Believe those who are seeking the truth. Doubt those who find it. Andre Gide


Monday, September 8, 2014

Who's Afraid of Deflation?

Everyone knows that deflation is bad. Bad, bad, bad. Why is it bad? Well, we learned it in school. We learned it from the pundits on the news. The Great Depression. Japan. What, are you crazy? It's bad. Here, let Ed Castranova explain it to you (Wildcat Currency, pp.160-61):

Deflation means that all prices are falling and the currency is gaining in value. Why is this a disaster? ... If you hold paper money and see that it is actually gaining in value, it may occur to you that you can increase your purchasing power--make a profit--by not spending it...But if many people hold on to their money, this can dramatically reduce real economic activity and growth...

In this post, I want to report some data that may lead people to question this common narrative. Note, I am not saying that there is no element of truth in the interpretation (maybe there is, maybe there isn't). And I do not want to question the likely bad effects that come about owing to a large unexpected deflation (or inflation).  What I want to question is whether a period of prolonged moderate (and presumably expected) deflation is necessarily associated with periods of depressed economic activity. Most people certainly seem to think so. But why?

The first example I want to show you is for the postbellum United States (source):


Following the end of the U.S. civil war, the price-level (GDP deflator) fell steadily for 35 years. In 1900, it was close to 50% of its 1865 value. In the meantime, real per capita GDP grew by 85%. That's an average annual growth rate of about 1.8% in real per capita income. The average annual rate of deflation was about 2%. I wonder how many people are aware of this "disaster?"

O.K., well maybe that was just long ago. Sure. Let's take a look at some more recent data from the United States, the United Kingdom, and Japan. The sample period begins in 2009 (the trough of the Great Recession) and ends in late 2013. Here is what the price level dynamic looks like since 2009:


Over this five year period, the price level is up about 7% in the United States and about 11% in the United Kingdom. As for Japan, well, we all know about the Japanese deflation problem. Over the same period of time, the price level in Japan fell by almost 7%.

Now, I want you to try to guess what the recovery dynamic--measured in real per capita GDP--looks like for each of these countries. Surely, the U.K. must be performing relatively well, Japan relatively poorly, and the U.S. somewhere in the middle?

You would be correct in supposing that the U.S. is somewhere in the middle:


But you would have mixed up the U.K. with Japan. Since the trough of the past recession, Japanese real per capita GDP is up 15% (as of the end of 2013)--roughly 3% annual growth rate. Is deflation really so bad? Maybe the Japanese would like the U.K. style inflation instead? I don't get it.

I have some more evidence to contradict the notion of deflation discouraging spending (transactions). The evidence pertains to Bitcoin and the data is available here: Blockchain.

Many people are aware of the massive increase in the purchasing power of Bitcoin over the past couple of years (i.e., a massive deflationary episode). As is well-known, the protocol is designed such that the total supply of bitcoins will never exceed 21M units. In the meantime, this virtual currency and payment system continues to see its popularity and use grow.


One might think that given the prospect of continued long run deflation--i.e, price appreciation (it's hard to believe that holders of bitcoin are thinking anything else)--that people would generally be induced to hoard and not spend their bitcoins. And yet, available data seems to suggest that this may not be the case:


Maybe deflation is not so bad after all?  Let's hope so, because we may all have to start getting used to the idea!

Additional readings:
[1] Good vs. Bad Deflation: Lessons from the Gold Standard Era (Michael Bordo and Angela Redish).

[2] Deflation and Depression: Is There an Empirical Link? (Andy Atkeson and Pat Kehoe).

[3] The Postbellum Deflation and its Lessons for Today (David Beckworth).

34 comments:

  1. Good point. Also Norway in the 1930s. And for your collection:
    http://www.mpls.frb.org/Research/sr/sr331.pdf

    ReplyDelete
    Replies
    1. Thanks Dave! I will add a link to the paper you cite (can't believe I hadn't seen it before!).

      Delete
  2. David, interestingly, I am one of those folks who have put some thought into postbellum deflation. I have published an article on it. See here: http://people.wku.edu/david.beckworth/postbellumdeflation.pdf

    ReplyDelete
    Replies
    1. Thank you for alerting me to your paper, David! Looks like we share Figure 1. So, what is your take on the Japanese experience? I presume that you would classify it in the "non-benign" category?

      Delete
  3. all right but only if you define deflation as a prices drop. Then, you can say that when prices decrease it's possible it's a good or a bad thing. I think deflation it's something else : prices drop AND contraction of economic activity. Best example: 1929-1932

    And as you write: it's necessary that it's "a period of prolonged MODERATE (and presumably expected) deflation"

    ReplyDelete
    Replies
    1. Sure, we can find examples where depression is associated with deflation. But see the Atkeson and Kehoe paper above.

      We should also keep in mind that contractions might be caused or exacerbated by things other than deflation. Deflation may only be a symptom and not a cause of things. Just wanted people to keep an open mind. So often it seems they fear deflation. Even moderate deflation.

      Delete
    2. Sergedago: you might want to look at the NBER Paper 10268: Atkeson/Kehoe "Deflation and Depression: Is There An Empirical Link". They found no significant statisical correlation between deflation and depression across a wide number of countries and timelines. Personally what I found most interesting was the fact that America was a statistical outlier during the depression of the 1930s. Many countries (i.e. UK, Nordic countries) had economic growth and deflation. America had higher deflation and worse decline in the economy. It certainly suggests that the government policies pursued by America in the great depression made matters worse rather than improving conditions.

      Delete
    3. I read the paper and note the 2 definitions: "we define a deflation as a negative average inflation rate and a depression as a negative average real output growth rate". So I can understand what it occurs (hoping that I am not totally wrong!). For me, the word deflation can be used only when there is depression (even weak growth). It is not only the prices drop. I know that the decrease of the prices can intervene in period of growth, the history shows it to us. But the word deflation is negatively connoted and it is the reason why I associate it with depression. So it's a question of definition and from this point of view the paper and your post are very clear.

      Delete
  4. An question that falls out of your piece:

    If Japan had a consistently positive real rate and similar per-capita GDP growth, does that mean the real Fed Funds rate could have been kept at 1% since 2009?

    ReplyDelete
    Replies
    1. You mean, like Canada?

      http://newmonetarism.blogspot.com/2014/07/monetary-policy-canada-and-united-states.html

      Delete
    2. I believe Canada's real rate was still <0. Higher than the U.S., but not as lofty as Japan's.

      The feedback between policy, real rates and growth is much more complex than various explanations (natural rate, safe assets, secular stagnation) purport it to be.

      Delete
  5. 1. I think you're too easy on the "common narrative." Could be true? You can't construct a theory for it, let alone find evidence, as you point out.
    2. The bitcoin evidence doesn't really apply. Transactions may be up, but if you look at the block chain (I'm told), most bitcoins were bought and held.

    ReplyDelete
    Replies
    1. Consider the OLG model studied by Bruce Smith (based on your own work). There is money and capital (linear technology). Too much deflation distorts wealth portfolios: banks are induced to hold too much in the way of real money balances; investment suffers.

      As for Bitcoin, doesn't matter what most people are doing with them. I see people purchasing things with BTC. What are these people thinking? When I talk to them, they expect long-term price appreciation. Yet they spend their bitcoins anyway. I do not get a sense that they are spending their BTC at an inefficiently slow rate that needs to be speeded up with an inflation.

      Delete
  6. Here is an easily understood piece on the deflation from Frank Shostak: http://mises.org/daily/4618/Is-Deflation-Really-Bad-for-the-Economy

    I think he gets at the deflation v. depression issue quite clearly.

    ReplyDelete
    Replies
    1. I don't really subscribe to his idea about deflation being part of a "healing" process, however.

      Delete
  7. I don't know, David. All those farmers whose debts skyrocketed during the Gilded Age might not hold deflation in such great esteem as you.

    ReplyDelete
    Replies
    1. CW, for sure, and as I take care to mention in my post, an unexpected deflation is going to impose a burden on debtors (including firms with fixed nominal wage obligations). But that's not what I'm talking about here. We're talking 35 years of low and steady deflation. I doubt very much that debt contracts had very long terms back then, but I could be wrong. Can you provide us with some evidence?

      Delete
    2. Interesting post, DA. Also related: Marcus Nunes compared the Panics of 1873, 1893, 1929, and 2008 in “Three panics and a nonevent”.

      Dunno about the term of debt contracts, but I have suggested that following the US Civil War, relatively low private debt opened a door to growth and also minimized a harmful sided-effect of deflation. Except among all those farmers, of course.

      Delete
  8. David,

    I think I recall you posting something about this before, and so I might be repeating myself, but I believe the booklet from Prof. George Selgin "Less than Zero" makes the case for what he calls a 'productivity norm' rate of money growth. It's from 1997, and it's been a long time, but I think he gets at this deflation issue quite effectively.

    ReplyDelete
  9. "And yet, available data seems to suggest that this may not be the case:"

    Part of the huge ramp up in transactions was due to the debut of Satoshidice, a bitcoin gambling platform, in April 2012, which would go on to account for 50-60% of blockchain activity (not sure if that's still the case). In any case, I'm not sure if gambling qualifies as spending bitcoin, I'm guessing you'd probably prefer if they were buying stuff with it.

    I'm also wondering what your charts of Japan, US, and UK would look like if they went back to the 1990s.

    My understanding of why deflation was bad always had to do with sticky prices, particularly for labour. Because prices can't fall to equilibrate, you get an excess supply of labour. Ensuring constant inflation is a way to ensure you don't get this sort of excess.

    ReplyDelete
    Replies
    1. You can well guess what the charts would look like going back further. But what would be the point? That Japan's relative underperformance was due to deflation? For 25 years? Wages may be sticky, but not *that* sticky!

      Delete
    2. This PPP-derived FRED chart shows the relative GDP per capita performance evened out about 1999.

      http://research.stlouisfed.org/fred2/series/PGDPUSJPA621NUPN

      Delete
    3. Sure, I am very familiar with this picture. But what are you suggesting? That inflation and deflation is responsible for this growth dynamic? Come on, my friend.

      Delete
  10. The answer to your blog post question is, of course, Paul Krugman, who seems to fear even "lowflation".

    "low inflation or deflation have the effect of prolonging slumps"

    http://www.princetonmagazine.com/paul-krugman/

    ReplyDelete
    Replies
    1. Sure. And one reply to PK is this: an economic boom under fiscal austerity and at the ZLB: http://andolfatto.blogspot.com/2013/09/another-look-at-koizumi-boom.html

      Delete
  11. The interesting thing would be to hear Ben Bernanke's response. His "BOJ" speech, I think in 2002, kicked off lowflation phobia. Little did he know Japan per-capita GDP growth would differ little from that of the U.S. over the next twelve years despite Japan's continuing deflation.

    ReplyDelete
  12. "We should also keep in mind that contractions might be caused or exacerbated by things other than deflation. Deflation may only be a symptom and not a cause of things."

    Thanks, I really wish more economists would realise this obvious argument.

    Also, with respect to Bitcoin, I would argue that there are still differentiating factors as opposed to traditional currencies that must be considered when analysing deflation. One, for example, is that Bitcoin is not used as a unit of account, so it does not affect pricing of goods. Or is the issue of parallel currencies, as this might cause a subsitution effect which in the typical macroeconomic analysis is missing. For example when I have to decide between holding bitcoins and fiat, I might spend fiat first and only use bitcoin when I run out of fiat.

    I'm not saying that you're wrong, I agree with you, but I'm just pointing out that the criticism of deflation is more nuanced and that the knee-jerk reaction "deflation bad" is not necessarily representative of all the criticism.

    On the other hand, there are reasons why the more advanced criticism of deflation are not relevant for macroeconomists either. With all due respect to Castronova, for example, games have fixed ratios of production structure and there is no flexibility in the composition of production plans (you cannot substitute producer goods against each other, e.g. labour vs. machinery), and for the company providing the game, virtual goods have zero marginal costs. MMORPGs also typically limit or prohibit entirely real money trading in order to prevent these zero marginal costs to be passed down to the players. This makes sense, because for the player, the utility of the game is in providing a challenge for the time spent playing rather than in the amount of virtual goods in the virtual economy. In real world on the other hand, decreasing the working time and increasing the number of consumer goods is in general good. In other words, in a game, the virtual goods are means and the experience while playing is the goal, while in the real world, the work is the means and the consumer goods are the goal.

    ReplyDelete
  13. An interesting post, but I fear you have missed the crucial factor: Japan’s population has been affected by a long-run decline in population since 2004.

    Given that real per capita GDP is calculated by dividing real GDP for a given year by the population in that year, and Japan's real GDP really has not been that spectacular, what does this suggest?

    There is no great mystery here. See my post:
    http://socialdemocracy21stcentury.blogspot.com/2014/09/japanese-real-per-capita-gdp-and.html

    ReplyDelete
    Replies
    1. Um...it suggests that the population is decreasing? Or maybe you'd rather live in China, where real GDP is high, but per capita GDP is low? I'm afraid I do not understand your point.

      Delete
    2. Your Japanese real per capita GDP growth of up to 15% is not because of spectacular real GDP growth under deflation, but mainly population decline.

      The whole notion that it is "beating" the UK or US is invalid. The US or UK have different population dynamics. If their population was stagnating or declining like Japan's, then their real per capita GDP growth would be unusually higher too.

      Delete
    3. How do you know? That is an unsubstantiated assertion on your part. Per capita growth comes from productivity gains. And these gains could well have occurred with the population growing (or shrinking). So your point is invalid.

      Delete
  14. Good piece!

    two remarks:
    1. what if the postbellum dynamics were driven by a tech shock - pulling down prices and up GDP? And could the price index suffer from a quality-adjusted and/or basket composition problem?

    2. On BTC, I also think that one of the reason for high value and number of transactions can be explained by the use of BTC for illegal activities such as bypassing capital control in china (since they were suspended the BTC stopped to rise) and other similar cross-border activities

    3. On Japan: no idea; although I dont know how they will manage the public debt

    ReplyDelete