Friday, September 6, 2019

September Macro Update: Rising Possibility of a Recession in 2020

Summary: The balance of the macro data remains positive. A recession starting in 2019 is unlikely, but, for the first time, a recession in 2020 is a rising possibility.

The bond market sees weakening growth. The yield curve has 'inverted' (10 year yields less than 2-year yields) ahead of every recession in the past 40 years (dots). The lag between inversion and the start of the next recession has been long: at least 7 months and in several instances as long as 2-3 years. Notably, the yield curve finally inverted in August; on this basis, the current expansion will likely last through 2019 but 2020 is now at risk (from JPM). Enlarge any image by clicking on it.

High yield spreads are normal and default rates remain below average. This part of the bond market is not signaling trouble (from JPM).

Similarly, real retail sales grew 1.6% and made a new all-time high (ATH) in July. The trend remains higher, in comparison to the period prior to the past two recessions.

Unemployment claims reached a 50 year low in April (just 4 months ago). Historically, claims have started to rise at least 7 months ahead of the next recession.

In July, housing starts rose 1% yoy, building permits rose 2% and new home sales rose 4%. In the past 50 years, a median of 28 months has lapsed between new home sales' expansion high (arrows) and the start of the next recession; so far, the cyclical high was in June, just 2 months ago, which suggests that the next recession is still a ways ahead.

New home sales have rebounded 4% in the first 7 months of the year compared to the same period last year. The 2017 peak in home sales came after mortgage rates troughed in late 2017, and the rebound so far this year corresponds with mortgage rates falling from 5% in November to under 4% this month.

The Conference Board's Leading Economic Indicator (LEI) Index reached a new uptrend high in July. This index includes the indicators above plus equity prices, ISM new orders, manufacturing hours and consumer confidence. This index can fluctuate during an expansion but the final peak has been at least 7 months before the next recession in the past 50 years, which also suggest that the next recessions is still a ways ahead (from Doug Short).

Why does any of this matter for the stock market?

Equity prices typically fall ahead of the next recession, but the macro indictors highlighted above weaken even earlier and help distinguish a 10% correction from an oncoming prolonged bear market. On balance, these indicators suggest a recession starting in 2019 is unlikely, but a recession in 2020 is a rising possibility (a recent post on this is here).

Here are the main macro data headlines from the past month:
Employment: Monthly employment gains have averaged 173,000 in the past 12 months, with an annual growth of 1.4% yoy, the weakest in 7 years.  Employment has been been driven by full-time jobs, which rose to a new all-time high in August. 
Compensation: Compensation growth remains firm. Hourly wage growth was 3.2% yoy in August, while the 2Q19 employment cost index grew 2.9% yoy.  
Demand: Real demand growth has been 2-3%. Real retail sales grew 1.6% yoy in July, making a new all-time high, and July real personal consumption grew 2.7% to a new all-time high. But 3Q19 real GDP is expected to fall to under 2%.
Housing: Recent housing data has improved, but growth is weak. In July, housing starts rose 1% yoy, building permits rose 2% and new home sales rose 4%. Multi-family units have been a drag on overall development for several years. 
Manufacturing: Macro weakness has been most apparent in manufacturing. Core durable goods fell 0.5% yoy while the manufacturing component of industrial production fell 0.5% yoy in July. 
Inflation: The core inflation rate remains near the Fed's 2% target. 

    Our key message over the past 6 years has been that (a) growth is positive but slow, in the range of ~2-3% (real), and; (b) current growth is lower than in prior periods of economic expansion and a return to 1980s or 1990s style growth does not appear likely.

    This is germane to equity markets in that macro growth drives corporate revenue, profit expansion and valuation levels. The simple fact is that when the economy is expanding, the historical risk of a greater than 10% annual (not intra-year) decline in the stock market is just 4% (from Goldman Sachs).

    The highly misleading saying that "the stock market is not the economy" is true on a day to day or even month to month basis, but over time these two move together. When they diverge, it is normally a function of emotion, whether measured in valuation premiums/discounts or sentiment extremes.

    A valuable post on using macro data to improve trend following investment strategies can be found here.

    While the macro data is fine, a wild card is the escalation in the trade war, which reduces global demand, raises prices and eventually leads to lower investment (from St Louis Fed).

    * * *

    Let's review the most recent data, focusing on four macro categories: labor market, end-demand, housing, and inflation.

    Employment and Wages

    The August non-farm payroll was 130,000 new employees minus 20,000 in net revisions for the prior two months.

    Employment growth is decelerating this year. The average monthly gain in employment was 240,000 in 2015, 211,000 in 2016, 189,000 in 2017 and 204,000 in 2018. In the past 12 months, the monthly average has fallen to 173,000, the lowest in 7 years.

    Monthly NFP prints are volatile. Since the 1990s, NFP prints near 300,000 (like in January) have been followed by ones near or under 100,000 (like in May). That has been a pattern during every bull market; NFP was negative at times during 1993, 1995, 1996 and 1997. This is normal, not unusual or unexpected.

    Why is there so much volatility? Leave aside the data collection, seasonal adjustment and weather issues; appreciate that a "beat" or a "miss" of 120,000 workers in a monthly NFP report is within the 90% confidence interval (explained here).

    For this reason, it's better to look at the trend; in August, trend employment growth was 1.4% yoy. Until spring 2016, annual growth had been over 2%, the highest since the 1990s.  Ahead of a recession, employment growth normally falls, a watch out (arrows).

    Employment has been been driven by full-time jobs, which rose to a new all-time high in August (blue line), not part-time jobs (red line).

    The labor force participation rate (the percentage of the population over 16 that is either working or looking for work) has stabilized over the past 5 years. The participation rate had been falling since 2001 as baby boomers retire, exactly as participation started to rise in the mid-1960s as this demographic group entered the workforce. Another driver is women, whose participation rate increased from about 30% in the 1950s to a peak of 60% in 1999, and younger adults staying in school (and thus out of the work force) longer.

    A better measure is the prime working age (25 to 54 year olds) labor force participation rate; it stands at 82%, down only slightly from its peak in 2000 at 84%, and much higher than anytime prior to the 1980s.

    Average hourly earnings growth was 3.2% yoy in August; February's rate of 3.4% was the highest in 10 years. This is a positive trend, showing demand for more workers. Sustained acceleration in wages would be a big positive for consumption and investment that would further fuel employment.

    Similarly, 2Q19 employment cost index shows total compensation growth was 2.9% yoy, near the highest in the past 10 years.

    For those who doubt the veracity of the BLS employment data, federal individual income tax receipts have also been rising to new highs (red line), a sign of better employment and wages (from Yardeni).


    Regardless of which data is used, real demand has been growing at about 2-3%, equal to about 4-5% nominal.

    Real (inflation adjusted) GDP growth through 2Q19 was 2.3% yoy. A pattern of deceleration is forming, another watch out.

    Stripping out the changes in GDP due to inventory produces "real final sales". This is a better measure of consumption growth than total GDP.  In 2Q19, this also decelerated, to 1.8% yoy.

    The "real personal consumption expenditures" component of GDP (defined), which accounts for about 70% of GDP, grew 2.7% yoy in 2Q19.

    On a monthly basis, real personal consumption expenditures grew 2.7% yoy in July to a new ATH.

    GDP measures the total expenditures in the economy. An alternative measure is GDI (gross domestic income), which measures the total income in the economy. Since every expenditure produces income, these are equivalent measurements of the economy. Some research suggests that GDI might be more accurate than GDP (here).

    Real GDI growth in 2Q19 was 2.3% yoy.

    Real retail sales grew 1.6% yoy in July, making a new ATH. Sales fell yoy more than a year ahead of the last recession which makes the negative annual growth in December notable.

    Retail sales in the past three years have been strongly affected by the large fall and rebound in the price of gasoline. In July, real retail sales at gasoline stations fell by 1.8% yoy after having fallen more than 20% yoy during 2016. Real retail sales excluding gas stations grew 1.9% in July.

    This expansionary cycle is not like others in the past 50 years. Households' savings rate typically falls as the expansion progresses; this time, savings have risen and remains at an elevated level.

    The next several slides look at manufacturing. It's important to note that manufacturing accounts for less than 10% of US employment, so these measures are of lesser importance.

    Core durable goods orders (excluding military, so that it measures consumption, and transportation, which is highly volatile) fell 0.5% yoy (nominal) in July. Weakness in durable goods has not been a reliable predictor of broader economic weakness in the past (arrows).

    Industrial production (real manufacturing, mining and utility output) growth was 0.5% yoy in July. The more important manufacturing component (excluding mining and oil/gas extraction; red line) fell 0.5% yoy. Growth has moderated for both, with the cycle peak (so far) in December 2018. Industrial production is a volatile series, with negative annual growth during parts of 2014 and 2016.

    Importantly, just 35% of industrial production groups are expanding. A drop below 40% has implied widespread weakness that typically precedes a recession, so this is a watch out (from Tim Duy).

    Weakness in total industrial production in 2015-16 was concentrated in the mining sector, with the worst annual fall in more than 40 years. It is not unusual for this part of industrial production to plummet outside of recessions. With the recovery in oil/gas extraction, mining rose 5% yoy in July. 

    Companies are often wrongly accused of underinvesting in lieu of greater share repurchases (buybacks). But capacity utilization is still under 80%, so there is plenty of room for production to expand within existing capacity. 

    For 3Q19, the Atlanta Fed estimates growth to be 1.4%.


    Macro weakness had been apparent in housing, but recent data has shown improvement.

    First, new single family houses sold was 635,000 in July, rising 4% yoy. The cycle high was in June, only 2 months ago: the cycle high has been a median of 28 months before the next recession (arrows). Overall levels of construction and sales are small relative to prior bull markets.

    Second, housing starts rose 1% in July, an improvement from an 11% yoy decline in FebruaryThe cycle high was in January 2018, 19 months ago; the cycle high has typically been well over a year before the next recession (arrows).

    Building permits (red line) rose 2% yoy in July after rising 5% yoy in July 2018.

    Single family housing starts (blue line) rose 2% yoy in July; multi-unit housing starts (red line) fell 5% yoy and have been a drag on overall starts for more than 5 years.


    Despite steady employment, demand and housing growth, core inflation remains near the Fed's target of 2%.

    CPI (blue line) was 1.8% last month. The more important core CPI (excluding volatile food and energy; red line) grew 2.2%.

    The Fed prefers to use personal consumption expenditures (PCE) to measure inflation; total and core PCE were 1.4% and 1.6% yoy, respectively, in July.

    Some mistrust CPI and PCE. MIT publishes an independent price index (called the billion prices index; blue line). It has tracked both CPI (red line) and PCE closely.


    On balance, the major macro data so far suggest continued positive, but slowing, growth. This is consistent with corporate sales growth.  SPX sales growth in 2019 is expected to be about 5% (nominal).

    Valuations are slightly above their 25 year average. The consensus expects earnings to only grow about 3% in 2019 but there is room for faster equity appreciation through valuation expansion (chart from JPM).

    If you find this post to be valuable, consider visiting a few of our sponsors who have offers that might be relevant to you.