Getting Better: Learning to Invest in an Emerging Stock Market by J.Y. Campbell, T. Ramadorai T, and B. Ranish
Stefan Nagel Stanford University, NBER, CEPR
April 2013
Stefan Nagel
Discussion of Learning to Invest
Origins of Investor Beliefs How to people come up with beliefs about equity premium? style return premia? their trading skill? importance of diversification? performance of active mutual funds? inflation & real interest rates?
Standard assumption in AP: Rational expectations. Leads to puzzles why why why why
are asset returns predictable? do people trade? do people invest in active mutual funds? are individual investors underdiversified?
Alternative: Learning
Stefan Nagel
Discussion of Learning to Invest
Anti-Learning
Stefan Nagel
Discussion of Learning to Invest
Stefan Nagel
Discussion of Learning to Invest
Anti-Learning
Investor Learning: Many Interesting Empirical Questions
Learning: From what? Educational offerings and introspection? Public information including all historical data? Life-time experiences of public information? Social network? Own-portfolio experiences?
Learning: How? Updating: probably some Bayesians, some boundedly rational, some immune to learning? Priors: probably heterogeneous priors
Stefan Nagel
Discussion of Learning to Invest
Learning of Investors in India This paper: Fantastic data set from India Comprehensive picture of stock investments (mutual fund holdings are low) by tax ID Limited demographic information but has account age In some ways comparable to Scandinavian data sets, but here, e.g., higher frequency observations than Calvet et al. (2007, 2009)
Do they learn? Yes. Older accounts outperform younger accounts partly due to better style tilts partly due to better stock picks unrelated to style tilts
Older accounts are better diversified, have lower turnover, smaller disposition effect
How do they learn? Learning from experience. Poor returns cumulative, or in single month induce lower turnover and disposition bias, and greater diversification Stefan Nagel
Discussion of Learning to Invest
Comment 1: How Do They Learn to Earn Higher Returns? Lower turnover and better diversification of older accounts do not explain why their returns are higher before transaction costs and risk-adjustment. Still an open question. How does better style tilt arise? From experience? relate style tilt to experience of market-wide style returns? Not much power with short time series
relate style tilt to experience of personal style returns? e.g., investor A bought value stocks that outperformed the market, investor B bought value stocks that underperformed. Does this heterogeneity induce differences in propensity to own value stocks?
foundation for Barberis and Shleifer (2003) style investing?
How do better stock picking skills arise? Perhaps to some extent unobserved style tilts? Expand style categories
Stefan Nagel
Discussion of Learning to Invest
Comment 2: Heterogeneity in Learning?
Young-old differences in updating: Young may be more sensitive to recent experiences Are there types of investors that are more adept at learning to invest? Are value investor “types” behaving different from growth investor “types”? Diversification “types” vs. gambling “types”? ...
Possibly non-monotonic Well diversified, value investors, ... have nothing to learn Not-so-well diversified ones, moderate growth investors, ... have something to learn Some extreme types immune to learning?
Stefan Nagel
Discussion of Learning to Invest
Comment 3: Does Past Investment Success Induce Bad Behavior?
Finding: Good own-portfolio performance in the past relative to the market/from trading/from disposition-behavior reinforces low diversification/high turnover/disposition-behavior Is this evidence of reinforcement of “bad” behavior? Test: Do own-portfolio performance measures predict future own-portfolio returns and Sharpe Ratio negatively?
Stefan Nagel
Discussion of Learning to Invest
Comment 4: Puzzling Factor-Adjusted Returns
Old-accounts minus young-accounts zero-investment portfolio in Table 6: why is hml loading negative even though older accounts seem to have a value tilt? odd consequence: factor-adjusted returns are even bigger than raw excess returns, even though some of raw excess return arises from apparent value tilt of older accounts due to value-weighting of stocks in construction of factors?
Stefan Nagel
Discussion of Learning to Invest
Summary
Great Data. New insights into investor behavior. Findings underscore that investor behavior evolves dynamically as function of experiences Potential to do a lot more with the data set Potential to connect to asset pricing questions (e.g. style investing).
Discussion of Getting Better: Learning to Invest in an ...
Public information including all historical data? Life-time experiences of public information? Social network? Own-portfolio experiences? Learning: How?
Jan 2, 2013 - can earn that also. To be spiritual ... money, a house, a family, children. You ... will make yourself miserable and go around with along face,.
Industry-specific experience = experience of (relatively) bad industry return while being invested. Finding: Managers with more experience in industry A than in ... Correlation between industry volatility and the number of negative industry shocks. D
tive of topic detection and tracking (Allan 2002) and dis- .... The decision trees' big advantage is its ability to handle .... It can not handle big amounts of data in an.
Outline. 1. Summary. 2. Comments ... Learning acts as coordination device. ⻠Similar to Sargent ... (Summary in Sargent and Söderström, 2000). ⻠Central ...
Counterparty-free, physical gold bullion is best defence against bail-ins. One of Italy's twenty ... This happens in companies of all scales and levels of success.
ÐаÑÑÐ¾Ñ Ñазглавного геÑоÑ,. СкаÑаÑÑ Ð±ÐµÑплаÑно игÑÑÐовÑй Человек-паÑк 2. Whoops! There was a problem loading this page. Retrying... Whoops! There was a problem loading this page. Retrying... Inv
Parents will need to access their email accounts and click the link provided by the system. 6. Once email validation is complete, the parent account is setup and ...
approaches, tree-based methods, support vector machines, and clustering. Related. The Elements of Statistical Learning: Data Mining, Inference, and Prediction ...
approaches, tree-based methods, support vector machines, and clustering. Related. The Elements of Statistical Learning: Data Mining, Inference, and Prediction ...
AllianceBernstein invites you to attend a presentation on summer internship opportunities with our firm: Launch your career with AllianceBernstein. Tuesday ...
Jan 30, 2015 - Under staggering and TI, firm has incentive to adjust price by more when it gets the chance. V. Lewis (). Discussion Weber. 30th January 2015.