Re-imagining Money and Banking

I’ve written a new paper motivated by my belief that the recent financial crisis was in no small part a failure of economic theory and therefore of economic thinking. In particular, there is a missing model of banking that was well understood a century ago, but is completely unfamiliar to modern scholars and practitioners. The goal of this paper is to introduce modern students of money and banking to the model of money that shaped the 19th century development of a financial infrastructure that both supported modern economic growth for more than 100 years and was passed down to us as our heritage before we in our hubris tore that infrastructure apart.

Another goal is to illustrate what I believe is a fundamental property of environments with (i) liquidity frictions and (ii) a large population with no public visibility but a discount factor greater than zero: in such an environment anyone with a notepad, some arithmetic skills, and some measure of public visibility can offer – and profit from – the account-keeping services that make incentive feasible a much better allocation than autarky for the general populace. Importantly collateral is completely unnecessary in a bank-based payments system.

This model has two key components. First, banks transform non-bank debt into monetary debt. Thus, the transformative function of banking is not principally a matter of maturity, but instead of the nature of the debt itself, that is, of its acceptability as a means of exchange. Second, monetary debt is money (contra Kocherlakota 1998). There is no hierarchy of moneys where some assets have more monetary characteristics than others. Instead there is only monetary debt and non-monetary debt. When we study this very simple model of money in an environment with liquidity frictions using the tools of mechanism design, we see that the economic function of the banking system is to underwrite a payments system based on unsecured debt and thereby to make intertemporal budget constraints enforceable or equivalently to make it possible for the non-banks in our economy to monetize the value of the weight that they place on the future in the form of a discount factor. Banking transforms an autarkic economy into one that flourishes because credit is abundantly available. In this model, constraints on the economy’s capacity to support debt are not determined by “deposits” or by “collateral”, but instead by the incentive constraints associated with banking.

In this environment, banking provides the extraordinary liquidity that is only possible when the payments system is based on unsecured debt. Underlying this form of liquidity is the banks’ profound understanding of the incentive structures faced by non-banks, as it is this understanding that makes it possible for banks to structure the system of monetary debt so that it is to all intents and purposes default-free. (This is actually a fairly accurate description of 19th century British banking. The only people who lost money were the bank owners who guaranteed the payments system. See Sissoko 2014.) Although this concept of price stable liquidity is unfamiliar to many modern scholars, Bengt Holmstrom (2015) has given it a name: money market liquidity.[1] In such a system the distinctions between funding liquidity and market liquidity collapse, because the whole point of the banking system is to ensure that default occurs with negligible probability. Thus, the term money market liquidity references the idea that in money markets, the process by which assets are originated must be close to faultless or instability will be the result, because the relationship between money – when it takes the form of monetary debt – and prices is not inherently stable (cf. Smith 1776, Sargent & Wallace 1982).

This paper employs the tools of New Monetarism, mechanism design, and more particularly the model of Gu, Mattesini, Monnet, and Wright (2013) to explain the extraordinary economic importance of the simplest and most ancient function of a bank: in this paper banks are account-keepers, whose services support a payment system based on unsecured credit. Unsecured credit is incentive feasible, because banks provide account-keeping services and can use the threat of withdrawing access to account-keeping services to make the non-bank budget constraint enforceable.

The basic elements of the argument are this: an environment with anonymity, liquidity frictions and somewhat patient agents is an environment that begs for an innovation that both remedies the problem of anonymity and realizes the value of the unsecured credit that the patience of the agents in the economy supports. I argue that the standard way in which economies from ancient Rome to medieval Europe to modern America address this problem is by introducing banking – or fee-based account-keepers – in order to alleviate the problem of anonymity that prevents agents from realizing the value inherent in the weight they place on the future. I demonstrate that in this environment, the introduction of a bank improves welfare. The improvement in welfare can be dramatic when the discount factor is not close to zero.

This paper uses the environment of Gu, Mattesini, Monnet, and Wright (2013) but is distinguished from that model, because here the focus is on a different aspect of banking. We study how the account-keeping function of banks serves to support unsecured credit, whereas GMMW studies how the deposit-taking function of banks is able to support fully collateralized credit.

The model of banking in this paper has implications that are very different from much of the existing literature on banking. This literature typically assumes the anonymity of agents and then argues – contrary to real-world experience – that unsecured non-bank credit is unimaginable (see, e.g., Gorton & Ordonez 2014, Monnet & Sanches 2015). In other words, the existing literature takes the position that in the presence of anonymity, no paid account-keeper will arise who will make it possible for agents in the economy to realize the value of unsecured credit that their discount factor supports. In the absence of unsecured credit, lending is generally constrained as much by the available collateral or deposits, as by incentive constraints themselves. This paper argues that standard assumptions such as loans must equal deposits (see, e.g. Berentsen, Camera & Waller 2007) or debt must be supported by collateral (see e.g. Gu, Mattesini, Monnet, and Wright (2013), Gorton & Ordonez 2014) are properly viewed as ad hoc assumptions that should be justified by some explanation for why banking has not arisen and made unsecured credit available to anonymous agents.

[1] While Holmstrom (2015) and this paper agree on the principle that money market liquidity is characterized by price stability, the mechanism by which that price stability is achieved is very different in the two papers: for Holmstrom it is the opacity of collateral that makes price stability possible.

Collateral and Monetary Policy: A Puzzle

A stylized fact about post-crisis economies is that asset markets have become segmented with “safe assets” trading differently from assets more generally. I have argued elsewhere that the collateralization of financial sector liabilities has played an important role in this segmentation of markets.

I believe that this creates a puzzle for the implementation of monetary policy that provides at least a partial explanation for why we are stuck at the zero lower bound. Consider the consequences of an increase in the policy rate by 25 bps. This has the effect of lowering the price of ultra-short-term Treasury debt, and particularly when combined with a general policy of raising the policy rate over a period of months or years this policy should have the effect of lowering the price of longer term Treasuries as well (due to the fact that long-term yields can be arbitraged by rolling over short-term debt).

A decline in the price of long-term Treasuries will have the effect of reducing the dollar value of the stock of outstanding Treasuries (as long as the Treasury does not have a policy of responding to the price effects of monetary policy by issuing more Treasuries). But now consider what happens in the –segmented — market for Treasury debt. Assuming that demand for Treasuries is downward sloping, then the fact that contractionary monetary policy tends to shrink the stock of Treasuries itself puts upward pressure on the price of Treasuries that, particularly when demand for Treasuries is inelastic, will tend to offset and may even entirely counteract the tendency for the yield on long-term Treasuries to rise. (Presumably in a world where markets aren’t segmented demand for Treasuries is fairly elastic and shifts into other financial assets quash this effect.)

In short, a world where safe assets trade in segmented markets may be one where implementing monetary policy using the interest rate as a policy tool is particularly difficult. Can short-term and long-term safe assets become segmented markets as well? Given arbitrage, it’s hard to imagine how this is possible.

These thoughts are, of course, motivated by the behavior of Treasury yields following the Federal Reserves 25 bp rate hike in December 2015.fredgraph


Finance and Economics: A Research Agenda

My research agenda employs deconstructive method to motivate a reconsideration of the meaning of neoclassical economics. Thus, an economic theory paper introduces a liquidity friction into a competitive model to study how standard models are constructed on the assumption of perfect intermediation (Sissoko 2007), an economic history paper demonstrates that in fact the markets of industrializing Britain relied on a carefully calibrated banking system that successfully stabilized money growth (Sissoko 2016a), an economic theory paper uses new monetarist methods to model banking and how it stabilizes the relationship between unsecured debt and the money supply (Sissoko 2016b), and another paper analyzes modern finance and explains how the growth of market-based lending has disrupted market liquidity by circumventing the stabilizing force of banks (Sissoko 2016c). Together these papers invert the mainstream view by arguing that starting in late 18th c Britain the banking system effectively stabilized the money supply, allowing it to be treated as a stable background condition: this made modern capitalism possible and neoclassical economics itself imaginable. This post explains the “big picture” of the role played by innovations in banking on European industrialization, and this chapter of my dissertation gives even more detail.

The 19th c. bank “bailout” that never happened

I’ve just read Eugene White’s Bank Underground post on the Baring liquidation in 1890. He is notable in getting the facts of what he calls the “rescue” mostly right. He accurately portrays the “good bank-bad bank” structure and the fact that the partners who owned the original bank bore the losses of the failure. What he doesn’t explain clearly is the degree to which the central bank demanded insurance from the private sector banks before agreeing to extend a credit line that would allow the liquidation of the bad bank to take place slowly.

These facts matter, because a good central banker has to make sure that the incentives faced by those in the financial community are properly aligned. In the case of Barings macroeconomic incentives were aligned by making it clear to the private banks that when a SIFI fails, the private banking sector will be forced to bear the losses of that failure. This brings every bank on board to the agenda of making sure the financial system is safely structured.

In the 19th c. the Bank of England understood that few things could be more destabilizing to the financial system than the expectation that the government or the central bank was willing to bear the losses of a SIFI failure. Thus, the Bank of England protected the financial system from the liquidity consequences of a fire sale due to the SIFI, but was very careful not to take on more than a small fraction (less than 6%) of the credit losses that would be created by the SIFI failure.

This is the comment I posted:

While this is one of the better discussions of the 1890 Barings liquidation, for some reason modern economic historians have a lot of difficulty acknowledging the degree to which moral hazard concerns drove central bank conduct in the 19th c. White writes:

The Barings rescue or “lifeboat” was announced on Saturday November 15, 1890. The Bank of England provided an advance of £7.5 million to Barings to discharge their liabilities. A four-year syndicate of banks would ratably share any loss from Barings’ liquidation. The guarantee fund of £17.1 million included all institutions, and some of the largest shares were assigned to banks whose inattentive lending had permitted Barings to swell its portfolio.

Clapham (cited by White), however makes it clear that the way the Bank of England drummed up support for the guarantee fund was by making a very credible threat to let Barings fail. Far from what is implied by the statement “The Bank of England provided an advance of £7.5 million to Barings to discharge their liabilities”, the Bank of England point blank refused to provide such an advance until and unless the guarantee fund was funded by private sector banks to protect the central bank from losses, Clapham p. 332-33.

In short, treating the £7.5 million (which is actually the maximum liability supported by the guarantee fund over a period of four years, Clapham p. 336) as a Bank of England advance may be technically correct because of the legal structure of the guarantee fund (which was managed by the Bank), but gets the economics of the situation dead wrong.

19th century and early 20th century British growth could only take place in an environment where central bankers in London were obsessed with the twin problems of aligning incentives and controlling moral hazard. Historians who pretend that anything else was the case are fostering very dangerous behavior in our current economic climate.

Note: Updated to make the last paragraph specific to Britain.

Lenders of Last Resort have duties in normal times too

I have a paper forthcoming in the Financial History Review that studies the role played by the Bank of England in the London money market at the turn of the 20th century. The Bank of England in this period is, of course, the archetype of a lender of last resort, so its activities shed light on what precisely it is that a lender of last resort does.

The most important implication of my study is that the standard understanding of what a lender of last resort does gets the Bank’s role precisely backwards. It is often claimed that the way that a lender of last resort functions is to make assets safe by standing ready to lend against them.

My study of the Bank of England makes it clear, however, that the duties of a lender of last resort go far beyond simply lending against assets to make them safe. What the Bank of England was doing was monitoring the whole of the money market, including the balance sheets of the principal banks that guaranteed the value of money market assets, to ensure that the assets that the Bank was engaged to support were of such high quality that it would be a good business decision for the Bank to support them.

In short, a lender of last resort does not just function in a crisis. A lender of last resort plays a crucial role in normal times of ensuring that the quality of assets that are eligible for last resort lending have an extremely low risk of default. This function of the central bank was known as “qualitative control” (although of course quantitative measures were used to predict when quality was in decline).

Overall, if we take the Bank of England as our model of a lender of last resort, then we must recognize that that the duty of such a lender is not just to lend, but also to constantly monitor the money market and limit the assets that trade on the money market to those that are of such high quality that when they are brought to the central bank in a crisis, it will be a good business decision for the bank to support them.

A central bank that fails to exercise this kind of control over the money market, can expect in a crisis to be forced, as the Fed was in 2008, to support the value of all kinds of assets that it does not have the capacity to value itself.

Note: the forthcoming paper is a new and much improved version of this paper.

Do “net financial assets” matter?

I’ve just read Steve Waldman’s post on “net financial assets” and am connecting it up with Michael Pettis’ excellent discussion. (See also Cullen Roche’s comment on the issue.)

Steve discusses the decomposition of financial positions on which MMT is based. He points out that the term “net financial assets” is used for the “private sector domestic financial position” which refers exclusively to the aggregate netted financial position of both households and firms and explicitly excludes “real” savings such as any housing stock that is fully paid up. By definition, if the “private sector domestic financial position” is positive, then it must be the case that on net the private sector holds claims on either the government or on foreign entities. Of course, the value of such claims depends entirely on the credibility of the underlying promises — this is the essential characteristic that distinguishes a claim to a financial asset from a claim to a real asset.

For Steve, there is a tradeoff between holding financial claims and holding real claims, and a principal reason for holding financial claims is to offset the risk of the real claims. Thus, Steve goes on to claim that to the degree that such a positive private sector financial position is due to claims on government, the government is using its credibility to provide a kind of insurance against real economy risk.

This is where I think Steve both gets what happened in 2008 right, and gets the big picture of the relationship between the financial and the real, and between the private sector and the public sector wrong. Steve is completely correct that in 2008 the issue of public sector liabilities played a huge insurance and stabilization role.  But Steve extends his argument to the claim that: “The domestic private sector simply cannot produce assets that provide insurance against systematic risks of the domestic economy without the help of the state.”

The key point I want to make in this post is this: the financial and the real are so interdependent that they cannot actually be divorced. The same is true of the private and the public sectors. Financial activity and real activity, public sector activity and private sector activity are all just windows into a single, highly-integrated economy. Thus, I would argue that it is equally correct to state that: “The domestic public sector simply cannot produce assets that provide insurance against systematic risks of the domestic economy without the help of the private sector.”

That financial activity and real activity are two sides of the same coin is most obvious when one considers that the credibility of private sector financial liabilities depends fundamentally on the performance of the real economy. But it is equally true that the credibility of public sector liabilities (when measured in real terms) depends fundamentally on the robustness of the real economy as well. Those countries that have very highly rated debt did not achieve this status ex nihilo, but because of the historical performance of their economies and the robustness of their private sectors.

Thus, it is entirely correct that the public sector can temporarily step in to provide insurance for the private sector when it is struggling, but the view that it is the public sector that is the primary provider of insurance fails to capture the genuine interdependence that lies at the heart of a modern economy.

Indeed, Steve recognizes the danger of framing the financial and the real and the public and the private in this way in his last paragraph, where he acknowledges that this publicly-issued insurance is in fact provided in real terms at the expense of a segment of the private sector — the segment that does not hold the claims on government.

Michael Pettis on Creating Money out of Thin Air

Now let’s turn to Michael Pettis (whom I’ve never met, so I’ll call him by his last name). Pettis has long stood out as an economist with a uniquely strong understanding of the relationship between the financial and the real. He argues that “When banks or governments create demand, either by creating bank loans, or by deficit spending, they are always doing one or some combination of two things, as I will show. In some easily specified cases they are simply transferring demand from one sector of the economy to themselves. In other, equally easily specified, cases they are creating demand for goods and services by simultaneously creating the production of those goods and services. They never simply create demand out of thin air, as many analysts seem to think, because doing so would violate the basic accounting identity that equates total savings in a closed system with total investment.”

His two cases are a full employment economy (without growth) and an economy with an output gap. He argues that it is only in the latter case that the funding provided by banks (or government) can have an effect on output. In a comment to Pettis’ post I observed that his first case fails to take into account Schumpeter’s theory of growth. An economy is at full employment only for a given technology. Once there is a technical innovation, the full employment level of output will increase. Schumpeter’s theory was that the role of banking in the economy was to fund such innovation. Thus, there is a third case in which bank finance in a full employment economy does not just transfer resources to a different activity, but transfers them to an innovative activity that fundamentally alters the full employment level of output. Thus, it is not only when the economy is performing below potential that bank funding can create the production that makes savings equal to investment. When banks fund fundamental technological innovation, it is “as if” the original economy were functioning below potential (which of course if we hold technology constant at the higher level, was in fact the case — but this deprives the concept of “potential GDP” of its meaning entirely.)

Schumpeter was well aware that the same bank funding mechanisms that finance fundamental technological innovation, also finance technological failures and a vast amount of other business activity. Indeed, he argued that even though the banking system was needed to finance innovation and growth, the consequences of the decision making process by which banks performed this role included both business cycles and — when banking system performed badly — depressions.

In short, there is very good reason to believe that even in a “full-employment” economy when banks create debt, some fraction of that process creates additional demand. The problem is that the fraction in question depends entirely on the institutional structure of the banking system and its ability to direct financing into genuine innovation. It’s far from clear that this fraction will exhibit any stability over time.

How Did We Get Here: The Fault Lies in Our Models

So why do economists fall into the trap of treating the financial and the real as separable phenomena? Why do macroeconomists of all persuasion look for solutions in the so-called public sector?

The answer to the first question is almost certainly the heavy reliance of the economics profession on “market-clearing” based models. In models with market-clearing everybody buys and sells at the same time and liquidity frictions are eliminated by assumption. Of course, one of the most important economic roles played by financial assets is to address the problem of liquidity frictions. As a result, economists are generally trained to be blind to the connections between the financial and the real. People like Michael Pettis and proponents of MMT are trying to remove the blindfold. They are, however, attempting to do so without the benefit of formal models of liquidity frictions. This is a mistake, because the economics profession now has models of liquidity frictions. The future lies in the marriage of Schumpeter and Minsky’s intuition with New Monetarist models.

The answer to the second question is that we have a whole generation of macroeconomic policy-makers who think that the principal macroeconomic economic debate lies between Keynesians and Monetarists, when in fact both of these schools assume that the government is the insurer of last resort. The only distinction between these schools is whether the insurance is provided by fiscal or by monetary means. (To understand why our economies are struggling right now one need only understand how the assumption that the government is the fundamental source of liquidity has completely undermined the quality of our financial regulation.)

The concept of liquidity as a fundamentally private sector phenomenon that both drives the process of growth and periodically requires a little support from the government (e.g. giving the private sector time to weather a financial panic without the government actually bearing a penny of the losses) has been entirely lost. Only the future can tell us the price of this intellectual amnesia.

On the “Low for Long” Report

Given that my preferred explanation for low real interest rates are the changes that have been wrought within the developed world’s financial markets — and in particular the growth in collateralized inter-bank lending, I read with interest the newly released report, Low for Long? Causes and Consequences of Persistently Low Interest Rates by Sir Charles Bean, Christian Broda, Takatoshi Ito, and Randall Kroszner.

The authors establish the basic facts:

the world long-term real risk-free rate has been drifting down remorselessly from around 4% in the late 1990s to just below zero today.


The very low level of long-term risk-free real interest rates in the advanced economies is historically most unusual. Real rates have rarely been so low; when they have been, it has almost invariably been during or after a war, when there was a degree of financial repression and/or inflation was elevated. The present configuration of low real rates with low inflation appears to be unprecedented.

In Chapter 2 the authors discuss the various explanations for why real interest rates have declined “remorselessly” from 1998 or 1999 on. While the authors do discuss shifts in preferences in favor of “safe assets,” they do not even mention what I would consider the most important explanation for such a demand shift: the increasing collateralization of financial exposures on the part of our biggest financial institutions. (As was noted in my previous post, data on this is available from the ISDA.) Indeed, the only source of an increase in demand for safe assets that the report cites which is consistent with the timing of the drop in rates is the increase in emerging market demand for foreign exchange buffers subsequent to the Asian financial crisis. All of the other explanations in this subsection refer to sources of increased demand subsequent to the crisis (post-crisis recognizing of the extent of possible bad outcomes, post-crisis regulatory requirements for banks to hold larger buffers of safe assets, and the demand for safe assets created by central bank quantitative easing policies.)

What is missing from the report is this: the latter stage of the Asian financial crisis coincided with the LTCM failure. The LTCM failure led to a significant increase in the collateralization of financial sector exposures. Collateralization ramped up continuously over the early years of the current millennium as laws supporting collateralization regimes were adopted in the US, the UK, and Europe. It is remarkable that this potential explanation of prolonged low rates on “safe assets” which has the same timing as the emerging markets savings glut explanation and therefore meets one of the most important criteria considered by the report has been entirely omitted from it.

This lapse strikes me as evidence that the financial sector is invisible to modern macroeconomists. I, of course, can’t help wondering whether this blindness is generated by the models they work with. In my view, we need to take a long and hard look at modern finance and how it has changed the way the real economy operates.