Behavioral Finance Meeting
November 5, 2011
Yen-cheng Chang, Shanghai Advanced Institute for Finance, and Harrison Hong, Princeton University and NBER
Using the popular Russell stock market indexes, Chang and Hong show that institutional portfolio rules induce regression discontinuity experiments in asset markets. Stocks are ranked each June on their market capitalization from 1 (largest) to 3000 (smallest). Those ranked just above the 1000 cut-off are in the Russell 1000, while those just below are in the Russell 2000. Because the indexes are value-weighted, smaller stocks just below the 1000 cut-off are heavily weighted in Russell 2000 and receive forced index buying. Larger ones just above the cut-off have negligible weight in Russell 1000 and are neglected by institutions. Smaller just-included stocks have discontinuously and significantly higher institutional ownership, price, liquidity, short interest, and market co-movement compared to just-excluded larger ones.
Dong Lou, London School of Economics
Lou provides evidence that managers adjust firm advertising, in part to attract investor attention and to influence short-term stock returns. Increased advertising spending is associated with a contemporaneous rise in retail buying and in abnormal stock returns, and is followed by lower future returns. There is a significant increase in advertising spending prior to insider sales, and a significant decrease in the following year. A similar pattern arises around equity issues and stock-financed acquisitions, but is absent around debt issues and cash-financed acquisitions. Additional analyses suggest that the hump-shaped pattern in advertising spending around equity sales is most consistent with managers' opportunistically adjusting firm advertising to exploit the return effect to the benefit of their own and that of their existing shareholders.
Huina Mao and Johan Bollen, Indiana University-Bloomington, and Scott Counts, Microsoft Research
Mao, Bollen, and Counts investigate whether the results of a variety of well-accepted socioeconomic surveys, generally obtained from opinion polling, that can be replicated (and predicted) by Computational Economic and Financial Gauges (CEFG) extracted from large-scale search engine and Twitter data. In particular, they examine the results of the Michigan Consumer Confidence Index, Gallup Economic Confidence Index, Unemployment Insurance Weekly Claims reported by U.S. Department of Labor and two investor sentiment surveys (weekly Investor Intelligence and Daily Sentiment Index). Their results show that CEFGs not only exhibit statistically significant correlations to many if not most existing socio-economic indexes, but also precede and thus predict survey data. They also find that a CEFG of investor sentiment obtained from Twitter may be a leading indicator of the financial markets, which existing surveys tend to lag.
David Hirshleifer, University of California at Berkeley, Irvine, and Jianfeng Yu, University of Minnesota
Introducing extrapolation bias into a standard one-sector production-based real business cycle model with recursive preferences reconciles salient stylized facts about business cycles (low consumption volatility and high investment volatility relative to output) and financial markets (high equity premium, volatile stock returns, and a low and smooth risk-free rate) with low relative risk aversion and an intertemporal elasticity of substitution in preferences of greater than one. Furthermore, the model matches several conditional stylized facts, such as return predictability based upon dividend yield, Q, and investment rates. These successes derive from the interaction of capital adjustment costs, extrapolative bias, and recursive preferences. Extrapolative bias increases the variation in the wealth-consumption ratio; recursive preferences cause this variation to be strongly priced; and adjustment costs make corporate payouts more pro-cyclical. Hirshleifer and Yu provide empirical support for the mechanism of the model.
Yigitcan Karabulut, Goethe University, Frankfurt
Using a novel and direct measure of investor sentiment, Karabulut finds that Facebook's Gross National Happiness (GNH) has the ability to predict changes in daily returns and trading volume in the U.S. equity market. For instance, a single standard deviation increase in GNH predicts an increase in market returns equal to 11 basis points over the next day. Moreover, the impact of GNH appears to be stronger among small-cap stocks, and in the face of turmoil.
Sebastien Pouget and Stephane Villeneuve, University of Toulouse
Pouget and Villeneuve propose a dynamic model of financial markets in which some investors are prone to the confirmation bias. Following insights from the psychological literature, these agents are assumed to amplify signals that are consistent with their prior views. In a model with public information only, this assumption provides a rationale for the volume-based price momentum documented by Lee and Swaminathan (2000). The results here are also consistent with a variety of other empirically documented phenomena such as bubbles, crashes, reversals, and excess price volatility and volume. Novel empirical predictions are derived: 1) return continuation should be stronger when biased traders' beliefs are more extreme, and 2) return continuation should be stronger after an increase in trading volume. The implications of the model for short-term quantitative investments are twofold: 1) optimal trading strategies involve riding bubbles, and that 2) contrarian trading can be optimal in some market circumstances.
Ulrike Malmendier, University of California at Berkeley and NBER, and Stefan Nagel, Stanford University and NBER
How do individuals form expectations about future inflation rates? Malmendier and Nagel propose that personal life-time experiences play a significant role in expectation formation. Unlike in the existing models of adaptive learning, here individuals put a higher weight on realizations experienced over their life times than on other "available" historical data. Averaged across cohorts, expectations resemble those obtained from constant-gain learning algorithms common in macroeconomics, but the experience model also predicts variations in learning speed based on cross-sectional heterogeneity between cohorts. Using 54 years of micro-data on inflation expectations from the Reuters/Michigan Survey of Consumers, the authors show that differences in life experiences strongly correlate with differences in subjective expectations. Young individuals place more weight on recently experienced inflation than older individuals, consistent with recent experiences making up a larger part of their lifetimes so far. The experience effect explains why there is substantial disagreement between young and old individuals about future inflation rates in periods of high surprise inflation, such as the 1970s. The experience effect also helps to predict the time-series of forecast errors in the Reuters/Michigan survey and the Survey of Professional Forecasters, and the excess returns on nominal long-term bonds.
Victor Stango, University of California at Davis, and Jonathan Zinman, Dartmouth College and NBER
For many households, paying lower borrowing costs is the surest, fastest way to increase net worth. Using administrative, credit bureau, and survey data on U.S. credit cards, Stango and Zinman find pervasive and systematic cross-individual variation in borrowing costs. Credit risk and product differentiation explain about one-third of that variation. The remaining risk-adjusted dispersion can materially affect wealth accumulation: moving heavy borrowers from the 75th to the 25th percentile of risk-adjusted borrowing costs increases their savings rates by more than a percentage point. Debt (mis)-allocation conditional on cards held could matter in principle, but appears to matter very little in practice, because most people allocate debt to their lowest-rate cards. Rather, similarly risky borrowers often hold cards with very different contract Annual Percentage Rates (APR). Heterogeneity in consumer search behavior appears to be an important factor in explaining that contract APR variation - a factor nearly as important as credit risk in explaining the cross section of borrowing costs.