When can banks create their own capital?

A commenter directed me to an excellent article by Richard Werner comparing three different approaches to banking. The first two are commonly found in the economics literature, and the third is the credit creation theory of banking. Werner’s article provides a very good analysis of the three approaches, and weighs in heavily in favor of the credit creation theory.

Werner points out that when regulators use the wrong model, they inadvertently allow banks to do things that they should not be allowed to do. More precisely, Werner finds that when regulators try to impose capital constraints on banks without understanding how banks function, they leave open the possibility that the banks find a way to create capital “out of thin air,” which clearly is not the regulator’s intent.

In this post I want to point out that Werner does not give the best example of how banks can sometimes create their own capital. I offer two more examples of how banks created their own capital in the years leading up to the crisis.

1. The SIVs that blew up in 2007

You may remember Hank Paulson running around Europe in the early fall of 2007 trying to drum up support for something called the Master Liquidity Enhancement Conduit (MLEC) or more simply the Super-SIV. He was trying to address the problem that structured vehicles called SIVs were blowing up left, right, and center at the time.

These vehicles were essentially ways for banks to create capital.  Here’s how:

According to a Bear Stearns report at the time, 43% of the assets in the SIVs were bank debt, and commentators a the time make it clear that the kind of bank debt in the SIVs was a special kind of debt that was acceptable as capital for the purposes of bank capital requirements because of the strong rights given to the issuer to forgo making interest payments on the debt.

The liability side of a SIV was comprised of 4-6% equity and the rest senior liabilities, Medium Term Notes (MTNs) of a few years maturity and Commercial Paper (CP) that had to be refinanced every few months. Obviously SIVs had roll-over (or liquidity) risk, since their assets were much longer than their liabilities. The rating agencies addressed this roll-over risk by requiring the SIVs to have access to a liquidity facility provided by  a bank. More precisely the reason a SIV shadow bank was allowed to exist was because there was a highly rated traditional bank that had a contractual commitment to provide funds to the SIV on a same-day basis in the event that the liquidity risk was realized. Furthermore, triggers in the structured vehicle’s paperwork required it to go into wind down mode if, for example, the value of its assets fell below a certain threshold. All the SIVs breached their triggers in Fall 2007.

Those with an understanding of the credit creation theory of banking would recognize immediately that the “liquidity facility” provided by the traditional bank was a classic way for a bank to transform the SIV’s liabilities into monetary assets. That’s why money market funds and others seeking very liquid assets were willing to hold SIV CP and MTNs. In short, a basic understanding of an SIV asset and liability structure and of the banks’ relationship to it would have been a red flag to a regulator conversant with the credit creation theory that banks were literally creating their own capital.

2. The pre-2007 US Federal Home Loan Bank (FHLB) System

In the early naughties all of the FHLBs revised their capital plans. For someone with an understanding of the credit creation theory, these capital plans were clearly consistent with virtually unlimited finance of mortgages.

The FHLBs form a system with a single regulator and together offer a joint guarantee of all FHLB liabilities. The FHLB system is one of the “agencies” that can easily raise money at low cost on public debt markets. Each FHLB covers a specific region of the country and is cooperatively owned by its member banks. In 2007 every major bank in the US was a member of the FHLB system. As a result, FHLB debt was effectively guaranteed by the whole of the US banking system. Once again using the credit creation theory, we find that the bank guarantee converted FHLB liabilities into monetary assets.

The basic structure of the FHLBs support of the mortgage market was this (note that I will frequently use the past tense, because I haven’t looked up what the current capital structure is and believe that it has changed):

The FHLBs faced a 4% capital requirement on their loans. Using the Atlanta FHLB’s capital plan as an example, we find that whenever a member bank borrowed from the Atlanta FHL bank, it was required to increase its capital contribution by 4.5% of the loan. This guaranteed that the Atlanta FHL bank could never fall foul of its 4% capital requirement — and that there was a virtually unlimited supply of funds available to finance mortgages in the US.

The only constraint exercised by FHLBs on this system was that they would not lend for the full value of any mortgage. Agency MBS faced a 5% haircut, private label MBS faced a minimum 10% haircut, and individual mortgages faced higher haircuts.

In short, the FHLB system was designed to make it possible for the FHLBs to be lenders of last resort to mortgage lenders. As long as a member bank’s assets were mortgages that qualified for FHL bank loans, credit was available for a bank that was in trouble.

The system was designed in the 1930s — by people who understood the credit creation theory of banking — to deliberately exclude commercial banks which financed commercial activity and whose last-resort lender was the Federal Reserve. Only when the FIRRE Act in 1989 was passed subsequent to the Savings and Loan crisis were commercial banks permitted to become FHLB members.

From a credit creation theory perspective this major shift in US bank regulation ensured that the full credit creation capacity of the commercial banking system was united with the US mortgage lending system making it possible for the FHLBs to create their own capital and use it to provide virtually unlimited funds to finance mortgage lending in the US.

 

Advertisements

In Defense of Banking II

Proposals for reform of the monetary system based either on public access to accounts with the central bank or on banking systems that are 100% backed by central bank reserves and government debt have proliferated since the financial crisis. A few have crossed my path in the past few days (e.g. here and here).

I have been making the point in a variety of posts on this blog that these proposals are based on the Monetarist misconception of the nature of money in the modern economy and likely to prove disastrous. While much of my time lately is being spent working up a formal “greek” presentation of these ideas, explaining them in layman’s terms is equally important. Thanks to comments from an attentive reader, here is a more transparent explanation. Let me start by quoting from an earlier post that draw a schematic outline of Goodhart’s “private money” model :

The simplest model of money is a game with three people, each of whom produces something another seeks to consume: person 2 produces for person 1, person 3 produces for person 2, person 1 produces for person 3. Trade takes place over the course of three sequential pairwise matches: (1,2), (2,3), (3,1). Thus, in each match there is never a double coincidence of wants, but always a single coincidence of wants. We abstract from price by assuming that our three market participants can coordinate on an equilibrium price vector (cf. the Walrasian auctioneer). Thus, all these agents need is liquidity.

Let the liquidity be supplied by bank credit lines that are sufficiently large and are both drawn down by our participants on an “as needed” basis, and repaid at the earliest possible moment. Assume that these credit lines – like credit card balances that are promptly repaid – bear no interest. Then we observe, first, that after three periods trade has taken place and every participant’s bank balance is zero; and, second, that if the game is repeated foerever, the aggregate money supply is zero at the end of every three periods.

In this model the money supply expands only to meet the needs the trade, and automatically contracts in every third round because the buyer holds bank liabilities sufficient to meet his demand.

Consider the alternative of using a fiat money “token” to solve the infinitely repeated version of the game. Observe that in order for the allocation to be efficient, if there is only one token to allocate, we must know ex ante who to give that token to. If we give it to person 3, no trade will take place in the first two rounds, and if we give it to person 2 no trade will take place in the first round. While this might seem a minor loss, consider the possibility that people who don’t consume in the first stage of their life may have their productivity impaired for the rest of time. This indicates that the use of fiat money may require particularized knowledge about the nature of the economy that is not necessary if we solve the problem using credit lines.

Why don’t we just allocate one token to everybody so that we can be sure that the right person isn’t cash constrained in early life? This creates another problem. Person 2 and person 3 will both have 2 units of cash whenever they are making their purchases, but in order to reach the equilibrium allocation we need them to choose to spend only one unit of this cash in each period. In short, this solution would require people to hold onto money for eternity without ever intending to spend it. That clearly doesn’t make sense.

This simple discussion explains that there is a fundamental problem with fiat money that ensures that an incentive compatible credit system is never worse and in many environments is strictly better than fiat money. This is one of the most robust results to come out of the formal study of economic environments with liquidity frictions (see e.g. Kocherlakota 1998).

In response to this I received the following question by email:

In your 3 person model, [why not allocate] a token to everybody? – I don’t understand how you reached the conclusion that “this solution would require people to hold onto money for eternity without ever intending to spend it”. If people have more units of cash than they need for consumption, the excess would be saved and potentially lent to others who need credit?

This question arises, because I failed in the excerpt from my post above to explain what the implications of “allocating a token to everybody” are when translated into a real world economy. In order for an efficient outcome to be achieved, you need to make sure that everybody has enough money at the start of the monetary system so that it is not possible that they will ever be cash-constrained at any point in time. In my simple model this just implies that everybody is given one token at the start of time. In the real world this means that every newborn child is endowed at birth with more than enough cash to pay the full cost of U.S. college tuition at an elite institution (for example).

Turning back to the context of the model, if the two people with excess currency save and lend it, we have the problem that the one person who consumes at the given date already has enough money to make her purchases. In short there is three times as much currency in the economy as is needed for purchases. What this implies is that we do not have an equilibrium because the market for debt can’t clear at the prices we have assumed in our model. In short, if we add lending to the model then the equilibrium price will have to rise — with the result that nobody is endowed with enough money to make the purchases they want to make. Whether or not an efficient allocation can be obtained by this means will depend on the details of how the lending process is modeled. (The alternative that I considered was that there was no system of lending, so they had to hold the token. Then when they had an opportunity to buy, choose to spend only one token, even though they were holding two tokens. This is the sense in which the token must be held “for eternity” without being spent.)

Tying this discussion back into the college tuition example. If, in fact, you tried to implement a policy where every child is endowed at birth with enough cash to pay elite U.S. college tuition, what we would expect to happen is that by the time these children were going to college the cost would have increased so that they no longer had enough to pay tuition. But then of course you have failed to implement the policy. In short, it is impossible to “allocate a token to everybody”, because as soon as you do, you affect prices in a way that ensures that the token’s value has fallen below the value that you intended to allocate. There’s no way to square this circle.

Connecting this up with bitcoin or deposit accounts at the central bank: the currently rich have a huge advantage in a transition to such a system, because they get to start out with more bitcoins or larger deposit accounts. By contrast in a credit-based monetary system everybody has the opportunity to borrow against their future income.

The problem with the credit-based monetary system that we have is that guaranteeing the fairness of the mechanisms by which credit is allocated is an extremely important aspect of the efficiency of the system. That is, in a credit-based monetary system fairness-based considerations are not in conflict with efficiency-based considerations, but instead essential in order to make efficiency an achievable goal.

Because of the failure to model our monetary system properly, we have failed to understand the importance of regulation that protects and supports the fair allocation of credit in the system and have failed to maintain the efficiency of the monetary system. In my view appropriate reforms will target the mechanisms by which credit is allocated, because there’s no question that in the current system it is allocated very unfairly.

The problem with proposals to eliminate the debt-based system is that as far as I can tell, doing so is likely to just make the unfairness worse by giving the currently rich a huge advantage that they would not have in a reformed and well-designed credit-based monetary system.

Corporate liability and the “crimes were committed” approach to law enforcement

Pursuant to Attorney General Loretta Lynch’s welcome change in DoJ policy, it occurred to me that an old draft post of mine might actually merit being posted, so here goes:

After listening to a presentation on the impressive growth in enforcement actions resulting in corporate criminal liability a few months ago, it occurred to me that people without legal training might not actually understand the reasoning behind the critique that individual prosecutions should almost always accompany corporate criminal liability. (The presenter at one point framed such critiques as claiming that prosecutors were colluding with management against the shareholders.)

The problem with corporate criminal liability is this: every crime has a mens rea or element of intent that must be proved as part of the prosecutor’s case. Negligence is one of the lower levels of mens rea, but many instances of negligence are not crimes. Often a “knowing” or “should have known” standard is applied in criminal law.

When a prosecutor chooses to seek corporate criminal liability, without bringing any cases of individual criminal liability, the problem is whether it makes logical sense to argue that the corporation had the mens rea for the crime, but no individual in the corporation had the mens rea (or the one with the mens rea managed not to take relevant action in promotion of the crime). Now one can dream up special circumstances where this position would actually be logical, but it seems to a lot of people that this situation should be rare.

Critics of corporate liability (I’m thinking of Judge Rakoff and Bill Black here, for example) would probably argue that pursuing corporate criminal liability, without pursuing individual liability is tantamount to stating that a crime was committed, but we don’t know by whom. (Note that the reverse where there is individual criminal liability without corporate criminal liability is likely to be much more common. Rogue employees and a genuine effort on the part of the corporation to avoid the criminal activity would both be good reasons – though not necessarily successful reasons – for not extending criminal liability from an individual to the corporation.)

Overall an important criticism of the growth of deferred prosecution agreements and non prosecution agreements is that finding this growth acceptable in the absence of individual prosecutions is essentially lowering the standards for what a prosecutor is supposed to do. “A crime was committed, but I don’t know by whom” should not be the normal stopping point for a prosecutor’s case.

The argument is, of course, not that there should never be corporate criminal liability without an accompanying case for individual liability, but simply that this outcome should be relatively rare. In general, we want our prosecutors to think of their jobs as going all the way to finding out “who done it,” and not stopping with “a crime was committed” and a fine was paid.

In short, the argument against treating a finding of corporate criminal liability as an end point is not about “collusion,” but instead goes to the heart of what it means to enforce the law.

A question for Martin Wolf: Was the crisis “unprecedented”?

In the acknowledgements to The Shifts and the Shocks (which I am currently reading) Martin Wolf has stated that friends like Mervyn King encouraged him to be more radical than initially intended, and I suspect, as I read Chapter 4, “How Finance Became Fragile,” that this was one of the chapters that was affected by such comments. In particular, I see a contradiction between the initial framing of financial fragility which focuses on Minsky-like inherent fragility, and the discussion of regulation. Was this a crisis like that in the U.K. in 1866 or in the U.S. in the 1930s, or was this an “unprecedented” crisis that was aggravated by the elimination of legal and regulatory infrastructure that limited the reach of the crisis in 1866 and the 1930s? I am troubled by Wolf’s failure to take a clear position on this question.

Because Martin Wolf understands that the government intervention due to the crisis was “unprecedented” (at 15), I had always assumed that he understood that the nature of the 2007-08 crisis was also unprecedented. He appears, however, to be of the opinion that this crisis was not of unprecedented severity. Martin Wolf really surprised me here by taking the position that: “The system is always fragile. From time to time it becomes extremely fragile. That is what happened this time.” And by continuing to treat the crisis as comparable to the 1930s in the U.S. or 1866 in the U.K. (at 123-24).

His treatment of how regulation played into the crisis could be more thorough. He concludes that “the role of regulation was principally one of omissions: policymakers assumed the system was far more stable, responsible, indeed honest, than it was. Moreover, it was because this assumption was so widely shared that so many countries were affected.” (at 141). Wolf is undoubtedly aware of the many changes to the legal framework that protected the U.K. financial system in 1866 and the U.S. financial system in the 1930s that were adopted at the behest of the financial industry in both the U.S. and the U.K. (Such changes include the exemption of derivatives from gambling laws, granting repurchase agreements and OTC derivatives special privileges in bankruptcy, and the functional separation of commercial banking from capital markets.) Is the argument that these changes were not important? or that these changes fall in the category of omission by regulators? Given the preceding section of this chapter, it would appear that he believes these changes were not important, but given the conclusion of the chapter, I am not so sure.

In the conclusion, Wolf takes a somewhat more aggressive stance than he does in the body of the chapter: “The crisis became so severe largely because so many people thought it impossible.” (at 147). So maybe the crisis is unprecedented compared to 1866 and the 1930s. (In 1866 at least the possibility of financial collapse seems to have been recognized.) He also adds two more points to the conclusion that I didn’t see in the body of the chapter: the origins of the crisis include “the ability of the financial industry to use its money and lobbying clout to obtain the lax regulations it wanted (and wants)” and the fact that “regulators will never keep up with” the ability of the financial industry to erode regulation (at 147).

Thus, once I reached the conclusion of the chapter, it was no longer clear that Wolf views this crisis as primarily an example of inherent fragility. He has laid out the argument for how the financial industry successfully removed the legal and regulatory protections that were in place in 1866 and the 1930s. So this is my question for Mr. Wolf: Was the crisis itself “unprecedented” in the course of the last two centuries of Anglo-American financial history, or was this just a Minsky moment like many that have come before?

Time priority is the key to fair trading

A true national market system would have the following property. There are clearly defined points of entry to the system: that is, when an order is placed on specific exchanges, ECNs or ATSs, they will count as part of the system. These orders are time-stamped by a perfectly synchronized process. In other words, it doesn’t matter where your point of entry is, the time-stamp on your order will put it in the correct order relative to every other part of the system.

Order matching engines are, then, required to take the time to check that time-priority is respected across the national market system as a whole.

This structure would eliminate many of the nefarious aspects of speedy trading, while at the same time allowing high-speed traders to provide liquidity within the constraints of a strictly time priority system. Speedy orders couldn’t step in front of existing orders, because time-priority would be violated. Cancellations couldn’t be executed until after the matching engine had swept the market to look for an order preceding the cancellation that required a fill. In short, speedy traders would be forced to take the actual risk of market making, by always being at risk of having their limit orders matched before they can be cancelled.

Overall, it seems to me that the error the SEC made was in creating a so-called “national market system” without a time-priority rule.

Note: this post was probably influenced by @rajivatbarnard ‘s tweets about this same topic today.

Update: Clark Gaebel explains very clearly that we don’t have anything remotely resembling a “national market system.” We have a plethora of independent trading venues and your trade execution is highly dependent on your routing decisions.

Is modern finance the source of secular stagnation?

The recent discussion of secular stagnation has once again brought up the question of whether there is a “savings glut” that is aggravating our problems. To the degree that a savings glut exists, it generally has the property that it is focused on the safest assets. That is, for reasons that remain unclear, the collapse of returns on safe assets has not be sufficient to turn this “savings glut” into a vast flow of funds into real-economy risky assets.

I believe that there has been too little discussion of the possibility that the marginal “investors” who have created the savings glut are to be found in the financial industry itself. Although it is certainly true that after the Asian crisis developing countries became net savers — and I do not discount this factor in the flow of savings — at the same time there was a significant transformation of the financial industry. The growth of derivatives was accompanied by the growth of the collateralization of derivatives and the latter phenomenon accelerated after the LTCM crisis which took place one year after the Asian crisis. Thus a non-trivial component of the “savings glut” is likely to be the demand for collateral of the financial industry itself. This source of demand can also explain the strong preference for “safe” assets, since risky assets can easily become worthless as collateral in a liquidity crisis.

If my thesis is right, then Basel III is probably aggravating the “savings glut” problem by increasing the demand for collateral on the part of financial institutions. Thus there has recently been discussion of the existence of a collateral shortage, which sounds to me like the mirror view of a savings glut. (Note that the question of a collateral shortage is complicated by the fact that collateral circulates just like deposits in a banking system, but this issue goes beyond what I want to address in this post.)

One problem with a “savings glut” that is generated in significant part by a demand for assets to be used as collateral is that it is likely to create a segmented markets effect: that is, a significant demand for highly rated assets can coexist with very tepid demand for typical, real-economy, somewhat risky assets that don’t have good characteristics as collateral. This kind of demand for assets is unlikely to play a part in economic recovery by supporting an increase in lending.

The basic problem is this: If the role of the banking system in the economy is to manage and to bear risk for the rest of the economy, then trying to make the banking system “safe” by requiring it to hold vast amounts of collateral and by making it distribute to others the risk that it is supposed to be bearing may actually prevent it from performing its role in the economy. If our banking system is no longer capable of bearing good old-fashioned credit risk, but must find others upon whom to lay that risk, then we should not be surprised that the outcome is low levels of lending to the real economy, low investment, and poor growth. In short, we cannot make the financial system “safe,” by discouraging it from carrying real economy risk, because that undermines economic growth and the performance of all assets.

Does the 2007-2008 crisis show that the universal banking experiment has failed?

The Anglo-American universal banking experiment started in 1986 with Britain’s Big Bang which was quickly followed by regulatory policies in the US that would lead to the formal repeal of Glass Steagall a decade or so later. The question that needs to be asked is whether the 2007-2008 crisis is evidence of the failure of this quarter-century of experimenting with universal banking.

In Germany universal banking has been successful over the long-run, but Germany has a civil, not a common law legal system and a social structure that ensures that companies are managed in the interests of many participants in addition to those of shareholders/management. When universal banking is combined with Anglo-American law and social norms, it is possible that it generates pathological behavior that is not evidenced by the German economy.

A standard objection to the claim that universal banking is the underlying source of the crisis is that the only banks that were allowed to fail were not universal banks, but investment banks. This objection ignores that the whole investment banking industry had been reshaped over the decades preceding the crisis by the need to compete with the universal banks, so the fact that it was the investment banks that failed tells us nothing. Furthermore there is significant evidence that one or two of the universal banks did not fail only because the government considered them too big to fail.

John Quiggin recently argued that Wall Street isn’t worth it and that we should put an end to the universal banking experiment:

The only remaining option is to separate these markets entirely from the socially useful parts of the financial system, then let them fail. Publicly guaranteed banks should be banned from engaging in all but the most basic financial transactions, such as issuing loans and bonds and accepting deposits. In particular, banks should be prohibited from doing any business with institutions engaged in speculative finance such as trade in derivatives. Such institutions should be required to raise all their funds directly from investors, on a “buyer beware” basis, and should never be bailed out, directly or indirectly, when they get into trouble.

Matthew Yglesias critiques this view arguing that “it’s a very hard concept to operationalize.” And then limits his focus to derivatives regulation. He writes:

But while it’s easy to say “we should allow derivatives trading for the purpose of hedging but not for the purpose of speculating” (certainly that’s what I think), it’s a lot harder to write precise legislative and regulatory language that accomplishes that goal. If you look at something like the Harvard interest-rate-swap fiasco, it’s difficult to say precisely where this crossed the line from a reasonable hedge to just gambling with endowment money.

Yglesias’ critique, however, misses Quiggen’s point: commercial banks shouldn’t be engaged in market making or in trading on financial markets at all. The difficulty of implementing the Volcker rule is that it’s trying to draw a line between trading that’s okay (e.g. market making) and trading that’s not okay (proprietary trading). Quiggen is stating that commercial banks should not be engaged in either of these activities. This is a much easier policy to implement (see Glass-Steagall).

This may leave open some room to allow commercial banks to be end-users of financial contracts like interest rate swaps for hedging purposes, but drawing this distinction is much less difficult than Yglesias implies. The distinction between the use of derivatives for hedging or for speculating is precisely the same distinction that is drawn in insurance markets between an insurable interest and the absence of one. Given that we know that drawing the distinction is not an insuperable problem in insurance markets, it’s far from clear why the problem suddenly becomes insuperable when the label “derivative” is placed on the financial contract.

[In addition the whole point of Felix Salmon’s post on the Harvard IRS fiasco is that it was clearly gambling at the time the swaps were entered into. Salmon states with barely veiled sarcasm “Larry was certain of two things: firstly that his beloved Allston project was a go — despite the fact that he hadn’t raised the funds for it, and secondly that interest rates would rise by the time construction started. Therefore, he decided to lock in funding costs by using forward swaps.” In short Salmon is stating that the contracts represents two gambles, first, on the future need for the funds, and, second, on the future path of interests rates. While ex post we know that in 2008 Harvard would have been better off holding on to its side of the bet rather than buying itself out of the contracts, the post is crystal clear about the fact that these swaps were never a “reasonable hedge.”]

While we can certainly debate whether or not the 2007-2008 crisis demonstrates that the Anglo-American experiment with universal banking has failed, arguments that it’s just too hard to reverse the experiment only play into the interests of the universal banks and probably should not be given much weight. If policies that were implemented at the tail end of the last century completely destabilized our financial system, it is clearly worth the effort to find a way to reverse those policies.