It is undisputed that the last 2 quarters have demonstrated an impressive jump in corporate earnings growth, if mostly due to a beneficial base effect from plunging 2016 earnings which pushed them below levels reached in 2014. And naturally, this rebound has been more than priced into a market which has seen substantial multiple expansion since the Trump election to boot. But what is much more important for the market is what corporate earnings look like in the future, and it is here that Bank of America has just raised a very troubling red flag.
According to BofA's Savita Subramanian, in November the S&P 500's three-month earnings estimate revision ratio (ERR) fell for the fourth consecutive month to 0.99 (from 1.03), indicating that for the first time in seven months, there were more negative than positive earnings revisions, needless to say a major negative inflection point in the recent surge in profits. The bank's more volatile one-month ERR also weakened to 0.94 (from 1.16).
A breakdown of EER by sector showed a sudden and broad-based deterioration, as the three-month ERR weakened across eight of the 11 sectors, with Materials, Health Care, and Financials seeing the biggest declines while, not surprisingly, Tech and Energy have the highest three-month ERRs, with Energy's ERR expanding the most on the back of rallying oil prices. Meanwhile, Telecom, Real Estate, and Discretionary have the weakest ratios, and November saw a drop in the Health Care and Industrials' EER ratio below 1.0 (meaning more cuts than raises to earnings forecasts) for the first time since March. Furthermore, while two sectors have been "improving" in recent months: namely Energy and Tech whose ERRs have been rising; all other sectors have seen their ERRs roll over.
Why is this significant?
As BofA explains, the three-month S&P 500 ERR is used by the bank as one of its 19 key "bear market signposts", and with the one-month ERR falling below 1.0 for the second time in six months, this marks the trigger for the 11th bear market signpost. BofA's ERR rule is triggered when, over a six-month window, all of the following criteria are met: 1) the one-month ERR falls from above 1.0 to below 1.0; 2) the one-month ERR is below 1.0 for two or more months; and 3) the three-month ERR falls below 1.1 for at least one month.
Incidentaqlly, the hit rate of the "ERR" bear market indicator, meaning its historical accuracy in predicting a bear market is 100%, the only question is how long it takes. The last time this trigger was set was mid-2003, and here is the punchline from Bank of America:
Since 1986, a bear market has followed each time that the ERR rule has been triggered. While individual signposts may not be useful for market timing (this one was triggered several years too early in the last two cycles), prior bear markets were preceded by a broader array of signals having been triggered.
This is shown in the chart below:
Ok so one indicator out of 19 now is flashing "bear market" dead ahead. That's hardly bad if the rest are all green, right? Well, they aren't.
As discussed two weeks ago, Bank of America recently compiled a list of bear market signposts that have always occurred ahead of bear markets. No single indicator is perfect, and as Subramanian wrote, "in this cycle, several will undoubtedly lag or not occur at all." And while single indicators may not be useful for market timing, they can be viewed as conservative preconditions for a bear market. In this context, the suddenly "triggered" ERR indicator is one of the bank's 19 bear market signposts.
Here a caveat is warranted: in the last two cycles, the ERR rule was triggered several years too early (Chart 3 above), although a bear market followed each time that the ERR rule has been triggered. As for timing, the more signposts triggered, the greater the risk of an imminent bear market, in BofA's view. And in November, the one-month ERR falling below 1.0 for the second time in six month marked the 11th trigger (out of 19). This is shown in the table below.
It also means that nearly two-thirds of Bank of America's bear market indicators have now been triggered. As Subramanian concludes "every cycle is different, but we expect to see more signposts triggered before the eventual market peak."
Then again, this remains a "market" where even if 12 out of the 11 indicators are triggered, it may just send the S&P limit up as there is no point in selling if all that does is guarantee another central bank bailout.