Thursday, September 24, 2015

Americans Aren't As Poor As Latest Census Poverty Data Suggest

Late last week, the Census Bureau released its latest data on poverty statistics (see full report here). According to the Census Bureau, the poverty rate is at 14.8 percent, or at 46.7 million in poverty. This poverty rate is 2.7 percentage points higher than it was in 2007, which was the year before the Great Recession. Based on these numbers, the current average income is roughly the same as inflation-adjusted income for 1999. Per the Census Bureau's table [below], I also see that the poverty rate has hovered around 15 percent since the War on Poverty in 1965, but I'll leave that one alone for now. Based on these numbers, it seems like the American government has barely made a dent in its poverty rate. 46.7 million people dealing with poverty sounds like an awful lot, but I have to ask whether the number is that high, so at the very least, we can have a contextualized conversation about poverty in the United States.

A good place to start is to see how the Census Bureau is measuring poverty. Looking at the metrics that the Census Bureau uses, the primary ones actually overstate the extent of poverty in this country. Let's take a look at what those are:
  1. Pre-tax income. If the government only provided a minimal, temporary safety net, then I could understand there being a negligible difference between measuring total income pre-tax and post-tax. However, that is not the world in which we live. There are many welfare programs that act as non-cash benefits for families in the United States, which brings me to my second point...
  2. Treatment of non-cash benefits. It is nice to see that unemployment benefits and Social Security are factored into the equation because they have a sizable impact on a family's budget. However, the fact that the Census Bureau does not include non-cash benefits in their figures. The average monthly food stamp benefits in this country is about $125 per month. Medicaid spending, which is health care dollars low-income and/or disabled individuals, is not a small amount. The child tax credit can be up to $1,000. The average Earned Income Tax Credit (EITC) refund was about $2,500 in 2014. Housing subsidies, school lunch subsidies, the list goes on. When we start to total these non-cash, post-tax benefits, they start to add up to a different amount (much like I alluded to back in 2012). If we adjust for post-tax income, much like economist Scott Winship does, households' incomes are actually higher than they ever have been.
  3. Poverty threshold statistics. While it's interesting to see that the Census Bureau has 48 different types of thresholds based on age and family size, does anyone find it peculiar that there is no geographic variation whatsoever? Cost of living varies from state to state, as well as between urban, suburban, and rural areas. The fact that geographic variation is not considered by the Census Bureau should give us reason to pause. 
  4. Inflation. The official poverty thresholds are updated by the Consumer Price Index (CPI-U). The CPI-U overstates inflation in a few ways. The first is substitution bias. If a price of a good, let's say oranges, increases substantially [relative to substitute goods], then consumers are more likely to choose the lower-price alternative. The second issue with the CPI-U is a quality bias. Over time, technological progress tends to increase the longevity and usefulness of a product. As an example, a computer in 1982 had much less capabilities and cost more than it does now. Third, there is a new product bias. Products are not introduced to the CPI's basket of goods until they become more commonplace. This means that the CPI misses the dramatic price drop with new technological products. Overstating inflation does not have implications just for poverty statistics, but with rate of return on investments and real GDP growth figures.

I don't mind having a conversation about the pervasiveness of poverty (particularly in comparison to other countries) and what we can do to mitigate it, but can we at least use statistics that tell a more accurate story so we know can properly diagnose?

No comments:

Post a Comment