Norges Bank’s Five-Factor Equity Risk Model
(May 3, 2013) – Norges Bank Investment Management (NBIM), which presides over $721 billion, including the Norway Government Pension Fund-Global, is living up to its reputation as a pioneer in risk-factor investing.
Two NBIM investment analysts, Tapio Pekkla and Nikolay Doskov, have teamed up with JP Morgan Executive Director Ruy Ribeiro to develop a factor model for public equities. The initial fruits of their labor are now available online: “Tradable Macro Risk Factors and the Cross-Section of Stock Returns,” a working paper.
The model has five factors: expected dividend growth, expected dividend level, equity risk, expected inflation, and interest rates. Doskov, Pekkla, and Ribeiro call them “tradable macro factors” because they capture the market expectations of aggregate cash flow and discount rate news, and are not based on information related to single stocks.
The authors applied their model to a data set covering the performance of US public equities from 2003 through 2011.
They found that expected dividend growth was the only factor consistently priced in to the test portfolios. Furthermore, that variable commanded a higher risk premium than the discount-rate factor. In the study, increases in expected inflation were associated with higher equity returns, but realized inflation was not.
Like any good investment managers, Doskov, Pekkla, and Ribeiro also tested their results against an established benchmark—or, rather, two benchmarks. The Fama-French three-factor model failed to explain equity returns for any of the study’s 25 test portfolios, except the one used in Eugene Fama and Kenneth French’s own study.
The single-variable Capital Asset Pricing Model (CAPM) also flopped when applied to the test portfolios for the 1963 to 1991 period. Unlike the Fama-French model, the authors did not reject CAPM outright because in most cases alpha was not significantly different from zero, which the model showed.
“Our model, which consists solely of macroeconomic variables, explains well the variation in returns of the 25 portfolios,” the authors wrote. “The outlier in the model is the (extreme) small-growth portfolio, which has the most significant (positive) pricing error. However, neither the CAPM nor the Fama-French model can explain the return of this portfolio. Overall, we find significant evidence that our tradable factor model provides a powerful characterization of equity returns.”
NBIM was an early adopter of risk-factor modeling. For the sovereign fund’s portfolio as a whole, it has identified 10 factors: Term, credit Aa, credit Baa, credit high yield, FX carry, liquidity, value/growth, small cap/large cap, momentum, and volatility.
“I first heard about risk-factor investing through Norway’s work on it,” said Melissa Brown—institutional risk veteran and senior director of applied research at Axioma, a New York-based risk analysis firm—during a recent interview with aiCIO. She has heard plenty about it since. “The fund is still cutting edge: they definitely seems to be one of the leaders in the area…What was the title of the paper? I’ve got to tell my boss about this.”
aiCIO is also hoping to be a leader in risk-factor investing. If you’ve gotten this far in the article, we encourage you to get in touch with columnist Angelo Calvello who’s leading a community of asset owners in exploring and developing risk-factor investing. The discussions center on four major themes: Definition, classification, implementation, and governance. The goal is to present the findings at the aiCIO Influential Investor Forum in New York on December 9.
Doskov, Pekkla, and Ribeiro’s full paper is available here, and the authors welcome comments on it.
Related profile: Yngve Slyngstad, chief executive officer of Norges Bank Investment Management