Friday, October 20, 2017

The Other Side of the Fed's Balance Sheet

Who controls the Fed's balance sheet? The answer may seem obvious. The Fed, after all, determines the size of its balance sheet. It also controls what happens to the asset side of its balance sheet. Its power over the liability side, however, is limited.

This diminished control arises because the public's demand for currency, bank regulations, and U.S. Treasury cash balances all influence the composition of the Fed's liabilities.1 These are exogenous forces that have the potential to create some economic bumps on the road ahead as the Fed normalizes the size of its balance sheet. 

So far, though, little attention has been paid to these liability-side issues. Most focus has been given to the asset side of the Fed's balance sheet. This focus, in my view, is misguided. I see the real dangers lurking on the liability side of the Fed's balance sheet--the very side where the Fed has less control. 

This post, then, is attempt to direct some attention to the liability side of the Fed's balance sheet. In it, I first explain how the public's demand for currency, bank regulations, and the U.S. Treasury cash balances are beyond the Fed's control but affect its balance sheet. I then explain how these forces and current Fed policy could interact in a disruptive manner.  

The Public's Demand for Currency
Consider first the public's demand for currency. It grows as the dollar size of economy grows. This growing demand for currency is met by banks turning reserves into currency. Some of this growing demand for currency comes from foreigners and some from domestic residents. As seen in the figure below, both have grown over the past decade. In total, currency grew from around $800 billion in 2007 to $1.58 trillion today. The almost $800 billion increase in currency means, ceteris paribus, a similar-sized decline in reserves over the same period.

This development matters because the nearly $800 billion loss of reserves and gain in currency, though fairly predictable, was largely beyond the Fed's control. Put differently, the Fed controls the initial level and form of bank reserves, but it loses control over time. If the Fed were trying to stabilize a certain level of reserves--say to keep its floor system working--this steady growth of currency demand makes it harder. This is a powerful force with which the Fed has to contend. 

Now the currency demand growth may be seen as plus since it reduces the amount of balance sheet reduction required going forward. But it also poses some challenges that I will discuss later. For now, note that the growth in currency demand is an exogenous force that steadily tugs at the composition of the liability side of the Fed's balance sheet. 

New Bank Regulations
There have been a lot of new bank regulations since the crisis, but here I want to focus specifically on the liquidity coverage ratio (LCR). It requires banks to hold enough "high-quality liquid assets" (HQLA) to withstand 30 days of cash outflow. Unsurprisingly, the LCR has increased demand for HQLA assets. In particular, it has increased demand for bank reserves and treasury securities which can be held at full value (other assets, like GSEs can be held but at a discount). The LCR was implemented in January 2015.

Per the LCR, bank reserves and treasuries are equally safe and banks should be indifferent between holding them for regulatory reasons. Per the banks, however, they are not the same. The IOER paid on excess reserves has been consistently higher than the yield on short-term treasury bills as seen below. This raises the demand for bank reserves relative to treasury securities.  

So the IOER being greater than other short-term market rates raises the relative demand for bank reserves and the LCR, which is beyond the Fed's control, intensifies this demand. 

U.S. Treasury Cash Balance
The U.S. Treasury Department is also an exogenous force affecting the Fed balance sheet. There are several channels of influence, but here I want to focus on Treasury's cash balances that are deposited at various places. These deposits can broadly grouped into two categories: (1) interest-earning deposits held at private financial firms and (2) non-interest earning deposits held at the Fed in the Treasury General Account (TGA).

Prior to 2008, Treasury kept most of its cash balances outside the Fed. This earned interest for the taxpayers and also made life easier for the Fed since the TGA was small. After 2008, this pattern reversed: Treasury parked most of its cash balances at the Fed and vastly expanded their size. This can be seen in the figure below:

The main reason for this big shift in 2008 was that the IOER rate has been higher than short-term market interest rates. Here is why this matters. Treasury funds deposited at private banks become reserves that get redeposited at the Fed so that banks can earn IOER. The IOER payments going from the Fed to the banks are payments not going to Treasury. Moreover, the IOER payments exceed what Treasury can earn on their deposits at private banks because IOER exceeds short-term deposit rates.2 

Treasury, consequently, was losing income after IOER. Treasury officials quickly realized they could minimize this loss by pulling their funds out of private financial firms and instead park them at the Fed in the TGA.By doing so, however, Treasury officials were pulling out reserves from the banking system and therefore shrinking the aggregate level of bank reserves. This change in the composition of the liability side of the Fed's balance sheet was beyond the Fed's control. 

Putting it All Together
So we have we have currency demand growth, LCR, and the TGA all exogenously affecting reserves on the liability side of the Fed's balance sheet. In addition, the Fed is affecting the demand for reserves with IOER being greater than market interest rates. The Fed is also draining reserves via its balance sheet reduction program and its overnight reverse repo program (ON-RRP). The table below summarizes these developments.

The figure below shows how these forces have affected the growth of the liability side of the Fed's balance sheet so far:

Potential Disruptions Ahead
These developments matter because, as seen in the table, they are pushing the supply and demand for reserves in opposite directions. 

Bank reserves have been shrinking because of currency demand growth, the TGA, and ON-RRP. At the same time the demand for reserves has been elevated because IOER is greater than short-term market rates. Since 2015, this demand has been elevated because of the LCR. And now the Fed is about to further heighten this imbalance by draining reserves as it begins shrinking its balance sheet. 

The figure below shows this is more than just a theoretical concern. It shows the level of bank reserves since 2015 (when the LCR started) plotted against the spread between the overnight dollar libor rate and the 1-month treasury rate. There is a negative relationship which means as reserves fall the libor rises relative to the treasury bill rate. The increases are not terribly large and this only explain a third of the variation in the spread since 2015, but it does suggest further tightening will occur moving forward if nothing else changes as the Fed shrinks its balance sheet. 

So what should the Fed do? It could curtail the shrinking of its balance sheet, but that is not the path I would chose. The balance sheet reduction is already baked into market expectations and it should continue for other reasons outlined here

What it should do, in my view, is aim to bring the IOER closer to market interest rates. I am not sure how easy this task would be, but it would reduce the heightened demand for reserves, lower the TGA, and soften the bite of the LCR.  One way to help push the IOER closer to market rates would be for the Fed to announce a corridor system as the final destination for the Fed's balance sheet reduction. This has not happened yet, but it should be something the FOMC seriously considers.

1 There are some other exogenous forces on the liability side that I ignore here. The biggest one being foreign official accounts at the Fed. These, however, generally are not too large and therefore less consequential.
2 This was due both to IOER being greater than short-term market interest rates, but also because by law Treasury can only earn 25 basis points less than the federal funds rate at depository institutions. Treasury can also park funds in repos and in term deposits, but even these earn less than the IOER. See this NY Fed piece for more on Treasury cash balances.
3The IOER incentive to park funds in the TGA was reinforced by new Treasury policy in 2015 that raised the minimum size of the account to $150 billion. 


  1. The error is in the Fed's definition of the money stock. It is inaccurate (for the cataloguer of economic statistics) to exclude the Treasury’s General Fund Account from the assets included in M1 (with the exception of WWII). No one has established any unique price effect of federal outlays, as compared to state and local government outlays, or expenditures by the private sector. Of course, the shifting of funds to and out of the Federal Reserve banks has a dollar for dollar effect on member bank reserves, but that is another problem that can be, and is dealt with through open market operations.

  2. The essence of our managed-currency system, is a system in which the volume of currency in circulation is impersonally determined by the total effective demands of the public or the amount which meets most closely the needs of trade.. The basic process by which currency is put into and taken out of circulation is through the banking system. The volume of currency held by the public needs no formal or specific regulation since it is impossible for the public to acquire more of a given type of currency without giving up other types of currency, or else bank deposits. In other words, under our managed system it is impossible for the public to add to the total money supply consequent to increasing its holdings of currency.

    This particular procedure distinguishes a managed currency system from a fiat currency system and makes our currency self-regulatory. While the money supply as a whole is not self-regulatory and must be kept under careful surveillance and control by the monetary authorities, the currency part of our money supply can be left free to seek its own level.

    Even in a managed system, however, governmental fiscal, monetary, and debt-management practices can influence the volume of currency held by the public, although the volume cannot be determined by any governmental edict, law, or administrative practice. The public’s holding of currency will be influenced by governmental practices which change public expectations in regard to such matters as prices, income, and employment.

  3. "What it should do, in my view, is aim to bring the IOER closer to market interest rates."

    Isn't that getting the causality wrong? Looking at e.g. your second chart (IOER, overnight Dollar Libor, 1 Month treasury yield) I get the distinct impression that IOER (or alternatively Fed Funds Target) is the anchor relative to which other market rates are priced. In other words, it seems to me that there is no such thing as a market rate that exists independently of IOER (or alternatively Fed Funds Target) and closer to which the Fed's rates can be brought.

  4. The DFIs once became unconstrained by the volume of legal reserves c. 1995. In 1990/1991 the Fed reduced reserve requirements by 40 percent and then thru reserve avoidance, sweep accounts further reduced reservable liabilities (transaction based accounts).

    See: “Are U.S. Reserve Requirements Still Binding?”

    But this trend has reversed since the GFC. Whereas the DFIs satisfied 85% of their required reserves on 12/2007 (virtually the peak), they now apply 34% of their vault cash towards satisfying legal reserve requirements (c. 1981 #s).

    So the Fed should publish a listing of their District Reserve bank master accounts (identified by a primary nine-digit Routing Transit Number, RTN), by member bank, by respondent bank's segregated pass thru accounts at their correspondent bank, by IBDD volumes (and differentiate domestic vs. FBOs).

    This skewing of IBDDs may be analogous to the 1920’s endogenous redistribution of deposits between former Central Reserve City Banks, vs., Reserve City Banks, vs. Country Banks, or the oligarchic skewing of IBDDs by money center banks and foreign bank branches, once they became remunerated). The geographical mismatch and redistribution of deposit-taking and lending activities should be minimized and banks should be required to make loans in the communities where they collect deposits.

    The maligned distribution of IBDDs held by the member banks, at their District Reserve bank, between the former Central Reserve City Banks, vs., Reserve City Banks, vs. Country Banks convoluted structure, precipitated bank panics, accumulated pressures between respondent/correspondent relationships, as the Fed’s country banks classification held upwards of ½ of all vault cash. And today, the FBOs hold roughly 40 percent of all IBDDs.

  5. The money stock (& DFI credit, where: loans + investments = deposits), can never be managed by any attempt to control the cost of credit [or thru a series of temporary stair stepping or cascading pegging of policy rates on “eligible collateral”; or thru "spreads", "floors", "ceilings", "corridors", "brackets", IOeR, etc.].

    Using a price mechanism (interest rates as a monetary transmission mechanism), to ration Fed credit is non-sense. The effect of current open market operations on interest rates (now via the remuneration rate), is indirect, varies widely over time, and in magnitude. What the net expansion of money will be, as a consequence of a given change in policy rates, nobody knows until long after the fact. The consequence is a delayed, remote, & approximate control over the lending and money-creating capacity of the banking system.

    Economists should have learned the falsity of that assumption in the Dec. 1941-Mar. 1951 period. That was what the Treas. – Fed. Res. Accord of Mar. 1951 was all about. Keynes' liquidity preference curve (demand for money) is a false doctrine. Reserve targeting worked well until William McChesney Martin Jr., abandoned the FOMC’s net free, or net borrowed, reserve targeting position approach in favor of the Federal Funds “bracket racket” beginning in c. 1965 (coinciding with the rise in chronic monetary inflation & in anticipated inflation).

    The only tool at the disposal of the monetary authorities, in a free capitalistic society, through which the volume of money can be controlled is legal reserves (not interest rate manipulation). And legal reserves ceased to be temporarily binding c. 1995 – at the start of the housing boom/bust. By using the wrong criteria (interest rates, rather than member bank reserves) in formulating and executing monetary policy, the Federal Reserve became the real-estate bubble’s engine. The "administered” or actual prices would not be the "asked" prices, were they not “validated” by money flows, i.e., “validated” by the world's Central Banks.

  6. Interest is the price of loan-funds (the market's bailiwick). The price of money is the reciprocal of the price-level (the Fed's bailiwick).

    See: Dr. Daniel L. Thornton, Vice President and Economic Adviser: Research Division, Federal Reserve Bank of St. Louis, Working Paper Series, “Monetary Policy: Why Money Matters and Interest Rates Don’t”

    And we knew this already:

    Monetary policy objectives should be formulated in terms of desired rates-of-change, roc's, in monetary flows, M*Vt (volume X’s velocity), relative to roc's in R-gDp. Roc's in N-gDp (though "raw materials, intermediate goods and labor costs, which comprise the bulk of business spending are not treated in N-gDp"), can serve as a proxy figure for roc's in all transactions, P*T, in Professor Irving Fisher's truistic: "equation of exchange".

    And Alfred Marshall's cash-balances approach (viz., a schedule of the amounts of money that will be offered at given levels of "P"), viz., where at times "K" is the reciprocal of Vt, or “K” has the dimension of a “storage period” and "bridges the gaps of transition periods" in Yale Professor Irving Fisher’s model. Roc's in R-gDp have to be used, of course, as a policy standard.