Home > Uncategorized > ‘DSGE’ macro models criticism: a limited round-up. Part 1. Money.

‘DSGE’ macro models criticism: a limited round-up. Part 1. Money.

On this site, we’ve occasionally criticized neoclassical ‘DSGE’ (Dynamic Stochastic General Equilibrium) macro models. Time for a round-up (comments welcome). This succinct round-up will not be an extensive discussion but will only provide some resources and, important, pay explicit attention to economic statistics which, as a rule, are conceptually much more consistent with Post-Keynesian and sometimes classical and Austrian economics than with neoclassical ‘macro’. I’ll use the next taxonomy to do this:

* Money
* Land, Labour, Capital
* Production
* Prices
* Interest and portfolio’s
* Market fundamentalism
* Equilibrium

Update: because of comments the next items are added:

* Ergodicity
* The intertemporral government budget constraint

Today: money.

Money. The way Eurozone statisticians estimate ‘money’ can be found in this older 1999 ECB manual, which is consistent with the way Japanese, British or USA statisticians estimate ‘money’ as well as with newer manuals, which however do take recent institutional changes into account (shadow banking). There is an internationally and historical consistent approach to estimating money – a hallmark of science. Money is seen as totally endogenous to the economic system (including asset transactions and mortgage credit). The banking system is essentially described as one big bank with the central bank at its core and with so called private banks as branches of the central bank. This system enabled statisticians to estimate how the housing market functioned as a machine providing ever more collateral (as money printing drove up house prices) for break neck speed private money creation. At the time, not enough people took notice of this (Godley did, however), but it was estimated and it is consistent with Post-Keynesian ideas about money. Statisticians however ignored securitization for too long – securitization and the neglect of shadow banking made statisticians underestimate balance sheets of and therewith money creation by for instance Irish and Dutch banks) These statistics are, however, not consistent with for instance the ECB macro models, which (quote!) ‘abandon‘ the use of money. No financial crises or overgrown banks in these models…. ‘Prices’ (even the ‘price’ between present labour and future leisure) are relative barter exchange rates and money (and banks, and monetary taxes, and companies building cash registers and whatever) does not exist, neither as part of the stock of capital (deposits) or as a means of exchange. Or, if it exists it is used as a kind of ‘good’ with different uses (‘loanable funds‘) instead of something which is routinely created and destroyed as part of market transactions all the time, for instance by people using a credit card, when a mortgage loan is initiated or when companies sell ‘on credit’ (mind that selling on credit is legally binding and creates a monetary asset). The most fundamental critique of this ‘abandoning money’/’loanable funds’ idea of money might have been voiced by Fieke van der Lecq.  In her thesis she states (my slightly interpretative summary) that monetary, by definition forward-looking, contracts not only often literally create money (case in point: mortgages) but that, in a market economy with fundamental uncertainty, historical time and a historical flow of intertwined transactions money is unavoidable as people try (sometimes in vain) to mitigate uncertainty by using these forward-looking monetary contracts. Mind that ‘tenure’ and even an implicit, informal ‘day labour’ contract are forward-looking monetary contracts, too. We’re stumbling forward in a dense fog and create a tangled maze of monetary contracts which enables but also constrains our actions and where the ability to create additional credit is often crucial for our plans, be it the plan to buy a candy bar with a credit card or the idea obtain a dream house. When no credit options available, owning credit tokens created by others might do – no matter if these are Euro’s or Pounds.

About the measurement: statisticians measure the total amount of deposit money (i.e. bank money) as well as chartal money but state that only part of the amount of deposit money is used for transaction purposes (M3-money). Money stacked away in longer term saving accounts is ‘bean-counted’ but not classified as M3-money. In the present day low-interest environment this distinction becomes increasingly blurred, while in countries like Greece there clearly is a ‘flight into liquidity’, i.e. from longer term accounts to cash. This criticism also holds for Divisia money, the neoclassical answer to the statisticians. Divisia money ignores the balance sheet approach (but accepts the results of the estimates) and gives different weights to different kinds of M3 money, which on one hand makes sense as not all constituent parts of M3 have the same velocity. In a zero lower bound situation with very low interest rates all across the board these weights will have to be recalculated – but we do not have the historical data to do this. A more fundamental critique: using these weights destroys the accounting relationship between money and credit and therewith makes us loose track of credit risk connected to money creation. Neoclassical economists seem to hate balance sheets, in my opinion because these show that economic entities are intertwinted to the core – via money creation. See this Koo article which shows that the increase the amount of money in the USA during the recovery phase of the Great Depression was caused not by private but by the government borrowing from money creating banks… When economists neglect the obvious – they always do this for a reason. Divisia and M3 money can however be used next to each other!

    April 11, 2015 at 11:41 am

    The first question you sh0ould ask is why are all transactions involving production and exchange organized via money denominated contracts? if you want the answer you should read my textbook POST BKEYNESIAN MACROECONOMIC THEORY

    April 11, 2015 at 11:44 am

    correcting previous comment typos:
    The first question you should ask is why are all transactions for production and exchange organized via money denominated contracts?
    for the answer see my text book POST KEYNESIAN MACROECONOMIC THEORY.

    Paul Davidson

    • merijnknibbe
      April 11, 2015 at 5:58 pm

      Dear Paul, I’ve added ‘ergodicity’ to the list. Could you write a little about these money denominated contracts?

  3. April 11, 2015 at 3:59 pm

    In your list, I would add the category “fiscal policy”. The inter-temporal governmental budget constraint is highly problematic.

    As for money, the standard DSGE treatment seems hard to reconcile with required reserves, or alternatively, how countries without reserves (Canada) operate.

  4. merijnknibbe
    April 11, 2015 at 5:57 pm

    Dear Brian, I’ve added it to the list. Are you willing to write a little about this (400-800 words?)

    • April 18, 2015 at 3:52 am


      I had not checked back here, and I just saw this. I would be interested; could you contact me (or let me know how to contact you, I did not see a contact address.) I have a “Contact me” control on my blog (bondeconomics.com)

  5. Macrocompassion
    April 12, 2015 at 7:57 am

    I am concerned with good representation of the macro-economy AS A WHOLE. If computational methods can do this then there is a good chance that the results will be satisfactory. But for many years the way this subject has been presented is as if each part was able to function independently. So please try to mention the way your particular descriptive feature interacts with all of the others, that is in my reckoning with Landlords, Households, Producers, Capitalists, Financial Institutions and Government. It would be a good thing to define these 6 entities first.

    • merijnknibbe
      April 13, 2015 at 9:41 pm

      Agree. landlords as such are not defined in the statistics (though recent statistiscs by Piketty and Rognlie make clear that we have to do this), the others more or less are (financial institutions not in the models). I’ll pay attention to this.

  6. April 12, 2015 at 10:45 pm

    ” ‘Prices’ (even the ‘price’ between present labour and future leisure) are relative barter exchange rates and money (and banks, and monetary taxes, and companies building cash registers and whatever) does not exist, neither as part of the stock of capital (deposits) or as a means of exchange. ”

    There’s much to not like about DSGE, but I’m not convinced on this point about the medium of exchange. I think it is very hard to interpret the results of these models without assuming that all exchange is monetary and there is no barter trade. Otherwise you either have a position where some agents are not acting optimally or no explanation of why relative prices are what they are. Do you have a reference for this?

  7. merijnknibbe
    April 13, 2015 at 12:20 pm

    Dear Nick,

    in the article about the ECB New Area Wide Model they literally state it: they (quote) *abandon* the use of money. Without money – no monetary prices. Which means: no longer term monetary contracts. In the same model they do have a ‘monetary authority’, which indeed makes things more complicated. But money as you and I know it does not play a role in the model.

    • April 13, 2015 at 4:10 pm

      I think all they mean by that is that they have removed any cash-in-advance-constraint features, i.e. people have a budget constraint, but no liquidity constraint. This means that the stock of money plays no role, rather than that the results of the model assume barter is taking place. Having no role for the stock of money would be in line with certain views on endogenous money.

      The point is that, with sticky prices and in response to shocks, you get different outcomes depending on what types of exchange can take place. Take an interest rate shock. With monetary exchange only, you get a position where households would like to provide more labour than they are able, so as to acquire additional monetary assets to consume more in the future (and there are monetary assets in that model, even if there isn’t a stock of money). With barter exchange allowed, households can trade labour for goods up to the point where they do not wish to provide more labour. What they would like to do is sell those additional goods for monetary assets, so they can consume tomorrow, but it is in this that they are frustrated. Neither outcome is optimal within the terms of the model, but they are different. I think (and maybe somebody can correct me) that the NAWM produces the former outcome.

      • merijnknibbe
        April 13, 2015 at 10:10 pm

        Dear Nick,

        in the EAGLE model (an offshoot of the NAWM) something called money is used. But we do have to pose the question: is this the real thing? it isn’t.https://www.ecb.europa.eu/pub/pdf/scpwps/ecbwp1195.pdf

        For one thing: there are no banks in the model, aside from the ‘monetary authority’ which, however, does not seem to be a bank, either as it’s the fiscal authority which creates money (mind that there is no Eurozone equivalent of a paper money creating USA ministery of finance).

        This is important as the real world the larger part of money is created by banks, using purchasing creating loans, while the fiscal authority only seems to produce and spend coins ínto existence’ and does not provide credit. Money and purchasing power is created out of thin air by banks plus (!) borrowers – as clearly shown by the quadruple accounting method which is the basis of the monetary statistics of the same central banks which publish models like the EAGLE model. Money creation is very much a private and not a government act.This *purchasing power* creating debt/money system is entirely absent from the EAGLE model which means that money as you and I use is is not covered by this model.

        Not that long ago a society (almost) without banks, as assumed by the EAGLE model, existed in fact not that long ago – we only have to look at the very monetized and commercialized Dutch or Flemish pre-1850 countryside (including villages and sometimes pretty large cities). But this does not mean that purchasing power creating credit is something new. Not at all. In those days households borrowed from each other but also and more importantly used amazing amounts of commercial credit to buy stuff from small groceries, cobblers, blacksmiths, whatever. This commercial credit solved liquidity problems and the problem of the mutual coincidence of wants, the modern credit card is in a very real sense the modern descendant of this system (old style commercial credit had almost always 0% interest, though…). Marcus Goldman, the founder of Goldman Sachs, in fact started by purchasing the ‘receivables’ owned by small groceries, cobblers and blacksmith at a discount them to larger banks at a smaller discount. But the point: even in these non-bank societies purchasing power creating credit existed, based upon ‘quadruple accounting’ relations between households and firms. Purchasing power creating credit is central to our monetary system and, to its very core, to our money and it has been like this for centuries. But it’s absent from the EAGLE model… Ironically, consumer debts owned by car companies (which still sell on credit, mind thise 0% car loans) are accepted by the ECB as collateral, mimicking the acts of marcus Goldman… but it’s not in the models of the same bank.

      • April 14, 2015 at 5:09 pm

        I certainly agree that the omission of any representation of lending is a big weakness in that model and ones like it.

  8. Lino N
    April 17, 2015 at 10:33 pm

    Thank you for undertaking this very worthwhile project. But, please, while attacking the details, let’s also be sure to identify the fundamental, underlying failures. A general failing of DSGE models, which supposedly describe dynamic economic behavior, is that their mathematical vocabulary is not capable, in any non-equilibrium situation, of describing either the changes provoked by institutions and players within an economy, or the way the various classes of actor interact, circumstances that are basic in any real economy. Mathematics is the language of quantification. Successful quantitative translation from English descriptions depends on choosing appropriate vocabulary to describe the phenomena of interest. The DSGE approach doesn’t provide it. Imposing equilibrium for the purpose of describing dynamic behavior distorts describing the dynamic behavior. Dynamic … General Equilibrium is an oxymoron.

  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s