Exhibits V and VI show the evolution of portfolio variances and their composition between individual security variance effects (main diagonal) and covariance (off-main diagonal) effects.
A further distinction is made between the proportion of portfolio variance due to variances,
Comparison of unhedged returns and hedged returns of local portfolios in Tables 3 and 4 shows that the hedging strategy leads to a significant reduction of the hedged portfolio variances
compared to those of unhedged portfolios.
We then use these idiosyncratic returns to estimate CEO portfolio variances, incentive ratios, percentage change in total compensation, and variance ratios, and re-estimate the results in Tables 2, 3, and 4.
The first column shows that the median CEO receives very few incentives from variation in his cash pay relative to the variation in his portfolio, and that cash pay variance is small relative to portfolio variance for most CEOs.
To address this concern, we re-estimate the regressions in Columns (1) and (2) using as the dependent variable the ratio of total pay variance to total portfolio variance summarized in Column (2) of Panel A of Table 2.
Table 3: Portfolio Variances of Low versus High Exposure Portfolios relative to Benchmark (1) (2) (3) (4) (5) In-sample B L (L-B/B) H (H-B/B) Global Portfolios global 19.4 16.1 (-17) 27.9 (44) country 19.4 14.2 (-27) 26.3 (35) industry 19.4 19.3 (-0) 20.0 (3) Country Portfolios U.S.
It gives the average portfolio variance as the number of stocks in a given portfolio increases from one to forty, expressed as a percentage of the average variance of all individual stocks in our sample.
Out-of-sample comparisons of portfolio variances indicate that the error correction hedging model outperforms the conventional method, with improvements averaging about 2%.
To formally compare the performance of each kind of hedge, we follow the procedure adopted in Kroner and Sultan (1993) by evaluating the following in-sample portfolio variance
(3) Figure 2 is constructed using portfolio variances for the full sample period and thus captures the average importance of region versus within-region country effects over time.
It gives the average portfolio variance as the number of stocks in a portfolio increases from 1 to 40, as a percentage of the average variance of all individual stocks.