Like the stealth bomber which is almost invisible to radar, the coming recession is almost invisible to leading indicators. The yield curve which has been reliable for decades is useless right now. Other indicators are hopelessly out of calibration indicating stronger than actual growth.
The Yield Curve
For almost six decades the yield curve has given warning about 18 months before a recession would come. Every recession from 1955 to the present was preceded by an inversion where the 1 year T-Bill yield rises above the 10 year T-Bond yield. As the gap between the yields narrow the yield curve is referred to as flattening. When the short term rate goes above the long rate it is referred to as inverted. In the chart below the red line labeled yield curve shows the difference of the 10 year T-Bond yield less the 1 year T-Bill yield. The red axes are for the yield curve; the black axes are for GDP growth. The red axis drawn at the value of zero is the warning indicator for recessions. Since this version of the yield curve tends to lead growth by 18 months the yield curve and its time scale are plotted 18 months forward. When the yield curve dips below zero showing an inversion a recession usually begins about 18 months later.
The yield curve gave recession warnings 10 times in the last 60 years. Nine of them were recessions and the other was a slow down in the annual GDP growth rate from 7.7% to 2.5%. No recessions came without a warning in last 60 years. However, we can't trust it this time.
The yield curve has also been an indicator of how strong the economy would grow. Normally when the 10 year yield exceeds the 1 year yield by more than 1% (100 basis points) it indicates the economy will be strong. but in the last 4 years it has indicated growth about 2.3% higher than actually occurred.
Despite the great record on the last 9 recessions the yield curve does not always work. An inversion did not warn of recessions beginning in 1937, 1945, 1948 nor 1953. It appears that when the T-Bill yield is below about 2% that the yield curve does not warn of recessions. The chart below shows the 3 months T-Bill yield and a version of the yield curve using the difference between the 10 year yield less the 3 month (would have used 1 year yields, but did not find 1 year data going back that far).
With this version of the yield curve the warning comes when the difference of the 10 year yield less the 3 month yield drops to about 0.3% (or 30 basis points). A red axis on the chart above shows this warning level. This warning worked for the last nine recessions, the Great Depression and the other 3 recessions of the 1920s. However, the four recessions between 1935 and 1955 did not come with a warning. In the period prior to those four recessions the 3 month T-Bill yield was less than 2% as shown by the purple axis.
Going into the recessions of 1937 and 1945 the Fed was holding short term interest rates near 0%, comparable to today's policy. Those two recessions began with a steeper yield curve than we have today. The blue axis on the chart makes it easy to see the difference in yield was greater going into those two recessions than the current difference. There is not a basis to conclude the yield curve is currently forecasting a recession. There is a strong basis to believe that a recession could come without any warning by the yield curve.
Note to those not fighting the Fed
While the Fed was holding short term interest rates near 0% from about 1933 to 1946 there was a 57% decline in stock prices from 1937 to 1942, even though the monetary base expanded 68% during that bear market.
ISM 50 is the new 40
The Purchasing Managers Index from the Institute of Supply Management had a stellar record of forecasting GDP growth up until 2008. The chart below shows how the three month moving average of the ISM number corresponds with annual GDP growth 6 months later. In 2008 something distorted the close relationship. The ISM started improving in January suggesting the economy should start improving in July. Instead the economy continued plummeting for 9 more months and the historically stellar correlation went to over forecasting growth by about 2% with perhaps a longer lead time.
The best correlation between ISM and the annual GDP from 2005 forward has an 18 month lead time and suggests growth will weaken to 0.4% around the end of 2014. Prior to 2005 an ISM at 50 suggested annual GDP would grow at 2.7%. Since 2005 it appears an ISM of 50 corresponds with growth of 0.5%. Typically when annual GDP slows below 2% the economy stalls into a recession.
While the correlation above looks good for this short period it could easily amount to over fitting the data and not actually be an indicator of growth about to weaken. The point is that the historical context of the ISM number may not apply to the present and thus I don't have any confidence the ISM will give a warning for any coming recession.
Auto Sales Level or Change?
Auto sales were a great leading indicator of annual growth until the 4th quarter of 2010. The chart below shows the correlation between the annual change in auto and light truck sales per 1000 population and annual GDP. In the last quarter of 2010 annual GDP growth reached 2.4% and went no higher, while auto sales implies growth should have rocketed higher to 4.5%. If auto sales were still connected to GDP growth in the historical way annual GDP would have hit 5% last quarter. Previously I used a percentage change in auto sales to estimate GDP growth, but since sales got so low, the low base made the percentage increase of sales over estimate growth even more. Using the historical correlation the recent percentage growth in auto sales implied annual GDP growth should have hit 5.5% in the first quarter of 2013 instead of 2.1%.
Auto sales per thousand is calculated using the seasonally adjusted annual rate of auto and light truck sales divided by the estimated population of the prior month and then multiplying by 1000 to get the annual rate per thousand population. I use the prior month population estimate because they are a month slower in being released than auto sales. I calculate the annual change by averaging the last 12 months and subtracting the average of the 12 months prior to that.
Another perspective is looking directly at the correlation of annual GDP growth with annual sales per thousand population.
Auto sales had at least a short term peak of 15.482 million in November. The 14.867 million in April equates to 47.1 annual sales per thousand population and would be consistent with annual GDP growth at 1.6% in July.
Usually the level and growth of auto sales have a more consistent estimate of where the economy is headed. In the last few years neither is very satisfactory and looking at either growth or level without the other could be misleading.
Does growth in housing starts or the level give a better forecast of GDP? The growth in starts suggests Annual GDP growth should have risen in the last two quarters rather than falling and should be accelerating to 4%. The annual change in housing starts per thousand population is calculated in the same manner as auto sales above.
Meanwhile the actual number of starts at 2.7 per thousand population suggests we are deep in recession territory.
Wages Still Connected
While leading indicators seem to have disconnected from economic growth the labor market has not. Real take home pay for workers has maintained its strong growth correlation with the economy. The chart below shows wage growth (in red) tightly fitting to GDP growth (in black). Since wages lag by two months GDP implies wage growth will slow, but wage growth by itself sheds no light on where GDP growth is headed.
While their growth rates still correspond, wages have been weaker than the economy. Below is the wage series I am using. GDP has risen to new highs in the current expansion, but the total amount paid in real wages is only back to the 2006 level.
If you were to adjust the series for population (not shown), real per-capita wages have not passed the 1999 high.
I am not confident right now about any indicators for wage or GDP growth for the next few months. The case for them both weakening substantially in the next two years is a bit stronger. The next chart looks at the long term correlation between wage growth and tax policy. Further down in a section called "Capital Overreach" I'll offer some possible explanations as to why the correlations are the way they are.
Wages and The Top Bracket
The top two graphs in the chart above show the correlation between the 7 year rate of wage growth and the top marginal tax bracket as a multiple of per-capita GDP. Each point in the scatter plot on the left shows the 7 year growth rate on the vertical axis and the top bracket at the beginning of the 7 year period on the horizontal axis. For example, the highest point on the chart, represents the 7 year period ending in 1956 where wages annualized growing 5.9% and top bracket of $400,000 in 1949 shown as 218 times per-capita GDP. The green upward sloping best fit line shows there is a positive correlation between wage growth over 7 year periods and the tax bracket at the beginning of that period.
This same data is shown in the time series graph on the right with wage growth in black and the top bracket in green. The last point on the green tax bracket line represents the top bracket going to $450,000 in 2013 which will presumably influence wage growth during the 7 year period of 2014 through 2020. This was such a small increase you may not be able to see that the bracket ticked up to 9 times last year's per-capita GDP from 8 times in 2012.
Wages and Capital Gains
The two charts in the middle look at the correlation of the five year moving average of the capital gains tax rate with wage growth over the next 7 years. The blue best fit curvilinear line in the scatter plot suggests a capital gains tax rate of 27.3% maximizes wage growth. The strongest wage growth in the actual data corresponds with a 25% capital gains tax rate. Wage growth substantially weakened as the five year average capital gain rate rose to 38.5%. Wage growth was also quite strong following the two times the five year average got to 28%. The 7 year wage growth rate has declined each of the last 12 years. This corresponds with the capital gains rate being cut from 28% to 15%.
The bottom chart shows a regression model of the two tax policy variables in red. In the last 7 years wage growth annualized 0.36%. The model suggests the rate will drop to -0.3% in 2013 and remain at -0.3% through 2018. This forecast has some implications. For instance it implies that in 2016 the total real employee compensation will be 5% below the 2009 low shown in the "Fred" chart above. Using the correlation from the chart above of annual GDP growth with annual wage growth a -0.3% wage growth implies GDP growth would only annualize 1.2% through 2018.
Now let's look at why these correlations might be the way they are.
A low capital gains tax rate appears to be very beneficial to those wanting to live off of financial investment income. It appears to shift a share of GDP away from rewarding labor toward rewarding capital, as pointed out in a previous article.
In the real economy supplying the needs and desires of the population is easier the greater the number of people that are working and the better their motivation to produce goods and services. In the financial economy short term profits get a boost when jobs are cut and remaining workers are paid as little as possible. Tax policy appears to have significant influence on whether short term financial interests run a muck or are channeled to serve the public good.
Cutting the capital gains rate to 15% may be like the flip side of a labor union being powerful enough to negotiate a contract with generous wages where the hours and number of employees cannot be cut. Everything looks rosy for the workers right up until the company goes bankrupt and the employees lose their jobs and pensions. The wage squeezing that comes with low marginal tax rates eventually means the amount paid to workers is not enough to buy what is produced and the economy stagnates then shrinks and the financial markets crash.
There is a dichotomy behind the weak average income growth. Those who get much of their income from capital have rapidly growing income even as the median real income continues to shrink and the purchasing power of workers declines. Those with the rising income consume a smaller portion of it and may be happily pushing stock prices higher with the remainder of their income. Their widening slice of the pie appears to insulate them from how slow the pie is growing. GDP has only annualized 1.0% for the last 7 years. Those with declining real purchasing power who consume a much larger share of their income may continue to push that growth rate lower.
While I did not chart it above, consider the effect of the top marginal tax rate on wages. If the marginal tax rate paid by a business owner were 10% the owner would get to keep 90 cents of every dollar he squeezes out of employee wages. In the 1950s and early 1960s when the top rate was 91% a very successful business owner would only get to keep 9 cents of each additional dollar squeezed from employees. Both wage growth and GDP growth were much faster when the top rate was 91%. The top marginal tax rate has a positive correlation with wage growth. I did not chart this above because when I controlled for the top tax bracket the correlation of the top rate and wage growth became statistically insignificant.
If the top tax bracket is too low the top tax rate may function as a tax on success. As the bracket goes up the top rate may function less like a tax on success and more like a tax on greed. If we are serious about restoring wage growth the top bracket should probably be at least 100 times per-capita GDP. This would be about $5 million dollars today. I estimate the appropriate marginal tax rate for the portion of income above $5 million would be around 55% to 60%. The average tax rate on business owners and the most talented among us must be low enough that they have plenty incentive to build businesses and the economy, but to distribute the incentive where the most prosperity is produced their marginal tax rate must be high enough that they have little incentive to squeeze those below them.
Corporate America is mostly run by CEOs rather than owners, but incentive pay agreements may give CEOs even stronger short term incentive to squeeze worker pay when marginal tax rates are low.
The last time we had capital overreach in the 1920s the capital gains rate went down to 12.5% and stayed there for 12 years. The recent bout is not as serious. The rate fell to 15% and only stayed there 10 years.
The worst consequence of capital overreach probably lie ahead. I expect the 7 year growth rate in GDP should bottom in 2014 at about 0.5%. There is a chart in this previous article suggesting it might go as low as 0.1%. The chart above suggest the labor market will remain weak through 2018. In the Great Depression the labor market also recovered slower. GDP set a new high in 1936, while the number of jobs did not surpass the 1929 high until about 1941.
The labor market appears to be weakening. The stock market rallied at the May 3 announcement the number of jobs grew 0.1% in April, as though this was an indication of strength. The total number of hours worked in April released on the same day showed a 0.4% decline. While both estimates are subject to revision it appears we replaced a lot of full time jobs with part time work.
Perhaps you don't trust my analysis based on tax policy. Consider the relationship of the 7 year growth rate with the trade balance.
The black line in the chart above shows the 7 year growth rate of GDP, which dropped to 1.0% in the first quarter of 2013. The trade deficit as a share of GDP, shown in red, leads eight and a quarter years and suggests the GDP growth rate will drop to 0.6% in the first quarter of 2014. To hit that target GDP growth would have to drop at a 1% annual rate in each of the next four quarters.
The large trade deficit that peaked in 2005 represents Americans consuming much more than they produced. The consequences of the associated accumulation of debt have long lead times. The full negative impact on growth will probably hit some time in the next two or three years and will almost certainly include a recession.
One caveat about this chart is that prior to 1980 when the nation largely ran current account surpluses the correlation between trade and growth was not nearly as strong as it has been the last 23 years.
Implications for Stock Market
The stock market appears to assume that the baseline growth rate, around which the quarterly and annual GDP growth numbers will fluctuate, is 2.5% to 3%. The assumption then goes on since growth has been running below baseline it should pick up to 2.5% this year and around 3% next year and corporate earnings should be strong. As outlined above the leading indicators upon which stronger growth forecasts are based may be mis-calibrated for the current environment. The baseline GDP growth rate may be less than 1%. Annual growth could turn negative soon to balance out with above baseline growth over the last three years.
Real as reported earnings of the S&P 500 (NYSEARCA:SPY) have become more volatile than any time in history and are very sensitive to weak economic growth. As shown in the chart below they dropped 54% from 2000 to 2002. This drop came with annual GDP growth only going down to 0.9% in the 2001 recession.
The earnings dropped 92% in the great recession. Yes they came roaring back, but not to a new real high. They are currently 25% above their long term trend represented by the green best fit trend line. It appears earnings could be starting back toward (or below) that trend. The peak appears to have been put in 3 quarters ago.
If annual GDP growth declines just a bit more corporate profits should start significant declines. The debt leverage of corporate America that allows double digit profit growth with GDP growth only a little above 2% can lead to big declines with GDP growth just a little under 2%. The slight weakening of real as reported earnings in the last three quarters suggests economic growth is right at the fulcrum point between rapid earnings growth and rapid decline. A slowing second quarter could be a tipping point.
The forward estimates put out by Standard and Poor's predict earnings will grow at a 16.6% rate to the end of 2014. If/when earnings start declining it should be a massive bear shock to stocks.
CEOs have been diligently trying to "earn" their bonuses. It might be more accurate to say they are trying to receive their bonuses. At any rate as companies have been reporting weaker than expected revenue this earnings season. They are also reporting better than expected "operating" earnings. Unlike operating earnings as reported earnings are coming in weaker than forecast. The cost cutting that allows operating earnings to still hit or beat the target falls largely on labor.
In the last reporting period industrial production, real personal income, the number of hours worked, new manufacturing orders and real retail sales excluding food services were all below their recent highs, although not all were down for the month. It will be several months before we know if growth rolls over and April was the beginning of a recession or if revisions make it not even a weak month.
I am not sure when it begins, but a stealth recession seems likely this year or perhaps in 2014. There may be almost no warning from the usual indicators. Estimates of future earnings and stock market gains could remain rosy right up until they crash.
Disclosure: I am short SPY. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.
Additional disclosure: There is no guarantee analysis of historical data their trends and correlations enable accurate forecasts. The data presented is from sources believed to be reliable, but its accuracy cannot be guaranteed. Past performance does not indicate future results. This is not a recommendation to buy or sell specific securities. This is not an offer to manage money.