Blog

Another Open Data Quick Dive: Metro Taxes

U.S. Income Tax Return Data, 2013
by Metropolitan Statistical Area
Sources: U.S. Census Bureau; Internal Revenue Service

By Kibichii Chelilim, Data Driven Detroit’s Data Manager and Programmer. His primary responsibilities allegedly include managing D3’s internal data warehouse and applying band-aids to various technical problems. Below is his first D3 blog. The accompanying map above was also his first attempt at data visualization using the D3 (no relation) JavaScript library. Go easy on him.

For many of you, this year’s tax season came and went quietly, and you’ve already spent your tax refund. For others, the days leading up to April 18th were met with a flurry of receipts, stamps, calculators, and tears, only to be rewarded with a mandate to cut a check to the government. And, of course, there are the select few among us who have chosen to blissfully ignore the W-2 omen from early February that signaled that Tax Season Is Coming.

Regardless of how you fared this past month, chances are you’re not a fan of paying taxes. According to a Gallup poll conducted in early April, most Americans think their taxes are too high. This doesn’t make sense for at least one writer, who cites that more than half of Americans don’t owe any income tax at all, and suggests that the survey participants might have been confusing income tax with other taxes, or were basing judgments on paycheck withholding and not considering their refunds. Is there any data available that can validate this claim?

In order to commemorate the holiday and to show how to answer questions like the one above, we decided to focus this Data Dive on one of the many datasets found on the IRS’s website. In the Individual Tax Statistics section of the site, you can find compiled data for almost every line item on tax returns submitted for a given year, broken down by state, county, or zip code. We took county data for the most recent year available (2013, i.e. all tax returns filed in 2014) and aggregated county data up to census metropolitan statistical areas (MSAs). We then joined this data to annual population estimates for these regions for 2013. We will look primarily at how the average income tax rates compare across metro areas in the country, as well as numbers related to the Earned Income Tax Credit and the educator expenses deduction. Particular focus will be given to the Metro Detroit area and how it compares to the metro areas surrounding the following Rust Belt cities: Baltimore, Buffalo, Philadelphia, Chicago, Cincinnati, Cleveland, Milwaukee, Pittsburgh, and St. Louis.

Before we begin number crunching, a few caveats need to be made about the numbers. First off, the goal of this data dive is to show what can be done with open data, so this analysis is neither rigorous nor exhaustive. Our main objective is to show the different ways in which data can answer questions, both by demonstrating what we can show with data as well as highlighting areas where more data are needed.

Regarding the numbers, this dataset comes with a few disclaimers:

  • The data reflect tax returns that were filed in 2014. While most of these returns reflect taxes for 2013, the data potentially include tax returns from 2012 and earlier.
  • The IRS determined county of residence based on the zip code on the tax return, which may not be where the filer actually lived. The IRS also excluded from the dataset returns without valid zip codes.
  • For a given county, if there were less than 20 filers who reported numbers for a particular line item on their return, that return was excluded from the data. If a particular tax return accounted for a certain threshold percentage of a value for a given line item, that return was not included in the data. (The IRS doesn’t disclose this threshold percentage.)

Detailed explanations surrounding these data caveats can be found here.

We’ll begin our data dive by determining what the average effective federal income tax rate is for a given metro area. For this blog, we define the average effective tax rate for a given region as the total income tax owed (before credits are applied) divided by total income reported for all filers. For Detroit, the effective tax rate is 13.4%. Among other Rust Belt metro areas, this does not stand out, but when compared to all metro areas in the country, this number is high. When accounting for state and local taxes, Detroiters reportedly had one of the highest tax burdens among cities in the U.S. What in this dataset could account for Metro Detroit’s high federal income tax rate?

MSA Title2013 PopulationNo. of ReturnsTotal Income (in thousands of dollars)Marginal Tax RateEffective Tax RateTaxable Income (% of Total Income)% of Returns with EITC% of Returns with EITC Refund% of Returns with Educator Expenses Deduction
Detroit4,294,9834,151,960$246,239,978 19.0%13.4%70.6%18.7%16.5%2.2%
Buffalo1,134,1151,098,920$61,226,090 17.5%11.8%67.7%16.1%14.2%3.3%
Milwaukee1,569,6591,534,260$97,506,166 19.0%13.1%69.2%15.7%14.0%2.5%
Cleveland2,064,7252,071,160$119,089,804 18.6%12.7%68.6%17.5%15.4%2.3%
Cincinnati2,137,4062,044,980$125,578,002 18.3%12.7%69.3%16.9%14.9%2.3%
Pittsburgh2,360,8672,377,940$146,793,804 18.6%13.3%71.3%13.3%11.5%2.4%
Baltimore2,770,7382,688,780$189,088,454 19.2%13.1%68.5%15.5%13.7%3.0%
St. Louis2,801,0562,697,720$171,129,836 18.9%13.2%70.2%17.5%15.4%2.7%
Philadelphia6,034,6785,762,560$408,649,620 19.8%14.1%71.3%16.3%14.3%3.2%
Chicago9,537,2899,157,360$646,175,692 20.4%14.5%71.2%17.1%14.8%2.8%

If you look at the most populated metro areas, the data suggest that metro areas with higher populations have higher effective tax rates. To offer a potential explanation, let’s take a look back at the definition of effective tax rate. The formula looks like this:

effective tax rate =   income tax
total income

Your income tax is based on the portion of your income that is taxable. To determine taxable income, the total income—that is, income from wages, investments, etc.—is adjusted by certain deductions. Total income minus these deductions (not including itemized or standard deduction) is your adjusted gross income (AGI). From here, the AGI is further reduced by 1) either the standard deduction or itemized deductions; and 2) the number of exemptions claimed (i.e., roughly the number of people you supported financially that year, including yourself; for 2013, the going rate was $3900/exemption). The end result is your taxable income, and it is from this amount that your income tax is determined. Therefore, if you have more deductions, more exemptions, or an itemized deduction amount greater than the standard deduction ($6100 for single filers in 2013), your taxable income is lower.

Another important measure to consider is the marginal tax rate, which can be best described as the tax rate applied to each additional dollar of taxable income earned. For federal income taxes, marginal tax rates take the form of our venerable income tax brackets, i.e. the applicable tax rate that depends on one’s income. The higher the income, the higher the tax bracket and marginal tax rate. So, if someone who already makes $100K earns an additional dollar, that person will pay more in taxes for that extra dollar than if that dollar were earned while making $30K. And, because that particular tax rate jumps each time one’s income ascends into a higher tax bracket, the relationship between income and marginal tax is not a smooth line.

If taxable income is equal to total income less deductions and exemptions, we can rewrite our equation as:

effective tax rate =   income tax
taxable income+deductions+exemptions

For the fraction on the right-hand side of the equation, if the top half (numerator) of the equation increases while the bottom half (denominator) stays the same or decreases, the entire quantity will increase. If the denominator increases while the numerator stays the same or decreases, the entire quantity will decrease. However, if both increase or decrease at the same time, the change in the quantity is not immediately evident.

If one’s total income increases (i.e., the denominator increases), this could be because either 1) only his/her taxable income increases; 2) only his/her deductions/exemptions increases; or 3) both increase. For situations #1 and #3, taxable income increases. As we described above, this will cause income tax—the numerator—to increase as well. Since both the numerator and the denominator are increasing, it is not clear whether or not the entire quantity increases or decreases. As a result, the effect of an increase in taxable income on the effective income tax rate is not readily apparent.

The situation, however, is much clearer for #2. If only deductions/exemptions increase, this means that taxable income does not change. If taxable income stays the same, income tax stays the same as well. This results in the denominator increasing while the numerator stays constant. This results in an overall increase in the effective tax rate. For example, suppose two individuals make the same salary each year (i.e., their total income is the same). If one takes advantage of more tax deductions/exemptions than the other, then that person’s effective tax rate will definitely be lower than the other person’s. Now consider our country’s metro areas. If people who live in more populated areas are less likely to claim deductions/exemptions, the effective tax rate, as a whole, may tend to be higher on average than that of less populated areas. Therefore, population could account for the apparently greater effective tax rates that more populated metro areas experience.

While a thorough treatment of this requires more than what is stated here, we will take a quick look at the percentage of total income that’s taxable (taxable income / total income). If a relationship between population and deduction/exemption amount exists, then we might see this reflected in the percentage of total income that’s taxable, which we calculated. If you take a look at the map and alternate between the variables for 2013 Population and Taxable Income (% of Total Income), there’s a noticeable shift in color. This casts doubt on a positive relationship existing between population and taxable percentage of income (if the relationship were stronger, there wouldn’t be as much a shift in color). Again, making any conclusive statements on this matter demands a more rigorous investigation. This exercise, however, helped us ask questions about the relationship between where people live and how they make a living. Even more important, however, is the fact that Kibichii made a really cool map all by himself.

Next, we’ll take a quick look at some data related to the Earned Income Tax Credit. Otherwise known as the EITC, this credit can be claimed by low-income workers who meet certain qualifications. Unlike other tax credits, which just reduce your tax liability, the EITC is almost like a negative tax. Under some circumstances, one could actually receive more in a tax refund than they paid out in payroll withholding the previous year (i.e. a negative marginal tax rate). The EITC is designed to encourage working by adjusting the amount received by one’s income earned. Some research, however, suggests that its ability to incentivize work is lessened when taking other taxes and low income benefit programs into account.

We calculated the percentage of filers in 2014 who claimed the EITC credit. Among our Rust Belt metro areas, Metro Detroit takes the top spot at 18.7%. All other Rust Belt cities were at least full percentage point lower. When we calculated the percentage of filers whose EITC amount was greater than the tax they owed, Detroit also ranked first at 16.5%. Interestingly, however, when compared with other metro areas, Detroit ranks right around the middle of the pack for both variables (170th and 172nd out of 381, respectively). In McAllen-Edinburg-Mission and Brownsville-Harlingen, TX, two out of every five filers claimed the credit.

If you look at the EITC percentage variables on the map, there’s a very distinct contrast among northern and southern MSAs. What could explain this difference? If incomes are, on average, lower in southern states, then it could be that a greater percentage of individuals are eligible for the EITC. If the cost of living was also proportionately lower, then these individuals would experience greater benefit from the EITC than their northern counterparts. The EITC benefit also increases with the number of children one has. Also, if someone decreases his or her income tax through claiming other deductions (e.g. mortgage interest deduction), this will increase the likelihood of one’s EITC being greater than his or her income tax. If measures of income, number of children, and/or number of deductions are, on average, different between northern and southern metro areas, this could account for the contrast in the percentage of individuals claiming the EITC and/or having an EITC amount greater than owed tax. Geographic data on cost of living, demographics, and other statistics relevant to tax deductions could offer valuable insight into this question.

Lastly, we’ll take a quick peek at the educator expenses deduction. K-12 teachers who make unreimbursed expenditures for materials in the classroom can deduct up to $250 of these expenses from their total income. Because only K-12 teachers qualify for this deduction, the percentage of filers who claimed it is low (1-5%). In Metro Detroit, 2.2% of filers claimed the deduction, ranking lower than three-fourths of all MSAs. Among Rust Belt cities, Detroit had the lowest percentage. Whether this is interesting to note is hard to say, especially given the small range of percentages. If we were able to compare this with the actual number of working teachers in each metro area for 2013 and/or the average teacher salary, we could stimulate some other interesting questions.

While we only looked at a few variables in isolation, a few calculations and a cool map enabled us to ask some questions about the federal tax regime and its comparative effects on metro areas in the country. If we had available the same data at the state and/or local level, we could gain even greater insight into how taxes affect the welfare of Detroiters, as well as how these effects compare across the nation. We were limited in both time and data as it relates to this blog; nevertheless, we hope that you gained an appreciation for how informative open data sources can be. And, if you happen to know of any open data sources on taxes, or if there’s any tax data you’d like to see made available, let us know at info@datadrivendetroit.org.