SEARCH

SEARCH BY CITATION

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. I. Data and Descriptive Statistics
  4. II. Multivariate Tests
  5. III. Conclusion and Policy Implications
  6. REFERENCES

We investigate the relation between organization structure and the information content of short sales, focusing on founder- and heir-controlled firms. Our analysis indicates that family-controlled firms experience substantially higher abnormal short sales prior to negative earnings shocks than nonfamily firms. Supplementary testing indicates that family control characteristics intensify informed short selling. Further analysis suggests that daily short-sale interest in family firms contains useful information in forecasting stock returns; however, we find no discernable effect for nonfamily firms. This analysis provides compelling evidence that informed trading via short sales occurs more readily in family firms than in nonfamily firms.

Critics observe that short sales allow traders to bet on a company's demise.1 Academic research suggests that short-sale activity anticipates poor firm performance and often reflects interest by informed traders with negative, nonpublic information on the firm (Diamond and Verrecchia (1987), Asquith, Pathak, and Ritter (2005)). Building on this influential literature, we examine how organization structure can influence the potential for informed trading. Our analysis focuses on short-sale activity in a specific organizational structure, founding family-controlled firms. Pérez-González (2006) notes the pervasiveness and prominence of family ownership in publicly traded U.S. firms while Shleifer and Vishny (1986) report that families control about one-third of Fortune 500 firms. Prior literature generally indicates that these controlling families are well-informed shareholders. Anderson and Reeb (2003) and James (2006), for instance, argue that the family's long-standing knowledge and information base can provide benefits to outside shareholders by facilitating firm monitoring and corporate interactions with capital markets. Consistent with the notion that family owners possess superior information, Demsetz (1986) and Chan, Chen, and Hilary (2010) report that family CEOs earn greater profits on their stock trades than CEOs of nonfamily firms.

Founders and heirs (family owners) arguably possess strong incentives to engage in short-selling activity.2 As investors with access to privileged information, these controlling shareholders may seek to earn profits in light of adverse information (Morck, Wolfenzon, and Yeung (2005)). Schulze, Lubatkin, and Dino (2003) observe that conflicts of interest among a myriad of family members with a claim on firm cash flows can lead family members not employed by the firm to take destructive or harmful actions. Family members not actively engaged in firm management or board decisions—unlike professional executives or current directors—potentially bear less scrutiny from regulators, thus minimizing concerns about trading on privileged information. Beyond families’ private motives to pursue short sales, family members could be a source of information leakage that provides outside investors with incentives to engage in short selling.3 Employees who are not members of the family group may also be disgruntled with family interference or family domination of senior managerial posts, leading to a leakage of negative information on firm activities. As a consequence, family firms may have a variety of linkages that could facilitate the leakage of material, nonpublic information regarding corporate activities.

On the other hand, family shareholders potentially seek to limit informed trading in their firm's shares. Overstock.com's founder (Patrick Byrne), for instance, attempted to use the courts to limit short sales in his firm's shares, arguing that such activities by hedge funds were increasing the firm's cost of capital. Consistent with this notion, Massoud et al. (2011) present evidence to suggest that hedge funds engage in short sales using material, nonpublic information. Other family owners take a more conventional approach to limit informed short sales by asking shareholders to move their shares from margin accounts into personal accounts or withholding their own substantial stakes from circulation, thereby increasing borrowing costs4,5). Families’ private or personal concerns over preserving reputation, limiting public visibility, and safeguarding wealth potentially deter family members from engaging in informed short sales. Overall, persuasive arguments exist for family shareholders to facilitate information leakage as well as exploit their private information advantages, or alternatively to limit active traders or other corporate insiders from trading on adverse nonpublic information. The question of whether family presence hinders or facilitates informed trading is thus an empirical issue.

We explore the relation between founding family control and informed trading using aggregate daily short-sale data for publicly traded U.S. firms. Our analysis focuses on short sales because—even with extensive regulations of corporate officers and strictures on fair, market-wide information disclosure—this activity appears to be influenced by nonpublic information. Christophe, Ferri, and Angel (2004) suggest that informed trading activity can be evaluated or captured by examining short sales prior to negative corporate events. Desai et al. (2002) and Diether, Lee, and Werner (2009) observe that short sales forecast negative abnormal stock returns and thus are indicative of informed trading. As a result, the short-sale market offers a robust environment for evaluating the influence of organizational structure on informed trading in large, publicly traded firms.

Focusing on whether firm ownership structure is associated with the presence of informed traders, we investigate two specific aspects of short sales for family and nonfamily firms. First, we examine short sales prior to, or in advance of, firms experiencing either negative or positive earnings surprises. Second, we examine whether short sales provide useful or valuable information in forecasting future stock returns. Our tests use the firm rather than the individual investor as the unit of analysis. As a consequence, we cannot unambiguously define the mechanism or the motivation of short sellers. Yet, by investigating these two aspects of short sales, we examine whether the potential for informed trading differs between family and nonfamily firms.

To examine the association between ownership structure and the informational content of short sales, we gather family ownership information on the 2,000 largest U.S. industrial firms (based on market capitalization) as of January 2004. We subsequently merge this information with the Securities and Exchange Commission's (SEC) REG SHO database. This database provides firm-level short-sale data on an intraday basis from January 2005 to July 2007 but, as a caveat to our study, does not provide the identity of the short seller. The merging of the 2,000 largest firms with the SEC REG SHO database provides a final sample of 1,571 firms spanning from January 2005 to July 2007. The sample captures 91.6% of the market capitalization of firms covered by Compustat. Our testing procedures employ quarterly and daily time periods to examine the relation between short sales and family ownership. In the quarterly tests, we examine abnormal short sales prior to negative quarterly earnings surprises and prior to positive quarterly earnings surprises. The 1,571 firms in our sample provide 4,702 negative quarterly earnings surprises and 5,491 positive quarterly earnings surprises from January 2005 to July 2007. The daily tests examine whether short-sale interest on day t predicts future stock returns on day t + 2. The 1,571 firms in our sample provide 310,720 firm-day observations for family firms and 523,264 firm-day observations for nonfamily firms from January 2005 to July 2007.

Based on a minimum 5% ownership threshold (Villalonga and Amit (2006)), family firms constitute 36% of the sample with an average ownership stake of 25.3%. The remaining 64% of firms are classified as nonfamily firms. Because family ownership and control characteristics potentially affect the firm's information environment, we gather data on CEO type (founder, descendant, and professional manager), family control of board seats, family ownership level, and dual-class share structure. We also consider the influence of other large blockholders (hedge funds, private equity funds, mutual funds, pension funds, and insurance companies) on the relation between family firms and short-sale activity.

Our analysis indicates that family firms experience substantially greater abnormal short sales prior to negative earnings surprises than nonfamily firms. Abnormal short sales are defined as [(average daily short sales prior to quarterly earnings announcements (day −30 to day −1) divided by average daily short sales for the year outside of preannouncement periods) −1]. The results indicate that short sales in family firms are over six times more sensitive to the magnitude of future negative earnings shocks than short sales in nonfamily firms. In total, family firms experience almost 17 times more abnormal short selling preceding negative earnings shocks than nonfamily firms, suggesting extensive informed trading. We include several different proxies for informed trading (bid-ask spread, Kyle's λ, and the probability of informed trading or PIN) and control for the presence of other large blockholders as well as the quality of corporate governance and continue to find similar results.6 The analysis also indicates that, prior to positive earnings surprises, family firms sustain marginally less abnormal short selling than their nonfamily counterparts. The results on negative and positive earnings shocks appear consistent with the notion that informed trading, at least in short sales, occurs more readily in family firms than in nonfamily firms.

Our investigation also indicates that family control characteristics bear a relation to abnormal short sales in advance of negative earnings surprises. Active family management by either the founders, or their descendants, greater family board involvement, lower ownership levels of cash-flow rights, and dual-class share structures appear to intensify abnormal short sales relative to other categorizations of family firms or nonfamily firms. One potential interpretation of these results is that greater family influence in the firm intensifies informed trading. Still, all categorizations of family firms, on average, experience greater abnormal short sales preceding a negative earnings shock compared to nonfamily firms.

Family owners are not the only type of large, influential shareholder. Other large blockholders arguably possess similar profit incentives as family shareholders and often have access to private information, suggesting that our results may be driven by blockholder organizational structure rather than the presence of a particular type of large blockholder, namely, family shareholders. To examine this conjecture, we contrast family ownership with that of hedge funds, private equity funds, mutual funds, pension funds, and insurance companies. Our analysis indicates that, except for hedge funds, these other classes of blockholders bear no significant relation to abnormal short sales in advance of negative earnings surprises. Notably, the inclusion of the other types of large blockholders in our analysis does not change the relation between family presence and abnormal short selling prior to negative earnings surprises. Firms with family owners experience significantly greater abnormal short sales in advance of negative earnings shocks than nonfamily firms. Interestingly, SEC enforcement actions over the past several years have extensively targeted hedge funds while few actions have affected family firms. For instance, from 2006 to 2008, 22 enforcement actions were brought against hedge funds and none against family owners.7

Our second set of tests examines whether short sales better predict short-term stock returns for family firms than nonfamily firms. In our tests, we calculate abnormal stock returns using Fama–French three- and four-factor models for value-weighted portfolios (based on short-sale rankings) for family and nonfamily firms. We rank the portfolios based on daily short-sale interest, which is measured as daily short-sale volume divided by daily stock trading volume. For family firms, we find that future abnormal returns are decreasing in short-sale activity. An investment strategy that buys the portfolio of family stocks with the lowest level of short sales and simultaneously shorts the portfolio with the highest level of short sales (long–short strategy) earns abnormal returns of 60.9 basis points per month (excluding portfolio rebalancing costs). In nonfamily firms, however, no clear relation exists between short-sale activity and future abnormal stock returns in either the three- or four-factor model specifications. The stark difference in abnormal returns between family firms and nonfamily firms appears to be large and economically significant.

Using a Fama–MacBeth methodology to corroborate the relation between daily short-sale interest and future short-term stock returns, we again note that daily short-sale interest provides useful or valuable information in predicting future returns for family firms. Short sales appear, however, to have little predictive ability in nonfamily firms. Additional analysis further suggests that family firm characteristics influence the relation between daily short-sale interest and future short-term returns. In particular, we note that active family management (founder CEO or descendant CEO), greater family board presence, low ownership levels of cash-flow rights, and dual-class share structures strengthen the relation between daily short-sale interest and future abnormal returns. Overall, the results indicate that daily short-sale interest provides useful information in forecasting abnormal returns for family firms but supplies little information in predicting returns for nonfamily firms. The evidence from our stock return predictability tests confirms that, of the earnings announcement tests, informed trading—at least in short sales—occurs more readily in family firms compared to nonfamily firms.

This study makes three important contributions to the literature. First, the analysis provides compelling evidence that informed trading accounts for a portion of short selling in family firms. The analysis suggests a potential conflict among investors that receives little attention in the literature or from regulators. Although our data do not allow for identification of the short seller, the results suggest the potential for well-informed traders to use negative information events to earn abnormal profits at the expense of atomistic shareholders. The results indicate that family firms, both founder or descendent controlled, sustain systematic and reoccurring informed trading via the short-sale market.

Second, the informed trading literature generally uses traded volume (Kyle's λ), bid-ask spreads (adverse information component), and trade classification data (PIN) to detect the presence of informed trading. In our approach, we focus on the firm's organizational structure to detect the presence of informed traders. Our analysis suggests that family control represents an important marker in understanding the entrance of private information, at least in short sales, into the stock market. In aggregate, the study suggests that short-sale activities and their effect on price discovery vary based on the firm's organizational structure.

Finally, we add to the literature on the regulation of trading by investors with access to material, nonpublic information. Even though informed trading can facilitate corporate transparency, U.S. regulators seek to restrict individuals from trading on material, nonpublic information. In 1984, the U.S. Supreme Court ruled that trading based on information received through breaches of fiduciary responsibilities, such as managers providing updates to family shareholders so that they can trade on the information, represents illegal insider activity (Seyhun (1992)). Market experts have questioned the general efficacy of insider trading regulations in limiting such activity by informed traders (Banerjee and Eckard (2001)). Our analysis suggests that regulations to reduce informed trading, while potentially limiting such activity in nonfamily firms, appear substantially less effective in family firms.

The remainder of this paper proceeds as follows. Section I provides a summary of the data and descriptive statistics. Section II presents the empirical results. Finally, in Section III, we conclude and discuss regulatory implications.

I. Data and Descriptive Statistics

  1. Top of page
  2. ABSTRACT
  3. I. Data and Descriptive Statistics
  4. II. Multivariate Tests
  5. III. Conclusion and Policy Implications
  6. REFERENCES

A. Sample

For our empirical investigation, we begin with the 2,000 largest nonutility, nonfinancial firms in the United States as of January 31, 2004. To gather the 2,000 largest firms, we extract all firms from Compustat and rank these based on market capitalization as of January 31, 2004. We exclude foreign firms, regulated public utilities (SIC codes 4812, 4813, 4911 through 4991), and financial firms (SIC codes 6020 through 6799) because government regulation potentially affects firm equity ownership structure.

We next merge the 2,000 largest U.S. firms with short-sale data based on SEC REG SHO from NYSE (including Archipelago, which merged with NYSE in early 2006), NASDAQ, AMEX, NSX, and FINRA. The SEC initiated a pilot program that requires self-regulated organizations, such as exchanges, to report any short-sales trades and make the information publicly available (see http://www.sec.gov/spotlight/shopilot.htm for details). Short-sale data are available from January 2005 through July 2007. Trades are also reported to FINRA because some trades are executed through nonexchange channels such as telephone. The short-sale data are of intraday nature and contain the trading symbol of the stock, the price, the short-sale size, the date, and the time of each trade. In addition, the data indicate whether trades are short-exempt. The data do not identify the individual or institution undertaking the short sale. We control for survivorship bias by allowing firms to exit our sample. After merging the 2,000 largest nonutility, nonfinancial firms with the SEC REG SHO database, our sample consists of 1,571 firms covering the period from January 2005 through July 2007. Of the 429 firms not included in the final sample, we find a similar percent of family firms (38.2%) and, on average, the 429 firms are smaller than the firms in the final sample.

The data cover only part of 2007; however, our principal tests use quarterly and daily data, allowing us to include the entire period. For the first set of tests, we examine short sales prior to quarterly earnings surprises. The 1,571 firms in our sample provide 4,702 negative quarterly earnings surprises and 5,491 positive quarterly earnings surprises from January 2005 to July 2007. For the quarterly earnings announcement tests, the control variables are measured at quarter-end for each firm and family ownership and control characteristics are set equal to their year-end values.

In our second set of tests, we examine whether daily short-sales interest on day t predicts future abnormal stock returns on day t + 2 for family and nonfamily firms. The 1,571 firms in our sample provide 310,720 firm-day observations for family firms and 523,264 firm-day observations for nonfamily firms from January 2005 to July 2007. We extract daily data on stock returns, prices, shares outstanding, and trading volume from CRSP. The control variables for the daily tests are computed using daily data and family ownership and control characteristics are set equal to their year-end values.

B. Primary Variable Measurement

For the quarterly earnings surprise tests, our primary dependent variable is abnormal short sales computed as [(average daily short sales prior to quarterly earnings announcements (day −30 to day −1) divided by average daily short sales for the year outside of preannouncement periods) −1]. In the daily test, we examine the relation between short sales and future stock returns. For the daily short-sale measures (daily short-sale interest), we use daily short-sale volume divided by daily share volume.

We define family firms as those where the family (founders or founders’ descendants) continues to maintain a 5% or greater ownership stake (Shleifer and Vishny (1986), Villalonga and Amit (2006)). To be classified as a family firm, a family member does not necessarily need to serve as the firm's CEO; rather, the classification refers to families maintaining a minimum 5% equity stake in the firm. The initial analyses use a binary variable that equals one when families hold a 5% or larger ownership stake and zero otherwise.

Family shareholders can gain an information advantage over other shareholders by maintaining an active role in firm management or serving on the board of directors (Anderson and Reeb (2004)). We subdivide family firms into those managed by founder CEOs, descendant CEOs, or professional CEOs. We measure family board presence as the number of board seats held by family members. Family ownership levels are computed based on cash-flow rights as the total number of shares held by families (over all classes of stock) divided by total shares outstanding. About 20% of the family firms in our sample employ dual-class share structures. To capture the effect of dual-class structures, we use a binary variable that equals one when the firm has two or more classes of stock and zero otherwise. To ascertain family ownership and its involvement, we examine corporate proxy statements and company histories for each firm in our sample to determine the founders, their subsequent lineage, and their involvement with the firm. Corporate histories for each firm in our sample come from FundingUniverse.com, ReferenceforBusiness.com, Gale Business Resources, and Hoovers, as well as from individual companies.

In our analysis, we compare family firms against firms with disparate ownership structures that are controlled by professional managers (the benchmark group, which we label as nonfamily firms). However, blockholders such as mutual funds, hedge funds, and other investment institutions often maintain stakes in many family and nonfamily firms (Tufano (1996)). We obtain blockholding information from Thomson ONE Banker for each sample firm. In the empirical analysis, we group blockholders into one of five categories based on Thomson ONE Banker's Investor Sub-Type. These five groups are hedge funds, private equity funds, mutual funds, pension funds, and insurance companies. We develop a binary variable that equals one if the blockholder holds a 5% or larger ownership stake in the firm and zero otherwise. In subsequent testing, we partition nonfamily firms into two subsamples based on ownership characteristics, namely, those with blockholders and those without blockholders.

Unexpected quarterly earnings for each firm are measured as the residual term from the following regression:

  • image(1)

where EPS is actual earnings per share of the announcement quarter (q), the prior quarter (q− 1), one year ago (q− 4), and two years ago (q− 8). The average unexpected earnings surprise for the sample is −$0.007 per share. In robustness testing, we also use analysts’ forecast errors, measured as the difference between analysts’ mean forecast and the firm's announced quarterly earnings. We drop observations with fewer than three analysts following the firm. Overall, we find similar results between short selling, earnings surprises, and family and nonfamily firms as those reported in our primary test Analyst forecast errors, at least at first blush, appear to be a better measure of earnings surprises as these incorporate current information into analysts’ estimates. However, using analysts’ forecast errors as a measure of earnings surprises requires assuming that analysts are randomly distributed across both family firms and nonfamily firms. Anderson, Duru, and Reeb (2009) note systematic differences between analyst following for family firms (about 5.3 analysts per firm) and nonfamily firms (about 7.1 analysts per firm).

C. Control Variable Measurement

Previous literature indicates that short-selling activity varies with firm characteristics. Diether et al. (2009) note that short selling tends to be higher for large-cap stocks, stocks with low book-to-market ratios, and stocks with high institutional ownership. Firm size is measured as the natural log of quarter-end total assets. We control for growth opportunities using the book-to-market ratio measured as the book value of common equity divided by the market value of common equity. Less liquid stocks potentially expose short sellers to substantial costs in the event these traders need to quickly cover their positions due to short squeezes, margin calls, etc. (Shleifer and Vishny (1997)). We capture stock liquidity using trading volume measured as daily trading volume averaged across all trading days in the quarter. To ensure that we are not simply capturing known aspects of informed trading, we control for differing levels of private information distribution among outside investors. Our tests use the stock's bid-ask spread to control for outside investor knowledge, calculated as the average of the daily bid-ask spread (over the quarter).8

Investors can also make negative bets on stocks by buying put options, thereby potentially drawing activity away from the shorts market. We control for put option activity as the quarterly average of daily put option volume divided by daily share volume. Firm performance is measured as return on assets from the prior quarter (t− 1), computed as income before extraordinary items divided by total assets from the prior quarter (t− 1). Firms with greater stock price volatility likely attract greater attention from short sellers. We capture stock price volatility as the standard deviation of daily stock returns for the quarter. Diether et al. (2009) note that short selling differs substantially between NYSE and NASDAQ stocks. We include a dummy variable that equals one when a stock is listed on NYSE and zero otherwise. We account for the quality of corporate governance with the Corporate Governance Quotient (CGQ) developed by Institutional Shareholder Services (ISS), as better governed firms may be less involved in informed trading.9 Stocks with greater uncertainty in future earnings arguably attract greater short sales than stocks with less uncertainty in earnings. To control for future earnings risk, we include analyst forecast dispersion, measured as the standard deviation of forecasted EPS scaled by the prior quarter-end stock price. Finally, the analysis includes dummy variables for each Fama–French industry to account for industry effects as well as quarter dummy variables to capture time effects.10Table I provides a description of the variables used in our analysis.

Table I.  Variable Definitions
Abnormal Short Sales: [(Average daily short sales prior to quarterly earnings announcements (day −30 to day −1) divided by average daily short sales for the year outside of preannouncement periods) − 1].
Unexpected Earnings: The residual term from the following regression:
EPSi,q=α+β1EPSi,q-12EPSi,q-43EPSi,q-8i,t,
where EPS is actual earnings per share of the announcement quarter (q), the prior quarter (q− 1), one year ago (q− 4), and two years ago (q− 8).
Family Firm: Binary variable that equals one when the family holds a 5% or larger ownership stake and zero otherwise.
Family Ownership: The percent of common equity held by the family.
Founder CEO, Descendant CEO, Professional CEO: Equals one when a founder, descendant, or professional outside CEO, respectively, holds the CEO position in a family firm and zero otherwise.
Dual Class: The percent of family firms that maintain a dual-class share structure.
Family Board Seats: The number of board seats held by family members.
Total Assets: Quarter-end total assets measured in millions of dollars.
Firm Size: Natural log of quarter-end total assets.
Leverage: Quarter-end long-term debt divided by quarter-end total assets.
Book-to-Market: Quarter-end book value of equity divided by the previous quarter-end market value of equity.
Forecast Dispersion: The standard deviation of analysts’ forecasts divided by the previous quarter-end stock price.
Bid-Ask Spread: Daily bid price less daily ask price, divided by the average of the bid price plus the ask price, averaged across each quarter.
Stock Return Volatility: Standard deviation of daily stock returns for each quarter.
Blockholders: Equals one when either a hedge fund, private equity fund, mutual fund, pension fund, or insurance company holds a 5% or larger ownership stake in the firm and zero otherwise.
Hedge Fund/Private Equity: Equals one when a hedge fund or private equity fund holds a 5% or larger ownership stake in the firm and zero otherwise.
Mutual/Pension Fund & Insurance Co.: Equals one when a mutual fund, pension fund, or insurance company holds a 5% or larger ownership stake in the firm and zero otherwise.
NYSE: Equals one when the firm is listed on NYSE and zero otherwise.
Trading Volume: Natural log of daily trading volume averaged across each quarter.
Put Option Volume: Daily put option volume divided by daily stock trading volume, averaged across each quarter.
Institutional Ownership: Fraction of common equity held by institutional investors.
Performancet-1: Prior quarter-end income before extraordinary items divided by prior quarter-end total assets.
Corporate Governance Quotient (CGQ): Institutional Shareholder Services’ CGQ for each firm at year-end.
Industry Dummy: Equals one for each Fama–French industry group and zero otherwise.
Quarter Dummy: Equals one for each quarter and zero otherwise.
ri,t+2: Total stock return in percent for firm i on day t + 2.
Shorti,t: Short-sales volume on day t divided by total stock trading volume on day t.
r(-5, -1)i,t: Total stock return in percent for firm i from day t− 5 to day t− 1.
Rank(r-5, -1)i,t: Rank of an individual stock's return based on (r-5, -1)i,t among all stocks in the sample, normalized to a range from 0.0 to 1.0.
Riski,t: Difference between the high and low stock price of stock i on day t divided by the high price of stock i on day t.
Turnover(-5, -1)i,t: Average share trading volume divided by shares outstanding (multiplied by 1,000) for stock i from day t− 5 to day t− 1.

D. Descriptive Statistics

Table II, Panel A, provides quarterly summary statistics for the 1,571 sample firms from January 2005 through July 2007. The statistics are calculated by aggregating firm characteristics for 4,702 quarters of negative earnings surprises and 5,491 quarters of positive earnings surprises.11 An Internet Appendix available at http://www.afajof.org/supplements.asp provides summary statistics for firm characteristics separately for the 4,702 negative earnings surprise quarters and 5,491 positive earnings surprise quarters.

Table II.  Descriptive Statistics Panel A: This panel provides quarterly summary statistics and difference of mean tests between family and nonfamily firms for the 1,571 sample firms from January 2005 through July 2007. The statistics are calculated by aggregating 4,702 negative earnings surprise quarters and 5,491 positive earnings surprise quarters (4,702 + 5,491 = 10,193). The Internet Appendix provides summary statistics, disaggregated, for the 4,702 negative earnings surprise quarters and the 5,491 positive earnings surprise quarters. Family firms comprise 3,670 firm quarters and nonfamily firms comprise 6,523 firm quarters. All data are measured at quarter-end for each firm except for family firm, family ownership, founder CEO, descendent CEO, professional CEO, dual class, family board seats, institutional ownership, and the Corporate Governance Quotient, which are annual measures. For data measured on an annual basis, we set their quarter-end measures equal to the corresponding year-end values. Panel B: This panel provides daily summary statistics and difference of mean tests between family and nonfamily firms for the 1,571 sample firms from January 2005 through July 2007, providing 310,720 firm-day observations for family firms and 523,264 firm-day observations for nonfamily firms. The statistics are calculated by aggregating family and nonfamily firms (310,720 + 523,264) and then averaging firm-day measures. The summary statistics in this panel apply to Fama–French portfolio tests (Tables IV and V) and the Fama–MacBeth regressions (Table VI).
Panel A: Summary Statistics of Firm Characteristics
 MeanMedianStd. Dev.Min.Max.Family (n= 3,670)Nonfamily (n= 6,523) t-test p-value
Abnormal short sales (%)−4.51−3.4027.13−80.8182.59−4.37−4.520.480
 Prior to positive unexpected earnings−4.40−3.1727.18−80.8176.20−6.03−2.220.124
 Prior to negative unexpected earnings−4.60−3.7027.10−65.6982.592.31−5.880.001
Unexpected earnings−0.0070.000.20−0.9810.661−0.004−0.0090.211
 Positive unexpected earnings0.1280.0730.1640.0010.6610.1270.1290.555
 Negative unexpected earnings−0.132−0.0650.201−0.981−0.001−0.136−0.1290.266
Family firm (%)36.290.0044.020.001.00
 Family ownership (%) (family firm only)25.3419.5718.295.7093.50
 Founder CEO (%) (family firm only)38.980.0048.780.001.00
 Descendant CEO (%) (family firm only)20.100.0040.080.001.00
 Professional CEO (%) (family firm only)40.920.0049.180.001.00
 Dual class (%) (family firm only)20.100.0040.080.001.00
 Family board seats (family firm only)1.732.001.000.004.00
Total assets ($ billions)7.331.8129.700.02795.343.629.460.000
Firm size7.697.501.462.9713.597.177.800.000
Leverage0.190.160.170.000.790.180.190.000
Book-to-market0.940.760.680.044.090.970.920.012
Forecast dispersion0.150.070.190.010.640.170.140.000
Bid-ask spread0.040.020.020.010.170.040.050.000
Stock return volatility0.180.170.080.050.770.190.180.000
Blockholders (%)83.47100.0037.140.00100.0086.1452.180.000
 Hedge fund/private equity (%)50.71100.0049.990.00100.0047.0257.220.000
 Mutual/pension fund & insurance Co. (%)76.36100.0042.480.00100.0078.0575.410.005
NYSE (%)77.03100.0042.030.00100.0072.6678.520.000
Trading volume13.3913.301.359.8816.8412.9313.660.000
Put option volume (%)5.782.052.120.449.775.326.050.000
Institutional ownership0.750.820.250.001.000.710.780.000
Performancet-10.060.060.16−0.510.350.060.060.082
Corporate governance quotient53.8555.7028.311.8099.2045.8758.390.000
Panel B: Summary Statistics for Fama–MacBeth Test
      FamilyNonfamily t-test
 MeanMedianStd. Dev.Min.Max.(n= 310,720)(n= 523,264) p-value
ri,t+2−0.015−0.0191.736−4.5144.569−0.014−0.0160.057
Shorti,t0.2290.2000.2150.0000.5300.2320.2270.007
r(−5,−1)i,t−0.0010.0010.085−2.3191.358−0.002−0.0010.087
Rank(r−5, −1)i,t0.5000.4990.2840.0041.0000.5010.4980.002
Riski,t0.0280.0250.0150.0070.0820.0280.0280.024
Turnover(−5, −1)i,t10.0317.8547.5121.16347.9869.71210.2360.000

Panel A indicates that our measure of short sales for the quarterly earnings surprise tests—abnormal short sales—has a mean of −4.51% with a standard deviation of 27.13%. Intuitively, this indicates that average daily short-sale volume is 4.51% less during the preannouncement period relative to daily short-sale volume outside of preannouncement periods. Family firms represent 36.3% of the total sample based on a minimum 5% ownership threshold (Villalonga and Amit (2006)). Families, on average, hold 25.3% of the firm's shares with a median stake of 19.57%. Founders serve as CEO in 38.9% of the family firms while descendants and outside professional managers hold the post for 20.1% and 40.9% of the family firms, respectively. Seventy-seven percent of the sample firms are listed on NYSE with the remaining 23% listed on NASDAQ or AMEX.

The last three columns in Panel A of Table II show the results of difference of mean tests between family and nonfamily firms using the total number of firm-quarters in our sample (4,702 negative quarters + 5,491 positive quarters = 10,193). Family firms constitute 3,670 quarterly observations with the remaining 6,523 observations characterized as nonfamily firms. The results of the difference of mean tests indicate that family firms are smaller, are less debt intensive, and have greater stock return volatility than nonfamily firms. Furthermore, we note that family firms tend to have greater forecast dispersion among stock analysts and lower bid-ask spreads than nonfamily firms. The difference of mean tests also indicate that family firms experience significantly greater abnormal short selling prior to negative unexpected earnings surprises than nonfamily firms. Notably, we observe that unexpected earnings—either positive or negative—do not differ between family and nonfamily firms (see footnote 3).

Table II, Panel B, provides daily summary statistics for the 1,571 sample firms from January 2005 through July 2007, providing 310,720 firm-day observations for family firms and 523,264 firm-day observations for nonfamily firms. We use daily stock return data to examine whether short-sale interest on day t predicts future stock returns on day t + 2. On average, daily short-sale interest for our sample firms is 0.229, indicating that short sales account for 22.9% of daily share volume. Difference of mean tests in the last three columns of Panel B, Table II, indicates that family firms experience significantly greater daily short-sale interest than nonfamily firms (0.232 vs. 0.227), suggesting that, on average, investors engage in more short sales for family firms than nonfamily firms. Average stock returns 2 days after our short-sale measure exhibit mean and median values of −0.015% and −0.019%, respectively.

II. Multivariate Tests

  1. Top of page
  2. ABSTRACT
  3. I. Data and Descriptive Statistics
  4. II. Multivariate Tests
  5. III. Conclusion and Policy Implications
  6. REFERENCES

A. Unexpected Earnings and Short Sales

Our central argument focuses on the information content of short sales in family and nonfamily firms. If family firms experience more informed trading than nonfamily firms, then we expect to observe greater short selling in advance of negative earnings surprises compared to nonfamily firms. In contrast, if informed trading does not differ between family and nonfamily firms, we do not expect to observe a difference in short sales prior to negative earnings shocks. We use the following specification to examine the informed trading argument:

  • image(2)

The variables are defined in Table I with X representing a vector of control variables. This analysis uses a binary variable to capture family presence. In Table III, Panel A, Columns 1 and 2 display the results for negative earnings surprises, Columns 3 and 4 display the results for positive earnings surprises, and Column 5 shows the results when combining negative and positive surprises into the same specification. We control for serial correlation and heteroskedasticity using the Huber–White sandwich estimator (clustered on firm-level identifier) for the standard errors on the coefficient estimates. For ease of interpretation, we use the absolute value of negative earnings surprises in the regression specification.

Table III.  Unexpected Earnings, Organizational Structure, and Abnormal Short Sales This table reports OLS results of regressing abnormal short sales on unexpected quarterly earnings, family presence, and the interaction of unexpected quarterly earnings and family presence for 4,702 negative quarterly earnings shocks and 5,491 positive quarterly earnings shocks. Panel A reports the results when using a binary family firm indicator variable. Panel B reports the results when examining family ownership characteristics (CEO type, board seats, cash-flow rights, and share classes). Panel C reports the results using a binary family firm variable and interacting blockholders with unexpected earnings.
Panel A: Unexpected Quarterly Earnings, Family Presence, and Short Sales
Dependent Variable: Abnormal Short Sales
 Negative ShocksPositive ShocksAll Shocks
Intercept−0.080−0.1070.1120.1220.124
 (−0.92)(−0.93)(1.10)(1.00)(1.61)
Family firm0.026***0.019**−0.025*−0.021*0.016*
 (2.69)(2.45)(−1.90)(−1.83)(1.90)
Negative unexpected earnings0.020*0.019*0.021*
 (1.78)(1.69)  (1.84)
Family firm * negative unexpected earnings0.109**0.095**
  (2.39)  (1.99)
Positive unexpected earnings−0.014−0.013−0.011
   (−1.60)(−1.52)(−0.84)
Family firm * positive unexpected earnings−0.008*−0.017
    (−1.91)(−0.93)
Hedge fund/private equity0.0140.013−0.002−0.0020.011
 (1.37)(1.51)(−0.50)(−0.46)(1.09)
Mutual/pension fund & insurance co.−0.001−0.002−0.009−0.009−0.004
 (−0.12)(−0.31)(−0.88)(−0.87)(−0.51)
Firm size−0.013*−0.012*0.0040.003−0.011**
 (−1.77)(−1.74)(0.66)(0.53)(−2.15)
Performancet-10.002*0.003*0.0010.0010.005
 (1.82)(1.80)(1.42)(1.50)(1.13)
Bid-ask spread−4.015−4.627−4.009−4.111−5.662
 (−0.88)(−1.02)(−0.53)(−0.50)(−1.52)
Forecast dispersion0.0200.021−0.029**−0.030**0.014
 (1.16)(1.23)(−2.28)(−2.22)(1.28)
Book-to-market0.0050.0060.0090.0090.009
 (0.89)(0.82)(1.20)(1.40)(0.86)
Stock return volatility1.0110.9412.317**2.330**1.095
 (1.22)(1.35)(2.44)(2.51)(1.50)
NYSE−0.126***−0.130***−0.111***−0.115***−0.116***
 (−9.46)(−9.50)(−8.30)(−8.32)(−8.50)
Trading volume0.0100.009−0.007−0.007−0.004
 (1.50)(1.42)(−0.99)(−0.90)(−0.79)
Put option volume−2.886−2.6771.9941.952−3.943
 (−1.32)(−1.33)(0.77)(0.73)(−0.84)
Corporate governance quotient−0.000−0.0000.0000.000−0.001
 (−0.33)(−0.26)(0.55)(0.72)(−0.40)
Industry dummy & quarter dummyYesYesYesYesYes
Observations4,7024,7025,4915,49110,193
Adjusted-R20.320.320.240.240.30
Panel B: Unexpected Quarterly Earnings, Family Characteristics, and Short Sales
Dependent Variable: Abnormal Short Sales
 Negative Shocks Only
Intercept−0.081−0.092−0.088−0.099
 (−1.09)(−1.03)(−0.97)(−1.05)
Unexpected earnings0.012*0.0110.012*0.012*
 (1.71)(1.57)(1.80)(1.70)
Family firm0.009**0.013*0.015**0.016**
 (2.35)(1.90)(2.21)(2.30)
Family firm * unexpected earnings0.052**0.067**0.070**0.084**
 (2.23)(2.16)(2.17)(2.29)
Founder CEO0.010*
 (1.91)   
Founder CEO * unexpected earnings0.049*
 (1.88)   
Descendant CEO0.031*
 (1.88)   
Descendant CEO * unexpected earnings0.114**
 (2.09)   
Family holds ≥2 board seats0.022**
  (2.21)  
Family holds ≥2 board seats * unexpected earnings0.074**
  (2.30)  
Family cash-flow rights <20%0.017**
   (2.52) 
Family cash-flow rights <20%* unexpected earnings0.061**
   (2.19) 
(Family votes >family cash flow)0.051**
    (2.35)
(Family votes >family cash flow) * unexpected earnings0.222**
    (2.37)
Same controls as in Panel AYesYesYesYes
Observations4,7024,7024,7024,702
Adjusted-R20.320.320.320.32
Panel C: Unexpected Quarterly Earnings, Family Presence, Blockholders, and Short Sales
Dependent Variable: Abnormal Short Sales
 Negative Shocks Only
 Family Firms versus Diffusely Held Nonfamily FirmsFamily Firms versus Block-Held Nonfamily FirmsFamily Firms versus All Nonfamily FirmsFamily Firms versus All Nonfamily Firms
  1. The t-values are reported in parentheses and are corrected for serial correlation and heteroskedasticity. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

Intercept−0.133*−0.118−0.098−0.100
 (−1.87)(−1.30)(−1.21)(−1.19)
Unexpected earnings0.011*0.0100.0080.008
 (1.92)(1.55)(1.50)(1.53)
Family firm0.023**0.021*0.022**0.020**
 (2.32)(1.94)(2.19)(1.99)
Family firm * unexpected earnings0.128***0.100**0.095***0.090**
 (2.70)(2.48)(2.62)(2.55)
Other blockholder0.012
   (1.33) 
Other blockholder * unexpected earnings0.015
   (0.77) 
Hedge fund/private equity0.013
    (1.22)
Hedge fund/private equity * unexpected earnings0.010
    (0.69)
Mutual/pension/insurance−0.001
    (−0.88)
Mutual/pension/insurance * unexpected earnings0.012
    (0.90)
Same controls as in Panel AYesYesYesYes
Observations2,0914,4514,7024,702
Adjusted-R20.360.300.320.32

The results indicate that, as the magnitude of unexpected negative earnings shocks increases, all firms—family and nonfamily—experience increased abnormal short sales prior to the shock. The stand-alone negative unexpected earnings terms in Columns 1 and 2 bear positive and significant coefficient estimates, suggesting that family and nonfamily short sales increase as the size of the negative earnings shock increases. Short sales appear to contain information useful in forecasting negative earnings surprises for all firms.

The stand-alone family firm terms (Columns 1 and 2) capture whether family firm short sales increase prior to negative earnings shocks regardless of the size of the shock. The positive and significant coefficient estimates on the stand-alone family terms indicate that family firms sustain greater short sales—regardless of the size of the negative earnings surprise—than nonfamily firms. From a practical perspective, family firm daily abnormal short sales increase by 1.9% preceding any negative earnings surprise. We calculate this percent as the coefficient estimate on the stand-alone family firm term multiplied by the family firm binary variable (0.019 × 1.0 × 100%= 1.9%).

Short sales in family firms also appear to be much more sensitive to the size of negative earnings shocks than short sales in nonfamily firms. The interaction term between family firm and unexpected earnings (Column 2) denotes the additional abnormal short sales that family firms experience (compared to nonfamily firms) based on the size of the negative earnings surprise. As an example, based on a $0.10 negative earnings shock and holding all other factors constant, daily abnormal shorts in a nonfamily firm increase by 0.19%. The same shock in a family firm results in daily abnormal short sales increasing by 1.28%. Family firm abnormal short sales are over six times more sensitive to the size of negative earnings shocks than abnormal short sales in nonfamily firms. We calculate this multiple as the sum of the coefficient estimates on unexpected earnings and the interaction of unexpected earnings and family firm, divided by the coefficient estimate on unexpected earnings [= (0.019 × 0.10 + 0.109 × 0.10)/(0.019 × 0.10) = 6.74].

The preceding analysis indicates that: (1) family firms experience greater daily abnormal short sales prior to any negative earnings surprise than nonfamily firms (about 1.9% more) and (2) family firm daily abnormal short sales are substantially more sensitive to the size of negative earnings shocks than short sales in nonfamily firms (over six times more sensitive). Based on a $0.10 negative earnings shock, our results indicate that daily abnormal short sales for a family firm increase by 3.18% (1.9%+ 1.28%) and daily abnormal short sales for a nonfamily firm increase by 0.19%. Thus, on an aggregate basis, family firms sustain nearly 17 times more abnormal short sales prior to a negative earnings shock than nonfamily firms (3.18%/0.19%= 16.7). This result indicates that short sales for a family firm increase by about 11,433 shares for each of the 22 days during the preannouncement period.12 Short sales for a nonfamily firm increase by about 683 shares for each of the 22 trading days during the preannouncement period. The analysis provides fairly compelling evidence that short sellers in shares of family firms anticipate negative earnings surprises, consistent with informed trading.

Columns 3 and 4 of Table III, Panel A, examine the level of short sales preceding positive earnings surprises. Although only significant at the 10% level, the results indicate that family firms experience less abnormal short selling in advance of a positive earnings surprise relative to nonfamily firms. The stand-alone family firm variable suggests that, prior to any positive earnings shocks, family firms experience significantly less abnormal short selling than nonfamily firms. The interaction term between family firm and unexpected earnings surprises also shows a negative and marginally significant coefficient estimate, indicating that, as the magnitude or size of the positive earnings surprise increases, family firms experience incrementally less short selling relative to nonfamily firms. Column 5 of Panel A in Table III combines negative and positive earning shocks into the same regression specification and yields similar inferences regarding informed trading in family-controlled firms.13 Our control variables have similar signs and significance levels as in prior short-sale studies. For instance, we note that stocks listed on NYSE exhibit less abnormal short sales than stocks listed on other exchanges and that firm size is negatively related to abnormal short sales. The results of the analysis on the relation between short sales and negative/positive earnings surprises, taken together, provide evidence consistent with the notion that informed trading affects substantially greater levels of short selling in family firms than nonfamily firms.

Although we control for a variety of firm-specific characteristics, a useful robustness check centers on comparing family firms to nonfamily firms that have similar traits. Specifically, we construct a propensity score–matched sample of family and nonfamily firms. Using a logit model with the family firm dummy as the dependent variable, we match family to nonfamily firms based on: unexpected earnings, hedge fund/private equity dummy, mutual/pension fund and insurance company dummy, firm size, lag ROA, bid-ask spread, forecast dispersion, book-to-market, stock return volatility, NYSE dummy, CGQ, stock trading volume, and put option volume. Our propensity score model uses one-to-one firm matching, a caliper of 0.1, and a common support range of 0.1 to 0.9 (Villalonga (2004), Caliendo and Kopeinig (2008)). The matching process yields a sample of 431 family firms and 431 nonfamily firms for a total of 2,343 firm-quarter observations for negative earning shocks and 2,434 firm-quarter observations for positive earnings shocks.

Consistent with our earlier results, the matched sample analysis, available in the Internet Appendix, suggests that family firms experience substantially greater short selling in advance of negative earnings surprises than nonfamily firms. The coefficient estimates on the stand-alone family firm variable and the interaction term between family firm and unexpected earnings are significant at the 5% level (t-values of 2.37 and 2.41, respectively). Notably, the matched sample results also indicate that hedge funds exhibit a positive and significant relation to abnormal short sales prior to negative earnings shocks, providing evidence consistent with Massoud et al. (2011) that hedge funds potentially engage in informed short sales. Overall, our analysis indicates that family firms exhibit substantially greater short selling in advance of negative earnings surprises than nonfamily firms, providing evidence in line with informed trading.

B. Family Characteristics and Short Sales

In our initial analysis examining the relation between short sales and unexpected earnings shocks, we use a binary variable to denote family presence. Prior literature, however, indicates that characteristics of family control and ownership potentially influence the firm's information environment (Anderson and Reeb (2003), Villalonga and Amit (2006), Bertrand et al. (2008)). For instance, a family member serving as CEO or family members holding multiple board seats arguably provide nonofficer/nondirector family members or other outsiders with access to private information on firm activities.

We examine four family attributes: active versus passive management, board presence, ownership levels, and divergence between ownership and control. In our regression specifications, we continue to include the family firm binary variable and add dummy variables for the family characteristic or trait under consideration (e.g., CEO type, number of board seats, etc.). This regression specification allows us to examine the incremental effect of the specific family characteristic on top of the general family effect. The Internet Appendix provides an alternative specification that excludes the family firm binary variable but includes dummy variables for all categorizations of the family characteristic under consideration. We find similar inferences with either specification.

Beginning with active family management, we segregate family firm CEOs into founder CEOs and descendant CEOs, with professional outside CEOs of family firms as the omitted or excluded variable. The results shown in Table III, Panel B, Column 1 continue to indicate that, in general, family firms experience significantly greater abnormal short selling in advance of negative earnings surprises relative to nonfamily firms. Notably, the results indicate that the presence of a founder CEO or descendant CEO appears to increase the level of short selling in advance of negative earnings relative to professional outside CEOs in family and nonfamily firms. That is, active family management, by either the founders, or their descendants, appears to be associated with increased abnormal short selling in advance of negative earnings surprises.

Column 2 of Table III, Panel B, examines the relation between family board presence and abnormal short sales. Due to difficulties in interpreting interaction terms that comprise two continuous variables (family board presence * unexpected earnings surprise), we create a binary variable that equals one when two or more family members serve as directors and zero otherwise. The mean (median) number of family board seats for a family firm in our sample is 1.73 (2.00). The analysis continues to indicate that, in general, family firms experience greater abnormal short selling prior to negative earnings surprises than nonfamily firms. The results further suggest that firms with a greater number of family directors experience significantly more abnormal short sales than family firms with fewer directors and nonfamily firms. The analysis thus supports the notion that close family association (i.e., management and director positions) is associated with greater abnormal short sales in advance of negative earnings surprises.

Family ownership level also potentially influences abnormal short sales. Families with a high ownership level, for instance, may be reluctant to draw attention to themselves or the firm due to short-selling activity. Column 3 of Panel B, Table III, examines the level of family ownership. We construct a binary variable that equals one when families hold less than 20% of the firm's cash-flow rights and zero otherwise. The median level of family ownership in terms of cash-flow rights for the sample is 19.57%. Consistent with earlier results, we continue to find that, in general, family firms experience more abnormal short sales in advance of negative earnings surprises than nonfamily firms. Family firms with low levels of cash-flow rights (<20%), however, tend to experience greater abnormal short sales prior to negative surprises than family firms with higher cash-flow rights and nonfamily firms.

Our analysis of family ownership level focuses on the level of family cash-flow rights. Yet, families often have control in excess of their cash-flow rights through the use of dual-class share structures. Firms with dual-class share structures allow families to maintain control through their superior voting class even if they sell their lower voting-class stake. In our sample, 20.1% of family firms use dual-class shares with these families, on average, holding 32.8% of the cash-flow rights and 61.0% of the voting rights. Column 4 of Panel B, Table III, investigates, the effect of dual-class share structure on abnormal short sales. We construct a binary variable that equals one when family voting rights exceed their cash-flow rights (dual-class firms) and zero otherwise. The analysis again indicates that, in general, family firms experience more abnormal short sales than nonfamily firms. Notably, however, we find that dual-class family firms exhibit significantly greater abnormal short sales than single-class family firms and nonfamily firms. The Internet Appendix repeats the analysis for the propensity score–matched sample and yields similar inferences.

The preceding analysis offers some insight into the effects of family attributes on abnormal short sales. Care should be taken, however, in interpreting any single family characteristic variable. Many family characteristic variables are not mutually exclusive of one another. For example, firms with a descendant CEO may have other family members serving on their boards, may use a dual-class share structure, and may have low cash-flow ownership rights. Many permutations of these variables exist.14 The Internet Appendix provides a correlation matrix for the family characteristic variables. Accordingly, we cannot unambiguously point to any single family attribute that accounts for our results, suggesting that an all-inclusive family firm binary variable potentially renders the most straightforward inferences. From our analysis, however, we note that active family management, greater family board presence, lower levels of cash-flow rights, and dual-class shares structures are associated with greater abnormal short sales in advance of negative earnings announcements relative to nonfamily firms.

C. Other Large Shareholders and Short Sales

The results thus far indicate that family firms experience greater abnormal short sales in advance of negative earnings surprises than nonfamily firms. However, our analysis suffers from the caveat that we do not know the identity of the short sellers. Although the short sellers could be family owners, other large blockholders arguably possess similar profit incentives to family shareholders and often have access to the same private information. The results could thus be driven by organizational structure (large blockholders) rather than the presence of a particular type of large blockholder (family owner). To investigate this possibility, we examine family firm short sales in advance of negative earnings shocks relative to two other types of organizational structures: nonfamily firms with no blockholders (diffusely held nonfamily firms) and nonfamily firms with blockholders (block-held nonfamily firms). Diffusely held nonfamily firms constitute 11.5% of the observations for the full sample and block-held nonfamily firms comprise 52.2% of the observations. Blockholders consist of hedge funds, insurance companies, mutual funds, pension funds, and private equity funds that hold 5% or more of the firm's shares.

Panel C of Table III presents the results. In Column 1, we examine the relation between abnormal short sales prior to negative earnings surprises for family firms compared to diffusely held nonfamily firms (the benchmark group). The results continue to indicate that family firms experience significantly more abnormal short sales prior to negative earnings announcements relative to diffusely held nonfamily firms. The coefficient estimates on the stand-alone family firm variable and on the interaction term between family firm and unexpected earnings surprises show positive and significant coefficient estimates.

Column 2 of Panel C shows family firms relative to block-held nonfamily firms. Again, the analysis indicates greater abnormal short sales prior to negative earnings shocks in family firms than in block-held nonfamily firms. As with our prior analyses, we note that the magnitude of short sales in family firms in advance of negative earnings surprises is several multiples larger than that observed in nonfamily firms. This result suggests that the presence of family owners rather than the blockholder organizational form is the primary driver of abnormal short selling in advance of negative earnings shocks.

Our results on family firm short sales could also potentially be driven by outside blockholders obtaining private information and assuming short positions prior to a negative earnings surprise. To investigate this conjecture, we examine the relation between abnormal short sales and the presence of outside blockholders. Column 3 of Panel C, Table III, shows the results. The stand-alone family firm variable and the interaction term between family firm and unexpected earnings shocks continue to exhibit positive and significant coefficient estimates. The outside blockholder variables, however, do not exhibit a significant relation to abnormal short sales, suggesting that our results do not appear to be driven by outside blockholders.

All blockholders, however, do not share similar investment styles. Anecdotal accounts indicate that hedge funds and private equity funds may be more aggressive in their trading strategies than mutual funds, pension funds, or insurance companies.15 Column 4 breaks outside blockholders into two classes: (1) hedge funds and private equity funds and (2) mutual funds, pension funds, and insurance companies. Again, we continue to find that family firms experience significantly more abnormal short selling in advance of negative earnings surprises than nonfamily firms. The analysis further indicates that the variables for the two subclassifications of outside blockholders do not exhibit a significant relation to abnormal short sales. The Internet Appendix repeats the blockholder analysis for the propensity score–matched sample. Overall, the presence of a specific type of large blockholder (family owners) appears to be a principal marker for our results rather than the blockholder organizational form.

D. Short Sales and Future Stock Returns

Asquith et al. (2005) argue that high short sales forecast low future stock returns. A question thus arises as to whether the greater levels of short sales in family firms that we observe provide valuable or useful information in predicting abnormal stock returns. In particular, we ask whether short sales in family firms better forecast stock returns versus short sales in nonfamily firms.

We answer this question by ranking and categorizing family and nonfamily firms (separately) into quintiles based on daily short-sale interest on day t. We then calculate future abnormal stock returns on day t + 2 for each of the five family firm portfolios and each of the five nonfamily firm portfolios using standard asset pricing models, that is, Fama–French three- and four-factor models. Daily short-sale interest is measured as daily short-sale volume divided by daily stock trading volume. The stocks in the portfolios are value-weighted and rebalanced daily. Daily return data come from CRSP and, to mitigate concerns about bid-ask bounce (Kaul and Nimalendran (1990)), we eliminate day t+ 1 from the analysis and focus our attention on the ability of daily short-sale interest on day t to forecast portfolio returns on day t + 2. The daily return factors for the Fama–French three- and four-factor models are obtained from Kenneth French's website. The intercept (α) terms in the three- and four-factor models are our measures of day t + 2 abnormal returns.

Table IV, Columns 1 through 4, presents the abnormal returns on day t + 2 for the five family firm portfolios and the five nonfamily firm portfolios based on short sales on day t. The analysis indicates that the family firm portfolio (Columns 1 and 2) with the lowest level of daily short selling (Low) generates positive abnormal returns of 3.7 and 3.3 basis points from the Fama–French three- and four-factor models, respectively. In contrast, the family firm portfolio with the greatest level of short selling (High) earns abnormal returns of only 0.9 (0.9) basis points from the three- (four-) factor model. The analysis further indicates that abnormal returns in family firms appear to be decreasing as a function of short sales. Specifically, moving from the portfolio with the lowest level of short sales to the portfolio with the highest level of short sales, we observe a generally decreasing trend in abnormal returns. An investment strategy of buying the family firm portfolio with the lowest level of short sales and shorting the family firm portfolio with the highest level of short sales (i.e., a long–short strategy) generates a daily abnormal return of 2.8 basis points or 60.9 basis points per month, excluding rebalancing costs.

Table IV.  Short Sales and Future Stock Returns This table presents daily value-weighted portfolio returns for day t + 2 for Fama–French three- and four-factor models. α3 and α4 are the abnormal daily stock returns in percent derived from the Fama–French three- and four-factor models. The portfolios are formed by ranking family firms and nonfamily firms, separately, into quintiles based on daily short-sale interest on day t. Low indicates the portfolio of firms in the bottom quintile of short selling. High indicates the portfolio of firms in the top quintile of short selling. Low − High is the difference in portfolio returns between the low portfolio and the high portfolio. Columns 1 through 4 provide the results for the full sample of firms and Columns 5 through 8 provide the results for the matched sample.
 Full SampleMatched Sample
Family Firms (n= 301,720)Nonfamily Firms (n= 523,264)Family Firms (n= 223,278)Nonfamily Firms (n= 218,277)
α3  factorα4  factorα3  factorα4  factorα3  factorα4  factorα3  factorα4  factor
  1. The t-statistics are shown in parentheses.

Low0.0370.0330.0170.0170.0310.0330.0140.015
 (1.68)(1.70)(1.93)(1.93)(3.33)(2.56)(2.98)(2.54)
20.0300.0290.0240.0230.0270.0310.0390.038
 (0.32)(0.31)(0.80)(0.80)(1.55)(1.44)(0.51)(0.55)
30.0210.0200.0440.0440.0250.0270.0020.002
 (0.88)(0.86)(0.33)(0.82)(1.30)(1.29)(0.86)(0.91)
40.0230.0230.0220.0230.0220.0210.0110.011
 (1.41)(1.33)(1.62)(1.57)(1.53)(1.40)(0.94)(1.01)
High0.0090.0090.0230.0230.0180.0160.0140.013
 (2.18)(2.15)(2.66)(2.73)(2.35)(2.70)(2.23)(2.23)
Low − High0.0280.024−0.006−0.0060.0130.0170.00030.001
 (2.89)(2.57)(2.78)(2.70)(5.44)(4.01)(4.46)(3.35)

Nonfamily firms present a substantively different picture. When ranking nonfamily firms based on daily short-sale interest (Columns 3 and 4), we note little difference in abnormal returns among the portfolios. The nonfamily portfolio with the lowest level of short sales generates a positive abnormal return of 1.7 basis points. In contrast, the nonfamily portfolio with the highest level of short sales earns an abnormal return of 2.3 basis points. The abnormal returns of the nonfamily portfolios do not exhibit any discernible pattern as daily short-sale interest increases. An investment strategy of buying the nonfamily firm portfolio with the lowest level of short sales and shorting the nonfamily firm portfolio with the highest level of short sales (i.e., a long–short strategy) generates a daily abnormal return of −0.6 basis points or −12.8 basis points per month. Overall, the analysis indicates that daily short-sale interest in family firms provides valuable information in forecasting future short-term stock returns. Daily short-sale interest in nonfamily firms, however, shows little, if any, power in predicting future returns. We interpret this evidence as suggesting that informed trading appears to drive a greater level of short sales in family firms than in nonfamily firms.

As an additional robustness check, we repeat the above analysis on the propensity score–matched sample outlined in Section II.A. The results are shown in Table IV, Columns 5 through 8. Similar to the full sample, we continue to find that family firm abnormal returns decrease as a function of short sales. A long–short investment strategy for the family firm–matched sample earns a daily (monthly) positive abnormal return of 1.3 (28.3) basis points based on the Fama–French three-factor model. A similar strategy for the diffuse shareholder firm–matched sample earns a daily (monthly) positive abnormal return of only 0.03 (0.61) basis points. The results from the matched sample provide evidence consistent with that from the full sample, indicating that short sales in family firms appear to contain more valuable information in forecasting future returns than short sales in nonfamily firms.

Our earlier analysis indicates that family ownership and control characteristics bear an association with short-sale activity. Table V presents the abnormal stock return analysis using the Fama–French three- and four-factor models for family CEO type, family board presence, family ownership level, and family dual-class share structure. The analysis indicates that the association between family characteristics and short sales appears to contain useful information in predicting future short-term stock returns. We find that: (1) a long–short investment strategy for founder CEO and descendant CEO firms provides daily positive abnormal returns of 2.1 and 2.6 basis points, respectively (a similar strategy for a professional CEO firm yields −0.1 basis points), (2) a long–short investment strategy for family firms with two or more family directors provides a daily positive abnormal return of 2.7 basis points (firms with zero or one family directors yield a return of 1.8 basis points), (3) a long–short investment strategy for family firms with low cash-flow rights (<20%) provides a daily positive abnormal return of 2.0 basis points (family firms with high cash-flow rights (≥20%) yield a daily positive abnormal return of 1.3 basis points), and (4) a long–short investment strategy for family firms with a single-class share structure provides a daily positive abnormal return of 1.4 basis points (family firms with a dual-class shares structure yield a daily positive abnormal return of 3.4 basis points).

Table V.  Short Sales and Future Stock Returns Categorized by Family Characteristics This table presents daily, value-weighted portfolio returns for day t + 2 for Fama–French three- and four-factor models. α3 and α4 are the abnormal daily stock returns in percent derived from the Fama–French three- and four-factor models. The portfolios are formed by ranking the family characteristic into quintiles based on daily short-sale interest on day t. Low indicates the portfolio of firms in the bottom quintile of short selling. High indicates the portfolio of firms in the top quintile of short selling. Low − High is the difference in portfolio returns between the low portfolio and the high portfolio.
 Founder CEO (n= 117,332)Descendant CEO (n= 60,328)Professional CEO (n= 124,060)Family Holds <2 Board Seats (n= 152,672)Family Holds ≥2 Board Seats (n= 149,048)
α3  factorα4  factorα3  factorα4  factorα3  factorα4  factorα3  factorα4  factorα3  factorα4  factor
Low0.0240.0240.0280.0290.0260.0200.0080.0080.0120.012
 (2.81)(2.80)(2.45)(2.34)(2.65)(2.64)(3.77)(3.75)(5.37)(5.37)
20.0210.0200.0250.0250.0190.0150.0050.0050.0060.006
 (1.90)(1.72)(1.92)(1.77)(2.31)(2.29)(2.55)(2.56)(3.33)(3.30)
30.0160.0150.0220.0210.0300.031−0.002−0.002−0.002−0.002
 (1.44)(1.03)(1.67)(1.80)(4.44)(4.44)(0.56)(0.56)(1.01)(1.02)
40.0100.0100.0150.0150.0330.029−0.003−0.003−0.005−0.005
 (0.86)(0.87)(0.90)(1.02)(4.32)(4.33)(1.99)(2.00)(2.41)(2.43)
High0.0030.0040.0020.0020.0270.028−0.010−0.009−0.015−0.015
 (2.87)(2.55)(2.66)(2.55)(3.87)(3.88)(3.01)(3.00)(5.92)(5.95)
Low−High0.0210.0200.0260.027−0.001−0.0080.0180.0170.0270.027
 (3.67)(3.57)(3.45)(3.33)(4.34)(4.56)(3.93)(3.89)(6.26)(6.33)
 Family Cash-Flow Rights <20% (n= 154,082)Family Cash-Flow Rights ≥20% (n= 147,638)Single Class (n= 241,175)Dual Class (n= 60,545)
α3  factorα4  factorα3  factorα4  factorα3  factorα4  factorα3  factorα4  factor
  1. The t-statistics are shown in parentheses.

Low0.0130.0130.0080.0080.0070.0070.0150.015
 (3.33)(3.30)(3.01)(3.00)(2.08)(2.10)(4.22)(4.21)
20.0050.0040.0050.0050.0080.0080.0090.008
 (2.42)(2.44)(2.00)(2.03)(0.41)(0.42)(3.28)(3.30)
3−0.001−0.0020.000−0.000−0.012−0.011−0.006−0.006
 (0.99)(0.97)(0.77)(0.78)(0.60)(0.65)(1.05)(1.10)
4−0.003−0.003−0.002−0.002−0.005−0.005−0.009−0.009
 (2.01)(2.00)(1.92)(1.90)(2.45)(2.49)(2.87)(2.89)
High−0.007−0.007−0.005−0.005−0.007−0.006−0.019−0.018
 (3.55)(3.54)(2.65)(2.66)(3.18)(3.17)(4.59)(4.59)
Low−High0.0200.0200.0130.0130.0140.0130.0340.033
 (3.92)(3.85)(3.66)(3.67)(3.77)(3.72)(6.01)(6.11)

The evidence indicates that short sales tend to be more informative in predicting future returns for family firms relative to nonfamily firms. Moreover, family ownership and control attributes exhibit a relation to daily short-sale interest and future returns. We note that active family management, greater family board presence, lower cash-flow rights ownership, and a dual-class share structure tend to magnify the relation between short sales and future returns. Overall, the evidence suggests that short sales better predict future returns in family firms than nonfamily firms, consistent with greater informed trading.

E. Future Returns: An Alternative Approach

The analysis generally suggests that short sales in family firms better forecast abnormal returns than short sales in nonfamily firms, consistent with an informed trading argument. Yet, the differing relation between short selling and future returns for family firms and nonfamily firms may also be affected by other factors, such as stock price momentum, share liquidity, or stock price uncertainty, that influence daily short-sale interest (Diether et al. (2009)). To further investigate our central finding and to control for other factors that potentially affect short selling, we examine future returns using the following specification:

  • image(3)

The variables are defined in Table I. The above regression specification, using a Fama–MacBeth methodology, examines whether daily short-sale interest on day t predicts future short-term returns on day t + 2.16 The returns in the model are raw unadjusted returns. The term rt-5  to  t-1 captures the effect of past short-term stock price movements on future returns, that is, a momentum effect. To allow for the likelihood of a nonlinear relation between past and future returns, we rank each of the individual stocks into quintiles based on their returns over the past 5 days and assign the stock a standardized value ranging from zero to one. Those stocks with a return in the top quintile receive a value of one and those with a return in the bottom quintile receive a value of zero. The variable Risk allows for the possibility that traders engage in greater short selling due to greater information uncertainty around the stock/firm. The variable Turnover controls for the possibility that high trading volume (liquidity) signals a demand shock that potentially leads to greater future returns (Llorente et al. (2002)).

Table VI presents the regression results. Columns 1 and 2 show results for family firms and nonfamily firms, respectively, across the full sample.17 Columns 3 and 4 show the analysis for our propensity score–matched sample of family firms and nonfamily firms. The full and matched samples yield the same inferences and, thus, for brevity we only discuss the full-sample results. After controlling for liquidity, risk, and momentum factors that potentially affect stock returns, the results of the panel regressions corroborate those from the Fama–French specifications. Specifically, current short sales on day t in family firms tend to be predictive of future negative stock returns on day t + 2. The negative and significant coefficient estimate on short sales (SHORT) in Column 1 indicates that higher short interest today forecasts a decline in future returns. Economically, a 10% increase in today's short sales for a family firm predicts a 2.88 basis point decline in returns (2-days hence). This corresponds to a monthly decline of 0.63%, suggesting an economically significant effect.

Table VI.  Regressions of Future Stock Returns on Current Short Sales This table reports Fama–MacBeth results of regressing unadjusted daily stock returns for firm i on day t + 2 against short-sale interest for firm i on day t. For the full sample, we have 310,720 firm-day observations for family firms and 523,264 firm-day observations for nonfamily firms: Columns 1 and 2, respectively. For the matched sample, we have 223,278 firm-day observations for family firms and 218,277 firm-day observations for nonfamily firms in Columns 3 and 4, respectively.
 Dependent Variable = ri, t+2
Full SampleMatched Sample
FamilyNonfamilyFamilyNonfamily
  1. The t-values are reported in parentheses and are corrected for serial correlation and heteroskedasticity. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

Intercept0.048***0.103***0.023***0.111**
 (2.94)(5.61)(2.60)(2.14)
Shorti,t−0.288**−0.106−0.292***−0.115
 (−2.04)(−0.45)(−2.67)(−1.22)
(r−5,−1)i,t−0.392***−0.338**−0.339***−0.392**
 (−4.56)(−2.30)(−4.50)(−2.33)
Rank(r−5,−1)i,t0.055**0.0370.054**0.037
 (2.62)(1.45)(2.20)(1.31)
Riski,t−0.597**−0.863***−0.595**−0.860***
 (−2.50)(−3.46)(−2.55)(−3.39)
(Turnover−5,−1)i,t−0.000−0.001*−0.000−0.001
 (−0.73)(−1.86)(−0.71)(−1.86)
Observations310,720523,264223,278218,277
Average Adjusted-R20.0270.0270.0280.027

In contrast, the results in Column 2 for nonfamily firms indicate that current short sales bear no significant relation to future stock returns. Although negative, the coefficient estimate on short sales (SHORT) for nonfamily firms is not significant at conventional levels. Short sales appear to have little ability to forecast future returns for nonfamily firms. The control variables generally indicate that past positive stock returns (r-5, -1) and information uncertainty (Risk) are associated with future negative returns.

The analysis from the panel regressions corroborates the results from the Fama–French three- and four-factor models. Specifically, short sales in family firms contain valuable information in forecasting future stock returns but short sales in nonfamily firms have little predictive ability. We interpret the evidence as suggesting that informed trading appears to affect a greater portion of short sales in family firms than nonfamily firms.

III. Conclusion and Policy Implications

  1. Top of page
  2. ABSTRACT
  3. I. Data and Descriptive Statistics
  4. II. Multivariate Tests
  5. III. Conclusion and Policy Implications
  6. REFERENCES

Focusing on whether firm ownership structure can lead to the presence of informed traders, we investigate the informational content of short sales for U.S. publicly traded family and nonfamily firms. Using daily short-sale data, our empirical results indicate that family firms experience significantly greater abnormal short selling in advance of negative earnings surprises relative to their nonfamily counterparts. The results further indicate that the magnitude of short sales in family firms prior to a negative earnings shock is quite large compared to nonfamily firms—about 17 times larger. The investigation also suggests that short sales in family firms contain valuable or useful information in forecasting future short-term stock returns. However, short sales exhibit no discernible relation with future returns for nonfamily firms. Characteristics of family control and ownership exhibit a relation to informed short-sale activities. In particular, we observe that active family management (founder CEOs and descendant CEOs), greater family board presence, lower cash-flow ownership rights, and the use of a dual-class share structure appear to be important markers in understanding short sales in family firms. Yet, even in the absence of these specific family characteristics, our evidence continues to suggest that informed trading drives a greater portion of short sales in family firms than nonfamily firms.

Although informed trading by investors potentially improves market efficiency (Manne (1966), Easterbrook (1981)), the SEC typically discourages traders from engaging in this activity. The U.S. courts have ruled that trading based on material, nonpublic information obtained during the performance of corporate duties or through a breach of trust by a corporate insider constitutes illegal insider trading (Chakravarty and McConnell (1999)). Managers violate their fiduciary duties to the firm's other shareholders when conveying nonpublic information to a select group of investors (Carlton and Fischel (1983)). The theory of misappropriation of corporate information continues to require a breach of fiduciary responsibility or duty but expands the realm of individuals covered by insider trading laws (Hoffman (2007)). However, courts have held that trading based upon nonpublic information obtained without a breach of fiduciary responsibility is legal (see US v. O’ Hagan, 1997).

The U.S. Congress and the SEC may impose severe monetary penalties for insider trading but do not explicitly define this activity in legislation. Rather, Congress and the SEC typically allow courts to define the scope of insider trading (Dolgopolov (2004)).18 Judicial definitions of illegal insider trading have been developed through a series of court cases (e.g., SEC v. Texas Gulf Sulphur (1966), Dirks v. SEC (1984), US v. Carpenter (1986), US v. O’Hagan (1997)) and require a breach of fiduciary responsibility or duty for trades to be considered illegal. Thus, even after the passage of Regulation Fair Disclosure in 2000, Congress, the SEC, and the courts maintain no legal expectation that all shareholders will have equal information. The U.S. Supreme Court made this explicitly clear in Chiarella v. US (1980), whereby the court rejected the principle of a “parity of information.” Yet, based on Dirks v. SEC, founding family shareholders with a 10% or less ownership stake might be categorized as constructive insiders and thus acquire the status of corporate insider. If ruled to be corporate insiders, then these family members would have a fiduciary responsibility to represent the best interests of the firm. As Section 16(c) of the Securities and Exchange Act of 1934 prohibits corporate insiders from short selling their firm's stock, family members would therefore be proscribed from short selling the firm's shares. Moreover, when a family member serves as CEO, the insider classification extends to his/her immediate family.19

In our analysis, we do not take a stand on whether informed trading yields costs or benefits to the firm's other investors. On the benefit side, informed trading can facilitate market efficiency through better price discovery and limit the effects of idiosyncratic risk on corporate investment policy. On the cost side, informed trading potentially undermines investor confidence in the market's ability to provide a fair and level playing field and/or limits capital market development. Our results at first glance, however, argue for a better understanding of the investors or groups of investors behind short sales in family firms. Interestingly, the enforcement of insider regulations appears to be quite stringent with respect to mergers and acquisitions (Huddart, Ke, and Shi (2007)), suggesting that short sellers can use quarterly earnings announcements as a relatively obscure and arguably safe route to engage in informed trading. For instance, U.S. regulations provide far more severe penalties for trading based on private information regarding mergers and acquisitions relative to routine earnings announcements. Regulators explicitly justify the differential treatment by arguing that long-term shareholders bear relatively little harm (or benefit) from short-term stock price fluctuations around earnings announcements (Coffee (2007)).

This study contributes to our understanding of the effects of organizational structure on short sales and informed short selling. In particular, our analysis suggests that short-sale activity bears a strong relation to the presence of large, influential, undiversified shareholders, that is, family shareholders. Our empirical results on the relation between short sales and family control suggest the need for a broader investigation into the effectiveness of regulations that limit informed trading, if that is the desired outcome of regulatory control. In aggregate, the analysis indicates that organizational structure plays a critical role in assessing the informational content of short-sale activity in large publicly traded U.S. firms.

Footnotes
  • 1

    Hogan, Mike, 2009, How to avoid getting short-shrifted, Barron's 89, 34.

  • 2

    One school of thought is that founder- and heir-controlled firms look similar but differ along several key dimensions (Bertrand et al. (2008)). Others emphasize that founders and heirs face similar economic incentives because they characteristically hold poorly diversified stakes in a single firm (Anderson, Duru, and Reeb (2009)). Bennedsen et al. (2007) note the need for an additional consideration, in particular, the use of professional nonfamily CEOs to run the firm.

  • 3

    U.S. regulations treat any individuals with more than 10% ownership as insiders of the firm, thereby limiting their trading activity and requiring these investors to disclose their trading in the firm. In family-controlled firms, ownership is often dispersed among several individuals, keeping individual owners below the 10% threshold. In Section III, we discuss U.S. regulations concerning insider trading.

  • 4

    Brewster, Deborah, and Jennifer Hughes, 2008, Negative sentiment: Short-sellers under ever closer scrutiny, Financial Times (June 23): 6.

  • 5

    Barret, Larry, 2001, E-business software provider MicroStrategy has declared war on short sellers, investors who profit by betting a stock will fall, but the company might be its own worst enemy. ZDNet News (April 20): 1–7.

  • 6

    To ensure that our results are not driven by differences in unexpected earnings between family and nonfamily firms (Ali, Chen, and Radhakrishnan (2007)), we examine whether family and nonfamily firms differ in the frequency of unexpected earnings surprises and the magnitude of unexpected earnings surprises, and whether stock price reactions differ in response to earnings surprises. The results indicate that family and nonfamily firms exhibit similar frequencies of earnings surprises, similar magnitudes of earnings surprises, and similar stock price reactions to earnings surprises (see the Internet Appendix, which can be found at http://www.afajof.org/supplements.asp).

  • 7

    See the 2007, 2008, and 2009 reports by the Division of Enforcement at the SEC, titled “Outlines of Recent SEC Enforcement Actions.” These findings highlight another potential interpretation, specifically, that nonfamily firms protect information so poorly that the informed trading occurs at earlier dates. Expanding our event horizon, however, leads to similar inferences as those found in our principal tests.

  • 8

    In additional testing, we use Kyle's Lambda (Kyle (1985)) and the PIN as defined in Easley et al. (1996) with similar results.

  • 9

    ISS uses eight core topics to develop the CGQ rating: (1) board structure and composition, (2) audit issues, (3) charter and by-law provisions, (4) laws of the state of incorporation, (5) executive and director compensation, (6) qualitative factors, (7) director and officer ownership, and (8) director education. Overall, 61 different governance attributes comprise the CGQ rating.

  • 10

    Prior literature notes that higher price stocks experience greater short selling. In our analysis, we substitute the natural log of stock price for the book-to-market ratio and continue to find similar inferences between daily short-sale interest, family firms, and nonfamily firms. The results are also robust to the exclusion of nine firms with stock prices less than $1.00 per share at year-end. In separate robustness testing, we use SIC codes (two- or three-digit) to account for industry effects versus the Fama–French industry groups. SIC codes provide similar inferences to those reported in our principal analysis (Villalonga and Amit (2010)).

  • 11

    In arriving at the final quarterly sample, we lose 802 (906) negative (positive) firm-quarter observations due to missing data for three control variables: book-to-market (290 negative obs./335 positive obs.), forecast dispersion (306/350), and put option volume (206/221). The Internet Appendix shows the results for Table III, Panel A, when replacing the missing observations with a value of zero, using predicted values, and excluding the control variables from the regression. The results are similar to those reported in our principal analysis.

  • 12

    We calculate the shares shorted for family and nonfamily firms as the average daily stock trading volume (1.57 million shares) times 22.9%, the volume attributable to short sales (1,570,000 × 0.229 = 359,530). Family (nonfamily) firms experience an increase of 3.18% (0.19%) in daily short-sale volume during the preannouncement period. Therefore, for family firms we obtain 359,530 × 0.0318 = 11,433 shares and for nonfamily firms we obtain 359,530 × 0.0019 = 683 shares.

  • 13

    Our analysis primarily centers on short sales and negative earnings shocks. Yet, based on material nonpublic information, family or other shareholders potentially increase their long positions prior to positive earnings surprises. We examine this conjecture using a measure of abnormal stock trading prior to positive earnings shocks. The analysis, available in the Internet Appendix, indicates that family firms experience a marginally greater increase in trading volume in advance of positive earnings surprises relative to the increase in trading volume for nonfamily firms.

  • 14

    To gain further insight into whether the number of family members with access to private information affects abnormal short selling, we use the firm's age (years since inception) as a proxy for the number of family members (family size). That is, older firms are more likely to have several generations of descendants associated with the firm than, for instance, younger founder-CEO firms (Anderson, Mansi, and Reeb (2003)). Again, to avoid the issue of two continuous variables comprising an interaction term, we develop two binary variables for family size: small family size (firms younger than 45 years) and large family size (those older than or equal to 45 years); 45 years is the median firm age for the family firms in the sample. Our proxies for family size suggest that firms with more family members experience greater abnormal short sales prior to negative earnings shocks than firms with fewer family members or nonfamily firms.

  • 15

    Clark, Richard, 2008, Success on a shoestring, BusinessWeek 4097, 42–44.

  • 16

    For these regressions, we use a Fama–MacBeth method of regressing daily returns on day t + 2 on daily short-sale interest on day t for family firms and nonfamily firms for each day of the sample period. The coefficient estimate on the short variable (and other controls) from the daily regressions are then averaged across all days of the sample period to yield the Fama–MacBeth coefficient estimates and t-statistics provided in Table VI.

  • 17

    In additional testing, we included a control variable for stocks trading on NYSE. With the inclusion of the NYSE dummy, we continue to find that daily short-sale interest forecasts future returns in family firms but not in nonfamily firms.

  • 18

    A recent case illustrates this murky issue (SEC v. Cuban (2009)). In this case, Mark Cuban received sensitive information about a firm in which he was a shareholder (case briefs are available at http://www.fedseclaw.com). As he did not work for the firm and owned less than 10% of the stock, he was not classified as an insider according to SEC regulations (http://www.sec.gov) and there was no breach of fiduciary duty in his receiving the information. The SEC maintained that Mr. Cuban allegedly made an oral agreement to keep the information confidential, which also limited his ability to trade on the information. A federal district court ruling on July 17, 2009 dismissed the SEC's case, stating that third parties who accept confidential information from the firm without a breach of duty are not restricted from trading on such information. The implication is that individual shareholders with less than 10% stakes can trade on material nonpublic information as long as the information is obtained such that a breach of trust is not violated.

  • 19

    The SEC requires officers and directors to disclose their trades through Forms 3, 4, and 5. In a review of a random sample of our firms, we find no reported trades by officers or directors prior to quarterly earnings announcements. Yet, reported trades of family and nonfamily corporate insiders could exhibit differing disclosure effects and thus have differing effects on stock price discovery, suggesting an avenue for future research.

REFERENCES

  1. Top of page
  2. ABSTRACT
  3. I. Data and Descriptive Statistics
  4. II. Multivariate Tests
  5. III. Conclusion and Policy Implications
  6. REFERENCES
  • Ali, Ashiq, Tai-Yuan Chen, and Suresh Radhakrishnan, 2007, Corporate disclosures by family firms, Journal of Accounting and Economics 44, 238286.
  • Anderson, Ronald C., Augustine Duru, and David M. Reeb, 2009, Founders, heirs, and corporate opacity in the United States, Journal of Financial Economics 92, 205222.
  • Anderson, Ronald C., Sattar Mansi, and David M. Reeb, 2003, Founding family ownership and the agency cost of debt, Journal of Financial Economics 68, 263285.
  • Anderson, Ronald C., and David M. Reeb, 2003, Founding-family ownership and firm performance: Evidence from the S&P 500, Journal of Finance 58, 13011328.
  • Anderson, Ronald C., and David M. Reeb, 2004, Board composition: Balancing family influence in S&P 500 firms, Administrative Sciences Quarterly 49, 209237.
  • Asquith, Paul, Parag A. Pathak, and Jay R. Ritter, 2005, Short interest, institutional ownership, and stock returns, Journal of Financial Economics 78, 243276.
  • Banerjee, Ajeyo, and Woodrow Eckard, 2001, Why regulate insider trading? Evidence from the first great merger wave (1897–1903), American Economic Review 91, 13291349.
  • Bennedsen, Morten, Kasper Meisner Nielsen, Francisco Pérez-González, and Daniel Wolfenzon, 2007, Inside the family firm: The role of families in succession decisions and performance, Quarterly Journal of Economics 122, 647691.
  • Bertrand, Marianne, Simon Johnson, Krislert Samphantharak, and Antoinette Schoar, 2008, Mixing family with business: A study of Thai business groups and the families behind them, Journal of Financial Economics 88, 466498.
  • Caliendo, Marco, and Sabine Kopeinig, 2008, Some practical guidance for the implementation of propensity score matching, Journal of Economic Surveys 22, 3172.
  • Carlton, Dennis W., and Daniel R. Fischel, 1983, The regulation of insider trading, Stanford Law Review 35, 857895.
  • Chakravarty, Sugato, and John J. McConnell, 1999, Does insider trading really move stock prices? Journal of Financial & Quantitative Analysis 34, 191209.
  • Chan, Lilian H., Tai-Yuan Chen, and Gilles Hilary, 2010, Insider trading and family firms, Working paper, University of Hong Kong .
  • Christophe, Stephen E., Michael G. Ferri, and James J. Angel, 2004, Short-selling prior to earnings announcements, Journal of Finance 59, 18451875.
  • Coffee, John C., 2007, Law and the market: The impact of enforcement, University of Pennsylvania Law Review 156, 229312.
  • Demsetz, Harold, 1986, Corporate control, insider trading, and rates of return, American Economic Review 76, 313316.
  • Desai, Hemang, K. Ramesh, S. Ramu Thiagarajan, and Bala V. Balachandran, 2002, An investigation of the informational role of short interest in the Nasdaq market, Journal of Finance 57, 22632287.
  • Diamond, Douglas W., and Robert E. Verrecchia, 1987, Constraints on short-selling and asset price adjustment to private information, Journal of Financial Economics 18, 277311.
  • Diether, Karl B., Kuan-Hui Lee, and Ingrid M. Werner, 2009, Short-sale strategies and return predictability, Review of Financial Studies 22, 575607.
  • Dolgopolov, Stanislav, 2004, Insider trading and the bid-ask spread: A critical evaluation of adverse selection in market making, Capital University Law Review 33, 83180.
  • Easley, David, Nicholas M. Kiefer, Maureen O'Hara, and Joseph B. Paperman, 1996, Liquidity, information, and infrequently traded stocks, Journal of Finance 51, 14051436.
  • Easterbrook, Frank H., 1981, Insider trading, secret agents, evidentiary privileges, and the production of information, Supreme Court Review 1981, 309365.
  • Hoffman, Drew, 2007, Martha Stewart's insider trading case: A practical application of Rule 2.1, The Georgetown Journal of Legal Ethics 20, 707718.
  • Huddart, Steven, Bin Ke, and Charles Shi, 2007, Jeopardy, non-public information, and insider trading around SEC 10-K and 10-Q filings, Journal of Accounting & Economics 43, 336.
  • James, Harold, 2006, Family Capitalism: Wendels, Haniels, Falcks, and the Continental European Model (Harvard Business Press, Cambridge , MA ).
  • Kaul, Gautam, and Gautam M. Nimalendran, 1990, Price reversals: Bid-ask errors or market overreaction? Journal of Financial Economics 28, 6793.
  • Kyle, Albert S., 1985, Continuous auctions and insider trading, Econometrica 53, 13151335.
  • Llorente, Guillermo, Roni Michaely, Gideon Saar, and Jiang Wang, 2002, Dynamic volume-return relation of individual stocks, Review of Financial Studies 15, 10051047.
  • Manne, Henry G., 1966, In defense of insider trading, Harvard Business Review 44, 113122.
  • Massoud, Nadia, Debarshi Nandy, Anthony Saunders, and Keke Song, 2011, Do hedge funds trade on private information? Evidence from syndicated lending and short-selling, Journal of Financial Economics 99, 477499.
  • Morck, Randall, Daniel Wolfenzon, and Bernard Yeung, 2005, Corporate governance, economic entrenchment, and growth, Journal of Economic Literature 43, 655672.
  • Pérez-González, Francisco, 2006, Inherited control and firm performance, American Economic Review 96, 15591588.
  • Schulze, William S., Michael H. Lubatkin, and Richard N. Dino, 2003, Toward a theory of agency and altruism in family firms, Journal of Business Venturing 18, 473490.
  • Seyhun, Nejat H., 1992, Why does aggregate insider trading predict future stock returns? Quarterly Journal of Economics 107, 13031331.
  • Shleifer, Andrei, and Robert W. Vishny, 1986, Large shareholders and corporate control, Journal of Political Economy 94, 461488.
  • Shleifer, Andrei, and Robert W. Vishny, 1997, The limits of arbitrage, Journal of Finance 52, 3555.
  • Tufano, Peter, 1996, Who manages risk? An empirical examination of risk management practices in the gold mining industry, Journal of Finance 51, 10971137.
  • Villalonga, Belén, 2004, Does diversification cause the “diversification discount”? Financial Management 33, 527.
  • Villalonga, Belén, and Raphael Amit, 2006, How do family ownership, control and management affect firm value? Journal of Financial Economics 80, 385417.
  • Villalonga, Belén, and Raphael Amit, 2010, Family control of firms and industries, Financial Management 39, 863904.