by Asset Allocation Committee

Why is inflation so low?  The persistence of low inflation, despite the long expansion and the decline in unemployment, continues to befuddle policymakers.  Standard economic theory suggests there is an inverse relationship between inflation and unemployment.  When the unemployment rate is low, firms should be experiencing reduced excess capacity.  As capacity is constrained, supply bottlenecks would be expected to develop which would eventually result in inflation.

Although the unemployment/inflation theory makes sense, it doesn’t work all that well in real life.

This chart shows the level of unemployment, with an 18-month lag, compared to the yearly change in CPI.  In the 1960s, it did appear there was a tradeoff; when unemployment fell, inflation rose 18 months later.  However, even during this period, the data relationship was merely directional as sub-4% unemployment led to 6% CPI.  In the mid-1970s, 4.5% unemployment led to CPI in excess of 12%.  Since the early 1980s, CPI has rarely moved above 4%, and in the current environment sub-4% unemployment has not yet triggered a notable inflation problem.

It turns out inflation is rather complicated.  Expectations play a major role; if households and businesses expect rising inflation, they take steps to protect themselves that exacerbate the inflation impulse.  Both sectors will build inventory levels, effectively changing their balance sheet allocation from financial instruments to real goods.  This action can lift the demand for goods and can trigger inflation.

There also appears to be at least two long-term factors that affect inflation.  Inequality seems to have an impact.

History shows that lower degrees of inequality are correlated with higher levels of inflation.  We think there are two factors that cause this outcome.  First, policies designed to expand supply, deregulation and globalization, tend to improve the efficiency of the economy at the expense of higher inequality.  Outsourcing and automation make production more efficient but also reduce the demand for domestic labor.  Simply put, one person’s efficiency is another person’s pink slip.  Second, CPI is designed to measure prices based on an average household’s consumption patterns.  With rising inequality, lower decile households may have less income to spend on basic items, making it difficult for firms to pass along price increases on such goods.  On the other hand, under conditions of inequality, prices on luxury items are likely to be priced higher simply because there is more spending power available.[1]

However, it is likely that the most potent reason inflation has stayed low is because of excess capacity in the economy.  Measuring excess capacity is profoundly difficult because it is something of a moving target.  A simple way is to regress long-term GDP against trend; the assumption is that the trend is a reasonable proxy for capacity.

The chart on the left is real GDP, log-transformed with a time trend regressed through it.  The lower line is the deviation from trend.  We are currently in a period where GDP is well below trend.  The only other period that exhibited such a negative deviation from trend was the Great Depression.  The chart on the right shows the GDP deviation line compared to CPI.  Note that inflation tends to be low during periods of below-trend GDP.  For the overall time frame (1921-2018), CPI averages 2.8%.  When GDP is above trend, inflation averages 3.8%; when GDP is below trend, CPI averages 0.4%.

This trend analysis could mean that it may take several years before inflation pressures become notable if policymakers become aggressive with stimulation (i.e., keeping monetary policy accommodative while lifting fiscal policy).  It also means the Federal Reserve can probably avoid raising rates for a considerable period.

View the PDF

[1] https://www.pewtrusts.org/en/research-and-analysis/issue-briefs/2016/03/household-expenditures-and-income

by Asset Allocation Committee

The employment data is closely watched by financial markets; although the data isn’t necessarily a leading indicator for the economy, it is probably the most important from a political and social perspective.  Weak employment data is a worry for political incumbents and concerning to policymakers.  However, beyond the headline data, there are usually interesting trends worth noting.  In this week’s report, we will examine two trends that have longer term implications.

Career paths were part of corporate culture three decades ago.  Large companies often had junior executive programs, where promising young talent was brought to the firm and would follow a rotation of positions in numerous departments before finding a permanent home.  In other situations, college graduates would join a company and follow a path of positions of increasing responsibility.  However, over the years, outsourcing jobs overseas and increasing industry concentration[1] have probably reduced the number of entry level professional positions in the U.S.  This chart shows the percentage of production and non-supervisory workers compared to total non-farm employment.

In the 1970s, this percentage declined to a low of 66%.  However, since the early 1980s, the percentage has steadily increased in each business cycle.  This data suggests that an increasing number of jobs are non-management positions.  We suspect that college graduates are being forced to accept non-management positions as fewer of them are available for an increasing number of graduates.  Such disappointment has the potential to cause social unrest.  At the same time, reversing industry concentration would tend to boost the number of management jobs in the economy (every firm needs HR, finance, etc.).  Thus, support for anti-trust actions could become more popular.

Second, initial claims, on a weekly basis, fell to 40-year lows recently.  However, the weekly data is “noisy” and can be affected by floating holidays and weather.  Another way of looking at claims is to scale to the civilian non-institutional population.  This data is at historic lows.

This low level of claims is likely due, in part, to firms holding on to workers because of tight labor conditions.  A rising number of retirees will lift the non-working civilian non-institutional population but fewer workers will tend to depress claims.  In any case, this level of claims compared to the population is remarkably low and would argue that wages should rise.

Overall, these two charts offer insights into longer term issues in the labor market.  They won’t have an immediate effect on financial markets, but both signal potential for further disruption.

View the PDF

[1] https://finance.eller.arizona.edu/sites/finance/files/grullon_11.4.16.pdf

by Asset Allocation Committee

Mortgage-backed securities have rather odd characteristics compared to Treasuries.  At their most basic level, mortgages are bonds—prices are inversely related to yields.  The pricing on mortgages assumes a certain level of refinancing activity.  However, when yields rise, expected mortgage duration tends to extend because mortgage holders are less likely to refinance.  When yields fall, expected duration shortens as mortgage holders replace their existing mortgages with new ones at lower rates.  Usually, mortgage bonds tend to act like options; they act “normal” within a certain range of yields, but duration adjustments occur if yields change above a given range.

Some bond fund managers are required to maintain a given duration level.  If the manager is holding mortgages and refinancing activity rises then the fund’s duration could shorten.  This would require the manager to take steps to re-extend duration.  A fast way to accomplish this goal would be to buy long-duration Treasury futures.  These instruments are liquid and generally have a set duration.

In general, refinancing tends to take place when interest rates fall below the level of the current mortgage, taking closing costs into account.

The chart on the left shows the 10-year T-note yield; we have regressed a time trend through the data.  Clearly, yields have been on a downward path since 1990.  The chart on the right shows the 10-year T-note yield less trend on the lower line and the mortgage refinancing index on the upper line.  Evidently, refinancing activity tends to rise when rates are below trend.  Since 2016, the steady rise in rates relative to trend has depressed refinancing.  However, we have seen a lift in refinancing recently.  On the one hand, the dip in rates doesn’t look like enough to overcome closing costs.  On the other hand, paying a higher rate on a new mortgage may be the only way a homeowner can capture price appreciation.

This chart shows the ratio of the new rate to the old rate.  When the old rate is higher than the new rate, the ratio is less than 1.0.  Note that this ratio is now at a new record high.  Essentially, buyers are refinancing at a less advantageous rate to extract home equity.

We suspect the most important factors of the recent bond rally have been changes in monetary policy expectations and reduced inflation fears.  But, the rise in refinancing has likely played a role as well.  We doubt that the attractiveness of locking in a higher mortgage rate will continue indefinitely, therefore this bullish factor should dissipate in the coming weeks.

View the PDF

by Asset Allocation Committee

Long-dated Treasury bonds have enjoyed a strong rally in recent weeks.  Fears about future U.S. economic growth, falling global economic growth and a reversal in monetary policy expectations have all conspired to lower yields.  The question now is whether or not yields have fallen more than is justified by the fundamental factors.

This is our model of the 10-year T-note yield.  Independent variables in the model include fed funds, an inflation proxy,[1] the yen/dollar exchange rate, German 10-year yields, WTI oil prices and the fiscal deficit as a percentage of GDP.  The standard error line is 70 bps; thus, at current fair value, the 10-year T-note yield would be “rich” around 2.15%.

In general, the first two variables have the most explanatory power.  The current yield is well below the fair value of 2.86%.[2]  One way to account for the decline in yields would be if the market was projecting easier monetary policy.  Leaving the rest of the independent variables unchanged, to reach a 2.40% 10-year T-note yield would require fed funds of about 1.50%, or a 100 bps reduction from current levels.  Since such a cut is unlikely in the near term, it is arguable that the 10-year yield has overshot to the downside.

However, another possibility is that inflation expectations are falling.  Inflation expectations have a powerful effect on long-dated debt yields.  Our current inflation proxy puts inflation at 2.10%.  To achieve current yields, inflation expectations would need to fall to 1.41%.  Although inflation expectations are not directly observable, we assume such expectations are slow to change.[3]  Thus, it would be unlikely that a drop in inflation expectations alone would account for the recent decline.  However, inflation expectations falling to 1.8% and fed funds at 2.00% would bring yields close to 1.40%.

It is quite possible that falling inflation expectations and forecasts of easier monetary policy can justify the current 10-year yields.  The risk is that the FOMC keeps policy steady and inflation eventually surprises to the upside.  Thus, we would not see the current environment as conducive for extending duration. 

View the PDF

[1] We use a 15-year moving average of the yearly change in inflation for inflation expectations.

[2] We use monthly averages in the above model; thus, the March yield is 2.63%.  The April yield will be much less.

[3] Milton Friedman assumed such expectations were built over a long period of time, perhaps two decades or more.