “Law and Finance,” Journal of Political Economy 106, 1113-55. Lambert, Richard; Larcker, David; and Robert Verrecchia (1991). “Portfolio Considerations in Valuing Executive Compensation,” Journal of Accounting Research 29, 129-149. 10
Slide 11: Massey, Cade and Richard Thaler (2006). “The Loser's Curse: Overconfidence vs. Market Efficiency in the National Football League Draft,” Working Paper. Mitchell, Mark; Pulvino, Todd and Erik Stafford (2004). “Price Pressure around Mergers”, Journal of Finance 59, 31-63. Oster, Sharon and Fiona Scott-Morton (2005). “Behavioral Biases Meet the Market: The Case of Magazine Subscription Prices,” Berkeley Electronic Journal Advances in Economic Analysis and Policy 5, No.1. Rajan, Raghuram and Luigi Zingales (1998). “Financial Dependence and Growth,” American Economic Review 88, 559-86. Roll, Richard (1986). “The Hubris Hypothesis of Corporate Takeovers,” Journal of Business 59, 197-216. Rosen, Sherwin (1981). “The Economics of Superstars,” American Economic Review 71, 845-858. Shleifer, Andrei and Robert Vishny (2003). “Stock Market Driven Acquisitions,” Journal of Financial Economics 70, 295-311. Stein, Jeremy (2003), “Agency, Information, and Corporate Investment.” in: G. Constantinides, M. Harris and R. Stulz (eds.), Handbook of the Economics of Finance, Amsterdam, North-Holland. 11