Earnings Pressure and Long-Term Corporate

Downloaded from informs.org by [194.221.86.200] on 29 March 2016, at 05:31 . For personal use only, all rights reserved.

Organization Science

Articles in Advance, pp. 1–19 ISSN 1047-7039 (print) ó ISSN 1526-5455 (online)

http://dx.doi.org/10.1287/orsc.2016.1056 © 2016 INFORMS

Earnings Pressure and Long-Term Corporate Governance: Can Long-Term-Oriented Investors and Managers Reduce the Quarterly Earnings Obsession? Yu Zhang China Europe International Business School, Shanghai 201206, China, [email protected]

Javier Gimeno INSEAD, 77305 Fontainebleau Cedex, France, [email protected]

R

ecent research has shown that managers in publicly traded companies facing earnings pressure—the pressure to meet or beat securities analysts’ earnings forecasts—may make business decisions to improve short-term earnings. Analysts’ forward-looking performance forecasts can serve as powerful motivation for managers, but may also encourage them to undertake short-term actions detrimental to future competitiveness and performance. To identify whether managerial reactions to earnings pressure suggest evidence of intertemporal trade-offs, we explored how companies respond to earnings pressure under different conditions of corporate governance that shape the temporal orientations of managers. Using data on competitive decisions made by U.S. airlines under quarterly earnings pressure, we examined the effect of earnings pressure on competitive behavior under different ownership structures (ownership by long-term dedicated investors versus transient investors) and CEO incentives (unvested incentives that are restricted or unexercisable in the short term, versus vested incentives). The results suggest that companies with more long-term-oriented investors and long-term-aligned CEOs with unvested incentives are less likely to soften competitive behavior in response to earnings pressure, relative to companies with transient investors and CEOs with vested, immediately exercisable stock-based incentives. Using a difference-in-differences (DiD) specification for stronger identification, we also found that firms respond to their rivals’ earnings pressure shocks by increasing capacity and prices, particularly when those rivals do not have long-term-oriented investors and CEO incentives. The evidence is more aligned with the view that the pursuit of short-term earnings as a result of earnings pressure may be detrimental to long-term competitiveness. Keywords: earnings pressure; corporate governance; strategic behavior; airline industry History: Published online in Articles in Advance.

Introduction

2005; Roychowdhury 2006; Cohen et al. 2008, 2010), marketing (Chapman and Steenburgh 2011), and strategy (Zhang and Gimeno 2010, Benner and Ranganathan 2012, Gentry and Shen 2013) have also found that managers change real business decisions to improve short-term earnings and meet analysts’ forecasts (Porter 1992). To draw more prescriptive implications from this research, it is important to understand whether responses associated with earnings pressure, and directed to improve short-term earnings, are aligned or misaligned with the long-term interests of the company. Clearly, additional pressure to achieve short-term performance targets can have strong motivational power on managers and organizations, as both agency theory and behavioral theory of the firm have argued (Jensen and Meckling 1976, Chen 2008). Therefore, if the actions associated with earnings pressure represent a stronger performance focus by management, with increased effort and productivity, then earnings pressure would provide a healthy discipline for managers to perform more effectively (Jensen 1986). In contrast, much of the research in accounting (Levitt 1998) and strategy (Porter 1992, Laverty 1996) has suggested that

The impact of securities analysts on managerial behavior has attracted the attention of both practitioners and academics in recent decades (Levitt 1998; Davis and Useem 2002; Davis 2005; Benner 2007, 2010; Benner and Ranganathan 2012; Litov et al. 2012). Acting as information intermediaries in the stock market, securities analysts’ buy/sell recommendations and earnings forecasts can influence investors’ expectations of a firm’s performance and consequently affect the trading and pricing of its stock (Womack 1996, Skinner and Sloan 2002) as well as its managers’ compensation and job security (Puffer and Weintrop 1991, Farrell and Whidbee 2003, Wiersema and Zhang 2011). Confronted with such potential consequences, managers try by various means to meet analysts’ forecasts, such as offering lower earnings guidance (Bernhardt and Campello 2007), managing earnings numbers (Degeorge et al. 1999), and using impression management and influence tactics (Westphal and Clement 2008, Westphal and Graebner 2010). More recently, research in multiple disciplines, including finance (Graham et al. 2005), accounting (Bhojraj and Libby 1


2

Zhang and Gimeno: Earnings Pressure and Long-Term Corporate Governance

managers’ short-term-oriented responses to earnings pressure are detrimental to the future of their firms, because earnings pressure encourages managers to engage in actions that increase short-term earnings at the cost of future competitiveness and performance (intertemporal trade-offs). Our study examines the impact of earnings pressure on competitive behavior, following on previous research on the impact of financial constraints and capital market pressures on competitive aggressiveness. Phillips (1995) showed that high levels of debt—which force firms to focus on short-term cash flow generation to cover the debt—reduced the intensity of competition (lower output, higher prices) in three of the four industries studied, whereas Chevalier (1995a, b) found that supermarkets undertaking leveraged buyouts (LBOs) in the 1990s also tended to soften competitive behavior, potentially undermining their market position vis-à-vis rivals. Focusing more on the pressures to meet investment analyst’s earnings expectations, Zhang and Gimeno (2010) found that U.S. electricity companies experiencing earnings pressure tended to restrict output in markets where they had market power, although this resulted in expansion by competitors. However, as both Chevalier (1995b, pp. 1098–1099) and Zhang and Gimeno (2010, pp. 761–762) pointed out in detail, these observations were not conclusive about whether the observed softening of competitive behavior represented a departure from agency-inspired “empire building” toward competitive choices more aligned with long-term shareholder value, or instead represented competitive choices that emphasized short-term cash flows or earnings at the detriment of future performance and competitiveness. Both alternative explanations are consistent with the existing evidence that high leverage and earnings pressure have a softening effect on competitive behavior. The objective of this paper is to leverage differences in corporate governance conditions that affect temporal orientation to better identify whether the effect of earnings pressure on softening of competitive behavior is aligned or not with long-term performance concerns. If responses to earnings pressure represent trade-offs between current and future performance, we would expect that companies with more long-term-oriented corporate governance would be more resistant to such pressure, whereas those with a shorter time orientation would be more responsive. On the other hand, if these responses intend to improve short-term performance without imposing intertemporal trade-offs, then we would expect similar responses among firms regardless of their temporal orientation. We propose that two specific dimensions of corporate governance may affect the temporal orientation of management and therefore, the willingness of managers to respond to earnings pressure: ownership structure, which influences the time horizon of investors, and managerial incentives,

Organization Science, Articles in Advance, pp. 1–19, © 2016 INFORMS

which determine the alignment of managers’ rewards with short-term and long-term performance. We examine our hypotheses using the competitive decisions (ticket price, flight frequency, flight cancellation) made by the U.S. domestic airlines industry from 1994 to 2000. Compare to the previous work by Zhang and Gimeno (2010) on earnings pressure and competitive behavior, this paper explores a novel industry with different competitive characteristics (price competition rather than output competition), and different competitive decisions (pricing, capacity allocation, service quality). We use three different empirical strategies to test our hypotheses. First, we use a dynamic panel data regression approach to understand the effect that a firm’s earnings pressure has on its competitive behavior in different markets. Second, because a firm’s earnings pressure and competitive behavior may be spuriously influenced by unobserved heterogeneity, we use a difference-in-differences (DiD) identification approach to examine whether and how firms respond to earnings pressure shocks experienced by their rivals in duopoly markets. This cross-rival approach leverages the fact that the same firm meets different rivals in different markets. Finally, we also assess the impact of earnings pressures on market-level competitive outcomes. Our paper is the first to examine the critical contingency of corporate governance conditions on the effect of earnings pressure from securities analysts on managerial decisions and competitive behavior. We show that the effect of earnings pressure on softening competitive behavior differs according to factors of corporate governance such as ownership structure and CEO incentives. Specifically, firms with long-term investors and unvested, long-term CEO incentives were less likely to soften competitive behavior when facing earnings pressure. The evidence is consistent with the view that earnings pressure encourages myopic managerial behavior, which was speculated but not tested in prior research (e.g., Chevalier 1995a, b; Zhang and Gimeno 2010). In addition, the paper also bridges two previously separated streams of literature about the influences of securities analysts and corporate governance on managerial decisions and contributes to each (Litov et al. 2012; Benner 2007, 2010; Connelly et al. 2010).

Managerial Responses to Earnings Pressure

Given the high stakes associated with missing earnings forecasts, managers may use different approaches to meet or beat them. However, given that managers are generally motivated to improve the business performance of a going concern, it is unlikely that traditional ways to improve ongoing performance (e.g., greater efficiency) will suffice to increase short-term earnings above the trend. On the other hand, if earnings management represents an intertemporal trade-off between current and future performance, the type of business decisions most likely


Zhang and Gimeno: Earnings Pressure and Long-Term Corporate Governance Organization Science, Articles in Advance, pp. 1–19, © 2016 INFORMS

to be affected would be those that (a) can generate a substantive and quick impact on earnings and (b) are not easily detected by analysts and investors. Consequently, delaying capital investment projects would have limited impact on current earnings, as those investments are capitalized in the balance sheet and amortized over time. Similarly, meeting earnings expectations by cutting R&D or advertising expenses, though possible, would likely be counter effective if easily detected by analysts and investors (Gentry and Shen 2013). One broad category of business decisions that involve intertemporal trade-offs that are not easily detected by investors involves investments in gaining current competitive position (e.g., market share) that could provide future benefits. Strategy research and consulting practice have argued that a strong competitive position in a market can have dynamic benefits for a firm through capability (lower future costs), demand (higher demand and revenues), or competition (lower future competition) trade-offs (Stern and Deimler 2006, Ghemawat 1991). One way to frame these effects is to consider two-stage models of intertemporal competition, such as models of learning-by-doing (Spence 1981, Fudenberg and Tirole 1984) or switching costs and brand loyalty (Klemperer 1987). In these types of models, a strong competitive position can lower future costs through scale and experience effects (Lieberman 1987, Ghemawat 1984). It may increase future demand and revenues through informational or reputational benefits of installed customer base, network externalities, or the lock-in effect of customer loyalty and switching costs (Klemperer 1987, Foster et al. 2016). Similarly, a strong competitive position may reduce future competition by deterring potential entrants and preempting expansion by existing competitors (Fudenberg and Tirole 1984, Ghemawat 1984, Goolsbee and Syverson 2008).1 These mechanisms would encourage long-termoriented competitors to be more aggressive than they would be if they were to maximize current earnings in a static way without regard to future consequences. An analytical model of our argument can be found in Online Appendix A (available as supplemental material at http://dx.doi.org/10.1287/orsc.2016.1056).2 On the other hand, companies could choose to engage in lower levels of competitive aggressiveness than that which maximizes the firms’ long-term performance (i.e., higher prices, lower quality of service)3 and enjoy higher short-term profits in the market, even though this could probably lead to a diminished market position as potential entrants and competitors may eventually take advantage of the reduced competitive conditions to expand their own presence. For example, in airline industry, the empirical context of our paper, more aggressive competitive behavior may take forms of lower ticket price, or higher flight frequency. Price competition is the most salient form of competitive behavior in the airline industry and has a strong effect on earnings. As profit margins are generally very low in the

3

industry (2%–5% of sales, on average), even a small price increase can affect earnings disproportionately. In contrast, a lower price could make a flight more attractive relative to competitors’ offerings and help to build market share (Lederman 2007). Similarly, flight frequency affects not only the capacity that an airline commits to a market, but also the scheduling convenience offered to customers. Airlines that dominate frequencies in a given market commonly carry a disproportionally high share of its traffic—an effect known as the “S-curve” (Holloway 2008). Such an effect may lead airlines to “overschedule” for strategic reasons, such as to deter entry of rivals. At the same time, maintaining a large number of frequencies in a market is costly, since airlines rarely operate at full occupancy and markets may be efficiently served with fewer, larger planes. In a model of dynamic competition like the one outlined previously, earnings pressure is likely to shape how firms weigh the pros and cons of achieving current earnings versus improving competitive position. Although most firms are likely to be motivated by both current earnings and competitive position, the relative emphasis on current earnings would likely increase in the face of earnings pressure. As a consequence, firms would be less motivated to engage in competitive aggressiveness for the sake of competition position if such aggressiveness sacrificed current earnings (e.g., Zhang and Gimeno 2010).4 Hypotheses The main goal of our paper is to examine whether the association between earnings pressure and competitive aggressiveness is contingent on the long-term or shortterm orientation embedded in the context of corporate governance. Specifically, we focused on two dimensions of corporate governance that should affect a firm’s temporal orientation: ownership structure, particularly the temporal orientation of the institutional investors in the firm, and CEO incentives, particularly the temporal vesting of CEO stock-based incentives. Ownership Structure. Managers’ concerns about analysts’ earnings forecasts and short-term financial performance may vary under different company ownership structures. Companies have different types of investors who differ significantly in terms of their investment time horizons and their influence on firms’ stock prices (Wright et al. 1996, Hoskisson et al. 2002). Among the typical investors in publicly traded firms, trading by large institutional investors has the most impact (Coyne and Witter 2002). Institutional investors are financial institutions, such as mutual funds, hedge funds, pension funds, banks, insurance companies, foundations, and endowments that hold significant portions of equity in publicly traded companies. During the 1990s, their aggregate holdings of U.S. equities increased to 60% (Securities Industry Association 2002). As a result, their trading can cause


4


significant fluctuations in stock prices (Brown and Brooke 1993) and consequently influence managers’ motivation to meet analysts’ earnings forecasts. Although some researchers argue that institutional investors in the aggregate generally focus more on firms’ short-term performance (Lang and McNichols 1997), the evidence suggests that institutional investors differ dramatically in their investment styles and temporal preferences. In particular, Porter (1992) argued that there are “dedicated” and “transient” institutional investors that have very different investment strategies and approaches. If earnings pressure implies intertemporal trade-offs, then we expect that ownership by these different institutional investors will be associated with different managerial reactions to earnings pressure. Dedicated institutional investors are institutions that have large, long-term holdings that are concentrated in a few firms (Bushee 1998). They are less constrained by liquidity needs and can use longer periods to evaluate investment managers’ performance. This allows investment managers to hold fewer stocks and to devote more effort to understanding each company’s business fundamentals. As a result, they are less sensitive to short-term earnings surprises. In contrast, transient institutional investors are institutions that take small equity positions in many firms and tend to trade frequently (Bushee 1998). Fund managers in these institutions typically hold a diversified portfolio of stocks (for liquidity needs) and rapidly change their investment decisions (to meet short-term performance expectations of the fund’s performance). As a result, they are sensitive to companies’ short-term earnings results and react dramatically to negative short-term earnings surprises.5 Although both types of institutional investors care about firm performance and may both exit if companies underperform, their time horizon of performance focus is different. That is, dedicated institutional investors are less focused on the short-term performance of the company. Hence, when there is intertemporal trade-off and the two types of competitive action are discrepant, managers in publicly traded companies in which dedicated institutional investors are prominent should be less concerned about missing analysts’ earnings forecasts. Therefore, if managerial reactions involve intertemporal trade-offs in performance, we would expect managers to respond less strongly when their company is owned more by dedicated institutional investors than when it is owned by transient institutional investors. Hypothesis 1 (H1). The effect of earnings pressure on the aggressiveness of competitive behavior will be less negative for firms under higher dedicated institutional ownership than for firms under higher transient institutional ownership.


CEO Incentives. Managerial incentives are viewed as important means to motivate managers to work in shareholders’ interests (Finkelstein and Hambrick 1988, Jensen and Murphy 1990). These incentives are especially important to the CEO, because the CEO has the largest impact on performance management within the company and sets objectives for the rest of the top management team and overall organization (Beatty and Zajac 1987, Zajac 1990). Various components of CEO incentives are designed to link CEO compensation (and efforts) to either accounting results (through bonuses) or stock price appreciation (through stock options and stock awards). In particular, stock-based incentive schemes, such as restricted stocks and stock options, are viewed as high-powered incentives that can help align the CEO’s interests with improvement of a company’s shareholder value because it makes the payoff function for the CEO more “convex” or risk-seeking (Murphy 1999), and more closely related to the company’s stock price performance and to the shareholders’ wealth (Gomez-Mejia and Wiseman 1997). On the other hand, researchers have also argued that stock options and restricted shares can make CEOs more sensitive to changes in stock prices because the changes will influence the value of their stock options and shares granted or owned (Hall and Murphy 2003). As a result, stock-based incentives may paradoxically give CEOs strong incentives to be risk averse and avoid decisions that may negatively affect share price in the short term, rather than focus on taking necessary risk and competitive actions to increase long-term shareholder value. For example, prior studies have found that CEOs who are given more stock-based incentives are more likely to use discretionary accounting accruals to manipulate reported earnings (Bergstresser and Philippon 2006) or even misreport earnings results (Harris and Bromiley 2007). Therefore, existing theories are inconclusive about the overall impact of stock-based incentives on managerial responses to earnings pressure. The constraints associated with the vesting of stockbased incentives (Devers et al. 2008, Souder and Shaver 2010, Souder and Bromiley 2012) may serve as a more effective dimension to identify the impact of stock-based incentives on responses to earnings pressure. Typical stock-based incentives involve a combination of stock ownership and awards of stock options. Companies usually will impose a period of time before CEOs can exercise, or vest, their stock-based incentives. These periods generally range from one to seven years. Four years is most typical, although sizable variations occur (Yanadori and Marler 2006). Over time, part of the awarded stock ownership and stock options become vested, whereas the rest and the newly awarded remain unvested. Unvested stock-based incentives may limit the ability of CEOs to liquidate or diversify their holdings as ways to benefit from short-term stock price appreciation. Because unvested stock-based incentives cannot be exercised or



cashed out in the near term (three years, five years, or even later), they will reduce the impact of shortterm performance and resulted stock price variations on managers’ decision making. Therefore, CEOs will benefit less from short-term maintenance or improvement of stock prices if they hold more unvested stock-based incentives. In this scenario CEO incentives and decisions will be driven more by the long-term value of the company. In contrast, CEOs with vested stock-based incentives may indeed have an incentive to diversify their risk after their incentives become vested (Bebchuk and Fried 2004). As a consequence, CEOs may differ in their policies for exercising vested incentives, ranging from selling immediately at vesting, structuring a schedule for selling over time, or keeping vested options and shares in their portfolio. It is likely that a CEO’s choice of retaining or selling vested stock-based incentives may be driven by his or her expectations of future performance or by a need to signal confidence in the future of the company (Malmendier and Tate 2005). Therefore, CEOs with vested incentives may not necessarily be myopic, given their choice not to exercise the incentives. However, if the size of the unvested stock-based incentive portfolio outsizes the vested portfolio, these CEOs will be less concerned by short-term stock price declines that could damage the value of their stock-based incentive holdings. As a result, earning pressure will affect less when CEOs hold more unvested stock-based incentives. Hypothesis 2 (H2). The effect of earnings pressure on the aggressiveness of competitive behavior will be less negative when CEOs have more unvested stock-based incentives (unexercisable stock options and restricted stocks) than when CEOs have more vested incentives (exercisable stock options and common stocks).

Methods Data and Sample We tested our theory and hypotheses in the context of the U.S. domestic airline industry from 1994 to 2000. This empirical setting had several advantages. First, a large proportion of the airlines were publicly traded and differed substantially in terms of ownership structure, managerial incentive schemes, and earnings pressure from the stock market. For example, U.S. Airways had 33% and 58% institutional ownership in the first quarter of 1994 and 1999 respectively, whereas AMR Corp. (parent company of American Airlines) had 84% and 69% institutional ownership in the first quarter of 1994 and 1999. In terms of executive compensation, American Airlines paid $8.7 million to its CEO in 1997, whereas Southwest paid $0.65 million the same year. Second, there was some anecdotal evidence that competitive behaviors in the industry might be influenced by a desire to meet analysts’ earnings forecasts. For instance, the New York Times attributed

5

successive price increases by major airlines in 1999 to pressure from Wall Street to maintain record profitability (Zuckerman 1999). More recently, Virgin American and Jet Blue were also pressured from Wall Street to increase ticket prices and baggage fees. For example, Jet Blue CEO Dave Barger was dismissed after analysts pressured the company to charge bag fees to increase earnings and share price, an action that the CEO opposed because of concerns for losing competitive position (Mutzabaugh 2014). Third, the U.S Department of Transportation (DOT) collects exceptionally rich information on competitive structures and behaviors that is both comprehensive and detailed. Such data includes a 10% sample of all tickets sold in the United States (the Airline Origin and Destination (O&D) survey), as well as comprehensive statistics at the market and firm levels. These data sources have been used widely in previous research on competitive behavior at the firm-market level (defined as an airline within a certain city-pair market) (Evans and Kessides 1994, Gimeno and Woo 1996, Prince and Simon 2009). Airline data was obtained from the DOT statistics and combined from three interrelated databases: the Airline O&D Survey (DB1B), the Air Carrier Statistics (Form 41 Traffic’s T-100 Segment), and the Air Carrier Financial Reports (Form 41 Financial Data). These sources allow analysis at the quarterly level, therefore matching the timing of earnings reports. The DB1B database was used to construct measures of competitive aggressiveness for firms in a market and a quarter. We used the T-100 Segment database to measure flight frequencies and market structure at the firm-market-quarter level, and the Form 41 Financial Data to create financial status variables at the firm-quarter level. Following previous studies on airline industry competition, we defined a market as a city-pair route between any given pair of cities. The 1994 starting date was selected as a result of data availability, and the 2000 ending date excluded the dramatic events of September 11, 2001. The period of the study represented approximately a full business cycle. We selected a sample of relevant markets that met the following conditions: (a) the origin and destination cities were at least small hubs according to the Federal Aviation Administration’s definition (i.e., carrying at least 0.05% of total U.S. traffic); (b) the origin and destination cities were at least 100 miles apart (to eliminate substitution from ground transportation); and (c) the markets had average daily traffic of at least 10 passengers. In those markets, we identified airlines as incumbents if they had at least 5% market share or carried at least 10 passengers a day. We identified companies as potential entrants if they were active at both cities of the city-pair market but did not qualify as incumbents. Because we were interested in competitive interactions, we eliminated cases of monopoly markets (i.e., markets with only one incumbent). There were 336,258 firm-market-period


6


observations satisfying those criteria, representing over 96% of the tickets available from the O&D survey. We obtained earnings forecasts and historical earnings of the companies from the Institutional Broker Estimates System (I/B/E/S) database provided by Thomson Financial/First Call. The I/B/E/S database has been widely used in finance and accounting for research on the causes and effects of earnings management. We then combined the I/B/E/S earnings data and DOT airline data by matching the airlines to their respective I/B/E/S tickers, based on industry and company histories. The DOT airline data are reported at the level of an “entity” (an operational unit flying under a single code). In the few cases in which publicly traded corporations owned multiple entities (e.g., American and American Eagle owned by AMR Corp.), we performed the analysis at the entity level but used relevant corporate-level earnings information.6 Although we used all entities to calculate market structure variables, the final sample includes only observations of entities of publicly traded corporations, because only those companies had analysts’ earnings forecasts. We obtained institutional investor ownership data from the Thomson-Reuters Institutional Holdings (13F) database, and we used categories of dedicated and transient institutional investors from the website of Brian Bushee (1981–2010), who operationalized Porter’s (1992) institutional investor typology.7 We compiled executive compensation data from the Compustat Executive Compensation database. Those data were also matched to the DOT airline data through the link established between I/B/E/S and the DOT database. Dependent and Independent Variables The dependent construct is the aggressiveness of competitive behavior, which we measured by two variables, ticket price and flight frequency by an airline in a market.8 We measured the price of an airline’s tickets in a market by its yield level, defined as the average ticket price that an airline charges in a city-pair route, divided by the route’s distance in miles. Yield is a commonly used measure of price or revenue generation in the airline industry, particularly because it allows more meaningful comparison of prices of routes of different length (Gimeno and Woo 1996).9 We measured the frequency of an airline’s flights in a market by the average number of flights per day served by the airline in a city-pair route. If a route comprised two segments or parts of a flight, we calculated frequency as the minimum daily flights of both segments. The independent variables measure the intensity of earnings pressure that a firm experiences for a particular quarter, and the two corporate governance dimensions: institutional investor ownership composition and CEO stock-based incentives. Earnings Pressure. We used Zhang and Gimeno’s (2010) measure of earnings pressure, calculated as the


difference between the consensus of analysts’ earnings forecasts and firms’ potential earnings evaluated at the beginning of quarter. This ex-ante measure is determined before any competitive actions taken in the course of the quarter, and therefore is not affected by reverse causality. Moreover, in contrast with other measures of earnings pressure based on final reported earnings or analyst consensus prior to earnings announcement, this measure is not affected by any form of earnings management during the quarter, such as earnings guidance or accounting accruals management. Therefore, the measure avoids some forms of endogeneity that may affect previous measures of earnings pressure. Specifically, we calculated earnings pressure for firm i at the beginning of quarter t as: Earnings pressureit = analysts0 consensus forecast it É potential earningsit 0

Analysts’ Consensus Forecast. This was calculated as the average of the last round of analysts’ earnings forecasts prior to the start of the quarter (Kasznik and McNichols 2002). For example, to measure the earnings pressure affecting a firm during 1999Q1, we collected the latest earnings forecasts for 1999Q1 from different analysts released prior to January 1, 1999, and average among them. Following Zhang and Gimeno (2010), we used the method developed by Matsumoto (2002) to estimate potential earnings, or the earnings that a firm would realize in the coming quarter. The method estimates a firm’s potential performance by using both historical earnings trends and the information contained in stock price returns before the analysts’ forecasts. This method is well accepted in the accounting field (e.g., Burgstahler and Eames 2006) and was recently adopted in management research (Zhang and Gimeno 2010). We adapted this method to estimate potential earnings in the context of the airline industry and standardized this measure by the firm’s stock price at the beginning of the quarter (Skinner and Sloan 2002).10 Given the relative novelty of the earnings pressure measure, we evaluated its validity in the two aspects suggested by Zhang and Gimeno (2010) that we reported in Online Appendix B of this paper. Institutional Investor Ownership. We used the categories of dedicated and transient institutional investors provided by Brian Bushee’s website and matched those categories with the institutional investor ownership data obtained from the Thomson-Reuters Institutional Holdings (13F) database. Bushee (1998) used both factor analysis and cluster analysis to classify institutional investors according to their investment behavior and time horizon in terms of such characteristics as sensitivity to current earnings, portfolio turnover, and portfolio concentration. These classifications have been used in both the accounting (Bushee and Noe 2000, Ke and Petroni 2004) and



7


management literature (Higgins and Gulati 2006, Connelly et al. 2010). The procedure results in three categories of institutional investors: dedicated, transient, and quasiindexer. Following Connelly et al. (2010), we focused our analysis on the dedicated and transient categories, which are the most differentiated of the three categories.11 Following Bushee (1998), we took the average percentage of dedicated or transient institutional investor ownership at the beginning and at the end of a quarter to measure the average level of dedicated or transient institutional investor ownership during that quarter. To avoid any reverse causality between competitive behavior and ownership changes, we used one-quarter lagged measures. CEO Stock-Based Incentives. Executive compensation data were obtained from the Compustat Executive Compensation database. Following the categorizations proposed by Devers et al. (2008), we used the dollar value of stock and exercisable unexercised stock options to measure the short-term, vested stock-based incentives owned by the pertinent CEOs. Similarly, we used the dollar value of restricted stocks and unexercisable unexercised stock options to measure the long-term, unvested stock-based incentives owned by these CEOs.12 We also used lagged measures for the analysis.13 Identification Strategy and Statistical Methods An ideal experiment to identify the contingent effects of earnings pressure on competitive behavior for different conditions of corporate governance would involve some random assignment of earnings pressure, or a natural experiment that would shift earnings pressure exogenously while keeping other things constant. Unfortunately, no such natural experiment could be identified in our empirical context. Analyst earnings forecasts are complex measures that involve interpretation of past performance and market trends that may affect future performance, social learning among analysts, and business interests of analysts and their institutions (Schipper 1991, Zhang and Gimeno 2010). As a consequence, there may be many observable and unobservable factors that influence earnings pressure and may also influence competitive behavior, leading to possible omitted variable bias or endogeneity problem. Similarly, the variables associated with ownership structure and CEO incentives are also endogenously determined by the decisions of investors (to hold or not the stock), boards (setting incentive structures), and managers (whether to hold or exercise vested incentives), potentially leading to biased inference about their effect on competitive behavior. Although there is no perfect solution for the identification problems outlined previously, we pursued a multipronged approach for inference, including dynamic panel data models to explore the effect of a firm’s earnings pressure on its own competitive behavior, DiD specification to explore how firms respond to earnings pressure shocks experienced by their rivals, and market-level

models to explore the impact of earnings pressure on market-level outcomes.14 Dynamic Panel Data Models. The first approach used a dynamic panel data regression method that allowed us to control for some sources of unobserved heterogeneity through fixed effects and lagged variables, and include some control variables and perform post-hoc tests to assess the robustness of results to inclusion of additional controls for omitted variables. We also developed an instrumental variable approach to reduce endogeneity bias in governance variables. We focused on explaining the quarterly levels of competitive variables (air fares and flight frequencies) for a given airline in a particular city-pair route over multiple quarterly periods. Because the airline industry is highly seasonal, we examined price and frequency levels each quarter (Yt ) as dependent variables while controlling for the lagged levels of those variables in the prior quarter and the same quarter in the prior year (YtÉ1 , YtÉ4 ) in order to account for autocorrelated time trends and seasonal effects.15 All independent variables and most control variables were measured at the most recent prior quarter (XtÉ1 ), to avoid any bias from simultaneity or reverse causality. This model structure provides a particularly demanding test for inference akin to Granger causality, since it examines whether lagged levels of the independent variables affect the dependent variable after controlling for the lagged levels of the same dependent variable. Sources of unobserved heterogeneity that are relatively stable over time would be neutralized by the autoregressive structure. Given the panel structure of the data, observations may share common sources of unobserved heterogeneity, associated with the firm, the market, or the time period. We controlled for those factors by including fixed effects for firms and for markets that absorbed any unobserved firm traits or market characteristics that were invariant over the sample period. The Hausman test suggested it was more appropriate to treat market effects as fixed rather than as random effects (p < 0001). In addition, we also included fixed effects for time periods (controlling for time trends that affected all observations within a period, such as oil prices or macroeconomic conditions). In practice, the interpretation of this model is that it controls the crosssectional differences across firms and markets and focuses on explaining the remaining covariance of variables within firms and markets. Formally, the models can be stated as follows: Let Xim1 tÉ1 represent the level of the independent variables for firm i in market m at time period t É 1, and let Yim1 t represent the level of the dependent variable for the same firm and market at time period t. Let â, å and í represent fixed effects for firm, market, and time periods, respectively. Then, our model structure follows the equation Yim1 t = Å0 + ÅYim1 tÉ1 + ÇYim1 tÉ4 + ÉXim1 tÉ1 + âi + åm + ít + òimt 0

(1)


8


Because both yield and frequency variables were skewed, we applied a log transformation to both, to make their distribution more symmetric. Because some hypotheses involved evaluation of interaction effects, we centered the interactions’ components around their sample mean before calculating the interaction items (Aiken et al. 1991). Such transformations helped us avoid multicollinearity between main effects and interaction effects and better interpret the interaction effects. We controlled for several factors that may influence both earnings pressure and competitive behavior, such as past performance (return on assets on previous quarter), and financial constraints (Kaplan-Zingales index of financial constraints; Kaplan and Zingales 1997). The latter takes into account different forms of financial constraint, such as lower cash balance, weaker cash flow, and higher leverage. We also controlled for product market factors that could affect competitive aggressiveness, such as changes in cost levels of inputs of air travel, including changes in airport fees, fuel, market size, market and airport concentration, and the presence of potential entrants. We also controlled for differences in service characteristics, such as the percentage of the firm’s tickets sold as round trips, the percentage of tickets in non-economy classes, and the percentage of tickets offering nonstop travel. Online Appendix D contains extended definitions of each of the control variables. Difference-in-Differences Models. The second approach uses a DiD method to identify the competitive effects of earnings pressure by observing how firms adjust their competitive behavior in response to earnings pressure shocks experienced by their rivals in different duopoly markets. This approach is inspired by Chevalier’s (1995a, b) crossrival analysis of the competitive impact of supermarket LBOs. The link between rival’s earnings pressure and focal firm’s competitive behavior is less likely to be biased by unobserved heterogeneity, since they are different firms with different sources of heterogeneity. Moreover, the analysis leverages the fact that a given firm in a given period may face different rivals in different markets, with some rivals experiencing an earnings pressure shock (treatment), whereas others not (control). By comparing treatment and control cases for a given firm and period, the analysis also controls for important sources of unobserved heterogeneity affecting the competitive behavior of these airlines. In contrast to the panel data model, which uses all observations available and tries to make observations comparable with the use of control variables, the DiD approach focuses on selecting the sample of observations that are as comparable as possible, except for the exogenously assigned treatment or control. Moreover, the identification strategy relies on the expectations that a rivals’ onset of earnings pressure would lead that rival to soften its competitive behavior in that market


(according to the baseline hypothesis), and that would lead the focal firm to respond. This expectation holds most strongly in a context of high competitive interdependence. Accordingly, we selected observations only from duopoly markets, where interdependence is at its highest. We selected situations where the rival firm did not experience earnings pressure in the previous period (t É 1), and either experienced (treatment) or did not experience (control) earnings pressure in the subsequent period. Earnings pressure was dichotomized based on the continuous earnings pressure variable being above zero (i.e., forecasts greater than potential earnings). To ensure that the pre- and post-treatment observations of the focal firm were stable and not polluted by the focal firm’s own earnings pressure (in case these conditions tend to be correlated), we selected observations where the focal firm was public but did not experience earnings pressure in the initial or subsequent periods, or the focal firm was privately traded and therefore not affected by analysts’ earnings pressure. This also allowed us to extend the number of airlines used for this analysis, by including private firms facing publicly traded rivals. Representing Yimt as the competitive behavior of firm i in market m at period t, and D_EPjmt as the dummy of whether rival j in market m faces an earnings pressure shock, the baseline DiD model is as follows: Yim1 t ÉYim1 tÉ1 = Å+É„D_EPjm1 t +òimt

6conditional on

D_EPim1 tÉ1 = D_EPjm1 tÉ1 = D_EPim1 t = 070 That model compares the pre–post change on competitive behavior of duopoly firm-market units whose rivals have experienced an earnings pressure shock versus those that have not. The coefficient É is equivalent to a t-test of the difference in changes in the subgroups. The previous analysis would compare treatment and control cases across different airlines, markets, and periods, making the assumption that those units were comparable with the exception of the rivals’ earnings pressure shock. An even more stringent identification could be achieved by leveraging the fact that a given firm in a given time period competes in different markets, and comparing treatment and control observations matched for the same focal firm and time period. Formally, this implied the following model: Yim1 t ÉYim1 tÉ1 = Åit +É„D_EPjm1 t +òimt

6conditional on

D_EPim1 tÉ1 = D_EPjm1 tÉ1 = D_EPim1 t = 070 We used this identification approach to indirectly test whether the effect of rival’s earnings pressure on focal firm’s competitive response was moderated by the rival’s ownership structure and CEO incentives. The expectation

0008 É0008

0011

0 015 0 000 É0 004 0 002 É0 005 É0 017 É0 017 0 022 É0 013 É0 012 É0 029 É0 012 0 009 É0 007 É0 015 0 015 0 019 0 008 0 000 É0 002 É0 003 0 004 É0 001 0 004 0 007 0 025 É0 002 0 004 É0 002 0 005 0 023 É0 001 0 010 É0 002 É0 006 0 000 0 009 0 008 É0 016 0 015 0 023 0 030 0 021 É0 018 0 014

É0 025 0 000 É0 003 0 001 0 009 0 003 É0 003 É0 018 É0 001 0 001 0 003 0 000 É0 008 É0 022 0 016 É0 005 0 007 É0 012 0 009 É0 002 0 023 0 035 É0 006 0 008 0 025 0 037 É0 010 0 009 É0 039 É0 054 0 033 É0 012 0 027 0 025 É0 010 0 005 0 027 0 028 É0 014 0 014 0 035 É0 023 0 021 0015 É0 045 0 016 É0012 É0067 É0 007 0054 0016 É0014 0052 0016 É0017 0043 0009 É0006 É0001 É0021 É0006 É0004 0000 0014 0 014 É0 005 0 033 É0 005 0 003 0 012 É0 019 0 004 0 016 0 029 É0 012 É0 016 É0 001 0 031 É0 014 É0 056 0 063 0 037 É0 064 0 037 0 035 É0 048 É0 027 É0045 0 025 É0010 0016 É0014 0010 0014 É0060 0040 0021 0044 É0027 É0024 É0011 0006 0035 É0012 0012 0031 0006 É0001 0011 0006 É0004 0011

(12) (11) (10) (9) (8) (7) (6) (5) (4) (3) (2) (1) Std. dev. Mean

Panel Data Regression Results The final sample for the panel data models included 145,355 observations representing 11 airline entities that were active in 3,835 city-pair markets for a maximum of 7 years (from 1994 to 2000).16 Despite attrition in the number of airlines in the sample, over 75% of the airline-market-period observations came from airlines active in the market for seven consecutive years (i.e., they constitute full panels). Firms active in a market were present for an average of 18 consecutive quarters. Since the sample eliminates monopoly markets, the incumbents per market ranged from 2 to 12, with a median of 4. The descriptive statistics and correlations of the variables are presented in Table 1. Tables 2 and 3 present the panel data analysis on price and frequency, respectively. Model 1 in the respective tables contains information about the effects of control variables on competitive aggressiveness. Both prices and frequencies have a strong autoregressive effect, although this effect is higher in frequencies than in prices (suggesting greater stickiness in frequency than in price). Both dependent variables also have a strong seasonal effect. Among the variables that had a significant effect on price levels, financial constraint, fuel cost, flight frequency, percentage of round trips, percentage of direct flights and

Variable

Results

Descriptive Statistics and Correlations for Sample in Dynamic Panel Data Models

In these models, we expect that a reduction of competitive aggressiveness of the rival (higher price, fewer frequencies) as a result of earnings pressure would imply an outward shift of the focal firm’s firm-specific demand in the market. From a point of view of expected response (Fudenberg and Tirole 1984), prices are strategic complements, whereas frequencies, as capacities, are strategic substitutes. The implication is that we should observe a response by focal firms to rival’s earnings pressure by increasing prices (triggered by rivals’ higher prices) or increasing frequencies (triggered by rivals’ fewer frequencies). These predictions leave substantial room for empirical differences in response by focal firms, since increasing prices in response to earnings pressure is a more accommodating response than increasing frequencies. In fact, Chevalier’s (1995a, b) research on competitive responses to supermarket LBOs found evidence of both higher prices and increases in capacity by competitors.

(13)

+ Ñ4„D_EPjm1 t ⇥ Gjm1 t 5 + òimt 0

É200616 004546 0 009 0 007 É0 004 É0 008 É0 004 É0 007 102440 005531 0024 É0 002 É0 001 0 002 É0 002 0 001 É000080 000207 0005 É0001 É0 020 É0 018 0 015 0 008 002213 001030 É0004 É0003 É0021 0 003 É0 029 É0 021 001767 000929 É0009 0002 É0022 0009 0 009 0 000 1305850 1705118 0000 É0004 0013 É0029 0005 0 066 1404769 1502291 É0004 0002 0005 É0032 0003 0067 200472 004695 0008 É0004 É0014 0032 0019 É0040 É0053 000075 000079 0001 0001 0006 É0018 0014 0061 0034 É106892 001700 0076 0028 0009 É0011 É0012 0021 0015 406646 007310 É0012 0037 É0001 É0006 0003 0011 0012 000853 002565 0003 0001 0002 É0033 É0021 0053 0051 008629 002072 É0013 É0004 0001 0027 0017 É0031 É0031 001465 003414 0019 0010 0000 É0007 É0001 0011 0011 204759 109081 0010 0026 0001 É0008 É0001 0009 0009 002445 001744 0010 É0004 0000 0003 0004 É0001 É0003 002749 001079 0025 0011 0000 0000 É0003 É0007 É0006

(14)

Yim1 t É Yim1 tÉ1 = Åit + ÇGjm1 t + É„D_EPjm1 t

(1) Ln (yield)t (2) Ln (frequency)t (3) Earnings pressuret (4) Dedicated institutional investortÉ1 (5) Transient institutional investortÉ1 (6) Vested CEO stock-based incentivetÉ1 (7) Unvested CEO stock-based incentivetÉ1 (8) Financial constraintstÉ1 (9) Past performancetÉ1 (10) Ln (cost)t (11) Ln (market size)tÉ1 (12) % high classtÉ1 (13) % round trip ticketstÉ1 (14) % direct flightstÉ1 (15) Potential entrantstÉ1 (16) Market HeifindahltÉ1 (17) Airport Herfindahlt

(15)

(16)

(17)

was that rivals with more dedicated investors and unvested CEO incentives would be less likely to soften competition in response to earnings pressure relative to rivals with more transient investors and vested CEO incentives, and therefore would lead to a weaker response by the focal firm. Formally, this was tested as follows:

Notes. n = 1451355. Correlations with absolute values greater than or equal to 0.01 are significant at p < 0005. Correlations in italics are within-group correlations by firm and market.

9


Table 1




10


market, and airport concentration affected yield level positively; past performance, market size, percentages of business and first class tickets, and number of potential entrants all had a negative effect. These effects were mostly consistent with previous findings in airline studies (Evans and Kessides 1994, Gimeno and Woo 1996) and price competition (Chevalier 1995a, b). In contrast, financial constraints, past performance, pricing level, market size, percentages of business and first class tickets, number of potential entrants, and airport concentration affected flight frequency negatively, but market concentration had a positive impact. Prior research has proposed that earnings pressure had a positive impact on prices of the focal firm and a negative impact on frequencies. Model 2 of Table 2 supports the effect on prices: It shows that earnings pressure had a positive and significant effect on price changes in the next quarter, which supports this hypothesis. An increase of one standard deviation in earnings pressure (about 2% of the stock price) would lead to an increase in yield of 0.2%, other things being equal. Although this number may seem small, it nevertheless amounts to about 50% of the average change in quarterly ticket prices in the sample (0.41%). It is also about 4.3% to 10.5% of the average net profit margin of the airlines in the sample period, which varied between 1.9% and 4.7% (Air Transportation Association 2007, p. 7). Table 3 reports the analysis of flight frequency. The results show that earnings pressure had a negative and significant effect on frequency, which agrees with our baseline hypothesis. An increase of one standard deviation in earnings pressure would cause daily flight frequency to decrease by 0.14%, which was about the average change in quarterly flight frequency in the sample. Moderating Effects of Governance Variables. Models 3 and 4 in Tables 2 and 3 examined the interactions with ownership structure and CEO incentives separately, whereas Model 5 in Tables 2 and 3 tested all interaction effects together. Because the results are similar across these three models, we report and interpret the results based on the full model, Model 5. Hypothesis 1 examined whether the baseline effect of earnings pressure in lowering competitive aggressiveness (i.e., higher prices, lower frequencies) was weaker for firms with dedicated investors than for those with transient investors. Model 5 of Table 2 shows that the interaction between earnings pressure and dedicated institutional investor ownership had a negative and significant effect on price-consistent with a dampening effect (b1), whereas the interaction between earnings pressure and transient institutional investor ownership was insignificant (b2). An F -test of the equality of coefficients in Model 5 of Table 2 showed that the coefficient estimate of the interaction between earnings pressure and dedicated institutional investor ownership was significantly less than that of


the interaction between earnings pressure and transient institutional investor ownership (H1: b1 > b2, F = 12089, p < 0001). A similar F -test in Table 3 showed that the interaction between earnings pressure and dedicated institutional investor ownership was higher than that of the interaction between earnings pressure and transient institutional investor ownership (H2: b3 > b4, F = 24077, p < 0001). This suggests that the effect of earnings pressure on frequency was less negative when the focal firm had more dedicated institutional investor ownership than when it had more transient institutional investor ownership.17 Hence Hypothesis 1 was supported for both price and frequency competitive decisions. Hypothesis 2 examined whether the effect of earnings pressure on competitive aggressiveness was weaker for firms with CEOs with unvested rather than vested incentives. Model 5 of Table 2 shows that the interaction between earnings pressure and vested CEO stock-based incentives had a positive and significant effect on price; whereas the interaction between earnings pressure and unvested CEO stock-based incentives had a negative and significant effect. These coefficients were statistically different (F = 123009, p < 0001). In terms of analysis of the frequency decision, Model 5 of Table 3 shows that the coefficient estimate of the interaction between earnings pressure and unvested CEO stock-based incentives was significantly larger than that of the interaction between earnings pressure and vested CEO stock-based incentives (F = 11096, p < 0001). This suggests that the effect of earnings pressure on frequency was more negative when a CEO had more vested stock-based incentives that could be immediately exercised, than when a CEO held more unvested stock-based incentives that could only be exercised in the future. Therefore Hypothesis 2 was also supported. Figure 1(a) illustrates graphically the interaction effects detected by using the method suggested by Aiken et al. (1991). Specifically, an increase in earnings pressure by one standard deviation increased ticket prices by 1.3% when a firm had predominantly transient investors (i.e., at one standard deviation below the mean of dedicated minus transient ownership), but decreased ticket prices by 0.4% when a firm’s investors were predominantly in the dedicated category (one standard deviation above the mean of dedicated minus transient ownership). Similarly, graphical representations of the interaction effects in Figure 1(b) suggest that an increase in earnings pressure by one standard deviation increased ticket prices by 2% when a CEO had an incentive package weighted more with vested incentives (evaluated at one standard deviation above the mean of vested minus unvested incentives). However, earnings pressure decreased ticket prices by about 1% when unvested incentives predominated in the CEO’s incentive portfolio (evaluated at one standard deviation below the mean of vested minus unvested incentives).



11


Table 2

Dynamic Panel Data Model Results for the Effects of Earnings Pressure on Yield

Ln (yield)t

Model 1

Model 2

Ln (yield)tÉ1

005086⇤⇤⇤ 40000205

Ln (yield)tÉ4

001214⇤⇤⇤ 40000205

Earnings pressuret

Model 3

Model 4

Model 5

005084⇤⇤⇤ 40000205

005067⇤⇤⇤ 40000205

005106⇤⇤⇤ 40000205

005088⇤⇤⇤ 40000205

001214⇤⇤⇤ 40000205

001212⇤⇤⇤ 40000205

001212⇤⇤⇤ 40000205

001214⇤⇤⇤ 40000205

001080⇤⇤⇤ 40002505

000261 40002605

001594⇤⇤⇤ 40002605

001114⇤⇤⇤ 40002705

Earnings pressure ⇤ Dedicated institutionaltÉ1

É102136⇤⇤⇤ 40024305

É103257⇤⇤⇤ 40024805

Earnings pressure ⇤ Transient institutionaltÉ1

É006120⇤⇤⇤ 40022605

000586 40023205

Earnings pressure ⇤ Vested CEO incentivetÉ1 Earnings pressure ⇤ Unvested CEO incentivetÉ1 Dedicated institutional investortÉ1

000043⇤⇤ 40000205

É000505⇤⇤⇤ 40000205

É000402⇤⇤⇤ 40000205

É001332⇤⇤⇤ 40000605

É001182⇤⇤⇤ 40000605

000494⇤⇤⇤ 40000805

000172⇤⇤ 40000805

Transient institutional investortÉ1 Vested CEO stock-based incentivetÉ1 Unvested CEO stock-based incentivetÉ1 Financial constrainttÉ1

000065⇤⇤⇤ 40000205

É000006⇤⇤⇤ 40000005

É000007⇤⇤⇤ 40000005

000002⇤⇤⇤ 40000005

000004⇤⇤⇤ 40000005

000131⇤⇤⇤ 40000305

000147⇤⇤⇤ 40000305

000086⇤⇤⇤ 40000305

000084⇤⇤⇤ 40000305

000082⇤⇤⇤ 40000305

É008155⇤⇤⇤ 40009005

É007489⇤⇤⇤ 40009105

É008466⇤⇤⇤ 40009305

É008618⇤⇤⇤ 40010705

É006656⇤⇤⇤ 40010905

Ln (cost)t

104322⇤⇤⇤ 40005805

104334⇤⇤⇤ 40005805

104300⇤⇤⇤ 40005805

104171⇤⇤⇤ 40005805

104206⇤⇤⇤ 40005805

Ln (daily frequencies)tÉ1

000122⇤⇤⇤ 40000105

000122⇤⇤⇤ 40000105

000123⇤⇤⇤ 40000105

000114⇤⇤⇤ 40000105

000116⇤⇤⇤ 40000105

Ln (market size)tÉ1

É000573⇤⇤⇤ 40000305

É000574⇤⇤⇤ 40000305

É000586⇤⇤⇤ 40000305

É000588⇤⇤⇤ 40000305

É000596⇤⇤⇤ 40000305

% high classtÉ1

É000165⇤⇤⇤ 40000505

É000170⇤⇤⇤ 40000505

É000292⇤⇤⇤ 40000505

É000264⇤⇤⇤ 40000505

É000343⇤⇤⇤ 40000505

% round trip ticketstÉ1

000562⇤⇤⇤ 40000305

000568⇤⇤⇤ 40000305

000644⇤⇤⇤ 40000305

000792⇤⇤⇤ 40000305

000822⇤⇤⇤ 40000305

% direct flightstÉ1

000052⇤⇤⇤ 40000205

000052⇤⇤⇤ 40000205

000051⇤⇤ 40000205

000042⇤⇤ 40000205

000042⇤⇤ 40000205

Potential entrantstÉ1

É000032⇤⇤⇤ 40000005

É000032⇤⇤⇤ 40000005

É000031⇤⇤⇤ 40000005

É000031⇤⇤⇤ 40000005

É000030⇤⇤⇤ 40000005

Market HerfindahltÉ1

000389⇤⇤⇤ 40000505

000390⇤⇤⇤ 40000505

000398⇤⇤⇤ 40000505

000394⇤⇤⇤ 40000505

000401⇤⇤⇤ 40000505

Airport Herfindahlt

000864⇤⇤⇤ 40001605

000880⇤⇤⇤ 40001605

000867⇤⇤⇤ 40001605

000774⇤⇤⇤ 40001605

000813⇤⇤⇤ 40001605

Constant

108670⇤⇤⇤ 40010305

108645⇤⇤⇤ 40010305

108492⇤⇤⇤ 40010305

108870⇤⇤⇤ 40010305

109028⇤⇤⇤ 40010305

Yes Yes Yes

Yes Yes Yes

Past performancetÉ1

Firm fixed effects Market fixed effects Time-period fixed effects

Yes Yes Yes

F for Hypothesis 1 (1 df ) F for Hypothesis 2 (1 df )

206† 209007

Observations R2 F -test Relative to model Note. Standard errors in parentheses; ⇤ p < 001,

Yes Yes Yes

145,355 009059

⇤⇤

p < 0005,

⇤⇤⇤

145,355 009059 1801††† 1

p < 0001, 2-sided, † p < 001,

145,355 009063 14602††† 2 ††

p < 0005,

⇤⇤⇤

145,355 009066 22909††† 2 †††

p < 0001, 1-sided.

Yes Yes Yes 12089⇤⇤⇤ 123009⇤⇤⇤ 145,355 009068 16009††† 2



12 Table 3


Dynamic Panel Data Model Results for the Effects of Earnings Pressure on Flight Frequency

Ln (frequency)t

Model 1

Model 2

Model 3

Model 4

Model 5

Ln (frequency)tÉ1

007815⇤⇤⇤ 40000205

007814⇤⇤⇤ 40000205

007812⇤⇤⇤ 40000205

007807⇤⇤⇤ 40000205

007805⇤⇤⇤ 40000205

Ln (frequency)tÉ4

002049⇤⇤⇤ 40000205

002050⇤⇤⇤ 40000205

002052⇤⇤⇤ 40000205

002057⇤⇤⇤ 40000205

002059⇤⇤⇤ 40000205

É000763⇤⇤⇤ 40002405

É000778⇤⇤⇤ 40002405

Earnings pressuret Earnings pressure ⇤ Dedicated institutionaltÉ1

É000283 40002405

105914⇤⇤⇤ 40022705

Earnings pressure ⇤ Transient institutionaltÉ1

É000482⇤ 40002505 108416⇤⇤⇤ 40023205 000465 40021705

É001645 40021105

Earnings pressure ⇤ Vested CEO incentivetÉ1 Earnings pressure ⇤ Unvested CEO incentivetÉ1

É000064⇤⇤⇤ 40000205

É000058⇤⇤⇤ 40000205

000073⇤⇤⇤ 40000205

000071⇤⇤⇤ 40000205

Dedicated institutional investortÉ1

000443⇤⇤⇤ 40000605

000312⇤⇤⇤ 40000605

Transient institutional investortÉ1

000128⇤ 40000705

000127⇤ 40000705

Vested CEO stock-based incentivetÉ1 Unvested CEO stock-based incentivetÉ1

É000004⇤⇤⇤ 40000005

É000005⇤⇤⇤ 40000005

000004⇤⇤⇤ 40000005

000004⇤⇤⇤ 40000005

Financial constrainttÉ1

É000054⇤ 40000305

É000066⇤⇤ 40000305

É000065⇤⇤ 40000305

É000035 40000305

É000049 40000305

Past performancetÉ1

É009164⇤⇤⇤ 40008405

É009634⇤⇤⇤ 40008505

É100153⇤⇤⇤ 40008705

É005427⇤⇤⇤ 40010005

É005423⇤⇤⇤ 40010205

Ln (cost)t

É000368 40005405

É000377 40005405

É000371 40005405

É000356 40005405

É000338 40005405

Ln (yield)tÉ1

É000171⇤⇤⇤ 40000205

É000169⇤⇤⇤ 40000205

É000166⇤⇤⇤ 40000205

É000167⇤⇤⇤ 40000205

É000165⇤⇤⇤ 40000205

Ln (market size)tÉ1

É000708⇤⇤⇤ 40000305

É000707⇤⇤⇤ 40000305

É000706⇤⇤⇤ 40000305

É000709⇤⇤⇤ 40000305

É000709⇤⇤⇤ 40000305

% high classtÉ1

É000207⇤⇤⇤ 40000405

É000203⇤⇤⇤ 40000405

É000177⇤⇤⇤ 40000405

É000214⇤⇤⇤ 40000405

É000198⇤⇤⇤ 40000405

% round trip ticketstÉ1

É000014 40000305

É000018 40000305

É000025 40000305

000004 40000305

000018 40000305

000021 40000205

000021 40000205

000022 40000205

000020 40000205

000020 40000205

Potential entrantstÉ1

É000004⇤⇤⇤ 40000005

É000004⇤⇤⇤ 40000005

É000005⇤⇤⇤ 40000005

É000004⇤⇤⇤ 40000005

É000005⇤⇤⇤ 40000005

Market HerfindahltÉ1

000357⇤⇤⇤ 40000505

000356⇤⇤⇤ 40000505

000355⇤⇤⇤ 40000505

000358⇤⇤⇤ 40000505

000357⇤⇤⇤ 40000505

É001654⇤⇤⇤ 40001505

É001666⇤⇤⇤ 40001505

É001688⇤⇤⇤ 40001505

É001639⇤⇤⇤ 40001505

É001659⇤⇤⇤ 40001505

002230⇤⇤ 40009605

002247⇤⇤ 40009605

002329⇤⇤ 40009605

002659⇤⇤⇤ 40009605

002672⇤⇤⇤ 40009605

Yes Yes Yes

Yes Yes Yes

Yes Yes Yes

Yes Yes Yes

Yes Yes Yes

% direct flightstÉ1

Airport Herfindahlt Constant Firm fixed effects Market fixed effects Time-period fixed effects F for Hypothesis 1 (1 df ) F for Hypothesis 2 (1 df )

25043⇤⇤⇤ 13076

Observations R2 F -test Relative to model Note. Standard errors in parentheses; ⇤ p < 001,

145,355 009447

⇤⇤

p < 0005,

⇤⇤⇤

145,355 009447 1003††† 1

p < 0001, 2-sided, † p < 001,

145,355 009447 1909††† 2 ††

p < 0005,

†††

⇤⇤⇤

145,355 009447 3503††† 2 p < 0001, 1-sided.

24077⇤⇤⇤ 11096⇤⇤⇤ 145,355 009448 2701††† 2

13


Figure 1

(Color online) Contingency Effects of Earnings Pressure on Price Levels

(a)

Effect of earnings pressure at high vs. low dedicated–transient investors

(b)

0.132 High (dedicated–transient)

0.128

Yield level

0.130

Effect of earnings pressure at high vs. low unvested–vested incentives

0.130

Low (dedicated–transient)

Yield level



0.128

0.126

0.126

0.124 Low (unvested–vested)

0.124 – 0.05

High (unvested–vested)

0.122 0

– 0.05

0.05

Earnings pressure

0

0.05

Earnings pressure

Note. Short-dashed lines are the 95% confidence intervals of the effects.

We also tried several post hoc analyses to ensure our results for the interaction effects were valid and strategically relevant.18 One alternative explanation for our results could be that there are some firm characteristics that correlate with ownership and incentives and also affect how companies respond competitively to earnings pressure. We tested whether the results were robust to the inclusion of additional controls and interactions in three ways. First, we tried to control for the possibility that stable, firm-specific factors may influence how firms respond to earnings pressure. We created firm-specific slopes for earnings pressure (interactions between earnings pressure and a set of firm dummies) to account for possible heterogeneity among firms in their response to earnings pressure. The interactions effects between earnings pressure and ownership or incentives remained statistically significant in the expected direction. Therefore, the results for Hypotheses 1 and 2 cannot be explained by differences across firms in their responses to earnings pressure. Second, we controlled for other time-varying variables that may be correlated to ownership or CEO incentive variables—such as past performance, financial constraints, stock price, CEO tenure, or service quality attributes—and may also moderate the effect of earnings pressure and have potential as an alternative explanation. We included interactions one at a time between earnings pressure and these time-varying variables. The hypothesized effects of ownership or incentives maintained their direction and significance in all cases. Third, we also used an instrumental variable approach to explicitly examine endogeneity concerns regarding ownership and incentives. Prior literature was used to identify factors that may influence ownership structure (Jiambalvo et al. 2002) and incentive structures (Rajgopal and Shevlin 2002).19 The interaction effects between earnings pressure and ownership or incentives remained significant in the hypothesized direction in the models with instrumented ownership and incentive variables.

Difference-in-Differences Analyses Results Table 4 reports the DiD analyses results for the baseline effect of earnings pressure, examined from a cross-rival angle. The subsample of duopoly markets that we use includes 13,154 observations composed by 31 airlines (including 20 private companies) in 2,057 markets over 29 quarters. Consistent with our expectation, we found that focal firms increased their prices (1.96% versus 0.55%, t = 2089) and flight frequency (4.57% versus 1.75%, t = 4053) more when their rivals faced earning pressure in the second quarter, and the differences between treatment and control were highly significant. This effect is also supported when we use an OLS regression in Model 1 and 4 with the treatment dummy as independent variable. The treatment effect remained positive and significant on focal firms’ prices (b = 00014, p < 0001) and flight frequency changes (b = 00028, p < 0001). For Model 2 and 5 in Table 4, we went further and matched treatment and control cases for the same focal firm and period (by adding firm-period dummies), but in different markets, as a way to block any firm-period specific differences. Hence these comparisons excluded cross-firm or cross-period comparisons for identification. In this specification, the treatment effects became statistically insignificant. However, one reason for this is that such model does not take into account that rival firm’s sensitivity to earnings pressure may depend on their governance characteristics such as ownership composition and CEO incentive horizons. Therefore, we expanded the DiD analyses further by exploring heterogeneity in the treatment effects in Model 3 and 6. Model 3 and 6 report the results of DiD analyses for interactions with rival firms’ ownership composition and CEO incentive horizons. The results showed that when the rival firm in a duopoly market experienced an earnings pressure shock, its effect on the focal firm’s ticket price and flight frequency were weaker when the rival firm had higher dedicated investor ownership (i.e., above the sample median) than when it had higher transient investor



14 Table 4


Difference-in–Differences Model Results DV = Focal firm yield change (1)

Rival’s earnings pressure shock dummy

(2)

000141⇤⇤⇤ 4000055

DV = Focal firm frequency change (3)

000075 4000085

(4)

(5)

000282⇤⇤⇤ 4000065

É000268 4000185

É000030 4000095

(6) É000204 4000225

Rival’s earnings pressure shock dummy ⇤ High rival dedicated ownership Rival’s earnings pressure shock dummy ⇤ High rival transient ownership Rival’s earnings pressure shock dummy ⇤ High rival vested incentive Rival’s earnings pressure shock dummy ⇤ High rival unvested incentive High rival dedicated ownership

É000246⇤ 4000145 000223 4000195 000394⇤⇤ 4000195 000023 4000225 É000070 4000065

É000846⇤⇤⇤ 4000175 000227 4000245 000635⇤⇤⇤ 4000235 É000127 4000275 000155⇤⇤ 4000075

High rival transient ownership

É000051 4000075 É000484⇤⇤⇤ 4000155 000122 4000085 009229⇤⇤⇤ 4001795

É000783⇤⇤⇤ 4000085 000245 4000185 É000043 4000095 007039⇤⇤⇤ 4002165

High rival vested incentive High rival unvested incentive Constant Firm-period dummies F -test for Hypothesis 1 (1 df ) F -test for Hypothesis 2 (1 df ) Observations R-squared Note. Standard errors in parentheses, ⇤ p < 001,

100055⇤⇤⇤ 4000025

009474⇤⇤⇤ 4001795

No

Yes

13,154 000006

13,154 000568

⇤⇤

p < 0005,

⇤⇤⇤

Yes 2082† 1053 13,154 000589

p < 0001, 2-sided, † p < 001,

ownership, and also weaker when the rival firm’s CEO had higher unvested stock-based incentives than when she had higher vested stock-based incentives. A focal firm would increase the price and flight frequency by 2.5% and 3.6% when the rival firm had high transient and low dedicated institutional investor ownership, but would decrease the price and flight frequency by 2.2% and 7.1% when the rival firm had high dedicated and low transient institutional investor ownership. In addition, a focal firm would increase the price and flight frequency by 1.5% and 0.9% when the rival firm’s CEO had higher vested and lower unvested stock-based incentives, but would decrease the price and flight frequency by 1.5% and 7.1% when she had higher unvested and lower vested stock-based incentives. These results are consistent with the cross-rival implications of Hypotheses 1 and 2: a response by focal firms to rival’s earnings pressure by increasing prices (triggered by the rival’s higher price as a result of its earnings pressure, since prices are strategic complements) or increasing frequencies (triggered by the rival’s lower frequency as a result of its earnings pressure, because frequencies or capacities are strategic substitutes) will be weaker when the rival firm had higher dedicated investor ownership or higher unvested stock-based incentives. This is because dedicated institutional investors or unvested stock-based incentives make the rival firm less subject to

††

100174⇤⇤⇤ 4000025

100083⇤⇤⇤ 4002185

No

Yes

13,154 000015

13,154 001508

p < 0005,

†††

Yes 10004††† 4042†† 13,154 001630

p < 0001, 1-sided.

the main effect of its earnings pressure, thus allows it to take more aggressive pricing or frequency decisions (i.e., lower price or higher frequency). The results control for unobserved effects of the focal firm, since both the treatment and control observations were matched for the same firm and period. Therefore, Hypotheses 1 and 2 are supported when we use the more rigorous DiD approach.

Discussion

In this paper we examined how corporate governance conditions influence managerial reactions to securities analysts’ earnings pressure. We employed a multiplemethod approach to overcome limitations of different empirical methods when examining the complex phenomenon of earnings pressure.20 Our baseline results show that, on average, airlines tended reduce competitive aggressiveness under earnings pressure. This is similar to what Zhang and Gimeno (2010) found with electricity firms: firms facing earnings pressure were more likely to restrict output and exercise market power as they focused more on capturing value to meet earnings expectations. In contrast, we found that airlines increased prices and reduced frequency to generate short-term earnings to meet earnings forecasts. Our post hoc results also validate Zhang and Gimeno’s 2010 finding that the effect of earnings pressure is stronger when firms have dominant



positions in the market or operate in more concentrated markets. These results suggest that, at least in these industries—both of which are characterized by capacity constraints and local oligopolistic structures—firms facing earnings pressure attempt to raise earnings by capturing more value from customers. Hence, our baseline results show that the findings of Zhang and Gimeno (2010) can be extended to other industry settings (airline) and new types of competitive behavior (pricing and service quality). However, the major finding and contribution of our paper is its evidence that managerial responses to earnings pressure are highly influenced by important dimensions of corporate governance, such as the time horizon of investors and CEO incentives, and that those contingencies may accentuate or dampen (or even reverse) the de-escalation behavior. Specifically, we found that earnings pressure had a more negative impact on aggressiveness of competitive behavior when airlines had higher transient, short-termoriented institutional investor ownership compared to when airlines had higher ownership levels by dedicated, longterm-oriented institutional investors. Similarly, different vesting conditions for stock-based CEO incentives can moderate the impact of earnings pressure. Companies with more vested or exercisable stock-based incentives were more likely to reduce aggressiveness in response to analysts’ earnings pressure, relative to companies with more unvested or unexercisable stock-based incentives. This perspective is new and has not been discussed or examined in the growing literature on the interaction between securities analysts and management decision about firm strategy, but carry important implication for our understanding of the intertemporal impact of earnings pressure on firm strategy and managerial decision. An ongoing debate in the literature on earnings pressure centers on whether earnings pressure curbs managers’ agency problems, such as excessive focus on market share and firm size, or encourages managers to focus on shortterm performance and behave myopically. Prior research (e.g., Zhang and Gimeno 2010; Chevalier 1995a, b) had speculated about the myopic effect, but could not exclude the alternative mechanism of reducing agency cost. We consider our findings more consistent with the view that earnings pressure encourages myopic managerial behavior. This is because if earnings pressure only serves to improve managerial efforts, then we should observe firms with different ownership structures or managerial incentives respond similarly to earnings pressure. Our results that firms were less likely to reduce aggressiveness in response to earnings pressure when investors and managers were long-term oriented (dedicated investors, unvested incentives) relative to when investors and managers were short-term oriented (transient investors, vested incentives) are consistent with the interpretation that earnings pressure encourages myopic managerial behavior. This interpretation is also consistent with the findings

15

in other empirical studies that securities analysts cause managers to reduce strategic investment (Benner and Ranganathan 2012) or to react slowly to disruptive technologies (Benner 2010). Our findings suggest that the phenomenon of earnings pressure exerted by securities analysts cannot be studied in isolation but must take into account factors of corporate governance. Therefore, together we have provided the first empirical evidence of how the effect of earnings pressure on managerial responses may vary in the context of different corporate governance, which has both theoretical and practical implications. Contributions and Implications Several aspects of our findings may interest both academics and practitioners. First, our study expands a growing line of research on how securities analysts’ pressure may lead to changes in business decisions aimed at increasing short-term earnings. These changes include such actions as reductions in advertising expenses, exploitation of market power, and cuts in strategic investment in innovations (Cohen et al. 2010, Zhang and Gimeno 2010, Benner and Ranganathan 2012). In particular, we find that these consequences may be mitigated under certain conditions of corporate governance, such as more dedicated shareholder ownership. On the other hand, transient investors and vested stock-based CEO incentives will accentuate this behavior. These findings show that ownership structure and stock-based incentives may influence not only accounting management of earnings, as suggested by Bushee (1998) and Cheng and Warfield (2005), but also real earnings management in the form of less aggressive competitive behavior during periods of earning pressure. Furthermore, our findings also provide evidence on whether earnings pressure leads to short-term versus long-term trade-offs by the managers that could potentially damage a firm’s long-term competitiveness, thus helping to solve a debate in prior literature about the intertemporal impact of earnings pressure on firm strategy and managerial decision (Jensen 1986, Porter 1992). Second, our findings contribute to the extant literature on corporate governance that has linked variables such as institutional ownership structure or CEO incentives to corporate and competitive behaviors (e.g., Hansen and Hill 1991, Kochhar and David 1996, Hoskisson et al. 2002, Sanders and Hambrick 2007, Connelly et al. 2010). Much of that literature has assumed a direct link between governance and strategic activities, such as R&D investments, diversification, or competitive activities, assuming that managers face at all times an intertemporal tension between short-term and long-term earnings. Our paper suggests that these intertemporal tensions will be particularly concentrated in periods of earnings pressure in which companies’ potential earnings based on ongoing activities are insufficient to meet analysts’ earnings expectations. For example, Connelly et al. (2010) explored the


16


link between ownership by heterogeneous institutional investors (dedicated or transient) and competitive moves. They found that companies with more transient ownership engaged in more tactical competitive moves to fine-tune strategy, such as pricing or service changes. Whereas Connelly et al. (2010) predict that such companies will increase tactical competitive moves at all times because of their more short-term focus, we found that the main differences in competitive activity between transient and dedicated investors are most noticeable during periods of earnings pressure and less so at other times.21 Our paper also suggests that periods of earnings pressure will be particularly productive windows for identifying the effects of corporate governance on corporate and competitive behaviors with intertemporal consequences. To a broader extent, our study and findings also suggest the importance of incorporating securities analysts into the research on corporate governance and managerial behavior. Much of the corporate governance literature has assumed that managers mainly face shareholders or board members. However, research has shown the significant influence that securities analysts may have on firms’ stock prices and on managerial behavior (Zuckerman 1999, 2000; Litov et al. 2012; Wiersema and Zhang 2011; Benner 2007, 2010; Benner and Ranganathan 2012; Zhang and Gimeno 2010). To the extent that securities analysts play an increasingly important role in the functioning of capital markets and governance in publicly traded companies, their role should be emphasized more in corporate governance research. More research is also needed to explore the role of professionals such as accountants and investment bankers in the corporate governance system (Davis 2005). Finally, our findings also carry strong implications and actionable knowledge for managers, boards of directors, and investors and regulators (Pearce and Huang 2012). Managers can benefit by understanding the influence of earnings pressure on firms’ competitive behavior and especially by understanding the contingent effects of different corporate governance structures. Although managers cannot easily change these contingencies, they may still exert a certain degree of influence via investor communications and relationship strategies or via commitment to certain governance practices. In addition, a better understanding of these effects can also help managers to react more effectively to pressures from the stock market and to take advantage of competitors when those competitors face different degrees of pressures. For boards of directors and investors, knowledge of the interactions between earnings pressure, corporate governance structures, and competitive behavior can help them better evaluate firm behavior and anticipate firm performance. Such knowledge can also help them form ownership structures and use managerial incentives more effectively so that managers can keep to a sound course of business actions when they face earnings pressure. Regulators’ awareness of the effect of earnings pressure on real business actions, such as


competitive behavior, can help them foresee the possible spillover effect of financial reporting regulations. Supplemental Material Supplemental material to this paper is available at http://dx.doi .org/10.1287/orsc.2016.1056.

Acknowledgments We appreciate the help from Senior Editor Gary Dushnitsky and three anonymous reviewers, the feedback from Raffi Amit, Mary Benner, Philip Bromiley, Weiru Chen, James Constantini, Parthiban David, Douglas Frank, Igal Hendel, Marvin Lieberman, Dawn Matsumoto, Steven Monahan, Vicente Salas Fumás, Edward Zajac, Peter Zemsky, and seminar participants at Arizona State University, Boston College, China Europe International Business School, Copenhagen Business School, INSEAD, London Business School, Wharton School, University of California at Irvine, University of Michigan, and University of Zaragoza. An earlier version of this paper was included in the 2010 Academy of Management Meeting Best Paper Proceedings, and won the BlackRock/NACD (National Association of Corporate Directors) “Global Challenge for Innovation in Corporate Governance” in 2013.

Endnotes

1

Such mechanisms have been invoked in prior literature linking financial leverage to competitive behavior (e.g., Chevalier 1995a, b; Phillips 1995), where the intertemporal trade-off is between generating short-term cash and investing in competitive positions. 2 We do assume that managers are making competitive decisions based on an objective function that is a combination of shortterm earnings and future competitiveness. We also assume that there are intertemporal effects of competitive behavior between current earnings and future competitiveness, and that those effects are negative (at the margin, competitive behavior that increases future competitiveness may reduce current earnings). There are several specific mechanisms consistent with such intertemporal trade-offs, as we showed later in this paragraph. Under these assumptions, “dynamic” marginal revenues would be higher than static ones, dynamic marginal costs would be lower than static ones, and firms managing for dynamic optimization (rather than static) would therefore be more aggressive than otherwise. 3 It is possible that firms may adjust one or several types of competitive actions at the same time. This is why we propose our hypotheses on the general competitive aggressiveness, and examined three different types of competitive actions, namely ticket pricing, flight frequency, and flight cancellation, in the empirical part. We do assume that traditional assumptions in economics about demand function holds, that is, the utility of the buyer is not dependent on price. As a result, demand decreases with the increase of product price. 4 The level of competitive aggressiveness may differ across different firms and markets, based on factors such as market conditions, demand elasticity, strategy, cost structure, etc. For example, firms with weak market positions in fragmented and competitive markets would likely engage in more aggressive competitive behavior, since not doing so would result in immediate loss of both market position and earnings. However, keeping



these conditions constant, if we compare the aggressiveness under EP versus the counterfactual without EP, the former would be less aggressive. 5 The difference between “transient” and “dedicated” is based on the time horizon of the investors and their sensitivity to short term performance, as Porter (1992) proposed, rather than the actions they can take when firms fall short of investors’ expectation, or the size or concentration of their stock holdings. 6 In our sample, we have American Airlines and American Eagle Airlines owned by AMR Corp., Continental Air Lines and Continental Express Airline owned by Continental Air Lines, and U.S. Airways and U.S. Air Shuttle owned by U.S. Airways. 7 Website http://acct.wharton.upenn.edu/faculty/bushee/IIclass .html. Data was retrieved November 17, 2009. 8 We also examined other strategies that could be used by airlines, such as flight cancellation. The effects are consistent with what we found using ticket price and flight frequency. 9 As a robustness check, we also used a log of ticket prices as the dependent variable, and our results hold up well. 10 Zhang and Gimeno (2010, pp. 745–746) also discussed in detail why analysts’ forecasts could diverge from firms’ potential earnings, including the effect of sell-side incentives in financial institutions. 11 In our sample, the mean of dedicated, transient, and quasiindex institutional investor is 22.1%, 17.6%, and 30.2%, respectively, also indicating that quasi-indexer institutional investors can be treated as a baseline category. 12 We also use the ratio of dollar value of stock-based incentives scaled by the amount of base compensation (cash and bonus), and the results are similar. 13 We also performed additional robustness tests using the relative difference between unvested and vested incentives and the difference between dedicated and transient institutional investor ownership, while controlling for their sum. The results are consistent with those using the measures here. 14 Because of the page limit, we reported market-level analyses and results in Online Appendix E. 15 Using a lagged variable for the same quarter of the previous year provides better fit than including quarter dummies, since different markets may follow different seasonal patterns. 16 The total number of airlines in our DiD and market-level analyses is 31, including 11 publicly traded and 20 privately held. We lost observations for the panel data models because of a lack of earnings data for private companies, missing observations for financial variables for some airlines, and the exclusion of outliers (top and bottom 1%). The final sample represented about 66% of the tickets available from the O&D survey. A comparison of the common variables between the final sample and the relevant markets that were excluded suggests that the final sample on average consisted of larger markets with lower concentrations and more potential entrants. These larger markets also had higher flight frequency, more direct flights, and more business and first class seats. These features were expected because publicly traded firms covered by securities analysts are typically the larger companies in an industry. However, all airlines (public and private) were included when constructing market structure variables. Our results also hold if we include in our analyses the outliers for earnings pressure. Details about sample airlines and sample attrition can be found in Online Appendix C.

17

17

The effect of earnings pressure on frequency is actually positive for firms with greater dedicated and lower transient ownership, which is surprising as we would expect that the marginal effect would vary from negative (reducing aggressiveness) toward zero. One possible interpretation is that changes in earnings pressure may be spuriously correlated with market conditions that encourage greater competitive aggressiveness (greater frequency), and that firms with greater ownership by dedicated investors respond to those conditions, whereas those with more transient investors are constrained by their concern with current earnings. We thank one of the reviewers point this out for us. 18 Although these analyses are not exhaustive, they do exclude a large number of commonly possible alternative explanations of our main findings. Because of the page limit, the results of post hoc analyses are omitted here but available from the authors upon request. 19 The instrumental variables used for ownership were firm size, liquidity of stock price, recent stock return, market-to-book ratio, and analyst coverage. The instrumental variables used for incentives were firm size, market-to-book ratio, debt level, cash compensation level, and cash balance. 20 For example, the dynamic panel data regression approach employed a much larger number of observations but only examined publicly traded companies that were covered by analysts. In contrast, the DiD approach used a smaller sample, but provided a stronger causality identification and could examine the impact of earnings pressure on firms without analyst coverage or privately held. 21 For example, an increase of one standard deviation in transient investors will increase the price by 0.7% when earnings pressure is high (one standard deviation above the mean), but only by 0.4% when earnings pressure is low (one standard deviation below the mean).

References Aiken LS, West SG, Reno RR (1991) Multiple Regression: Testing and Interpreting Interactions (Sage Publications, Newbury Park, CA). Air Transportation Association (2007) Economic report. Accessed September 6, 2013, http://airlines.org/wp-content/uploads/2014/08/ 2007.pdf. Beatty RP, Zajac EJ (1987) CEO change and firm performance in large corporations: Succession effects and manager effects. Strategic Management J. 8(4):305–317. Bebchuk LA, Fried JM (2004) Pay Without Performance: The Unfulfilled Promise of Executive Compensation (Harvard University Press, Cambridge, MA). Benner MJ (2007) The incumbent discount: Stock market categories and response to radical technological change. Acad. Management Rev. 32(3):703–720. Benner MJ (2010) Securities analysts and incumbent response to radical technological change: Evidence from digital photography and internet telephony. Organ. Sci. 21(1):42–62. Benner MJ, Ranganathan RAM (2012) Offsetting illegitimacy? How pressures from securities analysts influence incumbents in the face of new technologies. Acad. Management J. 55(1):213–233. Bergstresser D, Philippon T (2006) CEO incentives and earnings management. J. Financial Econom. 80(3):511–529. Bernhardt D, Campello M (2007) The dynamics of earnings forecast management. Rev. Finance 11(2):287–324.


18


Bhojraj S, Libby R (2005) Capital market pressure, disclosure frequencyinduced earnings/cash flow conflict, and managerial myopia. Accounting Rev. 80(1):1–20. Brown KC, Brooke BA (1993) Institutional demand and security price pressure: The case of corporate spinoffs. Financial Analysts J. 49(5):53–62. Burgstahler D, Eames M (2006) Management of earnings and analysts’ forecasts to achieve zero and small positive earnings surprises. J. Bus. Finance Accounting 33(5–6):633–652. Bushee BJ (1981–2010) Institutional investor classification data. Accessed November 17, 2009, http://acct.wharton.upenn.edu/ faculty/bushee/IIclass.html. Bushee BJ (1998) The influence of institutional investors in myopic R&D investment behavior. Accounting Rev. 73(3):305–333. Bushee BJ, Noe CF (2000) Corporate disclosure practices, institutional investors, and stock return volatility. J. Accounting Res. 38(3): 171–202. Chapman CJ, Steenburgh TJ (2011) An investigation of earnings management through marketing actions. Management Sci. 57(1): 72–92. Chen WR (2008) Determinants of firms’ backward-and forward-looking R&D search behavior. Organ. Sci. 19(4):609–622. Cheng Q, Warfield TD (2005) Equity incentives and earnings management. Accounting Rev. 80(2):441–476. Chevalier JA (1995a) Capital structure and product-market competition: Empirical evidence from the supermarket industry. Amer. Econom. Rev. 85(3):415–435. Chevalier JA (1995b) Do LBO supermarkets charge more? An empirical analysis of the effects of LBOs on supermarket. J. Finance 50(4): 1095–1112. Cohen DA, Dey A, Lys TZ (2008) Real and accrual-based earnings management in the pre- and post-Sarbanes-Oxley periods. Accounting Rev. 83(3):757–787. Cohen DA, Mashruwala R, Zach T (2010) The use of advertising activities to meet earnings benchmarks: Evidence from monthly data. Rev. Accounting Stud. 15(4):808–832. Connelly BL, Tihanyi L, Certo ST, Hitt MA (2010) Marching to the beat of different drummers: The influence of institutional owners on competitive actions. Acad. Management J. 53(4):723–742. Coyne KP, Witter JW (2002) What makes your stock price go up and down. McKinsey Quart. 2002(2):29–39. Davis GF (2005) New directions in corporate governance. Ann. Rev. Sociol. 31(1):143–162. Davis GF, Useem M (2002) Top management, company directors, and corporate control. Pettigrew AM, Thomas H, Whittington R, eds. Handbook of Strategy and Management (Sage, London), 232–258. Degeorge F, Patel J, Zeckhauser R (1999) Earnings management to exceed thresholds. J. Bus. 72(1):1–33. Devers CE, McNamara G, Wiseman RM, Arrfelt M (2008) Moving closer to the action: Examining compensation design effects on firm risk. Organ. Sci. 19(4):548–566. Evans WN, Kessides IN (1994) Living by the “golden rule”: Multimarket contact in the U.S. airline industry. Quart. J. Econom. 109(2):341–366. Farrell KA, Whidbee DA (2003) Impact of firm performance expectations on CEO turnover and replacement decisions. J. Accounting Econom. 36(1–3):165–196. Finkelstein S, Hambrick DC (1988) Chief executive compensation: A synthesis and reconciliation. Strategic Management J. 9(6): 543–558. Foster L, Haltiwanger JC, Syverson C (2016) The slow growth of new plants: Learning about demand? Economica 83(329):91–129. Fudenberg D, Tirole J (1984) The fat-cat effect, the puppy-dog ploy, and the lean and hungry look. Amer. Econom. Rev. 74(2):361–366.


Gentry RJ, Shen W (2013) The impacts of performance relative to analyst forecasts and analyst coverage on firm R&D intensity. Strategic Management J. 34(1):121–130. Ghemawat P (1984) Capacity expansion in the titanium dioxide industry. J. Indust. Econom. 33(2):145–163. Ghemawat P (1991) Commitment: The Dynamic of Strategy (Free Press, New York). Gimeno J, Woo CY (1996) Hypercompetition in a multimarket environment: The role of strategic similarity and multimarket contact in competitive de-escalation. Organ. Sci. 7(3):322–341. Gomez-Mejia L, Wiseman RM (1997) Reframing executive compensation: An assessment and outlook. J. Management 23(3):291–374. Goolsbee A, Syverson C (2008) How do incumbents respond to the threat of entry? Evidence from the major airlines. Quart. J. Econom. 123(4):1611–1633. Graham JR, Harvey CR, Rajgopal S (2005) The economic implications of corporate financial reporting. J. Accounting Econom. 40(1–3): 3–73. Hall BJ, Murphy KJ (2003) The trouble with stock options. J. Econom. Perspect. 17(3):49–70. Hansen GS, Hill CWL (1991) Are institutional investors myopic? A time-series study of four technology-driven industries. Strategic Management J. 12(1):1–16. Harris J, Bromiley P (2007) Incentives to cheat: The influence of executive compensation and firm performance on financial misrepresentation. Organ. Sci. 18(3):350–367. Higgins MC, Gulati R (2006) Stacking the deck: The effects of top management backgrounds on investor decisions. Strategic Management J. 27(1):1–25. Holloway S (2008) Straight and Level: Practical Airline Economics, 3rd ed. (Ashgate Pub., Aldershot, UK). Hoskisson RE, Hitt MA, Johnson RA, Grossman W (2002) Conflicting voices: The effects of institutional ownership heterogeneity and internal governance on corporate innovation strategies. Acad. Management J. 45(4):697–716. Jensen MC (1986) Agency costs of free cash flow, corporate finance, and takeovers. Amer. Econom. Rev. 76(2):323–329. Jensen MC, Meckling WH (1976) Theory of the firm: Managerial behavior, agency costs and ownership structure. J. Financial Econom. 3(4):305–360. Jensen MC, Murphy KJ (1990) Performance pay and top-management incentives. J. Political Econom. 98(2):225–264. Jiambalvo J, Rajgopal S, Venkatachalam M (2002) Institutional ownership and the extent to which stock prices reflect future earnings. Contemp. Accounting Res. 19(1):117–145. Kaplan SN, Zingales L (1997) Do investment-cash flow sensitivities provide useful measures of financing constraints? Quart. J. Econom. 112(1):169–215. Kasznik R, McNichols MF (2002) Does meeting earnings expectations matter? Evidence from analyst forecast revisions and share prices. J. Accounting Res. 40(3):727–759. Ke B, Petroni K (2004) How informed are actively trading institutional investors? Evidence from their trading behavior before a break in a string of consecutive earnings increases. J. Accounting Res. 42(5): 895–927. Klemperer P (1987) Markets with consumer switching costs. Quart. J. Econom. 102(2):375–394. Kochhar R, David P (1996) Institutional investors and firm innovation: A test of competing hypotheses. Strategic Management J. 17(1): 73–84. Lederman M (2007) Do enhancements to loyalty programs affect demand? The impact of international frequent flyer partnerships on domestic airline demand. RAND J. Econom. 38(4):1134–1158.



Lang M, McNichols MF (1997) Institutional trading and corporate performance. Working paper, Stanford Graduate School of Business, Stanford, CA. Laverty KJ (1996) Economic “short-termism”: The debate, the unresolved issues, and the implications for management practice and research. Acad. Management Rev. 21(3):825–860. Levitt A (1998) The numbers game (September 28). Speech, NYU Center for Law and Business, New York. Lieberman MB (1987) Excess capacity as a barrier to entry: An empirical appraisal. J. Indust. Econom. 35(4):607–627. Litov LP, Moreton P, Zenger TR (2012) Corporate strategy, analyst coverage, and the uniqueness paradox. Management Sci. 58(10):1797–1815. Malmendier U, Tate G (2005) CEO overconfidence and corporate investment. J. Finance 60(6):2661–2700. Matsumoto DA (2002) Management’s incentives to avoid negative earnings surprises. Accounting Rev. 77(3):483–514. Murphy KJ (1999) Executive compensation. Handbook Labor Econom. 3:2485–2563. Murphy KJ (2002) Explaining executive compensation: Managerial power versus the perceived cost of stock options. Univ. Chicago Law Rev. 69(3):847–869. Mutzabaugh B (2014) Coming soon on JetBlue: Bag fees, more seats. USA Today (November 19), http://www.usatoday.com/ story/todayinthesky/2014/11/19/coming-soon-on-jetblue-bag-fees -more-seats/19266701/. Pearce JL, Huang L (2012) Toward an understanding of what actionable research is. Acad. Management Learn. Ed. 11(2):300–301. Phillips GM (1995) Increased debt and industry product markets: An empirical analysis. J. Financial Econom. 37(2):189–238. Porter ME (1992) Capital disadvantage: America’s failing capital investment system. Harvard Bus. Rev. 70(5):65–82. Prince JT, Simon DH (2009) Multimarket contact and service quality: Evidence from on-time performance in the U.S. airline industry. Acad. Management J. 52(2):336–354. Puffer SM, Weintrop JB (1991) Corporate performance and CEO turnover: The role of performance expectations. Admin. Sci. Quart. 36(1):1–19. Rajgopal S, Shevlin T (2002) Empirical evidence on the relation between stock option compensation and risk taking. J. Accounting Econom. 33(2):145–171. Roychowdhury S (2006) Earnings management through real activities manipulation. J. Accounting Econom. 42(3):335–370. Sanders WG, Hambrick DC (2007) Swinging for the fences: The effects of CEO stock options on company risk taking and performance. Acad. Management J. 50(5):1055–1078. Securities Industry Association (2002) Securities Industry Fact Book (Securities Industry Association, New York). Schipper K (1991) Analysts’ forecasts. Accounting Horizons 5(4): 105–121. Skinner DJ, Sloan RG (2002) Earnings surprises, growth expectations, and stock returns or don’t let an earnings torpedo sink your portfolio. Rev. Accounting Stud. 7(2–3):289–312.

19

Souder D, Shaver JM (2010) Constraints and incentives for making long horizon corporate investments. Strategic Management J. 31(12): 1316–1336. Souder D, Bromiley P (2012) Explaining temporal orientation: Evidence from the durability of firms’ capital investments. Strategic Management J. 33(5):550–569. Spence AM (1981) The learning curve and competition. Bell. J. Econom. 12(1):49–70. Stern CW, Deimler MS (2006) The Boston Consulting Group on Strategy, 2nd ed. (John Wiley & Sons, Hoboken, NJ). Westphal JD, Clement MB (2008) Sociopolitical dynamics in relations between top managers and security analysts: Favor rendering, reciprocity, and analyst stock recommendations. Acad. Management J. 51(5):873–897. Westphal JD, Graebner ME (2010) A matter of appearances: How corporate leaders manage the impressions of financial analysts about the conduct of their boards. Acad. Management J. 53(1): 15–44. Wiersema MF, Zhang Y (2011) CEO dismissal: The role of investment analysts. Strategic Management J. 32(11):1161–1182. Womack KL (1996) Do brokerage analysts’ recommendations have investment value? J. Finance 51(1):137–167. Wright P, Ferris SP, Sarin A, Awasthi V (1996) Impact of corporate insider, blockholder, and institutional equity ownership on firm risk taking. Acad. Management J. 39(2):441–463. Yanadori Y, Marler JH (2006) Compensation strategy: Does business strategy influence compensation in high-technology firms? Strategic Management J. 27(6):559–570. Zajac EJ (1990) CEO selection, succession, compensation, and firm performance: A theoretical integration and empirical analysis. Strategic Management J. 11(3):217–230. Zhang Y, Gimeno J (2010) Earnings pressure and competitive behavior: Evidence from the U.S. electricity industry. Acad. Management J. 53(4):743–768. Zuckerman EW (1999) The categorical imperative: Securities analysts and the illegitimacy discount. Amer. J. Sociol. 104(5):1398–1438. Zuckerman EW (2000) Focusing the corporate product: Securities analysts and de-diversification. Admin. Sci. Quart. 45(3):591–619. Zuckerman L (1999) 3 more airlines raise fares in 2d industry move this year. New York Times (March 15) 15.

Yu Zhang is assistant professor of management at China Europe International Business School. He received his Ph.D. from INSEAD. His research interests include the interaction between firm strategy and capital markets, competitive strategy, and corporate governance. Javier Gimeno is professor of strategy and Dirk Verbeek Chair in International Risk and Strategic Management at INSEAD. He received his Ph.D. from Purdue University. His research interests include competitive dynamics in multimarket contexts, the effect of organizational incentives and delegation on competitive interaction, entrepreneurship, and strategic risk management.