Manipulating consensus forecasts
Most of my readers will remember the Libor scandal uncovered in the aftermath of the financial crisis. Back then it emerged that fixed income traders of several banks conspired to submit artificially high or low Libor rates to enhance their chances of making a profit on their trades. Because Libor was and is a truncated mean of all the submissions of participating banks, issuing an egregiously high or low number (even though there was never a single trade executed at that level) meant that the published Libor rate could be shifted ever so slightly.
This scandal not only led to large fines for banks but also to the abolition of Libor as a reference benchmark rate for mortgages, swaps, and other fixed-income instruments as of 2022. In the UK, the Libor will be replaced by SONIA, in the United States by SOFR and in the Eurozone by ESTR, to name a few.
But apparently, Libor is not the only rate that is actively being manipulated. Andrew Call and his colleagues are the first to document efforts by bearish equity analysts to manipulate consensus forecasts. They looked at c. 370,000 individual analyst forecasts between 2004 and 2018 and noticed that there are some bearish analysts who submitted an extremely high forecast just before the earnings announcement of a company. Even though these analysts are bearish on the stock, they revised their forecasts for sales, earnings or cash flow upwards above the most optimistic forecast by other analysts. What this does is shift the consensus number higher and thus makes it more likely that the company will miss consensus forecasts and suffer a decline in its share price.
Number of unbeatable forecasts between 2004 and 2018
Source: Call et al. (2020).
And these efforts to manipulate consensus forecasts aren’t rare either. They found 28,524 instances where an analyst has a sell or underweight rating on a stock but issued the highest forecast of all analysts covering the stock. That’s close to 20% of all bearish forecasts in their sample. They then subtract instances where the majority of analysts covering a firm made upwards revisions in their earnings forecasts and instances of upward revisions where the company eventually beat the bearish analysts’ forecasts anyway and ended up with 7,064 unbeatable forecasts (about 5% of all bearish analyst forecasts). They found that about 12% of all firm forecasts were subject to at least one such unbeatable forecast. The likelihood of an analyst making such an unbeatable forecast rises if the analyst is under pressure to justify his or her bearish view or when the stock is covered by fewer analysts (and hence the impact of an unbeatable forecast is bigger).
IBES and other aggregators of analyst views try to catch these unbeatable forecasts and eliminate forecast revisions that look like a data error or are too egregious. Yet, according to Call and his colleagues, they exclude only about 14% of all unbeatable forecasts from their data. In the end, issuing an unbeatable forecast means raises the likelihood of a company missing consensus estimates by a whopping 21%. So, if you use analyst consensus forecasts in your investment decisions, you might want to review all individual forecasts, especially for companies followed by fewer analysts.