What S&P 500 Earnings Estimates Are Not

by: Brian Gilmartin, CFA

Bob Doll, Nuveen's excellent strategist here in Chicago, was on CNBC Wednesday here and he talked about the S&P 500 possibly not seeing as a much as a 10% correction given that "earnings are too strong".

Bob could be right, but judging from some comments I've read on this blog and some readers from Seeking Alpha, it's important for readers to understand that the S&P 500 earnings and earnings data is NOT a market timing tool.

S&P 500 earnings data or the forward estimate won't suddenly turn down and then the S&P 500 will correct 10%.

Analogous to an airline pilot that assiduously studies the weather map for the planned flight path, S&P 500 earnings data give me and readers a good idea of what's ahead (noting both the degree and timing of changes in estimates) while remaining vigilant en route.

Truthfully, I know very few indicators that really are perfect timing tools.

Here are some some aspects to watching earnings data that have occurred to me over the years that readers might benefit from:

1.) Timing: The two most glaring examples of strong S&P 500 growth and contradictory S&P 500 price action are 1994, when the US economy was very strong, S&P 500 earnings grew 20% in calendar 1994, and the S&P 500 was flat on the year. In 2011, the S&P 500 corrected 20% mid-year, peak to trough, while the S&P 500 earnings grew 15% that year, thanks to Energy and Technology.

Maybe the best example is that from Jan. 1, 2012 through Dec. 31, 2016, the S&P 500 rose 94%, while the total cumulative earnings growth for that same period was roughly 20%, with the collapse of Energy starting in late 2014 and the serial write-downs in Financials post 2008.

S&P 500 earnings and market action can diverge markedly over shorter periods of time.

2.) Exogenous shocks aren't captured well: I've talked about this before, on this blog (but it has been a few years), but the 2008 Financial Crisis didn't start to show up in the "forward 4-quarter S&P 500 estimate" until July 2008, or within 3 months of the Lehman collapse. The S&P 500 has peaked in October 2007. The severity of that Financial Crisis never showed up in the numbers until it was too late. The S&P 500 earnings estimates are good predictors of S&P 500 earnings recessions, or slowdown, but some exogenous event that comes from outside the system can't really be predicted by the earnings data. This also gets to the point - when talking to clients - about what can and cannot accurately be predicted, but that's a topic for another time.

3.) EPS estimates themselves can be distorted: Trinity keeps a couple of hundred excel spreadsheets on individual companies (not all are kept constant) and using two recent examples, J.C. Penney's (NYSE:JCP) EPS estimates were updated after reporting in November 2017. Note how for JCP, the sum of the quarterly bottom-up estimates do not add up to the mean consensus for fiscal 2018 and fiscal 2020.






















The quarterly estimates sum to ($0.25) for fiscal 2018 and $0.19 for fiscal 2019.

Source: Thomson Reuters I/B/E/S estimates as of 12/9/2017

Another example is General Motors (NYSE:GM)















Source: Thomson Reuters I/B/E/S estimates as of 12/9/2017

For GM, the sum of the quarterly estimates is $6.30, while the mean consensus estimate of 16 analysts for 2019 is $5.83.

As I've talked to Thomson Reuters analysts over the years that have compiled the data (even John Butters, who is now at FactSet, was running "This Week in Earnings" for Thomson for many years), that mean consensus estimate for the full year (whether calendar or fiscal) does or can contain top-down estimates, although that is more prevalent for the S&P 500 itself. (The many sell-side strategists throwing out an S&P 500 earnings estimate for a calendar year without perhaps supporting it with quarterly estimates.)

4.) Trends in revisions usually matter more than absolute estimates: One thing I've learned over the years is that trends in revisions (analyst raising or lowering forward estimates, particularly at the sector level) are far more important than the absolute earnings numbers for the S&P 500.

The concept of "PE expansion" and "PE contraction" has a "yuge" (sic) impact on market returns while overall trends in underlying estimates remain constant.

Summary/conclusion: It may seem strange to attack the basic premise of this blog, but readers need to be aware that simply following S&P 500 earnings estimates and trends is helpful, there is no "magic formula" for market timing or stock picking.

When using any kind of quantitative model for investing, it is always best to be aware of the limitations.

Thomson Reuters data (by the numbers, 12/8/2017)

  • Fwd 4-qtr estimate: $142.54
  • PE ratio: 18.6x
  • PEG ratio: 1.7x
  • S&P 500 earnings yield: 5.38%
  • Year-over-year growth of the forward estimate: +10.92% vs. last week's +10.79%

The moral of the story today is to use S&P 500 earnings data as a helpful guide, but as always, price and price trends of individual stocks, securities, ETFs, and asset classes matter. This weekly blog forces me to stay focused on an important individual stock and valuation tool, but not rely on it exclusively.

Pay attention to sector info, too. Estimates and changes to forward estimates are very telling in my opinion, at the sector level.

What I've found over the years is that when the "forward 4-quarter estimate" is rising steadily as it is now, more often than not it portends positively for S&P 500 "PE expansion" and the bull market.

One other aspect to this that has been thought about is that with the constant rotation within Wall Street itself, and the constant moving of analysts you/we probably have another flock of analysts and portfolio managers on Wall Street that have never seen a real bear market. I do wonder whether sell-side analysts today are still leaning bearish after last decade, and are reluctant to raise numbers or most have now seen a 9-year bull market and are more prone to lean bullishly on estimates.

It still feels like more the former than the latter.

Hope this helped.