The demand for safe liquid assets and the implications of issuing a Central Bank Digital Currency for bank funding instruments

Müller, Carola Research Report The demand for safe liquid assets and the implications of issuing a Central Bank Digital Currency for bank funding instruments Staff Memo, No. 8/2020 Provided in Cooperation with: Norges Bank, Oslo Suggested Citation: Müller, Carola (2020) : The demand for safe liquid assets and the implications of issuing a Central Bank Digital Currency for bank funding instruments, Staff Memo, No. 8/2020, ISBN 978-82-8379-171-6, Norges Bank, Oslo, https://hdl.handle.net/11250/2690368 This Version is available at: https://hdl.handle.net/10419/246151 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/4.0/deed.no STAFF MEMO The demand for safe liquid assets and the implications of issuing a Central Bank Digital Currency for bank funding instruments NO. 8 | 2020 CAROLA MÜLLER 2 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS The papers in the Staff Memo series present reports and documentation written by staff members and other authors affiliated with Norges Bank, the central bank of Norway. The views and conclusions expressed in these papers do not necessarily represent those of Norges Bank. © 2020 Norges Bank This paper may be quoted or referenced provided the author and Norges Bank are acknowledged as the source. ISSN 1504-2596 (online) ISBN 978-82-8379-171-6 (online) 3 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS The demand for safe liquid assets and the implications of issuing a Central Bank Digital Currency for bank funding instruments Carola Müller1 The pros and cons of issuing a Central Bank Digital Currency (CBDC) is currently debated by Norges Bank and other central banks. A CBDC would grant access to the central bank balance sheet to a broader set of economic agents, including for example companies or individuals. How such access would be designed or administered are topics of the discussion. One dimension to consider is the potential impact on the stability of the financial system through the effect of CBDC issuance on bank funding markets. This article provides an overview on research related to the demand for financial institutions’ funding instruments and derives implications for the issuance of a CBDC. A key message is that CBDC has the potential to substantially crowd out bank funding instruments due to its superior safety features. Central bank digital currencies, safe assets, bank funding markets, financial stability, money demand. 1. Introduction The issuance of Central Bank Digital Currency (CBDC) can conceptually be perceived as the offer of a new asset to the general public. As such, it can potentially change the asset holdings of households, Non-Financial Companies (NFCs), and financial companies. Depending on the degree of substitutability with other assets such as cash or bank deposits, it can have additional repercussions on the financial sector or the central bank that act as counterparties to these holdings. Since there are no real world examples of CBDC at hand, it is hard to assess how the introduction of CBDC would affect the demand for other risk-free or relatively safe assets, such as bank deposits and other assets that are important funding sources for Norwegian financial institutions. Inferences have to be drawn from comparisons of other liquid assets with low credit risk. Comparable examples of these safe liquid assets are government bonds, secured or unsecured bank deposits, and money1 This staff memo should not be reported as representing the views of Norges Bank. The views expressed are those of the author and do not necessarily reflect those of Norges Bank. I thank Ragna Alstadheim, Magdalena Riiser, Knut Sandal, Haakon Solheim and Ylva Søvik for comments. 4 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS market funds. The following should give a comprehensive overview of existing studies from which such inferences can be made. This review aims to extract information on the demand for safe and liquid assets and its determinants, which can be used to build reasonable scenarios for the introduction of a CBDC. In what follows, the first section introduces a schematic of safe liquid assets that classifies financial instruments according to their degree of liquidity risk and credit risk, and describes recent insights into the demand for safe assets. Section 3 examines the demand for wholesale funding instruments of financial institutions. Section 4 reviews literature on the demand for money and deposits from a macroeconomic, household-oriented, and bank-oriented perspective. The section ends by showing findings on deposit demand in turbulent times. Section 5 briefly reviews the substitutability between wholesale and retail funding sources for financial institutions. Section 6 concludes. 2. Classification of safe liquid assets A Central Bank Digital Currency (CBDC) is a new financial asset that with a few experimental exceptions (see for example the case of Ecuador or Uruguay) has not been issued. The exact features of this potential new asset are therefore not yet defined. As a new form of money, however, two features are crucial: CBDC should neither have liquidity risk nor credit risk. As such, it competes for demand with other safe and liquid assets. Golec and Perotti (2017) provide an overview of how assets with almost no liquidity risk and almost no credit risk can be classified. Figure 1 below provides a further simplification of their framework. This review follows the classification. CBDC can be regarded as a very safe and very liquid asset, similar to cash in this setup. If central banks were to issue a new asset in form of a CBDC, the supply of other safe assets and therewith the funding possibilities of governments and financial institutions would be affected. The following gives an overview about what we know about the elasticities of substitution between these instruments. The supply of safe liquid assets mainly comes from three sources: central banks (cash), governments (treasuries and government-backed securities), and financial institutions (deposits, wholesale funding instruments). 2 For governments and financial institutions the main motivation for issuing these assets is to fund their activities. The demand for safe liquid assets comes from households, financial institutions, non2 In principle, highly rated corporate debt from non-financial companies could also fall within the category given a sufficiently deep secondary market. The following analysis, however, as well as monetary aggregates (M1-M3) do not consider corporate debt. 5 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS financial corporations, governmental bodies, and central banks. They can have diverse interests in holding safe liquid assets, which will be discussed where it becomes relevant (store of value, medium of exchange, unit of account). The low credit risk of the securities discussed here not only relies on the safety of the cash flows on which these claims are written but also on the quasi non-existent counterparty credit risk of the issuers. In normal times, issuers are generally believed not to default or at least that default is highly unlikely. In crises, however, the values of underlying claims may deteriorate and issuers, especially financial institutions and occasionally governments, become more likely to fail as recent experience in the Global Financial Crisis or Sovereign Debt Crisis in Europe has shown. As soon as any of these securities are not considered safe anymore, we expect much more dramatic portfolio adjustments than under normal circumstances. Private safe assets are thus susceptible to runs. The following analysis therefore differentiates between elasticities of demand for safe liquid assets in normal times, and elasticities in crisis situations or run behaviour. 2.1. Demand for safe assets A Gorton et al. (2012) document that the share of safe to risky assets has been surprisingly constant (at around 32%) in the US for at least half a century. Gourinchas and Jeanne (2012) attest a similar relationship for other developed countries (Japan, United Kingdom, Germany, France). Furthermore, Gorton et al. (2012) show that while the share of safe assets was constant, its composition was not. They find a strong negative relation between the share of financial sector’s debt and the share of government debt. Within financial sector’s debt the share of deposits has declined since 1980 relative to the share of wholesale funding instruments. The authors suggest to view safe assets as a production factor for total assets, although little is known about this function. There seems to be a stable volume of safe assets relative to risky assets, implying that whenever one issuer offers less, another steps in. The supply of treasuries is mainly driven by budgetary needs (Greenwood et al., 2014; Gorton, 2017). Hence, it is the financial sector that adapts its supply of safe assets whenever needed. Moreover, the financial sector uses marketable safe assets rather than deposits to adapt. In the run-up to the Global Financial Crisis, a fitting substitution pattern was observed between treasury debt supply and shadow banking debt (Sunderam, 2015; Krishnamurthy and Vissing-Jørgensen, 2015). 6 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS Figure 1 Classification of safe and liquid assets based on Golec and Perotti (2017) The money-likeliness of safe assets offers an explanation. With respect to the functions of money, safe assets are not only held for investment purposes but to make transactions (medium of exchange) or as a store of value until transactions are made. This implies the existence of nonpecuniary returns, a convenience yield. How money-like an asset is depends on how liquid it is and whether it is accepted as a means of payment. As Figure 1 points out, deposits are most money-like or the closest to cash because they are often accepted as means of payment. Treasury securities and other liquid wholesale funding instruments of financial institutions are considered near-money since they can easily be converted into cash or deposits. The demand for safe assets has different motives. The demand for money comes from different agents than the demand for near-money. Money (i.e. cash and deposits as in the monetary aggregate M1, and M2) is primarily held as an asset by households and non-financial corporations supposedly with the purpose of making real transactions. Near-money (i.e. treasuries and private securities) is primarily held as an asset by financial investors, supposedly with the purpose of making financial transactions, and to a lesser extent by households and firms, at least in Europe. Implicitly based on such a distinction, Gorton (2017) draws the conclusion that the “central bank needs to satisfy two clienteles (retail and wholesale) with demands for different kinds of money, each with different implications: one related to inflation (cash) and the other to financial stability (Treasuries)”. 7 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS Several conclusions can be drawn for the introduction of a central bank digital currency. First, the aggregate demand for safe assets is rather constant. A newly issued safe asset will therefore most likely crowd-out other safe assets, given a certain acceptance. Second, demand for safe assets produced by financial institutions is different for retail and wholesale funding instruments reflecting the differing needs of the respective counterparties. Hence, which kind of safe asset will be most affected by the introduction of a CBDC is based on the features of the CBDC and whether it will be a better substitute for retail or wholesale funding. The following analysis therefore distinguishes these two cases. 3. Demand for wholesale funding instruments Financial institutions are able to issue safe debt without relying on a government guarantee. They use collateral, securitization, or short maturities to make the cash flows secure. Beside commercial banks, shadow banks issue safe debt. Examples of shadow banks are securitization vehicles, money market funds, markets for repurchase agreements (repos), and mortgage companies. Commercial banks use several wholesale funding instruments. They issue secured and unsecured debt with long-term maturities and they use liquid short-term securities such as brokered deposits. Furthermore, they often own and control shadow banks thereby exposing them to developments in the markets for shadow bank debt. Gorton et al. (2012) show that the volume of government debt and privately-produced safe assets in the US are strongly negatively correlated. An explanation is that investors view them as substitutes. Following this substitution hypothesis, Krishnamurthy and VissingJørgensen (2015), Greenwood, Hanson, and Stein (2015), Sunderam (2015), Carlson et al. (2016), and Carlon and Wheelock (2018) show how private safe assets are issued when the convenience yield is high indicating that government-safe assets are in short supply. Table 1 gives an overview of estimates of the elasticity of substitution between government-backed safe debt instruments and privately produced safe debt that appear in these studies. They implicitly assume that the demand for wholesale funding instruments corresponds to a residual demand for safe assets consisting of the part of demand not met by the supply of government securities. 8 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS Table 1 Estimates of the crowding-out effect of government safe debt on private safe debt. 3 Paper Data Elasticity of substitution1 Krishnamurthy and Vissing-Jørgensen, 2015 1875-2014, q/m/w -51 to -57 cents (net short-term debt issued by the financial sector per GDP) Greenwood, Hanson, and Stein, 2015 1952-2009, q/m/w - 14 to -24 cents (financial commercial paper) Sunderam, 2015 2001-2007, weekly -3.5% (asset-backed commercial paper) due to 1% increase in T-bills/GDP Carlson et al., 2016 1975-2007, monthly -30 cents (commercial papers) / -47 cents (deposits) / -13 cents (MMFs) Carlson and Wheelock, 2018 1965-1979, quarterly -10 cents (Eurodollars issued) / -9 cents (negotiable CDs) 1 Numbers show the USD change in private safe asset after 1 USD (or 1%) increase in T-bills to GDP. All of the studies look at US treasury supply relative to distinct bank wholesale funding instruments or shadow bank debt securities. Krishnamurthy and Vissing-Jørgensen (2015) group several instruments that were of varying importance over the almost 150 year period they study into short-term debt issued by the financial sector and show that a one dollar increase in treasury debt decreases private-safe debt by up to 57 cents (deflated with nominal GDP). Carlson et al. (2016) study substitution between treasuries and several wholesale funding instruments and find a reduction of private-safe debt between 13 and 47 cents. Overall, all studies confirm the negative relationship but each study uses different time windows and different private-safe assets to evaluate substitution effects. Concentrating on the most comparable numbers, the elasticities of substitution resulting from a one dollar 3 The comparability of the estimates presented in a highly condensed format in this table and the other tables of the paper is subject to limitations set by each cited studies’ specific setting. These limitations include among other factors differences in the sample, time period, control variables, estimation method, and regression specification. Estimates should therefore be read with the necessary caution as rough hints at a quantification of the respective elasticities. 15 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS Table 2 Deposit demand cross-price elasticity and switching cost estimates. 1 Numbers show a % change in market share after 1% increase of competitor’s deposit rate. 2 Numbers show the percentage of costs relative to an average account balance. Depending on the design and way CBDC would be marketed, CBDC could potentially resemble a current account at a commercial bank. Banks would then face an outflow of deposits if households substitute bank deposits for CBDC. The extent of such substitution would depend, among other factors, on the perceived servicesand price-differential between a central bank and commercial bank account. By viewing the central bank as a new competitor in the deposit market, inferences can be drawn from structural models of bank competition in consumer markets. These studies are able to estimate own-price elasticities, crossprice elasticities, and switching costs. Paper Data Cross-Price Elasticity1 Switching Cost Estimate2 Shy, 2002 Finland, 1997 0% small bank 20% large bank Ho, 2015 China, 1994-2001 0.8% Egarius and Weill, 2016 FR, DE, IT 2006-2012 3.9% 5% Stenbacka and Takalo, 2019 Finland, 2017 2% small bank 15% large bank Adams et al., 2007 U.S., 1990-2001 - 0.16 rural - 0.0079 urban Ho and Ishii, 2011 U.S., 1994,2000, -0.048 average -0.089 close bank -0.033 distant bank Molnar et al., 2013 Italy, 2003-2007 - 0.01 16 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS While the own-price elasticities of retail deposit demand can be informative about banks’ potential to attract deposits despite competition of CBDC, cross-price elasticities and switching cost estimates give an impression about the likely outflows. Table 2 gives an overview of estimations of own-price elasticities. The numbers illustrate how a bank’s market share would increase if the bank offers a 1 pp. increase in its deposit rate to customers and range from 0.9 to 21 pp. Table 3 lists estimates of switching costs for retail deposit accounts as well as crossprice elasticity estimates that show how a bank’s market share would decrease if a competitor offers a 1 pp. deposit rate increase. Given the existence of switching costs in some markets, cross-price elasticity estimates are rather low, ranging from 0.79 to 16 bps indicating that deposits might indeed be rather sticky. Amel and Hannan (1999) find that the short-run interest rate elasticities of deposit demand are very small and mid-run (up to 2 years) still low enough to indicate a market power of banks’ against competing offers. A randomized control trial, where a new deposit product at randomized rates was offered to Philippine households, analysed by Karlan and Zinman (2018), confirms rather inelastic interest rate sensitivity of retail deposit demand. However, Brunetti et al. (2016) use a biannual household survey in Italy from 2006 to 2012 according to which on average a quarter of households changes banks between two survey rounds. Further, they estimate that a household is on average 3.5 pp. less likely to switch to another bank if the household makes use of one additional service of the bank, such as payments. Overall, offered interest as well as services are important factors of retail demand, which can determine the potential crowding-out effect of CBDC. Non-pecuniary considerations seem to play an important role in consumers’ choice of the supplier of financial services. This suggests a sluggish adoption of a CBDC that is perceived akin to a deposit account. Switching costs might protect banks from outflows but can be eradicated in case CBDC accounts are mandatory or automatically accessible. Another factor could be the perceived safety of deposit holding which will be discussed in the following section. 4.4. Deposit demand in uncertain times All features of deposit demand described above apply only under one condition, i.e. as long as deposits are perceived as safe. Depositors depend on the promise that they can withdraw their money at any point in time in order to satisfy idiosyncratic liquidity needs. They can only do so, if the bank does not default and has enough liquid assets. Fears about banks’ liquidity or solvency can lead to a bank run situation in which every depositor wants to liquidate. In order to avoid bank runs, most countries have a deposit insurance scheme in place that guarantees repayment in 17 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS case of bank failure up until a certain amount. Insured deposits are therefore much less information-sensitive than uninsured deposits. Several studies find evidence that deposit withdrawals react to banks’ performance. Gorton (1988) studies seven historical banking panics in the US. He shows that withdrawals are triggered by the arrival of negative information about aggregate risk of a recession arguing that in lack of bank-specific information depositors could only extrapolate the overall economic situation to judge the soundness of their investment. Relying on more recent data of US commercial banks from 1994 to 2013, Chen et al. (2019) show that deposit flows are more sensitive to bank-specific information if the information is of better quality. This highlights the role of depositors’ information set for their withdrawal behaviour, which consists of hard information as well as beliefs. Saunders and Wilson (1996) catch up on the difference made in theoretical bank run models between informed and contagion (or panic) runs. They identify a period of informed runs during the US Great Depression where failing banks experienced deposit outflows (and surviving banks deposit inflows) and a period of contagion runs where banks experienced outflows independent of their individual health. In the aggregate, informed runs lead to a reallocation whereas contagion runs lead to flight into currency and a contraction of available funds. Deposit insurance should reduce the link between bank solvency and withdrawal behaviour by making insured depositors less sensitive to negative news. By looking at the Savings & Loans crisis in the U.S. in the 1980ies, Goldberg and Hudgins (1996) point out that uninsured deposits are indeed more information-sensitive as they retract from banks prior to failure. Egan et al. (2017) confirm that uninsured deposits in particular reacted to financial distress of banks during the recent financial crisis. Further, Martinez Peria and Schmukler (2001) document for banking crisis in Argentina, Chile, and Mexico that struggling banks saw withdrawals of insured and uninsured deposits alike and offered higher interest rates on their deposits. However, deposit insurance might not have been credible in these samples as the effectiveness of a deposit insurance depends ultimately on the solvency of the government. Hence, deposit insurance might lose its vigour if a bank run is accompanied with a sovereign debt crisis. For example, Artavanis et al. (2019) estimate that the probability for early withdrawal quadrupled as policy uncertainty led to a sharp increase of Greek sovereign bond CDS prices. Finally, deposit insurance might attract funds in crisis times. Uninsured deposits might only be turned into insured deposits by slicing them up and distributing them at several banks. By studying deposit flows at one failing US bank in the recent crisis, Martin et al. (2018) document that this 18 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS bank is largely able to offset the loss of uninsured deposits through gains in insured deposits as it approaches failure. Iyer et al. (2017) affirm that such reshuffling took place when Denmark reduced the threshold amount of insured deposits. They further point out that especially large and systemically important banks profit from a reshuffling of uninsured into insured deposits as they could attract relatively more flows than their smaller competitors. This might imply that depositors value implicit bailout guarantees, here for too-big-too-fail banks, as well. A Norwegian study by Lie (2011) of deposit movements during the financial crisis found no dramatic movements in deposits, and that depositors mainly reshuffled uninsured deposits to other banks to keep their total deposits insured. Further, Gatev and Strahan (2006) show that banks gained deposit inflows in insured accounts as stress emerged in alternative liquid asset markets, such as the commercial papers market. Besides government-sponsored deposit insurance, customer relationships can mitigate withdrawal pressures as well. Brown et al. (2020) made a survey among Swiss households which revealed that an exclusive bank-client relationship significantly reduced the probability of withdrawal at two large Swiss banks that experienced substantial outflows during the financial crisis. A similar pattern is found by Iyer and Puri (2012) in an example of a run on a small Indian community bank. Overall, depositors show similar preference as investors in wholesale instruments for government-safe assets over private-safe assets which gains particular strength in crises. Deposit insurance is credible in most developed economies and helps mitigating outflows by incentivising a reallocation of deposits within the banking system. Demand for CBDC as a safe haven, however, could potentially be very high if uninsured deposits are being withdrawn and can be transferred to CBDC accounts at great convenience. An important feature of CBDC is therefore if and by how much it extends the insured or government-safe amount of deposit holdings. In this respect, CBDC design interacts with the framework of the deposit insurance system. 5. Substitution between bank wholesale funding and deposit funding Based on the notion that CBDC could potentially crowd-out either banks’ wholesale funding opportunities or deposit funding or both, an important question is how banks are able adapt to a change in their funding structure. 19 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS The flip side of the stability that retail funding offers is an inflexibility to adapt to sudden funding needs. Because deposit supply is highly priceinelastic, especially in the short-run, banks often reach out to the wholesale funding markets when they wish to expand their balance sheets. Although government-backed supply of safe assets might crowdout bank wholesale funding opportunities as described above, Carlson and Wheelock (2018) point out that bank balance sheets stay constant. They study a time period from the 1960ies to 1970ies in the US when banks started to use certain forms of wholesale funding: Eurodollar deposits and negotiable certificates of deposits (CDs) indicating that banks substituted deposits for wholesale funding. Using a more recent time period from 2006 to 2009, Cornett et al. (2011) confirm a substitution between wholesale and deposit funding. Banks with more core deposits, especially small banks, rely less on cash holdings and have to increase liquid assets less in times of crisis. Choi and Choi (2020) study how banks adjust their funding structure to exogenous shocks. They show that changes in monetary policy rates cause banks to shift between wholesale and retail funding. A tighter monetary policy induces deposit outflows and banks try to compensate by issuing marketable debt. Xiao (2020) supplements this finding by showing that depositor demographics determine the severity of the deposit outflow, where older clients are generally less likely to withdraw their funds, and that bank size as a proxy for funding frictions determines the ability of banks to replace deposit with wholesale funds. These findings indicate that banks are able to substitute funding sources. Due to the higher flexibility financial institutions often use wholesale funding markets to adjust to sudden fluctuations. Nevertheless, literature on funding constraints mostly document a negative credit supply effect indicating that banks shrink their balance sheets in response to funding shocks. If CBDC would have a strong crowding-out effect, the financial sector’s ability to raise funds in wholesale markets will be crucial. 6. Conclusions This review gives an overview of several strands of literature that jointly can give an impression of how the introduction of a Central Bank Digital Currency might impact demand for retail and wholesale funding instruments of financial institutions. It illustrates that CBDC has the potential to crowd-out funding sources of the private financial sector. Thus the introduction of CBDC can have repercussions on financial stability through its effect on funding markets. However, the exact characteristics of CBDC are yet to be defined and part of this policy debate is how and to what extend CBDC is likely to differ from other safe 20 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS liquid assets. Any inferences drawn therefore highly depend on the design of CBDC and the economy it is designed for. An important limitation of this study is that its findings concentrate on the demand for U.S. safe assets since most analyses from which inferences are drawn rely on U.S. data. Several factors should therefore be kept in mind when applying the conclusions that are presented here. For example, demand for safe assets in U.S. dollars might be particularly strong due to the important role of USD in international financial markets. In order to assess how the introduction of a CBDC might impact the banking sector of a particular country, the funding structure of domestic banks as well as the asset portfolio structure of banks’ creditors must be considered. Furthermore, this study did not comment on particularities of demand for safe liquid assets from non-financial corporations nor the footprint of recent trends in digital banking that might affect deposit demand of certain age or income cohorts of retail investors. Taking these caveats into consideration, some tentative assessments can be made. The current economic environment has a high demand for safe liquid assets as part of the growing amount of total assets. While the share of safe to total assets appears rather stable, its composition is not. A CBDC would have to find its place in this composition as agents substitute other safe assets for CBDC. The portion of safe assets issued by financial firms has increased since the 1980s reflecting a growing demand for money-like assets from institutional investors in quest of a safe store of value. Institutional investors are counterparties to most of banks’ wholesale funding instruments. Retail funding instruments are predominantly held by households and non-financial corporations in need of the payment services that these products provide. These groups of investors have different motives for holding safe liquid assets and arguably behave differently. Therefore, we would expect demand for a CBDC to show different sensitivities when it is designed to appeal to retail investors/customers than when it is used by institutional investors. Conceptually, we can distinguish three factors that determine demand for CBDC: the interest rate relative to alternative assets, the liquidity services, and the safety premium.  Money demand depends negatively on the yield spread between money and alternative investment opportunities. Similarly demand for privately issued near-money assets decreases as yields on 21 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS alternative assets rise. Wholesale investors display a high demand for safe assets. An important factor could be how and if international and institutional investors can make use of CBDC liquidity services. An easily accessible CBDC is likely to attract their attention. Nevertheless, these investors are interest rate sensitive given their motive to find safe store of value and the abundance of alternative investment opportunities. Retail investors on the other hand display low interest rate sensitivity, especially in the short run.  Retail investors value non-pecuniary benefits and liquidity services that tie them to their supplier of safe assets. This could indicate low demand from retail investors due to a sluggish adoption of CBDC when they face relevant switching costs relative to the average account balance. However, once they have implemented CBDC as part of their asset holdings, demand might show stable patterns.  Since CBDC is issued by a central bank, it is likely to have a higher safety premium than financial institutions’ debt. CBDC is able to offer a safe store of value even in times of uncertainty or crisis. For this reason, wholesale as well as retail investors have a strong preference for government-issued safe assets over privately issued safe assets. Institutional investors show excess demand for government-back safe assets, like treasuries. Furthermore, where deposit insurance is in place, bank debtors flock into insured deposits at signs of turbulences. This indicates a strong potential for CBDC to crowd-out financial sector debt instruments. These reallocation incentives will intensify in times of market stress or uncertainty. Hence, whatever demand CBDC can attract in normal times, it may increase immensely if a crisis is lingering. Overall, this review illustrates there could be a high elasticity of substitution between CBDC and other safe liquid assets that are used by financial institutions as sources for funding their activities. Assuming CBDC were generally accessible and offered similar liquidity services as do financial sector debt instruments, the strong preference for government-backed safe assets indicates a high potential for crowdingout. CBDC can be designed in ways that mitigate a crowding out effect by limiting the access to CBDC, reducing the services connected to CBDC use, or requiring safe collateral for the use of CBDC. 22 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS References Safe Assets Barro, Robert J. and Anthony M. Santomero (1972): Household Money Holdings and The Demand Deposit Rate. Journal of Money, Credit, and Banking, 4(2), pp. 379— 413. Carlson, Mark, Burcu Duygan-Bump, Fabio Natalucci, Bill Nelson, Marcelo Ochoa, Jeremy Stein, and Skander Van den Heuvel (2016): The Demand for Short-Term, Safe Assets and Financial Stability: Some Evidence and Implications for Central Bank Policies. International Journal of Central Banking, 12(4), pp. 307—333. Carlson, Mark, and David C. Wheelock (2018): Near-money premiums, monetary policy, and the integration of money markets: Lessons from deregulation. Journal of Financial Intermediation, 33, pp. 16—32. Golec, Pascal and Enrico Perotti (2017): Safe assets: A review. ECB Working Paper 2035. Gorton, Gary (2017): The History and Economics of Safe Assets. Annual Review of Economics, 9, pp. 547—586. Gorton, Gary, Stefan Lewellen, and Andrew Metrick (2012): The Safe-Asset Share. AER Papers & Proceedings, 102(3), pp. 101—106. Gorton, Gary and Andrew Metrick (2009): Haircuts. NBER working paper No. 15273. Gorton, Gary and Andrew Metrick (2012): Securitized banking and the repo run. Journal of Financial Economics, 104(3), pp. 425—451. Gourinchas, Pierre-Olivier and Olivier Jeanne (2012): Global safe assets. BIS Working Papers No 399. Greenwood, Robin, Samuel G. Hanson, and Jeremy C. Stein (2015): A comparativeadvantage approach to government debt maturity. Journal of Finance, 70(4), 1683—1722. Kacperczyk, Marcin T., Christophe Perignon, and Guillaume Vuillemey (2017): The private production of safe assets. CEPR Discussion Paper No. DP12395. Krishnamurthy, Arvind and Annette Vissing-Jørgensen (2012): The Aggregate Demand for Treasury Debt. Journal of Political Economy, 120, pp. 233—267. Krishnamurthy, Arvind and Annette Vissing-Jørgensen (2015): The impact of Treasury supply on financial sector lending and stability. Journal of Financial Economics, 118 (3), pp. 571—600. Sunderam, Adi (2015): Money creation and the Shadow Banking System. The Review of Financial Studies, 28(4), pp. 939—977. 23 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS Money demand Aastveit, Knut Are (2005): The demand for cash in Norway. Dissertation, University of Oslo. Bårdsen, Gunnar (1992): Dynamic modeling of the demand for narrow money in Norway. Journal of Policy Modeling, 14(3), pp. 363—393. Bårdsen, Gunnar, Jansen, E. S., Eitrheim, Øyvind, and Nymoen, R. (2005): The econometrics of macroeconomic modelling. Oxford University Press, Oxford. Benati, Luca, Lucas Jr, Robert E., Nicolini, Juan Pablo, and Warren Weber (2020): International evidence on long-run money demand. Journal of Monetary Economics, in press. Chetty, V. Karuppan (1969): On measuring the nearness of near-moneys. The American Economic Review, 59(3), pp. 270—281. Eitrheim, Øyvind (1998): The demand for broad money in Norway, 1969–1993. Empirical Economics, 23(3), pp. 339—354. Jadidzadeh, Ali, and Apostolos Serletis (2019): The demand for assets and optimal monetary aggregation. Journal of Money, Credit and Banking, 51(4), pp. 929—952. Knell, Markus, and Helmut Stix (2006): Three decades of money demand studies: differences and similarities. Applied Economics, 38(7), pp. 805—818. Lucas Jr, Robert E., and Juan Pablo Nicolini (2015): On the stability of money demand. Journal of Monetary Economics, 73, pp. 48—65. Nagel, Stefan (2016): The Liquidity Premium of Near-Money Assets. The Quarterly Journal of Economics, 131(4), pp. 1927—1971. Poterba, James M., and Julio J. Rotemberg (1986): Money in the utility function: An empirical implementation. NBER Working Paper No. 1796. Wang, Zhu and Alexander L. Wolman (2016): Payment choice and currency use: Insights from two billion retail transactions. Journal of Monetary Economics, 84, pp. 94—115. Money demand – household perspective and financial intermediation persepective Amel, Dean F., and Timothy H. Hannan (1999): Establishing banking market definitions through estimation of residual deposit supply equations. Journal of Banking & Finance, 23(11), pp. 1667—1690. Arrow, Kenneth J. (1971): The theory of risk aversion. Essays in the theory of riskbearing, pp. 90—120. Berger, Allen N., and Astrid A. Dick (2007): Entry into banking markets and the early‐ mover advantage. Journal of Money, Credit and Banking, 39(4), pp. 775—807. 24 NORGES BANK STAFF MEMO NO. 8 | 2020 THE DEMAND FOR SAFE LIQUID ASSETS AND THE IMPLICATIONS OF ISSUING A CENTRAL BANK DIGITAL CURRENCY FOR BANK FUNDING INSTRUMENTS Berlin, Mitchell, and Loretta J. Mester (1999): Deposits and relationship lending. The Review of Financial Studies, 12(3), pp. 579—607. Brunetti, M, Ciciretti, R. and Lj. Djordjevic (2016): The determinants of household’s bank switching. Journal of Financial Stability, 26, pp. 175—189. Choi, Dong Beom, and Ulysses Velasquez (2018): Hidden cost of better bank services: carefree depositors in riskier banks? FRB of NY Staff Report No. 760. Cornett, M. M., McNutt, J. J., Strahan, P. E., & Tehranian, H. (2011). Liquidity risk management and credit supply in the financial crisis. Journal of Financial Economics, 101(2), pp. 297—312. Dick, Astrid A. (2008): Demand estimation and consumer welfare in the banking industry. Journal of Banking & Finance, 32(8), pp. 1661—1676. Flannery, Mark J., and Christopher M. James (1984): Market evidence on the effective maturity of bank assets and liabilities. Journal of Money, Credit and Banking, 16(4), pp. 435—445. Flannery, Mark J. (1982): Retail bank deposits as quasi-fixed factors of production. The American Economic Review, 72(3), pp. 527—536. Hochguertel, Stefan, Rob Alessie, and Arthur Van Soest (1997): Saving accounts versus stocks and bonds in household portfolio allocation. Scandinavian Journal of Economics, 99(1), pp. 81—97. Karlan, Dean, and Jonathan Zinman (2018): Price and control elasticities of demand for savings. Journal of Development Economics, 130, pp. 145—159. Kiser, Elizabeth K. (2002): Household switching behavior at depository institutions: evidence from survey data. The Antitrust Bulletin, 47(4), pp. 619—640. Lie, Aud Ebba (2011): Finanstilsynets innskuddundersøkelse under finanskrisen – en kort oppsummering. Penger og Kreditt, 2, pp. 14—18. Lin, Leming (2020): Bank deposits and the stock market. The Review of Financial Studies, 33(6), pp. 2622—2658. Perraudin, William RM, and Bent E. Sørensen (2000): The demand for risky assets: Sample selection and household portfolios. Journal of Econometrics, 97(1), pp. 117—144. Sharpe, Steven A. (1997): The effect of consumer switching costs on prices: A theory and its application to the bank deposit market. Review of Industrial Organization, 12(1), pp. 79—94. Vissing-Jorgensen, Annette (2002): Towards an explanation of household portfolio choice heterogeneity: Nonfinancial income and participation cost structures. NBER Working Paper No. w8884.