What Determines the Holding Period of the Average Investor?

What Determines the Holding Period of an Investor?
This study examines the relationship between stock characteristics and the average holding period of an investor.
by Lawrence Kryzanowski and Bin Li
 

In the security market, both individual investors and professional money managers make investment decisions on what and when to buy, sell or hold. An investor's investment performance is determined by the ability to correctly make these types of investment decisions. In this paper, we assess various potential determinants of average holding period behaviour for Toronto Stock Exchange (TSE) 300 stocks, using a measure of investor holding period behaviour that works for the thinner Canadian market.

Our findings should be helpful to investors interested in which stocks are likely to be more liquid because of their greater trading activity; to all market participants, because empirical research suggests that analyst following of stocks (and thus, information generation) is related to the relative level of trading activity1; and to clients of portfolio managers who follow a particular investment style, since investment styles and holding periods are related.

The Measure
In a study of the relationship between investor holding period behaviour and stock (holder) characteristics, we can only observe the holding period of the average (or so-called consensus) investor for each firm. We measure the average holding period for a firm by dividing that firm's shares outstanding by the sum of that firm's shares outstanding and trading volume over a pre-specified time horizon. This relative measure of the average holding period never exceeds the time horizon over which trading volume is measured. This measure always falls in the range between zero and one, where zero indicates continuous trading and one indicates no trading over the period used to measure trading volume.2

Methodology
The average holding period for a firm is determined by the factors that affect the average investor's expected return from investing in that security. In this study, we classify the factors that potentially affect expected stock returns into five families: relative price level, past return performance, trade costs and liquidity, risk, and stockholder characteristics. We do not capture the effect of tax-induced trading on average holding periods. This is left for future study.

We examine the stocks in the TSE300 index from July of 1986 to June of 1996. We consider the beginning of July as the beginning of the studied year, and the end of June of the next year as the end of the studied year. Thus, each year runs from July of this year to June of the next year. The financial ratios for this year for a firm are taken from last year if the firm's fiscal year ended during the third and fourth quarters of the calendar year, and from this year if the firm's fiscal year ended during the first and second quarters of the calendar year.3

Due to the frequency of missing data for some of our determinants of the average holding period, we had to balance off the number of observations used with the number of determinants examined. To assess the robustness of our results to this choice and the impact of multicolinearity among the potential determinants, we ran nine combinations of possible determinants of the relative average holding period for our sample. The potential determinants are presented and described in Table 1.

The Determinants of Average Holding Periods
The regression results are summarized in Table 2.4 To minimize the presentation of statistical jargon, we state that we find a relationship between the average relative holding period and some determinant (for example, value stocks) if the estimated coefficient for this determinant is significant at the 0.05 level.

Determinants Based on the Relative Price Level
T he choice of investment style is considered to be an important step in the investment decision-making process. Equity investing generally is grouped into a value or growth style. Empirical research finds that value strategies outperform growth strategies. The traditional explanation is that higher returns are required to compensate for the higher systematic risk associated with value investing. Some authors argue that book-to-market (and size) proxy for distress, and that distressed firms are more sensitive to certain business cycle factors, such as changes in credit conditions.5 Others argue that investors who incorrectly extrapolate past growth rates in firm earnings generate the high returns associated with high book-to-market (or value) stocks.6 According to the overreaction hypothesis that is based on behavioural finance, investors allegedly are overly optimistic about firms that have done well in the past and are overly pessimistic about those that have done poorly. Since low book-to-market (or growth) stocks supposedly are more glamorous than value stocks, they may attract naive investors who push up prices and lower the expected returns of these stocks.

Studies in behavioral finance find that, when faced with a complex purchase, decision-makers tend to anchor on prices and prices changes as indicators of value. Decision-makers overweight more recent evidence, and over-rely on past trends when formulating future expectations. Being loss averse, investors tend to focus on negative information when under stress by overweighting the probability of negative events. They become more loss averse as downward price movements remind them of their incomplete personal control. Thus, behavioural finance suggests that the average holding period will be longer for growth compared to value stocks, or in-favour compared to out-of-favour stocks. We define value stocks as those with market-to-book values less than one, and growth stocks as those with price/earnings (P/E) ratios higher than a benchmark (for more details, see Table 1). In-favor (out-of-favor) stocks are defined as those with high (low) ratios of market-to-book values, P/E, price-to-cash-flow ratios and price-to-dividend yields.

We find that this year's average holding period for a stock is negatively related to whether or not it is a value stock this year. This suggests that a significant number of investors become overly pessimistic about value stocks, and that their selling activity causes the average holding period to shorten. We find that this year's average holding period is not negatively related to whether the stock was a value stock last year. If buyers or holders of last year's value stocks are value investors, they are expected to keep these value stocks longer. However, the results suggest that last year's value stock buyers and holders do not keep these stocks longer on average during this year. By buying growth stocks at relatively expensive prices (compared to earnings per share), growth investors hope to make an abnormal return through faster stock price growth. If true, rational growth investors are expected to keep growth stocks longer to realize their price appreciation expectations. As expected, we find that this year's average holding periods are longer for this year's growth stocks.

We find that this year's average holding periods become shorter for last year's growth stocks. A possible reason is that a large number of last year's growth stockholders sell their holdings during this year. Since the correlation between last and this year's growth stocks is only 0.411, another possible explanation is that continued high growth from one year to the next year is quite uncertain.

We find that this year's average holding periods are not related to last year's price-to-book ratios, this and last year's dividend yields, and this and last year's price-to-cash flow ratios. However, this year's average holding periods are negatively related with this year's earnings yields, and positively related to last year's earnings yields. Thus, the average investor tends to hold stocks that are in-favour this year for a shorter period of time.

Determinants Based on Past Return Performance
Many papers document the continuation or reversal of stock returns over various periods of time. These U.S. studies find, for example, that momentum strategies which buy stocks that have performed well in the past and sell stocks that have performed poorly generate significant positive returns over three to 12 month holding periods.7 Over multi-year holding periods, other authors find a winner-loser effect, where winners (losers) in one multi-year period are losers (winners) over the subsequent multi-year period.8 The profitability of these momentum strategies is attributed to overreaction induced by positive feedback trading strategies.9 Momentum investors trade stocks based on the trend of return movements and not on fundamental information. "Trend-chasers" reinforce movements in stock prices even in the absence of fundamental information. If "trend-chasing" plays an important role in the performance of winners and losers, then the average holding period will be shorter for both winner stocks and loser stocks than for other stocks.

We find that this year's average holding periods are negatively related with this year's winners and with last year's losers. One possible reason is excess buying of winners and excess selling of losers by investors. These findings are consistent with 'trend-chasing' behaviour by investors. Thus, we find that both winners and losers are traded in Canadian markets. This differs from investor behaviour in U.S. markets where winners are traded and losers are held. We do not find that this year's average holding periods are positively related to this year's market-adjusted excess returns, and to last year's market-adjusted excess returns. Our findings do not support the hypothesis that the average investor prefers to hold stocks with good current or past return performances.

Determinants Based on Trade Costs and Liquidity
Although brokerage commissions are material, the bid-ask spread represents the major part of trade costs. Theory suggests that investors who expect to hold a security longer (shorter) will own stocks with larger (smaller) bid-ask spreads.10 If this is true, then stocks with higher (lower) bid-ask spreads should have longer (shorter) average holding periods. We find that the average holding periods are positively related with spreads. This supports the notion that an average holding-period clientele effect exists based on the relative spreads of various stocks. Determinants Based on Risk

We use various proxies to measure risk; namely, standard deviation of returns, earnings and cash flow stability, number of years included in the TSE 300, debt/equity ratio, firm size and low price. While also can be a risk measure, we do not include it due to a lack of data for its estimation. The standard deviation is based on monthly returns over a one-year period. We use the Stock Guide definition of earnings stability, or the coefficient of determination, as a measure of how well earnings are related over a period of five or ten fiscal years depending upon data availability. This ratio gives some indication of the stability of earnings and the reliability of earnings growth. We use a similar definition for cash flow stability. We expect the debt/equity ratio to be positively correlated with the risk of common equity across firms.11

After controlling for differences in spreads and institutional trade activity, we expect larger firms to have longer holding periods. If we assume that low price stocks, defined as less than $5, are more risky, then such stocks will have shorter holding periods. Since firms with longer listing periods are allegedly less risky because they are more seasoned, we expect such firms to have longer holding periods. We use a similar logic for stocks that are included in the TSE 300.

We find no relationship between this year's average holding periods and the standard deviations of stock returns after accounting for factors like spread and size. As expected, we find a positive relationship between this year's average holding periods and this year's cash flow stabilities, and the number of years that firms have been included in the TSE300 index. We find no relationship between this year's average holding periods and this year's earnings stabilities or this year's debt/equity ratios or this year's low price stocks. As expected, we find a positive relationship between holding periods and size, after controlling for spread and institutional trade activity.

Determinants Based on Stockholder Characteristics
Institutional investors play an increasingly important role in the security market. The allegedly short-term investment horizon of money managers causes them to trade more frequently than those individual investors who have long-term investment horizons. Thus, our expectation is that a stock with larger institutional investor ownership will have a shorter average holding period.

As expected, we find that this year's average holding periods are negatively related with the intensity of institutional trading activity (as measured by large share trades to total trades). This supports the belief that institutional investors trade more frequently than other investors.

Conclusion
Our major findings are that the average holding periods (that is, the holding periods of the average investor) tend to be shorter for:
  • Growth compared to value stocks,
  • Stocks that are winners or losers in the current or in the prior year, and
  • Stocks that currently have higher earnings yields, lower bid-ask spreads, smaller firm sizes (after controlling for differences in spreads and institutional trade activity), less stable cash flows, lower numbers of years included in the TSE300 index, or higher proportions of institutional trade activity.

The findings of this study support the belief among market practitioners that an examination of the stock market must consider the strong possibility that investors are not fully rationale in their trade behavior.

 

Table 1: Possible Determinants of the Relative Average Holding Period
This table lists the potential determinants of the relative average holding period and how they are measured in this study. It also provides our expectation of the nature of the relationship between the relative average holding period and each determinant. Our expectations are positive, negative or unknown.
Determinant Measurement Expected Relationship
Value stock Dummy variable equal to 1 if price-to-book value is less than 1 and 0 otherwise. Both this and last year's values used. Negative (this & last year)
Growth stock Dummy variable equal to 1 if price-to-earnings ratio is greater than the market and equal to 0 otherwise. Both this and last year's values used. Positive (this & last year)
Price-to-book ratio Firm's price per share divided by book value per share. Both this and last year's values used. Positive (this & last year)
Earnings yield Firm's earnings per share divided by price per share. Both this and last year's values used. Negative (this & last year)
Dividend yield Firm's dividends per share divided by price per share. Both this and last year's values used. Negative (this & last year)
Price-to-cash flow Firm's price per share divided by cash flow per share. Both this and last year's values used. Positive (this & last year)
Loser stock Dummy variable equal to 1 if stock is a loser and equal to 0 otherwise. Both this and last year's values used. Negative (this & last year)
Winner stock Dummy variable equal to 1 if stock is a winner and equal to 0 otherwise. Both this and last year's values used. Negative (this & last year)
Market-adjusted Firm's return minus market return. Both this and last excess return year's values used. Positive (this & last year)
Relative spread Firm's weighted-average relative spread for this year. Positive
Standard deviation of return Firm's standard deviation of monthly return for the last year. Negative
Earnings stability Firm's coefficient of determination for earnings over the five-year period prior to this year as calculated by Stock Guide. Positive
Cash-flow stability Firm's coefficient of determination for cash flows over the five-year period prior to this year as calculated by Stock Guide. Positive
Years in TSE300 Number of years that firm has been included in the TSE300 index up to this year. Positive

 

Table 2: Estimated Coefficients of the Determinants of the Relative Average Holding Period Estimates for Nine Regressions
The dependent variable in the regressions is the relative average holding period calculated using our estimator. The definitions of the independent variables are provided in Table 1. The regressions use data for stocks included in the TSE300 index over the 1986-1996 period. * indicates significance at the 0.05 level.
Determinant/Statistic Regression Run
  1 2 3 4 5 6 7 8 9
Intercept 1.207* 1.202* 1.199* 0.206* 0.105* 0.130* -0.070* 0.610* 0.380*
Value stock, this year -0.042* -0.035* -0.041* -0.035* -0.028*     -0.034*  
Value stock, last year -0.000 -0.019* -0.002 0.000 -0.009 0.011   0.013  
Growth stock, this year 0.025*   0.031* 0.039*   0.040*   0.041*  
Growth stock, last year -0.047*   -0.044* -0.044*   -0.042*   -0.038*  
Price-to-book, this year           0.004* 0.004* 0.003* 0.004*
Price-to-book, last year           0.000 0.000 0.000 0.000
Earnings yield, this year   -0.003*     -0.004*   -0.003*   -0.003*
Earnings yield, last year   0.005*     0.003*   0.005*   0.004*
Dividend yield, this year   0.002* 0.001 -0.001 0.001 -0.001 0.000 -0.001 0.000
Dividend yield, last year   0.002* 0.002* 0.001* 0.001 0.001* 0.001 0.001 0.000
Price/cash flow, this year               0.000 0.000
Price/cash flow, last year               0.000 0.000
Loser stock, this year -0.047* -0.045* -0.046* -0.045*   -0.052*   -0.045*  
Loser stock, last year -0.052* -0.046* -0.047* -0.027*   -0.028*   -0.032*  
Winner stock, this year -0.025* -0.032* -0.028* 0.006*   -0.010   0.000  
Winner stock, last year -0.035* -0.034* -0.034* -0.025*   -0.030*   -0.033*  
Market-adjusted return for year, this year         0.026*   0.014   0.014
Market-adjusted return for year, last year         -0.013   -0.019*   -0.031*
Relative bid-ask spread, this year 0.088* 0.086* 0.086* 0.202* 0.194* 0.246* 0.224* 0.229* 0.190*
Firm size, this year       0.322* 0.336* 0.366* 0.390* 0.215* 0.242*
Return standard deviation, this year       -0.090 -0.110* -0.102 -0.131* 0.000 -0.064
Earnings stability, this year               0.005 0.003

 

Endnotes
  1. For example, see Richard Chung and Lawrence Kryzanowski, 2000, "Tests of Investor Cognizance Using Earnings Forecasts of North American Analysts," Working Paper, Concordia University, March.
  2. The commonly used measure of the average holding period for a firm is computed by dividing the number of shares outstanding for a firm by that firm's trading volume over some period of time (typically one year). The value of this absolute measure of the holding period approaches zero when trading volume becomes large relative to shares outstanding, and approaches infinity when trading volume approaches zero. Thus, studies that use this measure generally eliminate thinly traded stocks. For example, see Allen B. Atkins and Edward A. Dyl, 1997, "Transactions Costs and Holding Periods for Common Stocks," The Journal of Finance, March, 309-325. Since Canadian markets are supposedly characterized by thin trading, the absolute measure of average holding period is not well suited for studying Canadian stocks.
  3. To obtain the required data (especially for delisted stocks), several data sources are used in this study. These include the TSE Monthly Review, Stock_Guide, TSE/Western database, Computstat and TSE8696 (i.e, a daily summary file for TSE listed securities).
  4. We obtain comparable results using the log of the absolute average holding period as the dependent variable in these regressions. We obtain similar inferences when we add cross-product terms to capture interaction effects between the determinants to these regressions.
  5. Eugene F. Fama and Kenneth R. French, 1996, "Multifactor Explanations of Asset Pricing Anomalies," The Journal of Finance, March, 55-84.
  6. Josef Lakonishok, Andrei Shleifer and Robert W. Vishny, 1994, "Contrarian Investment, Extrapolation, and Risk," The Journal of Finance, December, 1541-1577.
  7. Narasimhan Jegadeesh and Sheridan Titman, 1993, "Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency," The Journal of Finance, March, 65-91.
  8. A number of studies on U.S. markets report that winners are traded and losers are held by investors. Examples include the examinations of turnover by Lakonishok and Smidt (1986) on the NYSE and AMEX and by Bremer and Kato (1996) on the Tokyo Stock Exchange. Josef Lakonishok and Seymour Smidt, 1986, "Capital Gain Taxation and Volume of Trading," Journal of Finance 41:4, 951-976. Marc Bremer and Kiyoshi Kato, 1996, "Trading Volume for Winners and Losers on the Tokyo Stock Exchange," Journal of Financial and Quantitative Analysis, March, 127-141.
  9. J. Bradford DeLong, Andrei Shleifer, Lawrence H. Summers and Robert J. Waldmann, 1990, "Positive Feedback Investment Strategies and Destabilizing Rational Speculation," Journal of Finance 45:2, 379-395.
  10. Yakov Amihud and Haim Mendelson, 1986, "Asset Pricing and the Bid-Ask Spread," Journal of Financial Economics, 17 (1986), 223-249.
  11. Laxmi Chand Bhandari, 1988, "Debt/Equity Ratio and Expected Common Stock Returns: Empirical Evidence," The Journal of Finance, June, 507-528.

Lawrence Kryzanowski is a professor in the Faculty of Commerce at Concordia University in Montreal, Québec. Bin Li is a consultant in Montreal, Quebec.

Transcontinental Media G.P.