I really, really enjoyed the 2003 book The Two-Income Trap by Elizabeth Warren and Amelia Warren Tyagi. (Yes, dork that I am, I even went so far as to review it.) But from the first day I read it, one part has really bothered me.

The second chapter of Two-Income Trap is entitled "The Over-Consumption Myth." In it, Warren and Tyagi posit that middle-class America's financial woes are not — as some authors would have us believe — largely based upon class-wide frenetic overspending and a vicious need to accumulate "stuff." Rather, Joe and Jane Middleclass are much more often done in by outrageously high home prices, skyrocketing medical costs, and exorbitant tuition bills. (This includes daycare tuition all the way to college tuition.) It isn't that Joe and Jane are spending more on food or clothing or appliances; in fact, the authors posit, in most of these areas middle-class America is spending less than they did in the past. (The authors tend to compare year 2000 consumers with those from 1972/1973, so as to make a sort of "previous generational" comparison.)

As I read and reread Warren and Tyagi's assertions that families are actually spending 44 percent less on major appliances today than a generation ago, for example, or that today's family of four is spending 22 percent less on at-home and restaurant food combined than a generation ago, I always got the nagging feeling that something was not quite right. I couldn't quite get my mental light to flick on, though, until Google dropped me on this page late last week.

First off, that review by Robert Atkinson is way better than mine. But here's what got me going. Check out his next-to-last paragraph:

Perhaps it was this that had been bothering me about Warren and Tyagi's numbers? Well, no, because I'm nowhere near smart enough to pick up on an error like this. All I know is that I had always had the distinct feeling that something was amiss with the "Over-Consumption Myth" numbers in Two-Income Trap. And of course, finance fiend that I am, I wanted to see the dollar signs for myself.

First things first — drag the book back out and flip to Chapter 2. "Throughout our discussion, in this chapter and elsewhere, all figures will be adjusted for the effects of inflation," they write on pages 16 and 17. Follow that statement to its corresponding note at the back of the book, and you read:

"All comparisons of expenditures and income are adjusted for inflation using the Inflation Calculator, U.S. Department of Labor, Bureau of Labor Statistics. Available at www.bls.gov/cpi/home.htm [1/22/2003]."

So Atkinson's review was right. If the authors were using the broad-based CPI (which is what the "Inflation Calculator" above does) to account for price changes in specific items (like food, clothing, and transportation) over the years, then they were using an inflation rate ("deflator") that didn't really apply. Over a few years, the mathematical differences caused by a percentage point or two may not mean much. But compound those rates and rate differences over ten years or more, and you begin to really muck up the picture.

For starters, consider that the CPI index is composed of a "market basket" of several consumer categories which are factored in a manner similar to this:

Food and beverages		16 %		Recreation		6 %
Housing		41 %		Education		3 %
Clothing		4 %		Communication		3 %
Transportation		17 %		Other goods and services		4 %
Medical care		6 %

It seems pretty logical:   If you want to measure consumption of food items across time, you should use the inflation rate for food items only (or as close to it as you can get), and not a rate which reflects a generalized "basket of consumer products." If you want to measure consumption of clothing items across time, you should use the inflation rate for clothing items only. And so on.

  Clothing / Apparel
From Two-Income Trap, page 17: "And yet, when it is all added up, including the Tommy sweatshirts and Ray-Ban sunglasses, the average family of four today [year 2000 family] spends 21 percent less (inflation adjusted) on clothing than a similar family did in the early 1970s."

So what exactly did families spend on clothing in the early 70s? We turn to the U.S. Consumer Expenditure Survey of 1972/1973. We find that under "Apparel and Services," the average urban consumer unit (read: family) spent $732 per year. Using the inflation rate computed by the CPI "Inflation Calculator," we see that to equal the 1973 clothing consumption level, a family in 2000 would need to spend $2,839 per year. And what does the 2000 family really spend? About $2,191. (Find this dataset at CES Standard Tables). That would mean the 2000 family is spending $648 (or 22 percent) less per year than its 1973 counterpart. This, it seems, nears the calculation Warren and Tyagi used for the book. Unfortunately, as Atkinson points out, it falls short of reality.

Rather than using the generic "Inflation Calculator," whose compounding rate applies to nothing in particular, let's crunch the numbers above with a more-correct inflation rate — that of "Consumer Price Index - Apparel." (Find this dataset at Consumer Price Index - Home). That index sported an annual value for 1973 of 64.6; its value for 2000 was 129.6. Over 27 years, that equates to an annual rate of inflation of 2.6 percent. From above, we know that the 1973 family spent $732 per year on clothing. Using our new apparel-only inflation rate, we figure that the 2000 family would have to spend only $1,468 (not $2,839 as above) to equal the 1973 family's consumption. Above, we found that the 2000 family reportedy spent about $2,191 per year on clothing. So, just from using apparel's specific inflation rate rather than an anomalous CPI rate, we go from the 2000 family spending 22 percent less than its 1973 predecessor to actually spending 49 percent more.

Changes the picture a little bit, doesn't it?

  Food
"The average family of four spends more at restaurants than it used to," Warren writes, "but it spends less at the grocery store — a lot less. Families are saving big bucks by skipping the T-bone steaks, buying their cereal at Costco, and opting for generic paper towels and canned vegetables. Those savings more than compensate for all that restaurant eating — so much so that today's family of four is actually spending 22 percent less on food (at-home and restaurant eating combined) than its counterpart of a generation ago."

Our 1973 family spent $1313 per year on food at home and $362 on food away from home, for a total yearly food expenditure of $1675. The CPI "Inflation Calculator" says that the 2000 family would need to spend amounts of $5092 (at home), $1404 (away), and $6496 (total food) to equal the 1973 consumption levels. The 2000 family actually spent a yearly total of $5158 on food, which is in fact $1338 (about 20 percent) less than the 1973 family on a CPI-adjusted basis. Change that to a "Food and Beverage" inflation basis — presumably a more-correct basis — and the total food savings drop to $484 per year, or spending 8.6 percent less than its 1973 counterpart.

Savings of $484 per year equals about $40.33 per month. That merits the term "big bucks?"

Food at home:   The 1973 family spent $1313. Using the "Food and Beverage" inflation basis, the 2000 family would need to spend $4423 to equal the 1973 family. In actuality, they spent just $3021. They spent $1402 less, or 31 percent.

Food away:   The 1973 family spent $362. Using the "Food and Beverage" inflation basis, the 2000 family would need to spend $1219 to equal the 1973 family. In actuality, they spent $2137. They spent $918 more, or 75 percent.

Yep . . . we're spending 31 percent less on at-home (cheaper) food, and 75 percent more for restaurant (expensive) food. That definitely does not hint at over-consumption, does it?

  Household Furnishings & Equipment
As far as measuring inflation, the BLS price index for "Appliances" only dates back to 1997. In order to reach farther back in time and find price inflation data for a comparable category, we can use the price index for "Household furnishings and operations." (This apparently includes appliances.) The "Household furnishings and operations" index value was 51.1 in 1973, and 128.2 in 2000. Thus it suggests an inflation rate of around 3.46 percent annually for these items.

The 1973 family spent $411 yearly on these household items. Adjust that to 2000 dollars with the 3.46 inflation rate, and they would be spending $1031. What did the actual 2000 families spend? About $1549. So our 2000 families spent $518, or 50 percent, more per year than their 1973 cohorts.

One troublesome aspect of measuring inflation which has always plagued economists is the "apples to apples" syndrome. Gauging price changes over time can be done, but we must also try to account for changes in the products themselves. In other words, prices of appliances may rise over time, but the quality and features of those appliances rise (hopefully!) over time, also, so a comparison of year 1973 appliance spending versus year 2000 appliance spending should consider differences in quality levels of the appliances themselves. One can only imagine how difficult it is to account for "quality and feature inflation" in monetary terms.

  Additional Sources
Here's an interesting PDF article on food prices. And a good article from the Federal Reserve, which provides a strong lecture on why we track inflation and why the widely-used CPI may overstate inflation by about one-half a percentage point per year.

Consumer Price Index - All Consumers
Standard Tables for Consumer Expenditures (1984-2002)
USDA:   "Food Expenditures by Families ..."
USDA:   "Food Expenditures and Income"


Michael | September 30, 2004

---

  Home     The Rules     Quotes     Play Great Defense  
  Reading List     Disclaimer  

---