An enduring truth about financial regulation is this: Given the discretion to do so, financial regulators will always do the wrong thing.

It's easy to explain why. In good times, regulators have every incentive to take banks at their optimistic word on asset valuations, and therefore on bank capitalization. It is almost impossible for bank regulators to be "tough" in good times, for the same reason it is almost impossible for mutual fund managers to be bearish through a bubble. A "conservative" bank examiner who lowballs valuation estimates will inevitably face angry pushback from the regulated bank. Moreover, the examiner will be "proven wrong", again and again, until she loses her job. Her fuddyduddy theories about cash flow and credit analysis will not withstand empirical scrutiny, as crappy credits continually perform while asset prices rise. Valuations can remain irrational much longer than a regulator can remain employed.

Bad times, unfortunately, follow good times, and regulatory incentives are to do the wrong thing yet again. When bad times come, overoptimistic valuations have been widely tolerated. In fact, they will have become very common. Overvaluation of assets leads to overstatement of capital. Overstatement of capital permits banks to increase the scale of their lending, which directly increases reported profitability. Banks that overvalue wildly thrive in good times. Fuddyduddy banks lag and their CEOs are ousted and The Economist runs snarky stories about what schlubs they are. The miracle of competition ensures that many of the most important and successful banks will have balance sheets like helium balloons at the end of a boom. Then, like a pin from outer space, somebody somewhere fails to repay a loan.

When this happens, bankers beg forbearance. They argue that the rain of pins will eventually pass and most of their assets will turn out to be fine. They ask regulators to allow them to write down assets gently, slowly, so that they can let ongoing earnings support or increase their regulatory capital. If that doesn't work, they suggest that capitalization thresholds be temporarily lowered, since what good is having a buffer against bad times if you can't actually use it in bad times? Knowing, and they do know, that their assets are crap and that they are on a glide path to visible insolvency, they use any forbearance they extract to "gamble for redemption", to make speculative investments that will yield returns high enough to save them, if things work out. If they don't, the bankers were going to lose their banks anyway. The additional losses that fall to taxpayers and creditors needn't concern them.

Here, wouldn't regulators draw the line? When the trouble is with just a few small banks, the answer is yes, absolutely. Regulators understand that the costs of closing a troubled bank early are much less than the costs after a delay. If a small bank is in trouble, they swoop in like superheroes and "resolve" it with extreme prejudice.

But when very large banks, or a very large number of banks are in trouble, the incentives change. Resolving banks, under this circumstance, will prove very expensive in terms of taxpayer dollars, political ill-will, and operational complexity. It will reveal regulators to have been asleep at the wheel, anger the public, and alienate nice people whom they've worked closely with, whom they like, who might otherwise offer them very nice jobs down the line. When a "systemic" banking crisis occurs, regulators' incentives are suddenly aligned with bankers: to deny and underplay, to offer forbearance, to allow the troubled banks to try "earn" their way out of the crisis. Regulators, in fact, can go a step further. Bankers can only "gamble for redemption", but regulators can rig the tables to ensure that banks are likely to win. And they do. A central bank might drop short-term interest rates very low to steepen the yield curve. It might purchase or lend against iffy assets with new money, propping up prices and ratifying balance sheets. It might pay interest to banks on that new money, creating de novo a revenue stream based on no economic activity at all. Regulators might bail out prominent creditors and counterparties of the banks, suddenly transforming bad bank assets into government gold. Directly or via those bailed out firms, regulators might engage in "open market transactions" with banks, entering or unwinding positions without driving hard bargains, leaving taxpayer money on the table as charity for the troubled institutions. They might even redefine the meaning of financial contracts in a subtle ways that increase bank revenues at the expense of consumers.

If all that stuff works out, regulators might be able to claim that they didn't do such a bad job after all, that the crisis was just a "panic", that their errors prior to the crisis were moderate and manageable and it was only the irrational skittishness of investors and the taunts of mean bloggers that made things seem so awful for a while. Regulators, like bankers, have everything to gain and little to lose by papering things over. And so they do. Besides, things here weren't nearly as bad as in Europe.

There was nothing new or different about the recent financial crisis, other than its scale. Yes, the names of the overvalued financial instruments have changed and newfangled derivatives made it all confusing about who owed what to whom and what would explode where. But things have always blown up unaccountably during banking crises. We have seen this movie before, the story I've just told you is old hat, and the ending is always the same. We enact "reform". The last time around, we enacted particularly smart reform, FDICIA, which was painstakingly mindful of regulators' incentives, and tried to break the cycle. It mandated in very strong terms that FDIC take "prompt corrective action" with respect to potentially troubled banks.

The theory of "prompt corrective action" was and is very sensible. It's pretty clear that the social costs increase and the likelihood of an equitable resolution to problems decreases the longer banks are permitted to downplay weakness, the more regulatory forbearance banks are granted, or the more public capital banks are given. (An "equitable resolution", in this context, means giving the shaft to bank managers, shareholders, and unsecured creditors to minimize costs to taxpayers and to sharpen the incentives of bank stakeholders to invest well. "Regulatory forbearance" and "public capital injection" are redundant: Under current banking practice, regulatory forbearance is economically equivalent to an uncompensated injection of public capital, like TARP but without the messy politics and with no upside for taxpayers. Make sure you understand why.)

So FDICIA tried to short-circuit our woeful tale by telling regulators they should have a twitchy trigger finger. If regulators intervene early and aggressively, the costs of the crisis will be moderate, and since the costs of the crisis are moderate, it should actually be plausible for regulators to intervene early and aggressively rather than playing the world's most expensive game of CYA. It was a great idea. Except it didn't work.

We've already told the story of why it doesn't work. Bank health and safety is a function of capitalization, capitalization is a function of bank asset valuation, and there is no objective measure of asset valuation. During good times "conservative" valuations are demonstrably mistaken and totally unsupportable as grounds for confiscating the property of respected, connected, and wealthy businesspeople. Regulators simply fail to take prompt corrective action until it is far too late.

As you read through the roughly 1400 pages of currently proposed regulatory reform, ask yourself what, if anything, would interfere with the (uncontroversial and long-understood) dynamic that I've described. Giving regulators more power doesn't help, when regulators have repeatedly failed to use the powers they had. Putting more bank-like institutions and activities under a regulatory umbrella seems sensible, as does eliminating opportunities for firms to choose among several regulators and shop for the most permissive. But even our most vigilant and competent regulator (hi FDIC!) was totally snowed by this and the previous two banking crises. It has become fashionable to suggest that the idea of "systemic risk" is novel, and that just having some sort of high-level, blue ribbon council explicitly charged with worrying about financial catastrophes will change everything. But financial meltdowns are not new, Timothy Geithner was giving smart, widely discussed speeches about systemic risk in 2006, exactly as the current crisis was building.

There is, unfortunately, almost no correlation between the degree to which an institution or sector is supervised by regulators and behavior or misbehavior during a financial crisis. Commercial banks, GSEs, and bond insurers were intimately regulated and are now toast. Mortgage originators, boutique securitizers, ratings agencies, and CDS markets were largely unregulated. They also clearly failed. The one trainwreck that the current round of proposals might have forestalled is AIGFP, whose unhedged, uncollateralized CDS exposure would make even the most lackadaisical regulator blush. But it is not at all plausible, especially in the US, that AIG was the linchpin without which the late troubles would not have occurred. If we had to refight the last war under all the regulations now proposed, we might have won one battle. But we'd very definitely have lost the war. Deregulation won't solve the problem. But neither will the sort of regulation now proposed by Barney Frank or Chris Dodd.

But what about Bernie Sanders? Is "too big to fail" the problem? Yes and no. Unsustainable bank-funded asset price booms can and do occur even among small banks. But systemic crises are more likely in a world with big banks, as only one or two need totter to take down the world. Chopping up banks reduces the frequency of major crises. Also, "prompt corrective action" does sometimes work for small banks. For big banks, PCA is just a joke — Citibank is and always will be perfectly healthy until it is totally a basket case. But regulators do stage early interventions in smaller banks, even during relatively quiet times, and that does help. Crises among small banks can lead to large fiscal costs (c.f. the S&L bailout). But even during serious crises, many small banks turn out to have been prudent, and small banks tend not to be so interconnected that a cascade of failures leaves us without a financial system. Small-bank-based systems fail gracefully. (See Felix Salmon.) Crises among small banks are less corrosive to incentives for careful capital allocation, and less offensive to distributive justice, than large bank crises, because regulators are willing to force preferred equityholders and unsecured creditors of smaller banks to bear losses while they hesitate to do so for big banks. Also managers of smaller banks can be perfunctorily defenestrated, while managers of megabanks somehow survive (and even when they don't, they are too wealthy to have to care). Again, one hates to be mean, but treating the managers of trouble-causing banks roughly is important both to get the incentives right and out of regard for justice.

We should insist upon a market structure in which financial institutions are universally small. But smallness cannot be defined by balance-sheet assets alone. We need to manage the degree of interconnectedness and the scale of total exposures (including off-balance sheet exposures arising from derivative market participation). Further, we need to ensure that no market participants are indispensable by virtue of controlling some essential market infrastructure. Clearing and payments systems, securities and derivatives exchanges, etc should be multiple and redundant if privately owned, or publicly managed if efficiency demands a monopoly provider. Essential infrastructure should be held by entities that are bankruptcy remote from firms that bear unrelated risks, although the stakeholders needn't be bankruptcy remote from the critical infrastructure. (For example, if the owner of a derivative exchange goes under, the exchange must be immune from liability, but if the exchange suffers losses due to insufficient collateral requirements, the owner could still be liable.)

Fundamentally (and a bit radically), for financial reform to be effective, regulators must actively target market structure. Financial systems are public/private partnerships, not purely private enterprises. It is perfectly reasonable for the state as the ultimate provider of funds and bearer of risk to insist on a robust and heterogeneous network of delegates. Regulators needn't (and really cannot) architect the breakup of today's destructive behemoths. All they need to do is identify dimensions along which firms become indispensable or threatening to financial stability as they grow, and tax measures of those attributes at progressively steeper rates. The taxes could be slowly phased in over a few years, to give existing firms time to arrange efficient deconglomerations. Importantly, legislators should characterize the target market structure, and empower regulators to define and alter tax schedules as necessary to achieve that target, rather than specifying them in law, to counter gaming by nimble financiers. (For example, a tax on balance sheet assets would lead to rapturous innovation in tricks to keep stuff off-balance-sheet.) Taxes should always be imposed in a nondiscriminatory way across the industry. (If you are not concerned about the role of political influence and favoratism in regulatory action, you haven't been paying attention.) "Taxes" could take the form of increased regulatory burdens, such as capitalization or reserve requirements (though the effectiveness of the latter is diminished if central banks pay interest on reserves).

Variations of these ideas are actually in both Frank and Dodd's proposed legislation. Regulators would have a fair amount of discretion, under the new laws, to do the right thing. They could ignore the terrifying shrieks of our banking overlords and force the monsters to break apart. But we come back to the first and most ancient law of banking regulation. Given discretion, banking regulators will always, always do the wrong thing. Only if Congress defines a verifiable target market structure and periodically audits the regulators for compliance will we eliminate "too-big-and-mean-and-rich-and-scary-and-interconnected-and-sexy-to-fail".

But what about the much vaunted "resolution authority". Doesn't that change the deadly dynamic of banking regulation described above? Sure, it will still be true that during booms, banks will make dumb mistakes and regulators will be unable to point out that the Swiss cheese they're calling assets is chock full of holes. But, you might argue, when the cycle turns, they'll no longer be helplessly forced to resort to CYA-and-pray! They'll have the tools to wind down bad banks ASAP!

Maybe. Resolution authority might be helpful. But I'm not optimistic. During the current crisis, there are two accounts of why we guaranteed and bailed existing banks — including creditors, managment, preferred shareholders, and financial counterparties — rather than resolving the banks and forcing losses onto the private parties who made bad bets. One account emphasizes legal constraints: we had laws that foresaw the orderly resolution of commercial banks, but not investment banks or bank-holding conglomerates. According to this view, regulators' only options were to permit Lehman like uncontrolled liquidations of financial firms or else make whole every creditor to prevent a formal bankruptcy. If this is what you think then, yes, resolution authority might change everything.

But another view — my view — suggests that despite the limitations of preexisting legislation, regulators throughout this crisis have had the capacity to drive much harder bargains, and have chosen not to. Despite having no legal authority to do so, the government "resolved" Bear Stearns over a weekend in almost precisely the same manner that FDIC resolves your average small town bank. Secretary Paulson could at that point have gone to Congress with a proposal for resolution authority to institutionalize the powers he clearly required. Instead, he had Treasury staff prepare the first version of TARP and put it on a shelf until an emergency sufficient to blackmail Congress arose. Whatever the legal prearrangements, regulators have always had sufficient leverage, over firms and firm managers, to push through any structural changes they deemed necessary and to create bargaining power for firms to insist on loss sharing. Finally, earlier this year, when bank nationalization was an active debate, opponents did not because they could not claim that authority would not be found if the administration decided nationalization was the way to go. Harsh measures towards banks would have been extremely popular. What authority the administration did not have by virtue of existing powers and informal leverage they could have achieved by purchasing common shares instead of preferred during "capital injections", or, in a pinch, by asking Congress for help. In my view, legal niceties were never the issue. Regulators opted to guarantee the banking system and bail out creditors because given the terrifying scale of the problem, the operational complexities and investor uncertainty associated with resolutions or nationalizations, the power of the banks and regulators' personal connections to them — our leaders simply opted not to pursue more hardball resolutions. If we don't change the structure of the financial industry, there's no reason to think that next time around regulators won't use the proposed resolution authority to do exactly what they opted to do this time. They won't even need to go to Congress for a new TARP, as Frank's proposal gives the executive branch carte blanche to provide financial firms unlimited guarantees and support. (I haven't read the text of Dodd's bill.)

Yes, I know that "living wills" are supposed to diminish the operational complexity and uncertainty associated with taking a harder line, and that, in theory, might encourage different choices. I also understand that some sort of industry-funded slush fund is supposed to bear future bail-out costs. My sense is that during a gut-wrenching financial crisis, hypothetical funeral plans will provide little comfort to terrified regulators, prepaid slush funds will prove to be laughably inadequate, and commitments to make firms pay for the mess ex post will be waived in order to shore up struggling bank balance sheets. These proposals are about providing the political cover necessary to get legislation passed, but they will prove to be utterly without substance when the next crisis occurs.

I'll end where I started. The one rule that you can rely on with respect to banking regulation is that whenever regulators have discretion, they do exactly the wrong thing. For very predictable reasons and despite the best of intentions, they screw up. Besides messing around with intragovernmental organization charts, the main proposals before Congress give regulators more power and more discretion. That just won't work.

Afterthoughts: It is really worth considering this excellent piece by Matt Yglesias, ht Mike Konczal.

I almost always disagree with Economic of Contempt on these issues. But despite being wrong, he is very smart, and a gentleman too. You should read this piece, which gets everything right, within the confines of supervisory regulation. We need structural changes most of all, but supervisory regulation won't be going away, and there his points are dead on. Also check out his very creative defense of Too Big Too Fail banks. I happen to think liquidity is as often vice as virtue, so I'm not persuaded. (Assets that are difficult to value should be illiquid. Otherwise investors fall prey to delusions of safety and rely upon risk-management by exit, which never works out well.) So EoC is wrong. But he's wrong in clever ways, and always worth reading.

Go Paul Kanjorski!

Commenter "reason" asks a question:

...it is not clear to me that it is well understood why inflation sometimes can be seen in consumer goods and sometimes is manifested in "asset price inflation". Do you have any ideas on this mechanism? I know some people deny there is such a thing as "asset price inflation". Do you have a theoretical basis for your ideas in this area?

I have a very simple answer to this question: Follow the money. Whether an economy generates asset price inflation or consumer price inflation depends on the details of to whom cash flows. In particular, cash flows to the relatively wealthy lead to asset price inflation, while cash-flows to the relatively poor lead to consumer price inflation.

Why? In Keynesian terms, poorer people have a higher marginal propensity to consume. The relatively poor include people who are cash-flow constrained — that is they cannot purchase what they wish to purchase for lack of green, so their marginal dollar gets immediately applied to the shopping list. Also, poorer people may be different, there may be a correlation between poverty and disorganization, lack of impulse control, inability to defer gratification etc. Think of Greg Mankiw's Spenders/Savers model.

Except when the world seems very risky, no one holds cash for very long. Poorer people disproportionately use their cash to purchase goods, while richer people disproportionately "save" by purchasing financial assets. If the supply of both goods and financial assets is not perfectly elastic, then increases in demand will be associated with increases in price. If relative demand for goods and financial assets is a function of the distribution of cash, what price changes occur will be a function of who gets what. [1]

This tale of two inflations helps to explain how we arrived at the unequal, credit-centric economy we have today. Central bankers are notoriously allergic to "wage pressure" as a harbinger of rising prices. Wages have two distressing properties: First, they are sticky. They represent repeated and persistent cash flows that cannot be downward adjusted en masse except during a serious crisis or dislocation. Second, a substantial fraction of wages goes to lower quintiles of the income distribution, who have a high marginal propensity to consume. Central bankers are not evil scrooges — they have nothing against consumption by poor people. But funding that consumption by wages limits the effectiveness of monetary policy. They'd prefer that the marginal dollar bound for consumption flow from a more malleable source.

During the "Great Moderation" in the US a variety of structural changes helped to increase the potency of monetary policy:

1. The wage share of GDP decreased significantly over the 1970s and 80s. Compensation did not decrease as much, but much of nonwage compensation is retirement savings that is saved rather than consumed.

2. Wage inequality increased, such that a growing fraction of wages went to "savers" rather than "spenders", limiting the direct impact of wage growth on consumption.

3. The growth and "democratization" of consumer credit provided consumers with an alternative source of purchasing power that was sensitive to monetary policy.

Prior to the Great Moderation, central bankers had to provoke recessions in order to control inflation. Broad-based wage growth led to increases in nominal cashflows by "spenders" that could only be tempered by creating unemployment or other conditions under which workers would accept wage concessions. In the post-Reagan world, growth in the sticky component of disposable income shifted to the wealthy, who tend to save rather than spend their raises. The marginal dollar of consumer expenditure switched from wages to borrowed money. The great thing about consumption funded by credit expansion, from a central banker's point of view, is that it is not sticky downward — no one who gets a loan today assumes that she will be able to expand her borrowing by the same amount every year. Credit-based consumption is susceptible to monetary policy with far less impact on employment than wage-based consumption. (One of Ben Bernanke's many claims to fame is his characterization of the credit channel of monetary policy transmission.)

By the middle 2000s, the credit economy was the air we breathed, and conventional wisdom held (and continues to hold) that economic growth and credit expansion are synonymous. We had those peculiar debates about the difference between "consumption equality" and "income equality", and which mattered more, since middle-class consumption had become significantly credit-financed. But from central bankers' perspective, we had stumbled into a good place, one where output growth was channeled into asset price inflation, but provoked consumer price inflation only indirectly and via a channel policymakers could regulate. This benign regime faced two threats, however. First, asset price inflation is unstable — while on any given day, price moves are determined by the flow of funds into assets, over time prices can become so unreasonable relative to the the asset's cash or service flows that arbitrageurs and nervous fundamentalists appear, creating the potential for a collapse. Second, credit expansion is unstable, as chronic borrowers may become unable to service existing debt, let alone borrow more to sustain aggregate demand. Unnervingly, sustaining consumption has required a secular downtrend in the policy interest rate, and eventually you hit that zero-bound. [2]

The Greenspan/Bernanke doctrine can be summed up by three familiar words, "Yes We Can!" Greenspan famously concluded that we can "mop up" asset price bubbles after they burst, rather than interfering with the dynamic whereby asset price inflation substitutes for consumer price inflation. Bernanke devoted his life to studying the role of credit in monetary policy and the hazards of deflation and credit collapse, and he famously concluded that we have the technology to prevent "it" from happening here. We are watching his experiment play out, in real time and from inside the maze. The outcome is not yet known.

I have my own normative view of "the great moderation", and it is not positive. I do not hope to see a return to the "good old days" of the 1990s and mid-2000s. But that isn't because the moderation dynamic cannot work, in principle. In principle, we can periodically reset the stage with a money-funded jubilee. It'd go like this: When credit expansion reaches its natural limit, let the debtors default, but make creditors whole with new money. "Moral hazard", rather than a problem, is the goal of the operation: Low marginal-propensity-to-consume "savers" are rewarded and encouraged to continue pouring their incomes into domestic financial assets, where any effect on goods price inflation is muted. Over several years, the balance sheets of debtors can be cured via some combination of bankruptcy, loan modifications, austerity, and youth. In the meantime, the Federal government adopts the role of consumer of last resort, in order to sustain nondeflationary levels of aggregate demand and limit unemployment. I think this is our current strategy. We are groping and stumbling towards the status quo ante, and it is not impossible that we will find it within a few years.

So what's the problem? First, in exchange for apparent stability, the central-bank-backstopped "great moderation" has rendered asset prices unreliable as guides to real investment. I think the United States has made terrible aggregate investment decisions over the last 30 years, and will continue to do so as long as a "ride the bubble then hide in banks" strategy pays off. Under the moderation dynamic, resource allocation is managed alternately by compromised capital markets and fiscal stimulators, neither of which make remotely good choices. Second, by relying on credit rather than wages to fund middle-class consumption, the moderation dynamic causes great harm in the form of stress from unwanted financial risk, loss of freedom to pursue nonremunerative activities, and unnecessary catastrophes for isolated families. Finally, maintaining the dynamic requires active use of policy instruments to sustain an inequitable distribution of wealth and income in a manner that I view as unjust. In "good times", central bankers actively suppress the median wage (while applauding increases in the mean wages driven by the upper tail). During the reset phase, policymakers bail out creditors. There is nothing "natural" or "efficient" about these choices.

The great moderation made aggregate GDP and employment numbers look good, and central bankers sincerely believed they were doing a good job. They were wrong. We need to build a system where changes in asset prices reflect the quality of real economic decisions, and where the playing field isn't tilted against the poor and disorganized in the name of promoting price stability.

Suppose that the Federal government were to offer sizable loan guarantees for any and all "green energy" companies. Any firm, including new entrants, would be eligible. The government would do some cursory due diligence, only to establish that the company in question would actually spend the capital it raised on real projects colorably linked to green energy (as opposed to, say, buying New Zealand dollars in a carry trade).

Wouldn't such a program constitute a stimulus to the economy? If sufficient leverage is allowed, it would lead in short order to a bunch of entrepreneurs founding companies on just a shoestring of equity and a whole lot of cheap, guaranteed debt. Firms with even a small likelihood of success would constitute real options worth more than the sliver of private capital at risk, so arbitrageurs would rush to create them.

Such a program would be a pretty direct form of fiscal stimulus. Although politicians and financiers enjoy pretending otherwise, contingent liabilities are still liabilities, and offering loan guarantees to all comers for risky projects is, ex ante, just a way of financing a government expenditure equivalent to the expected losses of the program. We shouldn't be surprised that an oddly financed stimulus would function as a stimulus.

But note that if, by good fortune, the artificially spurred new firms do surprisingly well and very few guarantees are actually paid, that wouldn't eliminate the ex ante stimulus effect of the program. It is not the actual transfer of Federal money that serves as the stimulus. The stimulus comes only and precisely form the certainty the program provides to investors that capital spent will be repaid, with interest.

So, suppose that the government does nothing, but "the market" becomes certain (correctly or not) that green energy companies are a sure thing. As long as the cost of capital to such firms falls sufficiently, precisely the same dynamic would take hold. We've just watched it happen, twice. When capital became very cheap to internet firms, entrepreneurs understood (and discussed quite openly) that there was an attractive lottery on offer, so why not get in? During the structured credit bubble, the market became convinced that some classes of debt yielding more than the "risk-free" interest rate were certain to be repaid. Entrepreneurs (both speculative borrowers and financial engineers) saw the arbitrage, and found ways of offering those classes of debt. In both cases, if the market had been right, everyone would have been happy. But when the market was wrong, it was someone else's cost. Many entrepreneurs walked away rich and happy. Others lost, but only a small amount relative to what they'd have made if things worked out differently. It was a good gamble for them ex ante.

Responding to Arnold Kling's "recalculation theory", Paul Krugman asks (as he has asked many times)

why [doesn't], say, a housing boom — which requires shifting resources into housing — ...produce the same kind of unemployment as a housing bust that shifts resources out of housing.

A housing boom, any kind of boom, is attended by an increase in certainty. Information is stimulus, confusion is contraction. A bust occurs when the market is unsure of everything, when market participants perceive better risk-adjusted return in holding government securities (or supply-inelastic commodities) than in financing real investment. Sectoral shifts per se have no clear implication with respect to variables like employment and output. But "hangovers" do happen, because powerful booms are periods when market participants make consequential decisions with great swagger and confidence, and busts are when we learn that despite their certainty, they were wrong. They are left not only impoverished and burdened by debt, but bereft of confidence in their ability to evaluate new opportunities. The best way to avoid the hangover is not to err so terribly in the first place. Easier said than done, perhaps, but that's no reason to cop out. We can build a better financial system, one in which degrees of certainty are attached and removed from economic propositions dexterously, rather than clinging like giddy leeches until a collapse.

Information is stimulus. As markets become more informed, money will be created and lent into the economy as surely as if the government printed and spent it. And stimulus is information, since governments do not spend randomly but do so in accordance with their own revealed certainties, which may or may not turn out to be wise. Poorly chosen stimulus and asset price bubbles are covert twins — only the identities of the people making bad decisions are different. Conversely good economic choices by governments can lead to outcomes as salutary as a healthy market. (See this very nice post by Bryan Caplan, and the articles cited.)

Information is a behavioral attribute, not an attribute of the external phenomena to which it may ostensibly refer. To say that an agent is informed means she behaves differently than an uninformed agent. Her behavior is less random, more predictable. To be informed does not imply ones information is accurate. (In general, accuracy is unknowable, both ex ante and ex post.) Information increases the volatility of outcomes, because it provokes larger and more concentrated bets than uncertain agents would take, creating large gains and losses depending on how adaptive the informed behavior turns out to be. It is often better, as a behavioral matter, to be uninformed than to be poorly informed.

But we do not always have the option of remaining uninformed. We cannot afford to hedge all of our bets. Whether via a great mis-recalculator in the sky or a political establishment largely captured by certain interests, new information will be manufactured. (I think it probable that government stimulus will substitute for market-generated information in the near term, as chastened capital market participants are more conscious of the hazards of certainty than policymakers are.) We will be spurred to take some actions and eschew others, and the structure of the economy will shift accordingly. Let's hope those choices are good, and do our best to help make them that way.

Update: While I was writing... Arnold Kling offers related and excellent "Thoughts on Probability and Uncertainty."

Note: I use the word "state" in this post to mean "the state", government generally, not state as opposed to federal government in the US. Thanks to Richard Serlin for pointing out the ambiguity.

Tyler Cowen writes

Now it's dead, everyone else has been blogging it...

The first thing to note is that vanilla is not dead. State-defined vanilla products are not an idea narrowly applicable to this moment's consumer finance challenges. They are, or could become, an important part of the regulatory arsenal in a wide variety of contexts. They are a tool whose development people with libertarian impulses (including, though you may not believe it, me) should view with cautious enthusiasm. At its core, the purpose of defining a vanilla option is to offer an additional choice, a well-understood default that helps consumers to weigh the purported benefits of exotic alternatives against the uncertainty costs they carry. Libertarians might reasonably object to a requirement that private enterprises offer vanilla products (although the objection is less compelling for industries that are and will continue to be state subsidized and regulated). But first and foremost, vanilla products are about cajoling into existence products that, despite their complexity, can be credibly certified as functional and nontoxic. The idea is in the "libertarian paternalism" / "nudge" tradition of not-exactly-regulation which, however objectionable, is less objectionable than the unqualified paternalism that may result from continued dysfunction and crisis. (Tyler might note a familiar form to this argument.)

The crucial question is whether governments could deploy vanilla products well. Both Tyler and Ezra Klein make the usual sad-but-true public-choice argument that, while vanilla might be a good idea in theory, what compromised politicians and bureaucracies actually offer might, in practice, not be so great.

They are undoubtedly right to be worried. But public choice concerns should always be the beginning, not the end, of a conversation. It's both incorrect and intellectually lazy to frame the argument as many "pro-market" commentators do (not Tyler or Ezra!), that sure, markets make mistakes, but politician-run governments make "worse" mistakes, so we are better off letting imperfect markets have free reign.

Governments and markets are dissimilar in the form and causes of their mistakes, and the badness of their errors is not uniformly rankable. Imagine that "what is to be done" is a radio signal being sent over two channels, both subject to maddening (but nonidentical) interference. A dumb strategy for extracting the signal would be to just decide that one channel is cleaner, and then throw away the other signal. Smarter strategies combine the information from both signals, and using information from each signal to help correct the errors of the other. There are domains where history and reason suggests that even terribly flawed markets provide a better signal than government. There are other domains where terribly flawed political processes generate a better signal than status quo markets. We should weight the different signals accordingly. But we should always be alert for opportunities to exploit the different strengths and weaknesses of markets and governments to produce better results than either could alone. How to do so, in real life rather than radioland, is an art. But vanilla products have the potential to be a masterpiece of the form. We have to insist on just one point.

A vanilla product must be defined by a uniform contract that regulators write and publish and that varies in a single dimension.

Of course regulators would consult with industry and consumer groups, and of course industry lobbyists would struggle to capture the process and embed their spicy "tricks and traps" into our vanilla creme. But they'll have a hard time doing so, as long as authorship and responsibility for the terms of the contract and its evolution over time rest with the regulatory agency.

Here's why. If the contract is written by the regulator, when consumers get screwed, they get screwed by the regulator as much as by the firm that sold them the product. Think about the politics of that. Suppose Struggling-Mom-Of-Four finds her credit card interest rate skyrocket because she missed a phone bill, and the industry had slipped a universal default provision in the "vanilla" credit card terms. When that story gets on the local news channel, it's no longer a just a story about megacorp screwing ordinary family. It's a story of the government screwing Mom-Of-Four, directly and on behalf of megacorp. When megacorp screws someone via a private arrangement, the populist "do something!" impulse is blunted by concerns about liberty and contract and personal responsibility. Citizens are genuinely divided about when government interference in private affairs is justifiable, so politicians get conflicting signals from their constituents, and unconflicted signals from megacorp not to interfere.

But citizens are not at all conflicted that the government should intervene to prevent the government from screwing people. If a contract written and owned by regulators has not-well-understood characteristics that leave consumers surprised and unhappy, representatives' "constitutent services" lines will ring off the hooks. Politicians that refuse to intervene against well-publicized injustices will be quite vulnerable, since they'll lack the usual philosophical justifications to defend inaction. So politicians will act. State ownership of vanilla contracts provide dispersed consumers a means to challenge concentrated industries for ownership of regulators, by virtue of the hypersensitivity of elected representatives to charges that they fail protect their constituents from rapacious government.

Another benefit of a uniform, regulator-owned contract has to do with the legal system. Every contractual arrangement is attended by legal uncertainty. A freshly written contract is quantum mechanical — words imbue a probability distribution that collapses to determinacy only when observed at the tip of a judge's gavel. (Usually no cats are killed.) Vanilla contracts offer an economy of scale in dispelling legal uncertainty. Disputes over vanilla arrangements would be quickly adjudicated, by public courts, not private arbitrators (binding arbitration would be a political nonstarter). Since there would only be one contract, legal precedents would be portable across providers. The characteristics of vanilla contacts would quickly become well-settled and widely known.

Vanilla contracts may or may not be dead this cycle, with respect to regulation of financial products. There are many other policy domains where vanilla contracts might be useful. I'd like a bit of vanilla with my health care, thank you very much. The Washington Post has an an excellent article about the too-little-discussed problem of tacit cherrypicking by insurers despite a formal "community rating" requirement (ht Dean Baker). This bit caught my eye:

A straightforward way to reduce gamesmanship is to standardize benefit packages, Precht wrote in a July report. One issue lawmakers must resolve is how much latitude to leave insurers over what they cover and how.

Sound familiar?

Extra vanilla

I realize I'm on time-delay, and that this is all so very last week. (When I am only a week behind, it's like living in the future.) Anyway, here's my RSS reader's dump of vanilla related links, in reverse chronologil-ish order. As always, if I've missed you, that's my lameness, not yours:

• Dupe Deluxe: The Sucker’s Market for Financial Innovation — Rob Horning
• Financial Innovation as a Tax — Karl Smith
• The plain vanilla option for financial competition — Tyler Cowen
• Confusing Disclosure and Information — Ezra Klein
• Vanilla — Robert Waldmann
• Commodities, Buying the Future — Mike Konczal
• Good-Bye, Vanilla Option — James Kwak
• Vanilla Option, Extra Scoop — Mike Konczal
• Assume You Don't Have a Can Opener — Ezra Klein
• Don’t believe the hype! Credit card sharks still hate the taste of plain vanilla — Dan Geldon
• While we weren't looking, Barney Frank sold us out on the Consumer Financial Protection Agency — lambert
• On Vanilla Products — David Merkel
• Why the vanilla option is necessary — Felix Salmon
• Vanilla Iced — Kady Liang
• And now for something completely different — The Epicurean Dealmaker
• Vanilla is a commodity — me
• Frankly I'm with Frank — Robert Waldmann
• Vanilla, iced — Ryan Avent (Free Exchange)
• Is There Any Consumer Protection Issue Blue Dogs Support? — masaccio (firedoglake)
• Vanilla fudge — Alyssa Katz
• Plain Vanilla — Kevin Drum
• Consumer Financial Protection Agency Gets Watered Down — Chris Walters
• Vanilla Products, Eulogy — Mike Konczal
• Financial Reform Already Takes A Hit — dday
• Obama caves to pressure on consumer financial protection — Edward Harrison
• The beginning of the end of meaningful regulatory reform — Felix Salmon
• Frank Waters Down Consumer Protection Proposal — Daniel Indiviglio

For the econgeeky, be sure to check out Karl Smith's piece (via Mike Konczal), which describes the costs of tutti-frutti into familar microecon diagrams, and this very nice paper by Gabaix and Laibson, that formally models much of the problem vanilla products are intended to solve (via Mike Konczal, via Tyler Cowen, via Christopher Shea who outlines the argument in accessible terms).

Do we have no fight left in us at all? Mike Konczal and Kevin Drum are excellent as always, but must we really write eulogies? Is one of the best regulatory proposals so far dead just because a single well-bought congressman says so?

Extracting the vanilla from the CFPA is not, as Felix Salmon put it "the beginning of the end of meaningful regulatory reform". It is the end of the end. Vanilla products were the only part of the CFPA proposal that was likely to stay effective for more than a brief period, that would be resistant to the games banks play. All the rest will be subject to off-news-cycle negotiation and evasion, the usual lion-and-mouse game where regulators are the rodents but it's the rest of us that get swallowed.

Wall Street's favorite comedian-politician, Barney Frank, was very savvy in framing the debate over the issue with his well-placed mischaracterization of vanilla products as "anti-market". That is bass-ackwards. The vanilla option is pro-market, because it is procompetitive. Of course, that is precisely why banks hate it: Vanilla products would turn basic financial services into a commodity business, and force providers to compete on price.

Ezra Klein is suitably depressed, but he's wrong when he writes that "the 'plain vanilla' provision was never likely to do that much." Vanilla products would be very popular, which is why they are so threatening. Financial services are an area where markets not only fail due to informational problems, but where participants are very aware of that failure. Consumers know they are at a disadvantage when transacting with banks, and do not believe that reputational constraints or internal controls offer sufficient guarantee of fair-dealing. Status quo financial services should be a classic "lemons" problem, a no-trade equilibrium. Unfortunately, those models of no-trade equilibria don't take into account that people sometimes really need the products they cannot intelligently buy, and so tolerate large rent extractions if they must in order to transact.

The price of assuring that one is not taken advantage of by financial service providers is not participating in the modern economy. You cannot have a job, because payments are by check or direct deposit. You cannot buy a home or a car, because for the vast majority, those purchases require financing. Try travelling with only cash for plane tickets, hotel rooms, and car rentals. People will "voluntarily" participate in markets rigged against them for the privilege of being normal. And we do, every day.

But define a reliable vanilla option, and the dynamic flips on its head. Instead of tolerating rent-extraction as a cost of participation, consumers put up with one-size-fits-all products in exchange for peace of mind. Most consumers benefit very little from exotic product features, and I suspect that many are made deeply nervous by the complex contracts they can neither negotiate nor understand, but nevertheless must sign. Vanilla financial products would be extensively vetted and and their characteristics would soon become widely known. Inevitable malfunctions would be loudly discussed in the halls of Congress, rather than hushed-up in rigged private arbitrations. Vanilla products would face discipline both from private markets (no one is suggesting we forbid other flavors) and from a very public political process. Politics and markets are both deeply flawed, but they are flawed in different ways, and we should take advantage of that. In Arnold Kling's lexicon, a market in which vanilla and exotic financial products coexist and compete offers the benefits both of exit and of voice. [Update: Chris Mealy notes in the comments that these terms are due to Albert Hirschman]

Rather than being anti-market, vanilla financial products would help correct very clear market failures that arise from imperfect information and high search costs. It is the status quo that is anti-market.

I'm sympathetic to the principled libertarian objection to having the government require that private parties offer a product they otherwise might not. No one should be forced to offer vanilla financial products. Small-enough-to-fail boutiques should be free to offer only the products they wish. However, if an institution wishes to avail itself of government-provided deposit insurance or to access Fed borrowing facilities, it is perfectly legitimate for the government to set requirements. The government can choose not to offer its safety net to institutions that don't offer vanilla products, just as banks currently choose not to offer me a credit card unless I sign up to binding arbitration and unilateral contract changes. I fail to see why one is coercive and the other not. (The government has no monopoly on deposit insurance. Private insurers are free to provide similar insurance, and do so for many financial service companies already.)

An Economist anonobloggeer has some peculiar non-compliments about the vanilla products proposal:

The vanilla offering seems to be intended to substitute for sophistication or research on the part of the customer, but I'm just not sure that's a good way to approach the issue. As best I can tell, the vanilla plan wouldn't mandate the price of the simple option; just because a bank would have to offer a vanilla mortgage loan doesn't mean it would have to offer a competitive vanilla mortgage loan. If that's the case, banks could easily use high rates on the simple products to steer individuals toward the complex offerings. Or, the vanilla rule could actually serve to direct bank collusion toward high-priced, high-margin products.

Just because a commodity exchange standardizes the quality of corn that must be delivered into a futures contract doesn't mean that any seller has to offer corn at a good price. So true! But sellers that offer commodities at above market prices don't usually find buyers. Since vanilla financial products would be commodities, banks would have to universally collude to offer them at inflated prices in order to bilk consumers. Competing vanilla project offerings would (at least they should) vary only on a single dimension (e.g. an interest rate). Points, fees, penalties, etc. would be homogeneous or uniformly pegged to the core price. Banks are very, very good at forming tacit cartels, but colluding on complicated terms and conditions is easier and less likely to attract the antitrust authorities than fixing a headline price.

More from the econoanonoblogger:

To me it seems like the more effective solution would be to require that financial institutions explain, in detail, each and every fee they are assessing (or might potentially assess) to customers. That would inform consumers of what's going on in the monthly bill, and it would create an incentive to reduce the number and complexity of fees, as lengthy explanations would be a hassle for all involved and would reduce business.

One of the great errors in modern policy is to confuse disclosure with information. It is not the case, currently, that banks secretly take your money without itemizing the charge on some statement. (Sometimes when they take your money they call it "service fee" or something equally nondescriptive, and it'd be nice if that practice went away.) Rather, banks intentionally define contracts in such a way that the cost to many customers of understanding and competitively shopping all the dimensions of the product seems higher than the cost of terminating the search and signing the dotted line. More detailed disclosure doesn't eliminate, and can sometimes exacerbate, the real information costs customers face, which derive from the complexity of the required analysis and lack of information about alternatives, not from an absence of product data. Of that we all have pages, with more arriving every month. You might think there'd be a market for ostentatious simplicity, and there might be. But no bank's lawyers would sign off on a single page, 12 point text, no-extratextual-incorporation-or-unilateral-modification contract. When routine contracts get more complex than that, it's just gibberish competing with gibberish for people who have lives. Some financial products are necessarily complex. But one way of managing complexity is standardization. It may be worth it for consumers to carefully study the one contract they will probably sign in a way that it would not be worth poring through 100 freeform contracts, 99 of which they will never sign.

The most serious objection I know to vanilla financial products is that they would be harmful precisely because they would catastrophically succeed. The theory is that nothing is more dangerous than a commodified bank, and the evidence is May Day, 1975, when the SEC ended fixed stock trading fees in the brokerage industry. Some commentators (e.g. Barry Eichengreen) claim that by eliminating a stable, cushy profit center, the May Day deregulation forced gentle investment banks to become hungry innovators, that the financial system has grown progressively less stable because under cut-throat competition risk-takers dominate (until they self-destruct and take the rest of us down with 'em). I don't buy the May Day story, but for the sake of argument, let's suppose it's true. Let's suppose that, in the name of stability, the best policy would be to ensure banks easy profits so that they needn't dabble in dangerous things. Then two conclusions follow:

1. If we are going to strike a policy bargain whereunder banks get a nice sinecure in exchange for a promise of stodgy mellowness, it seems reasonable that they should commit to the stodgy mellowness. Dull, subsidized banks should be heavily regulated banks, or, to use the term of art, "narrow banks".

2. If we are going to impose a regime that ensures bank profitability, we ought to do so in a reasonably equitable way. Business models that hide profit generators in complex contracts, or that extract fees especially from the disorganized and naive, are not reasonable instruments of public policy for keeping banks healthy. If we do go with the coddled but heavily regulated model of banking (not my preference!), and we're not willing to have the Treasury end the capitalist charade and just cut checks to its payment-systems subcontractors, then a decent approach would be to have narrow banks offer only vanilla products and provide monopoly rents by putting floors under fees and ceilings above deposit interest rates (as existed in the US until the 1980s). Under either a competitive or "regulated utility" model, the fairness and informational case for defining standardized vanilla products remains compelling.

I think people like Barney Frank, when they try to sleep at night, have been sold on the "we need healthy banks, so let's protect their profit centers" story, although they'd never admit to it while scoring points comparing powerless people with furniture. I wonder if it even occurs to Mr. Frank that maybe something serious should be demanded of banks in return for state protection of market power at the expense of the weak and disorganized. But then Mr. Frank has already gotten very much in return.

So I've belabored the distinction between transactional and revolving credit quite enough, I think. And I'm pleased, reading around the intertubes, that people seem comfortable with that distinction, and with the idea that it might be good public policy to treat these two forms of credit differently, despite attempts by credit card issuers to blur the lines. Yay!

But my previous piece seems to have left readers with two pretty big WTFs:

1. Don't we already have a ubiquitous and perfectly good transactional credit product called debit cards?
2. What would be the point of having the government provide a charge card?

Let's take those in turn.

1. Are debit cards good enough?

Debit cards are indeed a transactional credit product. Specifically, they offer overnight transactional credit which is automatically repaid from a designated bank account. Problem solved, right? Everyone should just use debit cards.

Debit card use is on the rise in the United States (and in other countries use of debit cards is often the norm). But very many of us — possibly even you, dear reader — have both debit and credit cards, but prefer the credit card, even though we pay our balances monthly. Why might that be?

Here's a minibarrage of reasons. Using a credit card...

1. is more convenient, since you needn't keep track of an account balance unless you are very near your credit limit;
2. is cheaper because it doesn't require keeping a substantial buffer of funds in a zero or low interest checking account;
3. is cheaper because you can get some of the interchange fee refunded via rewards programs;
4. is safer, because it offers the opportunity to review and occasionally repudiate transactions prior to transferring funds;
5. is safer, because stronger consumer protections are offered in the event of theft or fraud;
6. is safer, because if the cardholder faces an unexpected liquidity crisis, she can shift from transactional to revolving credit. Cardholders have the option of making partial payments with little penalty.

If we want a purely transactional product, we'll have to do without that last, the option to revolve. A competitive market for transactional credit might or might not offer kickbacks on transaction fees (American Express does). But to be conservative, let's suppose not. That still leaves four pretty good reasons why an American-Express-style, pay-at-the-end-of-the-month "charge cards" are superior to debit cards. Banks have tried to address the consumer safety and dispute resolution issues that disadvantage debit cards, but there is still a big gap in perceived and I think actual safety. But the first two issues are the zingers. To get the same payment flexibility with a debit card that you would get with a charge card, you need to maintain a large checking account balance, which costs you in interest foregone. If you keep your buffer savings in an investment account, then you have to track your account balance carefully and transfer funds between accounts frequently. Even with a $95 annual fee (the going rate for a basic American Express card), most families would be better off earning interest on their savings and paying the fee than keeping substantial savings in a no interest account. (Right now this may not be true, because interest rates are so low that you don't lose much by holding funds in a checking account. But it has usually been true when we've not been in ZIRP mode.)

There are arguably behavioral disadvantages to the buffered debit card strategy. If you spend unusually much in a month, there is nothing that forces you to replenish your buffer. But if your savings are held in a separate investment portfolio, transferring funds to cover a shortfall may feel like raiding the piggy bank. Avoiding that might encourage people to not spend more than current income. If you keep substantial savings in a debitable checking account, the mechanics of dissaving are indistinguishable from the mechanics of ordinary spending, and there is no third party demanding that you make yourself whole at the end of the month. With a charge card and a separate investment account, spending beyond income compels painful, explicit transfers from what you thought was savings, which might motivate you to replenish the overdraft quickly. (This entire paragraph is bullshit if you don't believe in things like "mental accounting" — money in an investment account is viewed differently than ordinary spending money — or "anchoring" — once you reach a level of "savings", it comes to define where you "ought to be", and you will strive to maintain or recoup that level. Do you believe?)

A lot of debit card enthusiasts will find all this comparing of "buffered" debit cards and charge cards artificial: Debit cards, they will claim, are good because you can't ever spend more than the cash you have. It's as simple as that, right? But we keep wealth we may wish to spend in a variety of different forms, and many of us even have multiple bank accounts. Debit cards without a cash buffer don't really do the job of transactional credit, that is letting you spend what you can near-term afford without having to much worry about how your (liquid) wealth is held. One might argue that extra friction is good. Perhaps people are so undisciplined that we should have to track balances and arrange transfers prior to even routine spending. That's an argument for an electronic "cash only" economy, and I think it's farther than we need to go. It's a judgment call, but my sense is that people handle short-term transactional credit pretty well, and benefit from the convenience of it, while indefinite-term, low monthly minimum revolving credit frequently becomes a trap.

2. What would be the point of government provided transactional credit?

In the previous post, I suggested that the government could offer a "Treasury Express" card, supplying access to transactional credit as a public good. To my surprise, some people seemed to actually like the idea (e.g. Ezra Klein, Matt Yglesias, Doug Singsen). Others understandably dislike the idea of more state involvement in a sphere that has been traditionally left to the private sector. Kevin Drum just doesn't see the point:

We already have "Treasury Express" cards: this is basically what debit cards are, and they provide the same benefits of transactional credit that regular Visas or Mastercards do. Why do we need the government for that?

That leaves us with the problem of limiting revolving credit, which is the same problem we have now. Do we need firmer rules on interest rates, fees, and penalties? Better bankruptcy protection? Bans on things like universal default? An end to tricks and gimmicks and fine-print-laden marketing come-ons? More sensible ways of setting credit limits? Maybe. Probably. But unless Steve is suggesting that we essentially ban credit cards entirely — and then create some kind of federal mega-authority to limit every other kind of consumer credit too — those are all the same issues we have now. I'm not really sure what his proposal would accomplish.

If it's true that the differences between a pay-at-the-end-of-month charge card and a debit card aren't very important, then Kevin's right, and there wouldn't be much point. But suppose that I am right, and people really value the efficiency, convenience, and safety of a charge card, even if they do not intend to run a balance. As things stand now, they have two choices: they can pay $95 per year for an American Express card, or they can get the same product for free with a credit card, as long as they accept a dangerous option not to pay in full at the end of the month. (If having to pay more for fewer options strikes you as odd, you are not alone. In two posts, the amazing Rortybomb tries to make sense of credit-card-pricing mysteries, including this one, using ideas we learned in finance class. Only his training in critical theory prevents a Scanners-style head explosion.)

A free basic transactional credit product would let people avoid signing on to temptation when all they want is a charge card.

A public option would also create political space for better regulating revolving credit. The obvious way to limit revolving credit to those most likely to use it well is to force rationing via price controls. In English, that means we should bring back usury laws. Credit card companies won't offer cards to financially insecure customers if interest rates, penalties, and transaction fees are capped. Under the present terms of debate, that would be a bad thing, "limiting access to credit". But the meaning of that phrase is very slippery. "Access to credit" is let to stand for participation in the modern economy, i.e. the ability to rent a car or hotel room, to make purchases or pay bills on-line, etc. If we unbundle those good things from credit cards, what usury laws would limit is "access to high-interest, high-fee unsecured revolving credit". Put that way, it doesn't sound so bad.

The existence of universal charge card accounts would offer some side benefits. Such accounts would provide uniform and convenient means for the government to make payments to citizens, rather than mailing out stimulus or social security checks. It would make it easier to implement flat transfers, which I consider a better form of fiscal policy than tax cuts or aggressive government spending.

Finally, it's worth thinking generally about when public sector competition helps to keep the private sector honest. Critical industries that are prone to concentration due to network effects or economies of scale, that are informationally opaque, or that have high barriers to entry may benefit from the implicit threat presented by even inferior public sector competition. I am a huge fan of UPS and Fedex, and generally prefer their services to those offered by the post office. But I wouldn't be at all comfortable disbanding the postal service, even though as a taxpayer I am forced to fund its losses. Perhaps I underestimate the magic of the marketplace, but if USPS weren't there to put under a floor under the quality and price of service offered by the private couriers, the degree of concentration and barriers to entry in that industry would make me nervous. There are other examples: State schools offer useful competition to private colleges and universities. A "universal Medicare" option in health care would serve as a low bar that private sector providers would have to overleap. Consumer banking services may be an industry where public sector competition would be useful. We don't want the government making fine-grained decisions about the allocation of business capital. There are good reasons to think that capital markets and profit-motivated relationship lenders do a better job of that than the state. But there's little reason why the public sector shouldn't provide basic transactional credit and checking account services. (Other countries have state-affiliated banks that compete with fully private institutions. A public/private banking ecosystem is not a radical idea.)

All that said, there are good reasons to oppose a "Treasury Express" card. I share the cynicism of libertarian critics. The only thing I trust less or want less involved in my life than the government are cartelized private corporations. To the degree that people choose to make payments with publicly provided cards rather than with cash or private credit products, the government would obtain detailed individual payment histories. This may (or may not be) worse than Chase or Citigroup having access to the intimate details of our lives. Universal access to transactional credit might do harm by altering the incentives of people on the margins of the economy. For example, someone who currently lives in the cash economy might max out, and not repay, their "Treasury Express" card. Even though the penalties they'd face for nonpayment would be mild, they would have created a new hurdle they'd have to overcome if they wish to reintegrate themselves into the mainstream. Dispute resolution might be a nightmare for a public program. Disputed credit card transactions leave either a customer or a merchant feeling screwed. What is now private cause to switch banks would suddenly become a contentious matter of public policy. The government might meddle in what people can buy and sell with the card, creating a nanny-state non-neutral form of money. Government entry into the banking sector with a limited, basic product could lead by increments of mission creep to a state-subsidized monopoly taking over financial services and credit allocation. These are all real concerns that would have to be addressed, if we were to give public sector transactional credit a try.

Neither this post nor the previous one has been intended as full-throated advocacy. A state-provided universal charge card is a speculative idea that merits further consideration. It might be worth doing. It might not be. But as we negotiate with the banking sector going forward, we will no doubt hear dire warnings about how this or that regulation will force thousands of widows and orphans into bartering chickens for shelter. We should keep in mind that if banks won't provide the simple, consumer-friendly financial products we require, we can create other options.

By the way, if you haven't read Felix Salmon's candid correspondence with former industry insiders about credit card business models, do that now. It is remarkable. For more of the weekend's credit card links, try Conor Clark on credit cards for college kids (and Richard Serlin's response), Michelle Singletary, and all the other brilliant writers I'm sure I've missed.

Megan McArdle...

[M]aybe it's worth remembering that the tyranny that credit scores exercise over our imagination have everything to do with the fact that we've built a society so utterly dependent on credit.

Kevin Drum responds...

[T]here's nothing per se wrong with the fact that modern economies are so dependent on credit. Widespread use of credit really does make life more convenient, really does make banking more efficient, really does enable useful advances like online shopping, and really does allow easier access to goods and services that would otherwise be difficult to get hold of. Used in moderation, it's good stuff. I sure don't want to return to the days of hauling around travelers checks whenever I fly off to Europe.

Speaking for myself, my jeremiads against the credit-industrial complex have never been meant as an attack on widespread access to credit itself. Used reasonably, credit cards are a boon and credit reporting is a necessary part of providing credit responsibly in a big, complex world. That said, credit is critically important to everyday living now, and that means that it needs to handled fairly and transparently.

We won't get very far in the debate about credit in the US economy if we fail to distinguish between transactional and revolving credit. These are two are fundamentally different products, and much ill has come from conflation of the two. All of the good things Kevin attributes to widespread credit access are benefits of transactional credit. Because credit cards have often bundled transactional and revolving credit together, it is easy to attribute these good things to revolving credit. That's a mistake. Transactional credit is essential, and might even be publicly provided. Revolving credit is a double-edged sword.

Transactional credit is a means of decoupling the process of making payments from the form in which ones liquid wealth is held. More simply, if you have the money to pay for what you are buying, but just don't want to carry cash or keep track of your checking account balance every day, you are making use of transactional credit. Many people only use credit cards for transactional credit. They pay off their entire balance each month.

Revolving credit is a different product. It provides a means by which people can spread the cost of purchases over an indefinite period of time. If you wish to go on a vacation, and can afford to pay for it from your next six months' salary, but could not easily come up with the money now, you are making use of revolving credit.

Both transactional and revolving credit are useful, and conventional credit cards give consumers access to both. But revolving credit is much more prone to abuse than transactional credit. Though economists hate to admit it, it is a fact of life that human beings do not, in general, seem to have time-consistent preferences. If preferences aren't time-consistent then people are prone to making short-term deals that they will seriously regret in the distant future. Revolving credit is like morphine: When used properly, it can be very useful. But experience has shown that it can cause great harm if used incautiously.

Revolving credit needn't be bundled with transactional credit. The traditional American Express card, for which you are charged a flat annual fee and pay your balance in full each month, is a transactional pure play. Transactional credit has different characteristics than revolving credit. In particular, consumers can get most of the benefits of transactional credit with low credit limits, perhaps twice a typical month's expenses. (If restricted to transactional credit, consumers may have to find other means of paying for occasional large purchases. They might need to put funds into a bank account in advance, and pay via debit. As long as such purchases are infrequent, this is not a terrible burden.)

If we set aside "tricks and traps", transactional and revolving credit products should inspire very different business models. Transactional credit resembles an insurance product, while revolving credit is like a traditional loan. The primary benefit a consumer receives from transactional credit is not the financing debt, but the option to purchase at any time without having to track specific account balances or coordinate transfers. This service is provided regardless of the size of any given month's balance. Consumers should be (and historically have been) willing to pay a flat fee for it. Even very modest interchange fees more than cover the cost of capital on the loans, so fees can be quite low and might go to zero for some customers. (A 1% interchange fee amounts to a 12% annualized rate for a low-balance, 1 month loan. Conventional interchange fees of 2-3% offers exorbitant returns on 1-month loans, part of which may be rebated via rewards programs.) There is no reason why transactional credit can't be a fine business, and it has been, both for American Express and every other credit card provider sending unsolicited offers to people whom they know pay their balances off monthly.

Revolving credit, when it is not about tricks and traps, is all about the interest rate. Revolving credit is prone to abuse, and should be made available cautiously, not automatically or indiscriminately. Credit card interest rates should simply be capped, which would prevent less creditworthy borrowers from gaining access to revolving credit lines. That is a feature, not a bug. In a world where agents have inconsistent time preferences, paying high interest rates for present consumption is prima facie evidence of selling out future selves for present goods. Rational high interest rate borrowers are either those who intend to default (thereby extracting a wealth transfer from the more responsible subset of the population), those who are financing goods that will yield benefits over time, and those who face an unusual emergency which requires the future be held hostage to the present. We want to deny credit to the first group, the rational defaulters. People who are financing goods that will yield benefits over time can usually get credit on much better terms by taking out securable, asset-specific loans. We should encourage the resurgence of secured vendor financing, because that form of credit can offer huge savings to less creditworthy borrowers, compared with ubiquitous unsecured plastic. Finally, the usefulness of high limit credit cards as kind of insurance is undeniable, but dangerous. Every day has its crises; that is the human condition. It is easy to err by taking on "absolutely necessary" debt today that leaves one absolutely destitute tomorrow. We should develop better forms of emergency insurance than unsecured debt. To the degree that we do rely upon access to credit as a reserve cushion, it shouldn't be attached to our instrument of casual commerce. It should be special, and have a "break-glass-in-case-of-emergency" feel to it.

Access to revolving credit should be rationed, but transactional credit should indeed be ubiquitous. Not having to carry and count cash, deal with paper checks, or even worry about some particular account's balance at the time of purchase are important benefits. Indeed, an efficient payments system is a public good. That's why states are in the business of establishing currencies, right?

In fact, while transactional credit provision is a perfectly good business, it might be reasonable for the state to offer basic transactional credit as a public good. This would be very simple to do. Every adult would be offered a Treasury Express card, which would have, say, a $1000 limit. Balances would be payable in full monthly. The only penalty for nonpayment would be denial of access of further credit, both by the government and by private creditors. (Private creditors would be expected to inquire whether a person is in arrears on their public card when making credit decisions, but would not be permitted to obtain or retain historical information. Nonpayment of public advances would not constitute default, but the exercise of an explicit forbearance option in exchange for denial of further credit.) Unpaid balances would be forgiven automatically after a period of five years. No interest would ever be charged.

Let's think about how this would work. For most people, access to various forms credit — transactional credit, auto and home loans, unsecured revolving credit, whatever — is worth more than $200 per year. Although people might occasionally fall behind, for the most part borrowers would pay off their government cards, simply because convenient participation in the economy is worth more than a once-in-five-years $1K windfall. However, people with no savings and irregular income (for whom transactional credit is a misnomer, since they haven't the capacity to pay) might well take the money and run. The terms of the deal amount to a very small transfer program to the marginal and disorganized, and a ubiquitous form of currency for everyone else. People with higher incomes would want more transactional credit, or revolving credit, which they would acquire from the private sector.

I've posited that people often have time-inconsistent preferences. Am I, them, inconsistent to suggest that most people wouldn't people take a free $1K today and be stuck without credit thereafter? No. The degree to which people underweight future costs varies between individuals, and is changeable. Most people do work to avoid present choices that will create future hardship. (Many people arguably overweight the future.) However, high-limit revolving credit is a particularly nasty trap for those who even slightly underweight future costs. One nice aspect of a low-limit, indiscriminate, mechanical public credit system would be educational. Many younger people would spend some period of time modestly in debt and shut out of the credit economy. This would provide a more gentle lesson than the current practice of running up revolving balances in college and working desperately for years to pay them down.

The notion of transactional credit as a state-provided public good is speculative. Maybe it's a terrible idea. Regardless, the distinction between transactional and revolving credit is crucial. A modern economy probably does require widespread access to transactional credit. But revolving credit is a different story entirely, and we would be better off controlling it more carefully. We shouldn't be shy about adopting policies that would curtail the provision of unsecured revolving credit, as long as transactional credit is protected.

Some of this was inspired by a conversation with the excellent keyholez over Twitter. (I have been playing on Twitter recently, and am still trying to decide what I think of it. If you're into that here's me and here's keyholez.)

I am continually amazed at the quality of comments that Interfluidity attracts. But commenters to the previous post truly outdid themselves. I proposed a "novel" security intended to make debt-to-equity conversions gradual and automatic, in hopes of avoiding disruptive "discrete" bankruptcies. Commenters quickly pointed out earlier work along the same lines, and some serious problems that I hadn't considered. All in all, I remain convinced that "continuous bankruptcy" is a goal worth pursuing. But I'm equally convinced that the specific security I suggested would probably not quite do it.

First, it turns out that better minds than mine have considered debt-like securities that would convert to equity as firms became undercapitalized. The excellent Economics of Contempt pointed out two antecedents, Mark Flannery's "Reverse Convertible Debentures", and a recent proposal by the "Squam Lake Working Group on Financial Regulation". The Squam Lake piece is a diamond in the rough, but Flannery's ideas are carefully developed. Flannery considers and tries to address most of the problems I would have missed without help from the comments. He also includes a detailed review of related work.

Flannery's Reverse Convertible Debentures and my proposed "IACCPEs" are both fixed income securities convertible to common stock at a price not set in advanced, but determined by the market price of the stock at the time of conversion. Commenters Alex R and Independent Accountant were reminded of the infamous death-spiral convertibles from the late 1990s. bondinvestor pointed out that a form of catastrophe bond, "catastrophic equity puts" are similar, in that they allow the issuer to convert a fixed income obligation to equity in the event of a prespecified bad event. (Here the bad event that would trigger conversion is running short of capital.) Thomas Barker and jck of Alea pointed out that these securities resemble Islamic "sukuk convertibles". jck suggests that these, like the death-spiral convertibles, have mostly not worked out so well. (I'd like to know more about Islamic convertibles, both the theory behind them and their history in practice — if you know of a good reference, please do drop a note in the comments.)

The trouble with my proposal is that it failed to adequately consider how both investors and firms would try to game them. Alex R pointed out that investors shorted the hell out of "death-spiral convertibles". Those who owned them gained if prices could be pushed downward, as lower stock prices meant ownership of a larger fraction of the firm upon conversion. Other traders had incentives to short as well, since dilution caused share prices to drop post-conversion. If the conversion trigger is predictable to investors, and the same dynamic might hold for IACCPEs. (beezer thought this likely as well.)

Also, Alex R and Awake noticed that conversions are effectively at the option of the firm, since "IACCPEs" are only convertible "in arrears" — that is, when a firm hasn't paid a dividend owed on the preferred shares. Since conversions (under my proposal) are made on terms favorable to IACCPE holders and disadvantageous to incumbent equity, managers might insist on paying dividends to the bitter end, even if overall firm value would be enhanced by preserving cash. Shareholders might prefer to "gamble for redemption" rather than transfer wealth up the capital structure and adopt a more conservative strategy. (In other words, the option value of an undercapitalized firm might be worth more to the original shareholders than the expected value of their fraction of a better capitalized firm.) Further, the fact that convertibility is at the option of firm management creates an incentive towards opacity and surprise non-payment of dividends. Firms would want to skip dividends when shares are overvalued (so that conversion would be on terms favorable to existing shareholders), and would want to hide any possibility of skipping dividends in distress (so that short-sellers don't front-run à la death-spiral convertibles. Modifying capital structure in a way that creates incentives for managers to reduce transparency and promote mispricing doesn't seem like a good thing.

Mark Flannery's "Reverse Convertible Debentures" try to avoid gaming by managers and short sellers in three ways: First, RCDs convert automatically, at the option of neither issuers nor investors, based on a debt-to-market-equity trigger. Secondly, the securities are designed to avoid any transfer of net wealth from equityholders to creditors (or vice versa) upon conversion. Conversions would be based on the market value, not the par value, of the RCDs at the time of conversion, and the market price of the stock. In theory, this would make for a perfectly even exchange, so that no party would have strong incentives to game the trigger. Finally (as foreseen by commenter Bill) "triggered" RCDs wouldn't all convert all at once. Only the minimal fraction required to bring a firm to a set level of capitalization would be affected, which again might make the threshold event less of an event worth manipulating.

I don't think that Flannery has solved all problems of strategic behavior — both the "automatic" trigger and the effective conversion ratio are susceptible to gaming, and I don't think it would be either possible or desirable to define terms under which investors would be indifferent to conversion. Nevertheless, Flannery's paper makes a serious attempt to address the problem of gaming the trigger, which really is the achilles heel of this genre of proposal.

But let's pull back. There might not even be a need for an explicit conversion scheme. Plain vanilla cumulative preferred equity has a built-in duality. As long as an issuer is paying out dividends, cumulative preferred equity is very debt-like. Investors expect a fixed coupon, and firms have strong incentive to pay it (so that common shareholders can take dividends, and because nonpayment of preferred dividends is taken as a bad signal by the market). When preferred goes "in arrears", it suddenly becomes very equity-like: Its value becomes dependent on the promise of dividend payments at some unspecified future date, and the probability that dividends will actually get paid is sensitive to the operating performance of the firm. Cumulative preferred equity has already been invented. So why do we still have a problem?

The first, easiest, and most important thing we could do to reduce the systemic risk and deadweight bankruptcy costs caused by brittle capital structure is change the %*$%&! tax code to eliminate the tax preference for debt over preferred equity. Long-term debt and cumulative preferred equity are basically identical, except for two things: 1) Taxpayers subsidize the issuance of debt while 2) debt contracts are enforceable via socially costly bankruptcy. Taxing preferred equity dividends but not interest on debt is like taxing people for the privilege wearing seatbelts, then wondering in gape astonishment at highway mortality rates. It is bass ackwards. We can either make dividends (at the very least cumulative preferred dividends) tax deductible, or we can make interest payments non-deductable (as Richard Serlin prefers). But we have got to stop tilting the capital structure scales towards bankruptcy-enforceable debt finance. The status quo is absurd, ridiculous, untenable.

(I have yet to encounter even an attempt to justify why interest payments should be deductible but CPE dividends not. There are arguments by definition — expenses are deductible, obligatory payments are expenses, obligatory means enforceable by bankruptcy, q.e.d. But the tax code needn't be slave to an arguable and legally alterable set of definitions.)

If cumulative preferred equity were not tax disadvantaged, firms might find that it is not much more expensive than debt. But there are two other problems with vanilla CPE as "natural convertibles". First, there is the question of control. When a firm runs into trouble and then deep goes into arrears on cumulative preferred equity, economically speaking, the preferred becomes equity-like, while the common stock becomes option like. Unfortunately, control usually in the hands of common equity, whose incentives may be to maximize the volatility rather than expected value of firm assets. Secondly, as a matter of convention, financial regulators and analysts often treat preferred equity like debt. When a financial firm is in jeopardy of skipping preferred dividends, regulators become inclined to intervene their "prompt corrective action" mandate. Markets may view skipping preferred dividends as tantamount to a default, and question firm viablity, potentially leading to self-fulfilling distress.

The second issue may or may not be easy to address. Economists pay too little attention to the role of convention in shaping corporate finance, and even less attention to the dynamics of convention. If preferred equity is going to serve as a means of strengthening firm capital structures, we need to move to an equilibrium where occasional periods of missed dividend payments are normalized, and not taken as a death knell for a firm. Sure, skipping preferred dividends is usually bad news, just as it is not a good sign when firms cut common stock dividends. But a skipped dividend needn't signal a death spiral for a well-capitalized firm. Perceptions might change naturally if preferred equity managed to shed its tax disadvantage. Firms might then lever themselves to the hilt with preferred stock, and over a period of time, investors would observe viable companies go into brief arrears and then emerge. Regulators could help quite a lot by treating preferred equity as equity full stop rather than as a kind of pseudo debt. For financial firms, regulators should commit not to intervene if preferred payments are skipped or threatened, so long as the value (primarily the market value) of a firm's total equity base remains high. Regulators should treat preferred equity as high risk capital. Regulated insurers, pension funds, etc. should be required to invest as if preferred shares are no less risky than common stock. Regulators should make it clear that during resolutions, preferred shares would go to zero as easily as common if any payout is made on guaranteed liabilities.

Control issues may be more difficult to address, but not impossible. In one of several very helpful e-mails, Economics of Contempt pointed out that

[D]ebenture indentures have included a wide range of covenants over the years that were designed to serve as early warning systems (e.g., the negative pledge clause). No need to reinvent the wheel.

Debt covenants are the means by which divergence of interests within firm capital structures are usually addressed. Lenders do not explicitly exercise control over firms, but they can attach very stringent conditions to loans that limit the ability managers and stockholders to gamble with creditors' money. Lenders hold a very big stick — if covenants are violated, they can demand immediate repayment of funds that will usually have already been spent. Debt covenants are violated quite frequently, but firms are not often forced to liquidate assets and cough up the cash. Instead, occasional violations allow lenders to renegotiate the terms of loans from a position of strength. So, even though lenders are not represented on a company's board of directors, their interests are protected both passively (firm managers strive not to violate covenants) and actively (when a covenant is breached — and they are designed to be breached easily — lenders can intervene to actively shape firm decisions).

Preferred stock can include covenants as well, but how to enforce them is a tricky question. It would defeat the purpose of replacing debt with preferred equity if preferred shareholders could force repayment upon a covenant violation. I know very little nothing at all about how preferred stock covenants are written and work in practice. (Hint, hint, amazing commenters!) However, via Google books, I came upon the notion of "contingent proxy" in Raising Entrepreneurial Capital by John Vinturella and Susan Erickson. Apparently there is precedent for clauses that transfer voting rights as "a penalty for breach of covenant". This idea could be very helpful. As preferred shares become more equity-like and common shares take on properties of options, control might be transferred quite directly to preferred shareholders without a formal debt-to-equity conversion. Transfers of voting rights could be architected so as to be gradual. They might be a function of the degree to which a firm is in-arrears, rather than the mere fact of in-arrear-edness, so that common shareholders would not much surrender control during occasional, brief lapses. Preferred shareholder control rights could depend upon the full mix of the solvency and health metrics typically included in debt covenants, so that there is little incentive by firm managers to game any one potential trigger. Formal options to convert to common equity could gradually be extended as well. Obviously, devils are in the details, and I don't yet know enough to attempt an exorcism. But, as EoC reminds us, we don't have to reinvent the wheel. Investors up and down firm capital structures have been eying one another warily for centuries. We have millions of paranoid legal documents to draw ideas from.

It seems to me that the right species of "trigger convertible", in David Murphy's coinage might well evolve from negotiation and experimentation, if fairly minimal changes in policy were made. First, we desperately need to level the tax playing field between debt and equity. At the very least, the differential treatment of debt and near-debt-but-safer cumulative preferred equity needs to be eliminated. Secondly, for financial firms, regulators should radically increase capital requirements for financial firms. Bankers will do what they do best, which is to seek the cheapest means possible of pretending to hold capital by inventing the most debt-like equity that they possibly can. Regulators should scrutinize these instruments carefully, but only to enforce two very simple requirements — that under no circumstances are dividend or principal repayments obligatory, and that no agreements made between the different classes of equity encumber firm assets or compel behavior that would compromise the interests of claimants higher in the capital structure. (Regulators, and creditors, would have to guard against "poison pill" arrangements, in which some form of "equity" is cheap because purchases have some means of sabotaging the firm if they are not paid. But this is not a new problem, see the negative pledge clause in EoC's comment.)

Finally, regulators would have to severely curtail the ability of regulated entities from holding preferred equity. All equity securities should be treated like common stock for risk-weighting purposes. No matter how the contract is writtem, when a security acquires a sufficiently high weighting in the portfolios of insurance companies, pension funds, and banks, it acquires an implicit state guarantee that issuers will aggressively exploit. One lesson of the current crisis is that the distinction between debt and equity has less to do with the legal characteristics of securities than with the political connectedness, financial interconnectedness, and risk-bearing capacity of the entities who hold them. Equity securities, preferred or common, must be restricted to parties whose losses the state would tolerate.

I'd like to propose a financial innovation that I think would actually be good (besides the ATM). It would be a new security that firms could market to investors, just like CDOs and all of that good stuff. But rather than being a means of expanding the supply of credit (the questionable purpose of most "financial innovation"), this investment product would change the character of credit provided by investors to firms. It would provide an alternative to the customary form of corporate debt.

True believers might argue that if what I suggest were a good idea, a free capital market already would have discovered it. I'm not a true believer, but I'll make common cause in part, and point out that securities like those I propose are presently tax disadvantaged, so capital markets have not been free to discover them. In particular, if dividends on preferred equity were tax-deductible to firms like interest, perhaps these securities would already be commonplace. But I'll reveal my cruel, statist heart by hinting that since firm managers may be myopic in their preference for cheap financing, and since distress costs are in part external to firms, an active policy tilt in favor of more robust capital structures might be worth considering. [1]

I'm suggesting a new financial instrument. Here's its catchy name: "In-arrears convertible cumulative preferred equity", or "IACCPE". ("Yak-pee" for short?) Let's chop that aromatic mouthful into tasty, digestible chunks:

Preferred equity is a form of investment that is like debt, in that the issuing firm promises to pay an agreed dividend level (like an interest rate), rather than a share of a firm's variable profits. However, preferred equity is unlike debt, and like stock, because if a firm for whatever reason does not pay the promised dividend, aggrieved investors cannot sue for bankruptcy. Preferred shareholders' only means of enforcing payment is priority: common equityholders cannot receive dividends if preferred shareholders' dividends have not been paid.

Cumulative means that if the issuing firm has skipped some dividend payments, the firm is said to be "in arrears", and must pay preferred investors all past skipped dividends before it can make any payout to common shareholders. "Cumulativity" ensures that, unless a firm goes bust before ever paying another dividend, preferred investors will eventually get all the payments they were promised, although they may suffer from delay. Cumulative preferred equity is more "debt-like" than noncumulative preferred equity, in that noncumulative investors permanently lose claim to some dividends if a company falls on hard times and suspends payments, while debtors always have claim to interest owed.

In-arrears convertible means that while payments on the preferred shares are "in arrears", when the firm had failed to pay some of the dividends that it had promised and not yet cured the failure, investors would have the option of converting the shares to common stock on favorable terms.

It's the in-arrears contingency that makes this security novel and interesting (I hope). But the feature requires some explanation. Usually, a convertible security has a par value and a conversion price that fix the number of shares of common stock an investor would get for converting a share of the security. This means that investors normally convert only when a firm is doing well. Suppose you have a share of preferred stock that (without the conversion feature) would be worth $100, but that can be converted to 10 shares of common stock. You would never exercise that right when the common stock price is less than $10, since the preferred share is more valuable than the stock you'd get. You would only convert when the common stock is doing well enough so that the value of the stock you would get on conversion exceeds the value of your preferred share. [2]

An "in-arrears convertible" would be pretty useless unless the conversion price were very low, since firms stop paying their preferred stock dividends in difficult times, when their stock price is depressed. So rather than fixing the conversion price in advance, these securities would be convertible at a discount to the market price of the stock at the time the preferred dividend was not paid. [3] That is, by going into arrears on the preferred shares, firms would open themselves up to dilutative preferred-to-common-equity conversions, at the option of the preferred shareholders. If a firm does have long-term, going-concern value, but is simply unable to meet the cash flow requirements of its capital structure, preferred shareholders could convert at a bargain rate during the limited in-arrears period. If a firm is not likely to be viable as a going concern, preferred shareholders could choose to hold tight. They'd be paid out in preference to common stock holders at the eventual liqudation.

Firms could issue multiple classes of "IACCPE", like they now offer multiple debt issues, each with a distinct priority in the capital structure of the firm. Ordinarily, these securities would be indistinguishable from debt, both to the firm and to investors. Investors would fork over a set amount of cash, and then expect to be repaid with interest (formally dividends) on a predetermined schedule. But in bad times, firms that fail to meet their obligations would be forced to offer equity, including control rights, to creditors on very favorable terms. (Non-payment of a dividend could also provoke a special shareholders meeting, and holders of the unpaid preferred could be given the right to propose replacement directors, thereby maximizing the value of converters' control rights.)

Substituting this kind of security for debt in firms' capital structure would enable a kind of bankruptcy in increments, an automatic and self-enforcing reorganization. I think this would improve value for all stakeholders compared to our present system. Chapter 11 bankruptcy was itself a great innovation, but it exposes even viable firms to large, indirect distress costs when capital structure and cash flows become misaligned. To the degree that a firm has widespread or important stakeholders outside its capital structure (customers, employees, financial counterparties, local governments, etc), Chapter 11 even at its best produces costly externalities, as stakeholders must provision for abrupt and unpredictable changes even when a firm is likely to survive and even thrive once arguments over who gets what are resolved. Because Chapter 11 bankruptcies (and receiverships for financial firms) are disruptive, governments sometimes intervene to prevent them or to the process with subsidies. The expectation of intervention causes investors in "systemically important" firms to over-lend and under-monitor. For large firms, the threat that contractually prescribed, preferred-to-common conversions might be triggered would be more credible than the threat of an uncontrolled bankruptcy without government subsidy. Investors would be forced to actually price the "lower tail", rather than hoping it will be truncated by the state. Common stockholders would face a steep penalty for missing "debt" payments, but the extent of their dilution would be predictably related to the scale of the obligations they fail to meet.

IACCPEs wouldn't replace or eliminate traditional bankruptcy, of course. Regardless of capital structure, firms that are not viable businesses will need to be liquidated. Sometimes firms have contractual arrangements other than straight debt that need to be modified if a firm is to become viable. Moreover, even if all "financial" debt were eliminated from firm capital structures (I think that would be a good thing), firms would still have transactional business creditors, for whom traditional "hard" debt makes sense. [4] This proposal does not directly address off-balance-sheet contingent liabilities or pension and health obligations, which are increasingly sources of firm distress. I think pension and health issues will have to be addressed on a national basis, that our employer-centric system of managing health and retirement issues will ultimately have to be, um, retired. But some contingent liabilities (uncollateralized derivative exposures) could and probably should be replaced by contracts that can be paid off in some form of equity (at punitive valuations) when they cannot be paid in cash.

Highly leveraged capital structures make individual firms, and networks of interdependent firms and communities, brittle. Replacing debt in firm capital structures with some form of preferred equity would serve as a shock absorber, allowing viable firms to survive transient cash flow shocks without affecting outside parties. It might be enough to simply level the playing field between debt and preferred equity by making preferred stock dividends tax-deductible for firms. But debt investors take some comfort in the fact that they have power, via the bankruptcy process, to enforce payment. That threat may reduce the cost to firms of debt finance. The sense of the present proposal is to define an instrument that gives fixed-income investors as much of the power they would have in a bankruptcy as is possible while reducing the likelihood of a "singularity" that creates far-reaching costs and uncertainties.

[1] The proposed "IACCPEs" would not necessarily be a more expensive form of financing than traditional debt: On the one hand, they take a weapon away from creditors, so creditors would want to be compensated for the additional vulnerability. On the other hand, by reducing the likelihood that a transient shock provokes an unnecessary bankruptcy, replacing debt with IACCPEs might reduce expected distress costs, and thereby increase overall firm value relative to a firm financed with traditional debt, which would be reflected in a lower cost of financing across the capital structure. Which of the two offsetting factors dominates would have to be an empirical question.

[2] Usually you would wait quite a bit longer than that, because the option of converting at any time makes the convertible security as valuable as the shares, but the agreed-in-advance payments of the unconverted security provides protection should the stock price tank. Exactly when it's worthwhile to exercise the conversion option on convertible shares is complicated, and in real life depends on the level of dividends paid by common shares and the relative liquidity of the market for common and preferred shares. In frictionless markets for a firm that issues no common dividend, it would only be worthwhile to convert an instant prior to maturity of the convertible security (if it is not perpetual). For our purposes, however, all that matters is that investors usually convert preferred stock only when the common shares are doing well.

[3] Getting the conversion option trigger right would be an important technical issue in defining these securities. Managers might try to manipulate their stock price around the time of the triggering nonpayment, in order to minimize the cost of dilution to existing shareholders (and themselves). The conversion price might be based on an average stock price over 30 days prior. Managers would not have very much scope to time the nonpayments, because they would be required to skip dividends on the most junior class of preferred shares, whose dividend schedule would have been set in advance.

[4] Broadly, "financial" investors should be expected to research and take some responsibility for the firms in which they invest, while customers and suppliers should be able to do business with a firm without worrying very much about its balance sheet. There is no bright line between transactional credit and financial debt, but it is nevertheless a distinction worth making, and even policing, in firms' capital structure. The "cash efficiency" movement, which encourages firms to maximize their use of transactional credit as a source of cheap financing, is in my view pernicious. But that's a rant for another day.

You just never know who you'll run into here in blogland.

Joe Peek is a professor of finance at the University of Kentucky (where I am a very poor excuse for a graduate student). He has a guest post over at The Hearing.

Dr. Peek's professional obsession is the Japanese banking system. He takes an unreconstructed view of the parallels between our response to the banking crisis and Japan's, and is particularly unhappy with the recent softening of mark-to-market rules. Here's a snippet:

Much like in Japan, U.S. policy makers have made efforts to avoid distinguishing among banks, for example, forcing all of the largest banks to accept billions of dollars of Troubled Assets Relief Program funds. The stress tests for the 19 largest banks provide policy makers with an opportunity for a "do over." The results of the stress tests must be based on market values and whether the banks are truly economically viable. Government capital should not be injected into banks indiscriminately; only the strong should survive. We need disclosure, as well as closure, if a bank either is not viable or cannot raise sufficient private-sector capital to become viable.

The time has come for transparency to replace the "parency" of government support of non-viable firms, financial or non-financial. The "convoy system" did not work in Japan during their "Lost Decade," and should not be expected to work here.

Do read the whole thing.

Would it have been better if Timothy Geithner had had the power to guarantee all bank debt early on? As James Surowiecki reminds us, that was part of the Swedish solution. Justin Fox plausibly suggests that we might have avoided a lot of pain with a fast, full guarantee.

But that's not the point. The question isn't whether we could have avoided this crisis, if only we had cut a big check. We could have, and that was not lost to any of us debating these issues more than a year ago. (See e.g. me or Mark Thoma.) Had we done so, the near-to-medium term fiscal costs might have been less than they probably will be now. So, with 20/20 hindight, would it have been a good idea?

How you answer that question depends upon how you view the crisis. Is it an aberration, a shock to a basically sound financial system, or is it a painful symptom of an even more dangerous condition? Under what circumstances would our political system be likely to impose reforms that would prevent large scale misallocations of capital and shifting of losses to taxpayers in the future?

If you think that our financial system just needs some tweaks, some consolidation of regulators' organizational charts and sterner supervision, then you should prefer that we had just cut a check, passed Sarbanes/Oxley Book II, and moved on. But that is not what I, or most proponents of nationalization temporary receivership for insolvent banks, believe.

If you believe, as I do, that we need a root-and-branch reorganization of the financial system, which must necessarily involve the dismemberment and intrusive restraint of deeply entrenched institutions, does that mean pain is the only way forward, "the worse the better" in the old revolutionary cliché? It need not mean that. But it does mean that palliative measures, like giving the banks money, would have to be attached to curative measures, like enacting capital requirements and imposing regulatory burdens that would force financial behemoths to break themselves up or become boring narrow banks. For almost two years, policymakers at the Fed and the Treasury, including Secretary Geithner, have offered bail-out after bail-out and asked for nothing serious in return.

Do I regret that Henry Paulson was not empowered to issue a blanket guarantee of bank assets early on, as the Swedes did? No, I don't regret that at all. Why not? Because I think that "Hank the Tank" was a crappy negotiator, not only for taxpayers in a fiscal sense, but also for the economy and the polity more deeply. He would have offered the financial system sugar without requiring it to make the medicine go down. He may believe, quite sincerely, that a cure would be worse than the disease. He may believe that, but he is wrong. If we "get past this crisis" by restarting a consumer-credit-based, indiscriminate-investor-financed, current-account-deficit-making, income-inequality-expanding economy, we will have increased, not diminished, the likelihood of a major collapse.

You may believe that we have learned our lesson, that if we can just get some stability and comfort for a while we are prepared to do what must be done. That's a respectable position. But I don't share it, and neither do the majority of Americans who are unwilling to allow their representatives to sign off on any more expensive aspirin. We want value for value, an ironclad commitment of root and branch reform in exchange for the unimaginable sums of money we are being asked to hand over.

Surowiecki has in the past suggested that those of us who favor nationalization would criticize any alternative simply because it is not precisely the policy we advocate. But it is not we who have refused to compromise. We've seen variations on the same basic proposal over and over again. Geithner's PPIP really does resemble Paulson's TARP, besides the part about actually asking taxpayers for the money. Each latest plan from our incestuous cadre of economic Mandarins demands only symbolic concessions from the dysfunctional organizations we are asked to support. The "moderate" political class goes on and on about how Geithner and Bernanke have to go all Enron, funding the banks via off-balance-sheet guarantees and special purpose vehicles, because "populist, childish" Congress won't put up the money. Setting aside how audaciously corrosive that sentiment is to Constitutional democracy, it is simply wrong. Congress would, because the public would, support large, explicit transfers, if they were attached to reforms sufficiently radical to prevent a recurrence, and suitably punitive towards the people who managed the system that brought us here. Value for value.

I am a true believer in American-style capitalism. So I would like to see people who earned profits lending to banks in good times bear the high costs of failing to monitor the organizations they funded. Investor fear is what is supposed to prevent the indiscriminate misuse of capital. To the degree that creditors have leaned upon "implicit" government guarantees, I think it would both be just and set a useful precedent if they were reminded that investors have to take responsibility for where they place the precious capital they steward.

That said, like Paul Krugman, I would be willing to hold my nose and tolerate a Swedish-style guarantee of bank creditors. I'd acquiesce to that even without formal nationalization. Nationalization is no one's idée fixée. It is a means to an end, and the desired end is a world in which too big to fail is too big to exist for any financial institution that originates or holds credit risk in any form. Secretary Geithner could send a bill to Congress today that would put all banks with a balance sheet of over $50B into run-off mode, while clearing away legal obstacles so that larger organizations could arrange their own breakups over time. I'd fax my Congressman and support a $2T on-budget buyout of bank creditors as part of that bill, as long as it had teeth. ("Teeth" would include making sure that off-balance-sheet and derivative exposures were included in the size cap, etc.)

It's not that us pitchfork-totin' populists are unwilling to pay the bill. It's that we want to know that in exchange for writing a very, very large check, the people that we are paying will actually deliver the goods. Given the behavior of bankers before the crisis and of shifty policymakers during, we have every reason to watch warily and to insist upon every precaution while we hand over suitcase after suitcase of freshly printed Federal Reserve notes.

I've been unimpressed with this oft-quoted bit from Phillip Swagel's insider account of the Paulson Treasury.

Legal constraints were omnipresent throughout the crisis, since Treasury and other government agencies such as the Federal Reserve must operate within existing legal authorities. Some steps that are attractive in principle turn out to be impractical in reality—with two key examples being the notion of forcing debt-for-equity swaps to address debt overhangs and forcing banks to accept government capital. These both run hard afoul of the constraint that there is no legal mechanism to make them happen. A lesson for academics is that any time the word "force" is used as a verb ("the policy should be to force banks to do X or Y"), the next sentence should set forth the section of the U.S. legal code that allows such a course of action—otherwise, the policy suggestion is of theoretical but not practical interest. Legal constraints bound in other ways as well, including with respect to modifications of loans.

Today's news (Clusterstock + source docs, WSJ Deal Journal, McArdle, Naked Capitalism, Calculated Risk, Marketwatch), that Henry Paulson, um, forced Bank of America's near suicidal merger with Merill Lynch kind of clinches the case. Pre-Merrill, BOA was viewed as relatively healthy among large banks. What's the statute under which a Treasury secretary unilaterally fires and replaces the board of a healthy bank? The Paulson Treasury talked up legal constraints whenever they were faced with something Paulson didn't want to do. When Paulson, or Bernanke, really did want to do something, they were very creative about bending the law to their will. The Fed's "special purpose vehicles" are clearly not lending in the sense that the architects of the Federal Reserve Acts "unusual and exigent circumstances" clause foresaw. The FDIC has no statutory authority to issue ad hoc guarantees of bank debt, but flexibility was read into the laws.

With respect to the banks, the Paulson Treasury could have forced any big bank into a bail-out or receivership scenario just by looking at it funny, or by having the Fed take a conservative view of bank asset collateral values under the special liquidity programs. It's worth noting that Treasury very ostentatiously forced banks to accept TARP capital, and Geithner's Treasury was able to persuade holders of Citi preferred to convert to common equity.

It's not exactly right to say that our don't-ask-don't-tell quasinationalization policy has given us "ownership but not control". An assertive Treasury secretary has tremendous leverage over zombie bank managers. Instead, what we have is is control without accountability. An informal, unauditable, hydra-headed set of private managers and public officials controls how quasinationalized banks behave. Neither taxpayers nor shareholders have reason to believe that decisions are being taken in their interest. The informality and disunity of control impedes the kind of hands-on, detail-oriented supervision and risk management that ought to be the core preoccupation of bank managers. Exactly as opponents of nationalization feared, America's large banks are poorly run behemoths that routinely make idiotic commercial decisions to satisfy tacit political mandates. No one really knows who is responsible for what.

Ironically, there might be less scope for political control if banks were in formal, least-cost-resolution receivership. A bank that has already failed cannot fail. If independent boards are appointed to oversee the receiverships, politicians might have very little leverage. Incumbent private managers face collapse, sacking, disgrace, and potential civil and criminal liability for improprieties that come to light during the post-mortem. New moderately paid, high reputation board members would bear no responsibility for what came before, and could very publicly resign in protest if pushed to act in a manner inconsistent with their charter. (Resignation in protest by long-affiliated board members of a zombie bank would have different reputational consequences, and it would be difficult to recruit high-reputation outsiders to serve on zombie bank boards.) Promoting insiders or recalling retired executives to run zombie firms leaves the leadership weak and compromised. A much higher caliber of outside talent could be recruited to oversee banks in receivership than would accept responsibility for banks that are insolvent but on government life support.

This is not to say that formal public control would be a panacea. The list of public and quasipublic organizations currently being gutted by politically motivated credit expansion includes Fannie Mae, Freddie Mac, FHA, FHLB, FDIC, and the Federal Reserve system. A bank in receivership managed by a weak board or not institutionally segregated from political bodies could easily join the list. But if received banks were put under strong boards, and given clear mandates to divide and sell their assets (maximizing taxpayer value subject to a scale constraint) while running off their lending books, there would be little hazard of politically directed credit or other shenanigans. That would imply that large insolvent banks would reduce their lending, contradicting the Administration's endless exhortations that banks should lend, lend, lend. My view is that public encouragement of expanding indebtedness is very bad policy (read Finem Respice). But if you misguidedly believe that "credit is the lifeblood of a modern economy", the thousands of well-run smaller banks in America are fully capable of taking advantage of today's deeply subsidized lending spreads to serve creditworthy borrowers. Whether in private or in public hands, the big, broken banks are simply too compromised to lend.

Commenting on Nassim Taleb’s provocative agenda for fixing the world, Felix Salmon notes that

Looking at the rest of the list, how on earth do you stop the financial sector from... creating complex products? Derivatives are, at heart, bilateral contracts: how can you ban two consenting adults from entering in to such a contract?

The only bilateral contract is a gentleman's agreement. Binding contracts involve an implicit third party, the state which (through its courts system) stands ready to enforce the terms of private arrangements. The state is not, and cannot be totally neutral in its role as contract enforcer: Communication between contracting parties is always imperfect; the universe presents an infinite array of unforseeable possibilities; even very clear contractual terms can be illegal, repugnant, or contrary to the public interest. The state makes affirmative decisions about how it will (or will not) enforce the terms of contracts. Libertarians may have perfect freedom to contract, if their agreements are self-enforcing or voluntarily adhered to. For the rest of us, every contract is a negotiation between three parties, the two who put a signature at the bottom the document, and the state which will be called upon to give force to the arrangement when disputes arise or someone fails to perform. I think of contract lawyers as two-bit psychics in fancy suits. Much of their job is to channel the voice of the incorporeal Leviathan, so that its quirks and predilections are taken into account whenever agreements are drafted.

This has something to do with derivatives, but even more to do with one of Taleb's broader concerns: debt. At present, the state enforces debt contracts by permitting lenders to force nonperforming borrowers into bankruptcy. That is not a natural or obvious arrangement. Bankruptcy evolved as an improvement over automatic liquidations or men with big necks and brass knuckles. It serves to balance the contractual right of a lender to be timely paid with a broader interest in preserving the overall value of enterprises and preventing extremes of immiseration. To some degree, bankruptcy lets debtors to escape the terms of their own agreements and limits the right of contract (though bankruptcy is onerous enough that debtors don't seek this sanctuary easily). The fact that creditors' rights are limited is socially useful: it encourages lenders to discriminate between good borrowers and bad, reducing the frequency with which resources are lent foolishly and then destroyed.

One thing I think that we are learning from the present crisis is that the logic of bankruptcy hasn't been taken far enough. Creditors' rights are too strong. Creditors have insufficient incentive to discriminate, especially when lending to "critical" organizations, because the bankruptcy that would attend a failure to pay is too disruptive and destructive to be permitted by the state. We have seen tremendous resources lent to banks thoughtlessly, and then squandered or stolen rather than carefully invested. Similarly, those who entered into derivative contracts often ignored credit risk when a counterparty was seen as too dangerous to bankrupt. If it were possible for borrowers and counterparties to welsh on their agreements without provoking consequences as disruptive as bankruptcy, creditors would have more reason to be careful of whom they do business with, and potential deadbeats (like large financial firms) might not be able to take levered risks cheaply.

I think that, going forward, the state will have to limit the right of debtors to enforce claims by bankruptcy. Creditors and counterparties who go to the courts would run the risk of having their claims converted into something like cumulative (and maybe convertible) preferred equity. This would ensure that no dividends are paid to stockholders until the disgruntled creditors are made whole, but would not otherwise disrupt the operation of firms or affect other claimants. (Such conversions could be combined with tight compensation limits, to prevent shareholders and managers from taking payouts as wages and bonuses while failing to pay creditors forcibly converted to equity.) Judges would weigh the rights of creditors against the costs to other stakeholders in deciding between formal bankruptcy and ad hoc conversions, so that the risk to creditors would increase with the size and interconnectedness of borrowers.

It may be hopeless to try to control what kind of contracts private parties write amongst themselves. But we can control how contracts are enforced. There is nothing natural or neutral about how we currently enforce debt contracts. We made up some procedures that seemed to work reasonably well. The current crisis has exposed some shortcomings. Nothing prevents us from modifying how we enforce contracts in order to improve the incentives of parties to manage their own risk, and to prevent collateral damage when private contracts come undone.

Divide the world into the consolidated financial sector and taxpayers. Under PPIP, each dollar a "public-private investment fund" overbids provokes a net transfer to the consolidated financial sector from taxpayers. The size of transfer to the financial sector increases with the degree to which bids are overpriced, and is maximized if the true asset value is very small relative to the price actually bid. If an asset is worth $6 dollars, and financial sector actors purchase a contract for $7 while the Treasury invests alongside, the consolidated finacial sector gains a dollar. But if financial sector actors pay $70 for the same asset, the financial sector would receive a net transfer from taxpayers of up to $118. (For more detailed arithmetic, see below.) The more financial sector actors are willing to overbid, the greater the net transfer from taxpayers to the financial sector. In theory the scale of the transfers is limited only by the quantity of asset purchases the government is willing to guarantee.

There are problems with this story. In real life there is not a consolidated financial sector, but a lot of different players who are usually in the business of competing with one another. PPIP includes rules and tools by which the government could prevent the use of taxpayer money to fund overpriced bids, and ensure that the parties who take small losses are distinct from the parties who make large gains, eliminating incentives to overbid. An important question is whether the government genuinely wishes to prevent backdoor transfers to the financial sector, or views such transfers as a desirable means of helping core financial institutions. (See Joe Weisenthal and Noam Scheiber)

It is worth noting that overcoming coordination problems so that diverse parties can collaborate on profitable ventures is precisely what the financial sector is supposed to be good at doing. Ideally, we would like the profitable ventures to be welfare-improving projects in the real economy, but there is little question that financial sector actors will gladly apply the same skillset to extracting transfers and rents when the opportunity presents itself. Attempts to regulate away intentional overbidding by cooperating parties will have to outwit some very clever professional deal makers.

A few more caveats — financial sector actors do pay taxes, so they are not distinct from the taxpayers from whom transfers are made. Qualitatively, this overlap would't change the story very much. (Quantitatively, it's interesting, you'd have to think hard about the realizability of "deferred tax assets" from losses the financial sector would absorb without the transfers.) The numbers I'm using are for the PPIP whole loan program. The degree of nonrecourse leverage that will be provided by the Fed towards the securities purchase program is as far as I know unspecified.

Some links:

There are way too many good links on this issue, but rortybomb's take is my all-time fave.

Restricting to the last 24 hours or so, see also Scurvon, Carney, Sachs, Nemo, Krugman, Free Exchange, Felix Salmon, Economics of Contempt, and Cowen. See also articles in the Financial Times (ht Conor Clarke, Calculated Risk) and the New York Post (ht Yves Smith). As far as I know, Karl Denninger is the first person to have pointed out the potential for gaming. My first take on this issue is here. Mish has a very similar take (here, here, and here). I'm sure I've left many great posts out of this linksplatter. My apologies to the unsung pundits.

Arithmetic:

Case 1: A private fund buys an asset for $7, but pays only $1, as the rest is borrowed from the bank via an FDIC guaranteed loan. The Treasury invests along side 1:1, purchasing the same asset on the same terms. The asset is really worth $6. The private fund loses its dollar, but this becomes a gain to the selling bank, causing neither a loss or gain to the consolidated financial sector. However, the Treasury also loses a dollar, which becomes a gain to the selling bank, and amounts to a transfer of $1 from taxpayers to the consolidated financial sector.

Case 2: A private fund buys the same asset for $70, of which it pays $10 cash and borrows the rest on an FDIC guaranteed nonrecourse loan. The Treasury invests along side 1:1, purchasing the same asset on the same terms. The asset is really worth $6, so each purchase amounts to a transfer of value to the bank of $64 dollars. When the value becomes known, the FDIC (indirectly) accepts the $6 asset in exchange for $60 of loan extinguishment, bearing $54 of the private investor's $64 loss. The net effect of the private investment is a transfer of $54 from taxpayers to the consolidated financial sector. Taxpayers bear the full loss of $64 on the Treasury's investment. So the total transfer fronm the public to the private sector is $118.

Not unusually, I was a bit incoherent in my previous post on bank size. On the one hand, I wrote...

...a sufficiently levered and inter-contracted microbank could take down the world as surely as the Citimonster.

On the other hand...

...limiting size defined by total asset base plus an expansive notional value of all derivative and off-balance sheet exposures limits both interconnectedness and leverage.

If limiting size constrains leverage and interconnectedness, how could a microbank get to be so interconnected and level as to bring on armageddon?

The key, of course, is in the definition of size. Entities that are small in terms of number of people involved and level of capitalization certainly can blow things up, by loading up on traditional leverage (debt) and untraditional leverage (derivative exposures, off-balance-sheet contingent liabilities). AIG might have been a big firm, but the unit that blew up the world amounted to a handful of people in a well-appointed London garage. The "bigness" of AIG mattered only insofar as it permitted that tiny operation to lever up, by taking on trillions of dollars in notional CDS exposure.

But if size is defined properly to mean the total scale of assets to which a firm is exposed, including balance sheet assets plus the notional value of any derivatives plus any off-balance-sheet commitments, then size is basically everything. Suppose there was a bank levered 10000:1. Sounds pretty bad, huh? But suppose the bank has precisely one penny of equity against a hundred dollars of assets. That might suck for the fool who lent the hundred bucks, but the rest of us can sleep easily. Leverage alone can't cause crises: it's only when an entity is levered up to some systemically troubling size that bad things happen.

That doesn't mean we could regulate bank size and then ignore leverage: If all banks had $100 on assets against a penny of equity, we'd end up with a lot of bank failures, creditor bailouts, and sleepless nights. What size limitations do is prevent mistakes or misdeeds at any one or few firms from becoming all of our problem. Smallness also reduces the likelihood of misdeeds, since dumb gambles have a bigger payoff for managers at big banks than at modest thrifts. Some banks will always slip through regulatory cracks. If they are small and few, that's not a problem. If they are big or many, we're screwed.

Size limits, like leverage and risk constraints, will inevitably be gamed. There will always be fuzziness surrounding what sort of contingent liabilities should fall under a size calculation, and bank lobbyists will work assiduously to create loopholes. Adding hard limits on size and trying to police them won't be a panacea. But faking small might be harder than faking well-capitalized. Plus, forcing banks to appear small may help to keep banks actually small, since counterparties get nervous about offering sneaky, uncollateralized leverage to banks that look like they are small enough to fail.

In the end, banks-as-we-know-them are flawed by design. They serve an important purpose, but do so in a manner that is predictably prone to failure. If your roof has a leak and water drips in, one way to handle the problem is with a bucket. That approach can be effective, but it demands constant supervision. There is always the danger of some lapse of attention, then whoops!, the bucket overflows and your floor is ruined. Resolving to watch the bucket very carefully by, say, titling yourself the integrated bucket super-regulator might help, a bit. But even super-regulators need the occasional bathroom break, and buckets are notorious for tempting super-regulators with songs of free water flows and offering cushy jobs if they look the other way. Imposing size and leverage constraints on banks is like replacing a small bucket with a big washtub: You'll still have problems if you don't watch the thing, but the occasional lapses in supervision are less likely to conjure the Great Flood. Of course, the best way to manage a leak in the living room would be to stop messing around with buckets and patch the roof instead. To do that, in my opinion, we'd have to separate the credit and investments function of banks from the payment and deposits function, and draw an enforceable bright line between guaranteed claims and risk investment.

But if we're too scared to climb that ladder, I suggest we get the biggest washtub, — I mean, the tightest size restrictions — that we can possibly manage. Promising to stare very sternly indeed at the same old bucket just won't cut it.

p.s. if you haven't seen it, I really like James Kwak's Frog and Toad post on the difference between supervisory and structural approaches to regulation.

Kevin Drum has nicely posed the question of whether it really is important to break up big banks. After all, he argues, even small-ish banks have proven to be too leveraged and interconnected to be permitted to really fail. He argues that maybe it's the banking industry, rather than individual banks, whose size and reach we need to constrain.

John Hempton has been arguing for the Australo-Canadian model of an oligarchic, heavily regulated, generously profitable banking system.

James Kwak offers a very nice discussion of the "too big to fail" problem in light of the absence of structural rather than supervisory approaches in Treasury Secretary Geithner's recent regulatory proposals. (And Drum responds.)

I think size does matter very much, but not because small banks are inherently small enough to fail. Drum is right about that: Like a dwarf with a suitcase nuke, a sufficiently levered and inter-contracted microbank could take down the world as surely as the Citimonster.

But in practice, a properly defined smallness could add a lot of safety to the banking system:

1. Very directly, limiting size defined by total asset base plus an expansive notional value of all derivative and off-balance sheet exposures limits both interconnectedness and leverage. (Defining size limits by capitalization would suffer from the same drawback as traditional leverage constraints — they encourage bankers to scheme secret ways of levering up.)

2. When a bank appears to be small enough to fail, creditor discipline will backstop regulatory supervision. If a bank is perceived as too big to fail, if its failure in "unthinkable", then clients and counterparties will be lax in managing or limiting their exposures, leaving always circumventable regulation as the only bulwark against becoming too levered and interconnected to fail. (Insured depositors, of course, won't provide discipline, and shouldn't be expected to. But bondholders and derivative counterparties will, if a bank's credit is potentially dodgy.)

3. Smaller banks, even very levered and interconnected ones, can be unwound, merged, or put into receivership. We've managed the failures of even large-ish banks like Drexel, Bear, Wachovia, WaMu, or IndyMac, and we could have managed Lehman in a costly but orderly unwind. But once banks have gone truly mega, we're not sure we can manage it. A bank that is too big too merge without overconsolidating the industry presents special problems. From a taxpayer perspective, we are generally able to unwind smaller banks without guaranteeing non-insured creditors, while we find haircutting the creditors of larger banks impossible, because these unsecured creditors regard failure as unthinkable and fail to adequately provision for the risk.

4. Political economy considerations mitigate against large banks (arguably more deeply in the United States than in Australia and Canada). Particularly if financial firms are segregated by scope (e.g. investment banking distinct from commercial banking distinct from brokerage distinct from insurance), groups of small firms with distinct industry agendas are likely to be less corrupting than huge, critical institutions with a unified management that acts strategically in political circles.

5. Scale breeds agency problems. Earning an extra five basis points on $100B in assets amounts to $50M in extra income a year, a fraction of which can make a manager very wealthy in an eat-what-you-kill bank. Making that same five basis points on a $100M portfolio earns a small bank 50K, a fraction of which amounts to a nice bonus, but not a lifestyle change. For both managers, the downside if something goes wrong is the same: they lose their jobs. The ability to leverage a large balance sheet tempts managers at larger banks to take risks that managers at smaller banks would not find at all worthwhile. (Drum points out that managers of small hedge funds earn huge sums too, but that's really apples to oranges. Hedge fund investors, like stock investors, are generally aware of and prepared to manage investment risk, while bank creditors expect that their money is safe. Hedge funds mostly present systemic problems when their use of leverage puts bank creditors at risk. That can and should be regulated, from the bank side and perhaps by eliminating the right of hedge funds to limited liability forms of organization.)

There are very few obvious reasons why large banks are useful at all, other than supervisory convenience if you think Hempton's regulated oligarchy is the right model. It may be annoying to have to pay other-bank ATM fees, but besides that, there are very few services or efficiencies a large bank can offer that a small bank cannot. Large banks can provide large loans more easily, which is convenient for corporate clients. But that may be a bad thing. Lending decisions can be mistakes. It's one thing if a lending committee misdirects $300K to a bad mortgage. It is much more costly if that same flawed body channels $3B to a crappy LBO. Raising large quantities of capital should require the separate assent of multiple independent parties. Misdirection of the resources represented by billions of dollars creates social as well as private costs.

My sense is that a lot of people think large banks are here to stay for precisely the reason they should be made extinct. Large banks feel modern, important, powerful. It seems nice, somehow, when you travel across oceans and find a branch of your own bank. It's like you are part of a winning team. There's that ubiquitous brand, and it's your brand. But "brand equity" is an important means by which banks build a mystique that makes their failure unthinkable, and charms bondholders and other uninsured counterparties into offering leverage on much too easy terms. Ironically, if banks felt a bit shabby and penny-ante, and if managed failures were regular events, the banking system as a whole would be much safer.

Size isn't everything: Bankers are famous lemmings, and a whole lot of small banks who pile into the same poor investment can fail together like one really big bank. But a thousand little banks are at least a bit less likely to make correlated mistakes than megabanks, which can turn a bad investment idea into a firm-wide mission. One goal of bank regulation, besides restricting size and leverage, should be to encourage independent lending decisions and supervising the diversity of the aggregate banking system's portfolio. Regulators should "lean against the wind" of booms that homogenize banks' asset base by restricting growth of overrepresented asset classes. If there is a good economic reason for a boom, nonbank equity investors can take advantage of the opportunity.

Note: Ideally I prefer a complete separation of the depository and payments function of banks from the lending and investment function. That is I'd prefer we create "narrow banks" that invest only in government securities, and define a new kind of explicitly at-risk investment fund to serve the traditional purposes of bank lending. But this piece is written under the pessimistic assumption that we'll leave the familiar structure of banking intact.

Is that it, then? You know, the "Public Private Investor Partnership" that the Treasury Secretary introduced on Monday. Are we doing that?

The plan involves the Treasury, FDIC, and Federal Reserve putting hundreds of billions, perhaps more than a trillion dollars, at risk. That should require some sort of Congressional approval, right?

I remember the whole TARP debate last fall. I thought that was a terrible plan. I faxed my senators and representative several times, and urged them not to pass it. I was gratified, and for a brief moment optimistic, when the bill was initially rejected in the House. I felt like it was a miscarriage of democracy that Congressional leaders staged a do-over on that vote, reintroducing substantially the same plan and passing it just a few days later. That battle was lost, but this is a democracy and I am an engaged citizen. There would be other battles, I thought.

In my view, the Geithner's PPIP includes two mechanisms intended to ensure that "private investors" offer substantially inflated bids for "legacy" assets, and the net cost of the plan will be comparable to that of TARP. I might be wrong about that, but I might be right. Much of the risk will be due to loan guarantees offered by the FDIC. Is there any legal basis for using the FDIC this way? Aren't the laws describing how the FDIC is and is not supposed to behave?

And isn't Congress supposed to have the power of the purse? A loan guarantee is a contingent liability, a cost in real terms. Can the US Treasury spend money without Congressional approval, as long as it promises to spend only if a coin flip comes up heads? That's exactly what the Geithner plan (along with the scandalous but already active "Temporary Liquidity Guarantee Program" program) does. Is that even Constitutional?

FDIC is a full-faith-and-credit agency of the Federal government. There's been a lot of commentary trying to explain the recently high CDS spreads on US sovereign debt. After all, wouldn't the government just print money to pay its debt rather tha default? Well, here's a scenario: Suppose the FDIC's loan guarantees come badly acropper, putting taxpayers on the hook for hundreds of billions of dollars. Suppose FDIC is short the cash, and has to come to Congress for an allocation. Given that neither Congress nor the public ever signed on to all these guarantees of bank assets, and that in fact FDIC is behaving in a manner precisely contrary to the laws under which it is chartered, the level of anger might be high enough that the public might just say no. Welcome to the world of full-faith-and-credit default.

Maybe that's why Chris Dodd wants Congress to give the FDIC a $500B loan commitment. Maybe it explains the apparently limitless appropriation of "such sums as are necessary" to the FDIC that Justin Fox noticed in a proposed bill that would actually authorize these sorts of liability guarantees. (The bill would also authorize FDIC receiverships of systemically important non-banks — yay! But it leaves out the "least cost resolution" stuff from the traditional FDICIA, and would give the Treasury Secretary and the FDIC complete discretion over whether firms are to be taken over or just bailed out in any of a number of ways.)

It seems to me that committing hundreds of billions of taxpayer dollars should still be considered a serious business. It seems to me that if Congress wouldn't approve the Geithner plan, in a democracy, that ought to have some meaning, and not just get written off as populist outrage and then extralegally ignored.

So I'll ask again, who passed the Geithner plan? What deliberative assembly gave the plan a pass? What's that you say? The stock market went up by nearly 500 points when it was announced on Monday? Oh. I guess the buys have it, then.

James Surowiecki has been pushing the idea (first mooted by John Hempton) that since bank financing always involves non-recourse by taxpayer (via government deposit insurance), it's no big deal that the Geithner Plan is built around generous non-recourse lending. Surowiecki:

There is one detail of the plan, though, that people are particularly bothered by, and that is the fact that the plan involves the FDIC guaranteeing loans to private investors. (The way the plan to buy pools of mortgages is set up, investors will be able to borrow six dollars for every one dollar they invest. If their bets go bad, they lose only the one dollar they invested—the FDIC is responsible for paying back all the borrowed money.) Paul Krugman, for instance, calls this the “central issue,” and argues that because the non-recourse loans are a massive subsidy to investors—which they are—the plan will distort the prices that investors are willing to pay for these assets, and therefore "has nothing to do with letting markets work." Ezra Klein, similarly, argues that because the plan relies on these "non-recourse” loans, the prices it will produce will be in some way “artificial." Their point is that the Geithner plan, among other things, is supposed to produce real market prices for these toxic assets, which will then give us a better picture of banks’ balance sheets and allow us to avoid valuing these assets at prices that the government thinks have become unduly low because investors are so risk-averse. But by creating a plan in which investors have only a small downside and a big upside, we’re supposedly creating fake prices.

There’s no doubt that the non-recourse loans constitute a big subsidy: while investors’ downside risk isn’t eliminated, since they can still lose all the money they invest, that risk is significantly limited, while their potential upside is significantly increased (since they’re leveraging every dollar they invest six-to-one). Yet for all the criticism of this subsidy, the truth is that the plan’s reliance on non-recourse loans is not an especially radical idea. In fact, it’s essentially the same kind of subsidy that the entire U.S. banking system has depended on for the last seventy-five years. What are FDIC-insured bank deposits, after all? They’re non-recourse loans to banks. You deposit money with a bank—that is, you lend it your money. The bank can then take that money, and leverage it up nine-to-one to make loans or acquire assets. If the loans are good, they keep all the profits. If the loans go bad, the most the bank can lose is the capital it’s invested. All the rest of the bank’s losses are paid for by the FDIC. This is precisely the same arrangement — down to the loans being guaranteed by the FDIC—that the Geithner plan sets up. In effect, it just extends to outside investors, for the purpose of acquiring toxic assets, the same subsidy that banks have been receiving since 1933.

Surowiecki is right in a superficial way, but he misses crucial details. His analogy breaks down in ways that I think are informative.

Non-recourse financing involves bundling a valuable put option along with a loan. The value of that option hinges on the details of the arrangement: An "at-the-money" option is more valuable than an "out-of-the-money" option. The "moneyness" of the implicit put option in a non-recourse arrangement is determined by the degree of leverage the borrower is allowed. The Geithner plan permits a maximum leverage of 7:1 assets to equity. As Surowiecki points out, that degree of leverage is less than the leverage of a traditional bank.

But notional "moneyness" is not the only thing that determines the value of an option. In particular, options are most valuable when the assets that underlie them are very volatile. In order to limit the degree to which banks maximize the value of the deposit insurance option at the taxpayers' expense, banks are supposed to submit to onerous, intrusive regulation that limits the riskiness, the volatility of the assets they can purchase.

Under the Geithner plan, the government will extend its non-recourse option to investors without preventing them from "swinging for the fences" on risk in order to extract value from the option. On the contrary, the government is insisting its loans be used to purchase assets that have already proved themselves unsuitable for purchase by a regulated entity, by virtue of being volatile and difficult to price. It's as if an insurance company that ordinarily refuses to cover homes in hurricane states suddenly offered policies only to purchasers looking to build homes on Gulf-coast barrier islands.

Sometimes an option is costly to exercise. Exercise costs reduce the value of an option to its owner along with the expected liability of the option writer. The traditional deposit insurance option was very costly for banks to exercise. Banks were only allowed to exercise the option by being put out of business. That sharply limited the value of the FDIC option to bank employees and shareholders. Usually the "franchise value" of a bank is greater than its regulatory capital. In order to extract value from the option, bank stakeholders must take bets whose (probability-weighted) payoff in a good outcome exceeds not only the loss of regulatory capital, but also the value of the business as an ongoing concern. Moreover, if we are valuing the "traditional" FDIC option, we should go back just a few decades to when most banks were privately held and run by lifers. A typical bank's owners and employees had an illiquid, undiversified exposure to the well-being of their institution. Calculating franchise value from the price/book of Citi pre-crisis would dramatically underestimate the value of a traditional bank to its controlling stakeholders. Exercise of the FDIC option used to be costly indeed for the owners and managers of banks. There were, if you'll excuse the terms, "synergies" between regulation and a high cost of exercise: If triggering a deposit insurance payout is very painful, strategies designed to monetize the option need a fly-to-the-moon upside to be worth the risk. But spaceports are more likely to be flagged by regulators than an extra 100 bps on a "AAA" CPDO.

Unlike deposit insurance, the non-recourse option offered to investors in the Geithner plan is completely costless to exercise, once it is in the money. Surrender of the collateral constitutes fulfillment of the lending contract full stop. Pace Megan McArdle, it is not like the non-recourse option implicit in the typical home mortgage, where exercising the option involves a hit to ones credit. There is no default of any sort involved in surrendering the assets rather than repaying the loans. The right to do so will be written into the deals.

Finally, the FDIC option didn't used to be free. Banks paid a fee in exchange for the deposit insurance. Under the Geithner plan, borrowers will be charged a fee as well, but then they'll be borrowing at rates dramatically lower than they could on their own. This works with Surowiecki's story — banks borrow very cheaply from depositors because of the FDIC guarantee. But, as always, the question is price: Given the volatility and uncertainty surrounding the underlying assets and the near-zero cost of exercise, will the FDIC charge a fee high enough to cover the expected liability of the option? I'll leave that for readers to decide.

Surowiecki's analogy does work pretty well if we compare the Geithner plan not to traditional banking thirty years ago, but to recent practices of the industry. Thanks to hands-off government and structured-finance shell games, banks were recently able to amass high risk, high yield investment portfolios despite being ostensibly regulated institutions. Institutional changes, including changing norms about the length and terms of bank employment and dominance of the industry by large, heavily-traded limited-liability corporations, decreased the expected cost of exercise to bank employees and informed shareholders, making volatility maximizing strategies more attractive. The option premium, the FDIC fee, was severely underpriced (it was reduced to zero from 1996 to 2006).

If Surowiecki wants to argue that the non-recourse option embedded in the Geithner plan would basically reproduce the subsidy to the banking system offered circa 2006, I'll readily agree. But it is not reasonable to argue that non-recourse loans offered on generous terms to unregulated investors for the express purpose of purchasing unusually volatile assets represent the "same subsidy that banks have been receiving since 1933."

The Compulsive Theorist has written a truly excellent post on bank bailouts (ht Mark Thoma). I'll excerpt a bit below, but do read the whole thing:

I sympathize with the point of view which says that the political window of opportunity is narrow and the need for action urgent, so let’s accept the bailout plan for now, and deal with... wider issues later on. But the very fact that political momentum is limited means that if these wider changes are to be brought about, the process has to begin in earnest at once. Does anyone seriously believe that in a years time, if following massive government support the banks are stable — or can be made to appear stable — there will be any political will to break up very large institutions, or any real change to underlying norms in the financial sector?

However, absent these deeper changes, it is entirely possible that we will see a replay of the crisis — but on a larger scale — in a few years time. Naturally, one cannot say with certainty that such a cataclysm (and if it were much larger than the current crisis, it really would be a cataclysm) will occur. But if it does, the resulting costs will be huge. Martin Wolf has written persuasively about the costs of major economic dislocation. Net of unemployment, political instability and even wars, the human costs of a sequel could dwarf even the current crisis. Then, the choice in the present between the "bailout" and "restructuring" plans hinges on whether expected cost (in the broadest sense), conditioned on the "bailout" strategy is higher than expected cost conditioned on "restructuring". One could formalize this argument as a decision problem, but it comes down to a judgement call on the relative probability of such a cataclysm under the two strategies and the magnitude of the dislocation. My feeling, admittedly subjective, is that the gloomy cataclysm scenario is substantially more likely under the "bailout" than "restructuring", and that the costs would be immense.

This case can be put very simply: if we do not use current political momentum to fundamentally reform a system which has shown itself to be unstable and even dangerous, a second opportunity may come at a very high price. And this is not a gamble I wish to see our leaders make.

I often wish I were Mark Thoma. If I were Mark Thoma, I could be smart and paying attention without being bitter.

So I am not wedded to a particular plan, I think they all have good and bad points, and that (with the proper tweaks) each could work. Sure, some seem better than others, but none — to me — is so off the mark that I am filled with despair because we are following a particular course of action.

Unfortunately, I have a darker temperament, a spirit less generous and optimistic than Mark's. I am filled with despair, not because what we are doing cannot "work", but because it is too unjust. This is not my country.

The news of today is the Geithner plan. I think this plan might work very well in terms of repairing bank balance sheets.

Of course the whole notion of repairing bank balance sheet is a lie and misdirection. The balance sheets we should want to see repaired are household balance sheets. Banks have failed us profoundly. We want them reorganized, not repaired. A world in which the banks are all fixed but households are still broken is worse than what we have right now. Too-big-to-fail banks restored to health are too-big-to-fail banks restored to power. The idea that fixing legacy banks is prerequisite to fixing the broad economy is a lie perpetrated by legacy bankers.

I think that critics of the Geithner plan are missing some of its tactical brilliance. My guess is that behind the scenes, Geithner has arranged a kind of J.P. Morgan moment. You know the story. During the Panic of 1907, J.P. Morgan locked a bunch of bankers in a room and insisted they lend to stave a panic. We've already seen one twisted parody of this event, when Henry Paulson locked a bunch of bankers in a room and insisted they borrow money from the Treasury. This second one is more clever. I don't think the scandal of the Geithner plan is going to turn out to be the subsidy to well-connected investors embedded in the non-recourse loan put option. On the contrary, I think that Treasury has already lined up participants for the "Legacy Loans Public-Private Investment Fund" and persuaded them to offer prices so high that despite the put, investors will expect to take a major loss. My little conspiracy theory is that the Blackrocks and PIMCOs of the world, the asset managers who do well by "shaking hands with the government", will agree to take a hit on relatively small investments in order first to help make banks smell solvent, and then to compel and provide "good optics" for a maximal transfer from government to key financial institutions.

Consider a hypothetical asset manager, PIMROCK. PIMROCK reviews a pool of loans held by the bank J.P. Citi of America, and its analysts determine they are worth 30¢ of par value. The bank holds them at 80¢ on its book. PIMROCK agrees to put down $10B to purchase loans from the pool at 82¢ thrilling stock markets everywhere. It was all just a bad dream!

Under Geithner's plan, PIMROCK's $10B permits a $10B equity investment from the Treasury. Then the FDIC levers the whole thing up, providing $6 of debt for every one dollar of equity. So, $140B of bad loans are lifted from J.P. Citi of America, nearly $90B of which is sheer overpayment to the bank.

Of course, as cash flows evolve, PIMROCK's $10B is wiped out entirely, as is the Treasury's investment. The FDIC gets repaid in a bunch of securities worth about $50B, taking a $70B loss. But, as Calculated Risk, likes to say "Hoocoodanode?" These were real market prices, Geithner or his successor will argue. Our private partners lost everything. There was no subsidy here.

Meanwhile, taxpayers will be out around $80B.

Why would PIMROCK go along with this? Because they feel it is their patriotic duty to work with the government for the good of the financial system, even if that involves accepting some sacrifices. And because they hold $100B in J.P. Citi of America bonds, and they've received assurances that if we can get the nation out of the financial pickle it's in, there will be no haircuts on those bonds. "Shaking hands with the government" means that nothing ever has to be put in writing.

Welcome to America, 2009. Change we can believe in.

The scenario I've presented is a variation on this by Karl Denninger (ht Tyler Cowen).

I liked this post today by Matt Yglesias:

My biggest concern about the PPIP approach to the banking system is that even if it works, what it does essentially is return us to the pre-crisis status quo — banks that are so large that they’re too politically powerful to regulate effective and too systemically important to be allowed to fail. That’s a recipe for dishonest transactions that produce short-term profits at the cost of blowups. One appealing element of nationalization is that it can easily be made to end in a world in which there is no institution named "Bank of America" or "Citi" and no such gigantic institution.

On the bright side, I'm thankful that we have people like Paul Krugman, Simon Johnson, and Willem Buiter, who fight the good fight while being too eminent to ignore.

On the dark side, try here, here, and here.