On The Transactions Costs Of Quantitative Easing

On The Transactions Costs Of Quantitative Easing

Francis Breedon
University of London, Queen Mary – School of Economics and Finance

Philip Turner
Bank for International Settlements (BIS) – Monetary and Economic Department

July 2016

Abstract:

Most quantitative easing programmes primarily involve central banks acquiring government liabilities in return for central bank reserves. In all cases this process is undertaken by purchasing these liabilities in the secondary market rather than directly from the government. Yet the only practical difference between secondary market purchases and bilateral central bank/Treasury operations is the transactions costs involved in market operations. This paper quantifies the significant cost of this round-trip transaction – government issuance of liabilities and then central bank purchase of those liabilities in the secondary market.

On The Transactions Costs Of Quantitative Easing – Introduction: Direct Finance Versus Monetary Financing

Although quantitative easing (QE) programmes vary in design, the four major ones – those of the Bank of England, the Bank of Japan, the European Central Bank and the Federal Reserve – have involved the secondary market purchase of government bonds by the central bank and the associated creation of bank reserves.

Such policies have had two main objectives. The first is to put downward pressure on longer-term interest rates by reducing the average maturity of government debt to sell on the market (ie not held within the central bank). In an open economy, lower domestic interest rates relative to those abroad tend to induce non-residents to sell local bonds (ie capital outflows). The currency depreciates. The second objective is to increase bank reserves and make the domestic banking system more liquid in the hope that bank lending conditions ease. Many studies have shown that quantitative easing has had a large measure of success in meeting these objectives.

The question we ask in this paper is one of implementation: why in all these cases has the government issued debt in the open market for its own central bank to then purchase in the secondary market? Would it not be simpler, and cheaper, for the government to issue the bonds directly to the central bank and avoid round-tripping via the markets?1 This question needs to be asked as central bank reinvestment of maturing government bonds is very large. And it would also be relevant during the exit phase.

Simple macroeconomic theory suggests that the two approaches are identical. The monetary and real effects of official balance sheet policy depend only on changes in the consolidated balance sheet of what used to be called the monetary authorities – that is the government and the central bank combined. Transactions involving only the government and the central bank have no direct effect on the private sector.

The simplest demonstration of this is to suppose the government issues bonds for its central bank to hold and puts the proceeds on deposit with the central bank. The central bank’s balance sheet rises (government bonds as assets and government deposits as a liabilities) but there has been no change in the consolidated balance sheet of the authorities (that is, central bank plus government). It is simply a matter of intra-public sector accounting with no direct economic effects – unless the behaviour of the government or of the central bank is altered (or perceived to be
altered) by this transaction.

It is this qualification – that the behaviour of the public sector entity is not affected by accounting conventions – which is key. Much of the controversy about balance sheet policies revolves around different views about how they could affect government fiscal decisions (eg central bank finance could mean bigger deficits) or central bank policy rate decisions (eg large short-term government debt might force the central bank to set its policy rate too low). In addition, there can be political constraints on governments and central banks that are not captured by the consolidated balance sheet of the authorities (Ueda (2003)).

How has recent quantitative easing changed the consolidated balance sheet of the monetary authorities? Stripped to its essentials, it has changed their liabilities – from long-term government bonds to very short-term bank reserves (i.e. banks’ deposits with the central bank). The effect on this consolidated position of any change in the central bank holdings of government bonds could have been replicated by a comparable change in government debt issuance. Consider the two key objectives mentioned above – lowering the long-term interest rate and expanding bank reserves.

(a) The long-term interest rate. Even assuming expected future short-term rates are given, central bank purchases of longer-dated government paper can drive down the term premium (ie that part of the long-term rate not explained by expected future short-term rates) through portfolio balance effects. quantitative easing has spurred a revival of James Tobin’s work on this mechanism (as well as the old preferred-habitat models, such as Modigliani and Sutch (1967)). Vayanos and Vila (2009) have developed a preferred habitat model to explain the term structure of interest rates. Gertler and Karadi (2013) have established that changes in the term premium play a significant role in US monetary policy transmission. Iwata et al (2016) develop a similar analysis for Japan.

There is hardly anything new in this. Buying long-term government debt was central to Keynes’s analysis in Treatise on Money of how central banks could combat slumps (Tily, (2010)). Worried that a central bank acting alone could run the risk of provoking excessive currency depreciation, he argued that the newly established BIS could encourage internationally coordinated central bank efforts to reduce long-term interest rates. Per Jacobsson, Economic Adviser at the BIS from 1931, strongly supported coordinated policies aimed at reducing long-term interest rates.

The Treasury could achieve exactly the same effects by issuing fewer long-term bonds and more short-term bonds. In the 1930s, HM Treasury was unconvinced by Keynes’s advice, and actually lengthened the maturity of gilts issued. By the mid-1930s, 86% of UK government bonds had a maturity in excess of 15 years. Susan Howson’s (1975) study of British monetary policy in the 1930s found that this limited the effectiveness of the cheap money policy instituted once Britain had left the gold standard: debt management policy ran counter to the monetary policy intent of low short-term rates.

US experience since the 1970s clearly demonstrates how issuing shorter-dated debt lowers long-term interest rates. A recent study over the period 1976 to 2006 (that is, before QE) found that lowering the average maturity of US Treasuries held outside the Federal Reserve – mainly in this period brought about by changes in US Treasury issuance because Federal Reserve purchases were small – reduced the yield of long-term US Treasuries by a significant amount (Chadha et al, 2013).

Since early 2009, the US Treasury’s policy of lengthening the average maturity of gross issuance has worked in the opposite direction as quantitative easing. Larry Summers argued recently that the Federal Reserve’s quantitative easing policies reduced dollar long-term rates by 1.37 percentage points while the increase in the average maturity of Treasury debt issuance added back 0.48 percentage points (Greenwood et al, 2014). Iwata and Fueda-Samikawa (2013) show how the rise in the average maturity of JGBs has at times run counter to the Bank of Japan’s quantitative easing policies.

(b) Expanding bank reserves. The government issuing bonds directly to the central bank with a given price – rather than having the central bank buy such bonds in the secondary market – would generate government deposits with the central bank equal in value to the price put on the bonds. It is only if the government draws down such deposits to finance its domestic spending that there would be monetary implications. As the government spends more, commercial bank deposits would rise leading to an expansion in bank reserves.

As already noted, the macroeconomic equivalence between bilateral central bank/Treasury operations and round-tripping through the markets depends on the assumption that the behaviour of the government or the central bank is not altered by the alternative financing arrangements. This assumption can of course be challenged. Fiscal discipline could be eroded if the Treasury were able to force the central bank to directly finance government spending at interest rates dictated by the Treasury (“direct finance”). Monetary policy independence would be undermined.

Nonetheless, simply banning direct purchases of government debt by the central bank would not, in itself, prevent such an outcome. If the Treasury could force the central bank to buy government debt in the secondary market, it would be relatively easy to find willing buyers for that debt in the primary market who could then quickly sell to the central bank. Indeed, other financing arrangements could equally undermine fiscal discipline – such as forcing commercial banks to finance the government (financial repression). In all cases, it is the ability of the Treasury to influence the behaviour of potential holders of government debt that is the key – not the mechanics of how that debt is sold.

The monetary financing of government deficits can take place without any direct finance. Indeed, the standard measure of monetary financing does not depend on the intentions of either government or the central bank. Rather it is based on the short-term liabilities on the consolidated balance sheet of the central bank and the government: currency, other central bank liabilities and short-term government debt. The short-hand view was that Treasury issuance of short-term debt was monetary finance, and therefore subject to limits.

For this reason, the Deutsche Bundesbank had for many years a veto on short-term debt issuance by the German government. Many central banks paid close attention to measures of monetary financing. Hoogduin et al (2010), for instance, describe the case of the Netherlands: “In the 1980s, government debt finance was an explicit part of the monetary analysis of De Nederlandsche Bank … when the government financed part of its debt in the money market, it was considered monetary financing, which would increase the amount of liquidity in the economy.” Treasury bills are close to money because their holders can cash close to the nominal value of the bill – which cannot be counted upon in the case of bonds. Implicit understandings that only the central bank should issue debt at the short-end of the market while government issued longer-dated paper took root in many countries, often to avoid having two official issuers “competing” in the same maturity tenor.

The definition of monetary financing in Table 1 follows Tobin (1963). Tobin argued that the monetary impact of government debt depends on the composition of the public’s holdings of government debt (that is, net of central bank holdings). He showed that, in 1969, it was the increasing reliance by the Treasury on short-term debt (not increased Federal Reserve obligations) that had made the economy more liquid.

Monetary financing was accordingly defined as currency, Federal Reserve liabilities plus US Treasuries with a maturity of less than one year. Non-monetary financing is simply longer-dated US Treasuries held outside the central bank – this can be altered either by the central bank buying bonds in the open market or by the Treasury issuing more short-term bills and fewer longer-term bonds. Likewise, according to this view, monetary financing can take the form of Treasury issuance of short-term bills.

Table 1 shows that such issuance in the United States was large in FY 2008 and FY 2009. Under the Supplementary Financing Program (SFP), the US Treasury did increase its issuance of Treasury bills – “separate from its current borrowing program” as its September 2008 press release put it – and put the proceeds on deposit with the Federal Reserve (which purchased mortgage-backed securities).

Although these terms are sometimes confused, monetary financing is not the same as direct financing by the central bank of the government on terms decided by the government. The central bank which purchases government bonds in open markets may protect itself from such pressure – but it is still engaged monetary financing. Central banks have historically been active in bond markets in implementing monetary policy. As Truman (2005) put it well before recent quantitative easing policies, “the proposition that a central bank should limit its purchases of long-dated government obligations in the open market because not to do so would impair its balance sheet and de facto independence [is incorrect]… as long as the central bank purchases long-dated government obligations in the open market and has no obligation to roll them over, the central bank should have no legislated or self-imposed limit on the amount of such obligations it may purchase.” Equally a government engaged in significant and regular market issuance in primary markets – so that market-determined new issue prices exist – can place new issues with the central bank at prices determined by the market. It can entirely avoid the traps that Truman warned about without the roundabout strategy of selling bonds in open markets that it knows the central bank – for its own monetary policy purposes – plans to purchase. This roundabout strategy is costly.

Of course, in the case of EU countries direct financing is legally problematic given the Lisbon Treaty on the Functioning of the European Union. Article 123 of the Treaty states:

“Overdraft facilities or any other type of credit facility with the European Central Bank or with the central banks of the Member States (hereinafter referred to as ‘national central banks’) in favour of Union institutions, bodies, offices or agencies, central governments, regional, local or other public authorities, other bodies governed by public law, or public undertakings of Member States shall be prohibited, as shall the purchase directly from them by the European Central Bank or national central banks of debt instruments.”

Some have argued this Treaty does not preclude direct central bank/Treasury operations, Bossone (2013), for example, suggests that such financing could be undertaken under the Emergency Liquidity Assistance (ELA) scheme. Such legal arguments are however beyond the scope of this paper which has a more limited objective – to examine whether the round-trip approach to quantitative easing involves significant transactions costs. If such costs are not incurred, then a further analysis of why the round-trip approach is undertaken would have little policy relevance. If, on the other hand, significant transactions costs are incurred (which is what we find), then the benefits of the round-trip approach that offset this cost need to be identified.

In this paper we take the UK quantitative easing programme as an example and estimate the transactions costs involved in that programme. We analyse the movements in bond yields around each type of operation (debt sales and debt purchases) to measure short-term underpricing of sales and overpricing of purchases – commonly termed the ‘concession’ – over the period when quantitative easing was in operation. In approach it is most similar to Lou et al (2013) who study US Treasury auctions, in a similar way and find similar results in terms of both the size and length of the concession window. On the QE purchase side, it is also similar to D’Amico and King (2013), though they use a somewhat different approach and a shorter window. In some senses, the approach taken here is also comparable with that adopted in a number of QE studies (see, for example, Krisnamurthy and Vissing-Jorgenson (2011) and Joyce et al (2010)) to measure the long-run and announcement effects of the policy. But the focus of this paper is very much on the short-run liquidity impact of individual operations (what D’Amico and King (2013) call ‘flow effects’) as a way of gauging the difference between the price government bonds could be sold at and the price they could be purchased at in large scale official operations over the QE period.