Tax Rates and Economic Growth – A Look at the Data on Marginal Tax Rates, Capital Gains Tax Rates, and Real GDP per Capita

This post looks at the relationship between tax rates and economic growth. Because this is a controversial subject, I will approach this methodically, and also (where possible) approach the problem (and the solution) from more than one angle. It makes for a long post, so be forewarned. If you want to read it, make sure your chair is comfortable.

Here’s the plan… First, look at how the top marginal income tax rate, the bottom marginal income tax rate, and the top capital gains tax rate evolved over time beginning in Ike’s term graphically. Next, look at the growth rate of real GDP per capita observed at these different tax rates. Where possible, when a “maximum” or optimal point is uncovered, use correlations between tax rates and growth rates to confirm whether that point truly is a maximum.

Real GDP per capita comes from NIPA BEA Table 7.1. The highest and lowest marginal tax rates come from this IRS document. Capital gains tax rates come from Citizens for Tax Justice. Regular readers are aware I prefer to get data directly from the source, but after 30 minutes I didn’t find the time series on the IRS website. I’m sure its there, I’m just not finding it. That said, the data does seem to match my recollections, and while Citizens for Tax Justice might be a biased source, I suspect they got the data right.

So let’s start by looking at how these tax rates have evolved over time…

We can group the top rates into a few categories…. There are many ways to skin a cat, so to speak, but I think a natural division (trying to keep bunches of tax rates together, and where possible, trying to keep the bunches more or less of the same size) might be the following:

Group 1: 90% to 93%, 11 observations
Group 2: 78% to 69%, 18 observations
Group 3: 60% to 50%, 5 observations
Group 4: 40% to 36%, 11 observations
Group 5: 36% to 25%, 9 observations

Similarly, the bottom marginal rates could probably be grouped as follows:

Group 1: 20% to 23%, 11 observations
Group 2: 15% to 20%, 14 observations
Group 3: 14% to 15%, 17 observations
Group 4: 11% to 14%, 6 observations
Group 5: 10% to 11%, 6 observations

Finally, we can group the capital gains tax into the following categories:

Group 1, 35% to 40%, 7 observations
Group 2, 30% to 35%, 5 observations
Group 3, 25% to 30%, 26 observations
Group 4, 20% to 25%, 12 observations
Group 5, 15% to 20%, 4 observations

Admittedly, other divisions are possible, but for the most part, unless one’s intent is some serious cherry picking, it won’t make much a different to much of what we’re going to do on the rest of the post. (I’ve tried a few breakdowns for grins but I’m not including them due to the length of the post.)

Let’s start by looking at the top marginal rate and how it compares to growth rates since 1953…

What we see is that when the rate is “too high”, growth rates are pretty low – presumably this is because the high tax rates discourage private enterprise. Similarly, when tax rates are “too low”, growth rates are low – this is probably because in such instances, the government shrinks too much to provide services essential to economic growth, or it must borrow heavily to finance its activities. (Put another way… taxes are too high when the private sector can make better use of a marginal dollar than the government, and they are too low when the government can make better use of a marginal dollar than the private sector.) So what’s the happy medium? Highest observed growth rates occurred when the top marginal rte was between 50% and 60%. And the shape of the distribution seems to bear out that this range is the optimum.

But this could be mere coincidence. Luck and whatever else. Its certainly heresy to anyone who has ever voted Republican. So its worth taking another bite at this apple. And here’s how we’ll do it. If there is a such a relationship, call it a “Laffer curve for growth”, then at points to the left of the optimum, we should expect, on average, that increasing the tax rates should lead to faster growth rates, and at points to the right of the optimum, we should expect, on average, that decreasing the tax rate should lead to faster growth rates. Put another way, when tax rates are below the optimum, the tax rate will be positively correlated with growth. Conversely, when tax rates are above the optimal level, we would expect to see the tax rate negatively correlated with growth.

Below is a table showing that relationship…

The first row can be interpreted as follows… when the top marginal tax rates are equal to or lower than 39%, the correlation between the top marginal tax rates and the growth in real GDP per capita is equal to –0.08. Similarly, the correlation between tax rates and growth rates is 0.02 when tax rates are above 39%. Anyway, none of the correlations are especially high – not surprising given the fact that we’re only looking at one variable (i.e., top marginal tax rate) to explain growth, but nevertheless, the signs of the correlations are as we expect for the range we’re observing, and the highest sum of absolute lower and higher correlations seem to occur in the range the earlier graph led us to expect.

Thus, given the historical data, it seems that a) the fastest growth rates were observed when marginal tax rates were between 50% and 60%, and a simple test of correlations provides a bit of support to the hypothesis that the tax rate that provides the fastest growth is indeed somewhere in the vicinity. Is this conclusive evidence? Heck no – I’m well aware people much smarter than I have used far more sophisticated tools to do this sort of work, but as time goes on I’m more and more convinced of the value of using simple tools. One thing I think is evident from the graph and the correlations… there isn’t much in the data to support having the top marginal tax rate at the current 35% rate if one wants to encourage growth, and there is even less in the data to support cutting the top marginal tax rates any more as some Presidential candidates have advocated.

Moving along, we do the same thing for the bottom marginal rate. Here’s the graph.

Notice that again we get the familiar bow shaped curve. And once again, the optimum is evident… somewhere in the neighborhood of 11% to 14%. Which in fact, is not too far distant from the range which we’ve seen in the not very distant past. (And for the economists among us… charging low income people lower rates than the high income individuals makes sense – the former have a higher marginal propensity to consume, so the multiplier effect is going to be much larger when their tax rates are very low relative to the tax rates faced by those with high incomes. Personally, I think taxes should go on wealth and not income, but that’s another story.)

Now the correlations between the lowest marginal income tax rates and real GDP per capita…

Seems the 11% to 14% range looks most plausible closer to the 11% rate.

Now what about capital gains tax rates? Here’s the picture.

Well, we have a problem… it seems we don’t have the familiar bow shape … it may be because we simply haven’t observed rates larger than the optimum, because while the graph suggests 20% to 25% may constitute a local optimum, it also suggests that the overall optimum is at least in the 35% to 40% range. Sadly, the shape of this curve precludes much of a correlation analysis – we simply don’t have observations above the 35% to 40% range.

What we learn from this is that regardless of how one chooses to interpret the graph, the current 15% capital gains tax is way too low – keeping tax rates that low is reducing economic growth and hurting us all, including those who pay taxes on capital gains. (In the short term they may be better off paying less in capital gains, but it comes at a cost of less economic growth which in turn decreases their income opportunities.)

This post is getting way too long, but I want to add in a few details… I want to check the results of tax cuts themselves, and compare them to tax hikes.

The graph below shows the median and average growth rates in the year of a cut in the highest marginal tax rate, the year after such a tax cut, two years after such a tax cut, three years after a tax cut, and four years after such a tax cut. It also shows the same after a tax hike. (Note… there have been instances where a change in tax rates in one direction was quickly followed by a change in tax rates in the other direction, one or two years later. For instance, in 1969, there was a tax hike, followed by a tax cut in 1970. I count the tax hike in 1969, but don’t count 1970 as one year following a tax hike or 1971 as two years following a tax hike since the more recent tax cut nullifies the hike. Note also.. there have been 12 cuts in the highest marginal rate, and four hikes in the sample.)

What does this show? Well, it shows that the economy has grown faster two years after a cut in the highest marginal tax rate than two years after a hike in the highest marginal tax rate. Similarly, it has grown faster three years after a highest marginal rate tax cut than three years after a highest marginal rate tax hike. But… growth rates are slower in the year of a highest marginal tax cut than the year of a highest marginal rate tax hike, as well as one year out and four years out. This isn’t much data, but its clearly not exactly evidence that cutting the highest marginal tax rates is any kind of a growth producer.

What about cutting the lowest marginal rate? Well, there’s only one year in which there was a tax cut, so it kind of not instructive.

What about cutting the capital gains tax rate? Seven cuts, eight hikes…

Results are similar to cutting the highest marginal rate. Put another way… in the five years beginning with a change to the tax rate, tax cuts are associated with faster growth than tax hikes 40% of the time, and with slower growth than tax hikes 60% of the time.

Cuts beat hikes 2 years out and 3 years out, and are beaten year of, year after, and are walloped 4 years out. Again… cutting the capital gains tax rate isn’t exactly a formula for super duper growth.

So… what can we conclude from all this? Well, there isn’t much evidence that cutting tax rates, at least to high income earners and on capital gains, will lead to faster growth. In fact, it would seem that since Nixon’s time, tax rates have been below the “optimal” level for generating growth, possibly even leading to slower growth over the past thirty five years or so.

This post may go some way toward explaining why since Ike took office, Democratic Presidents have generally produced faster rates of growth in real GDP per capita than their Republican counterparts. Some time soon I’m going to look at that in more detail.

(Note to regular readers who may be wondering why I did the entire analysis for real GDP per capita, as opposed to my favorite measure, real (GDP – change in the national debt) per capita. Basically, I wanted this to be as easy to replicate as possible. Also, those parts of the analysis I did using that measure of growth were even more damning of the “tax cutting” philosophy.)