20 August 2012

Welfare Cliffs

Jamul Blog | Welfare Cliffs...

The example on the chart below starts with a single mom earning (through a job) $29,000 a year. She also qualifies for several kinds of welfare payments, which (added to her wages) brings her net income and benefits to over $57,000 a year. Now consider what happens if she gets a raise to $32,000 a year in wages: the welfare payments she qualifies for drop steeply (the “welfare cliff”), and her total income and benefits drop to about $50,000 a year. So...she gets a raise of $3,000 a year, but gets a total income cut of over $7,000 a year. If you were in that single mom's place, would you ask for a raise?

The example in the chart below illustrates a slightly different point. That same single mom would have to get a raise from $29,000 to over $69,000 before she saw higher net income and benefits.

(0) This is a great chart format for this data. These should be used more widely. I'd like to see similar charts for different types of people (single mother; married with two children; etc.). That would be a good way of helping to cut through the bogus descriptions politicians give for their tax/welfare policies.

(1) I still think a lot of the "cliffs" — marginal tax rates of >100% — could be eliminated if legislators stopped using heaviside step functions. Such thresholding makes systems, from image processing to credit monitoring to tax law, more susceptible to noise. Linear — or better yet sigmoidal — phase outs should be used.

Yes, this makes things more mathematically complicated, but the tax or benefits systems are already hellishly Byzantine, so I count this as only a small incremental complication. Either design a worksheet to guide people through the calculations, or compile tables of pre-calculated values, or *gasp* get with the 21st century and put up a web app to do it for people.

(2) In the following, I'm implicitly referring to the withdrawal of a benefit as the equivalent to a tax. Philosophically I don't like doing this, but I don't know of other nomenclature which will make this easy to discuss and I'm in a hurry, so I'm going to lump together actual taxes and withdrawal of benefits under the heading "taxes." (See postscript for more.)

(3) Even with more gradual phase-outs and strong coordination between programs, I've concluded there's just no way to avoid very high marginal tax rates on low income people without reducing benefits.

Note that a horizontal line implies a 100% tax rate. Lowering marginal rates is equivalent to increasing the slope of the curve in this chart. Geometrically, there's only two ways to do that: lower the curve on the left side or raise it on the right side. The first means lower benefits for the poor, the latter means higher benefits for the middle class, which sure as hell better be DOA. It's a non-starter fiscally before you even get into moral considerations.

I wish there was another option, but there's no way to reconcile giving someone with no market income $40,000+ of benefits with them having low marginal tax rates. If you give someone that much money, they're going to have to experience high marginal taxes (and lowered incentives to work) until their market income is well north of $40k.

(4) This is not a normative argument against large redistributive programs, it's simply a description of what must follow from having such programs.

PS Greg Mankiw gives some figures on federal taxes as a percentage of income:
Because transfer payments are, in effect, the opposite of taxes, it makes sense to look not just at taxes paid, but at taxes paid minus transfers received. For 2009, the most recent year available, here are taxes less transfers as a percentage of market income (income that households earned from their work and savings):
Bottom quintile: -301 percent
Second quintile: -42 percent
Middle quintile: -5 percent
Fourth quintile: 10 percent
Highest quintile: 22 percent
Top one percent: 28 percent
The negative 301 percent means that a typical family in the bottom quintile receives about $3 in transfer payments for every dollar earned.

The most surprising fact to me was that the effective tax rate is negative for the middle quintile. According to the CBO data, this number was +14 percent in 1979 (when the data begin) and remained positive through 2007.
This, by the way, is why moving the slope of the curve in the first chart up on the right is not an option.

Here's a chart of similar data from Steven Landsburg for Obama's tax plans (blue) and Romney's (Red):

I bring this up because I have seen Blue Team partisans argue that Obama wants to make those three blue bars on the left lower, and that means he's cutting taxes. That's bullshit prevarication, and they have to know it. It's only cutting taxes to lower one of those bars if it's already positive. Otherwise you're increasing spending, not cutting taxes. The fact that we can't even get people to agree on what's a tax and what's spending is a Very Bad Sign.


  1. Great post - thanks for passing all this along. That first chart alone is worth its weight in gold.