Where Money Comes From — Do Banks Lend Deposits, or Create Them?

From $20 billion to $1 trillion

Before the financial crisis, the total amount of money that U.S. banks deposited with the central bank, the Federal Reserve (the Fed), was just over $20 billion. A few years later, that figure comfortably surpassed $1 trillion. That is more than a fiftyfold increase.

The name for this money is reserves. Reserves are money that banks keep in an account opened with the central bank. When you transfer 1 million won from a bank A account to a bank B account, this is the money that moves between the two banks. Because it is cash used exclusively among banks, you will never touch it in your lifetime.

According to the textbooks, banks should have used this money as a foundation to expand loans severalfold, and the money supply in the economy should have ballooned accordingly. The name for that multiple is the money multiplier. The money multiplier is the multiple indicating how many units of commercial money in the economy are generated by a single unit of money put out by the central bank.

A working paper by the Fed itself investigated this part. A working paper is a discussion paper released before undergoing academic journal peer review. This paper examined the relationship between reserves, money, and bank lending using a time-series model (a statistical model that tracks changes over time) incorporating aggregate data and bank-level data. The conclusion is this. That mechanism does not work through either the standard money multiplier model or the bank lending channel. Even if reserve balances affect the economy, the standard multiplier story is unlikely to explain that effect (DOI: 10.17016/feds.2010.41, Federal Reserve working paper).

Then the sentence to examine is this.

"Banks gather deposits entrusted by people and lend out a portion of them."

Before reading, two things should be made clear. First, this is not a fraud exposure, but a debate between economic schools of thought. Second, the grades of the cited literature are uneven. Peer-reviewed papers, working papers, and preprints (papers made public before peer review) are mixed together. The very first citation was a Fed working paper. This article draws the line between where the agreed-upon facts end and where the battleground begins. The grade is noted at every heavy-load juncture.

Blackboard Arithmetic — And Where That Arithmetic Was Right

If you took an economics class, you have seen this math on the chalkboard. One million won, 900,000 won, 810,000 won, 729,000 won……

This is the calculation of the money multiplier mentioned earlier. Suppose banks only need to keep 10% of deposits as reserves. This ratio is the reserve requirement, and the practice of lending out all but a portion of deposits is fractional reserve banking. When 1 million won comes in, 900,000 won is lent out, and that 900,000 won is deposited again to generate 810,000 won. Add it all up to the end and it becomes 10 million won, and the multiplier is 10.

There is a hidden premise in this arithmetic. It is that deposits must come first before loans can go out. The name of the picture built on this premise is the loanable funds model. The loanable funds model pictures the bank as an intermediary that takes in savings and lends those savings out. Think of it as a warehouse keeper standing between people depositing grain in a village warehouse and people borrowing it. Grain that is not in the warehouse cannot be lent out. That single sentence is the entirety of this picture.

That does not mean this picture was always wrong.

There is a paper that makes this point. It is a paper analyzing the balance sheets (ledgers recording assets and debts) of Russian banks and the determinants of loan decisions from 2005 to 2017. This paper describes multiple deposit creation, in other words that chain on the chalkboard, as a feature of an underdeveloped financial system. The authors' conclusion is that in that era, the volume of loans was tied to the supply of reserves, whereas today's main constraint is not the supply of reserves but credit demand, in other words the demand of people looking to borrow. This can be read not as saying the picture was false, but that the constraints have shifted. The evidence is an analysis of a single country, Russia (DOI: 10.32609/0042-8736-2018-11-50-69).

There is also a preprint that pushes back against the side wanting to discard the multiplier. The rebuttal is as follows. The multiplier is not useless. That is because it measures how much intermediation the financial system is achieving. When measured using an indicator that captures the money supply broadly, US financial intermediation peaked in 1985 (DOI: 10.31235/osf.io/zusqa, preprint). Here, one feels tempted to summarize: "Read as an indicator it survives, and read as a chain it collapses." But that is this article's summary, not the authors'. The authors do not concede anywhere that the multiplier collapses.

A similar refinement has appeared in a peer-reviewed journal. People have commonly interpreted the reason the multiplier slumped as "because banks are simply sitting on new reserves." However, adding two things to the traditional multiplier model changed the picture. They are the sharply increased demand for reserves and capital constraints since 2008. Then, both the low multiplier and the surge in deposits observed in the US were explained. What the model implies is this. It is true that the demand for reserves has increased greatly. However, it is not to the level where banks passively absorb all newly created reserves (DOI: 10.1111/jmcb.12975).

"Banks Are Not Intermediaries of Loanable Funds"

A 2015 working paper made its argument right in the title. It is "Banks Are Not Intermediaries of Loanable Funds".

The authors place two models side by side. The bank in the loanable funds model receives already existing physical resources from savers and lends them out. The contrasting model's bank creates deposits made of new money as it makes a loan. A constraint is attached in the very next sentence. Even while doing so, banks are primarily constrained by profitability and solvency (DOI: 10.2139/ssrn.2612050, working paper). The authors' conclusion is as follows. If you keep other conditions the same in both models and apply the same shock, a difference emerges. The money creation model predicts that changes in lending are much larger, much faster, and have a much bigger impact on the real economy.

The sentence stating they are "constrained" is just as important as the sentence stating they "create". The common summary that "banks print money out of thin air" erases the constraint the authors noted in the same abstract. That they can create, that regulations permit them to create, and that creating is profitable are different sentences.

One working paper lists three conventional views and refutes only the last one. The first is the conventional view that without excess reserves, banks are "unable to lend more". Excess reserves are reserves that exceed the share required to be set aside. The second is the conventional view that money creation begins with new deposits coming in. The third is the conventional view that small banks cannot create money (DOI: 10.2139/ssrn.2120439, working paper).

This view also has a name. Endogenous money is the view that the amount of money circulating in the market is not given from the outside, but is determined as much as lending occurs within the economy. Instead of a picture where a central bank turns a faucet on and off, it is a picture where new pipes are laid wherever water is needed.

Tests Done with Real Data — And Where That Data Came From

Let us look at Russia first. One peer-reviewed journal article matched the three theories against monthly data from 2005 to 2019. The three theories are theories explaining money creation. Credit money theory argues that loans create deposits, fractional reserve theory argues that deposits must come first, and financial intermediation theory argues that banks intermediate savings. The best-supported one was credit money theory. What the authors call "robust evidence" is not the sequence, but a constraint. What mainly binds bank lending in Russia today is credit demand, which is consistent with what has been observed in the United States (DOI: 10.1093/cje/beae036).

Poland's data from 2001 to 2016 tested which side moves first. Bank loans move before deposits. They move before broad money (the money supply cast broadly to include cash and deposits) and even before the monetary base (money supplied directly by the central bank). The author summarizes this as "loans create deposits, not the other way around." There is one more thing. The broad money multiplier does not move ahead of broad money (DOI: 10.4337/roke.2019.03.09).

There are also studies with a broader scope. One working paper examined the relationship between savings (time deposits and savings accounts) and bank credit using panel data from 20 countries. Credit growth was not linked to prior changes in savings. However, the author attaches a qualifier. At least in the short run, that is. The author considers this to indicate that the intuition of loanable funds theory does not work well to explain macroeconomic dynamics. They also found that the share of savings deposits decreases during credit booms. From this result, the author deduces a major flaw of fractional reserve banking. It is the tendency to drive the market interest rate away from the natural interest rate (the interest rate when the economy is in equilibrium) (DOI: 10.33119/kaewps2017024, working paper). We note separately that this deduction belongs to the author and came within the framework of Austrian Business Cycle Theory.

The direction of the three is the same. We must also look at the scope. Two are single-country tests of emerging and transition economies, and the one using the broadest data is a working paper. That there are multiple tests pointing in the same direction and that it is a law for every country and every period are different statements.

Where Consensus Ends and Debate Begins

A 2022 paper in the «Journal of Economics» takes this debate to a strange place.

The baseline the authors set is this: When there is no uncertainty, the two processes yield the same allocation. They write that in such cases, using the much simpler loanable funds approach as a shortcut does not lose generality. They add that a limited equivalence result holds even when there is aggregate risk and markets and contracts are complete. Equivalence means the two pictures give the same answer (DOI: 10.1007/s00712-021-00747-7).

This result is neither "loanable funds is right" nor "endogenous money is wrong." Nor does it mean "uncertainty is the fork in the road." The second case shows why. If equivalence survives in a limited way even with aggregate risk, this paper has never told us where the divergence happens. What this theorem establishes is not the winner of the contest, but the zone where the simpler side can be used as a shortcut.

Some research suggests the distance between the two depends on the monetary regime. A paper modeling a representative bank writes this: Exogeneity is the view that the quantity of money is determined from the outside. The more banks fear liquidity pressure—a situation where they fall short of cash needed right away—the more the money supply process leans toward the logic of exogeneity. The paper assumes banks had no such fear back then (in 2021). This is because central banks operate under a floor system overflowing with reserves following quantitative easing, where reserves are not pulled back. A floor system is an operating framework that sets a lower bound on interest rates using the interest paid on reserves (DOI: 10.4337/roke.2021.01.03).

There are voices on the other side, too. A paper in a financial regulation journal disagrees with the critique that post-crisis regulatory reforms misunderstood the nature of banks. It argues that viewing banks as engaging in relatively unbridled money creation is inaccurate. It is true that fractional-reserve banking creates the flow of money, but that flow is tied to banks' funding requirements. The author goes further, arguing that the revised regulatory framework contains several mechanisms that critics of money creation fail to fully appreciate, that the framework goes much further than critics allow, and that once completed, it will resolve many of the concerns they raised. At the same time, the author concedes one point: whether the countercyclical mechanisms in the new regulatory framework are sufficient remains to be seen (DOI: 10.1108/jfrc-06-2016-0046).

Papers discussing the flaws of fractional reserves are even sharper. They write that the recently fashionable idea—that commercial banks create rather than intermediate the money they lend—is not entirely correct. As evidence, they cite a 2014 explanatory article by the Bank of England (the UK's central bank) (DOI: 10.14738/assrj.77.8572). This is not from a top-tier venue, and as we will see later, it is a piece advocating for a specific reform proposal.

Then What Makes Banks Stop

Before the Fed existed, New York banks squared up the debts they owed each other every day.

There is a paper that reconstructs this scene from historical source materials. The traditional view holds that money creation arises spontaneously from individual market transactions. This paper chose the daily clearing practices of the New York Clearing House as the subject to test that view. It is the environment that looks the most like a free market because there was no central bank. Yet what the primary sources showed was the opposite. Daily clearing was a severe challenge for member banks. Handling that challenge required both state (public authority) governance and non-market agreements among the banks (DOI: 10.2139/ssrn.4090961, working paper · law).

The point is that how you settle the money you create comes before whether a bank can create money. The study that puts this distinction most cleanly is one that puts commercial banks, central banks, the interbank market, and the real sector into a single model. The banks in this model execute loans through deposit creation. Therefore, they never face funding risk, meaning a situation where they lack money to fund new loans. What banks face is only refunding risk. This is the risk of failing to settle the net outflow of deposits with reserves (DOI: 10.2139/ssrn.5034464, working paper).

The same paper reports that permanently reducing the assets bought by the central bank produces significant negative effects on financial and real variables. The pathway in the model is the rising cost of money creation in the banking sector, which lacks reserves. It also notes that not just the amount of reserves but their distribution can be a problem—even in a regime where the total volume is ample.

What happens when settlement gets blocked? Today's bank runs are electronic transfers by large depositors. The balances merely move between banks, which does not correspond to depleting scarce reserves. Panic runs do not happen as long as interbank lending runs efficiently. However, when that breaks down, a coordination failure occurs. Banks simply refuse to lend to banks in trouble. That then spirals into a run, followed by falling prices and contagion. That is why prescriptions like deposit insurance or payment suspensions are neither necessary nor sufficient for this crisis. What is needed is the central bank's role as the lender of last resort (DOI: 10.2139/ssrn.676053, working paper).

There is one more thing that binds today's banks. Capital and regulation. One modeling study identified seven situations under the four regulations of Basel III, an international banking regulation. It records that for each situation, there is a multiplier that translates changes in capital into changes in credit supply, and a corresponding multiplier effect on the money supply as well (DOI: 10.2139/ssrn.4093065, working paper).

Here a summary emerges, and this is not the author's sentence, but this article's. The author does not compare the two multipliers, nor claim that one replaces the other. Even so, placing them side by side makes it visible. The textbook multiplier uses reserves as its base, while this model's multiplier uses capital. So this is a place where you could read that the multiplier has not vanished, but moved.

That does not mean reserve requirements are a dead tool either. Lebanon raised its foreign currency deposit reserve requirement much higher than its domestic currency deposit reserve requirement. A peer-reviewed journal paper used this event for a natural experiment, which is an analysis that uses a policy change like an experiment. All banks scrambled to adjust their portfolios and cut lending. Banks with a high reliance on dollar funding and shallow dollar liquid asset buffers took a harder hit. Exposed domestic banks adapted more slowly compared to foreign banks. On the corporate side, technical matching evidence shows that small businesses took a disproportionate hit (DOI: 10.1111/jmcb.12113). The fact that it is a policy event in a single country remains true. Even so, the summary that "reserves no longer bind anything" does not hold up on the strength of this single paper alone.

So What Does the Central Bank Move?

In October 2008, the Fed began paying interest on the reserves that banks keep with it.

Let us clarify a term. Quantitative easing is a policy where the central bank purchases financial assets in large quantities and creates new reserves as payment, depositing them into bank accounts. The numbers that opened this piece are the traces of unconventional policy tools, including quantitative easing. Going the other way by shrinking and pulling back assets is quantitative tightening.

A working paper cuts the period from January 1967 to October 2018 in the United States into two sections. The boundary is the point when interest began to be paid on reserves. First is the period before the interest payment. A one-standard-deviation shock to the monetary base raised output by about 0.31% at its peak. It raised the price level permanently by about 0.29%. The state where the output response was significantly greater than zero lasted for about 55 months. In the subsequent period, both responses to the same shock were weak and statistically imprecise. The share of monetary base fluctuations explained by monetary policy shocks dropped from 86% to 44% at the time of the shock. That difference was absorbed by money demand shocks. The author's wrap-up is not a definitive assertion. It merely writes that these results are consistent with the weakening of the aggregate channel through which the expansion of the monetary base transmits to the real economy (DOI: 10.2139/ssrn.6809784, working paper, single U.S. country).

Did the new reserves then do nothing? Another study in a peer-reviewed journal points in the opposite direction. It estimated causality by using regulatory changes that strongly influenced the distribution of reserves as an instrument. The creation of reserves led to an increase in total lending, and risk-taking increased along with it (DOI: 10.1111/jmcb.12781).

The two results look contradictory, but they measure different things. The former looks at the aggregate channel, while the latter looks at bank-level behavior. There is research connecting the two. It is a working paper using U.S. data from the COVID-19 period. It is a study asking when excess reserves turn into loans. It used the difference in mortgage-backed securities holdings by bank and per capita disaster relief funds by region as the source of variation—in other words, the axis of comparison that would determine the outcome. Quantitative easing increased reserves overall. However, the extent to which that led to credit expansion depended on the magnitude of the demand shock that the region received. Yet what this paper puts forward is not a passive condition, but an amplification channel. In states more exposed to quantitative easing, credit expansion and house price increases following the shock were larger. The conclusion of the authors is that quantitative easing amplified the effects of the pandemic-era fiscal stimulus program (DOI: 10.2139/ssrn.4318709, working paper).

If we go by these COVID-era results, reserves are closer to an amplifier than a trigger. Simply increasing them does not automatically cause loans to go out, but when someone appears who wants to borrow, it magnifies the response. The quantity itself is not powerless either. That earlier model study deals with a quantity rule that adjusts the supply of reserves according to economic conditions. It writes that the benefit this rule gives to society as a whole—that is, its welfare contribution—can reach a magnitude comparable to a policy interest rate rule (DOI: 10.2139/ssrn.5034464, working paper).

What Stays Unsolved — Money Outside the Banks

In September 2008, the Fed failed to block the shock of the crisis.

The answer in a paper analyzing why it failed is not incompetence, but wiring. The safety structure between banks and the Fed back then did not reach out to shadow banking. In the meantime, shadow banking had already become the backbone of the global financial system (DOI: 10.2139/ssrn.3432365, working paper).

Shadow banking refers to institutions that are not banks yet do things similar to banks. It is the work of taking in short-term money and putting it into long-term assets, and in the process, producing claims close to money. A representative example is a market where large brokerages and asset managers exchange day-to-day money backed by bonds. Those claims do not show up in a bank account, but to the side holding them, they are just like cash.

According to the same paper, after the crisis, the Fed built a new structure that opens up liquidity and collateral to shadow banking participants as well. The author writes that it was the first time in central bank history that they opened their reserves to places that are not banks.

Then what does the number called "money supply" count? A study that mapped short-term funds around the world through an accounting framework argues this. Official monetary aggregates, in other words the M0, M1, and M2 indicators that count money, do not include the instruments that asset managers use like money at all. Ultra-short-term transactions backed by bonds are left out in particular. Those aggregates and the financial accounts of the U.S. do not properly reflect the institutional reality of today's financial ecosystem. The authors propose satellite accounts that separately capture the flows of collateral, risk, and Eurodollars (DOI: 10.2139/ssrn.2476415, working paper).

There is also a candidate to shake this picture once more. It is digital money issued directly to citizens by a central bank. The first evidence from a real-world case is India's digital rupee.

This is a report from a working paper that used the phased method of granting eligibility and promotional campaigns as sources of variation. Granting eligibility reduced retail deposits by 2.7%. That drop was concentrated in savings accounts that can be withdrawn at any time, and fixed deposits were not affected. Credit shrank briefly and then recovered. There are also points that went against expectations. What this money pushed out was not cash, but digital payments based on deposits. The authors' welfare calculations show losses in the short term and gains in the long term (DOI: 10.2139/ssrn.7301398, working paper).

An International Monetary Fund working paper models the points where results diverge across countries. In the authors' judgment, places where this money easily drains deposits are developing and emerging economies. These are countries with many low-income households and expensive access to bank accounts. A caveat is also attached. Even if deposits leak out and bank intermediation shrinks, if banks have other funding means, the impact on lending is quantitatively small. Wholesale funds or central bank funds are examples (DOI: 10.5089/9798400259029.001, working paper).

Prescriptions — Just the Authors' Arguments, Not the Conclusion of This Article

There are proposals put forward in the 1930s by Henry Simons, Lauchlin Currie, and Irving Fisher.

Full-reserve banking is a proposal to have banks keep 100 percent of the demand deposits they receive as reserves. This completely decouples the act of creating money from the act of lending it. Demand deposits are deposits that can be withdrawn at any time. This means that 1 million won in a bank account must remain locked up as is, and the money that banks lend out must come exclusively from separately raised funds.

After the crisis of 2007–2008, interest in this proposal revived, and confusion returned along with it. One paper sums it up like this: "100% money" has been confused with abolishing bank intermediation. It has been alternately confused with establishing a currency board system and with more recent narrow banking proposals (both of which are other reform plans not covered in this article) (DOI: 10.3917/redp.325.0835).

What would actually happen if we did that? A study in a peer-reviewed journal examined this. It is a model built so that money balances and flows do not mismatch. It compared when money is created through government spending under full-reserve banking versus when the same spending is done under the current system. The result is anticlimactically bland. Output, employment, and inflation move almost identically. What changes is the consolidated government debt, which is the sum of the government's and the central bank's debts. When money is created under full-reserve banking, this is permanently reduced. Monetary policy is also transmitted effectively, as an increase in central bank reserves translates into an almost equal-sized increase in demand deposits. The same model delivers a different verdict on quantitative easing. Money creation through quantitative easing affects only monetary aggregates and interest rates, but not the real economy (DOI: 10.1093/cje/bey034).

The opposite side is also noted. In all money-creation experiments, banks were able to fully meet loan demand. However, under full-reserve banking, temporary credit crunches can occur. Whether they occur depends on changes in private sector behavior, economic policy, and the safety margins that banks adopt, the author describes.

There are also arguments from the advocating side. The author of the paper discussing the flaws of fractional-reserve banking points out two things. One is that deposit insurance and bailouts backed by taxpayers are a special privilege given only to a single type of lender, namely banks. Compared to other lenders such as peer-to-peer loans or commercial credit, this is said to be a distortion of resource allocation. The other is that depositing and growing money in a bank is also a commercial transaction. Protecting participants from those losses is not the job of taxpayers, they say. Their prescription is to strip away state support. At the same time, they suggest letting people who want their money to be completely safe deposit it with the state (DOI: 10.14738/assrj.77.8572). This is the author's argument.

There are prescriptions in the exact opposite direction as well. A working paper asks right from its title: "Why Do Banks Create Money?". In an intermediated system, to the extent that households provide deposits, banks try to limit leverage (the limit to which they can take on debt) so that they properly monitor loans. However, monitoring costs differ across banks, and those differences may not be visible from the outside. The author points out that, as a result, total lending to bank-dependent firms becomes inefficiently low. The author shows through a model that if banks can create deposits without relying on household funds, that problem is mitigated. With appropriate leverage regulations in place, the gains from increased lending on the side of banks with a high probability of repayment outweigh the losses on the side of banks with less monitoring. They note that attaching risk-sensitive capital requirements can also reduce or eliminate those losses (DOI: 10.2139/ssrn.4095335, working paper).

Changing the Question Changes the Shape of the Answer

There are three points where the cited literature does not significantly disagree.

First, when an individual bank issues a loan, a deposit is created on that bank's ledger, and small banks are no exception. This is supported by a working paper (DOI: 10.2139/ssrn.2120439). However, a paper on fractional reserve flaws notes that this idea is "not entirely correct."

Second, under overflowing reserves, reserve supply is not the primary constraint binding lending, a direction shared by a Federal Reserve working paper (DOI: 10.17016/feds.2010.41) and a single-country analysis of Russia (DOI: 10.32609/0042-8736-2018-11-50-69).

Third, taking that place are settlement capacity, capital and regulation, and the demand from the borrower. This is supported by two modeling studies and one single-country empirical study on Russia.

There are also three points of contention.

First, is the money multiplier a framework to be discarded, or a framework whose scale varies depending on the regime? The argument against discarding the multiplier came from a preprint, while a sophisticated explanation for a low multiplier appeared in a peer-reviewed journal.

Second, do the two pictures truly yield different answers? On one side lie equivalence theorems and regime-dependent analyses. The opposing side (a working paper model) predicts much larger and much faster loan changes and much greater real effects in response to the same shock.

Third, how much does newly created reserves alter bank behavior? A single-country US working paper contains results consistent with a weakened aggregate channel (DOI: 10.2139/ssrn.6809784). Yet causal estimates at the bank level also exist in a peer-reviewed journal showing increases in both lending and risk-taking (DOI: 10.1111/jmcb.12781).

"Where does money come from?" looks like it is asking for a location, but when rephrased into an answerable form, it becomes this: What stops money?

The blackboard arithmetic wrote down reserves as the answer. The materials gathered here place other names. Is there a means to settle? Does capital remain? What does regulation permit? Is there a borrower? The summary that "banks print money out of nothing" is wrong not because of the words "out of nothing." It is because the latter four are entirely omitted. None of those four are fixed. Entities outside the banking system that create things close to money are already outside the monetary aggregates (DOI: 10.2139/ssrn.2476415, working paper). Money issued directly by central banks to citizens has begun to shift the raw material of deposits little by little (DOI: 10.2139/ssrn.7301398, working paper).

The article "The Real Cause of Inflation — Money Supply, Expectations, or Supply Chains?" asked whether the quantity of money pushes up prices. This article's answer to where that quantity is determined is that it is not determined in just one place. The collapses in "Why Financial Crisis Keep Coming Back — Fat Tails, Black Swans, and Systemic Risk" are also variations of the same question. The alternatives in "Does Bitcoin Have Value? — Digital Gold, or a Giant Bubble" are no different.

This was the problem New York banks faced every morning before the Federal Reserve was established: "With what shall we settle the promises we made?" That problem is not over yet. Only the names of the clearinghouses and settlement means have changed.

Source Papers

A mix of peer-reviewed journal articles, central bank and international organization working papers, and pre-peer-review preprints. Ratings are noted in the text as they appear.

• "Money, Reserves, and the Transmission of Monetary Policy: Does the Money Multiplier Exist?" (2010, Federal Reserve Working Paper) — DOI: 10.17016/feds.2010.41 — From reserve balances in the $20 billion range to over $1 trillion. Aggregate and bank-level time-series analyses show that neither the standard multiplier model nor the bank-lending channel operates.
• "Money multiplier in the context of modern views on money creation: theory and facts" (2018) — DOI: 10.32609/0042-8736-2018-11-50-69 — Multiple deposit creation is a feature of underdeveloped financial systems. Back then, reserve supply was the constraint; today, credit demand is. Russia 2005–2017.
• "The Money Multiplier and Other Measures of Financial Sector Performance" (2022, preprint) — DOI: 10.31235/osf.io/zusqa — The multiplier is not useless—it measures the degree of intermediation in the financial system. U.S. intermediation peaked in 1985 (based on a broad monetary aggregate).
• "A Model of QE, Reserve Demand, and the Money Multiplier" (2022) — DOI: 10.1111/jmcb.12975 — Incorporating higher reserve demand and capital constraints explains the low multiplier and deposit growth. However, this does not mean banks passively absorb all new reserves.
• "Banks are Not Intermediaries of Loanable Funds – And Why This Matters" (2015, working paper) — DOI: 10.2139/ssrn.2612050 — Banks create new deposits through loans, and in doing so are primarily constrained by profitability and solvency. Under the same shock, the money creation model predicts much larger and faster changes in lending and much larger real effects.
• "Money Creation: Misconceptions: A Small Bank Cannot Create Money" (2012, working paper) — DOI: 10.2139/ssrn.2120439 — Without excess reserves you cannot lend more, money creation starts from new deposits, small banks cannot create money—three misconceptions and a rebuttal to the last one.
• "The money multiplier and competing theories of money creation: empirical validation for Russia" (2025) — DOI: 10.1093/cje/beae036 — Monthly data for Russia from 2005 to 2019. Credit money theory receives the strongest support, and bank lending is primarily tied to credit demand (similar to the U.S.).
• "The Post-Keynesian endogenous-money supply: evidence from Poland" (2019) — DOI: 10.4337/roke.2019.03.09 — Poland 2001–2016. Loans lead deposits, broad money, and base money, whereas the broad money multiplier does not lead broad money.
• "The role of fractional-reserve banking in amplifying credit booms: evidence from panel data" (2017, working paper) — DOI: 10.33119/kaewps2017024 — Panel of 20 countries. Credit growth is independent of prior savings changes, at least in the short run. The share of time and savings deposits decreases during credit booms. Inferences regarding the natural rate of interest are the author's own.
• "Loanable funds versus money creation in banking: a benchmark result" (2022) — DOI: 10.1007/s00712-021-00747-7 — In the absence of uncertainty, the two processes yield the same allocation. Under total risk, complete contracts, and complete markets, there is limited equivalence.
• "Money creation in the modern economy: an appraisal" (2021) — DOI: 10.4337/roke.2021.01.03 — As fears of liquidity squeezes mount, the money supply process leans toward the logic of exogeneity. Right now, we are in a floor system with excess reserve supply, so that fear is absent.
• "Banks, private money creation, and regulatory reform" (2018) — DOI: 10.1108/jfrc-06-2016-0046 — The view that banks engage in runaway money creation is inaccurate, and flows are tied to funding requirements. The revised regulatory framework has gone further than critics concede and, once completed, will resolve many of the concerns. However, the adequacy of countercyclical mechanisms remains to be seen.
• "Two flaws in fractional reserve banking." (2020) — DOI: 10.14738/assrj.77.8572 — The modern idea that banks create the money they lend is not entirely correct. The author's prescription is the withdrawal of state support and the permission of state deposits.
• "Money Creation and Bank Clearing" (2022, working paper, law) — DOI: 10.2139/ssrn.4090961 — Research on historical records of the New York Clearing House prior to the Federal Reserve. Daily clearing posed severe challenges, requiring state (public authority) governance and non-market agreements.
• "Quantitative Easing and Quantitative Tightening: The Money Channel" (2024, working paper) — DOI: 10.2139/ssrn.5034464 — Banks face refinancing risk rather than funding risk. The quantity and distribution of reserves matter even in an ample-reserve regime, and the welfare contribution of countercyclical quantity rules can match that of policy-rate rules.
• "Money and Modern Bank Runs" (2005, working paper) — DOI: 10.2139/ssrn.676053 — Modern runs involve interbank balance transfers and do not amount to the depletion of scarce reserves. If interbank lending collapses, runs occur due to coordination failure. Deposit insurance is neither a necessary nor a sufficient condition.
• "How does bank equity affect credit creation? Multiplier effects under Basel III regulations" (2022, working paper) — DOI: 10.2139/ssrn.4093065 — Seven scenarios under four Basel III regulations. In each situation, there is a multiplier translating changes in equity into changes in credit supply, along with a corresponding multiplier effect on the money supply.
• "Reason for Reserve? Reserve Requirements and Credit" (2014) — DOI: 10.1111/jmcb.12113 — An increase in the reserve requirement on foreign currency deposits in Lebanon. All banks reduced lending, hitting dollar-funding-dependent banks and (with evidence of technological matching) small firms disproportionately.
• "Interest on Reserves and Monetary Policy Transmission" (2026, working paper) — DOI: 10.2139/ssrn.6809784 — U.S. 1967–2018. Prior to the payment of interest on reserves, monetary base shocks resulted in an output peak of about 0.31%, a permanent price level increase of about 0.29%, and a significant span of about 55 months. The subsequent period is weak and imprecise. Variance decomposition drops from 86% to 44%.
• "Quantitative Easing and Bank Risk Taking: Evidence from Lending" (2021) — DOI: 10.1111/jmcb.12781 — Causal inference using regulatory changes as an instrument. Reserve creation led to an increase in total lending and risk-taking.
• "Quantitative Easing, Bank Lending, and Aggregate Fluctuations" (2023, working paper) — DOI: 10.2139/ssrn.4318709 — COVID-19-era U.S. The degree to which excess reserves boost lending capacity and lead to credit expansion depends on the magnitude of the demand shock. The author's conclusion is that quantitative easing amplified the effects of pandemic fiscal stimulus.
• "The Politics of Shadow Money: Security Structures, Money Creation and Unconventional Central Banking" (2019, working paper) — DOI: 10.2139/ssrn.3432365 — The Federal Reserve failed to stem the shock in September 2008 because its safety structures did not extend to shadow banking. Three tiers: central bank money, quasi-money, and shadow money.
• "Shadow Banking: The Money View" (2014, working paper) — DOI: 10.2139/ssrn.2476415 — Monetary aggregates entirely exclude instruments used as money by asset managers (especially collateralized ultra-short-term transactions), and neither aggregates nor financial accounts adequately reflect institutional realities. Proposed subsidiary accounts.
• "Central Bank Digital Currency (CBDC), Bank Deposits, and Digital Payments: First Evidence from a Live Retail CBDC" (2026, working paper) — DOI: 10.2139/ssrn.7301398 — India's digital rupee. A 2.7% decline in retail deposits (concentrated in savings accounts, with time deposits unchanged), a temporary decline followed by recovery in credit, and substitution toward deposit-based digital payments rather than cash. Short-term welfare losses and long-term gains.
• "Central Bank Digital Currency and Bank Disintermediation in a Portfolio Choice Model" (2023, IMF Working Paper) — DOI: 10.5089/9798400259029.001 — Conditions under which deposit flight dominates, with the caveat that if alternative funding sources exist, the impact on lending remains quantitatively small.
• "Divorcing money creation from bank loans: Revisiting the '100% money' proposal of the 1930s" (2022) — DOI: 10.3917/redp.325.0835 — "100% money" is neither the elimination of bank intermediation, a currency board, nor narrow banking. A conceptual cleanup returning to Simons, Currie, and Fisher.
• "Money creation under full-reserve banking: a stock–flow consistent model" (2018) — DOI: 10.1093/cje/bey034 — When money is created through government spending under full-reserve banking, output, employment, and inflation move almost identically, and consolidated government debt permanently declines. Money creation via quantitative easing affects only monetary aggregates and interest rates. However, temporary credit crunches may occur.
• "Why Bank Money Creation?" (2022, working paper) — DOI: 10.2139/ssrn.4095335 — Under an intermediated regime, total lending to bank-dependent firms is inefficiently low. Bank money creation mitigates this, and leverage regulations and risk-sensitive capital requirements can reduce or eliminate losses.