Saturday, May 5, 2012

FDA for Financial Innovation?

Eric Posner and Glen Weyl are making a big splash with their proposal for An FDA for Financial Innovation

As you might guess, I think it's a terrible idea. But let me try not to be predictable. I do think there are financial products that need to be regulated if not banned. Interestingly, Posner and Weyl completely miss these elephants in the room. (What are the dangerous products? I'm going to make you wait so you'll read more of the post.) That observation alone seems like a good argument against their FDA as a structure for financial regulation. 

That's the real question. The question is not, "should there be some financial regulation?" The question is, "what form should it take?"  "What institutional structure should it follow?" For example, see a previous blog post distinguishing law, rules-based regulation, and regulatory discretion. The question is, "does it make sense to legislate an FDA-like structure, in which all products are presumed guilty until proved safe to the satisfaction of a regulatory agency's discretionary judgement?" 

The two missing ingredients

Though the FDA is not immune from criticism, the real FDA has two things going for it that the Financial FDA can only dream of: A clear and objective definition, and an objective method for testing products against that definition. Drugs either help patients to get better, with few side effects, or they don't. And we can evaluate that ability with randomized clinical trials. The Financial FDA has neither.

Posner and Weyl want the financial FDA to separate products that are used for "investment" or "hedging" from those used only for "speculation," and ban the latter.
The agency’s fundamental standard would be whether the welfare gains from insurance allowed by a new product exceeded the likely costs created by the speculation it facilitates.
But there is no consensus on "hedging" vs. "speculation" for  existing securities like stocks and options, after 400 years of actual experience!  And, not having that experience, or an objective method like  clinical trials, Posner and Weyl propose that panels of experts can make the call and decide how much "speculation" vs. "hedging" a new untried security will give rise to.

Posner and Weyl aren't really able to express what's so terrible about "speculation" anyway.  OK, some people lose money.  For example, they decry "heuristic arbitrage-based speculation". 

A large literature establishes that people’s trading strategies often reflect simple heuristics (buy a stock that has recently increased in price) that can be easily exploited by hedge funds. By considering such heuristics and how they interact with the product’s characteristics, the agency could project demand based on heuristic arbitrage.
"Easily?" I know a lot of hedge funds that are losing money. About 1 in 5 goes bust every year. They (and our endowments that invest in them) only wish it were so easy to "project demand." And even if so, where is the social problem here? It's a zero sum game played among grownups.

Yes, some people think "speculation" makes prices too volatile. The puzzle is,  by definition "excess volatility" provide opportunities for others to speculate against them and make money. Another 400 year argument with no consensus, at least not one ready to be written into Federal Law after selectively citing only one side of the debate. 


I've got bad news for Posner and Weyl. Almost all stock and option trading is "speculative." Exchanges exist to facilitate "speculative" trading.  Options were designed and invented purely for "speculation." Their use for "hedging" was a much later discovery, and remains a minor part of trading.

Let me explain. A call option gives you the right to buy a stock at a given price, but not the obligation to do so. For example, you might buy for 5 pounds the right to buy East India Company stock for 100 pounds. (A deliberately 1700s example to emphasize how long this has been with us.) Now, if East India company stock goes up to 120, the value of your option will increase dramatically, maybe to 22 pounds.

So, suppose your spies see the latest boat floating up the Thames, deep in the water with spices. What do you do? You want to buy lots of stock. But you only have 5 pounds. You could try to borrow 100 pounds to buy the stock, but the lender doesn't want to do that, because if the stock goes down you won't pay back the loan. Buying the option lets you speculate on the stock as if you borrowed 100 pounds, but you can only lose the 5 pounds. The trader on the other side (whose spies say the boat is just leaking) is perfectly happy to enter that contract. It is a perfectly designed security... for speculation

"Speculation" has important social functions, as everyone since Adam Smith has recognized. Suppose you want to sell some stock in a hurry. If "speculators" are banned, it becomes much harder to find a willing buyer. It is "speculation" that provides "price discovery" and "liquid markets" for the rest of us.

Posner and Weyl recognize this, to some extent: 

An investor who buys Facebook stock is making a bet as to how much money Facebook will earn by providing a service in the real economy. If people could not buy stock in this “speculative” manner (or make loans, etc.), then businesses with good ideas would have a great deal of trouble implementing those ideas and thus providing benefits to consumers, while companies with poor ideas might receive capital because no one would ensure that the price of their stocks or bonds remained low. Thus, financial market activity that helps prices adjust to their true value can influence the allocation of capital among potential
products and thus improve economic efficiency.
So far so good. But then they go on...
However, improving the informational efficiency of prices is only useful to the extent that it reflects the fundamental (social) value of the asset and affects the allocation of capital in the real economy. When fluctuations are too unpredictable, too driven by expectations of other traders’ behavior or shifts in prices over too short time-scales to have any impact on the real economy, they cannot have value under this argument.
Very nice. But which commissar can tell the line between "fundamental (social) value of the asset" and excesses?  Given we can't do it for existing assets, with 400 years of data, how is a panel of experts supposed to figure this out, ahead of time, for new products that we have not seen yet? 

The hopeless task for the panel of experts  

You really need to read the paper, not just the opeds, to get a sense of how pie-in-the-sky this faith in experts is. Remember, we are talking here about evaluating categories of new products, like "stock" or "call options," products that don't exist yet.
The crucial step would be to determine the speculative costs of the new instruments, based on how many individuals would be interested in speculating on them and at what volume. The key to a careful analysis is to break down speculative demand itself into several categories: disagreement-based, regulatory arbitrage-based, tax arbitrage-based, and heuristic-exploiting.
That is indeed the "crucial step." Sort of like "Here is where the magic happens." "The crucial step would be where the Easter Bunny comes in the middle of the night and gives the children chocolates."

Continuing, they offer a taxonomy of speculation to be forecast:
Pure disagreement-based speculation. This is perhaps the hardest of all the forms of speculation to project demand on, as so much depends on what catches the imagination of potential participants. Luckily, a large historical track record of past products offers a rich data set on which regressions using ex-ante characteristics of products can be run to project ex-post speculation, which can be measured fairly easily based on observed volumes compared to the demand accounted for by the other sources demand (both hedging and other speculative forms discussed below).
Nobody has ever done this for existing products.
For example, one natural predictor of speculative demand, proposed by Simsek is to survey professional forecasters for their estimates of the value of the security. If, for example, the forecasters agree on the value of the security, then it cannot be used to speculate. If the distribution of estimates is sufficiently wide, however, it can be used to speculate.
Give 100 analysts all the company data you want but not the stock price, and see how many of them can come within a factor of 10 of the actual price.
 Other predictive factors may relate to how prominent the phenomenon that the derivative is based on is in  the public mind or in commonly used financial models. These can be quantified using new tools of automated text analysis, such as Google’s nGrams Viewer.41 By harnessing data on past products and the speculative demand they generated, indicators like this could be used to form clearer expectations of likely speculative demand, in conjunction with documents that the proposer will submit about the sources of demand they anticipate and projections by similar but disinterested market players.

OK, this went on a bit, but is the pie in the sky nature of this clear? Nobody has ever credibly run even one of these regressions for existing securities, let alone proposed securities.

Let me try to be positive. The main thing Posner and Weyl could do is to actually produce one such evaluation, that survives widespread scrutiny and determines the amount of "speculation" vs. "hedging" in even an existing security, to the level of certainty required for us to bring down the heavy power of the Federal Government to ban it. 

Regulatory and tax arbitrage: Catch 22

Posner and Weyl make one good point: some financial innovation is undertaken for regulatory or tax arbitrage. Mortgage backed securities were bundled into special purpose vehicles with off-balance sheet guarantees as a dandy way to get around capital requirements. Institutions required to hold AAA securities found ways to construct such securities to hold more risk than regulators wanted.

Alas, a Financial FDA blessing new securities would be hopeless to stop this sort of thing. In these cases, as well as Posner and Weyl's other examples, the securities had perfectly valid other uses. They were invented for other uses. Securitized debt also goes back hundreds of years. (When you get bored here, go explore Geert Rouwenhorst's History of Financial Innovation website)

Most of all: Catch 22. Posner and Weyl's complaint is that financial engineers are one step ahead of regulators, who can't see how they're using financial products to get around regulations and taxes. Well, if the bank regulators and tax authorities can't figure out, often for 10 years or more after the fact, how a product is used to avoid regulation and taxes, how in the world is the Financial FDA's panel of experts supposed to figure it out ahead of time? If that were possible, the regulators themselves would be able to stop the practices! This is an airtight logical proof that the idea can't work.

Junk bonds are another good example. Poser and Weyl write of them approvingly, 
A financial instrument may lower the cost of capital to firms and individuals. Such reductions in the cost of capital result from the ability to spread the risk more evenly. For example, prior to the securitization of “junk bonds” in the 1980s, many small  firms could draw only on very wealthy investors for financing.
But junk bonds were also used for regulatory arbitrage. In the 1980s, savings and loans wanted to add a lot of risk. Regulation said they could only buy "bonds" but the S&Ls wanted to double or nothing by taking on the risk and return profile of "stocks." Junk bonds fit the bill perfectly, and let the S&Ls evade risk regulation. Posner and Weyl didn't notice this after the fact. Good luck to their Financial FDA to notice it ahead of time.

More regulatory arbitrage just gets around silly regulations. We subsidize debt by making interest payments tax deductible to companies while dividends are not. No surprise, companies do a lot of engineering to take advantage of this tax deduction. Big banks want huge leverage, then engineer their way around capital ratios. But it would be a whole lot easier to remove the subsidy for debt in the first place.  

The Nature of Innovation

 Who is going to run all these regressions?
This information should be available from the firm seeking approval; after all, it should be incorporated in the demand analysis the firm uses to decide whether to market the financial product in the first place.
This quote reveals a deep confusion on how markets work and how new products are invented. New products typically start as one-on-one agreements. Company A calls Goldman Sachs asking for a new kind of swap contract. Others find it useful, over the counter trading increases, some contracts get standardized and traded on exchanges, then clever ducks figure out how to use them to get around regulation, and traders start "speculating."

Stocks, bonds, credit default swaps, catastrophe bonds, insurance itself, reinsurance, mortgage backed securities, securitized debt  all started this way. They did not start with some big "firm" planning to "market" some new security like an iphone. Some consumer financial products start that way, but there's no "speculation" in credit cards.

Trading vs. Products

Posner and Weyl go on a bit on the evils of high freqeuency trading. This is particularly curious. Their FDA is supposed to analyze products. But high frequency trading is a practice, for a product that's been around 400 years.   Or is the Financial FDA supposed to preemptively approve or disapprove every "trading practice" whatever that could mean?
 Assuming all transactions that only occur when possible at sufficiently low cost are wasteful, one can combine this “elasticity” with the expected reduction in cost created by the new instrument to estimate the number of harmful transactions likely to be created
That's an interesting assumption to be written in to the Federal Register. 
A Reflection on Law.

A financial contract is a contract. It's just an agreement between two parties: if X happens, I pay you money. If Y happens, you pay me money. That's the most basic kind of contract there is.

In essence, Posner and Weyl are advocating a dramatic change to contract law as we've understood it for hundreds of years. (OK, I'm not a legal historian,  but you know what I mean.)

At heart, their proposal is to declare that any private contract involving money is illegal until the Federal Government authorizes it. As you see in their discussion of high speed trading, financial practices are  illegal if Posner and Weyl can't understand their social function.

And the determination of "social utility" comes from a politically-appointed regulatory body with wide discretion, no clear definitions, no clear procedure, and thus essentially no recourse. (If the panel says no, how can you prove your security is useful?)

Aside the issue of basic liberties, constitutional limitations, and all that old-fashioned stuff, it doesn't take much to imagine how quickly such an operation would become politicized, captured, or corrupt. Even the FDA doesn't do so well when big political interests are involved.

That's rather astonishing, especially coming from across the Midway at the University of Chicago

(The dangerous contracts? Short term debt, demand deposits, and now broker-dealer relationships. Contracts which induce runs. These have plain externalities: if I run, it makes the institution less liquid, so you have an incentive to run. We just had a big run in the shadow banking system. That is the problem to be fixed, not "heuristic-exploitative" hedge funds, no?)