At CTData, we are committed to building data literacy across the state. Data literacy is the ability to systematically and ethically ask and answer real-world questions with data, which includes collecting and finding data, critiquing and interpreting data, and analyzing and applying data.

In this series, we explore four variables associated with economic prosperity through the lens of different data sources: Income, unemployment, student loan payments, and mortgage interest payments. As you will see, each data source paints a slightly different picture of the state depending on how the question was asked, the data source, and the respondent population. We hope that through this series you will be reminded that we must all continue to be curious about the information we are presented with and work to be critical consumers of data in our everyday lives.

The analyses in this series were completed by our Fall 2019 Wesleyan Intern, Spencer Arnold. We are grateful for his contribution toward and enthusiasm for this project.

Part 1: Income

Connecticut is a state with many facets. At first glance, Connecticut appears well-resourced and prosperous, having long been cited as one of the wealthiest states in the nation. Those of us that work in our local communities know that there is a more nuanced story behind this facade of wealth. Connecticut continues to rank in the top 3 states for the highest income inequality in the nation.

Given that these income statistics are touted regularly in local and national media, we sought to understand how the data varies depending on the source and the extent to which this shapes the story. Being a critical consumer of data is a key step in building data literacy. To explore this question, we analyzed income data from two 2017 public data sources: the U.S. Census Bureau’s American Community Survey and the U.S. Department of Internal Revenue Service’s (IRS) Individual Income Tax Form 1040.

We found that mean household income estimates from the American Community Survey were higher than those reported on IRS Form 1040. We also found that Connecticut has higher mean household incomes than the United States, across both data sources.

Data Sources

American Community Survey

The American Community Survey (ACS) is an ongoing survey product of the U.S. Census Bureau that collects detailed information about the U.S. population. Questions cover many topics, including: educational attainment, employment status, income and earnings, industry and occupation, health insurance coverage, housing characteristics, and demographics. ACS data is widely used by planners, policymakers, and organizations to help communities plan for programs, businesses, and services to meet the needs of their residents.

IRS Form 1040

The U.S. Department of Internal Revenue Services (IRS) collects annual income tax returns from individual taxpayers using Form 1040. This form includes information such as annual income, sources of income, tax credits and deductions, and unemployment compensation. Data from Form 1040 has been historically underutilized in the social sector.

2017 At-a-Glance Findings

The ACS 2017 1-year estimates of 1,356,762 households report a mean annual household income of $107,420. To account for a greater proportion of CT households concentrated in higher income brackets overall, we also explored the median household income. The American Community Survey 2017 1-year estimates report a median household income of $74,168 in Connecticut.

Using Form 1040, we found that, in fiscal year 2017– 1,766,090 income tax returns were filed in Connecticut. Data from these returns suggests that Connecticut taxpayers are wealthier than average. In 2017, Connecticut had the highest average adjusted gross income of all states, with a mean income of $100,564. We were unable to calculate a median income since only aggregate, rather than individual-level, data is made available to the public.

As you can see from these 1-year findings, across the two data sources, there was about a $7,000 difference in 2017 mean annual household income in Connecticut.

Over Time Analysis

When looking at the data over time, we found similar differences for mean income, with Connecticut consistently ranking wealthier than the national average for both the ACS and IRS data sources. Data in the figure represents “real income,” which means that the current dollars as reported on the ACS and IRS Form 1040 were adjusted to account for inflation using 2017 rates.

To see how the state recovered from the recession, we also looked at capital gains reported on IRS Form 1040 from 2007-2017. The chart below displays over the past ten years the percentage of taxpayers who realized a net gain on investments within a particular tax year. Overall, CT mirrors the national trend with declines between 2007 and 2009 and then a gradual increase from 2009 onward. The biggest differences are that Connecticut has a larger percentage of taxpayers who capital gain earnings and Connecticut taxpayers saw steeper losses during the recession years (about 5% in CT versus 3% nationally) but bounced back more quickly, seeing approximately $7,000 more in earnings between 2007-2009 than the national average.

Why are there differences in reported income?

When working with data, we tend to desire clean data sets and findings so, given these differences, our first response might be one of concern. Upon closer examination, it is clear that the differences in reported income between the two sources are not due to errors in the data but rather how the data is collected and reported. For example:

Motivations for reporting accurate income data is different for IRS Form 1040, which is verified and audited by the government, versus the ACS which is a survey and not verified.
Surveys, like the ACS, tend to have higher non-response rates for questions about income (Moore, Stinson, & Welniak, 2000).
The ACS is a survey, which means that it requires respondents to recall their income rather than pull it from formal documentation. People often guesstimate their income since it can be challenging to know whether to report net or gross income, stigma associated with how much people earn, and being asked to report on a time period different from payroll records (Johnson & Moore, 2008).
The ACS has an estimation error (margin of error) because it is a survey and is therefore a sample, whereas IRS Form 1040 captures all the reported earnings and realized investment gains.
IRS Form 1040 only captures income data from tax filers whose gross income is above previously defined thresholds while there are no income thresholds for completing the ACS.
Total income reported on IRS Form 1040 counts more than just wages (it also includes interest earned on investments, dividends, and social security benefits) while the ACS requires people to remember to include these additional earnings in their income (Johnson & Moore, 2008).

Income in Connecticut: The Source Shapes the Story – A Series