Technical details for my analysis of 2014 poverty data

I use eight categories of people to break down the poverty data. They are defined as follows:

  1. Children — Ages 0-17.
  2. Elderly — Ages 65+.
  3. Disabled — Either 1) had one of the serious disabilities the Census tracks (PRDISFLG = 1), 2) was out of the labor force for the entire year because of an illness/disability, or 3) was out of the labor force for part of the year because of an illness/disability.
  4. Student — Either 1) was out of the labor force for the entire year because of school, or 2) was out of the labor force for part of the year because of school.
  5. Carers — Either 1) was out of the labor force for the entire year because they were taking care of home/family, or 2) was out of the labor force for part of the year because they were taking care of home/family.
  6. Unemployed — Either 1) was out of the labor force for the entire year because they could not find work, 2) was out of the labor force for part of the year because they could not find work, or 3) was in the labor force but officially unemployed for part of the year.
  7. Fully Employed — Worked 50+ weeks during the year.
  8. Other — Everyone else.

Some people fall into more than one categories. Those people are assigned to the first category they qualify for in the numerical order above.

Here is exactly how the Census supplemental poverty calculation is derived

Previously I wrote the smashing hit “Here is exactly how the Census poverty calculation is derived.” It explains in great detail how the Official Poverty Metric (OPM) works. Here, I do the same thing for the Supplemental Poverty Metric (SPM).

All poverty metrics require some way of drawing lines around sets of people and grouping them together into a unit (also sometimes called a “family,” though that can be misleading). The members of each unit are assumed to share their incomes with one another. Based on that assumption, either everyone in the unit is in poverty or none are, depending on whether the collective income of those in the unit is above the poverty line or below it.

The SPM defines a unit as all related individuals who live in the same household plus any unrelated children cared for by the family plus any cohabitors and their children.

Poverty Line
The poverty line for each SPM unit is derived accordingly. First, get the 5-year moving average of the 33rd percentile of expenditures on food, clothing, shelter, and utilities (FCSU) for households with two children in them. Second, multiply that sum by 1.2. Third, adjust that sum for each unit according to 1) the unit’s size and composition (how many children and adults), 2) whether the unit lives in a rental, a home with a mortgage on it, or a home without a mortgage on it, and 3) geographical differences in costs.

Out of this complicated muck, each unit will have a specific poverty line. If the collective income of the unit is less than that line, then the unit and everyone in it is in poverty.

The income for each unit is arrived at by adding together a list of positive resources and subtracting from it a list of negative resources.

Positive Resources:

  1. Cash Income (any source whether job, tax credits, or benefits like Social Security)
  2. Food Stamps
  3. Free/Subsidized School Lunch
  4. Housing Subsidies (e.g. Section Eight)
  5. Energy Subsidies (e.g. LIHEAP)
  6. Women, Infants, and Children Program

Negative Resources:

  1. Federal Income Tax
  2. Federal Payroll Tax (aka FICA)
  3. State Taxes
  4. Child Support Paid (by the unit to someone else)
  5. Medical out-of-pocket expenses
  6. Work expenses
  7. Child care expenses

(Note that work and child care expenses are capped at a certain level such that expenses over that cap are not counted against the unit’s income.)

For each unit, you just add up all the positive resources and subtract all the negative resources to get the income. Then, you take that income and see if it is above or below the poverty line for that unit. If it is below the line, then that unit and all its members are in poverty.

There seems to be a good deal of consensus among experts on this topic that this is a better poverty measure. Unlike the OPM, the SPM includes in-kind benefits that goes towards food, clothing, shelter and utilities (like food stamps, housing subsidies, energy subsidies, WIC, and so on). I am inclined to agree that this is better, as a relative matter, than the OPM.

There are still difficulties with it though. The metric assumes that in-kind benefits are equal to their cash equivalent. So a unit that had $500 in food stamps is deemed to have received $500 of income. But food stamps and other in-kind benefits are not cash and, in at least some cases, are not worth their cash equivalent to the unit that receives them.

If a unit had intended to spend cash on the in-kind benefits already, then the in-kind benefits free up that cash and are worth to the family exactly their cash equivalent. But if the family had not intended to spend cash that way, the benefits are worth less to them than the cash. People who already have ways of getting their food needs met are not made $300 better off when they get $300 of food stamps. That food stamps are occasionally sold for others to use at a discount from their face value strongly suggests that their face value is not the same as the value their recipients get out of them.

So, while it is good that in-kind benefits are being accounted for, counting them as cash equivalents likely overstates their actual value to the units that receive them.

Dumb Graph Shows Nothing

This graph purports to show an increase in regulation activity in the United States:

The Federal Register is the place where administrative agencies put notices of proposed rulemaking (NPR). The assumption underlying the graph is that more pages in the Federal Register means that there are more proposed rules and therefore more regulatory activity. But that is only true if the page length of a given NPR has held constant over time. It has not.

The required content for an NPR that is placed in the Federal Register is governed by the Administrative Procedure Act of 1946. But over time, courts took it upon themselves to ramp up the APA’s requirements for an NPR. Richard Stewart explains in the Harvard Law Review in 1978:

In order to develop a more adequate basis for effective review, lower courts began to require administrators to disclose the factual underpinnings of proposed rules and to respond in detail to commenting parties’ factual and analytical criticisms of the proposed rules and of the supporting materials mustered by the agency. In response to these rulings, agency lawyers sought to bolster the agency’s position by elaborate documentation, while respondents and intervenors submitted contrary documentation which they themselves developed or obtained from agency files through Freedom of Information Act litigation. These various documents provided an elaborate record for judicial review, and courts required agencies to provide a detailed explanation for final adoption of rules by reference to this record.

To survive increasing levels judicial scrutiny, it became necessary for administrative agencies to disclose in their NPRs all of the important and relevant information that underlies their proposed rules. This made the length of NPRs longer, which also increased the number of pages in the Federal Register. Using Federal Register pages to measure the level of regulatory activity therefore makes no sense whatsoever. More pages does not reflect more regulations. It reflects longer explanations and justifications for regulations, which courts forced administrative agencies to provide right around the time you see page length spikes.

Although counter-intuitive, the ballooning page length of the Federal Register is a consequence of it becoming much harder to make a regulation. It is a consequence of the courts clamping down on regulatory agencies and forcing them to do more research, explanation, and justification to avoid having their regulations struck down in judicial review.