Earnings Expectations In The COVID Crisis

The coronavirus crisis may seem to have dragged on for an age, but the reality is that the initial shock hit markets and most economies less than 2½ months ago. Since then we’ve witnessed massive volatility in the equity market. The equally weighted S&P 500 fell 39% from its 15 February high to a March low, and then recovered 26.7% by 20 April, for an all-in decline of 22.9%.

A recently released paper by HEC’s Augustin Landier and MIT’s David Thesmar, looks at analysts’ downward revised earnings forecasts during this period, and it considers to what extent these changes were a factor in the massive equity market volatility during this period.

It finds that 10% of the S&P 500’s all-in decline is attributable to lower earnings outlooks, and the rest to investors demanding higher risk premia or discount rates.

Analysts Have Been Cautious But Rational About Revising Earnings

The study analyses annual earnings forecasts for the period 2020-24 for about 1,000 companies. At the beginning of the study period, earnings per share (EPS) was projected to grow about 13.5% annually. By April, EPS growth for 2020 was -4%, then gradually recover to 5% in 2021, 10% in 2023, and 11% in 2024. Analysts, in other words, are projecting only a gradual recovery in earnings and growth rates over several years – hardly a V-shaped recovery.

Earnings revisions have been relatively smooth over the study period, which the authors believe implies that analysts have probably underreacted to the COVID news. Other studies have shown that analysts tend to overreact to so-called salient news developments (of which COVID is a good example) that result in sharp market volatility.[1] The authors of the new paper document considerable variation in earnings forecasts over the next two years, implying a high degree of uncertainty about outlooks. There is less dispersion about longer term earnings forecasts. Normally, the relationship is the opposite – little dispersion about year-ahead forecasts, and more dispersion for longer term forecasts.

Chart 1: Dispersion of Forecasts

Source: Page 6 of “Earnings Expectations in the COVID Crisis“

Even if earnings forecasts appear sluggish and dispersed, they still make intuitive sense. Earnings forecasts by industry sector show the worst performers are real estate, industrials, financials and consumer discretionary; and the best performer are utilities and consumer staples. The authors also show that more leveraged companies are more impacted by downward earnings revisions (because fixed interest expense will be a bigger share of lower EBITDA).

Panel Regression Suggests Earnings Forecasts Do Matter

The heart of the paper is a panel regression analysis of company-level evolving earnings forecasts on equity prices over the study period. Essentially, earnings projections are discounted back to equity prices.

The regression results suggest that, based on earnings revisions alone, the unweighted SPX should have fallen about 10% over the study period instead of the actual 23%. They attribute the difference between forecast and actual prices to changes in the discount rate.

Chart 2: Implicit Discount Rates

Source: Page 13 of “Earnings Expectations in the COVID Crisis“

The authors go further and decompose the changes in the discount rate into three parts: change in the risk-free rate, change in risk premium, and change in leverage factor. As of the end of the study period the discount rate was about 1% higher than in February, due to higher risk premium (1%) and higher leverage (1%), offset by a lower risk-free rate (-1%). As the chart above shows, however, the relative contributions of these components has varied over time.

Watch For Updates!

The business press has attributed much of the market volatility over the past two months to swings in investor sentiment and extraordinary policy actions. This paper reminds us that there is an underlying rationality behind market moves – that market prices do reflect ongoing changes in earnings outlooks and risk preferences.

The paper implicitly assumes that equity prices are responding to analysts’ earnings revisions and other risk factors (through changes in the discount rate). Given the highly unusual (if not unprecedented) nature of this economic shock, one might wonder if causation might run the other way – that analysts are reacting to changes in equity prices. But that is probably a topic for another study.

Considering how volatile incoming information continues to be about the economy, companies and earnings, this paper is likely to be a living document in the form of ongoing revisions and updates.

[1] It is worth noting that as of the end of the study period most companies had not released first quarter earnings and had said little about the ongoing and potential impact of the coronavirus crisis, in part because of uncertainty but also because of the quiet period before releasing earnings. The paper does not mention this timing factor.