s c o r e o n 2 i s g r e a t e r t h a n 2 . T h e c o n t r i b u t i o n o f t h e m a r k e t p o r t f o l i o i n t h e r e g r e s s i o n s i panels D and E is smaller and less significant than that of the TVPFM predicted returns. n
These tests lead to two initial conclusions. First, for the sample period of weekly returns examined, there is little evidence that the CAPM provides a good measure of the variation in sector returns. Second, failure to allow for time variation in return sensitivity to macroeconomic information can lead to the spurious conclusion that lagged macroeconomic information is not
predictive component that can be captured in part by the TVPFM model developed here.
A Dynamic Trading Strategy Based on the TVPFM
Description of Trading Strategy
Given the strong evidence of a predictable component in sector returns priced by the TVPFM, it is of interest to see if the step ahead model forecasts can be exploited profitably. To test for this possibility, we propose a basic trading strategy of sorting the ten S&P 500 sectors based on the predicted returns of the TVPFM. At the end of each period t, a long position will be purchased in the weighted constituents of the sector with the highest predicted positive return. Like wise, a short position is taken in the weighted constituents of the portfolio with the lowest predicted negative return. We impose the constraint that no long position will be taken in a sector portfolio with a negative predicted return and no short position will be taken in a sector portfolio with a positive predicted return.
Returns for the dynamic sector allocation strategy are calculated assuming available capital at each period t is evenly distributed between a long portfolio that is purchased, and in collateral against the portfolio of shares being sold short. In the event of a vector of all positive (negative) predicted returns, only a long (short) strategy will be pursued with one half the amount of capital at risk as in the case of balanced long and short portfolios.
In addition to testing the trading strategy on the model developed in Section 2, we also investigate the profitability of four additional lagged factor model specifications. The alternative