Thursday, October 11, 2012

Income, expenditure, and endogenous money

My supervisor Ray Streater used to quote an example of a physicist's version of proof by reductio ad absurdum that goes more or less like this: "Assume that asymptotic completeness doesn't hold in quantum field theory. What an absurd!"

I was reminded of that as I read this take down of my paper with Steve Keen, where we claim that expenditure is income plus change in debt. In an attempt to show that we are wrong, Ramanan shows that our model violates the "savings equals investment" identity. What an absurd!

Hmm, actually, not only this is not an "inconsistency" of the model (we never claim that this was true, or rely on it to show anything else in the paper), but it is rather essential: in our model, investment is equal to savings plus change in debt. This is the essence of equation (1.5) in the paper, and we were always acutely aware of it.

So much for the rather bombastic conclusion that we must be wrong because we violate the sacrosantity of "savings = investment" -- this is a feature of the model, not a bug!

Having said that, it is nevertheless a feature that ought to be defended, together with the other criticisms  raised in both the take down mentioned above and its predecessor.

As it turns out, without engaging in a point-by-pint reply (which would be a very boring read for anyone not called Grasselli, Keen, or Ramanan), it suffices to say that all the criticisms can be defended in two words: endogenous money!


  1. I think Ramanan is right. Change of debt changes your assets (and liabilities) - expands the balance sheet. Nothing says that they money will be spent, it may be borrowed for liquidity purposes. I think you confuse spending with how it is financed. Yes, it is mostly financed by the new debt, hence correlations spending-credit_impulse are very high. But credit impulse is not spending, it finances the spending. In a closed economy w/o govt Y=C+I=C+S, that is the definition of S. Yes, endogenous money means that new C and I can be financed not from the stock of existing financial assets, but from new debt. But this is not an accounting identity but a model.

    1. Try looking forwards through the model rather than backwards.

      Accounting is a historic reporting system and is designed to look backwards.

      We accuse the neo-classicals of sticking rigidly to one viewpoint and yet many on the PK side suffer from the same problem!

      There is always more than one viewpoint. Changing the viewpoint on a problem yields new insights.

    2. Neil,

      Viewpoints and models are fine.

      I can have a model of how income is driven by prior expenditure and debt:

      Income(t+1)=Expenditure(t)+change in debt(t)

      That is not accounting (which is instantaneous), that is a model of how present variables (t) will translate into future ones (t+1). This has to be stated very clear.

      When we are talking about accounting then we are talking about the same instance of t, there is no model involved, just counting:


      which is an accounting identity, not a model and has to be true in any sensible model.

      Using that we have:

      Income(t+1)=Income(t)+change in debt(t)

      or in a continuous notation:


      units match: Y=[$/sec], Debt=[$].

      That is a very sensible model and a good point of view. This has to be complimented by an accounting identity:

      But if someone writes:

      then it is wrong. You cannot have an economy in which in a given time period, however long or short income is unequal to expenditure (it is true in every tiniest transaction: cash changes hands, one person spends, the other earns income this same instant), it is impossible with the present definitions of "income" and "expenditure".

    3. "Change of debt changes your assets (and liabilities) - expands the balance sheet."

      Don't understand what you are implying here when you say it expands th ebalance sheet.

      Net worth is unchanged.

      Net assets is unchanged.

      It does expand the balance sheet in that there are two additional entries, but I don't think that's what you meant.

      And if you hold the loan for a month or so without spending it it contracts the balance sheet.

    4. Paul,

      This is exactly what i meant: more assets and liabilities, no net wort change. And importantly no spending. If you say that a loan automatically translates into spending then this is a model (which is fine), but it is not accounting that has to check no matter what (simply because *every transaction* generates exactly as much income as the amount that was spent).

  2. income = expenditure financed by existing debt + expenditure financed by new debt?

    If anyone knows his endogenous money, it's Ramanan. His quibble is with your ad hoc mangling of an identity not with the observation that an increase in debt is correlated with an increase in spending. The identity relies on the exact definition of the terms that it is comprised of, so a redefinition of the identity questions the terms themselves and the whole edifice of standard accounting with it. And although anyone is welcome to reinvent accounting, I doubt it's in any way helpful to the point you're trying to make. Standard accounting is fully compatible with endogenous money, so the accountants tell me.

  3. Neil, I think most people understand what's meant but don't agree with the way it's presented. Accounting is a precise business, predicting is not. You can account for things that have happened and predict by extrapolating from there. You cannot account for things that haven't happened nor predict away things that might happen (e.g. specualating, i.e. not spending, with new debt).

  4. I think I see what's happening here…correct me if I'm wrong…

    Traditional economics accounting looks at GDP from the perspective of the agent RECEIVING the spending, and spending = income.

    It looks like you are looking at income from the perspective of the SPENDER, in which case debt adds to one's ability to spend, and thus has an impact on th eagent receiving it. You aren't trying to redefine GDP, just observe how flows make it through the system.

    1. I disagree. We look both from the perspective of spender and earner at the same time

      Every transaction looks the same: A gives X of money to B. Income (that B earns) =X. Spending (that A does) =X.

      Spending= income.

      In every transaction. Irrespective of where A's money came from (existing assets or a new loan).

  5. Endogenous Money - I've probably have heard of it. :-)

    The important point of my post was to pinpoint "black hole" in the accounting in the most basic model and which carries over in your other models.

    You could get away by saying S is not equal to I is a feature of your model but this error will reflect elsewhere. You have capitalists increasing their debt but there you don't identify the corresponding asset. If the funds were borrowed from a bank, then the bank has an additional asset and a corresponding liability and the latter should reflect as increase in money deposits of workers. But this is contradictory to your assumption that workers do not save.

    1. Thanks for the comment Ramanan, but did you read my most recent (and more detailed) post, where I explain a bit better what I meant by the "two words" remark?

      About your last comment, an increase in debt for the capitalists is reflected in a corresponding increase in assets for banks (in the form of loans), which do not need to correspond to money deposit for workers at all, but rather as capitalists deposits (so assets and liabilities match for the bank). The capitalist then use the new debt to buy more capital (that's what we mean by investment), presumably from other capitalists, so this stays squarely in the deposit accounts of capitalists themselves, not workers. No inconsistencies that I can see.

    2. Let us do some accounting here Matheus to show you still have a black hole.

      Some of us have a good grasp of national accounts to spot things easily.

      Assume initial net worth of capitalists initially.

      Workers' net worth is 0 at all times as you assume.

      Now do you know of an identity which says

      Change in Net Worth = Saving?

      Capitalists' saving is Π_R


      change in Net Worth = Π_R

      At the end of a period, you think capitalists are left with both money and debt - which cancel out.

      So, capitalists' net worth increase by the change in fixed capital which is investment - here I

      So change in net worth = I.

      The two can only be compatible with each other if

      Π_R = I.

      But in your model, these two differ by "change in debt"!

      So you are still left with a black hole.