Alex Douglas says that economics “makes very few successful predictions”: that “nothing from economics has improved economic policy in the way that the integrated circuit has improved computing”: and accuses economists of making “logical mistakes.”
I don’t intend to reject his claims, but I’ll give a counter-example from my day job. Back in the early 00s, I read Narasimhan Jegadeesh’s and Sheridan Titman’s paper which claimed that stocks which had risen in previous months tended to continue to rise. I interpreted this as a prediction – that momentum stocks would tend to out-perform in the UK. So I tested it. At the end of each calendar quarter, I took an equal-weighted basket of the best-performing shares* from the FTSE 350 to see how they would perform in the following quarter. And they have hugely out-performed. Since I began the exercise in 2004, UK momentum stocks have tripled in price, whilst the 350 has risen only just over 50 per cent.
Jegadeesh and Titman therefore made a successful prediction – and a useful one too. In fact, researchers have uncovered momentum effects pretty much everywhere they have looked: in commodities, currencies, international stock markets and sports betting.
This is not the only useful and successful prediction in financial economics. The efficient market hypothesis might be much derided (and the great performance of momentum stocks is exhibit A against it**) but it has made a prediction – that investors shouldn’t expect active managers to beat the market after fees – which has been successful.
Or to take another example, back in 1985 Hersh Shefrin and Meir Statman claimed (pdf) that investors lost money because they were too quick to sell winning stocks and too slow to sell losers. Subsequent research by Brad Barber and Terrance Odean has found that this prediction has also been right.
These, I would contend, are only a few of the ways in which financial economics has provided us with good practical advice – successful predictions, if you like.
All this raises the question. How come these predictions have been successful and useful when Alex claims otherwise?
One big reason is that (aside from the EMH) these predictions haven’t been derived in the way that Alex characterises economics – as logical deductions from theoretical principles. Instead, they were gleaned from a careful search for facts. Were Jegadeesh and Titman doing data-mining? I don’t give a damn: they discovered a useful and replicable result.
In fact, this is how economics in general has progressed in recent years: it has become increasingly empirical.
This is not to say that Alex is attacking a straw man. There are still some theory-driven, fact-blind economists lingering on. : I fear that some heterodox economists with their ambitious theoretical attacks upon “mainstream economics” fall into this camp as much as some dogmatic neoclassicals.
However, insofar as economics is useful and progressive, it is not because of the development of grand theories but because of humble fact-gathering. In economics, I am a hardline Blairite: what matters is what works.
* Initially I measured the best performers over the previous three months, but latterly over the previous 12. Consistent with Jegadeesh and Titman’s claim, it doesn’t much matter which you take.
** Exhibit B is the out-performance of defensive stocks – a fact which explains much of Warren Buffett’s success (pdf).
who knows? I mean really, who knows what influence economists have had over economic policy, and what economic policy and outcomes would have looked like in the absence of that influence?
Posted by: Luis Enrique | March 08, 2016 at 04:40 PM
Isn't the problem that predictions that result from this kind of empiricism have the risk that they work ... until they don't.
I mean it was empirical data that showed that we had entered the Great Moderation, but that didn't stop the subsequent crash.
Posted by: gastro george | March 08, 2016 at 04:49 PM
@ Gastro - yes, that's a danger. It can be mitigated by both data and theory. Eg in the case of the Great Moderation, we had a theory that it mightn't last (Minsky's financial instability hypothesis and/or the "it was just luck" theory) plus decades of empirical evidence from around the world that stability doesn't last.
In the cases I've taken, nobody believes that any stock-picking strategy always works; the best we can do is out-performance on average over the long run. Momentum has, for example, under-performed so far this year.
Posted by: chris | March 08, 2016 at 05:47 PM
Gaussian cupola - Mathematics valid, conclusions drawn from the cupola incorrect, result financial crisis.
As Sir Mervyn King says:
The Biggest Monetary Stimulus the world has ever seen, eight years later still no sustainable recovery.
An epic fail, for economics.
But as for Mervyn's insurance. One word AIG. AIG provided insurance at cents on the dollar for Mortgage Back Securities (MBS) and AIG collapsed when the crisis struck.
We still face the prospect of Zero Interest Rate policy for several more years, even without another financial crisis, and the warning lights are flashing red (e.g Dry Bulk Shipping Index).
Your example seems relatively trivial, and may be down to human behavior. The best prediction of what will happen tomorrow, is what happened today. But when it doesn't it can be catastrophic.
What makes you think the concentration of power would produce socialism rather than feudalism. After all feudalism is about the massive inequality of power.
Posted by: aragon | March 09, 2016 at 01:35 AM
I'm with gastro george here. It isn't sufficient to say that stocks/currencies that are going up one month are likely to go up the next, you need to know what happens when they go down.
Some currencies famously go up by the lift and down by the lift shaft. Sop most of the time they are going up, but when they go down they go down really quickly. This means the value of the stock/currency is influenced by the risk of a quick collapse. Has this been covered in the analysis?
Posted by: Dipper | March 09, 2016 at 10:23 AM
@ Gastro, Dipper. You're right that crash risk can matter. But it doesn't seem to be a factor in momentum. Eg, looking at the ten worst three-monthly performances, the average fall in my momentum basket has been 22.4%, but the average fall in the ten worst three month performances for the FTSE 350 was 24.8%. Something else therefore explains momentum's outperformance. If you're looking for risk factors rather than behavioural ones, benchmark risk is a strong contender.
Posted by: chris | March 09, 2016 at 10:45 AM
N.B [empirical economics] are like oracles: they make a pronouncement of truth, yet give no explanation of that truth.
I have a couple of shibboleth for economists.
The issue Alex raises: The balanced budget.
"Willem Buiter’s "Joys and Pains of Public Debt" explores the question of fiscal sustainability within the framework of standard economics."
The state gives money value (fiat money), why is the state constrained by money (a balanced budget)?
What the state giveth, the state taketh away. The state has the power of coercion.
We can ignore the last trump and the infinite case as and the mathematics/logic as the state is temporal (relating to worldly as opposed to spiritual affairs; secular).
Where do you stand on intellectual property ?
The current patent/copyright regime is clearly a nonsense.
It turns out that Coases theorm does not apply as the externalities are positive.
(I may find the paper if necessary)
I could come up with more...
Intellectual property, and the Investor State Dispute in TTIP. Economics is just a smoke screen for the exercise of power.
Not a source of enlightenment.
Posted by: aragon | March 09, 2016 at 11:40 AM
I think it's great that empirical economics is outpacing theoretical economics. But... what if the empirical data is on flimsy foundations? There have been big issues raised about psychological studies for instance:
http://andrewgelman.com/2016/03/05/29195/
Posted by: Doc at the Radar Station | March 09, 2016 at 01:32 PM
chris - I have two points in reply to your comment but they do contradict each other.
The first is to whether the period you have measured is sufficient to arrive at a conclusion. If I run across a motorway ten times and survive, that does't mean running across a motorway is safe.
The second is that governments round the world acted to support the economy in the banking crisis, so momentum investing relies on governments bailing you out when your stocks crash, rather than on some magic formula.
Posted by: Dipper | March 09, 2016 at 05:41 PM
The throw back to Alex Douglas is to say, well you come up with economic theories that do result in the sort of gains seen in computing!
And then we will see that economic problems are of a different magnitude and type to that of computing problems.
The major logical error of Alex Douglas is to think the 2 are in any way comparable.
Posted by: An Alien Visitor | March 09, 2016 at 06:42 PM
@doc
It's funny you should mention that, because replication of empirical economic research has been similarly unsuccessful: http://www.federalreserve.gov/econresdata/feds/2015/files/2015083pap.pdf
I agree with gastro that we have to be cautious about extrapolating previously successful predictions. I'd also point out that 'economics' as a field contains a lot of different theories about financial markets and little ex ante reasons to prefer one theory or another. Sure, knowledge of an area combined with an eye to empirical evidence and some intelligence/intuition - all of which you no doubt have, Chris - will help you to pick out better ideas. But such an ability is not inherent to the practice of 'economics'.
Posted by: UnlearningEcon | March 13, 2016 at 09:22 PM
What a curious post. Your blog is brimming with insights from economic theory that are non-empirical in nature! And for the better; I probably would have stopped reading a long time ago if you were a simple pattern-sniffer.
Posted by: Sam G | March 14, 2016 at 09:23 PM