Marginal Revolution: Uncovered interest parity -- why does it fail?

November
S	M	T	W	Th	F	St
1	2	3	4	5	6	7
8	9	10	11	12	13	14
15	16	17	18	19	20	21
22	23	24	25	26	27	28
29	30

« The Caroline Hoxby vouchers debate | Main | Thomas Schelling on Iranian nukes »

Uncovered interest parity -- why does it fail?

This is one of the most puzzling puzzles in macroeconomics: that foreign-exchange speculators are not very good at linking domestic money and bond markets to the foreign exchange market. Not enough money seems to be engaged in betting that a currency with a high nominal interest rates will not decline in value fast enough to make investing in its securities unprofitable. Why not? It's an easy thing to do.

James Hamilton [correction: *Menzie Chinn*] adds more. What are the main hypotheses for why high nominal interest rate currencies appear to outperform the market?

1. The so-called "peso problem." Someday the roof will cave in on these currencies. The Asian financial crisis was only a small disruption compared to what will happen someday. Our current data set is incomplete and does not represent the real population.

2. Holding crummy currencies is riskier than CAPM and variants indicate. Most likely, the relevant investors are not and cannot be well-diversified. So their high pecuniary returns are offset by the risk they bear. If thirty percent of your wealth is in the South African Rand, the relevant measure of your risk may be the variance of that currency, not the covariance of that currency with some broader international market portfolio. Economic theory usually measures risk by looking at the latter.

3. Many investors stay away from crummy, high nominal interest rate currencies for fear of what their wives (or bosses, consider an agency problem at a financial firm) would say, should they lose money. The relevant markets are segmented. This is linked to #2.

4. This used to be a puzzle, but now the profit opportunity has been identified. The supposed additional risk of the high nominal interest rate currency is phantom. People now jump into high nominal interest rate currencies, at least when such investments are appropriate to restore an equilibrium of risks and returns. We should expect the paradox to disappear in future data sets.

Observation: Do not ever write or say "CAPM model." Do not ever write or say "ATM machine."

Posted by Tyler Cowen on October 31, 2005 at 07:03 AM in Economics | Permalink

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d8341c66b253ef00e5508350c08833

Listed below are links to weblogs that reference Uncovered interest parity -- why does it fail?:

Comments

I don't think DeLong has a clue. The link is known to every currency
speculator and some do make money. But the foreign currency markets are
fairly "efficient," although some are more so than others (the market for
the New Zealand dollar is much less deep than the market for the U.S.
dollar). It's one thing to know the theory, quite another to bet real
money on currency changes.
The head currency trader at the New York office of the Bank of New Zealand
was a client of mine when I was a broker.
He once commented on the connection--and disconnection--between textbook
theory and realtime trading.
If DeLong thinks it's so easy, why isn't he running a hedge fund and making
big currency bets?

Posted by: Bill Stepp at Oct 31, 2005 7:35:15 AM

By the way, the Econbrowser post to which Brad refers was written by Menzie Chinn, not me.

Posted by: James Hamilton at Oct 31, 2005 8:24:43 AM

Or ATM's ubiquitous sidekick "PIN number".

Posted by: Junxiong at Oct 31, 2005 9:45:35 AM

2 economists are walking down the street. They see something on the ground.

Econ 1: Is that a $100 bill?

Econ 2: It can't be. If it were a $100 bill, somebody would have picked it up.

Posted by: Robert Schwartz at Oct 31, 2005 10:45:43 AM

I disagree. The objection towards repeating a word from an acronym is merely a shibboleth. "ATM machine," "PIN number," and "CAPM model" are quite obviously types of machines, numbers, and models, respectively. That's the only time that this particular usage is done-- the only word ever repeated is the word which describes the general class of which the acronym is a specific type. Certainly it's odd when you spell out the acronym, but I don't see the point of objecting to it when the acronym is used.

Posted by: John Thacker at Oct 31, 2005 11:05:36 AM

I agree with John. My understanding is that
an acronym can, over time, become a word in
its own right. Although it has the same meaning as before,it no longer implicitly spells out the
constituent words of the original acronym.
To take another example I wouldn't see any
objection to referring to RADAR detection
even though the 'D' in RADAR originally
stood for detection. It might be worth putting
this question to the people on languagelog
(Link below)

http://itre.cis.upenn.edu/~myl/languagelog/

Posted by: Chris Tracey at Oct 31, 2005 12:09:10 PM

Also, never say "PIN number"

Posted by: Ivan Kirigin at Oct 31, 2005 12:20:32 PM

When I say an acronym, or write it, I am not thinking out individually all the words. Redundancy should be avoided if you're actually using the same words, perhaps for style more than anything else. But that's not the case with acronyms.

One might as well tell someone not to conjugate their English verbs to show number, since it's "redundant" with the subject already changing to show whether it's singular or plural.

Phrases like "CAPM model," "PIN number," and so on help with clarity, because it is not always obvious from usage what an acronym refers to without context. Yet, completely spelling out an acronym wastes time and space; that's why we have them in the first place. These sorts of usages provide the best marriage of economy of space and time with maximum clarity. They don't sound inappropriate, either, not like real redundancy.

Posted by: John Thacker at Oct 31, 2005 1:13:38 PM

When I was 9 and I heard Vanilla Ice refer to the VIP Posse, and I knew that VIP stood for Vanilla Ice Posse, that bothered me. I guess I just expected more from Vailla.

Posted by: joshg at Oct 31, 2005 2:39:21 PM

Add "HIV virus" to the list.

Posted by: Bryan at Oct 31, 2005 2:49:41 PM

What about RBI vs. RBIs for baseball???

Posted by: Dan at Oct 31, 2005 4:04:30 PM

The basic argument by Chinn in the original econbrowser piece was the
superiority for forecasting of the random walk over virtually all
structural models. Foreign excange investing is essetially dart
throwing. However, that said, it is easy to show within that
context that most people are way underinvested in foreign assets.

Posted by: Barkley Rosser at Oct 31, 2005 4:31:29 PM

Such examples are widely known as "RAS syndrome" (for "redundant acronym syndrome").

How about "SCUBA gear"? I hardly think using a contextual synonym saves it from the criticism.

Posted by: J. Goard at Oct 31, 2005 4:40:35 PM

Not all shibboleths are "mere". A person who says "ATM machine" or "PIN number" is demonstrating that they are not conscious of the meaning of the acronym which they utter at the moment of utterance. But having observed that these terms have become "blank cheques", is it not fair to question whether or no, "It matters not, the funds upon which they are drawn are empty"?

Please excuse me if I do not excuse others for sleep exposition.

Posted by: Nathan Zook at Oct 31, 2005 4:45:21 PM

I see no reason not to call a machine which operates in Asynchronous Transfer Mode an ATM machine.

Posted by: triticale at Oct 31, 2005 8:35:02 PM

A person who says "ATM machine" or "PIN number" is demonstrating that they are not conscious of the meaning of the acronym which they utter at the moment of utterance.

Poppycock. They are demonstrating that an acronym, while it is a symbol for its meaning, is not merely the same thing as writing down the same words a different way. It can have the same meaning as those words, yes, but it is different. Surely no style proscribes all redundant meaning, such, as previously mentioned, banning the conjugation of verbs for number simply because the noun already denotes number. Is someone who conjugates the verb correctly truly ignorant of the meaning of the plural form in English? Hardly.

In addition, sometimes one must ascertain what is meant by a single acronym that can have multiple meanings-- and spelling it out by using the entire phrase on which the acronym is based is clearly a waste of time and effort compared to merely mentioning the important class to which the acronym belongs.

Someone who stands on this point is no different, IMO, from someone who insists upon not splitting infinitives or ending a sentence with a preposition in English. (Some prepositions are prepositions, but some prepositions are parts of verbs in English.)

In addition, I find the whole prescriptivist thing in this case rather technocratic and socialist, insisting on a artificial rule with no firm justification that actually impedes clear speech. The occasional use of acronym + class word lends maximum clarity with the minimum necessary words and it is only right and proper that actual language would have evolved to employ it. There will always be those who seek to bend society and behavior to their artificial standards, though, trying to destroy the careful compromises and standards that cooperative use has formed. They have to sweep away the existing spontaneous order, not understanding or caring how it works, how it facilitates comprehension, in order to replace it with their abstract theories. Indeed, actual language use is Hayekian spontaneous order, and is oft given as an example of it. It has careful and sensible rules which can be discovered by anyone taking the time to observe them.

Posted by: John Thacker at Nov 1, 2005 1:16:05 AM

Do we also instinctively rule out the occasional "grand slam homerun," because "grand slam" (in the proper context) implies homerun? When using a different language, is "Takeshima Island" (or "Dokdo Island") out, since "shima" and "do" both mean island in their original languages?

Posted by: John Thacker at Nov 1, 2005 3:12:09 AM

I find it inconsistent that you insist on rendering my statements in such a wooden fashion while defending anything-goes flexibility on the part of everyone else. There are of course situations where genuine confusion between decodings exists, and the best resolution is likely unsatisfying.

Tell me, what is your opinion of people who confuse adverbs with adjectives or nouns with verbs? If your position is that ignorance is an acceptable basis for linguistic evolution, then we have little to discuss. If you hold that such ignorance is acceptable in an exposition, which is the basis of this digression, they you are truly hopeless.

Posted by: Nathan Zook at Nov 1, 2005 9:42:23 AM

It seems like Brad was refering to the fact that putting this trade on is easy to execute and seems to make money easily and that people still refuse to do this in large amounts. He probably should have written something like "It's an easy thing to do. Why aren't traders doing this?" rather than what he wrote. I think he is genuinely curious as to why they are not doing it, and believes there must be a legitimate reason.

He sees a pile of $100 bills in a room crowded with thieves, cheats and villians, and people are ignoring this pile of bills. Why?

One reason is that the sharpe of this trade would probably be quite low, and the amount of time you have to hold this trade to make it work is far outside the usual parameters of most traders.

In line with reason 1, this is essentially a short options trade. Make decent money 9 out of ten times, lose the farm that 10th time. If you are wrong, you won't be able to get out until liquidated by your prime broker.

Combine this style of risk with the holding time necessary to make it work and many traders see a career ending trade, not one that makes money.

Posted by: Mike Sankowski at Nov 1, 2005 9:45:21 AM

Brad's comment assumes mean reversion. If there random walk is supreme is true, then there is no mean reversion. There remains weak evidence of long run mean reversion, but it is very weak and slow, not much to bet large sums of money on.

Posted by: Barkley Rosser at Nov 1, 2005 10:50:35 AM

Random walk beats fundamentals-based forecasting models only at short and medium horizons.

Exchange rates are very wobbly relative to fundamentals: There are studies out there suggesting that the forward rate does as well as professional forecasts.

Mark and Moh argue that UIP works reasonably well except during periods of central bank intervention. They argue that the risk-premium notion doesn't seem to do a great job explaining things, but then again, our models of risk are non-stellar.

I think Mark has another paper in which he argues that UIP works well but this is difficult to detect in discrete-time data.

If the market is generally populated by chartists, and fundamentalist strategies are only used when the exchange rate is CLEARLY inconsistent with fundamentals, then over short horizons the exchange rate will behave unpredictably -- consistent (again) with chartists dominating trading.

Posted by: randy at Nov 1, 2005 2:02:48 PM

bisexueel journey ^^^ babes dronken in bad ^^^ generalista immagini ^^^ amatoriali dildo nel bagno ^^^ peu commun asiatique strip tease ^^^ dans l antichambre mariee ^^^ froid vingt deux ^^^ filles quarante et un ^^^ amore amatoriali anale fotti ^^^ di bell aspetto infermiera ass to mouth ^^^ kallare tonaring alska ^^^ fager skolflicka suga av den ^^^ amore ragazze doppio penetrazione ^^^ grazioso giovane ubriache ^^^ usynlig fatter ^^^ mannlig kysk ^^^ hetest utmerket geisha ^^^ kort ben ^^^ epifylaktikos efivos katourima ^^^ mora skata sta skalia ^^^ plus frais papa porno ^^^ mama chaude ^^^