Sneaking up on reality after being mugged by it: Neoclassical attempts to understand bank created money
 In 2014, the Bank of England published a seminal paper Money creation in the modern economy confirming the argument of heterodox economists that bank lending creates money, and rejecting the mainstream economics argument that banks are simply "intermediaries" between savers and borrowers. Since then, a few Neoclassical economists have attempted to understand what impact that bank creation of money has on conventional macroeconomics.

David Andolfatto from the St Louis Federal Reserve is the latest to do so, in a paper entitled "Reconciling Orthodox and Heterodox Views on Money and Banking". In it he cites the exchange Krugman and I had before the Bank of England paper came out, and the paper by me that Krugman criticized, "Instability in Financial Markets: Sources and Remedies", and he asked me and several others for feedback. He's a good guy, who does engage with non-orthodox thinkers like me, so I said I'd take a look.

I don't have time for a detailed response, but I'll note four things, and then link to my current models of Loanable Funds versus Bank Originated Money and Debt.

Firstly, I can't help but write this with a profound sense of ennui. I'm glad to see mainstream economists finally attempting to grapple with the roles of money and credit in macroeconomics, but they're doing it long after non-mainstream economists started to do so, and with far clumsier tools for the task.

I developed my first complex systems model of the role of credit in macroeconomics in 1992 (which was published by the Journal of Post Keynesian Economics in 1995). Wynne Godley & Francis Cripps developed the stock-flow consistent modelling technique before then, and in 1999, Wynne and Randall Wray used it to predict that the "Goldilocks Economy of the 1990s would come to a catastrophic end, largely because it was reliant on a level of credit-financed demand by households that could not be sustained. By 2009, I'd worked out how to build monetary models of the economy, and was doing so initially in the engineering mathematics program Mathcad.

By 2012, I'd started work on Minsky, which uses a double-entry bookkeeping I call a Godley Table to develop models of banking and monetary dynamics in an overall system dynamics framework.

Now, after the Bank of England paper, Neoclassical economists are trying to work out the importance of credit in macroeconomics, using their conventional tools. I'm glad they're doing it, but I wish they'd also question whether their tools played any role in them not anticipating the crisis of 2007. Maybe different tools be better, and maybe the results they get be a product of the tools themselves, and not the real-world phenomena they're attempting to understand using them?

Secondly, David didn't cite Michael Kumhof's work—probably because he wasn't aware of it (I've let him know). Michael Kumhof is a Senior Research Advisor to the Bank of England, and previously he was Deputy Division Chief of the Economic Modelling Division of the IMF.

Michael is a first-rate DSGE model builder, who also knew that banks create money long before that Bank of England paper came out, because he was a banker before he became an academic, and he understood what he did then (this is not something that can be said of many bankers). Michael also finds, in contrast to David, that bank creation of money results in a very different macroeconomy to one in which banks are just intermediaries. Michael's paper "Banks are not intermediaries of loanable funds - and why this matters" is the only Neoclassical paper to date to argue that the fact that banks create money does indeed matter in macroeconomics.

In the context of this debate, Michael's work is more significant than mine, because his work shows, using a Neoclassical methodology, that bank-created-money matters a great deal. So David's draft paper, which shows that it doesn't matter all that much, may say more about the techniques David employed (and the assumptions he made), rather than the issue itself.

Thirdly, while David's characterisation of the heterodox view is pretty accurate, there is one error when he states that "Banks do not in fact lend reserves–they lend their deposit liabilities (which are incidentally made redeemable for cash)". The first half of that sentence is correct; the latter is a statement of Loanable Funds, not the Bank created money approach. A better statement would be that:

"Banks do not in fact lend reserves–their lending creates their deposit liabilities (which are incidentally made redeemable for cash)".

Fourthly, my modelling of money has moved on since the paper that David cites, which pre-dates my debate with Krugman ( you can find a good overview of that debate, with links to the papers, on the Unlearning Economics blog at The Keen/Krugman Debate: A Summary). In particular, after that rather acrimonious exchange, I realised that the best way to show why the bank creation of money matters was to build a model of Loanable Funds which can be quickly converted into a model of Bank Created Money. In the former, not amazingly, bank lending has little macroeconomic impact. In the latter, its impact is huge.

My results are consistent with Kumhof's using a DSGE framework, but contrast with both David's, and also Faure and Gersbach's "Loanable funds vs money creation in banking: A benchmark result" where they both conclude that Loanable Funds versus Bank Money Creation is no big deal.

Not coincidentally, both David and Faure and Gersbach use the very peculiar (in both senses of the word) "Overlapping Generations" (OLG) approach. They also make assumptions that pretty much guarantee their conclusions, as David admits early on in his paper:

In the model, banks and financial markets are competing mechanisms for allocating credit. Banks are "special" only to the extent they are better than markets at funding investment. This specialness is not (in the model) logically rooted in their ability to create money…
Because bank-finance coexists with other forms of financing, it is not entirely clear why monetary/fiscal policies should make special consideration of the banking sector per se. The fact that commercial banks create liabilities that are used widely in payments is not likely a first-order concern. We saw in the financial crisis of 2008, in particular, how highly-leveraged shadow banks were perfectly capable of disrupting the financial market without the capacity to create money. A case can be made that the ingredient missing from conventional macroeconomic models was leverage in general, not banking in particular. This is the sense in which the analysis below, while sympathetic to the heterodox perspective, largely ends up supporting the mainstream view.

Of course, the specialness of banks arises precisely because they can create money, and this credit-financed demand adds to demand from the circulation of existing money. The logic here is quite simple: lending by one non-bank to another redistributes existing money, but doesn't create any more; it facilitates spending by the borrower, but reduces the spending potential of the lender.

Lending by a bank, on the other hand, creates new money; it facilitates spending by the borrower, but doesn't reduce the spending potential of any other agent. It's the explanation of why credit matters so much in explaining the actual economic phenomena of the last thirty years. If OLG modelling concludes that there's no significant difference between a world in which banks create both debt and credit, and one when it does, then it's the form of analysis that is wrong, and not the data.

I illustrate the macroeconomic importance of banks creating money with one Minsky model where banks simply intermediate between a consumer sector that lends and an investment sector that borrows, and another where the bank lends directly to the investment sector.

The contrast with the model where banks originate money and debt is huge. Changes in bank lending change the amount of money in the economy, and change the level of aggregate demand as well.

The models are linked at the bottom of this post, and anyone who wants to check them out can download Minsky from here and run them, check the equations, etc.