PDA

View Full Version : M3 and inflation, deflation & ideology


bart
08-22-08, 06:33 PM
M3 and inflation, deflation & ideology


The recent news from Lombard Research about "the biggest one-month fall since modern records began in 1959" in M3 could easily have been misunderstood and misinterpreted. M3, by the way, is simply the most inclusive measure of total money supply (excluding any credit measures) that exists. To add some broad perspective, here's a weekly M3 chart which includes an annualized 13 week rate of change line (in light blue).


http://www.nowandfutures.com/images/m3b_13week.png


As can easily be seen, the drop is not at all that large since a very similar and actually slightly larger drop happened just last year.

Additionally, drops have been much larger drops where M3 has been contracting at almost a -5% rate in various short periods in 1991, 1992, 1993 and 1994. It has also been below zero in the early '70s, early 1987, 1998, 1999, 2000, 2002 and 2003.

Is it significant and important? Yes... but in reviewing all the times that it has happened in the last 38 years of weekly Fed and reconstructed data, there isn't much that can be concluded that we don't know already. Those periods were ones of recession or slow growth or world financial crises or housing recoveries, or bubble tops or even U.S. financial crises.
(Chart since 1970 for the curious - http://www.nowandfutures.com/images/m3b_13week_long_term.png )







More important though is what it doesn't say, and especially what some bloggers and others who have a questionable understanding of M3 and other money measures have not been saying (or even outright avoiding due to ideology issues or vested interests, etc.).

Here's the chart that most clearly paints the story with both M3 and M2, and why they can be so useful. The obvious way that M2 and M3 track with inflation, as measured by CPI-U with corrections from John Williams of shadowstats.com, is unmistakable.

http://www.nowandfutures.com/images/m2m3_cpi_money_supply.png


The simple view and truth is that we're in an inflationary cycle, similar to the '70s, and real deflation is extremely unlikely in the foreseeable future.

A key though is what M2 is currently doing. It has been close to flat for some time while M3 is still growing, again in a similar manner as 1968-70 or 1974-75 or even 1999-2000. So its quite likely that, barring black swans, we'll see inflation dropping some in the short to intermediate term future.

Also note that neither M2 nor M3 tell anywhere near the entire story. There are many other ways to look at money supply and its effects on inflation and investing.
One quick example is M1, which some use as an inflation or deflation hint. Its true that M1 is expanding now, but when a certain monetary element called sweeps is added back in to help ensure that M1 is being compared and defined on an apples-to-apples basis, M1 plus sweeps has actually declining for a while now.



Hopefully, now that my opinions have been clearly stated and an additional important chart has been clearly linked to M3, certain bloggers and others will no longer mis-represent my views and opinions about M3 as they have been doing for months, regardless of my comments and requests.

jk
08-22-08, 08:21 PM
the clearest and best charts ever, bart. thanks.

metalman
08-22-08, 11:09 PM
excellent charts. question: i see the correlation but don't get the causation. if output is falling faster than the money supply, you get inflation. where's your output component?

bart
08-22-08, 11:57 PM
the clearest and best charts ever, bart. thanks.

Thanks jk, much appreciated.


excellent charts. question: i see the correlation but don't get the causation. if output is falling faster than the money supply, you get inflation. where's your output component?

If I understand your question, the output is (extremely roughly) implied. The gap between CPI and money supply is output.

Note also that by virtue of the extreme smoothing due to a 10 year moving average, monetary time lags have been almost eliminated.

metalman
08-23-08, 12:03 AM
Thanks jk, much appreciated.

If I understand your question, the output is (extremely roughly) implied. The gap between CPI and money supply is output.

Note also that by virtue of the extreme smoothing due to a 10 year moving average, monetary time lags have been almost eliminated.

thanks. i mean the old v=pq/m. never seen an economist try to correlate inflation with m except in relation to output. is it meaningful to try? v is a function of money and output. inflation correlates to v because inflation rises and falls with v but not necessarily m depending also on pq. you're missing the nominator in the equation.

bart
08-23-08, 10:19 AM
thanks. i mean the old v=pq/m. never seen an economist try to correlate inflation with m except in relation to output. is it meaningful to try? v is a function of money and output. inflation correlates to v because inflation rises and falls with v but not necessarily m depending also on pq. you're missing the nominator in the equation.

The money velocity boogie is one of the many areas in which I'm still in research (ok, confused ;) ) mode. The overall concept makes lots of sense but trying to find a practical way of expressing it is very non-trivial.

The chart isn't meant to be a be-all and end-all too, its just intended to show the very clear relationship between money supply and inflation... and secondarily of course expose a certain blogger for mis-representing me.



Just for grins though, here's a longterm chart showing many of the ways to view velocity.

http://www.nowandfutures.com/images/velocity_long_term.png

jk
08-23-08, 11:54 AM
what would be really useful is something quickly available that would function as a proxy or predictor for velocity. to rely on a gdp measure implies waiting for gdp calculations, and revisions, to be able to finally postdict velocity 6 months later. so is there a function of, e.g., retail sales, personal income, corporate profits, and so on - relatively contemporaneously reported variables - that gives us a quick [if dirty] measure of velocity?

bart
08-23-08, 12:28 PM
what would be really useful is something quickly available that would function as a proxy or predictor for velocity. to rely on a gdp measure implies waiting for gdp calculations, and revisions, to be able to finally postdict velocity 6 months later. so is there a function of, e.g., retail sales, personal income, corporate profits, and so on - relatively contemporaneously reported variables - that gives us a quick [if dirty] measure of velocity?

If there is something like that, I'm unaware of it... but am sure willing to listen to any ideas.

University of Michigan Inflationary Expectations may be useful and even watching currency (M0) growth can give hints though.
Maybe those would help a bit?

metalman
08-24-08, 10:10 AM
what would be really useful is something quickly available that would function as a proxy or predictor for velocity. to rely on a gdp measure implies waiting for gdp calculations, and revisions, to be able to finally postdict velocity 6 months later. so is there a function of, e.g., retail sales, personal income, corporate profits, and so on - relatively contemporaneously reported variables - that gives us a quick [if dirty] measure of velocity?

excellent point. gdp is reported quarterly, cpi monthly, mzm et al daily.

these guys claim to have a monthly gdp... (http://www.e-forecasting.com/US_Monthly_GDP.html)

http://www.e-forecasting.com/Charts/August_1_2008_rtgdptable.gif

http://chart.apis.google.com/chart?chs=450x300&chxt=x,y&chds=-2,2&chd=t:0.088,0.59,2.91,1.42,0.59,-0.07,-0.15&cht=lc&chl=Jan%7CFeb%7CMar%7CApr%7CMay%7CJun%7CJul
v = gdp monthly / mzm own rate

Finster
08-24-08, 12:27 PM
what would be really useful is something quickly available that would function as a proxy or predictor for velocity. to rely on a gdp measure implies waiting for gdp calculations, and revisions, to be able to finally postdict velocity 6 months later. so is there a function of, e.g., retail sales, personal income, corporate profits, and so on - relatively contemporaneously reported variables - that gives us a quick [if dirty] measure of velocity?

You could get the US mint to embed a microchip in each bill and track them via satellite to see how fast they're moving. Only problem with that is so much money is virtual bits. Maybe to capture those you could get the Treasury and Fed to team up with the CIA and FBI and conduct comprehensive Internet surveillance.

c1ue
08-24-08, 02:25 PM
Actually there is probably another way:

Identify 500 or 1 thousand existing paths for money, and track the progress of 1 dollar through these paths.

For example:

Path 1: Joe Blow the auto worker. Paid by auto company. Buys Food from grocery store. Grocery store pays checker. Checker buys gas. etc until somewhere someone saves the result.

Path 2: Bank lends to a small manufacturing business. Business pays workers. Workers spend on some specific good (i.e. restaurant). Restaurant buys supplies from wholesaler. Wholesaler buys from commodity broker. etc.

The goal of having many paths is to get an idea on 'typical' velocity in said path, and tracking over time would tell you (in aggregate) velocity changes.

Heck of a lot of work though.

bart
08-24-08, 02:55 PM
what would be really useful is something quickly available that would function as a proxy or predictor for velocity. to rely on a gdp measure implies waiting for gdp calculations, and revisions, to be able to finally postdict velocity 6 months later. so is there a function of, e.g., retail sales, personal income, corporate profits, and so on - relatively contemporaneously reported variables - that gives us a quick [if dirty] measure of velocity?

I just had another thought, and I think its key.

Velocity changes happen very slowly, as can be seen on that long term chart, so I think that there will be enough warning with the normal chart - especially since M1/Base data has very little to no reporting lag.

jk
08-24-08, 03:40 PM
my motivation is that i could imagine velocity plunging in certain scenarios- say a couple of big banks implode. in a crisis, it would be worth knowing in real time just how clenched people were feeling.

bart
08-24-08, 03:53 PM
my motivation is that i could imagine velocity plunging in certain scenarios- say a couple of big banks implode. in a crisis, it would be worth knowing in real time just how clenched people were feeling.

Currency in circulation?

http://www.nowandfutures.com/images/currency_in_circ.png


Lines at ATMs and similar could be useful too.

metalman
08-24-08, 07:19 PM
my motivation is that i could imagine velocity plunging in certain scenarios- say a couple of big banks implode. in a crisis, it would be worth knowing in real time just how clenched people were feeling.

true, but it's just as important to watch the numerator... output. what if banks crap out but output still crashes faster than the money supply?