Ask 100 users of demographics to name the 5 must-have variables they use in their analyses and income will almost always be listed. Ask those same users to define the term, and you will hear a stunning range of answers. In fairness, if you ask 100 economists to define income, you will get an even larger range. The Merriam-Webster dictionary defines income as “a gain or recurrent benefit usually measured in money that derives from capital or labor.” This is accurate, but likely not overly helpful.
A quick comparison of national scale income estimates is instructive. Income estimates even within the U.S. Department of Commerce (BEA, BLS, Census) differ significantly, and these often look nothing like the income reported by the IRS. A good overview is found in an IRS publication “A Comparison of Income Concepts” (https://www.irs.gov/pub/irs-soi/05henry.pdf).
What constitutes income depends very much on why you are trying to measure it in the first place. Some aspects are easy. The gross salary paid to an employee is clearly income, as is the interest payment from a savings account. One is from labor, the other from capital.
From there, things quickly deteriorate. If I extract money from my IRA, is this income? Or am I merely drawing down on my accumulated capital? What about if I take some of my wage income and invest in an IRA, should that investment amount be deducted from income? The deduction for social security from my paycheck doesn’t diminish my taxable income but yet when I collect my monthly check in retirement, it is taxed as income.
If I am making the choice between two vehicles with the same MRSP, but one comes with a federal electric vehicle subsidy, is that subsidy income? Depends on who you ask.
Consider also the treatment of capital gains and losses. If I invested heavily in a stock and it doubled in value, the gain is taxed as income the year I sell the stock. Yet if I sold the stock for less than I paid for it, my deduction (negative income) is limited to an annual maximum. The census income concept doesn’t even see this transaction.
And what about the company that offers membership in the gym to its employees. Technically, this is a non-cash benefit that is often treated as income. The IRS would tax that benefit but ignore it because of the complexity involved – if I don’t partake, am I receiving a benefit?
If the definition of income was simple, the IRS tax forms would be equally simple: enter your income here and give us our cut. But the tax code is longer than War and Peace (and, not surprisingly, even more boring) and the IRS itself estimated it to be over 3.7 million words in length (https://www.irs.gov/pub/tas/08_tas_arc_msp_1.pdf). Note the use of the word ‘estimated’ – even the IRS can’t figure out the complete contents of its own tax code.
Income is at best a fuzzy concept, and yet most users of demographics will look at the ‘median household income’ line and not give any thought as to either what it means or how it relates to the problem they are trying to solve.
For the purposes of most demographic analysis, the Census Bureau is the default standard. It is not because it is the best estimate or the most accurate. It is simply because no other source attempts to break income down to small areas such as block groups. While the AGS income models use the 1-year and 5-year ACS series as their base, we do supplement these with selected components of the IRS and Bureau of Labor Statistics estimates.
Next week we will explore more closely the Census definition of income, but more importantly, the systematic biases that lurk behind that map of median household income.
Recent Comments