André de Souza, assistant professor of finance, Fordham University Schools of Business

André de Souza

Assistant professor of finance, Fordham University Schools of Business

Curriculum Vitae: CV

Contact information: Department of Finance,
Fordham University Schools of Business,
1790 Broadway
Suite 1327
New York, NY 10019.

Phone: +1-212-636-6716

Email: adesouza1@fordham.edu

Research:

A mispricing-based explanation of how flow affects mutual fund performance
I study the effect of flow on mutual fund performance. I use the fact that investors respond to raw return, controlling for alpha, to create fund portfolios sorted along the two dimensions of skill (proxied for by alpha) and flow (proxied for by raw return). Performance increases in skill and, as conjectured by earlier work, decreases in flow. The high alpha-low raw return fund portfolio has a Carhart alpha, net of fees, of 2.65% over the next year. I construct an extension of the model of Berk and Green (2004), in which investors learn about fund manager ability from raw return, which predicts this pattern.
I obtain the same pattern in alphas when I track the portfolios held by the funds on the date at which they are ranked by performance through the subsequent year. This implies that the effect of flow is on the prices of the stocks the funds hold rather than on the funds themselves. Outperforming funds reinvest the flow they receive in the same stocks they already hold, driving up their prices and causing a lack of subsequent outperformance. Flow thus affects performance by correcting mispricing in the underlying stocks.

How would bondholders vote?
When voting in a company's shareholder meetings, mutual fund families which hold more debt than equity in that company vote differently on certain types of propositions than families which hold relatively more equity. I conclude that these classes of propositions do not affect bondholders and stockholders in the same way, and infer how bondholders would vote on these propositions if they had the vote. I find that there are three types of propositions which affect bondholders differently from equityholders: the authorization of new common and preferred stock, the approval of pay-for-performance schemes, and the removal of anti-takeover defences.

Average return and cross-sectional variance of return of mutual funds (in progress)
Skilled funds trading in a segment of the market in which some stocks are overpriced and some stocks underpriced will make returns, on average, that are far in excess of those made by unskilled funds trading in the same segment. On the other hand, in a segment where all stocks are more or less correctly priced, skilled funds will make returns that are only slightly higher than those made by unskilled funds. Therefore, the cross-sectional variance of fund returns will be higher in the former segment than in the latter. Moreover, since unskilled funds should make zero alpha on average in both segments, the cross-sectional average return of funds in the former segment will also be higher than it is in the latter. I write down a simple model that makes this relationship plain. Aside from linking cross-sectional dispersion in fund returns to average fund returns, this model can be used to estimate how predictable returns are in various segments of the stock market. Examples of such segments are large stocks versus small stocks, or stocks with high analyst coverage versus those with low analyst coverage.

Does mutual fund performance vary over the business cycle? (with Anthony Lynch)
Conditional factor models allow both risk loadings and performance over a period to be a function of information available at the start of the period. Much of the literature to date has allowed risk loadings to be time-varying while imposing either the assumption that conditional performance is constant or the assumption that conditional betas are linear in the information. We develop a new methodology that allows conditional performance to be a function of information available at the start of the period but does not make assumptions about the behavior of the conditional betas. This methodology uses the Euler equation restriction that comes out of the factor model rather than the beta pricing formula itself. It assumes that the stochastic discount factor (SDF) parameters are linear in the information. The Euler equation restrictions that we develop can be estimated using standard GMM, which does not use all available data when the mutual fund data starts at different times for different funds and later than the factor and instrument data. We also use econometric techniques developed by Lynch and Wachter (2007) to estimate the Euler equation restrictions taking account of all available factor return, instrument, and mutual fund data. These techniques allow us to produce more precise parameter estimates than those obtained from the usual GMM estimation. We use our SDF-based method to assess the conditional performance of fund styles in the CRSP mutual fund data set. Using dividend yield and term spread to track the business cycle, we find that conditional mutual fund performance relative to conditional versions of the Fama-French and Carhart pricing models moves with the business cycle, and this business cycle variation in performance differs across large-NAV and small-NAV funds for many of the fund styles. We find that the conditional performance is sensitive to whether return is measured in excess of the riskfree asset or a Fama-French 25 portfolio that is matched to each fund style on the basis of Fama-French loadings: performance is typically more cyclical after adjusting for the conditional performance of the underlying stocks. There is some evidence that conditional performance relative to the riskless rate using the Fama-French model of the four styles with the most data, (growth and income, growth, maximum capital gains and income) is countercyclical. However, after adjusting for the conditional performance of the underlying stocks and using the dividend yield as the instrument, only the conditional performance of the two maximum capital gains portfolio remains countercyclical, while the conditional performance of both growth portfolios and the small-fund income portfolio becomes procyclical. When conditional performance is measured using the Carhart model, the evidence for variation in conditional performance for these 4 fund styles over the business cycle is quite weak. Turning to the other fund styles, after adjusting for the conditional performance of the underlying stocks, the energy-sector and utilities-sector portfo- lios typically exhibit countercyclical performance, while the financial-sector, small-cap growth and flexible portfolios typically exhibit procyclical performance.

Are stock return correlations too high? (with Anthony Lynch and Sinan Tan)
(in progress)
Lynch (2003) presents a general equilibrium paper that solves for prices in an economy with predictable cash flows. We are concerned with the high correlations between the returns of stock portfolios relative to the much lower correlations between their dividends. We examine whether any of the major utility and consumption specifications that have been advanced to explain other empirical pricing regularities (like the equity premium puzzle) can explain this disconnect between the correlations of returns across stock portfolios and the correlations of the underlying cash flows across the same portfolios. Lynch (2003) examines CRRA utility and both i.i.d. and predictable cash flows, and calibrating the cash flow correlations to data, he finds that the return correlations in the model are much lower than those in the data, which is a puzzle. The question we address is whether Epstein-Zin, habit preferences or a long-run risk specification for consumption can generate return correlations that are comparable to those in the data.

	Curriculum Vitae:	CV
	Contact information:	Department of Finance, Fordham University Schools of Business, 1790 Broadway Suite 1327 New York, NY 10019.
	Phone:	+1-212-636-6716
	Email:	adesouza1@fordham.edu

	Research:
	A mispricing-based explanation of how flow affects mutual fund performance I study the effect of flow on mutual fund performance. I use the fact that investors respond to raw return, controlling for alpha, to create fund portfolios sorted along the two dimensions of skill (proxied for by alpha) and flow (proxied for by raw return). Performance increases in skill and, as conjectured by earlier work, decreases in flow. The high alpha-low raw return fund portfolio has a Carhart alpha, net of fees, of 2.65% over the next year. I construct an extension of the model of Berk and Green (2004), in which investors learn about fund manager ability from raw return, which predicts this pattern. I obtain the same pattern in alphas when I track the portfolios held by the funds on the date at which they are ranked by performance through the subsequent year. This implies that the effect of flow is on the prices of the stocks the funds hold rather than on the funds themselves. Outperforming funds reinvest the flow they receive in the same stocks they already hold, driving up their prices and causing a lack of subsequent outperformance. Flow thus affects performance by correcting mispricing in the underlying stocks.
	How would bondholders vote? When voting in a company's shareholder meetings, mutual fund families which hold more debt than equity in that company vote differently on certain types of propositions than families which hold relatively more equity. I conclude that these classes of propositions do not affect bondholders and stockholders in the same way, and infer how bondholders would vote on these propositions if they had the vote. I find that there are three types of propositions which affect bondholders differently from equityholders: the authorization of new common and preferred stock, the approval of pay-for-performance schemes, and the removal of anti-takeover defences.
	Average return and cross-sectional variance of return of mutual funds (in progress) Skilled funds trading in a segment of the market in which some stocks are overpriced and some stocks underpriced will make returns, on average, that are far in excess of those made by unskilled funds trading in the same segment. On the other hand, in a segment where all stocks are more or less correctly priced, skilled funds will make returns that are only slightly higher than those made by unskilled funds. Therefore, the cross-sectional variance of fund returns will be higher in the former segment than in the latter. Moreover, since unskilled funds should make zero alpha on average in both segments, the cross-sectional average return of funds in the former segment will also be higher than it is in the latter. I write down a simple model that makes this relationship plain. Aside from linking cross-sectional dispersion in fund returns to average fund returns, this model can be used to estimate how predictable returns are in various segments of the stock market. Examples of such segments are large stocks versus small stocks, or stocks with high analyst coverage versus those with low analyst coverage.
	Does mutual fund performance vary over the business cycle? (with Anthony Lynch) Conditional factor models allow both risk loadings and performance over a period to be a function of information available at the start of the period. Much of the literature to date has allowed risk loadings to be time-varying while imposing either the assumption that conditional performance is constant or the assumption that conditional betas are linear in the information. We develop a new methodology that allows conditional performance to be a function of information available at the start of the period but does not make assumptions about the behavior of the conditional betas. This methodology uses the Euler equation restriction that comes out of the factor model rather than the beta pricing formula itself. It assumes that the stochastic discount factor (SDF) parameters are linear in the information. The Euler equation restrictions that we develop can be estimated using standard GMM, which does not use all available data when the mutual fund data starts at different times for different funds and later than the factor and instrument data. We also use econometric techniques developed by Lynch and Wachter (2007) to estimate the Euler equation restrictions taking account of all available factor return, instrument, and mutual fund data. These techniques allow us to produce more precise parameter estimates than those obtained from the usual GMM estimation. We use our SDF-based method to assess the conditional performance of fund styles in the CRSP mutual fund data set. Using dividend yield and term spread to track the business cycle, we find that conditional mutual fund performance relative to conditional versions of the Fama-French and Carhart pricing models moves with the business cycle, and this business cycle variation in performance differs across large-NAV and small-NAV funds for many of the fund styles. We find that the conditional performance is sensitive to whether return is measured in excess of the riskfree asset or a Fama-French 25 portfolio that is matched to each fund style on the basis of Fama-French loadings: performance is typically more cyclical after adjusting for the conditional performance of the underlying stocks. There is some evidence that conditional performance relative to the riskless rate using the Fama-French model of the four styles with the most data, (growth and income, growth, maximum capital gains and income) is countercyclical. However, after adjusting for the conditional performance of the underlying stocks and using the dividend yield as the instrument, only the conditional performance of the two maximum capital gains portfolio remains countercyclical, while the conditional performance of both growth portfolios and the small-fund income portfolio becomes procyclical. When conditional performance is measured using the Carhart model, the evidence for variation in conditional performance for these 4 fund styles over the business cycle is quite weak. Turning to the other fund styles, after adjusting for the conditional performance of the underlying stocks, the energy-sector and utilities-sector portfo- lios typically exhibit countercyclical performance, while the financial-sector, small-cap growth and flexible portfolios typically exhibit procyclical performance.
	Are stock return correlations too high? (with Anthony Lynch and Sinan Tan) (in progress) Lynch (2003) presents a general equilibrium paper that solves for prices in an economy with predictable cash flows. We are concerned with the high correlations between the returns of stock portfolios relative to the much lower correlations between their dividends. We examine whether any of the major utility and consumption specifications that have been advanced to explain other empirical pricing regularities (like the equity premium puzzle) can explain this disconnect between the correlations of returns across stock portfolios and the correlations of the underlying cash flows across the same portfolios. Lynch (2003) examines CRRA utility and both i.i.d. and predictable cash flows, and calibrating the cash flow correlations to data, he finds that the return correlations in the model are much lower than those in the data, which is a puzzle. The question we address is whether Epstein-Zin, habit preferences or a long-run risk specification for consumption can generate return correlations that are comparable to those in the data.