Too Good to Ignore? A Primer on Listed Penny Stocks

Transcription

1 Too Good to Ignore? A Primer on Listed Penny Stocks Qianqiu Liu Shidler College of Business University of Hawaii qianqiu@hawaii.edu S. Ghon Rhee Shidler College of Business University of Hawaii rheesg@hawaii.edu Liang Zhang Monash University zlmailbox@yahoo.com 1st draft: August nd draft: March rd draft: November 2012 This draft: March

2 Too Good to Ignore? A Primer on Listed Penny Stocks ABSTRACT This paper identifies characteristics of U.S. penny stocks that are listed on NYSE, AMEX, and NASDAQ. Consistent with our expectation, these stocks exhibit high return, high beta, high idiosyncratic volatility, small capitalization, high BM ratio, and poor liquidity. Contrary to our expectation, however, the average institutional ownership of penny stocks is surprisingly high at 23% and each penny stock is on average owned by 27 institutional investors. The abnormal returns of penny stocks become insignificant in the asset pricing model framework only after the illiquidity factor is introduced to the Fama and French (1993) three-factor model and the Carhart (1997) four-factor model. However, penny stocks with high institutional ownership generate significant abnormal returns, whereas penny stocks with low institutional ownership do not show these abnormal returns. In addition, zero-cost penny stock portfolios that are constructed on firm size, BM ratio, and institutional ownership can yield significant profitability. JEL classification: Keywords: G14; G18 Listed Penny Stocks; Asset Pricing Model; Trading Strategies; Liquidity; Gibbs Effective Transaction Costs; Dimson Beta; Idiosyncratic Volatility; Institutional Ownership 2

3 According to the US Securities and Exchange Commission (SEC), the term penny stock generally refers to low-priced securities of small companies (securities with prices that are less than $5.00 per share). At the end of 2010, 1,083 penny stocks were listed on the New York Stock Exchange (NYSE), American Stock Exchange (AMEX), and National Association of Securities Dealers Automated Quotations (NASDAQ). Penny stocks account for approximately 25% of the total number of stocks that are listed on the three exchanges. The negative connotations of penny stocks are well-known. In fact, the pejorative perceptions that are associated with penny stocks include (but are hardly limited to) the following issues: the extreme illiquidity and high volatility of these securities; pump and dump and short and distort schemes; the propensity of these stocks to be involved in spams and other types of internet fraud; attempts to sell penny stocks through boiler-room operations involving cold calls; and the reputation of these stocks as gambling-like investments. Many websites and newsletters that promote investment in penny stocks even insinuate that they can help investors make enormous returns of 500% or even 1000% in a short period of time. 1 A careful review of SEC s prior announcements about penny stocks suggests that market regulators are more concerned about unlisted penny stocks that are traded in the over-thecounter (OTC) market, particularly stocks that are quoted in the Pink Sheets and Bulletin Board (OTCBB), than about listed penny stocks. 2 As a result, a number of academic papers investigate various issues that relate to OTC penny stocks. These studies may be grouped into four broad categories: (i) market manipulation [Hanke and Hauser (2008); Bӧhme and Holz (2006)]; (ii) financial disclosure [Jiang et al. (2012); Leuz et al. (2008); and Bushee and Leuz (2005)]; (iii) 1 For example, Global Penny Stocks ( provides a list of penny stocks that, it claims, have performed extremely well; this list includes returns ranging from 328% (for Iomega) to 3,487% (for Netegrity). 2 Please refer to SEC s announcements: (i) (ii) and (iii) 3

4 market microstructure [Bollen and Christie (2009); Harris et al. (2008); Macey et al. (2008); Angel et al. (2004); Marosi and Massoud (2004)]; and (iv) IPOs [Brav et al. (2009), Beatty and Kadiyala (2003)]. Two recent studies by Ang et al. (2011) and Eraker and Ready (2011) are closely related to our study because both of these investigations focus on the pricing aspect of a comprehensive set of over-the-counter (OTC) stocks. Ang et al. (2011) find that the cross sectional return patterns of OTC stocks and listed stocks are similar. They also report that the premiums that are associated with value/growth and size dimensions of OTC stocks are consistent with the corresponding premiums of exchange-traded stocks; however, the illiquidity premium of OTC stocks reaches 19.2% which is significantly greater than the illiquidity premium of 1.1% for comparable listed stocks. Eraker and Ready (2010) calculate an average return of -32% for OTC stocks. These researchers note that this return cannot be explained by the traditional valuation models but that the behavioral model of Barberis and Huang (2008) is able to explain the large negative returns. 3 Because low-priced listed stocks (that are listed on a stock exchange but traded for less than $5.00 per share) are also called penny stocks, these stocks suffer from the same types of degradation as unlisted penny stocks. To maintain their listing status, listed penny stocks must file their financial reports with the SEC. These stocks must also meet the listing maintenance requirements (most notably, a minimum quantity of net assets and a minimum number of shareholders) that are imposed by their stock exchanges. 4 Certainly, shares of certain listed companies are traded below $5.00 because these firms are experiencing financial distress. 3 Kumar s (2009) investigation yields similar results to the findings of Eraker and Ready (2012); however, there is an important difference between these two research studies. In particular, Kumar s sample stocks are limited to listed stocks with gambling-like features, such as high idiosyncratic skewness, low price, and idiosyncratic risk. 4 Companies quoted on the OTCBB or the Pink Sheets have no obligation to meet these listing standards. Companies that are quoted in the Pink Sheet do not have to file financial reports with the SEC, whereas OTCBB stocks are obligated to submit these types of filings. 4

5 However, except for stocks with jeopardized listing statuses due to financial distress, these listed penny stocks should not be maligned. For instance, consider the second and third largest banks in the US. Bank of America s stock traded below $3.00 per share during the recent subprime mortgage crisis. The stock of Citigroup Inc. also traded below $5.00 between 2009 and May 2011; in May 2011, the firm implemented a 1-for-10 reverse stock split that significantly increased the price of its stock shares. Unfortunately, we know very little about listed penny stocks because academic researchers typically exclude these stocks from consideration. However, one out of every four listed stocks belongs to this penny category. In particular, among the three stock exchanges, penny stocks represent 64% of the stocks that are listed on the AMEX, 31% of the stocks that are listed on the NASDAQ, and 7.4% of the stocks that are listed on the NYSE. Another little-known fact is that institutional investment in penny stocks is surprisingly extensive. On average, 23% of the shares of listed penny stocks are held by institutional investors, and 73% of listed penny stocks are held by one or more institutional investors. The implications of this larger-thanexpected institutional investment in listed penny stocks warrant careful scrutiny. In this study, we examine listed penny stocks. Our objectives are fourfold: first, we examine the characteristics of these penny stocks and compare them with those of comparable, low-priced non-penny stocks; second, we identify the factor in the framework of the asset pricing model that can explain penny stock returns; third, we construct investment strategies that allow for the performance of penny stocks to be evaluated relative to non-penny stocks; and fourth, we examine the impact of institutional holdings on the asset pricing and investment strategies for penny stocks. Our results are summarized as follows. Consistent with general perceptions, penny stocks typically exhibit small market capitalization, high beta, high idiosyncratic volatility, high BM ratio, poor liquidity, and high transaction costs. If an additional illiquidity factor is 5

6 introduced into the Fama and French (1993) three-factor model and the Carhart (1997) four-factor model, the abnormal returns of penny stocks become insignificant. We further examine the profitability of penny stocks through the use of zero-cost portfolios that have been constructed based on not only the level of institutional ownership of each stock but also other firm characteristics (such as firm size, value, momentum, idiosyncratic volatility). Motivated by an intriguing finding that zero-cost portfolios built on the level of institutional ownership yield positive returns, we examine the impact of institutional ownership on penny stock performance in the asset pricing model framework. After sorting penny stocks into two groups (one group with high institutional ownership and another group with low institutional ownership), the five-factor model suggests that penny stocks with high institutional ownership possess positive and significant intercept term (0.48% per month) whereas penny stocks with low institutional ownership have the insignificant estimated intercept term of 0.10%. The paper is organized as follows: Section I identifies the distinct characteristics of penny stocks relative to non-penny stocks. Section II examines whether penny stocks yield higher returns than non-penny stocks in the context of one- to five-factor models. Section III investigates different trading strategies for penny stocks. Section IV analyzes the impact of institutional ownership on abnormal returns from penny stocks in the asset pricing model framework. Section V presents conclusions. I. Characteristics of Penny Stocks A. Data CRSP is the source of the price data that are used in this study, which include the daily and monthly NYSE, AMEX, and NASDAQ stock returns from July 2001 to December Because penny stock returns are sensitive to the minimum tick size, which can range from $1/32 to $1/8, we conduct our study in the post-decimalization period, which involves price quotations 6

7 that are provided in one penny increments. 5 NYSE and AMEX shifted all of their stocks to decimal prices on January 29, 2001, and NASDAQ completed its decimalization of stock prices on April 9, COMPUSTAT is the source of the accounting data for individual stocks. For each examined firm, we match all of the data for the fiscal year that ends in calendar year t-1 with the returns from July of year t to June of t+1. This matching scheme allows stock returns to be explained by accounting variables with a degree of time lag. In particular, we use a firm s market equity at the end of December of year t-1 to compute its BM ratio for year t-1, and we match this ratio to the returns from July of year t to June of t+1. The July 2001 beginning of the study period is chosen to facilitate this matching scheme. Firm size, as measured by market capitalization, is updated on a monthly basis. Finally, we use NYSE/AMEX/NASDAQ index returns as the market returns and one-month Treasury bill yields as the risk-free rates in this study. Although the SEC uses the $5.00 benchmark price in its definition of penny stocks, we must exercise particular caution in identifying penny stocks. We adopt two selection rules. First, at the beginning of each month during the study period, we examine the preceding one-year period to compute the average price for each examined stock. If the average price of a stock is less than $5.00 per share during this retrospective examination, then the stock in question is regarded as a penny stock. A stock s price may certainly increase beyond $5.00 in the months following this computation; if this stock were subsequently excluded from the penny stock portfolio, the portfolio return may be understated. To address this consideration, a second rule is introduced that states that once a penny stock is identified, we grant it a one-year grace period; during this grace period, the stock remains a penny stock regardless of its price. To confirm the robustness of our results and to examine the effects of price level on stock performance, we 5 Please refer to the SEC s Report to Congress on Decimalization (2012) for more information regarding the impact of the decimalization process. 7

8 construct three portfolios of penny stocks: (i) penny portfolio 1 (Price $1); (ii) penny portfolio 2 ($1 < Price $3); and (iii) penny portfolio 3 ($3 < Price $5). To avoid the over- and underestimation of portfolio returns, the one-year averaging rule and the one-year grace period rule remain applicable to all three portfolios based on the examined prices per share of the component stocks of each portfolio. To make the comparison between penny and non-penny stocks more meaningful, we create a category of low-priced non-penny stocks, which are stocks with share prices that are between $5.00 and $ The one-year averaging rule and the one-year grace period rule are also applied to non-penny stocks. A total of 1,307 non-penny stocks is identified. To compare these non-penny stocks with the three penny stock portfolios, we create three non-penny stock portfolios: (i) non-penny portfolio 1 ($5 < Price $8); (ii) non-penny portfolio 2 ($8 < Price $11); and (iii) non-penny portfolio 3 ($11 < Price < $15). B. The Characteristics of Penny Stocks Panel A of Table I reports the returns and prices of penny and low-priced non-penny stocks: three penny portfolios and three non-penny portfolios. During the study period, a total of 4,621 common stocks are listed on the three exchanges, including 1,083 penny stocks and 1,307 low-priced non-penny stocks. The average share prices of these stocks are $2.57 for the penny stocks and $9.57 for the low-priced non-penny stocks. Penny portfolio 1 contains a total of 152 component stocks and featured an average share price of $0.67. This average may appear unusual given the $1.00-delisting rule that has been imposed by the NYSE, AMEX, and NASDAQ since September 1991; however, average prices below $1.00 are possible because a grace period (of 60 days or longer) is granted to listed stocks that trade below $1.00 per share and because the stock 8

9 exchanges occasionally lift their delisting rules in response to market conditions. 6 For example, this rule was suspended between October 2008 and July [Insert Table I] The equally-weighted (EW) average returns are 1.66% per month for penny stocks and 0.87% for low-priced non-penny stocks during the one-month holding period (month t) that immediately follows the portfolio formation (which occurs at month t-1). The annualized return for penny stocks is 19.92%, which dramatically contrasts with the large negative returns (-31% to -38%) that are calculated by Eraker and Ready (2012) for unlisted penny stocks. The EW average monthly returns of the three penny portfolios ranges from 1.09% to 2.60%; penny stocks with lower price levels produce higher average returns. By contrast, the three non-penny portfolios do not demonstrate the same trend as their penny stock counterparts; in particular, the average monthly returns during the examined one-month holding period were 0.87% for nonpenny portfolio 1, 0.81% for non-penny portfolio 2, and 0.90% for non-penny portfolio 3. The EW returns in month t and month t-1 indicate that return momentum exists for all three penny portfolios but not for the non-penny portfolios. One possible cause of this phenomenon is the limits to arbitrage that are expected for the penny stocks. The last three columns of Panel A report the statistics on institutional holdings. These data are from the Thomson/CDA database. 7 On average, each firm that issues penny stocks has 27 investors, whereas each firm that issues low-priced non-penny stocks has 68 institutional investors. Each of the firms in penny portfolios 1, 2, and 3 has an average of 15, 21, and 36 institutional investors, respectively. By contrast, we find that the non-penny firms have a much greater number of institutional investors than the penny firms; in particular, each of the firms in non-penny portfolios 1, 2, and 3 has an average of 52, 67, and 85 institutional investors, 6 Please refer to Rhee and Wu (2012) for more details regarding the $1.00 delisting rules. 7 The SEC requires institutional investors with $100 million or more in equity securities to file certain information on a quarterly basis. 9

10 respectively. The average percentage of a firm s stock that is held by institutional investors is 22.64% for firms that issue penny stocks and 47% for firms that issue non-penny stocks. This average percentage increases from 8.81% for penny portfolio 1 to 30.83% for penny portfolio 3, demonstrating that institutional investors own greater percentages of firms that have stocks with higher share prices. A similar trend is observed for non-penny stocks, as this average percentage increases from 41% for non-penny portfolio 1 to 52% for non-penny portfolio 3. In general, institutional investors have shifted their preference away from large, conservative stocks and towards smaller, riskier stocks since the 1990s [Bennet, Sias, and Starks (2003) and Blume and Keim (2011)]. 8 We summarize the risk and liquidity measures of the examined penny and non-penny stocks in Panel B of Table I. Dimson s (1979) beta is estimated for these stocks because they are infrequently traded. The average estimated values of Dimson s beta are 1.99 for penny stocks and 1.46 for non-penny stocks. The Dimson s beta estimates decline as the price level increases. Idiosyncratic risk is measured by idiosyncratic volatility (IV) from the residuals of the Fama- French (1993) three-factor model regression, following the approach of Ang et al. (2006). The average idiosyncratic volatilities are 17.99% for penny stocks and 10.13% for non-penny stocks, suggesting that the returns of penny stocks are approximately 75% more volatile than the returns of non-penny stocks. Firm size is measured by market capitalization. The median market capitalization of the penny stocks is estimated to be $134 million; by contrast, the median market capitalization for the examined non-penny stocks is estimated to be $573 million. Firm size is positively related to price level; high-priced stocks typically exhibit larger market capitalization. The average BM ratio of the penny stocks is higher than the average BM ratio of the examined non-penny stocks 8 Blume and Keim (2011) report that portfolio allocations to stocks in the lowest ten percent of market capitalization increased from 3.5% in 1980 to 10.4% in

11 (1.16 for penny stocks vs for non-penny stocks). All of the penny stock portfolios have average BM ratios of at least 1; thus, for each of these portfolios, the average book value of the component firms is higher than or equal to the average market value of these firms. The BM ratio monotonically declines as stock price increases. Illiquidity problems such as large bid-ask bounces and high transaction costs are particularly severe for penny stocks. To assess the poor liquidity of penny stocks, we calculate four metrics for each assessed set of stocks: (i) the percentage of zero-return days; 9 (ii) the Amihud (2002) measure; (iii) the Gibbs effective transaction costs; and (iv) the Corwin-Schultz (2012) bid-ask estimates using daily high and low prices. On average, the examined penny and non-penny stocks exhibit zero returns on 12.13% and 9.21% of trading days, respectively. 10 The Amihud (2002) liquidity measure is estimated by the square root of the quotient obtained by dividing the average daily absolute return ( r d ) for a security by the security s daily dollar volume (VOL d ): [ r d /VOL d ] 1/2. The advantage of this measure is that it can be calculated for days involving no price changes, which occur relatively frequently for penny stocks. A low value for this metric implies high liquidity by indicating that large trading volumes can be accommodated with small price changes; by contrast, a high value of this metric implies low liquidity by indicating that a large volume cannot be absorbed without a large price change. By the Amihud liquidity measure, the degree of illiquidity of penny stocks is approximately three times greater than the degree of illiquidity of low-priced non-penny stocks (1.3 vs. 0.48). The Amihud ratio declines as the price level increases; in particular, penny portfolio 1 demonstrates an Amihud ratio of 2.24, whereas non-penny portfolio 3 has the ratio of Ang et al. (2011) use the proportion of non-trading days to measure the degree of illiquidity. 10 Among the three exchanges, AMEX demonstrates the largest percentage of zero-return days; in particular, for the examined penny and non-penny stocks, 15.43% and 13.34%, respectively, of AMEX trading days are zero-return trading days; 11.15% and 9.03%, respectively, of NASDAQ trading days are zero-return trading days; and 10.03% and 7.10%, respectively, of NYSE trading days are zero-return trading days. 11

12 Hasbrouck (2009) advocates the use of the Gibbs effective transaction cost as an alternative measure of liquidity. The Gibbs effective transaction costs can be estimated from daily data during the calendar year and exhibit a high correlation with the bid-ask spread. 11 To accurately estimate these costs, stocks of interest must be traded on at least 50 trading days during a particular calendar year. The Gibbs effective transaction cost is a Bayesian version of Roll s (1984) transaction cost metric. 12 The Gibbs effective transaction cost of penny stocks is 1.42% of their share price, whereas the Gibbs effective transaction cost is only 0.72% of the share price for low-priced non-penny stocks. The cost estimates suggest that the round-trip trading cost is 2.84% of the average share price for penny stocks and 1.44% for non-penny stocks, which is consistent with the findings of previously published reports. Gibbs round-trip transaction costs are greatest for penny portfolio 1 (at 4.64% of share price) and lowest for non-penny portfolio 3 (at 1.24% of share price). 13 This extreme illiquidity among listed penny stocks allows us to examine the role of liquidity as an additional factor in the asset pricing model framework, a topic that will be discussed in the following section. We use the Corwin and Schultz (2012) method of estimating bid-ask spreads using daily high and low prices. Their bid-ask spread estimates exhibit a cross-sectional correlation of with high frequency TAQ effective spreads in the post-decimalization period which is comparable to our study period. The last column presents high-low spreads. Penny [non-penny] stocks show the average spread of 2.63% [1.42%]. The bid-ask spread estimates declines as the 11 Hasbrouck (2009) demonstrates that the Gibbs effective transaction costs from daily CRSP data and the bid-ask spread estimates from high-frequency Trade and Quote Database (TAQ) data have a high correlation of over a time period ( ) during which both high-frequency TAQ and daily CRSP data are available. 12 Please refer to Joel Hasbrouck s website: 13 Previous literature has documented that the round-trip trading costs for large-capitalization stocks are generally between 1% and 2% of their share price, whereas for small-capitalization stocks, estimated round-trip trading costs are much larger, ranging from 5% to 9% of share price [Stoll and Whaley (1983); Keim and Madhavan (1998)]. 12

13 price level increases, ranging from 4.28% for penny portfolio 1 with $0 < P $1 to 1.23% for nonpenny portfolio 3 with $11 < P $15. II. Do Penny Stocks Earn High Abnormal Returns? We run time-series regressions against the three Fama-French (1993) factors and the Carhart (1997) momentum factor, which captures the medium-term continuation of returns that was originally documented by Jegadeesh and Titman (1993). If the intercept (Jensen s alpha) of these regressions is significantly different from zero, then the loadings of these three or four factors are not sufficient to explain the observed portfolio return: where r p p p p = a + β MKT + β SMB + β HML + β UMD + ε, (1) p, t p MKT t SMB t HML t UMD t p, t r p, t is the excess return on the portfolio of penny stocks or non-penny stocks at month t, MKT is the market excess return, SMB is the difference between the return on a portfolio of small-cap stocks and the return on a portfolio of large-cap stocks (the size premium), HML is the difference between the return on a portfolio that is composed of high BM stocks and the return on a portfolio that is composed of low BM stocks (the value premium), and UMD is the difference between the return on a portfolio that is composed of stocks with high returns from t - 12 to t - 2 and the return on a portfolio that is composed of stocks with low returns from t - 12 to t - 2 (the momentum premium). Table II reports the results of time-series regressions of the EW monthly excess returns on penny stock portfolios and non-penny stock portfolios in the one-, three-, and four-factor models. The intercept terms that are estimated for the penny stocks as a whole are 1.08% for the single-factor CAPM, 0.68% for the three-factor model, and 0.74% for the four-factor model. 14 From an economic perspective, these values are rather large. Moreover, the intercept increases 14 We observe that the estimated intercept terms are smaller over the course of our study period than over the course of the longer study period, , that was examined in an earlier version of this paper. We believe that decimalization must have impacted the profitability of penny stocks because the new time period of this study is limited to the post-decimalization period of July December

14 and becomes statistically more significant for stocks with lower prices. For example, within the penny portfolio 1, which features stocks with share prices less than $1, the alphas from all of the conventional models are greater than 1.50% at the monthly level; these alpha values are significant at the 10% level in the CAPM and four-factor models. The significant coefficients on the size and momentum factors indicate that penny stocks behave similarly to small stocks and loser stocks. Among all of the regressions for penny stocks, none of the adjusted R-squares from these conventional asset pricing models is greater than 86%. By contrast, the three- and fourfactor models can explain at least 91% of the return variation of all of the examined non-penny stocks; most of these models produce R-square values greater than 95%. The alphas in these models are both statistically insignificant and economically small. [Insert Table II] Although penny stocks earn abnormal returns relative to the single-factor CAPM and the four-factor model, it remains premature to conclude that penny stocks are truly profitable. In fact, in the presence of high transaction costs, an apparently profitable investment strategy may not be economically feasible. Because penny stocks are illiquid, they must generate high expected returns to compensate investors for their inability to engage in timely trades [Amihud and Mendelson (1986), Brennan et al. (1998), and Datar et al. (1998)]. To examine this issue, we add one additional liquidity factor to the conventional four-factor model: r p,t = α p + β MKT (MKT t ) + β SMB (SMB t ) + β HML (HML t ) + β UMD (UMD t ) + β LIQ (LIQ t ) + ε p,t (2) Where, LIQ t is the aggregate liquidity risk factor, which is constructed using the difference between the return on the portfolio with highest Gibbs transaction cost (lowest liquidity) stocks and the return on the portfolio with lowest Gibbs transaction cost (highest liquidity) stocks. The bottom row of each panel in Table II reports the results of time-series regressions that are performed after the incorporation of the liquidity factor into the four-factor model. We observe 14

15 that penny stocks load heavily and positively on the liquidity factor: the coefficients of this factor are strongly significant in all of the regressions for the penny stocks as a whole and for penny portfolios 1, 2, and 3; in each of these cases, the t-statistic values are at least 9.52, indicating that penny stocks are extremely illiquid. After the liquidity factor is included in the four-factor model, the intercepts of all penny stock portfolios become very small and insignificant. More importantly, the new five-factor model demonstrates a substantially increased adjusted R-square value relative to prior models; in particular, a total of 97% of the return variations of all penny stocks are explained by the extended five-factor model, whereas the conventional four-factor model explains only 75% of these variations. The results from the penny portfolios reveal that liquidity risk is very important for understanding the high returns of penny stocks. All three of the examined penny portfolios demonstrate similar results to the penny stocks as a whole. By contrast, the intercept term that is estimated for the low-priced non-penny stocks is 0.35% in the single-factor CAPM, which is only about one third of the intercept term that is estimated for the penny stocks (1.08%). The estimated intercept terms become negative and insignificant in the three- and four-factor models. The addition of the liquidity factor to the model does not produce substantial changes for the low-priced non-penny stocks; in particular, the estimated intercept term remains negative and insignificant in the five-factor model. Moreover, although these non-penny stocks typically involve significant loadings on the liquidity factor, the five-factor model evinces an adjusted R-square value that is less than 5% greater than the R-square values of the traditional three- and four-factor models. In summary, we document that the extended five-factor model can explain the returns of the penny stock portfolios very well, whereas the additional liquidity factor contributes very little to the characterization of the examined non-penny stock portfolios. We conclude that the positive abnormal returns of penny stocks from the conventional models are largely explained by the 15

16 absence of the liquidity factor from these models. Penny stocks do not earn abnormal returns after this additional risk factor is incorporated into model estimations. 15 III. Penny Stock Trading Strategies Many studies have shown that certain investment strategies based on firm characteristics could produce considerable abnormal profits. For instance, published approaches have sought to exploit the size effect [Banz (1981)]; the BM effect [Basu (1983)]; the short-term return reversal effect [Jegadeesh (1990)]; the momentum effect [Jegadeesh and Titman (1993)]; and the volatility effect [Ang et al. (2006)]. These investment strategies could generate abnormal returns by holding zero-cost portfolios that take long and short positions in subsets of stocks that are ranked on the basis of firm characteristics. In this section, we examine whether investors can make abnormal profits from penny stocks and comparable non-penny stocks by employing these types of characteristics-driven investment strategies instead of holding long positions in all penny stocks, as assumed in the asset pricing model framework. In this study, for the creation of zerocost portfolios, we consider not only the traits listed above but also the additional firm characteristic of the amount of institutional ownership. For the purposes of comparison, we also examine the profitability of the same set of investment strategies for low-priced non-penny stocks. We construct trading strategies based on J = one-month formation period and K-month holding periods where K = one-, three-, six-, nine- and 12-month periods. 16 At time t, we sort penny stocks into quintile portfolios according to their characteristics, such as firm size, BM ratio, the previous one-month return, momentum return, idiosyncratic volatility or institutional ownership, we then hold those portfolio for the next K-months. The portfolios are rebalanced 15 The time series regressions using value-weighted average excess returns for both penny and non-penny stocks provide qualitatively similar findings; however, these results are not reported because space limitations. 16 The momentum strategy is the only exception with the formation periods of J = six- and 12-months considered. For the momentum strategy, no one-month holding period abnormal returns are compiled because past empirical evidence suggests that no abnormal returns for a one-month holding period in the US market. 16

17 each month; moreover, after the portfolios are formed, the EW return of each portfolio is calculated every month in percentage terms over the course of different holding periods. Subsequently, we run the time-series regressions of the excess returns on each quintile portfolio against the 5-factor model and report the abnormal returns of zero-cost portfolios under each trading strategy. The average monthly abnormal returns are the differences in abnormal returns of two quintile portfolios with largest and smallest firm-level characteristics. The Newey and West (1987) heteroskedasticity and autocorrelation consistent t-statistics are computed for the significance tests. The same procedures are repeated for low-priced non-penny stocks. The results are summarized in Table IV. The upper [bottom] panel reports the results for penny [non-penny] stocks. Column 3 reports the results of firm-size based trading strategies. The zero-cost portfolio that buys the smallest penny quintile portfolio and sells the largest penny quintile portfolio provides the monthly EW abnormal return 1.82% with a t-ratio of 4.52 during the one-month holding period. As the holding period increases, the abnormal returns of size-based investment strategy remain significant although the magnitude of these returns declines. By contrast, no abnormal returns are found for the zero-cost portfolios of non-penny stocks. Column 4 reports the results of trading strategies based on the BM ratios. The zero-cost portfolio (which is long quintile 5 portfolio of value penny stocks and short quintile 1 portfolio of growth penny stocks) yields EW abnormal returns that range from 1.05% (for a 12-month holding period) to 1.29% (for a one-month holding period) while these strategies do not work for non-penny stocks with the exception of one-month holding period. [Insert Table III] Column 5 reports the results for the monthly abnormal returns of contrarian trading strategies based on one-month short-term return reversals. No significant returns are detected for the zero-cost portfolios of both penny and non-penny stocks. Column 6 presents the results based on the idiosyncratic volatility-based strategy for penny well as non-penny stocks. The volatility- 17

18 based zero-cost investment strategies earn significant abnormal profits for penny stocks in the one- and three-month holding periods, but not for the longer holding periods of six- through 12- month. In contrast, non-penny stocks show significant abnormal returns for all five holdings periods considered in support of Ang et al. (2006). 17 Column 7 reports the abnormal return for the zero-cost portfolios constructed on the basis of the magnitude of institutional ownership. For penny stocks, zero-cost portfolios provide positive abnormal returns for all five holding periods; these abnormal returns are not observed for comparable zero-cost portfolios of non-penny stocks. A number of factors may be responsible for this interesting finding. The institutional investors may possess superb stock selection skills and information advantages relative to individual investors that allow these institutions to successfully identify high-quality firms among a selection of financially distressed companies. Another possibility is that limits to arbitrage may create profitable opportunities for institutional investors. The last column presents the results for the momentum strategies. To examine if the momentum strategy works for penny stocks, at the beginning of month t, the penny stocks are ranked in the ascending order on the basis of the past six- and 12-month returns. The performance of zero-cost portfolios is examined in the next three-, six-, nine- and 12-month periods. We examine a total of 8 strategies, where J = six or 12 months and M = three, six, nine, and 12 months, respectively. Penny stocks yield insignificant returns for the four holding periods. By contrast, non-penny stocks show positive momentum profits, but only for the six-month holding period. 17 Huang et al. (2010) demonstrate that the omission of the previous month s stock returns can lead to a negatively biased estimate of the examined relationship. In a follow up study, Huang et al. (2011) demonstrate that the dominance of loser stocks in December and the reversal effect in the subsequent month create a positive relation between idiosyncratic volatility and portfolio returns in January; however, for the remainder of the year, a negative relationship arises between idiosyncratic volatility and subsequent monthly returns for portfolios composed of the best-performing stocks of relatively large size and the worst-performing stocks with smaller market values. 18

19 In summary, it is possible to obtain statistically and economically significant returns from penny stocks. Investment strategies based on the firm size, BM ratio, and institutional holding can generate considerable profits over both short and long holding periods. However, the short term contrarian investment strategies, the momentum strategies, and the volatility-based investment strategies are only profitable in the short run and cannot persist. Naturally, further analyses are warranted to understand why the results vary depending on the weighting scheme and the different holding periods. IV. The Impact of Institutional Holdings on Asset Pricing Motivated by an intriguing finding on the positive profitability of trading strategies built on the level of institutional ownership in Section III, we examine the impact of institutional ownership on penny stock performance in the asset pricing model framework. We sort penny stocks into two groups: one group with high institutional ownership (greater than the median value) and another group with low institutional ownership (less than the median value). We rerun the 5-factor model and Table IV presents the results. As summarized in Panel A of Table IV, one striking result emerges: penny stocks with high institutional ownership possess positive and significant intercept term (0.48% per month) whereas penny stocks with low institutional ownership have the insignificant estimated intercept term of 0.10%. This trend becomes more conspicuous as the share prices of the examined penny stocks decrease. Penny portfolio 1 (with $0 < P $1) demonstrates intercepts of 1.40% and % for the subgroups with high and low institutional ownership, respectively. Penny portfolio 2 ($1 < P $3) possesses intercepts of 0.66% and 0.21% for the subgroups with high and low institutional ownership, respectively. Only penny portfolio 3 which includes stocks with share price ranges of $3 < P $5 produces insignificant intercept estimates for these subgroups. Similarly, non-penny stocks do not show any significant intercept terms as shown in Panel B of 19

20 Table IV. We also observe that the liquidity factor has significant loadings for all penny stocks but the significance level declines as the share prices of these stocks increase. [Insert Table IV] V. Conclusion Although many industry practitioners believe that penny stocks are high-risk, highreward investments, there is limited academic research to support this view. This paper comprehensively examines the characteristics and pricing behaviors of penny stocks. Our efforts are valuable because a significant portion of US listed stocks are penny stocks (as much as one quarter of all listed stocks) that are traded below $5.00 per share. Penny stocks are characterized by high return, high beta, high BM ratio, high idiosyncratic volatility, and low liquidity. Our time series analyses suggest that penny stocks do not earn abnormal positive profits in the five-factor asset pricing model framework (which includes the factors of size, BM, momentum, and liquidity). However, penny stocks with high institutional ownership generate significant abnormal returns, whereas penny stocks with low institutional ownership do not produce these abnormal returns. We conduct further analyses to investigate whether popular investment strategies for non-penny stocks that have been developed on the basis of firm characteristics are effective for penny stocks. Although the aggregate penny stock portfolio does not earn abnormal returns in the context of the five-factor model, investors can nonetheless realize abnormal profits from penny stocks through the use of certain characteristics-based investment strategies. We have considered firm size, value, contrarian, momentum, and idiosyncratic volatility investment approaches. Our paper contributes to the available understanding of the characteristics and pricing behavior of penny stocks; however, this study represents only an initial foray that must be followed by future research to elucidate the characteristics of listed penny stocks. 20

21 REFERENCES Angel, J. J., Harris, J., Panchapagesan, V. and Werner, I. R., 2004, From pink slips to pink sheets: Liquidity and shareholder wealth consequences of Nasdaq delistings, Washington University Working Paper. Amihud, Yakov, 2002, Illiquidity and stock returns: cross-section and time-series effects. Journal of Financial Markets 5(1): Amihud, Yakov and Haim Mendelson, 1986, Asset pricing and the bid-ask spread, Journal of Financial Economics 17, Ang, Andrew, Assaf A. Shtauber, and Paul C. Tetlock, 2011, Asset pricing in the dark: The cross section of OTC stocks, Columbia University Working Paper. Ang, Andrew, Robert J. Hodrick, Yuhang Xing and Xiaoyan Zhang, 2006, The cross-section of volatility and expected returns, Journal of Finance 61, Banz, R.W., 1981, The relationship between return and market value of common stocks, Journal of Financial Economics 10, March, 3-l 8. Barberis, Nicholas and Ming Huang, 2008, Stocks as lotteries: The implications of probability weighting for security prices, American Economic Review 98, Basu, Sanjoy, 1983, The Relationship between earnings yield, market value, and return for NYSE common stocks: Further evidence, Journal of Financial Economics 12, Beatty, Randolf and Padmina Kadiyala, 2003, Impact of the penny stock rreform Act of 1990 on the Initial Public Offering Market, Journal of Law and Economics 46, Bennett, James A., Richard W. Sias, and Laura T. Starks, Greener pastures and the impact of dynamic institutional preferences. Review of Financial Studies 16, Blume, Marshall E., and Donald B. Keim, 2011, Institutional investors and stock market liquidity: Trends and relationships, University of Pennsylvania Working Paper. Ekkehart Boehmer and Eric K. Kelley, Institutional Investors and the Informational Efficiency of Prices, Review of Financial Studies 22, Bӧhme, Rainer and Thorsten Holz, 2006, The effect of stock spam on financial markets, University of Mannheim Working Paper. Bollen, Nicolas P.B. and William G. Christie, 2009, Market microstructure of the pink sheets, Journal of Banking & Finance 33, Brav, Alon, Roni Michaely, Michael Roberts, and Rebecca Zarutskie, 2009, Evidence on the trade-off between risk and return for IPO and SEO firms, Financial Management, pp

22 Brennan, Michael J., Tarun Chordia, and Avanindhar Subrahmanyam, 1998, Alternative factor specifications, security characteristics, and the cross-section of expected stock returns, Journal of Financial Economics 49, Bushee, Brian J. and Christian Leuz, 2005, Economic consequences of SEC disclosure regulation: evidence from the OTC bulletin board, Journal of Accounting and Economics 39, Carhart, Mark M., 1997, On persistence in mutual fund performance, Journal of Finance 52, Corwin, Shane A. and Paul Schultz, 2012, A simple way to estimate bid-ask spreads from daily high and low prices, Journal of Finance 67, Datar, Vinay T., Narayan Y. Naik, and Robert Radcliffe, 1998, Liquidity and asset returns: An alternative test, Journal of Financial Markets 1, Dimson, Elroy, 1979, Risk measurement when shares are subject to infrequent trading, Journal of Financial Economics 7, Eraker, Bjørn and Mark Ready, 2011, Do investors overpay for stocks with lottery-like payoffs? An examination of the returns on OTC stocks, University of Wisconsin Working Paper. Fama, Eugene F., and James D. MacBeth, 1973, Risk, return, and equilibrium: Empirical tests, Journal of Political Economy 81, Fama, Eugene F., and Kenneth R. French, 1992, The cross-section of expected stock returns, Journal of Finance 47, Fama, Eugene F., and Kenneth French, 1993, Common risk factors in the returns on stocks and bonds, Journal of Financial Economics 33, Fama, Eugene F. and French, Kenneth R., 1996, Multifactor explanations of asset pricing anomalies, Journal of Finance 51, Hanke, Michael and Florian Hauser, 2008, On the effects of stock spam s, Journal of Financial Markets 11, Harris, Jeffrey H., Venkatesh Panchapagesan, and Ingrid Werner, 2008, Off but not gone: A study of Nasdaq delistings, The Ohio State University Working Paper. Hasbrouck, Joel, 2009, Trading Costs and Returns for U.S. Equities: Estimating Effective Costs from Daily Data, Journal of Finance 64, Huang, Wei, Qianqiu Liu, S. Ghon Rhee, and Feng Wu, 2012, Extreme downside risk and expected stock returns. Journal of Banking and Finance 36, Huang, Wei, Qianqiu Liu, S. Ghon Rhee, and Liang Zhang, 2010, Return reversals, idiosyncratic risk and expected returns, Review of Financial Studies 23,

23 Jegadeesh, N., 1990, Evidence of predictable behavior of security returns, Journal of Finance 45, Jegadeesh, N. and Titman, S., 1993, Returns to buying winners and selling losers: Implications for stock market efficiency, Journal of Finance 48, Jegadeesh, N. and Titman, S., 2001, Profitability of momentum strategies: An evaluation of alternative explanations, Journal of Finance 56, Kumar, Alok, Page, Jeremy K., Spalt, Oliver G., Religious beliefs, gambling attitudes, and financial market outcomes. Journal of Financial Economics 102, Kumar, Alok, 2009, Who gambles in the stock market? Journal of Finance 64, Jiang, J., Petroni, K., Wang, I., 2012, Did Stop Signs Stop Investor Trading? Investor Attention and Liquidity in the Pink Sheets Tiers of the OTC Market, Michigan State University Working Paper. Keim, D.B., Madhavan, A., The cost of institutional equity trades. Financial Analyst Journal 54, Lakonishok, J., A. Shleifer, and R. W. Vishny, 1994, Contrarian investment, extrapolation, and risk, Journal of Finance 49, Leuz, C., Triantis, A., and Wang, T., 2008, Why do firms go dark? Causes and economic consequences of voluntary SEC deregistrations, Journal of Accounting and Economics 45, Liu, Qianqiu, S. Ghon Rhee, and Hong Vo, 2012, Institutional investors and short-term return reversals, University of Hawaii Working Paper. Macey, Jonathan, Maureen O Hara, David Pompilio, 2008, Down and out in the stock market: The law and economics of the delisting process, Journal of Law and Economics 51, Marosi, András and Nadia Massoud, 2004, Why do firms go dark? University of Alberta Working Paper. Newey, Whitney K., and Kenneth D. West, 1987, A simple positive-definite heteroskedasticity and autocorrelation consistent covariance matrix, Econometrica 55, Rhee, S. Ghon and Feng Wu, 2012, Anything wrong with breaking a buck? An empirical evaluation of NASDAQ s $1-minimum-price maintenance criterion, Journal of Financial Markets 15, Roll, Richard, 1984, A simple implicit measure of the effective bid-ask spread in an efficient market, Journal of Finance 39, Securities and Exchange Commission, 2012, Report to Congress on Decimalization: As Required by Section 106 of the Jumpstart Our Business Startups Act. 23

24 Stoll, H.R., Whaley, R., Transactions costs and the small firm effect. Journal of Financial Economics 12,

25 Table I Characteristics of Penny and Non-Penny Stocks This table reports the characteristics of penny stock portfolios and low-priced non-penny stock portfolios. We adopt two decision rules. First, at the beginning of each month during the study period, we look back the past one-year period to compute the average price for each stock. If the average price of a stock is below $5, then it is considered a penny stock. It is quite possible that its price moves up beyond $5 in the following months. The exclusion of this stock from the penny stock portfolio may understate the portfolio return. Hence, the second rule is introduced. Once a penny stock is identified, we give it a one-year grace period. During this grace period, it remains as a penny stock. To confirm the robustness of our results and to examine the effect of price level on stock performance, we construct three portfolios of penny stocks: (i) penny portfolio 1 (Price $1); penny portfolio 2 ($1 < Price $3); and penny portfolio 3 ($3 < Price $5). The one-year averaging rule and the one-year grace period rule remain applicable to the three portfolios based on sorts on price per share of component stocks to avoid the over- and under-estimation of portfolio returns. To make the comparison between penny and non-penny stocks more meaningful, we identify lowpriced non-penny stocks if their prices are between $5.00 and $ The one-year averaging rule and the one-year grace period rule are also applied to non-penny stocks. A total of 1,307 non-penny stocks is identified. To compare with three penny stock portfolios, we create three low-priced non-penny stock portfolios: (i) non-penny portfolio 1 ($5 Price $8); non-penny portfolio 2 ($8 Price $11); and nonpenny portfolio 3 ($11 Price < $15). The institutional ownership data are obtained from Thomson/CDA. The Dimson (1979) beat is estimated based on the equation R j,t = β1*rm t +β2*rm t-1 +β3*rm t-2, where the beta is the sum of coefficients of three (lagging) market returns (β = β1 + β2 + β3). Size is market capitalization in $million during the formation period. B/M is the book-to-market ratio. IV is the equally weighted idiosyncratic volatility of the portfolio in the formation period. The idiosyncratic volatility is estimated using regression residuals from the Fama and French (1993) three-factor model using daily return data over the previous month, following Ang et al. (2006). The Amihud (2002) liquidity measure is estimated by the square root of the average daily absolute return, r d, over the daily dollar volume, VOL d : 1000*[ r d /VOL d ] 1/2. The Gibbs effective transaction cost for each stock is constructed based upon daily data during the calendar year. For the estimation, we require that stocks have at least 50 trading days during that calendar year, following Hasbrouck (2009). The Corwin-Schultz (2012) spread estimates are computed using daily high and low prices. 25