I loaded up some data from the St. Louis Fed (http://research.stlouisfed.org/fred2/) and thought it would be interesting to see the relationship between the CPI inflation measure and growth in the money supply.
Unfortunately M3 data stopped a few years back, so I have collected CPI vs M3 from 1960 to 2006 and CPI vs M2 from 1960 till the present day. I was struck by the amazing non-correlation - perhaps there's even an inverse correlation!
Can anyone explain what is happening? Is there an article discussing this relationship anywhere?
Its a shame the website only has M3 going back to 1960, is there somewhere that published M3 for earlier dates? - preferably going back to before the great depression.
What Went Wrong with Economics
CPI is one big shell game.
Oil goes up 100% over three years --> CPI (ex-energy) is only 2%/year.
Commodities go up 100% over three years --> CPI (ex-food) is only 2%/year
Real-estate goes up 100% in three years --> CPI (ex-housing) is only 2%/year.
Stocks go up 100% in three years --> CPI (ex-stocks) is only 2%/year
The money created out of thin air (M1, M2, M3, money supply) must find its place SOMEWHERE.
The "Law of CPI": CPI (ex-SOMEWHERE) is always <= 2%/year. That's why (according to CPI) there's no inflation while prices magically keep rising all around us.
Z.
Do you have a reference to the "law of cpi"? Google doesn't really come up with anything.
mickanomics:Do you have a reference to the "law of cpi"? Google doesn't really come up with anything.
It was my own attempt at sarcasm.
Think about it. It's impossible to measure the price of EVERYTHING consistently. Even without the "ex-something" shell game one could still play infinite shell games by "adjusting" the CPI "basket" of relevant goods. And this basket wouldn't even begin to address the productivity gains (a calculator today has more computing power than a 60's mainframe computer). CPI is wrong on too many levels, and it's only used as an instrument for hiding the fact that your cash savings are evaporating in value MUCH faster than you're told.
Just think logically. If there's more of ANYTHING (or money, in this case as shown by M1, M2, and M3) what usually happens to its price? Inflation IS money supply and nothing but. CPI is a mere propaganda veil.
z1235: It was my own attempt at sarcasm. Think about it. It's impossible to measure the price of EVERYTHING consistently.
Think about it. It's impossible to measure the price of EVERYTHING consistently.
Actually I'm in complete agreement with you. I was just after a more formal/detailed discussion/analysis of it. Is there something on mises.org?
mickanomics:Actually I'm in complete agreement with you. I was just after a more formal/detailed discussion/analysis of it. Is there something on mises.org?
An index can never capture fluctuations in the purchasing power of money--it's literally impossible. The Austrians wrote about this extensively in the early part of the last century, and now the mainstream is figuring it out.
"If we wish to preserve a free society, it is essential that we recognize that the desirability of a particular object is not sufficient justification for the use of coercion."
I must be reading this wrong:
http://www.nowandfutures.com/key_stats.html#money_cpi
Check my blog, if you're a loser
meambobbo:I must be reading this wrong:
Wow, your graphs look incredibly different!
My data is the combination of this CPI data:
and this M3 data or this M2 data.
Did I make a mistake?...
Always take into account the money multiplier. This bad boy has mostly been on a downward trend ever since it has been recorded, and has thus offset some of the consequences of an increasing money supply.
The CPI is changed regularly. For example, they might change Steaks to Hamburger meat because if the price of steaks gets too high people will just buy hamburgers. bla bla bla. But if you compare this year's hamburger to last year's steak, you're not going to show inflation properly :P
mickanomics:Did I make a mistake?...
Probably not - the link I provided says M3, M2, CPI-U, and CPI-U* are on 10 year moving averages. Also, the links you provided only date back to 1959. He uses a Fed paper for 1910-1959. Also note that the BLS changed reporting methodologies in 1982. I'm betting you didn't take all those things into account, no?
Using YoY growth rates can be deceiving, with numerous ups and downs that obscure the larger, general trends. When you smooth the data, there's a clear correlation, with money supply growth leading price inflation. Without smoothing, it just looks like chaos. Also, it makes sense that prices increase at a lower rate than money supply, due to economic growth.
Also, WWII is obscure due to price and wage controls and rationing. Price inflation would likely have been much larger without such.
inb4: gov't statistics are full of shit generally. Now, why did I post that? Why, to illumine a new generation of libertarians with this timeless truth.
Freedom of markets is positively correlated with the degree of evolution in any society...
You can't accurately measure inflation in a meaningful way. It's quite literally impossible.
meambobbo: mickanomics:Did I make a mistake?... Probably not - the link I provided says M3, M2, CPI-U, and CPI-U* are on 10 year moving averages.
Probably not - the link I provided says M3, M2, CPI-U, and CPI-U* are on 10 year moving averages.
Ok, so it seems that in the long term there is a nice correlation, but in the shorter term the correlation is a mess. Indeed to my eye it looks like a short term inverse correlation. Am I "seeing canals on Mars" or has anyone else noticed? -> Look here. I can even think of an explanation: the points where M2 are growing fastest are when stocks/housing/derivatives are rising fastest. At these points investors have the greatest incentive to leave there money in the sky-rocketing investments and not go and spend their money on consumer goods. Alternatively, particularly with housing, when their prices are rising fastest, that's when people will try their hardest to join the bandwagon and forgo consumer spending in favour of getting a bigger mortgage.
Sphairon:Always take into account the money multiplier. This bad boy has mostly been on a downward trend ever since it has been recorded, and has thus offset some of the consequences of an increasing money supply.
But we're analysing M2/M3 so we don't need to "take in to account" the money multiplier. If I was comparing M0 to CPI then the money multiplier may need to be taken in to account.
A low money multiplier is a sign that the money supply has become disconnected from the monetary base. The limits on money creation are now much more related to the general propensity to take out new loans vs. the propensity to pay back/default.
The monetary base is now a tiny insignificant part of the money supply - you can halve it or double it and it makes little difference to M3.
mickanomics: But we're analysing M2/M3 so we don't need to "take in to account" the money multiplier. If I was comparing M0 to CPI then the money multiplier may need to be taken in to account. A low money multiplier is a sign that the money supply has become disconnected from the monetary base. The limits on money creation are now much more related to the general propensity to take out new loans vs. the propensity to pay back/default. The monetary base is now a tiny insignificant part of the money supply - you can halve it or double it and it makes little difference to M3.
Fantastic. I don't understand what you're trying to do.
Esuric:Fantastic. I don't understand what you're trying to do.
I'm trying to understand better the relationship between the money supply and CPI-inflation. The mainstream textbooks, and many economists say "printing money leads to inflation"... so I thought I'd have a look at the raw data and see how true (or not) this statement is in practice.
mickanomics:I'm trying to understand better the relationship between the money supply and CPI-inflation. The mainstream textbooks, and many economists say "printing money leads to inflation"... so I thought I'd have a look at the raw data and see how true (or not) this statement is in practice.
The CPI doesn't measure inflation, so why are you trying to find an imaginary correlation? The raw data will tell you nothing because it doesn't consider human action. You're looking at meaningless numbers when you should apply some logic.
Esuric:The CPI doesn't measure inflation, so why are you trying to find an imaginary correlation?
But the 10 year moving average data looks like a very nice correlation - how do you explain that?
mickanomics:But the 10 year moving average data looks like a very nice correlation - how do you explain that?
The CPI to m3 is not close at all, and who knows what it looks like for MZM. Furthermore, your charts cover a 50 year period, where there are major fluctuations (nevermind the fact that it stops at 2006). Either way, you misunderstand my position: the prices of the goods chosen for the CPI do not tell you anything about overall relative price distortions, or the misdirections of capital and labor. When there is a correlation it is an arbitrary one (correlation does not mean causation).
Esuric:The CPI to m3 is not close at all
In the long term it looks pretty good to me: http://www.nowandfutures.com/key_stats.html#money_cpi
Esuric:When there is a correlation it is an arbitrary one
Arbitrary? What do you mean by that?
Esuric:correlation does not mean causation
I know that very well. But I think you are overstating your case if you claim the apparent long term (10yr m.a.) correlation is just coincidence.
mickanomics:In the long term it looks pretty good to me: http://www.nowandfutures.com/key_stats.html#money_cpi
The chart you provide in the OP looks entirely different.
mickanomics:Arbitrary? What do you mean by that?
mickanomics: Esuric:correlation does not mean causation I know that very well. But I think you are overstating your case if you claim the apparent long term (10yr m.a.) correlation is just coincidence.
The price of commodities rose by 100% in 2008 and the CPI was at 3%. The curve for housing prices was practically vertical and the CPI remained considerabaly low throughout the 90s and the last decade.
Global supply rose 270 kb/d in December to 86.2 mb/d, on both higher OPEC and non-OPEC output.
Forecast global oil demand remains virtually unchanged at 84.9 mb/d in 2009 (-1.5% or -1.3 mb/d year-on-year) and 86.3 mb/d in 2010 (+1.7% or +1.4 mb/d versus the previous year)
http://omrpublic.iea.org/
Jan. 21 (Bloomberg) -- China’s growthaccelerated to the fastest pace since 2007 in the fourth quarter, capping Premier Wen Jiabao’s success in shielding the nation from the global recession....true???
http://www.bloomberg.com/apps/news?pid=20601087&sid=asqZE.UsivdE
i am not sure about the commodity price increases you speak of.
if the above info is true it seems that supply increased and demand stayed flat....occurences a 100+ year old global industry by this point should be able to cope with via competition rather easily...without massive price increases.
with various products and processes that are developed that offer different levels of increased productivity i dont think you can make any real correlation between cpi and money supply increase.
certain credit-trails can possibly be traced to malinvestments, but i am not certain of the specific level of harmful price-inflation that would cause...if any. if a few items went up but most went down and an overall budget is less for many once adjusted for inflation then the various cpi modes may have some validity regardless of monetary-inflation.
33n119w: Global supply rose 270 kb/d in December to 86.2 mb/d, on both higher OPEC and non-OPEC output. Forecast global oil demand remains virtually unchanged at 84.9 mb/d in 2009 (-1.5% or -1.3 mb/d year-on-year) and 86.3 mb/d in 2010 (+1.7% or +1.4 mb/d versus the previous year) http://omrpublic.iea.org/ Jan. 21 (Bloomberg) -- China’s growthaccelerated to the fastest pace since 2007 in the fourth quarter, capping Premier Wen Jiabao’s success in shielding the nation from the global recession....true??? http://www.bloomberg.com/apps/news?pid=20601087&sid=asqZE.UsivdE i am not sure about the commodity price increases you speak of. if the above info is true it seems that supply increased and demand stayed flat....occurences a 100+ year old global industry by this point should be able to cope with via competition rather easily...without massive price increases. with various products and processes that are developed that offer different levels of increased productivity i dont think you can make any real correlation between cpi and money supply increase. certain credit-trails can possibly be traced to malinvestments, but i am not certain of the specific level of harmful price-inflation that would cause...if any. if a few items went up but most went down and an overall budget is less for many once adjusted for inflation then the various cpi modes may have some validity regardless of monetary-inflation.
I don't know what you're talking about, and neither do you.
33n119w:i am not sure about the commodity price increases you speak of.
Gold, silver, copper, tin, corn, wheat, oranges, oil. You know?
Esuric:The chart you provide in the OP looks entirely different.
Indeed true. My chart is the raw data, the other chart (posted by meambobbo) is the 10 year moving average of M3 growth vs. the 10 year moving average of CPI growth
It looks like in the short term the correlation is very poor (or even negative!), but in the long term the correlation is very good.
I'd expect short term correlation to be negative actually. When CPI is rising fast, CB is tightening. When CPI is falling or is stable, CB is inflating.
I thought i'd contribute since I'm working on an econometric analysis of inflation, or was... my hard drive is a few hundred miles away... long story short, I can't show you any data because I don't have it with me, but here's my 2 cents:
1. Are you actually regressing anything, or just plotting graphs? You're not going to get very far that way.
2. CPI should be used as an index in absolute terms, not as a percentage change or anything like that. Moreover, if you model CPI linearly you'll probably misspecify the model. CPI grows exponentially, so that means you're probably better off modeling it as a logarithm. For that matter, the same applies to the money supply (Hayek's accelerationism, right?). We've only just started being taught formal tests for functional form misspecification so I haven't done any of those yet though, so don't take my word for it.
3. I believe I used M2, which seemed to yield the highest coefficient of determination between the money supply and CPI (having also tried M3 and MZM).
4. I think my model was a multiple regression one. My other explanatory variables were levels of credit (can't remember which series exactly) and I think something else.
5. I haven't played around with dummy variables a lot yet, but the disturbance term grows in fluctuations over time, starting around the 80's I believe. Moreover there is a cyclical pattern in the error term, which means I've under-fitted my model (I'm not including some important variable).
6. I was thinking that just looking at money supply was pointless, since a lot of that money is sitting abroad doing nothing. This is especially true now that China has been hoarding dollars, and of course eastern europe, especially Russia and Belarus, use a lot of dollars locally, so this has to be taken into account somehow.
7. We haven't been taught how to use lag time yet, but I figure that has got to be important, if it does what I think it does (I have no idea). But basically, we would expect there to be a lag between money supply creation and price inflation, so that can probably be modeled too, and might strengthen the relationship.
8. If I remember correctly, my coefficient of determination was above 0.8 in the first single variable regression model I tried, which is pretty high, although we're dealing with time series so it's not amazing.
When my housemate brings my computer back, I'll post some graphs, but I've got lots more work to do on it...
Regards,
Fred
meambobbo: Probably not - the link I provided says M3, M2, CPI-U, and CPI-U* are on 10 year moving averages. Also, the links you provided only date back to 1959. He uses a Fed paper for 1910-1959. Also note that the BLS changed reporting methodologies in 1982. I'm betting you didn't take all those things into account, no?
Could you please elaborate on what changes the BLS introduced in 1982? I think the early 80's was when the residuals in my model started fluctuating more wildly.
azazel:I'd expect short term correlation to be negative actually. When CPI is rising fast, CB is tightening. When CPI is falling or is stable, CB is inflating.
Excuse my ignorance, but what does CB stand for?
mickanomics: Excuse my ignorance, but what does CB stand for?
Probably central bank.
mickanomics: I loaded up some data from the St. Louis Fed (http://research.stlouisfed.org/fred2/) and thought it would be interesting to see the relationship between the CPI inflation measure and growth in the money supply. Unfortunately M3 data stopped a few years back, so I have collected CPI vs M3 from 1960 to 2006 and CPI vs M2 from 1960 till the present day. I was struck by the amazing non-correlation - perhaps there's even an inverse correlation! Can anyone explain what is happening? Is there an article discussing this relationship anywhere? Its a shame the website only has M3 going back to 1960, is there somewhere that published M3 for earlier dates? - preferably going back to before the great depression.
As a small side-note that I don't believe has been mentioned here:
http://mises.org/content/nofed/chart.aspx
Mises.org calculates its own money supply figures...
Latest Projects
"Even when leftists talk about discrimination and sexism, they're damn well talking about the results of the economic system" ~Neodoxy
Fephisto: http://mises.org/content/nofed/chart.aspx Mises.org calculates its own money supply figures...
That's excellent! I'll be able to use those now, instead of relying on Fed data, thanks :)
Fred Furash:1. Are you actually regressing anything, or just plotting graphs? You're not going to get very far that way.
The brain is a brilliant pattern matcher, and can spot at a glance all sorts of things that statistical procedures struggle with. Don't get me wrong, I'm not against doing statistics at all, but its important to use your eyes in addition to doing the stats. For example if the short term fluctuations are inversely correlated and the long term changes are correlated (perfectly possible) that would probably be missed altogether if you never sat and had a good look at the graphs in the first place. You'd probably just do some kind of "correlated or not" test and leap to an oversimplified conclusion.
Fred Furash:2. CPI should be used as an index in absolute terms, not as a percentage change or anything like that. Moreover, if you model CPI linearly you'll probably misspecify the model. CPI grows exponentially, so that means you're probably better off modeling it as a logarithm. For that matter, the same applies to the money supply (Hayek's accelerationism, right?). We've only just started being taught formal tests for functional form misspecification so I haven't done any of those yet though, so don't take my word for it.
I agree that if comparing absolute CPI vs. M3 then you should use logs. I think you can learn things from both comparing logs of absolute values as well as looking at rates of change (no logs required). There are all sorts of things you can see easily with one type of comparison that are very hard to spot in the other.
Fred Furash:6. I was thinking that just looking at money supply was pointless, since a lot of that money is sitting abroad doing nothing. This is especially true now that China has been hoarding dollars, and of course eastern europe, especially Russia and Belarus, use a lot of dollars locally, so this has to be taken into account somehow.
Good point. If you manage to get together some data for M3 kept within the US I'd be very interested to see it.
Fred Furash:When my housemate brings my computer back, I'll post some graphs
I look forward to seeing them.
mickanomics: Fred Furash:1. Are you actually regressing anything, or just plotting graphs? You're not going to get very far that way. The brain is a brilliant pattern matcher, and can spot at a glance all sorts of things that statistical procedures struggle with. Don't get me wrong, I'm not against doing statistics at all, but its important to use your eyes in addition to doing the stats. For example if the short term fluctuations are inversely correlated and the long term changes are correlated (perfectly possible) that would probably be missed altogether if you never sat and had a good look at the graphs in the first place. You'd probably just do some kind of "correlated or not" test and leap to an oversimplified conclusion.
I agree, you have a point there. The problem however is that mainstream economists won't recognize any sort of proof based on "look at how this correlates". They like to see formal statistical tests. But yes, I ought to pay more attention to plots too, you're right that I would have otherwise missed such a correlation.
mickanomics:The mainstream textbooks, and many economists say "printing money leads to inflation"...
Printing money (growing M1, M2, M3) doesn't merely LEAD to inflation. It IS inflation, by definition. Defining inflation as (an easily manipulated and constantly "updated") CPI is only done for the purpose of throwing dust up in the air and in everyone's eyes.
z1235: mickanomics:The mainstream textbooks, and many economists say "printing money leads to inflation"... Printing money (growing M1, M2, M3) doesn't merely LEAD to inflation. It IS inflation, by definition. Defining inflation as (an easily manipulated and constantly "updated") CPI is only done for the purpose of throwing dust up in the air and in everyone's eyes.
When they say "printing money leads to inflation", I think perhaps they mean:
A: Increasing the monetary base or lowering interest rates leads to an increase in the money supply
B: An increase in the money supply leads to an increase in the "cost of living"
Unfortunately I think that both A and B are very tenuous and crude links in the short term, though in the longer term it becomes more true.
I think we need more than one word for "inflation", perhaps we could use the labels "M0-inflation", "m3-Inflation","CPI-inflation" etc.
z1235:Printing money (growing M1, M2, M3) doesn't merely LEAD to inflation. It IS inflation, by definition.
If M1 was growing while M3 was shrinking would that be inflation or deflation?
mickanomics:If M1 was growing while M3 was shrinking would that be inflation or deflation?
My understanding of the processes is not affected by semantics (varying definitions of inflation). Perhaps neither should yours.
z1235: mickanomics:If M1 was growing while M3 was shrinking would that be inflation or deflation? My understanding of the processes is not affected by semantics (varying definitions of inflation). Perhaps neither should yours.
What are you saying? Are you trying to make a virtue out of being imprecise?