What determines the level of short-selling activity?

Financial Management (Financial Management Association), Winter, 2007 by Hung Wan Kot

2. Overpricing Hypothesis: If an investor has private/inside information that shows the firm's future perspective is poor, i.e., the stock is overpriced, then short selling would be one of his choices. Diamond and Verrechia (1987) suggest that the short-seller will not trade unless he expects the price will decline enough to compensate for the additional costs and risks of short selling. Dechow et al. (2001) find a strong relation between the trading strategies of short-sellers and ratios of fundamentals (such as earnings and book values) to market prices.

3. Arbitrage Hypothesis: An investor might short to arbitrage a price differential between the stock and convertible securities into the stock (Dechow et al., 2001; Arnold, Butler, Crack, and Zhang, 2005), or short an acquirer's stock from a merger and acquisition announcement (Mitchell, Pulvino, and Stafford, 2004; Arnold et al., 2005). An additional security (such as stock options, stock futures) linked to the stock price is necessary also, to obtain arbitrage profits (Brent, Morse, and Stice, 1990).

4. Taxation Hypothesis: An investor will realize tax benefits when he shorts the stock while holding a long position, but does not deliver the stock to cover the short position. A short sale against the box allows the investors to lock in the profit, but defer the capital gain to a subsequent tax year (Dyl, 1978; Brent et al., 1990). However, after 1997, the Taxpayer Relief Act eliminated the opportunity to defer capital gains' taxes (Arnold et al., 2005).

II. Data Description and Variables Construction

In this section, I describe the data sources, and the variables, and predict their relations with level of short-selling.

A. Data Description

I obtain the data used in the paper from the following sources: The NYSE provides monthly short-interest data for the January 1988 to December 2002 period. Nasdaq provides monthly short-interest data for the June 1988 to December 1994 period. I download data from January 1995 to December 2002 directly from the Nasdaq website. I obtain common stock returns, trading volume, number of shares outstanding from CRSP daily and monthly databases. Compustat provides me with annual equity book value (Item 60), convertible debt and convertible preferred stocks (Item 39), and S&P primary index maker (CPSPIN). I obtain options availability data from Berkeley Options Database (1988-1995) and Option Metrics Database (1996-2002), and institutional holding data from Thomson Institutional Holding Database. SDC platinum provides me with merger and acquisition data. I obtain the NYSE composite index and the Nasdaq composite index from Datastream.

Both the NYSE and the Nasdaq report short-interest data on the 15th of each month, or the first business day before 15th of each month if the 15th is a holiday. Normally, it takes three business days to settle trades, so the short-interest number reflects the short sales that occur three business days prior to the 15th.

Because short-interest data for the Nasdaq are not available for February 1990 and July 1990, I follow Chen and Singal (2003) by using the average short interest for January and March to estimate the February short interest, and the average short interest for June and August to estimate the July short interest. The sample period in Options Metrics Database begins in 1996. Berkeley Options Database contains only options listed in the CBOE, but the Options Metrics Database covers options listed in all exchanges. Therefore, from 1988 to 1995, I can only identify common stocks with options listed in CBOE.