Equity Valuation and Portfolio Management

STUDYING THE NAIVE PORTFOLIO

We next examine what kinds of country bets result from the naive stock selection strategy. Specifically, we measure the extent to which the resulting country exposures are consistent (i.e., line up) with the country selection portfolio. We obtain these measures by decomposing the portfolio's holdings into three components: country selection, country noise, and pure stock selection.

For each of the two investment strategies, value, and momentum, we perform the decomposition by a two-stage regression. In the first stage, at each month t, we run a cross-sectional regression of the stock selection portfolio's holdings on the holdings of the country selection portfolio, across all stocks j:

The intercepts are all zero since the weights are constructed to sum to zero. In this regression, b_naive(t) measures the degree to which the portfolio makes country bets that are correlated with those of the country selection portfolio.

The residuals, e_naive(j,t), represent the positions in the portfolio that are not correlated with the country selection portfolio. If we assume that the country selection portfolio represents the bets that are most consistent with the investment approach (i.e., choosing country baskets based on value and momentum), then any other country bets may be regarded as “unintended.”

The second step of the decomposition isolates these unintended ...