Liquidity effect on stock repurchases
The influence of market liquidity on the number of stock repurchases executed by firms in the
United States of America
Master Thesis
Abstract: Contrary to theoretical viewpoint on stock repurchases motives, this research finds
that the acquisition hypothesis, stock liquidity stimulation hypothesis, dividend substitution
hypothesis and age differentiation in the Executive board do not influence investment
decisions on stock repurchases. However it is found that the undervaluation hypothesis,
gender Executive Board differentiation and liquidity have a significant effect on stock
repurchase investments. Market and stock illiquidity negatively affect stock repurchases.
Undervalued stock and the increasing percentage of women in the executive Board
significantly stimulate stock repurchases. With the illiquidity effect on stock repurchases,
larger firms cope with a decreasing illiquidity effect on stock repurchases. Finally, it is
confirmed that new debt or excess earnings finance stock repurchases.
Key words: liquidity, stock repurchases, under-pricing hypothesis, gender Board
diversification, sector adjusted Tobin’s Q, commonality, fixed effect model, linear regression,
panel data.
Student Stefanie de Waard Number 6117309
Study Master Business Economics Track Finance
Supervisor Rafael Perez Ribas Date 7 July 2016
STATEMENT OF ORIGINALITY
This document is written by student Stefanie de Waard who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of Contents
1. Introduction 3
2. Theoretical framework and Literature review 6
2.1. Stock repurchase 6
2.2. Liquidity 8
2.3 Liquidity Commonality 11
3. Data and descriptive statistics 12
3.2. Descriptive statistics 17
4. Empirical methodology 17
5. Empirical results 19
5.1 Market, Sector and stock illiquidity effect on stock repurchases 20
5.2 Cash and Debt intermediation effect 26
6. Robustness check 27
7. Assumptions 30
8. Conclusion 30
9. Discussion 31
REFERENCES 32
APPENDIX 35
Appendix A - list of variables 35
1. Introduction
In 1954 the magnitude of the stock repurchase, measured as a percentage of total New York Stock Exchange (NYSE) trading, accounted c.1% (Guthart, 1967). In the US in 1971 stock repurchases accounted for 11.82% of total payouts, whereas in 2000 this value increased to 44.42% (Dittmar & Dittmar, 2002). This increasing trend in stock repurchases is based on several motivational frameworks for firms. Theoretical research provides knowledge on firm’s stock repurchase motives, however no economist, to the writer’s knowledge has linked the theoretical research to an empirical research. Understanding stock repurchase behaviour is important for predicting future stock repurchases behaviour but also to prevent liquidity dry-ups. This research provides in-depth knowledge on the effect of illiquidity on stock repurchases and contributes to theory on stock repurchases motives.
Firms repurchase stock as a response to stock undervaluation hypothesis, acquisition hypothesis, reduce equity base hypothesis, distribute excess cash hypothesis, as a substitute to dividend payments (substitution hypothesis), or to increase liquidity (liquidity stimulation hypothesis). However, how have firms been triggered by the stock repurchase motives over the last 15 years, and most importantly, how does market illiquidity effect the stock repurchase behaviour of firms? Are firm’s still approaching the repurchase strategy to increase their stock liquidity during economic events that cope with illiquidity waves? Are stock repurchases still applied as a substitute for dividend payments? Do acquisitions cause a significant increase in stock repurchases? What theoretical motives for stock repurchases have we actually seen in the last 15 years, and how were they funded?
The purpose of this research is to investigate the relation between illiquidity and stock repurchases. The objective research question of this study is therefor: ‘how does illiquidity effect the number of stock repurchases firms execute?’ To test the illiquidity effect on stock repurchases this research tests (1) the market illiquidity effect on the number of stock repurchases, (2) the sector illiquidity on the latter, and (3) the stock illiquidity effect on the number of stock repurchases. To ultimately come to an answer the market, sector and stock illiquidity, accompanied by control variables’ relation to stock repurchases is tested.
This research tests the relation from 2000 to 2015 of all listed firms on the American Stock Exchange (AMEX), New York Stock Exchange (NYSE), and the NASDAQ excluding the financial, real estate, insurance and public administration firms. Data is collected that include daily volume and stock return data to derive the illiquidity measure of Amihud (2002)1. Furthermore Executive Board
characteristics, exempli gratiā mean and standard deviation of the age and the gender female
percentage to total Executive Board members are taken into account. Board diversification effects investment decisions. Christiansen at al. (2016) document a positive association between corporate return on assets and firm’s performance with the share of women in senior position in line with previous finding by Erhardt et al. (2003). Gender diversity in addition contributes to innovation (Miller & del Carmen Triana, 2009). Age diversity is positively linked to donations (Siciliano, 1996), sales growth lending support for information and decision-making theories, and is found curvilinear with sales growth; positively at low and medium levels and negatively to sales growth at high levels (Richard & Shelor, 2002). Gurthart (1967) elaborated on theoretical incentives for firms to repurchase stock, based on a market research. All incentives for firms to repurchase stock are taken into the regression, such as the undervaluation hypothesis, expressed by a sector adjusted Tobin’s Q dummy variable, acquisition dummy variable, and the dividend yield, to check for the substitution effect of stock repurchases versus dividend payments. Amihud (2002) acknowledges firm size as a risk variable. The interaction effect of firm’s size that could potentially limit the illiquidity effect on stock repurchase expenses is further tested by including a standardized interaction term between illiquidity and size. Larger firms are more diversified, therefor it is expected to be less volatile to market illiquidity fluctuations. Initially, small cap firms on aggregate have relative lower stock liquidity than large-cap stock. In addition small-cap firms have less access to capital, funding or financial resources, are more vulnerable to consumer preference changes, and transparency of information on the stock is relatively lower than larger-cap firms. Reinganum (1990) distinguished the illiquidity effect between NASDAQ and NYSE by the difference in the meaning of the trading volumes on both stock exchanges. Trading on the NASDAQ is on aggregate done by market makers, and on the NYSE directly between buying and selling investors. Reinganum (1990) concluded that for this reason the NASDAQ and NYSE cope with different microstructure effects on stock returns. Furthermore to personal knowledge financial data are included that characterize the firm. These characteristics are divided in two subcategories. The first include the investment variables, and the second the solvency variables. The investment group includes research and development; property, plant and equipment; cash, and capital expenses. The investment variables are scaled to total assets. The second group, solvency, include the solvency ratio, current (working capital) ratio, and interest coverage ratio. To check the data for the effect of the financial crisis, a dummy variable is included that represents this period. In addition, yearly differences are regressed by including year dummies. The year-dummies capture the influence of aggregate (time-series) trends, to control whether stock repurchases have increased due to popularity or other external factors. Initially, the data set consists of c.4.2 million observations that account for 1,739 firms. The data include yearly and daily data, resulting that this research applies a fixed effect panel regression including cluster firm standard errors.
The results of this research verify the large negative relation of market illiquidity on stock repurchases. The firm size decreases the market and firm stock illiquidity effect. Additionally, the undervaluation
hypothesis and Board gender diversification hypothesis are accepted. The acquisition hypothesis is not considered significant. The undervalued stock motives to repurchase stock are mainly funded by extracting either debt, or via excess earnings. This relation is verified by the effect of the undervaluation variable on debt, cash and the interest coverage ratio (see chapter 3 on variable derivation and chapter 5 & 6 for the empirical results). The investment variables PP&E and CAPEX have a positive effect on stock repurchases. Increasing long-term investments either affect undervaluation that increase stock repurchases, or is correlated to other factors. PP&E is funded by debt and CAPEX (excluding the PP&E effect) is on aggregate funded by cash. Both variables reflect the firm’s aspiration for growth, executing stock repurchases that increase economic performance indicators. The diversification hypothesis is verified by the increasing ratio of women in the Executive Board that has a positive effect on the number of stock repurchases executed. The microstructure effects on return (Reinganum, 1990) are not significantly verified to be distinctive amongst stock exchanges.
Market liquidity can suddenly dry up, has commonality across securities, is related to volatility, is subject to ‘’flight to quality’’, and co-moves with the market (Brunnermeier and Pederson, 2008). Trading speculators are affected by market liquidity. These downward liquidity spiral effects are maintained when speculators hit their capital constraints effecting funding liquidity. The opposite encounters when speculators capital is abundant, proceeding insensitivity to marginal changes in capital and margins (Brunnermeier and Pederson, 2008). Additionally, a liquidity spiral obtains when initial losses occur on the market that cause funding problems for speculators, reducing positions, prices move away from fundamentals, causing higher margins and losses on existing positions that further cause for more funding difficulties for speculators and round the liquidity loss spiral and margin/haircut spiral goes.
To improve stock liquidity, firms can execute a stock repurchase strategy (Vermaelen, 1981; Comment & Jarrel, 1991; Ikenberry et al., 1995; Grullen & Michaely, 2002, Kahle, 2002; Lie, 2005). However how does a firm behave when the market is illiquid? Will it increase stock repurchases to increase stock liquidity or do funding or earning constraints debilitate this strategy, and hereby contribute to the liquidity spiral by decreasing stock repurchases?
For this reason, when speculators and market makers are unable to trade, thereby unable to improve liquidity, and it is significant accepted that firms additionally decrease stock repurchasing behaviour as a response to illiquidity; other players such as the Federal Reserve (FED) have to step in to increase liquidity to prevent liquidity dry-ups. An example is the implementation of the Quantitative Easing (QE) policy by the FED during and post crisis.
Theory explains stock repurchase increases as a response to excess capital payments, substitute for dividends, undervaluation hypothesis, and to increase liquidity (Guthart, 1967) however this empirical
research does not significantly accept the substitution hypothesis and the acquisition hypothesis. In addition, the theoretical motives for stock repurchases such as undervaluation hypothesis, distribution of excess cash are significantly verified in line with theoretical knowledge on stock repurchases. Simultaneously, age diversification does not significantly determine stock repurchase decisions, however gender board diversification is significantly accepted. The increasing females in Executive Boards contribute to the number of stock repurchases. Empirical research describes stock repurchases as a long-term investment in the company as it increases performance indicators, price, and stockholder’s gain.
The second order hypothesis and robustness chapter test the intermediate variables cash, debt and interest coverage ratio to see how stock repurchases are financed, and thereby affected by illiquidity. Debt and interest coverage ratio are intermediate variables with illiquidity and cash is not significantly considered an intermediate variable with illiquidity. The three tested variables all affect the undervaluation hypothesis variable, in mean of funding relation. In addition, the three tested variables are affected by the investment variables property, plant, and equipment, and capital expenditures. This relation initially describes how they are funded. Overall, the results explain the fundamental reason why stock repurchases decrease as a response to illiquidity up movements.
This research links the theoretical incentive for stock repurchases to empirical research that indicate which incentive has played a significant part in the last economic developing 15 years.
This research confirms the undervaluation hypothesis, distribution of excess cash hypothesis, however does not significantly find that stock repurchases are a substitution for dividend payments, and does not find a significant relation between acquisitions and stock repurchases. Additionally, this research contributes to economic behaviour theory on gender diversification that sees a weakly significant relation on female ratio increasing long-term investment strategies.
2. Theoretical framework and Literature review
The theoretical framework and literature review elaborated on the topic stock repurchases and liquidity that are required in understanding both individual and interfering behaviour. Section 2.1 comments on stock repurchase behaviour, section 2.2 on the empirical history of liquidity measures, and section 2.3 ends the chapter by explaining liquidity commonality that causes for liquidity dry-ups. 2.1. Stock repurchase
Stock repurchase refers to a company’s reacquisition of its own current outstanding shares. Share repurchase distributes capital to the shareholders that decide to sell. The remaining shareholders gain by the effect of the reduced number of shares outstanding. It accordingly increases earnings per share (EPS), cash flow per share (CFS) and other performance indicators e.g. return on equity (ROE). In
time stock repurchases stimulate the market share price, increasing shareholders’ capital gain. Repurchase programs often have maturities of several years or more (Ikenberry and Vermaelen, 1996)
Firm’s motives behind stock repurchases are either to reduce the equity base, prevent equity expansion due to e.g. limited investment opportunities or optimizing leverage, or because of control incentives. Also referring to the excess capital hypothesis, undervaluation hypothesis, takeover deterrence (acquisition) hypothesis, optimal leverage hypothesis and the management incentive hypothesis. The signalling hypothesis is the response hypothesis of the market’s thoughts behind the firm’s stock repurchases incentive.
Under investment incentives, this research considers the stock repurchase decision based on the concept that stock are considered undervalued, and in case of mergers and acquisitions (M&A) tax incentives. Control incentives include stock repurchase activity to increase ownership control or decrease minority shareholders. The prevention of equity expansion is to foresee equity dilution, e.g. in case executive equity vesting options are exercised, in case of conversion of outstanding convertible debentures and the exercising of warrants, prior to M&A, or to limit the market shock in case a majority shareholders desire to liquidate (Guthart, 1967).
Dunsby (1995) acknowledged that in mid-1980 when a high activity in share repurchase occurred, this was mainly caused by the distribution of earnings. Additionally, Dunsby (1995) addressed that firms that repurchase do not pay less dividends than other repurchasing comparable firm, which opposed literature that recognize stock repurchases as a substitute to dividend payments.
Opponents argue that stock repurchases are considered an alternative to the payout of dividends as a result of the considerable tax advantage in the US market. It is argued that stock repurchase is a tax-efficient way firms use to distribute wealth compared to cash dividends (Ikenberry and Vermaelen, 1996). Dunsby (1995) refers to the tax advantage into two ways, first by the transformation of the ordinary income to the capital gains, often taxed at lower rates, and secondly by allowing investors to defer taxes to future periods. However (1) companies that repurchase shares do not generally pay out fewer dividends than comparable companies that do not repurchase (Dunsby, 1995); (2) Ikennerry and Vermaelen (1996) find that the tax hypothesis can be considered seemingly inconsistent due to the fact that some firms that announce a stock repurchase however do not execute them, or partly; and (3), in 1986 after a tax reform act, abolishing the dividend and long-term gains, stock repurchase activity did not decrease (Bagwell and Shoven, 1989). Indicating hesitant on the substitution effect of stock repurchase and dividend payout caused by the tax hypothesis.
Singh et al. (1994) find that open market stock repurchase programs are followed after a period of steep decline in stock prices.
Dittmar (2004) provides evidence that stock repurchases are replacing dividend payments. Both dividend payment and stock repurchases are a means to distribute recurring earnings and therefore considered substitutes, additionally stock repurchases are also a means to distribute
temporary earnings. Fama and Fench (2001) find that the percentage of firms paying out dividend has decreased over time. However this is not completely linked to the tax hypothesis.
The signaling hypothesis is based on a scenario of imperfect information, where stock repurchase is seen as a market signal of the firms’ future prospects and that the stock is currently undervalued (Ikenberry and Vermaelen, 1996). Signaling hypothesis is one explanations of the market response to open market buyback announcement.
2.2. Liquidity
Liquidity is an incomprehensible concept. Liquidity is not observed directly, and copes with diversified aspect that cannot be captured into one single measure (Amihud and Medelson, 1991b). Illiquidity captures the impact of order volume, resulting from adverse selection costs and inventory costs, on price (Amihud and Mendelson, 1980; and Glosten and Milgrom, 1985). The price impact, for standard-size transactions, is reflected by the bid-ask spread. Amihud (2002) point out that the bid-ask spread is the preferred liquidity measure, however due to data limitations Amihud applied a different illiquidity measure. Larger excess demands encourage greater impacts on prices (Kraus and Stoll, 1772; Keim and Madhaven, 1996), reflecting a likely action of informed traders (Easley and O’Hara, 1987). Because market makers are not able to distinguish order flows generated by informed traders from liquidity (noise) traders, prices are set as indicators originated from informed trading (Kyle, 1985), creating a positive relationship between order flow and transaction volume and the price change, called price impact (Amihud, 2002).
Amihud and Mederlson (1986), and Eleswarapu (1997) applied the quoted bid-ask spread as a measure for illiquidity. Bid-ask spreads are positively affected by return volatility due to higher adverse selection and inventory risk (Stoll, 1978).
Chalmers and Kadlec (1998) applied the amortized effective spread as a measure for liquidity.2
Brennan and Subrahmanyam (1996) measure stock illiquidity by the price impact, inspired by measurement models of Kyle (1985), Glosten and Harris (1988), and Hasbrouck (1991). The former illiquidity measure is derived taking the intra-day continued data on transactions and quotes to regress the price response to the signed order size and by the fixed costs of trading.3
Easly et al. (1999) applied a microstructure risk measure, called the probability of information-based trading. It reflects the adverse selection costs resulting from asymmetric information between traders and reflects the risk that is involved with stock price deviation from its full-information value. This measure included the intra-daily transaction data.
2 The amortized effected spread is obtained from quotes and subsequent transactions. It is the absolute difference
between the mid-point of the quoted bid-ask spread and the transaction price that follows, classified by a buy or sell transaction, divided by the stock’s holding period, obtained from the turnover rate on the stock, to obtain the amortized spread (Amihud, 2002).
3 The slope coefficient of the regression is the transaction-by-transaction price change on the signed order size
Amihud (2002) applied the stock illiquidity measure defined as the average ratio of daily absolute return to the (dollar) trading volume on that day averaged over some period of time. This ratio represents the average absolute percentage price change per dollar per daily trading volume, or differently phrased, the daily price impact of the order flow over some period of time (Amihud, 2002). In principle this measurement is based on Kyle’s concept of illiquidity - the response of price to order flow - and on the thinness measure of Silber (1975). The latter is defined as the ratio of absolute price change over the absolute excess demand for trading. Amihud points out that the former illiquidity measure, measures the consensus belief among investors about new information, due to the effect of agreed or disagreed investors. For one, the effect of news on agreed investors cause stock price to adjust while volumes to stay the same, the news effect however on disagreed investors in addition to the price impact, cause an increase in trading volumes. Illiquidity is therefor positively and strongly related, with a R-squared of 0.3, to variables that measure illiquidity via the microstructure data (Brennan and Subrahmanyam, 1996).
Amihud and Mendelson’s (1989) measure turnover, the ratio of the trading volume to the number of shares outstanding, to be negatively related to illiquidity costs.
Stock expected returns are negatively related to size (Banz, 1981; Reinganum, 1981; Fama and French, 1992). Size, or the market value of the stock, is also related to liquidity, because a larger stock issuance has a smaller effect on price for a given order flow and a smaller bid-ask spread (Amihud, 2002). The higher return on small firms’ stock is considered a compensation for the information asymmetry, or less available information on small firms that have been listed for a shorter period of time (Barry and Brown, 1984). Brennan et al. (1998) confirm the negative effect on size related to the stock dollar volume’s significant negative effect on the cross-section of stock returns. Illiquidity costs increase due to occurring asymmetric information between traders (Glosten and Milgrom, 1985; Kyle, 1985), justifying a consistent proxy for liquidity (Amihud and Mendelson, 1986). Berk (1995) interprets that the negative return and size relationship could have been a result from the size variable being akin to the complementary expected return function. Stock returns are decreasing in stock turnover, which is consistent with a negative relationship between expected return and liquidity (Haugfen and Baker, 1996; Hu, 1997a; Datar et al., 1998; Rouwenhorst, 1998; Chordia et al. 2001). Empirical research proves that less liquid stocks cope with higher expected returns, due to the increased transaction costs and that the premium commoves with the bid-ask spread (Amihud and Mendelson, 1986; Eleswarapu, 1997). Bakaert et al. (2003) tested the liquidity effect on expected returns for emerging markets, stating additionally that the liquidity measure predicts future return and that turnover measures do not (Bakaert et. Al, 2003). In this research it is found that unexpected liquidity shocks have a positive effect on return and a negative effect on dividend yield. Additionally it
is found that liquidity market reaction to an open market repurchase is on average c.3%4 (Ikenberry and Vermaelen, 1996; Grullen & Michaely, 2002, Kahle, 2002; Lie, 2005).
The quoted (1) and effective (2) bid-ask spread is an effective measure for liquidity (Amihud, 2002). The Amihud illiquidity measure is shown in formula (3). The quoted bid ask spread is derived by the ratio of the bid-ask spread (𝐴𝑠𝑘!!− 𝐵𝑖𝑑
!
!) over the midpoint of the price (𝑚
!!), averaged over 𝑛!! periods. 𝑄!! = 1 𝑛!! 𝐴𝑠𝑘!!− 𝐵𝑖𝑑!! 𝑚!! !!! !!! (1) The effective spread measure is equivalent to the ratio of the absolute difference between price and mid-price (𝑝!!− 𝑚!!) over the mid price, averaged over some period of time 𝑛!!.
𝐸!! = 1 𝑛!! |𝑝!! − 𝑚 !!| 𝑚!! !!! !!! (2) The Amihud (2002) illiquidity measure, as earlier describes, is derived by dividing the absolute return (𝑅!"! ) by the dollar volume ($𝑉𝑂𝐿𝑈𝑀𝐸
!"! ), averaged over some period of time 𝐷!!. The dollar volume is equivalent to the daily closing price of stock (i) multiplied by the daily volumes trades for i.
𝐼𝐿𝐿𝐼𝑄𝑈𝐼𝐷𝐼𝑇𝑌!! = 1 𝐷!! |𝑅!"! | $𝑉𝑂𝐿𝑈𝑀𝐸!"! !!! !!! (3)
The measures for (il)liquidity mentioned above all measure different aspects of illiquidity. It is debatable whether there is one measurement that is able to capture all aspects. All described measures value the effects on stock return or stock repurchases. Due to the fact that no research has been done on the illiquidity effects on stock repurchases, we value the above methodologies based on its key measurement characteristics.
Hasbroyck (2006) justifies the relation between the Amihud illiquidity measure and the microstructure-based measures. Edision and Warnock (2003), Bekart et al (2003), Manzler (2005), Hashbrouck (2006), Acharya and Petersen (2005), Fang et al. (2006), and Choi and Cook (2007) further apply the Amihud illiquidity measure in the true or converted form.
Goyenko et al. (2009) test the liquidity measures across many countries. It is found that Amihud’s illiquidity, Past and Stambaugh’s Gamma, and Amivest’s illiquidity are not fitting as proxies for the effective or realized spreads. To capture price impact, Amihud shows a significant positive correlation contrary to the insignificance of Past and Stambaugh’s Gamma, and Amivest’s illiquidity measures. It is suggested by Goyenko et al (2009) to apply Amihud’s illiquidity or one of
4 Other confirming stock repurchase liquidity effect see: Comment and Jarrell (1991), Dann (1981); Ikenberry,
Goyenko at al.’s own effective spread measurement that divided the effective spread by volume when measuring price impact. However the latter methodology by Goyenko would require a special computer and for this research the latter methodology is too advanced.
For larger transactions, the price impact of the bid-ask spread dominates illiquidity costs (Kraus and Stoll, 1972; Keim and Madhaven, 1996). However due to the fact that we also take smaller transaction into the regression the illiquidity measure by Amihud (2002) that not only captures the cost associated effect of the bid-ask spread but also the price impact (tightness and depth) of both small and large trades, and additionally because the illiquidity measure is strongly related to the Brenman and Subrahmanyam’s (1996) microstructure measure of the price impact (Amihud, 2002) the Amihud illiquidity is given preference to all other methodologies in measuring liquidity. Especially since previous findings state that overall price improvement and reduction in volatility, rather than changes in market dynamics, cause improvements in liquidity (Nayar et al., 2008), it is decided in this research to make use of the most price effective liquidity measure addressed by Amihud (2002).
There is no empirical research on the effect of liquidity on stock repurchase. Dittmar and Dittmar (2004) find that stock repurchase waves are recurring because of corporate pay-out policies. Stock repurchases are valued as substitutes to dividend payment due to the fact that they are both alternatives in extending permanent earnings to shareholders and empirical research show an increasing trend in stock repurchases as a pay-out strategy for temporary earnings. So stock repurchases are dependent on firm’s earnings. Since earnings are dependent on liquidity it can be speculated that illiquidity will ultimately have a negative effect on stock repurchases.
2.3 Liquidity Commonality
Traders provide market liquidity, which is alternatively dependent on their funding availability. Traders’ capital and the margins they charge depend on asset market liquidity (Brunnermeier and Pederson, 2007). Liquidity not only affects stock returns, it also co-varies across stocks. This liquidity co-variation is called liquidity commonality that can arise on both the demand and supply side. The recognition of the existence of commonality is key in addressing suggestive evidence on the effect of inventory risk and asymmetric information on intertemporal changes in liquidity (Chordia et al., 1999).
Hameed, Kang, and Anathan (2010) find that the higher levels of commonality in liquidity are associated with liquidity dry-ups during the Asian financial crisis (1997), long-term capital management (LTCM) crisis (1998), and the 9/11 (2001) terrorist attacks. Additionally these periods cope with large negative market returns, displaying the event’s nature of illiquidity. Hameed, Kang, and Anathan (2010) concluding results on the commonality liquidity are that illiquidity becomes more correlated across all assets following market declines. Engle (2002) yield supportive findings. Eagle’s (2002) dynamic conditional correlations methodology find that the conditional correlations in
illiquidity (spreads) among size-sorted portfolios are significantly higher following large market declines contrary to market increases. However Hameed, Kang, and Anathan (2010) find that while liquidity commonality is driven by market supply and demand for liquidity, the demand factors cannot explain the asymmetric effect of market returns on liquidity. Additionally, investors become more risk averse in volatile markets, increasing the demand for liquidity (Vayanos, 2004). The former causes a market volatility jump, that further increased liquidity commonality.
Chordia et al. (2002) reemphasize on liquidity demand, where high levels of aggregate order imbalances increase liquidity commonality. Hameed et al. (2010) find consistency that the increased commonality liquidity in down market conditions is consistent with a decrease in liquidity supply. Karolyi et al. (2011) find that commonality in liquidity is greater in countries with and during times of high market volatility, this particularly holds during large market declines, greater presence of international investors and more correlated trading activity. Karolyi et al. main focus is on the funding liquidity hypothesis and mainly elaborated on the demand-side explanation of the latter.
The importance of commonality is that it explains liquidity dry-ups. The essence of this chapter is to address the commonality supply and demand response stream in order to understand the liquidity dry-up, and thereby contemplate the liquidity effects on stock repurchases. The next chapter describes the data selection and descriptive statistics
3. Data and descriptive statistics
The following section will elaborate on how data for this research is collected, cleaned and merged to excess the dataset that enables expressive regression clarifying market liquidity’s correlation with stock repurchases.
3.1. Data collection and sample selection
All data are collected via CRSP, Compustat, Execucomp, and the US Bureau of Economic
Development for the period from 2000-2015. This time frame is selected to obtain pre, current, and post financial crisis years to reflect the most recent period and behaviour characteristics in the
economic market. In obtaining all financial reported company data the dataset Compustat is accessed, Execucomp to collect executive data, CRSP to receive all daily stock data, and the US Gross Domestic Product (GDP) is collected from US Bureau Of Economic Development.
Exclusion criteria: companies with missing data such as ticker and fiscal year are excluded from the dataset. Additionally, division 8 and 10, as shown in table 1, are excluded from the data by their Standard Industrial Classification (SIC) codes. The SIC code, established by the US Government in 1937, represent four digit Sic codes per industry, and by two digit SIC coding the division classification, see table 1. This research will from hereon refer to the term Sector as a substitute term for the US government justification division codes, see column 1.
Daily stock information
The stock illiquidity ratio per company is calculated by the Ahimud (2002) illiquidity measure, see formula (3) to (6). The daily absolute return is divided by the dollar trading volume of that day averaged over some period of time p, i.e. trading days (𝐷!!) occurred in the period selected in year y. 𝑅!"! is the stock return of company i in year y on day d, 𝑅!"! is the average sector stock return in year y Table 1: Division based sector classification
Division SIC code Division Observations
1 0100-0999 Agriculture, Forestry and Fishing 0
2 1000-1499 Mining, incl. Oil and Gas 239865
3 1500-1799 Construction 88085
4 2000-3999 Manufacturing 2399178
5 4000-4999 Transportation, Communication, Electric, Gas and Sanitary Services 566458
6 5000-5199 Wholesale trade 176630
7 5200-5999 Retail trade 439455
8 6000-6799 Finance, Insurance and Real Estate 1007557
9 7000-8999 Services 847051
10 9100-9729 Public administration 0
10 9900-9999 Not Classifiable 8479
n.a. 9730-9899 n.a. 0
n.a. 1800-1999 not used 0
on day d, and 𝑅!"! is the average market stock return in year y on day d. Stock return is equivalent to the daily stock price growth rate of year d+1 and day d. Dollar volume is equivalent to the closing price times the volume traded for day d. Three illiquidity measures are applied in this research, stock illiquidity (4), sector illiquidity (5) and market illiquidity (6) on daily averages represented by the period d. Outliers are eliminated at the highest or lowest 1% tails of the distribution per year before running the illiquidity measures.
𝐼𝐿𝐿𝐼𝑄!!_𝑓𝑑 = 1 𝐷!! |𝑅!"! | $𝑉𝑂𝐿𝑈𝑀𝐸!"! !!! !!! (4 ) 𝐼𝐿𝐿𝐼𝑄!!_𝑠𝑑 = 1 𝐷!! |𝑅!"! | $𝑉𝑂𝐿𝑈𝑀𝐸!"! !!! !!! (5) 𝐼𝐿𝐿𝐼𝑄!!_𝑚𝑑 =𝐷1 !!, |𝑅!"!| $𝑉𝑂𝐿𝑈𝑀𝐸!"! !!! !!! (6) The new variables are created.
Company’s financial reported data
Company’s financial reported dataset include all reported financial data. Leverage is considered positively related to asset liquidity (Sibilkov, 2009). Costs of financial distress and inefficient liquidation affect capital structure decisions. It is for this reason, in order to derive the company’s
ability to meet its short and long-term liabilities that the following three ratios shown by formula (7), (8), and (9), are taken into the regression as control variables.
The current ratio, also known as the working capital ratio, is derived by the full year reported current assets (𝐶𝐴!!) divided by the full year reported current liabilities (𝐶𝐿
!
!) for firm i, year t, and created as a new variable (7).
𝐶𝑈𝑅𝑅𝐸𝑁𝑇!! =𝐶𝐴!!
𝐶𝐿!! (7) The debt to equity ratio, the degree of financial leverage used by the company, is applied to measure solvency ratio. The derivation divides the full year reported debt (𝐷!!) by its equity (𝐸!!) for firm i in year t, to create the new firm specific yearly solvency variable (8).
𝑆𝑂𝐿𝑉𝐸𝑁𝐶𝑌!! =𝐷!!
𝐸!! (8) Volatility return literature states that a drop in stock prices positively impact financial leverage, making stock riskier (increased solvency ratio) and increases subsequent liquidity (Black, 1976; Christie, 1982). Implying negative returns may increase spreads by the subsequent effect of volatility.
The firm specific interest coverage ratio is a debt and profitability ratio that determines the company’s ability to pay interest on outstanding debt. The interest coverage ratio shown underneath divides the full year reported earnings before interest and taxes (𝐸𝐵𝐼𝑇!!) by the company’s full year reported interest expenses (𝐼𝑁𝑇𝐸𝑅𝐸𝑆𝑇𝐸𝑋𝑃!!) for year t. The higher the coverage ratio, the better the company ability to cover its interest expenses.
𝐼𝐶𝑂𝑉𝐸𝑅𝐴𝐺𝐸!! = 𝐸𝐵𝐼𝑇!!
𝐼𝑁𝑇𝐸𝑅𝐸𝑆𝑇𝐸𝑋𝑃!! (9) Ratios (7), (8), and (9) are good indicators for a company’s financial strengths. Since stock repurchases can either be invested by excess cash or by taking on more debt, it is mandatory to take the above three ratios as control variables to control for a firm’s financial position. Note that during economic downfall these ratios are under pressure.
To measure firm size the following valuation methodology (10) is applied (Dalbor, 2004). Amihud (2002) confirms firm size, the value of stock, as a risk variable. To receive more in-depth knowledge on the relationship of illiquidity on stock repurchases this research for controls the interaction effect of firm’s size on illiquidity.
𝑆𝐼𝑍𝐸!! = 𝐿𝑁(𝐴𝑆𝑆𝐸𝑇𝑆
!!) (10) Tobin’s Q is applied to interpret the stock market mispricing. To measure the exact over- or undervaluation of the stock’s market value, the Tobin’s Q firm specific value difference to the sector Tobin’s Q is derived. Consequently because a stock can either be under- or overvalued the scenario can also be present that the whole sector is over- or undervalued. Therefore the firm specific Tobin’s Q is corrected for the sector Tobin’s Q. The stock is considered undervalued when it is lower than zero
(negative value), accurately valued when equal to zero, and overvalued when larger than zero (positive value). The Industry is based on SIC two digit values that represent the sector (see table 1). Assuming market value liabilities are equal to book value liabilities the formula for adjusted Tobin’s Q is shown underneath. Tobin’s Q is further created into a dummy variable, that is equal to one in case of undervaluation, and equal to zero otherwise. The adjusted Tobin’s Q is the measured by dividing the market value of the firm (𝑀𝑎𝑟𝑘𝑒𝑡𝑐𝑎𝑝 + 𝑑𝑒𝑏𝑡)!! by the book value of the firm’s equity (𝐵𝑉 𝐸 !!) ratio corrected for the Tobin’s Q sector ratio average, right-hand side of the subtraction operator, see formula (11). 𝑇𝑂𝐵𝐼𝑁𝑆𝑄!! =(𝑀𝑎𝑟𝑘𝑒𝑡𝑐𝑎𝑝 + 𝑑𝑒𝑏𝑡)!! 𝐵𝑉 𝐸 !! − 1 𝑛!"#$%&'( (𝑀𝑎𝑟𝑘𝑒𝑡𝑐𝑎𝑝 + 𝑑𝑒𝑏𝑡)!! 𝐵𝑉(𝐸)!! ! !!! (11) The dividend yield is representative to quantify how much cash flow the investor is receiving for each dollar invested in the stock. Additionally, the dividend yield can be interpreted as the return on investment for a stock. Amihud and Mendelson (1991a) find that Treasury notes with higher coupon provide lower yield to maturity. Amihud (2002) links this to higher dividend yields being recognized by investors, ignoring taxes, to providing a higher liquidity. The dividend yield may have a negative effect on stock return across stock if it is negatively correlated with an unobserved risk factor; say e.g. stocks with higher dividends are less risky (Amihud, 2002). Another negative effect could be caused by the preference of larger investors for companies with high liquidity, and for receiving dividends (Redding’s, 1997). However on the other hand, dividend yield could also have a positive effect on stock returns, due to the requirement of investors to be compensated for the higher tax rate on dividends contrary to the tax rate on capital gains that is lower (Amihud, 2002). The relation factor with stock return could indicate a correlation with illiquidity, since absolute return is taken in the numerator for the illiquidity measure. As for the effect on stock repurchase, literature elaborates that stock repurchase is applied as a substitute for dividend. The dividend yield (𝐷𝑖𝑣𝑖𝑑𝑒𝑛𝑑𝑌!!) is derived by dividing the dividend per share (𝐷𝑃𝑆!!) by the price per share (𝑃𝑃𝑆!!), see formula (12).
𝐷𝑖𝑣𝑖𝑑𝑒𝑛𝑑𝑌!! = 𝐷𝑃𝑆!! 𝑃𝑃𝑆!!
(12) The financial variables CASH, DEBT, CAPEX, R&D and PP&E are scaled to total assets, and the total amount of stock repurchases are scales to the total amount of shares outstanding represented by the variable NSR.
Executive Data
To measure the effect of decision-based executive incentive characteristics on stock repurchases, the executive characteristics are taken into account. A broader diversified board member characteristic has been found to help improve organizational performance (Siciliano, 1996). Gender diversification in Boards, has been proved in favour of social performance however less on the level of fund raisings
(Siciliano, 1996). Psychologists and scientists found that women are more long-term investing minded, which help them achieve more long-term success in the market (Loften, 2011). For this reason the female to board ratio (13), and the standard deviation and mean of the age of the board members are derived and used as control variables because they correlate to Board Member’s decision-making characteristics on stock repurchases.
This dataset includes data on all firm executives. First the ratio of female board members are calculated by deriving the quantity of female board members (𝐹𝐵𝑀!!) by the quantity of total board members (𝑇𝐹𝐵𝑀!!), see formula (13).
𝑃_𝐹𝐸𝑀𝐴𝐿𝐸!! = 𝐹𝐵𝑀!!
𝑇𝐵𝑀𝐿!! (13) Secondly, the spread and the mean of the age in the executive board is measured and created as new variables. Data are further cleaned: only CEO information remains, and other executives are taken out of the data set. A ticker to merge the executive dataset with theprevious two mentioned is created by combining the ticker and year and all three datasets are merged.
Dummy variables
To control for the recession period that occurred during the tested timeframe, the dummy variable 𝑅𝐸𝐶𝐸𝑆𝑆𝐼𝑂𝑁!! is created. A financial recession is defined by a period of 6 months, two consecutive quarters, or longer of decrease in Gross Domestic Product (GDP). The data collected quarterly and half yearly GDP information. When one year is characterized by one recession period, the year is interpreted as a recession year and the dummy variable is equal to one. Figure 1 visualized the period defined as 𝑅𝐸𝐶𝐸𝑆𝑆𝐼𝑂𝑁!!.
Figure 1 U.S. Gross Domestic Product and recession analysis
10,000 11,000 12,000 13,000 14,000 15,000 16,000 17,000 18,000 19,000 20,000 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 GDP ($bn) Year Recession US GDP ($bn)
As mentioned in the literature review, acquisitions are valued as indicators for stock repurchase motivations. It is for this reason that the variable 𝐴𝐶𝑄𝑈𝐼𝑆𝐼𝑇𝐼𝑂𝑁!! is taken into the regression as a control variable. The dollars expensed on acquisitions are non-quantifiable towards stock repurchases, therefore if an acquisition occurs in year y, dummy variable acquisition is equal to one.
Reinganum (1990) distinguished the illiquidity effect by the difference in the meaning of the trading volumes from NASDAQ versus the NYSE, due to the fact that trading on the NASDAQ is almost entirely done by market makers, and for the NYSE most trading happens directly between buying and selling investors. Therefor NASDAQ and NYSE cope with different microstructure effects on stock returns (Reinganum, 1990). It is for this reason that the effect of the firm’s traded Stock Exchanges on stock repurchase is tested by adding two dummy variables to the regression: NYSE=1 for the New York Stock exchange and AMEX=1 for AMEX and the third unexpressed dummy for NYSE=AMEX=0 for NASDAQ.
Dummies per two digits SIC codes are created, as shown in table 1. 3.2. Descriptive statistics
The descriptive statistics of the regression analysis variables are shown in table 3. The kurtosis test for normality is tested significant for all individual variables. Given the numbers of observations, all variables are significantly normal distributed. Stock repurchases is expressed as the firm specific number of stock repurchases to the total outstanding stock (NSR). Because the data set including very diverse firm characteristics, data are very spread, resulting in large standard deviations. The interest coverage ratio shows a large maximum, this consists for firms that have very low interest expenses to earnings. All regression variables are normally distributed; the dependent variable is not normally distributed, however also not required. The average number of observations is c.5.8 million.
4. Empirical methodology
The empirical methodology section elaborates on the hypothesis development and initially debates the hypotheses expectation outcome.
Liquidity affects volatility. As a response to an increasing volatility, investors demand a higher rate of return. In illiquid markets investors face higher transaction costs and lower return on securities that requires an offsetting liquidity discount. On aggregate the responsiveness of the investor’s demand increases the companies cost of capital, which further undermines real investments and economic growth. Illiquidity negatively affects stock price discovery, as a result of the decreasing trades, affecting investment portfolios errors, which ultimately also decrease company’s real investment choices.
In this research liquidity covers market liquidity and funding liquidity. The market liquidity reflects the ability to trade quickly at prices close to fundamental and the funding liquidity the ability to obtain Table 2: Descriptive statistics NYSE, AMEX, and NASDAQ
The descriptive statistics describe from left to right per column the number of observations (N), the mean, the median, the standard deviation (St. Dev.), the minimal value and the maximal value, and the 25 and 75 percentage tails of the observation. The reason that the dummy variables are included in this table is because the mean value of a year dummy will provide information on the ratio of observation to total observations, e.g. mean Y2000 equal to 0.05 mean that 5% of all observation occurred in 2000. (NSR100=NSR*100)
Statistic N Mean Median St. Dev. Min Max Pctl(25) Pctl(75)
NSR100 5,278,375 2.41 0.38 4.83 0.00 201.79 0.00 3.12 ILLIQ_md 5,787,699 0.0003 0.0003 0.0002 0.0001 0.002 0.0002 0.0004 ILLIQ_sd 5,787,699 0.0004 0.0003 0.0004 0.00 0.08 0.0002 0.0005 ILLIQ_fd 5,787,699 0.03 0.001 0.51 0.00 74.07 0.0002 0.004 SOLVENCY 5,088,925 0.65 0.34 1.60 -12.42 26.53 0.02 0.78 ICOVERAGE 4,749,028 63.25 6.60 306.20 -267.86 5,993.07 2.57 18.25 CURRENT 4,832,487 2.45 1.91 1.88 0.40 25.45 1.30 2.88 TOBINSQ 5,787,699 0.57 1 0.50 0 1 0 1 SIZE 5,784,929 7.75 7.64 1.71 4.07 12.52 6.49 8.89 DIVIDENDY 5,626,697 0.01 0.01 0.02 0.00 0.29 0.00 0.02 CASH 5,694,510 0.11 0.06 0.12 0.0000 0.95 0.02 0.15 CAPEX 5,568,709 0.05 0.03 0.05 0.00 0.74 0.01 0.06 PP&E 5,210,669 0.49 0.39 0.40 0.00 8.34 0.18 0.74 R&D 3,184,948 0.05 0.02 0.07 0.00 0.89 0.002 0.07 DEBT 5,089,177 0.20 0.17 0.20 0.00 3.68 0.02 0.31 P_FEMALE 5,787,699 0.08 0.00 0.12 0.00 1.00 0.00 0.17 RECESSION 5,787,699 0.15 0 0.36 0 1 0 0 ACQUISITION 5,787,699 0.47 0 0.50 0 1 0 1 Y2000 5,787,699 0.05 0 0.22 0 1 0 0 Y2001 5,787,699 0.05 0 0.22 0 1 0 0 Y2002 5,787,699 0.05 0 0.22 0 1 0 0 Y2003 5,787,699 0.06 0 0.23 0 1 0 0 Y2004 5,787,699 0.06 0 0.23 0 1 0 0 Y2005 5,787,699 0.06 0 0.24 0 1 0 0 Y2006 5,787,699 0.07 0 0.25 0 1 0 0 Y2007 5,787,699 0.08 0 0.26 0 1 0 0 Y2008 5,787,699 0.08 0 0.27 0 1 0 0 Y2009 5,787,699 0.08 0 0.26 0 1 0 0 Y2010 5,787,699 0.08 0 0.26 0 1 0 0 Y2011 5,787,699 0.07 0 0.26 0 1 0 0 Y2012 5,787,699 0.07 0 0.26 0 1 0 0 Y2013 5,787,699 0.07 0 0.26 0 1 0 0 Y2014 5,787,699 0.07 0 0.26 0 1 0 0 Y2015 5,787,699 0.01 0 0.09 0 1 0 0
funding on acceptable terms. The market liquidity as earlier brought to attention is expressed by the Amihud (2002) measure, shown in formula (3) to (6), that encounters the price and return effect. As a result of the investment’s behaviour responsiveness to increasing illiquidity it is expected that market illiquidity has a negative effect on the number of stock repurchases. The funding illiquidity effect on NSR is measured by including the SOLVENCY, ICOVERAGE, CURRENT, and DEBT control
variables. To test the effect on stock repurchases, the previous three variables are included in the regression to control for this effect. The effect of the investors demand, affecting the company’s real investments and costs of capital are determined by including the CASH, CAPEX, R&D, and PP&E as control variables. The recession period of 2008-2009 should be reflected by a significant increase in the market illiquidity, however the dummy variable RECESSION is included to control for this relation. The ACQUISITION variable test the theoretical motive described by Guthart (1967). The year dummies included control for time-trends.
H1: Market illiquidity has a negative effect on firm’s stock repurchases
a. Larger firm size will decrease the negative effect of market illiquidity on the number of stock repurchases
H2: Sector illiquidity has a negative effect on firm’s stock repurchases
a. Larger firm size will decrease the negative effect of sector illiquidity on the number of stock repurchases
H3: Stock illiquidity has a positive effect on firm’s stock repurchases
a. Larger firm size will increase the positive effect of stock illiquidity on the number of stock repurchases
H4: Debt is an intermediate variable H5: Cash is an intermediate variable
The first hypothesis tests the market illiquidity effect on stock repurchases, the second hypothesis tests the sector illiquidity effect on the latter, and the third hypothesis tests the firm specific stock illiquidity effect.
Hypothesis 3 explores the individual firm stock illiquidity relationship with stock repurchases. As earlier discussed, theoretical literature states that firms buy back stock to improve stock liquidity (Guthart, 1967). Hypothesis 3 tests whether the market has behaved in line with economic theory over the last 15 years.
Hypotheses 4 and 5 test whether cash and debt are considered intermediate variables, and are applied to fund stock repurchases.
5. Empirical results
The firm’s specific financial and executive data, based on annual data from the fiscal year-end reporting’s are regressed together with daily illiquidity measures by the use of a fixed effect panel regression with cluster standard errors by firms.
For all scenarios the errors are correlated with the regressors in the fixed effect model and the f-test is significant, justifying that all coefficients in the model are different from zero.
The ILLIQ measures are classified by market (_md), sector (_sd), or firm (_fd), based on daily (_id) data. The dependent variable NSR is multiplied by a hundred, such that the independent betas are easier to interpret (NSR=NSR100). The interaction_size_ii variables are classified by market (_md), sector (_sd), or stock (_fd) on daily inputs.
The sector dummies (Di) are excluded from the regressions due to omitted variable bias. The RECESSION dummy variable implicates the exact theoretical financial crisis years. However including both the year dummies and recession dummies cause for one-year dummy to be omitted due to the fact that RECESSION represents two years. It is decided that the year dummies provide the reader and the research with more interesting output when 2008 is the reference group, and initially control for time-trends– recession is dropped.
Subparagraph 5.1 elaborates on the regression results for testing hypothesis 1, the market illiquidity effect on stock repurchases; hypothesis 2, the sector illiquidity effect on the latter; and hypothesis 3, the stock illiquidity effect on NSR; Subparagraph 5.2 debates the result for hypotheses 4 and 5 that test the intermediation effect of debt and cash on illiquidity and stock repurchases.
5.1 Market, Sector and stock illiquidity effect on stock repurchases
The following section elaborates on the results testing the three illiquidity measures. First, the market illiquidity impact on stock repurchases is tested; see regression output (1), table 1. Secondly the sector illiquidity effect on stock repurchases is tested; see regression (2), table 1; and, thirdly the stock illiquidity effect on stock repurchases is tested to control whether firms repurchase stock to stimulate stock liquidity, see regression (3) and (4), table 1. Stock and market illiquidity are verified and significantly accepted. Market illiquidity, sector illiquidity and, stock illiquidity are correlated and interchangeably affected. Regression (5) to (7) show the results for the relation of the three illiquidity measures with stock repurchases. The results are discussed per topic on investment, solvency, substitution, Executive Board diversity, stock exchange, and aggregate illiquidity effect. The number of stock repurchases ratio (NSR) is multiplied by hundred for simplicity of the regression betas.
Investment
On aggregate level the undervaluation hypothesis is confirmed by the significant TOBINSQ variable. The results show in table 1 that the firms that are undervalued repurchase more stock. Results imply that when firms value their stock to be undervalued, they on aggregate level repurchase 0.6% more NSR. As earlier indicated, firms repurchase stock when considered undervalued, to increase it’s price, and to sell their stock again on the market when the market price is considered overvalued or accurately valued. This buy and sell strategy is treated as an investment strategy (Guthart, 1967), because it is profitable to buy low, and sell high. How the market alternatively responses to a firm’s reissuance of stock, dependents on the information transparency of the firm. In perfect market circumstances it is pledged that stock issuance does not affect stock prices, however due to the fact
Table 3: Summary statistics – Illiquidity effect on stock repurchases
Regression (1) test the effect of daily market illiquidity (ILLIQ_md) on the number of stock repurchase over total outstanding stock (NSR). Regression (2) test the effect of daily sector illiquidity (ILLIQ_sd) and regression (3) and (4) the daily stock illiquidity (ILLIQ_fd) on the latter. Regression (5) to (7) shows the result for the aggregate three illiquidity measures on NSR. NSR is expressed in percentages.
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
(1) (2) (3) (4) (5) (6) (7) VARIABLES NSR NSR NSR NSR NSR NSR NSR ILLIQ_md -242.2*** -351.9*** -194.3** -162.6*** (48.20) (103.9) (94.45) (39.85) Interaction_Size_mi -0.172*** -0.169*** -0.197*** -0.182*** (0.0297) (0.0479) (0.0439) (0.0326) ILLIQ_sd -24.63 100.6 35.02 (99.33) (82.05) (81.33) Interaction_Size_sd -0.141*** -0.00501 0.0496 (0.0536) (0.0647) (0.0538) ILLIQ_fd -0.0576** -0.0688 -0.0714 -0.0583** -0.0604** (0.0252) (0.0463) (0.0483) (0.0266) (0.0260) Interaction_Size_fd -0.0439*** -0.0260 -0.0203 -0.0377*** -0.0384*** (0.0196) (0.0199) (0.00532) (0.00530) SOLVENCY 0.000187 (0.000142) ICOVERAGE 0.000179 -0.0646 0.000201 (0.000140) (0.0592) (0.000141) CURRENT 0.0589 (0.0460) TOBINSQ 0.580*** 0.582*** 0.572*** 0.583*** 0.578*** 0.583*** 0.567*** (0.103) (0.104) (0.112) (0.103) (0.103) (0.115) (0.112) SIZE -0.132 -0.133 -0.167 -0.130 -0.128 -0.231 -0.164 (0.145) (0.145) (0.147) (0.144) (0.143) (0.150) (0.146) DEBT 3.350*** 3.367*** 3.588*** 3.373*** 3.348*** 3.716*** 3.562*** (1.253) (1.253) (1.369) (1.254) (1.253) (1.429) (1.368) CAPEX 2.826** 2.869** 3.958** 2.890** 2.844** 4.054** 3.939** (1.412) (1.414) (1.634) (1.418) (1.413) (1.649) (1.626) PP&E 1.381*** 1.366*** 1.100** 1.347*** 1.388*** 1.132** 1.129** (0.508) (0.508) (0.474) (0.505) (0.508) (0.471) (0.474) P_FEMALE 1.003* 1.026* 1.045* 1.032* 0.999* 1.004* 1.020* (0.534) (0.535) (0.566) (0.536) (0.534) (0.569) (0.563) Y2000 -2.071*** -2.081*** -1.940*** -2.071*** -2.076*** -2.016*** -1.933*** (0.292) (0.292) (0.301) (0.292) (0.292) (0.303) (0.301) Y2001 -3.496*** -3.504*** -3.360*** -3.500*** -3.495*** -3.432*** -3.350*** (0.234) (0.236) (0.254) (0.234) (0.234) (0.252) (0.254) Y2002 -3.166*** -3.176*** -3.010*** -3.170*** -3.159*** -3.066*** -3.001*** (0.241) (0.243) (0.263) (0.241) (0.241) (0.264) (0.262) Y2003 -3.315*** -3.316*** -3.137*** -3.323*** -3.311*** -3.181*** -3.128*** (0.227) (0.228) (0.245) (0.227) (0.227) (0.242) (0.245) Y2004 -2.742*** -2.734*** -2.589*** -2.754*** -2.739*** -2.637*** -2.581*** (0.222) (0.222) (0.239) (0.222) (0.222) (0.240) (0.239) Y2005 -1.855*** -1.838*** -1.640*** -1.861*** -1.853*** -1.699*** -1.640*** (0.246) (0.245) (0.267) (0.245) (0.246) (0.268) (0.267) Y2006 -1.277*** -1.254*** -1.092*** -1.275*** -1.276*** -1.160*** -1.098*** (0.304) (0.304) (0.339) (0.303) (0.304) (0.345) (0.339) Y2007 -0.178 -0.153 -0.00124 -0.174 -0.177 -0.0892 -0.0101 (0.269) (0.270) (0.293) (0.269) (0.269) (0.289) (0.293) Y2009 -3.182*** -3.181*** -3.084*** -3.181*** -3.179*** -3.107*** -3.085*** (0.214) (0.214) (0.233) (0.214) (0.214) (0.237) (0.233) Y2010 -2.308*** -2.282*** -2.129*** -2.297*** -2.307*** -2.148*** -2.145*** (0.225) (0.225) (0.248) (0.225) (0.225) (0.254) (0.248) Y2011 -1.144*** -1.117*** -0.946*** -1.130*** -1.143*** -0.967*** -0.965*** (0.230) (0.229) (0.252) (0.230) (0.230) (0.257) (0.252) Y2012 -1.736*** -1.698*** -1.554*** -1.711*** -1.735*** -1.579*** -1.586*** (0.245) (0.244) (0.269) (0.245) (0.245) (0.274) (0.269) Y2013 -2.180*** -2.127*** -1.988*** -2.138*** -2.180*** -2.060*** -2.040*** (0.238) (0.237) (0.257) (0.238) (0.238) (0.261) (0.257) Y2014 -1.649*** -1.587*** -1.499*** -1.594*** -1.649*** -1.548*** -1.565*** (0.249) (0.248) (0.266) (0.248) (0.248) (0.267) (0.267) Y2015 -1.206*** -1.138*** -1.144*** -1.139*** -1.204*** -1.216*** -1.212*** (0.377) (0.376) (0.391) (0.377) (0.377) (0.392) (0.391) Constant 3.443*** 3.356*** 3.405*** 3.352*** 3.408*** 3.836*** 3.441*** (1.200) (1.214) (1.224) (1.193) (1.188) (1.247) (1.211) Observations 4,223,194 4,223,194 3,704,176 4,223,194 4,223,194 3,622,144 3,704,176 R-squared 0.066 0.065 0.061 0.064 0.066 0.063 0.062 Number of firms 1,739 1,739 1,625 1,739 1,739 1,600 1,625
that no market is perfect reissuing stock to the market, when prices are accurately priced can signal that prices are considered to be overvalued causing prices to drop. Therefor reissuing stock when prices are accurately valued is done by exception or last resort when the company is unable to receive more debt financing. Repurchasing stock improves stock liquidity, and improves stockholder’s gain by the increased investment ratios, e.g. ROE and ROA, because of the decreasing number of shares outstanding. Under imperfect market circumstances stock repurchases can stimulate the market stock price. Encapsulating, the undervaluation hypothesis is confirmed, meaning that undervalued stock encourages firms to repurchase more stock. Note however that the variable TOBINSQ is a dummy variable that the relation does not imply the level of undervaluation, just that the stock is undervalued. According to Guthart (1967) firms repurchase stock during M&A activities. It is for this reason that the expectation is set on a positive ACQUISITION beta. However the relation with NSR is not significantly confirmed. The fact that acquisitions do not increase the number of stock repurchases in this research is not a surprise. Merger waves are namely linked to over- or under valuation of the market (Rhodes-Kropf and Viswanathan, 2004). Overvalued firms gladly engage in acquisitions, especially engaging the deal via a stock for stock exchange. The undervalued firms are either the target firm, or the ones that do not execute acquisitions, on aggregate level. Firms that are undervalued that however decide to participate in a stock-for-stock exchange M&A deal might repurchase stocks as an attempt to manipulate and increase the price. This behaviour is however captured by the TOBINSQ variable- repurchasing undervalued stock. Initially, when stocks are overvalued companies are excited to participate in stock-for-stock exchange M&A deals. However, the company does not want to fund capital to repurchasing its own stock. That does not mean that the company will not participate in M&A activities, it will however tender its shareholders to participate in the deal. Bearing this in mind, the insignificancy of this variable is not surprising. For this sake the insignificancy does not fully reject Guthart’s (1967) theory. Further in-depth research on this topic is required to either reject or verify this relationship.
The effect of the investors demand on liquidity, affecting the company’s real investments and costs of capital are determined by including the CASH, CAPEX, R&D, and PP&E as control variables. PP&E and CAPEX is a measure of capital (fixed-assets) intensity. Both variables differ significantly across industries and are related to earning timelines and volatility (Beaver & Ryan, 1993). Both variables are highly correlated since property, plant, and equipment is part of the capital expenditures of a firm. The main difference is however that capital expenditure includes the depreciation of previous property, plant and equipment expenses. Both variables are part of total assets therefor determine firm’s size. It is for this reason that the two variables are included in the regression, and that the firm’s size does not show a significant effect on stock repurchases. The reason why capital expenditures, including property, plant, and equipment affect NSR, is because illiquidity affect investment behaviour. The beta relation implies that when investment behaviour is increasing, firms also increase their repurchasing activities. The fact that property, plant, and equipment, and capital
expenditures positively effect stock repurchases can be explained by the undervaluation hypothesis, funding hypothesis, or correlation to omitted variables by industry characteristics. When firms invest in longer-term assets, the firm knows its growth potential. The scenario can occur that the market is not aware of the investment potential, therefor not reflected by the market stock price: the undervaluation hypothesis. The industry characteristics are linked to the fact that firm’s with relative larger PP&E and CAPEX are firms that require more fixes assets, e.g. manufacturing, utility exploration and production, or telecommunication firms. It can be the case that these firms on aggregate are more active in stock repurchasing strategies, also for omitted reasons. On aggregate, these variables indicate longer-term investment opportunities and behaviour for firms. Firms that are characterized by a higher capital expenditure level including property, plant and equipment level relative to total assets on aggregate repurchase more stock (beta of c.3% and 1.4%).
Another investment indicator is the research and development expense. R&D increases a firm’s expected intangible assets. Research and development costs for pharmaceutical or technology firms are long-term investments, however not always treated under the capital expenditures but under operating expenses in US GAAP. It is for this reason that the research and development costs are included in the regression as a control variable. The fact that this variable is not considered a true estimator for stock repurchases confirms that the recurring expenses -the operating expenses- do not effect decisions on stock repurchases. Recurring expenses are very transparent expenses and do not harm any pricing disbelief for outstanding stock. If the research and development is however long-term, and accounted under capital expenditures, the relation is given by the positive CAPEX beta.
Cash is directly affected when consumer behaviour changes (demand) or the supply for firm’s products or services changes. Cash or debt is used for investments. If cash to total assets decline the ability to pay liabilities (all payment obligations) declines. It is expected that CASH positively affect investments, therefor stock repurchases. This research however finds contradicting results, namely no significant relation to NSR.
Solvency
Solvency determines whether the company is able to meet its short-term liabilities (CURRENT), meet its debt obligations (ICOVERAGE), or represents how growth is financed (SOLVENCY).
The larger ICOVERAGE, the better a company is able to pay its interest expenses and the less risky the company is considered. In addition, a relative high ICOVERAGE improves the excess reserves or cash available. It is expected that the interest coverage ratio would represent the relation to excess earnings, and firm riskiness, therefore positively correlated to stock repurchases. This effect is however not significantly verified. Interestingly, the stock illiquidity measure is only significant if ICOVERAGE is included in the regression, see regression (3) and (4). Apparently stock illiquidity is dependent on the liabilities the firm has, thereby correlated to volatility. Excluding the volatility by excluding the interest coverage ratio that cause volatility fluctuations, result that the stock illiquidity
