17 Nov 2014

“NGDP” Is An Artificial Construct Invented By Economists

Scott Sumner, Shameless Self-Promotion 40 Comments

Don’t let Scott Sumner tell you otherwise. The meat from my latest Mises CA post:

Regardless of the usefulness of (nominal) GDP as an economic statistic, Scott Sumner (and perhaps some of his fellow Market Monetarists) are incorrect when they lead their readers to believe that this is a “raw” empirical fact, which requires no conceptual judgment calls by the economist (the way that real GDP requires an adjustment for the “price level”). In the real world, you could give two teams of economists the same data on nominal expenditures in an economy, and they would come up with different answers for “nominal GDP” if they weren’t allowed to communicate with each other. That’s because the dividing line between “final” and “intermediate” goods and services is somewhat arbitrary.

For example, in a textbook treatment they would say that if a baker buys flour that is completely turned into bread and sold to his customers, then the expenditure on the flour doesn’t count in GDP. But if that same baker buys a new oven, then the expenditure counts in GDP as part of investment. But what about the fact that a fraction of the oven depreciates during the year, and effectively “turns into” the loaves of bread, in terms of accounting? This is the distinction between fixed and working capital, but the proper way to apply it in the real world is somewhat arbitrary. I am quite sure that the average fan of Market Monetarism has not thought through the subtleties involved.

Note that this complication affects the equation of exchange, too. People routinely thinkMV=PQ involves multiplying the number of dollars by the amount of times a dollar bill turns over, on average. But if you want the “Q” on the right hand side to refer to GDP–which economists usually mean–then you have to talk about the velocity regarding a final good or service. In other words, “How many times does the average dollar bill change hands, in a transaction that involves a final good or service?” That makes “V” much weirder than it already is, as a variable to be used in a discipline that ostensibly embraces methodological individualism rather than mindless aggregates.

40 Responses to ““NGDP” Is An Artificial Construct Invented By Economists”

  1. gienon says:

    Thanks for pointing it out, Mr. Murphy. I’m in awe of those nuances that I’ve assumed authomatically, despite “heavy-dose austrianism”, because of my mainstream training and which are misleading or utterly wrong. In terms of making me go “damn, how could I not thing of that earlier? you’ve come close to Mr. Herbener’s explosion of the horizontal demand schedule in the PC model. Keep it up.

    • Bob Murphy says:


    • Ag Economist says:

      ” you’ve come close to Mr. Herbener’s explosion of the horizontal demand schedule in the PC model.”

      I’d love to see that one. Do you have a link?

      • gienon says:

        It’s in his latest course for Tom Woods’ Liberty Classroom, which requires a paid membership. I’ll summarize the point, though:

        A flat demand curve implies that the quantity demanded is positive at a given price (equal to the marginal cost), and is zero at every other price (higher or lower). This, however, violates the law of demand and, if one company decided to sell its products below marginal cost, surely it would be able to find buyers, so demand would not be zero. Obviously, this company won’t have any incentive to do that, because the profit-maximizing price is equal to the marginal cost.

        If there is anything to salvage from the flat curve, it’s that it doesn’t represent demand but rather the only price at which exchanges will occur.

        • Ag Economist says:

          Thanks! It’s all very obvious when one thinks through the logic. Sadly, though, a lot of I.O. is internally consistent but rather silly.

      • gienon says:

        I should have prefaced the post above with the following: I have yet to relisten to the entire course and thing through it more deeply, so my explanation here might diverge from that of Mr.Herbener in certain aspects, but I don’t think I’m misrepresenting him.

  2. skylien says:

    I am arguing since years that economists still forget to impute the GDP value of me sorting the socks of my girl friend after they were washed and dried… No wonder that GDP is an artificial construct without counting that! …

    • gienon says:

      Surely you must be aware of the attempts to come up with a valuation of home cooking and other nonmarket houseworks.

      • skylien says:

        No I am not. Well, if already hookers and illegal drugs are imputed, then this isn’t that far off either I guess. There are no limits to folly..

        Imagine if economists don’t stimulate demand (=housework and cooking!) enough, then I might starve to death and have the wrong socks on my feet.

        • gienon says:

          We need to incorporate money into houseworks to get a good measure of how valuable they are to us. Plus if we did that, we might even stumble upon a whole new batch of “fixprice markets” for Mr. LK and his fellow Keynesians to equillibrate for us.

    • Daniel Kuehn says:

      Google “home production, macroeconomics”. Better yet, Google Scholar it.

      It has consequences for macro that people have been talking about for a long time, but there’s a good reason it’s not included in GDP.

      For anyone interested in this I have an article that should be coming out in Journal of Family and Economic Issues soon on home production and the Great Recession (accepted conditional on minor revisions… which I think more or less means accepted).

  3. Transformer says:

    Do Market Monetarists strongly disagree that “NGDP Is An Artificial Construct Invented By Economists”’ ?

    I think they see it just a convenient way of monitoring variations in AD so they can use monetary policy (adjusting the size of the monetary base) to stabilize it and minimize the effects of demand-shocks on the economy.

    Many Market Monetarists see the targeting of other nominal variables (such as the aggregate spending on wages) as equivalent or better than NGDP targeting as a means to this end.

    • Bob Murphy says:

      Transformer wrote:

      Do Market Monetarists strongly disagree that “NGDP Is An Artificial Construct Invented By Economists”’ ?

      Scott Sumner does. You can click the link I gave in that post, but more generally, he has repeatedly said on his blog that NGDP is a “real thing” (meaning it exists in the world) while RGDP is not.

      • Transformer says:

        Well, he says “The definition of NGDP is clear—the dollar value of all final goods produced in a quarter, or a year. “. He doesn’t explicitly address the issues with the definition of final goods that you raise but I suspect he might acknowledge your point

        What about this ?

        I think many market monetarist might believe that

        1) If you use a nominal variable that is closely correlated to AD
        2) You measure it in a consistent way

        Then targeting that variable will enable you to minimize the impact of demand shocks – even if the variable is “an artificial construct”.

        • Bob Murphy says:

          OK Transformer I had read that and forgotten about it. I’m in the middle of something right now but I’ll come back to it.

          However, in my defense, even here note that Scott himself is opening the post by admitting that he’s taking a break from his normal schtick to talk about nuances.

          • Transformer says:


            And to correct my own comment – I said “If you use a nominal variable that is closely correlated to AD” , but on second thoughts just being “closely correlated” is not enough – it actually needs to have a direct impact on AD.

        • Bob Murphy says:

          Transformer, OK I went back and skimmed that Sumner post. Besides admitting upfront that there are some conceptual problems with NGDP, Sumner doesn’t discuss any of them. You would think, after reading this post, that the drawback to thinking of NGDP as something “out there waiting to be measured” is just the fact that rampant inflation can drive it up. I don’t think Scott has ever gone out of his way to make sure his readers don’t think of NGDP as “total spending.” The closest I can think of is when he sometimes contrasts NGDP targeting with “total wages” targeting or something, but even there I don’t think he makes it clear that neither is “total spending.”

        • Bob Murphy says:

          Last thing: Part of the issue, Transformer, is that Scott with his “musical chairs” analogy makes you think that there’s just not enough total spending going on, so people have enough slack with which to make their mortgage payments etc. I’m not sure if switching to NGDP screws that up, but it’s clear that the commonsense appeal of his “total spending” arguments is hurt when he’s not actually recommending a target of “total spending.”

          • Transformer says:

            I’ve been thinking about it like this:

            An effective economic policy should aim to optimize the output of goods and services.

            The dollar value of this output is NGDP, and it seems pretty obvious (to me anyway) that it should disregard the kind of double-counting you highlight in your post for it to be meaningful.

            Market Monetarists believe that a simple way to optimize the output of goods and services is to stabilize spending on it, as represented by NGDP.

            Clearly there is loads of other spending in the economy beside NGDP. As well as spending on intermediate goods, there’s sales of used goods, financial transactions etc. I’m sure these dwarf NGDP in dollar amount. But as we’re interested in optimizing the output of goods and services its spending on this that we should target.

            • Ben B says:


              In order to optimize the output of final goods and services it’s important to optimize spending in the intermediate production stages.

              I don’t see how you can ignore (or at least place less emphasis on) spending in the higher stages of production since ultimately the spending on final goods depends on these higher stages.

              • Transformer says:

                I’m not an expert on NGDP calculations – but I think that spending in the intermediate production stages is counted, its just not double-counted.

              • Ben B says:

                Hah. Yeah, I’m no expert either; however, I was basing my post on what you wrote, but maybe I misunderstood you.

                In your final paragraph you make it seem like intermediate stages aren’t counted. Yet, even if they are counted, it still seems like you are saying only the spending on final goods should be targeted, and I’m not sure why that is necessarily so if one is basing their policy recommendations on NGDP, and if NGDP includes BOTH consumer spending and total investment spending.

              • Transformer says:

                Here’s an example that I hope clarifies what I meant (I agree above was confusing)

                A wheat farmer grows wheat and sells it for $1000. That’s $1000 added to NGDP.

                Its bought by a baker who makes bread that he sells for $2000. If this full $2000 was added to NGDP then the wheat would have been double-counted. Only the $1000 value-add is included.

                The reason this is important is as follows: Suppose the CB decided to target total spending (not just spedning on final goods). It targets $3000 total spending.

                Someone has the idea of buying the bread from the baker and selling it in a retail store. This retailer buys all the bread from the bakers for $2000 and sells it for $2,200.

                This would significantly increase total spending and if the CB tried to target it they would cause a recession (if prices were not very flexible).

                Final spending has also gone up (by $200 in real terms) but this is an increase in real output. The CB will either accommodate this by increasing the money supply (if it has an increasing NGDP target) , or by allowing the price level to drop if it has a static target (and theory shows that prices will drop more easily after an increase in real output than a decrease in demand or the money supply).

  4. John Becker says:


    What happened to the debate between you and Sumner? I’d be very interested to see that debate because I think that you are right but his case is easier to make.

    • Bob Murphy says:

      John, the postponement of the debate was 100% my fault/call. I got swamped with day job stuff, and continue to be swamped. So I told Scott I would check in with him when I felt I had had adequate time to prepare. Of course, he’s gotten more famous in the interim, so he might not want to deal with me any more.

  5. Tel says:

    Is an iPhone a tool (investment) or a toy (consumption) ?

  6. Bob says:

    Hi Bob,

    In defense of Sumner: When he says “real” he doesn’t mean that there is one NGDP that everyone will naturally calculate.

    I think when says says that NGDP is real he is saying that it is an attempt to express statistically something we all actually experience in our real lives. Every week I receive a paycheck, but it is not split up between the price index and real gdp. Its just a nominal number. The same goes when I spend my money. People never experience price levels and real gdp when receiving income or spending it. They only experience a nominal sum. It is only after that economists tell you that it was split up between price levels and real gdp. This especially makes sense because before you calculate real gdp you need to calculate NGDP.

    I find nothing incorrect in your thoughtful post, but its not really addressing his point. I’m sure in debate (which would be wonderful!) he would be the first to acknowledge your basic point about GDP.

  7. Bala says:

    Is it not as important to note that GDP per se (with whatever prefix) is a meaningless, artificial construct (for the same reasons you have identified) and that it deserves to be banished from all economic theorising and discussion? If so, doesn’t this just become a subset of that broader discussion?

  8. Maurizio says:

    Great post, however I think Sumner could still reply that stabilizing NGDP would work under _any_ definition of NGDP.

    • skylien says:

      There is no reason to believe that the delta between each different way of calculating NGDP has to stay the same. So each way would trigger different policy responses to keep this NGDP “stable”. Of course Sumner can argue the delta will not vary that much so it is close enough for his purpose.

      • Maurizio says:

        I mean, maybe all these different policy responses would work. In other words, maybe it does not really matter if an Iphone is considered an investment good or a consumption good. Maybe NGDP targeting would work with a target of 3%, 4%, 5%, 6%.

        • Scott H. says:

          I agree with you Maurizio. At some point you could jab back at Bob Murphy and question ALL economic measurement. In Bob’s Mises post, he references an easily measured price of apples. What kind of apples? Where? When? And under what circumstances? Does Bob just want to pick the methodology of measuring apple prices that suits his politics and biases?!? I think we’ve really got to think about this whole price of apples thing.

          • Bob Murphy says:

            What are you guys talking about? I’m not saying, “Sumner is dumb because NGDP is an artificial construct.”

            On the contrary, I’m saying, “Sumner should stop selling his policy proposal as being reality-anchored because it measures nominal expenditures and thus avoids the arbitrariness of price levels that economists must construct. His proposal too rests on artificial and somewhat arbitrary constructs devised by economists.”

  9. Andrew_FL says:

    In Fisher’s conception of the Equation of Exchange (Indeed, the normal way of thinking about it before the National Income and Product Accounts came about explicitly for Keynesian analysis) it was not MV=PQ but MV=PT. T was the total volume of transactions, and made no distinction between final or intermediate goods.

    I’ve been thinking this makes much more sense, actually, than arbitrarily considering some transactions to “count” and others not.

    It would still be an aggregate and conceal the important mechanisms of change, but it would at least be less arbitrary.

    • Tel says:

      Actually not clear cut, suppose one firm buys out its supplier. Then the transactions that used to exist between the firm and its supplier no longer exist (other than as internal stock control).

      Now we have less transactions, but the same production.

      • Major.Freedom says:

        You raise a very important point, but does it undercut the point about reducing arbitrariness by including all transactions between property owners?

        What you are saying is akin to me saying “Tracking transactions is not clear cut in general. After all, we don’t include parents giving allowances to their children (because it is money retained under one “household”), so one is as good as any other.”

        In terms of production you’re right, but I’m inclined to agree with Andrew’s point that tracking transactions is least arbitary, or at least less arbitrary than PV, when we set the exchange constraint to separate private property entities.

        But your point is interesting because it means the proliferation of multinational companies and the verticalization of business models seems to reduce the “velocity” and “number of transactions” in an economy, even if actual production is greater in real terms.

        That’s good stuff to think about…

        • Tel says:

          Parents giving pocket money to their children will probably be roughly consistent without any sudden change of behaviour. Admittedly, some fashionable new book about good parenting might hit the market or something… let’s presume that’s unlikely.

          Companies merging and/or splitting can happen at any time, and will result in a sudden dislocation of the metric.

          Maybe it will average out, if the economy consisted of lots of small firms and the random variation averaged out, but with a few big firms that isn’t the case.

      • Andrew_FL says:

        I think it depends what you want to measure. If you want to measure production, you need a measure of production (“Q”). If you want to measure the “effective money stream”-the term Hayek used in Prices and Production-I think you want neither Q nor T, but PT.

        Quoting from one of the footnotes therein:

        “From now onward the term ” amount of money in circulation ”
        or even shortly ” the quantity of money ” will be used for what should more exactly be described as the effective money stream or the amount of money payments made during a unit period of time. The problems arising out of possible divergences between these two magnitudes will only be taken up in Lecture IV.”

        Hayek is fairly clear here that the idea of the “money stream” is equivalent to total payments, rather than payments in only certain, “final goods” transactions.

  10. Major.Freedom says:

    Anyone hear about the head of the Ukrainian central bank recently admitting that almost all of their gold is now gone?


    Perhaps it was the price the new leaders of Ukraine paid to the US.

    • skylien says:

      The gold stuff is really crazy. All the secrecy. No audits in decades, Germans want to get some gold back and they plan to need 7 years! After managing only 5 tons from NY in over a year, Germans just give up in Summer because it is fine how it is anyway… On the other hand Ukraine gold of over 40tons was managed to transport to US within few days and obviously without any bureaucracy… No clear information of CBs how much is leased out… Etc etc etc

      If gold really doesn’t play a role, apart from lying around due to “tradition” in some vaults, then why all this? For heavens sake just let CBs sell all their gold and stop it! Right? If there is so much gold that we don’t need to mine it for the next 40 years then so be it…

Leave a Reply