It is increasingly obvious that the market's obsession with short-term earnings has pushed many companies into unwise or cosmetic business practices. Tactics such as accelerating revenue recognition and managing accounting reserves to smooth earnings are based on the assumption that investors will pay a premium for steady and predictable earnings growth. As confirmation, some point to Jack Welch, whose management of GE's accounting reportedly smoothed the company's earnings and earned its stock a so-called Jack Welch premium.1 Executives today regularly cite stable earnings growth as a reason for strategic actions. For example, the CEO of Conoco justified its pending merger with Phillips Petroleum in part by asserting that the merger would offer greater earnings stability throughout the commodity price cycle.2
It would seem to make sense that the market should place a higher value on companies that exhibit steady profit growth quarter after quarter, year after year. Yet in today's environment, in which financial accounting is closely scrutinized, emerging anecdotal evidence indicates that the market is increasingly suspicious of overly smooth and predictable earnings. When companies beat analyst estimates by exactly one penny, for example, quarter after quarter, it is typically their lack of variability that raises eyebrows.3
So the fundamental question remains: does the market value steady and predictable profit growth, and if so, how much? We examined how earnings stability was related to total returns to shareholders (TRS), the market-to-book value ratio, and the price-to-earnings (P/E) ratio. We found that variability in earnings growth rates had little effect on TRS. It is true that as earnings stability decreases, TRS declines, but the magnitude of these effects are small, especially when compared to the effect of fundamental drivers like earnings growth and return on invested capital (ROIC). Similarly, greater earnings stability results only in a small increase in market-to-book ratios and has no effect on a company's P/E ratio.
We also found that going through accounting contortions just to meet analyst expectations is overrated. While the short-term effects of meeting or missing projections can be significant, fundamental performance drivers are much more important over time. Comparing actual earnings results to the consensus predictions from analysts, we found that while positive earnings surprises led to positive excess returns4 and negative surprises to negative excess returns, the relationships to excess returns were very weak.
Most firms have earnings growth variability between 0% and 40%.
Earnings stability and shareholder returns
To understand the link between earnings stability and shareholder returns, we measured the variability of growth in earnings per share (EPS) for more than 550 companies5 between 1996 and 2000 (Exhibit 1). We compared each company's annual earnings growth to its average earnings growth over the entire period and noted each company's growth variability.6 Walgreens' earnings per share, for example, rose steadily from $0.32 in 1995 to $0.77 in 2000. Its average annual growth rate over that time was 19 percent, with variability of four percentage points, i.e., 15 percent to 23 percent. The higher a company's variability range, the more its annual earnings vary up or down from its average growth over time.
When adjusted for underlying performance—that is, the effect of earnings growth and ROIC—the relationship between earnings stability and market performance measures (TRS, market-to-book ratio, and P/E ratio) was weak. Long-term earnings growth and ROIC combined explained 51 percent of TRS for the entire sample over a five-year period, but earnings stability explained only an additional 2 percent of TRS.
We then sorted the data into low- and high- earnings growth groups to isolate the effects of earnings stability from that of earnings growth. For both groups, firms with higher variability do see a decline in TRS (Exhibit 2), but the effect is slight. For example, if a firm with an annual TRS of 13 percent and earnings growth variability of 30 percent were to increase its variability even by as much as 20 percent,7 its annual TRS would decrease only to 12.4 percent.
Increased earnings growth variability does not affect TRS significantly in low-growth firms or high-growth firms.
Nor do companies whose earnings vary more widely receive lower valuation multiples. An increase in variability of about 20 percent (as in the previous example) would result in only a 2 percent reduction in the market-to-book value multiple. There was no measurable effect on the P/E ratio.
Meeting analyst expectations and shareholder returns
Managers often lament having to set aside attractive investments in long-term growth opportunities in order to meet near-term earnings expectations and avoid a possible decline in share price, despite having themselves set expectations in the first place. We tested 735 companies8 between 1991 and 2000, comparing the effect on TRS of discrepancies between actual results and analyst forecast estimates at 12 months and 3 months out.
As would be expected, forecasts are more accurate the closer they are to the release date of actual results9 and investor response to over- or underperformance indicates their awareness of the impact of timing. Indeed, if a company sets earnings expectations of $1.00 a few weeks prior to the actual earnings release, and then delivers only $0.90, investors should be disappointed. Such a shortfall calls into question the internal controls and systems of the company more than anything else. Yet while the effect of missing a three-month forecast can be significant, the effect of missing a forecast made a year ago should be weaker, since negative performance can be ascribed to many factors such as changes in the economy, interest rates, technology, and competition. A 10 percent gap between actual results and three-month forecasts leads to approximately a 4 percent drop in the share price over the three-month period of the forecast, whereas a shortfall of 10 percent between actual results and 12-month forecasts leads to approximately a 1 percent drop in the share price over the 12-month period of the forecast (Exhibit 3).
The relationship between excess returns and predictability is very weak.
Yet over- or underperformance relative to forecasts still does not explain most of a company's actual share price change. Obviously, meeting expectations is only one of many factors driving share prices and investors do look beyond a company's performance relative to expectations to the reasons behind its performance. For example, when Arrow Electronics announced first-quarter earnings of 30 cents per share at the end of April 1999, it fell short of analyst estimates by 34 cents. The company attributed its underperformance to oversupply in its sector, to pricing pressures, and to higher interest expenses following two acquisitions. Yet in the three-month period from April to June, Arrow had excess returns of 14 percent compared to the S&P 500. The market, presumably, was focused on expected performance improvement as oversupply conditions eased.
McKinsey research has found that the effect of missing analyst forecasts is shaped by three factors: whether or not the deviation was expected; the degree of surprise (or the amount by which forecasts were missed); and whether the deviation was anticipated to be temporary or long term. Indeed, the market reacts three times as strongly to unanticipated news with long-term impact than to anticipated news with long-term impact or unanticipated news with short-term effects. This has important implications for company executives whose communications with analysts, investors, and the media are a crucial part of shaping expectations and perceptions of long-term performance. Managers should typically be more forthcoming rather than less in order to ameliorate the impact of surprises.
Meeting short-term expectations may be important, therefore, but it is not paramount. The market constantly evaluates a business's long-term potential, and investor reaction to short-term earnings surprises is really linked to whether the underlying cause has the potential to alter long-term performance.
TRS and long-term earnings predictability
When we examined the effect of long-term earnings predictability on TRS, we found that while companies whose earnings are more predictable did earn higher cumulative returns, the correlation was weak. We defined a long-term measure of predictability as the average difference between actual results and 12-month estimates over ten years.10 We then compared it with ten-year cumulative returns.
Among nearly half of the firms we analyzed, actual earnings performance averaged within 20 percent of annual predictions over ten years. A 20 percent difference in average performance relative to annual predictions corresponds to only an 8 percent difference in the cumulative excess return over the same period, or 0.8 percent annually.
When we isolated earnings predictability from earnings growth, we found that predictability explained only an additional 4 percent of shareholder returns beyond what EPS growth and ROIC change could explain. The weakness of that correlation is easily illustrated by high-predictability firms with low returns and vice versa. For example, both McDonald's and Sherwin Williams produce notably predictable earnings, with each company's earnings varying on average less than 3 percent from annual forecasts over the ten years we studied. But they have underperformed the S&P 500 on a risk-adjusted basis, with ten-year cumulative excess returns of -21 percent and -57 percent, respectively. On the other hand, Southwest Airlines and Biogen are both considerably less predictable, varying annually by nearly 30 and 50 percent, respectively, on average from forecasts. Southwest fell below one-year forecasts in three and Biogen in five of the ten years. Yet both of those companies have outperformed the S&P 500 with ten-year cumulative excess returns of 109 percent and 164 percent.
Companies should pay attention to short-term earnings, but they should not go through accounting contortions to do so. Even for firms renowned for delivering steady year-on-year double-digit earnings growth, shareholder returns can be largely explained by the magnitude of their overall long-term earnings growth rate. Only a small portion of the TRS is assignable to a premium for the lack of variability in their earnings. Actions to improve short-term results that involve borrowing from next quarter's performance perpetuate a short-term mind-set, and put companies in an increasingly perilous wager that performance in future quarters will not only cancel previous quarters' deficit, but will also show growth. Markets will increasingly call their bluff.