Thursday, February 11, 2010

The Fed's Exit Strategy

So Ben Bernanke confirmed that the Fed may have to turn to interest paid on excess reserves as its main policy tool rather than targeting the federal funds rate. Bernanke says such a move would only last until normalcy returns to the Fed's balance sheet or, equivalently, excess reserves return to their pre-crisis levels. Mark Thoma provides an informative discussion of this process that shows how the discount rate, federal funds rate, and interest paid on excess reserves interact in a supply and demand graph for bank reserves. Here is what some Wall Street economists had to say about this potential change in Fed policy. And here is what Josh Hendrickson had to say on the matter:
As I previously discussed, the interest on reserves methodology is a rather crude way to solve the problem. If the problem is with excess reserves, then the reserves should be removed from the system using normal open market operations. So why isn’t the Fed employing this method? Well, I have long suspected that the reason the Fed was employing this strategy was because of the change in the composition in the Fed’s balance sheet away from traditional Treasury holdings and toward mortgage backed securities. This view is confirmed in the WSJ:

Plans for the Fed’s portfolio of mortgage-backed securities are another element of the internal debate over the exit strategy from super-cheap money. The Fed is on course to buy up to $1.25 trillion of the securities, in an effort to hold down mortgage rates and buoy housing.

Over time, officials want to reduce these holdings and return to holding U.S. Treasury securities as the Fed’s primary asset. But they are reluctant to take steps that might push mortgage rates higher and damage the still-fragile housing market. Eventually, they could gradually sell mortgage securities, but such a move would be unlikely in the early stages of tightening.

So, ultimately, the Fed is conducting fiscal policy by subsidizing mortgage rates. What’s more, given that open market operations would necessarily require not only open market sales of Treasury securities, but also mortgage backed securities, the Fed finds itself in a position in which open market operations are politically and practically infeasible.


  1. The issue regarding the “choice” between reserve interest and the funds rate as the policy target is a total red herring. The only reason for choosing (temporarily) reserve interest as a target is to quell the anxiety of those who don’t understand the reason for volatility of the effective fed rate around the fed funds target. RDQ economics identifies the primary reason for the discrepancy in your link. GSEs don’t earn interest on their Fed balances. Also, market arbitrage between the effective fed rate and reserve interest is less than perfect where participants are concerned about non risk weighted leverage optics.

    It is a fact is that fixed reserve interest and a fluctuating effective fed rate will co-exist so long as excess reserve balances remain outstanding and the two factors noted above continue. Whether you announce a fixed reserve interest target or a fed funds target doesn’t matter a wit to what actually happens in the market. It’s pure theatre to acquiesce those who don’t understand the reason for effective fed volatility.

    As far as the issue of draining excess reserves, or announcing targets for doing so, this is also irrelevant and is again being done for theatre. Banks do not take on non-zero risk weighted assets (e.g. loans) as a substitute for zero risk weighted assets (e.g. excess reserves), not without referencing capital criteria that are completely independent of reserve conditions. Banks expand their risk according to capital allocation discipline and capital adequacy. Reserves have nothing to do with it. Banks operationally do not lend reserves to non-banks, and banks do not lend as a function of their current reserve position in any event.

    The purpose of reserves for banks is that of settlement balances. The additional purpose from the central bank’s perspective in this particular environment is that excess reserves have provided a useful liability management option for a central bank that is engaged in credit easing and asset expansion. That particular central bank purpose has nothing to do with their usefulness for commercial banks. From that perspective, excess reserves in this environment are simply a risk free asset that is forced onto the collective balance sheets of the commercial banking system.

    The idea that excess reserves pose some “threat” for overly aggressive bank lending is simply an erroneous application of the textbook “multiplier” theory, which itself is hogwash. Banks lend according to the availability of credit worthy customers and the adequacy of their own capital. Banks do not lend reserves and they do not lend from reserves. Reserves are a settlement mechanism and nothing else from a commercial banks perspective.

    The Fed is engaged in PR theatre in order to quell the fears of those who misunderstand the nature of monetary operations.

  2. JKH's comment sounds remarkably like a variant of the efficient markets hypothisis.

  3. JKH's excellent comments have nothing to do with efficient market hypothesis. Even the BIS (the central bank's central bank) recognises that the manner in which JKH describes banking operations are correct. A recent BIS paper ("Unconventional monetary policies: an appraisal") discusses the reserves – bank lending – inflation nexus and concludes the following:

    "[A]n expansion of bank reserves endows banks with additional resources to extend loans, adding power to balance sheet policy. Second, there is something uniquely inflationary about bank reserves financing."

    The BIS correctly points out that those who think that an expansion of bank reserves provides banks with additional resources to extend loans assumes that “bank reserves are needed for banks to make loans”. Accordingly, mainstream economists (including, it appears, Ben Bernanke, although he should know better) think that the “bank lending is constrained by insufficient access to reserves or more reserves can somehow boost banks’ willingness to lend.”

    The BIS authors go on to say that:

    "An extreme version of this view is the text-book notion of a stable money multiplier: central banks are able, through exogenous variations in the supply of reserves, to exert a direct influence on the amount of loans and deposits in the banking system."

    JKH rightly rejects these propositions. Bank reserves are not required to make loans and there is no monetary multiplier mechanism at work as described in the text books.

    The BIS authors concur and say that:

    "In fact, the level of reserves hardly figures in banks’ lending decisions. The amount of credit outstanding is determined by banks’ willingness to supply loans, based on perceived risk-return trade-offs, and by the demand for those loans. The aggregate availability of bank reserves does not constrain the expansion directly."

    Bill Mitchell notes why this is obvious why this is the case. He has illustrated repeatedly that loans create deposits which can then be drawn upon by the borrower. No reserves are needed at that stage. Then, as the BIS paper says, “in order to avoid extreme volatility in the interest rate, central banks supply reserves as demanded by the system.”

    The loan desk of commercial banks have no interaction with the reserve operations of the monetary system as part of their daily tasks. They just take applications from credit worthy customers who seek loans and assess them accordingly and then approve or reject the loans. In approving a loan they instantly create a deposit (a zero net financial asset transaction).

    The only thing that constrains the bank loan desks from expanding credit is a lack of credit-worthy applicants, which can originate from the supply side if banks adopt pessimistic assessments or the demand side if credit-worthy customers are loathe to seek loans.

  4. JKH:

    In economics, there is something called unintended consequences. Banks may think that they are lending based upon the credit worthiness of their borrowers and their capital conditions, but the unintended consequence of their actions might be something quite different. The argument that "banks don't see it that way," suggests a basic failure to understand economic reasoning.

    If banks fund loans with nonmonetary assets, this is irrelevant to macroeconomics. If banks sell CDs and use them to fund loans, then there is an increase in bank credit and a decrease in whatever those buying the CD's would have purchased instead. For example, less credit provided to the government, if people buy CD's instead of government bonds.

    Now, I am sure the allocation of funds between various parts of credit and financial markets have many important effects. Like, for example, the allocation of consumer expenditure between the washing machine and restaurant industry.

    If, on the other hand, banks fund loans with checkable deposits, those receiving the checkable deposits don't necessarily want to hold them. That is because they serve as the medium of exchange.

    In equilibrium, of course, they must want to hold them, but the whole point is the possibility of disequilibrium--an imbalance between the quantity of money and the demand to hold it.

    The Fed believes it can deal with this issue by making gradual adjustments in short term interest rates. Well, not always as we have seen over the last two years.

    But aside from these simple truths about monetary theory, I am puzzled-- who could possibly care about variation of the effective federal funds rate relative to the target? Who cares if the Fed has control over the Federal Funds rate? Why?

    It is evident that the Fed is quite concerned about its control over the Federal Funds rate (and presumably other short and lower risk interest rates.) Why is this so important to them?

  5. JKH and Marshall:
    I agree the capital adequacy and credit worthiness are important factors in banks making loans. But are they the only factors? Along this line, here are few questions:

    (1) Is it not possible for the Fed to exogenously change bank reserves? That is, why can't the Fed suddenly change the quantity of reserves beyond the point warranted by the existing bank demand for reserves? If so, what happens next?

    Imagine for example, Bernanke unexpectedly drops me a million dollars from his helicopter. I then take it to my bank and deposit the cash. I now have a million dollar deposit and the bank has a million dollars in reserves. Why wouldn’t some of these reserves be turned into loans?

    (2) If bank reserves are inconsequential to bank lending, then this would imply the Fed could pull out all reserves in the banking system and lending would not be affected. Do you believe this?

    (3) If bank reserves are inconsequential to bank lending , then how do we explain the 1937-1938 recession that almost everyone believes was caused by the Fed increasing the required reserve ratio to offset the increase in bank excess reserves? For that matter, why does a required reserve ratio matter at all?

    (4) There is a large empirical literature that bank reserves matter to total spending in the economy. From Friedman and Schwartz'classic to the numerous vector autoregressions that use nonborrowed reserves, there is a lot of evidence that shows that for the importance of bank reserves. Note, that many of these studies look at movements in bank reserves controlling for other changes in the economy. In short, they show non-forecasted or unexpected changes in reserves have real economic effects. How do you handle these studies? Is there some other story beside the money multiplier story one can tell that links the unexpected bank reserves growth with real economic activity?

    (5) Along the lines of (4) is there any systematic evidence for the JKH and Marshall view?

    (6) You make the case that the bank system only creates more loans in response to a change desire for them by credit worthy borrowers. But what is driving this changed desire for loans? Is it a change in demand for loans or a change in the quantity demanded of loans? This is an important distinction. Your story makes sense for the former case: if the economy is booming and the demand for funding increases then the banking system would accommodate this change. This would be a change in demand for loans. The Fed had nothing to do with this. However, what if the Fed unexpectedly lowered interest rates (by increasing bank reserves unexpectedly) which signaled to folks sitting the fence that borrowing costs were now low enough to take out a loan? Here the total quantity of loans change based on their price, not because incomes got higher, risk preferences changed, or some other outside factor. My impression is that you are assuming the change in loans is only driving by a change in demand not a change in quantity demanded. Why can't both be true?

  6. David,

    I would argue that the 1937 relapse was much more a function of tighter fiscal policy. If you recall, that was the year that taxes on Social Security were introduced for the first time (another tragic case of mistakenly using taxes to "fund" a social program). The President, having been advised by his Treasury Secretary that the recurrent budget deficits of his first term were "bankrupting" the country, duly listened and submitted a balanced budget in 1937 (sound familiar?). Anyway, we all know how well that ended up.
    On your other point, yes, if the FDIC were to exercise some regulatory forbearance (as the Japanese FSA is doing right now on its banks), then that could affect the lending postures of the banks insofar as they are less capital constrained. But that doesn't change the fundamental position: if banks can't find creditworthy customers, then won't lend and borrowers are unlikely to take on more debt if they can't service existing debt. Which brings us back to the role of fiscal policy. Fiscal policy ought to be designed to enhance CREDITWORTHINESS, not "unblock" credit flows. Credit follows from improving creditworthiness. So easy that even a banker can figure it out.
    Is there any systematic evidence for our view? Well, not if you read the Economics 101 textbooks, but if you ask any banker how it's really done, then yes, there is ample evidence to support our description. You can also read books by people like L. Randall Wray, "Understanding Modern Money", as this gives systematic evidence. But ask any banker today about a "money multiplier" and he'll tell you (if he's honest) that the idealised models are nonsense. In the real world banks make loans independent of reserve positions, then during the next accounting period borrow any needed reserves. The imperatives of the accounting system, as previously discussed, require the Fed to lend the banks whatever they need. Bank managers generally neither know nor care about the aggregate level of reserves in the banking system. Bank lending decisions are affected by the price of reserves, not by reserve positions. If the spread between the rate of return on an asset and the fed funds rate is wide enough, even a bank deficient in reserves will purchase the asset and cover the cash needed by purchasing (borrowing) money in the funds market. This fact is clearly demonstrated by many large banks when they consistently purchase more money in the fed funds market than their entire level of required reserves. These banks would actually have negative reserve levels if not for fed funds purchases i.e. borrowing money to be held as reserves. If the Fed should want to increase the money supply, devotees of the money multiplier model (including numerous Nobel Prize winners) would have the Fed purchase securities. When the Fed buys securities reserves are added to the system. However, the money multiplier model fails to recognize that the added reserves in excess of required reserves drive the funds rate to zero, since reserve requirements do not change until the following accounting period. That forces the Fed to sell securities, i.e., drain the excess reserves just added, to maintain the funds rate above zero. If, on the other hand, the Fed wants to decrease money supply, taking reserves out of the system when there are no excess reserves places some banks at risk of not meeting their reserve requirements. The Fed has no choice but to add reserves back into the banking system, to keep the funds rate from going, theoretically, to infinity.

    In either case, the money supply remains unchanged by the Fed’s action. The multiplier is properly thought of as simply the ratio of the money supply to the monetary base (m = M/MB). Changes in the money supply cause changes in the monetary base, not vice versa. The money multiplier is more accurately thought of as a divisor (MB = M/m).

  7. Bill Woolsey:

    Money does not sprout wings and fly away to spend on its own. The Fed’s willingness to expand its balance sheet in the face of deflationary pressures is one acknowledgement of this fact. Its actions were intended to offset a pace of deleveraging in the private sector that was abrupt enough to be lethal.

    Bank lending creates deposits. Government spending creates deposits and bank reserves in the first instance, which governments then typically drain by issuing bonds. The idea of “loanable funds” and “crowding out” is completely wrong when viewed against the reality of how balance sheets actually come into being.

    The Fed controls the policy rate because that is the way that monetary policy works in the modern world.

    As I explained already, I’m not convinced that the Fed is all that concerned in this case about tracking error in the effective fed funds rate, relative to the target. The sources of that tracking error are not strong enough for it to become a very material issue, in my view, and probably in their view. Still, those pundits and participants who don’t understand these things are still capable of becoming counterproductively noisy about it. So a policy announcement framework that captures the importance of interest on reserves is worth considering, so long as interest on reserves remains critical to the Fed’s interest rate control architecture.

  8. David,

    My answers:

    (1) The bank has a risk free asset and your deposit liability. That has no bearing on the bank’s capacity or willingness to make a new loan. The bank will make a new loan (which creates a new deposit) based only on the availability of a credit risk it wishes to assume and whether or not it has the capital to assume that risk.

    (2) Provided that the banking system was adequately capitalized and that the settlement system was operating smoothly, yes, the Fed could pull out almost all EXCESS reserves. In normal (i.e. historic) circumstances, the Fed has run a very modest excess reserve position of about $ 10 billion, in order to alleviate operational frictions that might otherwise push up the funds rate. The current excess is about $ 1.1 trillion by comparison.

    (3) A required reserve ratio does not matter, in that sense that reserves are not required for lending. It does matter in the sense that uncompensated required reserves are a tax on bank net interest margins. Therefore, required reserves affect the pricing of loans and deposits (the absorption mix is up to the banks). But required reserves do not affect the amount of credit that can be extended, provided capital is available to assume the risk. The taxation pricing effect of reserves is different than the “multiplier” quantum constraint assumed by (erroneous) theory.

    (4) , (5) Don’t know.

    (6) The Fed controls the policy interest rate. That is supposed to have some effect on the demand for credit. BTW, the Fed does not need to change bank reserves in any material or permanent way in order to change the policy rate. The mere announcement will do it. That’s because the market knows the Fed can overpower any potential market resistance by changing reserves. So the market arbitrages that contingency immediately. If that weren’t the case, you’d have been able to observe discernable trends in excess reserves pre autumn 2008, and that isn’t the case.

  9. JKH,

    (1) You say in response to my Q1:

    "The bank has a risk free asset and your deposit liability. That has no bearing on the bank’s capacity or willingness to make a new loan."

    Really? Take two identical banks with the same balance sheets, capital cushions, and credit worthy borrowers. Now suddenly one gets that million dollar deposit from me. I find it hard to believe the bank that just got the million dollars in reserves will continue to act the same as the other bank. Yes, it will need to consider capital,risk, and loan demand issues, but at least it is now in a position to use those excess reserves to earn a return unlike the other bank. It may not do so, but it might.

    (2) You acknowledge the fed controls the policy rate and that affects the demand for loans. However, the Fed controls the policy rate by adusting bank reserves. Take the case of lowering the policy rate. This requires creating more bank reserves. The low rates also signal borrowers to acquire more loans. The banks use the newly acquired excess reserves to make the loans being demanded. Note the flow: Fed creates more reserves => rates fall=> quantity of loans demanded increase. I am sure the banks could find other ways to finance these new loans, but surely some of them are created using the excess reserve.

    Yes, I know the causality can also go the other way (loan demand goes up => rates increase => fed accommodates ). But it doesn't rule out an exogenous increase in bank reserves by the Fed.

    (3)The big question behind our discussion is this: will the excess reserves lead to more lending once the economy recovers? In other words, once banks have better capital cushions and borrowers are better credit risks and seeking loans, what happens to all those excess reserves? It seems to me that if the Fed leaves them in the banking system they get turned in loans. You disagree. Tell me what you believe would happen to them in such a case (i.e. the Fed doesn't pull them out). Do they just sit idly in the banking system?

  10. Marshall:

    You say "Changes in the money supply cause changes in the monetary base, not vice versa." Again, I would agree with that to some degree. Some of th money supply is endogenously determined. But it cannot all be so. For example, it were always true that changes in the money supply cause changes in the monetary base then hyperinflation would never be the fault of the government. (Robert Mugabe would be victim not a pertpetrator!)Or take the example of FDR in the Great Depression. He devalued the dollar-to-gold ratio and allowed the Hitler-induced gold flows coming from Europe not to be sterilized. These two measures significantly increased the monetary base. The money supply just so happened to increase after this development too. It would be hard to argue here that the increase in the money supply caused the gold supply to increase which increased the monetary base. More recently, look at Paul Volker's bloody recessions in the early 1980s. Did the slowdown in the monetary base growth result from a slowdown in M1? The standard story is no--he purposefully engineered a recession by tightening monetary policy.

    Again, I agree changes in money demand can influence the total quantity of money, but so can exogenous policy changes in the monetary base.

  11. David,

    Excess reserves are risk free assets. Risk free assets earn the risk free rate of interest.

    Risky assets earn the risk free rate plus an expected spread. The purpose of capital is to underpin the risk around the expected spread.

    Banks must allocate capital to risk. Therefore, the decision to take risk depends on the availability of capital.

    A bank with adequate capital is itself an acceptable credit risk. That means it does not have a problem in attracting reserves required for settlement purposes. It can do this in the normal course by selling liquid assets or attracting new liabilities.

    Therefore, the risk taking exercise does not depend on the availability of a stock of reserves. Any bank that formulates a strategy of risk taking based on reserves is simply being irrational and reckless. The textbook multiplier theory is an outrage in terms of its ignorance of risk and capital issues.

    The Fed controls the supply of aggregate system reserves. When individual banks do engage in new lending, the best they can do is push the aggregate reserve distribution in the direction of other banks as a result. That is simply the result of the settlement mechanism. It does not reflect a process of “using” reserves.

    Any bank CEO that would formulate a strategy of risk taking primarily in an effort to “use” reserves would be foolish. The best the bank could do is contribute in a minor way to the conversion of system excess reserves to system required reserves. The effective reserve ratio of the system prior to the crisis was less than one per cent, due to the preponderance of non-reservable time deposits in the total checking/savings/time deposit mix. Based on that mix tendency, the US banking system would have to create more than $ 100 trillion in new loans and deposits in order to “use up” excess reserves. This line of thinking is just preposterous, and those who pursue it should spend more time on their numbers. Moreover, any strategy of risk taking designed simply to push existing system reserves over to the competition, regardless of normal risk analysis and capital considerations, is just outrageously reckless.

    The proper strategy has to be risk and capital based on a loan by loan basis. So yes, I do expect those banks to behave in the same way.

    .... continued

  12. .....

    There is one other important pricing issue. Given that the system has $ 1.1 trillion in excess reserves and that the system has it until the Fed decides to take it away, you can bet that the system is being careful in pricing its deposits at the margin to ensure that margins on risk free reserves are no narrower than necessary in the circumstance. But they have to consider issues like depositor relationships in doing that.

    David, you’ve not read my previous comment if you think the Fed controls the policy rate by adjusting bank reserves. That is not correct in more than a momentary sense. Have a look at the historic series on system excess reserves prior to 2008. It is absolutely stable. The reason is due to the arbitrage pricing dynamic I mentioned in my previous comment, whereby traders move all market rates in sympathy with any fed funds announcement. There is no way that the Fed leaves new excess reserves in the system for any material length of time when easing. They’re gone within a matter of days. That has no operational effect on bank credit and capital strategies per se, other than short term money market type activity.

    Banks do not “use” reserves to create new loans. Banks for one thing do not lend reserves to non-banks. Reserves are for the exclusive access of the banking system, not its clients. Banks do not “lend” reserves; they use reserves to make PAYMENT through the interbank settlement system for client transactions. Payment is not lending.

    The Fed is the one who will decide the destiny of excess reserves. In part, you’ve answered your own question there. So long as the Fed leaves them in the system, they sit there by definition as a system aggregate. They get pushed around a bit due to normal settlement imbalances and various asset liability strategies of banks. But they sit there for the system as a whole by definition. The only change might be a tortoise like conversion of excess to required reserves. But assuming the banking system retains its current mix of desired deposit characteristics as between checking, saving and time deposits, and its current reserve ratios, and assuming the nominal banking system balance sheet grows by 7 per cent a year, I estimate it would take roughly 45 years to convert $ 1.1 trillion in excess reserves to required reserves. Such is the absurdity of the multiplier argument in this or any other environment.

    Finally, the Fed won’t leave the excess reserves in forever. But the reason it won’t leave them in is not as expected in the conventional wisdom. There is no issue with respect to overheating of lending due to excess reserves. The reason the Fed will withdraw them is that the banking system won’t need the Fed’s credit risk taking support any more. The reason for the excess reserves in the first place is because of the Fed’s intervention in the form of its extraordinary asset initiatives, which created the reserves as a by product. That intervention was designed to relieve the banking system of some of the credit risk burden that it normally assumes. As the economy gains better footings and as the banking system becomes better able to function as it normally does, the Fed will withdraw the reserves. The exit strategy is an asset strategy, with the corresponding reserve effect as a by product. Its exactly the same causality as was the case with the Fed’s “entrance strategy”, where asset expansion created the corresponding reserves.

  13. David,

    I agree with JKH that Excess Reserves are essentially meaningless to lending. If the Fed decided to pull out all ER's tomorrow, and THEN the banks decided they wanted to lend $1tr more to clients, then banks would simply do so and the Fed would provide the $1tr in needed reserves to the system at a zero percent rate. The level of Excess Reserves has no bearing on lending in a Fed Funds targeting system because the Fed supplies any funds necessary to keep the target rate from rising. Banks should be indifferent as to whether reserves come from existing Excess Reserves or from new reserves supplied by the Fed to accommodate increased lending by the system at the target rate. In this sense the whole discussion of interest on reserves is a red herring: at any given Fed Funds target rate, it doesn't matter what the level of Excess Reserves is.

    Excess Reserves are meaningful as an artifact of past Fed policy. That is, they fund the Fed's acquisition of MBS. The question is, what will happen to long term rates when the Fed unwinds this "trade"? Perhaps more important is this question: what will happen to l.t. interest rates once the Fed completes its MBS purchase program? Not enough people recognize that on the margin, the cessation of purchases represents an end to the subsidy of long term rates, and therefore constitutes, on its own, tighter -- and not "neutral" -- monetary policy. To make a long story short, the problem with ER's occurs when you stop accumulating them to finance a subsidy to the long end of the curve while housing is still in a fragile recovery.

  14. David Pearson,

    I concur with the general thrust of your comments, but offer a couple of technical qualifications.

    The banks won’t need that $ 1 trillion in reserves in order to lend $ 1 trillion. They create deposits by lending. From there it’s just a matter of managing the reserve distribution that results from such activity. Individual banks achieve their desired share of aggregate reserves on a daily basis through normal asset liability management procedures. If an individual bank lends $ 1 billion, it can sell liquid assets or raise deposits to offset any interbank settlement and reserve effect associated with the lending. Then as you say it is merely a matter of the Fed maintaining the system excess reserve setting at a level that ensures a reasonably steady fed funds level. Prior to the crisis, that aggregate system excess was typically less than $ 2 billion (I checked the numbers, and was overly generously on my previous estimate of $ 10 billion). That’s a $ 2 billion fulcrum for a system with $ 7 or 8 trillion in deposits.

    The problem is that the word “reserves” itself is a misnomer, a vestige of gold standard thinking. The purpose of reserves from the banking system perspective is to provide a system of interbank bank payment and settlement for asset liability management transactions of all types. It is not to provide an inventory of some sort to prepare for new lending. The most obvious evidence of this is those countries that moved to zero reserves some time ago, e.g. Canada. The normal purpose from the central bank perspective is to facilitate interest rate control via settlement balance conditions. That technique has morphed into a broader liability management tool to accommodate central bank asset expansion in a credit crisis. In this sense, reserves during the crisis have become far more important to the central bank as a liability than they have to the banks as an asset. That (temporary) change has required the Fed specifically to modify its architecture for interest rate control through reserves by paying interest on a very bloated excess position.

    And you’re right that at this point reserves are the Fed’s liability management tool for accommodating MBS purchases. And its primary objective in those purchases is to provide interest rate support to that market (i.e. lower rates). In that sense, the risk that the Fed has absorbed into its balance sheet at this stage has a lot to do with interest rate risk, in addition to the residual credit risk that still exists in this late stage of some of the previous programs.

  15. JKH,

    Thanks for the clarification.

    BTW, one answer to David Beckworth's third question above is that the ER's in 1937 were essentially "precautionary balances" -- liquidity reserves the banks wanted to maintain to survive future runs. When the Fed raised reserve requirements, banks may have wanted to maintain precautionary balances at those levels, which means that the increase in RR's had to come from loan liquidation. This is a case of "asymmetry": the desire to hold a fixed level of precautionary balances means the Fed can cause lending to decline by increasing reserve requirements, but not rise by lowering them -- at any given level of the Fed Funds rate and underwriting standards.

  16. Some fascinating insights into the monetary mechanism in these comments. I think we are seeing an all-out assault on the monetarist framework espoused by the likes of Friedman and the modern-day disciples like Bill Woolsey and Scott Sumner in the US and the likes of Tim Congdon in the UK. They are seeing their paradigm systematically refuted by the modern monetary theorists like Bill Mitchell and Randall Wray. Almost every single thing that is taught by the monetarists is wrong, from the money multiplier model of money supply to the theory that inflation is caused by exogenous increases in money growth engineered by the central bank.
    Actually anyone who paid attention knew this was wrong even before the MMTers. First there was Friedman's hopelessly wrong predictions about inflation resurgence in the 1980s. And I remember a careful econometric paper by David Hendry in the 1990s modeling UK inflation dismissed the monetarist hypothesis.

  17. Got here through JKH's comment as Moslers .. Good thread.

    Basil Moore's textbook Horizontalists and Verticalists says it all: Quoting the opening of that book (with the author's permission)

    The central message of this book is that members of the economics profession, all the way from professors to students, are currently operate with a basically incorrect paradigm of the way modern banking systems operate and of the causal connection between wages, prices, and monetary developments, on the other hand. Currently, the standard paradigm, especially among economists in the United States, treats the central bank as determining the money base and thence the money stock. The growth of the money supply is held to be the main force determining the rate of growth of money income, wages, and prices.

    ... This book argues that the above order of causation should be reversed. Changes in wages and employment largely determine the demand for bank loans, which in turn determine the rate of growth of the money stock. Central banks have no alternative but to accept this course of events. Their only option is to vary the short-term rate of interest at which they supply liquidity to the banking system on demand. Commercial banks are now in a position to supply whatever volume of credit to the economy their borrowers demand

  18. Sorry for the delay folks. First, my reply to JKH:

    You obviously know a lot about this issue and have forced me to think more about it, but I am not convinced by your story. First, I find it hard to believe that two banks would continue acting the same after the unexpected reserve increase for the one bank (i.e. my Bernanke helicopter drop story from above). Yes, they may both be able to get bank reserves for settlement purposes anyway, but now one bank has a new deposit account and excess reserves to manage. In other words, one bank now has explicit costs in managing the new deposit account as well as the opportunity costs of excess reserves sitting around idly versus earning a return. Yes, the banker has to manage risk but he/she also has to manage these new costs. How can the two banks possibly act the same thereafter?

    Second, you claim bank reserves only are important for settlement purposes, but settlement issues themselves can influence how much lending a bank can do. For example, a bank that has insufficient reserves to meet settlement will be less likely to issue new loans (and may even call in existing ones). That bank reserves are used for interbank payments which in turn may influence bank lending amounts to bank reserves being indirectly capable of influencing lending. (See this NY Fed study for more on this point).

    Third, I agree that the Fed controls total bank reserves, but I am surprised at your claim that excess reserves turn into required reserves at a tortoise pace. I thought excess reserves in normal times were minimal as the banks would quickly get rid of them given their opportunity cost. I am even more surprised at your claim of it taking 45 years to convert the current excess reserves into required reserves. Are you using numbers that would hold in normal times or crisis times? The distinction is important since it is a fear that the excess reserves will lead (indirectly) to more loans once conditions return to normal. (BTW, I think a better indication of the Fed's potential influence on banking lending would be to look at non-borrowed reserves not excess reserves.)

    Fourth, there is a large literature on how monetary policy affects the broad economy. There are many transmission mechanisms or channels (see here for overview). One of them is called the the bank lending channel which basically says that monetary policy can drain reserves from the banking system which in turn leaves fewer funds to lend . Now this is not the most important channel but many empirical studies have found it to be important for smaller banks with less access to other funding sources. The big point here that there is a large literature with lots of empirical evidence behind it. (Here is one recent study; there are numerous others.) All I have heard from the skeptics so far are stories but no systematic empirical evidence. It is hard to convince me just with stories.

    Fifth, your original assertion that all of these concerns about the threat of excess reserves is just theater is simply wrong. Most economist believe it. There is abundance of empirical evidence to support it (see above). Bernanke, for one, believes it because he wrote the book on the bank lending channel. There is no conspiracy here to give the misinformed masses what they want to hear. You could make the case that most economist are simply wrong in their view but I would wrestle with the empirical evidence first.

  19. ECB:

    Et tu, Brute? You have gone and joined the other side? I guess I shouldn't be too surprised given your plugs for endogenous money and the RBC theory in an earlier post.

  20. ECB & Ramaan:

    Are you sure you want to claim all money supply changes are endogenous and that the central bank simply responds to the economy? If you believe that then (as I told Marshall) you also must believe Robert Mugabe was a victim of hyperinflation not the perpetrator of it.

  21. I'd like to thank all the commenters on this thread for their insightful and civil discourse. Terrific stuff.

    Special thanks to JKH who is the man, both in terms of his clarity of explanation and patience (not in this thread so much, as all discussion has been very mature, but I have read him, in other places, carefully debating these concepts with plenty of dismissive snark coming back).

  22. Also, JKH and Marshall Auerback (and others), I wonder whether it would be a useful endeavour at this point to engage in the "money multiplier" wikipedia page discussion? Would think this is likely the most effective method for informing layfolk, such as myself, of these facts. There is a "Loans first" segment on the page, but getting "above the fold" would be likely create a large shift in awareness.

  23. Yes, for the most part I think it is correct as a first approximation to model the money supply as endogenous. The private sector creates the money supply it needs to finance a business cycle expansion. But clearly central bank base money is important as part of the payment and clearing system, so the central bank's role as lender of last resort has a part in the story. Episodes such as the 1980-83 recession mean we cannot dismiss the ability of the central bank to affect the level of real activity via interruptions to the financial intermediation process. Exactly how this occurs is still very much an open question.
    The chartalists such as JKH and Bill Mitchell and the real business cycle guys don't have all the answers, but at least they are forcing us to question conventional outmoded thought patterns.

  24. ECB, because central banks can create Volker-type disruptions or hyperinflation one should never rule out exogenous policy changes to the money supply. I get the impression, though, that these folks assume these possibilities away.

  25. JKH and others:

    Just to be clear, the bank lending channel is probably not the main transmission channel most mainstream economists have in mind when thinking about the threat of excess reserves creating inflation down the road. Many, I suspect, believe the excess reserves if not removed would eventually create higher inflation expectations which would lower real interest rates that in turn would lead to more borrowing. This interest rate channel would work by increasing the quantity of loans demanded and the excess reserves would be there to (indirectly) support the loan expansion. There are other stories one could tell (e.g. balance sheet channel, exchange rate channel, wealth effect channel) but regardless the key poiont is that the Fed's buildup of reserves would either directly or indirectly lead to the expansion of the money supply and ultimately inflation.

  26. Along the lines of whether there is theater or not see this recent post by David Altig of the Atlanta Fed. His view is typical and based on what he believes. No theater here, but conviction. Could all these folks like Altig really be so wrong?

  27. David,

    I used to watch the Fed's H.4 report weekly for signs that Excess Reserves are being deployed by the system (I was watching for "early warning signs" of inflation). That sparked me to think that I would probably never be able to tell: ER's would fall too slowly to be observable.

    Say you have $1tr in ER's the first week, and the banking system buys $100b in securities, creating an equivalent amount of deposits -- in one week. The impact would be to reduce ER's and increase RR's by something much less that the required reserve ratio (given sweep accounts). Say its 5%. So you would see Excess Reserves decline by $5b. At this rate of buying, of course, the change would be quite noticeable in the H.8 Report (balance sheet of the banking system). However, it would also increase the money supply by $1tr in ten weeks, which of course the Fed would never allow. So even a $50b decline in Excess Reserves will not be observable in a period of say, six months.

    I don't know where JKH came up with 45 years, but it does appear that $1tr in ER's would expand the money supply by so much if completely deployed into RR's that the period is much more likely to be infinite -- the Fed will never allow them to be converted.

    BTW, David, one issue I'd like to hear you address is the equivalence, at a give Fed Funds rate, between the banking system holding $1tr in Excess Reserves to accommodate lending and the Fed supplying those funds to maintain the target rate. If the two are equivalent, then why do Excess Reserves matter?

    One last point: think of a bank as a generic "firm". We have many examples of firms -- in the tech sector for instance -- which accumulate cash balances as a result of operations. Rarely have I heard a Microsoft analyst argue that the firm will initiate a new line of software products PURELY as a direct function of the level of that cash. In fact the two decisions -- investment projects and cash management -- are separate. The former is a function of demand for particular types of software, and this is independent of the level of cash. (I acknowledge that firms tend to make balance sheet structure decisions -- acquisitions, for instance -- based on cash levels, but these are different from the operational decisions).

    The set of cash management decisions a bank faces involve acting in the inter-bank market for loans or buying short term securities. The question is why don't banks deploy the ER's into T-bills? The reason is they probably do, and this drives the T-bill rate to at or below the interest rate on reserves, which in turn eliminates that option. Given that the inter-bank market is broken, and that formerly liquid s.t. securities (Agencies) cannot be counted on, the banks' cash management choices are severely limited.

  28. "Are you sure you want to claim all money supply changes are endogenous and that the central bank simply responds to the economy? If you believe that then (as I told Marshall) you also must believe Robert Mugabe was a victim of hyperinflation not the perpetrator of it."

    I am probably (not probably, definitely) the least qualified here to chime in on this discussion but I do have some very strong thoughts on this comment and since in this internet age I do not have to raise my hand and ask to speak I will give my opinion. Do with it what you will. I ask the "elders" of MMT here to correct me if I am astray, I am eager to be chastised and learn.

    The use of the ZImbabwe situation is a very popular rhetorical tool of those opposed to the ideas of MMT. Bill Mitchell does a splendid job of addressing it here
    and I think the partial answer to your question is YES Mugabe was a victim. He was also a perpetrator because the problem was not simply one of "money printing"

    The fact is because of political responses to social inequalities (real and perceived) a real shortage of the most critical economic output was created. This PRECEDED the printing of money. The printing of money was an improper response to the real supply shock which forced up the real price of the agricultural goods. I will not try to navigate what Mugabe should have done after his political move, which was driven by an (I think justified ) attempt to "right" a long standing social "wrong", backfired and made things worse for the people he wanted to help. That is for greater economic minds than mine but I dont think its wrong to state that Mugabe WAS a victim in a sense. Yes he made the original mistake of giving productive land to less productive farmers, but this CLEARLY was not a situation where Mugabe simply "printed more money" and added to an already healthy supply/demand ratio, yet this is how it is improperly understood by most of the people who bring up ZImbabwe.

    The supply shock came first ( are result of a Mugabe political decision) driving the REAL price of the goods up and the money printing was an improper response to this dire situation.

  29. David,

    Zimbabwe suffered hyperinflation because of supply side issues arising out of poor government (non-fiscal) policies. The increase in government spending can just make an already existing problem worse.

    Bank lending is always and everywhere an endogenous phenomenon. Borrowing is demand led. Of course, the regulatory authorities should take care that there are no Ponzis around because it seems, in the housing boom, everyone became a Ponzi and the banking system supported it.

    Except a few risky banks - which may face capital constraints - there is never a shortage of supply of loans. It is always demand determined. If banks have less reserves to satisfy reserve requirements, the central bank lends them reserves. The only exception was the recent turmoil when banks faced capital constraints. Every rule has an exception, and the rule that banks can always lend also has an exception.

    The mainstream view is that banks are quantity setters and the market is the price setter. However it is just the opposite - banks are price setters and quantity takers.

  30. David Pearson:

    You say "it does appear that $1tr in ER's would expand the money supply by so much if completely deployed into RR's that the period is much more likely to be infinite -- the Fed will never allow them to be converted." We agree then--the ER could be problematic down the road. As you note, though, the Fed does not want the potentially explosive money supply that would result if they were left as is and therefore will intervene. The only area where we see things differently is that you feel with certainty the Fed will move to remove them in a timely fashion where I am not as certain.

    On your second question of equivalence I am not sure I completely understand it. Could you restate it for me? (I suspect it can be answered using the supply and demand diagram M.Thoma demonstrates in the link above in the original post.)

    Finally, I think the analogy of microsoft and the bank breaks down because for banks investment projects and cash management are linked. Controlling for risk, when a a bank invests it has to make a trade off between liquidity (have asset they can quickly sell to meet their obligations) and return. Obviously, the most liquid asset is cash. In terms of my discussion with JKH, there is an definite opportunity costs to holding excess reserves. They could be earning a (risk-adjusted) return in the interbank market and this has to be weighed against holding the excess reserves as precautionary balances.

  31. Greg and Ramaan:

    For the record, I have never engaged in this discussion before nor heard of MMT. So I am not falling back on a "popular rhetorical tool" but pointing out one of the glaring problems with this understanding that came to my mind. With that said, your responses to my Zimbabwe hyperinflation question were (1) a negative supply shock started it all and that was followed by the printing press to solve government problems (2) an increase in government spending caused the problem. On (1) printing money means the central bank was creating monetary base and on (2) the same is implied (i.e. government spending was finance by the central bank). Either way the central bank still caused the hyperinflation! It wasn't the private banking system that suddenly demanded more monetary base or some some unexpected change in economic conditions that "forced" the central bank to print money. It starts (and ends) with the central bank making a conscious choice to create excessive monetary base and that in turn creates the hyperinflation--not the other way around. (Yes, expectations at some point matter too as velocity takes off but this too is triggered by the initial actions of the central bank.)

    Note, for this discussion it is irrelevant that the monetary authority was politicized or was driven by political pressure. The question is whether the central bank has the ability on its own to create enough extra monetary base (beyond that justified by the banking systems demand for reserves and the existing economc conditions) to create hyperinflation. This should adamantly clear when one does a simple counterfactual: what if there had been a stable government in Zimbabwe during this time? Most observers would say no--there would have not have been excessive creation of monetary base and no hyperinflation.

  32. David,

    Thanks for engaging. You have a size limit on comments, so I’ll do a separate response for each question. Several responses also required more than one page. Sorry for the chop job.

    “First, I find it hard to believe that two banks would continue acting the same after the unexpected reserve increase for the one bank (i.e. my Bernanke helicopter drop story from above). Yes, they may both be able to get bank reserves for settlement purposes anyway, but now one bank has a new deposit account and excess reserves to manage. In other words, one bank now has explicit costs in managing the new deposit account as well as the opportunity costs of excess reserves sitting around idly versus earning a return. Yes, the banker has to manage risk but he/she also has to manage these new costs. How can the two banks possibly act the same thereafter?”

    There is no opportunity cost associated with excess reserves provided that excess reserves are earning the risk free rate. This is because excess reserves are zero risk weighted for capital allocation purposes. Accordingly, they require no capital allocation on a risk weighted basis.

    It is irrational for any bank risk/capital manager to allocate capital to risk simply on the basis of perceiving a nominal opportunity cost when earning the risk free rate on a risk free asset. It is wrong to view earning a higher risk adjusted return on risky assets as a substitute for lower earning risk free assets, without reference to capital requirements. A higher compensation for risk is a requirement with or without the alternative presence of risk free assets. So the presence of risk free assets is a red herring in terms of opportunity cost.

    In terms of deposit costs, I referenced the fact that in this environment banks will be careful to review the interest rate structure (and the fee structure) on deposit accounts to ensure that net interest margins and related costs are reasonable compared to the risk free rate. 

  33. “Second, you claim bank reserves only are important for settlement purposes, but settlement issues themselves can influence how much lending a bank can do. For example, a bank that has insufficient reserves to meet settlement will be less likely to issue new loans (and may even call in existing ones). That bank reserves are used for interbank payments which in turn may influence bank lending amounts to bank reserves being indirectly capable of influencing lending.”

    This is not the case for a bank that has adequate capital. A well capitalized bank has capital for two purposes. First, it allows it rationally and prudently to support the risk it chooses to take. Second, it enhances its own credit quality in the market place. A bank that has good credit has liability liquidity. It can raise deposits from institutions and clients that perceive it to be a good credit. There is no reason not to expect this against the backdrop of a central banking environment that provides sufficient reserves in aggregate in order for individual banks to meet their respective reserve requirements through normal bank reserve management and money market functions. And that’s what central banks in fact do.

    Moreover, all well run banks have liquidity specific policies for managing both asset liquidity and liability liquidity. These policies enable them to adjust their settlement positions quite easily in normal course, in response to the more strategic changes in their balance sheet profiles, such as those created by taking on new credit risk and new lending. The central bank reserve position is merely the facilitating operational mechanism through which these broader liquidity policies ensure adequate settlement balances in the normal course. The reserve position is not viewed as a strategic warehouse for either liquidity or lending purposes in normal banking operations. And it is not properly viewed that way in the current environment, notwithstanding the Fed’s creation of an outsized excess reserve position. Again, that position is the by product of Fed asset activities, and not an objective in itself.

  34. “Third, I agree that the Fed controls total bank reserves, but I am surprised at your claim that excess reserves turn into required reserves at a tortoise pace. I thought excess reserves in normal times were minimal as the banks would quickly get rid of them given their opportunity cost. I am even more surprised at your claim of it taking 45 years to convert the current excess reserves into required reserves. Are you using numbers that would hold in normal times or crisis times? The distinction is important since it is a fear that the excess reserves will lead (indirectly) to more loans once conditions return to normal. (BTW, I think a better indication of the Fed's potential influence on banking lending would be to look at non-borrowed reserves not excess reserves.)”

    The Fed controls excess reserves tightly in normal times in order to control the effective fed funds rate. Excess reserves are normally minimal (about $ 2 billion), because that’s the amount that is adequate to control upside pressure on the effective fed funds rate, given the inevitable interbank reserve adjustments and settlements that banks need to undertake at the margin in order to meet their individual requirements. By construction of the facilitating excess position, any material amount more than that would put unwanted downward pressure on the rate.

    Uneven reserve distribution results from the normal day to day asset liability activities of banking and their associated interbank payment and settlement requirements, as agents on behalf of their customers, and as principals on their own behalf. Excess reserves must be sufficient (but not more) in order to allow surplus banks to make adjustments through their money market operations and to allow deficit banks to make corresponding adjustments through their operations. These adjustments are all money market oriented and short term in nature. This is about short term tactical liquidity management. It has nothing to do with balance sheet expansion over the longer term.

    ... continued...

  35. ...

    The idea that banks use pre-supplied reserves to expand loans and deposits is an erroneous fantasy put forth by textbook multiplier analysis. In fact, banks make loans that create deposits, and the central bank supplies required reserves in response (with a data recognition lag). This pattern is factual at the operational level of central banking. It is supported statistically in the attached data series as well:

    Looking closely at this data series, one sees that in normal historical times, there is no correlation between changes in the monthly level of required reserves, and changes in excess reserves. In fact, the monthly required levels are volatile, while the monthly excess is extremely quiet and stable by comparison. There are some interesting points at which the monthly excess is volatile, including September 2001 and August 2007, which were both months during which there was extreme volatility in money and credit markets. The Fed poured excess reserves into the system in both cases to respond to extreme liquidity disruptions and market turbulence as well as aberrant reserve distribution patterns among banks. Its reason for doing so in both cases was to attempt to maintain an effective fed funds rate reasonably close to the target, in markets that were extraordinarily stressed. The expansion of excess reserves due to central bank “credit easing” and asset expansion began in earnest in September 2008. But the historic pattern is completely consistent with the behaviour of a central bank that is supplying required reserves in response to statutory needs, and excess reserves in response to market conditions and their effect on the effective fed funds rate.

    As described already, today’s crisis level of excess reserves is driven by an entirely different motivation than sufficiency for interest rate control. It is driven by a balance sheet requirement for liability management corresponding to the excess funds the Fed itself has created through its proactive asset expansion. In that mode, the Fed must pay interest on excess reserves in order to manage a lower bound for the policy rate. The policy rate and interest on reserves become inextricably linked at an operational level - the point I made in my initial comment on this thread.

    ... continued ...

  36. ...

    The multi-year calculation is a matter of arithmetic extrapolation, as I described it:

    E.g. January 2009:
    Required Reserves held as deposits at the Fed: $ 66 billion
    Commercial Banking System Deposits: $ 7.6 trillion
    EFFECTIVE reserve ratio: 1 per cent
    Excess Reserves: $ 1 trillion

    Applying a (arbitrary but not unreasonable) growth rate of 7 per cent per annum to the aggregate balance sheet of the commercial banking system, including deposits, and maintaining the same sort of deposit mix and effective reserve ratio, one arrives at a time period of about 40 years in order for $ 1 trillion in excess reserves to become $ 1 trillion in required reserves. My previous number of 45 years incorporated a slightly lower estimate of the pre-crisis effective reserve ratio, which is probably weighted slightly more to savings and time deposits (I can’t seem to extract the historic data to make that more precise). There’s obviously not much difference in the order of magnitude using the current effective ratio.

    The purpose of such a calculation of course is to demonstrate the absurdity of the idea that banks will use $ 1 trillion in excess reserves to begin expanding their balance sheets in such a way as to attempt to “use up” a reserve position that has been created for entirely different reasons. Combining that wayward idea with the facts of how banks assume new risks by deploying capital, one should see the error in this type of thinking and that the cause for alarm on that basis is misplaced.

  37. “Fourth, there is a large literature on how monetary policy affects the broad economy. There are many transmission mechanisms or channels (see here for overview). One of them is called the bank lending channel which basically says that monetary policy can drain reserves from the banking system which in turn leaves fewer funds to lend. Now this is not the most important channel but many empirical studies have found it to be important for smaller banks with less access to other funding sources. The big point here is that there is a large literature with lots of empirical evidence behind it. (Here is one recent study; there are numerous others.) All I have heard from the skeptics so far are stories but no systematic empirical evidence. It is hard to convince me just with stories.”

    And I wouldn’t expect you should be convinced so quickly.

    One can question stories in a number of directions here. The economics profession can question itself and its own stories in fundamental ways. I think of Krugman’s Dark Age of Macro, which seems to me to be about a story as much as it is about empiricism. Outsiders can have relevant knowledge of some of the working pieces behind a particular story as well. To cut to the chase, let me pose a contentious proposition:

    My contention would be that mainstream economics does not have a good understanding of how banks work at their core. The core function of banking is the allocation of capital to risk. Once that is understood, one can begin to think more clearly about the distinction between capital and liquidity. Once that is understood, one can begin to think more clearly about the role of central bank reserves in the context of bank liquidity. And once those things are understood, one can begin to think more clearly about how the modern monetary system works as the monetary environment for bank risk and capital management. And once that is understood as the relevant monetary architecture, one can move on to think more clearly about how to go about formulating macro policy over a wide variety of areas, including fiscal policy, monetary policy, and financial system regulation.

    ... continued ...

  38. ...

    The second piece of this theme is that mainstream economics specifically does not have a good understanding of how the reserve system functions at an operational level. The fact that it has not had an inadequate grasp on the function of capital has been the catalyst for this dual error. My own personal story is based on some direct experience in the area. But it’s not an isolated story and not one without substance or support. About a year ago I discovered quite by accident that a small sub-group of economists somehow had sunk their teeth quite accurately into the facts of how the banking system actually works in this area – e.g. Randall Wray, Scott Fullwiler, Bill Mitchell, and Marshall Auerback and Warren Mosler. This group is identified variously with Chartalism, or “Modern Monetary Theory”. It concerns itself first at the analytical level with the facts of monetary operations and balance sheet accounting. Post Keynesian circuitist Steve Keen has also done macro empirical statistical studies demonstrating the error of the textbook multiplier theory in particular.

    You referenced a few papers.

    Here’s one from the BIS:

    It’s mostly correct and not bad for a mainstream interpretation of bank reserves. It borrows heavily (although without direct attribution) from the work of the Chartalists, whose websites are replete with thorough background papers and blogs on bank reserves and related issues in much more depth.

  39. “Fifth, your original assertion that all of these concerns about the threat of excess reserves is just theatre is simply wrong. Most economists believe it. There is abundance of empirical evidence to support it (see above). Bernanke, for one, believes it because he wrote the book on the bank lending channel. There is no conspiracy here to give the misinformed masses what they want to hear. You could make the case that most economist are simply wrong in their view but I would wrestle with the empirical evidence first.”

    I’ll back off on the theatre comment. The only reason for me to make such a comment in the first place is to give the Fed some benefit of the doubt. Perhaps I have a respect for Bernanke that I need to overcome. I hope not. I think of him as learning on the job, and that it was pretty much impossible for him to co-author TARP with Paulson, without galvanizing a pretty good understanding about the relationship between risk taking and bank capital, and the fundamental disassociation of that relationship from the issue of excess reserves.

    But your final question gives me pause about being so charitable:

    "Could all these folks like Altig really be so wrong?"

    Yes. They can be wrong in their failure to understand how the banking system allocates capital to risk taking and balance sheet expansion, and in their confusion of the reserve management function with the capital management function.

    Miraculously, as if higher power were helping to make the case, the words “risk” and “capital” never appear in Altig’s piece. Managing bank capital is a specific function in a big commercial bank. The capital manager’s job is an executive function that reports up through to the senior bank risk operatives, including the CEO, CFO and CRO. It is a critical function in the overall strategic risk taking and balance sheet growth of a bank. The reserve manager’s job by contrast is a much more reactive operational set of tasks, based on the normal ebb and flow of regular payment transactions throughout the financial system. And although the capital management function is at the center of everything, one of the less important operating relationships it has is that which it has with the reserve manager.

  40. JKH:

    I promise to stop with the comments and questions soon. I am after all, an untenured professor who needs to spend time publishing. Oh, the dangers of running a blog.

    I am guessing you think the Swedish's central bank policy of penalizing banks for holding excess reserves is an exercise in futility. Along those same lines the Fed increasing required reserves in 1937 to prevent banks from turning excess reserves was another case of misunderstanding (although the standard story is it worked too cause a recession). What do you think of these event?

    I talked to one of my colleagues who specializes in banking and they agreed with you that capital considerations and risk issues are important. However, she said that monetary policy matters too and yes, that excess reserves do pose a problem down the road. I am going to make it a point to chat with some other colleagues who are finance professors that specialize in banking on this issue as well as some bankers. I may even call the Fed settlement system. Honestly, though, it seems that more folks would be on board with your view it it were correct. There are plenty of smart people out there--much smarter than me--who according to you get it wrong.

  41. David,

    I’m aware of the Swedish move, but haven’t followed it closely. Haven’t seen much written about it either, but can’t say that I’ve looked for it. And I’m not a student of the Fed in the 30’s unfortunately.

    But here is a two part piece by Scott Fullwiler that you might find interesting:

    Note that I’ve said there’s no reason for the Fed to leave excess reserves in the system once they’ve served their purpose. That purpose is to serve as the liability management offset for the Fed’s extraordinary asset management activities - not to motivate commercial bank asset management. Once Fed objectives have been achieved in that regard, they should withdraw the excess reserves, at some pace. Leaving them in is a pointless distortion otherwise.

    I’ve also said elsewhere that there may be an element of Fed hedging involved here, whether deliberate or inadvertent, something along the lines of my “theatre” comment. What I’ve described about capital and risk taking is the rational approach to bank asset expansion and risk taking in any circumstances, with or without excess reserves. If the Fed were to leave outsized excess reserves in the system for no reason at all, it’s almost a temptation for irrational behaviour or rogue risk taking by banks that may be poorly managed. But apart from that, as I say, the correct reason to withdraw excess reserves is as a natural consequence of the Fed withdrawing from its credit risk and interest rate curve risk support activities. Hedging against irrational outcomes is a secondary by product. And I still insist that well managed banks such as JP Morgan for example won’t take such an irrational approach.

    I’m not sure you’ll get back anything too different from your own view with the type of survey you’re considering. My preferred survey would consist of posing a public question about how bank capital managers respond to their bank’s excess reserve position, particularly in this environment. Imagine if the economics community were to take out a full page add in the Wall Street Journal or the New York Times on this question, and put the following to the capital manager of JP Morgan or Wells Fargo or any number of other banks:

    “Does your bank’s excess reserve position affect your assessment of credit and other risks, and your willingness to allocate capital to those risks?”

  42. JKH:

    I think most everyone who is concerned about the large stock of ER being a problem down the road understands their creation was a byproduct of the Fed's attempt to repair/prop up the financial system. No one has said the ER were created to "motivate commercial bank asset management". What people are saying, however, is that they may just do that in the future when the economy recovers. Your emphasis on capital and risk issues makes more sense for the present and that seems to be the thrust of your posts . The ER concerns, though, are focused on the future when normality (if it ever) returns. Thus, you want to ask bank managers how ER affect them now and I want to ask how it will affect them in the future.

    Also, you believe that the ER will simply disappear once the Fed withdraws its support for the financial system. I believe this is not so certain for two reasons: (1) political pressure may be too much to remove the support and (2) even if there is no political pressure the Fed's securities may have lost value and not be sufficient to cover the withdrawal of all ER. This, of course, is a minor point; the big issue is what would the ER do to the system in the future if they remained. I believe they will both directly (via the bank lending channel) and indirectly (via, for example, a change in inflationary expectations that keeps the real interest rate inordinately low which in turn increases the quantity demanded of loans) lead to an increase in the money supply.

    On this last point it is worth noting that the capital and risk concerns themselves are not independent of Fed's actions. For example, if the Fed states it is unexpectedly dropping interest rates the market may suddenly expect higher asset prices. If so, the capital position of a bank could improve and at the same time there would be an increase in the demand for loans given the expectation of higher spending in the future. All of this, of course is predicated on the Fed being able to lower rates which it can do through actual or perceived adjustment of bank reserves. In short, the Fed states its intention to adjust bank reserves to meet a new lower target interest rate, the demand for reserves simultaneously goes up in expectation of this change, and banks facilitate this process since their capital and risk position have improved. Note, though, that the causality starts with the Fed indicating they will adjust bank reserves as need to change the target interest rate. [This here is a prime example of why it is important to distinguish between a change in the demand for bank loans originating from say an increase in economic activity versus a change in quantity of loans demanded originating from a change in the price of the loans. ]

    Finally, I hate to belabor this point but it seems hard to fathom how so many smart people who study the banking system for a living say at the Fed or Riksbank or any other central bank could get it so wrong.

  43. David,

    "Thus, you want to ask bank managers how ER affect them now and I want to ask how it will affect them in the future."

    I’m asking both.

    “Also, you believe that the ER will simply disappear once the Fed withdraws its support for the financial system.”

    They disappear as quickly as the Fed chooses to settle incoming cash flow from any assets without redeploying it (other things equal). That’s an operational fact. I haven’t specified anything about the pace at which they should attempt to do that.

    “On this last point it is worth noting that the capital and risk concerns themselves are not independent of Fed's actions.”

    I disagree with your paragraph there.

    Should be interesting to see how it goes! Thanks for the discussion, David.

  44. David,

    Thanks for responding to the many comments.

    The "equivalence" argument is that the Fed supplies any reserves the banking system needs at the target Fed Funds rate. If the banking system needs reserves to support $1tr in new lending, demand for reserves would push up the Fed Funds rate, and the Fed would inject reserves to push it back down.

    Therefore, the banking system is not constrained by the level of Excess Reserves. The Fed will support -- at a given Fed Funds target rate -- any amount of loans the system wants to create without the existence of these reserves.

    So yes, Excess Reserves could be "problematic" if loan demand increases, but no more problematic than the injection of reserves that would result from a too-low Fed Funds target rate. Therefore, the level of Excess Reserves is really irrelevant for supporting lending, the money supply, or inflation. It is only the target rate that matters.

  45. "Conventional" monetary theory involves an excellent understanding of how interest rate targeting by the central bank makes the quantity of reserves endogenous.

    The money multiplier analysis, instead, looks at what happens when the Federal Funds rate is left to adjust according to market forces, and the quantity of bank reserves is given.

    Determining what the Fed needs to do in order to keep its interest rate on target in the face of changes in the quantity of reserves or the demand to hold them, can hardly be answered by saying, "the Fed will keep the interest rate on target." How, exactly?

    The whole of monetary theory is about the distinction between what the individuals do and how it impacts the economy.

  46. David: Your post (including its quotes) makes good sense.

    Reading through the comments has increased my respect for the importance of teaching the standard textbook money multiplier model, in particular the dynamic aspects of that model.

    Take for example the case where banks have a 10% desired reserve ratio. In that case, the textbook story takes great pains to explain that an individual bank with $10 excess desired reserves will expand deposits by only $10, not $100, even though the banking system as a whole can expand deposits by $100 in equilibrium.

    Even if the desired reserve ratio were 0%, it is as if the desired reserve ration were 100% at the margin for an individual bank.

    This very important insight has been overlooked by your commenters.

    The celebrated Paradox of Thrift reminds us that what is true for each individual at the margin is not (always) true for the system as a whole in equilibrium. To say otherwise is a Fallacy of Composition.

    To argue that (under, say 0% desired reserves) since the banking system as a whole does not need additional reserves to expand deposits, therefore no individual bank needs reserves to expand deposits, therefore the supply of reserves is irrelevant, is to commit a Fallacy of Composition in reverse. A Fallacy of Decomposition, if you like.

    And when I say "supply of reserves" in the above, I do NOT mean the quantity of reserves. I mean the supply CURVE of reserves, of course. That is a second important distinction that gets overlooked by "horizontalists".

    Maybe I'll do a post on this sometime, and try to draw their fire!

    (By the way, none of the above is to deny that JKH in particular knows far more about money and banking in general than I ever will. And these are indeed guys who commendably try to engage, and do so well.)

    Off-topic: Winterspeak (since I guess you might be following this thread): why don't you allow comments on your blog (or have my eyes missed something)? I couple of times in the past I wanted to post in agreement (and once in disagreement).

    Must fix my daughter's car's brakes first, but should be able to post soon. Then they'll all come after me! (I've just about recovered from the 100+ comments on my last post on this subject ;-) )

  47. David: Good discussion. As others have done before, you kill me by bringing up "Zimbabwe" while ignoring the fact that 1) Mugabe contracted real output by about 50% and 2) Zimbabwe runs a mixed currency regime with very limited ability to tax. The similarity between this situation and the US is zero.

    "Zimbabwe" has become this totemic verb monetarists like to utter in discussion as an alternative to thinking. I throw "Japan" right back at you.

    Bill & Nick: Same old same old, eh? At the very least, you should take some comfort in David pointing out that the ER mechanism may be entirely psychological -- since it proven tricky (to be kind) to find anything else.

    Nick: I had comments on my blog briefly, but Blogger did not support it correctly so I took them off. They're shutting down FTP hosted sites, so once I migrate over to a new system, I'll consider turning them on again.

  48. Winterspeak:

    One doesn't need Zimbabwe, but it is the most recent experience so I referenced it. And I already responded to the supply shock question. Here goes again...

    Supply shocks do not create hyperinflation. Creating excessive amounts of monetary base does. Negative supply shocks make a one-time increase in the price level, they not alter the inflation rate. Even if there were a series of negative supply shocks they would never be able to generate the size of sustained price increases to create hyperinflation. Mugabe may have created negative supply shocks but they were not the reason for hyperinflation. In November 2008,for example, inflation was 79.6 billion percent. It is simply implausible that a negative supply shock(s) could cause such a high rate. It takes an excessive and sustained increase of the monetary base to get there. (There is no evidence of any hyperinflation experience ever being caused by supply shocks.)

    What is key here is whether a central bank can exogenously increase the monetary base to create hypeinflation. It does not matter why it would do so--political influence, bad analysis,mistakes etc.--just whether could it do so. If so, then it significantly undermines the MMT.

    I would be careful using Japan as a counterexample. Many, if not most, observers believe Japan's monetary authority did not do enough to fight the deflation.

  49. If the expansion of the monetary base only results in inflation if the Central Bank successfully influences inflation expectations, then doesn't that mean the base expansion (i.e. Excess Reserve creation) is at best a precondition for creating hyper-inflation?

    The risk of the Fed's "exit" is not whether Excess Reserves are lent out in a growing (real) economy. It is that, in the absence of the QE subsidy to term rates, the Treasury will be seen as needing to pay higher rates to finance what are arguably structural long term fiscal deficits. If this view takes hold, term Treasury rates will rise, and the Fed will feel the pressure to return to QE or risk deflation (this is essentially a continuation of "asymmetric" policy). Ultimately, this dynamic might persuade the market that the Fed will consistently finance structural deficits. The Latin American experience is that this fear of monetization is the number one driver of hyperinflationary expectations.

    Is the above scenario farfetched? If you assume a real recovery, then yes. If you assume a reasonable chance of stagnation or double dip, then its quite possible that term Treasury rates may become, as they are all over the world in high-debt countries, inversely related to RGDP expectations. This is essentially happening in the European periphery today, and it may happen next in the UK and Japan.

  50. David Pearson,

    I think you’ve got the exit risk about right (one of the few).

    There are two elements to the exit risk – the policy rate and the balance sheet.

    It turns out that the balance sheet strategy is mostly about influencing long (mortgage) rates at this point.

    Together, it’s about the entire curve.

    The balance sheet/long rate piece is done entirely through the asset side of the balance sheet.

    Given the Fed’s ability to pay interest on reserves, and the irrelevance of excess reserves for bank commercial bank credit expansion, the least important part of the whole exit equation is excess reserves.

  51. David: Help me out here.

    You clearly believe that interest rates have a clear net effect on the non-Govt sector's desire to spend or save, as a consequence of "inflation expectations".


    Can you list 3 other things you believe influence, at a sector level, the desire to spend/save/"inflation expectations"?

    Then, can you please assign a rough % as to how material you think each of those are. Your %s can sum to far less than 100% if you believe there are hundreds of factors, each of which (by itself) has a small impact.


  52. David: My last comments seems to be eaten, but I'm continually struck by how the "expectations crowd" (monetarists) are so awful at actually understanding how real people think about the future.

    Of course observers "believe Japan's monetary authority did not do enough to fight the deflation" -- if they believed any differently they would have to admit that their monetary theories are nonsense!

    Have you ever spent time in a third world country? I have. All monetarists should spend time their to see their theories in action. You would see how frankly ridiculous your comment is "Negative supply shocks make a one-time increase in the price level, they not alter the inflation rate."

    A population adopts a native currency because the rulers can create and maintain a tax obligation in that currency. For Zimbabwe, that was tenuous to say the least because the Govt is so weak.

    Zimbabwe's supply shock was not just a one-time destruction of capital, it was a policy that undermined what little savings desire there was for ZWL. Third world countries de facto run very very mixed currency regimes and have terrible tax collection ability, so their ability to maintain demand for their currency is weak. The wealthy inevitably have (large) foreign fx bank accounts (which solves yet another macroeconomic "riddle" -- can you guess which one)? The poor accept fx when they can, but also diversify into real assets, barter, and buy gold (in some parts of the world).

    A 50% supply shock that also undermined whatever remaining savings desire there was for the ZWL would create hyperinflation no matter what monetary policy was, and no matter how large the monetary base was, and no matter whether monetary base was expanding or contracting. It doesn't matter what interest rates are, and it doesn't matter how much people have in their bank accounts. If you have a massive switch out of a currency into anything else, you will have hyperinflation.

    Ranaman, I believe, might have some insight into this as he talked about India's monetary/Treasury system and I'm sure most Indian merchants are pretty happy accepting certain foreign fx instead of rupees. And they "save" in gold.

    Given the very very limited sovereignty Mugabe's Government enjoys, their CB could have been shut the printing presses off and they would have still had hyperinflation.

  53. In case you want more of this discussion go see Nick Rowe's new post on bank reserves and the money supplyhere.

  54. Winterspeak:

    Your first comment was never eaten; I just didn't get around to moderating comments until late in day.

    Let me speak first to your second post. There is a reason why some countries end up with mixed currency regimes: people don't trust the domestic currency. And this lack of trust arises because governments are debasing the currency in some form. In the hyperinflation context it is excessive creation of the monetary base. In short, mixed currency regimes or defacto dollarizations are the consequence, not the cause, of currency problems like hyperinflation. Yes, defacto dollarization will worsen the existing hyperinflation problem but it is not the original cause. The story I am telling would have the following causal chain: excessive money creation => prices increase => velocity picks up (i.e. substitution out of domestic currency into other currencies or real goods and services)=> prices increase more => more money created to keep up with hyperinflation cycle => repeat. The velocity uptick, shaped by expectations of hyperinflation, can become a big contributor to hyperinflation cycle and I am guessing you saw some of this. But, again, it is dependent on the original spark: excessive money creation. By the way, there is abundant evidence that excessive money creation was a crucial part of Zimbabwe's hyperinflation experience. Here is one example.

  55. Winterspeak:

    On your second question let me begin by acknowledging that the quantity of the monetary base (and bank reserves by default) is endogenously determined if the central bank is targeting an interest rate. Clearly, if the Fed targets a fixed interest rate is must passively accommodate changes in demand for the monetary base and thus the quantity becomes endogenously determined at this fixed rate. Now what happens if the Fed suddenly and unexpectedly changes its target interest rate? If it were truly unexpected, then there will be an exogenous change in the monetary base to hit this new target. Now how does this move lead to an increase in the broader money supply? There are many channels, only one of which is a change in inflation expectations. In other words, the Fed can change desire for loans through many channels. Here a few:

    (1) Interest rate channel. This channel works by lowering the real interest rate which in turns increases the quantity demanded of loans--it pulls new borrowers into the loanable funds market. In normal times when inflation expectations are grounded, a change in the nominal interest rate will not change the expectations component. As a result the real interest rate will adjust. For example, if the Fed lowers the (nominal) fed funds rate and if it has sufficient inflation fighting credibility such that its inflation expectations don't budge, then the real federal funds rate will drop too by default (using the fisher equation, nominal rates = real rates + expected inflation). In the present context, when the nominal federal funds rate is close to being stuck (can't go below zero) then the real federal funds rate could still drop if the Fed were able to increase inflation expectations. (This latter case is what i was referring to above, though its relevance would be limited to the current environment only)

    (2) Balance Sheet Channel: here the Fed lowers the federal funds rate which, in turn, increase asset prices by (i) lowering the discount rate and (ii) increasing expected earnings. An increase in asset prices makes for more better collateral and thus more borrowing (e.g. home equity withdrawal during housing boom). Banks may also be more willing to lend since the increase in asset prices would improve their capital position. In short, improved balance sheets increase the demand and supply of loans.

    (3) Wealth Effect Channel: the above increase in asset prices also means one's net worth is now higher and there is less urgency to save. As a result spending increasing and this will increase the demand for money and loans.

    (4) Exchange Rate Channel: the lower interest rates will cause capital to leave the U.S. and thus cause the dollar to depreciate. Suddenly our exports become more affordable to foreigners and they buy more. This increase in domestic spending by foreigners will increase the demand for money too.

    There is also the bank lending channel, but you have heard that story before and empirically it appears to be less important than the channels above.

    Three other important factors that can influence money demand would be expected income, expected price level, risk aversion. All of these can move independent of Fed policy, so yes some of the money supply will be endogenously determined. How important these factors are as compared to exogenous Fed moves working thru the above channels would depend on the circumstance.

    Note, all of this discussion hinges on the Fed targeting interest rates. What if the Fed changed is approach and started targeting a quantity like the monetary base? In that case, interest rates become endogenously determined for a given target monetary base target. The quantity of money is now largely shaped by the Fed even in the short run.

  56. All:

    This has been interesting discussion, but I need to move on to other matters. Consequently, at this point I closing comments to this post. If you need to continue this conversation I recommend you do so at Nick Rowe's new post linked to above. Thanks to everyone for their comments.