- The U.S. economy will likely grow less than 1% this year with a recession starting in 2014 or 2015.
- Economists have missed the lesson about marginal income tax rates they should have learned from The Great Depression and great recession.
- This misunderstanding likely makes 2014 the 10th straight year mainstream economists over forecast growth for the coming year.
- The misunderstanding of marginal tax rates also prevents seeing bubbles.
- There is a stock market bubble and it will pop with the coming recession.
Federal Tax policy has a long history of influencing the economy and investments. Understanding the actual influence may be crucial to navigating the next few years. To paraphrase Upton Sinclair:
It is difficult to get [an economist] to understand [the influence of marginal tax rates] when his [consulting fee or salary] depends on not understanding it.
The common belief is that lower marginal income tax rates lead to stronger growth. This would be an inverse relationship where the growth rate and tax rate move in opposite directions. The empirical reality differs starkly.
Below we will look at logical framework of why low marginal tax rates are bad for growth and then show the empirical evidence that they are. Using this evidence we'll build regression models that have predicted long-term growth with much better accuracy than most economists. Finally, we will look at why low marginal tax rates are a bigger influence behind bubbles than low interest rates.
Growth benefits from business owners having a low average income tax rate with a high marginal tax rate. Their average rate has to be low enough that it is worthwhile for them to run and/or fund a business. Their marginal tax rate has to be high enough they avoid taxation by building value in a business.
If marginal tax rates are too low, business owners take more personal income out of business revenue, leaving a smaller portion to be spent on wages, equipment, training, research, inventory and marketing. Dollars spent on deductible items within a business have a larger growth effect than dollars pulled out as personal income. Many of the dollars pulled out may do little for growth and only serve to bid up prices of rare art works, beachfront property, gold, stocks, bonds or luxury goods.
If low marginal tax rates are in place long enough the diversion of business revenue away from expenditures with a high growth effect eventually results in a Great Depression or great recession.
The Empirical Case
Marginal tax rates appear to have a curvilinear relationship with growth. If the top marginal rate is above or below that optimal rate, growth suffers.
The chart below shows annual GDP growth from 1920 through 2013 with the top marginal tax rate leading two years. We'll examine other lead times below, but for now let's look at the lead time with the best correlation.
Each point on the scatter plot (left graph) represents what growth was one year shown on vertical scale and what the tax rate was two years before on the horizontal scale.
The highest point represents 18.9% growth in 1942 and a top marginal rate of 81.1% in 1940. The lowest point represents the economy shrinking 12.9% in 1932 and the top rate of 25% in 1930. The second lowest point on the chart depicts GDP shrinking 11.6% in 1946 and a 94% marginal tax rate in 1944. The green curved line is the best quadratic fit of the data. The red line is the best linear fit. This same data is shown in the time-series plot on the right. The blue line shows annual growth. The green and red lines are estimates of the influence the top rate has on growth based on the best linear and curved fits in the scatter plot.
Here is a closer look at the scatter plot with a theoretical inverse correlation shown in pink. Most people seem to believe in something like the pink line.
In the chart above, the red line shows a positive relationship where growth and the tax rate move in the same direction. The green line shows a positive relationship through the marginal tax rate rising into the mid 60s percentile rage. Those who are uncomfortable with this positive relationship correctly point out that correlation does not prove causation. For example, it is true that rising ice cream sales do not cause warmer temperatures. On the other hand, if there were a theory that warmer temperatures caused ice cream sales to decline (an inverse correlation) while the empirical data showed a positive correlation, the theory would be proven false. And so it is with the theoretical pink line in the chart above.
Economists and others, who promote this belief that lower marginal rates lead to faster growth, never specify when the influence is supposed to occur. To get specific would make it too obvious the theory is wrong.
To make the point, let's examine lots of lead times and multiple time periods. For the time period from 1920 through 2013, I have examined all the lead times from zero years (concurrent) out to fifteen.
Below are two scatter plots, one with no lead time and the other with a three year lead time.
Note the red linear best fit correlations have a positive slope in both charts. The best fit linear correlation was positive with every lead time up to 8 years. It was inverse for the lead times from 9 to 14 years, but not statistically significant, particularly if the tax variables outlined below are accounted for. The strongest inverse correlation had a 12 year lead time with an R-square of only 0.04. The inverse correlation is probably just an artifact of the volatility of growth from the 1920s through the 1940s. If we analyze growth from 1950 through 2013, the correlation between the top rate and annual growth is positive for every lead time from zero through 23 years.
Now let's take a closer look at the correlation/influence low marginal rates had in The Great Depression and great recession.
The chart above shows the correlation/influence of the top rate on growth during the 1920s and 1930s. Annual GDP growth is shown in the grey bars using the black scales. The top rate is shown in the dashed red line using the red scales. Note the red time scale is pushed forward 2 years to show the leading influence. During the 1920s, the top rate was cut in a series of steps from 73% to 24%. It was only at 24% in 1929 and then went to 25% in 1930. The top rate was at 25%, or less, for seven years. These seven years correlate with the growth from 1927 through 1933. This was the weakest 7 year period in history where GDP annualized declining 3.2% a year as symbolized in the lowest point on the black line above.
In June 1932, the top tax rate went from 25% on the portion of income above $100,000 to 63% on income above $1 million. After the two year lag, GDP grew 11% in 1934. GDP and industrial production set new highs in 1936. However, non farm jobs did not fully recover until 1940.
This tax increase appears to have started the recovery from the Great Depression. A few days after the tax increase, the stock market hit its depression low in early July. It rose 135% from June 1932 to July 1933 for the best 13 month period in U.S. history.
Strong growth in the early 1920s corresponded with high marginal tax rates. The decline into The Great Depression followed low marginal rates. Recovery came with high marginal tax rates. Crucial to the high marginal rates being effective were high brackets that allowed business owners to have a low average tax rate if they plowed revenue back into the business.
We will look more at the importance of tax brackets later. Let's go ahead and see the tax policy around the great recession. For better context the chart below starts in 1984 and looks at growth related to both the top tax rate and the capital gains rate.
In the chart above the top rate (in green) leads growth by two years as in previous charts. The capital gains rate (in blue) leads growth by four years. The model (in red) combines the influences of the two tax rates. The tax rate axes are scaled so that a move up or down in the tax rate matches the move up or down in the model.
Higher marginal tax rates are good for growth in the period and range covered in the chart above. The great recession started in 2007, four years after the capital gains rate was cut. The other two recessions in the chart also correspond to a cut in either the capital gain rate or top rate. The eight years from 2007 through 2014 are influenced by the top rate of 35% and the capital gains rate of 15%. The model estimates growth of 0.9% for these eight years. By contrast, the strongest year of growth, 1984, corresponds with a 50% top rate and a 28% capital gains rate.
The bottom red line in the chart showing the estimated 7 year growth rate suggested the 2007-2013 period would grow at 0.9%. Actual growth was 1.1%. I am expecting the rate for the 7 year period ending this year to head down toward or below the 0.9% estimate.
Growth volatility during a business cycle is mostly smoothed out by looking at the 7 year growth rate. The close fit between the model and 7 year growth rate implies most of the variation in long run economic growth in the last 30 years has been determined by the incentive marginal tax rates give business owners to either plow money into deductible expenditures within the enterprise to avoid taxes, or to pull it out as personal income.
In other words, prosperity benefits when business owners build wealth within their business or in starting new businesses, rather than taking large personal incomes and making financial investments. The growing wealth within a business can remain untaxed until the owner takes it out as personal income, the business is sold or it gets taxed in the estate. Tax brackets play a pivotal role in balancing the incentives to build a business and avoid taxes.
Top Bracket and Growth
High tax brackets are the key to having a low average tax rate with a high marginal rate. For a person in the top 0.01% of society it would make a lot of difference to their average tax rate if a marginal tax rate of 60% kicked in at $200,000 or $200 million. Below we will look at what the top bracket has been, then see its positive relationship with growth and finally at how it interacts with the top rate and capital gains rate to influence growth.
The chart above shows how the bracket has changed since 1920. The black line in the chart above shows the top bracket has ranged from $5 million (1936-1941) to $29,750 in 1988. In Constant 2012 dollars (green line) that range is $83.6 million to $60,000. My research suggests adjusting the bracket by per-capita GDP gives more significant results. The red line shows the top bracket this year is just under 9 times last year's per-capita GDP. The highest multiple was 8,571 in 1936 which would be the equivalent of about $450 million today.
In 1965 the top bracket was cut from $400,000 to $200,000. Since 1965, the bracket has never been above 56 times per-capita GDP. Prior to that, it had never been below 118 times. The top bracket has its strongest influence on growth with a three year lead time. The next several charts will examine the difference between the high bracket era and the low bracket one. The dividing point will be 1968 to account for the 3 year lead time.
The chart below shows the linear correlations between growth and the top tax bracket in the high and low tax bracket eras.
Each dot in the scatter plots represents the annual growth rate one year and the top bracket as a multiple of per-capita GDP three years earlier. The strongest growth on the chart, 18.9% in 1942, corresponds with a top bracket that was 7,435 times per-capita GDP in 1939. The high bracket era scatter plot on the left has a positive sloping dark orange best fit line. The low bracket scatter plot in the middle has a positive sloping light orange best fit line. This influence on growth is also shown in the time series plot with dark and light orange lines.
The high bracket era appears to have a steeper sloping best fit line, but if the low bracket scale went from zero to 10,000 rather than zero to 60, it would be obvious that the low bracket era actually has a steeper sloping best fit line.
Top Rate and Growth
The top tax rate has curvilinear correlations with growth in both the high and low bracket eras. The best curvilinear (quadratic) fits in the chart below suggest the growth optimizing top rate is 64% in the high bracket era and 54% in the low bracket era.
In the high bracket era the highest growth corresponds with an 81% top tax rate. Conventionally, this strong growth is attributed to WWII. I have come to believe it is mainly due to the highest tax brackets in history and a favorable capital gains tax rate. The strongest growth in the low bracket era came with a top rate of 50%.
While the 50% rate appears to have improved growth, it probably could have been better. In the five years the top rate was 50%, the top bracket averaged 8.7 times per-capita GDP. This low bracket pushed many of the near affluent into the arms of tax shelter salesmen. There were a lot of junk investments where uneconomic oil prospects were drilled. Office and apartment buildings were completed with no tenants ready to move in. There is not enough data to make a precise estimate, but I suspect the bracket that would give the 50% tax rate the optimal balance between a low average rate and a high marginal rate would be about 30 times per-capita GDP, or about $1.6 million based on last year's GDP. If the top rate got into the mid 60 percentile range, the optimal bracket is likely in the hundreds of millions.
Capital Gains Tax Rate and Growth
The capital gains tax rate like the top rate appears to have curvilinear relationships with growth where the growth optimizing capital gains rate in the high bracket period is around 54% and the low bracket's growth optimizing tax rate is around 27.5%.
Unlike the top tax rate, the capital gains rate appears to differ in its lead time between the high and low bracket eras. In the High bracket period the dark blue best fit curve is based on the capital gains rate leading growth by 5 years. In the low bracket period, the light blue best fit curve is based on leading growth by 4 years. The difference in lead time may have more to do with different time periods or some other factor than with the level of the bracket.
In the high bracket period the best growth, 1942, corresponds with a 39% capital gains rate, while the Great Depression was influenced by a 12.5% tax rate.
In the low bracket period the best growth, in 1984, lines up with a near optimal 28% tax rate. The great recession corresponds with the 15% tax rate. The back to back recessions of 1980 and 1981-82 were influenced by the above optimal capital gains rate of 39.9%.
The capital gains tax rate around 39% highlights the importance of a low average tax rate with a high marginal rate and of the tax bracket. When the top bracket was over 7000 times per-capita GDP, the rate of 39% appears to be favorable to growth. When the bracket was less than 20 times per-capita GDP this rate appears to be detrimental to growth.
Combining Tax Policy Influences
In the chart below the black line shows growth. The dark red line shows a regression model combining the high bracket influences of the top tax rate, the top tax bracket and the capital gains tax rate. The lighter red line show a regression model for the tax policy influences in the low bracket era.
The low bracket model estimates a growth rate of 0.8% for the years 2007 through 2014. Estimated growth picks up to 1.4% for 2015 and 2016, assuming the top rate remains at 39.6% for the rest of the year.
The chart below shows the 7 year growth rate and the 7 year estimate based on the high and low bracket models.
I have found no other theory of what influenced long-term growth over the last 95 years that comes close to the explanatory accuracy of the above interpretation of tax policy.
It appears popular to blame the Fed and low interest rates for bubbles in asset prices, but this finger pointing does not withstand empirical analysis. The culprit is low marginal tax rates.
If low interest rates caused financial bubbles, we would have had bigger bubbles in the 1950s than we have had so far in the 21st century and the biggest U.S. stock bubble of all time would have been preceded by low interest rates.
The table below shows the conditions in seven year periods leading to bubbles in 1929, 2000 and 2007, as well as a control period 1950 to 1956 when interest rates were unusually low while marginal tax rates were very high and there were no bubbles to speak of.
Avg. Treasury Yield
Avg. Marginal Rate
7 year period
1923 - 1929
1950 - 1956
1994 - 2000
2001 - 2007
The period leading into the 90% stock decline from 1929 to 1932 had the highest interest rates in the table and the lowest marginal tax rates. The financial crisis was preceded by the second lowest marginal tax rates and fairly normal interest rates.
Bubbles happen when the price of an asset is bid up well beyond the underlying intrinsic value of the asset. For stocks, this process appears to be influenced by marginal tax rates. As discussed above when marginal tax rates on personal income are low, more revenue, or even equity, is pulled out of businesses as personal income. Some of this money pulled out is then used to buy stocks, thus bidding up the price. So low marginal rates tend to weaken companies on the inside and bid up the price on the outside: practically the definition of a bubble.
Low interest rates may contribute to bubble formation if marginal tax rates are low. If marginal rates are high, low interest rates more likely contribute to investment in productive capacity inside a business.
While I don't have a precise measure of bubbles such that I could calculate a lead time with which marginal tax rates lead to inflated stock prices, I suspect the 15% capital gains rate in place through 2012 is still influencing stock prices into bubble territory. There are various measures such as Tobin's Q, Shiller's PE, the ratio of stock value to GDP and my PEses that all point to stocks being quite overvalued.
The PEses is approaching the 2007 high and is higher than any time prior to 1997. Valuation has a strong correlation with real total returns for periods of about 15 to 20 years. While Dr. Robert Shiller popularized looking at returns over 10 year periods, the correlation is stronger with longer periods.
Here is the correlation of total real S&P 500 returns over 17 year periods (black line and scales) and the inverted PEses on a log scale (red line and scales).
The chart depicts that when valuation is high (red line is low using inverted red scale) that return over the next 17 years is low. Where low valuation (the red line is high on chart) leads to higher returns the next 17 years.
The last point on the black line plotted at March 2014 is based on an estimate of the average closing price for the first 21 days of March. It shows a 17 year annualized real return of 4.62% from March 1997. The PEses of March 1997, 41.1, plotted directly below estimated the return would only annualize 0.1%. This is biggest underestimate on the chart.
The correlation since 1960 has been very strong, but in the last two years, the 17 year return has held up in the 4% to 5% range while the estimate plunged to about 0%. In the past when the actual return was far above the estimated return, such as 2001 and 2007, the market fell enough to bring the actual return down to or below the estimate. It remains to be seen if or when that will happen this time. There is more information on PEses in this article.
The last point on the red line shows PEses at 45 and implies the return over the next 17 years after inflation and dividends will annualize losing about 1%. This would be comparable to the period ending August 1982. I expect all of that loss to come in the next few years and set up an extraordinary buying opportunity.
Valuation is not a timing tool and can easily be off three years or more. Just because this or any other measure of value shows a market to be overvalued does not mean it will not be more over valued next year.
While stocks are probably in bubble territory, the deflation may not come until economic and/or earnings growth turns down. The signs pointing to this are quite mixed: a few suggest the decline has started, some that it is several months away, while others don't even show a decline coming.
Prosperity depends on the wealthy building businesses that create more wealth and employment, as opposed to living a life of ease off existing wealth. The tax policy that most encourages prosperity appears to be a low average income tax rate on the wealthy with a high marginal rate.
The lagged effect of low marginal tax rates suggests the baseline growth rate is around 1% this year and 1.5% in 2015 and 2016. Growth closer to the long-term average of 3.3% should begin in 2017 when the lagged effect of increasing the capital gains tax rate should begin influencing growth.
With an expected growth rate near 1%, normal fluctuations in growth mean deeper recessions and a larger percentage of time in recession than during times where the normal growth rate was above 3%.
Forecasts that growth will return to 3% this year or next probably assume that 3% is still the normal baseline growth rate. If that were true, the last eight years all of which have been weaker than 3% should make growth bouncing up to 3% a sure bet. If a sub 1% growth trend is the reality, the last four years of above trend growth make a decline in growth a strong probability.
Current market valuations rest on the assumption that we are returning to "normal" growth. If/when growth falls back to 1% or turns negative this year or next, the jarring gap between expectation and reality could create a stock market crash.
The combination of weak growth, high valuation and the shift to unfavorable demographics create the potential for a 40% to 80% decline in U.S. stock indexes (NYSEARCA:SPY) sometime in the next one to five years. Protecting capital should be the primary concern.
Diversification abroad should play a key role in protecting and growing capital. The U.S. has had the reputation of the least dirty shirt in the hamper the last few years and U.S. stock valuations rose relative to international indexes. A decline in the growth rate and especially a recession could reverse the relative gain of the last few years. However, if the U.S. market goes down, other stock markets will likely follow. International bonds probably merit a larger-than-normal slice of the portfolio in the next year or two. American Century's International Bond Fund (BEGBX) has outperformed the U.S. stock market so far this year. It was also a strong performer during the last two bear markets.
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. I am long BEGBX