Is It Time to Get Rid of Earnings-per-Share?

Do U.S. companies focus too much on short-term profits at the expense of long-term investments, profits, and growth? There is considerable debate among academics, practitioners, and politicians about the relevance of short-termism, its possible sources, and potential mechanisms to mitigate it.

In a recent article, I propose that earnings-per-share (EPS) targets are a very likely driver of short-termism and discuss what we can do to break the link between EPS targets and short-termism. Earlier survey evidence shows that managers admit to short-termism driven by earnings management—they are willing to sacrifice positive net present value long-term investments to meet earnings goals.[1] Recent empirical research in corporate finance and accounting shows that managers do as they say. EPS targets affect stock repurchases, R&D investment, capital expenditures, employment, and M&A deals.

For example, suppose that analysts expect a company to report EPS of $3.00 a share, but the company’s managers learn during the quarter that EPS will actually turn out to be $2.98 a share. One might imagine that managers would not care much about this small shortfall in profits. After all, an EPS of $2.98 a share is economically very similar to an EPS of $3.00 a share. However, rather than reporting an EPS of $2.98 (or guiding analysts towards the lower number), some managers prefer to make sure that reported EPS is above $3.00 a share by going beyond simple accounting tricks such as managing accruals. Empirical research shows that managers also engage in opportunistic cuts in R&D investments and increase stock repurchases to make sure reported earnings are above the analyst consensus forecast.[2] These EPS-driven stock repurchases are also associated with subsequent cuts in capital expenditures and employment, presumably to help finance these repurchases, which are very substantial.

EPS targets other than the consensus analyst forecast also seem to affect corporate decision-making. Payouts or vesting schedules for equity and non-equity grants are commonly tied to achieving a specific EPS goal. These goals create “kinks” in managerial compensation—compensation jumps if performance is just above the target specified in the contract.  A recent paper shows that firms are much more likely to exceed EPS goals by a small margin than to miss the EPS goal by the same margin.[3] EPS targets also seem to affect M&A decisions. Deals that increase the acquirer’s EPS (e.g., accretive deals) appear to be more attractive to acquirers. Recent research shows for example that acquirers shift from stock to cash financing to assure EPS accretion.[4]

This literature also suggests that the practice of chasing EPS with changes in real investments leads to long-term under-performance and lower economic growth. Even opportunistic cuts in investments that are reversed in the future can have significant economic costs. A recent paper shows that because of decreasing returns to scale in innovation, the practice of cutting R&D to meet the current EPS target and then increasing R&D in the future (that is, an increase in the volatility of R&D) leads to lower economic growth and welfare losses that are of similar magnitude to those engendered by business cycles, trade barriers, or inflation.[5]

What can be done to break the link between EPS targets and short-termism? The low-hanging fruit is to change executive compensation practices. Not only is EPS the most common target in compensation contracts, but a significant fraction of compensation contracts also specify EPS as the only performance target. Other researchers have formulated specific proposals to address this problem.[6] For example, it seems clear that companies should use compensation targets that are also based on sales and profit rather than just EPS. It would also be useful to specify a smooth link between performance and pay, rather than discrete targets at which either the level of compensation or the pay-for-performance sensitivity jump.

However, it is not clear that eliminating EPS-linked compensation will completely break the link between EPS targets and inefficient long-term investment decisions. That is because missing EPS targets has large negative effects on stock prices. Evidence suggests for example that the cumulative abnormal stock return from one day before to one day after the earnings announcement is 0.6% higher for firms that just meet the analyst EPS forecast than for firms that just miss the forecast (independent of the magnitude of the surprise). Given this evidence, it is no wonder that managers have strong incentives to meet the analyst EPS forecast. As long as executives care about stock prices, they will continue to have incentives to inflate EPS and meet EPS targets such as the analyst consensus forecast, even if EPS-based compensation is somehow scrapped.

Thus, I believe it is time to consider whether we should get rid of earnings-per-share (EPS) altogether. Analysts, investors, and companies should move away from focusing on EPS as a measure of performance. That focus is a bit puzzling. I am not aware of any theoretical or empirical evidence that explains why EPS should be the  measure of performance. It is much easier to point out problems with EPS, as academics tend to do in their MBA and other business school courses. This focus on a relatively poor measure of performance is not just a harmless blunder—it has real consequences for long-term investments, growth, and welfare.

My article discusses three related proposals for shifting away from EPS targets. First, we can replace EPS with superior measures of profitability such as ROA (return on assets). The most important step in defining ROA is to use lagged assets in the denominator. For example, one would measure profitability in the quarter by dividing end of quarter operating income (or another measure of profits) by assets measured at the beginning of the quarter. This measure of profitability cannot be manipulated using stock repurchases, since repurchases during the quarter do not reduce assets measured at the beginning of the quarter.

Second, analysts and companies can focus on multiple measures of performance, rather than trying to summarize performance using only EPS. For example, analysts can summarize their forecasts in two or more estimates (say ROA and sales), rather than in just one. If firms also have to meet a specific revenue target rather than just a profit target, adjusting R&D is no longer enough to meet the targets.

Third, analysts and companies can start focusing on forecasting ranges rather than point estimates. For example, rather than saying that EPS should be exactly equal to X, each analyst could summarize his or her forecast by saying that ROA should be between Z and Y. The consensus forecast could also be reported using a range that reflects the distribution of analyst forecasting ranges. Naturally, there may still be incentives to manipulate performance when firms are about to miss the lower end of the range. In that case, having multiple measures of performance would help. If analysts, companies, and investors replace EPS targets with multiple targets based on other profit and sales range forecasts, the link between performance targets and short-termism is likely to be weakened.

How would we convince managers, analysts, and investors to implement these changes in financial forecasting? I believe that academic research has a role to play. My article discusses several open topics for research on EPS targets and short-termism. Our own teaching in business schools can be another lever to change this equilibrium. Many academics have been emphasizing for years that EPS is a poor measure of performance. However, the evidence on the negative consequences of EPS targets is relatively recent and has not yet made its way into finance textbooks and most business school classes. It is striking for example how prevalent the usage of NPV in capital budgeting has become. One possible explanation is the uniform and consistent focus that business schools have given to NPV techniques. I hope that my research will convince some of my colleagues to discuss these issues in greater detail in their classes and textbooks and help steer future managers and analysts into a better equilibrium.

ENDNOTES

[1] Graham, J., C. Harvey, and S. Rajgopal, 2005. The economic implications of corporate financial reporting. Journal of Accounting and Economics 40: 3–73.

[2] Almeida, H., V. Fos, and M. Kronlund, 2016. The real effects of share repurchases. Journal of Financial Economics 119: 168–185.

[3]Bennett, B., C. Bettis, R. Gopalan and T. Milbourn, 2016. Compensation goals and firm performance. Journal of Financial Economics 124 : 307-330.

[4] Dasgupta, S., J. Harford and F. Ma, 2018. EPS Sensitivity and Merger Deals. Unpublished manuscript.

[5] Terry, S., 2017. The macro impact of short-termism. Unpublished manuscript.

[6] Gopalan, R., J. Horn, and T. Milbourn, 2017. Comp Targets that Work. Harvard Business Review, 95: 102 – 107.

This post comes to us from Professor Heitor Almeida at the University of Illinois at Urbana-Champaign. It is based on his recent article, “Is It Time to Get Rid of Earnings-per-Share (EPS)?”, available here.

1 Comment

  1. RJ

    Interesting analysis. For the short-term, requiring analysts to publish a range of EPS estimates is possible and the SEC can easily do it. However, only relying on ROA might encourage firms to undervalue their assets to show a higher ROA.

Leave a Reply

Your email address will not be published. Required fields are marked *