Thursday, June 11, 2015

Exchange rates and irrational markets

An anonymous commenter asked if I could do a version of this post for the Euro/British Pound exchange rate and the model comes out with an interesting result/prediction. The exchange rate has been falling over the past five years (1 Euro buys fewer British Pounds today than five years ago).

The information transfer model (ITM) essentially says that the exchange rate for two currencies is roughly proportional to the ratio of NGDPs for the two countries issuing those currencies (X12 ~ NGDP1/NGDP2). However in general there are large fluctuations around the model -- exchange rates are very volatile. I speculate that the reason is that traders have an incorrect theory of what sets exchange rates in this post.

Anyway, here are the model's results:


This model works well -- up until the financial crisis, when the Pound loses value against the Euro even though the Great Recession was about the same strength in the two regions. This probably reflected the incorrect view that the UK's rather large level of quantitative easing (QE) would devalue the Pound. However, in the ITM, QE does not impact NGDP so there was no commensurate rise in UK NGDP relative to EU NGDP: the "fundamental" value (given by the ITM) has only slowly drifted upward due to stronger growth in the UK relative to the EU. Ever since that incorrect (according to the model) move by the markets, the exchange rate has been drifting back to the "fundamental" value.

The trends indicate that exchange rates will return to their fundamental value about early to mid-2016. The Greek crisis (e.g. if Greece leaves the Euro) could push exchange rates away from this, and in any case, markets will probably over-shoot this equilibrium (making the Pound stronger than it "should" be).

Much like in interest rate markets, there are large fluctuations in currency markets that deviate from the ITM trend. It is possible these "irrational" periods represent real genuine economic effects or the impact of market expectations. However it seems perfectly consistent with episodes of market herding behavior resulting in non-ideal information transfer and market failure. That is to say there might not be any economic model underlying these effects -- just a sociological one.

That's also why I don't put too much stock in Scott Sumner's claims of exchange rates as indicators of the effect of monetary policy. They're indicators of what people think monetary policy does, but not necessarily of what monetary policy actually does.

Then again, the ITM may be wrong ... I definitely haven't put my money where my mouth is! But then again, the ITM says that markets don't really process information so there's no theoretical reason to believe bets reveal beliefs.

11 comments:

  1. All well and good, but wouldn't you really have to set both the Euro and GBP up in a relationship with the Dollar (i.e. vs. a basket of worldwide currencies) to determine the relative level of each, if you wanted to really look at devaluation? It may make a difference, as I recall the Euro actually went up against the Dollar early on in the financial crisis, so that the GBP may not have moved as much relative to the Dollar (or I may be wrong).

    ReplyDelete
    Replies
    1. Hi Todd,

      It's not necessary in the model -- any two currencies are essentially the ratio of their NGDPs. You could probably construct a "world GDP" by adding together (or averaging) a bunch of GDPs and compare to that, though. That might be interesting.

      Delete
    2. -- *the exchange rate between* any two currencies ...

      Delete
  2. Thank you for doing that Jason very kind of you.

    ReplyDelete
  3. What about modelling it without a crash or any QE.

    Using a ten year period of say calm in the markets ?

    ReplyDelete
    Replies
    1. Unfortunately the Euro only goes back to 1999, so the graph above represents all of the data available. However, a period like you describe is essentially the period from 2000 to 2007 in the graph above.

      Delete
  4. It is very weird to me to say that an economic model is not a sociological one or that a sociological model is not an economic model....if it involves people in markets interacting it is both economic and sociological....

    ReplyDelete
  5. This comment has been removed by the author.

    ReplyDelete
  6. "The information transfer model (ITM) essentially says that the exchange rate for two currencies is roughly proportional to the ratio of NGDPs for the two countries issuing those currencies (X12 ~ NGDP1/NGDP2). "

    But isn't the euro area GDP much bigger than britain's GDP? Why are exchange rates so close to 1?

    Also, proportional in the same currency or each GDP in its own currency?

    It could be interesting to update this prediction, I think it performed very well.

    ReplyDelete
    Replies
    1. When I wrote X ~ G1/G2 it was representing a proportional relationship X = c G1/G2 + s(t) for some constant value c and stochastic shocks s(t) with approximately zero mean. That overall factor c would account for the relative size of the countries (and would change if countries entered or exited the EU/Euro Area).

      And yes, it'd be good to check back in here ... I did about a year after this post but I'll need to do it again.

      Delete

Comments are welcome. Please see the Moderation and comment policy.

Also, try to avoid the use of dollar signs as they interfere with my setup of mathjax. I left it set up that way because I think this is funny for an economics blog. You can use € or £ instead.

Note: Only a member of this blog may post a comment.