Policy Implementation with a Large Central Bank Balance Sheet

Policy Implementation with a Large Central Bank Balance Sheet Antoine Martin The views expressed herein are my own and may not reflect the views of the Federal Reserve Bank of New York or the Federal Reserve System

Outline Monetary policy implementation before 2008 Monetary policy implementation since 2008 The leaky floor Why are banks holding so many excess reserves? Monetary policy normalization Framework Tools Some implications for the long-term framework Is it costly to have a large balance sheet? 2

Monetary policy implementation before 2008 Interest rate DW rate Target rate Demand for reserves 0 Required reserves Target supply Reserve balance 3

Pre-Crisis Operational Framework The level of reserve balances was quite low Averaged around $10 billion in 2006 Desk and Board staff forecast factors driving supply of and demand for reserves Almost every day, conduct a repo operation to add enough reserves to hit the fed funds target rate Averaged about $5-10 billion per operation As demand for currency grew, banks reserves would decrease, and the Desk would add reserves through purchases of Treasuries (long-run) and repos (short-run) 4

Framework was similar to a corridor system Interest rate DW rate Demand for reserves Target rate IOER 0 Required reserves Target supply Reserve balances 5

The crisis lead to a huge increase in reserves 3000 Reserve Balances with Federal Reserve Banks 2500 (Billions of Dollars) 2000 1500 1000 500 0 2007 2008 2008 2009 2011 2012 2012 2013 2015 7

Monetary policy implementation since 2008 Interest rate Supply of reserves is not linked to target rate DW rate IOER 0 Supply Reserve balance 8

Interest on excess reserves (IOER) Relatively recent authority granted to the Fed in 2008 Allows the Fed to pay interest to banks on the balances that they leave in their account with the Fed How does it work? If banks can earn, say, 0.25% leaving money in their Fed account, they won t have an incentive to lend it out below that rate Should provide a floor on interbank rates 9

Current Framework The level of reserve balances is much higher Around $2.7 trillion; a 27,000% increase! Little need to forecast factors driving supply of and demand for reserves each day But market interest rates, including the fed funds rate, have been below IOER No need to conduct a repo operation to set the supply of reserves to the desired level Relatively small-scale adjustments to the supply of reserves might do little to move the fed funds rate Large-scale asset purchase programs have meant that currency is no longer our largest single liability 10

Reserves and Money Market Rates Rate (Percent) 0.1.2.3 GC Treasury Overnight, Effective Fed Funds, IOER, Eurodollar, and Total Reserves 01 Jan 2010-30 Aug 2015 Jan 10 Oct 10 Aug 11 Jun 12 Apr 13 Jan 14 Nov 14 Sep 15 Date -3000-2500-2000-1500-1000 Reserve Balances (Billions) GC Treasury Overnight Fed Funds (FF) Eurodollar (ED) IOER - (Total Reserves) Sources: FRED, Federal Reserve Economic Data, from the Federal Reserve Bank of St. Louis, Bloomberg, and Federal Reserve Data Releases, H15 11

The leaky floor Interest rate Supply of reserves is not linked to target rate DW rate IOER 0 Supply Reserve balance 12

Why are market rates below the IOER? Why is the federal funds (FF) rate below IOER? GSEs and FHLBs have Fed accounts but cannot earn interest on reserves Why are other market rates below IOER? Most cash lenders in U.S. money markets are non-banks, including MMFs, and do not have a Fed account 13

Why doesn t competition lift market rates Absent frictions, we would expect competition to pull market interest rates close to IOER Two frictions appear to limit the pull of IOER: Banks face balance sheet costs, related to new Basel III regulation U.S. money market appear to be imperfectly competitive 14

Why are banks holding excess reserves? Short answer: Because they have no choice Reserves are injected in the system when the Federal Reserve purchases assets Reserves must be held by institutions that have a Fed account Banks can redistribute the reserves among themselves but cannot change the aggregate amount Except for turning reserves into currency The analysis that follows is based on Keister and McAndrews (2009) 15

Before 2008: Low amount of reserves Suppose there are two banks, A and B Balance sheets in normal times: Bank A Bank B Reserves 10 Deposits 100 Reserves 10 Deposits 100 Loans 50 Loans 130 Due to Due from Bank A 40 Bank B 40 Securities 10 Capital 10 Securities 10 Capital 10 Note: Interbank lending promotes the efficient allocation of resources 16

The crisis Suppose Bank A is no longer willing to lend to Bank B May have concerns about credit risk or about its own funding needs If Bank B cannot replace this funding, it must reduce lending Bank A Bank B Reserves 50 Deposits 100 Reserves 10 Deposits 100 Loans 50 Loans 90 Due to Due from Bank A 0 Bank B 0 Securities 10 Capital 10 Securities 10 Capital 10 17

Central bank response to the crisis Suppose the central bank lends to Bank B The decrease in lending can be avoided This is the lender of last resort function of the central bank Bank A Bank B Reserves 50 Deposits 100 Reserves 10 Deposits 100 Loans 50 Loans 130 Due to CB 40 Securities 10 Capital 10 Securities 10 Capital 10 Note: total reserves are now $60, $40 of which are excess The Central Bank s policy is highly effective, even though it generates a large amount of excess reserves 18

Bank lending and reserves But isn t there a sense in which the $50 of reserves in Bank A s account are idle? Don t excess reserves represent unused capacity in the banking system? Suppose Bank A lends $20 to Firm X, which is a customer Bank A Bank B Reserves 50 Deposits 120 Reserves 10 Deposits 100 Loans 70 Loans 130 Due to CB 40 Securities 10 Capital 10 Securities 10 Capital 10 No change in reserve positions! 19

Now suppose Firm X buys $20 in equipment from Firm Y Firm Y holds its account at Bank B Bank A Bank B Reserves 30 Deposits 100 Reserves 30 Deposits 120 Loans 70 Loans 130 Due to CB 40 Securities 10 Capital 10 Securities 10 Capital 10 Reserves (and deposits) are transferred to Bank B Again, no change in total reserves The total level of reserves is determined entirely by the actions of the central bank (almost) reveal nothing about the lending behavior of banks 20

Monetary policy normalization Questions: Why are money market rates below IOER? Do we have the tools to raise market rates? If so, what is the most effective way to use our tools? 22

Exit and Patching the Leaks Last September the FOMC published its Policy Normalization Principles and Plans. Fed intends to Keep a target range for the FF rate Move the FF rate primarily by adjusting IOER Use overnight reverse repos and other tools as needed Reduce securities primarily by ceasing reinvestments 3 tools have been discussed recently: IOER, overnight RRP, and TDF What are they? 23

A framework We build a simple model with households, firms, banks, and nonbanks In our model: Households use the financial system to save for consumption at a later date Firms produce consumption goods Banks lend to firm, offer deposits, and have access to IOER Nonbanks hold government debt on behalf of households and cannot earn IOER 24

Key Frictions Absent frictions, all money market rates would be equal to the IOER There are three main frictions: Banks face convex balance sheet costs: Explains why money market rates are below IOER Market for reserves may not be perfectly competitive Banks face convex interbank monitoring costs (reserve scarcity): Explains why interbank rates are above IOER Nonbanks do not face balance sheet costs because they are more transparent 25

A Simple Economy Households Sell endowment to firm Save through banks and non-banks Buy goods for consumption later Banks may experience a liquidity shock They may use reserves to absorb the shock or borrow in the interbank market Firms Buy household endowment and use it as input for production Sell goods to the households when they want to consume 26

Two polar cases When the supply of reserves is small Interbank market is active Interbank market frictions are high Balance sheet costs are low Because of scarcity, banks compete for reserves When the supply of reserves is high Interbank market is inactive Interbank market frictions are low Balance sheet costs are high Banks don t need to compete for reserves 27

New Federal Reserve Tools We study two tools (tested recently) Reverse Repurchase agreements (RRPs) Banks and non-banks, such as MMFs, can lend to the Fed (against collateral) We focus on fixed-rate overnight RRPs Term Deposit Facility (TDF) Banks can deposit reserves with the Fed for a term maturity We study other tools in the paper and could adapt the framework to study additional tools 28

Overnight RRP A tool we have used for decades New wrinkle: instead of operating in fixed size, operate at a fixed rate with a broader set of firms How does it work? In tri-party repo system, investors give the Fed cash overnight and we give them Treasury securities as collateral (essentially, a collateralized loan to the Fed) Used to support a floor under rates Similar to IOER, if an investor can earn, say, 0.05% investing at the Fed, why invest with a private counterparty below that level? 29

TDF A relatively new tool Similar to IOER in many ways, but instead of overnight, deposit is for a specified term How does it work? Banks can choose to place a fixed amount of their reserves in the TDF Used to support a floor under rates--why lend below TDF rates?--and to drain reserves--transforms one type of liability (reserves) into another (TDF) Typically needs to be offered above IOER; otherwise, why lock up your money for longer than you need to? 30

How do the tools affect the frictions? The tools work by affecting the two frictions that determine interest rates: Balance sheet costs: Reducing balance sheet size raises market rates by decreasing the spread to IOER Competition: Scarcity makes banks compete for reserves Interbank market frictions: Increasing the interbank market activity raises interbank market rates Assessment: TDF: Creates reserve scarcity but does not affect balance sheet size RRPs: Create reserves scarcity and reduce balance sheet size (if RRP rate < IOER) 31

Do we have the tools to raise rates? Yes, TDF and RRPs will help raise rates Large reserves RRPs are more effective RRPs (but not TDF) reduce balance sheet size Creating scarcity through TDF only would require massive drain of reserves 32

How should we use our tools? Welfare in our economy can be measured as the (negative of the) sum of the three frictions TDF: Increases rates in part by increasing interbank market costs Lower welfare RRPs: Increase rates by lowering balance sheet costs and increasing interbank costs Trade-off: Effect on welfare can be ambiguous but likely positive when reserves are large ON RRPs also stabilize rate by absorbing liquidly shocks 33

Implications for long-term framework What is the optimal supply of reserves? Trade-off: Reserves alleviate interbank market cost but increase bank balance sheet costs Optimal amount is likely larger than pre-crisis supply Additional potential benefits: Improves timing of payments Increases public supply of safe assets 35

One option: A floor system Interest rate DW rate IOER Supply of reserves is still not linked to target rate 0 Supply reserve balance 36

Is it costly to have a large supply of reserves? Consider a three period example The central bank that holds 2 period assets and 1 period liabilities Assets bought at t pay R t Bank pays i t on reserves Upward sloping yield curve R t = i t + 1 Central Bank New Assets 100 Reserves 200 Old Assets 100 Central Bank Profits: π t = 100(R t + R t 1 2i t ) 37

Central bank profits in steady state Interest on reserves is Low i 0 = i 1 = 0.25% Central Bank New Assets 100 Reserves 200 Old Assets 100 Assets pay according to yield curve R 0 = R 1 = 1.25% CB profits are positive π 1 = 100 1.25% + 100 1.25% 2 0.25% 100 = 2 In steady state, profits increase with size of balance sheet 38

Remittances have increased since the crisis SOMA Net Income and Federal Reserve Remittances to Treasury Billions of U.S. Dollars 0 20 40 60 80 100 2002 2004 2006 2008 2010 2012 2014 SOMA Net Income Remittances 39

Interest rates increase could lead to losses Interest on reserves goes up i 2 = 4.25% New asset purchases pay higher rate R 2 = 5.25% Old assets pay period 1 rate R 1 = 1.25% Central Bank New Assets 100 Reserves 200 Old Assets 100 CB profits are now negative π 2 = 100 5.25% + 100 1.25% 200 4.25% = 2 40

Interest rate decrease would lead to high profits Interest on reserves goes back down i 2 = 0.25% New asset purchases pay higher rate R 2 = 1.25% Old assets pay period 1 rate R 1 = 5.25% Central Bank New Assets 100 Reserves 200 Old Assets 100 CB Profits are now negative π 2 = 100 5.25% + 100 1.25% 200 0.25% = 6 41

Conclusion Challenge of normalization in the US is to raise market rates with a large supply of reserves Our current set of tools is likely effective to raise market rates RRPs can reduce bank balance sheet costs TDF and RRPs can create scarcity if used in high enough quantity Our framework suggests benefits of (fairly) large reserves in the long run 42

Questions? 43