Saturday, August 3, 2013


According to the BEA's first version (the "advance estimate") of the second-quarter GDP numbers, which were released this past week, the US economy grew at a lackluster 1.7% annual rate in April-June.  That's an acceleration from the first quarter (1.1%) and the last quarter of 2012 (0.1%). But its still pretty disappointing - a treading-water-ish growth rate at best when we need faster-than-normal growth to make meaningful progress towards something resembling "full employment".

Although the unemployment rate has been falling, other measures like the employment-population ratio highlight how the labor market is far from fully recovered from the 2008-09 recession, and a 1.7% growth rate isn't nearly fast enough to really help.
Second-quarter growth was boosted by investment, which increased at a 9% rate (including 13.4% growth rate in residential investment; inventory accumulation also contributed).  Net exports were a minus, with imports growing faster than exports (9.8% versus 5.4%).  Government expenditures were a small drag, falling 0.4% (the federal government was at -1.5%, while state and local grew 0.3%).  It was the third consecutive quarter that the federal government has made a negative contribution (though less so than in the previous two).

The interesting thing in the release was not the second-quarter numbers, but rather the "comprehensive revision" of all the past data that came with it.  The BEA does this about every 5 years to incorporate changes in definitions and methodology.  By its new reckoning, US GDP is about $550 billion (3.4%) larger.

This shows how much the revisions have added to GDP over time (the percent difference between the new and previous versions of the series, constructed using vintage data from Alfred):
The main changes are to treat research and development as well as the creation of movies, TV programs, books and music (which the BEA is calling "entertainment, literary and artistic originals") as forms of investment.  These appear in the tables as a new subcategory of investment called "intellectual property products" which also includes software (formerly part of "equipment and software"; now there's a separate "equipment" subcategory).

Slate's Matthew Yglesias had a nice explanation of rationale for the changes:
Government statisticians draw a distinction between money a company spends as the cost of doing business and money a company spends on investing for the future. When a moving company buys a new truck, that’s an investment. You expect the truck to last for years and generate an ongoing stream of income. The truck purchase is part of GDP. Each year the truck depreciates in value, with the depreciation subtracting from GDP as what’s known as “consumption of fixed capital.” When a moving company buys gasoline to fuel the truck, that’s just the cost of doing business. The expense incurred is subtracted from the company’s income when calculating how profitable it was in any given year.

Right now we treat filming a season of Game of Thrones or having researchers work on a new pharmaceutical as being similar to filling up the truck’s gas tank. The new method is to treat them as similar to buying the truck.

Conceptually, the new way is clearly correct. Game of Thrones is to HBO as a truck is to a moving company: part of its stock of capital. Accounting for artistic originals in that way will add about half a percentage point to the size of America’s overall economy. Doing the same for research and development spending will add slightly more than two percentage points. And here, too, the new system is more sound in theory. A drug company is going to own plenty of physical capital goods, but its most important investments are the ones it makes in researching new products.
At Econbrowser, James Hamilton used a "Robinson Crusoe" story to explain the changes, Dean Baker and Jared Bernstein used the occasion to consider some of the other limitations of GDP in a NYT op-ed.

Overall, the revisions do not significantly change the "macro" picture of the behavior of the US economy.  With the new series, the average growth rate is slightly faster: 3.16% versus 3.11% since 1947 (when the quarterly data begins) and 2.50% versus 2.42% since 1993.  The standard deviation of growth is a tiny bit lower in the new series, and inflation, measured by the GDP deflator, is reduced a smidge by the revisions.  The "extra" output added in the revision is acyclical for the 1947-2013 period as a whole - the correlation of the addition to GDP with the growth rate is -0.04 - but slightly pro-cyclical over the last 20 years, with a correlation between growth and the addition of 0.19.

The new series makes the 2008-09 recession look slightly less bad, and the subsequent expansion a little bit better, as can be seen looking at both series normalized to 100 at the previous business cycle peak of fourth-quarter 2007:
At his FT blog, Gavyn Davies puts the new numbers in the context of economic policy.

Another change was to treat traditional, defined-benefit pensions on an "accrual" basis, which means they will be counted as income as the obligations to workers rise, rather than when employers put money in their pension funds.  This boosts the measured saving rate, which the Washington Post's Jim Tankersley thinks is problematic:
That money isn’t necessarily real. The Bureau of Economic Analysis didn’t find hundreds of billions of dollars stuffed in Americans’ mattresses. It decided to start counting all pension promises as savings in the bank....

The promises that aren’t backed by an income stream are called unfunded liabilities, and by changing how it counts them, the government added almost $200 billion to the nation’s personal savings for 2012.

The catch is, what if those promises don’t come true? The accounting change was made shortly after Detroit became the largest U.S. city to file for bankruptcy, in part due to the unfunded liabilities in its pension plan, raising questions over whether pensioners will actually receive the benefits they’ve been promised. Signs point to more strains on state and local government pension funds down the road.
(It's worth noting that some, including Paul Krugman, argue that the hand-wringing about under-funded pensions is overblown).

Another change is that some expenses associated with property transfers like attorney's fees and title insurance are also going to be treated as investments (the idea is that when you pay for these, you're not consuming a service, but rather paying for something that generates a flow of services - keeping the rain off your head - over a period of time).

The BEA also moved the "base year" to 2009 - i.e., real GDP is now expressed at 2009 prices (before it was 2005).

The BEA has more information about the comprehensive revision here; I found this article from the March Survey of Current Business (pdf) particularly helpful.

The revised "second estimate" of second-quarter GDP is due on Aug. 29.

1 comment:

The Arthurian said...

OMG i just got it.
GDP:TNG, The Next Generation.
How the hell did I miss that???