Money Creation in Fiat and Digital Currency Systems
  • 1, International Monetary Fund
  • | 2, International Monetary Fund

Contributor Notes

Authors’ email addresses:,

To support the understanding that banks’ debt issuance means money creation, while centralized nonbank financial institutions’ and decentralized bond market intermediary lending does not, the paper aims to convey two related points: First, the notion of money creation as a result of banks’ loan creation is compatible with the notion of liquid funding needs in a multi-bank system, in which liquid fund (reserve) transfers across banks happen naturally. Second, interest rate-based monetary policy has a bearing on macroeconomic dynamics precisely due to that multi-bank structure. It would lose its impact in the hypothetical case that only one (“singular”) commercial bank would exist. We link our discussion to the emergence and design of central bank digital currencies (CBDC), with a special focus on how loans would be granted in a CBDC world.


To support the understanding that banks’ debt issuance means money creation, while centralized nonbank financial institutions’ and decentralized bond market intermediary lending does not, the paper aims to convey two related points: First, the notion of money creation as a result of banks’ loan creation is compatible with the notion of liquid funding needs in a multi-bank system, in which liquid fund (reserve) transfers across banks happen naturally. Second, interest rate-based monetary policy has a bearing on macroeconomic dynamics precisely due to that multi-bank structure. It would lose its impact in the hypothetical case that only one (“singular”) commercial bank would exist. We link our discussion to the emergence and design of central bank digital currencies (CBDC), with a special focus on how loans would be granted in a CBDC world.

I. Introduction

Three structural features of monetary systems form the starting point for this paper. They include (1) a two-layer structure comprising a private sector agent deposit system with commercial banks, parallel to a commercial bank deposit (reserve) system with central banks, with the two being separate and not allowing transfers between the two, that is, reserves cannot be “lent out” to the private sector (Sheard 2013); (2) the fact that money is created upon the creation of bank loans (Werner 2014/16), while repayment implies money destruction, both of which are an accounting reality, and implying that banks would better not be called “intermediaries”; and (3) the fact that the money stock is endogenously and elastically driven by demand and constrained loosely by regulation. The constraints to the provision of new credit include capital regulation, banks’ conditionality on an incremental profit prospect, and, eventually, demand (McLeay et al. 2014)1.

Against this background of economic realities, some misled views of how bank lending functions have come about and persisted since long. These include the financial intermediation and the fractional reserve theories of banking, which can be subsumed under a generic “loanable funds” header. They have a micro perspective in falsely interpreting the need for liquid funds as meaning that banks “pass deposits (money) on” when creating loans2. This deceptive understanding is, moreover, related to the misleading notion that “savings finance investment,” which should rather be phrased as “savings (tautologically) equal investment” (Lavoie 1992, Davidson 1993, Shapiro 2005).

The money creation understanding matters for seeing the role of monetary and macroprudential policy in qualitative and quantitative terms. The intermediation view of banking has implied the loss of relevance of banks and bank credit in macroeconomic models and effectively led to them being dropped in dynamic stochastic general equilibrium (DSGE) models for long. Banks were and still are regularly implemented as if they were nonbank financial intermediaries; which implies the risk of underestimating financial accelerator effects.

Against this combined background, our objective is to highlight that the money creation reality and the notion of “liquid funding needs” are compatible3. We sense that the reason for why some skepticism with the money creation view persists may lie in the presence of another economic reality, that is, the need for “liquid funding” with which the money creation view may seem to be incompatible at first sight. Our aim is to emphasize that they are compatible. We follow two routes to that end.

First, we develop a series of balance sheet examples based on double-entry bookkeeping principles. Unlike in the existing related literature, we do not focus on only bank lending (money creation), but also lay out the lending process through centralized nonbank financial institutions as well as decentralized market-based intermediation. After presenting a series of required definitions in Section III, we put forward the balance sheet examples in Section IV.

Second, we develop a simple stock-flow consistent agent-based model (ABM) to illustrate the monetary system dynamics related to the loan and money creation process. The ABM methodology lends itself usefully to implementing the money creation view correctly, as it is built on a stock-flow consistent integrated balance sheet structure; in the case of our model for a population of private sector agents, banks, one central bank, and optionally a sovereign4. Based on the model, we highlight that liquid funds are required to back the transfer of newly created, initially illiquid loans and deposits in a multi-bank system. Liquid funding needs do not negate the fact that banks create money “out of nothing.”

Based on the ABM, we simulate the hypothetical case of compressing the commercial banking system down to a “singular” one. When considering only one commercial bank and no physical cash, liquid funding (reserve) needs would vanish, liquidity risk cease to exist, and conventional, interest-based monetary policy lose its impact on the economy. The reason is that central bank reserves would not be needed anymore, as all cross-agent transfers would happen on the one commercial bank’s electronic deposit pool. It is a very hypothetical, yet insightful in our view, additional insight. We present the ABM and the related simulations in Section V.

Two closely related topics are discussed: they concern bond versus loan finance for governments and the emergence of central bank digital currencies (CBDCs). The two topics are placed in Sections IV and VI, respectively. Concerning CBDCs, we emphasize two points: (1) the emergence of CBDCs should not be seen as a result of central banks’ fear of losing their bearing on the economy if physical cash was vanishing, that is, we argue that the monetary policy transmission with and without physical cash is virtually identical; and (2) an important question is how CBDC systems would be designed in terms of credit provision, which, if backed 100 percent by the new digital currency, would resemble the Chicago plan of the 1930s (a topic that we link to in the literature review section as well).

II. Literature

Figure 1 presents a series of papers and books spanning the last century, which we have categorized in three groups. These groups include the intermediation, the fractional reserve, and the money creation view. The intermediation view assumes that banks “pass on money” when granting loans, so neither individual banks nor the banking system create money when doing so. The fractional reserve view posits that individual banks do not, but the banking system does, create money when granting loans. If its unrealistic assumption of new loans being converted 100 percent to physical cash5 and redeposited elsewhere was dropped, it would collapse to the third view, the money creation understanding under which both individual banks as well as the banking system create money when granting loans.

Figure 1.
Figure 1.

Literature Review

Citation: IMF Working Papers 2019, 285; 10.5089/9781513521565.001.A001

Source: The authors.

The intermediation versus money creation debate is inherently intertwined with the opposing views of monetarists and Post-Keynesians. Monetarists (led by Friedman) think that the money supply is exogenous, with central banks setting the monetary base and thereby being believed to be acting as though the effective money stock can be controlled through the money multiplier process. Money supply endogeneity, on the other hand, has its origin in Keynes (1930) and was made more explicit by Robinson (1956), Davidson (1978), Kaldor (1982), and Moore (1986). The institutional reality is in line with the latter, as documented in the numerous references, including from central banks, under the money creation view in Figure 16.

A significant literature stream that was in line with the money creation view had evolved between the end of the 19th century and the Great Depression in 1929. Useful entry points to the early literature include the references by MacLeod, Wicksell and Schumpeter, and others in Figure 1 under the money creation view. Annex II contains a series of quotes that corroborate the money creation and endogenous money view.

The money creation view came to the fore (of policy) when the Chicago plan was proposed as a consequence of the Great Depression in 1929. The idea was to deprive commercial banks of their credit and hence money creation ability by introducing a 100 percent reserve system (Knight 1933, Currie 1934, Douglas 1935, Fisher 1936, Graham 1936, Simons 1934/36). The rationale for doing so stemmed from the understanding that the money creation ability of commercial banks was, allegedly, the root cause of the Depression at the time, and of recurrently evolving endogenous business and financial cycle dynamics more generally. The initial Chicago economists’ proposal was revived a second time toward the end of the Great Depression period in 1939 (Douglas et al. 1939), when a first post-depression, smaller-scale recession occurred that was conceived to be the result of the FED raising reserve requirements that year. Neither the initial nor the second attempt to implement the 100 percent reserve plan ever succeeded7.

The money creation view then became increasingly repressed during the 1940-50s, with the intermediation view fading into mainstream economics instead. This development seems to have started with books and related papers by Gurley and Shaw (1955/56/60). The intermediation view implied that the role of bank credit was significantly diminished, which one may see as the reason why DSGE models largely dropped banks and bank credit toward the end of the 20th century.

After the Global Financial Crisis (GFC) of 2007-09, the money creation view re-surfaced, including through central bank communication. A series of bulletin articles by the Bank of England, more papers from Werner, the Bundesbank, and others (Ryan-Collins et al. 2011, McLeay et al. 2014a/b, Werner 2014a/2016, Deutsche Bundesbank 2017) all aimed at reviving the money creation understanding. In the long intermittent period, a few names that kept being aligned with the money creation view included, for example, Minsky (1978), Moore (1979/83), Bourva (1992), and Werner (2005).

Past and contemporary DSGE models were and still are largely based on the intermediation view, which implies the risk of underestimating the role of banks, credit, and financial accelerator effects. A sizeable set of DSGE model papers could be cited under the financial intermediation view in Figure 18. Any paper that refers to banks as being “intermediaries” usually tends to thereby show its affiliation with the intermediation view. An assessment of the consequences of falsely assuming an intermediation view in DSGE models is presented in Jakab and Kumhof (2015).

How cross-border monetary flows are understood in relation to domestic (versus global) money supply is a matter of scope. Monetary flows into a country, for example, in the form of foreign currency deposits (liability) along with reserves (assets) to domestic banks, add to the money stock, that is, mean money creation, as long as one assumes a local economy perspective. From a global perspective, it depends on whether the money inflows are generated via foreign bank loans (money creation at the global level) or the channeling of existing funds (no money creation at the global level). Ponomarenko (2017) focuses on this one form of leakage, the “foreign sector,” and discusses the importance of banks’ net foreign asset-liability stock position and related flows for influencing deposit stock dynamics, in particular in emerging market economies.

We conclude our literature review with two observations. First, we observe that the money creation understanding is to an extent cyclical, in the sense of it reappearing and becoming more widely understood after “heavy recessions” such as the Great Depression in the 1930s and the GFC in 2007-09. “Normal” recessions of intermediate levels of severity do not seem to have sufficed to let the understanding resurface, as only deep recessions imply drastic needs for rethinking and possibly significant policy action. Second, the money creation literature has, in our reading, not sufficiently emphasized how the “need for liquid funding” squares with the money creation view. This is a gap we aim to fill with this paper.

III. Definitions

Nine definitions related to money and credit form the basis for the subsequent work in this paper. They are presented in Box 1. The first five are well known. The sixth and seventh pertain to our differentiation of bank and nonbank lending, including a distinction between centralized financial institution-based nonbank intermediation versus de-centralized bond market intermediation, and a link to the notion of shadow banking. The eighth definition stipulates the different components of the total money stock to which we refer in the examples that we present later. The ninth definition is central to our discussion, relating to the notion of liquid funding.

The notion of “at par” appears in some of the definitions. By “at par” we mean the initial nominal value that a money holder deposited on a bank’s deposit account, that is, without it being at risk of fluctuations in that nominal value. This in turn means that no secondary market has to be approached for redeeming such at par holdings before transferring them, whether into physical cash or in electronic form to other banks.

The distinction between bank and nonbank lending is important from a system perspective. Both bank lending (Definition 6) and nonbank lending (Definition 7) create leverage for the borrower in the same way, while bank lending does, and nonbank lending does not, however, increase system-wide money stocks upon the creation of debt through these two channels (and likewise decrease and do not decrease through repayment in the two categories).

We can equate our notion of nonbank lending (Definition 7) with shadow banking. The Financial Stability Board (FSB), for instance, defines shadow banking as a “system of credit intermediation that involves entities and activities fully or partially outside the regular banking system” (FSB 2014, page 4). We will come back to some details of this definition and what components it effectively comprises in Section IV. We define “nonbank agents” (which appear in some of the definitions below) as including households; nonfinancial firms; nonbank financial firms such as pension funds, insurance companies, investment funds, etc.; as well as state governments (both central and local). Nonbank financial institutions, furthermore, include firm types such as money market funds (MMFs), investment funds, hedge funds, captive financial institutions and money lenders, broker-dealers, structured finance vehicles, trust companies, and real estate investment trusts (REITs). They all have in common that they temporarily channel existing money stocks from one subset of economic agents to others9.

The notion of “disintermediation” is misleading. Given that bank lending does not mean intermediation (but money creation), while non-bank financial firms’ lending implies pure intermediation, the terminology “disintermediation” the way it is used is misled. It conceives bank lending as an intermediation process and the move toward non-bank (shadow bank) lending as a move away from this intermediation.


Definition 1: Money. Following standard textbooks (Mankiw 2010), we define money as any medium that is commonly considered to have the following three properties by a sufficient portion of the population10: (1) store of value, for money to allow its holder to conserve purchasing power over time; (2) unit of account, for money to serve as a reference in which the value of goods and services is measured; and (3) medium of exchange, as money is meant to flow for the purpose of settling transactions. This is a social definition that hinges on agents’ acceptance of a given medium as money.

Definition 2: Legal money. Legal money is a medium that is prescribed to be money by laws and regulations, that is, imposed by a government. The most tangible criterion of legal money is whether a medium can be used to pay taxes. It may not bear any own intrinsic value, say, when existing in paper or electronic form. The term “legal money” may be used synonymously with “fiat money” and “legal tender.”

Definition 3: Commercial bank money. We define commercial bank money as the portion of the total money stock that nonbank agents hold in the form of electronic bank deposits. Nonbank agents can “swap” commercial bank money into physical cash back and forth (equivalently transfer funds electronically across banks), while thereby leaving system-wide money stocks unchanged.

Definition 4: Bank. A bank is a firm whose liabilities consist at least in part of legal money (Definition 2) for other economic agents, who conceive these holdings to be redeemable at par value at all times, to serve the functions of money as outlined under Definition 1.

Definition 5: Nonbank financial firms. They collect and channel on the existing money stocks of other economic agents who serve as “fund providers.” Such fund providers’ invested amounts of money may not be retrievable at par, that is, be subject to market risk and involve approaching a secondary market to redeem the investment, and hence do not serve the functions of money (Definition 1) while being invested.

Definition 6: Bank lending. Bank lending occurs when banks (Definition 4) create a deposit (money) through granting a loan. Bank lending increases total money stocks, while bank loan repayments destroy total money stocks accordingly.

Definition 7: Nonbank lending (= shadow bank lending). Nonbank lending occurs when an economic agent passes existing legal money stocks on to another agent—subtracting from its own money holding and adding to another’s—through a debt contract that stipulates its repayment in the future.

Definition 7a: Nonbank lending through centralized nonbank financial institutions. Denotes the nonbank lending process (Definition 7) carried out by nonbank financial firms (Definition 5).

Definition 7b: Nonbank lending through decentralized primary bond markets. Denotes the nonbank lending process (Definition 7) through decentralized primary bond market issuances of debt. They entail the collection of multiple lenders’ existing funds to channel them to one borrower.

Definition 8: Components of the total money stock. Total money is the combined value of all units falling under Definition 1. It consists of the following quantities: (1) Commercial bank money (Definition 3); in our examples: deposits. (2) Central bank money: central bank liabilities that are socially accepted money and that are not held as an asset by a bank; in our examples: cash not held by banks, government reserves. (3) Nonbank money: all forms of socially accepted money that are not included in the quantities listed in (1) or (2)./1

Definition 9: Liquid funding needs. This notion refers to the need for commercial banks to hold reserves at their central bank accounts for the purpose of cross-bank settlement of customer deposits. From an individual bank perspective, liquid funding needs can be satisfied in three ways: (1) by receiving existing deposits (liability side) along with reserves (asset side) from other banks; (2) by borrowing them through money markets from other banks; or (3) by borrowing them from the central bank. /2 /3

1/ Banks’ reserves in their central bank accounts shall not be considered part of broad money, because the reserves can be used by commercial banks only for backing cross-bank transfers of flows between nonbank public deposit accounts. For example, banks cannot transact directly with the private sector using such reserve stocks; they cannot be spent for either the purchase of financial or nonfinancial products. It should be kept in mind, however, that in the US, commercial banks may use their central bank reserves to purchase newly issued treasury bonds, which implies a liability swap from the bank reserve account at the FED to the sovereign account at the FED (US Treasury 2004).2/ It is the first option among these three, the pull of existing deposits (along with reserves), that supposedly contributes to the prevalence of the deceived intermediation view.3/ Our definition of liquid funding needs is compatible with more generic definitions of the kind put forward in the IMF’s GFSR (2008), whereby “funding liquidity is the ability of a solvent institution to make agreed-upon payments in a timely fashion.” Such payments (likely to other banks) require the availability of reserves.

Examples of nonbank lending through centralized nonbank financial institutions (Definition 7a) include peer-to-peer lending, investment funds, hedge funds, trust companies, etc. Examples of decentralized nonbank lending (Definition 7b) include primary corporate and sovereign bond markets. Secondary bond markets then give bond holders (investors) the opportunity to exchange ownership of existing bond contracts and the associated right for receiving their principal and interest payment flows in the future.

IV. Balance Sheet Stock and Flow Mechanics under Bank and Nonbank Lending11

A. Initialization

We begin with a “tabula rasa” economy, that is, all agents start with empty balance sheets. To show how money can be introduced into an economy without money, we first assume that the private sector agent and the bank agent each gains possession of some real assets, for example, gold (for the sake of our example, it does not matter whether these assets have the quality of non-fiat money or not). So far, there is no fiat money in the system.

Both agents then transfer their assets to a central bank agent in exchange for newly created currency. These initial currency endowments of the agents correspond to the outside (fiat) money in the system, since all other assets (including all non-fiat money) were transferred to the central bank. The assumption of an initial asset transfer ensures that the central bank starts with zero equity.

In principle, a central bank could also distribute currency without receiving any assets in exchange, in which case it would start with negative equity. This negative equity would equal the amount of currency, unless it gained possession of assets in a different manner, for example, through endowment by a sovereign. The one bank agent can be seen as the consolidated banking system. Figure 2 shows the booking statements (flows) along with the resulting balance sheet positions (stocks) after initialization.

Figure 2.
Figure 2.

Initializing the Balance Sheets

Citation: IMF Working Papers 2019, 285; 10.5089/9781513521565.001.A001

B. Granting and Repayment of Loans to the Private Sector

The creation of loans implies the creation of deposits. These may not initially be backed by central bank reserve amounts of the bank that grants the loan, and in that sense may not be “liquid” in case the newly created deposit would want to be transferred across banks in an electronic manner (or converted into cash, an option we abstract from without loss of generality of our exposition). Figure 3 shows the booking statements and resulting balance sheet structure.

Figure 3.
Figure 3.

Loan Creation for the Private Sector

Citation: IMF Working Papers 2019, 285; 10.5089/9781513521565.001.A001

The bank has three options for obtaining reserves to accomplish a cross-bank transfer of the newly created deposit. First, it may borrow reserves from other banks through the interbank market. Second, it may incentivize depositors of other banks to move their deposits to itself, which would flow along with reserves. Third, it may borrow reserves from the central bank.

Various interest payment flows are involved, which we do not present in detail here12. Loan interest rates imply interest income for the banks, which can in turn be assumed to be distributed back to private sector agents through dividends, which are an expense along with deposit interest. Net interest income flows with the private sector add to residual equity of banks; they leave the size of the banking system’s consolidated balance sheet unchanged. Interest expense flows for central bank borrowing let the consolidated reserve stock shrink (all else equal), while remuneration on reserves increases reserve stocks13.

C. Bank versus Nonbank Lending (Example Involving a Sovereign Borrower)

Lending by nonbank financial institutions does not change total money stocks (Figure 4). It implies a temporary reallocation of existing monetary funds from one agent to another. In the example, the household sector invests part of its existing money holdings (deposits) in sovereign bonds. The total money stock remains constant throughout the example and would be no different conceptually if we had set it up as a nonfinancial corporate firm’s bond issuance instead of a sovereign one.

Figure 4.
Figure 4.

Nonbank Lending (Primary Sovereign Bond Market Auction)

Citation: IMF Working Papers 2019, 285; 10.5089/9781513521565.001.A001

The purchase of bonds issued by banks destroys money temporarily. Bonds issued by banks are excluded from our—and from most typical—definitions of the money stock, rightly so because bank bonds are not necessarily redeemable at par for immediate transaction purposes (see Paragraph 7 and Definition 1). Hence, if bonds are purchased by private sector agents with deposits, this transaction has the effect of reducing the total money stock temporarily. Their repayment means a subsequent feed of that money back to the system. Interest paid on top of the bond principal would increase the money stock to levels exceeding the nominal money stock before the bank bond issuance.

The question arises as to whether the choice between bank loan and bond finance would make a difference for macroeconomic dynamics at large. It appears that not much academic research has been conducted on this question so far.

When private sector agents’ money holdings are invested in sovereign bonds, they cannot be spent otherwise. In that sense, they subtract from the purchasing power of private sector agents, who chose to do so with their savings stocks, which, however, they do not plan to spend momentarily anyhow. From that perspective alone, system dynamics may not differ much when comparing the loan versus bond finance for the sovereign case.

Banks granting loans to the sovereign implies a stronger sovereign bank nexus. The banking system’s solvency risk would be more tightly connected to variation in sovereign risk. This additional tie would come on top of explicit and implicit guarantees that the sovereign may carry with regard to the banks’ liabilities to the private sector.

D. Securitization

The purchase of loans through securitization implies the destruction of money stocks (Figure 5). The purchase price is paid in the form of bank money, which is destroyed in the process of the purchase. The destruction of money through securitization is preceded by the creation of money through loans (see subsection B above), which are about to be securitized. Hence, the eventual outcome of the securitization (here assumed to occur to 100 percent) would render the provision of bank credit equivalent to the nonbank financial intermediation case (see subsection C above)—that is, the initially newly created money will be neutralized by an investment based on existing money stocks, invested by nonbank financial or nonfinancial agents14.

Figure 5.
Figure 5.

V. Agent-Based Model Simulation

We present an ABM to illustrate how banks—even when they are unconstrained by regulation in their ability to create money—can face a situation of liquid funding shortfalls. In the model presented in the following, the central bank is willing to refinance the banks unconditionally. In reality, the central bank may be constrained by its mandate or unwilling to do so, thus making it possible that solvent banks fail due to illiquidity.

Our monetary ABM comprises one central bank along with banks and private sector agents. Our ABM comprises one central bank (CB), b=1,...,B commercial banks and n=1,...,N private sector agents. There is no physical cash by assumption, for all money in the hands of the N private agents being held in electronic form in deposit accounts at the B banks. Private agents can be seen as consumers and firms at the same time. Figure 6 shows a schematic of the components of the ABM, which are described briefly below. A form of annotated pseudo code for the ABM is presented in Annex I, along with an outline of how the simulated model outcome behaves as a function of the core model parameters.

Figure 6.
Figure 6.

Agent-Based Model Structure

Citation: IMF Working Papers 2019, 285; 10.5089/9781513521565.001.A001

Note: The schematic summarizes the main features and building blocks of the ABM. See text for details.

The model is very stylized, involving hardly any behavioral elements for the agents in the model. The stylized nature of the model and the absence of behavioral rules mean that the model is not meant to be fitted to real world data. The model’s sole purpose lies in making some conceptual points about money creation and liquid funding and transfer processes, to reflect reality in these specific respects.

The three agent sets’ individual balance sheets are initialized using a random drawing procedure and integrated for all balance sheets to be cross-consistent and connected. Private sector agents’ money stocks are drawn from a uniform distribution. The N private sector agents are assigned permanently to the B banks. The private agents’ money holdings are held in electronic form in their bank deposit accounts.

The spending process in the model is random. Each period, all private sector agents receive random signals to spend, in which case they get a randomly chosen other private agent assigned as a recipient. The random spending signal is based on a uniform random draw, which is compared to a threshold parameter, and which we refer to as the “spending propensity.” Spending flows are uniform random and set as a percentage of money holdings. The latter parameter we call the “spending fraction.”

A cross-bank net settlement process follows. Banks compute their net deposit transfer needs, which can be positive or negative in a given period, and which sum to zero at the system level at all times (net financial flows sum to zero just as net financial asset stocks do, since each one agent’s financial asset must be another one’s liability). If a bank’s liquid funding needs exceed their available reserve amount at the CB agent’s account, they borrow the required residual reserve amount from the CB. In the model, the bank agents do not have the option to borrow reserves from other banks or to actively incentivize private sector agents from other banks to move deposits, along with reserves, into a bank.

A random loan creation and subsequent repayment process follows. There are random loan granting signals for the banks to create uncollateralized credit for the private sector agents at an endogenous loan interest rate i. The random signal is again based on a uniform random draw and a threshold model parameter on the [0,1] interval, which is referred to as the “loan granting propensity.”

On the liability side of the banks, deposits imply a zero expense by assumption. The only expense flows that banks face in the model is for CB reserve borrowing, with r denoting the CB funding rate. We will later see that the relation between the loan interest rate i and the CB rate r depends on how many banks there are and how frequent intrabank transfers are. The policy rate follows a white noise random Normal process15.

The model abstracts from numerous other economic realities. First, none of the agents can default. Second, there is no money market. Should non-zero net transfers need to be backed by liquid funds that are temporarily unavailable at the bank, the CB’s standing borrowing facility will have to be called. Third, there is no explicit distinction between consumption and investment flows. The term “spending” is meant to be the neutral expression comprising both household consumption and firm investment flows. Fourth, profit margins are set to zero in the model, although they would be positive and a function of competition in the system (the number of banks) in reality. All these features are on purpose excluded from the scope of the model for its structure to be reduced to the essentials that are needed to make our points. Importantly, the addition of any of such “relevant realities” would not change the conclusions that we draw from the model.

A baseline simulation was run with B=30 banks, N=500 private agents, and T=600 periods. The threshold spending propensity parameter was set to 25 percent, the spending fraction to 50 percent, and the loan granting probability to 75 percent. The system aggregates for various stocks and flows for the aggregate private and banking sectors are depicted in Figures 7 and 8.

Figure 7.
Figure 7.

Exemplary Simulation Path for Selected Model Variables (Stock Variables)

Citation: IMF Working Papers 2019, 285; 10.5089/9781513521565.001.A001

Figure 8.
Figure 8.

Exemplary Simulation Path for Selected Model Variables (Flow Variables)

Citation: IMF Working Papers 2019, 285; 10.5089/9781513521565.001.A001

Figure 9 shows how a reduced subset of model variables behaves as a function of the number of banks. The CB funding over bank equity ratio, the percentage of banks pulling CB reserves, and the loan interest rate all fall monotonically with a decreasing number of banks, while the policy rate remains at a mean of 2 percent with some stochastic fluctuation (not shown in Figure 9). The size of the aggregate banking system and private sector balance sheets remain unaffected, just as all spending frequencies, spending flows, loan flows, loan stocks, etc. The correlation between the policy rate and the loan interest rate falls to zero. Hence the only effect that the variation of the number of banks has is that the same amount of monetary flows through the system need to be settled in an increasingly fine-grained banking system in which a rising share of transfers must be backed by liquid funds that are not initially available in the system and have therefore to be pulled from the CB. Importantly, we do not argue that the effective loan interest rate would fall toward zero in reality if the number of banks was compressed toward zero; the effect of falling competition, and hence more sizeable profit margins, may dominate and do the opposite to effective loan rates. The conclusion in terms of the vanishing correlation of policy rates with loan interest rates, however, would keep holding up, which is the main point we aim to make.

Figure 9.
Figure 9.

Compressing the Banking System Down to a “Singular” One

Citation: IMF Working Papers 2019, 285; 10.5089/9781513521565.001.A001

Note: The figure illustrates the effect of reducing the number of banks in a banking system on selected variables. Blue lines depict the median, gray lines the 25th/75th percentiles from 500 simulation rounds of the model, each for 300 periods forward in time. The model does not take account of elements of competition. The loan interest rate is defined as the periodic loan interest payment flows over the outstanding principal before an annuity payment is instructed.

VI. Central Bank Digital Currency (CBDC)

Numerous monetary authorities around the world consider introducing (experiment with) retail CBDCs: their introduction would mean that private sector agents’ money holdings could be held as an electronic central bank liability. Commercial banks have held their reserves in electronic form at their central bank accounts for decades, which is conform with the CBDC definition and just being confined to banks. Useful entry points to the recent discussions around retail CBDC design, its underlying rationale, costs, benefits, etc., can be found in Mancino-Griffoli et al. (2018), Bindseil (2019), and Barontini and Holden (2019)16.

An unregulated introduction of CBDC creates the risk of a cliff-effect threatening to undermine financial stability. It is conceivable that the benefits of CBDC will not immediately be internalized by private sector agents. In a healthy economy, old conventions with regard to the treatment of bank deposits as money may continue to hold, in particular when commercial bank deposit interest rates would be set at a margin above a CBDC rate (whether itself zero or positive). This may change abruptly at the onset of a crisis, however. If the viability of private banks would be called into question, agents may disregard the higher interest on bank deposits and collectively flee into CBDC. This would effectively constitute an “electronic” bank run, with all the known consequences for banks’ liquidity (reserve shortage/insufficiency). Given the relative inconvenience of physical cash compared to bank deposits, the promise of deposit insurance is an adequate tool for mitigating this risk in the current monetary system. This may, however, no longer be the case in a system where a convenient alternative to deposit money is available.

Regulations regarding the use of CBDC can address some of the financial stability risks posed by its introduction. While a CBDC would conceivably always be treated as legal tender, the sovereign still can restrict its use as social money, per Definition 1. By imposing limits on how much CBDC can be held by any one individual (stock constraints), the sovereign could effectively curtail its use as a store of value. Further imposing limits on transaction volumes (flow constraints) can be used to reduce its use as a medium of payment, restricting its use to smaller transactions and thereby rendering it more similar in usability to physical cash.

Conventional central bank policy tools can further mitigate the risks from introducing CBDCs. The interest rate on CBDC deposits can be an effective tool for managing the desirability of CBDC holdings relative to bank deposits, and even negative interest rates on such deposits could easily be implemented. Small differences in interest rates are unlikely to prevent a bank run, however. In the case of a bank run, the central bank in principle has unlimited resources to refinance banks. In practice, however, it may be constrained by its mandate to do so: if banks with solvency problems are hit by bank runs, then these may in fact not have sufficient assets of high enough quality to be used as collateral in emergency liquidity assistance programs.

As with deposit money, a crucial question is how the lending process under a CBDC system would work. To the best of our knowledge, this question is not yet explicitly addressed in the ongoing discussions around CBDC. In principle, the introduction of a CBDC could occur entirely through open market transactions, for example, the purchase of government bonds on secondary markets, or even through more radical options amounting to “helicopter money” distributions. In an economy that fully transitions to central bank money, it seems unlikely, however, that such actions could sustain sufficiently high money creation flows to avoid deflationary tendencies. We are thus interested in how lending that is denoted directly in CBDC could occur, and what role commercial banks would play in such a system. Figure 10 presents two possible implementations of such a system17.

Figure 10.
Figure 10.
Figure 10.

CBDC-Denoted Lending

Citation: IMF Working Papers 2019, 285; 10.5089/9781513521565.001.A001

If the risks of transitioning to a CBDC are well managed, it constitutes a viable option to switch the economy to a “full money” system. As noted in the discussion on the Chicago plan (Section II), removing private banks’ money creation ability may offer potential benefits (a more detailed discussion of related pro and contra arguments is beyond the scope of this paper). We merely caution here that the transition, if not well managed, could induce serious unintended consequences. If well managed, however, a CBDC may be a useful tool for ensuring a smooth and gradual transition towards a full money system.

The move to a 100 percent CBDC reserve system can come into existence in three ways. First, if imposed by a central bank when requiring loans granted in commercial bank deposit form to be backed 100 percent by central bank reserves (CBDC); second, by granting loans in CBDC on the central bank balance sheet right away; or third, when private sector agents would choose to shift virtually all their commercial bank deposits (created through commercial bank loans and deposits) to their CBDC accounts. In the first two cases, the move to the “full money” system would be imposed, while the third scenario could come about without even being intended by a regulator.

VII. Conclusions

We have highlighted that money creation and liquid funding needs are two realities that coexist. Liquid funding needs in the form of commercial banks’ reserves in their central bank accounts are relevant to back payment transfers in a multi-bank system, which is a point that some readers may argue is trivial. Liquid funding needs, whether addressed ex ante or ex post relative to the loan creation process, and whether based on pulling deposits along with reserves from other banks, through interbank markets, or standing borrowing facilities from a central bank, do not negate the fact that banks create deposits and hence money “out of nothing” upon the creation of loans.

We have considered a hypothetical thought experiment, assuming that a commercial banking system was reduced to a singular system, in which case the central bank would lose its handle on economic dynamics. In the presence of only one bank (and the absence of physical cash), liquid funding needs would cease to exist and the central banks’ handle via conventional policy instruments over bank funding costs, hence bank loan interest rates and in turn macroeconomic conditions, be lost18. We have illustrated that fact based on a simple, stock-flow consistent monetary ABM.

From our discussion related to CBDCs, we conclude that the mere disappearance of physical cash would not imply that monetary policy’s bearing on economic dynamics be in any way reduced. The disappearance of cash cannot be the argument for calling for CBDCs from that perspective. More relevant arguments for digital instead of physical cash include instead the objective to be better able to counteract money laundering and combat the financing of terrorism and tax evasion.

An important question that should be explored in detail concerns how a loan-granting process would be designed in a system with a CBDC parallel to commercial bank money. A 100 percent backing of loans by digital CBDC, if implemented as such, would indeed imply an implementation of the Chicago plan as considered in the 1930s in the aftermath of the Great Depression. The shift to such a full money system could also come about without regulations, if private sector agents decide to prefer CBDC to bank deposits as social money. It would in any case mark a highly significant change to the financial system structure.

Pure financial intermediary business, which is flourishing as part of FinTech developments in many countries (e.g. in the form of peer-to-peer lending), may not be able to fully replace bank lending. This is a one of two conclusions that are tangent to our main conclusions above. Pure intermediaries’ lending potential is bound by standing money stocks in the system that are not required for transaction purposes (that is, saved), while banks’ lending implies money creation, and hence is more elastic and bound only by regulatory requirements and demand.

The merits of financing government expenditures based on bank loans as opposed to bonds should warrant deeper conceptual and quantitative research. This second tangent conclusion concerns the differential effects that the two forms of government finance may entail and relate to possible crowding-out effects that bond finance (intermediation) might imply, which bank loan-based financing (money creation) would not. On the other hand, the stronger sovereign-bank nexus that bank lending to the government would entail may be seen as a concern.

Money Creation in Fiat and Digital Currency Systems
Author: Mr. Marco Gross and Christoph Siebenbrunner