Corporate Cash Holdings and the Cross-Sectional Variation In Asset Returns1 Shane Shepherd Anderson School at UCLA First Draft: June 2004 This Draft: June 2007 Abstract: The level of cash holdings varies significantly across firms. As a low-returning asset, previous literature has theorized that holding excess cash should harm firm value. We find that this appears to be true amongst large firms. However, amongst small firms, we find that market returns rise with cash ratios (cash/assets or cash/market cap). These higher returns are not due to any known risk factors. Additionally they are accompanied by superior operating performance – higher growth in assets and growth in cash flow. Evidence supports the theory that the outperformance results from the easing of financial constraints upon these firms. The results are most robust for small, growth-oriented firms, who have the most difficulty accessing capital markets and could benefit the most from holding cash. 1 Shane Shepherd is with The Anderson School at UCLA and Research Affiliates, LLC. This paper serves as part of my dissertation at UCLA. I would like to acknowledge the helpful comments of Antonio Bernardo, Walter Torous, Marc MartosVilas, Brad Barber, Terrance Odean, Richard Roll, and Mark Garmaise. 1 Electronic copy available at: http://ssrn.com/abstract=1084552 1. Introduction A high ratio of cash on a firm’s balance sheet is typically viewed as a poor investment and a drag on returns. However, in this paper we present evidence that a firm’s cash stockpile, or lack thereof, can strongly and positively influence its market returns and operating performance. Particularly amongst small and growth oriented firms, and those with poor access to the capital markets, we find that firms with high levels of cash significantly outperform their cash-poor peers. A discrepancy exists in the theoretical motivations for cash holdings. Jensen’s (1986) famous free cash flow theory explores the agency conflict between stockholders and management. Managers have many incentives, both personal and organizational, to increase the size of their firm. This widens the resources under their control, often increases their compensation, and delivers greater public recognition. In general, managers have much more control and discretion over cash holdings, as opposed to outside equity and debt financing, which is subject to board scrutiny. Jensen’s theory predicts that managers will be likely to squander leftover cash on negative-value projects, such as ill-advised takeovers or empire-building attempts. If this behavior is prevalent then stockholders would clearly prefer their managers to return excess cash through dividends or share repurchases, and either penalize cash-rich companies with lower valuations, or attempt to oust the current managers. 2 Electronic copy available at: http://ssrn.com/abstract=1084552 However, other theories explore the benefits from holding liquid assets. This benefit primarily arises because of the wedge between internal and external finance. Myers and Majluf (1984) construct a transactions cost model in which firms face time-varying and perhaps prohibitive costs to raising capital. In some cases, firms may be entirely shut out of the external capital markets and will have to forego positive NPV projects due to lack of financing. In this case, a cash stockpile acts as a buffer for rainy days, allowing firms to smooth their investing behavior over time. This theory predicts that firms more likely to face constraints on their financing capabilities should be more like to hold high levels of cash reserves. Of course, these two stories are not mutually inconsistent. In a large cross section of firms, cash could hold value for some firms and be wasted by others. Firms most likely to extract value from holding cash are small growth firms, firms with volatile growth options, firms with a high cost of external capital, and firms subject to financial constraints by the capital markets. Larger firms that do not face transactions costs to raising capital, and have fewer growth options, should have lower benefits to holding cash. It is not obvious that cash-rich firms are the ones most prone to the agency costs discussed by Jensen. High free cash flow does not necessarily equate to high cash holdings. Consider three types of firms: the first type is cash-constrained and cannot fully invest in its positive NPV projects. This firm will not hold excess cash. The second type has more than enough capital to invest in positive NPV projects and then, in keeping with 3 Jensen’s prediction, wastes its excess cash on negative NPV projects (or, under a happier scenario, distributes it to shareholders). This firm will also not hold much cash, but for a different reason. A third class of firm also enjoys excess cash flow, but accumulates the capital rather than wasting it on the present span of negative NPV projects – perhaps waiting to invest in future positive NPV projects. In this situation, only the third type of company becomes cash-rich. Rather than predicting imminent agency costs, high cash holdings may instead indicate a history of skilled and disciplined management; firms with high cash holdings find themselves in that situation precisely because they have skillfully created excess cash flow in the past, and (more to the point) not invested it poorly but possibly saved it in anticipation of future investment opportunities. 2. Literature Overview Previous empirical investigations of cash-rich firms find that, in general, cash-rich firms are precisely the ones that stand to benefit from financial slack. Opler et al. (1999) find that firms with strong growth opportunities and risky cash flows tend to hold a large percentage of their total assets in cash, and larger firms and firms with better credit ratings tend to hold low levels of cash. Firms with time-varying growth opportunities put these cash hoards to good use; Harford, Mikkelson, and Partch (2004) find that cash-rich firms tend to invest more than their industry average during economic downturns, and this results in higher post-downturn sales growth and superior operating performance. Looking across the business cycle, Mikkelson and Partch (2003) find that firms with large persistent cash reserves (those that held more than 25% of their assets in cash and 4 cash equivalents for a five-year period) have operating performance comparable to or better than a group of matched firms. Recent literature also holds evidence for agency costs to holding cash. Harford (1999) uses a probit model to predict takeovers and finds that excess cash holdings (defined as the residual from a regression which predicts optimal holdings) significantly increase the probability of a firm undertaking a value-decreasing merger. Furthermore, he finds that the stock price reaction to the announcement of a bid is decreasing in the amount of excess cash held by the bidder. Investigating firms that hold cash but are not necessarily financially constrained, Blanchard, Lopez-de-Silanes, and Shleifer (1994) find that their small sample of firms who receive an unexpected windfall tend to waste this excess cash. If cash holds greater value for some firms than for others, then the market value of liquid assets should differ across firms. Indeed, Pinkowitz and Williamson (2002) find a large spread in the cross-sectional value of cash holdings. The market prices the average dollar of corporate cash at $0.97, but they find a tremendous spread, with the value of a dollar ranging from $0.27 to $1.84 across firms. Cash is valued below par for firms with poor growth opportunities – those with predictable investment options – and those with large stockholder-bondholder conflicts (for example, firms close to bankruptcy). These are the firms likely to fall prey to Jensen’s cash-wasting hypothesis. Those with high and variable growth options have their cash valued at a premium. These are the firms most likely to benefit from the advantages provided by financial slack from their internal capital markets. But perhaps surprisingly, the market values cash at a strong discount for 5 firms that appear to have poor access to the capital markets, and at a premium for firms that have good access to external finance. In the first direct study of the effect of cash upon returns, Greenwood (2004) finds aggregate corporate cash holdings to be a negative predictor for future market returns. He argues that cash accumulation is a sign of active market timing by managers, in which overvalued firms issue equity to take advantage of inflated valuations. However, this study only tangentially addresses the issue of returns to cash-rich firms in the crosssection. 2. Data A firm’s level of cash holdings obviously depends largely upon firm size. Thus, to compare firms in the cross-section, some normalization is needed. We employ three metrics, with largely similar results: cash and short-term investments divided by book equity, cash and short-term investments divided by market capitalization, and cash and short-term investments divided by total assets. From the Compustat industrial quarterly files, we gather information on cash and short-term investments (Data item 36), stockholder’s equity (Data item 60), and the report date for the quarterly earnings. We restrict our analysis to firms for which we can identify the report date, to ensure that all conditioning information is publicly available at the portfolio formation date. We further restrict the sample to non-financial firms (eliminating all SIC codes between 6000 and 7000) because financial ratios are largely meaningless for such firms, and to United States common equity (share codes 10 and 11 in CRSP). From CRSP, we gather monthly 6 returns and end-of-month market capitalizations for each company in the sample. We calculate each of the three cash ratios for each firm in month t using the most recently available information, and rank firms into deciles based upon these ratios. Summary results appear in Table 1. These rankings provide a sizeable spread of cash holdings. The smallest decile of firms holds an average of 3 million in cash and short-term investments, while the largest holds around 400 million (although somewhat less for the cash/assets metric). The mean (median) ratio increases across all metrics from zero to 215% (103%) of equity, 129% (74%) of market cap, and 60% (60%) of total assets. However, there are also obvious differences in the size and book to market loadings across the cash deciles. Both the lowest and the highest decile are composed of smaller stocks, with the middle deciles drawing larger market capitalizations. The cash/assets ranking gives the clearest view of the book-to-market effects. Because book value serves as the denominator of the cash/equity group, and market value serves as in the denominator of cash/market cap, the book value ranking will mechanically push low book-to-market stocks into the high deciles, and the market cap ranking will mechanically push low market cap stocks (and therefore high book-to-market stocks) into the high deciles. This pattern is clear in Table 1. Since the cash/assets ranking is less intricately involved with the book-to-market variables, it provides a clearer view of the true book-to-market effects. As the Myers/Majluf theory predicts, the book-to-market variable steadily decreases across the deciles, and therefore the growth options (measured 7 as Tobin’s Q) of the company appear to increase with cash holdings. Firms identified as cash-rich tend to be small, growth-oriented companies – precisely the sort that would face frictions in raising external capital and that would benefit the most from financial slack provided by cash holdings. We provide direct evidence of this later. To be sure that the portfolio returns are not picking up the well documented size or bookto-market effects, we do a second series of rankings that will mitigate the differences in these characteristics across portfolios. Each month, we place each firm into one of three size groups based upon the Fama/French monthly 30/40/30 NYSE breakpoints, and also into one of three book-to-market groups based upon the 30/40/30 NYSE breakpoints. We then construct three new decile rankings with a two-stage sort, ranking within each of these nine subgroups on each of the three cash ratios. This controls for a good portion of the size and book-to-market effects found across the deciles. Table 2 shows the descriptive statistics for these second-stage rankings. Unless otherwise specified, our analysis is conducted using these second-stage rankings. The second stage ranking is still very much determined by the cash holdings. Table 3 shows the correlation matrix for the six different decile rankings with one another and with the size and book-to-market rankings. The denominator for the normalization of cash holdings, and the order of ranking, makes little difference; the lowest correlation between the ranking deciles is 77%. And the correlations of the decile rankings with market cap and book-to-market ratio show that the second-stage sorts control much better for those effects. For example, the correlation of the book-to-market ratio with the cash 8 ranking variables drops from –14% (cash/equity), 20% (cash/market cap), and –14% (cash/assets) to –2%, 7%, and –3%. 3. Returns Results A. Calendar Time Returns We analyze the returns to the cash portfolios by estimating a calendar time series of returns to each portfolio. Stocks are placed into portfolios based upon the first-stage cash rankings each month from January of 1970 until December of 2001. Equally weighted monthly returns are then estimated for various holding periods (1 month, 3 months, 1 year, 3 years, and 5 years) 2. For all holding periods, the average monthly returns increase nearly monotonically as we move from low-cash firms to cash-rich firms. The largest impact comes in the first month, although cash-rich firms outperform even at long horizons. For a one-month holding period, the high-cash portfolio outperforms the lowcash portfolio by 88 basis points (equity ranking), 115 basis points (market cap ranking) and 64 basis points (assets ranking), and this difference is statistically significant at the 1% level for the first two groups. The one-year mean monthly outperformance is 54 basis points (equity), 98 basis points (market cap), and 48 basis points (assets). These results are not, in general, robust to value-weighted returns. Table 5 shows the value-weighted returns for each ranking strategy, and only one of the nine high minus low portfolios provides returns significantly different from zero at the 5% level. This suggests that small stocks are primarily driving the observed excess returns. 9 B. Three-Factor Model Although the equally weighted cash-rich firms clearly outperform the low-cash firms, it may be due to the correlated size and book-to-market factors. To control for beta, size, and book-to-market effects, we run a three-factor model time series regression on the monthly portfolio returns, following Fama and French (1993). We estimate the equation (R pt − R ft ) = α j + β j ( Rmt − R ft ) + s j SMBt + h jVMGt + ε jt , where Rpt is the monthly portfolio return, Rft is the monthly return on T-Bills, Rmt is the monthly return on a value-weighted market index, SMBt is the return on a value-weighted portfolio of small stocks minus the return on a value-weighted portfolio of big stocks, and VMGt is the return on a value-weighted portfolio of high book-to-market stocks minus the return on a value-weighted portfolio of low book-to-market stocks. The results appear in Table 6. The alphas from this regression increase nearly monotonically from the low-cash decile to the high-cash decile, with (for the 1-month holding period) the high-cash decile outperforming the low-cash decile by 97 basis points per month (equity ranking), 120 basis points per month (market cap ranking) and 97 basis points per month (asset ranking). Quite strikingly, the alphas increase nearly monotonically across the deciles, but somewhat surprisingly, betas do not significantly vary across the deciles 3. For all three rankings, the high-cash portfolios are weighted more toward small stocks than either the low or middle deciles, indicating that large 2 Results for the 3-month, 3-year and 5-year horizons are not reported. This result is puzzling, as extant theory predicts that an equity beta should reflect the weighted sum of the component asset betas. If we assume the beta on cash holdings to be zero, equity betas should decrease across the portfolio deciles as their cash holdings increase, but this is not found in the data. One possible explanation is that the beta of the noncash assets of firms increases along with the firms’ cash holdings. 3 10 stocks tend to hold lower levels of cash. And under all three rankings, the low-cash firms have a higher loading on the HML factor, indicating that they tend to be composed more of value stocks, while the high-cash portfolios contain more growth firm. But even so, the differences in these factor loadings cannot explain the returns results, as the risk-adjusted alphas remain highly significant and monotonic. C. Ordered sorts As a second method to control for alternative factors, we examine returns to portfolios formed on the ordered sort rankings. Table 7 shows the nominal returns to the secondstage ranking portfolios. The effect still increases nearly monotonically across deciles, and the cash-rich portfolio outperforms the cash-poor portfolio at the 5% significance level in all strategies except the cash/assets ranking, 1-month holding period. As before, using value-weighted returns weakens the results. Controlling for the size and book-tomarket factors actually strengthens the observed returns patterns. Rather than being the cause of the excess returns, the size and book to market factors seem to diminish the effect. While the cash-rich firms tend to be smaller (thus benefiting from the small-stock effect), they also tend to be growth firms (thus being handicapped by the book-to-market effect). The three-factor model regressions reported in Table 9 show similar risk-adjusted outperformance for the higher-cash deciles on the second-stage sorts. The highest and lowest decile portfolios both weight more heavily towards small stocks than the middle deciles, although the differences in loadings are not statistically significant, and the high 11 cash portfolios are heavily tilted towards growth stocks. However, the alphas still increase monotonically across the deciles and are highly significant. Neither the ordered sorts nor the three-factor model can explain the return performance of the portfolios; higher cash portfolios consistently outperform lower cash portfolios, and the returns grow nearly monotonically under every metric. D. Size and Book-to-Market Subgroups These outperformance patterns are not robust to value-weighted returns. This implies that the effect is much stronger for small stocks than for large cap stocks. If financial slack is valuable, it should be most valuable for firms that face the most frictions to external financing. As large firms have more resources, generally better credit ratings, more contacts and longer-lasting relationships with the banking industry, it seems likely that they will have an easier time gaining outside financing in difficult market conditions or poor credit environments. It also seems likely that small companies would benefit the most from holding cash to avoid the transactions costs of external financing. In this section, we directly examine the nominal returns to cash holdings amongst the Fama-French 30/40/30 size bins and the 30/40/30 book-to-market bins. Table 10 shows the returns to the size portfolios for the one-month holding periods. As expected, the returns to holding cash decreases as companies grow larger. Small cash-rich firms outperform their cash-poor counterparts by nearly one percent per month. By contrast, the gap is only 64 basis points amongst medium sized firms, and large, cash-rich firms actually underperform their cash-rich counterparts. Table 11 shows the returns to the 12 decile portfolios broken down by the book-to-market groups. While the 10 minus 1 spread does not decrease monotonically as we move from growth to value stocks (the middle group actually shows the greatest spread), there is evidence that the premium does not exist for value companies. While high-cash value stocks outperform low-cash value stocks, the difference is not statistically significant; the middle group is the only one that shows reliable outperformance. Consistent with the value effect, the value stocks outperform their relevant growth-stocks counterparts in every cash decile. E. Momentum Jegadeesh and Titman (1993) document a short-run momentum effect in stock prices, showing that previous winners tend to continue to outperform while previous losers tend to underpeform at the three-to-twelve month horizon. It is possible that companies that hoard cash tend to outperform prior to the ranking date, and that my rankings simply pick up this well documented momentum effect. To check the robustness of our results, we examine risk-adjusted alphas from a Fama-French three factor regression plus a momentum factor. 4 We report results for the one-month holding period, cash/equity and cash/market cap rankings in Table 12. (Results for other specifications are similar.) All of the equally weighted portfolios show a negative weighting on the momentum factor. This is likely due to the mechanical result of smaller stocks correlating negatively with momentum. (Given two otherwise equal stocks, the smaller one will be more likely to have experienced a lower return in the past.) The value-weighted results (not reported) show a 13 positive loading on momentum for the cash-poor portfolios and a negative loading for the cash-rich portfolios. Most importantly, in all specifications, the high-cash portfolios are generally less subject to the momentum factor than the low-cash portfolios, and the 10-1 difference portfolio shows a negative weighting on the momentum factor. As such, this raises the alphas from the three-factor Fama-French regressions. Momentum cannot explained the observed returns to the cash-rich portfolios. F. Industry controls Another possible explanation is that cash-rich firms tend to gather in disproportionately successful or disproportionately risky industries. If industries have heterogeneous and time-varying costs of equity, and if high-return industries tend to require high cash holdings, then our results may merely reflect this historical industry outperformance. We categorize each firm into one of the 48 industry groupings specified in Fama and French (1997). An examination of this breakdown shows that certain industries do indeed tend to require more liquidity than others: for example, manufacturing, defense, raw materials, and wholesaling tend to be low-cash industries, while pharmaceuticals, business services, computers, and healthcare are disproportionately represented in the high-cash industries. To explore whether this accounts for the observed return patterns, we rerun the calendar time portfolios and correct each firm’s monthly return by subtracting the corresponding industry return for that month. The results in table 13 show that industry differences cannot account for the return patterns; cash-rich firms tend to outperform their industry peers, while cash-poor firms tend to underperform. The 4 The momentum portfolio returns are taken from Ken French’s data library. 14 difference portfolios still deliver sizeable and statistically significant returns. Industry congregations do not seem to explain the observed returns. G. Post-earnings Announcement Drift Because the conditioning date for portfolio formation coincides with the quarterly earnings reports for each firm, information contained in the earnings announcement could influence the observed returns. However, the calendar-time return method we implement examines returns starting with the month subsequent to the month containing the earnings announcement date; there is a lag of at least one day and up to thirty days between the announcement date and the beginning of the return examination period, so any effect from the announcement itself will be excluded. Furthermore, the outperformance is robust to longer time periods; the longer holding periods show that the excess returns persist long after any news announcement should be incorporated into prices. However, the companies in the cash-rich deciles to seem to have, on average, more positive earnings reports than those in the cash-poor deciles. Returns to the portfolios for the three-day event window around the earnings announcement grow across the portfolios (see Table 14). The presence of a large cash position on a firm’s balance sheet could indicate that profits were unexpectedly high in the previous quarter, and that the firm has not yet invested or distributed the proceeds. Thus, the cash effect may be correlated with the post-earnings announcement drift effect documented by Bernard and Thomas (1990). Still, it seems doubtful that one positive earnings surprise could be enough to move a firm into the higher cash deciles. The median firm in the highest decile 15 holds 75% of its market cap in cash. Assuming an average price/earnings ratio of 20, a back-of-the-envelope calculation shows that a quarter’s worth of retained earnings would amount to only 1.25% of a firm’s market capitalization. At least on the high end, the deciles are clearly not determined by one quarter of earnings surprises. Additionally, there are at two important distinctions between the cash effect and postearnings announcement drift: first, the cash effect is not concentrated around earnings announcement dates; and secondly, the post-earnings announcement drift phenomenon predicts a return reversal in quarter t+4. To examine these conditions, we conduct an event-time analysis of the returns to the decile portfolios, examining market-adjusted returns for each day from 505 days before the earnings announcement until 505 days afterwards. Each day following the earnings announcement, we assign the portfolios a 1 if their market-adjusted return is positive, and a 0 for a negative market-adjusted return. The percent of days with positive marketadjusted returns appears in table 14. The returns are not concentrated in earnings periods, but rather spread across all days in the analysis. From day t+2 to t+275, 66% of the daily returns to the cash-rich portfolio are persistently greater than the market average, while only 39% of the cash-poor returns are greater than the market average. And the percent of outperforming days rises almost monotonically across deciles. Secondly, the outperformance does not reverse after a year. The event time study shows that the returns to the high-cash firms remain high around the earnings window four 16 quarters after the ranking; from day t+250 to t+275 (the month-long period one year after the original earnings announcement, which should include the next year’s earnings announcement), the high-cash portfolios systematically outperform the low-cash portfolios. This appears to be a persistent, long-run phenomenon. H. Subperiods and Seasonalities To be sure that a small subsample is not driving the returns, we examine three subdivisions of the data: calendar time returns by decades, by year, and by month. The results for the cash/market cap ranking are in table 15. The portfolio returns retain their general monotonic pattern in all three of the decade subgroups (although the difference between the top and bottom deciles is sizable but not statistically significant in the 1970s). The yearly returns show that the difference between the top and bottom deciles is positive for 23 out of the 30 years sampled 5. The seven negative return years average minus 78 basis points per month, while the 23 positive return years average 159 basis points. A binomial test rejects the hypothesis that the top and bottom deciles are equally likely to outperform one another with a p-value of 0.0026. For the monthly returns, only June and December give negative returns (averaging 75 basis points per month) while the remaining ten months give positive returns of 144 basis points per month. Even excluding January’s outsized returns, the remaining eleven months return an average of 58 basis points per month. A binomial test rejects the hypothesis that the extreme deciles are equally likely to outperform one another over the months with a p-value of 0.0193. It appears that this pattern persists across time and is not driven by seasonal effects. 5 The top decile ranked by cash/equity outperforms the bottom in 18 out of 22 years, and the top decile when ranked by cash/assets outperforms the bottom in 15 out of 22 years. 17 I. Takeovers Another possible explanation for the outsized returns could be due to a differing takeover rate across deciles. If cash-rich firms are more attractive takeover candidates, the premiums paid to these firms could explain their high future returns. Although Jensen (1986) theorizes that firms with large cash reserves are likely takeover targets, this hypothesis runs counter to the prevailing empirical literature: Pinkowitz (2004) finds that, over the 1985-1994 period, cash-rich firms are actually less likely to be targeted by hostile takeovers, and that when they are acquired, the premium is no higher than for their cash-poor counterparts. And Harford (1999) finds that not only are cash-rich firms less likely to be targeted, but they are more likely to make acquisitions, and this on average hurts shareholder value. We gather data on 2933 acquisitions of publicly traded companies from the Mergerstat database from 1992 through 1999. We first examine the takeover rate across deciles, and then estimate a logit model where the dependent variable is 1 if a company announces within the next 12 months that it has acquiesced to a takeover offer, and 0 otherwise. The independent variables are the cash/equity ratio 6, the log of market cap, and the book-tomarket ratio. Table 16 reports the results. In this sample, the cash-rich deciles are taken over at a higher rate than the other deciles, but the magnitude is small. While 4.8% of firms in the lowest decile receive a takeover offer within the next 12 months of the ranking, 5.7% of the highest decile firms receive an offer. Furthermore, the takeover rate 6 The results are robust to cash/market cap ratio. 18 does not increase monotonically across the deciles. The premium paid is lower than average for the low-cash companies, but does not seem to be related to cash holdings. (The takeover method, however, clearly depends upon cash holdings; stock buyouts are much more frequent for cash-rich firms than cash buyouts, and the opposite holds true for cash-poor firms.) If returns to the high and low cash deciles were otherwise equal, the difference in takeover rate and the premiums paid explains only 22 basis points per year of the difference between the extreme decile returns. While takeover premiums undoubtedly account for a small part of the outperformance of the highest decile over the lowest decile, they explain neither the magnitude nor pattern of the return differences. The logit analysis shows that, holding size and book to market constant, a cash position actually decreases the probability of receiving a takeover offer in the next 12 months. The coefficient on the cash/equity ratio is negative, although not significantly so. The bookto-market factor is positive and significant, and the log of market cap is negative but insignificant. Judging from the takeover rates across deciles, it is possible that the cash effect on takeovers is not linear. To test this possibility, we run the same logit regression with the log of the cash level and the decile ranking as independent variables. Both of those coefficients are also reliably negative. Any takeover premium given to the highcash portfolios seems more likely due to their status as smaller stocks, and might supplement the returns to the highest decile but cannot explain the pattern of returns. 4. Financing Constraints 19 One of the main reasons that firms may hold cash is to avoid costly external financing. At times, some firms may be so financially constrained that external financing is not available. These financially constrained firms may be financially sound, without any risk of default or bankruptcy, but lack the funds necessary to invest in all of their positive NPV projects. Recent literature has documented significant return patterns amongst financially constrained firms. Lamont, Polk, and Saa-Requejo (2001) use the Kaplan and Zingales (1997) measure to classify manufacturing firms, and construct a financially constrained risk factor. Firms with lower loadings on this factor tend to have higher returns. Somewhat surprisingly, facing financial constraint does not increase a firm’s risk premium, but rather lowers the expected return. Gomes, Yaron, and Zhang (2004) identify financially constrained firms through a GMM analysis on an Euler equation of optimal investment behavior. They find evidence of a common financing constraints factor, and that the costs of financial constraints are procyclical, and thus financial frictions are more important during good economic conditions. Finally, Pinkowitz and Williamson find that the market values a dollar of cash in a financially constrained firm at a significantly lower value than a dollar of cash in an unconstrained firm. A. Measuring Constraints Although investment-cash flow sensitivities have been widely used to identify financially constrained firms, beginning with Fazzari, Hubbard, and Peterson (1988), there is good reason to believe that this not a useful method. Kaplan and Zingales (1997) point out that 20 the relationship between cash flow and investment is likely nonmonotonic, and lacks good theoretical justification. They develop a measure based on qualitative information gleaned from SEC filings, augmented with quantitative data, which appears to successfully identify the degree of financing constraints faced by firms. We apply Lamont, Polk, and Requejo’s adapted method of Kaplan and Zingales (1997) to explore whether financing constraints influence firm returns in my sample. Lamont et al. report in Appendix A the results from an ordered logit model, based on Kaplan and Zingales’ measure, that identifies financially constrained firms using four ratios: cash flow to capital, debt to total capital, dividends to capital, and cash to capital. Taking data from the yearly COMPUSTAT files, we score all firms in our sample on this KZ metric. We then rank firms into quintiles each year based upon this score. Kaplan and Zingales report that around 15% of firms are financially constrained at any given time; therefore, we consider firms in the highest quintile for each year to be financially constrained, and the other 80% of firms to not be financially constrained. While this distinction is somewhat crude, it reflects the fact that firms in the highest quintile are more financially constrained, and is adequate for comparison purposes. Not surprisingly, the high-cash firms tend to be less financially constrained than low-cash firms. We find a –0.30 correlation between the KZ quintiles and the cash decile rankings. We report the full breakdown in Table 17. The percentage of each decile classified as financially constrained decreases monotonically, except for an uptick in the highest 21 decile. However, only 15% of the highest decile firms are categorized as constrained, below the global average. Lamont et al. find that unconstrained firms earn, on average, higher returns than constrained firms. Since the level of financial constraint drops as cash holdings rise, this factor could be in part driving the portfolio returns. Table 18 confirms that unconstrained firms in my sample earn higher returns. There is very little difference in returns for the first four quintiles, but a decline in returns for the fifth quintile; the groupings of constrained firms in the bottom 20% may be appropriate. The least constrained firms earn, on average, 1.46% per month; the most constrained firms return an average of 1.29%. This difference is not statistically significant. B. Financial Constraints Risk Factor We next construct a time-series of returns to a financial constraints (FC) factor by subtracting the lowest quintile return (least constrained) from the highest quintile return (most constrained) each month. We then augment a Fama-French three-factor model with this factor to see if the superior performance of unconstrained firms can explain some of the variation in returns. The results are reported in Table 19. While all portfolios show a positive loading on the FC factor, indicating all portfolios are at least somewhat constrained, the low-cash portfolios have the highest loadings, and the loadings generally decline as cash holdings rise. The difference portfolio has a strong and statistically significant negative loading on the FC factor. However, this loading does not seem to 22 make much difference in the excess returns; the alphas are little changed from those reported in the standard three-factor model in Table 9. C. Avoiding Financial Constraints with Cash Holdings Since cash-to-capital is one of the four inputs into the KZ measure, firms with high cash holdings should be, on average, less constrained than other firms. However, it may be that these firms choose to hold large amounts of cash because they would otherwise face financial constraints. To test this hypothesis, we rerun the KZ measure based only on the first three ratios, excluding cash holdings from the estimation. We then rerank firms into quintiles based on this secondary KZ score. Results are reported in Panel B of Table 17. The differences are striking. The percentage of financially constrained firms grows, instead of shrinks, across the cash holdings deciles. Furthermore, the percentage of highest decile firms that are considered financially constrained more than doubles, to 32.35%. There are 10,171 firm/quarters that would be considered financially constrained without their cash holdings, but use their financial slack to avoid this categorization in the full KZ measure. There is a significant overlap between these constraint-avoiders and the cashrich firms: a third of the constraint-avoiders fall into the highest cash decile, and 17.5 percent of the highest cash decile is a member of this group. If a benefit accrues to companies for holding cash to avoid financial constraints, it should accrue most strongly to these firms. To explore whether these firms are driving our results, we conduct some further analyzes upon this interesting subgroup. 23 First, Table 20 shows that the constraint-avoiders, as a group, experience superior market returns. An equally weighted calendar-time portfolio returns 1.71% per month with a one-month holding period, and 1.42% per month with a year holding period. This is superior to all of the cash holdings portfolio returns. Table 21 examples the alphas from a three-factor and five-factor regression of these portfolio returns on the market, size, and book-to-market factors (panel A) and the first three factors plus a momentum and financial constraint factor (panel B). The risk-adjusted alphas, with the exception of the one-year horizon value weighted portfolio, are highly positive and significant. Furthermore, these alphas are higher than those for any of the cash portfolios. Because the constraint-avoiders are concentrated in the higher cash deciles, any excess performance will have the strongest impact on the higher cash deciles. To examine the extent of this impact, we evaluate the performance of the cash portfolios after removing these constraint-avoiders from the sample. Table 22 shows that, while the general pattern of returns persists, the effect is much weaker and the monotonicity fails. Since our categorization of financial constraints is imperfect, and firms certainly face varying degrees of financial constraints, we would not expect to have removed all firms that benefit from cash holdings in this manner from the sample. However, it seems clear that removing those most likely to benefit deteriorates the results significantly. 5. Operating Performance 24 We next provide evidence that these cash-rich companies use their wealth to finance growth options, leading to higher earnings. In Table 23 we present operating performance figures for one year and two years following the ranking date. Investment is calculated as capital expenditures divided by lagged property, plant, and equipment; asset growth is the ratio of total assets to year-ago total assets; sales growth is calculated as the ratio of sales in year t to sales in year t-1; and cash flow growth is calculated as cash flow (income before extraordinary items plus depreciation) during year t to cash flow in year t-1. The cash flow growth figures are conditional on the firm having positive cash flow in both year t and year t-1. All ratios are negatively skewed, so we winsorize at the 1% tails before calculating means. There is strong evidence that higher cash levels lead to higher capital expenditures. Both the mean and median investment levels rise with cash ranking deciles, and the highest cash firms invest, on average, almost twice as much as the lowest cash firms. This trend continues into the following year. Furthermore, it seems these firms are investing in good projects; both sales and cash flow grow at faster rates as the portfolios increase in cash holdings. Finally, we present weaker evidence that this leads to a growth in assets. While the mean asset growth rises nearly monotonically with the cash deciles, there is little difference in the median asset growth. This additional investment allows some firms to do very well, but most do not show better growth in assets then their cash-poor peers. Additionally, we examine the operating performance of the constraint-avoiders group. Their investment level matches that of the top cash decile, and their sales and cash flow 25 growth is much higher than any other group. Their mean asset growth and median asset growth for year t also ranks as the highest, although there is a sharp dropoff in median asset growth in year t+1. The group of firms for which cash holds the most importance to investment policy also shows the best operating performance going forward. 6. Implementing a Trading Strategy Going long the cash-rich portfolios and shorting the cash-poor portfolio could deliver profits of around one percent per month. Trading costs to this strategy are relatively low compared to some arbitrage strategies. First, turnover is minimal. While a 1-month holding period delivers the strongest returns, the results are robust to 3-month and 12month holding periods. Since the conditioning data (quarterly earnings announcements) arrives every three months, a 3-month holding period strategy seems appropriate: this will reduce turnover and maximize the recency of the information. Secondly, there is an 83% autocorrelation amongst the cash rankings. This means that maintenance of the strategy will require some rebalancing every three months, but not major turnover. The bid/ask spread generates the largest cost to this trading strategy. As the effect is primarily manifested in small stocks, they may be illiquid and require heavy costs to trade. Using data from the CRSP NASDAQ-NMS file, we measure the end-of-day bid/ask spread for each NASDAQ stock in the portfolios for the first month following the ranking date. The bid/ask spread drops as cash holdings increase, ranging from 5.5% for the low-cash firms to 4.7% for the high-cash firms. A 40% turnover four times a year (20% on both the long and short portfolio), and a 5% bid/ask spread will result in costs of 26 8% per year to the portfolio. However, the 5% spread, based on historical NASDAQ figures, probably overestimates the true trading costs going forward. Rebalancing the portfolio yearly, with a spread at 3%, reduces returns by only 25 basis points per month. Finally, it may not be feasible to properly hedge the long position. The lowest decile, which should be sold sort, has an average market cap of 800 MM (cash/equity ranking) and 1.4 BB (cash/market cap). If sufficient shares of these smaller stocks are not available to short, the strategy will bear considerable market risk. Furthermore, the difference portfolios load heavily on the small and growth factors. Although the literature considers growth stocks to be less risky than value stocks, neither this exposure nor the small stock risk can be avoided. 7. Conclusions Undoubtedly, some firms use spare cash to increase shareholder value, and others waste it. However, it appears that firms who accumulate cash tend to be exactly those whose operations stand to benefit the most from this financial slack – small, possibly financially constrained firms, with time-varying growth options. Previous studies have shown that firms subject to financial constraints earn lower returns, on average. By holding cash, financially constrained firms raise both their operating performance and their expected market return. However, Pinkowitz and Williamson show that the market places a low value on cash held by firms subject to financial constraints. Financial slack improves firms’ operating performance, and at least amongst small firms, this financial slack may not be properly priced in the market. 27 References Almeida, Heitor, Murillo Campello, and Michael Wesibach, The Cash Flow Sensitivity of Cash, 2003, working paper. Bernard, Victor, and Jacob Thomas, 1990, Evidence that Stock Prices do not Fully Reflect the Implications of Current Earnings for Future Earnings, Journal of Accounting and Economics, 13, 305-340. Blanchard, Lopez-de-Silanes, and Andrei Shleifer, 1994, What do Firms do with Cash Windfalls? Journal of Financial Economics, 36, 337-360. Bond, Stephen, and Costas Meghir, 1994, Dynamic Investment Models and the Firm’s Financial Policy Fama, Eugene F., and Kenneth R. French, 1993, Common risk factors in returns on stocks and bonds, Journal of Financial Economics, 33, 3-56. Fama, Eugene, and Kenneth French, 1997, Industry Costs of Equity, Journal of Financial Economics, 43, 153-193. Faulkender, Michael, and Rong Wang, Corporate Financial Policy and the Value of Cash, working paper, Washington University. Fazzari, Steven, R. Glenn Hubbard, 1988, Bruce Peterson, Financing Constraints and Corporate Investment, Brookings Papers on Economic Activity, 1, 141-206. Gomes, Joao, Amir Yaron, and Lu Zhang, 2004, Asset Pricing Implications of Firms’ Financing Constraints, working paper. Greenwood, Robin, 2004, Aggregate Corporate Liquidity and Stock Returns, working paper, Harvard Business School. Harford, Jarrad, 1999, Corporate Cash Reserves and Acquisitions, Journal of Finance, 6, 1969-1997. Harford, Jarrad, Wayne Mikkelson, and Megan Partch, 2004, The Effect of Cash Reserves on Corporate Investment and Performance in Industry Declines, working paper. Jegadeesh, N. and Sheridan Titman, (1993), Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency, Journal of Finance, 48, 65-91. Jensen, Michael, 1986, Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers, American Economic Review, 76, 323-349. 28 Kaplan, Steven, and Luigi Zingales, 1997, Do Investment Cash-Flow Sensitivities Provide Useful Measures of Financing Constraints? Quarterly Journal of Economics, 112, 169-215. Kim, Chang-Soo, David Mauer, and Ann Sherman, 1998, The Determinants of Corporate Liquidity: Theory and Evidence, Journal of Financial and Quantitative Analysis, 33, 335-359. Lamont, Owen, Christopher Polk, and Jesus Saa-Requejo, 2001, Financial Constraints and Stock Returns, Review of Financial Studies, 14, 529-554. Mikkelson, Wayne, and Megan Partch, 2003, Do Persistent Large Cash Reserves Hinder Performance?, Journal of Financial and Quantitative Analysis, 38. Myers, Stewart, and Nicholas Majluf, 1984, Corporate Financing Decisions when Firms have Investment Information that Investors do not, Journal of Financial Economics, 13, 187-221. Opler, Tim, Lee Pinkowitz, Rene Stulz, and Rohan Williamson, 1999, The Determinants and Implications of Corporate Cash Holdings, Journal of Financial Economics, 51, 3-46. Pinkowitz, Lee, 2004, The Market for Corporate Control and Corporate Cash Holdings, working paper, Georgetown University. Pinkowitz, Lee, and Rohan Williamson, 2002, What’s a Dollar Worth? The Market Value of Cash Holdings, working paper, Georgetown University. Whited, Toni, and Goujun Wu, 2004, Financial Constraints Risk, working paper 29 Table 1: Portfolio summary statistics, unconditional rankings Firms are sorted each month from 1980-2001 into deciles based upon previous month cash ratio. Cash holdings includes all cash and short-term investments. Panel A: Deciles sorted by cash/book equity Decile Cash (MM) Cash/equity (mean) 1 3.8 0.007 2 13.8 0.025 3 29.0 0.050 4 50.0 0.087 5 72.9 0.140 6 94.8 0.212 7 99.9 0.306 8 110.1 0.442 9 109.4 0.665 10 425.8 2.159 Cash/equity (median) 0.007 0.025 0.049 0.084 0.136 0.207 0.301 0.439 0.662 1.036 Book value (MM) 443.8 522.7 553.0 535.9 499.9 432.3 316.4 244.1 157.6 257.0 Market Cap (MM) 817.1 1100.9 1317.4 1396.6 1390.2 1522.2 1086.5 830.8 604.7 743.7 BTM 0.92 0.89 0.87 0.85 0.80 0.75 0.70 0.64 0.57 0.53 Cash/mkcap (median) 0.003 0.014 0.027 0.046 0.071 0.105 0.151 0.221 0.353 0.743 Book value Market Cap BTM 459.2 575.9 583.7 530.1 414.6 315.8 276.8 270.0 232.6 279.7 1457.9 2021.3 1991.5 1557.9 1067.9 674.0 497.7 425.7 330.8 365.1 0.60 0.61 0.64 0.65 0.66 0.68 0.72 0.81 1.00 1.55 Book value Market Cap BTM 424.3 562.9 617.4 561.2 549.0 460.5 381.1 245.1 167.1 88.4 811.1 1155.6 1558.7 1338.7 1331.6 1407.9 1372.8 874.1 669.3 450.7 0.91 0.90 0.88 0.87 0.83 0.83 0.71 0.65 0.58 0.50 Panel B: Deciles sorted by cash/market cap Decile Cash (MM) 1 2 3 4 5 6 7 8 9 10 6.7 28.9 54.5 72.3 78.3 71.9 78.2 98.6 131.8 412.8 Cash/mkcap (mean) 0.004 0.014 0.028 0.048 0.074 0.109 0.157 0.233 0.396 1.290 Panel C: Deciles sorted by cash/total assets Decile Cash (MM) 1 2 3 4 5 6 7 8 9 10 3.4 13.9 32.4 47.2 80.8 96.2 107.5 94.3 89.1 73.8 Cash/assets (mean) 0.002 0.009 0.017 0.029 0.048 0.080 0.133 0.214 0.342 0.605 Cash/assets (median) 0.002 0.008 0.016 0.028 0.047 0.078 0.127 0.203 0.327 0.606 30 Table 2: Portfolio summary statistics, second-stage rankings Firms are placed each month into one of three size bins based on the 30/40/30 NYSE market cap breakpoints and one of three book-to-market bins based on the 30/40/30 NYSE book-to-market breakpoints. A second-stage ranking is then done by cash ratio within each of the nine bins. Panel A: Deciles sorted by cash/book equity Decile Cash (MM) Cash/bk (mean) 1 3.2 0.008 2 9.7 0.028 3 17.2 0.056 4 28.5 0.095 5 44.1 0.145 6 62.7 0.209 7 83.4 0.292 8 109.9 0.403 9 166.5 0.582 10 399.2 1.832 Cash/bk (median) 0.006 0.023 0.043 0.072 0.112 0.168 0.246 0.353 0.526 0.931 Book value (MM) 359.7 407.8 406.9 420.4 440.1 443.4 436.7 417.1 370.6 332.0 Market Cap (MM) 882.3 970.6 997.7 1083.9 1295.0 1317.6 1318.5 1197.2 945.7 888.3 BTM 0.78 0.79 0.78 0.79 0.78 0.77 0.75 0.74 0.73 0.71 Cash/mkcap (median) 0.003 0.012 0.025 0.042 0.065 0.094 0.132 0.181 0.266 0.495 Book value Market Cap BTM 329.8 386.4 391.5 402.5 431.2 447.0 427.2 423.2 401.1 395.5 932.1 1044.1 1176.8 1242.5 1366.7 1323.2 1094.2 1019.0 889.4 810.4 0.70 0.70 0.71 0.72 0.73 0.74 0.75 0.77 0.80 0.98 Book value Market Cap BTM 345.9 423.9 453.8 428.0 414.8 438.3 461.1 428.1 383.9 263.6 883.4 1100.3 1211.3 1138.7 1076.6 1220.2 1402.2 1290.7 1014.1 763.3 0.77 0.79 0.79 0.78 0.79 0.77 0.76 0.73 0.72 0.69 Panel B: Deciles sorted by cash/market cap Decile Cash (MM) 1 2 3 4 5 6 7 8 9 10 3.1 9.4 17.8 28.4 44.0 60.8 75.1 104.2 162.9 419.3 Cash/mkcap (mean) 0.004 0.015 0.029 0.047 0.073 0.108 0.155 0.224 0.357 0.953 Panel C: Deciles sorted by cash/total assets Decile Cash (MM) 1 2 3 4 5 6 7 8 9 10 3.4 12.0 22.2 30.2 43.4 62.5 87.5 110.1 127.7 143.0 Cash/assets (mean) 0.003 0.010 0.022 0.039 0.065 0.100 0.151 0.222 0.328 0.555 Cash/assets (median) 0.002 0.009 0.017 0.028 0.046 0.074 0.119 0.187 0.295 0.552 31 Table 3: Correlation matrix for decile ranks, market capitalization, and book-to-market ratio. C/E is the cash/equity rank, C/ME is the cash/market cap rank, and C/A is the cash/assets rank. SS refers to the second-stage sorts. Panel A: Pearson Correlations C/E C/ME C/A C/E (SS) C/ME (SS) C/A (SS) C/E 1 0.80 C/ME 1 C/A C/E (SS) C/ME (SS) C/A (SS) Market Cap Panel B: Spearman Correlations C/E C/ME 0.92 0.77 1 0.91 0.87 0.86 1 0.86 0.92 0.83 0.92 1 0.85 0.82 0.92 0.92 0.88 1 C/A C/E (SS) C/ME (SS) C/A (SS) C/E C/ME C/A C/E (SS) C/ME (SS) C/A (SS) Market Cap 0.93 0.78 1 0.91 0.87 0.86 1 0.86 0.92 0.83 0.92 1 0.86 0.83 0.92 0.92 0.88 1 1 0.81 1 Market Cap -0.02 -0.08 -0.02 0.00 -0.01 -0.00 1 Market Cap -0.05 -0.22 -0.04 -0.01 -0.02 -0.00 1 BTM -0.14 0.20 -0.14 -0.02 0.07 -0.03 -0.06 BTM -0.26 0.23 -0.24 -0.05 0.03 -0.04 -0.34 Table 4: Average monthly calendar time returns to deciles formed on cash ratios. This table includes equally weighted returns for holding periods from 1 month to 1 year. (Standard errors in parentheses.) Panel A: Cash/Equity Decile 1mo 3mo 1yr 1 1.09 0.95 1.12 (0.32) (0.29) (0.28) 2 1.11 1.02 1.13 (0.33) (0.30) (0.29) 3 1.20 1.09 1.20 (0.35) (0.32) (0.30) 4 1.42 1.16 1.25 (0.37) (0.32) (0.31) 5 1.27 1.23 1.35 (0.36) (0.32) (0.32) 6 1.42 1.32 1.43 (0.36) (0.34) (0.33) 7 1.68 1.45 1.46 (0.39) (0.35) (0.35) 8 1.81 1.52 1.52 (0.44) (0.39) (0.38) 9 1.73 1.53 1.49 (0.47) (0.44) (0.43) 10 1.98 1.63 1.66 (0.46) (0.42) (0.43) Diff 10 - 1 0.88 0.67 0.54 (0.27) (0.23) (0.25) 32 Panel B: Cash/Market Cap Decile 1mo 1 0.84 (0.34) 2 1.03 (0.35) 3 1.00 (0.35) 4 1.16 (0.38) 5 1.45 (0.39) 6 1.37 (0.39) 7 1.41 (0.42) 8 1.54 (0.41) 9 1.90 (0.43) 10 2.00 (0.43) Diff 10 - 1 1.15 (0.27) Panel C: Cash/Assets Decile 1mo 1 0.99 (0.33) 2 1.06 (0.33) 3 1.10 (0.34) 4 1.21 (0.36) 5 1.26 (0.34) 6 1.48 (0.39) 7 1.52 (0.41) 8 1.50 (0.45) 9 1.69 (0.47) 10 1.64 (0.54) Diff 10 - 1 0.64 (0.38) 3mo 0.95 (0.28) 1.04 (0.30) 1.10 (0.31) 1.21 (0.33) 1.21 (0.33) 1.30 (0.36) 1.42 (0.39) 1.45 (0.42) 1.62 (0.49) 1.61 (0.49) 0.66 (0.30) 1yr 0.91 (0.30) 0.97 (0.30) 1.12 (0.32) 1.21 (0.34) 1.34 (0.36) 1.35 (0.37) 1.45 (0.39) 1.61 (0.40) 1.76 (0.41) 1.89 (0.39) 0.98 (0.21) 3mo 0.88 (0.29) 0.99 (0.30) 1.04 (0.32) 1.15 (0.32) 1.28 (0.33) 1.36 (0.35) 1.44 (0.38) 1.50 (0.43) 1.58 (0.47) 1.65 (0.54) 0.77 (0.36) 1yr 1.07 (0.28) 1.09 (0.29) 1.10 (0.30) 1.30 (0.31) 1.36 (0.32) 1.38 (0.34) 1.51 (0.37) 1.53 (0.41) 1.52 (0.45) 1.56 (0.52) 0.48 (0.35) 33 Table 5: Value-weighted average monthly calendar time returns to deciles formed on cash ratios. This table includes monthly returns for holding periods from 1 month to 1 year. Returns are weighted by market capitalization. (Standard errors in parentheses.) Panel A: Cash/Equity Decile 1mo 3mo 1yr 1 0.97 0.88 1.04 (0.31) (0.26) (0.24) 2 0.87 1.04 1.08 (0.32) (0.26) (0.26) 3 1.10 1.14 1.17 (0.32) (0.28) (0.27) 4 0.99 1.28 1.23 (0.36) (0.28) (0.27) 5 1.17 1.22 1.25 (0.35) (0.28) (0.27) 6 1.31 1.15 1.19 (0.36) (0.31) (0.30) 7 1.08 1.46 1.47 (0.42) (0.33) (0.31) 8 1.29 1.41 1.41 (0.47) (0.35) (0.33) 9 1.16 1.49 1.39 (0.52) (0.43) (0.40) 10 1.18 1.47 1.39 (0.48) (0.40) (0.38) Diff 10 - 1 0.21 0.58 0.34 (0.34) (0.25) (0.24) Panel B: Cash/Market Cap Decile 1mo 3mo 1yr 1 0.99 0.89 1.02 (0.34) (0.26) (0.27) 2 1.07 1.08 0.97 (0.33) (0.25) (0.28) 3 0.76 1.13 1.19 (0.34) (0.27) (0.28) 4 1.00 1.19 1.33 (0.39) (0.29) (0.31) 5 1.31 1.22 1.45 (0.40) (0.28) (0.33) 6 1.15 1.26 1.52 (0.40) (0.32) (0.34) 7 1.38 1.43 1.39 (0.42) (0.33) (0.34) 8 1.15 1.35 1.33 (0.42) (0.39) (0.34) 9 1.40 1.67 1.39 (0.42) (0.48) (0.33) 10 1.39 1.33 1.39 (0.37) (0.42) (0.34) Diff 10 - 1 0.39 0.43 0.36 (0.33) (0.30) (0.22) 34 Panel C: Cash/Assets Decile 1mo 1 0.86 (0.32) 2 0.85 (0.31) 3 1.13 (0.32) 4 1.12 (0.33) 5 0.93 (0.32) 6 1.10 (0.39) 7 1.38 (0.41) 8 1.01 (0.41) 9 1.66 (0.52) 10 1.49 (0.55) Diff 10 - 1 0.62 (0.48) 3mo 1.07 (0.24) 1.04 (0.26) 1.21 (0.280 1.27 (0.29) 1.11 (0.28) 1.25 (0.29) 1.30 (0.33) 1.46 (0.39) 1.50 (0.49) 1.81 (0.55) 0.73 (0.46) 1yr 1.14 (0.24) 1.11 (0.25) 1.22 (0.27) 1.29 (0.28) 1.18 (0.27) 1.22 (0.29) 1.30 (0.32) 1.61 (0.38) 1.31 (0.46) 1.54 (0.52) 0.40 (0.42) Table 6: Results from a Fama-French Three Factor regression. This table analyzes the time series of returns from the 1-month holding period, equally weighted portfolios. (Standard errors in parentheses.) Panel A: Cash/Equity Decile Alpha Beta Small Value 1 -0.26 0.95 0.72 0.43 (0.16) 2 -0.25 0.97 0.76 0.40 (0.17) 3 -0.21 1.03 0.74 0.46 (0.20) 4 0.01 1.06 0.79 0.37 (0.19) 5 -0.08 1.01 0.87 0.32 (0.16) 6 0.08 0.98 0.93 0.29 (0.17) 7 0.41 0.99 0.88 0.09 (0.18) 8 0.48 1.07 1.09 0.08 (0.20) 9 0.56 1.06 1.03 -0.29 (0.19) 10 0.70 1.07 1.11 -0.03 (0.21) Diff 10 - 1 0.97 0.11 0.39 -0.47 (0.22) 35 Panel B: Cash/Market Cap Decile Alpha 1 -0.52 (0.16) 2 -0.28 (0.17) 3 -0.33 (0.13) 4 -0.07 (0.16) 5 0.21 (0.16) 6 0.10 (0.62) 7 0.18 (0.21) 8 0.24 (0.19) 9 0.60 (0.22) 10 0.68 (0.28) Diff 10 - 1 1.20 (0.28) Panel C: Cash/Assets Decile Alpha 1 -0.46 (0.19) 2 -0.39 (0.17) 3 -0.33 (0.17) 4 -0.21 (0.20) 5 -0.12 (0.16) 6 0.08 (0.19) 7 0.23 (0.19) 8 0.20 (0.24) 9 0.53 (0.21) 10 0.51 (0.23) Diff 10 - 1 0.97 (0.27) Beta 1.00 Small 0.73 Value 0.36 1.00 0.71 0.22 1.05 0.69 0.19 1.00 0.79 0.01 1.02 0.81 -0.01 1.03 0.83 0.02 1.01 0.86 -0.03 1.04 0.93 0.08 1.06 0.94 0.06 0.99 0.89 0.22 -0.01 0.15 -0.14 Beta 1.00 Small 0.72 Value 0.52 1.02 0.71 0.48 1.04 0.70 0.42 1.05 0.65 0.33 1.02 0.70 0.31 1.06 0.83 0.20 1.04 0.86 -0.04 1.04 1.01 -0.05 1.02 1.00 -0.32 1.03 1.20 -0.52 0.03 0.47 -1.05 36 Table 7: Equally weighted returns to deciles formed on second stage cash ratio rankings; monthly returns for holding periods for 1 month and 1 year; returns equal weighted. (Standard errors in parentheses.) Cash/Equity Cash/Market Cap Cash/Assets Decile 1mo 1yr 1mo 1yr 1mo 1yr 1 1.04 1.01 0.94 0.95 1.01 0.99 (0.33) (0.29) (0.34) (0.30) (0.36) (0.30) 2 1.08 1.05 0.90 1.01 1.07 1.01 (0.34) (0.30) (0.34) (0.31) (0.34) (0.30) 3 1.14 1.18 1.30 1.13 1.02 1.12 (0.35) (0.31) (0.35) (0.32) (0.35) (0.32) 4 1.26 1.21 1.29 1.22 1.23 1.20 (0.36) (0.32) (0.36) (0.33) (0.36) (0.33) 5 1.23 1.26 1.21 1.23 1.25 1.29 (0.36) (0.33) (0.37) (0.34) (0.37) (0.34) 6 1.51 1.32 1.35 1.33 1.30 1.34 (0.39) (0.35) (0.38) (0.35) (0.37) (0.35) 7 1.46 1.44 1.54 1.48 1.42 1.41 (0.39) (0.37) (0.39) (0.36) (0.39) (0.37) 8 1.41 1.52 1.46 1.53 1.55 1.55 (0.41) (0.40) (0.40) (0.39) (0.42) (0.39) 9 1.63 1.60 1.44 1.66 1.74 1.62 (0.44) (0.42) (0.42) (0.41) (0.47) (0.42) 10 1.66 1.67 2.01 1.74 1.57 1.64 (0.46) (0.43) (0.46) (0.42) (0.47) (0.47) Diff 10 - 1 0.62 0.65 1.01 0.78 0.55 0.64 (0.27) (0.23) (0.23) (0.21) (0.32) (0.29) Table 8: Value-weighted returns to deciles formed on second stage cash ratio rankings; monthly returns for holding periods for 1 month and 1 year; returns weighted by market cap. (Standard errors in parentheses.) Cash/Equity Cash/Market Cap Cash/Assets Decile 1mo 1yr 1mo 1yr 1mo 1yr 1 0.76 1.00 0.81 0.99 0.91 1.04 (0.34) (0.26) (0.33) (0.27) (0.35) (0.27) 2 1.02 1.03 1.00 0.92 0.92 1.08 (0.32) (0.26) (0.33) (0.27) (0.32) (0.27) 3 1.03 1.08 1.16 0.97 0.84 1.18 (0.32) (0.27) (0.34) (0.29) (0.33) (0.27) 4 1.14 1.15 1.10 1.14 1.25 1.10 (0.32) (0.27) (0.34) (0.28) (0.33) (0.27) 5 0.94 1.13 1.11 1.15 1.15 1.12 (0.36) (0.28) (0.38) (0.27) (0.36) (0.28) 6 1.30 1.20 1.03 1.36 1.19 1.20 (0.36) (0.29) (0.36) (0.29) (0.35) (0.28) 7 1.12 1.26 1.17 1.29 0.82 1.13 (0.41) (0.29) (0.38) (0.29) (0.39) (0.30) 8 1.39 1.44 1.20 1.43 1.22 1.32 (0.44) (0.32) (0.42) (0.33) (0.45) (0.33) 9 1.54 1.44 1.25 1.63 1.21 1.43 (0.45) (0.38) (0.42) (0.35) (0.47) (0.39) 10 1.19 1.38 1.31 1.28 1.47 1.50 (0.45) (0.37) (0.43) (0.33) (0.51) (0.45) Diff 10 - 1 0.43 0.37 0.49 0.28 0.56 0.46 (0.34) (0.23) (0.33) (0.18) (0.42) (0.34) 37 Table 9: Results from a Fama-French Three Factor regression, second-stage sorts. 1-month holding period, monthly returns equally weighted. (Standard errors in parentheses.) Panel A: Cash/Equity Decile Alpha Beta Small Value 1 -0.32 0.96 0.74 0.43 (0.18) 2 -0.29 1.00 0.78 0.38 (0.16) 3 -0.16 1.03 0.66 0.16 (0.15) 4 -0.07 1.02 0.71 0.25 (0.18) 5 -0.06 1.00 0.75 0.15 (0.16) 6 0.24 1.00 0.84 0.08 (0.18) 7 0.14 1.07 0.85 0.10 (0.17) 8 0.16 1.04 0.85 -0.05 (0.17) 9 0.47 0.99 0.96 -0.20 (0.20) 10 0.43 1.06 0.99 -0.13 (0.22) Diff 10 - 1 0.76 0.09 0.25 -0.56 (0.22) Panel B: Cash/Market Cap Decile Alpha Beta Small Value 1 -0.40 0.97 0.81 0.37 (0.16) 2 -0.44 0.99 0.73 0.34 (0.16) 3 -0.00 1.01 0.69 0.17 (0.16) 4 -0.05 1.05 0.83 0.20 (0.16) 5 -0.08 1.03 0.81 0.12 (0.15) 6 0.07 1.01 0.88 0.10 (0.16) 7 0.26 1.04 0.76 0.04 (0.19) 8 0.21 1.00 0.89 0.02 (0.19) 9 0.24 1.03 0.80 -0.13 (0.20) 10 0.76 1.05 0.94 -0.06 (0.25) Diff 10 - 1 1.17 0.08 0.13 -0.44 (0.23) 38 Panel C: Cash/Assets Decile Alpha 1 -0.26 (0.16) 2 -0.25 (0.17) 3 -0.21 (0.20) 4 0.01 (0.19) 5 -0.08 (0.16) 6 0.08 (0.17) 7 0.41 (0.18) 8 0.48 (0.20) 9 0.56 (0.19) 10 0.70 (0.21) Diff 10 - 1 0.96 (0.22) Beta 0.95 Small 0.72 Value 0.43 0.97 0.76 0.40 1.03 0.74 0.46 1.06 0.79 0.37 1.01 0.87 0.32 0.98 0.93 0.29 0.99 0.88 0.09 1.07 1.09 0.08 1.06 1.03 -0.29 1.07 1.11 -0.03 0.12 0.39 -0.46 Table 10: Monthly returns to deciles formed on cash ratio (unconditional sorts), grouped by size based on Fama/French NYSE breakpoints; 1-month holding period, returns equal weighted. Panel A: Sorted by cash/equity Decile small med large 1 1.04 1.02 1.28 (0.35) (0.33) (0.35) 2 1.02 1.06 0.72 (0.37) (0.35) (0.38) 3 1.25 0.63 1.26 (0.39) (0.37) (0.39) 4 1.50 1.09 0.81 (0.40) (0.37) (0.42) 5 1.29 1.13 1.57 (0.38) (0.38) (0.45) 6 1.46 1.13 1.87 (0.39) (0.38) (0.41) 7 1.81 1.09 1.27 (0.42) (0.40) (0.48) 8 1.98 0.89 1.29 (0.46) (0.49) (0.58) 9 1.76 1.25 1.04 (0.47) (0.54) (0.63) 10 1.98 1.71 1.09 (0.47) (0.54) (0.55) Diff 10 - 1 0.94 0.64 -0.07 (0.26) (0.44) (0.51) 39 Panel B: Sorted by cash/market cap Decile small med 1 0.82 1.15 (0.39) (0.37) 2 0.94 1.27 (0.40) (0.37) 3 1.06 1.08 (0.39) (0.38) 4 1.21 1.25 (0.42) (0.43) 5 1.55 1.30 (0.42) (0.45) 6 1.42 1.21 (0.41) (0.45) 7 1.41 1.64 (0.44) (0.51) 8 1.64 1.04 (0.43) (0.48) 9 1.98 1.50 (0.45) (0.48) 10 2.04 1.57 (0.45) (0.50) Diff 10 - 1 1.21 0.42 (0.29) (0.42) Panel C: Sorted by cash/assets Decile small med 1 1.06 1.20 (0.38) (0.34) 2 1.08 1.06 (0.38) (0.33) 3 0.95 1.39 (0.37) (0.36) 4 1.32 1.04 (0.40) (0.37) 5 1.41 0.97 (0.38) (0.37) 6 1.58 1.22 (0.43) (0.41) 7 1.64 1.38 (0.45) (0.44) 8 1.46 1.23 (0.47) (0.46) 9 1.64 1.69 (0.47) (0.54) 10 1.82 1.09 (0.57) (0.61) Diff 10 - 1 0.75 -0.11 (0.39) (0.52) large 1.08 (0.42) 0.89 (0.41) 1.22 (0.43) 1.40 (0.45) 1.63 (0.65) 1.04 (0.57) 1.67 (0.64) 1.48 (0.61) 1.51 (0.56) 1.21 (0.49) 0.12 (0.48) large 0.62 (0.38) 0.89 (0.37) 1.32 (0.37) 1.09 (0.41) 1.21 (0.40) 0.79 (0.51) 1.57 (0.53) 0.88 (0.58) 1.79 (0.72) 1.71 (0.75) 1.08 (0.69) 40 Table 11: Monthly returns to deciles formed on cash ratio (unconditional sorts), grouped by book-to-market based on Fama/French NYSE breakpoints; 1-month holding period, returns equal weighted Panel A: Sorted by cash/equity Decile Low mid high 1 0.73 0.93 1.57 (0.40) (0.29) (0.39) 2 0.75 1.46 1.36 (0.39) (0.37) (0.41) 3 0.51 1.25 1.56 (0.38) (0.33) (0.37) 4 1.21 1.39 1.41 (0.45) (0.37) (0.39) 5 0.84 1.20 1.98 (0.43) (0.35) (0.40) 6 1.00 1.59 2.26 (0.46) (0.40) (0.44) 7 1.15 1.46 1.91 (0.48) (0.42) (0.52) 8 1.53 1.82 2.21 (0.50) (0.44) (0.50) 9 1.46 1.86 2.25 (0.55) (0.48) (0.51) 10 1.28 1.89 1.64 (0.56) (0.52) (0.54) Diff 10 - 1 0.56 0.95 0.08 (0.43) (0.42) (0.50) Panel B: Sorted by cash/market cap Decile Low mid high 1 0.44 1.00 1.63 (0.41) (0.31) (0.50) 2 0.78 1.21 0.99 (0.44) (0.35) (0.38) 3 0.77 1.30 1.12 (0.46) (0.33) (0.40) 4 1.05 1.21 1.44 (0.49) (0.37) (0.41) 5 1.37 1.37 1.80 (0.53) (0.35) (0.41) 6 1.50 1.38 1.37 (0.52) (0.38) (0.36) 7 1.46 1.54 1.81 (0.60) (0.41) (0.43) 8 1.00 1.74 2.18 (0.54) (0.41) (0.44) 9 1.72 1.86 2.10 (0.63) (0.45) (0.44) 10 3.17 1.88 2.03 (1.91) (0.55) (0.42) Diff 10 - 1 2.72 0.88 0.40 (1.84) (0.45) (0.44) 41 Panel C: Sorted by cash/assets Decile Low 1 0.67 (0.41) 2 0.43 (0.40) 3 0.99 (0.45) 4 0.38 (0.39) 5 1.07 (0.41) 6 1.08 (0.49) 7 1.19 (0.50) 8 1.15 (0.49) 9 1.47 (0.53) 10 1.34 (0.59) Diff 10 - 1 0.76 (0.45) mid 0.86 (0.32) 1.18 (0.35) 1.10 (0.33) 1.28 (0.38) 1.07 (0.36) 1.71 (0.38) 1.52 (0.41) 1.57 (0.47) 1.94 (0.46) 1.82 (0.55) 0.95 (0.45) high 1.57 (0.42) 1.27 (0.37) 1.27 (0.38) 1.64 (0.43) 1.57 (0.39) 1.94 (0.46) 2.25 (0.47) 1.69 (0.48) 2.35 (0.50) 2.39 (0.56) 0.75 (0.54) 42 Table 12: Results from a Fama-French Three Factor regression plus momentum. This table regresses the time series of returns from the 1-month holding period, equally weighted portfolios, upon the three FamaFrench risk factors, plus a momentum factor. (Standard errors in parentheses.) Panel A: Cash/Equity Decile Alpha Beta Small Value Momentum 1 -0.109 0.96 0.75 0.37 -0.21 (0.17) 2 -0.078 0.99 0.79 0.33 -0.20 (0.16) 3 0.048 1.02 0.68 0.10 -0.20 (0.14) 4 0.178 1.01 0.73 0.19 -0.24 (0.17) 5 0.170 0.99 0.77 0.10 -0.22 (0.15) 6 0.597 0.98 0.86 0.00 -0.33 (0.16) 7 0.399 1.06 0.87 0.04 -0.24 (0.16) 8 0.408 1.03 0.86 -0.11 -0.23 (0.16) 9 0.810 0.98 0.98 -0.28 -0.32 (0.19) 10 0.732 1.05 1.01 -0.20 -0.29 (0.22) Diff 10 - 1 0.842 0.09 0.26 -0.58 -0.07 (0.23) Panel B: Cash/Market Cap Decile Alpha Beta Small Value Momentum 1 -0.214 0.96 0.82 0.33 -0.19 (0.16) 2 -0.283 0.99 0.74 0.30 -0.15 (0.15) 3 0.170 1.00 0.70 0.13 -0.16 (0.16) 4 0.197 1.04 0.84 0.14 -0.24 (0.15) 5 0.082 1.02 0.82 0.08 -0.16 (0.15) 6 0.290 1.00 0.90 0.05 -0.21 (0.15) 7 0.602 1.03 0.78 -0.03 -0.32 (0.17) 8 0.551 0.99 0.91 -0.05 -0.32 (0.17) 9 0.623 1.02 0.82 -0.23 -0.37 (0.18) 10 1.184 1.04 0.97 -0.16 -0.40 (0.23) Diff 10 - 1 1.398 0.07 0.14 -0.49 -0.21 (0.23) 43 Table 13: Calendar-time portfolio returns to cash decile portfolios, correcting for industry returns. This table contains equally weighted returns for holding periods from 1 month to 1 year. Firms are categorized into the 48 Fama-French industry groupings and then the relevant industry return is subtracted from each firm’s return each month. (Standard errors in parentheses.) Cash/Equity Cash/Market Cap Decile 1mo 1yr 1mo 1yr 1 -0.11 -0.07 -0.19 -0.13 (0.13) (0.07) (0.12) (0.07) 2 -0.02 -0.06 -0.23 -0.12 (0.12) (0.07) (0.11) (0.06) 3 -0.01 0.03 0.13 -0.02 (0.11) (0.06) (0.13) (0.06) 4 0.11 0.05 0.16 0.07 (0.12) (0.06) (0.11) (0.05) 5 0.10 0.12 0.04 0.08 (0.12) (0.06) (0.12) (0.06) 6 0.33 0.19 0.15 0.17 (0.11) (0.05) (0.11) (0.06) 7 0.23 0.24 0.36 0.29 (0.12) (0.05) (0.12) (0.05) 8 0.20 0.31 0.22 0.31 (0.13) (0.06) (0.13) (0.06) 9 0.40 0.38 0.21 0.45 (0.14) (0.07) (0.13) (0.06) 10 0.42 0.44 0.82 0.53 (0.16) (0.08) (0.16) (0.07) Diff 10 - 1 0.53 0.52 1.01 0.67 (0.20) (0.11) (0.19) (0.09) Table 14: Event study results. Panel A shows the equally weighted return to each firm for the 3-day window around the earnings announcement date. Panel B shows the percent of the subsequent 252 days for which each portfolio’s excess return is positive. Panel C shows the portfolio returns over the next-year (days 250 to days 275) earnings announcement period. Decile 1 2 3 4 5 6 7 8 9 10 Diff 10 - 1 Announcement window return 0.10 0.16 0.18 0.34 0.26 0.23 0.32 0.32 0.32 0.29 0.19 % days > 0 39.7 35.7 37.2 47.4 54.0 62.0 65.3 64.9 60.2 66.7 27.0 Day t+250 to t+275 0.10 -0.01 0.10 0.04 0.06 0.37 0.28 0.36 0.47 0.41 0.31 44 Table 15: Returns by year and month. This table shows equally weighted returns from the unconditional cash/market cap ranking. (Standard errors are in parentheses.) Panel A: Returns grouped by Decade 1 2 3 1970s 0.95 1.03 1.08 4 1.16 5 1.51 6 1.48 7 2.16 8 1.39 9 1.34 10 1.80 1980s 0.98 1.25 1.33 1.30 1.65 1.53 1.53 1.64 2.07 1.57 1990s 0.71 0.84 0.99 1.11 1.44 1.48 1.42 1.61 2.15 2.66 Panel B: Returns grouped by Month 1 2 3 Jan 5.62 6.37 6.77 4 7.00 5 8.59 6 7.42 7 9.17 8 8.72 9 9.79 10 12.12 Feb 1.76 1.56 1.47 1.55 1.87 1.78 2.03 2.91 2.43 1.77 March 0.97 1.55 1.52 1.41 1.06 1.44 1.10 1.53 1.58 1.65 April 0.41 0.96 0.64 0.69 1.09 0.97 0.83 0.68 1.49 1.66 May 0.81 1.00 1.34 1.81 1.26 1.46 2.02 1.96 2.21 1.21 June 0.81 0.75 1.18 0.91 1.89 1.27 1.60 0.86 1.15 0.60 August -0.05 -0.20 -0.35 0.45 0.42 0.33 0.86 0.62 0.16 1.85 September -0.80 -1.00 -0.41 -0.47 -0.57 -0.08 -0.31 -0.25 -0.08 -0.27 October -2.20 -1.02 -2.57 -1.94 -1.27 -0.82 -1.52 -1.81 -1.31 -1.06 November 1.58 1.09 1.24 1.48 1.99 1.91 2.30 1.87 2.38 3.21 December 1.44 1.03 1.21 1.36 1.04 1.09 0.87 0.40 0.98 0.14 Feb-Dec 0.44 0.55 0.50 0.63 0.82 0.85 0.93 0.84 1.02 1.03 45 10-1 0.84 (0.68) 0.58 (0.23) 1.94 (0.36) 10-1 6.50 (1.61) 0.33 (0.99) 0.68 (0.62) 1.24 (0.67) 0.40 (0.55) -0.21 (0.65) 1.15 (1.29) 0.53 (0.53) 1.13 (0.69) 1.63 (1.07) -1.29 (0.99) 0.58 (0.24) Table 16: Takeovers. Panel A shows the takeover rate for the next 12 months for each decile (ranked on cash/market cap), the average premium for the target company, the amount by which each decile’s yearly return is increased from takeover activity, and the frequency of stock and cash offers by decile. Panel B shows the results to a logit model that predicts the probability of becoming a takeover target based on cash holdings, book-to-market, and size. Panel A: Percent of companies taken over and premiums within the next 12 months Rank 1 2 3 4 5 6 7 Takeover % 4.8 5.1 5.1 5.0 5.0 5.2 5.0 Average 14.4 15.8 15.6 16.5 13.8 17.1 16.8 premium Yearly return 0.69 0.80 0.79 0.82 0.69 0.88 0.84 due to takeovers (%) % Stock 33.2 35.8 36.2 37.6 39.5 42.6 42.7 % Cash 50.9 46.1 46.7 47.5 46.3 43.2 44.3 Panel B: Logit Model Intercept Regression 1 2.93 Regression 2 2.88 Regression 3 2.98 Cash/equity -0.042 (1.64) Log(Cash) -0.020 (6.71) -0.011 (6.42) 46 BTM 0.049 (4.32) 0.048 (4.56) 0.042 (3.61) 8 5.0 19.6 9 5.1 17.0 10 5.7 16.1 0.98 0.86 0.91 44.3 42.4 46.4 42.2 51.3 37.1 Log(ME) -0.008 (1.75) -0.011 (2.74) -0.009 (1.75) Table 17: Percentage of firms in each cash holdings decile classified as financially constrained under the KZ measure. Panel A includes the full KZ scoring criteria: cash flow to capital, Tobin’s Q, debt/total capital, dividends/capital, and cash/capital. Panel B scores companies on only the first three factors. The bottom 20% of firms in any given year are classified as financially constrained. Panel A: Percent of firms categorized as constrained under KZ measure, by decile. Cash Decile 1 2 3 4 5 6 7 8 9 10 Constrained 5 21.92 19.49 19.29 18.71 16.70 14.15 13.12 11.57 10.94 15.72 4 33.28 30.77 28.26 25.11 22.17 18.49 15.60 12.60 10.18 8.03 3 23.97 25.24 24.29 24.35 24.07 24.08 22.35 19.63 1600 10.14 Unconstrained 2 14.48 17.14 19.19 21.41 23.36 25.32 26.46 27.29 25.05 19.84 1 6.35 7.35 8.97 10.42 13.69 17.95 22.47 28.91 37.83 46.27 Panel B: Percent of firms categorized as constrained under the modified KZ measure, ignoring cash holdings. Cash Decile 1 2 3 4 5 6 7 8 9 10 Constrained 5 11.99 11.62 12.77 14.07 13.88 13.58 14.92 15.93 20.28 32.35 4 28.15 25.91 24.22 22.39 21.08 18.86 17.08 16.05 15.13 14.18 3 25.67 24.70 23.97 22.78 21.79 21.38 20.92 19.64 18.25 13.45 Unconstrained 2 20.81 22.74 22.27 22.83 23.34 24.27 23.70 23.92 20.92 15.32 1 13.38 15.02 16.77 17.93 19.91 21.91 23.38 24.46 25.42 24.69 Table 18: Performance of financially constrained firms, by KZ ranking quintile. This table presents calendar-time returns for one-year holding periods to portfolios grouped into quintiles based on the KZ scoring. Firms in the bottom quintile are considered financially constrained. Qunitile 1 year 1 year (equal) (value) 1 (unconstrained) 1.463 1.394 (0.38) (0.28) 2 1.476 1.305 (0.35) (0.28) 3 1.509 1.300 (0.31) (0.26) 4 1.404 1.246 (0.31) (0.25) 5 (constrained) 1.299 1.200 (0.38) (0.33) 5-1 -0.164 -0.194 (0.15) (0.21) 47 Table 19: Results from a Fama-French Three Factor regression plus financial constraints factor. This table regresses the time series of returns from the 1-month holding period, equally weighted portfolios, upon the three Fama-French risk factors, plus the financial constraints factor. (Standard errors in parentheses.) Panel A: Cash/Equity Decile Alpha Beta Small Value FC 1 -0.177 0.94 0.68 0.23 0.53 (0.16) (0.07) 2 -0.124 0.96 0.71 0.18 0.53 (0.15) (0.06) 3 -0.078 1.01 0.61 0.03 0.37 (0.14) (0.06) 4 0.135 0.97 0.66 0.03 0.52 (0.17) (0.07) 5 0.047 0.98 0.71 0.02 0.34 (0.16) (0.06) 6 0.388 0.96 0.81 -0.05 0.34 (0.18) 7 0.307 1.01 0.81 -0.05 0.36 (0.16) (0.07) 8 0.283 1.00 0.82 -0.16 0.23 (0.17) (0.07) 9 0.532 0.95 0.93 -0.30 0.23 (0.20) (0.08) 10 0.559 1.02 0.98 -0.25 0.26 (0.23) (0.09) Diff 10 - 1 0.736 0.07 0.30 -0.49 -0.26 (0.22) (0.09) Panel B: Cash/Market Cap Decile Alpha Beta Small Value FC 1 -0.281 0.95 0.75 0.20 0.51 (0.15) (0.06) 2 -0.261 0.95 0.68 0.14 0.52 (0.14) 3 0.108 0.99 0.62 0.14 0.45 (0.15) (0.06) 4 0.037 1.03 0.79 0.08 0.32 (0.16) (0.06) 5 0.025 1.01 0.78 0.01 0.28 (0.15) (0.06) 6 0.226 0.97 0.85 -0.04 0.37 (0.15) (0.06) 7 0.427 0.99 0.73 -0.09 0.30 (0.19) (0.08) 8 0.367 0.95 0.87 -0.11 0.32 (0.19) (0.08) 9 0.303 1.00 0.75 -0.24 0.26 (0.20) (0.08) 10 0.943 0.99 0.91 -0.25 0.39 (0.25) (0.10) Diff 10 - 1 1.224 0.04 0.16 -0.45 -0.12 (0.22) (0.09) 48 Table 20: Performance of constraint-avoiders This table shows the calendar-time monthly returns to firms that avoid being financially constrained because of their cash holdings. Results are presented for both one-month and one-year holding periods, and raw returns and market-adjusted returns (subtracting the CRSP equal or value-weighted index from each return). Raw returns Market-adjusted returns 1mo 1yr 1mo 1yr Equally 1.719 1.422 0.659 0.363 weighted (0.64) (0.52) (0.54) (0.39) Value 1.578 0.823 0.517 -0.23 weighted (0.71) (0.57) (0.60) (0.42) Table 21: Fama-French three factor regressions for constraint-avoiders. This table presents risk-adjusted alphas from a factor regression for the subset of firms that avoid being financially constrained via their cash holdings. Panel A regresses portfolio returns on the Fama-French three factors, and Panel B regresses portfolio returns on these three factors plus a momentum factor and the financially-constrained factor. Panel A: Fama-French three-factor model Equally weighted Value weighted 1mo 1yr 1mo 1yr Alpha 0.946 0.473 1.205 -0.127 (0.45) (0.28) (0.67) (0.32) Beta 0.96 1.02 1.07 1.24 Size 1.46 1.61 1.18 0.97 Value -0.71 -0.52 -1.23 -0.91 Panel B: Five-factor model Equally weighted Value weighted 1mo 1yr 1mo 1yr Alpha 1.585 0.750 1.783 0.137 (0.44) (0.28) (0.68) (0.33) Beta 0.98 1.03 1.06 1.25 Size 1.44 1.60 1.15 0.97 Value -0.85 -0.56 -1.46 -0.94 Momentum -0.56 -0.26 -0.42 -0.26 FC -0.11 -0.10 0.15 -0.14 49 Table 22: Performance of cash decile portfolios after removing firms that avoid financial constraints via their cash holdings. Decile 1 2 3 4 5 6 7 8 9 10 Diff 10 - 1 1mo 0.468 0.875 1.265 0.945 1.090 1.154 1.135 1.036 1.631 1.236 0.832 (0.39) Cash/Equity 1yr 0.915 0.936 0.956 1.022 1.013 1.174 1.143 1.236 1.432 1.137 0.221 (0.20) 1mo 0.531 0.921 1.288 0.977 1.030 1.122 1.287 1.207 1.384 1.245 0.761 (0.42) Cash/Market Cap 1yr 0.774 0.940 0.820 1.041 1.017 1.159 1.198 1.310 1.491 1.230 0.455 (0.16) Table 23: Operating performance of cash deciles, based on cash/equity rankings. Decile means are reported, and decile medians appear in parentheses. Panel A: Investment and Asset Growth Investment Asset Growth Decile Year t Year t+1 Year t Year t+1 1 27.2 (19.0) 25.4 (18.0) 12.9 (7.0) 11.1 (6.2) 2 26.8 (19.0) 24.8 (18.1) 13.3 (6.9) 10.2 (5.9) 3 27.8 (19.5) 26.1 (18.7) 13.5 (7.1) 11.0 (6.3) 4 29.2 (20.2) 27.5 (19.6) 13.6 (7.0) 11.3 (6.2) 5 30.7 (21.2) 28.7 (20.1) 13.7 (7.3) 11.4 (6.3) 6 32.7 (22.1) 30.1 (21.4) 14.1 (7.2) 11.7 (6.5) 7 34.7 (23.3) 31.7 (22.0) 14.5 (7.5) 11.9 (6.5) 8 38.7 (24.9) 33.6 (23.3) 15.1 (7.3) 12.9 (6.8) 9 45.1 (26.9) 37.5 (24.5) 15.7 (6.8) 13.2 (6.5) 10 50.0 (28.0) 43.1 (26.0) 15.8 (5.7) 14.1 (5.9) Constraint- 48.3 (28.1) 45.1 (25.6) 16.7 (9.7) 17.8 (1.6) avoiders Panel B: Sales Growth and Cash flow growth Sales Growth Cash Flow Growth Decile Year t Year t+1 Year t Year t+1 1 12.0 (8.3) 10.4 (7.4) 30.4 (12.7) 29.1 (11.9) 2 12.0 (8.0) 9.8 (7.0) 30.0 (12.6) 27.6 (11.7) 3 12.1 (8.2) 10.6 (7.4) 29.6 (12.8) 29.5 (12.5) 4 12.7 (8.2) 10.6 (7.3) 31.2 (13.1) 28.7 (12.1) 5 13.5 (8.4) 10.8 (7.4) 32.1 (13.0) 30.6 (12.7) 6 13.9 (8.8) 11.3 (7.7) 31.6 (13.2) 30.6 (12.5) 7 15.2 (9.2) 12.4 (8.1) 31.8 (13.2) 31.8 (12.6) 8 17.6 (9.8) 14.4 (8.7) 33.7 (13.9) 33.7 (13.0) 9 22.7 (10.7) 17.2 (9.4) 35.0 (14.2) 35.3 (13.7) 10 27.5 (11.2) 21.9 (10.3) 43.1 (16.2) 43.9 (15.6) Constraint- 40.0 (13.7) 31.9 (12.0) 103.6 (32.5) 91.1 (28.5) avoiders 50