Wednesday, November 11, 2009

Goldman Sachs On the Trillion Dollars in Excess Reserves

Senior Goldman Sachs economist Ed McKelvey this week put out a comment on the trillion dollars in excess reserves that banks now have sitting at the Federal Reserve. A good portion of his perspective will not be new to EPJ readers, as I have scattered similar comments across a broad spectrum of posts. I am not necessarily as sanguine about the outcome as McKelvey and he is too Keynesian for me, but he does touch all the bases.

McKelvey's review ties everything together in one nice analytical piece. Here it is courtesy of Goldman Sachs (My comments are in blue):

US Daily: A Trillion Dollars in Excess Reserves – Why We’re Not Worried (McKelvey) (Nov 10 2009)

Many market participants and economists worry about the large volume of excess reserves in the US banking system, which just crossed the $1trn mark on the way to $1.3-$1.4trn by next March. Some see this as evidence that the banking system is not lending enough while others worry about inflation implications.

In our view, excess reserves are not the best indicator for adequacy of bank lending; more direct data are available from bank balance sheets and from the senior bank loan officers’ survey. Both of these sources confirm that bank lending is weak.

As for the inflation risk, we remain deeply unconcerned given the huge amount of spare capacity in the US economy. This gives Chairman Bernanke and his colleagues plenty of time to manage excess reserves, and they appear to be working on the tools to accomplish this. McKelvey is a bit Keynesian here and believes inflation can not occur with significant excess capacity. Has he never heard of stagflation?

The surge in excess reserves in the US banking system continues to worry and confuse many participants in the financial markets. In this comment, we explain the mechanics behind this surge and offer our interpretations about whether it is something to worry about. Our one-word answer is "no." We also outline the circumstances that would change this judgment and review and update our thinking on the tools the Fed has to manage the situation.

Q: What is the history of excess reserves?

Until September 2008, excess reserves were quite low – virtually zero by latter day standards. During the preceding 12 months they averaged $1.9bn, or about 4% of total reserves, and they never exceeded $4.6bn (in mid-March, during the Bear Stearns phase of the financial crisis). This was in keeping with the longer-term history of excess reserves. Except for unusual circumstances – in the wake of 9/11, for example – excess reserves were routinely only 1%-2% of total reserves.

In the most recent two-week reserve accounting period that ended on November 4, excess reserves were $1.06 trillion (trn). Most of this increase occurred between early September 2008 and year-end 2008, when excess reserves reached nearly $800bn. They fluctuated between $800bn and $900bn during most of 2009 with no clear direction. Lately they have moved up again. Absent offsetting transactions, we expect excess reserves to reach about $1.3-$1.4trn by March 2010 as the Federal Reserve completes its program of purchasing agency debt and MBS. Nothing new for EPJ readers here. I pointed this out two days before McKelvey notes it here.

Q: What’s behind the sudden change?

The surge in excess reserves is a byproduct of a surge in the Fed’s balance sheet, from about $900bn prior to mid-September 2008 to about $2.2trn today. When the Fed decides to boost the volume of assets it holds, via expansion of an existing facility (e.g., making more loans in the discount window), creation of a new one (from the Term Auction Facility created in late 2007 to those created in the wake of the Lehman bankruptcy to support the commercial paper market), or direct purchases of assets (Treasury securities, agency debt, or agency MBS), its liabilities must go up by a similar amount.

The Fed’s two main liabilities are currency (Federal Reserve Notes) and bank reserves, which together constitute the monetary base. The Fed has other liabilities – for example, deposit accounts for foreign central banks and the Treasury Department. However, these items are not under the Fed’s control, and they do not move predictably in response to its other balance-sheet operations. Thus, a decision by the Fed to expand its assets typically results in a parallel increase in the monetary base.

Mechanically, this usually occurs by the Fed’s crediting a bank’s reserve account if that bank is the recipient of a loan (from the TAF, for example) or the seller of securities to the Fed. (For examples of how this works, see Todd Keister and James McAndrews, “Why are Banks Holding So Many Excess Reserves,” Staff Report No. 380, Federal Reserve Bank of New York July 2009.) In cases where the Fed’s counterparty is not a bank (another financial institution selling securities to the Fed, for example) the transaction will increase bank reserves as the seller deposits the proceeds at the bank where it does business. Conceivably, though, the Fed’s purchase of securities could result in an increase in currency instead.

The reason behind the latest increase in excess reserves, from about $850bn in mid September to more than $1trn currently, is a bit different. In late September the US Treasury Department announced that it would allow balances in its Supplemental Financing Program (SFP) to run down from $200bn to $15bn to provide more flexibility to finance ongoing operations as the it approaches the statutory debt limit. The SFP had been introduced a year earlier as a device to permit the Fed to shield the monetary base from its balance-sheet expansion; to do so the Treasury sold special cash management bills, effectively draining reserves in the process and holding the proceeds on deposit at the Fed.This is some of the sterilization that I have mentioned from time to time. Here, McKelvey is going into this type transaction in much more detail. As the Treasury more recently paid down most of the remaining $200bn in bills that had financed this program, reserves were put back into the banking system. This increase in reserves was thus due to a reduction in one of those other liabilities that the Fed does not directly influence – Treasury deposits – rather than an outright expansion of its balance sheet.

Q: Is it therefore correct to say that the Fed determines the volume of excess reserves in the banking system?

Technically, the Fed’s balance-sheet operations determine the size of the monetary base, as indicated in the preceding answer, rather than excess reserves, which are just one component of the monetary base. The other two components are: (1) currency in circulation (which includes coin issued by the Treasury as well as Federal Reserve Notes) and (2) required reserves, which are a function of the size and compositions of deposits held at banks. However, the public’s demand for currency and coin has been quite stable in recent years, even during periods of extreme uncertainty, McKelvey is just wrong here. Since the crisis intensified in September of 2008, currency in circulation (if that is what he means by "demand" for cash) has climbed by 10.3%. Further, I would argue that "demand" for cash is the demand to hold cash, which has been climbing dramatically during the crisis as evidenced by the very low price inflation. Whatever measure you use, demand is clearly has been growing quite rapidly. and required reserves are a function of the size and composition of deposits held at banks that are members of the Federal Reserve System. They also tend to move slowly over time. I don't know where he gets that required reserves only move slowly. That is not the case, and most recently they have skyrocketed. Take a look for yourself. Thus, it is reasonably accurate to say that the Fed’s balance-sheet operations also determine excess reserves in the short run, but it is important to keep the other two components in mind when working through the mechanics of how changes in the monetary base can ultimately affect financial transactions, the economy, and inflation.

Q: But aren’t excess reserves evidence of the banking system’s unwillingness to lend?

The answer to this is not straightforward. If we focus on a single bank, then the answer would clearly be “yes.” The bank has chosen to hold excess reserves, which currently earn only ¼%, instead of lending the money out at a much higher return. Under these circumstances, one can only infer that the risk/reward assessment is unfavorable for the loan – i.e., that the bank sees too much risk of default to make the loan worth more than the certainty of the small yield on excess reserves. I would add that this is probably an indication that the pure riskless real interest rate is extremely low right now. (Lest one conclude that the payment of interest on reserves has therefore gotten in the way of bank lending, note that the bank could draw this same conclusion with zero interest on reserves if the risk of an outright loss on the loan looked too high.)

But when we focus on the banking system as a whole, then the judgment is not as clear-cut, as the extension of a loan by one bank will usually just transfer reserves to other banks. The borrower uses proceeds from the loan to pay workers, suppliers, etc., who in turn deposit the money in their accounts at various banks (possibly including the bank that made the original loan). As this occurs, reserves simply move from one bank to another. Although some can “leak out” of the banking system as cash, this leakage tends not to be significant.

Those who believe that excess reserves are evidence of a lending problem will be quick to note that while the volume of reserves does not change, the composition does. As the funds that the original bank has lent circulate through the banking system, the volume of deposits also rises. This in turn raises the amount of reserves that banks are required to hold. In theory, banks could keep lending out excess reserves until this expansion of their balance sheets converted all the excess reserves into required reserves – a process well known to students of money and banking as the “money multiplier.” In this framework, excess reserves still reflect unexploited lending opportunities.

However, such an inference ignores two important caveats. First, reserve requirements are quite low – only 3% on transactions accounts exceeding $10.7 million and 10% on those exceeding $55.2 million, and none at all on other deposits. Given these low requirements, excess reserves could still be quite high even if lending were strong; put differently, the money multiplier implied by these requirements is huge (perhaps one reason people worry about inflation). Second, long before bank lending reached the money multiplier limit, capital requirements are apt to limit lending, especially at a time when the consequences of too much leverage are so fresh in bankers’ minds. This is really a very strong argument why you must look at money supply versus other components, including the monetary base, since the money supply is the only money that is out there bidding up goods and services, and the other factors, especially because of Bernanke's new tools, don't always work like they used to.

More to the point, there are more direct ways of assessing whether bank lending is adequate, namely from the banking system’s balance sheet and from the quarterly survey of bank lending officers. From the former, we learn that the total outstanding volume of commercial and industrial loans is down 16½% over the past year; moreover, this decline has accelerated – to annual rates of about 20½% and nearly 25% over the past six and three months, respectively. Real estate loans exhibit a similar pattern of accelerating weakness, though at more moderate rates of decline (from about -2% year on year to -10% at an annual rate in the past three months). As for the bank lending survey, terms continue to tighten for business loans, albeit at a lesser pace, while terms on mortgages have tightened more significantly.

Q: Won’t the huge volume of excess reserves spark inflation once banks do start to lend?

The short-term answer to this question is “no;” the long-term answer is that “it depends on the Fed.” In our view, the difference between short-term and long-term, while not precise, is more likely to measured in years rather than months or quarters.

Our lack of concern about inflation in the short run reflects the large amount of slack that is currently in the US economy. Argh! So Keynesian. We have published extensively on this elsewhere, most completely and recently in Andrew Tilton’s “Deflating Inflation Fears,” Global Economics Paper No. 190, published on September 29, 2009. Suffice it to say that with the unemployment rate now at 10.2% and still likely to rise from here, bank lending of excess reserves would have to occur on a massive scale to prompt the changes in behavior that would have to occur to close the gap that now exists between GDP and its potential.

To put the point differently, excess reserves are not some secret sauce that creates inflation out of thin air, as many seem to imply when they worry about the inflation consequences without specifying how those consequences materialize. While it is always possible that inflation expectations could be influenced by the large volume of excess reserves, large changes in behavior are needed to transform those expectations into actual inflation. To facilitate those behavioral changes, lending would have to change on an equally impressive scale. And if this were to occur, Fed officials would be the first to see it – in their informal contacts with bankers, in the lending survey, on banks' balance sheets, and eventually in the data on spending (retail sales, factory orders and shipments, construction outlays, and the like) that we all look at every month. If the Fed is the first to see it, that would be something new, since they were the last to see the housing bubble.

As for the long term, excess reserves do have the potential to create inflation if left in the system for too long. This is where the Fed’s exit strategy comes into play. While we believe, and most Fed officials seem to agree, that it is far too soon to be implementing the exit strategy, it is not too soon to be planning it – or to be discussing it to ease public concerns.

Q: What is the exit strategy?

Given that the surge in the balance sheet that has given rise to the surge in excess reserves has a highly illiquid aspect to it, the strategy focuses on steps that either live with the excess reserves (step 1) or reduce it via a restructuring of the liability side of the Fed’s balance sheet (steps 2 through 4). In this regard, the comments by Fed officials in recent months reveal the following broad outlines, in probable order of implementation:

1. Increase interest rates, after laying the appropriate groundwork. The groundwork would come in changes to the rate commitment language as Fed officials see developments in resource utilization, the trend of inflation, or inflation expectations – the bill of particulars that was just added last week to this part of the FOMC statement – that tell them inflation risks are rising. As for the rate increases themselves, Fed officials appear convinced that they can lift their target for the federal funds rate and the interest rate on excess reserves (IOER) in tandem with one another, most likely by first reestablishing a small spread with the funds rate over the IOER. (Right now, IOER is 25bp while the funds rate is targeted in a zero to 25bp range and has generally traded about 5-10bp below IOER.)

The theory is that at higher levels of interest rates banks will have an incentive to arbitrage any tendency for the funds rate to drop below the IOER – why lend to another bank at a lower rate when you can lend to the Fed without risk at this rate? This theory did not work particularly well last fall because the GSEs, who are not eligible to receive interest on reserves, lent cash to banks at a federal funds rate well below the IOER. This goes to my point that Bernanke has created so many new tools that we don't really know how they will work during crunch time. Fed officials believe that the unusually stressful circumstances that gave rise to this situation will not recur, at least to a significant degree, in circumstances where financial conditions are stable enough to start increasing interest rates. Fed officials believe a lot of things.

2. Conduct large-scale reverse repurchase agreements. Fed officials have referred several times to such transactions, which would draw reserves out of the banking system as the Fed sold securities out of its massive portfolio with agreements to repurchase those securities at a later date. Reverse repos have been used before to fine-tune the volume of reserves in the system, but on a much smaller scale and for much shorter periods than would now be needed if they concluded that excess reserves should be withdrawn from the system in greater quantity. For example, the largest reverse repo position with dealers to date has been $25bn, and this was in the period immediately following the Lehman bankruptcy. Prior to that, the largest was $4.4bn. Most of these transactions were overnight or a few days at a time. In contrast, using this tool to reduce $1trn in excess reserves obviously requires both larger size and, presumably, longer time periods. Fed officials are reportedly hard at work determining how they would make the necessary adjustments. I'll bet they are. I don't see how this could be conducted without causing interest rates to skyrocket, if they suddenly have to pull reserves quickly. They will have to have kids on the street hawking T-Bills.

3. Offer banks time deposits. The idea would be to “lock up” excess reserves, much as a commercial bank locks up deposits in a certificate of deposit, by offering higher rates on longer maturities. While Fed officials have mentioned this possibility off and on, it has not gotten as much attention as the reverse repos. The key question is how to price it so as to induce banks to give up the liquidity and discourage market participants from reading too much about the Fed’s intentions concerning future interest rates into the maturity profile of its offerings. This would presumably mean either pricing off of market curves such as the LIBOR curve or fixing specific spreads relative to the spot IOER.

4. Sell assets. When this idea first surfaced in the spring, Fed officials were quick to downplay it as an option, noting that such a move would have significant implications for market interest rates and stressing that they were a “buy and hold” institution. (The New York Fed has reportedly backed these words up with actions by hiring portfolio managers for its burgeoning portfolio of MBS.) More recently, however, it has been mentioned as an option, though it appears to be a last resort. If the Fed actually finds someone that will buy the junk they bought off the commercial banks, I want to talk to that person. Boy, do I have some invetments for him.

Many observers assume that asset sales would incur losses for the Fed as well as create problems of sending market signals. A full assessment of this concern is beyond the scope of this already long comment, but we offer two quick observations. First, the Fed has a fairly thick cushion of earnings power. In the fiscal year that just ended, the Fed remitted $34bn in profits to the Treasury. With an expanded balance sheet, this cushion is apt to grow. (Although we’re a bit surprised that the FY 2009 figure was not higher, the September remittance was a record $6.5bn.) Second, and perhaps more importantly, the Fed does own a significant amount of securities on which losses would not be an issue. At a minimum, those maturing within the next year can be redeemed at par as they come due. In this regard, the Fed's Treasury holdings include almost $100bn due within a year, with another $23bn of such paper in its agency portfolio.

Fed and Treasury officials could also reactivate the SFP once the debt limit has been lifted. In short, the Fed has a number of options to manage a high volume of excess reserves. At this point, however, we see no reason why they should be in hurry to do so.

Unless the banks start to aggressively loan out aginst those reserves, I don't see the Fed drainning much of those reserves at all. Thus, although I am not as confident that things are going to go as smoothly as McKelvey believes, McKelvey's scenario is one possibility. But, the real value of his comment is in understanding the true nature of the excess reserves and how they fit into the big picture monetary puzzle.
Special thanks to Anna Greenwood at Goldman Sachs, who provided me a copy of McKelvey's comment.


  1. The FED's policy tool, interest on reserves (IOR's), the FOMC's interest rebate on member bank reserve balances (which Congress made the taxpayers responsible for), has induced widespread dis-intermediation among the non-banks (the most important economic sector in this recession/depression -- or 82% of the lending market, Z.1 release, sectors, e.g., MMMFs, GSEs, overnight repos, etc.).

    I.e., the intermediaries have shrunk in size & the size of the member banks has remained essentially the same.

    The non-banks are financial intermediaries - intermediaries between saver & borrower. The member banks are new money and credit creators (they always create new money in the lending process, member banks do not loan out existing deposits).

    A trillion dollars + in monetary savings (if you count just the verifiable portion in excess reserves), was siphoned out of the non-banks (e.g., hedge funds, investment banks, finance companies, insurance companies, mortgage companies, pension funds).

    The financial press has attributed this to deleveraging. However, the member banks (18% of the lending market, Z.1 release), has suffered no dis-intermediation.
    Monetary savings (savings held beyond the income period), are impounded within the banking system. They are lost to investment, consumption, or to any type of payment (if held in this form).

    I.e., savings held within the monetary system have a transactions velocity of zero, and are a leakage in the Keynesian national income concept of savings. Such a "cessation of circuit income" has adverse effects on production and employment, and requires large dosages of money to counter-act.
    Thus under one view, the quantitative easing performed by the FED (an increase in legal reserves), has been substantially erased.

    But we are not done. If the FOMC raised the average reserve ratios on member bank deposits, the volume of required reserves would increase (which if large enough, could induce bank credit contraction), ceteris paribus.

    This process is the same as if the FOMC raised the remuneration rate on excess & required reserves, vis a’ vis other competitive instruments and yields. It would also increase the volume of legal reserves, ceteris paribus (which also acts to reduce the monetary system's lending capacity).

    Quantitative easing was tried, but there were opposing forces that rendered it immeasurable.

    The solution is to redirect savings to the non-banks, and velocity (consumption & investment), will rebound, without unnecessarily forcing prices (stagflation), higher. This re-routing was successful in the housing crisis of 1966 (such targeted redirection is used in a command economy). In 66, both the member bank's and non-bank's profits were revived, and the housing market (and the economy along side it), recovered thereafter.

  2. I.e., disintermediation is an outflow of funds, deposits, or a negative cash flow out of the non-banks (financial intermediaries). Disintermediation for the member commercial banks can only exist in a situation in which there is both a massive loss of faith in the credit of the banks, and an inability on the part of the Federal Reserve to prevent bank credit contraction, as a consequence of its depositor’s withdrawals.

    The last period of disintermediation for the member CBs occurred during the Great Depression, which had its most force in March 1933. Ever since 1933, the Federal Reserve has had the capacity to take unified action, through its "open market power", to prevent any outflow of currency from the banking system.

  3. In Reuters:

    "There is also a lively debate about whether the Fed should extinguish the reserves or permanently maintain a larger balance sheet, according to Marvin Goodfriend, a former director of research at the Richmond Fed who now teaches at Carnegie Mellon University in Pittsburgh." November 17, 2009

  4. This process is the same as if the FOMC raised the remuneration rate on excess & required reserves, vis a’ vis other competitive instruments and yields. It would also increase the volume of legal reserves, ceteris paribus (which also acts to reduce the monetary system's lending capacity).

    Quantitative easing was tried, but there were opposing forces that rendered it immeasurable.
    The solution is to redirect savings to the non-banks, and velocity (consumption & investment), will rebound, without unnecessarily forcing prices (stagflation), higher. This re-routing was successful in the housing crisis of 1966 (such targeted redirection is used in a command economy).
    Tax Reserves