by Asset Allocation Committee

In the past few years, we have generally avoided allocations to non-U.S. markets for our asset allocation portfolios due to two primary concerns.  First, the dollar was rising as a result of an improving U.S. economy and policy divergences between the U.S. and the rest of the world.  The Federal Reserve was raising rates and tapering its balance sheet while the majority of the other nations were still adding monetary stimulus.  Second, we have had secular concerns about the stability and attractiveness of foreign investing in a world where the U.S. is seemingly reducing its hegemonic role.

This quarter we have added foreign allocations into our portfolios.  The primary reason is that we believe the dollar’s bull market is probably coming to a close.  On a relative valuation basis, the dollar has become rather expensive.

This chart shows four purchasing power parity models for the euro, yen, British pound and Canadian dollar.  In all four cases, the dollar is trading “rich” by more than one standard error, and in two cases, nearly two standard errors from parity.  Purchasing power parity uses relative inflation to value currencies.  As the models show, currencies are rarely at parity.   Although purchasing power parity is the oldest way to value currencies, it isn’t the most accurate.  However, it tends to be useful at extremes, and exchange rates tend to move around the parity measure.  In other words, parity is something of an anchor around which the actual exchange rate vacillates.

History also suggests that exchange rates can diverge from parity for long periods of time, which usually means that a catalyst is required to move currencies from extremes of under-and over-valuation.  For example, in the above euro and pound models, the deep undervaluation in the mid-1980s ended with the Plaza Accord.  Yen overvaluation in the early 1990s was ended by the Halifax Accord.  Changes in policy or governments can trigger an end to valuation extremes as well.

This leads us to two additional reasons for this repositioning.  We believe there are two catalysts which will pressure the dollar in the coming quarters.  First, the Federal Reserve appears to be hedging on future tightening.  Although the dots charts still indicate rising rates, a number of FOMC members have raised worries about raising rates while inflation remains depressed.  Chair Yellen appears to be offering a trade to the hawks on the committee; instead of raising rates, balance sheet reductions can act as policy tightening.  It is possible reducing the balance sheet will be dollar bullish.  However, that isn’t supported by the data.

Although economic theory would suggest that boosting the size of the balance sheet should depreciate a currency, all else held equal, the pattern on the above chart developed likely because QE raised hopes of stronger growth.  Our expectation is that reducing the balance sheet probably won’t matter, but if the pattern is consistent, balance sheet reduction may actually be dollar-bearish.  So, we expect the anticipated monetary policy trade of fewer rate hikes for balance sheet shrinkage to be bearish for the dollar.

The second catalyst is policy disappointment from the Trump administration.  Earlier this year, we were worried about the dollar’s overvaluation. And, proposed tax cuts and fiscal expansion from the Trump administration would have been dollar bullish, even at lofty valuations.  However, the likelihood of major policy actions have declined and each month that passes without a policy adjustment means that political capital is being lost and it becomes less likely anything will pass.  As disappointment grows, we would expect the dollar to retreat.

Although we retain our secular concerns about the stability and attractiveness of foreign investing in a world where the U.S. is seemingly reducing its hegemonic role, these concerns are overshadowed by the changing dynamics of monetary stimulus. Essentially, the near-term effects of policy and the exchange rate outweigh our secular concerns.

Finally, the issue of recent performance needs to be addressed.  Both emerging and foreign developed equity markets have performed very well this year and there is a concern that we may be shifting too late into these asset classes.  Although possible, if our expectations of a weaker dollar are accurate, history would suggest a longer period of foreign outperformance.

The red line on both charts shows the JP Morgan real dollar index. The blue line on the left chart is the ratio of dollar-denominated EAFE and the S&P; on the right chart, it’s the dollar-denominated ratio of emerging markets and the S&P.  Both have been rebased to their respective start periods.  On both charts, a lower ratio indicates U.S. equity outperformance relative to foreign markets; a rising ratio shows the opposite.  In both cases, periods of dollar strength coincide with domestic outperformance.  Thus, even with good foreign performance this year, a weaker dollar may result in a longer period of foreign outperformance.

View the PDF

by Asset Allocation Committee

One of the mysteries of this expansion has been the slow pace of wage growth.  Despite the plethora of evidence that labor markets are tight, including hires-to-openings ratio below one, low unemployment, low initial claims and a low unemployment rate, wage growth has remained stunted.   The chart below is one we have used often in the past; it suggests with unemployment at current levels, previous episodes would have brought wage growth closer to 4% compared to the current growth of 2.3%.

One of our thoughts was that, perhaps, the national unemployment rate was masking pockets of high regional unemployment.  In other words, if there was a wide dispersion of labor market activity, the weak regions of the country could be holding down wage growth.

To test this, we looked at the level of state unemployment relative to the Congressional Budget Office’s calculation of the natural rate of unemployment.[1]  We calculated the number of states with unemployment below the natural rate.  What we found is consistent with the above graph.  In other words, it doesn’t appear that regional issues were holding down wages.

Since 1987, average wage growth is 2.98%; in the periods where the percentage of state unemployment rates below the natural rate exceeds 65%, wage growth was 3.89%.  The current combination is unusually low.

However, as part of this research, we did find another interesting factor.  There has been speculation that low inflation rates may be affecting wage growth.  To test that thesis, we subtracted annual wage growth for non-supervisory workers against the three-year average of the yearly change in CPI.  Our reason for using the average is that businesses and workers tend to react to the recent trend in inflation and not necessarily the most current number.  We found results that were more consistent with this theory.

Over the same time frame, real wage growth was 0.26%, but when the percentage of states with unemployment below the natural rate exceeds 65%, real wage growth is 1.04%.  As the chart shows, real wage growth is consistent with tight labor markets as defined by the percentage of state unemployment rates below the national natural rate.

So, the puzzle of why wage growth is slow may simply be due to low inflation.  The three-year average of inflation may act as an anchor in the wage bargaining process and wage growth will probably remain stalled without rising inflation.   A cursory glance at the annual rate of inflation relative to the three-year average suggests that the average rate will likely remain stable for the foreseeable future.  Thus, wage growth will probably remain stable, increasing the risk that tightening monetary policy will have an adverse impact on the economy and markets.  Of course, the wage issue is a matter of debate within the FOMC which could mean that the path of tightening is slow and measured, more in line with the financial market’s expectations surrounding future policy.

View the PDF

---

[1] This rate is sometimes called the Non-Accelerating Inflation Rate of Unemployment, or NAIRU.  The idea is to determine what level of unemployment is the lowest rate attainable before higher inflation is triggered.

by Asset Allocation Committee

One of the relationships we persistently monitor is the expectations of 10-year Treasury yields compared to the actual level of yields.  To do this, at the beginning of each year, we plot the rate forecasts from the Philadelphia FRB’s Professional Forecaster Survey.  The record of the forecasters has not been stellar.

The blue line on the chart shows the level of yield; the other lines show the consensus forecasts.  The open boxes on the red lines are “misses” and the filled dots are “hits.”  The forecasters are wrong about 59% of the time.

What is interesting to us is not the error rates but the fact that the direction of the error is consistent; the forecasters expect higher rates.  We suspect a major part of this was due to forecasters overestimating inflation.  It has been our position that globalization and deregulation are responsible for low inflation, not monetary policy.  We had little faith in central bank ability to cause inflation and so we have tended to be more dovish on long-term rates in our asset allocation portfolios.

Still, we have noticed that the downtrend in rates has mostly ended in 2012, while rates have been rangebound for the past five years.  Clearly, the forecasters are looking for an upside “breakout” in yields that has failed to materialize…so far.

Our 10-year T-note yield model puts fair value at 2.10%; thus, current yields are a bit high.  The model uses fed funds, long-term inflation trends, the yen/dollar exchange rate, oil prices and the yield on German bonds.  What is troubling for us is that if we just use the first two variables of the model, the fair value yield jumps to 3.03%, which is in the neighborhood of the current Philadelphia FRB forecast.  Simply put, international factors appear to be holding down fair value yields.  A weaker yen, higher oil prices or rising German sovereign yields will likely have a negative effect on the fair value yield.  Of the three, German yields are probably the most critical.  If the Eurozone avoids financial and political problems, it appears the ECB is prepared to begin slowly withdrawing stimulus.  If that’s the case, the forecasters may have a chance of being right this year.

View the PDF

by Asset Allocation Committee

For equity investors, there is always a concern about the major market declines; being able to reduce market exposure prior to crashes like the ones in 2000 or 2008 is always desirable.  Although large declines have occurred outside of economic recessions, they have become increasingly rare.  The last major market pullback absent a recession was the 1987 crash.  Thus, an indicator that can use high frequency economic data (data that is available on a weekly basis) and relate it to equities should have some value.

This chart shows one of our recent efforts.  The upper line is the monthly average for the S&P 500; the lower line is an indicator built of three economic numbers—initial claims, the CRB commodity index and the Conference Board’s Consumer Confidence data.  We have standardized[1] the economic data and created an indicator, shown on the bottom of the graph.  In general, a positive reading is generally bullish for equities.  We have placed vertical lines on the chart to indicate when the indicator turned negative with persistence.  These are usually periods of falling equities.

Although useful, it is clear that the indicator is somewhat “late” in that the equity decline is well underway by the time it becomes negative.  That would suggest that a momentum number based off the economic indicator might be helpful.

To achieve this, we calculated the 18-month change in the above indicator and set a “crash line” at -1.0.  The adjustment improves the indicator’s value, indicating an investor should reduce equity exposure sooner than the unchanged index would suggest.  At the same time, using a -1.0 reading eliminated the false exit signals.

There is an element of “data mining” here and we would not recommend using this indicator as a “precision instrument.”  However, it does show that these three standardized data points  are fairly good coincident indicators of the economy and, using an 18-month change and a filter, do offer reasonably good warnings when one should reduce exposure and when a market pullback is a mere correction.

We use the period since 1997 because it makes the data easier to observe.  But, we did look at data from 1980 and the performance is similar.  As we would expect, the indicator did not help at all during the 1987 crash, confirming that its usefulness is as a gauge of the interaction between equities and the economy, not as a pure market indicator.

What is it telling us now?  The economy is doing well enough that market declines will probably not be more than normal pullbacks.  Of course, this sort of indicator won’t necessarily be helpful if a geopolitical event triggers a major market decline.  But, the most common cause of bear markets in equities are recessions.  For now, that doesn’t appear to be on the horizon.

View the PDF

---

[1] Standardizing entails subtracting the raw data by its average and dividing by its standard deviation.  The resulting number is then adjusted by how far it is from average and how far it is from its normal deviation.  Standardizing allows one to combine unlike indicators and essentially balance their relative effect.  In this case, the formula is Indicator = (standardized CRB + standardized Consumer Confidence) – standardized Initial Claims.  We subtract the claims data because lower claims indicates a stronger economy.  Subtracting that number thus allows the indicator to rise when economic data are improving and vice versa.