QE: Exit-path implications for collateral chains

By Peter Stella

This article first appeared at Vox.

QE is still on, but central banks are pondering exit pathways. Exit requires vacuuming up excess reserves, winding down massive securities holdings, and restoring normal interest rates – all without killing the recovery. This column points to the importance of a seemingly technical issue – the impact of the exit on the supply of high-quality collateral. This matters since collateral plays a critical role in today’s credit and money creation processes. When reducing excess reserves, the ‘how’ matters as much as the ‘when’ and ‘how much’.

Five years after the demise of Lehman Brothers there remains considerable uncertainty as to how and when central banks will exit their extraordinary interventions. Nevertheless, a few aspects of the road ahead have come into sharp focus.

Central banks to exit by raising interest rates, not by shrinking balance sheets

To raise policy rates without reducing their balance sheets, central banks will proceed down one of three paths:

  • Adopt an administrated policy rate target and raise it by fiat;
  • Return to the globally predominant pre-crisis operational interest-rate target – the overnight interest on borrowed bank reserves – and raise it by first tightening reserve supplies; or
  • Adopt an alternative market-based operational target such as a general collateral repurchase rate.

Each exit path would be accompanied by its own operational environment and implications for the financial system. In order to concentrate the discussion I focus on the challenge facing the US Fed.

What needs exiting?

The Federal Reserve balance sheet is much simpler than at the height of the crisis.1

  • If Rip Van Winkle awoke today after a five-year nap, he would see only one key development in the Fed’s balance sheet – securities holdings higher by $2.9 trillion and deposits of depository institutions (banks) higher by $ 2.2 trillion.2

If he asked how this happened, Rip would be given a very simple answer.

  • The Fed bought securities to lower interest rates; it paid for them by creating bank reserves.

That is, the Fed credited the securities seller’s commercial bank with a deposit at one of the 12 Federal Reserve Banks, and the commercial bank then credited the seller’s account. On net, privately held securities were exchanged for Fed deposits.

If pressed further as to why banks are holding enormous reserves at the Fed, Rip would get an equally simple answer: Banks have no choice.

  • It is – for all intents and purposes – technically and legally impossible for a bank to transfer deposits at a Federal Reserve Bank to a nonbank.

Reserves in the US are defined to comprise bank deposits held at one of the 12 Federal Reserve Banks (plus qualifying vault cash).

  • Fed deposits may be transferred only to entities entitled to hold Fed accounts.

This is a key point when thinking about the various exit paths ahead. Fed deposits are not fungible outside the banking system, but Treasuries are.

A massive portfolio effect

Large Scale Asset Purchases (LSAPs) have inadvertently caused a significant change in the composition of assets available in the open market.

  • The stock of marketable, highly liquid, AA+ collateral fell by trillions (disappearing into the Fed’s portfolio, i.e. System Open Market Account).
  • The stock of assets available only for interbank trade (bank reserve deposits at the Fed) rose by trillions.

The essence of this change has nothing to do with credit risk. Treasuries and Fed deposits are equally safe. But they differ significantly in their marketability. Anyone can trade Treasury securities; only banks can exchange Fed deposits. The massive importance nonbanks play in today’s markets means that this portfolio effect can be important. To understand why marketability beyond banks matters, it is necessary to understand the modern money and credit creation process.

Reserves and money supply: Probably not what you learned in school

One cannot think straight about the future impact of different exit strategies without understanding of the role of bank reserves in today’s financial markets.

  • Banking and money creation has not worked for at least two decades in the way that most people learned in school.

The old system was rather simple in the textbooks. The basic assumptions were (i) all credit was provided by banks; (ii) all bank credit (assets) were funded by the issuance, or creation, of depository liabilities (money) subject to a reserve requirement; and (iii) central banks controlled credit/money/inflation by rationing bank reserves. A stable ‘money multiplier’ was hypothesised to allow central banks to accurately predict the eventual impact of changes in bank reserves on money and credit.

The problem with the old theory of monetary operations is that none of the three assumptions has been true for at least a generation.

Most credit in the US is created by nonbanks; virtually all bank lending is funded by the creation of liabilities that are not subject to reserve requirements,3 and central banks do not ration reserves. In fact they take great pains to provide banks with the amount of reserves they desire. Central banks influence credit not by rationing the quantity of reserves but by altering the interest rate that banks must pay to obtain the quantity of reserves they desire.

  • Today, credit creation in general and money creation in particular are no longer tied to the stock of reserves (i.e. the stock of banks’ deposits at the Fed).

Today, bank deposits at the Fed have only one real role – to facilitate management of the payments system. They are used to settle transactions among banks. Thus:

  • The old notion that the quantity of bank reserves constrains lending in a fiat money world is completely erroneous.
  • Traditional monetary policy has virtually nothing to do with money.4

This is clearly seen in the long-term evolution of reserves and credit.

  • In 1951, total commercial bank deposits at the Fed were $20 billion larger than they were at the end of 2006.
  • Over the same period, total US credit-market assets rose by over 10,000%.5

Plainly the stock of reserves is no longer connected to credit or meaningful measures of “money” via the old-notion of a reserve-ratio-based money multiplier.

Excess Fed deposits as ‘deadwood’ on bank balance sheets

One of the unintended consequences of Fed LSAPs has been the withdrawal of high quality liquid collateral such as US Treasuries from the financial markets paid for by crediting commercial bank reserve accounts. As discussed above, the banking system as a whole cannot dispose of these assets (reserves). At the same time, banks are under massive pressure world-wide to deleverage. This can take place either by increasing capital (a bank liability), which is costly to shareholders, or by reducing assets. Thus banks’ massive holdings of reserves at the Fed are ‘deadwood’ as far as the banks and their credit-creation capacity are concerned. They may crowd out credit.

The deadwood problem will get worse if the US tightens regulatory leverage ratios – that is, reduces the maximum ratio permitted between a bank’s total assets and capital.6

There is a great irony in the journalistic history of monetary policy. What many are calling central bank “money creation” “helicopter money” or “rolling the printing presses” may – in combination with tighter leverage ratios – lead to a tightening of bank credit and deflationary pressures. And all this is occurring while the spectre of uncontrolled credit expansion and monetary debasement are being decried countless times by those who have not recognized that yesteryear’s monetary paradigm is defunct.

Exit and collateral: The operational challenges

The operational exit challenge facing the Fed is how best to engineer an increase in interest rates while:

  • Alleviating the collateral shortage in the nonbank sector; and
  • Reducing the deadwood on the asset side of commercial-bank balance sheets called ‘reserves’.

We have had several indications as to how the challenges may be met.

  • In September 2008, Federal Reserve Bank of New York economists outlined one exit path (Keister, Martin, and McAndrews 2008).7

The Fed would pay interest on bank reserves and simply adopt the rate paid as the operational target. Note that this paper was released two months before the Fed started paying such interest – known as ‘interest on reserves’, or IOR to specialists.8

  • Since the rate would simply be announced by the Fed, it would not be the result of a market process and therefore could be raised regardless of the level of bank reserves in the system.

It would, from an operational standpoint, be stunningly simple.

Weighing against the operational simplicity of this interest-on-reserves-centred plan are many drawbacks.

  • There is a major unknown associated with shifting to a rate determined by fiat and thus divorced from market conditions.
  • It might have little or no predictable impact on the term structure of interest rates – especially longer term interest rates which are the ones that matter most for the real economy.

The key point here is that the quantity of reserves in no way establishes a technical obstacle to using interest-on-reserves as a target.
While paying interest was the first exit path mooted, recent Fed actions seem to indicate that they intend to target something other than ‘interest on reserves’ during the exit.

The Fed’s likely interest target during exit

Given the time and effort the Fed has spent developing and refining new reserve-draining tools, it seems that the Fed intends to either:

  • Return to the fed funds effective interest rate as its operational target; or
  • Adopt an alternative market-determined interest-rate target.

In this scenario, the interest-on-reserves tool would only support the achievement of those targets.

In order to re-establish the fed funds target rate as an effective policy signal, the huge excess bank reserves overhang must be eliminated. The tools with which the Fed has been experimenting include:

  • Term deposits.

Federal Reserve Banks would offer ‘term deposits’ to banks through the Term Deposit Facility. When banks place funds into such term deposits, reserves are removed from the Fed accounts of participating institutions for the life of the term deposit. This is therefore a way of draining reserve balances from banks.9

  • Reverse repurchase (repo) transactions.

With these, the Fed sells Treasury bonds to financial market players and – simultaneously – agrees to buy them back on a set day in the future. Since the Fed would pay a slightly higher price when it repurchases the bonds, this is a way for the Fed to effectively pay interest in the same way it would on a term deposit, with the important additional flexibility that it could agree a reverse repo not only with banks but with nonbanks who could otherwise not place cash at term with the Fed. That is, in both cases the Fed would reduce the quantity of bank reserves but with reverse repos it could replace its liability to a bank (which from the bank’s viewpoint is an asset at the Fed) with a liability to a nonbank.

The Fed has spent much time and effort to develop and refine these tools including expanding its list of approved counterparties for reverse repos – and this suggests that the Fed intends to use them. That is, the Fed seems set to drain reserves from the system, so the Fed Funds rate starts to again reflect general credit-market conditions.

Collateral matters: Critical differences between term deposits and reverse repos

Term deposits and reverse repos might appear very similar reserve draining tools to the proverbial Rip Van Winkle who slept through the radical changes in credit creation, collateral chains, and expansion of the shadow banking system.

  • To Rip Van Winkle, term deposits and reverse repos are both simple transformations of overnight deposits into 7, 14 or 28 day “term” deposits, or to 7-, 14-, or 28-day reverse repos.

Well informed observers, however, will immediately see the critical difference (a good start on getting informed is to read Singh and Stella 2012).10

  • A reverse-repo has a portfolio effect that term deposits do not.

The reverse-repo takes cash out of the market and replaces it with high-quality collateral. Thus a key difference is the enlargement of the market supply of good collateral that banks and nonbanks would posses and thereby be able to use to fund their asset positions and create credit via ‘collateral chains’. (Technically this re-lending of collateral is called re-hypothecation.) Thus reverse repos add to financial lubrication in a way term deposits do not.11

This ‘sweetener’ would soften the bitter medicine of rising interest rates and alleviate the so-called ‘high quality collateral shortage’ pointed out by Treasury Borrowing Advisory Committee members (US Treasury 2013).

Rough magnitude of the task: $2-3 trillion

What is the rough magnitude of the task if the Fed balance sheet were to remain at its current size?

  • In the two weeks ending 27 August 2008, average daily reserves held by depository institutions (banks) were $46.1 billion; required reserves were $ 44.1 billion.12 Of this, vault-cash used to satisfy required reserves was $36.4 billion and reserve balances held at FRB were $9.7 billion.
  • In the two weeks ending 21 August 2013, average daily reserves held by banks were $2.2 trillion – that’s trillion with a ‘t’; required reserves were $115 billion – that’s billion with a ‘b’. Of this, vault cash used to satisfy required reserves was $53.4 billion and reserve balances held at FRB were $2.1 trillion.13
  • Thus bank excess reserves rose by $2.123 trillion during the last five years.

Of course, the Lehman bankruptcy convinced banks that they want more excess reserves than before, so the Fed will not need to drain the full $2.123 trillion. Let us say the task is to eliminate about $2 trillion in excess reserves. But of course the world does not stand still. The number could be closer to $3 trillion by the time LSAPs end.

What this means is that the change in the availability of good collateral could be quite significant.

Alleviating deadwood bank assets

The deadwood problem on commercial bank balance sheets would not be alleviated if the Fed’s cash draining operations were confined to banks. 

  • Bank reserves would merely be replaced with term deposits or repo lending to the Fed on bank’s balance sheets.
  • Fed transactions with nonbanks, however, are a different story.

Fed deals with nonbank financial institutions would transform their net claims on banks into claims on the Fed (repo financing).

  • As agents of the nonbanks, banks would execute payments for winning bidders at the repo auctions.
  • Reserves would fall on the commercial bank asset side and net bank liabilities to nonbanks would decline by an equivalent amount.

In this way, banks’ deadwood assets would be pruned, enabling an expansion of credit to the real economy – despite rising interest rates.
In sum, the intermediary role of banks between nonbank financial-market players and the Fed would be lessened.

  • Bank-leverage ratios would be less binding;
  • High-quality collateral would be provided to the market; and
  • Central-bank control over an operational interest rate would be re-established.

Only the latter would be accomplished by providing term deposits to banks only.

Concluding remarks

Many major central banks are thinking strategically about exit pathways – how best to return to normal central banking. The main point of this column is to point to a key issue – the role of collateral – that has been under appreciated by many economists who are not in daily contact with financial markets.

When economies strengthen and central banks begin to drain reserves from the system, they will inevitably alter the composition of private sector asset portfolios.

  • If good collateral is swapped for reserves, banks and nonbanks can use the collateral to fund create credition via what are known as collateral chains.
  • If only term deposits are swapped for reserves, or if interest rates are raised only through IOR, the opportunity to lengthen collateral chains will be missed.

In today’s financial world, these chains are critical sources of money and credit creation – the days of textbook money-multipliers are long gone.

When it comes to reducing excess reserves, the ‘how’ matters as much as the ‘when’ and ‘how much’. Understanding this point requires mastery of the brave new world of shadow banks and re-hypothecation – a world that either did not exist or was truly in the shadows when most of us were taught about money and credit creation.


Stella, Peter (2009). “The Federal Reserve System Balance Sheet — What Happened and Why it Matters” IMF WP 09/120.

Stella, Peter (2013). “Bank of Japan Theater: Is this Kabuki after Noh?” https://stellarconsultllc.com/blog/wp-content/uploads/2013/05/Bank-of-Japan-Theater.pdf

Carpenter, Seth and Selva Demiralp (2012). “Money, Reserves, and the Transmission of Monetary Policy: Does the Money Multiplier Exist?”, Journal of Macroeconomics 34:1.

Claessens, Stijn and Lev Ratnovski (2013). “What is shadow banking?”, VoxEU.org, 23 August.

Keister, Martin, and McAndrews (2008), “Divorcing Money from Monetary Policy” Federal Reserve Bank of New York Quarterly Review, September.

US Treasury (2013). “Minutes of the Meeting” Treasury Borrowing Advisory Committee of the Securities Industry and Financial Markets Association, US Treasury web site, https://www.treasury.gov/press-center/press-releases/Pages/jl1923.aspx, 30 April 2013.

Singh, Manmohan and Peter Stella (2012). “The (other) deleveraging: What economists need to know about the modern money creation process”, VoxEU.org, 2 July.

Singh, Manmohan (2011). “Velocity of Pledged Collateral: Analysis and Implications”, IMF WP/11/256.

1 The alphabet soup of special programs and facilities — the TALF, TAF, CPFF, SFP, and the three Maiden Lanes – have largely been replaced by Treasuries, US Agency, and Mortgage Backed Securities. The counterpart on the liability side has been an enormous increase in bank reserves – deposits at Federal Reserve Banks.

2 Most of “shortfall” in liability growth is explained by the increase in currency – $ 367 billion.

3 See Carpenter and Demirlap (2012), also Claessens and Ratnovski (2013) who discuss bank funding from money market funds, hedge funds and other nonbank elements of the “shadow banking” system.

4 Traditional monetary policy is more accurately thought of as the control of the interest rate at which the central bank provides the reserves (over which it has a monopoly of creation) to banks.

5 See Stella (2009). Stella (2013) discusses similar issues in the Japanese context.

6 Large holdings of reserves do not cause any problem for banks with respect to their risk-weighted asset to capital ratio – the main regulatory ratio—because the weight on risk-free assets, such as reserves, is zero. Therefore reserves effectively fall out of the numerator of the ratio. Leverage ratios treat all assets the same so that reserves take up just as much balance sheet “space” as a risky loan. Although leverage ratios are “crude”, regulators are emphasizing leverage ratios post-crisis owing to doubts about their ability to accurately gauge risk and determine appropriate risk weights for opaque assets.

7 Keister, Martin, and McAndrews (2008), “Divorcing Money from Monetary Policy” Federal Reserve Bank of New York Quarterly Review, September.

8 See https://www.federalreserve.gov/monetarypolicy/reqresbalances.htm.

9 See https://www.federalreserve.gov/monetarypolicy/tdf.htm

10 Singh and Stella (2012) explain: “One of the financial system’s chief roles is to provide credit for worthy investments. Some very deep changes are happening to this system – changes that surprisingly few people are aware of. This column presents a quick sketch of the modern credit creation process and then discusses the deep changes that are affecting it”. Also see Claessen and Ratnovski (2013).

11 See Singh (2011) Velocity of Pledged Collateral: Analysis and Implications IMF WP/11/256.

12 Fed Board release H.3 Table 2, September 25, 2008.

13 Fed Board release H.3 Table 2, September 4, 2013.

Comments are closed.

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More