Friday, August 26, 2011

Aaron Was Right -- And Some Great Graphs To Prove It

Awhile back, I 'corrected' Aaron that in the course of the business cycle there isn't really a problem with 'overconsumption.'  As it turns out, I was the one that needed correction.

In defense of myself, I will here blather uselessly for a few moments about why it did (and in some ways, still does) not seem reasonable to think this way about the problem.  Firstly, the business cycle is caused by the issue of excess credit above the rate of savings.  The extra credit is used to increase demand for capital goods, shifting the ratio of expenditures in favor of capital goods versus consumer goods.  Thus, how could there be an excess of both investment and consumption at the same time, in a universe in which materials are generally understood to be conserved?

Secondly, the partition of demand between present (consumption) and future (production) goods is given by the time preference of the society in question.  This is an intrinsic property of the population.  It is time preferences which determine interest rates.  The hypothesis of overconsumption would tend to suggest that FED policies had in fact influenced time preferences -- that the manipulation of interest rates had changed the time preferences of society.  This is a case of the tail wagging the dog, and begs the chicken-and-egg question.  Do interest rates determine time preferences, or time preferences determine interest rates?

Finally, I had had my delicate ego crushed rather cruelly not so long ago by a particular passage in The Theory of Money and Credit:
In drawing this conclusion, their doctrine implicitly denies the existence of interest. It regards interest as compensation for the temporary relinquishing of money in the broader sense—a view, indeed, of insurpassable naivety. Scientific critics have been perfectly justified in treating it with contempt; it is scarcely worth even cursory mention. But it is impossible to refrain from pointing out that these very views on the nature of interest hold an important place in popular opinion, and that they are continually being propounded afresh and recommended as a basis for measures of banking policy.

This had fairly well destroyed my view that I knew anything about the phenomenon of interest at all.  I had (stupidly) believed that people saved principally in response to interest rates, as if they were passive beings being led about by the nose by whatever the market 'told them to do.'  If the rate changed, they would find something else to do with their money.

Actually, people save mostly because they feel the need to, not primarily in a sort of passive response to 'compensation.'  They will do it even if they are punished by markets for doing it.  Thus, it was firmly fixed in my mind that real savings follow from qualities intrinsic to people, and whatever the effects banking might have on appearances, they wouldn't really change, except possibly over the very long term as the mentality of society itself was warped by the experience of inflation.  Thus, time preferences stay fixed over the course of the business cycle, while the market adjusts to a new apparently lower time preference illusion created by inflation, only to be smashed at the end when the old preference reasserts itself because inflation is no longer distorting things.

Be all that logical or illogical as it may, a short time later I encountered Robert Murphy making the same claim as Aaron.  Dr. Robert Murphy, that is.  So, even though this didn't make any sense to me at all, I decided to investigate for myself, and I'm glad I did.

This passage pretty much set me straight:
In Mises's subsequent exposition in Human Action (1966), forced saving takes on a substantially different meaning. Mises articulates a theory of boom and bust that is largely compatible with Hayek's formulation but not obviously so. The essential problem of the credit-induced boom is repeatedly summarized by Mises with variations of the phrase "malinvestment and overconsumption." The aspect of the problem identified by the first term of this phrase is recognized and emphasized by Hayek. The misallocation of resources—too many resources committed to the early stages of production—is the malinvestment. But what about the "overconsumption"? Can a macroeconomy experience forced saving and overconsumption at the same time or, at least, during the same boom? On the surface it would seem that the two terms are virtual antonyms. A sorting out of their various meanings and applications to the different phases of the boom-bust sequence, however, can almost fully resolve the seeming contradiction and in the process produce a more thorough understanding of the Mises-Hayek theory of the business cycle.
The notion that the boom is characterized by overconsumption also follows straightforwardly from the loanable-funds theory. The market for loanable funds—or, more inclusively, for investable resources—is equilibrated by movements in the interest rate, broadly conceived. An increase in saving would be depicted by a rightward shift in the supply of loanable funds. The market would take the economy down along the demand for loanable funds to a new equilibrium at which the interest rate is lower and both saving and investment are greater.
While creating similar incentives for the business community, a credit expansion in the absence of an increase in saving would have ultimate consequences that are fundamentally different. With this policy-induced change in market conditions, the apparent rightward shift of the supply curve represents an increase in credit in the absence of an increase in saving. Saving is simply augmented by credit creation. Nonetheless, the rate of interest would fall and the business community would be enticed, at least initially and to some extent, to undertake greater investments and would tend to allocate the credit-financed resources to the early stages of production. But since saving, as still represented by the unaugmented supply curve, has not changed, the lower rate of interest means that the amount saved actually decreases.
Only in the extreme and unlikely case of a perfectly inelastic supply of loanable funds would there be no decrease in saving. With an upward sloping supply, credit expansion causes the volume of saving to decrease—which is to say, it causes consumption to increase. This increase in consumption associated with a policy-induced decrease in the rate of interest is justifiably labeled by Mises as "overconsumption." Workers and other factor owners receiving increased incomes as a result of credit expansion will be induced to consume more than is implied by their pre-expansion intertemporal choices.
You get all that?

Actually, it is pretty simple.  The saving/consumption partition is principally set by time preferences, as I had thought.  But there is a secondary influence by interest rates -- the reward to save, whether phony or not -- as Aaron said, that will have a smaller effect over the course of the business cycle and nudge behavior this way or that depending on whether rates were higher or lower than the natural rate that would prevail in the absence of manipulation.  As inflation generally pushes interest rates down, over the course of the business cycle people will save slightly less than they otherwise would have, leading to an effect of 'overconsumption.'  I had discounted this effect, because I was taking "the extreme and unlikely case of a perfectly inelastic supply of loanable [sic] funds" perspective.

That would seem to settle the matter, but I didn't stop there.  You see, I'm such an arrogant and bull-headed twit that I refused to take on authority something that seemed counterintuitive to me.  I thought I'd show all of these guys they were wrong, including Mises.  Instead, I think I proved them right.

I decided to see if I could actually graph time preferences using FED data, and prove for myself what really happens.  I was 'inspired' to do this by of one of the charts Murphy used to support his claim, which I thought misrepresented things.  It is a graph of Personal Savings (PS), which is defined by the FED as Personal Income (PI) minus Personal Consumption Expenditures (PCE).  But PS doesn't really equal that, because 'savings' in the view that we are taking here -- the preference for future goods over present goods, the putting back of things for the future -- must include investments.  PCE includes Durable Goods (DG), which by definition cannot be consumer goods precisely because they are durable.  These expenditures are investments and should be included as savings.  So, I thought I'd try to tease the statistics apart and reconstitute them according to something closer to the Austrian take and see what they said then.

Unfortunately, a second component of PCE is Services (S) -- a label which implies that all services, whether production oriented and therefore to be categorized as savings, or consumption oriented and therefore to be categorized as consumption, had been uniformly lumped together in the same category, with no possibility of teasing them apart.  So, I couldn't make it a simple matter of shuffling things around.  And since the S component of PCE is pretty large, it can't exactly be neglected.

So, thought I, if I could just tease out the pattern I was looking for in the other quantities as a ratio -- the proportion if savings to consumption expressed as a fraction -- I could assume that whatever was going on in S expenditures would mirror this and effectively neglect them.  First, I stuck to the goods market, and simply assumed that whatever priorities were reflected there would probably be reflected elsewhere.  A graph of PCEDG (PCE Durable Goods) divided by PCENDG (PCE Non-Durable Goods), while not quite time preference because it neglects savings, would be expected to show how future good demand compared to present good demand -- and any fluctuations in this proportion over the course of the business cycle.

Here is the graph:

The grey bars denote periods of recession, as determined by the NBER.

I could not believe my eyes when I saw this graph. Though it wasn't exactly what I was going after for this argument, I have never seen a better illustration of the business cycle.  It shows perfectly the rising demand for capital goods in the boom, which fades and begins to fall towards the end of each cycle.  There are not many metrics that show the periodicity of recessions as well as this.  Unemployment is probably the best I know of, and by the time changes start showing up there, it is usually pretty much into the bust.  Note that little molehill out where we are in 2011, and how similar it looks to the one around 1981.  I think that is Bernanke's green shoot.  Not the best case I have ever seen that recovery is around the corner...

This graph shows the periodic surge in demand for producer goods that produces the business cycle, but it does not actually show time preference because it neglects savings.  I decided "what the hey, why not add in the savings and forget about services and just see what happens?" 

So, I did --
Again, I was amazed.  Adding back in savings, I got a graph just like I would expect if the 'overconsumption' hypothesis were true.  The curve flattens out -- note the y-axis scale.  Most of what is going on is just 'noise,' the basic curve is flat, and very slightly falling over the course of the business cycle, from about 1983 to 2003, to signify that ever slightly more expenditure is devoted to consumption rather than savings over the period of inflation.  During periods of recession, it turns back up.  Amazing.

But I couldn't stop there, because as I said, I had neglected services, and obviously their contribution might be significant.  So, I decided to make the approximation I had made before -- that whatever was going on in goods markets was probably reflected in the service market as well, so that if I assumed the proportion of production and consumption spending on services was the same, I could approximate the outcome as if I had the real data.  I just multiplied the durable goods/non-durable goods ratio by the spending on services, and put that part of services to the numerator and the rest to the denominator.  I also flipped the ratio, so that the graph of expressed time preference would follow the convention that 'high' time preference favored consumption, and 'low' time preference favored savings.  Basically, I put consumption in the numerator and savings in the denominator.  And to see if the time preference responded to interest rates, I also graphed it versus the 10-yr Treasury yield, which I chose rather arbitrarily.

Here it is --

The blue line is time preference (as expressed/approximated by my metric) and the red line is the 10-yr yield.

It is not perfect, but it does largely show falling time preference as interest rates rose from 1959 to 1982, followed sometime later by a broad inflection and rising time preferences as rates fell from 1982 until today.  Overall, the curve is fairly flat (note the scale of the right y-axis), while being nudged up and down by the interest rate.  All pretty consistent with the notion of 'overconsumption.'  Note especially the period from 2003 to the present, a rather steep rise in the face of the monumental inflation of that period.  If it really is savings and investment that gets the economy out of the slump -- which the Austrian school and common sense would suggest that it is -- it does not bode well for us...

Altogether, I thought these graphs were excellent and interesting confirmations of some of the basic assertions of the Austrian school, if not proofs, and I never would have found them if I hadn't stubbornly clung to an incorrect idea.

Sometimes it is good to be wrong.

No comments:

Post a Comment