Ramanan Lyer on the Keen Change in AD Function – What Lies Between Identities

A fecund debate is taking place at Mike Norman Economics, 119 comments in less than two days, about a post on Ramanan’s blog criticising the new and extended  definition and justification of Steve Keen’s formula for aggregate demand i.e. AD=Income + delta Debt.  The number of comments indicates that either something important has been disocvovered or a major fallacy is out there.


In a recent talk (video here), he claims that income is not equal to expenditure due to debt creation. Keen also claims in the video that Schumpeter and Minsky claimed that is the case.

Keen is right about an individual sector but not an economy as a whole when it is closed.

In other words the accusation is of a fallacy of composition error.

There is no error.  I will outline the case Ramanen makes, some of the key exchanges in this thread and what I hope is a firm refutation of the claim of a fallacy.

This is not the same time we have tackled this conceptual issue.  Just as Ramanen argues that by accounting definition income=expenditure at every point in time we have argued that savings = investment (for loan generated investment) as an entity is only true ex poste over the period of a good loan. An article Steve keen tweeted support over.    It is the same point re accounting, but before getting into too much detail on this point, and I hope in not too mathematical.


Keen forgets that expenditure creates income

He applies the Godley-Cripps approach to budget constraints

 since the possibility of borrowing is included as a source of funds for spending, our formal representation of the budget constraint for any individual or institution including the government, is


any excess of income over spending must equal the acquisition of financial assets less the acquisition of debts. As this is true for all individuals it must also be true for the economy as a whole.

But since total national income equals total national expenditure (i.e., Y ≡ E) it must follow for the economy as a whole the change in financial assets must be equal to the aggregate change in debt, i.e.,


..Keen forgets that consumption is income for firms and his accounting has black holes. The whole thing can be done right by creating a Transactions Flow Matrix, so that one is sure that nothing is missed out….

The right definition of expenditure does not include purchases of financial assets. For Keen if  a household purchases financial assets, it will be counted as “expenditure”. ..Keen’s definition of expenditure itself is different to begin with from standard ones and obviously he gets the paradoxical claim that Income ≠ Expenditure!

But in a closed economy with no government sector modelled Y=C+I.  Ramamam seems to be accusing Keen of confusing consumption (which does not include purchase of financial assets) with investment (which does).  In fact Keen extends the standard Y=C+I identity to Y=C+I+A where A = Assets, of all kinds. I take this as splitting the I element into net income from investment (which contra to Keen is the total of all factor returns from investment in retained assets not simply capital goods – easy danger of a capital theory slip up here) and net income from asset sales and purchases.  This is a useful distinction and allows contrast between the two income streams – and the sources of investment – undistributed profits from retained assets  or increase in net asset sales.

The accusation that ‘Keen forgets that consumption is income for firms‘ No we are talking about the difference between an ex-ante income before a discontinuous injection of debt and then the income (of the economy as a whole) after that injection of debt.

As Keen put it himself in a tweet.

They’re confusing ex-ante & ex-post. Simplest def’n: Expenditure=Income before debt injection plus debt injection.

And Neil Wilson perceptively

Express it like a computer variable assignment: Income = Income + Debt Injection.

Neil had this useful comment on the Mike Norman Thread

Accounting measures income and expenditure at the end of the day using discrete function snapshots.

Steve is measuring income and desires to spend in excess of income at the beginning of the day and following those through the day using a continuous change function.

So income at the beginning of the day + desire to spend realised by borrowing = expenditure at end of day = income at end of day = income at the beginning of the following day.

Essentially you walk into a shop with a credit card and $100 in your wallet. You walk out with a pair of trainers, $100 in your pocket and a credit card debt of $100.

You can still spend $100 during the rest of the day. Your purchasing power hasn’t been diminished by the debt creation and neither has anybody else. Yet a real transaction happened, expenditure generated and income produced.

Between the beginning and the end of the day a number of things will have happened.  The economy will have grown (or shrunk) and the money supply will have grown (through debt) or shrunk (through deleveraging).

If we start with an economy in position T0 at the beginning of the day and T1 at the end of the day then if we have a income Yo at T0 and the economy grows by X then at T1 Y1=Y0+x.  Looking back over the day I1=Y1, but both are greater than I0=Y1.  Because we are used to dealing with discrete time and fallaciously treat accounting identities as true at every point in time, rather than over a period looking backwards in time this is easily misunderstood and to be fair Keen needs to be careful with his terminology here.  Keen’s dynamic approach is based on hunting for the source of the difference, what gets added.  As Marx put it M-C-M’.  So where does the ‘ come from? A pure identities approach will not tell you that – it is platonic and backward looking – the source of the change needs to be identified   Keen has focussed rightly on debt , but it is I hope just the beginning of a wider investigation as the ‘mystery’ of profits also needs explanations of causes in the ability to extend debt, the source of interest, the speed of transactions and the source of successful investment (profit) in order to fully close the monetary circuit.  Each a subject tackled, or about to be, on this blog.

Two of the great rules of the stock-flow consistent approach are that every expenditure is somebody else’s income, and that every asset is matched by an accompanying liability.  But these are necessary but not sufficient axioms for modelling an economy.  Two more must be added, and Ramanen’s mistake is only applying the previous axioms.  These axioms being that at any point in time money can only be spent on one transaction at a time – however quickly it is spent; and that for the economy as a whole the monetary circuit only ever flows in one direction, it can never flow in reverse.

These together mean that if you have a series of t0-t1 accounting identities it does not mean you can reverse them and assume that the t1-to-t0 identities would be the same.  In that time we have seen capital revaluation and destruction, commodities have been created, transformed and consumed in an irreversible process.

How does this apply.  Lets take the specific claims that the ΔFA ≡  ΔD identity is ignored and that Keen has no transactions flow matrix.  The latter accusation is slightly unfair, it is there in his Minsky computer model it just is not generated automatically yet in the Godley and Lavoie format Ramanen would like to see.  the change in financial assets is clearly there in Keens formula’s however these do not distinguish between physical and financial assets.  This allows Ramamen to make the same claim found in (almost) every macro test. Debt and asset values created by debt cancel so debt doesn’t matter.

The flaw in this reasoning is that this equalisation is based, as we have set out on this blog before, on the capitalisation of future anticipated income streams on the (risk and collateral adjusted) assumption that the debt will be repaid in full with interest.  At any point in time the income stream from debts and income streams from repayments of debts will not be in balance.  This is a key insight, not unique to Keen, I have seen various versions of it in Torrens, Holden, Marx, Minsky and Hawtry, it has been key to different theories of the credit cycle for over 150 years, but Keen is the first to express the idea in formal mathematical terms and attempt to model it.

As Hawtry put it.

[producers of intermediate commodities] do not wait for the retail sales, but are paid at the moment of sale with money created by the banks, and then when the final sale to the consumers takes place, the money advanced by the banks has to be paid off. That part of the proceeds of sale is simply destroyed. For just as a bank advance creates money, so the repayment of an advance extinguishes money…

If we suppose the production and sale of [goods] in all the successive stages, to form an isolated operation, then at the beginning there will be an excess of purchasing power and no goods to buy, and at the end an excess of goods and a shortage of purchasing power…

In order that the goods produced in any interval of time may be sold, what is needed is that the incomes occurring in that same interval of time should be sufficient to buy the goods at remunerative prices…

Incomes are the source of demand. But it cannot be assumed that the amount of demand in any interval of time must be equal to the aggregate of incomes. The expenditure of the individual is not exactly equal to his income; he may leave part of his receipts unspent in his cash balance, or he may draw on his cash balance (or overdraw) for expenditure in excess of his receipts

the payments made by the group of traders being in respect of services rendered toward production and other economic activities, will be increased or diminished according as the traders accelerate or retard production. If they accelerate production, they must pay out more in respect of the greater productive activity. There will result an excess of the traders’ disbursements over their receipts, an excess which may be described as a “release of cash”.
The cash released goes to pay additional incomes, and then reappears as additional demand. The additional
demand evokes a still greater productive activity and a further release of cash.
When the traders retard production, there occurs an excess of their receipts over the disbursements or an
absorption of cash. There is a shrinkage of incomes, of demand, of sales, and then a still greater shrinkage of

The release and absorption of cash play an important part in the regulation of credit. What is commonly called an expansion of credit is really a device for inducing a release of cash, while a contraction of credit is a device for inducing an absorption of cash. The release of cash may be effected either with money drawn from existing balances or with money lent by the banks. Similarly an absorption of cash may mean either the accumulation of idle money or the repayment of bank advances. The majority of traders avoid holding idle balances, and borrow just so much from their bankers as their varying needs for working capital require from time to time.

Although Hawtry was at times held back by a Knightian continuous production fallacy where the law of large numbers cancels everything out you will see from the extended quote that he sensed it was the discontinues driving and driven by the change in debt that drove the credit cycle.

In comments Ramenen introduced new objections.

If I (or a lot of people) take on debt to purchase financial assets, it doesn’t add to spending power to the extent Keen supposes. The effect is indirect if people buy assets and if creates holding gains due to price rise in financial assets. But the effect is limited to capital gains (and the propensity to consume out of capital gains) and not to the debt actually incurred.

For Keen it doesn’t matter if the liabilities incurred is for purchases of nonfinancial assets/ consumption or for purchases of financial securities. Both add to the same purchasing power.

In the extreme case (to illustrate) think of my taking a loan and doing nothing afterwords. But for Keen it adds to aggregate demand (now effective demand).

Debt not spent and kept in idle balances does subtract from aggregate demand.  Keen does not cover it but its effect is small.  We have talked about in on this blog before as ‘Davenport’s k factor’ the proportion of granted credit remaining in balances   It is important as we demonstrated through modelling it effects the extent to which idle balances are transmitted bank to bank and so expand net lending power.  But it is small because businesses tend to minimise idle balances especially when they accrue interest and most consumer debt tends to be spent very quickly.

But to state ‘If I (or a lot of people) take on debt to purchase financial assets, it doesn’t add to spending power to the extent Keen supposes.’ is just plain wrong.  if you borrow to buy a house the debt immediately increases the spending power of the developer and, in being used to secure development finance, the incomes of the brickie, chippies, sparks etc. who constructed it.

 In [keens]  models, loans are taken to purchase financial assets. I don’t know how that “adds to demand”.

To which Kis Rosberg perfectly replies

The sellers of these assets buy goods and services.

Osborne Announces New Right to Attack Planners

George Osborne is to clarify government policy to allow ministers and homeowners to attack planners.

Previously Ministers could only use ‘disproportionate bullshit’ to attack planners.

‘I want to make the approach absolutely clear’

Homeowners defending monster extensions and ministers seeking diversionary targets will now be able to use ‘grossly disproportionate bullshit’.