Economic Theory does not Predict that Markets Produce Efficient Outcomes

Ingrid Robeyns (h/t Steve Waldman) in Economics as a Moral Science questions economists’ use of Pareto efficiency arguing that it is not value-neutral, but I actually think that adding normative analysis to economics is of distinctly secondary importance to simply insisting that economists and those who make policy on the basis of economic principles get their positive analysis of economics right.

First, in most real-world economic environments economic theory very clearly fails to predict that market trade will produce efficient outcomes. The prediction of efficient prices as a market outcome relies fundamentally on the assumption that all participants in the market are “price-takers” — in other words, all market participants must honestly reveal their private information about the their inventories and their desires, or economics does not predict efficient prices. As soon as the issue was clearly framed, economic theorists determined that efficient market equilibria are incentive compatible for the participants in the market when there are very many (technically infinite) market participants on both sides of every single market, but in general makes no predictions about efficiency in other circumstances. (See, e.g., John Geanokoplos “Arrow-Debreu Model of General Equilibrium” p. 122.) In short, because there is no reason to believe that efficient market equilibria are likely to be incentive compatible with real-world behavior except when all market participants are very small relative to the size of the market, game theory is extremely popular among economic theorists.

Second, the concept of Pareto efficiency divides social states into only two categories: states where the allocation is efficient and states where it is not. Thus, once we have reason to believe that the market is unlikely to produce an efficient outcome, the whole set of Pareto efficient outcomes comprises an appropriate target for government intervention.  For this reason, Pareto efficiency states that if we can design a policy that takes everything (or something or nothing) away from the rich, but ends up at an allocation where nobody’s welfare can be improved without reducing the welfare of someone else, then the government policy is an improvement over the market. In short, the second theorem of welfare economics makes it crystal clear that government policies that have redistributive aspects are entirely consistent with economic efficiency.

Overall, the problem with economics is not the use of Pareto efficiency, but the failure to acknowledge the implications of economic theory for the importance of the structure of our markets. Economics predicts efficient prices only when markets are carefully structured to make the revelation of private information incentive compatible. (See Mechanism design and  Auction theory.) An ill-defined concept of a “market” is not predicted to produce the same result, but instead to induce strategic behavior about when and how to reveal information.  When market participants are behaving strategically in an environment not designed like an auction to induce the revelation of truthful information, economic theory does not predict that the outcome will be efficient.

The fundamental problem with modern economics does not lie in the use of Pareto efficiency, but in the failure of both the broader economics profession and policy-makers to incorporate the implications of economic theorists’ formal economic analysis into their intuition about how the economy works.

Why banks are special: A model

In a post titled “Commercial Banks as  Creators of Money” Paul Krugman asserts the “non-specialness of banks.” Cullen Roche responds that “banks are the oil in our monetary machine.”

I think Krugman’s problem is that he doesn’t have a formal model that can help him understand what makes banks so important. Instead he refers to Diamond-Dybvig and Tobin-Brainard.  Diamond-Dybvig is an excellent workhorse model, but it’s a model of why banks face runs and doesn’t really attempt to take a close look at the role banks play in the broader economy.  Since I’m not very familiar with Tobin-Brainard, I’m going to discuss Krugman’s description of it:

if we ask, “Is the volume of bank lending determined by the amount the public chooses to deposit in banks, or is the amount deposited in banks determined by the amount banks choose to lend?”, the answer is once again “Yes”; financial prices adjust to make those choices consistent.

But this discussion abstracts from the fact that the funds the public is able to deposit in banks depend fundamentally on whether or not the bank is willing to lend those funds.  A classic example is the difficulty of borrowing against one’s future wage income. For purposes of consumption smoothing members of the public would often like to hold as deposits cash representing future wage income that was posted to a checking accounts against a loan that has to be paid back to the bank.  However, when the banks refuse to provide this form of lending, the public doesn’t get to hold these deposits in banks or take advantage of the consumption smoothing opportunities that such deposits would allow.  In short, the criteria that banks use for making lending decisions are (rather obviously) crucial inputs determining the size of the deposit base.  Because of these dynamics, it appears that “the answer is ‘Yes'” only as long as the banks’ lending criteria are held constant.  This, of course, is a unexceptional modelling decision, but it is incumbent on Krugman to understand that Tobin-Brainard depends fundamentally on not modelling the details of bank lending. [Note that Steve Waldman gives a more general explanation of the problem with Tobin-Brainard: “it must be reasonable to model the nonbank private sector as if it were a unified actor with preferences independent of the behavior of the banking system.”]

For a better understanding of the central role that banks play in the monetary system, I suggest that Krugman take a look at a paper I wrote half a decade ago, An Idealized View of Financial Intermediation. This paper looks only at the role banks play in facilitating payments and abstracts entirely from portfolios and the banking system’s role in long-term investment. All banks have is creditworthiness, but by adding this to the economy banks are able to offer credit lines and make it possible for everybody’s debts to be accounted for and made enforceable.

(More specifically, I introduce a monetary friction into a general equilibrium model of a repeated endowment economy and demonstrate that credit lines underwritten by banks are a better solution to the monetary problem than cash: as long as loss of access to bank lending is a sufficient penalty to induce bank customers to repay debt, banks need only set a credit limit for everyone that’s high enough for the richest person to overcome the monetary friction and then everyone will choose to use the right amount of credit. The model also shows why bank customers will choose not to default: loss of access to bank services restricts your ability to participate in the broader economy, and leaves you with a very bad set of choices. Because there is no investment in this environment the only role played by the infinite horizon is to provide for a future punishment for bank customers who do not repay debt.)

This is a formal model, so I abstract from issues like investment, portfolios, and bank default.  On the other hand, since Krugman seems to need a model to explain to him  in what way banks add something both unique and extraordinarily important to the economy, this paper provides an answer: banks make credit and credit-based transactions possible, because they are trustworthy. In short, banks are the cornerstones of the economy’s circulatory system, which is, I think, also Cullen Roche’s point.

What do banks do? A response to Krugman and Rowe

Paul Krugman’s critique of Steve Keen’s work has set off a debate in the econoblogosphere over the degree to which bank creation of money is constrained by central bank policy.  The clearest explanation of what, I think, is also Krugman’s view is presented by Nick Rowe:

An individual commercial bank can create money out of thin air simply by buying something. But the money it creates may not be its own. Its money may subsequently be redeemed for the money of another bank, or the central bank. The individual commercial bank that wants apermanent increase in its stock of money may need to persuade people not to redeem its money. Whether or not it needs to do anything to persuade them all depends on how the other banks, especially the central bank, react.

I can imagine a world in which an individual commercial bank can permanently create money. All it needs is that the central bank, which is free to do what it wants, allows the supply of its own money to increase in proportion to the supply of commercial bank money. This is what would happen if the central bank targeted a rate of interest, for example. One commercial bank creates money, and all the other commercial banks, and the central bank, respond to the hot potato process by increasing their own money supplies in response to the rising demand for loans and deposits and currency.

I read these two paragraphs as a statement that the velocity of money (with respected to the monetary base or currency) tends to be constant.  It’s far from clear to me why this would be the case — probably because the basic model of a credit-based monetary economy that I work with has infinite velocity; that is, the model demonstrates that the optimal solution to a monetary problem is an institutional structure with a monetary base of zero.  (See here for a formal model and here for a heuristic discussion of the model.)  The basic idea of this framework is that if a monetary problem exists, “good” credit will always be a better solution to the problem than money.  The question then becomes how do you establish a “good,” i.e. very low default, credit system.

My model of banks has them financing working capital (which they have been doing for centuries).  If one thinks of bank lending as the means by which the value of inalienable assets — the entrepreneur’s personal knowledge — are realized, then when banks are lending wisely and true value is being realized via bank lending while at the same time the monetary base is being held constant, the velocity of that base should increase since true new value is being created.  The problem arises when banks miscalibrate and lend unwisely, then there can be a temporary increase in velocity before the mistake is recognized as such, followed by a decline back to the “true” velocity given the “true” value of output.

I think the difference between my model and the one that Krugman and Rowe are working with is that they assume that economic output is not a function of bank finance, that is, output simply comes into being frictionlessly* and banks are then tacked onto this economy.  (In their posts Krugman‘s banks “offer a better tradeoff between liquidity and returns” and Rowe‘s banks “buy something”.)

By contrast, I assume that any output that exists, exists because it is financed by the banks — thus, an increase in “good” bank activity necessarily increases real GDP.  If the monetary base is held constant, the velocity of money will increase.  If this model is correct, then there is no need to “persuade” anyone to hold additional bank money, because the economy demands that money in order to pay for the real goods that were produced using bank financing.

*  Update 4-3-12:  Since a “friction” has many definitions in the economic literature, I should be specific that I am talking about monetary frictions here that create a role for a means of exchange.  In my lexicon that’s the missing friction that makes general equilibrium models very hard to relate to the real world — and by assumption eliminates the obvious role of banks.

Incrementally reducing your negative equity is not saving

… at least as long as the goal is to assess the effect of the activity on the macroeconomy today.

Rortybomb directs us to Matt Rognlie‘s critique of “deleveraging.”  Rortybomb correctly points out Rognlie’s error in using aggregate data to discuss consumer behavior and links to a year-old analysis explaining that it’s the middle class that’s overleveraged.  I have a different bone to pick.

While Rognlie nominally acknowledges that the deleveraging problem is specific to the aftermath of an asset bubble (“consumers and businesses experienced an enormous hit to net worth”), he restates the problem as something that is not precisely the same, that consumers desire to spend less and save more.

The aftermath of an asset bubble implies that many economic participants owe more than the value of the assets securing the loan.  This has the immediate implication that either (i) defaults must take place, transferring full ownership of the assets to the lenders or (ii) assets are “locked in” and those who hold title to the assets cannot sell them (because the assets undersecure a lien, so in some sense this title is only nominal) or (iii) some combination of (i) and (ii).

The theory to which Rognlie refers tends to assume that (i) takes place instantaneously:  After a bad economic realization,  borrowers who owe more than the asset is worth strategically default, and losses are immediately recognized as lenders sell the foreclosed assets off to the highest bidder == highest value user.  Economic recovery takes place quickly because the model doesn’t allow for assets to be held and used by low value users.

As I understand his argument, Richard Koo sees solution (i) as so full of costs that typical economic models don’t recognize that it can be rejected out of hand.  And indeed most policy-makers seem to agree with him.  Banks are not asked to recognize their losses, borrowers are induced through temporary payment reduction plans like HAMP to continue making payments on loans that would lead to strategic default in an economic model, etc.

The balance sheet recession theory focuses on a world, like the one we are experiencing now, where strategic default does not take place on a large scale, and assets continue to be held by entities that are paying more money for them then they are worth on the market.  Such debtors have the use of the asset, but cannot sell it because the market value is insufficient to pay off the debt.  They are “locked in” at least until such time as they choose to default and transfer ownership to the lender.  This implies that we are now are in a world where assets are not as easily transferred to their highest value use (whether due to policy decisions, social pressures, or other concerns) as economic models tend to assume.

Observe, in addition, that spending less in order to have the means to make a debt payment on an underwater loan to a financial institution is very different in macroeconomic terms than spending less in order to invest the money in some asset.  The reason for this is a simple matter of accounting:  banks assume when they lend that the debt will be repaid (at least until such time as there is a significant default and they transfer the loan to an impaired asset category).  Thus the payment of debt has already been accounted for by the bank in its assets and adds nothing to the economy’s capacity to lend.  Savings/investment are a very different matter, because these are sums that an entity sets aside to provide itself with future income, but there aren’t many realistic future housing market scenarios in which reducing the negative equity in your home from 50% to 49% by making a year’s worth of payments is going to result in future income within the next decade or two.

Thus, it’s not true that all the two-earner households who have become one-earner households and are now cutting back on consumption in order to make their mortgage payments on underwater homes are “saving” in a meaningful macroeconomic sense — because many of these households don’t expect these payments to result in home equity for at least a decade (and since they are likely to lose the house in the end anyhow, they are really just renting == consuming housing services), and the banks’ current balance sheets are founded on the assumption that these payments will be made.  These households are cutting back on their economic activity, but the “savings” from doing so added to economic activity in the year in which they took out the loan and bought the house, not now.  Only if you want to argue that when banks don’t have to recognize losses on the bad loans they made, that also constitutes savings for macroeconomic purpose, can you claim that the vast amounts currently being paid on underwater mortgages are savings.  This is, however, at best a disputable position.

In short, in order for Rognlie’s theory to apply to our current situation he needs to consider the effectiveness of monetary policy in an environment where a principal goal of policy-making is to protect financial institutions from experiencing (or at least realizing) losses and where the policy is implemented at the cost of obscuring the valuation information available to shareholders and of encouraging deeply underwater consumer-mortgagors to continue making payments on their loans (e.g. HAMP).   Confusing the macroeconomic effects of paying off debt for the purpose of incrementally reducing negative equity (and in many cases insolvency) with the macroeconomic effects of saving is a serious error, but one that is easily made by those who work with models that don’t take insolvency and the bankruptcy process into account.

Misunderstanding data on economic growth

According to Paul Krugman, we know that the growth estimates of the naughties are close to accurate because:

On the financial side, the point is that we measure growth by output of final goods and services, and fancy finance is an intermediate good; so if you think Wall Street was wasting resources, that just says that more of the actual growth was created by manufacturers etc., and less by Goldman Sachs, than previously estimated.

This just shows how little economists have tried to understand the nature of recent financial innovations.  Wall Street can now create synthetic assets.  That’s what a synthetic CDO is — it goes on someone’s balance sheet as an asset and there’s no requirement in accounting conventions that it go on somebody else’s balance sheet as a countervailing liability.  AIG is just an illustration of how the accounting for such CDOs takes place.

In an environment where Wall Street can fabricate assets that enter into financial accounts in this way, it’s not realistic to claim that “fancy finance is just an intermediate good”.  That used to be true in the good old days, when there was no reason to believe that the CDS contracts underlying synthetic CDOs would be enforceable contracts, but after the CFMA of 2000 (and other changes in the law), that isn’t true anymore.

For this reason, the economic assumptions underlying analyses like Krugman’s and Steindel’s do not reflect reality — and in fact they function as a blinder that prevents these economists from seeing and understanding what’s actually going on.  By assuming that financial institutions can’t do what they did do, economists hobble their understanding of the nature of the current economic malaise.

If economists ruled the financial world

The negative reaction by many in the financial industry to this New York Times article critiquing the structure of over the counter derivative markets motivates me to make a very simple point:  Economic theory makes it clear that there is no reason to believe that trading activity will be socially beneficial when the market structure is as described in the article.

Elementary economics tell us that in order for the invisible hand of the market to work, pricing (including bids and offers) must be transparent.  That is, economics is explicit that it is only where there is public information about prices that self-interested behavior is socially beneficial or that the ability to make money on the market is an indicator that a valuable service is being provided to the economy.  [Ask any economist to explain a competitive equilibrium to you and you will find that good price information is a necessary condition for “first-best” social welfare to be achieved.]

Because transparent pricing is a pre-requisite to market trade having positive implications for society as a whole, if economists ruled the financial world these would be the foundational principles of the marketplace:

1.   [Enforceability of pricing regulations]  In general contracts that are not traded on SEC or CFTC regulated trading forums are unenforceable.  (Note that this is a revitalization of the reforms enacted during the Depression by the passage of the Securities Exchange Act and the Commodities Exchange Act.  Note also that I recognize the need for limited exceptions to this rule.)

2.  [Price transparency pre-trade]  Best bid and offer prices posted by market makers and other traders on all trading forums are publicly available in real time.  Market depth information is also publicly available in real time.  The cost of providing this data to the public is covered by a fee proportional to trade on the market (e.g. a penny per $100 traded).

3.  [Price transparency post-trade]  All completed trade data (item, time, price and quantity) on all trading forums is publicly available.

4.  [Meaningful prices]  All bids and offers have a minimum duration of one minute.

A critique of standard macro

I looked over Michael Woodford’s “Financial Intermediation and Macroeconomic Analysis” and the first thing I noticed was that his whole conception of financial intermediation is completely different from mine.  I think that Woodford’s approach to financial intermediation is fairly standard in macroeconomics, and thus that he is following well recognized norms when he avoids addressing the complex relationship between banks, money and the macroeconomy.  The habit of divorcing the study of banking and financial instruments from the analysis of monetary and macroeconomics seems to be so deeply ingrained that even in a paper that seeks explicitly to integrate the two, the bias in favor of the existing literature somehow manages to overwhelm the goal of taking financial intermediation seriously.

Unpleasant as this conclusion may be, I think much of modern macroeconomics needs to be understood as fair weather macroeconomics which produces accurate results only when the financial system is operating optimally;  genuinely integrating financial intermediation with macroeconomics will require reconsideration of the fundamentals of macroeconomics.  Afterall banks and money market funds issue liabilities that are counted in basic measures of the money supply, so it is a mistake to discuss “money” without first developing a thorough understanding of the relationship between financial institutions and money.

The biggest conceptual flaw in Woodford’s article comes from viewing the financial system as a place where trust-worthy intermediaries take in deposits and lend out the depositors’ funds.  This most definitely is not the traditional role of a bank in an economy.  Instead, because banks are trust-worthy – and their liabilities are accepted as a means of payment – banks originate loans to creditworthy borrowers as a means of creating liabilities to put into circulation.  That is, loans generate deposits, not vice versa.  In historical environments without capital requirements, banks potentially had unlimited leverage.  (Arguably this is the same conceptual error that Henry Thornton’ 1801 Paper Credit accused Adam Smith of making in the Wealth of Nations.)  In practice, of course, bank leverage was limited by the cost to the banker of failing – such costs usually included loss of all personal assets and sometimes the death penalty.

Once one understands that loans are the means by which banks put their liabilities into circulation, one also understands that banks for the most part are not constrained by some need to have funds flow into the bank before the bank can make loans.  That is, banks can be capital-constrained only to the degree that they face capital requirements on their lending, they are never capital constrained because they have yet to receive an inflow of funds from depositors.  (I will not discuss reserve requirements because the Fed’s interest rate target ensures that reserves can be borrowed at a rate well below the bank’s lending rate and therefore reserves are almost never a binding constraint on bank lending.)

I think that the best way to think about financial intermediaries is that in a modern economy there are public and private issuers of money.  Both the public and the private sector are able to issue money only because they are perceived by the public as trustworthy:  government is able to issue money because the general populace trusts that inflation will be kept within reasonable bounds, and financial intermediaries are able to issue money because the public views them as unlikely to default on their liabilities.

Because financial intermediaries are entrusted by the public with the job of issuing money, they have extraordinary access to leverage.  Their leverage is usually constrained by regulation in the form of capital requirements.

To make my description of financial intermediation a little more concrete, let’s look at the constraints faced by JP Morgan when making a loan.  Currently JP Morgan’s tier 1 leverage ratio is 5.97%, while it is required to have a leverage ratio of at least either 3 or 4%.  Given that JP Morgan has total assets of $1.7 trillion, this implies that per regulatory requirements JP Morgan has somewhere in the realm of $800 billion of free lending capacity.  Now, JP Morgan probably wants to maintain a comfortable cushion of capital to ensure that the bank doesn’t need to worry about falling below regulatory capital requirements even in adverse circumstances, but we can be confident that if some entrepreneur walks in the door with an extremely strong business plan that just needs $1 million of funding, JP Morgan can write that entrepreneur a check today, simultaneously increasing both assets (the loan to the entrepreneur) and liabilities (the check) at the cost of an incremental decline in tier 1 capital.

So how do we reconcile this view of financial intermediation with what Woodford calls the “market-based financial system”?  The first thing to recognize is that the “market based financial system”, at least to the degree that the phrase is meant as a contrast with a “bank based financial system” is much smaller than it appears.  Although the bond market is a true non-bank financial market, in my view most asset backed securities markets, the commercial paper market and the repurchase agreement market – and the money market funds that invest in these assets – should actually be classed as sub-categories within the “bank based financial system”.  (I will not discuss mortgage backed security markets here, because these markets are complicated by the important role played by government-sponsored financial intermediaries like Fannie Mae, Freddie Mac and the Federal Housing Administration, and require detailed analysis that is far beyond the scope of this post.)

Let me illustrate how lending takes place in the asset backed securities market.  A bank originates an auto loan, just as JP Morgan originated the entrepreneur’s loan, by simultaneously creating an asset (the loan) and a liability (the check written to the dealer for the purchase of the car).  It packages this loan along with hundreds of others and sells it to an asset backed commercial paper conduit, replacing loans in the conduit that have matured.  The conduit raises the funds needed to buy the loans on the commercial paper market, so one might be inclined to say that this is market-based lending.  There is, however, a catch:  in order for the conduit to have an investment grade rating (which is required if it is to finance itself on the commercial paper market), a bank must provide the conduit with a liquidity backstop.  Because the conduit’s liabilities mature much earlier than the conduit’s assets, this liquidity backstop protects commercial paper buyers in the event that the conduit is unable to roll over its commercial paper issues.  Even though the bank provides a liquidity, not a credit, guarantee, as a practical matter the short maturities of commercial paper all but ensure that the conduit will experience a liquidity crisis months before its credit rating is downgraded so low that the bank is no longer obliged to honor the liquidity guarantee.  Thus, asset backed commercial paper (ABCP) should be viewed as bank lending with the ABCP conduit as an additional intermediary, that in theory can carry credit losses, but in practice serves to reduce the capital a bank must hold against the loans – since liquidity backstops have very small capital requirements.

The multi-tiered bank based financial system

Money market funds also play a role in this process, because they are important purchasers of asset backed commercial paper.  But the fraction of money market fund assets that do not carry a guarantee from either the federal government or the banking system is probably quite small.  According to the Investment Company Institute at the end of 2009, the funds in non-government money market funds were invested as follows:  14% government and agency issues, 39% certificates of deposit and bank notes, 27% commercial paper, 9% repurchase agreements. [1] Certificates of deposit are explicit bank liabilities.  The vast majority of commercial paper carries either an explicit (in the case of bank issued commercial paper) or implicit (in the case of a liquidity backstop) bank guarantee.  Data on repos is hard to find, but it seems highly unlikely that money market funds are deliberately choosing hedge funds and other risky firms as counterparties.  Most likely, a large majority of money market fund repos are repos with bank counterparties.

In short, it’s a mistake to look at lending by money market funds and assume that this lending can be classified as “market based” in contrast with “bank based” lending, because the vast majority of assets that money market funds choose to invest in are protected by a bank or government guarantee.  The role of money market funds is (i) to channel money into the banking system via financial commercial paper, certificates of deposit and repos and (ii) to reduce the capital banks are required to hold against their loans by removing loans from bank balance sheets via non-financial and asset backed commercial paper, protected by bank backstops.

In fact, I find it odd when people claim that the fact that banks are financing themselves via commercial paper, repos and CDs is evidence of the growth of “market based” lending.  To me this is just evidence that money market funds should be viewed as an adjunct to the banking system.  It seems to me that the only reason money market funds are more successful than banks at raising money from the public is that they have lower costs, because they are not required to hold any capital whatsoever to protect against the possibility that they cannot repay account holders.  I have addressed the fundamental illogic of the regulation of money market funds elsewhere.

In my view the correct way to characterize the US financial system is to describe it as a multi-tiered bank based financial system.  In the 1950s the US had a simpler financial system where the circular flow of funds went from banks to firms to households and back to banks.  Now-a-days that channel still exists, but in addition there are a number of multi-tiered channels, for example:  from banks to firms to households to money market funds to asset backed commercial paper conduits to banks.

One of the reasons this multi-tiered structure has developed is because it is an effective means of circumventing bank capital requirements and thereby increasing the leverage in the economy and the availability of credit.  Minimizing capital has the effect of increasing the short-run profitability of financial intermediaries at the expense of their long-run solvency.  When the financial industry’s competitive forces are not restrained either by regulation or via punitive costs of failure, the profitability of undercapitalized intermediaries can easily force older well-capitalized firms to find a way to participate in undercapitalized lending.  It is important for regulators to understand that one consequence of an aggressively competitive financial market is likely to be undercapitalization and instability – as the bad banks drive out the good.

Why don’t those who lend to financial intermediaries constrain leverage to levels consistent with financial stability?  I can see two main reasons.  In some cases it is for the same reason that depositors cannot be relied on to discipline banks in an economy without deposit insurance:  the lenders have neither the information nor the sophistication necessary to evaluate the behavior of the intermediary.  They know there are dangers to maintaining a deposit account, but want to take advantage of the convenience of deposits as circulating currency. Knowing that what they do not know can hurt them, these depositors are apt to react strongly to rumors with the result that bank runs are endemic.

I would say that this is a fair description of what happened with investors in money market funds:  they were informed of the risks and chose to take them because of the convenience value of the accounts.  However, as soon as there was a hint of danger these same investors sought to withdraw from the accounts.

On the other hand, another reason for investors not to constrain leverage to levels consistent with financial stability is that there are explicit and implicit guarantees from the government to prevent instability in the banking system.  In the ABCP market, money fund managers relied on bank liquidity backstops to give them time to exit the commercial paper and transfer any losses to the banking system.  Government guarantees ensure that the money fund managers don’t need to worry about the failure of the bank-guarantor rendering those liquidity backstops valueless just when they are most needed.  As money fund managers generally choose to invest in bank guaranteed assets, it seems very likely that the government’s commitment to the stability of the banking system is an important driving force behind these decisions. Even the rating agencies take the government’s commitment to the stability of the banking system into account explicitly when they determine that an asset is investment grade.  In such an environment, it seems a bit odd to expect investors to constrain leverage.

Once one realizes that bank guarantees – and through the banks, government guarantees – are essential to most of the assets in the “market-based” financial system, one is forced to recognize that this nomenclature is misleading.  This is why I propose that our current system should be called a multi-tiered bank based financial system.

Financial capital is an endogenous variable

We have found that, because macroeconomists choose not to carefully evaluate financial instruments and their relationships to each other, they have failed to correctly characterize the type of financial system we have – and have effectively assumed that it is “market based” despite abundant  factual evidence to the contrary in the form of bank guarantees.  There is, however, another serious error that derives from a failure to carefully evaluate the role of financial intermediaries in the money supply.  By ignoring the fact that financial institutions issue money, macroeconomists generate capital constraints that do not in fact exist.

Woodford claims:  “Probably a more important limit on credit supply derives from the limited capital of intermediaries — or, more fundamentally, the limited capital of the “natural buyers” of the debt of the ultimate borrowers — together with limits on the degree to which these natural buyers are able to leverage their positions.”

I believe that the fundamental misconception that deposits create loans rather than vice versa is leading Woodford astray here.  Once one accepts that financial intermediaries issue money, one realizes that there are no absolute constraints on the amount of capital in the system.  Just as the monetary authority can choose a helicopter drop of cash, so the financial intermediaries can create “capital”.  Think for a moment:  there is a borrower whose income for years was viewed as limiting him to a $200,000 mortgage, but in our competitive overleveraged financial system this borrower – and all his neighbors – is suddenly able to afford a $400,000 mortgage.  As one might expect house prices rise – for simplicity, let’s say that they double.  The seller of each house has twice as much “capital” as in the past (or even more if the house was not fully paid off).  Even though half of his capital was created by the fact that lending standards have collapsed, the seller still has the ability to invest the full amount as a “natural buyer” into whatever asset he chooses.

In short, a collapse in lending standards is inflationary, just like the government’s printing press.  It just so happens that our recent inflation showed up more in asset prices, than in goods (thanks most likely to the Chinese).  Thus the claim that financial intermediaries as a group are constrained by “the limited capital of the ‘natural buyers’” is incorrect, because it fails to take into account the fact that the amount of that capital is endogenous to the behavior of the intermediaries.  When credit is easy, “capital” will be relatively abundant and when credit is tight it will be relatively scarce.  This is part and parcel of an understanding that banks issue money.

For example, ABCP conduits and money market funds are intermediaries that hold no capital to protect investors in the event that assets go into default.  Because the loans that are funded by the interaction of money market funds and ABCP conduits go back into the system in the form of either money market deposits or bank deposits, there are very few constraints on the leverage these two financial intermediaries can jointly create.  The constraints enter in at the level of the underwriters:  In a world where nobody originates bad loans, the two intermediaries will probably facilitate very efficient allocation of credit.  However, in a world where lending is not sound, the fact that ABCP conduits buy bad assets and money market funds are indirect investors in these bad assets will mean that these two intermediaries are facilitating the growth of an overleveraged financial system.

Furthermore, the absence of capital means that the value of ABCP and money fund liabilities collapses much more easily than that of bank deposits, and these intermediaries are even more subject to runs than the uninsured banks of the 19th century.  It is far from clear that it makes sense for these financial intermediaries to exist in an environment where underwriting standards can break down – it is instead clear that if market forces had been allowed to operate both of these two intermediaries would have seen their market share plummet, and possibly even be extinguished entirely.

In short, I think that a careful analysis of the role of financial intermediaries in the macro economy will conclude that credit supply depends most importantly on the credibility of intermediaries, not on their capital.  As long as both money market funds and ABCP conduits were viewed as conservatively managed, safe investments, credit was abundant, but the moment that their management was doubted, credit evaporated.  In practice, it seems to be easy for the public to grant infinite leverage to intermediaries who are trusted with the issue of the circulating money supply, like banks and money funds – only to draw back in distrust when that easy access to leverage results in flawed lending policies.  Thus in order to understand what constrains credit supply, I think one must look towards (i) coordinated behavior by financial intermediaries who recognize that it is in their interests to avoid a collapse of credit, (ii) underwriting norms which depend on the penalties to intermediary failure and (iii) government regulation.

Update 7-1-10: I’ve been reading the comments on this post over at Interfluidity and wanted to mention that the idea that deposits come from loans is an idea that’s been around for a very long time.  I would source it back to Henry Thornton, but wouldn’t be surprised if earlier bankers turned theorists had also reached the same conclusion.  More recently it has been used as a tenet of Randall Wray’s endogenous money theory and of “Modern Monetary Theory”.


[1] For reference at the end of 2006 non-government money market funds were invested as follows:  3.5% government and agency, 20% CDs and bank notes, 39% commercial paper, 11% repurchase agreements and 21% corporate notes.  It is possible – though I do not in fact know what the norms were in this market – that some of the corporate notes were protected by bank liquidity guarantees.

The problem with negative real interest rates

The “Taylor rule” is representative of a large school of macroeconomic thought which takes the view that monetary policy-makers should set the policy rate as a function of the inflation rate and the unemployment rate (or an alternate measure of economic slack).  Those who have been schooled in this framework have difficulty understanding why anyone would recommend raising the policy rate when inflation is extremely low and the economy is weak.

Monetary policy is viewed as a “neutral” policy because it works through interest rates.  As long as the big banks pass changes in the policy rate on to borrowers, then all participants in the economy are able to benefit from any decreases in the policy rate and to face the costs of any increases.  Fiscal policy by contrast will always favor some beneficiaries of the government’s largess over others, and thus cannot be considered “neutral”.  It should be noted, however, that the neutrality of monetary policy depends on the competitive structure of the monetary transmission mechanism and if the banks that trade in the federal funds market, where policy rates are set, do not always pass on changes in the policy rate, monetary policy will act as a subsidy to the big banks.  In fact, it is my view that when the policy rate approaches the zero lower bound — and real interest rates are negative — the incentive for financial intermediaries to arbitrage their negative cost of funds distorts lending markets and the monetary transmission mechanism.

Models that support the use of a Taylor rule rarely specify the structure of the financial system explicitly.  And I think that the best way to understand them is that they express a very simple view of financial intermediation along these lines:  A vast number of the transactions that underlie economic activity are dependent on the extension of credit — many firms depend on short-term credit to finance working capital and on long-term finance to make large scale projects possible and most consumers are dependent on credit to finance large purchases like homes and automobiles and other consumer durables.  For this reason a decrease in the interest rate tends to reduce the cost of financing transactions and by reducing their cost increase the number of transactions — and it is of course these transactions that are used to calculate GDP.  The role of financial intermediaries in this process is to hold as deposits the income received by the consumers (who also staff the firms) and sales receipts of the firms and recycle those deposits into the loans — at an interest rate sensitive to the policy rate — that allow this economic activity to take place.  In short, the model is based on old-fashioned banks that (i) hold loans on their balance sheet — and because they carry long-term exposure to these loans underwrite them very carefully and (ii) set interest rates on the loans they offer in a competitive market.

There are many historical time periods in which this simple model of financial intermediation is accurate enough.  However, the drive towards profitability in the financial system has a tendency to generate a process of financialization in the economy.  A multi-tiered financial system is created in which banks lend to other financial intermediaries who in turn lend to other financial intermediaries or the real economy.  One of the most common consequences of this process is a reduction in the capital supporting the issue of loans.  In fact the profitability of the multi-tiered system is generally driven by the fact that more loans can be issued with a given capital base.  (It is precisely because there are norms for bank capital requirements that a new kind of intermediary has to be created.)

Some examples of this process are the bill-brokers of mid-19th century England, who are discussed in detail by Walter Bagehot, and whose role in the financial system was greatly reduced after the failure of Overend and Gurney in 1866.  In the early 20th c US, trust companies and investment trusts were important vehicles for financialization — until the reforms of the ’30s forced a return to “plain vanilla” banking.

Of course another wave of financialization has taken place over the past few decades and it is this multi-tiered financial structure that interferes with the simple operation of the Taylor Rule.  Modern banks don’t hold the loans they originate, but instead either sell them off — to for example an asset backed commercial paper conduit — or finance them on the repo market.
(i)  Because the loans are not carried on the balance sheet, but instead have their value realized at the date of origination, the immediate value of the loan becomes more important than the life-long value of the loan.  This is an environment which is primed to issue as many loans as possible.  Furthermore,
(ii) Each tier of the financial system is profiting from an interest rate spread and when nominal interest rates are below the rate of inflation, economic incentives push these intermediaries to borrow as much as possible and finance as many loans as possible.  While, in theory, these should all be “good” loans with little risk of credit loss, (i) ensures that the incentives are not necessarily aligned to guarantee the quality of the loans.

Another aspect of modern financialization is that financial intermediaries are able to use their borrowed funds to trade on secondary markets — an act that has traditionally been prohibited for banks.  Thus, the unlimited demand for funds created by negative real interest rates is easily channeled into purchases on secondary markets.  The natural consequence of this process is that negative real interest rates tend to drive up the value of risk assets like stocks and junk bonds.  Of course, to the degree that the demand for these assets is driven by the negative cost of borrowing, normalization of interest rate policy will necessarily have the effect of reducing demand for these assets and the value of risk assets.  Thus, negative real interest rates will tend to create increases in the prices of risk assets that are unsustainable over the long-run.

In short, in an environment where new financial intermediaries have been created that attenuate the relationship between deposits and loans, negative real interest rates create an unlimited demand for funds by the financial sector which can easily flow into loans of dubious quality and markets for risky assets as intermediaries gamble on their ability to successfully arbitrage the negative cost of funds.  This view implies that in an economy with a multi-tiered financial system the Taylor rule needs to be adjusted for the additional economic costs created by a policy rate that falls below the rate of inflation — and that as we near the zero bound the optimal policy rate will be significantly higher than that implied by the Taylor rule.  On the other hand these concerns will not necessarily apply to an environment where a law like Glass-Steagall (i) restricts the activities of financial intermediaries in a way that protects the profitability of simple commercial banks, while at the same time allowing for interbank competition, and (ii) precludes the flow of funds borrowed by intermediaries into secondary markets.

The Knightian uncertainty meme is back …

The Knightian uncertainty meme is back.

[During the asset backed commercial paper crisis of 2oo7] Knightian uncertainty took over, and pervasive flight to qualities plagued the financial system. Fear fed into more fear, and caused reluctance to engage in financial transactions, even among the prime financial institutions

In response to this let me just quote an old post of mine:

Apparently the reason for this abrupt outbreak of Knightian uncertainty is that bankers have suddenly realized that there is a difference between reality and their models. As long as the world behaves according to model, bankers want to claim that they are earning profits from managing “risk,” and as soon as their models fail, risk becomes uncertainty and necessitates a government bailout.

In short whenever you read that bankers can manage risk, but uncertainty requires government intervention, you should hear: “Privatize the profits and socialize the losses.”

The truth is that bankers always have to price assets in the face of Knightian uncertainty, they have just chosen to spend the last decade pretending that this was not their job.

In response to the view that the government should insure the financial system against systemic tail risk:

… the main failure during the crisis was not in the private sector’s ability to create triple‐A assets through complex financial engineering, but in the systemic vulnerability created by this process. It is important to preserve the good aspects of this process while finding a mechanism to relocate the systemic risk component generated by this asset creation activity away from the banks and into private investors (for small and medium size shocks) and the government (for tail events).

I will refer readers to an earlier post that distinguished between systemic liquidity risk and systemic credit risk.

Isn’t modern economic logic circular by design?

Economic analysis relies on models.  It has no choice but to rely on models for the same reason that useful maps ignore a lot of information that is valuable to someone — there’s just too much data in the real world to analyze in one sitting.

Models start with assumptions and then investigate whether the implications of those assumptions contradict the facts.  It is generally acknowledged that the value of a model is open to question when the facts contradict the implications of the model.  (However, it is also common to see a spirited defense of models that are not consistent with the facts.)  When the model is consistent with the facts, the modeller has a winner.  The modeller does not, however, have the winner.  The modeller has just presented us with one of many stories that can explain the facts.

Economics faces the same problems as the field of epidemiology (amongst others):  the model underlying the data analysis assumes causality and statistical methods can be used to show that the assumption of causality is not contradicted by the data.  In fields where controlled studies are impossible there is, however, little hope that statistics can be used to positively confirm that the assumption of causality is a good one.

Thus, economists are frequently found making arguments along the lines of:  Given my assumptions I reach these conclusions.  Since these conclusions are consistent with what we know about the world, my assumptions are good.

To which the logical response is:  Maybe you’re right and maybe you’re wrong; maybe your assumptions are good and maybe they’re bad.  You still haven’t justified to people who have doubts about your model that there is any reason to believe in it.

In short, until economists recognize that they need to defend the assumptions underlying their models at the level of pure theory, they will continue to be accused of circular logic.  In my view the problem with economics today is that only a small subset of economists see the need to defend the value of their arguments on a theoretical level.

Despite my view that most modern economists have yet to come to terms with the basic principles of logic, I actually think that Ted Gayer‘s defense of “market based mechanisms” is correct.  The reason that I think Gayer is correct is because I find Hayek‘s explanation of why market based mechanisms are better convincing — that is, because somebody took the time to convince me at the level of verbal argument and pure theory that market prices are directional signals that are essential to the ability of a successful economy to function.

Too many economists seem to take the view that:  Well, Hayek convinced me, so I feel free to take it as a given that Hayek is correct and expect my readers to do the same.  They should not be surprised if a huge number of people who have never read Hayek find this approach arrogant and illogical.  Any economist who is above using verbal argument to demonstrate why the pure theory underlying his or her approach is correct — even if this just means that the economist paraphrases Hayek —  deserves to be accused of circular logic.