Regarding my last post on the Merton-Scholes-Grumpy Fallacy: No free lunch on covered interest parity, I have received a thoughtful email from Tim Streufert. What follows below is our exchange. It illustrates the importance of terminology and some of the conditions for fair trade or arbitrage regarding the supposed paradox identified by the great John Cochrane on his post on the covered interest rate parity. Tim makes a good point that my comparison between LTCM and the present CIP paradox is weak. Finally, if you are not following Cochrane’s blog, you should.

Tim: I read your response to John Cochrane’s blog post on deviations from covered interest parity and I have a couple questions.  You reference LTCM and argue that their strategies were similar to the CIP trade in that both pick up a liquidity risk premium.  I’m not so sure this is the case.

Even LTCM’s safest trades (like shorting the on-the-run 30yr treasury and buying the 29.5 year off-the-run treasury) had risk of spreads widening, long maturities where “fair value” would be forced to converge, and no true arbitrage parity condition that had to hold true.  The trades you mentioned (US/EUR rate relative value) were riskier still in that their “fair values” had no clear link (different default, inflation, currency risks, for starters).

Covered interest parity is very different from these trades.  It has an arbitrage condition that is guaranteed to hold true by the maturity of the securities in question which is very short in practice (1w, 1m, 1y all provide opportunity).  It’s true that the position could generate margin calls for levered players (hedge funds) if spreads widened.  For unlevered investors just trying to enhance cash yields or banks without these leverage constraints the trade is riskless.  As long as your counterparties can make good, the trade is the same as investing in CD, but with a higher yield than a domestic currency CD. Let me know where I am going wrong here.

Rodrigo: I really like your points. Margin calls by itself will make them non-arbitrage (so many companies went under from ignoring this simple fact, ). Will edit the original post to reflect you excellent analysis. One question: from what I recall LTCM “sold” its strategy as arbitrage, and hence “risk-free”, right?

I concede I may have been wrong in equating the “weak” arbitrage strategy from LTCM with the “strong” arbitrage “opportunity” implied in the graphs in Cochrane’s blog. Regardless, there were are aren’t true arbitrage opportunities but the difference between the two cases is enough that I should rewrite my points. (note: instead of changing the original post, I am displaying our whole exchange).

Tim: The term “arbitrage” has changed meaning over time as the marketing departments on Wall Street have commandeered it.  While it used to mean a true risk free profit (covered interest parity, put-call parity, etc.), it now is used to describe most relative value trades.  It is important to note that the key for true arbitrage is not that the trade can’t go against you in the interim.  Even with true arbitrage like CIP, spreads can widen (markets get more irrational) from where you put on the trade.  The difference between true arbitrage and the relative value trading now called “arbitrage” is that true arbitrage has a binding link or mathematical condition that guarantees the intrinsic value of the trade while relative value trades bet on what fair value “ought” to be.  Relative value often works but can be foiled by any number of shocks (like liquidity) while true arbitrage like CIP will always converge by maturity.

LTCM did call their trading “arbitrage”, but I don’t think they misled investors that their strategies were riskless.  They knew there were risks and disclosed their best guess of the risks.  However, their models significantly underestimated how much the trades would move against them when there was a significant shock to the markets which ultimately led to their demise.

Rodrigo: It is really interesting how terms evolve over time. I have done my fair share of consulting in Latin America. There, the term arbitrage still has its original meaning. Most likely, the reason is that companies and financial institutions have experienced first hand what happens when risks are misunderstood (I wrote about this for the case of derivatives and non financial companies: https://rzeidandotcom.files.wordpress.com/2014/12/2013-afe.pdf and https://rzeidandotcom.files.wordpress.com/2014/12/2015-zeidan-and-mullner-jmfm.pdf)

As for CD, even if you are unleveraged, you still have margin calls that can destroy cash flows. Imagine the situation you described, as a complete riskless proposition in which trades are on CD with similar maturities (1w, 1m, 1y all provide this opportunity, as you pointed out). During the period of the trade there is the risk of margin calls, which means that there has to be a cash reserve put aside for this. The opportunity costs of managing uncertain cash balances may destroy any extra yield you get from going short on one end of the trade.
Let me give you a personal example. I now reside in Shanghai but still have a fair amount of expenses in Brazil. So I hedged part of my future expenses with futures of USD against Reais. I did that when USD 3.4 to 1BRL. I was getting the Brazilian interest rate in return (at the time 14% a year, a good deal). The Real exploded with the impeachment process and volatility went through the roof. Every day I either got a margin call or extra money in my account. It became a hassle to leave a lot of money aside to cover whatever margin call I could eventually face. I was unleveraged. It did not matter. I ended the trade with a modest loss but without the hassle of managing volatile cash balances.
The impact on cash balances of increased volatility make this a fair trade but not an arbitrage one. In other words, I think we both agree with everything, the difference that remains is due to terminology.