Progress in economics

I’m getting cheesed off with the economic culture war. One peeve I have with it is that economists do so many things that it’s easy to find examples of both bad work and good. Rather than speak in generalities, I’d prefer critics and defenders to focus more upon specific aspects of economics.

As an example of what I mean, I’ll take a bit of financial economics, as this might highlight what’s both good and bad in economics.

For years, we’ve had a mainstream theory here. The efficient market hypothesis says that all information is embedded into share prices and so you cannot beat the market without taking extra risk. And the capital asset pricing model tells us that the risk that matters is a share’s covariance with the market (or beta); idiosyncratic risk doesn’t matter.

Now, there is a distinction between these theories. From the off, the EMH arose from an empirical observation, that share prices seemed (pdf) random and the market was hard to beat*. Way back in 1953 Maurice Kendall pointed out the former (pdf). And in his excellent book Fischer Black and the Revolutionary Idea of Finance Perry Mehrling points out that Black had good evidence in 1967 that mutual funds under-performed the market. Even men who otherwise disagreed with the EMH shared this view. Benjamin Graham, for example, wrote (pdf) in 1972 that “the majority of the investment funds, with all their experienced personnel, have not performed so well as the general market.”

The CAPM, however, was less rooted in evidence and more a theory of equilibrium prices: in Bill Sharpe’s** paper (pdf) introducing the theory, there is pretty much no empirical evidence.

And in fact, there never has been any. Eugene Fama, the high priest of the EMH, has said that it “has never been an empirical success”.  Others have been less kind. Eben Otuteye and Mohammad Siddiquee say it has been “60 years of wasted effort". Economists have long known that beta isn’t the only determinant of returns, believing that size and value also matter – hence the so-called three factor model.

In fact, economists claim to have discovered countless “anomalies” – exceptions to both the CAPM and EMH. Many of these, however, are fragile: Campbell Harvey has said they are "likely false". And many others have diminished since they were pointed out – at least in the US and UK if not elsewhere – which is sort of consistent with a weaker version of the EMH: money-making opportunities don't last for long.

There seem to me, however, to be two pretty robust anomalies.

One is the tendency for low-beta stocks do better than they should and high beta ones worse. This was first pointed out by Fischer Black, Michael Jensen and Myron Scholes in 1972 but has since been corroborated by lots of other evidence (pdf). The other is that stocks which have done well in the past tend on average to subsequently out-perform. This has been found not just in shares but in other assets too. In my day job, I have constructed very simple defensive and momentum portfolios in real time. Both have for years beaten the market.  

There are, I think, some general messages to take from all this.

One is that the distinction between mainstream and heterodox doesn’t mean much. The first evidence against the “mainstream” CAPM came from “mainstream” economists.

Secondly, bad theories such as the CAPM are beaten not by competing theories but by facts. Of course, behavioural finance is an alternative to the CAPM. But what gives it credence is not the plausibility of its assumptions but the existence of some facts that are easier to square with behavioural finance than the CAPM. The momentum effect, for example, might well be due in part to investors’ tendency to under-react to news.

Thirdly, all this is a counter-example to what Jason Smith says. He claims that of both heterodox and mainstream economics that “neither appear to value empirical data. Neither appear to make accurate forecasts.” But this is not true in this case. There’s a vast library of research on empirical asset pricing, some of which – momentum and low-beta – survives the replicability crisis. And we do have some (so far!) successfully accurate forecasts, that momentum and low-beta stocks generally out-perform.  

Fourthly, my story is of a retreat from a single unified theory. Empirical evidence has undermined an elegant parsimonious theory (the CAPM) and left us with a few hard facts and a few mechanisms to explain them, such as constraints upon short-sales and leverage, and the presence of at least some cognitively-constrained investors. I think there’s been progress, but it’s away from grand theory and towards messy reality. Maybe this is – or will be – true of other parts of economics.

* I’m speaking here of the EMH as applied to individual stock prices rather than the overall level of share prices. It’s quite possible, as Paul Samuelson said, that stock markets are “micro efficient but macro inefficient.”

* Sharpe was only one of several fathers of the CAPM.

Comments

[Peter K.]

a footnote -
Myron Scholes, who went on to co-found the hedge fund Long Term Capital Management with Noble prize winner Robert Merton.

LTCM has to be bailed out by the New York Fed, lest its failure spur a crisis in the late 90s. This presaged the epic housing bubble and 2008 financial crisis.

The ruling ideas come from the ruling class. They must be doing something right b/c they have the money and power.

[Luis Enrique]

one small literature that I follow is misallocation, which is one attempt to add substance to that 'measure of what we don't know' - the Solow residual. It's been interesting to watch this evolve and get closer to the data, some recent papers suggesting that what was hitherto thought of evidence of misallocation is actually measurement error. Here are some examples:

http://www.econ.pitt.edu/sites/default/files/Udry_Heterogeneity%2C%20Measurement%20Error%2C%20and%20Misallocation.pdf

http://klenow.com/misallocation-mismeasurement-paper.pdf

https://sites.tufts.edu/neudc2017/files/2017/10/paper_325.pdf

this stuff is what my day to day experience of economics is like. It's hard to square with what I read about economics in the culture wars

[Jason Smith]

In my defense, I was thinking of macro models (GDP, unemployment rates, inflation) when I wrote that, not financial models (which are often agnostic about the underlying causes of the empirical regularties — in a good way!). In fact, I have taken a look at financial econ attempting to understand the empirical regularities with my own model:

https://informationtransfereconomics.blogspot.com/2016/12/stocks-and-k-states.html

[Blissex]

«I was thinking of macro models (GDP, unemployment rates, inflation) when I wrote that, not financial models»

Here lies an important reason why I think that our blogger's post here is entirely misguided, the argument that there is “both good and bad in economics”.

The orthodoxy in Economics is pretty much pervasive and dominating, and shapes careers with the necessity if "internal consistency". Consider for example:

http://www.bradford-delong.com/2018/05/warning-reading-marxs-capital-can-introduce-serious-bugs-into-your-wetware-hoisted-from-2006.html
«reading Karl Marx's Capital — something that, I am becoming convinced, should only be done by somebody with immunity to the mental virus — by a trained intellectual or social or economic historian, or by a trained neoclassical economist....»

Then there are many "something economy" fields that have been largely created to allow people to avoid challenging the orthodoxy and thus being ostracized by "top journals" and "tenure committees" and donors for having "anti-american" attitudes.
For example "behevioural economics" is the same as "microeconmics", but labeled differently; and various "something economics" studies are done in business schools that don't necessarily conform to the JB Clark orthodoxy.

If there is a claim that research into "something economics" show that there is something good in orthodox Economics that is at best naive and most likely propaganda, because those studies are done to work around and usually in opposition to Economics.

As to the specific case of finance economics, large parts, those that are used for right-wing propaganda, actually rely on and are meant to repeat Economics as in “more a theory of equilibrium prices”, other parts are empirical (mostly in business schools), fairly interesting and again done to work around Economics.

As to the details, there is not one EMH, but at least a dozen, and several are drastically different in terms of their import, never mind the generic EMH but “applied to individual stock prices rather than the overall level of share prices”.

[Nanikore]

Diane Coyle writes

"Our models are written with mathematical notation as a shorthand and to enforce logical consistency, just as historians use models—”the causes of the First World War”—but write them out with lots of words."

I would dispute this. Historians do not use models for very good reasons. An historian builds the story from the ground up. They use archival material, which critically includes unquantifiable information. There's a minefield of information out there - in fact there always has been for the genuinely inquisitive. Then you put the pieces of evidence together - and build up your story. You NEVER start with a model or theory - even if your finding confirms a model or theory later. But almost certainly your new material (including unquantifiable evidence) will bring new perspectives. And almost certainly the most interest things cannot be modelled.

I am pleased to see that one person was constructing a large data set on colonial railways - but this is the norm in history departments - not the exception. I have not seen this particular paper, but I wonder how deeply it really goes into the history.

[Blissex]

«constructing a large data set on colonial railways - but this is the norm in history departments - not the exception»

But the study of the "longue duree" and quantitative history ("cliometry") have some important advantages:

* They are dull boring work, so it does not have a direct political value to potential sponsors. Economists useful to sponsors like L Summers get $600,000 base salary at Harvard, or like G Hubbard get £1,200 per hour consultancy fees, it is hard for historians to be so richly rewarded. N Ferguson has had to veer into Economics to achieve his usefulness.

* That kind of work originates in continental Europe, where professors are usually civil servants and content to draw an upper-middle class salary and to be "public intellectuals", and where sponsors don't have an overwhelming influence on the funding of departments (reminder: Ken Lay of Enron "fame" funded 35 [thirty five] endowed chairs in his name).

* "Internal consistency" with JB Clark's "3 fables" (or clever working around that requirement by doing "something economics") is not essential to career progress in the field.