# The quantity equation, the quantity theory, and Bennett McCallum

I’ve always found it interesting that the quantity equation (M*V=P*Y) is linked to the quantity theory of money.  Obviously there is no logical relationship between the two, as one is almost always defined as an identity, while the other is a theory.  But there certainly is a perception that the two are somehow linked.

As an aside, I am talking about the modern version of the quantity equation, aka the equation of exchange.  I believe that in earlier times there was an assumption that velocity could be estimated independently, and thus the equation was sometimes treated like a theory that may or may not hold true.  Today “velocity” is nothing more that the ratio of nominal GDP to whatever monetary aggregate you are interested in examining.  This is one reason I have always preferred the Cambridge version of the equation (M = k*P*Y), as k really is the ratio of money to gross income, whereas V isn’t really the average number of times a given dollar is spent.  In addition, k better reflects the concept of the “demand for money,” which lies at the heart of the quantity theory of money.  A better term for velocity would be “the multiplier,” but of course that term is already used for different ratios—the ratios of monetary aggregates to the monetary base.

In the past, even some textbooks were confused about the quantity equation, which was often mislabeled “the quantity theory of money.”  I don’t see that mistake as often today.  Although there is no necessary relationship between the quantity equation and the quantity theory, it so happened that proponents of the QT tend to use the equation to illustrate the intuition behind the theory.  As an analogy, Keynesians often use the C+I+G=GDP identity to illustrate the intuition behind Keynesian economics, although there is nothing really “Keynesian” about the equation.

So what is the quantity theory of money?  I suppose there are many versions:

1.  The ratio of P and M is relatively stable.

2.  The ratio of P*Y and M is relatively stable.

3.  An exogenous, one time, permanent increase in M causes a proportional rise in P*Y

4.  In the long run an exogenous, one time, permanent increase in M causes a proportional increase in P.

I prefer the second, third, and fourth versions, which I think each capture something of the spirit of the QT.  But I also think the quantity equation can get in the way of clear thinking.  For instance, people worried that current Fed policy will lead to much higher future inflation sometimes cite the quantity theory.  But this is a misuse of the theory.  It does not imply that any increase in the money supply is inflationary, but rather that permanent, exogenous increases are inflationary.  For instance, suppose the Fed adopted a policy of targeting the expected inflation rate at 2%.  Assuming their policy was efficient, i.e. the errors were unforecastable, then there should be zero correlation between the money supply and inflation.  Of course the Fed doesn’t have a precise 2% inflation target, but they certainly have some inflation target in mind.  If so, then changes in the money supply are partly endogenous, and the QT does not predict much correlation between the money supply and inflation.

Another example occurred a few years back when econometricians developed a new plaything called “cointegration.”  Two time series are cointegrated if they tend to move together over time, rather than drift apart.  Some economists got the bright idea to test the QT by seeing whether M and P*Y were cointegrated.  And at least some of the studies showed that the two series were not cointegrated—thus the QT was refuted.  I recall reading these studies with a sense of frustration, as it didn’t seem to me that their tests showed much of anything.  I never recall David Hume, Irving Fisher or any other quantity theorist arguing that M and P*Y were cointegrated.  Yes, they certainly argued that changes in M and changes in P*Y were correlated, but that is very different concept.

For instance, suppose that velocity was a random walk, slowly evolving over time as payments technology changed.  Also suppose that the average annual change in velocity was 0.1%.  I think almost everyone would admit that if velocity was that stable, usually changing by only a small fraction of one percent, then the QT would be a very useful theory.  But if velocity behaved this way then M and P*Y would not be cointegrated, despite the extremely close correlation between M and P*Y.

Later I recall reading an article by Bennett McCallum where he made the same argument.  Since then I have read more articles by McCallum than by any other economist.  He is my favorite monetary economist.  One thing I began to notice is that every couple of years some hotshot would come up with a new theory or empirical finding that supposedly refuted the QT.  And each time Bennett McCallum would pour cold water on the idea.  And in each case I found his argument convincing.  In some cases I had reached the same conclusions on my own, while in other cases I didn’t know enough about the subject to have an intelligent opinion, but his arguments always seemed more persuasive to me that those of the anti-quantity theorists.  Here are a few more examples:

1.   Until the 1970s most economists had visualized RGDP as “trend stationary” which meant that it was assumed that real GDP fluctuated around a stable trend line.  Indeed that was a prediction of the so-called natural rate models, which argued that a nominal shock would temporarily push RGDP away from the natural rate, but once wages and prices adjusted RGDP would return to a stable trend line.  Then some studies in the 1970s and 1980s showed that real GDP actually seemed more like a “difference stationary” process.  This means roughly (unless I am mistaken) that if the average rate of RGDP growth is 3%, and if RGDP suddenly falls by 5%, then the expected rate of growth from that point forward is still 3%.  This is puzzling, as it suggests that we never recover from recessions.  Indeed it is similar to studies showing that the stock market is a random walk, which also imply that the stock market never recovers from a crash.  When stocks fell 23% early last October, the expected level of the stock market 20 years out in the future should have also fallen by roughly 23%.

If true, this wouldn’t just imply that the monetarist theory of business cycles was wrong; it would also imply that the Keynesian model, and indeed any demand-side model, was equally wrong.  I found this exceedingly implausible, but the empirical evidence was pretty strong.  Strong enough to boost “real business cycle” models that could explain why RGDP was a difference stationary process.  Bennett McCallum was also skeptical, and pointed out that the data could not distinguish between two hypotheses, one where all cycles were real, and RGDP was difference stationary, and another where there were both real and nominal shocks, and the nominal shocks had no permanent impact on RGDP.

In my view one reason RGDP seems difference stationary is that real and nominal shocks are “entangled.”  So when you have a long stretch of positive productivity developments, the (unobservable) Wicksellian equilibrium real interest rate goes up.  This makes monetary policy (unintentionally) more expansionary.   Once a negative productivity shock hits, the Wicksellian equilibrium real rate falls and monetary policy may tip the economy into a recession just when the real underlying rate of growth was falling anyway.

2.  Another group of anti-QT models went under names like “the backing theory,” or the “fiscal theory of the price level.”  I don’t know a lot about these theories, as they always seemed implausible to me.  But the backing theory argued that what mattered wasn’t the quantity of money, but rather what sort of assets backed the money stock.  The fiscal theory argued that what mattered was the total supply of government bonds, not the quantity of base money.   McCallum also found these theories implausible.  He argued that they either had the same predictions as a modern QT model featuring rational expectations, or they produced inferior predictions.

Some proponents of the backing theory argued that it could explain the Colonial American experience better than the QT.  In Colonial America the colonies issued some temporary paper money to finance the French and Indian Wars.  Because these injections were backed, and the intention was to eventually redeem the money for gold or silver, the large increase in the money supply was not inflationary.  But I pointed out in a 1993 study that even the modern QT would predict little or no inflation after a temporary currency injection, backed or unbacked.  Later I found out that McCallum was also skeptical of these theories.

3.   I sometimes like to imagine economies with no banks, interest rates, or financial system.  Just money and goods.  Money is used to avoid the hassles of barter.  And that’s it.  I find these models really useful for thinking about how monetary policy impacts NGDP.  The QT still holds in that sort of world, even though there are no interest rates or loans or other mechanisms that people often see as central to the money supply process.  Michael Woodford is exactly the opposite of me.  He looks at monetary policy exclusively from the interest rate perspective.  He doesn’t even put money in his models.  Woodford once developed what he claimed was a “cashless economy” model, to emphasize that interest rates were central and money was peripheral.  But this made no sense to me.  I could imagine a monetary economy with no interest rates, but how could you have a monetary economy with no money?  What would the price level refer to?  If you don’t have a medium of account, then there is no price level.  McCallum pointed out that Woodford’s model did have a MOA, bank reserves used to settle interbank debts.  So even if all ordinary transactions were settled with electronic payments, there still must be some sort of medium of account (MOA), or else there would be no way to measure prices.  Of course you can make a real asset like gold the MOA, but then gold would essentially assume the role of “money” even if it did not circulate.  Indeed toward the end of the gold standard very little gold did circulate, but nevertheless the price level was determined by changes in the supply and demand for gold.

4.  You have seen me argue strongly for a policy of “targeting the forecast.”  Bernanke and Woodford once argued that this led to a potential “indeterminacy problem.”  If I’m not mistaken this means there might be multiple money supply paths that are all consistent with the inflation target.  As the following abstract shows, McCallum doesn’t think this is much of a problem in practice:

Contemporary literature on monetary policy analysis concludes that use of an interest rate policy rule that responds to expected inflation in some future period may generate indeterminacy””a multiplicity of stable rational expectations (RE) solutions. By contrast, this article argues that in these analyses only one of the solutions possesses the property of learnability, which is necessary for the plausibility of any RE solution since its absence implies that there is no way for individuals to obtain enough information to form expectations that would support the solution in question. Thus indeterminacy of the type discussed is not an actual problem for actual policymakers.

If someone can explain to me what “learnability” means, I would be very grateful.

McCallum also had some interesting things to say about rational expectations.  Many economists were initially skeptical of rational expectations theory.  They would ask “if brilliant macroeconomists cannot even agree on what the correct model of the economy is, then how can we possible expect the public to fully understand the model?”  McCallum pointed out that this was the wrong way to think about rational expectations.  He didn’t even like the term ‘rational expectations,’ insisting the concept should be called ‘consistent expectations.’  McCallum said the real question was whether when you develop a model, you assume that the public has expectations that are consistent with that model.  He pointed to the absurdity of developing a model that assumed the world was X, but then assuming that the public believed the world was Y.

Here’s how I think of his argument.  It may seem arrogant to think you know how the public forms expectations, but it is far more arrogant to assume that you know how the public should form expectations, but the public doesn’t know this.  And yet any non-Ratex model implicitly adopts exactly that sort of arrogant stance toward the public.  McCallum also pointed out that if you didn’t assume rational expectations, you would not be able to get any kind of useful policy advice out of the model.  Why not?  Because an optimal monetary rule would take advantage of the public’s lack of rational expectations, it would try to fool the public.  But as we learned in the 1960s and 1970s, that can only work for a short period.

About 20 years ago I met David Glasner at a conference, and we discovered that we had a lot of similar interests.  One of those interests was Ralph Hawtrey, an interwar economist who has always been in the shadow of Keynes and Fisher.  Glasner knew more about Hawtrey that I did, but we both agreed that he had really good judgment.  I’ve notice that some very brilliant economists, or brilliant people in any field, can have very poor judgment about which hypotheses are plausible and which are not.  It is something that can’t be taught.  I see Bennett McCallum as a Hawtrey-type economist.  He’ll never win a Nobel Prize for any theoretical breakthrough, but I’d trust his judgment on monetary economics more than anyone else alive.  He has an unerring ability to get to the heart of the matter, to see whether a new idea is not just logically sound, but plausible, or useful.

I recall that he sometimes criticized other economists who developed some highly stylized model of the economy, and then solved for the “optimal monetary policy” in that imaginary economy.  McCallum pointed out that we really don’t have an agreed upon model of the microfoundations of business cycles.  He once listed no fewer than 10 different theories of wage and price stickiness–each of which had different policy implications.  He argued that we should look for a robust policy rule, one that would work reasonably well under a wide range of model specifications.  At the time, most mainstream macroeconomists favored some sort of flexible inflation target.  But McCallum came to a different conclusion.  So what did the world’s most sensible macroeconomist decide was the optimal monetary target?

You guessed it—NGDP!

At one time McCallum favored a feedback rule where the base would be adjusted each period to offset changes in velocity.  But this policy is still backward-looking, and hence may react too slowly in a crisis.  Aaron Jackson and I developed the idea of a velocity futures market, where traders would forecast the ratio of next period’s NGDP and this period’s monetary base, as a guide to policymakers.  Of course many Keynesians can’t stand the sound of the word “velocity,” so we also developed a similar proposal using the ratio of next period’s NGDP and this period’s fed funds target.

Unfortunately, this still left the central bank with too much discretion for my taste.  The central bank would still have to decide which instrument was better, the monetary base or interest rates.  Most economists prefer interest rates, but on the other hand interest rates don’t work well when the nominal rates hit zero.  So I developed a different version of futures targeting where the Fed wouldn’t have to choose any monetary instrument at all.  The market would essentially become the open market committee, and purchases and sales of NGDP futures would trigger offsetting open market operations in T-debt.  The futures traders could look at the monetary base, interest rates, asset prices, or any other indicator they choose.  It would be a post-ideological monetary policy, one that moved beyond the old-fashioned debate about interest rates vs. the money supply.  We could finally get rid of the quantity equation.  Not because the QT was no longer true, but rather because it was no longer useful.  Money would be fully endogenous.

[I hope you didn’t think I was going to admit that there was any “fact of the matter” as to whether the QT was true.  Not after confessing my conversion to Rortian pragmatism.]

PS.  A lot of the work Bennett McCallum does is highly technical and goes over my head.  But he also explains the intuition of each concept very well.  Nevertheless there are probably a few technical errors in this post, as my research is at a lower level than his.

Tags:

73 Responses to “The quantity equation, the quantity theory, and Bennett McCallum”

1. Gabriel
11. August 2009 at 18:06

Learning/learnability: http://press.princeton.edu/titles/7097.html ?

2. Mattyoung
11. August 2009 at 19:41

In the real world every goods exchange is costly, so transactions are minimized, not money velocity. Consumers settle into the optimum transaction rates for long, medium and short term transactions. The rate of trips to the grocery store, intervals between car purchases, frequency of house purchases, etc; and all coherent rates designed so the economy acts as a frequency sampled system.

Price, quality and quantity adjust to keep the same transaction rates. Along the yield curve there are a finite set of terms and associated inventories. Quantity and substitutability are adjusted to keep the sample rates constant and optimally spread across the curve.

The constant transaction rate hypothesis best explains stationary trend and “snap back” as Milton called it.

3. Richard A.
11. August 2009 at 20:22

Wikipedia gives a good writeup on the McCallum rule.
http://en.wikipedia.org/wiki/McCallum_rule

The advantage of a McCallum-like rule is that it is usable when interest rates try to go below 0%.

4. pushmedia1
11. August 2009 at 22:35

Learnability is a way to select RE equilibrium. The idea is that if agents don’t know the true parameters of the economy, they should be able to learn them (using least squares methods). Only some RE equilibria have this property.

5. Current
12. August 2009 at 03:40

That was interesting. I now have a huge amount of reading to do.

6. Nick Rowe
12. August 2009 at 04:05

Yes. Very much agree.

I like to think of the Quantity Theory and the Neutrality of Money as twin theories that go together.

The QT says a change in M will cause an equi-proportionate change in all P; and the NM says it won’t change anything else (real). That way, the two theories stand and fall together.

It is theoretically possible that monetarists and Keynesians are right, and that real GDP have a unit root. If a monetary-induced recession causes a drop in real investment, the consequences of that for real GDP could be permanent in an AK growth model, for example.

7. ssumner
12. August 2009 at 04:14

Gabriel, Thanks. I read the blurb on the front page. My initial reaction is that if they are assuming that people learn by developing ad hoc rules for interpreting data in asset markets, then the theory may be useful for looking at issues like nominal wage determination. If they are assuming that the data people look at is only variables like money, inflation, and real GDP, then they may be missing something important. But I don’t know enough to have an intelligent opinion.

Mattyoung, I’m not sure I follow your argument about transactions. Does it have any implications for the quantity theory of money? That might help me understand it better.

Thanks Richard. I agree about the advantage you cite, but I still think that any backward-looking rule would have been too slow for this crisis. Of course the key issue is level targeting. If he proposes level targeting in NGDP than we probably would have done OK.

pushmedia1, The problem I have is that when I read the sort of statement you make, it doesn’t trigger any reaction in my slow brain. Perhaps if I could see an example of a REE that was not learnable. I have trouble visualizing the sort of scenario someone has in mind. Are they talking about actual real world data? I.e we don’t see explosive price level movements so that we can infer that people using rules of thumb in forecasting would not expect any REE that implied an explosive price level movement?

Again, this is my fault not yours, it’s just that I am used to absorbing abstract concepts by picturing real world examples. Perhaps you can tell me this: Am I right in inferring that the learnable REEs are those with normal, mundane-looking price level paths, and the non-learnable REEs are the one’s with crazy-looking price level paths?

Current, I wish McCallum would write a book explaining his take on all of modern monetary theory in layman’s terms. Or at least “layman economist” terms. He does have an older money textbook on the market, maybe I should take another look at it.

8. ssumner
12. August 2009 at 04:21

Richard, I just checked the Wikipedia link. It looks like it is halfway between level targeting and a “memoryless” rate targeting regime. If I am not mistaken, McCallum is basically saying that if you have a 5% NGDP target, and in the previous period you had 7% NGDP growth (an overshoot of 2%) you try to make up for one half of the mistake over the next period. I.e. you try for 4% growth.

9. Current
12. August 2009 at 04:22

Nick Rowe: “I like to think of the Quantity Theory and the Neutrality of Money as twin theories that go together.

The QT says a change in M will cause an equi-proportionate change in all P; and the NM says it won’t change anything else (real). That way, the two theories stand and fall together.”

I don’t think that’s the right way to look at it. The Quantity theory is much older than the modern idea of money neutrality, and really quite separate.

Old mechanical Quantity theory says that if M rises then there is an equi-proportionate change in P after a period of time. It does not say that this change doesn’t affect any real variable, that is a separate theory.

10. ssumner
12. August 2009 at 04:43

Nick, You said;

“It is theoretically possible that monetarists and Keynesians are right, and that real GDP have a unit root. If a monetary-induced recession causes a drop in real investment, the consequences of that for real GDP could be permanent in an AK growth model, for example.”

That’s good point, and it could also explain why unemployment rates do seem trend reverting while RGDP doesn’t. But I still don’t think it is likely. My intuition tells me that technological progress (not investment) drives the long run trend of RGDP in the US, and Alexander Field showed that technological progress actually accelerated in the Great Depression, for instance. I like the explanation that a 1% real shock today causes RGDP to change by more than 1 percent 10 or 15 years out, and a 1% nominal shock causes almost no change in RGDP 10 or 15 years out. When you mix the two together you have a messy, complex time series, for which we don’t have enough data to disentangle qualitatively different shocks.

I do think the unit root stuff is an interesting way to think about countries like Argentina. Did you or Stephen do a post comparing them to Canada recently? It seems they fell way behind around the 1930s, and the gap has been fairly stable ever since.

On your other point about money neutrality, it best applies to my 3rd and 4th definition of the QT. If the income elasticity of money is less than unity, then velocity should rise with real GDP growth, and thus the 2nd version of the QT that I provided might not hold even if money was neutral. In practice I think the income elasticity has been pretty close to unity, despite the Baumol-Tobin model’s prediction of 1/2.

BTW, A point I keep emphasizing is that the neutrality of money only applies to monetary injections expected to be permanent. If they are expected to be permanent, they even apply in a liquidity trap. And if they aren’t expected to be permanent, then they don’t even apply to non-liquidity traps.

11. Current
12. August 2009 at 05:08

Scott: “My intuition tells me that technological progress (not investment) drives the long run trend of RGDP”

This seems odd to me, clearly these are related. Spending on technology is investment.

12. Current
12. August 2009 at 05:20

On a sidenote Scott wrote:
“As an analogy, Keynesians often use the C+I+G=GDP identity to illustrate the intuition behind Keynesian economics, although there is nothing really “Keynesian” about the equation.”

Actually, real GDP is quite a Keynesian idea. If we are concerned with final output then consumption is what we look at. If we are concerned with the level of investment for the future then clearly investment is important. What do we get from adding them together? We get a gross output, but why is that figure important?

In Keynesian economics investment and consumption output rise and fall together, so GDP is particularly important.

13. ssumner
12. August 2009 at 05:26

Current, David Hume is pretty old, and he said no real variables were affected in the long run (after prices adjusted.)

14. Current
12. August 2009 at 05:57

Scott: “David Hume is pretty old, and he said no real variables were affected in the long run (after prices adjusted.)”

Well, if you go that far back you’re right. I should say that the ideas have been separate for some time.

15. Mike Sproul
12. August 2009 at 06:50

Scott:

Hardly anyone denies the truth of the backing theory when money is physically convertible. If some bank issues 100 currency units (shillings) that are each convertible into 1 oz. of silver, and if the bank’s assets consist of either 100 oz of silver, or various assets worth 100 oz., then each shilling will trade for 1 oz. It should also be clear that if the bank issued another 20 shillings, while getting additional assets worth 20 oz., then each shilling will still be worth 1 oz. The quantity theory and the equation of exchange are clearly not applicable to this case.

The colonial currencies covered by McCallum were all physically convertible, just in a different way. Instead of a colonist being able to take a paper shilling to a bank and get 1 oz of silver for it, the colonist could take a paper shilling to the tax man, and the tax man would accept it in lieu of 1 oz. If the colony issued another 20 shillings, while getting additional assets worth 20 oz., then the value of the shilling would stay at 1 oz. The quantity theory is just as inapplicable in this case.

If you agree that the backing theory is correct for convertible moneys, then the next step is to recognize that there are many different types and degrees of convertibility. Convertibility can be instant or delayed, physical or financial, certain or uncertain, at the customer’s option or at the bank’s option, done with taxes or with pay-outs of metal. The moneys that are traditionally called unbacked fiat moneys are actually backed by the issuing bank’s assets. The fact that they are sometimes not physically convertible at the customer’s option gives them the false appearance of being unbacked.

16. Philo
12. August 2009 at 07:53

ï»¿”We could finally get rid of the quantity equation.” If the equation is an identity–perhaps implicitly a definition of ‘velocity’–why would you want to get rid of it?

“Not because the QT was no longer true, but rather because it was no longer useful.” Would it be “useful” to believe the *negation* of QT? Was it *ever* useful to believe QT (as opposed to believing its negation, or suspending judgment)? If so, when, and why?

By the way, many–probably most–of the things I believe, which in ordinary parlance are “true,” aren’t particularly *useful*. Under certain circumstances they might become so, but those circumstances aren’t going to occur.

The philosophizing may be doubtful, but the economics is (to this non-economist) fascinating.

17. Bill Woolsey
12. August 2009 at 08:23

Current–

Adding consumption and investment together makes sense because of scarcity.

Both are spending on current output and there are only limited resources available to produce current output.

In the simplest situation, adding investment and consumption and getting output is the same thing as saving equals investment. (Saving is income less consumption.)

The opportunity cost of producing additional future consumer goods is the current consumer goods that must be sacrificed in order to produce capital goods.

As for investment and technology, as old capital goods wear out, they can be replaced by new and better ones.

This relates to issues of gross and net output. The bulk of gross investment replaces capital good. If there were no technological change, and the same capital goods were replacing worn out ones, then this wouldn’t incease output at all.

18. Current
12. August 2009 at 09:22

Bill,

Your view of GDP does make sense. But, how many times have you read an economist bring up GDP in order to make the points that you have made? What you have explained is the conceptual significance of GDP in the way Hayek would. That still leaves investment as being more important in the long term.

Bill: “As for investment and technology, as old capital goods wear out, they can be replaced by new and better ones.

This relates to issues of gross and net output. The bulk of gross investment replaces capital good. If there were no technological change, and the same capital goods were replacing worn out ones, then this wouldn’t incease output at all.”

Yes. My point is though that technology is not exogeneous. Scott said “My intuition tells me that technological progress (not investment) drives the long run trend of RGDP”

Investment and technology are conceptually distinct. But technology is the outcome of a production process for making technology. It is the outcome of investment in technology which is a sort of investment.

19. StatsGuy
12. August 2009 at 11:45

This is a really dumb question (because I haven’t read the paper), that I’m sure you’ve handled elsewhere:

In your NGDP futures targeting regime, what mechanism do you use to ensure that traders on the NGDP futures market do not manipulate the market in order to induce the Fed to respond with favorable monetary policy (“favorable” meaning favorable in a rent seeking way). Or did you presume the NGDP futures market would be sufficiently large that no such incentive would exist?

20. Paul
12. August 2009 at 12:43

This is somewhat unrelated, but doesn’t NGDP level targeting increase your chances of stagflation greatly. ie if real GDP falls quite a bit, you are guaranteeing very high inflation? I am sure this is a very simplistic understanding but the thought just popped into my head and I couldn’t figure out how it wouldn’t happen.

21. Paul
12. August 2009 at 12:46

P.S. Sorry to just bombard your discussion with that random side note, usually I read all the comments before I post anything, but today was a long day, and I couldn’t stop thinking about the stagflation comment.

22. Greg Ransom
12. August 2009 at 20:05

Explain the absurdity here, because I don’t see it:

“He pointed to the absurdity of developing a model that assumed the world was X, but then assuming that the public believed the world was Y.”

23. OregonGuy
12. August 2009 at 20:37

Without reading the other comments, learnability, as directed by “expected inflation in some future period may generate indeterminacy””a multiplicity of stable rational expectations (RE) solutions. By contrast, this article argues that in these analyses only one of the solutions possesses the property of learnability, which is necessary for the plausibility of any RE solution…” that what we would look for is the ability of the observer to uptake the information that is available and the observer’s ability to transform that information into usable data.

Now, let’s see what other commenters had to say.

M’kay. I pretty much hit that one. What is interesting is the lack of discussion of prices. Of course, I am a determinist that seeks price info for supply. But, I’m an old guy.
.

24. Gabriel
12. August 2009 at 20:43

Re: OLS learning… agents hold a model of the economy (that might or might not be the true model) and they estimate its parameters via (recursive) OLS using the t-1 dataset. They act based on those parameter estimates and then, via the true model, the actual outcome is realized at time t. Agents observe the outcome and reestimate the parameters, and [repeat].

When agents use the true model, the learning process converges to some R.E. equilibrium of the model (under some regularity conditions).

It turns out that this dynamic will never converge to some equilibria and, happily, it’s usually those equilibria we don’t like/can’t make sense of.

People in this literature looked also at the case when agents use a misspecified model.

25. Bill Woolsey
13. August 2009 at 02:11

Paul:

NGDP level targetting means that aggregate supply shocks involve stagflation. This would be as opposed to a scheme where aggregate demand drops, which generally results in a deeper recession in output and employment and less inflation. Or else, some scheme where aggregate demand grows more quickly, inflation is more servere, but the disequilibrium impact on output offsets the adverse aggregate supply shock.

If the adverse aggregate supply shock is permanent, then the stagflation will be termporary. That is, there will be inflation (or higher inflation with 5% growth,) but then the price level will settle at a higher level (or growth path with 5% growth.) Output is on a lower growth path and prices are higher (or on a higher growth path.) The growth rate of both output and prices return to where they were.

If the aggregate supply shock were temporary, then the price level will fall back as output recovers. (Or there will be disinflation as the price level returns to its past growth path.)

26. Bill Woolsey
13. August 2009 at 02:27

Statsguy:

I think it should be a matter of concern. I suppose the simplest scenario is you sell the futures and buy T-bills. The Fed responds to its matching purchases of the futures by purchasing T-bills. You sell T-bills to the Fed at a profit. Of course, you are left with a short position on the future, and the Fed’s policy is going to create higher nominal income, and you will take a loss on the future.

I would note that if you think nominal income will be below target, and you sell the futures, it may well be sensible to buy T-bills. The speculation in T-bills, lowering their yields, is stabilizing.

I do think some of Sumner’s more mechanical schemes, where a new futures contract is traded each day, and the Fed is forced by a rule to trade T-bills equal to some multiple of its net futures position, manipulation is more possible than with a looser scheme… contracts traded for a period of time and the Fed seeking to keep its net position on the contract at zero.

27. Mesa
13. August 2009 at 04:31

It’s absolutely true that there is not enough data in the GDP time series to decide what it’s properties are. So, better to think of what is actually happening. It seems reasonable to consider that large changes in output or prices create conditions favorable to their correction – ie mean reversion. Lower prices encourage consumption and/or capital disinvestment in that product or industry. Higher prices the converse. A recession creates pain and incentives for harder work and technological breakthroughs. Way above trend growth creates bubbles, waste and misallocation of resources. All these cause feedbacks act to correct the change or error. This can either be called mean reversion or perhaps another example of Le Chatelier’s principle in action. Almost all physical systems work this way to some extent.

28. ssumner
13. August 2009 at 05:08

Current, You said;

“Scott: “My intuition tells me that technological progress (not investment) drives the long run trend of RGDP”

This seems odd to me, clearly these are related. Spending on technology is investment.”

Let me give you an example of what I am thinking of, and then you tell me whether you agree or not. Suppose there is a recession and the entire chipmaking industry stops making chips and chip factories for 18 months. There are two possibilities about what would happen when they resumed:

1. They start making the same sort of chips as before.
2. The new factories produce chips twice as powerful as before (Moore’s law.)

In other words the debate is over the issue of whether Moore’s law is based on time, or the amount invested in chip making. I say time. And I think the Great Depression massively supports this perspective. Technological progress was really fast during the 1930s despite low investment, so when we started investing heavily in the private sector again after the Depression was over, the goods were far more sophisticated than in 1929. We got back onto the old trend line, instead of a newer and lower trend line. We never skipped a beat.

Current; You said;

“On a sidenote Scott wrote:
“As an analogy, Keynesians often use the C+I+G=GDP identity to illustrate the intuition behind Keynesian economics, although there is nothing really “Keynesian” about the equation.”

Actually, real GDP is quite a Keynesian idea. If we are concerned with final output then consumption is what we look at. If we are concerned with the level of investment for the future then clearly investment is important. What do we get from adding them together? We get a gross output, but why is that figure important?”

I disagree for several reasons. First of all I said GDP, which in the Keynesian model is nominal GDP, not real. Keynesian AD theory explains changes in nominal spending. Keynes said that when the economy is at capacity then increases in GDP meant higher prices.

Second, non-Keynesians such as Austrians and monetarists also use GDP. So although GDP is important to the Keynesian model, there is nothing specifically Keynesian about GDP. Indeed M*V is just another way of describing GDP, and few people would say that it is a Keynesian concept.

Mike, The question in my mind isn’t whether the backing theory or the QT are “true” in which case neither are literally true, but rather which theory is more useful. The usefulness of the QT depends on the circumstances. When money is partly or completely endogenous (as under a commodity standard) then the QT is of limited usefulness. In that case I think it makes more sense to view gold as money, and model the gold market. After all, under a gold standard the price level is nothing more than the value of gold.

In a modern fiat money regime, I believe the QT is much more useful than the backing theory. The QT does a great job of explaining the differences between inflation rates in Argentina, Brazil, Israel, the UK, the US, etc, during the high inflation 1960s-1980s. I don’t believe that “backing” adds any useful information to the current and expected future changes in the quantity of money. In my view, modern citizens really don’t care if fiat money is backed. Almost everyone I talk to thinks money is still backed up with some sort of asset like gold. They don’t realize that modern central banks can just print money and pay off the national debt.

Philo, I am not saying anything radical in a philosophical sense in my “get rid of the QT or QE” comment. Pick up any old economic textbook and you will see a different set of tools from those in a modern textbooks. For instance, many modern textbooks no longer have the “Keynesian cross” diagram. It is not a question of whether they think the graph is “true” or false, it is a question of whether they think it is useful.

1. Studies show that market manipulation is typically not a problem in prediction markets.

2. The Fed could limit the size of the position taken by each trader (I generally would oppose this unless there was good reason to be worried.)

3. The Fed could monitor the market for suspicious activities, and enter the market taking the opposite position of any trader than it felt was manipulating the market.

I might add that I think it would be incredibly hard for a trader to move the equilibrium far enough to have significant macroeconomic consequences. Thus they would almost certainly lose more in their market manipulation than they would gain in their side bets on NGDP sensitive investments. As soon as they moved the expected NGDP significantly away from the price, other traders would flood into the market seeking easy profits. Remember, there is a fixed 12 month maturity. It’s not like the Hunt brothers driving up silver prices when rational investors had no idea how long the bubble would last.

Paul, It depends what you mean by “stagflation.” Many people have in mind what we experienced in the 1970s. But that is not a problem, as NGDP grew very fast in the 1970s. I favor a slow rate of growth in NGDP. But yes you could have a touch of stagflation, say 1% RGDP growth and 4% inflation. But were that to occur, I would still view it as superior to a price level target. Remember that the long run inflation rate is EXACTLY the same under either NGDP or price level targets, but the volatility of RGDP is larger under price level targets. And RGDP volatility is much more harmful than price level volatility, as long as the average rate of inflation is low.

Greg, Perhaps I could explain the absurdity with an example. Suppose your model suggested that inflation was predictable. But suppose your model also assumed that people did not anticipate inflation. Indeed this is exactly what Keynesian models did in the 1960s. We later found out that people did indeed predict inflation, so it was a mistake to assume that people simply assumed that the price level next period would be the same as this period. On the other hand gold standard models predict that inflation is almost unforecastable. So in that model it would be a mistake to assume that people could predict inflation.

Oregonguy, Thanks, but the only way I can understand macro concepts is through examples. What information do people need to know? Do they need to know the capital of South Dakota? The birthday of the Queen of England? I’m just kidding of course. But that is my point. I have no idea what people are supposed to be “learning” in these models. Are they learning actual macro time series data, are they learning the parameters of the model itself. I’m sure it’s a mental block on my part, but I just don’t get it.

Gabriel. OK, that’s more like it. Now I think I understand.

Bill, I don’t agree that the daily trading would make manipulation more likely. Remember that the daily trades have only a small impact on future NGDP. The problem with market manipulation is that to get numbers that are MACROECONOMICALLY meaningful, you’d have to have an implausibly large amount of manipulation. And I couldn’t care less about market manipulation that doesn’t move future expected NGDP at least a few tenths of a percent away from the target.

29. StatsGuy
13. August 2009 at 05:30

Scott, it’s obvious you thought of this, but I’m slightly less confident this is enough.

1) There _have_ been allegations of people manipulating prediction markets. For example, there have been allegations (on both sides) of manipulation of InTrade futures for Obama/McCain. Some of these sound semi-credible.

2) The incentives to manipulate are huge. If a hedge fund were to own \$20 billion in bonds (or shorts on bonds, or options, or currency options, etc.), then the ability to _induce_ a change in federal policy is enormously lucrative.

3) Expecting the Fed to intervene (against the “market wisdom” no less) is even more hopeful than expecting the Fed to do the right thing without having to contravene a futures market (which, allegedly, contains the best wisdom of the crowd). The goal is to remove as much Fed discretion (and their one-sided conservativeness) as possible from the equation.

4) I think your final point – the NGDP futures market would be too big to move – may be the best chance. But in that case, the NGDP futures market would need to be REALLY BIG. It would need to act as more than a simple prediction market… It would need to be big enough to absorb hedging positions. Only in this way would it be large enough that it would be too dangerous to try to waste money to shift the prediction market in the hope of making back more money on bets in other markets.

30. StatsGuy
13. August 2009 at 05:45

Scott, it’s obvious you thought of this, but I’m slightly less confident this is enough.

1) There _have_ been allegations of people manipulating prediction markets. For example, there have been allegations (on both sides) of manipulation of InTrade futures for Obama/McCain. Some of these sound semi-credible.

2) The incentives to manipulate are huge. If a hedge fund were to own \$20 billion in bonds (or shorts on bonds, or options, or currency options, etc.), then the ability to _induce_ a change in federal policy is enormously lucrative.

3) Expecting the Fed to intervene (against the “market wisdom” no less) is even more hopeful than expecting the Fed to do the right thing without having to contravene a futures market (which, allegedly, contains the best wisdom of the crowd). The goal is to remove as much Fed discretion (and their one-sided conservativeness) as possible from the equation. By the time the Fed figured out it should have intervened, it will be too late.

4) I think your final point – the NGDP futures market would be too big to move – may be the best chance. But in that case, the NGDP futures market would need to be REALLY BIG. It would need to act as more than a simple prediction market… It would need to be big enough to absorb hedging positions. Only in this way would it be large enough that it would be too dangerous to try to waste money to shift the prediction market in the hope of making back more money on bets in other markets.

Bill:

Your argument that the looser scheme might be less vulnerable to manipulation seems to make sense. The less liquid the market, the more vulnerable it will be to manipultion.

What about a very simple scheme without any contract length… or, more accurately, an infinite contract length and a coupon that is issued periodically?

The Fed could generate a lot of income in the initial public offering, and could have a fixed schedule of IPOs (wherein it sells a steady 3% (or 5% or whatever) of futures each year to keep nominal prices of the futures contracts stable over time, and to fund the coupon.

31. StatsGuy
13. August 2009 at 05:47

er, fixed schedule of POs…

32. Jon
13. August 2009 at 05:59

Scott writes:

In other words the debate is over the issue of whether Moore’s law is based on time, or the amount invested in chip making. I say time.

I don’t see how you can distinguish the two. Time is money. R&D costs for a process generation run close to 10B. Thousands of people are deployed solving scientific and technical problems. Those people have to eat TODAY but the factories run TOMORROW. So the rate of progress is dependent upon investment levels.

So maybe you can restate your thesis. Let me put one forward to the expert (you): In an environment of massive unemployment, being employed is a significant wealth disparity. We know that ‘wealthy’ people have a higher propensity to invest. Therefore isn’t it plausible the rate of investment among ’employed’ people was higher not lower, and further that the distinctions between employed and unemployed were not entirely arbitrary but related to skill and ingenuity. In this scenario, societies excess resources are concentrated among the best, who then make substantial unit-labor productivity gains.

Here is a less incendiary version: we’re seeing the same things now. Productivity growth at firms has skyrocketed but overall investment is down. This is due laying off less efficient staff and focusing on measures that improve productivity without lots of capital–surely the best way to gear-up ROI! My suspicion is that normally the curves are individually too flat. People know how to be more efficient but they choose not to do so from the bottom up because of poor-incentives.

One of the miracles of Chinese manufacturing is genuine dedication to cost reduction. You don’t see that from your contract manufacturers here in the US.

33. StatsGuy
13. August 2009 at 07:00

“In other words the debate is over the issue of whether Moore’s law is based on time, or the amount invested in chip making. I say time.”

Isn’t the obvious response that

Technology is an output…

Time, Capital, and Labor are inputs that are subject to diminishing marginal returns?

If you consider another line of research, say silicon based solar cells, there was minimal progress made from the mid-80s through 2000/2005, and indeed much of the R&D over the last few years picked up where researchers left off in the late 70s and early 80s.

Moore’s Law works for chips because investments are large and stable. They are stable precisely because companies _know_ that there is decreasing marginal return on investments at any point in time, so that lost innovation due to failure to make investments today cannot be inexpensively made up in the future.

34. Current
13. August 2009 at 07:53

I fairly much agree with Statsguy here (it had to happen once). I’ve spent a large portion of my life in electronics development. Development does not just happen over time. It happens because huge sums of money are spent on it. It is the product of a production process.

The same sort of things apply to it as apply to other processes of production. As Statsguys says “Time, Capital, and Labor are inputs that are subject to diminishing marginal returns”. So, the capital prices and labour prices and the rate of interest all matter.

I don’t think it’s defensible to say that in the long term technology is more important than investment. Technology requires investment. We may agree that in 50 years time the reason we can maintain a better lifestyle is because of technology. This misses the point, we can create the technology because of the investment earlier. That investment was made possible by the capital we had at that time. So distinguishing in the long term what comes “from investment” and “from technology” is impossible.

Statsguy: “Moore’s Law works for chips because investments are large and stable. They are stable precisely because companies _know_ that there is decreasing marginal return on investments at any point in time, so that lost innovation due to failure to make investments today cannot be inexpensively made up in the future.”

Something like that. The problem for silicon companies is that… On the one hand, it is very much more expensive to move faster than the competitors than to move at the same speed. On the other hand, there is high demand and high profits in having the most up-to-date process. The profits in older processes are much lower. So, a decision not to keep up can be very costly and can imperil the ability of the company to invest further and catch back up.

35. Jon
13. August 2009 at 08:19

Moore’s Law works for chips because investments are large and stable. They are stable precisely because companies _know_ that there is decreasing marginal return on investments at any point in time, so that lost innovation due to failure to make investments today cannot be inexpensively made up in the future.

Moore’s Law works because of cartelization. Funds are pooled into an organization called the ITRS which establishes a multiyear roadmap and coordinates activities. The discipline of the cartelization is the 18mo-2yr period which discourages cheating. Pushing faster on your own is incredibly expensive. Only Intel with substantial market volume to itself is able to push in advance of the broad industry coordiantion.

Second, Moore’s law is not just an abstract technological advance. It is very specifically about cost reduction–cost reduction which is also used to deliver more at the same price, but critically equally delivers the same thing at half the cost. Once the technology is in place, the fixed cost of retool a Fab (~\$2B) is relatively riskless, whereas the cost of not retooling is a substantial cost-competitiveness impairment.

I don’t think there is an element of not being able to make-up the investment from the perspective of an individual firm. Yes, for the industry as a whole, it may be difficult to make up for lost years, but for a given firm? no. Its the competitive cost pressures which keep every ganged together.

See for instance page 16 of

Each line-item refers to a specific engineering challenge, each column refers to a quantitative goal. The color-coding reflects the status of the problem.

36. 123
13. August 2009 at 12:33

“The fiscal theory argued that what mattered was the total supply of government bonds, not the quantity of base money. McCallum also found these theories implausible. He argued that they either had the same predictions as a modern QT model featuring rational expectations, or they produced inferior predictions.”
If fiscal theory and QT model can have the same predictions, then it is strange that you so easily dismiss fiscal theory.

“You have seen me argue strongly for a policy of “targeting the forecast.” Bernanke and Woodford once argued that this led to a potential “indeterminacy problem.” ”
Woodford and Bernanke in 1997 paper proved that market forecast targeting is imperfect. Current crisis has proved that discretionary policy is much more imperfect. Woodford and Bernanke found flaws in EMH, but forgot to look for the flaws in EGH (efficient government hypothesis).

37. Current
13. August 2009 at 16:03

Jon is fairly much correct here. The “cartel” effect of the ITRS is overestimated though. If it paid the companies involved to go faster or slower they would. In fact they have gone slower in the past.

As he says it’s not hard for a company to “catch up”. The problem is what you lose in profit if you aren’t at the cutting edge.

Statsguy may not be wrong about this though, just out of date. A few generations ago the fab owners determined a much about how good a process was by their tweaks. Today though the folks who supply equipment to fabs are much more important. I understand that it is mostly their equipment that determines what can be acheived these days.

38. Current
13. August 2009 at 16:14

Scott: “I disagree for several reasons. First of all I said GDP, which in the Keynesian model is nominal GDP, not real. Keynesian AD theory explains changes in nominal spending. Keynes said that when the economy is at capacity then increases in GDP meant higher prices.”

I think you may be right. In Keynesian economics the nominal consumer spending added to the nominal investment spending is important.

Scott: “Second, non-Keynesians such as Austrians and monetarists also use GDP. So although GDP is important to the Keynesian model, there is nothing specifically Keynesian about GDP. Indeed M*V is just another way of describing GDP, and few people would say that it is a Keynesian concept.”

I agree to some extent. Hayek used GDP as Bill describes, in a descriptive sort of way. A price production function was made from it.

In monetarism it’s used too, but that’s quite similar to Keynesian economics. In earlier economics breaking off trade in second-hand goods and capital from other sorts of trade was quite alien. Remember the money quantity equation was MV=PT where V involved all transations using money.

I see what you and Bill are saying though.

39. Greg Ransom
13. August 2009 at 20:45

OK, let’s go back to “philosophy” and “methodology”.

How does this differ from Friedman’s position?

“For instance, many modern textbooks no longer have the “Keynesian cross” diagram. It is not a question of whether they think the graph is “true” or false, it is a question of whether they think it is useful.”

Isn’t your account of “Rorty” pretty much what is already found in Friedman?

And if they are different, it what way that matters?

40. Current
14. August 2009 at 00:46

Going back to the original point about capital and growth….

Some forms of economics that claim that recessions are caused from the demand side, as Scott points out.

However, such a recession will have supply side consequences and lasting consequences for investment. I think we have cleared up this problem about a suppose capital/technology split. Given that it seems clear to me that all trade cycle theories should predict that losses from a recession will be permanent to some degree.

41. ssumner
14. August 2009 at 03:20

Mesa, Yes, I agree that there is some mean reversion. I might add that I think this is especially true for demand shocks (i.e. where prices and output go in the same direction. Less so for productivity shocks, although even in that case there may be some mean reversion.

42. ssumner
14. August 2009 at 04:04

Statsguy, I bet on the election in the Iowa electronic market (Obama of course), and I know they limit trading to \$500. So monetary limits are a possible solution. But that is not my first choice. I am no expert, but those who are suggest that studies show manipulation of prediction markets is unlikely to be a problem. I really don’t care if a hedge fund tries to manipulate the market in order to make money on its bond portfolio, even if it succeeds. I only care if it moves the expected NGDP prices significantly far away from the target. And it is really hard to come up with a plausible scenario for that happening. Why wouldn’t other hedge funds take the opposite position? They could make even more money that the one trying to manipulate the market. I also envision the Fed subsidizing this market, so it would be very deep.

I believe Fed intervention would be less of a problem than many people think. Suppose the Fed had intervened last fall to try to hold down the monetary base because they suspected market manipulation by short sellers. I would have been thrilled—it would have been like free money for people like me who wanted to go short. After all, everyone knew that NGDP growth was going to come in way short of 5%. Once you start thinking of things that way you begin to realize that life isn’t that easy, there is no free money out there for the taking, and that’s precisely because even if the Fed’s intervention was in the wrong direction, they would be overwhelmed by traders like me trying to take advantage of any perceived profit opportunities.

Jon, Statsguy and Current. I may be wrong about chip making, but I am not convinced I am. I agree that new chips require a lot of investment, but that is a different issue from whether the long run rate of progress will be affected by a temporary recession. Suppose in 1987-88 there was a bad recession and we missed a generation. Would we still be one generation behind 22 years later? I’m not so sure.

In addition, the experience of the Great Depression suggests that technological progress is not very dependent on the rate of investment. I agree that science requires some investments, but my impression is that investments in basic science are not very cyclical. The frontier of knowledge keeps moving out in both good times and bad.

123, I view the fiscal theory as having implausible assumptions. So it’s not going to move me away from the QT unless it can produce superior predictions. In fact, I don’t think it can even produce equal predictions.

I agree about Bernanke and Woodford’s 1997 paper.

Greg, You said;

“Isn’t your account of “Rorty” pretty much what is already found in Friedman?

And if they are different, it what way that matters?”

I am basing this view on just one essay by Friedman, where he adopted a fairly dogmatic stance as to the one “right” method of testing theories. For Friedman the only question was: Did the theory predict well? Rorty seems much less dogmatic. He acknowledges that persuasion occurs in many different ways, and the best way may be dependent on the field of inquiry.

I’m not sure why the Keynesian cross was removed from textbooks, but I don’t think it was simply because it didn’t predict well.

Current, You said;

“However, such a recession will have supply side consequences and lasting consequences for investment. I think we have cleared up this problem about a suppose capital/technology split. Given that it seems clear to me that all trade cycle theories should predict that losses from a recession will be permanent to some degree.”

You may be right, but I don’t think this is obvious. Certainly mainstream Keynesian and monetarist models assume just the opposite, that losses are temporary. Of course many people understand those models as an approximation to reality.

43. StatsGuy
14. August 2009 at 06:21

I will, in the future, endeavor to avoid writing anything that Current might agree with. 🙂

Scott:

“Suppose in 1987-88 there was a bad recession and we missed a generation. Would we still be one generation behind 22 years later? I’m not so sure.”

While there is some tendency for technology to advance without any visible investment, a lot of this happens due to “invisible investment” (for example, university research), “learning by doing” effects, and reverse engineering competitor products.

Even so, visible investment is critical – if we were to stop R&D investments in a sector for 20 years, we would not lose the entire generation of R&D because there is still a level of residual invisible investment occurring (and at the margin, the value of this investment is high). But we would lose part of that generation.

Scientists continually scratch their heads when they see economists modeling technology as an “exogenous factor” that “just happens”. For predictive purposes, this often yields decent models (particularly in the short to medium term). But prediction and causation are two very separate things. If you apply that “insight” to policy, you end up gutting the national R&D infrastructure. And even many conservatives agree that basic R&D investment is one of the handful of govt. expenditures with the highest long term ROI.

Consider the following: You get frustrated when you see other economists writing models where the economy magically recovers all by itself. Why? Because if the Fed takes the view that recoveries “just happen” (without active Fed loosening), then the Fed will take a passive role, and the recovery may not happen at all.

The same is true with science investment. If everyone assumes the benefits of R&D will “just happen” because historically they have, then no one takes the steps that made R&D historically happen.

44. Don the libertarian Democrat
14. August 2009 at 07:38

Scott,

I have another question which, if you have time, I hope that you can answer:

If I believe that in Sept. 08 we avoided a Debt-Deflationary Spiral, and that we’ve had low grade Debt-Deflation since, following Fisher/Friedman, is it even possible that Monetary Policy was too loose prior to Sept. 08?

Have a nice trip,

Don

45. Current
14. August 2009 at 07:56

Scott: “Jon, Statsguy and Current. I may be wrong about chip making, but I am not convinced I am. I agree that new chips require a lot of investment, but that is a different issue from whether the long run rate of progress will be affected by a temporary recession. Suppose in 1987-88 there was a bad recession and we missed a generation. Would we still be one generation behind 22 years later? I’m not so sure.

In addition, the experience of the Great Depression suggests that technological progress is not very dependent on the rate of investment. I agree that science requires some investments, but my impression is that investments in basic science are not very cyclical. The frontier of knowledge keeps moving out in both good times and bad.”

I don’t think you understand quite what we are saying.

I agree that investment in basic science isn’t very cyclical. But basic science isn’t all there is too it. Technology requires a whole lot more investment. Even if basic science research is funded by government it may be cyclical. Even if it is funded by government that portion of basic science it developed in the private sector may be cyclical. But, most importantly, engineering and technology take place largely in the private sector and they are very cyclical.

Now, put yourself in the shoes of the owner of a technology business. Broadly speaking the staff of that business can be used for research tasks or on other sorts of tasks. Management can decide on an split, what I’m discussing here is a simple production-possibilities frontier situation. It isn’t obvious what split will be chosen. Research will not necessarily yield more than other activities. (Research isn’t necessarily even a long term activity it may yield a return faster than building a shopping arcade would for example.)

We are talking about what happens when investment is cut. Since we don’t know about the particular form of the split I mention above we can’t tell if research will be cut or not. It may have been other capital spending that was cut in the Great Depression, that doesn’t mean it will be next time.

In fact we can’t even tell if research will be more useful to the future than making other capital goods would be. Marginal research may well be less useful than a building that lasts a hundred years, for example.

Scott: You may be right, but I don’t think this is obvious. Certainly mainstream Keynesian and monetarist models assume just the opposite, that losses are temporary. Of course many people understand those models as an approximation to reality.

Yes. I don’t really understand how they justify that. In both involuntary unemployment is something avoidable. Doesn’t that clearly mean that it is beneficial for society in the long term if it is avoided?

46. George Selgin
14. August 2009 at 08:53

Scott: I share you’re fondness for Ben McCallum’s work. I had long hoped for the appearance of a graduate text combining his various (terrific) Carnegie-Rochester papers, along with his writings on monetary rules, and was sorry that the monetary textbook book he eventually did write was something rather different.

Indeed, the more I read your blog, the more I’m inclined to think that the only thing separating our macroeconomic ideas is about 4.5 percentage points of nominal GDP growth!

47. Mike Sproul
14. August 2009 at 14:04

Scott: You said “Mike, The question in my mind isn’t whether the backing theory or the QT are “true” in which case neither are literally true, but rather which theory is more useful. The usefulness of the QT depends on the circumstances. When money is partly or completely endogenous (as under a commodity standard) then the QT is of limited usefulness. In that case I think it makes more sense to view gold as money, and model the gold market. After all, under a gold standard the price level is nothing more than the value of gold.

In a modern fiat money regime, I believe the QT is much more useful than the backing theory. The QT does a great job of explaining the differences between inflation rates in Argentina, Brazil, Israel, the UK, the US, etc, during the high inflation 1960s-1980s. I don’t believe that “backing” adds any useful information to the current and expected future changes in the quantity of money. In my view, modern citizens really don’t care if fiat money is backed. Almost everyone I talk to thinks money is still backed up with some sort of asset like gold. They don’t realize that modern central banks can just print money and pay off the national debt.”

1) I’d like to think that when it comes to theories, true=useful.
2) You seem to agree that when money is convertible, the backing theory is applicable and the quantity theory is not. But as I’ve said, convertibility can be instant or delayed, physical or financial, at the customer’s option or the bank’s option, etc. The US dollar can be described as convertible, just not instantly physically convertible at the customer’s option.
3) One example where the backing theory fits the data better: When central banks used to try to support the value of their currencies in international markets, they would use some foreign currency to buy up their own currency. For example, if the pound is worth \$2, and the Bank of England wanted to push that up to \$2.01, the Bank would start paying \$2.01 per pound in the market. According to the backing theory, this will fail, because the Bank loses \$.01 on each purchase, and the resulting loss of assets drives the pound down, not up. Dean Taylor’s ‘Bet Against the Central Bank’ paper showed that this policy always failed. That result supports the backing theory and refutes the quantity theory.
4) Printing money doesn’t pay off the national debt. It just replaces one government liability (bonds) with another (paper dollars).

48. ssumner
16. August 2009 at 19:04

Statsguy, You said,

“While there is some tendency for technology to advance without any visible investment, a lot of this happens due to “invisible investment” (for example, university research), “learning by doing” effects, and reverse engineering competitor products.”

I suppose this is part of what I was thinking of. In addition to the Alexander Field article on the Great Depression (he argued technology expanded faster than any other decade of the 20th century) I might have also been thinking in an international context—the way a country like Japan can sort of “catch up” through fast growth as long as another country (say the US), was continuing to advance while Japan was set back by war. On the other hand the current recssion is worldwide, so the German and Japan after WWII example may not be appropriate. Overall you make a lot of good points, and I’d say I have an open mind on this issue, with my hunch being that the truth is somewhere between the 2 extremes.

Don,

I don’t see money as being too loose then because I focus on NGDP growth, which was subpar even before September. I have acknowledged that the Fed faced public pressure due to high oil prices. And I understand why at first they were reluctant to ease. But they then needed to do a quick about face, and they didn’t.

I’m not sure I understood your question totally, so I’ll try to revisit this if you want to clarify it a bit.

Current, Also check out my answer to statsguy. I have an open mind on this. One issue that I struggle with is if the basic science keeps chugging along, then in a sense if we skip a generation we have fallen further behind the scientific frontier. I don’t think you catch up right away, but perhps over several decades. Again, I’m not sure either way, but you can’t just look at a real world example of a high tech firm’s behavior over a few years, and necessarily answer the deeper question. Growth is still something of a mystery.

I agree that those natural rate models are questionable. But they still may be servicable for modelling business cycles.

Thanks George. I look forward to discussing the NGDP growth rate issue in November. I was also a bit disappointed with McCallum’s textbook. Too much IS-LM for my taste. I guess he is better at doing scholarly articles. I’m guessing you have benefited as I have from some of his explanations of cutting edge technical work in macro. He has reassured me that I am not missing all that much, although that may not have been his intent.

Mike; You said;

“1) Id like to think that when it comes to theories, true=useful.
2) You seem to agree that when money is convertible, the backing theory is applicable and the quantity theory is not.”

I meant that neither theory is literally true, in that it can perfectly explain reality. There are so many ceteris paribus issues to deal with. Even a gold standard will depend on how credible the peg is, which can be partly a psychological issue–how determined are the authorities to keep the peg? Because expectations play such an important role in both models, it’s often unclear exactly what one is debating. If you say “suppose the government does this” and expect that there is a clearcut answer as to which theory is more “true” you are wrong. After the government takes some policy step its effect will partly depend on how it affects the future path of other government steps.

I suppose you could say that the backing theory applies to the gold standard. But in my Great Depression research I had a lot of success ignoring the backing theory and just treating gold as if it was “money.” I modeled the price level as a function of the supply and demand for gold.

Regarding your third point. It doesn’t always fail as China proved in 2005 when it started gradually rasing the value of its currency.

Regarding your fourth point. Dollar bills are not debt. The government is not required to redeem them for anything but themselves. That’s not debt. After my 30 year mortgage is due I can’t go to the bank and pay it off by issuing another 30 mortgage. The bank must agree. Not so for Federal Reserve notes.

49. Mike Sproul
17. August 2009 at 10:35

Scott:
“I had a lot of success ignoring the backing theory and just treating gold as if it was “money.” I modeled the price level as a function of the supply and demand for gold.”

You weren’t ignoring the backing theory. You were using it in an appropriate way.

“Dollar bills are not debt. The government is not required to redeem them for anything but themselves. That’s not debt. After my 30 year mortgage is due I can’t go to the bank and pay it off by issuing another 30 mortgage. The bank must agree. Not so for Federal Reserve notes.”

This is the real sticking point, and I want to focus on it. It is a very critical accounting error.

1) Dollars are just as much debt as government bonds are. If a bond comes due, the government re-issues it. If the government is financially sound, it will be able to sell its new bonds (or its newly-printed dollars) for a high price. If the government is stretched, it will be able to sell its bonds (or dollars) for a low price. But there will always be willing buyers (i.e., lenders) for either.
I can do the same thing, by the way. I can buy a loaf of bread from my local grocer by handing him my IOU. He accepts it because he has employees who rent a house from me, and they accept ‘mike dollars’ in wages, since I accept them for rent. I can also buy another house by issuing a \$400,000 bond to a local mortgage broker. He will accept the bond because I give him a lien on the house. Finally, I can issue new mike dollars and use them to buy back part of my own mortgage bond. The mike dollars just replace the mike bond on the liability side of my personal balance sheet.

2) I’ve mentioned this before, but what if the demand for money falls? Everyone would agree that the government would normally respond by buying back some of the outstanding dollars, in order to keep their value stable. The fact that the government bought them back means that those dollars were a debt of the government all along. If the government wouldn’t buy them back, then of course the dollars would lose value. But if everyone knew that the government wouldn’t buy them back, nobody would value them in the first place. In other words, nobody would value the dollars unless they were the government’s debt.

50. Current
18. August 2009 at 01:11

Statsguy: “While there is some tendency for technology to advance without any visible investment, a lot of this happens due to ‘invisible investment’ (for example, university research), ‘learning by doing’ effects, and reverse engineering competitor products.”

Scott: “I suppose this is part of what I was thinking of. In addition to the Alexander Field article on the Great Depression (he argued technology expanded faster than any other decade of the 20th century) I might have also been thinking in an international context””the way a country like Japan can sort of ‘catch up’ through fast growth as long as another country (say the US), was continuing to advance while Japan was set back by war. On the other hand the current recssion is worldwide, so the German and Japan after WWII example may not be appropriate. Overall you make a lot of good points, and I’d say I have an open mind on this issue, with my hunch being that the truth is somewhere between the 2 extremes.”

Scott: “Current, Also check out my answer to statsguy. I have an open mind on this. One issue that I struggle with is if the basic science keeps chugging along, then in a sense if we skip a generation we have fallen further behind the scientific frontier. I don’t think you catch up right away, but perhps over several decades. Again, I’m not sure either way, but you can’t just look at a real world example of a high tech firm’s behavior over a few years, and necessarily answer the deeper question. Growth is still something of a mystery.”

What I’m saying is definitely not intended to be tied to recent behaviour of tech firms.

I’m pointing out the long running characteristics of the activity. That includes the trade off between investment in research and that in other things. At no point in history has it been certain that this breakdown will always be a certain way.

Also, it is true always and everywhere that research and development cost money. Newton, for example, was funded by his own personal wealth.

Bohm-Bawerk looked at machines like this…. Suppose we have a machine that can produce 100 widgets with 10 hours of labour. Making 100 widgets manually would require 200 hours of labour. Suppose that the machine makes a widget for £1 and sell it for £2. The machines lasts 10 years. In that time it can make £36500 for it’s owner. Due to time-preference the machine must cost substantially less.

Now, compare the situation where the machine exists to that where it doesn’t. Clearly the gain that comes from the machine is not illusory, without it fewer goods could be produced.

Time-preference does not directly affect the long-run gains of the community. The widgets don’t get less useful because of it. So, the long-term benefits and the short-term ones are not firmly related. If another type of machine could make widgets for 5 years it would be worth less, but not half as much as the original machine. Time-preference would make it worth more than half as much. Other factors may make it worthwhile to make the first machine or the second, we can’t say with any certainty.

Now, everything I’ve said above that applies to machines also applies to technology itself. I’ve known technologies that have become useless in less than 5 years. That need not happen by a technology being superceded, something else may change.

Both investments and technology work in similar ways. They raise the amount of services we can obtain from the permanent resources available. Hence both can provide a permanent rise in the level of income. The fact that both have finite lives doesn’t affect this. It is the services they give within their finite lives that are important.

Regarding “catch up”. I agree with you that we would not remain one year behind if we lost a year. Similarly, though if a building was burnt down in 1959 we would not necessarily be one building down by 2009. In both cases though the costs of the loss will be spread across the economy.

51. Current
18. August 2009 at 01:29

I have a feeling after that post that someone will point out that “basic research” is different. It is since it gives permanent services to humanity, not non-permanent ones. It’s like the sun (and may be more permanent than that 😉 ).

This doesn’t mean though that investment in basic research will always yield more to society than investment elsewhere. Basic research itself uses all of the services of the rest of society. High-technology, for example, may be transient, but it is very useful in doing basic research.

The very elaborate and expensive particle accelerators that have been made in the past few decades are not only the outcomes of basic research. The technology needed to construct them came from many sources. Without that technology much basic research would be much slower.

52. Don the libertarian Democrat
18. August 2009 at 10:41

Scott,

I asked Nick the same confusing question. Here’s our exchange:

“Don: “If I believe that in Sept. 08 we avoided a Debt-Deflationary Spiral, and that we’ve had low grade Debt-Deflation since, following Fisher/Friedman, is it even possible that Monetary Policy was too loose prior to Sept. 08?”

That’s a strange question. If you had said “too tight” instead of “too loose” I would have found it easier to understand.

Since we want to avoid debt-deflation, and if you believe we have had low grade debt-deflation since Sept 08, then it follows almost by definition that monetary policy was too tight. (OK, it’s not really by definition, since the economists’ view that tighter money causes lower inflation and looser money higher inflation is built in.)”

“You answered my question, even if I phrased it poorly. I was trying to say that monetary policy must have been too tight, and so couldn’t have been too loose. I hope that I make more sense now.”

“Don: Yep, that makes sense.

I expect that someone who disagreed could say that monetary policy was already as loose as it could be, with 0% interest rates. Or could say that it was looser than “normal” as measured by interest rates (to which Scott would reply “But interest rates are a poor measure, and what matters is not relative to ‘normal’, whatever that might mean, but relative to what is needed to keep NGDP growing at 5%!”)”

Here’s an interesting post:

http://modeledbehavior.com/2009/08/17/monetary-policy-in-the-aftermath-the-case-for-a-5-inflation-target/

Take care and have fun,

Don

53. ssumner
19. August 2009 at 02:37

Mike, Suppose you were told that that the demand for cash would drop 20% every single year, and the government was going to do nothing about it. (Implying 20% inflation.) I say people would still hold cash, you seem to suggest they wouldn’t.

Current, You said;

“Also, it is true always and everywhere that research and development cost money. Newton, for example, was funded by his own personal wealth.”

But I think that basic R&D may continue at a fairly high rate during recessions, even if temporary. Why? Because when the recession is over you want to be first out of the gate with a state of art new computer chip.

I’m no expert in growth theory, but I vaguely recall that the “new growth theory” (Paul Romer) emphasized that technology is not just like investment. Technology has a time component that investment lacks. By that I mean if you spend 10 times as much on building a machine, you can build 10 times as many. But if you spend 10 times as much on R&D there are diminishing returns. You can’t just manufacture new techology at will, if the basic science isn’t there yet.

Don, Thanks, but I couldn’t open the post. There’s lots of stuff I can’t open here, like marginalrevolution.com and Mankiw’s blog. But I do get Krugman’s. I don’t know if it is my computer (DSL line) of the Chinese government likes American leftists more that rightwingers. (I’m kidding, Mankiws blog is carried here in Chinese.)

54. Current
19. August 2009 at 03:14

Scott: “But I think that basic R&D may continue at a fairly high rate during recessions, even if temporary. Why? Because when the recession is over you want to be first out of the gate with a state of art new computer chip.”

I agree. I myself have used this argument to defend my own job.

Scott: I’m no expert in growth theory, but I vaguely recall that the “new growth theory” (Paul Romer) emphasized that technology is not just like investment. Technology has a time component that investment lacks.

Investment has a time component too though. If I want a factory built then that is going to take time. More workers can be employed at higher cost. However there is still a minimum time needed. Employing more workers will not lead to a proportionate increase in productivity. It will lead to a less than proportionate increase.

Scott: “By that I mean if you spend 10 times as much on building a machine, you can build 10 times as many. But if you spend 10 times as much on R&D there are diminishing returns.”

Yes, definitely.

Scott: You can’t just manufacture new techology at will, if the basic science isn’t there yet.

But the same is true of the pre-requisites for any other sort of production. You can’t build a brick wall unless there are bricks ready. I argue with my boss “you may as well employ me because by the outputs of my work are ready we’ll be in a boom”. Some other people in time consuming capital manufacturing are probably in the same situation. Those who sell vines to vineyards for example, or those who sell brick furnaces.

I broadly agree with what you’re saying. But I don’t think it gets you to the conclusions you are making above. If you want to claim that we will catch up it’s not just necessary to show that much technology research continues in a recession. You must also show that the rest of capital accumulation isn’t that useful. Now it may be that it isn’t that useful in the long term. But it is impossible to clearly tell that, the economy is too complex.

55. Mike Sproul
21. August 2009 at 08:36

Scott:
“Suppose you were told that that the demand for cash would drop 20% every single year, and the government was going to do nothing about it. (Implying 20% inflation.) I say people would still hold cash, you seem to suggest they wouldn’t.”

Since I believe the dollar is backed by the Fed’s assets, I’d say that the 20% reduction in demand would be offset by the Fed buying back 20% of its outstanding cash at par, thus implying no inflation. But if, as you say, the government will do nothing about it, that announcement would be equivalent to a complete loss of backing, so the dollar would immediately become worthless.

On backing theory principles, one way that the dollar could lose 20% of its value each year would be if the fed lost 20% of its assets each year, and reduced the convertibility peg by 20% as well. If this happened, some people would still hold dollars, just less than before.

56. Current
21. August 2009 at 09:03

Mike Sproul,

Do you have an answer to me in the “Princeton rules the World” thread?

http://blogsandwikis.bentley.edu/themoneyillusion/?p=1791

57. ssumner
23. August 2009 at 22:58

Current. There’s probably not much more I can say here, as we either agree or the issues are too complex to resolve.

Mike, I have no idea why the dollar would become worthless if demand fell 20% and the Fed didn’t reduce the number of dollars in ciculation. People in many countries gladly held cash which was depreciating at 20% per year for decades. Why did they do so?

58. Mike Sproul
25. August 2009 at 18:42

Scott:

If some bank had issued 100 shillings, backed by 10 oz. of silver plus bonds worth 90 oz., then suppose, as you say, that the public demand for shillings fell from 100 to 80. On backing theory principles, the bank should buy back 20 shillings with 20 oz. of its bonds. That would leave the bank with 80 oz of assets backing 80 shillings, so each shilling is still worth 1 oz. That’s the backing theory view.

Now you’re asking what happens if the bank refuses to buy back the 20 unwanted shillings, or any future unwanted shillings either. That is tantamount to an announcement by the bank that the 100 shillings are no longer backed by the bank’s 100 oz. worth of assets. The bank has removed all backing for the shillings, so they lose all value. Again, that’s the backing theory view.

But then you look at it from a quantity theory view, implicitly assuming that those shillings were unbacked all along, and it seems natural to you that a 20% fall in demand would cause 20% inflation. You are not giving the logic of the backing theory a chance to work.

On backing theory principles, one way that the dollar could lose 20% of its value each year would be if the fed lost 20% of its assets each year, and reduced the convertibility peg by 20% as well. If this happened, some people would still hold dollars, just less than before.

59. ssumner
27. August 2009 at 02:19

Mike, You said;

“On backing theory principles, one way that the dollar could lose 20% of its value each year would be if the fed lost 20% of its assets each year, and reduced the convertibility peg by 20% as well. If this happened, some people would still hold dollars, just less than before.”

In standard monetary theory it doesn’t matter whether the backing was reduced at all. Suppose the public’s demand for cash falls by 20% (perhaps due to the spread of ATMs). And suppose the amount of cash in circulation stays the same, despite this drop in cash demand. Then standard monetary theory says prices will rise 20%. If the Fed has maintained the same level of assets as before, then I presume backing theory says the price level doesn’t rise at all. So who is right? My hunch is that the standard theory is right. But it is hard to find perfect experiments.

60. Mike Sproul
28. August 2009 at 07:31

Scott:

That assumes that people don’t look forward. If people see that the Fed failed to soak up the unwanted 20%, they will expect more of the same in the future, so the value of money will fall by more than 20%, even on quantity theory principles. A few quantity theorists might even expect the money to lose all value, especially if some superior foreign money is available.

Of course, the fed would normally use its bonds to soak up the unwanted 20%, so both the quantity theory and the backing theory would say that in this case the value of the dollar would not change. But that brings us back to a point I made earlier (which you did not answer): What if the fed has no assets with which to buy back the dollars? You’ve said several times that the fed’s assets don’t matter at all, but now you’d have to admit that the fed’s assets do matter. If the fed has assets, and uses them to buy back the unwanted 20%, then you’d say that the dollar won’t lose value. But if the fed has no assets, and can’t buy back the unwanted 20%, you’ve already said that the dollar would fall by at least 20%, although you’d probably admit that if people are forward-looking, the fall would be more than 20%.

The backing theory gives coherent, consistent answers to all these points, while the quantity theory just becomes more and more of a tangled web.

61. Current
28. August 2009 at 09:10

Mike, I see what you mean about Quantity theory being more complicated.

However, I think there are complications with backing theory too. For example, why is value of Fed assets not printed in the newspaper sections on finance? Surely it is the most important piece of information?

Also, we all know that if the Fed’s assets run out the government will bail it out. Really the Fed/State’s assets are whatever they can realistically tax from the populace.

Do you agree with what I’ve said in the “Princeton rules the world” thread?

62. Mike Sproul
28. August 2009 at 15:21

Current:

2 reasons I can think of for the lack of publication:

1) Everybody’s a quantity theorist, so they don’t realize that fed assets matter.
2) As you said, the fed’s assets aren’t the end of the story, since the government might bail it out. So the thing that really determines the value of the dollar is the government’s assets. But that’s not the end of it, since the government can, if it wants, take all the private wealth in the country. To the extent that’s true, the dollar is actually backed by the wealth of the entire US. Of course, the US could, if it wanted, steal all the world’s wealth, so…

Sorry I’m getting a little slow with replies. Things are piling up at work.

63. ssumner
29. August 2009 at 20:09

Mike, Two points;

1. When I said demand fell 20% I implicitly assumed that included the effects of expectations, otherwise demand would have fallen by more than 20%. Thus I assumed a one-time change.

2. The quantity theory has no problem with a situation where the central bank lacks funds to prevent inflation. That is the hyperinflation scenario well-described by the modern QT. In that case the government’s budget constraint simply becomes a factor influencing the expected future path of M.

64. Mike Sproul
30. August 2009 at 18:53

Scott:

But you’ve been claiming that backing is irrelevant. Now it sounds like you’re saying that if the fed has the funds to buy back (20% of) its dollars, the 20% fall in money demand won’t cause inflation, whereas if the fed lacks the funds, the 20% fall in money demand will cause inflation. That sounds like you’re saying that backing matters.

65. ssumner
2. September 2009 at 15:20

Mike, I have always agreed that the fiscal theory is relevant for the case where the government must print money to pay its bills. My argument is that this factor does not constrain the monetary policy of most of the world’s major central banks. The fiscal theory was very relevant to Germany circa 1920, and explains why the Reichbank printed so much money.

66. Mike Sproul
6. September 2009 at 09:06

Scott:

The fiscal theory is completely different from the backing theory. The fiscal theory says that Germany’s financial trouble ‘forced’ them to print more money, and that the resulting inflation was caused by ‘more money chasing the same goods’. The fiscal theory is just a special case of the quantity theory.

The backing theory says that the resulting inflation was caused by the big increase in Germany’s liabilities (money+bonds), combined with a drop in Germany’s assets (taxes receivable).

That still leaves the question:

“Now it sounds like you’re saying that if the fed has the funds to buy back (20% of) its dollars, the 20% fall in money demand won’t cause inflation, whereas if the fed lacks the funds, the 20% fall in money demand will cause inflation. That sounds like you’re saying that backing matters.”

67. ssumner
7. September 2009 at 17:48

Mike, I am not saying it won’t cause inflation if they have the proper backing, I am saying it might not. It will still cause inflation if they don’t reduce the money supply.

Sorry for the fiscal/backing mixup. Are you sure the fiscal theory is just a special case of the QT? I thought it’s proponents thought it contradicted the QT.

68. Mike Sproul
14. September 2009 at 08:57

Scott:
I expect there are many versions of the fiscal theory. I was only referring to the version that you seem to believe: that a government at the end of its rope is forced to print more money, and it is the rise in the quantity of money relative to goods that causes inflation.

Suppose the fed has issued a total of \$100, and holds various assets worth 100 oz. of silver. Each dollar is worth 1 oz. If money demand falls by 20%, and the fed uses 20 oz. worth of its assets to buy back \$20, then both the quantity theory and the backing theory imply no inflation. If the fed refuses to buy back the unwanted \$20, then both theories imply 20% inflation. The backing theory would view the refusal as an effective default by the fed on 20% of its obligations.

But what if the fed did something crazy, like buying back the \$20 in exchange for the full 100 oz. worth of its assets? The backing theory says the 80 dollars left in public hands would lose all value, since there is nothing backing them and the fed will never be able to buy back the \$80 remaining in public hands. The quantity theory, as I read it, says there would be no inflation because the value of the dollar is maintained not by backing, but by the limitation of the supply of dollars, combined with the existence of a demand for them.

But of course, the fed’s complete loss of assets means that the fed will never be able to buy back any dollars in the future, should demand for dollars fall any more. I think any quantity theorist would have to admit that the fed’s decision to throw away its assets would be inflationary. In other words, quantity theorists are forced to admit that backing matters. But if that’s the case, why not just accept the backing theory in the first place?

69. ssumner
16. September 2009 at 04:25

Mike, I think the truth is somewhere in between. In the case you cite the value of money would not fall to zero, so I think the backing theory is wrong. At the same time prices would rise, so I think the crude QT is also wrong. I think the lack of backing would change the expected future path of the money supply. It would remove the possibility of further decreases in M as money demand fell. This is logically equivalent to an increase in the expected future money supply (by removing one tail of the distribution.) The sophisticated QT says expected future changes in M also affect the current price level.

So I’d say both the backing and simple QT are wrong, but the sophisticated QT is true.

70. mbk
16. September 2009 at 18:27

Taking a leaf from the Austrian book, in some regard. I am still trying to forge a useful model for myself of how money works, even in a simplified way.

If money was any other good, then its “tightness” (demand vs. supply situation) would be measured by a rising price, maybe stabilizing at a high equilibrium price. And “loose” money would mean, money demand falling behind money supply, which means the price of money would be falling as well, and maybe stabilize at a low equilibrium price.

To me there are two ways of looking at the price of money – one, its exchange rate with goods (changes in the “price of money” are what most people, though not say, Mises, would call inflation), and two, the price you pay for renting it (interest). Rent and exchange price of money are linked, but in that network of links there are several other parameters so the link is not direct. In that network of valuation we have at least, quantity of money now, expected quantity of money in the future, quantity of exchangeable goods for money now, quantity of exchangeable goods for money expected in the future, demand for money now, and demand for money expected in the future. All of these are interdependent with the two possible prices of money, rent and echange value for goods. And this interdependence cuts both ways.

Keeping with the rental price: if interest rates are the price for renting money then they depend on both supply and demand. Setting interest rates then amounts to price fixing. And as in all acts of price fixing, the result can be scarcity (tight money) or overabundance (loose money). So, interest rates will make money loose or tight, but not as many assume, in their absolute value, but only in their difference to the “natural” interest rate. (empirically, even now, similarly industrialized countries have wildly different absolute value interest rates).

So if money was just an ordinary good, a good for rent, but subject to price fixing, then by demand and supply laws, if interest rates were fixed higher than what the natural interest rate woud have been, money would effectively be loose. And if interest rates were set lower than what the natural interest rate woud have been, money would effectively be tight. Regarding Friedman then, low interest rates in some situations might actually _cause_ tight money.

So out of 3 parameters (money demand, money supply, money price) you need 2 to determine the third in the purely monetary realm. Crucially, whether high interest rates really mean tight money or not depends on where demand vs. supply stands, taken now as actual (effective) demand by individuals and buinesses to rent money vs actual (effective) supply willingly offered as loans by financial institutions. How is effective money supply related to the total quantity of money in the economy? Well, through interest rates, assuming fixed demand: I can easily imagine a situation where actual demand is acutely high, but price fixing towards a low interest interest rate leads to artificial scarcity, i.e. low _effective supply_ being offered (as loans): hoarding by banks. Regarding Friedman then, low interest rates in some situations might actually _cause_ tight money.

Ultimately of course the real world comes in with the real exchange rate of money vs. goods. I could well imagine that the real world quantity of goods mechanically reacts to monetary distortions. A key problem is that (loosely tying in Mises with Sumner) interest rates are in themselves forward looking because their value depends at least in part on expectations for the future.

The problem with any policy here, and beyond the problem of fixing what should be natural prices, is the multiple interacting variables. We have quantity of goods, now and expeced future, quantity of money, now and expected future, and interest, both real and fixed. So we are now in a complexity situation where none of the nice intersecting trend lines of economy text books would lead to some kind of equilibrium of anything – more likely, a host of local suboptimal maxima, but on the bright side, probably a lot of different ways to achieve a similarly positive outcome.

I hope this makes at least partial sense to someone here.

71. ssumner
17. September 2009 at 17:52

mbk, Much of what you say is right. but not this:

“Keeping with the rental price: if interest rates are the price for renting money then they depend on both supply and demand. Setting interest rates then amounts to price fixing. And as in all acts of price fixing, the result can be scarcity (tight money) or overabundance (loose money).”

The Fed doesn’t fix rates, rates are determined by the market. The Fed targets rates by adjusting the money supply. So unlike with ordinary price controls, the money market is not out of equilibrium. S=D.

You are right about the complexity of the problem however. Expectations of the future are key, and add an extra layer of complexity.

72. TheMoneyIllusion » Laidler on Friedman’s puzzling views on the Phillips Curve
5. November 2010 at 15:27

[…] a number of famous monetarists, particularly Bennett McCallum, who I described last year as my favorite monetary economist.  One of the conference readings was written by David Laidler, who is an outstanding monetary […]

73. (MV=PQ) â‰  QTM | It Don't Mean Much, These Seats are Cheap.
10. March 2013 at 17:23

[…] the quantity theory. In many places, the equation is confused to represent the theory, but that is wrong. But first, a little […]