Banks are not intermediaries of loanable funds – and why this matters

Zoltan Jakab & Michael Kumhof

Problems in the banking sector played a critical role in triggering and prolonging the Great Recession. Unfortunately, standard macroeconomic models were initially not ready to provide much support in thinking about the role of banks. This has now changed, with many new papers that study the interaction of banks with the macroeconomy. However, as emphasized by Adrian, Colla and Shin (2013), there are many unresolved issues. In our new paper “Banks Are Not Intermediaries of Loanable Funds – And Why This Matters” (Jakab and Kumhof (2015)), we argue that many of them can be traced to the fact that virtually all of the newly developed models are based on the intermediation of loanable funds (ILF) theory of banking.

In the simple ILF model, bank loans represent the intermediation of real savings, or loanable funds, between non-bank savers and non-bank borrowers. Lending starts with banks collecting deposits of real resources from one agent, and ends with the lending of those resources to another agent. In the real world, however, banks never intermediate real loanable funds, an activity that, correctly understood, can only amount to barter.

Rather, the key function of banks is the provision of financing, meaning the creation of new monetary purchasing power through loans, for a single agent that is both borrower and depositor. Specifically, whenever a bank makes a new loan to a non-bank customer X, it creates a new loan entry in the name of customer X on the asset side of its balance sheet, and it simultaneously creates a new and equal-sized deposit entry, also in the name of customer X, on the liability side of its balance sheet. The bank therefore creates its own funding, deposits, through lending. It does so through a pure bookkeeping transaction that involves no real resources, and that acquires its economic significance through the fact that bank deposits are any modern economy’s generally accepted medium of exchange. This understanding of the function of banks, which we will refer to as the financing and money creation (FMC) model, has been repeatedly described in publications of the world’s leading central banks—see McLeay, Radia and Thomas (2014a,b) for an excellent summary. What has been challenging is the incorporation of these insights into macroeconomic models.

Our paper therefore builds examples of simple DSGE models with FMC banks, and then contrasts the predictions of these models with the predictions of otherwise identical models with ILF banks. Figure 1, in simple schematic form, contrasts the barter function of banks under the ILF model with their monetary function in the FMC model.

Figure 1: The Role of Banks in LF and FMC Models

Figure 1: The Role of Banks in LF and FMC Models

This shows the simplest possible case of an FMC model, where banks interact with a single representative household. More elaborate setups with multiple agents are possible, and one of them is presented in the paper.

The main reason for using FMC models is therefore simply that they correctly represent the role of banks in the economy, while the banks of the ILF model have no identifiable counterpart in the real world. But in addition, the empirical predictions of the FMC model are qualitatively much more in line with the data than those of the ILF model. The model simulations in our paper show that, compared to ILF models, and following identical shocks, FMC models predict changes in bank lending that are far larger, happen much faster, and have much larger effects on the real economy. Compared to ILF models, FMC models also predict procyclical or acyclical rather than countercyclical bank leverage, and an important role for quantity rationing of credit, rather than an almost exclusive reliance on price rationing, in response to contractionary shocks.

The fundamental reason for these differences is that savings in the ILF model of banking need to be accumulated through a process of either producing additional resources or foregoing consumption of existing resources, a physical process that by its very nature is gradual  and slow. On the other hand, FMC banks that create purchasing power can technically do so instantaneously, because the process does not involve physical resources, but rather the creation of money through the simultaneous expansion of both sides of banks’ balance sheets. While money is essential to facilitating purchases and sales of real resources outside the banking system, it is not itself a physical resource, and can be created at near zero cost.

The fact that banks technically face no limits to instantaneously increasing the stocks of loans and deposits does not, of course, mean that they do not face other limits to doing so. But the most important limit, especially during the boom periods of financial cycles when all banks simultaneously decide to lend more, is their own assessment of the implications of new lending for their profitability and solvency. By contrast, and contrary to the deposit multiplier view of banking, the availability of central bank reserves does not constitute a limit to lending and deposit creation. This, again, has been repeatedly stated in publications of the world’s leading central banks.

Another potential limit is that the households and firms that receive the newly created money after it is spent may wish to use it to repay an outstanding bank loan, thereby quickly extinguishing the money and the loan. This point goes back to Tobin (1963).  This, however, is a surprising argument, because it makes the implicit assumption, which would never be made for other macroeconomic shocks, that every shock to one set of agents must be accompanied by an equal and offsetting shock to another set of agents. If this strange assumption is dropped, an increase in loan supply to one group of agents leads to higher aggregate money balances, which in turn stimulate additional economic activity, by encouraging spending through reductions in transactions costs. This in turn increases the extent to which all households want to keep the additional money to carry out spending transactions, rather than to repay existing loans. This phenomenon is prominent in our simulations.

As an illustration, Figure 2 shows impulse responses for a shock whereby, in a single quarter, the standard deviation of borrower riskiness increases by 25 percent. This is the same shock that is prominent in the work of Christiano et al. (2014). Banks’ profitability immediately following this shock is significantly worse at their existing balance sheet and pricing structure. They therefore respond through a combination of higher lending spreads and lower lending volumes. However, ILF banks and FMC banks choose very different combinations.

Figure 2: Credit Crash due to Higher Borrower Riskiness


Figure 2: Credit Crash due to Higher Borrower  Riskiness

Figure 2: Credit Crash due to Higher  Borrower Riskiness

ILF banks cannot quickly change their lending volumes. Because deposits are savings, and savings are a predetermined variable, they can only decline gradually over time, mainly by households increasing their consumption or reducing their labour supply. Banks therefore keep lending to borrowers that have become much riskier, and to compensate for this they increase their lending spread, by over 400 basis points on impact.

FMC banks can instantaneously and massively change their lending volume, because lending and deposits are jump variables. We observe a large and discrete drop in the size of banks’ balance sheets, of around 8 percent on impact in a single quarter (with almost no initial change in the ILF model), as deposits and loans shrink simultaneously. Because, ceteris paribus, this cutback in lending reduces borrowers’ loan-to-value ratios and therefore the riskiness of the remaining loans, banks only increase their lending spread by around 200 basis points on impact. A large part of their response, consistent with the data for many developed economies, is therefore in the form of quantity rationing rather than changes in spreads. This is also evident in the behaviour of bank leverage. In the ILF model leverage increases on impact because immediate net worth losses dominate the gradual decrease in loans. In the FMC model leverage remains constant (and for smaller shocks it drops significantly), because the rapid decrease in lending matches the change in net worth. In other words, in the FMC model bank leverage is acyclical (or procyclical), again as in the data, while in the ILF model it is countercyclical.

As for the effects on the real economy, the contraction in GDP in the FMC model is more than twice as large as in the ILF model, as investment drops more strongly than in the ILF model, and consumption decreases, while it increases in the ILF model.

To summarize, banks are not intermediaries of real loanable funds, they do not collect new deposits from non-bank savers. Instead they provide financing, they create new deposits for their borrowers. This involves the expansion or contraction of gross bookkeeping positions on bank balance sheets, rather than the channelling of real resources through banks. Replacing intermediation of loanable funds models with financing and money creation models is therefore necessary simply in order to correctly represent the role of banks in the macroeconomy. But it also addresses several of the empirical problems of existing banking models.

Authors: Zoltan Jakab works at the International Monetary Fund and Michael Kumhof works in the Bank’s Research Hub.

Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.

If you want to get in touch, please email us at