February 10 NYU Tax Policy Colloquium (with Mick Keen), part 2

Before reading this, you may want to check out Part 1 of this multi-part post.

Part 1 left the damsel on the railroad tracks (so to speak) in the sense that I had mentioned Weitzman's "prices versus quantities" framework as giving us a potential handle on how to address the problems with financial institutions that led to their unleashing devastating harm on the world economy in 2008.

Before going further, however, I should note a certain Rashomon quality to discussions of what wrong. In these discussions (whether or not in Rashomon), often all the interpretations appear to be correct. E.g., should we think the banks were mispricing risky assets because they had bad incentives ("heads I win, tails you lose") that led them to embrace investments with an above-normal return in most circumstances but a hideous downside from "tail risk" that was bound to manifest sooner or later? I'm inclined to say yes. But wait a second, wasn't there a bubble going on, such that everyone was mispricing the bubble assets, and the key thing about a bubble is that it's hard to be sure one really exists until it in fact pops. Not a bad theory either.

But moving right along: what are some of the tax versus regulatory choices that we face with regard financial institutions? (From now on, I'll call them "banks" although I mean financial institutions more generally.) Here are a few:

--How much regulatory capital should banks be required to keep on hand? In the notes I prepared for the colloquium discussion, I dramatized this as the little bank in It's a Wonderful Life deciding how much cash to keep in the strongbox so they could satisfy any demand depositors who came in. In a more realistic modern setting, it has to do with "regulatory capital" that is effectively equity rather than debt, hence generating no right to demand either one's principal back or periodic payments. A banker who was utterly indifferent to the risk of a bank run would want to keep zero in the strongbox so that he could invest every penny he had at an expected positive rate of return. More realistically, while he might want to keep something in the strongbox so that the bank doesn't collapse the moment the first depositor shows up, the amount he decided to keep there might be socially too low, as he would give weight only to his aversion, if any, to having the bank collapse, as opposed to a full measure of the social harm this would cause. And even if the depositors can monitor what he does, everyone else who would be affected presumably can't.

--How riskily should banks be allowed to invest? My notes analogized this to the question of whether they should be restricted to making safe but boring loans, or instead allowed to go to Vegas and bet everything on red (apart from the cash in the strongbox). Here we face a familiar incentive problem, since bankers may have an incentive to make "heads I win, tails you lose" bets that are only appealing because they don't face the downside. (The incentive may arise either from their owning shares in the banks they operate, or from incentive compensation packages.) The same moral hazard issues arises, for example, whenever a borrower has limited liability (whether literally, as in the case of a corporation, or effectively, as in the case of an individual who might end up declaring bankruptcy). But for banks the issue may be especially dire given both (a) the externality problem, extending beyond depositors, and (b) the fact that derivatives permit bankers to make huge bets, even with very limited capital, that are very opaque so far as any potential monitor is concerned.

--What about other decisions that may increase systemic risk, such as becoming "too big to fail" and thereby increasing the failure externality, with the possible consequence that the government will have to become an implicit guarantor (thus substituting an increased bailout externality for some or all of the added failure externality).

Each of these three margins can be addressed through "quantity" regulation (using that term very broadly) and/or by a "pricing" approach that uses taxes or subsidies.

Thus, in addition to or instead of requiring a given level of "cash in the strongbox" via capital adequacy regulations such as those that have existed for decades and are being revised, one could address bankers' incentives by treating such cash more favorably than that which the bank takes out of the strongbox and invests (i.e., by issuing debt rather than equity, if we can define these terms meaningfully enough, in relation to the underlying objective).

Likewise, one can reenact Glass-Steagall-style restrictions on permitted investments by relevant financial institutions, or alternatively one can tax-penalize risky relative to safe investment. (Note, BTW, that various institutions that never would have been subject to Glass-Steagall have nonetheless proved to create failure externalities given their role in the financial system.) The tax approach does not require identifying which investments are in fact the unduly risky ones. Instead, one can get at the problem through unfavorable treatment for what are observed ex post as unusually high returns. An arguable example of this approach is the "Financial Activities Tax" or FAT that a recent IMF Staff Report outlined. The FAT would tax above-normal returns, such as by the mechanism of allowing a deduction for the ordinary rate of return (or equivalently, applying expensing to all cash flows) and also disallowing for this purpose the high-end executive compensation that reflects bankers' paying out these above-normal returns to themselves. The rationale offered for the FAT is that the extra-normal returns may be rents, which in theory can be taxed away without creating inefficiency. But an alternative or additional rationale would be that observed above-normal returns ex post may actually not have been rents, but simply bets with nasty tail risk that simply didn't eventuate in the observed period.

Finally, firms that appear to be "too big to fail" can be broken up under a regulatory approach, or alternatively simply tax-discouraged. Indeed, if the pricing problem were easier than it is, one could simply tax them for the value of the implicit guarantee that they force on us (via the potential magnitude of the failure externality), thus in effect pricing it into their decisions.

Okay, one last word on all this before I turn to how one might think about the taxes versus regulation choice given the analysis in Mick Keen's paper along with that in the Weitzman paper on taxes versus subsidies. A whole lot of regulatory effort, and I gather indeed the main effort, has focused on capital adequacy. But is this a misdirection of effort - not in the sense that one shouldn't do this, but rather that it may be inadequate unless a lot of other things are done as well? I am inclined to say yes, though admittedly not an expert on financial institutions.

Consider AIG. My crude rendering of what it did is as follows. By holding so much of the downside risk that credit default swaps faced from the prospect of a nationwide decline in real estate prices, AIG exacted very much like the hypothetical individual who shows up in New Orleans in 2004 and issues hurricane insurance to everyone in town. By the way, that fictional New Orleans guy, like AIG, has a fabulous business model. He gets people to pay him all these hurricane insurance fees. If no hurricane hits, it's free money. If Katrina comes, the party is over but at least he doesn't have to pay on the policies (because he can't). So this is equivalent to betting everything on red because heads he wins, tails other people lose.

Would capital adequacy rules help enough with such an individual? As stated, not unless (a) each person's insurance premium covered 100% of the expected harm that a big hurricane the next year would cause (which seems unlikely) and (b) the capital adequacy requirement was 100% - none of the premiums can be reinvested.

Again, the point isn't that capital adequacy regulations aren't important and valuable, but that by themselves they're inadequate and arguably have received relative over-emphasis - not because they should be done less, but because other and additional things should be done more.

But this post has also gotten long enough, so I'll finish my observations in Part Three, which is available here.