This copy is for your personal, non-commercial use only.

What the Numbers Tell Us
Unemployment And Marginal Tax Rates

We, as a nation, are now in our 5th year since the beginning of the greatest recession we’ve seen since the great depression (Federal Reserve Bank of Minneapolis, 2012). Indeed, at least one prominent economist and Nobel Prize winner has made the argument that we are in a second depression (Krugman, 2011). At the same time, perhaps predictably, U.S. school systems have been increasingly under funded with well over one hundred thousand teachers having been laid off nationally, even by the most conservative estimates (Kessler, 2012). But unemployment has been stubbornly high, above 8% for four years running (U.S. Department of Labor, 2012). People are out of work, therefore tax revenues are down, so schools, law enforcement, fire departments, etc. will have to survive on smaller budgets, while they are simultaneously expected to improve their services under increasingly austere conditions. Or so the argument goes. Tax revenues and tax rates cannot be augmented until we have a strong economy once again, so we will have to make due. We cannot possibly consider increasing taxes on anyone during a recession. This is a logical and persuasive argument, or so it would seem, one that many of us have heard trumpeted loudly in recent years.

This argument suggests that the strength of the economy is the driving force that determines the amount of revenue available to fund public services across the country. In other words, a stronger economy will bring in more tax dollars because more people will be employed. Simple enough. Except that for the last three decades, politicians, think tanks, and special interest groups have been making the case that lower taxes will strengthen the economy because it frees up capital for job creators and those in the private sector to then spend, thus keeping businesses thriving and people employed. This argument presupposes a cause and effect relationship. It suggests that low tax rates lead to more money circulating through the system, creating a stronger economy and lower unemployment. To test this claim we can examine, empirically, two essential parts of this equation: We can test the relationship between marginal tax rates and unemployment (with low unemployment acting as an indicator of a strong economy). We can ask the question; do low tax rates correlate with low unemployment and vice versa? If indeed they do, then it would lend validity to the argument that increased tax revenues should only take effect after the economy recovers and unemployment drops.

The Longitudinal Trend in Tax Rates: 1932 – Present

Before we answer this question it is worthwhile to examine recent trends in one part of this equation: marginal tax rates. Pundits, politicians, media personalities, and the person on the street may make the case that current tax rates are “sky high”, suggesting that Americans now pay a higher percentage of their incomes than the historical norms. But does the data support this notion? When we consider the top marginal tax rates since prior to World War II, the answer is an emphatic no (Tax Foundation, 2012). There has been a clear and continuous downward trend in the top marginal tax rates since 1932, and Americans now enjoy the lowest tax rates they have in three generations. The current marginal tax rate for those in the highest bracket is 35%, the same as it’s been for the last decade and the lowest it’s been in 80 years, with one brief exception that we will discuss shortly. When President Clinton was in office the highest bracket was 39.6%. Interestingly, when Reagan was president and tax reduction became a staple of the Republican platform, the highest bracket was 50% for most of his 8 years in office. The two decades prior to that it was 70%, and from 1963 back until 1945 it was 91%. In 1945 it was 94%. So the point is clear when you examine the actual numbers: Taxes have never in modern history been lower in the U.S., except for the following caveat.

There was an interesting anomaly from 1988 to 1992 when the highest tax rate dipped to between 28% and 31% (Tax Foundation, 2012). And what happened to the economy during that time of low taxes? There was a large recession beginning in 1990, one that pales by today’s standards, but a significant one by historical standards (The Economist, 2011). President G.H.W. Bush saw the harm this was doing to the economy and raised taxes, breaking his “Read my lips- no new taxes” pledge. This of course was one of the factors that caused him to lose the election in 1992. Prior to that, in 1982, there was another tax cut (Tax Foundation, 2012) and another deep recession (The Economist, 2011), leading to the two highest back-to-back yearly unemployment rates we have seen since the great depression- 9.7% in 1982 and 9.6% in 1983 (U.S. Department of Labor, 2012). Our latest tax cut went into effect in 2003 and within five years the Great Recession was well underway. It would seem that tax cuts correspond with big recessions. When you subtract the substantial amount of money that wealthy individuals and large corporations contribute to our federal government and the overall economy, bad things tend to happen to that economy.

But while the tax part of our equation shows a clear pattern- a consistent downward trajectory for 80 years, with tax cuts tending to correspond with recessions- the second part is less clear. Since World War II the unemployment rate each year has fluctuated with no discernable pattern to the naked eye, from a low of 2.9% in 1953 to a high of 9.7% in 1982, and a wide variety of levels across the years (U.S. Department of Labor, 2012). So it was determined that inferential statistics would be needed to analyze the relationship between the top marginal tax rates and unemployment since 1948, when the first unemployment statistics where available through the U.S. Department of Labor.

Analysis: Correlating Marginal Tax Rates and Unemployment

The data for the top marginal tax rates (Tax Foundation, 2012) and the unemployment rates for each year (U.S. Department of Labor, 2012) from 1948 to 2011 were compiled. These numbers were entered into a Pearson product-moment correlation analysis, two tailed, to test for the strength and direction of correlation and for statistical significance. The results indicated that there was a statistically significant negative correlation, r(64) = -.31, p = .013, in the relationship between top marginal tax rates and the unemployment rate in the 64 years from 1948 to 2011. This means that when taxes were high, during that same period unemployment tended to be low, suggesting a stronger economy. And when taxes were low, during the same period unemployment tended to be high, indicating a weaker economy. It is important to note that this is not a political argument; it is a mathematical one. This is what the numbers tell us when this statistical analysis is conducted.

Now there are a number of issues to consider in this analysis. First, 64 years is a relatively small sample size (N = 64). With a sample size this small we would often not expect to see a statistically significant correlation. In many cases there simply would not be enough data for the probability to reach .05, much less .013. But even with this relatively small sample size, the association between tax rates and unemployment did prove to be significant, which suggests that longer trend lines, perhaps 80 or 100 years, would reveal a more pronounced relationship between those variables. The more data you have, the clearer the relationship often becomes, as long as that relationship is not due to random chance. And this analysis suggests the relationship between the top marginal tax rates and unemployment is not due to random chance, or at least we are 98.7% certain that it is not.

Another thing to keep in mind is one of the first concepts we teach students in introductory statistics and research courses: correlation is not causation. We cannot make the claim that one variable in this equation causes the other variable, even though they clearly seem to be associated with one another. In fact, we know that unemployment rates do not cause the top marginal tax rates to be what they are at any given time. Top marginal tax rates are set (caused) by the laws implemented by state and federal legislatures. Even if one were to argue that law makers’ decisions on tax policies are influenced by unemployment rates, the election cycle and legislative cycle normally play out over a number of years, sometimes decades, when unemployment often fluctuates a great deal, so it is not reasonable to contend that unemployment rates cause the marginal tax rates to be what they are. However, it is quite plausible that top marginal tax rates have a causal effect on the unemployment rate, particularly since those tax rates have shown a steady and consistent downward pattern and may only change once or twice per decade. In other words, it is possible that the top marginal tax rates may be one of the primary factors that dictate the unemployment rate and the strength of the economy at any given time. But since we are dealing with a correlation, we cannot claim to have isolated that variable as a cause, and it is quite probable that other factors are in play despite the clear relationship between the two variables.

However, a correlation does not rule out causation, of course, and if there is causation in this relationship, then it can only be unidirectional. The unemployment rate cannot dictate tax rates. We know what causes tax rates to be what they are: laws enacted by legislatures. So if there is a cause and effect relationship present, it could only be in the opposite direction, with tax rates influencing unemployment rates. But a critic could legitimately argue that examining tax rates and unemployment rates during the same year is ineffective because if tax rates were indeed affecting fluctuations in the unemployment rate, the impact would not appear until the following year or later. A proponent of the hypothesis that low tax rates allow job creators to expand their businesses could credibly make a case that when tax rates are reduced it takes at least a year or two for the effects to be seen in the unemployment rates, making a year-to-year comparison invalid. Essentially, the possible positive effects of the tax reduction haven’t had a chance to take hold yet in the same year that rates are reduced. The critic could assert that while comparing tax rates to unemployment rates in the same year may show a negative correlation, subsequent years may show a positive correlation once the job creators have had a year or two to put that extra capital back into the system.

For this reason a Pearson correlation was conducted comparing the top marginal tax rates for each year to the unemployment rates the following year for every year from 1948 until 2010 (as of 2012 when the analysis was done, the tax rates for 2011 could not be compared to unemployment rates from 2012 because that data had not been released yet). Put another way, the tax rate from 1948 was compared to the unemployment rate for 1949 and so on until 2010 to account for the possibility that any effect of the tax rates may not appear until the following year. The results again indicated that there was a statistically significant negative correlation between the top marginal tax rate and the unemployment rate the following year, r(63) = -.267, p = .034. Just as in the first analysis, these findings suggest that when top marginal tax rates are low, the unemployment rate the following year tends to be high, and when tax rates are high, the unemployment rate the following year tends to be low.

The question remained as to whether this pattern would hold true if the top marginal tax rates were compared to unemployment rates two years after the fact. In essence, did tax rates appear to influence the strength of the economy two years after those revenues were collected? Based on the initial findings, and to extend the analysis even further, the decision was made to explore the relationship between the top marginal tax rates and the unemployment rates two years, three years, and four years after the fact. When tax rates were compared to unemployment rates two years later, a statistically significant negative correlation again emerged, r(62) = -.259, p = .042. When tax rates were compared to unemployment rates three years later, a statistically significant negative correlation was also revealed, r(61) = -.265, p = .039. For the analysis of four years after the fact, a negative correlation appeared, but in this case it was not statistically significant, r(60) = -.245, p = .059. This final analysis did not meet the criteria for significance for two reasons: First, with each subsequent analysis the sample size was reduced by one year. For instance, for the 2011 tax rates, unemployment figures do not yet exist for two years after that time, so 2011 had to be removed from that analysis and so on for each succeeding analysis.

Likewise, for 2010, unemployment rates do not yet exist for three years after that time. The second and more important reason for the lack of significance in the final analysis is related to the first. With each analysis, when a year was removed for lack of an unemployment statistic to compare it to, the year removed was the next most current one, so the tax data for the years 2011, 2010, 2009, and 2008 were removed with each respective analysis. This was noteworthy because these were all years with historically low tax rates (35%), and when those extremes were removed from the data set the results gravitated towards the historically higher tax rates and the correlation appeared less significant. There is no doubt, however, that when the data for 2012-2017 become available in the coming years and the current low tax rates are compared to unemployment rates for four and five years later from 1948 until the present, there will indeed be a statistically significant negative correlation, just as in the other analyses.

Interpreting the Results

What we see when examining the whole of this data is a consistent pattern: When the top marginal tax rates are compared to unemployment rates for the same year, one year later, two years later, and three years later, nearly identical results emerge. Not only is there a negative relationship in each case, with low tax rates correlating with high unemployment and vice versa, the magnitude of each relationship is nearly identical. So between 1948 and 2011, there appears to be a clear and consistent relationship between top marginal tax rates and the unemployment rate. And since unemployment rates cannot dictate tax rates, any influence must go in the opposite direction, with tax rates influencing the unemployment rates. Because we are dealing with correlations, there is a possibility that a third variable or more variables are also at play, particularly in a dynamic as complex as the U.S. economy. Indeed, it is almost a certainty that other factors are involved. But the unmistakable and highly uniform pattern revealed in the analyses reported here would lead us to believe that the relationship between top marginal tax rates and unemployment is in fact present, even if other factors are also involved.

What we can say with absolute confidence, though, is that there is no evidence here that low tax rates are associated with low unemployment, and by extension, a healthy economy. Similarly, there is no evidence that high tax rates are associated with high unemployment, and by proxy a weak economy. There is simply no empirical basis to make those claims based on this historical data. In fact, everything we see here suggests that just the opposite is true. Low marginal tax rates do not appear to be beneficial to employment rates, and if they are in fact detrimental to employment rates one would be hard pressed to make the case that they are helpful to the economy. In the most basic terms, a healthy economy is one in which the vast majority of citizens who want to work can find that work.

If one were to accept the common contention these days that we must wait until we again have a strong economy before we are able to collect the tax revenues needed to adequately fund public sector services, the data simply does not support that claim. These numbers tell a far different story. They instead suggest that while tax rates remain at historical lows we will continue to have a weak economy and high unemployment. There is no data to suggest that by keeping top marginal tax rates low it will improve the economy or decrease unemployment. For those who insist on low taxes at all costs, it would be worthwhile for them to look at the numbers and realize that pursuing low marginal tax rates, and gutting education and other social services in the process, is not the answer to a weak economy. It may be one of the causes of it, and certainly appears to be a prime factor in the equation. If we continue on the trajectory that we as a country have been on for more than 30 years of demanding lower and lower tax rates in the hopes that it will keep money in our pockets and food on the table, the data tells us we are more likely to have empty pockets and less on the table.

JOSHUA A. CUEVAS Ph.D. is an Educational Psychologist and Assistant Professor, College of Education, North Georgia College and State University.

Federal Reserve Bank of Minneapolis. (2012). The recession and recovery in perspective [data file]. Retrieved from
Kessler, G. (2012, June 12). Spinning the number of teacher layoffs. The Washington Post. Retrieved from
Krugman, P. (2011, December 11). Depression and democracy. The New York Times, pp. A23. Retrieved from
Tax Foundation. (2011). U.S. federal individual income tax rates history, 1913-2011 [data file]. Retrieved from
The Economist (2011, July 29). Recessions compared. The Economist online. Retrieved from
U.S. Department of Labor, Bureau of Labor Statistics. (2011). Labor force statistics from the current population survey [data file]. Retrieved from