Tim Harford wants forecasters to take wagers: most of them already bet their livelihoods

In a nice post, Tim Harford explains why it would be desirable for economic pundits or forecasters to place bets based on their punditry and forecasting, citing a recent £1000 wager between Jonathan Portes and Andrew Lilico.  It would put a premium on precision, and it would incentivise the forecaster to make the best forecast possible, not simply one that is sexy or dramatic, and grabs the headlines.  As forecast consumers, the rest of us, worried about whether to get a fixed or floating rate mortgage, buy a car or not, would have forecasts that were more useful.  In fact, my last post covered a recent example.  We can view Mark Carney as a ‘pundit’.  And his recent remarks speculating about the timing of the next interest rate rise as a forecast.  Wouldn’t it be better if the BoE took bets that the forecasts these words were presumably based on were right?

Well, yes.  But what Tim doesn’t say in his post that most forecasts are underpinned by bets of sorts.  Many forecasters work for retail banks, investment banks, economic consultancies or hedge funds.  Their forecasts are backed by two kinds of bets. The first is a reputational bet.  Some of the motive for employing high-profile economic commentators from the outward-facing of these institutions (ie not hedge funds) is about showing off that you have great wise staff on your team.  You clearly don’t want to get a reputation for employing dumb forecasters who always say something that turns out wrong.  The second kind of bet is quantitative, no different really from the bet you would place at the Bookmakers on a horse.  Many of these forecasters are employed, directly, or indirectly [the forecasts are bought and consumed by] traders who are placing bets all the time on the prices of government bonds in different countries, or anything really, which are sensitive to how macro data turn out.  So good forecasts are essential.  Many of my former colleagues at the Bank of England left to engage in this kind of activity and stress a great deal about the calls they make from week to week about what the BoE or other central banks do, because they know they can be quickly fired if they have a run of bad calls.

I tweeted this point back to Tim, and Tomas Hirst replied with the very reasonable point that reputational discipline isn’t necessarily good enough to guarantee the best possible forecasts.  In fact there’s an old and interesting literature arguing that in these reputational competitions there can be an incentive to make a deliberately bad, but striking forecasts, to differentiate oneself and cultivate a brand, presumably when verifying whether the forecast was good or bad ex post is a murky science, as it often is.  [This would be one way of making sense of the otherwise puzzling fact that highly qualified data analysts can coexist that either always make doveish remarks (Blanchflower) or always make hawkish remarks (Sentance) no matter what the data release].

So granted, reputational bets may not be sufficient to give us the forecasts we would like.  But these bets are ubiquitous, and may be helpful.   Indeed, I think it’s precisely because institutions like the Bank of England know that they would be betting their reputations that they are wary of publishing an interest rate forecast, and instead wrestle with signalling what they are doing through more opaque tactics.

[Postscript:  one contact responded to this post saying that traders in his firm regularly invited him, Harford-style, to back his in-house forecasts with a bet.  His standard reply was that if he were to bet, he should bet against his forecast, to hedge against the enormous reputational risk of his forecast turning out wrong.  Likewise, Jonathan Portes and Andrew Lilico ought to be betting on the other to win, to compensate themselves for the loss of face, and associated future earnings, in case they lose the forecast horserace.]

 

Advertisements
This entry was posted in Uncategorized. Bookmark the permalink.

3 Responses to Tim Harford wants forecasters to take wagers: most of them already bet their livelihoods

  1. Roger Sealey says:

    Maybe you should read: ‘The Three Rs of Economic Forecasting Irrational, Irrelevant
    and Irreverent’ by Edgar R Fiedler CB in Economic Research Across the board pp 62-63 published in June 1997. In which he states:
    15. A forecaster’s best defence is a good offence, so: If you have to forecast, forecast often.
    16. But: If you’re ever right, never let ’em forget it.

  2. Claudia Sahm says:

    I too am skeptical that money wagers on forecasts will help improve quality much. I think the money would be better spent on independent forecast evaluation. Point forecasts are often uninteresting (as they are always conditional on something) … it is the model, the covariance structure, the distribution of outcomes that matters a lot more. Having a better sense of which models or frameworks have a better track record in which settings or time horizons would be more helpful. Also I think de-personalizing the forecasts would be better but wagers go in the opposite direction. As some money managers know there is this thing called “luck” and we need not pay professional forecasters for that too.

  3. daniels says:

    There are potentially similar problems here to trying to directly derive probabilities of economic events from financial derivative prices (Hartford’s proposal amounts to requiring forecasters to trade Arrow Debreu securities), also based on the idea that financial markets provide the best baseline forecasts for the likelihood of e.g a market crash. Heterogeneity in risk aversion either at the psychological level or at the institutional level in terms of varying degrees of market incompleteness make it hard to extract physical event probability assessments as opposed to risk neutral probabiliities. And what if there is a positive correlation between proximity to rational expectations and risk aversion? After all, a more risk averse analyst may have be more motivated in “to cover all the bases” by paying the mental costs of gathering more information and processing more intensely. Then, better forecasters. may be systematically more cautious in their bets. Is there any way to adjust for that?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s