by Asset Allocation Committee

Foreign exchange economics has become something of a backwater in economic theory.  There are four predominant valuation methodologies; if one were any good, the others wouldn’t exist!  The four are purchasing power parity, real equilibrium theory, interest rate differentials and productivity equalization (unit labor cost equalization).   The general idea is that under flexible exchange rates, currency values adjust to eliminate differences between nations.  The oldest of the four is purchasing power parity, which assumes exchange rates move to equalize prices across nations.  Real equilibrium theory suggests the exchange rate adjusts to equalize real current account differences.  Interest rate differentials suggest the exchange rate adjusts to equalize interest rates, and productivity equalization normalizes unit labor costs (labor costs adjusted for productivity) across nations.

In practice, the macro data used to calculate fair value for exchange rate models is usually not granular enough to capture differences between nations.  For example, with purchasing power parity, all goods and services in an inflation index are not tradeable and so these non-traded products cannot be adjusted via exchange rates.  And so, all the models tend to be useful only at extremes.  In other words, wide deviations from calculated fair value can offer useful signals, but, like all valuation models in finance, they are not helpful for timing.[1]  At the same time, long-term investors can find value in such models in that they do signal when a relationship is cheap or rich.  This is helpful but only if (a) the investor is truly patient, and (b) something significant hasn’t changed.

Our favorite model is purchasing power parity because long-term inflation histories are usually easy to access and, theoretically, inflation is a very important variable.  As we have been noting for some time, the dollar is expensive based on these models.  For example:

This is a parity model for the EUR, using German inflation against U.S. inflation.  The fair value exchange rate is $1.3008 compared to the current rate of $1.1200.  This deviation from fair value, which in historical ranges is where reversals usually occur, has led to the belief that, at some point, the dollar will weaken.  The model also suggests that once a reversal occurs it is common that the exchange relationship will overshoot.  Thus, a turn in the exchange rate is an important event; in markets, for U.S. investors, foreign stocks and commodities are two areas where one can historically find outperformance.

But, the Trump administration has moved to the use of tariffs to combat persistent trade deficits.  The use of tariffs had fallen out of favor, in part because the U.S. fostered open trade, and because tariffs tend to be less effective under flexible exchange rates.  Why?  Because the nation targeted by tariffs can simply allow its currency to weaken, offsetting the price effect of the tariff.  For example, if China is a target for tariffs, the expected response would be CNY depreciation.

This chart shows how tariffs have become less of a factor.

As the chart shows, after the 1920s, the U.S. has steadily abandoned tariffs…until now.

If the administration continues down this path of using tariffs on a widespread basis, our position on future dollar weakness has to be reconsidered.   Although the tariff policy may not last once Trump leaves office, one cannot necessarily assume that to be the case.  Of course, the U.S. could act against depreciation too, but that might prove difficult to stop.

At some point, we expect the administration (or some future administration) to realize that tariffs are a poor tool for reducing the trade deficit.  Currency weakness is more effective but the most effective method is capital controls.  The reason the U.S. runs persistent trade deficits is because the dollar is the reserve currency.  If the U.S. denied access to the U.S. capital markets, then foreigners would have less reason to engage in policies to promote exports.  Would there be collateral damage from such policies?  Yes, but we may be approaching a point where Americans are willing to accept those side effects in order to reduce inequality and lift domestic wages.

View the PDF

---

[1] Or, said another way, “markets can stay stupid longer than you can stay solvent.”

by Asset Allocation Committee

While the financial industry is rife with performance comparisons to selected benchmarks, the most important investing goal for the majority of clients is a return above inflation that avoids catastrophic losses.  Although beating the S&P 500 is a nice goal, solely focusing on that outcome may lead an investor to accept more risk than appropriate.  This is an age-old issue where one confuses ends with means.  Benchmarking is a means to an end.  A benchmark gives investors some insight into how their investments are doing but should never be considered an end in itself.  Sadly, measurement of performance seems to have eclipsed, and even replaced, the principal goals for clients.  In other words, the benchmark has become the goal.

A good example of the problem with benchmarking is found in academia.  Students have been told that the “4.0” is the clear marker of academic success.  Now, getting all “As” is a good thing.  But, anyone who has been to college knows that the GPA can be gamed.  Students can fill their electives with easy courses.  They can select the easiest professors in their major’s hardest courses.  And, they can cheat.  Or, perhaps equally as perverse, they can “know it for the test.”  In other words, they can memorize the necessary information but fail to really understand it.  Resolving this issue is part of hiring new graduates.  There are ways to ferret out who knows their stuff and who gamed.  Checking transcripts is a good way to look for clues—what were the electives and how did the candidate do in the hard classes?  Another is to ask questions about the most basic components of a discipline but in a way that is rarely presented in class.  An example for economists is, “Assume all drug users are addicts; what is the best way to reduce illicit drug consumption?”[1]  However, how many positions are filled by candidates who are screened by GPA?  In other words, how many good candidates never get an interview because their GPA fell below 3.5 because they took more challenging course work?

Let’s suppose that instead of attempting to help clients accumulate wealth within their acceptable risk tolerance, the goal was to outperform the S&P 500 Index and the criteria was what outperformed the index over the past seven calendar years.  Out of the 34,468 U.S. dollar-based indices in Morningstar’s database, the sole index that met this criteria was the S&P HealthCare Equipment Select Industry Index.  Naturally, exposure exclusively to this single index would be a poor investment strategy for the vast majority of clients, as it would expose them to very specific risks, yet it underscores the notion that simply striving to outperform the return of a particular index is fraught with the potential risk of a permanent impairment of capital.

Another problem that arises in making relative performance the principal objective is the potential for miscreants to juggle benchmarks in order to appear successful.  As an example, the table below illustrates the divergence that is associated with popular small cap growth benchmarks.  While all four of the benchmarks in the table are from highly reputable providers with well-documented methodologies for the U.S. small cap growth stocks included in their indices, the variance among these four in any given year can be profound.  Note that even with indices from the same vendor the differences can be significant as evidenced by the MSCI U.S. Small Cap Growth Index varying from the MSCI U.S.A. Small Growth by 244 basis points in 2018.

A further complication that may be encountered is the utilization of benchmarks that incorporate significant complexity in the myriad sub-asset classes that roll up to major asset classes.  The resulting information will be a hash of statistics that are of little use to either investors or advisor supervision.

These potential pitfalls do not obviate the necessity of monitoring and evaluating relative performance as part of proper due diligence.  The evaluation of a manager’s portfolio or asset allocation strategy against an appropriate benchmark is an essential tool for validation of an investment thesis, which can lead to the achievement of the client’s goals.  However, this can be taken to extremes.  Our industry’s all-consuming fascination with performance measurement has the potential to cause actions that are perpendicular to the goal of inflation-adjusted wealth creation, such as performance chasing.

What, then, is the correct approach to ensure a manager is properly positioned in a client’s portfolio or an asset allocation strategy is appropriate to help attain the client’s goal?  The most straightforward means is to evaluate a manager or asset allocation strategy against a benchmark that is objective, possesses a sound methodology, recognizable, germane to the asset class represented and free of unnecessary complexity.  For this last facet, the notion of Occam’s razor applies.  This approach will naturally yield significant tracking error; however, tracking error should not only be expected, but embraced, for an active manager.  While on a quarter-to-quarter basis investors may observe divergent returns relative to the benchmark, during discrete, representative periods and especially through a full market cycle, an uncomplicated and recognizable benchmark will represent a solid barometer against which to measure the risk-adjusted return of a manager or investment strategy.  This will serve to evaluate whether the manager or strategy is contributing to the overarching client goal.  But, ultimately, the key point to remember is that a benchmark is a means to an end, not an end in itself.

View the PDF

---

[1] If all users are addicts, then the demand curve is highly inelastic.  Reducing supply merely drives up the price, but reducing demand (drug rehab, substitution) could reduce demand and have the biggest effect on reducing consumption.

by Asset Allocation Committee

Covered interest rate parity is a basic concept that, at its heart, says all interest rates are equal after hedging exchange rate risk.  It is one of the theories in finance that is beyond dispute—it works as long as capital markets are open (free of capital controls) and short-term money markets are liquid.  In general, if foreign interest rates are higher than domestic rates, the forward exchange rate will trade at a premium high enough to absorb the interest rate difference.  Here is a simple example:

In our example, one-year interest rates in the Eurozone are 5.00% compared to U.S. rates at 2.50%.  If the forward rates were equal, a U.S. investor could exchange dollars for euros, invest at the 2.50% spread and hedge the currency risk, earning a risk-free extra 2.50% compared to dollar-based interest rates.  However, in the process of doing so, the forward rate on euros would be bid up to 1.0250, which eliminates any arbitrage opportunities.  A U.S investor should be indifferent to either investing in the U.S. at 2.5% or in the Eurozone at 5%, at 1.025 $/€ one-year forward exchange rate.  In a year, if nothing changes, that euro purchased at 1.025 will be at 1.00, eliminating the entire interest rate spread.

However, this doesn’t necessarily mean that hedging opportunities don’t exist.  An investor could decide to buy into longer duration fixed income abroad and roll the hedge periodically.  Using the above example, an investor could buy a foreign bond and hedge each year, selling the bond if the hedging costs become excessive.

The current spread between U.S. and German 10-year sovereigns clearly favors the U.S.  German yields are around zero and U.S. yields remain significantly higher even though U.S. yields have declined.  Thus, it would seem that German investors would have an incentive to buy longer dated Treasuries.

However, for a German investor to make this investment, he would have to either accept the currency risk or attempt to hedge the risk.  As shown above, the short-term rate spread determines the currency forward discount/premium relationship.

This chart shows the 10-year sovereign spread and the three-month LIBOR spread.  When the bond spread exceeds the LIBOR spread, the bond spread exceeds the cost of hedging.  Even with a very wide sovereign spread, a foreign investor cannot, under current conditions, profitably hedge the currency risk.

This has led foreign bond investors to (a) accept U.S. credit risk by purchasing corporate bonds and hedging the exchange rate risk, or (b) accept the currency risk by leaving the transaction unhedged.  However, that situation is less than ideal and may discourage foreign investors from investing in U.S. fixed income markets.

Although the hedge relationship with regard to the euro isn’t overly strong, there is a modest tendency for the currency to strengthen against the dollar when the hedge spread is negative.  Clearly, there are other factors that affect exchange rates beyond mere interest rate differences.  Although the euro has been weakening recently, it hasn’t declined significantly despite unusually wide sovereign spreads; the likely culprit is the cost of hedging.  If the U.S. yield curve were to steepen, perhaps with policy easing, the window for hedging might reopen and, paradoxically, could lead to a stronger dollar.

View the PDF

by Asset Allocation Committee

One of the age-old problems of analysis is the problem of correlation versus causality.  Correlations simply show the degree of relation; the range runs from -1 (perfectly negative) to +1 (perfectly positive).  In any introductory statistics class, the instructor will discuss the difference between relation and causality.  Here is an example:

Although it might be possible to concoct a reason why chicken consumption would be related to crude oil imports, in reality, the relationship is spurious.

The advent of cheap computing power has given rise to data mining that allows researchers to scan massive amounts of data and find relations.  In fact, some in the tech industry tend to believe that theory is unnecessary—simply find the relationships and assume they will continue.  However, without theory, if the relationship breaks down then there is no way of knowing why it fell apart.

This chart has started making the rounds in reports:

This chart shows the 10-year Chinese sovereign yield with the yearly change in the S&P 500 and the MSCI World Index.  Casual observation would suggest they are closely related; in fact, the correlation between the Chinese bond yield and the MSCI is 66.7% and 72.1%with the S&P 500.  What’s implied is that developed world stock markets are dependent on Chinese yields.  One would expect Chinese yields to rise with better growth in China, and this would suggest that foreign economies, and thus their markets, depend on Chinese growth.[1]

Perhaps.  But it is also possible that Chinese interest rates are sensitive to world growth; therefore, when the developed world economy does better so does the Chinese economy, and the improvement leads to higher yields in China.  In other words, the direction of causality is difficult to ascertain.

Often, relationships such as this occur because of some other, unnamed variable.  Consequently, in the above case, all three are related to a fourth variable.  Since 2014, for example, Chinese bond yields have been closely correlated to the dollar.

On this chart, the JPM Dollar Index is shown on an inverted scale.  When the dollar strengthens, Chinese yields tend to fall; during periods of dollar weakness, the CNY yields rise.  Since China tends to manage the CNY/USD rate, it’s plausible that when the dollar weakens China enjoys the benefits of that weakness by allowing the CNY to weaken as well.  A weaker currency boosts China’s exports and leads to higher interest rates.  And, a weaker dollar tends to support both the above stock indices.  Therefore, the dollar’s impact probably explains the relationship between Chinese bond yields and developed market equities.

Finally, one last word of caution.  In regression analysis, the dependent variable = intercept + independent variable + error term.  The error term, in theory, contains a myriad of variables that we assume, under the majority of circumstances, balance each other out and thus allow for the model to be stable.  In reality, the variables not specifically delineated in the error term can emerge and affect the dependent variable; sometimes this occurs only for a short time but can make forecasting a challenge.  Given the short history of the above, it is possible that conditions could change and the relationship completely breaks down.  Thus, the moral of the story is that conditions do change and investors need to be cognizant of that fact and try to adjust accordingly.  And, one needs to be careful of seemingly clear charts.

View the PDF

---

[1] Granger causality testing tends to support the idea that Chinese yields drive equity market performance.  Although statistically sound, Granger causality testing is atheoretical and thus not necessarily causal.