An update on the small-cap effect

  • Written by 

    Written by Christian Theis 20 May 2014

As the Russell 2000 index celebrates its 30th birthday this year, we assess the small cap effect through the lens of the benchmark used by most institutional US small cap products. The performance of smaller stocks has been cyclical, outperforming in bursts, but not enough for long-term returns to compensate for their extra volatility.

Macro update: UK Groundhog Day? 1 of 2

Residents of the United States, please read this important information before proceeding

The long debate about the small cap effect

The idea of a small-cap premium goes back to 1981, when Rolf Banz published “The Relationship between Return and Market Value of Common Stocks” in the Journal of Financial Economics. Among his findings was: “In the 1936-1975 period, the common stock of small firms had, on average, higher risk-adjusted returns than the common stock of larger firms.” Professors Eugene Fama and Kenneth French looked at the evidence in their famous 1992 paper, “The Cross-Section of Expected Stock Returns.”, covering the period 1963-1990. While they did find that small stocks had higher average returns, they believed the higher returns were compensation for risk.

Since the late 1990s, the small-cap thesis has been questioned. There has been criticism of Banz’s original work: the small-cap premium that Banz discovered was based on a handful of performance bursts, not consistent outperformance. There is also the question to what extent the excess returns of small-company stocks disappear when taking higher transaction costs into account. Finally the data that Banz used fails to account for stocks that are delisted by stock exchanges for performance-related reasons.

So it has been questioned if the anomaly of a higher risk-adjusted return for small cap firms really existed in the past. But, even if it did, one economist comments that: “After they are documented and analysed in the academic literature, anomalies often seem to disappear, reverse, or attenuate.” How has the small cap effect evolved after its documentation?

Annual leadership by small cap or large/mid cap has been fairly evenly split

Results for the US

For the US market, a look at the Russell 2000 index is probably most appropriate to answer this question. Russell Investment introduced this index in 1984 as the first small cap benchmark and it has since then been widely adopted by both institutional and retail investors for measuring performance in the small cap US equity market. As of Dec 31, 2013, approximately 97% of institutional US small cap products were benchmarked against the Russell 2000. The index represents the 2000 smallest stocks of the Russell 3000 index, and the large/mid cap Russell 1000 index the other 1000 stocks. Figure 1 shows the annual returns for those large/mid cap and small cap stocks since the start of the backfilled indices in 1979. For this 35-year history, there is not much evidence of a compensated risk premium – overall annual returns are both 12% and leadership was split evenly. The small cap led in 18 years, while the large/mid cap led in 17 years. Small cap had a run of consecutive years of leadership, from 1979 through 1983, in the time before publication of the index. Since then, the US small cap index actually underperformed the large/mid cap by 1.3% annually.

Figure 1 shows a bar chart of the Annual large/mid and small cap US equity returns.

Relative performance between large/mid cap and small cap indices is further illustrated in Figure 2, which shows the rolling 12-month total returns of the Russell 2000, minus the Russell 1000, from 1979 through 2013. In the past, there have been prolonged leadership cycles with small cap stocks leading in the early 1980s, early 1990s and for most of the first half of the next decade. Before 2000, there were several peaks with relative annual performance in absolute terms exceeding 30%. Since 2007, on the other hand, in the “risk on/off” environment, leadership cycles by capitalisation have been less persistent and relative annual returns stayed comparably muted, with peaks around half that. Figure 3 shows that small cap stocks have however consistently exhibited higher volatility than large/mid cap stocks. On average, the small cap index was 1.3 times more volatile than the large/mid cap, while over the time-period from 1978 as a whole, this additional risk has not been rewarded.

Results for Europe and the United Kingdom

How does the small cap effect vary across regions? The look at indices from a global family of equity indices based on a consistent methodology might give some insights. MSCI provides such indices with a total return history for small cap since 2001 and we now compare the results for developed markets, USA, Europe ex-UK and UK. It cannot be expected that the MSCI USA results completely match the previous ones of the Russell indices. One key issue is which stocks are considered large/mid cap and small cap by the index providers. Here, Russell’s simple 1000+2000 stocks approach compares to a more elaborate MSCI methodology.

MSCI considers the two separate goals of “global size integrity” and consistent market coverage in its methodology for splitting stocks into large/mid cap and small cap. “Global size integrity” means that the split should ideally occur at the same company size for all developed countries and at a second common company size for all emerging countries. Consistent market coverage means that the percentage of a country’s overall market capitalisation that is considered large/small cap is the same for all countries. As both goals can’t be achieved simultaneously, MSCI gives precedence to global size integrity but allows for ranges at which the split between large/mid cap and small cap occurs. The ranges are derived from an overall target coverage for large/mid cap - the developed market range is currently between 2.41 and 5.54 billion. Similarly there is a range from 236 million to 542 million for stopping developed market companies from being included in the small cap index.

Figure 2 shows a graph of the rolling US large/mid cap versus small cap returns. Figure 3 shows a graph of US Equity annualised volatility.
Figure 4 shows a bar chart of MSCI sector weights.

As a consequence, there are currently 611 stocks in the MSCI USA index, 1823 stocks in the MSCI USA Small Cap index and the small cap market capitalisation is about 17% of that of the large/mid cap. This compares with a ratio between Russell 2000 and Russell 1000 of 11%. For the MSCI definition of small cap, the size effect has been considerably larger than for the Russell indices – since 2001 the MSCI USA Small Cap index has outperformed the MSCI USA by 5.4% annually compared to 3.0% for the Russell indices. Figure 5 shows the 12 month rolling relative return of small vs. large/mid cap for developed markets and the regions USA, Europe ex-UK and UK. It demonstrates that the small cap effect for these regions – while varying in intensity – is highly correlated and strongly cyclical. Since 2001 the size effect has varied between 5.3% annually for developed markets to 6.3% for Europe ex-UK; correlations exceed 0.75.

Figure 6 shows the relative volatility of small cap compared to large/mid cap for main regions. The chart demonstrates that there is considerable volatility in this ratio over time and regions. In the US, small cap has consistently been more volatile, as with the Russell Figure 6 shows the relative volatility of small cap compared to large/mid cap for main regions. The chart demonstrates that there is considerable volatility in this ratio over time and regions. In the US, small cap has consistently been more volatile, as with the Russell indices, and occasionally by twice as much as large/mid cap. In contrast, for Europe ex-UK, there have been prolonged periods when small cap was less volatile than large/mid cap. More often than not, however, small cap are more volatile than large/mid cap across regions.

Figure 5 shows a graph of rolling MSCI large/mid versus small cap returns. Figure 6 shows a graph of MSCI large/mid versus small cap volatility.
Figure 7 shows a bar chart of the 12 month forward MSCI PE ratios. Figure 8 shows the trailing MSCI price to book ratios.

One reason might be the sector exposure. Figure 4 shows the current sector weights of the MSCI indices, showing that the small cap index has consistently less weight in the low volatility Consumer Staples and Health Care sectors, while there is more weight in the Industrials and Consumer Discretionary sectors, which have above average volatility. We might expect some possible anomalies in the UK at some stage, as its large cap indices have a disproportionately high weighting in the potentially volatile oil sector.

Annual leadership by small cap or large/mid cap has been fairly evenly split

Figures 7 and 8 show some valuation measures – the 12 month forward price to earnings (PE) ratio, and the trailing price to book (PB) ratio. Comparing current levels with 10-year averages, the relative valuation of small vs. large/mid caps differs depending on the ratio. The PE ratio looks less stretched for small caps, while, on a PB basis, small caps look more expensive. The only relative reading that sticks out is that US small caps have become rather expensive (more than one standard deviation) on a price to book basis. As a summary regarding the small cap effect, the recent 5.3% to 6.3% effect for the MSCI indices looks promising, but this is based on roughly one decade of history only. The longer history for the Russell indices demonstrates that size leadership is probably still cyclical and it remains to be seen when large cap leadership re-emerges. On the other hand Figure 9 demonstrates that regarding relative performance, the Russell 2000 is 50% below its 1983 peak.

Figure 9 shows a graph of small cap versus large/mid cap total return performance.