Sunday, June 21, 2009

Misleading Uses of Statistics, Chapter 57,942

A conservative interlocutor on a facebook thread challenged me the other day:
...Let's just use numbers without liberal or conservative pundit spin, okay? Here's how things stood at the end of 2006, no commentary:

Economic expansion was in its 74th month. The country had a rolling average of 103,000 job increases for the last three months of 2006. Unemployment stood at 4.7 percent. The household job count, which picks up small businesses, posted a 303,000 average gain.

Worker wages rose 3.8 percent in 2006, a full percentage point ahead of inflation. U.S. productivity surged 6.3 percent in the third quarter, its best pace in four years. Business inflation fell from 3.5 percent a year ago to 1.5 percent.

U.S. household net worth hit a record high of $58.6 trillion, and household wealth increased 43 percent from 2001-2005.

Now give me numbers showing me how "delusional" I am. No spin, just facts.
No problem...

First, I have to say I find your periodization of data bizarre. Aside from anything else, no government spending makes its presence felt in the economy for about a year, regardless whose it is. It just takes a while to get the ball rolling. This was noted in the Obama team's projections when they introduced the stimulus plan, by the way. It is absurd to draw any conclusions about the efficacy of the stimulus plan (or any of the others, for that matter) 5 months into a new administration.

Secondly, your reliance on 2006 data (I'd like to know your sources, by the way--I'm relying on the Bureau of Labor Statistics and several economists, as noted in the links below) is a bit odd, given that Bush was in office for 8 years. I think it's much more objective to compare apples to apples, so I'm comparing 8 years of Clinton to 8 years of Bush. Surely with 8 years of data for each, there can be no question as to the onset of the economic policies of each president or responsibility for same. Sound fair to you? I knew it would...

One last point: I also think it's unreasonable to compare the economic performance of any president in the midst of a major economic crisis (this one's generally considered the worst since the end of the Great Depression) with that of a president not facing a similar situation. Compare Obama in a few years to FDR if you like. Or if you really want to stretch way back and account for lots of data problems, try the panic of 1873 or, even worse, the economic crisis of the early 1820s. But again, apples to apples seems a reasonable approach.

1. Employment Rates:

Type of data: Percent
Age: 16 years and over

1993 199419951996199719981999200020012002200320042005200620072008

The graph and data above show remarkable growth in the employment-population ratio from 1993 (61.4%) through 2000 (64.4%), followed by a sharp drop in employment from 2001 (64.4%) through 2004 (62.4%). This was followed by a modest increase in employment from 2004 (62.4%) through 2007 (62.7), followed by a dive off a cliff from 2007 (62.7%) to present (59.7%) coinciding, of course, with the economic crash.

Incidentally, the reason I use the employment-population ratio instead of raw employment numbers is that, unlike the latter, the former illustrates the extent to which employment rate changes reflect, by lagging, equalling or exceeding population growth rates, actual increases or decreases in the availability of jobs. The economy gained somewhere in the neighborhood of 22 million new jobs under Clinton. I forget offhand how many jobs we lost under Bush, but it was a lot.

2. Wages:

From 1993 to 2001, wages were flat for the first 2 years, followed by net changes in percentage points of .4%, .3%, 2.6%, 2.1%, .9%, .7% and 2.0%. From 2001 to 2007, net changes in wages in percentage points were .6%, 0.0%, -1.0%, -0.4%, 1.8% and -0.7%. By any measure, net increases in wages were greater from 1993-2001, than from 2001-2007.

3. Productivity:

Percentage changes in productivity from 1993 to 2008 were as follows:

1993 199419951996199719981999200020012002200320042005200620072008

In other words, productivity increased almost continually from 1993-2002, then declined consistently until 2008.

4. Business inflation:

Business inflation rose slowly from 1993-2001, increasing faster from 2001-2007, then experiencing a levelling out.

1993 199419951996199719981999200020012002200320042005200620072008

5. Household Net Worth:

As far as household net worth is concerned, let’s disaggregate your statistics. The wealthiest 1% of the population experienced a 40% increase in wealth during the Bush years; everyone else either had level net worth or lost ground. Adjusted for inflation, the real wages of the average American worker has actually decreased slightly since the 1970s. Use of an aggregate household net worth number obscures these differences.

To understand why consider a hypothetical case in which you have 5 income levels: $10,000, $20,000, $30,000, $40,000 and $1,000,000. Assume the number of households at each level is, respectively, 10, 30, 50, 40, 10--a very rough bell curve. Multiplying the number of households at each income level by its income, then adding up the subtotals (140 households, $13,800,000) and dividing the total income by the number of households yields the average income of the whole group: $98,571.43. Now consider another group with the same number of households, but with these 5 income levels: $10,000, $18,000, $29,500, $39,000, $1,500,000. For the sake of simplicity, lets keep the distribution of incomes at each level the same as before. We still have 140 households, but the total income is now $18,675,000--solely because of the increase in income in the highest quintile. The average income per household is now $133,392.90, simply because the highest quintile value is so much higher than it was in the previous example. Did most households experience an increase in income from the first example to the second? Of course not--only those at the highest quintile did, but their increase was so great it pulled up the average for the whole group. The numbers I used were obviously not taken from the U.S. Census, but the phenomenon I've explained in simplistic terms is roughly what has happened to American households since the 1970s--those in the highest income brackets experienced a dramatic increase in income, especially over the past 8 years, while the average working household (I'm referring here to the median, or midpoint value in an index of U.S. incomes) either experienced no change or a slight decline in income during that period.

(Additional source data available at U.S. Department of Commerce website—fee required for access.)


By the way, the contrast turns out to be even more stark than the numbers above indicate. It turns out that, no matter how you slice it, the economy always has done better under Democratic than Republican administrations—with the exception of every 4th year of a GOP president’s term in office (the exception to the exception being, of course, 2008 under Bush).

It’s interesting, by the way, to see how heavily my conservative challenger relies on 2006. Bush was in office for 8 years—why not data for his whole term in office? His economic record was anemic compared to Clinton’s (see charts above), and it was topped off with a global economic collapse, which, contrary to claims from the right, was not caused by Fannie Mae and Freddie Mac, but by wildly irresponsible behavior by mortgage brokers, ill-informed consumers, many of this country’s biggest investment banks, and the insurance company they relied on to hedge their bets via credit default swaps, AIG. (The latter, incidentally, didn’t think it necessary to back those financial instruments with their own money. And of course, they were unregulated (thanks largely to GOP Sen. Phil Gramm), so who was going to notice?) And yes, it’s true, Bush didn’t cause the collapse all by his lonesome, and some Democrats were complicit in the mess, but any honest measure of the economic history of the US from 1980 to present will unavoidably conclude that the GOP was the prime mover in the rush to deregulate, reflexive hostility to government, and the near-religious devotion to the idea that unfettered “free” markets would naturally police themselves. We’ll all be living with the results of that folly for a long time to come.

And people who try to blame Obama for the mess are being disingenuous. Sorry—5 months just isn’t enough time for any president to have a significant effect on the economy. It just takes time for programs to start & for money to flow through the pipeline. I realize my conservative challenger is a libertarian and particularly aggressive in his belief in the sanctity of the marketplace. But fair is fair—and if anyone attacked a conservative or libertarian politician with the intensity, obsessiveness and peculiar one-sidedness he exhibits toward Obama, he’d have no trouble recognizing how unreasonable that sort of behavior really is.

Incidentally, while I can't speak for your sources, not knowing what they are, I don't look to pundits when forming my opinions. Nor do I have one iota of patience for spin, regardless who does the spinning.


Michael J. Hayde said...

Having viewed the Facebook thread in question, I'd say you're the one not comparing apples to apples. The "conservative interlocutor's" choice of 2006 as the demarcation point is due to the fact that, in that year's elections, Democrats assumed the Congressional majority.

In simplistic terms, put Democrats in congressional control and businesses (especially small businesses) pay through the nose, as taxation leads to layoffs and investment initiative goes right out the window.

If, as you say, it takes about a year for government spending to impact the economy, then we're looking at 2007. Thus your own numbers and words ("a dive off a cliff from 2007") for employment, worker wages, and so forth actually support the original thesis.


Sign Sans Signified said...

Well, you certainly did put it simplistically. Do you have one iota of evidence that the Democratic return to control of Congress in 2006 led to (a) higher taxes on small businesses; (b) layoffs; or (c) a reduction in investments? Or would you be substituting a half-baked (albeit sometimes popular) pseudo-theory for actual evidence?

As far as 2007 is concerned, in order to successfully claim, as you do, that the recession was caused by resumption of Democratic control of Congress, you'd have to disprove the alternate explanations--which are very well established. Namely, widespread reliance, primarily by non-bank financial institutions operating in the absence of any regulatory oversight (thanks mostly to the GOP), in unscrupulous mortgage lending practices via complex, risky financial instruments, created a massive economic house of cards which collapsed when overextended mortgage holders became unable to keep up with their mortgage payments. I don't know what you've been reading, but I can suggest a whole list of economists. You might want to test your assumptions--they're wrong.