Philips curve, stimulus

If stimulus works, if Keynesianism is a good enough approximation to the modern American economy, there should be a relationship between unemployment and inflation, the Philips curve.  Higher unemployment should lead to lower inflation, and vice versa.

There is no relationship.

**

Philips curve

Keynesianism says that unemployment is caused by lack of aggregate demand, which should also manifest as falling prices, deflation.

Supposedly, when there is insufficient aggregate demand, stimulus is needed, there is unemployment, but no inflation.  Stimulus will not cause inflation, it will cause people to be employed.  Only when just about everyone is employed, when there is sufficient aggregate demand, will increased aggregate demand cause inflation.

Of course prices never seem to fall in practice, so Keynesians then say that lack of inflation is deflation.  And if it looks like prices are rising at an alarming rate, Keynesians then say that lack of core inflation is deflation.

Now you might think that core inflation is stuff like food and fuel, stuff where the price is easy to define and measure, stuff you cannot do without.  But no, core inflation is stuff like ipods, washing machines, and laundry detergent.  And since you cannot really say how the value of yesterday’s ipod compares with today’s ipod, the wise folk at the Bureau of Labor Statistics will make a hedonic adustment.

Now ipods are certainly getting better, but you might find it hard to say how much better.  The wise folk at the Bureau of Labor statistics, are, however, wise, being civil servants, and have no difficulty saying how much better ipods are getting

Now you might think that washing machines and laundry detergent are getting more expensive and markedly worse, since they are no longer very effective in getting clothes clean, possibly thanks to regulations restricting power usage, water usage, and the use of phosphates in laundry detergent, but the wise folk at the Bureau of Labor statistics know that they are getting better and better, because, after all the purpose of those regulations was to make them better (not better for getting clothes clean, but better nonetheless).  So even though your clothes are not as clean, modern washing machines get a big hedonic adjustment.  Today’s washing machine is supposedly more washing machine than yesterday’s washing machine.

In short, core inflation is inflation in those items were no one can possibly make an accurate estimate of inflation.  And, unsurprisingly, core inflation tends to be a lot more politically correct, which is to say lower, than non core inflation.

If there is massive unemployment, there must be insufficient demand, which can be remedied by stimulus.  And if stimulus fails to remedy it, obviously there was not enough stimulus.  And if stimulus leads to inflation – well, it cannot lead to inflation while there is massive unemployment, so there is none.

The flaw in Keynesianism is that “aggregate demand” involves adding apples to oranges.  It ignores, as trivial, the questions of what will be produced, what jobs will people do.  It simplifies the economy enough that government can manage it.  That Keynesianism is a simplification does not make it wrong.  All theories are necessarily simplifications.  What makes it wrong is that the world does not behave as described.

5 Responses to “Philips curve, stimulus”

  1. Bill says:

    The Philips Curve story is even more interesting than that. If you break the graph into 1948-70 and 1970-present, the PC does pretty well in the earlier period and disastrously badly in the latter period. The Philips curve paper was published in 1958, and it fully entered the consciousness of policymakers and analysts in the 60s. So, the PC was a regularity which held up until policymakers started trying to use it (publicly) as a knob to adjust the economy.

    The breakdown of the PC led to a flurry of activity in macroeconomics in the 70s and 80s. Unfortunately, nobody came up with anything particularly good to replace Keynesianism with. So, we have the two current cults in macro: the “saltwater” or Keynesian camp and the “freshwater” or real business cycle camp. Each retails a (different) false theory with various ex-post fixes glued on. The “recalculation” theory some libertarians seem to like also does not seem to sport much evidence. Recessions don’t tend to be associated with big sectoral shifts in employment or production, and they tend to hit all sectors. The business cycle is pretty mysterious, and, for that reason, tends to spawn charlatans.

    The claims about inflation seem false to me. On core vs non-core inflation, you can go to the BLS website to see the all-items index (CUUS0000SA0), all-items less food&energy (CUUS0000SA0L1E), energy (CUUS0000SA0E), food&beverages (CUUS0000SAF). Their data extraction tool does not seem to provide for convenient perma-linking, so I can’t just give you a link. From the 1982-84 baseline to end 2010, cumulative inflation in these four indexes has been 118%, 121%, 111%, 120%, i.e. not different. I also think you are wrong about the size of the hedonic adjustments. They are very large for computers. The price of computers is calculated (effectively) in units of something like $/MFLOP instead of $/computer. This strikes me as the right think to do. Quality change outside computers (and cars) is still mostly unaccounted for, AFAIK (though I am not a specialist in price indexes). If you think quality is mostly increasing, then inflation is overstated for this reason. If you think that quality is mostly decreasing, then inflation is understated for this reason. Most experts still think that inflation is, on balance, overstated though. Do you have a link for the size of the hedonic adjustment for washing machines?

    • jim says:

      PC does pretty well in the earlier period and disastrously badly in the latter period. The Philips curve paper was published in 1958,

      We would expect Keynesianism to be true to the extent that floor prices are set by government for many wages and/or goods, to the extent that government enforces unions and cartels as was the case during the great depression.

      I don’t have data for wage supports and agricultural price supports in the period 1948-1970, and if I had data, would find it difficult to know if the price supports were effectual or merely symbolic. To my recollection, however, the big change in the economy that happened around 1970 was not a reduction in price supports or state enforced cartels and unions, so to the best of my recollection, that beautiful theory (that Keynesianism used to be true because of government intervention) is killed by ugly data – or at least ugly anecdote, I don’t have real data.

      An alternate theory that I find more plausible is that high levels of inflation kill employment by disrupting economic calculation – that the stimulus medicine works as Keynesians describe, up to a point, and governments tend to exceed that point.

      This strikes me as the right think to do. Quality change outside computers (and cars) is still mostly unaccounted for, AFAIK (though I am not a specialist in price indexes)

      In my own area of expertise, they make hedonic adjustment for custom software and mass market software, which is inappropriate. Computers are getting better. Mass market software is not getting better, merely cheaper. Today’s Microsoft Word is not a significant improvement on much earlier versions of Microsoft Word. Custom software is neither getting better nor markedly cheaper. The hedonic adjustments within my area of expertise are merely for the purpose of fudging the numbers.

      • Bill says:

        Yeah, I believe that, though I might chalk it up to herd mentality rather than intentional lying. When I tell people that Microsoft Word 4, which took up <1MB on disk and loaded fast, was just as good or better a program than current MW for 99% of users, they tend to look at me as if I am a bit off. Even people who used both programs tend to think I am making a funny rather than a true assertion. And I think I am saying something about as disputable as "the sky is blue."

        The most recent iteration of MW represents substantial regression. There is a real time penalty associated with trying to figure out where they have hidden everything in the bizarre menu/toolbar/thingy reorganization. Adjusting for this kind of thing would be quite daunting, though.

    • nyan sandwich says:

      >The Philips Curve story is even more interesting than that. If you break the graph into 1948-70 and 1970-present, the PC does pretty well in the earlier period and disastrously badly in the latter period. The Philips curve paper was published in 1958, and it fully entered the consciousness of policymakers and analysts in the 60s.

      Another brick in the 1973 wall. Do we know what happened there yet?

      • jim says:

        Something very bad happened in 1972-1973, but hard to say, or even define, what is cause and effect.

        The song “I’m living in the seventies” would suggest that we stopped believing in manliness.

Leave a Reply for Bill