In the paper, Multinationals and the High Cash Holdings Puzzle, which was recently made publicly available on SSRN, we investigate whether the cash holdings of American companies are abnormally high after the financial crisis and whether these cash holdings can be explained by the theories summarized in the previous paragraph. We show that the extent to which cash holdings are unusually high after the crisis depends critically on the measure used. We would expect larger firms to hold more cash. Since corporate assets tend to grow over time, the dollar amount of cash holdings would grow even if the ratio of cash to assets stays constant. Consequently, at the very least, cash holdings should be measured relative to a firm’s assets. Using all non-financial and non-regulated public firms with assets and market capitalization greater than $5 million per year, the average cash/assets ratio is 20.18% in 2009-2010 compared to 20.50% in the 2004-2006 pre-crisis period. However, when we consider the median ratio, it is higher by 0.87% in 2009-2010 than in 2004-2006. Similarly, the asset-weighted ratio is higher by 0.74% in the recent period. The larger increase in the asset-weighted ratio than in the equally-weighted ratio suggests that large firms increased their holdings more and we show that this is the case. However, the changes in cash holdings from 2004-2006 to 2009-2010 are dwarfed by the changes in cash holdings from 1998-2000 to 2004-2006. Over that latter period, the average cash/assets ratio increases by 3.77%, the median by 6.39%, and the asset-weighted average by 3.62%. When we distinguish between private and public firms, we show that there is no evidence of an increase in the cash/assets ratio for private firms.
Having established that cash holdings are much higher during the 2000s than in the late 1990s, and higher after the crisis than before for some measures, we investigate whether this increase in cash holdings is a U.S. phenomenon or a worldwide phenomenon. Cash/assets ratios increase across the world in the 2000s compared to the late 1990s. However, the increase in average cash holdings across the world is smaller than the increase in average cash holdings in the U.S. A striking result is that median cash holdings in the US in the late 1990s are lower than median cash holdings in foreign countries, but the opposite is true by 2010. As for U.S. firms, there is little evidence of an increase in average cash holdings from the pre-crisis period to the post-crisis period for foreign firms.
While univariate statistics of the cash/assets ratio are useful to understand how cash holdings have changed over time and differ across countries, they are of little help in assessing whether the cash holdings are somehow abnormal or unusual. There is a vast literature that explains that cash holdings are related to firm characteristics. Among other results, this literature shows that firms with more growth opportunities hold more cash, firms with more uncertain prospects hold more cash, and firms with higher capital expenditures hold less cash. Over time, firms drop out or enter the sample, the characteristics of firms that stay in the sample can change, and macroeconomic conditions evolve. As a result, comparisons of cash holdings that do not take into account firm characteristics may be comparisons across firm populations with different characteristics, so that it is possible that cash holdings have changed simply because firm characteristics have changed. If this were the case, there might be nothing abnormal about the large cash holdings of American firms in recent years. Bates, Kahle, and Stulz (2009) show that changes in firm characteristics explain much of the increase in cash holdings in the 1980s and 1990s.
To evaluate whether the cash holdings of American firms are abnormal, a benchmark is required. We have to estimate what the cash holdings of American firms would be under normal circumstances. Ideally, we would have good theoretical models that would predict cash holdings given the known motivations for firms to hold cash. Unfortunately, such models do not exist. The alternative approach is to use an empirical model that captures cash holdings well before the 2000s. The approach of using an empirical model to assess normal cash holdings is widely used in the literature on cash holdings following its introduction by Opler, Pinkowitz, Stulz and Williamson (1999). With such a model, the empirical analysis evaluates how cash holdings differ from the cash holdings predicted by that model. One approach would be to estimate such a model of cash holdings for U.S. firms in the late 1990s and then use that model in the 2000s to estimate abnormal cash holdings. The problem with this approach is that it does not make it possible to evaluate whether the high recent cash holdings of U.S. firms are an American phenomenon or a global one. The implications of the high cash holdings of American firms would be quite different if American firms behaved like similar firms in other countries. If that were the case, many of the explanations for the high cash holdings of U.S. firms would make no sense. In particular, features of the U.S. tax code or of the U.S. regulatory system could not explain these holdings.
We therefore use a model that explains cash holdings of firms across the globe. We find that at the end of the 1990s, the holdings of cash of foreign firms are lower by 1.01% of assets than the holdings of comparable U.S. firms. Estimating this model every year, we find that as the 2000s evolve, the gap between foreign firms and comparable U.S. firms increases steadily, so that in 2009-2010 foreign firms hold 3.19% less of their assets in cash than comparable U.S. firms. Strikingly, U.S. firms increased their abnormal cash holdings relative to comparable firms from the Eurozone. We would expect that macroeconomic uncertainty concerns would be greater within the Eurozone, so that this result makes it questionable that the high cash holdings of U.S. firms can be explained purely by U.S. macroeconomic uncertainty. A concern with this approach is that the regression model will fit cash holdings over time. On average, cash holdings will not be abnormal across the globe. Hence, such a model could distort the extent to which patterns of cash holdings have changed as it will try to explain them using firm characteristics.
Our sample has 45 countries and for 82% of these countries we find that in the 2009-2010 period firms have lower cash holdings than comparable U.S. firms. While there are some types of countries where firms held as much cash as U.S. firms at the end of the 1990s, there is no type of country other than tax havens where firms hold more cash than U.S. firms after the crisis. In particular, U.S. firms hold more cash than comparable firms whether these firms are in developed countries or not, in common law countries or not, in countries that tax income worldwide or not, and in Eurozone countries or not. Additionally, we use the cross-country comparisons to investigate whether the change in cash holdings of firms in a country is related to how much that country was affected by the financial crisis. We find that cash holdings increase more in countries that were less affected by the crisis.
An alternative way to use the regression model is to estimate over time the cash holdings firms would have if they held cash as they did in the late 1990s. This approach produces estimates of abnormal cash holdings relative to the cash holdings firms would have if the determinants of cash holdings were the same as in the late 1990s. With this approach, we again find that U.S. cash holdings become abnormal over time. Strikingly, foreign firms hold abnormally low cash during much of the 2000s. While American firms have abnormal cash holdings of 1.86% of assets on average during 2009-2010 relative to the 1998-2000 benchmark, the average abnormal cash holdings of foreign firms are -0.14%.
We investigate various hypotheses that have been advanced for the increase in cash holdings of U.S. firms. The hypotheses discussed in the first paragraph of this section imply that different types of firms have different changes in cash holdings. We therefore focus much of our analysis on comparing the evolution of cash holdings for different types of firms. For each firm-type classification, we assign firms before the start of the period we consider. Except for large firms, high capital expenditure firms, firms with poor governance and financially constrained firms, all types of firms we consider have positive abnormal cash holdings after the crisis. Since firms with high capital expenditures do not have abnormal cash holdings, our results would seem to suggest that low investment is part of the explanation for high cash holdings. However, the high investment firms do not experience a change in abnormal cash holdings throughout the sample period, so that their lack of abnormal cash holdings after the crisis is not different from their lack of abnormal cash holdings before the crisis. The agency hypothesis for the increase in cash holdings suggests that firms with poorer governance should have a greater increase in cash holdings. We find the opposite result when we measure governance by the GIM index. Further, firms whose stock price was more adversely affected by the adoption of Sarbanes-Oxley did not increase their cash holdings more from the late 1990s to the period after the adoption of Sarbanes-Oxley than the other firms. While abnormal cash holdings increase for multinational firms, they do not increase for purely domestic firms. Consequently, it is possible that the increase in abnormal cash holdings of U.S. firms could be explained by the tax treatment of repatriations.
Before examining the cash holdings of multinationals in more detail, we investigate how the cash holdings of various types of U.S. firms compare to the cash holdings of similar types of foreign firms. We find that typically the cash holdings of U.S. firms are higher than those of comparable foreign firms. For instance, American firms that have high R&D expenditures have higher abnormal cash holdings than similar foreign firms. Similar results obtain for multinational companies. However, the abnormal cash holdings of American firms are not higher than those of foreign firms for the most profitable firms and for the firms that have the highest capital expenditures. Further, large foreign firms have higher cash holdings than similar U.S. firms. For all types of firms except for the firms with the highest cash flows, with the highest capital expenditures, and the most reliance on equity financing, abnormal cash holdings of American firms increase during the 2000s compared to abnormal cash holdings of foreign firms. Strikingly, the increase in cash holdings of U.S. multinationals compared to foreign multinationals is comparable to the increase in cash holdings of U.S. multinationals compared to U.S. purely domestic firms.
We find that U.S. multinationals held comparable amounts of cash than purely domestic firms in the late 1990s, but now hold significantly more cash than similar purely domestic firms. Since the increase in cash holdings of U.S. firms is concentrated among multinational firms, we investigate why the cash holdings of U.S. multinationals increased so much and whether the increase is explained by the tax treatment of repatriations. Foley, Hartzell, Titman, and Twite (2007) show that the tax treatment of remittances makes it advantageous for multinationals to keep their earnings abroad and they find that firms for which repatriation is more costly hold more cash. Our findings suggest that the tax costs of repatriation are not the whole story for the increase in cash holdings of U.S. multinationals in the 2000s. We find that the Homeland Investment Act of 2004, which was designed to reduce momentarily the tax cost of repatriation, failed to reduce the cash holdings of multinational firms. There are at least two explanations for this finding. First, it could be that the incentives of the Act were insufficient to affect firms’ cash holdings. Second, the repatriation tax costs could affect more where firms locate their cash rather than how much cash they hold. We expect that the tax cost of repatriation would be more important for high cash flow multinationals, but empirically these multinationals do not hold more cash than low cash flow multinationals. The increase in cash holdings of multinationals is strongly related to their R&D intensity, so that multinationals with no R&D expenditures do not have an increase in abnormal cash holdings compared to domestic firms with no R&D expenditures. Further, a striking result is that, among high R&D spending firms, firms that were already multinationals before 1998 do not hold more cash now than firms that were purely domestic firms before 1998. Among these firms, cash holdings increase sharply for multinationals relative to purely domestic firms, but that is because the cash holdings of multinationals are becoming more comparable to the cash holdings of purely domestic firms. Finally, and perhaps most importantly, we find no evidence that firms that become multinationals start holding more cash after they become multinationals. It appears that firms that become multinationals are firms with attributes that lead them to hold large amounts of abnormal cash even before they become multinationals.
The full paper is available for download here.