(Mis) leading indicators, data surprises and Keynes' beauty contest

This is a very long and somewhat belated post, inspired by a 17 August blog by Mark Thoma on The Use of Leading Economic Indicators in Economic Forecasting. Mark cites a paper by Stock and Watson and a Bloomberg column by Caroline Baum, which asked why economic forecasters don’t use leading indicators? It's a good question.

Baum asked a slightly different one, which is why are almost all US market economists seemingly bullish on growth when the Index of Leading Economic Indicators (LEI) "is signaling slower, not faster, growth ahead"?

The direct but partial answer to Baum's question is that many market economists haven't been giving the Conference Board series much credence lately. Barry Ritholtz, for example, thinks their Index is juiced data:

The Board was apparently not pleased that 8 of the 10 past LEIs were negative. Hey, if you don’t like what the indicators are suggesting, than why not just change the model?  And that’s exactly happened. .. the Conference Board reduced the utility of LEIs for investors. Their work now falls into the category of economic cheerleading.

In a subsequent post he lambasts the Conference Board for their revisions, saying "this fantasy-based indicator is now less than worthless - it is actively misleading."

On the strange and bizarre planet Conference Board, a flattening yield curve is somehow economically stimulative. For this crowd of absurdists, it takes a full yield curve inversion to start raising concerns about any potential slowing.

See James Hamilton's 4 August post for a good discussion on why the Board changed its treatment of the yield curve in its index. The broader question raised by Thoma and Baum, though, is why so much weight seems to be given to 'high frequency' data? Why are market economists swayed by the flow of daily and weekly data? There are several plausible explanations.

First, timeliness. Most market economists and strategists will argue that monthly indicators like the LEI just aren't timely enough. Financial markets react almost instantly to data releases, and the main LEI components are published beforehand. So, in theory at least, by the time the monthly LEI is released they should already have been priced in. It is 'old news'.

Second, frequency. A monthly series may be OK in some markets, but in a place like the United States which has weekly measures of jobless claims, they want more, sooner thanks. Hence the obsession with weeklyinitial jobless claims. There is also the ICSC weekly US retail chain store sales index, and my personal favourite, the Economic Cycle Research Institute Weekly Leading Index.

Why use a monthly leading index when you have a weekly one to hand? Of course weekly data has more 'noise' and is hence more volatile. But it is still used. That's because when you're on a trading floor if you are the first to identify a turning point - or mispricing by the market - you can make a lot of money. Conversely, if you too often miss those shifts in sentiment you can lose plenty.

So how to make sense of the torrent of data? One commonly used approach is to construct a measure of the upside or downside 'surprise' contained in data releases - how the numbers vary from the market consensus. Many banks construct such measures, which are a form of diffusion index. They do not attempt to measure economic activity itself, but rather the momentum behind the flow of news on activity. If they continually print stronger (more pro-growth) than expected the currency may perform better than previoulsy expected, and market views about future central bank moves could become more hawkish. The opposite effects obviously apply if the 'surprise index' shows weaker-than-expected data.

Helpfully, RBC Capital Markets have posted a note explaining how they contruct their surprise index. Quantifying and Trading Economic Surprises (PDF), written by currency strategist Adam Cole in September 2004, explains the rationale thus:

Our aim is to find a way of spotting turning points in market sentiment and to develop systematic trading strategies based on economic surprises. ...Economists have a natural tendency to defend existing views as the evidence begins to contradict them and only capitulate when the evidence is really convincing. ...Our contention here is that inertia in economists’ views can lead to a tendency to dismiss news as noise and to miss turning points, in market sentiment if not in activity.

The real surprise with such measures is how often they work, considering how crude they are. I have used similar measures in my high-frequency trading models, and they had a higher success rate than models using monthly leading indicators. This is not because they are methodologically or conceptually superior. They aren't. It's because they usefully flag potential changes in the market's economic expectations, which in turn effects sentiment. Moreover, they do it sooner than less frequent measures such as monthly leading indicators.

John Maynard Keynes put it best in The General Theory (1936, p.156) when he likened playing the stock market to judging a beauty contest. What he he had in mind were contests that were popular in England at the time, where a newspaper would print 100 photographs and people could write in and say which six faces they liked most. Everyone who picked the most popular face was automatically entered in a raffle, where they might win a prize. Keynes wrote:

It is not a case of choosing those [faces] which, to the best of one’s judgment, are really the prettiest, nor even those which average opinion genuinely thinks the prettiest. We have reached the third degree where we devote our intelligences to anticipating what average opinion expects the average opinion to be. And there are some, I believe, who practise the fourth, fifth and higher degrees.

Those who are able to figure out where market sentiment on the economy (and interest rates) is shifting before most of their competitors do can make money. Lots of it. The current post-Katrina period, where there is considerable divergence of opinion about the outlook for the US economy and the Fed funds rate, is a time where the run of data will be more carefully scrutinised than usual for that very reason.

It is important to note that for traders, whether market sentiment is ultimately proved right or wrong is secondary. What really matters is to anticipate shifts in market sentiment - to figure out what view the market is likely to arrive at, not what view it ought to have.

This is why economists working on a trading floor are frequently exasperated by the seemingly irrational or ever-changing views held by some traders. Unlike economists, their job is not to work out where the economy is actually headed. It is to try to work out where the market thinks it's going, and in particular to anticipate changes in economic sentiment ahead of the rest of the market. For that very iterative task you need a flood of data and some pretty smart analysts. Surprise indexes sometimes point the way for a shift in market sentiment; leading indexes seldom do. They just aren't timely enough.

Comments

[Edward Hugh]

The really interesting thing is the way people react to data which are then subsequently revised. ie they are practicing fine-tuning on a pretty rough-and-ready (blurry) version of reality. At the same time those who are glued to their screens watching all the up- and down-ticks probably have some implicit virtual model in their head that they are working from :).

Don't you know that all the good billiard players are notoriously bad at formal geometry?

[spencer]

Having wall street economist make estimates of what an individual economic release will be is a "suckers" game created by the brokerages houses to generate volume.

The brokerage houses make their money on volume and do not really care if the volume is in an up or a down market.

I know many people who are paid to generate such forecast,
but I have never heard of a single one who had enough confidence in their own forecast to go out and start their own firm betting on individual economic releases.

This strongly implies that no one is consistently right
and the original statement that it is a suckers market is accurate.

[Roehl Briones]

In (what I think is an important) paper, Pierre Monnin shows evidence that such "higher order beliefs", i.e. beliefs about beliefs, are responsible for introducing volatility into the stock market.