Importance of Catering Incentives for Growth Dynamics

Abstract

This paper tests whether the dynamics of firm growth metrics, such as sales and investment growth, are consistent with firms catering to the market by delivering growth when stock prices are more sensitive to growth-related news. After developing growth valuation premium measures, we document four main results consistent with catering theory. First, time periods of high growth premium are followed by higher-than-expected growth indicators. Second, catering to the premium is more pronounced for firms whose managers care more about maximizing short-term stock prices. Third, firms whose managers care more about short-term stock prices have higher time-series volatility of median sales, investment, and PPE growth. Finally, conditional trading strategy based on timing the revenue growth premium yields 26 basis points per month after adjusting for risk and postearnings announcement drift.

Keywords:

• Catering
• Revenue surprises
• Growth
• Market reactions

ACKNOWLEDGMENTS

We would like to thank Aydogan Alti, Itztak Ben-David, Michael Boldin, Keith Brown, Lorenzo Garlappi, Jay Hartzell, Eric Hirst, Rabih Moussawi, Luis Palacios, Paul Tetlock, Sheridan Titman, Roberto Wessels, Brandon Cline, Kelly Carter, and all UT Austin Finance and WRDS Research Seminar participants for valuable suggestions.

Notes

1. There is also evidence that the stock market affects financial decisions, such as dividend policy (Baker and Wurgler [2004a, b], Lie and Li [2006]).

2. “In July of last year, Amazon.com reached an important way station. After four years of single-minded focus on growth, and then just under two years spent almost exclusively on lowering costs, we reached a point where we could afford to balance growth and cost improvement, dedicating resources and staffed projects to both” (Amazon's 2001 letter to shareholders). See Aghion and Stein [2008] for detailed discussion.

3. The Economist, February 3, 2011. “Never had it so good: How much longer can corporate America keep on delivering bumper increases in profits?”

4. Note that we study sensitivity of stock prices to growth information, not the relation between the level of stock prices and investment. Unlike in case of financial variables where one can imagine a scenario when investors prefer dividends to capital gains or vice versa, all else equal, investors always want growth. However, the emphasis investors place on incremental increase in growth might be time-varying. Indeed, our measures of growth premium vary over time but are always positive, while the dividend growth premium in Baker and Wurgler [2004a, b] takes on negative values.

5. See Narayanan [1985], Stein [1988], Bebchuk and Stole [1993], and Holmstrom [1999] for discussions of managerial short-termism.

6. The source of time variation in the premium is a separate research question and is beyond the scope of this paper.

7. This effect compares well with the explanatory power of aggregate GDP growth: contemporaneous one-standard-deviation change in GDP growth is associated with roughly 43% of one-standard-deviation change in sales growth changes.

8. We tried different time horizons and obtained qualitatively similar results.

9. See Stein [1996], Subrahmanyam and Titman [2001], and Shleifer and Vishny [2003].

10. Our primary aggregate measure of time varying relative investor demand for high growth firms is constructed using market response to revenue surprises. A similar measure using investment surprise is noisy because investment growth is not known at the time of an earnings announcement. Polk and Sapienza [2009] use the difference in price-to-book, which does not fully reflect market response to investment, in their time-series tests.

11. The positive relationship between revenue surprises and earnings announcement returns is well documented in the literature. However, this study shows that this relationship changes overtime. Revenue growth premium used in our analysis captures this time-varying preference for growth. Using the level of revenue surprises in our cross-sectional tests in place of revenue growth premium would have made our analysis directly comparable to cross-sectional tests in Polk and Sapienza [2009].

12. As Mary Sammons, president of Rite Aid, a leading drugstore chain, says: “You have to do the right things to keep and grow market share, and yet you have to do it smartly so you aren't giving away all the margin to do it” (http://www.findarticles.com/p/articles/mi_m3374/is_1_24/ai_82137618).

13. We test all of the predictions of the A-S model (other than those pertaining to expected returns) only for growth regimes and refer to the remaining regimes as “nongrowth.” A-S theory assumes when the firm is not in growth regime, it is in the margin regime: “the firm's manager must decide how to allocate her effort between two business strategies: (i) a “growth” strategy of expanding production and sales and (ii) a “margins” strategy of maintaining sales while improving profit margins (say by reducing unit costs)” (Aghion and Stein [2008], p. 1028). However, firms can pursue many other strategies aimed at sustaining current levels of growth and/or profitability and therefore would not fall under either “growth” or “margins” strategy envisioned by Aghion and Stein [2008]. For example, a firm can find it optimal to focus on product quality, consumer relations, supply chain management, change of product/service mix, change of production technology, employee development and retention, and so forth.

14. However, the paper does not examine whether market reaction is consistent with market knowing more than the manager. Their evidence can be also consistent with managers catering to potentially not fully rational market reaction.

15. Ultimately, a firm's growth efforts in terms of increased research and development expenditures and acquisitions are meant to boost total revenues. Therefore, we look at the final outcome of different managerial actions aimed at higher growth: growth in revenues.

16. See Ertimur and Livnat [2002] for further motivation of using change in the profit margin as a measure of cost controls.

17. Jegadeesh and Livnat [2006] document the correlation of 0.26 between earnings and revenue surprises in their 1987–2003 quarterly sample.

18. All firms with an available market value (size) at the beginning of a quarter are classified into ten groups according to size. The size-adjusted return is the return on an individual company minus the equally-weighted return on the portfolio of all firms in the same size decile. We exclude firms with prices below $5.00 on the date of earnings announcement to avoid the bias caused by outliers. All independent variable (SUE, SUREV, and ΔIncToSales) are standardized to mean 0 and standard deviation of 0.065 in order to make regression coefficients comparable over time.

19. Market value is computed as a product of the price of firm's stock the day following earnings announcement and shares outstanding as reported by Compustat, whereas the book value is calculated at the end of fiscal quarter (see the appendix for detailed variable definitions).

20. Polk and Sapienza [2009] employ similar approach for constructing investment premium.

21. High positive correlation between and by itself can be viewed as supporting catering theory. Catering theory suggests that the manager should care about growth (or dividend) premiums only if the market rewards her for considering the demand for growth (dividends). However, Baker and Wurgler [2004] do not find significant correlation between the average announcement effect of recent dividend initiations (which is similar in spirit to our ) and their primary measure of dividend premium (which is analogous to in our framework). Lie and Li [2006] note that the lack of positive correlation between these two measures of dividend premium “raises doubts about the empirical validity of the catering theory” (p. 294).

22. Revenue growth premia proxies and future returns difference continue to be consistently negatively correlated (though, insignificant for one-year horizon) if future returns are calculated over one- and three-year horizons (Futcumret12 and Futcumret36).

23. Note that by construction, does not suffer from this problem because it captures marginal effect of revenue surprises controlling for changes in profit margin.

24. The results are robust if instead of five years, we chose other intervals such as three or four years.

25. Aghion and Stein's [2008] theory assumes that when the firm is not in a growth regime, it is in a margin regime: “the firm's manager must decide how to allocate her effort between two business strategies: (i) a ‘growth’ strategy of expanding production and sales and (ii) a ‘margins’ strategy of maintaining sales while improving profit margins (say by reducing unit costs)” (p. 1028). However, firms can pursue many other strategies aimed at sustaining current levels of growth and/or profitability and therefore would not fall under either “growth” or “margins” strategy envisioned by Aghion and Stein [2008]. For example, a firm can find it optimal to focus on consumer relations, supply chain management, change of product/service mix, change of production technology, employee development and retention, etc.

26. Unexpected sales growth is defined as the difference between its actual and expected value given a set of firm characteristics. Expected sales growth is obtained using Fama-MacBeth methodology. Every year from 1964 to 1980, we estimate cross-sectional regression of sales growth on concurrent characteristics such as age, size, market-to-book, earnings and industry dummies, and compute the average of cross-sectional coefficients for each firm characteristic across these 16 years. The expected sales growth is computed for the firm i in year t as the sum of product of these coefficients and firm characteristics from 1964 to 2003. The residuals from these regressions are aggregated across all firms in the cross-section to obtain yearly time-series of unexpected sales growth.

27. Economic significance of revenue growth premia is understated because revenue growth premia have only time series variation while all other variables have both time series and cross sectional variation.

28. We perform our tests on two measures of investment: (a) change in Capex scaled by lagged PPE (Panel A) and (b) change in CapEx scaled by lagged Assets (Panel B). Results are qualitatively similar.

29. Polk and Sapienza [2009] find a positive relation between investment levels and discretionary accruals controlling for investment opportunities and financial slack and argue that this supports the idea that overpriced (underpriced) firms tend to overinvest (underinvest). In their framework, firms are catering in their investments to firm-specific levels of mispricing, whereas we test whether the overall market's response to revenue surprises (rather than individual firm's mispricing) influences firm's real-side activity.

30. Yermack's sample includes all firms which qualified for at least one of the four Forbes magazine lists of the 500 largest public U.S. corporations (the lists rank firms by sales, profits, assets, and market value) in at least four of the eight years between 1984 and 1991. A firm must also have been publicly traded for four consecutive full fiscal years in the 1984–1991 period. We sincerely thank Prof. Yermack for generously providing this data.

31. Unexpected sales growth is defined as residuals in the cross-sectional regression of lead sales growth on a set of firm characteristics such as current sales growth, (log) market-to-book ratio, (log) total assets, (log) age of the firm, ROA, cash flow, median analyst forecast of ROAt+1 and a set of firm-specific dummies.

32. This relation is robust to controlling for current asset base, age, and market-to-book ratio.

33. See Ball and Brown [1968], Foster, Olsen, and Shelvin [1984], and Bernard and Thomas [1989] for detailed description of PEAD.