Writing in 1967, before he had yet formulated his famous Financial Instability Hypothesis, the American monetary economist Hyman Minsky identified the starting point for his analysis. “Capitalism is essentially a financial system, and the peculiar behavioral attributes of a capitalist economy center around the impact of finance on system behavior.”1 From this point of view, the key institutions of modern capitalism are its financial institutions, which make a business out of managing the daily inflow and outflow of cash on their balance sheets. And the quintessential financial institutions are banks, whose daily cash inflows and outflows are the mechanism of the modern payments system.

Everyone else—households, businesses, governments, even entire nations—is also a financial institution since, in addition to whatever else they do, they must attend to the consequences of their activities for their own daily cash flow. Indeed, this daily cash flow, in and out, is the crucial interface where each of us connects with the larger system. This interface provides the cash that makes it possible for us to pursue today dreams for the future that would otherwise be impossible; but it does so at the cost of committing us to make future payments that can, if our dreams do not work out, constrain our independence more or less severely. The seductive allure of present credit and the crushing burden of future debt are two faces of the same creature.

The Inherent Instability of Credit

The two faces of credit show themselves not only at the level of each individual, but also at the level of the system as a whole because one person’s cash inflow is another person’s cash outflow. If the allure of credit induces one person to increase spending, the immediate result is income somewhere else in the system, which income is then available for additional spending. Similarly, if the burden of debt induces one person to decrease spending, the immediate result is reduced income somewhere else in the system, and thus possibly also reduced spending. This interaction of balance sheets is the source of what the British monetary economist Ralph Hawtrey called the inherent instability of credit.2 In his view, the main job of the central bank is to prevent a credit-fueled bubble from ever getting started, in order to avoid the collapse that inevitably follows.

But, from another point of view, the inherent instability of credit is not entirely a bad thing. On the way up, real things get built, new technologies get implemented, and productive capacity expands. The Austrian economist Joseph Schumpeter always insisted that credit is critical for the process of “creative destruction” that is the source of capitalism’s dynamism, because it provides the crucial mechanism that allows the new to bid resources away from the old. Instability is, from this point of view, inseparable from growth, and a central bank that intervenes to control instability runs the risk of killing off growth by stifling the new on the way up and coddling the old on the way down.3

In any concrete case, the question therefore arises: are we looking at a Hawtreyan speculative bubble that we want to rein in, or at Schumpeterian dynamic growth that we want to let run? One reason this question is hard to answer is that a credit-fueled boom typically involves a bit of both. That is why we seem always to be tempted to draw a distinction between speculative and productive credit, and to look for ways to channel credit preferentially to the latter. But in practice the distinction is often difficult to draw and, even more problematic, discrimination in credit allocation is often impossible to implement. In this latter regard, the institutional structure of finance, including the regulatory structure, is crucial. If potential borrowers and lenders can find one another and do business outside the reach of the authorities, then it will be impossible to allocate credit preferentially to socially desirable uses, even assuming they could be identified and agreed on. (In such a situation, even control of aggregate credit can be quite difficult.)

In the last analysis, the only dependable source of leverage over the system as a whole is the role of the central bank as a banker’s bank. If banks are the quintessential financial institution because of their management of the retail payments system, then the central bank is the quintessential bank because of its management of the payments system that banks themselves use. When one bank makes a payment to another, the mechanism involves changing entries on the balance sheet of the central bank; there is a debit to the account of the bank paying and a credit to the account of the bank being paid. Here, in the requirement to settle net payments every day on the books of the central bank, we find the location of the ultimate discipline for the entire system.

Hyman Minsky called this requirement the “survival constraint”—cash inflows must be sufficient to meet cash outflows— and we all face such a constraint. For banks, the survival constraint takes the concrete form of a “reserve constraint” because banks settle net payments using their reserve accounts at the central bank. The leverage that the central bank enjoys over the larger system arises ultimately from the fact that a bank that does not have sufficient funds to make a payment must borrow from the central bank. In such a circumstance, the central bank must lend or else risk a breakdown of the payments system, but the lending does not have to be cheap or easy. It is the central bank’s control over the price and availability of funds at this moment of necessity that is the source of its control over the system more generally.

Opportunities for such control arise naturally from time to time, simply because of fluctuations in the pattern of payments, but the central bank can also create such opportunities as the need arises. Just so, when the central bank “tightens money” by selling Treasury bills, the consequence is that the banking system as a whole has to make payments to the central bank, which amounts to tightening the survival constraint that all bankers face. Alternatively, when the central bank “loosens money” by buying Treasury bills, the consequence is that the banking system as a whole receives payments from the central bank, thus relaxing the survival constraint. The effects of these central bank interventions show up in the short-term rate of interest that banks pay as the cost of putting off to the future a payment that is due today. Historically, the art of central banking was all about the choice of whether to raise or lower that cost.

The central bank’s ability to influence the degree of discipline or elasticity faced by banks at the daily clearing provides some control over the credit system as a whole, but that control is by no means absolute. Private credit elasticity is always a substitute for public credit elasticity. In its attempt to impose discipline, sometimes the most the central bank can do is to force banks to find and use alternative private credit channels. Similarly, in its attempt to impose elasticity, sometimes the most a central bank can do is to offer its own public credit as an alternative to collapsing private credit.

That’s why Hawtrey referred to the “art” of central banking, rather than the “science” or the “engineering.”4 The central bank can use its balance sheet to impose a bit more discipline when the private market is too undisciplined, and it can use its balance sheet to offer a bit more elasticity when the private market is imposing excessive discipline. But it is only one bank and ultimately small relative to the system it engages, especially so in the modern globalized financial system in which private credit markets are all connected into an integrated whole. Because the central bank is not all-powerful, it is especially important that it choose its policy intervention carefully, with a full appreciation of the origins of the instability that it is trying to counter.

According to Hawtrey, the inherent instability of credit has its origin in the way that credit-financed spending by some creates income for others, not only directly but also indirectly by pushing up the price of the good being purchased, thus producing an upward revaluation of existing inventories of the good. The capital gain for holders of inventories tends to stimulate additional spending, in part to buy ahead of rising demand in order to earn additional profit from rising prices in the future. Because revaluation of existing inventories tends to improve creditworthiness, this additional spending is easy to finance, even easier than the initial spending. The feedback loop of rising asset prices and credit expansion is the source of the inherent instability of credit emphasized by Hawtrey.

The price-credit feedback mechanism is also the reason that credit-fueled bubbles are so difficult to control, because it means that central bank interest rate policy can sometimes have very little traction. The question for the speculator is only whether the rate of appreciation of the underlying asset is greater than the rate of interest, and that is a condition often quite easily satisfied. If house prices are appreciating at 15 percent a year, it takes an interest rate of greater than 15 percent to stifle the bubble. Even supposing that the central bank is able to impose such a high interest rate, 15 percent would stifle a lot of other things as well. Conclusion: if you don’t catch the bubble early, it may be impossible to do anything using interest rate policy.

Meanwhile, the larger the bubble grows, the greater the distortion in the allocation of credit and in the allocation of real resources commanded by that credit. Not only does a bubble prospect of 15 percent attract new credit disproportionately, but also it bids up the price of credit across the board. Borrowers and lenders find one another at a rising market rate of interest, and the central bank must raise its policy rate merely to keep up. Eventually, and long before interest rates reach 15 percent, the effects of higher market interest rates are felt on nonbubble balance sheets throughout the economy, and it is these effects that bring the bubble to an end.

The way it works is this. Higher interest rates mean greater cash outflows for debtors, and eventually the most vulnerable among them find their cash outflows exceeding their cash inflows. If you are one of those vulnerable debtors, Minsky’s survival constraint begins to bind for you. Logically there are only three ways out. First, you can spend down any cash balances you may have, but these balances are finite and quickly exhausted. Second, you can borrow to cover the shortfall, but credit lines are also finite, and even possibly contracting in the face of declining creditworthiness. Third, you can sell some of your earning assets, for whatever price they will fetch on the market. Typically these three ways out are used sequentially, as debtors hold on for as long as they can, hoping that some other balance sheet in the system will prove to be the weakest link. The important point is that sooner or later asset prices come under pressure, not just the prices that were rising at 15 percent but all asset prices, and especially the price of the assets held by the most vulnerable debtors, who are forced to liquidate first.

When that happens, liquidity problems (the survival constraint) become solvency problems, and especially so for highly leveraged financial institutions. Even if they are not forced to sell assets in order to make promised payments, they may be forced to write down the valuation of those assets to reflect current market prices. For highly leveraged institutions, with financial liabilities many times larger than their capital base, it doesn’t take much of a write-down to produce technical insolvency. And even before insolvency, asset write-downs can quickly generate serious liquidity problems as credit lines shrink to fit reduced collateral valuations. Liquidity and solvency problems thus reinforce one another on the way down, just as credit expansion and asset valuation do on the way up. This is the downside of the inherent instability of credit.

On the way up, as has been emphasized, the central bank tends not to have much traction, since borrowers and lenders share an interest in avoiding central bank discipline. On the way down a similar mutual interest, now in avoiding market discipline, brings both borrowers and lenders back to the central bank as the last available source of credit elasticity. “Lender of last resort” intervention involves the central bank extending credit when no one else will (or can); in effect, the central bank relaxes the survival constraint by providing current cash inflow to allow borrowers to delay the day of reckoning. Used wisely, such intervention can control the downturn and prevent it from turning into a rout. Used unwisely, such intervention can foster further continuation of unhealthy bubble conditions. In a crisis, as in normal times, the art of central banking is all about walking the fine line between providing too much discipline versus too much elasticity.

The Old Lombard Street

The impact and effectiveness of central bank control both depend crucially on the institutional organization of the banking system, and on its articulation with the financial system more generally. Walter Bagehot’s Lombard Street explored these questions in the context of the London money market of his day, a set of institutional arrangements different in important respects from modern arrangements, but nonetheless a good starting point because the conclusions that Bagehot drew continue to shape the way we think today. The Bagehot principle that guided central bankers in the current crisis has its origin in that nineteenthcentury book.

Today we summarize the Bagehot principle as “lend freely but at a high rate.” Here are Bagehot’s own words (1906 [1873], 197): “The end is to stay the panic. And for this purpose there are two rules:—First. That these loans should only be made at a very high rate of interest.  .  .  . Secondly. That at this rate these advances should be made on all good banking securities, and as largely as the public ask for them.” Why did Bagehot think this was wise policy for his world, and is it still wise policy for our own very different modern world?

Bagehot’s world was based on a short-term commercial credit instrument known as the bill of exchange. Firms issued bills in order to buy inputs for their own production processes, and they accepted bills as payment for their own outputs. The bill of exchange was a promise to pay at a specific future date, perhaps in ninety days. For a fee, banks would “accept” bills, which meant guaranteeing payment. For another fee, banks would “discount” bills, which meant buying them for less than face value, the difference amounting to a rate of interest to be earned over the term to maturity. As payment for the bills, banks would offer either currency or a deposit account credit. Either way, the proceeds of the discount were most typically not held as idle balances but rather spent in payment of other maturing bills. In this way, the discount mechanism was crucial for British firms’ management of their daily cash flow, in and out.

Ideally, over the ninety days between issue and maturity, the firm that issued the bill would use the inputs so acquired to produce output for sale, and then use the sale proceeds to pay the bill as it came due. Timely repayment thus depended on timely sale of the production financed by the bill. Assuming timely repayment, the banking business was all about managing one’s portfolio of bills in order to match up the timing of cash inflows (from maturing bills) with the timing of cash outflows (for new discounts). If ever a firm failed to pay, however, then the accepting bank would experience a cash shortfall.

In this system, banks managed their own daily cash flow by managing the discount rate they quoted to their customers. If requests for discount were depleting one’s cash reserve, one had merely to raise one’s discount rate and the business would go elsewhere; if maturing bills were swelling one’s cash reserve, one simply lowered the discount rate to attract additional interest-paying business. In this way, the market rate of interest fluctuated according to supply and demand. The rate of interest was high when requests for new discount were running ahead of repayments, and low when the balance went the other way.

It was in this institutional context that the Bank of England developed the principles of central bank management that laid the foundations for modern monetary theory. At first, so Bagehot relates, the Bank thought of itself as simply one among other banks, responsible to its shareholders for the profitability of its operations, and with no larger responsibility for the system as a whole. In accordance with this way of thinking, the Bank moved its discount rate in line with the market in order to attract its rightful share of the discount business.

The experience of periodic financial crises, however, eventually taught the lesson that the Bank was not like other banks insofar as it was the central repository of cash reserves for the entire system. In times of general crisis, all banks looked to the Bank of England for help, and in order to prepare for that day the Bank had to safeguard its own reserve. That meant keeping its own discount rate ordinarily somewhat higher than the market rate, even at the cost of sacrificing some discount business and thus shareholder profit.

In this context, the Bagehot principle can be understood as the distillation of hard-won practical wisdom about how to deal with a crisis when one comes. The proximate origin of the crisis could be many things, but from the point of view of the Bank it always took the form of a large, often sudden, demand for cash. Any hiccup in current sales would mean that maturing bills could not be paid by their issuer. As a consequence, the accepting bank would be called on to make good from its own resources, which involved drawing down reserves held at the Bank of England and then, should that prove insufficient, borrowing more.

If the Bank of England failed to lend in such a circumstance, the needy bank would be unable to meet its commitments and those who had been expecting payment from that bank would similarly find themselves unable to meet their own commitments, and so on and so on as the cascade of nonpayment spread throughout the economy. The Bagehot principle was designed to stop the potential cascade by providing completely elastic lending to needy banks against any security that would be acceptable in normal times. But it was also designed to provide discipline by charging a high rate of interest. Only those who really needed the cash would borrow at the high rate, and the high rate would also provide incentive to repay the loan as soon as possible.

The problem with elastic lending in time of crisis was that it tended to drain the note reserves of the Bank of England. Under the provisions of Peel’s Act of 1844, the note issue was fixed, and any additional notes had to be backed 100 percent by additional gold reserves. In normal times, the Bank held a significant fraction of the note issue as reserve against deposits in the Banking Department, and it was these deposits that served as reserves for the banking system at large. During a crisis, the demand for cash was met both by paying out cash reserves (notes) and by expanding the supply of cash substitutes (deposits). When the crisis was over, the emergency loans would be repaid, the emergency supply of cash substitutes would be extinguished, and the Bank’s cash reserve would be built up again. That is how it was supposed to work, and how in fact it did work, so long as the crisis remained within the confines of Britain itself.

The policy of elastic lending ran into trouble, however, whenever the crisis assumed international dimensions, which more often than not it did, given the centrality of the pound sterling in the world trading system. The same bills of exchange apparatus that merchants used to finance domestic production was used also to finance foreign trade, trade not only between British merchants and their foreign counterparties but also between different foreign parties themselves. No matter where you were in the world, if you wanted to import goods, you were likely to pay by issuing a bill of exchange payable at some London bank and your counterparty was likely to present that bill of exchange for discount prior to maturity in order to raise cash to meet his own payment obligations.

The problem was that foreigners did not consider either notes or deposits to be acceptable means of payment; they wanted gold. (Mechanically, payment would be demanded in notes, and those notes would be presented to the Issue Department at the Bank of England for payment in gold.) The effect of a foreign demand for cash was thus to reduce the supply of currency in Britain and also, more important, to drain the Bank’s holding of gold, which served as reserve for the nation as a whole.

Not only firms and banks but also nations have to look after their daily balance of cash inflows and outflows, and for nations on the gold standard that meant the daily balance of gold flows. For Britain, gold flows were mostly about the balance between payments on maturing international bills of exchange (gold inflows) versus requests for new discounts (gold outflows). The money rate of interest in London was thus a symptom of international as well as domestic balance and imbalance, and the central position of the Bank of England in the London money market meant that its reserve was essentially the international as well as the national reserve. In normal times, if gold was flowing out of Britain, the Bank raised its discount rate in order to make new discounts less attractive, thus shifting the balance of payments back in its favor. The high rate of interest recommended by Bagehot for times of crisis was intended not only to limit the supply of funds to those most in need, but also to safeguard the nation’s gold reserve in the face of a potential external drain.

By 1873, when Bagehot was writing, the Bank had gotten used to its role as lender of last resort domestically, and this was the main focus of the Bagehot principle. But the Bank had not at all gotten used to its role as lender of last resort internationally, nor did Bagehot endorse such a role. For him, elasticity was all about domestic lending—here the Bank should not safeguard its reserve but rather mobilize it, down to the last farthing. But once those farthings come into the hands of foreigners who ask gold for them, the Bank has to stop. It can create more deposits to meet an internal drain, but it cannot create more gold to meet an external drain. In a crisis, the Bank could and did suspend the gold reserve requirement for notes, thus freeing up its gold holdings for payment to foreigners. But if that buffer was ever exhausted, there would be no choice but to suspend convertibility.

Clearly, the ideal solution would be to get foreigners to behave like domestic residents, which is to say to accept sterling balances (deposits or securities) as substitutes for gold. Britain’s most significant colonial possession already did so, as the young John Maynard Keynes pointed out in his first book, Indian Currency and Finance (1913). According to Keynes, the case of India showed that a gold-sterling exchange system was a workable arrangement for international monetary affairs more generally. But World War I, the Great Depression, and World War II dashed that dream. What we got instead, after the dust cleared, was a gold-dollar exchange system established at Bretton Woods in 1944, which became a plain dollar standard in 1973 after the United States abandoned gold convertibility.

The New Lombard Street

Our modern world is not Bagehot’s world, and not only because the dollar and the Federal Reserve have replaced the pound and the Bank of England, and the dollar standard has replaced the gold standard. For us, the most important money market instrument is not the bill of exchange but rather something called a “repurchase agreement,” or repo. Repos are issued not to finance the progress of real goods toward final sale, as in Bagehot’s world, but rather to finance the holding of some financial asset.

Formally, the underlying financial asset serves as collateral for a short-term loan, often as short as overnight. The “repurchase” refers to a legal construction whereby the short-term loan is arranged as the sale of an asset combined with an agreement to repurchase the asset at the original sale price plus some rate of interest. The original sale price is lower than the market value of the asset by an amount known as the “haircut”; the purpose of the haircut is to provide extra collateral for the loan, so the size of the haircut varies with the perceived riskiness of the asset being used for collateral. The lowest repo rates, and the lowest haircuts, apply when the collateral for the loan is a Treasury bill.

In our world, the Treasury repo market plays a special role as the main interface between the money market and the Fed. (I speak here of the way things worked before the crisis.) The Fed enters that market typically as a lender, offering short-term loans of highpowered money (deposits at the Fed) in return for Treasury bill collateral. On a daily basis, the Fed might “tighten money” by allowing outstanding repo loans to mature without replacement, or it might “loosen money” by offering new and larger loans. The immediate counterparties to these loans are the “primary dealers,” so called for their commitment to bid for Treasury securities whenever the Treasury wishes to borrow. In normal times, the funds that the dealers borrow from the Fed at the daily repo auction are a low-cost source of finance for their main business of making two-way markets in Treasury securities by posting offers to buy and sell.

The special position of the primary dealers can be considered a legacy of World War II, when the U.S. government issued vast volumes of Treasury securities not only to finance its own war effort but also to finance the war spending of its allies. When the war was over, the war debt remained, on the balance sheets of households that would use it to purchase houses and cars, on the balance sheets of corporations that would use it to fund conversion from wartime production, and on the balance sheets of banks that would use it to fund private loans. All of these debt holders depended on the ability to convert government debt readily into spendable cash, which is to say on the existence of the two-way markets provided by security dealers.

During the war and its immediate aftermath, the Fed directly fixed the price of government debt, and directly backstopped the convertibility of government debt into cash at that fixed price. After the Fed-Treasury Accord of 1951, the Fed no longer fixed the price of Treasury securities but it did continue to provide liquidity support to the Treasury market. Eventually, even that responsibility passed on to the primary dealers, with the Fed backing up the dealers by providing liquidity support to them through its daily operations in Treasury repo.

Here then is how the New Lombard Street works. Whereas Bagehot’s central bank used the discount rate to manage the system, the Fed focuses its attention on the price of overnight lending in the federal funds market, which is an interbank market for deposits at the Fed. (An overnight federal funds loan involves receipt of reserve funds today in return for payment of reserve funds tomorrow.) The Fed does not directly lend or borrow in the federal funds market, so the “effective” federal funds rate fluctuates depending on supply and demand. Instead the Fed uses the Treasury repo market to control the supply of the underlying deposits that are borrowed and lent in the federal funds market.

The Fed’s monopoly supply of bank reserves gives it considerable control over the federal funds market, but there is quite a bit of slippage between conditions in the federal funds market and funding liquidity more generally. The Fed is only a small player in the enormous general collateral repo market where security dealers fund most of their activity. And it is not a player at all in the offshore market in Eurodollar bank deposits, which is always available to banks as an alternative to federal funds and, indeed, has grown up to be the most liquid money market in the world. In both repo and Eurodollar markets, borrowers and lenders find one another and do business outside the reach of the Fed.5 As always, private credit elasticity is a substitute for public credit elasticity, indeed often an attractive substitute.

Nevertheless, it remains true that balance sheet operations by the Fed affect funding liquidity, and thus also market liquidity, through the risk calculus of security dealers. Dealers post prices at which they are willing to buy and sell a particular security—the buy (bid) price lower than the sell (offer or ask) price—and then they adjust those prices depending on customer response. If they find themselves accumulating a large position in a particular security, they lower their posted prices. The main idea behind this practice is to control risk by allowing their exposure to increase only if it comes at an attractive price. But the effect of lowering price is also to control cash flow by attracting more buyers and fewer sellers, hence more cash inflow through net sales and less cash outflow through net purchases.

Actual dealing operations are more sophisticated than this, but even this simple account is enough to make clear that security dealers provide a sensitive link between conditions in the money market and conditions in broader financial markets. At one end of the chain of causation, we have the Fed setting the federal funds rate; at the other end, we have private dealers seeking profit by making markets. Private dealers borrow in the money market in order to finance their market-making operations in capital markets; that is the way that “funding liquidity” in money markets gets translated into “market liquidity” in capital markets.6 The market for Treasury securities is the first place this market liquidity shows up, but then it gets spread by means of arbitrage more or less quickly and efficiently to other related markets such as those for corporate bonds and, more recently, residential mortgagebacked securities. (I remind the reader again that I speak of the way things worked before the crisis.)

By contrast to Bagehot’s time, under modern conditions the Fed’s discount window has fallen into disuse. When individual banks need money to meet their commitments at the daily clearing, they usually raise it from other banks in the wholesale money market. And when the banking system as a whole needs money, that money is usually raised by selling security holdings into liquid markets. Both channels are backstopped ultimately by the Fed’s commitment to stabilize the federal funds rate around a chosen target, and by its intervention to make good on that commitment by lending in the Treasury repo market. Put starkly, under modern conditions the Fed is always lending freely, but only to primary security dealers, only against Treasury security collateral, and only at the Treasury repo rate that corresponds to the target federal funds rate.

This practice was supposed to prevent crisis. The way it was supposed to work is that the Fed would lend freely to the dealers, and arbitrage would do the rest, modulo some term spread between Treasury bills and longer-maturity issues, and some credit spread between Treasuries and nongovernment issues. By raising the federal funds rate, the Fed would raise the funding cost of making markets and thus induce some deleveraging and push around the spreads. By loosening, the Fed would lower the funding cost and thus lessen the pressure to liquidate, again pushing around the spreads. That is how it was supposed to work and, in fact, how it did work until the recent crisis.

In the crisis, this system broke down. As asset valuations came into question, haircuts for secured borrowing rose sharply, even for Treasuries but especially for non-Treasury securities, and the result was forced deleveraging and disordered markets.7 The problem was that, in private credit markets, collateral is marked to market, not to fundamental value. Bagehot’s admonition to lend freely against any security that would be acceptable collateral in normal times is a principle for central banks only. Individual banks have always followed the save-yourself rule of lending only against securities that can be readily liquidated in current extraordinary times. This time was no exception.

In response to the severe contraction in private liquidity, the Fed stepped in, widening the category of counterparties to which it was prepared to lend, and widening also the category of collateral it was prepared to accept. Borrowers and lenders who had previously found each other in the wholesale money market now found each other only through the intermediation of the Fed. The result was, first, a hollowing out of the Fed’s balance sheet as it sold off its Treasury securities (to the former lenders) to fund new loans (to the former borrowers), and then an explosion of the Fed’s balance sheet as it expanded its deposit liabilities (to the former lenders), and used the proceeds to fund additional lending (to the former borrowers).

The Fed’s response to the crisis can be understood as a modern adaptation of the Bagehot principle, at least in part. Rephrased in terms that connect up with modern institutional arrangements, Bagehot can be understood as arguing that the central bank should act as money market dealer of last resort, providing both borrowers and lenders with what they want but at prices that are worse than they would be getting if they were meeting directly rather than on the balance sheet of the Bank. In line with Bagehot’s conception, not only would the borrower pay a high borrowing rate, but also the lender would accept a low deposit rate. It is the gap between the borrowing and lending rates that provides incentive for borrowers and lenders to find one another again once the storm dies down. In effect, the Bagehot principle can be understood as recommending that the central bank post a wide bidask spread in the money market and use its balance sheet to absorb the resulting flow of orders.

That is more or less exactly what the Fed did in the various emergency liquidity facilities that it opened in response to the crisis. The Fed’s bid-ask spread was not always as wide as Bagehot might have wished—the Fed charged only a small spread over the federal funds target for its Term Auction Facility (TAF) lending facility, and it also paid interest on its deposit liabilities. But other facilities had wider spreads, and as a consequence wound down rather quickly—to wit, the commercial paper funding facility and the central bank swap facility. So far, so Bagehot.

What was not Bagehot was the level of interest rates, which fell almost to zero. This was possible only because the Fed, unlike the nineteenth-century Bank of England, faces no reserve constraint in terms of gold. The whole world treats dollar deposits at the Fed not only as good as dollar currency, but also as the ultimate world reserve in a time of crisis. That means that the Fed, unlike the Bank of England, can create both more domestic dollars to meet an internal drain and more international dollars to meet an external drain. The Fed has no need to safeguard its holding of world reserves by keeping the federal funds rate high, since world reserves are its own liability.

But just because the Fed can evade the reserve constraint that others must obey does not mean that it should. There are reasons to question whether such evasion is the correct policy even for crisis times, and a fortiori for normal times. From a Hawtreyan point of view, the very fact of the crisis stands as an indictment of Fed policy in the years leading up to it. Hawtrey would have had no trouble understanding the present crisis as a consequence of the central bank losing control of a runaway credit expansion; at root the boom must be a problem of excessive elasticity and insufficient discipline. How did it happen that the inherent instability of credit was allowed to play itself out as it did? Where was the Fed?