Center for Financial Innovation and Stability

Notes from the Vault
Larry D. Wall
December 2016

Prior to the 2007–09 financial crisis, the primary tool of central banks in developed countries for controlling aggregate nominal demand was short-term interest rates. As central banks changed these rates, market participants would revise their expectations for the future path of short-term rates, which would induce them to change rates all along the yield curve. These changes in the yield curve would then induce consumers and businesses to adjust their spending. However, as central banks reached what was called the "zero lower bound" on their policy rates, their ability to increase aggregate demand via reductions in the short-term rate waned. In response, these central banks adopted a variety of extraordinary measures to reduce interest rates and encourage lending as a way of boosting demand. These measures included large-scale asset purchases, programs to facilitate increased bank lending, and even dropping the rate they paid banks below zero (thereby proving that zero was not the effective lower bound on rates).

The Atlanta Fed recently hosted a workshop to discuss these measures, The Impact of Extraordinary Monetary Policy on the Financial Sector.¹ In this Notes from the Vault post, I review some of the contributions of papers looking at how extraordinary monetary policy affected bank lending and the markets for collateral and primary dealers in government securities. A companion macroblog post reviews the papers on how extraordinary policies have affected interest and foreign exchange rates.

Impact of liquidity provision on bank lending
An important objective of many central banks during and after the U.S. and European crises has been to encourage bank lending as a way of stimulating economic activity. In November 2008, the Federal Reserve announced a program to buy mortgage-backed securities (MBS) issued by the housing-related government-sponsored enterprises with the intent "to reduce the cost and increase the availability of credit for the purchase of houses." The Federal Reserve initiated further programs in November 2010 to buy Treasury securities and in September 2012 to purchase additional MBS and Treasuries. The Federal Reserve referred to these programs as large-scale asset purchases (LSAPs), but they are often referred to by the less accurate name of quantitative easing (or simply, QE).

The impact of the Federal Reserve's purchases on bank lending is analyzed in a paper by Indraneel Chakraborty from the University of Miami, Itay Goldstein from the University of Pennsylvania, and Andrew MacKinlay from Virginia Tech. They find that banks that were more active in the MBS market increased their activity relative to other banks during those periods when the Federal Reserve was purchasing MBS. However, they do not find comparably strong evidence that banks with large securities portfolios expanded their lending when the Federal Reserve was purchasing Treasury securities.

As the financial shocks coming from the United States abated, several countries in the Eurozone encountered sovereign debt problems that spilled over into the broader European financial system. In response to ongoing problems, the European Central Bank (ECB) announced a package of measures in December 2011 intended to "support bank lending and liquidity in the euro area money market." Two papers presented at the conference addressed the impact of two of these measures: a relaxation of the ratings requirements to allow lower-quality asset-backed securities (including MBS) to be used as collateral, and two long-term refinancing operations (LTROs) that would provide banks with 36-month loans. The banks were promised "full allotment" on the LTRO, which meant they could borrow as much as they wanted subject only to having sufficient acceptable collateral.

The impact of the December 2011 loosening of eligible collateral and a further loosening in June 2012 is the subject of a paper by Sjoerd Van Bekkum and Marc Gabarro from the Erasmus School of Economics and Rustom M. Irani from the University of Illinois at Urbana-Champaign. In particular, they examine the impact of the ECB accepting MBS with lower credit ratings on mortgage lending in the Netherlands. One advantage of using Dutch data is that the ECB's actions were in response to financial crisis in other parts of the Eurozone, thus the authors could examine the change in collateral policy without having to worry that the borrowers or banks were being adversely affected by the crisis. They find the banks that were active issuers of lower-quality MBS prior to the ECB's actions increased their production of lower-rated MBS after the policy change and that the loans backing these securities subsequently had worse repayment performance.

In contrast to the Netherlands, banks' willingness to lend had been adversely affected in Italy prior to the ECB's December 2011 announcement. The problem was particularly acute for Italian banks that had relied on wholesale funding. From June to December 2011, there was a drastic fall in the supply of wholesale funds to Italian banks, led by a decrease in supply from U.S. money market funds. Thus, Italy provided an opportunity to observe both the effect of a sudden drop in the availability of funding and a subsequent increase in the availability of stable (longer-term) funding through the ECB's LTRO.

In their paper, Luisa Carpinelli of the Bank of Italy and Matteo Crosignani from the Federal Reserve Board examine the impact of both the drop in wholesale funding and the subsequent provision of stable provision through the LTRO. They use cross-sectional differences in Italian banks' precrisis reliance on wholesale funding to help them identify the changes in lending arising from the drop and subsequent increase in the availability of funding. They find that the drop in liquidity resulted in a reduction in lending by Italian banks that were more exposed to the decrease in availability of wholesale funding. They further find that Italian banks took advantage of the opportunity to borrow from the LTRO, with uptake equal to approximately 9.9 percent of their total assets.

Somewhat surprisingly, however, Carpinelli and Crosignani find that individual banks' borrowing relative to their assets was independent of an individual bank's precrisis reliance on wholesale funding. The difference in banks was in the collateral they provided and what they did with the funds. Banks that were exposed to the drop in wholesale funding also appeared to have been short of eligible collateral. These banks took advantage of a special Italian government program that helped them convert loans into securities that were acceptable collateral for borrowing at the central bank. Further, those banks that were more exposed to the drop in wholesale funding responded by increasing their lending after the LTRO, whereas banks that had not suffered reduced access used their funds to invest in securities.

The Van Bekkum, Gabarro, and Irani paper and the Carpinelli and Crosignani paper highlight an important point about central bank provision of liquidity. A central bank can provide liquidity to financial institutions, but it has limited ability to control how the funds are subsequently used. As a result, such programs may have unintended consequences for the allocation of credit. A paper by Alexander Bleck of the University of British Columbia and Xuewen Liu of Hong Kong University of Science and Technology examines one way in which increased liquidity could have an unintended adverse impact on the allocation of credit.

Bleck and Liu develop a theoretical model in which firms' ability to borrow depends on having sufficient collateral value to back their loans. To examine the implications of this reliance on collateral, they split firms into two types: (a) a sector in which the firms' assets are specialized and lose value if they are taken from the firm and (b) a sector in which the firms' assets are not specialized so could be bid up by potential acquirers. They find that sufficiently large liquidity injections by the central bank can drive up the value of the assets used by firms that are not so specialized, thereby increasing the value of their collateral and, thus, their ability to borrow. Indeed, they find that in some circumstances, the increase in collateral value of the less specialized firms may attract such a large increase in bank lending to that sector that it crowds out lending to more specialized firms.

Impact of negative rates on financial intermediaries
Many central banks, including the Bank of Japan and the ECB, have supplemented their asset purchase programs with the payment of negative rates on bank deposits (instead of the central bank paying banks that deposit money with it, the banks must pay the central bank). Moreover, in some cases these negative rates have spilled over into the pricing of government and private debt. The intent of these central banks is to further support borrowing for consumption and investment, and also in some cases to limit exchange rate appreciation (especially for Denmark, Sweden, and Switzerland). However, these negative rates may adversely affect financial institutions whose business model was based on positive rates and are constrained in their ability to respond to the change.

In theory, banks would not be exposed to negative rates, as their earnings are derived from the spread between what they earn on their assets and what they pay on their liabilities. As banks tend to have shorter-term assets and liabilities that adjust to changes in market rates, banks should remain profitable as long as they reduce the rate they pay on liabilities in proportion to the reductions in what they earn on assets. However, banks have proven reluctant to reduce their deposit rates below zero.

A paper by Florian Heider and Glenn Schepens of the European Central Bank and Farzad Saidi of the Stockholm School of Economics analyzed the lending behavior of banks operating in areas with negative interest rates. They find that those banks that are more reliant on deposit funding respond to decreases in rates by increasing their risk, a response that is not observed when rates are positive. They further find a switch in bank borrowers, with high-deposit banks taking on riskier customers and safe customers switching to low-deposit banks. Consequently, they find that the increased lending to riskier customers relaxes financial constraints on these firms, leading to more investment.

Another industry potentially exposed to negative rates is the insurance industry, especially life insurers that promised their policyholders positive rates of return over the long run. As with banks, insurers could have offset this risk by holding sufficiently long-dated (higher-duration) assets to provide high returns to the policyholders. Whether they are adequately hedged is the topic of a paper by Elia Berdin of the International Center for Insurance Regulation and Cosimo Pancaro and Christoffer Kok of the European Central Bank. The authors analyze the life insurance industry in five large European countries operating under negative rates. They find that life insurers' assets did not have a sufficiently long duration given their liabilities and as a consequence, these insurers are exposed to losses from negative rates. This exposure is positively correlated with the weighted average interest rates promised policyholders and the mismatch of the asset and liability durations—as is the case with the German insurance industry. This risk could, however, be mitigated to the extent the insurance firms are diversified into non-life lines of insurance that are less exposed to interest rate risk (such as property and casualty insurance).

Impact of extraordinary policy through the markets for collateralized lending
Central bank large-scale asset purchases do more than supply banks with liquidity and take duration out of the market. They also take high-quality, liquid assets that can be held by anyone and replace them with reserves at the central bank that can only be held by banks. The reduction in high-quality, liquid assets is important because they have been used by nonfinancial corporations as a higher-yielding, possibly lower-risk substitute for bank deposits. These assets are also used by financial institutions to provide collateral for their borrowing of cash and securities, and also to provide margin for their exposure to over-the-counter derivatives. Thus, the positive benefits of these purchases may, to some degree, be offset by their adverse impact on the functioning of financial markets.

The paper by Manmohan Singh of the International Monetary Fund observes that the reduction in high-quality, liquid assets has been accompanied by changes in regulatory policy that have the effect of discouraging banks and money funds from participating in wholesale collateral markets—especially reusing the collateral obtained in such transactions. The result has been a dramatic reduction in the amount of pledged collateral received by major U.S. and European banks. Further, Singh observes that the Federal Reserve is inserting itself into what had been largely private markets through the Fed's reverse repurchase program. His concern is that the Fed's expanded role may end up "rusting" the plumbing that had helped financial assets reach their best owner in the precrisis financial system.

In their paper, Gary Gorton of Yale University and Ping He of Tsinghua University develop a theoretical model to analyze central bank monetary policy in an environment where the availability of collateral is a determinant of the productivity of the private sector. In their model, collateral consists of Treasury securities and a privately produced, liquid but risky asset (which they call mortgage-backed securities or MBS). Firms use Treasury securities as collateral for obtaining loans. They can also use MBS, but the MBS is subject to a haircut (lenders will supply less funds for a given dollar of MBS than they would for an equal valued amount of Treasury securities). If insufficient Treasury securities are available, an incentive exists to create privately issued liquid securities. However, as the amount of MBS increases, the risk of a financial crisis also rises. Thus, in Gorton and He's model, the central bank optimally trades off the benefits of purchasing Treasuries to increase the supply of money with the cost these purchases incur in the form of an increased risk of a financial crisis.

Impact of large-scale asset purchases on primary dealers
Most economic models of LSAPs simply assume the central bank buys a given quantity of bonds at the prevailing market rate and express little concern about how this is done. However, in practice U.S. Treasury securities come with a variety of maturities and interest rates. Thus, when the Federal Reserve undertakes LSAPs, the Fed must choose which securities it will buy and from which dealers. The Federal Reserve uses an auction to purchase its securities, so if two dealers offer the exact same security at different prices, the Fed will buy the security at the lower price. However, if more than one security is offered for sale, the Fed has the problem of how to select the most favorably priced securities.

Zhaogang Song from Johns Hopkins University and Haoxiang Zhu from the Massachusetts Institute of Technology analyze the Fed's purchases of Treasury bonds. They observe that the Fed has its own model for pricing securities, which allows it to buy those bonds that appear cheapest by the Fed's model. However, the Fed's model is likely to differ from that used by market participants. Thus, to the extent that dealers can predict how the Fed's pricing model will evaluate bids, they are in a position to offer securities that are more highly valued by the Fed than by the market. Song and Zhu test this hypothesis by comparing what the Fed paid with prices observed on these securities later the same day. They find that the Fed paid moderately more (0.7 to 2.7 cents per $100 par value) than the price later in the day. They observe that this might be fair compensation to the banks for the risks they take in participating in an auction. However, they find that the variation around this estimate is large. They also find that while some large dealers appear to have earned large profits from these transactions, other dealers suffered losses.

Conclusion
Central banks in the developed countries have taken a variety of extraordinary measures since the financial crisis to boost nominal demand in order to meet their inflation (and, in some cases, employment) targets. Moreover, extraordinary policy may be with us for many more years, considering that these central banks currently have little or no ability to respond to future adverse shocks using traditional tools. Given that central bank policy is necessarily conducted through the financial system, it is important monetary policy and prudential regulatory authorities understand the many ways these policies may affect the financial sector. A number of papers evaluating these impacts were discussed at a recent Atlanta Fed workshop. These papers shed important new light on a variety of ways in which extraordinary policy impacts financial markets and institutions.

Larry D. Wall is executive director of the Center for Financial Innovation and Stability at the Atlanta Fed. The view expressed here are the author’s and not necessarily those of the Federal Reserve Bank of Atlanta or the Federal Reserve System. If you wish to comment on this post, please email atl.nftv.mailbox@atl.frb.org.

_______________________________________

¹ The workshop was cosponsored by the Center for the Economic Analysis of Risk (CEAR) at Georgia State University and the Center for Financial Innovation and Stability (CenFIS) at the Atlanta Fed.