The explanatory variables thus far are all measured at the firm level. In many cases, this is because they measure the firm’s access to the capital market (e.g. firm size, asset tangibility, and the existence of a bond rating). However, the intensity and type of investment of the subsidiary may differ from that of the entire firm. To test if the firm-level variables are such noisy measures that they drive the precautionary savings variables to zero, we rerun the regression only using observations where the foreign income is above the median (see Table 3, column VIII). For these observations, the firm-level variables are more closely associated with the subsidiary-level variables. The explanatory power of the precautionary saving variables does not improve, and in two cases, it shrinks.