Earnings quality and short sellers.
Keywords: accruals; earnings quality; returns; short sellers.
Data Availability: Data are available from sources identified in the text.
Earnings quality is often defined in terms of persistence and sustainability. Revsine et al. (1999, 224-225) state that earnings are considered to be of high quality when they are sustainable. Bodie et al. (2002,628) speak of the quality of earnings in terms of the "extent to which we might expect the reported level of earnings to be sustained." In this paper, earnings that are more persistent are viewed as higher quality. Examples of low-quality earnings include insufficient allowance for doubtful accounts, insufficient provisions for obsolete inventory, and aggressive revenue recognition practices that bring future revenues into the current period. Common to these examples of low earnings quality is the fact that current earnings are temporarily inflated due to accounting choices, but cash flows are unaffected.
The lower persistence of earnings resulting from high levels of accruals does not have to be a direct result of earnings management activity. The nature of accrual accounting is to accrue and defer past, current, and anticipated future cash receipts and disbursements. The accrual process involves a significant amount of estimation of future cash receipts and payments, and a subjective allocation of past cash receipts and payments. In doing so, the accrual process creates accounts of varying reliability. For example, recording the net realizable value of receivables involves estimation of default risk across a portfolio of debtors. Other examples include estimating recoverable amounts of inventories, depreciating and amortizing long-lived assets, and estimating post-retirement benefit obligations. Estimation errors for the various asset, liability, and associated revenue and expense accounts (either intentional or unintentional) will all lead to lower persistence in earnings. Collectively, these estimations ma nifest themselves in the magnitude of reported accruals.
Sloan (1996) documents that firms with reported income greater than operating cash flows (i.e., high accruals) experience a decline in earnings performance in the following year. In addition, stock prices fail to impound the implications of current accruals for future earnings, leading to predictable return patterns for firms with high levels of accruals. Furthermore, firms reporting earnings with large accruals are more likely to be subject to SEC enforcement actions (Dechow et al. 1996; Bradshaw et al. 2001) and earnings restatements (Richardson, Tuna, and Wu 2002). Collectively, this suggests that the magnitude of accruals is a good indicator of earnings quality.
In this paper, I examine whether short sellers are able to identify and trade on earnings quality information embedded in accruals. Short sellers have particularly strong incentives to utilize measures of earnings quality because they can directly profit from the lower future performance of high-accrual firms. Previous research finds that short sellers are an informed subset of investors who are able to identify over-priced securities (see e.g., Figlewski 1981; Dechow et. al. 2001; Desai et. al. 2002). Short sellers have the ability and the financial incentive to trade on the basis of accrual information. They represent a sophisticated group of investors who can directly benefit from price declines.
The returns to the accrual anomaly are realized over the next 12 months (Sloan 1996), as opposed to other anomalies such as market-to-book where the returns are realized over longer horizons (e.g., 3 to 5 years). Given that short sellers desire to keep short interest positions open for the shortest time possible to minimize associated costs, the accrual anomaly is a natural candidate for short sellers.
The evidence in this paper suggests that short sellers do not appear to utilize the information in operating accruals about future earnings. One interpretation of this result is that investors, in particular short sellers, are ignoring valuable information. Thus, a potential reason why the accrual anomaly continues to exist is the absence of traders taking short positions in high-accrual firms. An alternative explanation is that high-accrual firms are riskier and more costly to short sell. Additional analysis reveals some merit for this alternative hypothesis. High-accrual firms tend to have low book-to-market ratios and high sales growth. This evidence suggests that it may be too costly for short sellers to exploit earnings quality information contained in accruals. Short sellers carry unlimited risk on the upside and will shy away from securities that have a chance for continued growth. Furthermore, I find that high-accrual firms tend to be smaller, less liquid securities. Dechow et al. (2001) suggest that it is more difficult to borrow stock and take short positions for these type of firms. It is therefore possible that transaction costs may be prohibiting short sellers from taking positions in high-accrual firms.
To provide further insights into whether short seller's trade on earnings quality, I provide a detailed analysis of the earnings restatements of Enron and WorldCom. These firms reported high levels of accruals in the late 1990s. Both firms are large, closely followed by analysts and actively traded. Hence, transaction costs are not likely to be inhibiting short sellers from taking positions in these securities. I find no evidence that short sellers anticipate the earnings-quality problems associated with the high levels of accruals. Increased short interest activity is only evident at the restatement announcement.
Collectively, I am unable to document that short sellers trade on the information in accruals about future earnings. The failure to reject the null of no association may be because trading on the accrual strategy is too risky or costly for short sellers, or that the tests are not powerful enough to detect such trading patterns. Alternatively, it may simply be the case that short sellers are unaware of the accrual strategy.
SAMPLE SELECTION AND VARIABLE MEASUREMENT
Short interest reports are compiled monthly by the U.S. stock exchanges. I use reports from 1990 to 1998 inclusive. Each report contains one record per month per security, identified by a stock ticker. Matching Compustat and CRSP tapes (and excluding firms in the financial services industry) yields a final sample of 12,195 finn-year observations, consisting of 2,402 firms. Consistent with prior work there is a substantial number of firms (23 percent) that do not have any short interest activity (Asquith and Meulbroek 1996; Dechow et al. 2001).
Short interest is calculated as the number of shares sold short in month X, divided by the number of common shares outstanding (Asquith and Meulbroek 1996). I use the number of shares outstanding as reported on the CRSP tapes for the month of the release of the short interest report, thereby reducing the possibility of a change in capital structure affecting the short interest calculation. I use the short position four months after the end of the fiscal year, ensuring that accrual information is available to investors.
Figure 1 illustrates the distribution of short positions. The first bar is all zero short positions. Each subsequent bar includes all firm-years up to the number reported beneath the bar, e.g., the second bar includes firm-years where the short position is greater than 0 but less than 0.25 percent, the third bar includes firm-years where the short position is greater than or equal to 0.25 percent but less than 0.5 percent, and so on. After 7 percent, I group all firm-year observations together; 23 percent of the sample has no short position four months after the end of the fiscal year. The figure documents that the distribution is highly skewed, with few firms having short positions in excess of 1.5 percent of shares outstanding.
Data for the accrual calculations are obtained from Compustat. I calculate two accrual measures. First, operating accruals is calculated as follows:
Operating Accruals = (Earnings -- CFO)/Average Assets.
I use earnings before extraordinary items and cash flow from operations (CFO) as reported on the statement of cash flows. The second measure, TotalAccruals, is calculated as:
Total Accruals = (Earnings -- CFO -- CFI)/Average Assets.
CFI is cash flows from investing activities as reported on the statement of cash flows. This measure of accruals incorporates both operating and investing accruals. The measure is different from that used in much of the prior literature. This measure is designed to capture all nonfinancing accrual transactions. Cash that is spent on capital expenditures and other acquisition-related activities create assets whose future economic benefit, while probable, is not certain. Richardson, Sloan, Soliman, and Tuna (2002) find that such investing accruals are not as persistent as operating accruals. Limiting the analysis to only operating accruals offers an incomplete analysis of earnings quality.
I expect to find a positive relation between the level of accruals and short interest positions. I expect short sellers to take advantage of the information in accruals about future profitability. If this behavior is systematic, then it will manifest itself in a positive association between the level of accruals and short positions.
Firm Characteristics Conditional on the Level of Reported Accruals
Panel A of Table 1 contains mean values for my accrual measures, short interest position, and other financial statement variables. The mean firm in my sample has operating (total) accruals equal to --4.9 (4.7) percent of its asset base. Consistent with prior research, operating accruals are negative, reflecting the impact of depreciation and amortization. Total accruals, on the other hand are positive, reflecting the inclusion of the originating accrual (acquiring the underlying asset).
The average short interest position is 1.2 percent of outstanding shares. The mean firm has change in earnings of about 1 percent of its asset base, market capitalization of $2.6 billion, a dividend yield of 1.8 percent, a book-to-market ratio of 0.64 and annual sales growth of about 9.5 percent. Trading volume is about 148,200 shares per day and the average stock price is about $23.
Panel B of Table 1 reports the mean values of the same variables after partitioning the sample based on the level of operating accruals. The lowest (highest) quintile of firms has operating accruals equal to -17.9 (6.9) percent of average total assets. The pattern of short interest across these portfolios is relatively constant (ranges from 1.24 percent for the lowest quintile to 1.25 percent for the highest quintile, with quintile 4 having the lowest short interest). The inter-quintile difference is not significantly different from zero (i.e., the 1.24 percent for the lowest quintile is not statistically different from the 1.25 percent for the highest quintile, t-statistic of -0.09). This finding does not support the primary prediction of the paper. Short sellers do not appear to be more active in firms with high accruals. This is surprising as there is a 14.4 percent hedge return from taking a short (long) position in firms in the highest (lowest) accrual quintile. Short sellers appear to be forgoing a prof it opportunity.
The remaining variables suggest that high-accrual firms tend to be smaller firms with lower daily trading volume, lower stock prices, higher sales growth, and lower book-to-market ratios (when compared to all other firms). Together this evidence suggests that high-accrual firms may be particularly costly to short. Short sellers are less inclined to short dividend-heavy stocks. This is because short sellers are obliged to pay the owner an amount equal to all dividends that accrue during the period of the short sale (and this can have tax implications). Second, short sellers are less inclined to short small, illiquid securities. This is because if there is little stock available to be borrowed to keep a short position open (i.e., the equity lending market is illiquid), then the short seller will have to make a purchase to cover the position. Third, if a stock has a chance for a large upward price movement (as evidenced by high sales growth and low book-to-market measures), this will make it less attractive to a short seller, as they bear potentially unlimited upside risk. The concentration of higher growth stocks in the high-accrual portfolio may be prohibiting short sellers from trading on the basis of earnings quality information, even though on average these stocks will underperform. Together the evidence in Panel B suggests that high-accrual firms are more costly to short sell, thereby discouraging short interest activity in high-accrual firms. This suggests a potential alternative explanation (transaction costs and risk) for the absence of a relation between the level of accruals and short interest positions.
I undertake a similar analysis in Panel C of Table 1 for total accruals. The lowest (highest) quintile has mean total accruals equal to -14.4(26.5) percent of average total assets. For quintiles formed on total accruals the level of short interest is reasonably flat across the first four quintiles, with a peak for the highest quintile (the difference between the 1.19 percent for the lowest quintile and 1.50 percent for the highest quintile is significant at the 1 percent level, t-statistic of -3.52). However, the economic significance of this increased short-selling activity is not great. In unreported tests, I examine the median value across the total accrual quintiles. The highest accrual quintile has the second lowest median short interest position (second only to the lowest accrual quintile).
The remaining columns in Panel C of Table 1 report the mean values of the risk and transaction costs across the total accrual quintiles. Similar to the operating accrual quintiles, I find that the highest quintile is comprised of smaller, less frequently traded firms, with lower book-to-market rations and higher sales growth. Again it appears that high-accrual firms may be particularly costly to sell short. This reduces the power of the tests designed to detect a positive relation between the level of accruals and short interest positions.
Are Short Sellers Only Active in High-Accrual Firms That Are Cheaper and Less Risky?
To address the alternative explanation of transaction costs and risk I examine the short-selling activity in firms reporting high levels of accruals. By examining those high-accrual firms where short sellers are most active, I will be able to examine the merits of the transaction cost and risk explanations for the absence of a relation between short interest and accruals. If transaction costs and risk are deterring short sellers from taking positions in high-accrual firms, then high-accrual firms where short sellers are not active should be characterized by less frequent trading, higher dividend yields, smaller market capitalization, higher levels of sales growth, and smaller book-to-market ratios.
I split high-accrual firms into two groups. The first group has all high-accrual firms with high levels of short interest (greater than 1.5 percent of outstanding shares have been shorted). The second group has all other firms. The 1.5 percent cut-off is the same used in Dechow et al. (2001) to examine the impact of high levels of short interest activity. In Panel A of Table 2, I report the analysis for operating accruals, Panel B has the results for total accruals.
The evidence in Panel A of Table 2 suggests that short sellers are more active in high-accrual firms that have large market capitalization (a test of difference on the logged value of market capitalization is statistically significant at the 5 percent level), greater trading volume, lower dividend yield, and lower book-to-market ratios. The finding that short sellers are most active in high-growth securities is inconsistent with the upside risk explanation for an absence of a relation between accruals and short interest. However, the finding that short sellers are most active in larger, more liquid securities is consistent with a transaction cost explanation. The findings for total accruals in Panel B are similar--short sellers are most active in high-accrual firms with greater trading volume and lowest book-to-market ratios.
To address the issue of risk and transaction costs more fully, I examine only those stocks that attract a high level of short selling. Presumably, for these stocks the cost of short selling is outweighed by the potential for price deterioration. Thus, focusing solely on firms with very high short positions may create a more powerful test to analyze whether short sellers trade on the basis of earnings quality measures. In the highest operating (total) accrual portfolio there are 482(578) out of 2,438 (2,438) firms with short interest in excess of 1.5 percent of shares outstanding. This translates to 19.7(23.7) percent of high operating (total) accrual firms having high short interest positions. The proportions across the other four accrual portfolios are 19.7 (18.7) percent, respectively. There is no evidence that short sellers are more active in firms reporting high levels of operating accruals. But there is some evidence that short sellers are more active in firms reporting high levels of total accruals. Thi s is consistent with recent research by Dechow et al. (2001), who find that short sellers are most active in growth firms (as measured by market-to-book ratios). The measure of total accruals used is strongly negatively correlated with book-to-market (Pearson correlation coefficient of --0.151), as compared to the weaker association between book-to-market and operating accruals (Pearson correlation coefficient of -0.035).
Detailed Analysis of WorldCom and Enron
In this section I examine short interest positions for two high-profile companies: WorldCom and Enron. Both companies have recently been subject to SEC investigation and subsequent earnings restatements. Both companies also reported unusually high levels of total accruals (in particular investing accruals) leading up to these events. I investigate their short positions in order to gain further insights into short-selling activities.
Panel A of Table 3 reports the accrual information for WorldCom for the years 1997-2001 leading up to the massive restatement of over $3 billion on June 25, 2002 (this number has been subsequently increased to over $7 billion). Total accruals (as broadly defined in Richardson, Sloan, Soliman, and Tuna (2002) were large in the latter part of the 1 990s for WorldCom. In 2000, the total accruals were more than 12 percent of the asset base (this placed WorldCom in the top quintile for reported total accruals as reported in Table 2). Operating accruals for WorldCom were consistently negative for the 1997-2001 period, but total accruals were positive (particularly so for 1997 and 2000). This difference is attributable to investing accruals. WorldCom was notorious for acquisition-intensive practices in the 1990s that manifested in large investing accruals. Investing accruals (the difference between total accruals and operating accruals) for WorldCom ranged from 15.5 percent of assets in 1997 to 9.9 of assets in 2001 .
A recent article in The Financial Times (2002) described the situation at WorldCom as follows:
"In 2000, they just pulled every rabbit out of the fraud book. Bad debt was manipulated. Tax was manipulated. Everything they could do they did. But by the end of 2000, they had run out of tricks." The large accruals that were common throughout the 1990s started to unwind in 2001, leading to the ultimate demise of the company.
The stock price movement and short interest positions of WorldCom for the 12-month period August 2001-July 2002 is seen in Figure 2. The stock price and short interest positions remained fairly stable until news of an "accounting scandal" was released in June 2002. WorldCom had been capitalizing line costs as noncurrent assets, which had contributed to the growth in noncurrent assets and had prevented significant charges from flowing through the income statement. Short sellers only increased their position around the time of the restatement and ensuing bankruptcy filing on July 23, 2002. WorldCom is a large, well-established firm for which it would have been relatively easy to borrow stock to take a short position. For this example, transaction costs are not likely to have limited short sellers from taking positions.
Enron is another example of a firm reporting high levels of total accruals in the latter part of the 1990s. Panel B of Table 3 reports the financial information used to calculate accruals for Enron for the years 1996-2000 (data was not available to calculate accrual information for the 2001 year). In 1998 (1999) Enron reported total accruals equal to 11.5 (10.5) percent of its asset base (placing it in the highest total accrual quintile as reported in Table 2). These large total accruals were primarily driven by noncurrent operating accruals. Richardson, Sloan, Soliman, and Tuna (2002) find that growth in noncurrent operating assets (i.e., investing accruals) is a leading indicator of subsequent earnings declines, future stock price, and subsequent SEC enforcement actions. Despite the usefulness of this accrual measure as an indicator of earnings quality, short sellers do not appear to have traded on this information. The stock price of Enron continued to rise throughout the latter part of the 1990s and into 2001 before concerns started to surface about SEC investigations into the accounting of Enron. Short sellers lagged this activity considerably and were not really active until late in 2001, by which time Enron was preparing to file for bankruptcy proceedings. Figure 3 depicts the stock price and short interest pattern for the 12-month period leading up to the bankruptcy announcement on December 2,2001. It was not until the firm announced a $586 million restatement on November 7, 2001 that short sellers started to take an active interest in the security. Meanwhile the stock price had started to fall throughout 2001 in response to unsustainable earnings projections. Enron is another example for which the cost of short selling is relatively low, so the transaction cost explanation has less merit. Enron, however, was viewed as a high-growth stock that had repeatedly won awards for developing innovative products. Such prospects for future growth may have deterred short sellers.
Additional Tests and Caveats
The results in this paper are robust to additional specifications. First, examining changes in short positions across years with changes in the level of reported accruals reveals no association between short selling and earnings quality measures. Second, examining changes in short positions around subsequent earnings announcements does not yield different results. Third, I have examined short positions immediately around the release of the 10-K. While, I do not have the exact filing date for all the firms in my sample, I am able to examine the change in the short position from the last month of the fiscal year-end (e.g., December for a December year-end firm) to four months after the fiscal year-end, thereby straddling the 10-K release date for most firms. The results of this analysis (unreported) still suggest that short sellers do not trade on the basis of accrual information.
The short interest data used in this study limits the power of tests on two dimensions. First, the frequency of the data is monthly, thereby limiting the precision to detect changes in short-selling activity. The costs of short selling make it undesirable to keep short positions open for extended periods. It is possible that short sellers are trading on the basis of earnings quality information but are taking positions for short periods. The additional tests designed to address changes in short interest activity around subsequent earnings announcements were an attempt to address this possibility. However, month-to-month changes may be too wide a window to detect the changes in short activity. The low frequency of short interest data may contribute to low power tests. Unfortunately, there is no data available at the daily or weekly level. Second, the data is aggregated across all classes of investors. Investors take short positions for a variety of reasons, including hedging. Ideal tests would exclude these tr ansactions. However, the monthly short interest data do not allow the researcher to identify individual trades. The data are reported in aggregate. This data limitation reduces the power of the tests.
This paper examines whether short sellers trade on the basis of earnings quality information. Specifically, I examine whether there is greater short interest activity for securities with high levels of accruals. Previous research has found accruals are a good leading indicator of subsequent earnings, making them a useful summary measure of earnings quality. Information in accruals about future earnings is ignored by (1) investors (Sloan 1996), and (2) analysts and auditors (e.g., Teoh and Wong 2002; Bradshaw et al. 2001). In this paper I find that short sellers do not appear to utilize information in accruals about future earnings.
The failure to reject the null could be due to several reasons. First, high-accrual firms may be more costly to short sell. I find evidence that high-accrual firms are smaller, less liquid firms for which it may be more difficult to borrow stock to take a short position. Furthermore, I find that short sellers are more active in the larger, more liquid high-accrual stocks, suggesting that transaction costs may be prohibiting short selling of some high-accrual firms. Second, high-accrual firms may be more risky to short sell. As short sellers carry potentially unlimited upside risk, they may shun stock exhibiting growth prospects. I find that high-accrual firms are, on average, high-growth firms as measured by book-to-market and sales growth. Together this suggests that costs and risk may be prohibiting short sellers from trading on the basis of earnings quality information. Third, the failure to reject the null could be because short sellers are unaware of the accrual strategy and are forgoing a profit opportu nity.
Finally, I conduct detailed analysis of WorldCom and Enron. I find that despite early warning signs of impending issues of earnings quality as measured by total accruals, there is no evidence that short sellers were unusually active in these stocks. Instead short sellers appear to wait until the firm restates its earnings. Overall, the paper is consistent with prior findings that the market does not impound earnings quality information, by documenting the absence of a relation between short sales and accruals.
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
[FIGURE 3 OMITTED]
TABLE 1 Analysis of Short Interest Positions across Accrual Portfolios Panel A Mean Values of Variables Number Operating Total Short Change in of Firms Accruals (a) Accruals (b) Interest (c) Earnings (d) 12,195 -0.0499 0.0466 0.0120 0.0091 Number Future Market Div Book-to- Trading of Firms Returns (e) Cap (f) Yield (g) Market (h) Volume (i) 12,195 0.0002 2621 0.0175 0.6357 148,200 Number Sales Stock of Firms Growth (j) Price (k) 12,195 0.0944 23.14 Panel B Mean Values of Variables across Operating Accrual Portfolios Accrual Number Operating Short Change in Portfolio of Firms Accruals (a) Interest (c) Earnings (d) Low 2437 -0.1789 0.0124 0.0316 2 2439 -0.0764 0.0121 0.0071 3 2442 -0.0466 0.0117 0.0065 4 2439 -0.0169 0.0115 0.0041 High 2438 0.0690 0.0125 -0.0037 Accrual Future Market Div Book-to- Trading Portfolio Returns (e) Cap (f) Yield (g) Market (h) Volume (i) Low 0.0796 2138 0.0115 0.6560 162,880 2 0.0229 3792 0.0202 0.6431 181,010 3 -0.0042 3255 0.0219 0.6397 152,180 4 -0.0331 2561 0.0206 0.6511 139,040 High -0.0645 1358 0.0132 0.5894 104,250 Accrual Sales Stock Portfolio Growth (j) Price (k) Low 0.0539 17.52 2 0.0740 26.33 3 0.0850 27.08 4 0.1015 25.62 High 0.1577 19.13 Panel C Mean Values of Variables across Total Accrual Portfolios Accrual Number Operating Short Change in Portfolio of Firm Accruals (b) Interests (c) Earnings (d) Low 2437 -0.1440 0.0119 0.0368 2 2439 -0.0105 0.0116 0.0061 3 2442 0.0338 0.0113 0.0026 4 2439 0.0884 0.0113 0.0025 High 2438 0.2653 0.0150 -0.0024 Accrual Future Market Div Book-to- Trading Portfolio Returns (e) Cap (f) Yield (g) Market (h) Volume (i) Low 0.0883 1669 0.0195 0.7728 143,480 2 0.0231 3226 0.0210 0.7248 150,440 3 -0.0145 3761 0.0226 0.6254 167,260 4 -0.0258 3055 0.0154 0.5659 163,080 High -0.0705 1393 0.0089 0.5009 114,730 Accrual Sales Stock Portfolio Growth (j) Price (k) Low -0.0262 16.06 2 0.0464 24.95 3 0.0741 27.54 4 0.1137 25.57 High 0.2642 21.57 Firm-year observations are ranked annually and assigned in equal numbers to quintile portfolios based on Accruals. (a)Operating Accruals is calculated as the difference between earnings before extraordinary items (item 123) and cash flows from operations (item 308) as reported on the statement of cash flows. This variable is scaled by average total assets (item 6). (b)Total Accruals is calculated as earnings before extraordinary items (item 123) less cash flows from operations (item 308) less cash flows from investing activities (item 311) as reported on the statement of cash flows. This variable is scaled by average total assets (item 6). (c)Short Interest is the ratio of the number of shares shorted four months after the end of the fiscal year deflated by the number of shares outstanding in the same month as reported by CRSP. (d)Change in Earnings is the change in operating income (item 178) from year t-1 to year t. (e)Future Returns are calculated from the start of the fifth month subsequent to the fiscal year-end, Returns are cumulated over a 12-month window and are reported after deducting the returns from an equally weighted market portfolio. (f)Market Cap is the market capitalization of the firm. It is calculated as the number of shares outstanding (item 25) multiplied by the fiscal year end price (item 199). (g)Div Yield is the dividend yield. It is calculated as the dollar value of dividends (item 21) divided by Market Cap. (h)Book-to-market is calculated as the book value of common equity (item 60) divided by Market Cap. (i)Trading Volume is the average daily trading volume as reported on the Monthly Short Interest Reports. (j)Sales Growth is calculated as log([Sales.sub.t]/[Sales.sub.t-1]). Sales is net sales as reported by Compustat (item 12). (k)Stock Price is the fiscal year end price as reported on Compustat (item 199). TABLE 2 Detailed Analysis of Transaction Costs for High-Accrual Firms Panel A: Operating Accruals (a) Number Market Book-to- Div Trading of Firms Cap (c) Market (d) Yield (e) Volume (f) High Short Interest ([greater than or equal to] 1.5%) 482 1647 0.4503 0.0090 210,020 Low Short Interest (< 1.5%) 1956 1288 0.6234 0.0142 68,200 Statistical Test for Difference Number Sales Future of Firms Growth (g) Returns (h) High Short Interest ([greater than or equal to] 1.5%) 482 0.1909 -0.080 Low Short Interest (< 1.5%) 1956 0.1495 -0.061 Statistical Test for Difference Panel B: Total Accruals (b) Number Market Book-to- Div Trading of Firms Cap (c) Market (d) Yield (e) Volume (f) High Short Interest ([greater than or equal to] 1.5%) 578 1386 0.3945 0.0079 217,200 Low Short Interest (<1.5%) 1860 1396 0.5337 0.0093 70,300 Statistical Test for Difference -0.06 -7.54 ** -1.22 7.05 ** Number Sales Future of Firms Growth (g) Returns (h) High Short Interest ([greater than or equal to] 1.5%) 578 0.2591 -0.093 Low Short Interest (<1.5%) 1860 0.2658 -0.064 Statistical Test for Difference -0.28 -1.18 ** Significant at the 1% level or better. Firm-year observations are ranked annually and assigned in equal numbers to quintile portfolios based on Accruals. This table only examines firms that fall into the highest accrual portfolio. This group of firms is then split into two groups based on Short Interest The comparison is for firms with high levels of short interest (greater than 1.5 percent of shares outstanding) to other firms. (a)Operating Accruals is calculated as the difference between earnings before extraordinary items (item 123) and cash flows from operation (item 308) as reported on the statement of cash flows. This variable is scaled by average total assets (item 6). (b)Total Accruals is calculated as earnings before extraordinary items (item 123) less cash flows from operations (item 308) less less cash flows from investing activities (item 311) as reported on the statement of cash flows. This variable is sealed by average total assets (item 6). (c)Market Cap is the market capitalization of the firm. It is calculated as the number of shares outstanding (item 25) multiplied by the fiscal year end price (item 199). (d)Book-to-market is calculated as the book value of common equity (item 60) divided by Market Cap. (e)Div Yield is the dividend yield. It is calculated as the dollar value of dividends (item 21) divided by Market Cap. (f)Trading Volume is the average daily trading volume as reported on the Monthly Short Interest Reports. (j)Sales Growth is calculated as log([Sales.sub.t]/[Sales.sub.t-I]). Sales is net sales as reported by Compustat (item 12). (h)Future Returns are calculated from the start of the fifth month subsequent to the fiscal year-end. Returns are cumulated over a 12-month window and are reported after deducting the returns from an equally weighted market portfolio. TABLE 3 Analysis of Accrual Information for WorldCom and Enron Panel A: Selected Data from Financial Statements for WorldCom (1997-2001, in millions) 2001 Net Income $1,407 Cash Flows from Operations 6,389 Cash Flows from Investing -8,818 Activities Average Total Assets 88,897 Operating Accruals (a) -0.056 Total Accruals (b) 0.043 Size of Earnings Restatement More than $3 billion (as at (as originally announced) November 2002 this number had increased to over $ billion) Date of Restatement Announcement June 25, 2002 Panel A: Selected Data from Financial Statements for WorldCom (1997-2001, in millions) 2000 1999 1998 1997 Net Income $2,609 $3,950 -$2,558 $383 Cash Flows from Operations 5,330 11,005 4,085 1,318 Cash Flows from Investing -13,612 -9,555 -9,433 -3,277 Activities Average Total Assets 88,482 88,736 54,395 21,126 Operating Accruals (a) -0.031 -0.080 -0.122 -0.044 Total Accruals (b) 0.123 0.028 0.051 0.111 Size of Earnings Restatement (as originally announced) Date of Restatement Announcement Panel B: Selected Ata from Financial Statements for Enron (1996-2000, in millions) 2000 1999 Net Income $979 $1,024 Cash Flows from Operations 4,779 1,228 Cash Flows from Investing -4,264 -3,507 Activities Average Total Assets 49,442 31,366 Operating Accruals (a) -0.077 -0.007 Total Accruals (b) 0.009 0.105 Size of Earnings Restatement $586 million (as originally announced) Date of Restatement Announcement November 7, 2001 Panel B: Selected Ata from Financial Statements for Enron (1996-2000, in millions) 1998 1997 1996 Net Income $703 $105 $584 Cash Flows from Operations 1,640 501 1,040 Cash Flows from Investing -3,965 -2,436 -1,230 Activities Average Total Assets 26,386 19,780 14,688 Operating Accruals (a) -0.036 -0.020 -0.031 Total Accruals (b) 0.115 0.103 0.0527 Size of Earnings Restatement (as originally announced) Date of Restatement Announcement Enron date was unavailable for the end of the 2001 fisal year. The source of the announcement date is drawn from Wu (2002). (a)Operating Accruals is calculated as the difference between earnings before extraordinary items (item 123) and cash flows from operations (item 308) as reported on the statement of cash flows. This varibale is scaled by average total assets (items 6). (b)Total Accruals is calculated as earnings before extraordinary items (item 123) less cash flows from operations (item 308) less cash flows from investing activities (item 311) as reported on the statement of cash flows. This variable is scaled by average total assets (items 6)
Asquith, P., and L. Meulbroek. 1996. An empirical investigation of short interest. Working paper, Harvard University, Boston, MA.
Bodie, Z., A. Kane, and A. J. Marcus. 2002. Investments. 6th edition. New York, NY: McGraw-Hill.
Bradshaw, M. '1'., S. A. Richardson, and R. G. Sloan. 2001. Do analysts and auditors use information in accruals? Journal of Accounting Research 39: 45-74.
-----, R. G. Sloan, and A. Sweeney. 1996. Causes and consequences of earnings manipulation: An analysis of firms subject to enforcement actions by the SEC. Contemporary Accounting Research 13: 1-36.
-----, A. P. Hutton, L. Meulbroek, and R. G. Sloan. 2001. Short sellers, fundamental analysis and stock returns. Journal of Financial Economics 61: 77-106.
Desai, H., S. R Thiagarajan, K. Ramesh, and B. V. Balachandman. 2002. An investigation of the informational role of short interest in the NASDAQ Market. Journal of Finance 57: 2263-2287.
Figlewski, S. 1981. The information effects of restrictions on short sales: Some empirical evidence. Journal of Financial and Quantitative Analysis 16: 463-476.
Financial Times. 2002. Comment and analysis-Ebbers' undoing. How the deals ran out, doubts set in and the house of cards fell down. (December 20): 13.
Revsine, L., D. W. Collins, and W. B. Johnson. 1999. Financial Reporting and Analysis. Upper Saddle River, NJ: Prentice Hall.
Richardson, S. A., R. G. Sloan, M. Soliman, and A. 1. Tuna. 2002. Information in accruals about earnings persistence and future stock returns. Working paper, University of Pennsylvania and University of Michigan.
-----, A. I. Tuna, and M. Wu. 2002. Predicting earnings management: The case of earnings restatements. Working paper, University of Pennsylvania.
Sloan, R. G. 1996. Do stock prices fully reflect information in accruals and cash flows about future earnings? The Accounting Review 71(3): 289-315.
Teoh, S. H., and T. J. Wong. 2001. Why new issues and high accrual firms underperform: The role of analysts' credulity. Review of Financial Studies 15: 869-900.
Wu, M. 2002. Earnings restatements: A capital market's perspective. Unpublished dissertation, Hong Kong University of Science and Technology.
Scott Richardson is an Assistant Professor at the University of Pennsylvania.
I thank Mark Bradshaw, Jeff Coulton, Patricia Dechow, Ilia Dichev, Andreas Gintschel, Irene Kim, Jim Largay, Venky Nagar, Bill Rees, Richard Sloan, Doug Skinner, Steve Taylor, Irem Tuna, Peter Wysocki, and participants at the University of Michigan and the 2000 European Financial Management Association Conference for their comments. All errors are my own.
|Printer friendly Cite/link Email Feedback|
|Date:||Jan 1, 2003|
|Previous Article:||Investors' perceptions of earnings quality, auditor independence, and the usefulness of audited financial information.|
|Next Article:||Measuring stockholder materiality.|