“Consensus…is something in which no one believes and to which no one objects” Margaret Thatcher
The financial world seems to place a huge amount of importance on consensus estimates. Following each set of results, the first reaction of every sell-side analyst and investor is to compare the results – particularly sales and EPS – to consensus expectations. Subsequently, when the results are reported in the financial media, the primary focus of the discussion is on comparison to “The Street”. To the outside world, this seems like a sensible way to judge the results of a company and hold management to account. Yet this veneer of respectability disguises a hot mess of shifting opinions with limited credibility or value. The closer you get the uglier it looks, leaving you to question if it’s worth paying attention to consensus numbers at all.
The Case Against Consensus
There is plenty of evidence to show that sell side analysts are a poor judge of the future. A McKinsey study published in 2001 showed that analyst forecasts are consistently too optimistic. Between 1985 and 2000, forecasts 12 months forward overestimated actual EPS by an average of 10%. And the further out the forecast, the bigger the gap. Forecasts 3 years out exceeded actual EPS by 22%. This corresponds with my own experience as a fund manager of too many bullish analysts who want to show strong growth and margins several years out, but are then forced to reign in their expectations as the future draws closer.
In fact, by the time the company comes to report, the same bullish analyst want to make sure their companies will meet or beat numbers. So they bring back their estimates to the point at which they know the company won’t disappoint. These analysts are constantly highlighting a beat in the recent quarter, then finding reasons to pare down overly-bullish numbers for the next quarter or next year. This is the equivalent of the old magician’s trick – get the audience to focus intensely on where they think the magic is happening, while the real trick is happening somewhere else entirely (the closer you look, the less you see!)
An experienced analyst can always find ways to downgrade numbers without drawing too much negative attention – by hiding it within a good quarter, referencing technical reasons (e.g. tax rate change or disposals), referencing externalities (e.g. commodity prices) or by presenting it as a positive (e.g. reinvestment of profits). Alternatively they do it when most investors are on vacation.
Experienced buy-siders know that consensus numbers are not always a true representation of what the market is thinking. And in many cases they form their own “buy-side consensus” that can differ significantly from the sell-side view. This can happen for two reasons
- The buy-side consensus reflects a more “pure” view of numbers. They have no interest in trying to promote the stock or in gaming earnings estimates. They are genuinely trying to guess where the numbers will land.
- The sell-side tends to be quite slow in updating their estimates. They don’t want to have to publish a revision every week. The buy-side has no such constraints and will automatically factor in new information as it becomes available.
The problem is that buy-side consensus is not a published number. The market might have a sense where it’s different to the sell side, but nobody can say exactly what it is.
In recent years platforms like Estimize have grown in popularity – by taking a sample from a wider group of market participants (including the buy-side) they are likely to give a more accurate view than sell-side consensus. This is not to say that the buy-side always does a good job of forecasting the future – generally speaking we humans do a lousy job of that. But it is more likely to be a fairer representation of what the market is current thinking.
Does a miss matter?
As suggested by the McKinsey research above, the accuracy of consensus data seems to becomes greater as we get closer to results day. By the time it comes to forecasting the next quarter, management has usually guided the market within a close range and analysts can hardly fail to get it right, meaning it’s unusual for a company to experience a large miss either way.
Another McKinsey study corroborates this view. In a study entitled “Avoiding the consensus-earnings trap”, they show that a small miss or beat versus consensus has little impact on the share price over the following 5 day period. Furthermore in 40% of cases they found that the share price moved in the opposite direction of the earnings miss. Some of this reflects that buy-side consensus may settle in a different place to sell-side consensus (as argued above). However much of it comes down to the fact that the beat/ miss is usually marginal and has little impact on the long-term value of the company.
Other research by McKinsey shows that the market’s reaction to results is far more impacted by changes to long-term expectations than by whether the company beat or missed that quarter’s numbers.
This makes intuitive sense– an approximate view of earnings in 3 years time is far more valuable than a precise view of earnings in the quarter just gone. This is why intelligent investors are focusing their attention on the change to longer-term expectations while the dumb money on CNBC is still deciding whether it was a miss or a beat.
So why bother with them?
So if short-term consensus doesn’t matter and long-term consensus is always wrong, why bother looking at it at all? This is a question that divides the value-investing community, but I would argue that it pays to have an awareness of where consensus numbers sit.
The reason for looking at estimates is less about their value as a forecast, but more as an indicator for the market mood. While bearing in mind that most long-term forecasts are over optimistic, there will be some forecasts that are clearly unachievable and some that set a much lower hurdle. The extent of extrapolation tells us a lot about the mindset of the market. In many cases consensus has set up these companies for failure. As it becomes painfully obvious that the company’s future is not as bright as expected, the numbers get rapidly downgraded and the stock falls out of favor. Conversely where expectations are low, it doesn’t take much to see upgrades come through and sentiment turn around.
You should exercise caution here – as we suggested above consensus estimates are not always a reliable indicator of what Mr Market is thinking. But we can use consensus as one indicator of the market’s view. Where other indicators are pointing the same way (extreme optimism or extreme pessimism) it may well indicate that an opportunity is at hand.
“You cannot ignore the market – ignoring a source of investment opportunities would obviously be a mistake – but you must think for yourself and not allow the market to direct you” Seth Klarman
What Does Management Think?
There is another reason you may need to be aware of consensus expectations – because management may also be paying attention to them. Some management teams go to such lengths to meet consensus expectations that they change entirely how the company is managed. A research study published in the Journal of Accounting and Economics in 2005 sampled 400 CFOs, 78% of who admitted to sacrificing long-term value to smooth earnings. 41% of the CFOs sampled would reject an NPV-positive project if it led to a short-term miss of $0.10c on consensus EPS of $1.90. If management is acting against the best interests of the shareholder in an attempt to meet short-term earnings, it’s a management team that you would do best to avoid.
Towards an Independent View
We may look back one day soon and wonder why we gave so much credence to a handful of analysts sat in New York and London with a proven track record for getting it wrong. Technology is now enabling a new generation of independent analysts to contribute to the debate in a way that wasn’t possible 5 years ago. This is something that we believe in passionately at StockViews – we encourage our analysts to come to a view that’s independent of what the sell-side is thinking. More often than not it will be a more accurate and a more valuable view. Over time we see sell-side consensus becoming less representative, and therefore less reliable. So pay attention to consensus estimates where it helps you to form a view of sentiment, but don’t get too attached. That’s best left to the pundits on CNBC.