How Big Should the Financial Services Industry Be ?

Robert Waldmann

James Surowiecki has an interesting article on this topic in the May 11 New Yorker.

He reaches rather precise conclusions including

“The desire to bring back the boring, small banking industry of the nineteen-fifties is understandable. Unfortunately, the only way to do that would be to bring back the economy of the fifties, too.”

and outsourced to Thomas Philippon, an economist at N.Y.U.,
” a normal financial sector would be about the size it was in 1996.”

He reaches these conclusions by confusing causation with correlation, deducing something about the location of an unobserved variable from a correlation, assuming that 19th and 20th century financial markets must have been efficient and reaching a conclusion by not considering any alternative. Altogether a methodologically interesting article.

The key phrase in the article is “accounted for” which is equated with caused by.
This is a dead give away. I review the article and point out the areas which I find unconvincing after the jump.

Before going on I must stress that I am commenting on Surowiecki not Philippon. I haven’t read Phillipon’s article.

The sentence in which the key phrase appears is “To be sure, deregulation was also a factor, but Thomas Philippon, an economist at N.Y.U., has shown that most of the increase in the size of the financial sector in this period can be accounted for by companies’ need for new capital.”

By the end of the next paragraph this becomes “Each wave, Philippon shows, was propelled by the need to fund new businesses, and each left finance significantly bigger than before. In all these cases, it wasn’t so much that the bankers had changed; the world had.” So “accounted for” has become “propelled by” and correlation has become causation.

Importantly, there is no claim that the volume of financial services was equal to the level required to fund new businesses. The same pattern would be found in the data even if the level of financial services was a billion times that needed to fund new businesses. So long as the ratio is constant it doesn’t matter what it is. A claim about the location of a variable (the actual level is about equal to the socially necessary level) can’t be based on a correlation.

In particular, even if the causation is technological progress leads to an increased number of new firms and then to an increase in financial activity, there is no evidence that the increased financial activity was necessary or helpful. An increase in the amount of prey causes an increase in the number of predators. This does not mean that the predators are helpful to the prey. If the financial services industry were purely parasitic, one would expect the same pattern.

So far this is obvious and neither Surowiecki nor especially Philippon would be the least bit discomfited if they read the above criticisms. I have just said that the evidence they present does not amount to proof.

The last two methodological errors are, however, interesting. They are also closely related. My first claim is that the argument that variables must reflect the effect of the real economy on the financial sector is based on the efficient markets hypothesis — that Surowiecki assumes what he seeks to prove. My second claim is that the most popular alternative to the efficient markets hypothesis explains the patterns.

The evidence surveyed by Surowiecki is partly broadly historical. “There have been three big banking booms in modern U.S. history. The first began in the late nineteenth century, during the Second Industrial Revolution, when bankers like J. P. Morgan funded the creation of industrial giants like U.S. Steel and International Harvester. The second wave came in the twenties, as electrification transformed manufacturing, and the modern consumer economy took hold. The third wave accompanied the information-technology revolution.” This is a very crude correlation. In particular, the 60s are left out, although they were a period of rapid economic growth. The late 19th century is vague. Surowiecki asserts that the 21st century is different as he considers the recent expansion of finance to be quite different from earlier ones.

The more quantiative evidence in Surowiecki’s article is the following

This meant that the economy became, and has remained, far more competitive, while corporate performance became far more volatile. In the nineteen-eighties, companies moved in and out of the Fortune 500 twice as fast as they had in the fifties and sixties. Suddenly, there were lots of new companies with big appetites for outside capital, which they needed in order to keep growing. And it was Wall Street that helped them get it. Companies like Turner Broadcasting, M.C.I., and McCaw Cellular used junk bonds to turn themselves into major businesses. Venture-capital investing took off, and so did the I.P.O. market; there were twice as many I.P.O.s between 1980 and 1999 as there were between 1960 and 1979.

So the evidence is that there were more I.P.O.’s and more firme entered and exited the Fortune 500. These are held to be real economy changes which cause changes in the financial sector. This would be a valid assumption *if* financial markets were efficient. However, the point of the exercise is to decide if the scale of financial sector changes because of changes in the level of irrational financial activity. To see if the correlations are evidence against this hypothesis one has to look at the hypothesis. How about looking at “Advances in Behavioral Finance” edited by Richard Thaler 1993. Oh my it has a chapter on IPOs.

The chapter notes that an IPO can be a win/win when a firm grows to a size such that its founders don’t want to bear all the risk *or* an IPO can be a win lose when suckers are willing to pay unreasonably high prices for the firms shares. If one considers the possibility that changes in financial activity are based on changes in the number or irrational optimism of irrational agents, then one notes that all indices of market sentiment are correlated with IPOs. Importantly future predictable long term returns are low when the optimisms index is high. IPO shares are a bad investment (I mean if you get them straight from the initial subscription they are OK but if you buy them second hand or later on the next day you get bad beta adjusted returns on average).

The idea that asset prices and trading volume move up and down with the flow of new suckers to be fleeced is the alternative to Surowiecki’s hypothesis. It fits the correlation with IPOs just as well.

Consider the new technologies leading to new firms leading to more finance “Second, the corporate world was transformed by revolutionary developments in information technology and by the emergence of new industries like cable television, wireless, and biotechnology.” OK I submit that the new industry called “television” was more revolutionary that the new industry called “cable television.” I think this is safe to say. So why did cable require more from the financial services industry ? I’d guess that cable used the financial services industry more, because there were more suckers to be fleeced.

Ohr how about “biotechnology”. I love biotechnology, but it remains minuscule compared to chemotechnology. Can anyone really pretend that the ability to synthesize new and interesting compounds increased more in the 80s and 90s than in the 50s and 60s. I mean anyone who is familiar with the state of organic chemistry in the 40s. What is a bigger innovation erithropoetin or penicillin ?

It isn’t that biotech started in universities and synthetic organic chemistry in large corporations. The pathbreaking work in synthetic organic chemistry was based in universities (does the phrase Woodward-Hoffman rules ring a bell ?).

Something else caused the chemists to let existing firms develop their technology while molecular biologists set up new firms. I don’t think it was the science. I think it was the huge number of suckers to be fleeced.

Here the evidence that changes in the financial services industry were caused by changes in non financial technology is entirely financial. It is not based on total factor productivity growth or any assessment of inventiveness other than an IPO.

Now I admit that the Fortune 500 evidence is less circular (I even admit that when I began typing this post I misremembered “Fortune 500” as “S&P 500” mega ooops). However, at the very least, one has to look at the largest *non financial* corporations. In 2009 the 500 included

11 Bank of America Corp. 113,106.0 4,008.0
12 Citigroup 112,372.0 -27,684.0
13 Berkshire Hathaway 107,786.0 4,994.0

I’m not sure about old names but I don’t recognise any financial services corporations in the top 100 in 1955

The entry of financial corporations into the Fortune 500 (and no I won’t look at the while 500) can’t be evidence of a cause outside of finance of financial hypertrophy.

More generally, membership has a lot to do with the price of products. Obviously the price of petroleum has a huge effect. I notice Anaconda Copper in the top 100 in 1955. How about the price of a kilometer of fibe optic cable ? How much of that real economy volatility is based on speculation ? There is no way of knowing.

Oddly Surowiecki seems not only to make assumptions about the direction of causation, but seems to make contradictory assumptions about the direction of causation. The statement “The desire to bring back the boring, small banking industry of the nineteen-fifties is understandable. Unfortunately, the only way to do that would be to bring back the economy of the fifties, too.” asserts causation from the financial sector to the real sector. The claim is that an intervention in fiance (much stricter regulation) will cause a change in the rest of the economy (it will become as boring as it was in the 50s). However, the interpretation of the evidence is always that changes in non financial technology caused changes in finance. So “are accounted for” becomes “propelled” which becomes “needed or else it would have been cut short.” Altogether not convincing.

I have a guess as to how this article came to be. Surowiecki’s conclusions are fairly radical. He advocates a major tightening of regulation and a huge reduction of the financial sector. The arguments which I find unconvincing are arguments against much more radical proposals. I think that the need to prove that he isn’t a DFH leads Surowiecki to make invalid arguments, which he considers gestures of a willingness to compromise with people who want no change in regulation.

Everyone here already knows that I am a DFH so I just note that the arguments are invalid.

update: From comments a correction.

EconProph says:Today, 17.46.08“Robert,
I agree wholeheartedly with your methodological analysis (and also I suspect what would be your policy recommendations). But, a data artifact: the entry of financial corps into the Fortune 500 was an editorial decision of the magazine, not an economic phenomenon. In decades past (70’s and earlier), the “Fortune 500” was properly called the “fortune 500 Industrials” (no banks, no transports, no retails) and a separate list of 100 largest banks/finance companies was published with it. By the 1980’s the magazine decide to consolidate the lists since the industry boundaries had blurred.