Tuesday, May 08, 2007

Fact Checking Elizabeth Warren

Elizabeth Warren is Gottlieb professor of law and faculty director of the Judicial Education Program at Harvard University. Her daughter and she are authors of The Two Income Trap. She wrote an article for the Harvard Magazine called The Middle Class on the Precipice which is based in part upon another article titled "Rewritting the Rules: Families, Money and Risk." In this article, she makes a few factual claims that I have attempted to verify. I will state up front that I had a difficult time verifying her sources, specifically because she does not list them in detail in these articles beyond "Source: US Census Bureau" and "Updated from sources cited in The Two-Income Trap". That being said, I think she makes some inappropriate data comparisons that render her conclusions in doubt. Throughout, I provide my backup data sources and analysis. Note: all data in this article is 2005 dollars converted using the Census Bureau's CPI-U-RS.

Issue 1: The inappropriate period comparisons median income for full-time males across time periods. Consider these two paragraphs from the two web articles:
"Today the median income for a fully employed male is $41,670 per year (all numbers are inflation-adjusted to 2004 dollars)—nearly $800 less than his counterpart of a generation ago. The only real increase in wages for a family has come from the second paycheck earned by a working mother. With both adults in the workforce full-time, the family’s combined income is $73,770—a whopping 75 percent higher than the median household income in the early 1970s."(1)
"Today a fully employed male earns $41,670 per year. After adjusting for inflation, that is nearly $800 less than his counterpart of a generation ago. The only real increase in wages for a family has been the second paycheck added by a working mother. With both adults in the workforce full-time, the family’s combined income is $73,770—a whopping 75 percent higher than the household income for the family in the early 1970s."(2)
Source (3)

Please notice a few things. In article 1, it is clear the discussion of a fully employed male's income is the median figure. It is unclear in the 2nd paragraph if the figure is the mean or median. This is important, because it is entirely unclear if the family’s combined income figure is the mean or median. If both are the same (median), then the comparison would be valid. If not, then this is an inappropriate comparison. I had a difficult time finding data on this until I found a Census report. Scroll to page 45 of the PDF (listed as page 38) and you will find data current as of 2005. One will note that in 2005, the fully-employed male earned a median wage of $41,386 versus $39,036 in 1970 or a full $2,350 more. In order to find a time period in which a fully employed male made $800 less in 2004 ($42,160), one would have the cherry-pick the earning figure from 1987 ($42,919) or from 1978 ($42,877). Since Warren does not list the year in which she makes the comparison to 2004 earnings, it is difficult to determine if she intentionally chose peak earnings for the comparison. This is a small deception, but one that is frequently performed by picking convenient starting and ending time periods for which to calculate comparisons.

Issue 2: The comparison of two-income earning families to all families. I attempted to replicate Warren's claim of $73,770 income which was calculated as a 73% increase. According to the Census Bureau, the median income for families with both man and woman in the labor force was $79,424 in 2004 (5). In order for it to be 73% higher than levels in the early 1970's, the figure would have to be $45,909. In 1970, the median income for two-income families was $53,447 (5). This equates to a 49% increase.

The only possible explanation is that Warren wishes to compare two-income families in 2004 with all married couples of the 1970's. True enough, married couples of all types had a median income of $45,785 (5) in 1970, but all married couples had a median income of $65,754 (5) in 2004 or an increase of 44%. This is clearly seen on her graph. I find it inappropriate to compare two different demographic groups and not identify them as such.

Issue 3: The use of the median wages of all males instead of the more significant married male. It is unclear to me why Warren chooses to use less specific data (all males) instead of data collected on married males. Lumping in single men who statistically make less money and, more significantly, do not head the majority of American families, diminishes the effect of the rise in married males' median income. In 1975 (the first year for which the census reports data), a married male working full time made $44,401 (6). By 2005, the median income had risen to $50,524 (7) - not the flat income as reported by Warren. This is a significant statistical deception as the fully employed single male's only increased from $28,976 in 1975 (5) to $30,015 in 2005 (6).

Source: US Census Data, Income Studies of Each Year in Notes. 1975 is the first year where income by marital status was available.

Issue 4: False attribution of the rise in family median income to the entry of women into the work force. She does so here:
The only real increase in wages for a family has come from the second paycheck earned by a working mother.
Of course, it has been demonstrated that the fully employed, married male's median income was anything but flat. I submit there are three other possible explanations in the rise in the family's median income:
  1. The percentage increase in married women working full-time
  2. The percentage increase in married men working full-time
  3. The increase in the fully-employed married female's median wage
This graph demonstrates that 33% of the married women worked full-time in 1975 (6). That increased to 40% by 2005 (7). Over the same time frame, the full-labor participation of married men increased from 63%(6) to 66% (7). So both numbers one and two above demonstratively played a role in the family's median wage increase.

The fully-employed married woman's wage also increase from $24,189 (6) to $35,267 (7) between 1975 and 2005 or a whopping $11,087. Consider this table:

I would submit that the significant increase in households led by single moms and dads has negatively affected family income trends. An increase of 9 percentage points for women-only head of households and 5 percentage points for men-only indicates a major demographic shift unaccounted for by Prof. Warren.

Warren also obliquely implies that hardly any women worked back in the 1970's.
This “added-worker effect” could buttress the safety net offered by unemployment insurance or disability insurance to help families weather bad times. But today, a disruption to family fortunes can no longer be made up with extra income from an otherwise-stay-at-home partner.

But today, a disruption to family fortunes can no longer be made up with extra income from an otherwise-stay-at-home partner.
As the data clearly indicates, 33% of married women already worked full-time by 1975. That only increase 7 percentage points by 2005, demonstrating that there are still 60% of married women that can either increase their hours or join the labor force should their partner falter.

Issue 5: The exclusion of benefits from the wage calculation. Throughout the comparison of the families of today with the families of the past, the discussion of the increasing role of benefits eludes Prof. Warren. According to the Bureau of Labor Statistics, the average benefits in 1986 (the first year the data is available) was $6.08 an hour or $12,167 for someone working a typical 2000 hour year. By 2005, that had risen to $7.09 or $14,180. That is a 16.5% increase over that time period. Since the data is the average, not median, I will not inappropriately compare median wages to average benefits increases. One, however, can draw the conclusion that it is inappropriate to discuss employee compensation without including benefits.

In conclusion, it is my belief that these documented factual errors demonstrate the flawed analysis methodology of the Two Income Trap mythology. The American family certainly is not on the edge of the cliff waiting to fall off. These errors and inappropriate comparisons render the two-income trap analysis invalid or severely damaged.

(1) The Middle Class on the Precipice
(2) Rewritting the Rules: Families, Money and Risk
(3) Rewritting the Rules: Families, Money and Risk
(4) Income, Poverty, and Health Insurance Coverage in the United States: 2005, P38 (Page 45 of PDF file)
(5)Historical Income Tables - Families
(6)Money Income in 1975 of Families of Persons in the United States P181
(7) Detailed Income Tabulations from the CPS Median income, married male, spouse present used as data for all married males not available in previous comparison periods. Note that the vast majority of married males had spouses present so this is not a significant data departure.

Data for individual years is pulled from the following tables.
1988 (Not available)
1993 (not available)
1994 Men, Women
1995 Men, Women
1996 Men, Women
1997 Men, Women
1998 Men, Women
1999 Men, Women
2000 Men, Women
2001 Men, Women
2002 Men, Women
2003 Men, Women
2004 Men, Women
2005 Men, Women


At Thursday, May 10, 2007 9:39:00 PM, Anonymous Anonymous said...

You have too much time on your hands.... lol!

-- ADS


Post a Comment

Links to this post:

Create a Link

<< Home