Tuesday, July 15, 2014

Spontaneous Combustion Theory of Inflation


There have been many posts trying to ridicule the Spontaneous Combustion Theory of Inflation.  The name is to make it seem ridiculous that inflation could just suddenly take off.  However, in the real world hyperinflation does just suddenly take off.   There is good theory as to how inflation suddenly takes off.  Making a silly name for something is a debating technique but it is not a sound logical argument.  Hyperinflation is a real thing.


26 comments:

  1. Also, I don't think the phrase "spontaneous combustion theory" sounds all that silly...(pile of oily rags in the garage... you know the story) but "spontaneous *human* combustion" does! (remember that?)

    ReplyDelete
    Replies
    1. I think most people think of that. People don't take hyperinflation seriously till it is too late.

      Delete
  2. Since hyperinflation is a lack of faith in the currency, it makes sense that it would happen fairly spontaneously. How does a man lose faith in his wife? He has his suspicions at first that she is being unfaithful, and he makes some minor changes in the way he conducts his affairs to coincide with that suspicion. But then one day he knows without a doubt that she is playing the whore, and he completely severs ties with her and distances himself from her as much as possible. This moment of clarity that the jaded husband experiences is what the market experiences when hyperinflation begins. The realization is the beginning of the stampede away from the object in which there is no more faith, and it happens very fast.

    ReplyDelete
    Replies
    1. I think there's a problem with this analogy: namely that there's a big difference between a huge economy with people doing all manner of things, and one guy and his wife. I think on one extreme you have an individual and his wife, and on the other we have a diffusion problem: you drop a single drop of colored dye into a container of water, and it spreads out evenly. What's the probability that all the dye molecules will regroup into a volume the size of a drop again? It's not zero!... but it's very very small. I think the actual probability of sudden loss of faith lies somewhere between those extremes... perhaps closer to the diffusion example... like the probability of 50 million men all suddenly losing faith in their wives... many many orders of magnitude more probably than the dye all going back to a single drop, but still a very small probability.

      Delete
    2. But if you read my stuff it is like an avalanche. Just takes a single rock falling and all the sudden you have a hillside coming down. It is not that all the rocks just decided at once, it is that one hit others, which came loose, which hit others, in a chain reaction. Avalanches, earth quakes (ground ripping after stress buildup), forest fires, and hyperinflation happen far more often than some dye all going back to a single drop.

      Delete
    3. Vincent, I figured somebody would point out that not all the probabilities are independent. Sure I accept that, however, they might be approximated as independent up to some tipping point (say 20% of the adult population, or 30%... I don't know... at least in 1st world economies in which no other disasters are happening).

      Re: earthquakes, forest fires, avalanches: we may not be able to predict when they happen, but we can assign probabilities. And we can use how we did (statistically) to tune the models that produce those probabilities. Here's a challenge for you: why don't you do the same: say for each country you can get any data for (maybe 100 of the 200 or so countries on Earth?), assign a probability of experiencing hyperinflation over the next 10 years. Then in 10 years you'd have some great data on how well you did!

      Jason doesn't have that precisely, but he does have a measure he calls kappa which indicates the degree to which the QTM applies for an economy. This allows him to make predictions (like his Canada prediction) for different economies.

      Shoot, if you're going to get anywhere with your hyperinflation theory, I think you need to set up a big matrix to grade yourself by.

      I'm sure people predicting earthquakes probably have an expected number of earthquakes that will occur over the next 10 years of varying magnitudes, over all the regions on Earth. You should be able to do something similar, don't you think?

      Delete
    4. "I'm sure people predicting earthquakes probably have an expected number of earthquakes that will occur over the next 10 years of varying magnitudes, over all the regions on Earth. You should be able to do something similar, don't you think?"

      Yes, I agree. But with fiat currencies you can use the Forex markets to grade your predictions. If I end up in a large yacht, then I get an "A".

      Delete
  3. Noah posted about gold, and I commented:

    http://www.bloombergview.com/articles/2014-07-16/gold-dreamers-face-harsh-reality

    ReplyDelete
    Replies
    1. See, if you had a numerical probability for each country on Earth, for the next 10 years, and you did good, then you could shut guys like this up (10 years from now):

      http://www.bloombergview.com/articles/2014-07-16/gold-dreamers-face-harsh-reality#comment-1490129826

      :D

      Plus you'd be laying your cards on the table: rather than a vague "high risk" you'd have concrete numbers: one for each country. Even if you just make the numbers up, at least it wouldn't be a vague statement. You don't want to come off like the guy on the street corner with the sign reading "Repent Now!! Before it's too late! The end is nigh!" You don't have to go and pick a date for the end of the world (remember that guy?), but I think that street corner guy would get a totally different reaction if his sign said "Probability for the world ending over the next 10 years, according to my calculations, is 0.52" ... people might stop and ask him about that.

      Delete
  4. Vincent, speaking of making inflation predictions, after presenting his model of P to Scott Sumner, Scott asked Jason how well it did against alternatives, such as TIPS spreads. Jason did a quick analysis today, and these are his results:

    http://informationtransfereconomics.blogspot.com/2014/07/better-than-tips.html

    I also really like how his results stack up against the Fed model here:

    http://informationtransfereconomics.blogspot.com/2014/07/notes-from-ben-bernanke-and-p-model.html

    Especially impressive to me is how he "predicts" P from 1990 to 2014 by fitting his three parameters to the 1960 to 1990 data only. In contrast the Fed model, with data over about the same time period (1955 to 1990) has 7 to 8 parameters to fit, and predicts deflation by 1992!

    ReplyDelete
  5. Shoot, Jason's on a roll: two more posts today on predicting inflation. Here's a link to the second one (specifically predicting inflation in the US out quite a ways):

    http://informationtransfereconomics.blogspot.com/2014/07/us-inflation-predictions.html

    In the middle one (the 1st post today), he make this bold statement at the end:

    "That is a pretty startling piece of information. It means you likely can't do any better than the information transfer model in predicting inflation in the medium term (5-15 years out)."

    He's just itching for someone to prove him wrong there Vincent! Maybe you ought to at least put your hat in the ring, just in case... wouldn't it be fun to look back 5 to 15 years from now, and remind him of the day it went wrong for his theory?

    ReplyDelete
    Replies
    1. Sorry, I know regular inflation is not your thing. You like the avalanche (hey, that's your comparison, not mine!) theory of hyperinflation. But I still think you need to expand into the regular inflation prediction business... I still don't have any reason to doubt what Sadowski wrote: that hyperinflation doesn't suddenly appearing after years of low to moderate inflation in a 1st world country: there's always multiple years of high inflation preceding it.

      Delete
  6. Commented on seekingalpha:

    http://seekingalpha.com/article/2324865-is-hyperinflation-possible

    ReplyDelete
    Replies
    1. I got a "the earlier poster who references japan makes the most sense but...". :-)

      Delete
    2. Here's a discussion of gold a bit off the beaten track:

      http://moneymarketsandmisperceptions.blogspot.com/2014/07/why-matt-yglesias-should-spend-more.html

      Delete
  7. Vincent, here's an inflation theory to add to your collection: "Three digit inflation by 1978!" I downloaded it so I could read it better... it's on page 33. (Don't laugh, he might be onto something!) :D

    ReplyDelete
  8. Vincent, Scott makes an interesting observation:

    "...the public and economists have a radically different view of what the word ‘inflation’ means. To average people inflation is something that reduces living standards, by raising the cost of living. Economists would call that scenario a supply shock, or a fall in real GDP. Economists think of inflation as something that raises both wages and prices, with no first order effect on real income."

    As does Jim Glass:

    "A while back I went to FRED and got the correlation between CPI and wage rate increases over the last 50 years, it was 99+%"

    ReplyDelete
    Replies
    1. The Keynesian idea is to give people a real pay cut by printing money. I commented.

      http://www.themoneyillusion.com/?p=27180&cpage=1#comment-359488

      Delete
    2. Well you got a response from three people on that, including Sumner.

      Delete
  9. One more for you: JP Koning did a post on using P*T instead of P*Q (=P*Y) in the exchange equation (I guess similar to how Irving Fisher originally had it). Nick Rowe seems to support this idea in the comments. It would be quite different though. How do you suppose that would affect your model?

    http://jpkoning.blogspot.com/2014/07/fedwire-transactions-and-pt-vs-py.html

    ReplyDelete
  10. Commented on Brad Delong's post.

    http://delong.typepad.com/sdj/2014/07/is-the-us-at-immediate-risk-of-rapidly-becoming-argentina-as-a-result-of-expansionary-counter-cyclical-policies-the-bet.html#comment-6a00e551f08003883401a511eb6feb970c

    ReplyDelete
  11. Someone asks about trying to calculate the probability of hyperinflation and I commented with a link to my simulator:

    http://quant.stackexchange.com/questions/14195/probability-of-hyperinflation-as-a-function-of-probability-of-soverign-default

    ReplyDelete
  12. Nick Rowe against repeats his statement about dedicated targeting of 0% inflation = communism:

    http://worthwhile.typepad.com/worthwhile_canadian_initi/2014/07/the-soft-fade-from-monetary-to-fiscal.html

    ReplyDelete

Looking for polite debate on ideas. Never attack a person. Be nice.