In my first post on the “two-income trap” and in my Wall Street Journal column, I took note of the peculiar way in which the authors presented their data, which led them to overlook the crucial role of the rise in tax liabilities between the two periods and the impact that has on the average family’s household finances. When the Warren & Tyagi compare the average family of the 1970s to the 2000s, they present the data on all expenses (mortgage, cars, health insurance) in terms of the change in the actual dollar expenditures between the two periods. The rise in the tax burden, however, is incongruously presented in terms of the percentage of household income dedicated to paying all taxes (federal, state, and local). Thus, the authors state that the percentage of household income paid in taxes increases from “24 percent” of household income to “33 percent” of household income.
As I noted, the difficulty with presenting the data in this manner is that it obscures the underlying dynamic of what is happening in the example. Adding the second worker increases household income by 75 percent–this is actually a greater increase than the expenditures on mortgage, automobiles, and health insurance, all of which increase by less than the income growth of 75 percent. The problem is that total tax obligations over this period increase from about $9000 to about $22,000–an increase of about 140 percent. Thus, assuming that the authors’ argument is theoretically sound (a proposition open to question) it seems clear that the increase in tax obligations is the driving dynamic in their example.
Nonetheless, the authors apparently even confused themselves, as they completely ignore this massive growth in the tax burden on the household budget, even though use the actual dollar values in calculating the “fixed costs” portion of the average family budget. Apparently this idiosyncratic presentation style confused most readers as well, as even though the book was reviewed and featured in a large number of mainstream media outlets, it appears that none of those commentators observed the growth in the tax burden either. A list of reviews of the book, with links, is available here.
Fortunately, the authors presented all of the raw figures necessary to convert the percentages to actual dollar values, which made it possible to figure out that the importance of the growth in the tax burden.
Now I see that in May 2007, Professor Warren provided testimony before the United States Senate Finance Committee on the same topic. At the hearing, Professor Warren presents all of the data in the book in a different format–but the end result is that she actually presents the data on taxes in an even more confusing and idiosyncratic style than in the book, making it even more difficult to understand the underlying dynamic.
In her Senate testimony, Professor Warren no longer actually presents the actual dollar values for the changes in expenditures between the two periods. Instead, she presents the data for everything but taxes in terms of the percentage change in the amount expended by the household on various categories of the household budget. So, for instance, she reports that the average family now spends “32% less” on clothing, “18% less” on food, “52% less” on appliances, etc., than the family of a generation ago. They also spend “76% more” on mortgage payments, “74% more on health care” and “52% more” on automobiles than in the past (the data here is apparently updated from the presentation in the book). She also notes that families are spending more on electronics, such as DVD players, televisions, and computers. As before, she then recalibrates all of these percentages into a figure for “fixed costs” and “discretionary expenditures” that actually uses the actual dollar values, rather than these percentages. Moreover, she never indicates what percentage of the household budget each of these categories comprise, so it is difficult to figure out what impact these percentage changes in isolation have on the overall household budget.
But here’s where it gets confusing. For each of these other expenditures, she present the percentage change in the actual amount of money the household actually pays for each of these obligations. If a similar measure were used for taxes, as noted above, it would indicate a massive increase of about 140% in tax obligations over the relevant period, which would make clear that this increase dwarfs the increase in every other expenditure.
But for the presentation in the change in taxes alone–Professor Warren presents the change in the “average tax rate”–the percentage change in the percentage of household income dedicated to paying taxes. So, under this approach, because tax obligations increase from 24% of household income to 33% of household income–and increase of 9 percentage points–she reports the change in “the average tax rate” between the two periods as being 25%. But had she applied the same methodology to taxes as she does to every other expenditures–i.e., the growth in actual household expenditures on various categories of goods and services, rather than the percentage change in the percentage of the household budget dedicated to a particular category of expenditures–the “apples to apples” comparison in the have been about 140%, not 25%.
If she applied to the other categories of expenditures the same methodology she applied to taxes (the percentage change in the percentage of the budget dedicated to those expenditures), then each would have actually declined or stayed approximately the same as a percentage of household income than previously, so they would be either zero or negative, while taxes would have increased by 25%. Instead, she presents a true apples to oranges comparison with absolutely no resemblance to one other for purposes of comparison.
In fact, this confusion leads to a misreporting of the data in Figure 3 of her testimony, which is labeled “Median Family Spending by Category, Percent Change, 1972-2005.” She does in fact reporting the percent change in “family spending” for each of the other categories, she does not do so for taxes. But what is reported for taxes quite plainly is not the percent change in spending on taxes–it is the percent change in the percentage of the household budget dedicated to paying taxes, a completely different and unrelated number.
I don’t know why all of this is presented in the way it is, but it doesn’t make any sense to me. Presumably Professor Warren understands that the data she reports for taxes is not based on the same methodology as for all expenditures. And I assume that she is aware that the change in the “median family spending” on the category of taxes is actually about 140%, not 25%, such that the number reported in her Figure 3 is incorrect. More fundamentally, I don’t understand why it is thought useful to put any expenditure in terms of the percentage change in the percentage of the household budget dedicated to a given category expenditure. And if that is useful, why would it be useful only for reporting taxes but not for any other category of expenditures?
Perhaps there is some logical reason why it makes sense to present the data in this peculiar and heterogeneous fashion. But if so, it is not obvious to me. As a result, it makes even more difficult to understand what is going on than even the original presentation, which already seemed to have been confusing to most readers (and perhaps even the authors themselves). I just don’t understand why this one obligation–taxes–is consistently presented in a unique fashion that invariably makes it more difficult to understand what is going on with that expenditure from the household budget and thus to understand how the change in the tax burden compares to the changes in the burdens of other expenditures. And in fact, it seems likely that the average Senate Finance Committee member or staffer would be likely to look at this presentation and be misled into concluding that the increase in the tax burden is a relatively small part of the overall change in the household financial burden, when in fact it is the largest change.
Furthermore, unlike the previous presentation of the data, where one could at least replicate what was going based on the information provided (however obscurely), in this iteration the actual dollar values expended on various budget items are never presented, nor are the percentages of the household budget dedicated to certain expenditure categories (which would allow one to back out those values from the total income figures which are presented). In fact, from what I can tell, Professor Warren never even presents in her testimony the 24% and 33% figures for the percentage of the family budget dedicated to taxes, which I had to infer from her book that was where her new 25% figure of the percentage change in the percentage of income dedicated to tax payments is coming from. As a result, it is unclear in her testimony exactly how much the tax obligations rise, but the underlying expenditures on taxes seem to be basically the same as the original data on this score.
By suggesting that the increase in household expenditures on taxes is only 25%, rather than its actual increase in value of about 140%–Professor Warren’s testimony unfortunately leaves the Senate with the impression that the growth in tax obligations is much smaller than the growth in categories such as mortgage and health care expenses, when in fact the growth in taxes is much, much larger and more important. Whatever the rationale for reporting the data in this fashion it appears to have once again confused the logical policy recommendations that follow. Professor Warren recommends five types of policy responses to her version of the “two-income trap,” that range from the affordability of health care and college education, to improvements in education and public schools, and a novel proposal for government “safety regulation for credit products.” All of these may or may not be sensible ideas, but they seem like they’d have a relatively minor impact on this particular problem when compared to the elephant in the room–taxes. This is especially so given that government at all levels could do something about the tax burden much more easily, with greater direct impact, and with fewer unintended consequences than trying to address these more difficult social problems.
Because Professor Warren either does not recognize or for some reason simply chose not to report the massive contribution of increased taxes to the overall household budget crunch, none of her policy recommendations address the dramatically increased tax burden, which as we have seen, is the underlying factor driving the whole two-income (tax) trap. Moreover, it goes without saying that if the average tax burden had increased at the same rate as income during this period (75% instead of 140%), then this would provide a huge amount of money for savings, for college, to pay for a home, or to pay for the other household expenses she enumerates. Or, at least, tax reform seems like it is worthwhile to at least consider in this context.
More generally, I still don’t really understand why this data is presented in such a heterogeneous and confusing manner, especially when it consistently leads to confusion about what it actually demonstrates. Furthermore, this confusion results in policy recommendations that don’t seem to follow from what the data actually show. It seems like it would be more effective to just pick a uniform presentation format and use that for all of the numbers, thereby permitting a more transparent comparison among them.