A Democratic think tank publishes a chart that would get laughed out of Econ 101.
Ezra Klein recently took up the question of whether the Democratic party is out of ideas. I tend to agree with Yuval Levin that both the Left and the Right haven’t modernized their agendas. But there might be a more pressing problem for Democrats: Are they running out of rigorous, clear-headed data analysts, too?
There are plenty of good and forthright data analysts who are liberal, Democrats, or both. But I believe that an influential set of politically oriented, activist researchers associated with think tanks on the left (and a few in academia) has steered the Democratic party to an overly pessimistic view about how the American middle class is doing. That misapprehension has contributed to the party’s inability to capitalize on the demographic changes that — for now — ought to give them a head start in elections.
For Exhibit A, you can’t do much better than the Center for American Progress report that the group’s president, Neera Tanden, cites in Klein’s article. Tanden says that Democrats need to think more creatively about “the vise of stagnating wages and rising costs” (in Klein’s words). “We did this report,” she says,
that showed that if you look at the prototypical family — double earner, two kids — their wages have stood still since roughly 2000, but their cost of living has gone up by about $10,000 because of things like child care and health care. We have had tax policy that has ameliorated that challenge by about $5,000. But they still have $5,000 less than they did before. So you can see why they’re getting kind of irritated.
Klein includes a chart from the report in his piece showing that “median income for all families” fell 8 percent from 2000 to 2012 while rent, medical care, child care, and higher education have grown more expensive.
Except this “vise” is invented. (The report actually indicates that the median income of Tanden’s double-earner-two-kid family rose slightly, but we’ll let that slide and focus on the median for all families shown in the chart.) There’s an amateur-hour mistake in this chart, which is the basis for the CAP report, which in turn is the sort of analysis driving the thinking of CAP’s (and therefore Democrats’) economic agenda. The chart shows that median household income has declined by 8 percent — but that’s the change after adjusting for the increase in the cost of living. It makes no sense to point out that income has decreased by 8 percent while the cost of living has increased, because the increased cost of living is already included in that 8 percent figure. (Update: See note below.) Tanden’s lament about a family getting hit from both sides is based on double counting.
If you do not adjust for the increase in the cost of living and instead look at nominal income, it rose by 23 percent. In other words, while middle-class income rose by 23 percent, the cost of living rose by 33 percent, so in the end real median income was lower.
Now to be clear, an 8 percent decline in cost-of-living-adjusted income is not a good trend. But it’s not nearly as bad as the same decline on top of a $10,000 increase in the cost of living, which CAP mistakenly thinks is what happened.
And in fact, that 8 percent decline is itself highly suspect. The cost-of-living adjustment used by CAP (and, to be sure, the Census Bureau) overstates the rise in the cost of living because it accounts incompletely for consumers’ ability to substitute one kind of purchase for another when something gets too expensive — for instance, if a cold snap causes orange-juice prices to rise, you can buy apple juice instead. Another adjustment — the “personal consumption expenditures deflator,” or PCE — fully accounts for substitution (or tries to), but even that adjustment is thought, by a lot of economists across the political spectrum, to overstate price increases. If you adjust the nominal Census Bureau figures for the rising cost of living using the PCE, the drop in median family income from 2000 to 2012 falls to 4 percent.
Then there’s the fact that the Census Bureau’s income measure does not count income from four of the five largest safety-net programs we have (Medicare, Medicaid, food stamps, and the Earned Income Tax Credit). Nor does it count as having any value health-insurance costs paid by employers, a growing share of Americans’ compensation. The Census Bureau measure also ignores tax cuts and doesn’t count as income the part of payroll taxes paid for by employers, which economists agree effectively comes out of employee paychecks.
The Congressional Budget Office publishes estimates that most closely approximate an ideal income measure that would include all these things (and also use the PCE deflator). It finds that from 2000 to 2011 (the most recent year available), median household income actually rose, by 13 percent. (Household and family incomes rose by similar amounts in the Census Bureau data, so the fact that the CBO figures are for households is inconsequential.)
“Factivists” on the left immediately respond to these sorts of estimates by saying that government redistribution is the only thing propping up incomes, an argument I’ve dealt with more times than I care to count. Briefly, it’s not possible to look meaningfully at median income before taxes and transfers if one includes retirees (who depend heavily on Social Security and Medicare), because the number of retirees in the middle class has steadily grown, a trend that pulls pre-tax-and-transfer income downward. But it’s possible in the CBO data to take a rough look at the pre-tax-and-transfer income of households with a nonelderly head. When I look at the average income before taxes and transfers for nonelderly households in the middle fifth of income (not, unfortunately, the middle fifth of nonelderly income), I find a decline of 2 percent from 2000 to 2011, which very well may have been recouped by 2012.
Furthermore, comparing 2012 (or 2011) to 2000 compares a recovery year with a peak in the business cycle. If one compares 2007 (also a peak) with 2000, median household income after taxes and transfers rose by 14 percent, and pre-tax-and-transfer income of middle-class, nonelderly households rose 7 percent.
These gains are small relative to middle-class income growth in the 1950s and 1960s, or to more recent growth in income at the top end of the scale, but they still amount to real gains. The CBO figures indicate that the overall post-tax-and-transfer gain — after accounting for the rise in the cost of living – was $7,700 per household from 2000 to 2011. A typical middle class household in 2011 would not have been indifferent between that and their 2000 income.
The slowdown in income growth is hardly unique to the U.S. and its distinctive political economy; middle-class income productivity growth has slowed throughout the developed world. But the Democratic party has no room for anyone who believes that the American middle class is doing remarkably well by any kind of world-historical standard and that we should instead focus on helping the poor and expanding their upward mobility. Trust me — I know this from experience. Instead, it has relied on bad data analysis and concluded not only that the middle class is struggling mightily but that income inequality is to blame.
This sort of perspective leads to health-care-reform legislation that incrementally expands coverage even as it disrupts a health-care system that strong majorities of Americans liked (and pretends to control costs via unrealistic provider cuts that will never occur). It means the Democratic party promises to retain senior entitlement programs in their current form, or even expand them, without any honesty about the degree of middle-class tax increases that would be involved (which would be the equivalent in political courage of Representative Paul Ryan’s past proposals to reform Social Security and Medicare without increasing taxes). In short, it misleads Democrats into the sort of agenda that has failed to resonate broadly with the electorate.
To be sure, the Republican agenda is not wildly popular with voters either. But given the nature and scope of our economic challenges – as revealed by rigorous data analysis — it is not unreasonable to think that appealing to the soft-heartedness of conservatives is likely to be more successful than appealing to the hard-headedness of liberals, if the goal is to actually help those truly in need in a practical, responsible way.
— Scott Winship is the Walter B. Wriston Fellow at the Manhattan Institute.
UPDATE: Embarrassingly, I misinterpreted the estimates in the chart shown on the first page of this piece as the change in prices for each category of expenditure when they actually show that change divided by an overall cost-of-living adjustment. In other words, the increases shown are relative to the overall increase in the cost of living — each grew more than other types of expenditures. Therefore, the chart does not double count. If nominal income growth (not adjusted for the cost of living) were compared with the change in prices in these expenditure categories, it would show that the prices have grown significantly faster than nominal incomes (see this tweet from CAP). I thank CAP’s economy team for pointing out my error, which I regret, and I apologize for my misinterpretation.
That said, Tanden does double-count when she says in the quote above, “their wages have stood still since roughly 2000, but their cost of living has gone up by about $10,000 because of things like child care and health care.” The cost of living adjustments in the chart from which the $10,000 figure comes do not adjust, say, child-care costs for changes in the cost of child care relative to overall changes in the cost of living — they simply adjust all costs by the overall change in the cost of living. That means this statement is rightly criticized in the way my earlier accusation of double-counting was not. Income has only “stood still” in real, cost-of-living-adjusted terms, so to then cite an increase in the cost of living as an additional problem is effectively double-counting.
And CAP similarly double-counts in the report when the author writes of double-earner two-kid families, “Put another way, in 12 years, this household’s income was stagnant — rising by less than 1 percent — while basic pillars of middle-class security rose by more than 30 percent.” It’s the same problem as with the Tanden statement — income was stagnant after adjusting for the cost of living (i.e., after the rise in “basic pillars of middle-class security” has been taken into account).
What’s more, it is expenditures that rose by more than 30 percent, not prices. Families could simply have chosen to consume more child care, for instance, rather than simply paying more for the same amount and quality of child care. Or they could have bought larger, better homes. Spending more on a category of consumption need not only reflect the fact that costs have risen.
More importantly, CAP’s report — by highlighting a few types of purchases with above-average price increases — still is misleading (not necessarily intentionally). When some price increases are higher than average, others are lower than average. That is why cost-of-living adjustment focuses on changes in the overall price of a “consumption basket” that the representative consumer purchases. Everything you need to know about a possible “middle-class squeeze” is embodied in the trend in real income, which has been adjusted not just for increases in the price of child care, rent, or health care but for the increase in all prices. CAP can argue that the consumption basket of a representative consumer and changes in it are not appropriate for looking at dual-earner two-child families, but that requires some evidence.
I’ll be writing more about other problematic aspects of the CAP report next week.