Table of Contents

New GDP Estimates GDP Odds and Ends Grapevine The Evidence File

---

The next day I retrieved a transcript of the briefing from the White House's Internet page. As you read the transcript below, keep in mind that the audience is the White House press corps, a group for the most part statistically challenged, though probably no more so than principles of economics students.

Stiglitz: The [growth] numbers look low mainly because we've switched the accounting framework, the way we calculate the numbers-this is the new chain weighted numbers.
Question: What does that mean, chain-weighted?
Stiglitz: Well, the way it used to be -- let me explain how it used to be. It used to be that to calculate the GDP, you used the relative prices of a particular year -- that's a fixed weight-used 1987. And what happens is over time those relative weights that you associate with different goods get out of line. And there are periodic revisions -- 1987 -- then we were going to go to 1992 this year. But instead of doing this periodically, which causes lots of disruptions, what we're doing now is having what's called chain-weighted. So every -- you continually change the weights as time moves on. And when you do that, you get lower growth numbers. The main reason that numbers look lower than they did before was because of computers.
Question: Because of computers?
Stiglitz: Yes. The computer prices that were used were the '87 prices, and the computer prices have come down a lot. So when you bought a computer, they used the prices that computer would have bought if you could have bought it in 1987. So these are technical issues, but they do have a big effect on how you estimate GDP.

As most know, for decades the Bureau of Economic Analysis (BEA) has estimated real GDP using a fixed-weight set of prices. Output has been weighted using prices in some base year-most recently 1987, so real GDP has been reported in "constant 1987 dollars." The main advantage of the fixed-weight system is its simplicity in calculating and reporting, not to mention ease of explanation in the classroom.

Using prices in a single base year yields accurate measures of real GDP as long as the year considered is close to the base year. But over time, relative prices in fact change and this change distorts the measure of output using fixed weights. Most notably, computer prices have fallen an average of 13% per year since 1982. As a result, real GDP estimates based on 1987 prices give computers too little weight before that date and too much weight thereafter. What's more, people tend to increase purchases of relatively cheaper goods and cut back on goods that become more expensive. This switching distorts even more the fixed-weight measure of output.

After delays caused by two federal government shutdown and fierce snowstorms, BEA has finally released details of its comprehensive revision of the national income and product account (NIPA). The most important change is moving from a fixed-weighted to a chain-weighted system. Annual changes in real GDP and in prices are "chain-type" measures based on the so-called "Fisher Ideal" formula that incorporates weights from two adjacent periods. The chain-type weighting system adjusts the weights more or less continuously, getting rid of much of the bias caused by a fixed system. BEA has adopted geometric averaging as the new method for calculating and reporting real GDP and its components. The geometric mean is found by multiplying the quantity index in, say, 1995 based on 1994 prices by the quantity index in 1995 based on 1995 prices, then taking the square root of that product.

Even though a chain-type index adjusts the weights from year to year, any index, by definition, must be anchored in some base year-that is, any index must answer the question "compared to what?" The BEA sets its quantity and price indexes equal to 100 in 1992, which then becomes the base year. Why 1992? The indexes are based in 1992 because that is the most recent year for which current-dollar estimates will not be subject to revision until the next comprehensive revision (in about five years from now).

In addition to calculating index numbers for quantities and prices, BEA has also decided to provide dollar-denominated measures of real GDP and its components, designated "chained (1992) dollars," a term that seems ungainly now but will likely become as familiar as "constant 1987 dollars." These chain-weighted estimates are computed by multiplying the 1992 current-dollar value of GDP (or a GDP component) by the corresponding quantity index number for the year in question. For example, the current-dollar GDP in 1992 is $6,244.4 billion. The output index increased from 100.0 in 1992 to 105.8 in 1994, so the chained (1992) dollar estimate of real GDP in 1994 equals $6,244.4 billion times 1.058, which is $6,604.2 billion (except for a tiny difference because the index, as reported, is rounded).

Because the new formula uses the geometric mean of weights and because each component of GDP is estimated separately, the chained (1992) dollar estimates for the detailed GDP components do not add up to the chained (1992) dollar estimates of GDP. Consequently, BEA reports a residual, which is the difference between real GDP and the sum of component estimates. For years close to 1992, this residual is tiny (for example, for 1990 and 1993 the residual is only 0.01% of GDP). The residual grows as the year moves further from 1992 (for example, in 1982 it was 1.42% of GDP).

The most important change (other than moving to chain weights) is an improved methodology for estimating depreciation -- the consumption of fixed capital. The change replaces straight-line depreciation patterns with new estimates based on the prices of used equipment and structures in resale markets. The total revision of depreciation is downward, with greater variability across different types of assets. This change affects current-dollar estimates for all years. Another major change is a redefinition of "government purchases" to "government consumption expenditures and gross investment." The new component now includes the consumption of government fixed capital.

Compared to previously published figures, GDP estimates in current dollars have been revised up for all years; the average upward revision is 2.8%. The increase in current dollar estimates is due almost entirely to the new definition of government spending and an upward revision of personal consumption expenditures. The recognition of government investment raises GDP by including as expenditures the services of government fixed assets, measured as depreciation, or consumption of fixed capital.

Between 1959 and 1994, the revised estimates of real GDP show an average annual growth rate of 3.2%, compared to 3.0% for previously published estimates. The revision period can be sorted into two subperiods-before and after what had been the fixed-weight year of 1987. Between 1959 and 1987 the average annual growth rate in real GDP is now 3.4%, compared to 3.1% for previously published estimates. The upward revision in the growth of real was due primarily to the shift to the chain-type measure. Between 1987 to 1994, the average annual growth rate is now 2.3%, compared to 2.4% for previously published estimates.

During the most recent completed expansion, which began in the third quarter of 1982 and ended in the second quarter of 1990, the average annual growth in real GDP is now 3.8%, which is 0.3 percentage point higher than previously published estimates. During the current expansion (from the first quarter of 1991 through the third quarter of 1995), the average annual growth in real GDP is now 2.5%, which is 0.6 percentage point lower than previously published estimates. So, compared to the fixed-weighted estimates, the chain-weighted estimates show a stronger expansion during the 1980s but a weaker current expansion.

I have only touched on some highlights of the revision. More detail is provided by studies referenced on the back page. But be forewarned, the chain-weighted approach is about three times more complicated than the current fixed-weight system. You will have to decide how much detail you want to get into in class, and how much you expect students to know. My own view is that in the principles course we might explain some intuition behind the new estimation procedure along with some implications of the new approach. But, given competing demands for the course (not to mention declining economics enrollments), a detailed examination of the new GDP estimation methodology may not reflect optimal use of scarce class time.

Return to Contents of Issue 11, Spring 1996

William A. McEachern, Editor

Burton Abrams	Dimitris J. Knaniou
Ali A. Ataiifar	John Jascot
Brad Stamm	Eric K. Steger
Jerry McElroy

---

• Burton Abrams of the University of Delaware says that his favorite illustration of the law of demand in science fiction comes from one of the original Star Trek shows. The crew of the Enterprise happen upon a planet that is waging war against another planet via computer, which estimates the impact of simulated attacks then project the number of casualties. Citizens selected as victims dutifully go to vaporization chambers. Because no actual weapons are used, no capital is destroyed, greatly reducing the cost of war -- a war that has been underway of decades. Captain Kirk, appalled by the antiseptic nature of the process, essentially pulls the plug on the planet's computer. This causes a treaty violation and forces the two planets to confront the implications of waging a real war. The suddenly higher cost of war forces both sides into peace negotiations. Professor Abrams raises two follow-on questions. Did the existence of nuclear weapons prevent the United States and the Soviet Union from going to war during the Cold War? And would the existence of a neutron bomb make the use of nuclear weapons more likely? All this reminds me of the movie War Games, where, in the end, the computer "realizes" that the only way to win the game of "Thermonuclear War" is not to play.

• Dimitris J. Knaniou of Point Park College in Pittsburgh says he has used science fiction to complement journal articles in his graduate business class. Material about extraterrestrial resource utilization that he has found provocative and conducive to further economic analysis include Greg Bear's Moving Mars, Ben Bova's Privateers, Michael McCollum's Thunder Strike, Frederick Pohl's Mining the Oort, and Kim Stanley Robinson's Red Mars and Green Mars. Food for thought includes ideas about the property rights of interstellar matter, "wormholes" as public goods, creating infrastructure on extraterrestrial facilities, and more generally, exonomics, which is what Professor Knaniou calls the "proper management of extraterrestrial resources."

• Ali A. Ataiifar of Delaware County Community College in Media, Pennsylvania, notes that micro appears more intuitive than macro because micro reflects individual behavior pursuing rational self interest. Students are, after all, individual actors pursuing their rational self interest. Macroeconomics, on the other hand, considers the entire economy, but few people have thought about the entire economy. To solve this perspective problem, Professor Ataiifar suggests that macroeconomic students think of themselves as visiting aliens from outer space, who have no self-interest on this planet. "They don't want our land, food, money, or anything else. Their sole mission is to help poor earthly beings put their economic house in order and prosper." This helps students to think not in terms of them versus the rest of society, but to think in terms of the best solution for a given situation involving the economy as a whole.

• The discussion of the economics of science fiction reminded John Jascot of Capital Community-Technical College in Hartford, Connecticut, of that new strain of fiction that is explicitly economic. He thought "a fun read…with lots of economic concepts thrown in" the fiction by Marshall Jevons, including Murder at the Margin (Princeton University Press), A Fatal Equilibrium (MIT Press), and A Deadly Indifference (Caroll & Graff). As most know, Marshall Jevons is the pseudonym for Ken Elzinga of the University of Virginia and Bill Breit of Trinity University in San Antonio. Another novel he recommends is Ben Stein's On the Brink, which describes the tumult created by a jump in oil prices and includes lots of macroeconomic issues. Ben, incidentally, is a prolific author and also an actor in roles that call for a dry, boring academic. He is the son of Herb Stein, former head of the Council of Economic Advisors and witty contributor to the Wall Street Journal.

• Brad Stamm of Nyack College in Nyack, New York, suggests that we expand the literary net cast for economic examples to include the Bible. He says the focus need not be on the spiritual side. A public finance example involves Christ's observation in Mark (12:41-44) that "the poor widow put in more than all the others contributors to the fund…. For they all have put in out of their abundance; but she out of her want has put in all that she had -- all that she had to live on". This passage implies that the poor widow contributed 100% of her income. More generally, we might consider the holy books of all religions as a source of economic examples.

• Eric K. Steger of East Central University of Ada, Oklahoma, explains in class how to use credit cards more rationally to take full advantage of the float. He shows the class that he has written the closing date on the back of his Visa and MasterCard. The Visa card closes the 6th of the month and the MasterCard, the 22nd. Which card should be used to pay for a purchase on the 12th day of the month and why? With Visa, the user can postpone payment until the 6th day of the following month. MasterCard would be billed as of the 22nd of the current month. By delaying payment, the credit card user has the use of money for a longer period.

• Jerry McElroy of Saint Mary's College in Notre Dame, Indiana, says that late in the class period, he selects a student to work something out at the blackboard based on the day's lecture. For example, the task may involve explaining how natural market forces operate in the context of aggregate demand and aggregate supply. To increase the probability of success during the first few classes, Professor McElroy takes care to select a student who has responded in class before-someone who seems to know what's going on. Once the term gets rolling, he selects students more randomly. This feature provides students a break from the lecture, a reason to pay closer attention, a fresh voice, and a chance to review the day's material. The feature provides the instructor a way to assess how well the material has been communicated.