Let’s not write the Fed a blank check

Last week, the Fed decided to ask Congress for the right to pay interest on bank reserves. (Hat tip Barry Ritholtz, see also William Polley, Mark Thoma, Brad DeLong) This is a very big deal.

Don’t be misled into thinking that the Fed’s proposal is just some arcane, technocratic change. The Federal Reserve is asking taxpayers for a big pile of signed, blank checks. That’s far too much power to put in the hands of a quasipublic organization with little democratic accountability. This authority should not be granted without some strong strings attached.

First, some background. There is a trend among central banks to move from old-fashioned, fractional-reserve banking to a system whereby interest rates are managed via a “channel” or “corridor”, and under which fixed reserve requirements might be dispensed with entirely. The basic idea is simple. The Fed currently manages interest rates indirectly, by manipulating the supply and demand for cash in the banking system. But the Fed could adopt a more direct approach. It could choose two interest rates, a “floor rate” at which the Fed would stand ready to borrow funds, and a “ceiling rate” at which the Fed would stand ready to lend. As long as there is no stigma attached to transacting with the Fed, banks would never lend for less than the floor rate or borrow for more than the ceiling rate. The interbank interest rate would necessarily lie within a “corridor” defined by these two interest rates. The Fed would continue to adjust the money supply to keep interest rates somewhere inside the desired range. But the corridor would serve as a fail-safe. When banks have more cash than would be consistent with the policy interest rate, they would lend the excess money back to the Fed, causing it to disappear in a poof of green smoke. When banks have too little cash, they would borrow more into existence, until the quantity on hand becomes consistent with the Fed’s desired interest rate. The level of borrowing from or lending to the Fed would provide feedback, telling central bankers whether they need to add or remove cash from the banking system to achieve their targetted interest rate, usually at the center of corridor.

A corridor system would represent a meaty change to how central banking is done in the US, but the approach seems to work okay in other countries. Advantages for central banks include more robust control of short-term rates, and the ability to fine-tune monetary policy by altering the “spread” between the central bank borrowing and lending rates without changing the core interest rate. A disadvantage, from taxpayers’ perspective, is that the loss of zero-interest reserves amounts to a stealth tax cut for banks. On the back of my napkin, the cost to taxpayers would be between $190M to $530M per year if the level of reserves is unchanged. (I’m assuming “floor rates” between 1.75% and 4.75% against reserves of $11B). The Wall Street Journal reports estimates of $150M and $280M per year. If one assumes that corridor interest rates will roughly match the Treasury’s average cost of financing over time, and that the Fed invests reserves in Treasuries, then the total cost of the program in NPV terms would be the value of the current (interest-free) reserves. This amounts to a one time cost of about $11 billion. A more serious drawback is that a channel system paves the way for the getting rid of reserve requirements entirely, which seems a perverse thing to do in a credit crisis caused by too much leverage. But reserve requirements have already been eviscerated, and nothing prevents regulators from maintaining or strengthening reserve requirements in a channel system.

So far, so good, then. As long as the Fed is conducting ordinary monetary policy, switching to a channel system offers modest benefits at a modest cost to taxpayers. But the Fed’s monetary policy has not been ordinary at all lately. In fact, it’s been quite extraordinary. It is in the context of this extraordinary policy that the Fed has asked Congress to accelerate its authority to implement a channel system, and it is in the context of this extraordinary policy that we must consider the change.

The core of the Fed’s new exuberance is a willingness to enter into asset swaps with banks. The Fed lends safe Treasury securities to banks, and accepts as collateral assets that private markets consider dodgy or difficult to value. (This is the direct effect of the Fed’s TSLF program, and the net effect of TAF and other lending arrangements that the Fed sterilizes in order to hold its interest rate target.) In doing so, the Fed puts taxpayer funds at risk. If a bank that has borrowed from the Fed runs into trouble, the Fed would face an unappetizing choice: Orchestrate a bail-out, or permit a failure and accept collateral of questionable value instead of repayment. Either way, taxpayers are left holding the bag.

In December, the Fed had $775 worth of Treasury securities. That stock will soon have dwindled to $300B, give or take. The difference, about $475B, represents an investment by the central bank in risky assets of the US financial sector.

$475B is an extraordinary sum of money. It is as if the Fed borrowed more than $1500 from every man, woman, and child in the United States, and invested that money on our behalf in Wall Street banks that private financiers were afraid to touch. For bearing all this risk, if things work out well, taxpayers will earn about what they would have earned investing in safe government bonds. If things don’t work out well, the scale of the losses is hard to predict. The Fed will claim to have done “due diligence” on its loans, to have valued collateral conservatively, and will point to strength of bank guarantees and the enormous diversity of collateral assets to convince us that its actions are safe and prudent. But rating agencies made the same claims about AAA CDO tranches, and turned out to have been mistaken. Correlations often tend towards one when asset values fall sharply. Central bankers struggling to manage day-to-day crises in financial markets might cut corners when trying to value complex securities. They might find it convenient to err on the side of optimism, as the ratings agencies did, albeit for very different reasons. And even if the Fed is cautious and sober-minded, are we sure that central bankers can value these assets more accurately than private investors?

If the Fed were to blow through the rest of its current stock of Treasuries, it would have invested more than $2500 for every man, woman, and child in America. Public investment in the financial sector would have exceeded the direct costs to date of the Iraq War by a wide margin. Would that that be enough? If not, how much more? Just how large a risk should taxpayers endure on behalf of companies that arguably deserve to fail, to prevent “collateral damage”? Have we considered other approaches to containing damage, approaches that shift costs and risks towards those who benefited from bad practices, rather onto the shoulders of taxpayers and nominal-dollar wage earners? Does this sort of policy choice belong within the purview of an independent central bank?

Now I don’t actually mean to be too harsh. Putting aside the years of preventable foolishness that got us here, in the new day that began last summer, a crisis emerged that had to be managed and the Fed was the only organization capable of stepping up to the plate. I don’t love the decisions that were made, but decisions did have to be made, and there weren’t very good options. But now we have a moment to reflect. If the credit crisis flares hot and bright again, how much more citizen wealth should be put at risk before other policy options are considered? That’s not a rhetorical question: We need to choose a number, a figure in dollars. My answer would be something north of zero, but not more than the roughly $300B stock of Treasuries that remains on the Fed’s balance sheet. But this is a decision that Congress needs to make.

And what does all this have to do with the question that will soon be put before the Congress, whether the Fed should be permitted to pay interest on deposits? Everything, as it turns out. Suppose the Fed decides it wants to swap more than the $300B in Treasury securities it currently has available in order to support the financial system. Given its current tools and practices, the Fed would have to print money in order to buy more Treasuries to swap. But if it did that, the extra cash would drive interest rates below the Fed’s target level, quite likely provoking inflation. The Fed cannot simultaneously swap away more than its existing stock of Treasuries and satisfy its legal mandate to promote price stability, unless it resorts to something weird.

But suppose Congress gives the Fed the authority to pay interest on reserves. Suddenly the Fed can print cash to buy all the Treasuries it wants to swap for troubled assets. When banks find they have more cash than they need, they lend the money back to the Fed, collecting the “floor” interest rate and removing the currency from circulation. Since interest rates can be held to any level by adjusting the “corridor”, the Fed would retain the flexibility to respond to inflation. At the same time, it would be able print cash in any amount that it pleases — “to infinity and beyond!” — in order to fund asset swaps (or outright purchases) at taxpayers’ risk. This strikes me as a delegation of Congressional authority that would not only be undesirable, but arguably unconstitutional.

So, should we simply refuse the Fed’s request? Probably not. Brad DeLong makes an excellent point:

The Fed may also want to raise the general level of interest rates in order to fight inflation–which requires that it sell its Treasuries for safe bank reserves rather than temporarily swap them for risky MBSs.

The Fed is already rubbing pretty close to its “balance sheet constraint”. If, after exposure to gamma radiation from televised images of food riots, Ben Bernanke were suddenly transformed into The Incredible Volcker, he might lack the tools he’d need to jack rates up into the muscular high teens, unless he’s given this new authority. So what should we do? James Hamilton has an answer:

Congress has a quite proper role in determining the magnitude of the fiscal risk that the Fed opts to assume… [A] statutory limit on the non-Treasury assets that the Fed is allowed to hold might make sense. Perhaps the outcome of a public debate on this issue would be a decision that the Fed needs the power to lend to private borrowers even more than the $800 billion or so limit that it would run into from completely swapping out its entire portfolio… Or perhaps after deliberations, Congress would decide that the business of swapping Treasury debt for private sector loans is one that is better run by the Treasury rather than the Federal Reserve.

I agree. I think that Congress should grant the Fed’s request, but it should simultaneously impose constraints on the composition of the Fed’s balance sheet that cannot be violated without express legislative consent. This will be a complicated exercise, unfortunately. Besides government debt, central banks quite ordinarily hold precious metals and foreign exchange, and limitations on non-Treasury assets will have to take this into account. Plus, restrictions would have to be written carefully to apply to off-balance sheet arrangements such as TSLF, and contingent liabilities like the insidious reverse MBS swap proposal. Finally, Congress must consider restrictions on the Fed’s ability to enter into derivative positions, whether directly or indirectly via special purpose entities, including how the bank’s existing derivative book should be managed and whether the bank should or should not guarantee the liabilities of current Fed-affiliated SPEs.

Congress might also limit the quantity of reserves on which the Fed will be permitted to pay interest.

The Fed can retain full independence for the purpose of conducting ordinary monetary policy, exchanging government debt for cash and vice-versa. But if the central bank wants to put ever greater quantities of public money at risk, it will have to accept a lot more public supervision. If the prospect of intrusive oversight is too much for the Fed, then, as James Hamilton hints, perhaps the roles of central bank and macroeconomic superhero should be moved to separate boxes on the organizational chart. If we are not careful, the next bank requiring a taxpayer bailout may be the Federal Reserve system itself.

Update History:
  • 12-May-2008, 2:20 a.m. EDT: Changed a “fine” to “okay” to avoid having “fine” too close to “fine-tune”.

37 Responses to “Let’s not write the Fed a blank check”

  1. RueTheDay writes:

    There are a couple of interesting, though different points here.

    I agree that enabling the Fed to pay interest on reserves is a step in the direction of getting rid of reserve requirements altogether. A number of countries have gone down this route – Canada, Australia, New Zealand, UK, etc. There are arguments to be made in this direction – statutory reserves do not really serve as the cushion they were intended to, since, by definition, they can’t be used to pay depositors in the event of a bank run. Thus, they really serve no purpose. Countries that have done away with reserve requirements generally manage risk through capital adequacy requirements. They also generally see greatly reduced actual reserves on deposit at the central bank, so in practice, central bank interest payments on reserves turn out to be almost nil.

    What really caught my eye in this post was the following:

    “…perhaps the roles of central bank and macroeconomic superhero should be moved to separate boxes on the organizational chart.”

    This is the important point. I’m not crazy about Paulson’s idea of a super-regulator, however, if we’re going to end up with one, then I do not believe it should be the central bank. Let the Fed focus on stability in the monetary system and on the lender of last resort function (to banks and ONLY banks) and put the other regulatory functions elsewhere. I can’t even wrap my mind around what it would mean for the Fed to assume such a function. Had it had such powers in 2000, would we have seen a TSLF variant where Treasuries were lent with dotcom stocks taken as collateral in an attempt to stabilize the NASDAQ? If we’re opening the door to radical regulatory change (not saying whether we should), then perhaps we ought to go even one step further back and rethink the structure of financial markets altogether.

  2. Steve,

    I’d like to comment on your April 6th post, “Central banks are dangerous”. The comment period for that post expired, so I hope you don’t mind my putting my comments here:

    I know you’re concerned that the fed chairman could be incompetent, and so if you give him a lot of power he could cause serious problems. Further, you state, “Our long-term plan, though, ought not be to canonize central banks, but to render them obsolete.”.

    What I say to that is this: it’s like saying our long term plan should be to have no U.S. government. U.S. Governments may sometimes be horribly incompetent and corrupt, relative to what we normally have , like the Bush administration and the Republican congresses of the conservative era. But having no government at all would be far worse. That would lead to far lower quality of life and wealth creation. It would mean no police, no military, no public education, and on and on.

    I know you are recently started a finance Ph.D. program. As you learn more and more about economics you will find out more and more the crucial importance of government in addressing pure free market problems like externalities, asymmetric information, inability to patent, large economies of scale and monopoly power, and much more.

    As with government in general, likewise for the Fed., because Greenspan, despite the mythology, was a horrible Chairman, doesn’t mean we should have no Fed. Or that we should not give it strong powers. The cost of a rare chairmen who acts largely based on Libertarian ideology rather than what’s best for the economy, is better than the immensely greater costs from having no Fed at all, or of a very weak one.

    Keep in mind that if a fed chairman is grossly incompetent for a long period of time the economic community will speak out en masse, and congress and the president can pass a special law to remove him. There are checks and balances.

    Why is the Fed so important and indispensible. You will learn this as you progress in your education, but to learn a great deal about this quickly and with great intuition, I strongly suggest you read two books by Princeton Economist and John Bates Clarke Medal winner Paul Krugman: “Peddling Prosperity” and “The Return of Depression Economics”. You will really enjoy, and learn from, these books.

    A final analogy is with modern scientific medicine. You can find terrible doctors, but that doesn’t mean it’s best to never use modern medicine and instead to go back to primitive herbs and folk remedies for everything. With all of the flaws in modern medicine, it’s more than doubled the human lifespan compared to the days of only non-scientific medicine, given us a Polio vaccine, etc., etc.

    And really, I’d say the same thing for academic economics and finance. It has a lot of flaws and inefficiencies, but it’s the only game in town that is largely trying to and succeeding in being scientific, and that thinks long and deep about issues, well beyond simpleminded sound bite “analyses” like those used by today’s conservatives.

    Academia may have serious flaws and waste, but it’s essentially the only game in town for so much of what it does, and the good that does come out, is so valuable, it’s immensely worth paying for the waste that’s hard to fight in this insular world of massively asymmetric information.


    Blog: http://richardhserlin.blogspot.com/

  3. RTD — One way of thinking about bank reserve requirements is that they’re like the bank-regulation version of the AMT.

    Sure, in theory, they’re a barbaric relic. As you say, they ain’t about preventing bank runs, and in theory, if banks had perfect risk management models, reserve requirements would simply block useful lending for no good reason.

    But like clever tax-avoiders, banks are always at finding ways to lend ever more freely, in an old-fashioned tragedy of the commons. (It is always in each bank’s interest to have a freer hand, but widespread loose lending creates widespread failures.) Simple, clear reserve requirements limit the effects of regulatory capture and risk-model cleverness, etc. Banks are deputized with the power to conjure from air funds the government will guarantee as money. Reserve requirements, if designed well, place a regulable upper bound on the public obligations that private banks are empowered to create.

    I’m not certain that comprehensive, fixed reserve requirements should make a comeback. But I wouldn’t take “great moderation” era experience, in the US or other countries, as definitive proof they are dispensable. The US largely abandoned them, and it worked until it didn’t. The same may hold true in the other countries you cite. Or not.

    Re the cost of paying deposits, it’s important to note that the real price to taxpayers of abandoning required, interest-free reserves isn’t the interest that has to be paid, but the free loan that is lost. Holding constant the size of composition of bank deposits, the required interest-free reserves represent a zero interest loan to the public that will rollover perpetually. That is, the reserves are effectively an asset of the public, not of the banks. (Sure the public can’t spend them, but taxpayers earn the interest banks don’t and can use that to finance an equivalent amount of other spending.)

    If, under a interest-paid-reserves-not-required regime reserves drop to near zero (and they would drop to near zero, unless the central bank as a policy choice wants to grow or shrink its balance sheet by borrowing from or lending to banks), the cost to taxpayers is identical to the case where zillions of dollars are deposited and interest is paid. To a first approximation, in either case taxpayers neither earn nor lose money from the arrangement. When there are no reserves, nothing happens. When there are lots of reserves, taxpayers earn interest on the reserves and remit that interest to banks. The “cost” in either case is zero in cash terms, but there is an opportunity cost relative to the alternative policy of requiring reserves and not paying interest them. It’s not that paying interest on reserves is costly, it’s that not paying interest and requiring reserves is lucrative. Our present arrangement is a tax on banks, worth a few hundred million dollars a year. Paying interest just means cutting that tax, regardless of the quantity of reserves we end up paying interest on (as long as the interest we pay and the interest we save by purchasing Treasuries is comparable).

    I agree with you on the last. The headline piece is sprawling and overlong, but one of the main points trying to escape is that there are policy domains in which central bank independence might make sense, and policy domains in which it does not, and that risking taxpayer fund for the purpose of discretionary bailouts is definitely not a function that should be left to central bankers alone.

  4. Richard — My critique of central banks isn’t so much about personality as structure. I don’t think one should lay the failings of central banking solely at Alan Greenspan’s feet, and I think Ben Bernanke is quite brilliant, in a tactical way. (It’s just that I disagree with the project he’s set for himself.)

    I think central banks, as an institutional arrangement, are liable to doing great harm, for very “Hayekian” reasons — they act based on this week’s latest theory, always professing to have learned from the past, but central bankers fundamentally cannot know enough to do even the smallest part of their job, defining currencies that serve as meaningful and sustainable yardsticks of value without hindering the real economy with high interest rates. Every generation of central bank has considered itself finally adequately prepared and cutting edge, and every generation has turned out to make grave mistakes (with the exception of Volcker, whose pain infliction was so obvious he is given a pass, since surely the awfulness of the medicine was intended).

    Moreover, the international dimension to currency valuation creates a whole game-theoretical dimension between central banks, that can become dangerous if central bankers aren’t careful enough or lucky enough to prevent it via some mix of cooperation or deterrence. Central-banking has become the heart an international monetary system that has failed gravely over the past generation, in my view, and that is continuing to store up devastating consequences. We’ll be lucky to escape mass tragedy. And that’s not because the individuals running the banks were incompetent or malicious. Some of them erred, and some of them played their games too well while others played poorly.

    That said, central banks all we have for the moment. But architecturally, I think the central bank based money system is a weak arrangement, and we’ll either have to invent something substantially better, or introduce constraints that prevent malfunctions as dramatic as we’ve seen, both recently and in the past.

    BTW, I’m not broadly speaking a libertarian. I think institutions matter very much, and that broad labels of “free markets” vs “government” generate more heat than light. We should be interested in defining institutional arrangements that work. In the end, how we label the institutions matters less. CFTC and ISDA are very different organizations, one a government regulator with coercive power, the other a voluntary association, but they sometimes serve similar purposes in similar ways. I don’t claim that we should ban central banks (and certainly don’t claim that governments should keep out of sacrosanct free markets), only that central banks as they are perform sufficiently poorly that we should constrain them some in the present and try to come up with better approaches to solving monetary and macroeconomic stability problems as we move forward. I won’t care much whether the improved institutions are incrementally developed offshoots of today’s organizations that we still call central banks, or whether we have very conspicuously obsoleted and retired existing institutions. As long as what we end up with works better, a lot better than what we have now.

  5. Alessandro writes:

    Thank you very much for clarifing such technicalities to non-experts (but willing to understand) as me.

    Keep up the good work!

  6. RueTheDay writes:

    What we need is structural reform to return the financial services sector to its proper role within a capitalist system, with that role being to enable firms to finance real investment (investment in the economic sense of the term, not the popular sense) and to thus promote the capital development of the country. It’s time to purge the casinos.

    To that end:

    Pass Glass-Steagall II. Re-instate the original separation between commercial banking and investment banking activities. Then take it a step further and, within the investment banking realm, wall off primary market activity from secondary market activity. In other words, separate the underwriting business from the trading/broker dealer/research/money management business. Let’s not forget the lessons of Enron just because we’re facing a different financial crisis now. The objective here is to create firewalls and eliminate feedback loops in nonlinear systems.

    Impose capital adequacy requirements on all large financial firms. Eliminate the mortgage interest deduction for individuals. Change the corporate tax laws that favor debt financing over equity financing. Rather than allowing corporations to deduct dividends paid, eliminate the ability to deduct interest paid. Set (rather low) caps on the interest rates credit card companies can charge, which will in turn restrict the amount of credit they willingly extend to poor credit risks (see Adam Smith’s discussion on usury laws). The objective of this disparate set of reforms is to reduce the degree of leverage in the overall economy, which has gotten completely out of control.

    Longer term, I would like to see some sweeping changes made to the tax system. First, the implementation of a very small Tobin-esque tax on all secondary market financial transactions. Second, and even more important, a Henry George style tax on the site value of land extended into a broader tax on economic rents in general. More on that in a later post.

  7. martinb writes:

    I don’t think there’s a balance sheet constrain in the short run for the FED since negative equity is a non-issue. Historically, afaik, the FED has made a profit every year so there are is an implied, relatively large, off-balance sheet asset in the form of the discounted “seignorage” that can cover their risk-exposure to a high degree. Now this could still rightly be seen as paid by the tax-payers since in the end the liabilities of the FED is backed by the states ability to tax. But I don’t think it necessarily means a monterization of the eventual losses.

  8. jwh writes:

    My knowledge of economics is extremely basic, so I had a question about the following two sentences.

    “But suppose Congress gives the Fed the authority to pay interest on reserves. Suddenly the Fed can print cash to buy all the Treasuries it wants to swap for troubled assets.”

    It seems to me these sentences say the following.

    authority to pay interest on reserves implies Fed can produce infinite amounts of money

    I do not understand how the first implies the second?

  9. martinb writes:


    It doens’t. But it can print to the amount of the swapped assets. Which could be considered sterelized money creation. In order for the FED to be able to print an unlimited amount of money under this sytem it would have to lower rates to stimulate the willingness to hold this new money. Otherwise it will return to the bank, where they’d have to pay interest on it. It could be argued that this actually limits the ability to print money.

  10. RueTheDay writes:

    One other thing, and this can be a bit of a touchy subject. Something must be done about the asymmetry in compensation that is endemic throughout Wall Street. During the boom, people are making 7 and 8 figure bonuses (there was a report released a few weeks ago regarding some hedge fund managers making in excess of a _BILLION_ dollars in comp each). However, there is no downside, only upside. To add insult to injury, every bust is accompanied by enormous losses and writedowns, meaning the boom time profits weren’t even real. Yet no one is being asked to give those boom time bonuses back.

    This has the potential to be an explosive issue, particularly in an election year. Whether you call the Fed’s recent actions a bailout or not, the popular perception is that it was one. It’s difficult to reconcile bailouts with stratospheric comp packages while paying lip service to free markets.

  11. jwh & martinb —

    So, I guess I didn’t explain very well how paying interest on reserves relieves the Fed of its balance sheet constraint. Here’s the deal in a nutshell:

    1) Mechanically, the Fed always has the power to print all the cash it wants, to buy or lend however it sees fit. (Theoretically the Fed is subject to statutory constraints on what it purchases and to whom it can lend, but it has interpreted those loosely.)

    2) However, without the ability to pay interest on reserve deposits, if the Fed it were to exercise its power to print and spend, it would force interest rates down towards zero (and stoke inflation). So, if the Fed wishes to hold its interest rate target, it is de facto unable to just print money to buy what it wants.

    3) But if the Fed paid interest on reserves, the Fed could print money to buy stuff while holding an interest rate target. The Fed would buy stuff, but banks would just lend the money back to the Fed at whatever rate the Fed pays. The extra cash becomes “one-time-use” money: it is spent once to buy what the Fed wants, but then disappears back into an account at the Fed, from which it will return only if market interest rates are higher than what the Fed is paying. The Fed can print and spend as much as it wants without forcing interest rates below the “floor” rate.

    4) The fact that the Fed has to pay interest on this boomerang cash is not a constraint, so long as the assets that it purchases perform and pay interest comparable to the deposit interest rate. If T-bill rates closely follow the Fed’s rate, the Fed can print cash to buy up a bazillion dollars worth of T-bills, even though banks will deposit that cash with the Fed and expect interest payments. The interest the Fed has to pay on a bazillion dollars in cash deposits will be the same as the interest it earns on the T-Bills it has purchased. The Fed just passes the money along.

    5) Of course, if it lends those T-bills to banks that then go bust, it might not get interest payments to pass along to depositors. But that’s precisely the sense in which letting the Fed expand its balance sheet enables it to expand its risk-taking, which is ultimately backstopped by taxpayers.

  12. RTD — Many of your suggestions, and the underlying way you are considering the problem, are very close to my perspective.

    I have a kind of random question: What do you do for a living? (I’ll explain why I wanna know afterwards, if you don’t mind saying.)

  13. martinb — Seignorage is by definition monetization, it is represents the difference between the purchasing power of newly produced money and the cost of production, which represents profit to the currency issuer.

    When a central bank prints money to cover a private loss, that is an opportunity cost to the taxpayer. If, given the fact of the loss, there was excess demand for cash in the economy so that money could be printed without driving interest rates below target, the newly printed money could have been used to purchase public debt rather than to make a private party whole. The monetary/inflation implications of both transactions are identical, there are distribution consequences. One option privatizes the seignorage profits while the other captures the benefit for taxpayers.

  14. JKH writes:

    SRW – you’ve covered considerable territory, with some interesting links. My comments refer mostly to the technical perspective, rather than the fundamental issue of Fed balance sheet expansion. Concerning the latter, there can be no doubt there must be some new transparent set of rules for the Fed’s expanded activities.

    There are different ways of defining or describing the Fed’s strategic mission. But its central operating objective is control of the short term interest rate – i.e. the Fed Funds rate. This was true even in the Volcker era. Volcker may have prioritized containing money supply growth, but he used the short rate as the operational weapon to get there.

    In order to control the short rate, the Fed must control the supply of reserves to the banking system. The Fed steers the short rate by fostering a competition among banks for a finite supply of reserves, the amount of which the Fed determines. The Fed determines the supply of “excess reserves” over and above the requirement mandated by the defined reserve ratio. The market forces of supply and demand move the short rate up or down depending on whether the supply of excess reserves is tight or loose. To the degree that the Fed wants to control the short rate, it actively manages this excess reserve setting. And the Fed always wants to control the short rate.

    As a result, the “multiplier” relationship between reserves and deposits doesn’t work at all as it’s described in textbooks. In fact, the true causality is the reverse of the textbook explanation. Bank balance sheet expansion is not a function of supplied reserves. Supplied reserves are a function of bank balance sheet expansion. This is the only way in which the Fed has proactive control over the level of excess reserves, which it needs to have in order to control the short rate. The Fed implements its strategic approach to inflation, real economy growth, and the banking system by plotting a path for the short term rate. It expects the banking system and bank balance sheets to respond to the short rate over time, not to a reserve multiplier accounting identity.

    The correct relationship can be illustrated using an option analogy. The reserve requirement is analogous to the strike price of the option. The supply of excess reserves becomes the intrinsic value of the option. The Fed wants to control the intrinsic value of the option in order to control the short term rate. When bank balance sheets expand, the Fed resets the strike price for the option – i.e. it increases the overall supply of reserves to ensure that banks have the opportunity to meet their new reserve requirements. The Fed still controls the intrinsic value component (excess reserves), and can steer the actual trading level of funds toward target by adjusting the excess setting. The option strike price (required reserves) has increased, the underlying (total reserves) has increased, but the intrinsic value (excess reserves) has been kept in line with short rate management requirements.

    If the Fed did not supply new actual reserves in response to changes in required reserves in this way, changes in required reserve levels (strike reset) would start to change the excess reserve setting (intrinsic value) by default, and the Fed would lose control over the excess reserve setting and therefore over the short rate.

    Thus, the classic textbook causality of the multiplier works in reverse. The Fed supplies the basic level of required reserves in response to changes in bank deposit composition. It then sets the excess reserve setting at the margin to put upward or downward pressure on the short rate as needed.

    The corollary to this is that the stipulated reserve requirement is irrelevant to Fed’s conduct of its operational goal of short term rate control. Using the option analogy again, it is the intrinsic value of the option (excess reserves) that gives the Fed its leverage over rate behaviour, not the strike price (reserve requirement).

    The effect of paying interest on reserves has been described as creating a rate “corridor”. I would describe it as an “administered rate corridor”, with open market operations being an “intervention rate corridor”. The latter arrangement is looser in the specification of target levels at which a “floor” is invoked, but the general idea is the same.

    The administered corridor idea does not in and of itself represent a departure from “fractional reserve banking”. The same required reserve level can operate under either administered or intervention corridors. This includes the possibility of a 0 reserve requirement under either system. Either type of “corridor” sets up bounds within which the Fed attempts to target the actual funds rate toward the target rate, which becomes the targeted “mean” for actual rate volatility.

    One should be precise in using the phrase the “elimination of reserve requirements”. It does not refer to a system in which the discipline of reserve management is in some sense abdicated, and where there is no longer any reputational stigma in borrowing from the discount window. It refers rather to a system in which reserve requirements are specifically set at a level of 0. This positions it as one possible setting out of many for a specific level of reserves within a framework that allows the Fed to remain in control of the short rate. Therefore, there is nothing unique about the 0 reserve level insofar as this most important characteristic of Fed management is concerned.

    Nor does anything change with respect to multiplier causality. Viewed in the textbook fashion, the multiplier becomes infinite. But viewed according to the true causality, it merely means that the Fed sets the supply of required reserves at 0, with any additional supply of reserves equal to the excess reserve setting. Setting the requirement to 0 becomes a specific case where, in the option analogy described above, the option becomes the underlying, with a strike price of 0 and an intrinsic value equal to the underlying (excess reserve setting). A targeted excess reserve setting is still operational in a 0 requirement system, in order for the Fed to continue to control the funds rate and steer it toward the target or mean value within the specified or assumed corridor.

    The volatility of the actual rate within either type of corridor can never be 0, simply due to the operational frictions within the bank clearing system. This is a function of such variables as the micro distribution of reserves and differentiation across individual banks in perceived credit risk (a major problem today). Multilateral reserve clearing at the target funds rate is difficult to obtain at a 0 excess reserve setting.

    This is all to say that it doesn’t matter if the system reserve requirement is $ 10 billion, $ 1 trillion, or $ 0 in order for the Fed to control the funds rate. But it must be some stated level to which the banks must conform in aggregate and according to their individual required shares. This control is necessary whether the corridor is administered or interventional. The corridor is intended to define a permissible range for short rate volatility, and it integrates to the target mean for the Fed funds rate.

    In summary, the stipulated level for required reserves is irrelevant to the Fed’s operating objective of short rate control. Higher levels of required reserves do provide banks with a degree of liquidity insurance in the event of a deposit run, but this would be short lived in scenarios that become seriously relevant. The most substantial economic impact of the required reserve level is the implicit taxation it imposes on the banking system when no interest is paid. This becomes one component of the seigniorage benefit associated with the Fed balance sheet.

    Other comments:

    The administered corridor may be a more efficient monetary mechanism than the intervention corridor in a legitimate sense
    . But it opens up the theoretical possibility of substantially increased financial intermediation by allowing the Fed to fund asset expansion with excess reserves. Any such activity would be highly transparent, at least after the fact. It would seem outrageous on the surface for the Fed to use a mechanism intended for efficiency improvement as the means with which to blow up its balance sheet in a credit crisis. The convenience would be useful, given reasonably framed policy objectives in dealing with the crisis, but the perception would not be good – that of a back door technique for an adventure in the socialization of credit risk.

    The payment of interest on current reserve levels would be a very small cost relative to what would continue to be free funding from notes in circulation, which is the much larger component of the monetary base. The interest cost is minimal compared to the continuation of the bulk of the seigniorage benefit.

    Finally, on seeing your link to the old Chicago article, I immediately flipped to the back, past the standard T-account examples, to check for the kind of final qualifications that typically appear in such articles, regarding how things work “in the real world”. This is how such articles obliquely reference the issue of the reversal of multiplier causality.

    Ironically, the Fed itself does a better job of admitting to this issue on its website:

    “In practice, the connection between reserve requirements and money creation is not nearly as strong as the exercise above would suggest … the Federal Reserve operates in a way that permits banks to acquire the reserves they need to meet their requirements from the money market, so long as they are willing to pay the prevailing price (the federal funds rate) for borrowed reserves. Consequently, reserve requirements currently play a relatively limited role in money creation in the United States.”



  15. RueTheDay writes:

    “I have a kind of random question: What do you do for a living? (I’ll explain why I wanna know afterwards, if you don’t mind saying.)”

    Steve – My degree is in economics and finance, but I work as a technology consultant designing analytical systems for large companies (mostly data warehousing and BI, some forecasting and predictive analytics). Several of my clients have been in the banking and capital markets sector.

  16. martinb — re: negative equity — The equity position of a central bank is a nebulous thing, and you are right to point out that there is nothing automatically apocalyptic about a central bank having negative balance sheet equity. As long as the central banks liabilities are in its own currency, it can always print more, so while “technically insolvent”, the CB would always be liquid, like GM.

    But in the end, the formal equity position of the CB is irrelevant, the flow of currency matters more than the stock. For a CB to print its way to negative equity, it would have to print money and give it away, consume it, or purchase assets for more than they’re worth. If it could do this without causing inflation, it could have given the money to the to pay off public debt Treasury instead, saving money for the taxpayer. Practically, the first question is the economy’s ability to absorb new money without causing inflation. Then comes the second, but most interesting question, cui bono?

    Of course, in practice, there are symbolic problems with central bank negative equity, and we’d never tolerate it. But it doesn’t matter. We could paper over CB negative equity by having the Treasury issue a bond, and give it to the Fed in a recapitalization. That’s precisely what a “taxpayer bailout of the Fed” would look like, and doing this makes it transparent that a decrease in the Fed’s equity position is a loss to the taxpayer. But the actual fact of issuing a Treasury bond or leaving the Fed with diminished book equity is an accounting choice, without real effect. When the central bank takes a loss, taxpayers absorb the cost, whether we allocate that loss to the Treasury via an explicit recapitalization or not.

    There are wrinkles, though: For example, China’s central bank might take a large loss on its US dollar reserves as its currency appreciates. But the Chinese government captured large seignorage gains by printing Yuan to buy dollars. Theoretically, China could have printed Yuan and bought Chinese government debt instead, saving money for taxpayers. But China’s ability to print money in large quantity without (until recently) provoking inflation depended on the Chinese economy’s phenomenal growth, which itself arguably would not have happened without China providing “vendor financing” via its currency peg. So, one could argue that the there was opportunity to the taxpayer, because under the alternative policy, the original opportunity for noninflationary seignorage would not have presented itself.

    Similarly, one could argue now, with respect to the Fed’s policy, that there is not really an opportunity cost to covering private losses with the government’s seignorage power, because without the Fed’s intervention, the economy would tank so badly that opportunities for future seignorage and/or taxation would be diminished (the excellent knzn has made something close to this argument).

    Fundamentally, though, I don’t think it works. A credit contraction creates a huge opportunity for seignorage on behalf of taxpayers, as both the reduction in credit and the reduction in the velocity of money are disinflationary. The only kind of economic growth that creates seignorage opportunities is the kind that would reduce the dollar prices of real goods and services absent central bank intervention. I don’t see how supporting credit bubble exposed institutions is likely encourage that sort of growth. On the contrary, privatized money creation invested nonperforming projects that provoked the credit bubble was inflationary, and seignorage opportunities were lost to taxpayers that should now be recaptured. (Absent private credit creation this decade, we would have experienced more disinflationary headwinds, which would have obligated the central bank to monetize some public debt in order to prevent deflations.)

  17. PrintFaster writes:

    This harks back to Andrew Jackson the US bank, and the events leading up to the Fed establishment.

    What happens with concentrated or central banking is the need for more and more secrecy because banking becomes more and more unstable to catastrophic runs.

    Before the Fed, there were runs on banks, and the effects were geographically and financially limited. Banks essentially lent only to people that they knew, and people put money in banks that they trusted. What is happening now is that banks are too large to fail and need the veil of secrecy to protect them from runs, eg. BofA, Citi, Morgan. On top of that the Fed needs more secrecy to protect it from a run on international confidence, exactly what this bill is intended to protect.

    Thinking, the panics before the Fed were all rather limited, but after the Fed we had a national depression. Under current circumstances, unless the fed and other G10 nations are decoupled, we will have an international depression.

    Banking must be decentralized to prevent a catastrophe, similar to the compartments on a ship, that can be sealed off to prevent the sinking of a ship if there is one hole. We have one hole, and one compartment. The ship will sink.

  18. RTD — Ha! That counts as IT!

    So, the reason I asked is that I have a theory, that software and IT people tend to a particular spin on financial issues by virtue of being involved in the management of systems that are too complex to control or predict in practice, but which we consider nonetheless more “designed” then “emergent”. Software people talks of single points of failure, redundancy, feedback loops, and data consistency, and think that by an ongoing process of “architecture” we can make a real difference in the behavior of such systems. A lot (though not by any means all) of non-IT finance types view markets and market misbehavior as fundamentally natural phenomena, that we might be able to mitigate at the margins, but not so much. The non-IT types tend to be more conservative — the institutional arrangements were currently have are the market, they have emerged “naturally” for the most part, and we should work within them, one way or another. IT-inspired finance types tend to be “radical” in that we see financial systems as not so different from enterprise scale information systems, where there is unbelievable uncertainty to manage and significant costs to change, but where we nonetheless find it often worthwhile to incrementally “rearchitect” in hopes of evolving towards more effective and better-behaved systems.

  19. martinb writes:


    I know some of my previous comments was a bit rushed but I’d just like to say that the result of any discussion about central bank financies is bound to end up in the realization that the balance sheet of the central bank only is an extension of the states’ and what matters is not technical but policy insolvency.

    If the loans that the FED has swapped fails on a big scale the question will be if the government is prepared to have a weak central bank or bail it out. If the seignorage is used to rebuild equity you’re of course right that it’s still paid by the taxpayer. In the end the thing that backs the libalities of the central bank is of course not it’s equity but the states power to tax.

    This is certainly the big test for the hailed inflation fighting central bank doctrine. Atleast the FED isn’t tied by a nominal anchor…

  20. JKH — An excellent comment, and one on which I find very little to take issue. I didn’t mean to imply that I viewed the elimination of reserve requirements as anything particularly apocalyptic. They are largely gone already in the US, and are formally eliminated elsewhere. This needn’t be a problem as long as banks have some need for cash even when no reserves are required. As you say, the absolute level of bank demand for cash isn’t so important for the day-to-day goals of a central bank, since by adjusting the supply side relative to that demand, CBs can achieve any desired interest rate.

    I don’t think it’s quite right to say that textbook causality is “backwards” though. Both endogenous cash needs and reserve requirements affect banking sector cash demand, while the Fed supplies cash. Both supply and demand conspire to form a price. It’s just that in practice, endogenous cash needs fluctuate much more than reserve requirements, and the Fed responds to fluctuating demand by changing supply in order to hold a price. In theory, the Fed could hold supply constant, and vary reserve requirements with high frequency to offset endogenous changes in demand. This would be cumbersome in practice, but it could be done.

    That reserve requirements are neither the binding constraint on bank lending nor the sole determinant of cash demand doesn’t imply they are superfluous, though. See my first response to RueTheDay here. In practice, at present, the “multiplier” is a fiction, because in aggregate reserve requirements are loose relative to bank lending (which does not, as simple models would suggest, automatically expand to the maximum permitted). But if reserve requirements were tighter, the multiplier would in fact bind. It’s fine if reserve requirements are usually loose relative to bank lending, but I think a case can be made for maintaining them in order to define an outer limit to bank lending. As efficiency in cash management increases, it’s perfectly possible that endogenous cash demand (by banks and the economy at large) would come to approach zero, at which point the Fed would lose control of both interest rates and the overall quantity of credit creation, absent reserve requirements. The greater the degree to which the public insures private credit creation, via FDIC or implicitly via a propensity to bail-out creditors, the greater the justification for public limits on the overall the quantity of lending. The only way to establish certain limits, not sensitive to changes in cash management in the economy, are with a combination of reserve requirements and supply management.

    That said, I like your view that zero is just a number. Central banks in theory have a bunch of free parameters, including reserves supplied, a lending rate, a borrowing rate, and reserve requirements. Under the present system, reserve requirements are (in theory) not permitted to go to zero, and the Fed’s borrowing rate is restricted to precisely zero. By eliminating restrictions, we increase the policy options available to central bankers. Assuming widely shared policy goals and technically capable central bankers, eliminating those restrictions could only be a good thing. However, without widespread agreement or perfect competence, it may be reasonable to impose restrictions. For example, Congress might impose a statutory max for reserves supplied and statutory minimum reserve requirement for insured categories of deposits. This eliminates the usual public choice problem with central bank non-independence, because legislators can only restrict money and credit, they cannot expand it (although there’s a danger the distinction becomes academic in practice if the restriction nearly always binds). Similarly, it strikes me as reasonable for the legislature to impose restrictions on credit risk taken within balance sheet limits.

    I also like your taxonomy of an administration corridor and an intervention corridor. I agree that these are just two tools, neither of which, as you say need be abdicated.

    While talking terminology, I do think that we need a different term than “fractional reserve” for a system where zero is a permissible value for a bank’s inventory of the commodity it owes. Sure, such a system might have evolved from fractional reserve banking and represent a generalization thereof, but without any substantive notion of fractions and multipliers the term “fractional reserve” misleads more than it reveals. Perhaps another way to recast your point is to say that we abandoned fractional reserve banking decades ago, if we ever practiced it at all, so that proposed changes are less radical than one might think. We should come up with an informative name for the banking system we actually have, and update the textbooks to describe it. Banking systems are important institutions, too important to be left as evolved practice understood only by professionals and insiders. Banking systems must in fact be sensibly arranged, and the theory of their arrangement must be accessible to those willing to put in the study time.

  21. PrintFaster — I wish I disagreed with you. I’m not sure about the secrecy part, but our response to the credit crisis thus far has been to pool risk and centralize it where taxpayers cannot easily to disown it. That strikes me as high risk even in a static situation, since we don’t know the scale of the losses. It strikes me as terribly misguided in our real, dynamic situation, as if we always do this, people have every incentive to take on wild risks, harvest the winners and pool the losers, until the scale of losses is catastrophic and can no longer be papered over. I don’t trust that regulation will prevent this.

  22. shrek writes:

    Hey Steve,

    No matter what the fed does it cant make all this risk disappear. As you note above all its doing is shifting risk which the banking system already got into massive trouble trying to do. This is a VERY broken system.

  23. Aaron Krowne writes:

    I think it’s funny that suddenly the inability to raise interest rates is seen to be a potential problem! This is not only funny because the Fed is pawning off its balance sheet at a rapid pace, but because the only real reason Treasuries have been expensive (and low-interest-rate) is because of the fundamentally paradoxical “recycling” of a handful of central banks into the bonds.

    It seems to me that if this central bank recycling dysfunction were addressed, the “problem” if having higher interest rates would certainly go away.

    But somehow I doubt the Fed really considers this a “problem” — they can always blame inflation on external factors. Everyone else does it.

    Another point I’d make is that too-low interest rates will likely take care of themselves: we are facing unprecedented looming supply of Treasuries due to the fiscal deficit. It would be pretty amazing if all of these extra hundreds of billions were lapped up by the safe haven play and we could maintain current low interestrates.

    Also, the risk of central banks getting out of the dollar — and hence ceasing their Treasury buying — seems to be greater than the odds that they will step up their buying from here, and put downward pressure on rates.

  24. Aaron Krowne writes:

    I’m still trying to wrap my mind around what it really means to eliminate reserves (acknowledging of course the fact that they are already basically gone).

    Say a bank makes a bunch of bad loans under this system. The loans fail to pay out and the bank enters a cash crunch. Their bank peers will not lend to them so they have to go to the Fed at the ceiling lending level. The Fed lends to them and further unleashes additional liquidity to bring the interest rate back down (this would especially be the case if the same thing happened to a few banks).

    Ok so now what have we achieved? The interest rate is lower, the effective quantity of money has gone up, and the bad loans are probably still on the books.

    Regarding those bad loans, in more detail, if they are written off, the bank will have to raise permanent capital, which is dilutive. Some loans may indeed get written off, but if the central bank is always willing to provide liquidity, the tendency will be to keep loans on the books.

    So as this system evolves we’ll have more and more bad loans (bogus capitalization), an interest rate which gradually falls lower and lower, and a quantity of money and credit which goes up and up (probably exponentially).

    Maybe this analysis is harsh, but I don’t see how central bank borrowing in a channel system WOULDN’T be associated with a stigma. If something appears wrong, bank peers will rightly shirk a troubled bank (and there will ALWAYS be troubled banks — the more so the greater “safety net” is though to be there for them). This architecture also seems to presume that it’s always good to provide liquidity to a distressed institution, and if they’re fundamentally insolvent, to effectively help recapitalize them. But why should bad institutions be recapitalized as opposed to phased out?

    This seems to be a super-bad idea for a system, culminating in currency collapse (doesn’t matter if it is inflationary or deflationary).

    The only way such a thing could be excusable is if all major central banks implement virtually the same thing, hence making all major currencies “just as bad”.

    If that is the world we are moving towards, then we need to be prepared for perpetually too-low interest rates, a perpetually distressed financial system, and perpetually too-expensive commodities.

  25. Aaron Krowne writes:

    I just thought of a corollary to my point about it being seen to always be desirable to provide liquidity to a distressed institution. The additional thought is that sure, central banks might see fit to NOT save some banks.

    But then what will bank survival be, other than a mostly politically outcome?

    That seems a bad way to run the central sector of a “capitalist” economy.

  26. PrintFaster writes:

    Hi Steve

    When I used the term secrecy, it was in reference to the current context of TAF/TSLF and discount window secrecy that shields the name of the bankrupt banks from public view.

    Think of Bear Stearns. All of the events leading up to catastrophe were shielded from public scrutiny, allowing business to go on with BSC in spite of a horrible balance sheet. Under a transparent system, the public would be the first to know, not the last, and BSC would have been resolved years ago.

    All the machinations of the banks with CDOs and SIVs should have all been in the public view, in a manner that is understandable. You can have secrecy in the open, if the items presented are hideously baroque, and I would classify the CDSs and CDOs as baroque. The public and bankers especially need to demand more transparency in what they enter into, but the grossly high returns cause everyone to look the other way.

    I guess when the CDSs and CDOs were put out there, we should have known that when no one could look at them, that we were dealing with a modern day Medusa.

  27. RFS writes:


    I realize this is relatively small potatoes, but re: the Fed starting to pay interest on reserves, any reason they couldn’t do that just on any excess over the reserve requirement and thereby not relinquish the entire “bank tax”?

  28. RFS — Yeah, they could. That’s one option on the table according to the WSJ article.

    PrintFaster — No argument, especially with respect to fancy securitizations explicitly designed to comfort by hiding “distracting” details from investors. Ironicaly, TAF/TSLF makes the Fed itself very much like a structured credit, a large pool of assets, claimed to be overcollateralized, conservatively valued, and so well diversified that investors needn’t (and can’t) trouble themselves with details of specific components.

    Aaron — In theory, the central bank would only lend to banks in good financial condition at the “ceiling rate”. In practice, well, the current experience makes one wonder, to say the least. However, at this moment, given the Fed’s shrinking Treasury stockpile, I think the Fed would print money to borrow from banks at the floor rate and then lend separately via TAF or TSLF, rather than waiting for banks to ask for funds at the ceiling rate.

    I too dislike the fact that bank failures, at least among large institutions, are now politically chosen examples rather than outcomes of economic decisions.

  29. NickF writes:


    Very interesting post and something I had to check back with other experts (I wrote to Lou Crandall about this and he generally agreed with you except the part about the reverse repo of MBS!).

    Without a reserve requirement, as you mention, the Fed cedes control of the money supply to the banks. There is some similarity to the UK where the BOE allows banks to set their own reserves (but enforces them quite strictly).

    The one thing that has become quite clear about this crisis is the Fed’s coming around to the ECB’s way of doing things: broadening the collateral and counterparties, lengthening the term, adding to its arsenal.

    The ECB had always maintained through the crisis (note past tense) that this issue was a liquidity issue, not a rates issue. And, they felt they had two instruments to control monetary policy, the rate (a minimum bid rate for their weekly auctions) and cash which they could add or subtract (i.e. through deciding how many of the bids to fill/where to set the marginal rate, when to add or subtract during the maintenance period and whether to do fine-tuning). The collateral is clearly getting worse (Govt GC is sub 4% but the Avg rate at the auction is closer to 4.25% now, so clearly banks are not posting govvies) but this is something that the ECB lets the market decide. They claim they are not taking on more credit risk and there is some element of truth since everything they lend out they get back in reserves anyway.

    Of course there are constraints on money supply in order to ensure that EONIA doesn’t drift too much from target.

    So money supply and level of rates, they are obviously highly correlated but now less so and much more clearly disentangled. Your point here is that the Fed also sees them as completely distinct. And they are having different effects.

    I know this is more a macro than a micro-of-the-banking-sector question but shouldn’t this all cause further revision of the standard neo-Keynesian framework? Shouldn’t this be a reprieve for the (slightly more) monetarist position that supply does matter? Do you see any new economic theory along these directions? Neo-Keynesian models are just so much part of the mainstream canon now.

    I know standard RE models which also include level of rates (i.e., Alvarez, Lucas and Weber), use segmented markets to impose the relationship between money and rates, but they too are also quite limited in their applicability.

  30. RueTheDay writes:

    Steve – I agree with your comment about technology folks and the “systems perspective” of designing for redundancy, scalability, fault tolerance, etc. As the old saying goes, a dog is fault tolerant because if it breaks a leg it continues functioning; a horse is not fault tolerant, if it breaks a leg you have to shoot it. So we need to design systems (including central bank systems) to anticipate the worst. The technology metaphor is one of many though; there is much the financial world could learn from biology as well. I believe it was Paul Ormerod, the evolutionary economist, who said that if economists designed humans, we’d only have one kidney and a much smaller liver, and the human species would have likely died out long ago. The point being that economists consider efficiency to be the sole “design goal” and do not consider criteria like resilience or even acknowledge the existence of design tradeoffs. This is not surprising, given that economic models generally assume a tendency toward equilibrium as the foundation of the model itself.

    Aaron – Regarding the elimination of reserves – remember, banks technically only need sufficient reserves to settle their daily net transaction position and cover depositors’ daily withdrawals, which are typically only a tiny fraction of total liabilities. The existence of bad (e.g., non-performing) loans on the balance sheet is a separate issue entirely, and one that (unless things are REALLY bad) doesn’t affect the bank’s daily reserve position. However, this is the absolute crux of the matter – injecting reserves can ONLY solve the liquidity problem (insufficient reserves to clear all daily interbank transactions and depositor withdrawals); there is nothing it can do to solve the solvency problem created by the non-performing loans on the balance sheet.

  31. JKH writes:

    SRW – thanks for your very thorough response.

    Some additional comments (with slight push back):

    As things have evolved, much of the precautionary liquidity function has been handled through liquid assets other than balances at the Fed, as well as guidelines/policies for prudent liquidity management (e.g. term extension on liabilities; cash flow matching; etc.). This requires a proper regulatory and supervisory framework, of course.

    The broader prudential balance sheet management function has been taken over by rules and regulations for capital allocation and management. The primary constraint on bank balance sheet expansion has become an imposition of capital ratios rather than liquidity ratios.

    Given this framework, the need for prudential liquidity in the specific form of Fed balances has declined.

    Accordingly, the primary purpose of Fed balances has been their utility as the exchange mechanism for the interbank clearing system. The question that remains from the banking system perspective is what level of Fed balances is needed to ensure the smooth operation of the clearing system.

    The answer I think is not much, given improvements in cash management technology and techniques.

    The question that remains from the perspective of the Fed is what level of system Fed balances is needed to ensure the Fed’s control over the Fed funds rate.

    Given the Fed’s control over the supply of reserves, the answer is any level that is set as an objective for banks to achieve, that results in banks incurring some penalty if it isn’t achieved.

    We referred to small positive or 0 balances. In theory, bizarrely, this required number could even be a negative system balance – i.e. a borrowed reserve target for the system. There is no point of course to such an aberrant construct, but the logic does suggest a whiff of artificiality in the requirement for Fed balances as prudential liquidity reserves, assuming prudential reserves are well covered otherwise in the maintenance of liquid assets, etc. in forms other than Fed balances.

    As to the reserve multiplier, consider a system that is absolutely continuous in real time, where there is a perfect feedback system (in terms of both information and operational capability) between reserve availability and deposit expansion. I would still maintain that the Fed will supply required reserves, with some marginal excess or deficiency to steer the Fed funds rate appropriately. At the same time, the Fed will consider broader balance sheet developments in the banking system, including deposit expansion, in the context of inflation risks and capital adequacy risks, and take this into account in setting strategy for changing the target Fed funds rate. In other words, to the extent the Fed wishes to influence deposit and balance sheet expansion (apart from capital ratios), it will attempt to do through the strategic management of the Fed funds rate going forward, rather than the blunt withholding of reserves required for the current level of deposits. This is perhaps a fine line, but there is a difference in degree between withholding excess reserves at the margin in order to affect the Fed funds rate directly, and more boldly withholding required reserves for existing deposits in order to affect deposit activity directly. Having said that, while the Fed is looking for an immediate interest rate response when it tightens reserve availability, deposit contraction may well occur as the indirect result of the banks calling in money market type loans. But I think it is the immediate interest rate response more than the induced deposit contraction that is the short term objective of such reserve tightening.

  32. I agree with Print Faster. This whole discussion seems to be “Much Ado About Nothing”. In the end, the Fed will protect the TBTF banks. How? The details don’t matter. Got gold? Get more!

  33. NickF — You ask a lot of good questions, a lot more than I’ve good answers to. (I am, by the way, just a dog on the internet, with no more claim to authority or expertise than my own voice.)

    First: “The collateral is clearly getting worse… but this is something that the ECB lets the market decide. They claim they are not taking on more credit risk and there is some element of truth since everything they lend out they get back in reserves anyway.”

    I don’t think that’s quite right. In the aggregate, if the rate at which ECB pays interest on deposits sufficiently high, then the funds it lend return to it. That in itself is a symptom of financial system distress, though. As JKH pointed out early in the discussion, ordinarily, in a “corridor system” central banks would target the midpoint of its corridor, and conduct open market operations until the level of reserves deposited or funds lent at the ceiling was very low.

    More importantly the borrowers (who by definition have cash needs) and the depositors (who have excess cash) are mostly different institutions. The central bank cannot use the returned deposits as a security against the loans. Should one of the borrowers default and surrender collateral that turns out to be trash, the deposited reserves are still a liability of the ECB that would have to be honored. I need to learn more about how the ECB works, but one way or another I gather that seignorage gains are shared among member-state governments. Should a bank default on the ECB, those gains would have been captured by the defaulter or its creditors, and would represent a cost to European taxpayers.

    I’ve just skimmed through the Alvarez, Lucas and Weber paper. It’s an ambitious little thing, isn’t it? I’d need to think a whole lot more to have much to say on the broad issues you bring up. But it’s worth noting that throughout this comment thread, we’ve basically assumed a segmented model at least superficially similar to the one in the paper. That is, we assume that there are banks and the rest of the world, and central banks target interest rates by matching cash supplied to bank cash demand, and that in equilibrium that somehow affects inflation in the broader economy. The model in the paper suggests that when interest rates are very sensitive to money growth, pegging an interest rate might be a more effective way to hold inflation constant than targeting money growth directly. But the model itself is monetarist by construction — money-growth and inflation track one another over times, it’s just that interest-rate targeting best minimizes short-term fluctuations.

    The model tells a nice story, but it was constructed to tell a story. I’d have to think a lot harder to decide if I think it captures reality or just rationalizes a practice. All that said, I’m not sure I have very much to say about broad disputes between neo-Keynesianism and monetarism, except that I’m skeptical of a lot of things that we’ve mostly taken for granted for the purpose of this discussion. It may not be true that control over cash interest rates is sufficient to control price inflation. “Money” may be a less special asset than we think.

    One of the earliest posts on Interfluidity was an attempt to capture my intuition about moneyless money supply. (You have to click through the “Read more” link to see anything interesting.) If you buy my reasoning in that post, then all assets are to some degree money, and what central banks are up to is trying to offset the losses in value with increases in liquidity in order to hold the total “moneyness” in the economy roughly constant.

  34. JKH — So you’ve lost me just a bit.

    In a world without reserve requirements, where regulation focuses on balance sheet capital rather than precautionary liquidity, in what sense is there any kind of reserve target, positive, negative or zero?

    “Given the Fed’s control over the supply of reserves, the answer is any level that is set as an objective for banks to achieve, that results in banks incurring some penalty if it isn’t achieved.”

    But aren’t we considering abandoning such a world for one in which the Fed manages the supply of money relative endogenous bank choices about the cash and reserve levels they require to conduct business smoothly, rather than by virtue of any target against which the central bank threatens a penalty?

    Okay… writing it out, I think I’m getting it. Even with a reserve requirement of zero, banks that have to borrow reserves for endogenous uses do end up paying a penalty in the form of the discount/ceiling rate spread (or interest premium on auctioned funds). Right now this penalty is quite small, but it needn’t be in general. So even with a reserve requirement of zero, the central bank could manage rates by providing less cash than banks endogenously require, forcing an adjustable level of borrowed reserves that would end up implying an interest rate. If the central bank were to set a negative target, would that imply two lending rates, so that banks could inexpensively borrow upto the targeted level, then pay a penalty rate to borrow more? Weird, but I can see how this would work.

    Still, with non-positive reserve requirements, the system begins to break down as bank money requirements go to zero. If we abandon ATMs entirely for plastic and private interbank settlement networks became the norm, the Fed would lose rate-setting traction as bank demand for Fed money disappeared. Presumably this possibility would be regulated away, but that would mean, in a way, a de facto reintroduction of reserve requirements. If the Fed enforces a monopoly on interbank clearing, and participation therein requires transactional balances, that’s just a reserve requirement by the backdoor, albeit without the fiction of being defined as some fraction of bank deposits.

    In the last paragraph, are you basically arguing that even in a textbook fractional reserve world, the Fed would choose to target interest rates rather than quantity? I wouldn’t argue with that, except to say that in such a world, wouldn’t the two approaches actually be equivalent? It “feels” blunter to target deposit contraction or expansion, but if there really were “a perfect feedback system… between reserve availability and deposit expansion” my sense is that the two approaches would pretty much be the same.

  35. JKH writes:


    I appreciate your penetrating feedback, but I get nervous when smart people say “you’ve lost me”. Also, I’m afraid I’m expanding on my points more than responding adequately to yours. So toward the end of this comment, I’ve attempted some responses more targeted to your points. The overall result is quite lengthy – sorry about that.

    I’ll begin with a simple example for a $ 0 system reserve requirement with a corridor.

    Requirements are $ 0 for the system and for each bank in it.

    Suppose Fed funds trade below target and the Fed wants to steer it back to target (I’ll assume some actual rates below, but it’s not necessary here).

    The Fed attempts this by tightening actual system reserve balances to the level of $ (1).

    This means the excess reserve position is also $ (1) because the required reserve position is $ 0.

    Suppose all banks manage their cash exactly as they’re supposed to, with one lone straggler – Bank A -who doesn’t.

    So we assume all non-A banks will have exactly met their $ 0 requirement by the end of the day.

    Suppose A’s only transaction for the day is a new loan for $ 10.

    (Assume the borrower pays that money to another bank in some other transaction. The example assumes all these other banks have flattened their Fed accounts by the end of the day.)

    A’s balance at the Fed, as a result of the newly extended loan, drops from (yesterday’s) $ 0 to $ (10) today.

    Bank A attempts to fund this shortfall in the Fed funds market.

    Bank A is short $ 10 in reserves. The system by assumption is only short $ 1. This means there must be $ 9 in reserves held by non-A banks in excess of what they require to square their Fed accounts. The example assumes those banks do flatten their accounts, which means Bank A will only be able to attract $ 9 in new reserves through the funds market. But A doesn’t know this. So it continues to bid for Fed funds in search of that final dollar, which doesn’t exist. As Bank A proceeds to have difficulty attracting the final dollar it needs to cover its Fed account, it bids up the Fed funds rate. If the target funds rate is 2 per cent, but it previously traded at 1.875 per cent, assume A’s activity moves the market back to 2 per cent. Bank A fails to find the last dollar it needs in the funds market and goes to the window for $ 1 at a rate of 2.25 per cent, say. The Fed is happy. Mission accomplished – the funds rate trading level has moved back to the target level. Bank A isn’t particularly happy due to the stigma of the window, but that’s life.

    We could also invert the problem to an example of the Fed supplying positive excess reserves: 2.125 per cent Fed funds trading rate, a Fed injected system excess of $ 1, a new customer deposit for A of $ 10, $ 9 in fed funds sold by A at 2 per cent, and a final balance at the Fed of $ 1 dollar, on which the Fed pays $ 1.75 per cent (more on corridor interest rate economics below in response to your points).

    Although the Fed doesn’t use open market operations in its administered corridor, it still requires a method for controlling the level of excess reserves in the system. Since the funds rate will be highly sensitive to any system excess (or deficiency), the level of excess (or deficient) reserves needed for tactical interest rate operations will be very small relative to the size of the Fed’s balance sheet. Such reserve adjustments are easily facilitated using transfers of funds between the government’s deposit account at the Fed and its accounts in the rest of the banking system.

    In this example, I’ve assumed no aggregate changes in the banking system deposit base or corresponding reserve requirements. But if the reserve requirement is $ 0, there is no sensitivity to this in terms of system reserves. When the reserve requirement is $ 0, it is a fixed requirement – hence the disappearance of sensitivity to changes in the deposit base. The examples show the difference between required reserves of $ 0, versus actual (excess) reserves of $ (1) or $ 1 respectively, and effect of that difference on interest rates.

    But the entire exercise of a $ 0 reserve requirement seems pathological. Why would we consider this to be a reserve requirement at all? The question itself invites the point that needs to be made about the positive balance reserve requirement system currently in place. The Fed is supplying excess or deficient reserves for the purpose of marginal Fed funds rate influence. Holding balances at the Fed is not terribly useful or meaningful for precautionary liquidity reserve management. Fed accounts are required as part of the funds clearing mechanism, which in turn allows the Fed to adjust the interest rate at which such funds clear. In other words, it is not the size of the specified level of balances held at the Fed that is important, but the idea that banks become disciplined to target the level that is specified, or incur a penalty interest rate. Banks need good reserve management for cost efficiency purposes and the Fed needs it to control interest rates.

    Finally, consider the ultra-pathological case of a “negative system reserve requirement”. (This is just to reinforce the interpretation of causality, not to suggest that this case is practicable or reasonable.) Instead of $ 0, suppose A’s individual requirement as a share of the system requirement is $ (1). We can just do a linear transformation of the balances in the above examples. E.g. in the Fed tightening scenario, Bank A ends up with $ (2) in its Fed account and goes to the window for $ 1. Causality unchanged, QED. This ultra-pathological case should again illustrate that it is the effective management toward the required reserve level (efficiency of funds clearing) rather than the level itself (precautionary reserve value) that is the essence of the Fed funds system. I believe this is consistent with the secular decline in the actual positive reserve balances that the Fed has required to be held.

    The (theoretical) aspect of continuity is only relevant in a system with positive reserve balances that are a function of the deposit base. It’s moot in the case of a fixed reserve of $ 0, or the pathological fixed negative reserve. With perfect continuity, the Fed reacts to changes in the deposit base by changing the required level of system reserves. Banks continue to compete for their share of those reserves. The Fed will continue to manage the excess or deficient reserve setting according to where Fed funds are trading relative to the existing target rate. Thus, the setting of the differential excess reserve position is independent of this continuity characteristic. It depends only on the level that funds trade at. Banks continue their deposit expansion activity knowing that required system reserves will be supplied continuously, apart from excess reserve adjustments that respond to a temporarily aberrant funds rate level relative to the target rate or a change in the target itself. That is a matter of judgment for both the Fed and the banks’ expectations for the Fed. Banks may well take such expectations into account in their deposit strategies, but rational banks won’t adjust their deposit strategies for fear that the Fed will not ensure that required reserves are available. Concerning the case of a change in the target funds rate, it is important to note that the Fed does not undertake a cumulative permanent withdrawal of required reserves if it increases the target rate. The rate change is a quick announcement following by a trading response and Fed reversion to the objective of managing deviations of the actual rate from the new target through the excess reserve setting. Target interest rate increases do not require a cumulative contraction in supplied reserves. They require only the resetting of the discipline of achieving the level of required reserves in the context of the new interest rate envi

    I could reformulate again the ultra-pathological case by noting that banks currently run intra-day short positions in their Fed accounts as a matter of course. There’s no mathematical reason why such short positions couldn’t become an end of day target, or that the Fed couldn’t set a system short position as the required level for system balances (see more on this in my response to your points below).The fact that today’s target Fed balances are positive is only a reflection of a somewhat artificial deposit ratio calculation that bears little relationship to the way in which banks manage their overarching liquidity strategies, and little relationship to the natural flexibility they have in managing the level of balances in their Fed accounts. The Fed account is merely an important detail in a much bigger liquidity management function for banks, where precautionary liquidity levels are managed through other asset-liability techniques. And again, I’m not suggesting that it makes sense for Fed account reserve requirements to become short positions. The point merely illustrates the nature of the Fed’s interest rate management techniques and their relationship to required Fed balances as being essentially detached from any real characteristic of prudential liquidity reserves.

    Finally, on some of your specific points:

    “In a world without reserve requirements, where regulation focuses on balance sheet capital rather than precautionary liquidity, in what sense is there any kind of reserve target, positive, negative or zero?”

    Liquidity management is still extremely important in its own right, and a vital complement to capital management. It’s just that Fed balances are merely the cash management front end interface for liquidity management. They don’t represent in any way the core strategy for precautionary liquidity. This requires stocks of liquid assets such as t-bills and precautionary term extension of liabilities. This larger liquidity management aspect will be a huge issue in the full post-mortem on Bear Stearns and for the future for the regulation of investment banks. Investment banks really have operated in a different liquidity management paradigm than commercial banks. That said, and on a tangent well beyond this comment, there is a pressing need for a more robust intellectual interface between liquidity and capital management frameworks. Just consider the intensity of the internet debate regarding the distinction between the two in the case of Bear Stearns.

    “Even with a reserve requirement of zero, banks that have to borrow reserves for endogenous uses do end up paying a penalty in the form of the discount/ceiling rate spread (or interest premium on auctioned funds). Right now this penalty is quite small, but it needn’t be in general. So even with a reserve requirement of zero, the central bank could manage rates by providing less cash than banks endogenously require, forcing an adjustable level of borrowed reserves that would end up implying an interest rate.”

    Agreed, I think.

    “If the central bank were to set a negative target, would that imply two lending rates, so that banks could inexpensively borrow up to the targeted level, then pay a penalty rate to borrow more? Weird, but I can see how this would work.”

    This idea was intended as an intuitive illustration, but your question helps me think through the detail. Assume the banking system operates with a required negative balance at the Fed, i.e. a short position, which is a long position from the Fed’s perspective. The Fed earns some specified rate on that long position. I’ll refer to this rate as the “required borrowing rate”, as opposed to the discount rate. The constraint on the required borrowing rate is that it must be consistent with corridor rates that constitute penalties as borrowing and lending rates. The upper corridor rate is the discount rate, which applies to banks that fail to meet their short position requirement and therefore must borrow from the window. Failing to meet their short position requirement means that these banks end up with short positions that exceed their required (i.e. allowable) short positions. The lower corridor rate is a rate which applies to banks that exceed their short position requirement, and therefore have surplus funds to deposit with the Fed. Exceeding their short position requirement means these banks end up with short positions that are less (in absolute terms) than their required short positions, or that are outright long positions. The upper and lower corridor rates should straddle the target fed funds rate so that the upper and lower halves of the corridor are symmetric, allowing the Fed unbiased flexibility to steer wandering funds rates toward target. Therefore, the lower corridor rate should probably mirror the discount rate in terms of its spread against the target Fed funds rate. I used a lower corridor rate of 1.75 per cent in the earlier example, relative to a target funds rate of 2.00 per cent, and an upper corridor discount rate of 2.25 per cent.

    Most interestingly, the “required borrowing rate” could be any rate that exceeds the lower corridor rate. This causes the lower corridor rate to be a penalty when banks with excess balances (relative to required short positions) deposit these balances with the Fed. It results in a negative interest margin on surplus balances held at the Fed. The required borrowing rate isn’t necessarily bound by a necessary relationship to the discount borrowing rate, since the discount rate is always a penalty rate so long as it exceeds the Fed funds rate. But it probably makes the most sense to set the required borrowing rate equal to the Fed funds rate itself, with a symmetry that makes the intended penalty costs in either direction equal. With such a construct, the case of a negative required reserve doesn’t affect corridor interest rate economics or the Fed’s ability to control the actual funds rate under the corridor mechanism.

    “Still, with non-positive reserve requirements, the system begins to break down as bank money requirements go to zero. If we abandon ATMs entirely for plastic and private interbank settlement networks became the norm, the Fed would lose rate-setting traction as bank demand for Fed money disappeared. Presumably this possibility would be regulated away, but that would mean, in a way, a de facto reintroduction of reserve requirements. If the Fed enforces a monopoly on interbank clearing, and participation therein requires transactional balances, that’s just a reserve requirement by the backdoor, albeit without the fiction of being defined as some fraction of bank deposits.”

    Whoa! I’ve been referring only to balances held at the Fed, which are the inter-bank clearing component of the monetary base. Banks do hold currency inventory for precautionary liquidity purposes. I hope I didn’t infer they don’t need to do this. The degree to which they do depends on the evolution of the payments systems for customers. I would classify the economic importance of these currency inventories as something that belongs in a bank’s larger liquidity management framework, apart from any interconnection it may have with Fed mandated Fed account reserve requirements. There is no necessary connection between the formulation of a required Fed balance (as stipulated by the Fed) and the formulation of a required currency inventory (logically determined by some economic inventory model). If the Fed wants to connect these things via reserve regulations, so be it. But the connection isn’t necessary. Currency inventories to satisfy customer demand have nothing to do with the economics of clearing interbank balances through the Fed or the Fed’s control of interest rates (“Orders” or “returns” of currency between the Fed and the banks do affect system Fed balances and individual bank positions, but such flows are integrated into the sterilization process.)

    “In the las
    t paragraph, are you basically arguing that even in a textbook fractional reserve world, the Fed would choose to target interest rates rather than quantity? I wouldn’t argue with that, except to say that in such a world, wouldn’t the two approaches actually be equivalent? It “feels” blunter to target deposit contraction or expansion, but if there really were “a perfect feedback system… between reserve availability and deposit expansion” my sense is that the two approaches would pretty much be the same.”

    Although it seems intuitive, I don’t believe so. Causality depends on differentiability, but not on continuity. Causality refers to the Fed’s propensity to supply required reserves in response to deposit expansion or contraction. Differentiability refers to the Fed’s ability to make marginal adjustments in the form of excess or deficient reserves, in order to affect interest rates through bank competition to sell or buy Fed funds. The size of this marginal adjustment is independent of either the size of required reserves or the continuity with which the level of the requirement changes. Continuity refers to an increase in the frequency of the Fed’s response in specifying required reserve levels as a function of a changing deposit base. Perfect continuity doesn’t invert or neutralize this causality, because banks expect the Fed to accommodate changes in system reserve requirements induced by changes in system deposit balances continuously. They may expect the Fed to change the funds target rate for a variety of reasons, but this is a fundamentally separate operation and decision. The Fed won’t change the target rate with an extraordinary change in the reserve setting per se. They’ll just announce it. There may be some excess reserve noise or anticipatory funds rate noise around the time of a change in the target rate, but the underlying causality of reserves supplied in response to deposit changes doesn’t change. The key is the provision of an excess or deficient reserve differential relative to what the Fed deems appropriate to move the actual funds trading rate toward the target. They might allow the trading rate to begin to move toward an expected new target prior to the precise time of the target change, but this is just a circumstantial modification in the actual/target rate relationship. The matching of the required relationship between deposits and reserves might become continuous, but the decision on the excess reserve position is independent of this underlying continuity. The economics around the differential alone motivate the iterative interest rate adjustments desired by the Fed.

  36. flow5 writes:

    The adoption of a Lombard facility is no exception. The Fed cannot control interest rates, even in the short-end of the market, except temporarily. The effect of a series of temporary pegging operations (the one-day cost-of-carry on Government bonds), as guide posts, is indirect, and varies widely over time, and in magnitude. And cost, obviously, is not a significant factor determining the volume of borrowing (see Jan 82 – Jan 83).

    Why inflationary? The effect of tying open market policy to a FFR target (or penalty rate) is to supply additional (and excessive) legal reserves to the banking system when loan-demand increases. Since the member depository institutions have no excess reserves of significance, the banks have to acquire additional reserves, to support the expansion of deposits, resulting from their loan-expansion.

    If they use the Federal Funds market, the rate is bid up and the “trading desk” responds by putting through an excessive volume of (1) Open Market Purchases, (2) Reserve Bank Credit, and (3) Legal Reserves.

    Soon a multiple volume of money is created on the basis of any given increase in legal reserves.

    And since the “money multiplier” has been emasculated, what the net expansion of the money supply will be – won’t be known until long after the fact. This is the process by which the Fed has financed the rampant real estate speculation which has characterized the 3d millennium.

  37. flow5 writes:

    The crux of the cause of our monetary mismanagement, esp. since 1965, is the assumption that the money supply can be managed through interest rates, specifically the federal funds rate (“bracket racket”). As any monetarist knows, it is impossible to control properly the money supply through the manipulation of interest rates.

    We should have learned the falsity of that assumption in the Dec. 1941-Mar. 1951 period. That was what the Treas. – Fed. Res. Accord of Mar. 1951 was all about.

    The Fed’s technical staff either never learned, or forgot, how Roosevelt got his “2 percent war”. This was achieved by having the Fed stand ready to buy (or sell) all Treasury obligations at a price which would keep the interest rate on “T” bills below one percent, and long-term bonds around 2-2 1/2%, and all other obligations in between.

    This was achieved through totalitarian means, involving the control of total bank credit and the specific rationing of that credit. Plus there were controls on prices and wages that kept the reported rate of inflation down.