I learned a new macro fact yesterday, from Vincent Geloso. Macroeconomists should always be on the lookout for new facts that might help us discriminate between macro theories (since they won't let us do randomised experiments with hundreds of countries). Here's a (clearer) graph that Vincent tweeted this morning:
It's US data. The red line is the familiar Nominal Gross Domestic Product. The blue line is Nominal Gross Output. What's the difference?
Think back to Macro 1000. GDP is production of final goods and services. The miller produces flour, which he sells to the baker who uses it to produce bread, which people eat. The bread is a final good that gets counted in GDP, but the flour used to make the bread is an intermediate good that does not get counted. Because that would be double-counting. And in practice, since it's not always easy to tell which goods are final and which are intermediate, we calculate the value added for each firm, by subtracting the goods it buys from other firms from its output, then add up the value addeds across all firms, to get the same answer.
"Gross Output" (I think I've got this right) is what you get when you ignore all that stuff about value added and final vs intermediate goods, and just count everything, double-counting and all. It's bread plus flour. If the miller and baker merged into one integrated firm that did exactly the same thing as the two original firms, Gross Output would fall, but GDP would stay the same.
Macroeconomists don't spend much (any?) time thinking about vertical integration. We leave it to the Industrial Organisation field. We ignore Gross Output (I had thought only a few very strange communist countries even published the data). But Vincent's graph shows that Gross Output fell much more than GDP in the US recession. Hmmm. Why did that happen? What might it tell us about recessions?
Most economists see business cycles primarily as fluctuations in output and employment. I don't. I see business cycles as fluctuations in monetary exchange. We should go back to the old-fashioned name, and call it "the trade cycle". It really doesn't matter whether apple sellers and banana sellers have an endowment of apples or bananas that fell as manna from heaven, or whether they produced those apples or bananas by the sweat of their brows. The apple sellers are better off if they can sell some of their apples and buy bananas. The banana sellers are better off if they can sell some of their bananas and buy apples. Standard gains from trade stuff. And if they use money rather than barter, an excess demand for the medium of exchange will cause an excess supply of both apples and bananas, and a reduction in the quantities of apples and bananas traded, making them both worse off. That's why recessions are Bad Things.
The fall in output and employment that we observe in a recession is just an example or symptom of the general fall in monetary exchange.
Barter, home production, and vertically integrated firms, are three very different things. But they have one thing in common. They are ways of organising economic activity that avoid the use of monetary exchange.
The electrician wants to sell her services to earn money so she can hire a plumber. The plumber wants to sell his services to earn money so he can hire an electrician. But if there's an excess demand for money both will be underemployed. They might resort to direct barter to resolve the problem. Or if barter is hard (which it is if the Wicksellian triangle is a very large circle with many traders who don't all know and trust each other) they might decide to forgo their comparative advantage and DIY instead. The electrician does her own plumbing; the plumber does his own electrics. I don't remember if Coase 1937 on the theory of the firm talked about the distinction between barter and monetary exchange (probably not) but it's the same thing. An integrated firm is like an electrician and plumber who get married. What used to be a monetary exchange now becomes home production. The metaphorical name is almost the same. We call it "in-house" production.
Money is an economic coordination device. But there are other ways of organising economic activity that don't require money: like barter; a household's home production; and firms' in-house production.
As far as I can tell, barter seems to increase in a recession, and home production seems to increase too. These are puzzling facts unless you have a monetary disequilibrium theory of recessions. Perhaps we should think about firms' in-house production the same way. In a recession, when some of the firm's resources are underemployed, it will be more likely to do jobs in-house than contract them out. And a large integrated firm should be less affected by an excess demand for money than an equivalent set of small firms that do the same things as the large firm and use money to trade goods and services between themselves. For exactly the same reason that an electrician and plumber would be less affected if they were married than if they were single.
And maybe, just maybe, that explains the fact that Gross Output declined by more than GDP during the US recession in Vincent's graph. (Though I can't explain why GDP started to fall first; and I have no idea if the same fact will also be a fact in other recessions in other countries.)
[Update: Please read Matt Rognlie's excellent comment, and click on his Fred graph, which seems to resolve the puzzle why GDP fell before Gross Output, and shows that the gap between the two has only partly closed.]
I wonder what recessions would look like if we added home production to GDP? I wonder what recessions would look like if we recognised that lots of "final" goods aren't really final goods but are intermediate goods used in home production. I wonder what recessions would look like if we treated "leisure" as a good and added it to GDP? I'm pretty sure that if we kept going along these lines we could disappear recessions altogether. But recessions are real. And recessions are Bad Things. Recessions are disruptions of monetary exchange, and monetary exchange is (usually) easier than barter exchange, and exchange is a Good Thing, just like we learned in Micro 1000. Macro is Micro 1000 plus money.
But does this mean central banks should target (the level path of) Nominal Gross Output rather than Nominal GDP, as Vincent proposes? Hmmm. There I am not so sure. I can certainly imagine a world in which targeting total nominal transactions would be better than targeting NGDP. Simply imagine a world where exactly the same quantities of apples and bananas are produced every year, but in recessions people are stuck eating their own apples or bananas because they can't trade them for money. But I can also imagine a world in which exogenous waves of mergers and de-mergers (changing Biz skool fads) caused NGO to fluctuate relative to NGDP. Dunno.
Perhaps my understanding of the definitions is a bit off, but aren't inventories also part of the story here?
In the lead-up to the 2008 recession, mills kept producing flour at their trend rate, but bakers turned less of it into cakes. That is reflected by gross production remaining on trend whereas final production fell.
Over time, the mills began reducing their new production, but their existing inventory remained. It took overcorrection -- reducing production below the bakers' demands -- to draw that inventory down in the latter half of the recession and into the recovery.
In the final-goods GDP statistics, these trends would be somewhat offsetting. In the earlier, inventory-building part of the recession the drop in cake consumption would show up to the negative, but the (unwanted) investment in the inventory of flour stocks would show up to the positive. In the later period of the recession and into the recovery, cake production made from flour inventory would show up as only the small-ish positive value-add of the final baking.
In this way (again if my interpretation is correct), the gross output statistic would front-load the recession's total production, causing the observed and eventual drop to be more pronounced than the value-added GDP statistic. Unfortunately, it appears this data is inaccessible for earlier recessions for comparison.
Posted by: Majromax | December 29, 2015 at 01:59 PM
Very interesting, Nick. I'm hoping that Vincent will himself have more to say regarding the significance of NGO and its merits as an indicator or monetary excess and deficiency.
Posted by: George Selgin | December 29, 2015 at 02:01 PM
Majro: you *might* be onto *something* there. Assume the Miller produces $100 of flour, and the Baker adds $100 value to produce $200 bread. GDP is $200 and GO is $300. If sales and production of bread halve, to $100, but the Miller keeps grinding away to add it to inventory, both GO and GDP drop by $100. But that's a 50% drop in GDP and only a 33% drop in GO.
Dunno. I'm still getting my head around it. But unsold inventories are normally included as part of output and expenditure ("you sold them to yourself!").
Geroge: thanks! Vincent tweeted he would get back to it January 2nd. He's probably shovelling snow, like me.
Posted by: Nick Rowe | December 29, 2015 at 04:41 PM
At one point I remember noticing the larger decline in gross output relative to value added and thinking (without investigating further) that it was probably caused by commodity prices, especially oil. The idea is that raw materials are counted many times in gross output, while value that is added downstream, closer to the end of the production process, is counted fewer times. (The value of crude oil is ultimately included in the gross output of the drillers, and the gross output of the refiners, and the gross output of any commercial or industrial users.)
So when the prices of these raw materials go up or down, assuming that short-term substitutability is close to zero, you should expect the ratio of nominal gross output to nominal value added to go up or down as well.
There's an easy correction for this, of course: just look at quantity indices rather than nominal values, getting rid of these relative price effects. Along these lines, this fred graph has:
- a green line (nominal gross output divided by the value added deflator, i.e. the GDP deflator)
- a red line (real value added, i.e. nominal value added divided by the value added deflator)
- a blue line (real gross output, i.e. nominal gross output divided by the gross output deflator)
Comparing the green line to the red line, gross output seems to spike right relative to value added in 2007 and early 2008 before plummeting in late 2008. But when we switch from the green line to the blue line, correcting for gross output prices, the pattern changes and there is no longer any pre-recession spike in gross output. Instead, gross output falls over roughly the same period as value added, with the decline in gross output shrinking a little, but remaining roughly twice as large as the decline in value added.
This confirms the hypothesis that commodity prices played a big role: you'd expect them to push up nominal GO/VA in 2007 and early 2008, then push it down in late 2008, which is exactly the effect that disappears once you control for prices. But there's still a big movement in GO/VA here -- so my initial guess that it was all about commodity prices seems very wrong. (As a side note, the fall in GO/VA seems to persist after the recession, which is curious: there may be some long-term changes mixed in here with the cyclical forces.)
This is a very interesting fact, and I'm anxious to look at it in more detail...
It looks like this point has been recognized in the literature. In section 4 of their paper, which argues for the importance of the "product market wedge", Bils, Klenow, and Malin (2015) point out that intermediate inputs are cyclical as a share of gross output. This is equivalent to saying that gross output is more cyclical than value added, as we're doing here. And Figure 7 in Kim and Shin (2013) shows shipments vs. value added in manufacturing, using the greater cyclicality of shipments as motivation for a theory where production chains can break down due to poor financing conditions.
Posted by: Matt Rognlie | December 29, 2015 at 06:05 PM
Matt: great comment. I have updated the post, so readers will see it.
Posted by: Nick Rowe | December 29, 2015 at 06:38 PM
So if the Baker hires a delivery boy who adds $5 of value then GO jumps to $405.
Sounds like an excellent system... unless I'm missing the obvious.
Peter Klein has some interesting talks on the obvious problem that happens there... money is indeed an economic coordination device and marriage is a different kind of economic coordination device, but the two are not entirely compatible (try bringing up the topic with your wife about a payment system).
Thus, the advantages of the price system when it comes to signalling and incentives, also becomes a limiting factor for vertical integration, along very similar lines as the socialist knowledge problem.
There's stuff here...
http://organizationsandmarkets.com/category/theory-of-the-firm/
I'm sure there's some good introductory talks on the concept, but I seem unable to locate them at the moment.
Posted by: Tel | December 29, 2015 at 09:33 PM
No disagreement there.
Not really, because any type of disruption to our normal system of trade and exchange would show similar symptoms. It could be some strange tax paperwork that people want to avoid, or some sort of employment regulation, or issues relating to trust and contractual agreements, or how the banking industry works, or price fixing by fiat, or just people being nervous about seeing so much government debt and expecting their future tax to go up in order to pay for it. Lots of things could be disruptive, you haven't found a way to isolate the exct causal factor here.
That is an interesting prediction. However, as above, I'll point to Peter Klein's theory that large integrated firms tend to use market prices from outside the firm as their internal benchmarks for decision making purposes (what other reference do they have?) so they could be disrupted at the information level of operations (i.e. planning, and coordination) regardless of whether they had the capacity to avoid disruption at the physical level of operations (all the necessary equipment on the shop floor).
Posted by: Tel | December 29, 2015 at 09:46 PM
Tel: "Thus, the advantages of the price system when it comes to signalling and incentives, also becomes a limiting factor for vertical integration, along very similar lines as the socialist knowledge problem."
Yes. This post is very much about the socialist planning debate and Hayek's Use of Knowledge in Society. It's implicitly underlying everything here.
Here's an old post where I talked about the connection with the Hayek stuff explicitly.
Posted by: Nick Rowe | December 30, 2015 at 05:36 AM
Nick:
What's the difference between "monetary disequilibrium" and more traditional "aggregate demand shortfall"? Is it possible to distinguish between these?
Taking your a/b/m model, you can simply recast it as a typical Keynesian "excess demand for saving" model, and I think all the implications remain the same.
I would expect an increase in barter and home production under the Keynesian story too: when you have more unemployed people, they do favors for each other and spend more of their time cooking at home.
I'm really not sure what the "monetary disequilibrium" angle adds, besides clarifying why you need a monetary economy to get aggregate demand effects in the first place.
Posted by: jonathan | December 30, 2015 at 10:15 AM
@Tel:
> Not really, because any type of disruption to our normal system of trade and exchange would show similar symptoms.
I disagree. I don't think we saw an increase in barter during the very supply-side oil shock of the 70s, for example.
> It could be some strange tax paperwork that people want to avoid,
That's money, or specifically cash being out-competed as a medium of exchange due to regulatory factors.
> or some sort of employment regulation
That could cause an increase in unemployment, but it wouldn't cause barter over under-the-table cash payments or restructuring towards subcontracting. (See for example the Alberta oil patch, where many workers were on paper single-employee subcontracting firms.)
> or issues relating to trust and contractual agreements
That's money, here reflected in the extent to which short-term debt behaves like money. Disruptions cause serious but strictly nominal problems when financing is disrupted and cash-in-advance becomes binding. See obviously the mess of collateralized mortgages or the ABCP market freeze-up in Canada.
> or how the banking industry works
That's vague but mostly money, since the 'banking industry works' by creating and managing beta money.
> or price fixing by fiat
Again money, but like the tax paperwork it involves ordinary money no longer being an ideal unit of account.
> or just people being nervous about seeing so much government debt and expecting their future tax to go up in order to pay for it
Money, here an anticipation of reduced future nominal income. Note that the response between 'expecting their future tax to go up' and 'expecting future inflation' differs in sign, despite both austerity and inflation being valid ways of reducing sovereign debt burden. This also meshes with monetary offset in history such as Canada in the 90s, when federal fiscal retrenchment did not result in dire economic times.
Posted by: Majromax | December 30, 2015 at 10:20 AM
jonathan: add a fixed stock of land to the model, so we can now talk about non-monetary saving and distinguish between deficient aggregate demand for newly-produced goods and excess demand for money. For simplicity, give everyone an equal initial endowment of land.
Suppose there is an insatiable "...craving for the ownership of land, independently of its yield.." (I'm quoting Keynes, arguing against Gesell). So what. Either the price of land increases to eliminate that excess demand for land, or it cannot, or it does not (because the price of land is sticky). Unless it spills over to cause an excess demand for the medium of exchange, it does not prevent people trading apples and bananas for money. It's exactly like an excess demand for mangoes in my barter economy. People want to buy land, but cannot. It's unobtainium.
Posted by: Nick Rowe | December 30, 2015 at 10:32 AM
jonathan: the fundamental difference between land and money is this:
There is only one way an individual can get more land -- buy more land. And if nobody else wants to sell land, there is nothing the individual can do.
There are two ways an individual can get more money: buy more money; and sell less money. Because money flows both into and out of our pockets. And when there is an excess demand for money no individual can buy more money, so the first way is barred, just like with land. But nothing prevents an individual selling less money, so he goes that second way, and we get a recession.
Posted by: Nick Rowe | December 30, 2015 at 10:41 AM
Nick:
Gotcha! That Keynes quote helped me understand by reverse engineering what he must have been arguing with Gesell about.
So let me make sure I understand the terms of the debate:
Keynes is saying that an excess demand for land produces deficient demand for other goods and services. Then since output is demand-constrained, output of other goods and services falls, causing a recession.
Gesell is saying that an excess demand for land produces rationing in the land market, so that some people who want to buy land can't find someone to sell it to them. But people rationally anticipate this, and so spend their money on other things instead of land, and so there is no shortfall in demand for goods and services.
Is that right?
(I'm still not sure what difference this makes in our understanding of recessions relative to the standard Keynesian model, however.)
Posted by: jonathan | December 30, 2015 at 11:47 AM
jonathan: that's exactly right! (Or rather, it's exactly right about how I interpret Keynes and Gesell, and I'm taking "Gesell's" side on this one, whether or not Gesell actual said anything like that!)
And that is part of a longer passage in Keynes where he is saying that Gesell's stamped money might not work, because "..if currency notes were to be deprived of their liquidity premium by the stamping system, a long series of substitutes would step into their shoes....." then gives the craving for land "...which served to keep up the rate of interest...". Which is quite wrong.
If the yield on land did fall, and if land and money were substitutes, Gesell might need to increase the stamp rate (negative interest paid on holding money) in proportion to compensate, but that is it.
Posted by: Nick Rowe | December 30, 2015 at 12:03 PM
I don't see how an increases in barter or home production is puzzling in a "conventional" view of recessions. Let's say investors suddenly change their expectations of future growth and stop investing. People employed making capital goods are out of jobs and with nothing better to do, fix up their houses. That story is not inconsistent with the shortage of money explanation, but it's not paradoxical in and of itself.
Posted by: ThaomasH | December 30, 2015 at 01:49 PM
Hello Nick,
Please find at the following link my reply to this post. As for the commenters here, I hope that I have answered the key point in my own reply.
Best
V
[Link here. It's well worth reading. NR]
Posted by: Vincent Geloso | December 30, 2015 at 05:50 PM
The supply side oil shock was not a disruption to our system of exchange, unless you believe the US dollar is primarily backed by Middle Eastern petroleum deposits... but that would be crazy talk.
A supply side shock caused by external forces is merely new information for the economy to digest and adjust to, it may take time but expecting instantaneous adjustment is unrealistic and unreasonable, so just stand back and let it take the time it needs. There's no fault here, fully normal operation. There may be a downturn in real GDP and loss of wealth... well if a terrible disease killed half the population it would also be a downturn in real GDP and loss of wealth, but the economy would simply adapt as best as it could under the circumstances.
Nick I sort of accept that barter competes with money, but for the vast majority of complex transactions money should win, unless someone has gone and broken the monetary system. My examples were of the many different ways in which it could become broken. You say "a monetary disequilibrium theory of recessions" but that suggests some kind of quantitative problem with currency... what I'm pointing out here is that all sorts of qualitative problems can also occur, and all of them will have a similar symptom which is "a general fall in monetary exchange" but the cause might be totally different.
You cannot print more trust, so if the real crisis is one of confidence, tipping more quantity of the fiat currency into the pool, when people are nervous about the pool itself simply does not help. Fiat currency itself is an IOU note anyway, ultimately all you can do with it is either exchange it or pay tax with it.
Yes, a large proportion of what we commonly call "money" consists of credit instruments (primarily "at call" which means less than 30 days), and that's why M2 is larger than M1.
The freeze-up comes because there's bad credit in the system, and no one knows where it is, so the only safe option is to avoid the credit markets (but as you say, "that's money"). Those RMBS were known to have nastiness in them (somewhere) because people were walking away from their mortgages, unable to pay. A correctly functioning system would sort the good credit from the bad, and of course the guy with the bad credit on his books gets a kick in the bum. Problem is we now have a belief that these systems can operate with no losers... no one will accept being the guy who takes the kick. Clearly that cannot work... and sure enough, it doesn't work.
The "mark to market" system was happily employed by the banks up until it became inconvenient to accept their poor position, at which point it got "temporarily" suspended, because it was an emergency and we just had to do it.
There's a lot written about how rottenness was able to creep into the system, for example Triumph of Banking by J. Edward Ketz
No, people behave based on their best estimate of real income and of course real wealth too. Shuffling nominals may confuse some people some of the time, but at the end of the day people don't want money, they want the things money buys. Say's Law just never goes away. I'm not planning my retirement based on eating money, and I doubt you would be either. If we both know for sure that food prices will be higher when we retire, then we will certainly need more nominal income to cover that.
I try to avoid the word "austerity" because there are too many definitions, shifting too quickly for me to keep up. In this case I presume you use "austerity" to mean "raising tax" and yes in terms of the money supply as a whole, raising tax (all other things being equal) should suck money out of the system to reduce M1 and M2; while inflationary strategy generally involves injecting money into the system (at some point, and who benefits generally depends on who is standing close to the injection point, allow me to borrow the printing press and I will demonstrate this).
However, in terms of an individual decision maker, there's no guarantee that inflation will make them better off. For many people, they have seen their nominal income go up, and their real income go down, and there's at least some expectation that inflation is picking up now (in the USA at least) and seems to be raising prices more than wages (same in Japan I might add). So a person who faces their real income going down in the foreseeable future would be pretty stupid to spend up big right now (IHMO).
Inflation is not necessarily good for the sovereign either, depends on where he borrows from. The sovereign might see short term advantage using inflation to effectively "soft default" on debts, but the sovereign probably is going to need to borrow again at some stage, quite possibly roll-over existing debt. Look at Argentina. Probably Greece would be same as Argentina if they weren't constrained by the Euro.
Posted by: Tel | December 30, 2015 at 07:51 PM
Try that link again: Triumph of Banking by J. Edward Ketz.
Posted by: Tel | December 30, 2015 at 07:52 PM
I was thinking of an example using oil, but in a different way than Rognlie. The stock market currently favors the integrated oils rather than the producers (for example). It seems reasonable to think that integration brings risk diversification and increases the probability of survival during a recession. E.g. an integrated oil company that may be hurting on production margins may at the same time be enjoying fatter refining and retailing spreads. Non-integrated producer firms may suffer by comparison because of too much risk concentration. This thinking is not quite the same as a monetary exchange argument, but correlates with it. Non-integrated firms facing greater risk may have a greater demand for money in recessions relative to integrated firms. And when the crunch comes, their operations be become more disrupted than those of an integrated firm with the capital and liquidity to survive. Maybe another way of thinking about it is that integrated firms replicate external monetary exchange within their internal funds transfer pricing systems. For example, an integrated oil company will show an actual transaction in which the refining division purchases output from a production division. Money is used as the medium of account, but not as the medium of exchange in this way. And the advantages of an integrated firm in internal co-ordination means less disruption within what might be interpreted as a virtual internal monetary exchange system.
Posted by: JKH | December 31, 2015 at 05:56 AM
@Tel:
> The freeze-up comes because there's bad credit in the system, and no one knows where it is, so the only safe option is to avoid the credit markets (but as you say, "that's money").
This is missing one critical feature of the credit seize-up: the short-term value of the paper fell below the long-term value because the market was suddenly illiquid.
An agent without a short-term focus could (and several did) make a mint by purchasing the seized credit at a modest discount and holding it to term. In return, the purchase would provide liquidity to the market by substituting base money for less-moneylike-than-expected credit.
The US Fed, of course, did just this with its MBS purchases.
> You cannot print more trust, so if the real crisis is one of confidence, tipping more quantity of the fiat currency into the pool, when people are nervous about the pool itself simply does not help. Fiat currency itself is an IOU note anyway, ultimately all you can do with it is either exchange it or pay tax with it.
Fiat currency itself almost never loses its trust, and doing so is limited to cases of observed hyperinflation. If anything, the issues in the major economies have been one of too much trust in the base money and not enough in derived (M2) money, so that the effective money supply has fallen despite seemingly neutral to stimulative actions of central banks.
> No, people behave based on their best estimate of real income and of course real wealth too.
If we're looking at a stable or predictable price level, these are equivalent. If we're not -- if we expect the government to inflate away its debt -- I don't want to be left holding the bag of cash. I'll prefer to meet my savings desire through real investment and not nominal instruments, in the worst case even accepting an expected real loss (such as a warehouse full of twinkies that needs climate control and maintenance) if that is less than the anticipated inflation loss. This is not compatible with a fall in observed aggregate demand, only a shift from immediate to durable goods.
In the narrower context of your original point about government debt, you're begging the question. Ricardian equivalence is an immediate and not delayed thing; if consumers are forward-looking about sovereign debt retrenchment then they should be collectively indifferent to deficits versus balanced budgets. The idea that they can develop concerns about debt is a point against the equivalence.
In turn, this is still a story about money. It's either a story about whether government bonds are suitable as collateral (risk of default), or it's a story about future tax revenues (nominal income), or it's a story about the future price level. It's just that the responses to these three stories are not the same.
Posted by: Majromax | December 31, 2015 at 10:42 AM
@JKH:
Even more than internal coordination, integrated firms don't experience counterparty risk. A restaurant that franchises, for example, typically requires its franchise owners to have sufficient liquid capital to cover expected losses over the start-up period. One that owns and operates its own locations, however, faces no such constraint at the level of an individual storefront.
Posted by: Majromax | December 31, 2015 at 10:48 AM
JKH, could you check my comment at the end here?
http://worthwhile.typepad.com/worthwhile_canadian_initi/2015/12/money-demand-and-supply-in-a-redgreen-world/comments/page/2/#comments
Thanks!
Posted by: Too Much Fed | December 31, 2015 at 06:10 PM