Large-/Small-Cap Rotation Model: We Now Favor Small-Caps

Large-/Small-Cap Rotation Model: We Now Favor Small-Caps

 Our model now favors the small-cap segment for the first quarter of 2012.
 Three out of five model factors favor small-caps.

Large-Cap/Small-Cap Outlook for the First
Quarter 2012
At the beginning of January 2012, our model was forecasting the Russell 2000
Index should outperform the S&P 500 through March 2012.

Historical Performance of the Model: Outperforming the Indices So Far
In Figure 1, we show the historical quarterly performance of the out-of-sample backtested
Large-Cap/Small-Cap Rotation Model from the first quarter of 1993 through the
fourth quarter of 2011. We plot the cumulative total return index level of the portfolio
obtained by investing its full value in the S&P 500 if the forecasted quarterly return
difference between the large-cap and small-cap indices was positive; and in the
Russell 2000 if the forecasted return difference was negative. For comparison, we
also show the quarterly total return of the S&P 500 and the Russell 2000 indices. We
assigned a value of 100 to each index at the end of the fourth quarter of 1992.

The average quarterly return of the portfolio for the 76-quarter period was 3.08%,
while the average quarterly returns of the S&P 500 and Russell 2000 indices were
2.26% and 2.51%, respectively.

Description of the Model: Five Key Inputs
Our Large-/Small-Cap Rotation Model uses multiple linear regression to predict
whether the large-cap S&P 500 Index will outperform (or underperform) the smallcap
Russell 2000 Index the next quarter. The five independent variables with the
most explanatory power, listed in order of statistical significance, are:

No. 1. Classification Tree Variable — Our Rotation Model uses both classification
tree analysis (CTA) and multiple regression analysis. An important feature of the
Rotation Model is its use of the output from the CTA as input to the multiple
regression analysis. We found that the following three variables had the greatest
power to explain the historical quarterly difference between the total returns of the
S&P 500 and Russell 2000 within the CTA:

1. The quarter-end dividend yield spread between the S&P 500 and the Russell
2000
2. Prior one-month Russell 3000 total return
3. The quarter-end yield curve, measured as the difference between the Ibbotson
Associates long-term U.S. Treasury bond and the three-month U.S. Treasury bill.

No. 2. The Russell 2000 to S&P 500 Forecasted Long-Term EPS Growth Rate
Ratio — We use the quarter-end ratio between the I/B/E/S median forecasted fiveyear
EPS growth rate of the Russell 2000 Index to that of the S&P 500. We find that
this variable is positively correlated with the next quarter’s S&P 500 and Russell 2000
return spread. This suggests that this variable is a contrarian indicator such that, when
the long-term earnings growth forecasts of the large-cap (small-cap) segment become
overly optimistic, the small-cap (large-cap) segment performs better.

No. 3. 12-Month Broad Market Momentum — This variable is the total return of
the Russell 3000 over the last 12 months. It is positively correlated with the next
quarter’s S&P 500 and Russell 2000 return spread, suggesting that large-caps are
more likely to outperform small-caps in a medium-term bull market.

No. 4. One-Month Change in Credit Spread — This variable is the month-end
difference between the yield of the Citigroup High-Yield Bond Market Index (10-plus
years to maturity) and the yield of the 10-year U.S. Treasury note minus the prior
month-end difference. It is positively correlated with the next quarter’s S&P 500 and
Russell 2000 return spread, which suggests that large-caps outperform small-caps
when the credit spread widens.

No. 5. One-Month Change in Short-Term Interest Rates — This variable is the
month-end yield of the 3-month U.S. Treasury bill minus the prior month-end yield. It
is positively correlated with the next quarter’s S&P 500 and Russell 2000 return
spread, which suggests that large-caps outperform small-caps when the short-term
interest rate increases.

The five independent variables in the model are:
1. Classification Tree Variable
2. The Russell 2000 to S&P 500: Forecasted EPS Growth Rate Ratio
3. 12-Mo. Broad Market Momentum
4. 1-Mo. Chg in Short-Term Int. Rates
5. 1-Mo. Chg in Credit Spread