Painting the tape: Aggregate evidence

Painting the Tape: Aggregate Evidence∗ Dan Bernhardt† University of Illinois Ryan J. Davies‡ Babson College April 27, 2005 Abstract We document systematic patterns in daily aggregate market returns around endof-quarters consistent with strategic fund manager behavior and the growing presence of mutual funds in the market. JEL Classification: D82, G2, G14. Keywords: turn-of-quarter effect, painting the tape, mutual fund performance, investment distortion. ∗ Dan Bernhardt acknowledges funding from the National Science Foundation research grant, SES0317700. † Department of Economics, University of Illinois, 1206 South Sixth Street, Champaign IL 61820. Tel: +1–217–244–5708. email: [email protected] ‡ Finance Division, Babson College, Tomasso Hall, Babson Park MA 02457–0310. Tel: +1–781–239–5345. Fax: +1–781–239–5004. email: [email protected] (Corresponding author) Painting the Tape: Aggregate Evidence We document systematic patterns in daily aggregate market returns around end-of-quarters consistent with strategic fund manager behavior and the growing presence of mutual funds in the market. JEL Classification: D82, G2, G14. Keywords: turn-of-quarter effect, painting the tape, mutual fund performance, investment distortion. 1 Introduction [We] found a great many suspicious rises at the point that determines how a share’s or fund manager’s performance will be judged. The price of a stock would jump right at year’s end and fall sharply at the start of the new year. [. . .] The percentage of stocks that became stars on the final trading day and turned into dogs after New Year’s was far greater than for [randomly selected mid-month days.] The same applied to mutual funds, where last-minute leaps beyond the normal market trends strongly suggested a round of ‘portfolio pumping’. [The Globe and Mail, July 7, 2000.1 ] Fund managers have the incentive to distort their investment of new funds at the end of an evaluation period towards stocks in which they hold positions. This trading behavior, sometimes referred to as “painting the tape,” causes a price impact which increases the value of the existing position, and thereby increases the overall fund return. The higher shortterm fund return leads to greater cash inflows (Sirri and Tufano, 1998), greater assets under management, and ultimately greater mutual fund manager compensation. Because the price impact of these trades decay over time, fund managers have especially strong incentives to augment existing holdings near the end of evaluation periods. That is, early in an evaluation period, fund managers may want to minimize investment distortions/price impacts. However, later in the evaluation period, enough of the price impact will persist through the end of the evaluation period to make it attractive to distort investments toward assets in place. An extreme manifestation of this investment distortion is the practice of “high closing”—submitting a buy order that blows through the sell limit order book at the close of a trading day at the end of an evaluation period. As the news magazine Maclean’s observes, “[...] nearly everyone seems to agree that high closing is common. ‘It’s caused by 1 “Hey, that’s my money!” Editorial (Editor: Richard Addis), The Globe and Mail, Thomson Canada Limited, Toronto, July 7, 2000, page A14. 1 the competitive nature of the business,’ says John Gilfoyle, an investment consultant [...] ‘They have to beat the guy across the street.’ ”2 Carhart et al. (2002) show that funds earn tremendous short-run returns near the end of an evaluation period, when trading behavior has the greatest impact on performance. This is especially so for for small-cap funds, which trade less liquid stocks. Carhart et al. find that 80% of funds beat the S&P 500 Index on the last trading day of the year (62% for other quarter-end dates), but only 37% (40% other quarters) do so on the first trading day of a new year. The difference is even greater for small-cap funds: 91% on year end, 70% other quarter end dates versus 34% for first quarter trading day. Carhart et al. (2002) reject benchmark-beating hypotheses in favor of strategic behavior similar to that motivated here. They find that this strategic effect is higher for better past-performing funds, which have more cash on hand. Funds that performed better in the past year earned 42 basis points higher returns on the last trading day and 29 basis points lower on the first trading day than funds with worse historical performance. In the next section, we uncover evidence that because Carhart et al. (2002) measure fund performance relative to the S&P 500 Index, they underestimate the impact of strategic mutual fund trade. Specifically, we find that strategic trading by fund managers appears to impact returns on aggregate market indexes, so that measuring the impact relative to the index understates the total effect. In particular, we find that the return on the equally weighted index on the last trading day exceeds that on other trading days, and especially exceeds that on the first trading day of a quarter. Efficient markets would imply that these return differences should be unpredictable white noise, but we show that the share of all equity in the economy held by mutual funds explains this return difference well. 2 “A Royal Bashing – Regulators accuse top Bay Street players of share manipulation,” by John Nicol, Maclean’s magazine, July 10, 2000, Vol. 113 No. 28, page 39. 2 2 Empirical Evidence Mutual fund managers have especially large incentives to manipulate their portfolio at the end of a measurement period (e.g., quarter). In particular, private investors target their mutual fund investments according to recent past performance (Chevalier and Ellison (1997), Ippolito (1992), Sirri and Tufano (1998), and Goetzmann and Peles (1997)), which creates the incentive for fund managers to take actions at the end of a measurement period, that raise perceived immediate mutual fund returns, at the expense of long-term performance. If this is so, and mutual funds are “large” players in the economy, the daily market return at the end of a measurement period may exceed daily returns on other days. We now document this fact. Indeed, we find that as the aggregate share of all equity held by mutual funds has increased over the years, the end-of-quarter daily return/other daily return difference on the aggregate market indexes has grown, strongly suggesting that fund manager portfolio manipulations have had increasingly-large systematic impacts on aggregate market outcomes. We use the Center for Research in Security Prices (CRSP) Equally- and Value-Weighted Index Returns to calculate (i) the difference in returns on the last trading day of the period (quarter or month) and the return on the first trading day of the following period, and (ii) the difference in returns on the last day of the period versus the average return on the other trading days.3 We obtain mutual funds holdings of corporate equity (Reference FL653064000) and market value of domestic corporations (Reference FL893064195) from the Federal Reserve US Flow of Funds accounts for the first quarter 1970 to third quarter 2001. Figure 1 plots the relation between share of mutual fund holdings of equity and the end-of-quarter abnormal return difference for the equally-weighted index. The figure reveals an extremely strong and significant positive relation between the mutual fund share of aggregate equity and the return difference on the equally-weighted index. The figure also 3 Our reported results are based on CRSP index returns including dividends. Numerically identical results are obtained using CRSP index returns excluding dividends. 3 shows that the end-of-quarter return is far more likely than not to exceed the return on the first day of the next quarter. Indeed, the end-of-quarter return distribution clearly stochastically dominates the beginning-of-quarter return distribution, implying even greater differences were we to adjust for risk. Table 1 presents the estimation results for the regression model ¶ µ Mutual fund holdings of corporate equityt +εt , Abnormal return = β0 +β1 Market value of domestic companiest ¢ ¡ εt ∼ N 0, σ 2 , where the abnormal return is measured as: (i) the turn-of-quarter return (panel A); (ii) last day of quarter versus other days (panel B); (iii) turn-of-month return (panel C); (iv) last day of month versus other days (panel D); and (v) turn-of-month, excluding end of quarter months (panel E). We use both the equally-weighted and value-weighted indexes to calculate returns. First, consider the regression based on the turn-of-quarter returns reported in panel A. The regression fit as measured by the adjusted R2 = 0.20 for the equally-weighted index is remarkable: efficient markets would predict that the return difference should be white noise, i.e., that the adjusted R2 should be zero. The estimate of the coefficient β1 is significant and sharply positive. This accords with the central investment forces that we characterize being large enough to have aggregate consequence. Figure 1 makes clear that this is not driven by outliers. The result is also robust to the exclusion of the last five years of data when both abnormal returns and the share of mutual fund holdings of equity were high. Note that the share of mutual fund holdings of equity does not rise monotonically over time and it has considerably more explanatory power than a simple time trend variable.4 Contrasting the regressions for the equally-weighted and value-weighted indexes reveals 4 Our findings are also robust to the exclusion of turn-of-year quarter ends and thus are not driven by the so-called “January effect”. Furthermore, we note that on an individual stock basis, there is nothing special from a risk-expected return perspective about other end-of-quarter dates (since, for example, earnings announcement dates can occur throughout the quarter). End-of-quarters only matter for measuring the aggregate return performance of fund managers. 4 that the impact of the share of mutual fund holdings on this return difference is only statistically significant for the equally-weighted index. Because the equally-weighted index emphasizes smaller, less liquid stocks for which the short-term price impact of a trade is greater and more persistent. The incentive to distort investment toward stocks in which a fund has a position is especially great if the price impact is more persistent. Thus, investment distortions should be higher for smaller and more specialized funds that disproportionately invest in less liquid stocks (see Bernhardt, Davies, and Westbrook (2005) for a complete theoretical development). Panel B of Table 1 presents the analogous regressions in which the dependent variable is now the difference between the end-of-quarter return and the average return on the other days in the next quarter. The same qualitative pattern remains, but to a reduced degree, and the adjusted R2 remains a substantial 0.09. The weaker relationship highlighted in Panel B relative to Panel A reveals that the market under performs on the first trading day of a quarter, as it “bounces” back from the previous day’s manipulation, as our theory would predict. That is, if fund managers’ presence is large enough to affect returns, then our theory predicts that the return distribution on the last day of the month should dominate that on other days of the month, which in turn should dominate returns on the first day of the month. Panels C–E investigate whether strategic mutual fund behavior explains abnormal endof-month return differences. Panels C and D reveal that while the coefficient estimate for the share of equity held by mutual funds is significantly positive, the model has less explanatory power, as measured by the adjusted R2 , when estimated at monthly intervals. In fact, when the return difference for the last month in a quarter is excluded, as in Panel E, the model has no explanatory power for equally-weighted index returns. Thus, we find that the crucial measurement period is a quarter, which is consistent with Carhart et al. (2002). The end-of-quarter return patterns that we document for the equally- 5 weighted index are consistent with fund managers manipulating purchases at the end of a quarter in order to raise short term returns, especially in illiquid stocks, and then realizing the negative consequences of the distortion the next day. That is, the collective impact of such behavior is so substantial that it affects aggregate outcomes in indexes that weight less liquid stocks more heavily. Indeed, it seems implausibly difficult to construct an alternative explanation for this combination of findings. 3 Conclusion We find that the equally-weighted index return on the last trading day of a quarter is significantly higher than both the return on the first trading day of a quarter and the average return on other trading days; and further that these return differences rise with the percentage of total equity that is held by mutual funds. This strong empirical evidence indicates that the incentives of fund managers to distort investments are so high at the end of a quarter that their behavior significantly alters aggregate market outcomes. A diligent reader might ask: (i) Does mutual fund (or unit trust) investment distortion impact aggregate index returns in other countries with different trading environments and different return-flow sensitivities?; and (ii) Can an individual stock level analysis reveal how much of the aggregate index results are driven by trading in illiquid, small capitalization stocks? Due to space constraints, we leave these interesting questions as topics for future research. References Bernhardt, D., R.J. Davies and H. Westbrook Jr., 2005. Smart Fund Managers? Stupid Money?. Mimeo, University of Illinois. Carhart, M.M., R. Kaniel, D.K. Musto, A.V. Reed, 2002. Leaning for the Tape: Evidence 6 of Gaming Behavior in Equity Mutual Funds. Journal of Finance 57(2), 661–693. Chevalier, J., G. Ellison, 1997. Risk Taking by Mutual Funds as a Response to Incentives. Journal of Political Economy 105(6), 1167–1200. Goetzmann, W. N., N. Peles, 1997. Cognitive Dissonance and Mutual Fund Investors. Journal of Financial Research 20(2), 145–58. Ippolito, R., 1992. Consumer Reaction to Measures of Poor Quality: Evidence from the Mutual Fund Industry. Journal of Law and Economics 35(1), 45–70. Sirri, E.R., P. Tufano, 1998. Costly Search and Mutual Fund Flows. Journal of Finance 53(5), 1589–1622. 7 Return on last day of quarter - return on first day of quarter 4% 3% 2% 1% 0% -1% -2% 0 5 10 15 20 (Mutual fund holdings of corporate equity) / (Market value of domestic companies) [%] Figure 1: Plot of (Return on last trading day of quarter − return on first trading day of next quarter) versus (Mutual fund holdings of corporate equity / Market value of domestic companies). Returns are based on the CRSP Equally-Weighted Index. The solid line is the ordinary least squares estimated fit. 8 Table 1: Abnormal end-of-month and end-of-quarter returns and the share of equity held by mutual funds. Let rt denote the difference between the return on the last trading day of the quarter t, let r̂t denote the return on the first trading day of the next quarter, and let r̄t denote the average daily return calculated over the remainder of the trading days in quarter t. Let rti denote the return on the last trading day of month i, let r̄t,i+1 denote the average daily return calculated over all of the trading days (except the last trading day) in month i + 1, and let r̂ti denote the return on the first trading day of month i + 1, where i = 1 indicates the first month in quarter t, i = 2 indicates the second month in quarter t, and i = 3 indicates the last month in quarter t. Monthly return differences are averaged over each quarter t and are expressed in percent per day. Return differences are expressed in percent. Number of observations: 127. The estimated´regression model is: ³ Mutual fund holdings of corporate equityt Abnormal return = β0 +β1 +εt , εt ∼ N (0, σ 2 ) Market value of domestic companiest Panel A: Turn-of-quarter return. Dependent variable: (rt − r̂t ) Share of equity held Constant by mutual funds (in %) Returns Coefficient P-value Coefficient P-value Equally-weighted −0.0952 0.412 0.0679 0.000 Value-weighted −0.0204 0.897 0.0182 0.258 adj. R2 0.203 0.002 Panel B: Last day of quarter vs. other days. Dependent variable: (rt − r̄t+1 ) Share of equity held Constant by mutual funds (in %) Returns Coefficient P-value Coefficient P-value adj. R2 Equally-weighted 0.168 0.047 0.0375 0.000 0.091 Value-weighted 0.0523 0.327 0.00221 0.426 −0.008 P Panel C: Turn-of-month return. Dependent variable: 13 3i=1 (rti − r̂ti ) Share of equity held Constant by mutual funds (in %) Returns Coefficient P-value Coefficient P-value adj. R2 Equally-weighted 0.100 0.152 0.0175 0.015 0.039 Value-weighted 0.183 0.081 −0.0165 0.121 0.011 Panel D: Last day of month vs. other days. Dependant variable: Share of equity held Constant by mutual funds (in %) Returns Coefficient P-value Coefficient P-value Equally-weighted 0.163 0.005 0.0195 0.001 Value-weighted 0.160 0.033 −0.00251 0.612 1 3 P3 i=1 (rti − r̄t,i+1 ) adj. R2 0.061 −0.007 P Panel E: Excluding turn-of-the-quarter months. Dependent variable: 12 2i=1 (rti − r̂ti ) Share of equity held Constant by mutual funds (in %) Returns Coefficient P-value Coefficient P-value adj. R2 Equally-weighted 0.197 0.045 −0.00776 0.438 −0.003 Value-weighted 0.284 0.042 −0.0339 0.018 0.037 9