Tuesday, February 3, 2009

Money Supply and Inflation: An Inverse Relationship?

Monetarism is a beautiful theory.  Economics is full of such beautiful theories, like Say's Law, Ricardian equivalence, Keynesian fiscal stimulus, and so forth.  These theories are generally internally consistent, predictive, and elegant.  They are also all at odds with empirical results.  That has led to the discard of some, while other ideas linger anyway.

I regularly hear that our expedition into the lands of quantitative easing and monetization will inevitably cause a rise in the dollar value of goods and services.  That makes a lot of sense intuitively: more dollars chasing fewer things.  It's certainly true in extremis.  But outside of that extreme, there's a wide variety in behavior.

Multiple authors have found that for US data series prior to 1961, there is, as expected, a strong positive correlation between M2 growth and inflation.  But after that, the relationship broke down, and an even stronger negative correlation emerged.


This breakdown makes little sense with money supply held as an independent variable.  We would expect more money to consistently lead to higher prices modulo GDP growth, by definition, but the opposite has clearly happened for half a century.

A strong inverse correlation is still correlation, so there's probably a good reason.

Perhaps we could invert the causality to make money supply a function of the inflation rate, and hence nominal interest rates.  A fall in interest rates would lead to a rise in credit outstanding as entities sought to maintain financial obligation ratios.  This would make sense in a world with a great deal of callable debt, e.g. mortgages.

This could also be a result of financial deepening and counter-cyclical monetary policy.  Agents would seek to lever up when inflation and interest rates are low, typically during recessions, anticipating future rises.  During a boom, higher inflation and interest rates deter borrowing in anticipation of rate cuts later.

These ideas are just stabs into a nebulous dark thus far, and I'd appreciate reader thoughts.

Does forcible money creation exhibit a different relationship?  Theoretically, to the extent that people believe quantitative easing will cause inflation during a liquidity trap, it could.  In practice, the inverse relationship seems to linger.  We might also see the inverse relationship persist, or suffer a strong snap back to monetaristic normalcy.  As M2 growth rates stab to new highs, the response in inflation and interest rates is unpredictable.

28 comments:

babar ganesh said...

i can't see the graph well enough -- can you point me to a bigger version?

JKH said...

Experiment development:

Milton Friedman may have been a great economist, but the bromide “inflation is always and everywhere a monetary phenomenon” (if that’s the correct quote), is in and of itself, profoundly useless.

First, in a similar way, breathing is always and everywhere an atmospheric phenomenon. But that doesn’t tell me a whole lot.

Second, to the degree it’s supposed to refer to the quantity of money, it ignores the equally important variable of the velocity of money.

Third, both quantity and velocity are plug items for a GDP monetary equation, when taken as functions of each other’s assumed level. There is no reason to assume that either quantity or velocity is the relatively more important variable.

Fourth, theoretical definitions of money are hopelessly detached from the reality of money complexity and its continously differentiated forms. This compounds the futility of centering the explanation of inflation dynamic on the basis of either the quantity or the velocity of money.

Fifth, the notion of the “money multiplier” in banking is a completely fraudulent explanation of the way the monetary system expands itself. Central banks actively control the short term interest rate, while passively supplying the reserves that the banking system requires in order to meet its statutory obligations as determined by deposit liability growth. Reserves follow deposits; not vice versa. How could any quantity theory of money be helpful when this basic fact of banking is so widely misunderstood?

Sixth, the quantity and velocity of money is dissipated not only through GDP but through non-GDP asset liability transactions. How can a quantity theory that plugs into a GDP equation but ignores the asset economy effect ever hope to explain anything usefully?

Sorry, I’m just in a sceptical mood about the usefulness of economic theory in general ... I think I’ll stop there.

JKH said...

I'll correct one thing from above.

Velocity is more important than quantity. And it's typically ignored. Also, velocity is fundamental to the notion of a liquidity trap, which is the problem we face now.

donebenson said...

There is a narrowness to the standard understanding of inflation, which I think is incorrect.

Richard Skinner in 1937 wrote a wonderful book entitled "Seven Kinds of Inflation." In it he described that inflation does not just show up in the standard measures of CPI. It also is very much in evidence in those 'bubbles' that form in stocks, real estate, commodities, and credit [even though the CPI figures are quiescence]. So, while in the 1920's and in most of the period after 2000, CPI inflation was modest, stocks, credit, real estate, and commodities showed major inflationary gains.

Skinner's book was instrumental for the original editors of The Bank Credit Analyst in developing their analytical ideas, and the concepts are used by equity strategists like Marc Faber, David Roche, and others.

If we think that inflation only shows up in the CPI figures, we will miss [or not understand] many major moves in other asset classes.

ndk said...

i can't see the graph well enough -- can you point me to a bigger version?

Sorry, Babar, still trying to figure out how to make Blogger go. I put one up on my personal site, and amended the post.

The graphs are most revealing in the tables and panels of the paper by Shelley and Wallace. I suck at pictures.

If we think that inflation only shows up in the CPI figures, we will miss [or not understand] many major moves in other asset classes.

I think that's a very good point, donebenson. I personally think we can be successful in driving inflation through credit creation, but it won't necessarily be in CPI or wages. It'll be in that with closer ties to the financial system, such as assets, commodities with futures, and luxury goods and services.

That kind of inflation is pretty much what we had over the last five or ten years, and I suspect that without any real trade wars, any resurgent money growth will float there again, as wage pressures are capped by international competition.

Sorry, I’m just in a sceptical mood about the usefulness of economic theory in general ... I think I’ll stop there.

I feel the same way, JKH, if the first paragraph of my post didn't make that clear. I try to rely heavily on empirical evidence for that reason. We're performing advanced surgery on the patient without even understanding where food goes in, much less where it comes out. But I still think attempting to understand and model it is critical, because eventually, we might learn something useful.

Sixth, the quantity and velocity of money is dissipated not only through GDP but through non-GDP asset liability transactions. How can a quantity theory that plugs into a GDP equation but ignores the asset economy effect ever hope to explain anything usefully?

I strongly believe asset prices and total credit levels relative to production need to be factored into future economic theory. If we don't learn that from the last 10 years, we're truly ideologically captive.

S said...

In short: Federal Reserve rate manipulation + wage labor arbitrage = artificial capital repatriation (rinse wash repeat). Inflation was reversed it was bottled up and exported, at the moment. Old relationships


1961 began the foray into Vietnam and represented the front end of the great Society to come...
...Gold Window tossed out thereafter...
...Federal Reserve embarks on decade long exercise in manipulating the yield curve - with 1987 marking the watershed...
...US engages China in a cold war ploy to profit off the sino soviet split..
...with unintended consequence of a massive wage labor arbitrage...
...Japan employs mercantilist policies to drive export led economy and is rumored to take over the USA...
Rate laundering scheme firmly entrenched....
Law of gravity reversed in housing, equities and other...
UNTIL

ndk said...

Velocity is more important than quantity. And it's typically ignored. Also, velocity is fundamental to the notion of a liquidity trap, which is the problem we face now.

It's certainly become more important post-1961, JKH. Before that, apparently it was less important.

This tautology leaves us without much of an explanation, though. Why would adjustment have primarily taken place in prices then, and in velocity now?

The relationship in the post also demonstrates that not only does money growth post-1961 not result in price rises, but it is inversely correlated with them. Why would velocity compress significantly more than necessary to maintain a steady second derivative of prices?

This one really has me scratching my head.

JKH said...

NDK,

I've put very little thought into this, and don't mean to be trite, but what I also believe more than ever is that inflation is not entirely a monetary phenonenom (any more than breathing is entirely an atomospheric phenomenon).

It would take a true monetary historian to explain the specific relationship inversion you note. I have no insight right now. But just note again that fabulous misconceptions abound about how the monetary process works at the level of base money alone. It seems reasonable to me that these sorts of errors in interpretation and explanation might be propagated to the outer reaches of money definition. Sorry for the negative type response.

JKH said...

Sorry. I didn’t check my spelling. No disrespect intended.

ndk said...

Sorry. I didn’t check my spelling. No disrespect intended.

None taken; I share your belief, and have been following with some interest the discussion at Interfluidity. I'm not convinced I can come up with a compelling explanation for these trends either, and I say as much in my post.

But the authorities and the majority of market participants are much more confident they know the effects of printing massive volumes of cash than we are. The only problem with the quantity theory of money is we see literally the inverse correlation persisting for the last 50 years, and we have absolutely no mechanism to explain why.

Irrespective of that fact, we're pouring phenomenal volumes of reserves into the system in an attempt to generate inflation and loosen financial rigidities, without the foggiest idea what might happen. That scares me.

It's a critical divergence of reality from theory that we really should be able to explain. Maybe Niall Ferguson will write about it, but until then, this rank amateur is going to devote a few neurons to thinking about it.

JKH said...

“Irrespective of that fact, we're pouring phenomenal volumes of reserves into the system in an attempt to generate inflation and loosen financial rigidities, without the foggiest idea what might happen. That scares me.”

My guess is that the market, when it attempts to, is underestimating the Fed’s mechanical ability to fully reverse this process at its chosen timing (including allowing first for what might well be an upcoming second and bigger wave of reserve expansion). The tricky part will be judging the best timing for the reversal, which will include a lot of anticipatory risk.

Marlowe said...

Steven Keen quoted by Bron Sucheki:

"The standard money multiplier model’s assumption that banks wait passively for deposits before starting to lend is false. Rather than bankers sitting back passively, waiting for depositors to give them excess reserves that they can then on-lend in the real world, banks extend credit, creating deposits in the process, and look for reserves later."

JKH said...

Steven Keen is exactly right.

ndk said...

WRT the historical observations by S, and JKH pointing out my total unfamiliarity with that history, Bordo's History of Monetary Policy provides really fascinating reading. Apparently, there was very little to no intervention by the Fed in the markets from the '30's to the late '50's, because of a prevailing view that the Great Depression was their fault. That coincides very well with the binary flip in this relationship.

"The standard money multiplier model’s assumption that banks wait passively for deposits before starting to lend is false. Rather than bankers sitting back passively, waiting for depositors to give them excess reserves that they can then on-lend in the real world, banks extend credit, creating deposits in the process, and look for reserves later."

I also think that profound insight is generally correct, and it featured prominently in the last post.

I think you're suggesting the same logic could apply to this problem as well, Marlowe. Such would be the lending half to releveraging businesses and consumers, as everyone seizes lower nominal interest rates they anticipate to be temporary.

I'm starting to favor this explanation above others. Money supply's inverse relationship with inflation and its rapid growth to GDP could be both artifacts of successful conventional modern monetary policy. It works very well once or twice, but when used too many times in series, there are problems.

I have a ton of data to look at before even proposing such, though.

JKH said...

NDK,

We know credit expansion was outsized; so much of this must have been monetized.

We also know that the asset economy (i.e. stocks, bonds, finance, existing houses, etc.) grew in relative importance, compared to the traditional GDP economy.

If the proportion of credit and money supporting asset trading grew (compared to that supporting GDP transactions), then GDP money velocity, measured using total money as the denominator, would have declined. This is because total money overstates the money denominator applicable to the GDP economy.

I.e., Assume total money M turns over V times within GDP. Then V has declined because M overstates the money basis for GDP, given the proportion of M that is required to support wealth trading, separate from GDP.

Anonymous said...

> We know credit expansion was outsized; so much of this must have been monetized.

is it possible that a lot of this money then has made its way back into treasuries, pushing down yields?

S said...

Not to edge in the conspriacy angle but hueristics and adjustment sclearly have a role. What doe the chart look like with shadow statistics post 1970? Also as pointed out per JKH, if indeed the GSDP economy diverged from the FIRE economy then how can anyone put any merit in the Krugman favorite Okun law or output gap theory's? Also JKH comments on velocity are well said, but doesn;t this just cycle back to the obvious that GDP is massivily overstated? Peak debt in other words!

think like a trader said...

JKH made many valid points. Here is an article that could help to understand this topic.
http://www.debtdeflation.com/blogs/2009/01/31/therovingcavaliersofcredit/

Hope your guys can read it.

ndk said...

We know credit expansion was outsized; so much of this must have been monetized.

We also know that the asset economy (i.e. stocks, bonds, finance, existing houses, etc.) grew in relative importance, compared to the traditional GDP economy.

If the proportion of credit and money supporting asset trading grew (compared to that supporting GDP transactions), then GDP money velocity, measured using total money as the denominator, would have declined. This is because total money overstates the money denominator applicable to the GDP economy.


This is all true, JKH, but this only gets us to the point where money can be created with limited impact on the rest of the economy, because it operates almost exclusively in a subset of the main economy.

If that were the complete explanation, we would expect to see growth in money supply growth and contraction occur without much impact on GDP, no?

But that's not what happened. Again, we see an extremely strong negative correlation(-.68 between M2 and CPI, for example). How could more money supporting asset trading affect the CPI negatively? I really appreciate your patience, because I really don't think that logic fits the data.

JKH made many valid points. Here is an article that could help to understand this topic.

I read the article and I think it's an excellent explanation, far better than that in most textbooks. But I still don't see how we get that negative correlation.

ndk said...

How could more money supporting asset trading affect the CPI negatively?

Now, if we invert the causality, we're back at the explanation Marlowe and I were using.

A negative shock hits the economy, causing output and inflation to drop. Interest rates are lowered in response. Investors, homeowners, and businesses alike respond to the forcibly lowered interest rates by levering up, creating money proportionately to the drop in interest rates, and hence the drop in inflation.

The economy recovers gradually, leading inflation and interest rates to rise a little bit, but not to the highs seen in the previous cycle, because the debt load is now correspondingly larger.

Rinse, repeat, zero bound, present day.

My next post will develop this hypothesis. Thanks for your readership and juicy talented brains.

Mikkel said...

"Interest rates are lowered in response. Investors, homeowners, and businesses alike respond to the forcibly lowered interest rates by levering up, creating money proportionately to the drop in interest rates, and hence the drop in inflation."

The reason why things are so bad now is because no one can lever up.

I'm just going to throw out there that I would be careful when characterizing correlations when looking at coupled oscillatory systems. You really need to have a hypothesis about propagation time between the components. What is a negative (or positive) correlation at one time delay may reverse at a longer time delay. This data supports the idea that an increase in supply leads to an increased GDP deflator at about a 12-18 month delay (unless I'm completely missing the boat).

What makes economics so difficult is that there is not enough data and it is too noisy to actually determine real mathematical relationships. If the series was 10x longer and a tiny bit smoother I could help calculate whether the series are indeed coupled and the optimal range of coupling. However, just eyeballing it I would not be convinced that there is anything noteworthy in that series, and I really like pointing out flaws in all models...so.

To give my two cents, the 70s was a messed up time. I've looked at many timeseries of economics data (CPI/debt loads/median wages/productivity/etc) and I would best characterize it that the 70s truly look like an external shock outside the "normal" (50s, 60s, 80s, 90s) dynamics. In fact if you just deleted that decade then the volatility and complexity looks pretty smooth.

I buy your argument that inflation and interest rates are tied to liquidity (in general, again the 70s were messed up) which in our system is primarily a function of debt. Friedman showed that inflation can take hold simply due to expectations, I would actually phrase it as inflation is a function of economic uncertainty that operates on positive feedback. It's possible that expectation changes drive debt creation itself as much as debt drives inflation. Does anything I'm saying make sense? It's very late.

Anyway, under my intuitive feel the liquidity keeps inflation low due to increased productivity, and then when debt gets too large to be supported it starts popping up in bubbles and destabilizing the system. When the bubbles take place in "real" assets/commodities then that increases volatility of inflation and we get the blowoff tops that are very common before deflationary collapse.

ndk said...

This data supports the idea that an increase in supply leads to an increased GDP deflator at about a 12-18 month delay (unless I'm completely missing the boat).

Your point is well taken. I've seen a tremendous variability in estimates, Mikkel, ranging from 2 months to 2 years. Regardless, the 2-8 year window used by Shelley and Wallace with such a high correlation coefficient makes me confident that this is indeed a strong relationship.

Does anything I'm saying make sense?

Much of it. I agree that inflation expectations are important, and I actually am starting to believe that money supply is to be targeted through the tool of interest rates, even though such action is completely impractical.

Anyway, under my intuitive feel the liquidity keeps inflation low due to increased productivity, and then when debt gets too large to be supported it starts popping up in bubbles and destabilizing the system. When the bubbles take place in "real" assets/commodities then that increases volatility of inflation and we get the blowoff tops that are very common before deflationary collapse.

I'm almost with you here, Mikkel. I diverge only in thinking it's not solely, or even primarily, increased productivity that keeps inflation suppressed, though it does play a role. I think it's also heavily dependent on that increased sensitivity of the real economy to fluctuations in the financial system that we discussed in the prior post. The latter half of that is why I believe traditional monetary policy is doomed to failure.

doggett said...

babar ganesh said...

"i can't see the graph well enough -- can you point me to a bigger version?"

I double-clicked on the graph (I'm on a Mac) and a larger, readable image opened in a separate window.

Anonymous said...

ndk writes: "A negative shock hits the economy, causing output and inflation to drop. Interest rates are lowered in response. Investors, homeowners, and businesses alike respond to the forcibly lowered interest rates by levering up, creating money proportionately to the drop in interest rates, and hence the drop in inflation."

The standard thinking would be that negative shocks reduce ecoomic activity and increase uncertainty, which would induce reduction in debt (in all segments of the non-gov't economy). If you're hypothesis is correct, it would be a significant surprise, so I'm eager to hear more. Is there an assumption about time lags here that is implied?

jult52

Mikkel said...

"I'm almost with you here, Mikkel. I diverge only in thinking it's not solely, or even primarily, increased productivity that keeps inflation suppressed, though it does play a role."

How about this formulation? There are various mechanisms that keep inflation suppressed but reducing sensitivity to the financial system and letting there be productivity growth is (more) sustainable than the other way around? I say more because I have problems with all growth based systems in general, since they are, you know, exponential and all.

" Regardless, the 2-8 year window used by Shelley and Wallace with such a high correlation coefficient makes me confident that this is indeed a strong relationship."

I strongly disagree with their methodology. They are just running different bandpass filters, not different time delays. Their methodology is looking at different timescales of behavior but their conclusion is not supported because they aren't looking at shifted forms of the data and those are two different qualities.

Even more importantly, the data is no where close to stationary as it's not driven by the same internal dynamics. It also isn't very long (doesn't have many cycles). So, I don't even agree with them using band pass filter to make any determinations.

There is really no way to accurately measure what they are trying to measure from empirical data...you'd have to construct a model and then just "eyeball" it. Even then, the 70s were such a weird time that there are Nobel prizes just for explaining special characteristics for how weird it was so that gets rid of most of the model validation and would need to be modeled separately.

The best you could do with what they have is look at cross correlations over different time delays and look at different timescales by averaging data points together. I'm still not sure it'd give you enough to make a strong argument.

I hate to sound like a sourpuss but I tend to push for pretty radical modeling/interpretations of data, and so I've become very sensitive to using tools properly and being honest about when no tools are appropriate. I've destroyed so many dreams that way..destroyed dreams are delicious.

mikus said...

JKH said:

"Fifth, the notion of the “money multiplier” in banking is a completely fraudulent explanation of the way the monetary system expands itself. Central banks actively control the short term interest rate, while passively supplying the reserves that the banking system requires in order to meet its statutory obligations as determined by deposit liability growth. Reserves follow deposits; not vice versa. How could any quantity theory of money be helpful when this basic fact of banking is so widely misunderstood?"

Isn't this basically what the Federal Reserve itself says? I quote from the Fed publication, "The Federal Reserve System Purposes and Functions":

"The Federal Reserve can try to achieve a desired quantity of balances at the Federal Reserve Banks or a desired price of those
balances (the federal funds rate), but it may not be able to achieve
both at once. The greater the emphasis on a quantity objective,
the more short-run changes in the demand for balances will influence the federal funds rate. Conversely, the greater the emphasis
on a funds-rate objective, the more shifts in demand will influence
the quantity of balances at the Federal Reserve. Over the years,
the Federal Reserve has used variations of both of these operational
approaches."

In other words, the Fed injects or removes the amount of reserves required to maintain the Fed funds rate at its desired target, and it is demand for reserves that determines the level of reserves. This is essentially the same thing as what Steve Keen argues, although perhaps with some different emphasis.

Of course, the money multiplier story is also used but I think it's just a heuristic tool for describing how deposits are created. If there's a real problem with the money multiplier story I'd say it's a failure to distinguish between currency and debt (deposits), not so much in its claim that a given amount of currency can support a greater growth of deposits.

think like a trader said...

NDK Said: "I read the article and I think it's an excellent explanation, far better than that in most textbooks. But I still don't see how we get that negative correlation."

Actually it's not difficult to figure out why such a huge money base increase could not pump up the inflation based on Steve keen's theory which I think has the best theoretic framework.

Using Steve's example, when Mary and Joe exchanged their products via the bank, they don't really care the money supply level at the bank. All they care is the $500 they get can buy a pig or 50lb copper. As long as this price level unchanged, no matter how much money the bank is printing. In other words, you can increase money base but as long as it's not affecting the price level, there will be no inflation. Now let's say the bank printed more than $500 in bills, it has incentive to turn the worthless extra bills into something valuable. The bank could loan the extra bills out even without interest since any repayment of that loan is something valuable generated from nothing. In order to encourage people to get the loan, the bank also put a bankruptcy law that allow anyone who couldn't repay the loan later basically walks away from their loan obligation. That makes sense since the money the bank loaned out is actually worth nothing. From here you will see inflation.

My thinking is as long as the banking system does not make loose loans as we witnessed during the sub-prime madness, M2 increase will mostly end up as inactive excess reserve. However the banks have the very incentive to put those reserves into work and the result will be another round of Sub-prime like lending in the future.

The $800B stimulus plan is highly inflationary. It is similar to dropping money from a helicopter. Unlike what the Fed does, this money will be injected into real economy. Its effectiveness will be short lived since the amount of the stimulus is much small to the amount of assets value lost due to weak demand. However it greatly increases the government debt and that will be destructive.

I already know there are only two choices for US and countries like US in high debts. Either going through a prolonged deflationary recovery or though an inflationary destruction. One choice is like to cut off your poisoned arm, the other way is to try to keep that poisoned arm and let the poison to circulate into your heart. I see many counties are digging their own graves right now.

JKH said...

Mikus,

You’re right. That is what the Fed correctly says. But the Fed isn’t the problem.

The problem is that the money multiplier paradigm obscures the actual nature of money creation in the banking system. That’s all. But as a result, fewer people will have a correct understanding of the Fed and banking operations.

The money multiplier concept can appear to be valid when the banking system is viewed as a snapshot at a point in time. But this is an illusion. The concept is not true as it is normally presented, which is as a process of money creation over time that depends on the availability of reserves at the beginning of the period.