Factor Investing

Factor investing is an approach to building portfolios in which securities are chosen and weighted according to specific, measurable characteristics, called factors, that research has linked to differences in risk and return. Instead of picking individual companies on judgment, or holding the whole market in proportion to size, a factor investor deliberately tilts toward, for example, cheap stocks, small companies or stocks with strong recent performance.

The approach sits between passive management and traditional active management. Like index investing it is rule-based, transparent and relatively cheap. Like active management it aims to do better than a market-weighted index. Products built on it are often marketed as smart beta. It is used by pension funds, sovereign wealth funds and quantitative managers, and is available to individuals through funds.

The first factor was the market itself. The capital asset pricing model of the 1960s held that a stock's expected return depends on one thing only, its beta to the market. In 1976 Stephen Ross proposed the arbitrage pricing theory, which allowed for several sources of systematic risk without naming them.

Empirical work then found patterns that beta did not explain. Small companies and companies with low prices relative to book value had earned more than the model predicted. In the early 1990s Eugene Fama and Kenneth French gathered these findings into a three-factor model of market, size and value. Mark Carhart added momentum in 1997, and Fama and French extended their model in 2015 with profitability and investment. Fama shared the Nobel Memorial Prize in Economic Sciences in 2013. The statistical tool that connects these ideas is the factor model.

• Value. Stocks that are cheap relative to earnings, book value or cash flow have tended to outperform expensive ones over long periods. This is the systematic form of value investing.
• Size. Small-cap stocks have historically earned more than large ones, though the effect has been weak in recent decades.
• Momentum. Stocks that have outperformed over the past six to twelve months have tended to continue doing so for some months. See cross-sectional momentum.
• Quality. Companies with high profitability, stable earnings and low debt have outperformed weaker ones.
• Low volatility. Stocks with low volatility or low beta have delivered better risk-adjusted returns than theory predicts, the basis of betting against beta.
• Carry. In currencies, bonds and commodities, assets with higher yields have tended to outperform those with lower yields, as in the currency carry trade.

There are two broad forms. Long-only factor funds hold the stocks that score well on a factor and leave out or underweight the rest. They keep full exposure to the market, so their returns are dominated by the market return, with the factor adding or subtracting a modest amount. Most smart beta products, including many of the exchange-traded fund type, work this way.

Long-short factor portfolios buy the top-ranked stocks and sell short the bottom-ranked ones in equal amounts. The market exposure roughly cancels, leaving the pure factor return. This is the academic definition of a factor and the form used by hedge funds, but it involves leverage, borrowing costs and higher turnover.

Because single factors can underperform for years, many investors combine several. Value and momentum tend to do well at different times, which makes a value and momentum strategy smoother than either alone. Holding several lowly correlated factors is a form of diversification across sources of return.

Three kinds of explanation are offered. The first is risk: factor portfolios do badly in bad times, and the extra return is a risk premium for bearing that. Value stocks, for instance, are often distressed companies that suffer most in recessions. The second is behavioral: investors overpay for exciting growth stories, react slowly to news, or chase lottery-like payoffs, and the resulting mispricing is slow to correct. The third is structural: constraints on leverage, short selling or benchmark deviation prevent institutions from trading the patterns away.

The distinction matters for the future. A premium that pays for risk should persist. One that arises from mistakes may fade as more capital pursues it.

The largest concern is data mining. Researchers have published several hundred factors, a situation described as a factor zoo. With so many tested on the same historical data, many apparent discoveries are likely to be chance. Studies have argued for a much higher statistical bar before a new factor is accepted and have found that factor returns tend to shrink after publication.

Factors also fail in practice for long stretches. The value factor underperformed badly for more than a decade up to 2020. Momentum is subject to a sudden momentum crash. In August 2007 many quantitative funds holding similar factor positions suffered sharp simultaneous losses as they all reduced positions at once, showing the risk of crowding. Returns reported in academic papers ignore transaction costs and often rely on small, illiquid stocks, so the return an investor can actually capture is lower. Past factor premiums are no assurance of future ones.