AlphaLetters Quant Paper Reviews

What is the Coase Theorem?

—

Goldman is preparing the largest hedge fund launch ever – which is surprising given how bad two of their flagships funds have performed.

—

Some readers emailed in with some questions on the foreign listed hedge funds. Here is a list of the largest individual funds and FOFs listed in London. The manager is listed in parenthesis if different from the fund name. You can find the quotes on Bloomy, Yahoo, or BigCharts. Some have different currency share classes.

Individual Hedge Funds
MW Tops (Marshall Wace)
BH Macro (Brevan Howard)
Boussard & Gavaudan
Third Point
RAB Special Situations

London Listed FOFs
Dexion Absolute (Harris Alternatives)
Alternative Investment Strategies (ABN Amro, formerly International Asset Management)
Goldman Sachs Dynamic Opps
CMA Global Hedge (CM Advisors)
Thames River Multi Hedge
AcenciA Debt Strategies (Sandalwood Securities)
Close AllBlue Funds (BlueCrest)
Dexion Equity Alternatives (K2)
Dexion Alpha Strategies (RMF)

—

Thanks to the good people at AlphaLetters here is another sample of their academic quant paper reviews:

Article 1.
Category: Beta, alternative measures
Title: Long-Term and short-term market betas in Securities Prices
Author: Gerard Hoberg and Ivo Welch
Source: Brown University
Link: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=987353

Summary:
Beta is commonly measured using 5 years of monthly data, and is not significant at predicting future stock returns. This paper proposes two alternative measures that can predict future one-year stock returns:
· a short term beta based on 1 year (year -1) daily stock returns
· a long term beta based on 9 year (year -10 to year -1) daily stock returns

Key findings:
· Long-term (short-term) beta is positively (negatively) related to future stock returns.
· A strategy that is long stocks with high long-term beta and low short-term beta, and short stocks with the opposite pattern, yields a risk -adjusted significant excess return of 7.5% per year
· The so-called “change in beta” (difference between long term beta and short-term beta) has significant predicting power of stock returns.
· Robustness tests confirm that these two betas are indeed different from the Fama -French variables.

Why can the long-term and short-term beta predict future returns, while the standard 5 year beta cannot?

The author proposes that
· Long-term beta is a proxy for investors hedging motives, while short-term beta seems to be a novel factor not captured in previous studies
· The standard beta (5-year) loses its predictive power because it mixes these two different effects which may move in opposite directions.

AlphaLetters comments:
1. Why important
This article builds support for the beta factor that Fama-French model discredits, and also presents a market timing strategy that may provide excess returns.

2. Data source
1962 – 2005 stock data are from CRSP/Compustat. Firms with less than 4 years of past return data are excluded. On average the data set has 3300 firms per month.

3. Discussions
We are not completely convinced by the authors’ explanations of findings (page 25 lists four of them: 1. Slow Adjustment to Changes in Beta, 2. Tax Effects in Up vs. Down Markets, 3. Relative Mean Reversion, 4. Novel Factor Exposure).

To exclude the possibility of data mining, we have to ask what is a typical stock with high beta changes(i.e., stocks with high past beta but low recent beta)? Intuitively, “out-of-favor” growth stocks migrating into value category are likely candidates, since they tend to have high volatility in the past but are no longer
“hot”. Evidence suggests that these stocks are likely to outperform. Does this mean that the change in beta is just a fancy name for “value”?

On a separate note, the authors have done a meticulous job of confirming their results and testing for robustness.

—-

Article 2.
Category: Treasury bond returns, stock returns, seasonality
Title: Opposing Seasonalities in Treasury versus Equity returns
Author: Mark J. Kamstra, Lisa A. Kramer, Maurice D. Levi
Source: Maryland University working paper
Link: http://www.rhsmith.umd.edu/finance/pdfs_docs/seminarspring07/Kramer.pdf

Summary:
This article documents an opposing annual seasonality in US Treasury (5, 7, 10, and 20-year US government bond) and equity monthly returns:
· Treasury monthly returns usually peak in October, and bottom in April
· The difference between treasury peak and trough is over 80+ basis points
· An opposing seasonal pattern can be found in equity returns: peak in April but trough in October.

A rather unique behavioral approach is used to explain the finding: the incidence of seasonal depression in North American populations (seasonal affective disorder, SAD, which influence 5% -15% population)usually peak in April and ebb in October. This may cause the pattern of investors’ risk tolerance level.

To rule out data snooping, the authors investigate 24 separate effects (including a broad range of seasonal macroeconomic factors), the four factors from Fama-French and the January effect. None of these effects seem to explain the seasonality better than the SAD variable.

AlphaLetters comments:
1. Why important
Various papers have confirmed that equity returns tend to be positive before summer time (hence “sell in May and go away”). This paper presents a new finding in treasury returns, which corresponds to a related pattern in equity returns. Such seasonal trends may help investors refine their risk budget during different time of the year, and it’s also meaningful for asset allocation.

2. Data source
1952 – 2004 treasury and Equity data are from CRSP.

3. Discussions
In explaining the finding, we are curious about the effect of number of trading days in a month. Because of holidays (i.e. November, December, and January) and shorter months (i.e. February), the number of trading days in a month is less from October – April. The smaller number of trading days could also affect the monthly return volatility which is one of the authors’ causes for the Treasury trend. We would like to see more analysis done on the monthly Treasury returns. For example, it would be
interesting to study the impact of standard deviation for the monthly returns, especially since one of the monthly jumps can be 40 bps.

The authors do not directly study capturing excess returns with a portfolio of Treasury bonds. The expected raw return would be approximately 1.6% (i.e. 80 basis points from October to April is approximately 1.6% per annum). The authors’ claim that the profit can be approximately 10% per annum (bottom of page 5) seems wrong.

—-