Wednesday, December 03, 2008

Income versus output 

Caution: Serious data geekery ahead. Proceed at own risk.

I was only mildly surprised by the NBER call of recession yesterday, certainly not as much as Brian Wesbury or Mark Perry appear to be. Still, there are going to be those who will wonder about the recession call in light of there not yet being two consecutive quarters of GDP decline (even though every economist who studies these things knows the committee does not define recessions that way.) What the NBER's Business Cycle Dating Committee (BCDC) said about that is interesting.
The committee believes that the two most reliable comprehensive estimates of aggregate domestic production are normally the quarterly estimate of real Gross Domestic Product and the quarterly estimate of real Gross Domestic Income, both produced by the Bureau of Economic Analysis. In concept, the two should be the same, because sales of products generate income for producers and workers equal to the value of the sales. However, because the measurement on the product and income sides proceeds somewhat independently, the two actual measures differ by a statistical discrepancy. The product-side estimates fell slightly in 2007Q4, rose slightly in 2008Q1, rose again in 2008Q2, and fell slightly in 2008Q3. The income-side estimates reached their peak in 2007Q3, fell slightly in 2007Q4 and 2008Q1, rose slightly in 2008Q2 to a level below its peak in 2007Q3, and fell again in 2008Q3. Thus, the currently available estimates of quarterly aggregate real domestic production do not speak clearly about the date of a peak in activity.
Jim Hamilton notes that the monthly measures that the BCDC uses instead have been shouting recession for awhile. This includes real personal income less transfers which BCDC says "places the data closer to the desired measure, real gross domestic income."
The resulting monthly measure of real personal income less transfers is an imperfect measure of monthly real output because of definitional differences between personal income less transfers and gross national income and because we use the interpolated price index. Our measure of real personal income less transfers peaked in December 2007, displayed a zig-zag pattern from then until June 2008 at levels slightly below the December 2007 peak, and has generally declined since June.

I recalled while reading this that the only time I had read before about GDI was a discussion of its volatility, that it's more volatile than GDP. �That's odd, I thought, that they'd use a more volatile measure. �This has lead me in the last 24 hours to spend time thinking about the measurement of both GDP and GDI.

Thayer Watkins gives a fairly nice exposition on the concepts involved. �The best part is to just think about the idea that the total payments an industry makes must equal their total sales. �Now often these payments are not for final goods and services but for intermediate products -- those items a firm buys from another firm that will be input to something eventually sold to a final user. �(See also this by Mark Skousen.) �If you assume intermediate sales made between firms is equal to intermediate payments they receive, you should get final payments equal to final sales. �Payments are the income people receive plus imports, depreciation and indirect business taxes. �In other words, GDI.

Is GDP mismeasuring the state of the economy? �The difference between GDP and GDI is considered to be a "statistical discrepancy", and the Bureau of Economic Analysis used to argue that GDP is the more accurate measure.

BEA views GDP as a more reliable measure of output than GDI, because it considers the source data underlying the estimates of GDP to be more accurate. For example, most of the annual source data used for estimating GDP are based on complete enumerations, such as Federal Government budget data, or are regularly adjusted to complete enumerations, such as the quinquennial economic censuses and census of governments. In addition, all the expenditure components of GDP are revised every 5 years to reflect BEA's benchmark input-output accounts, which are prepared within an internally consistent framework that tracks the input and output flows in the economy. For GDI, only the annual tabulations of employment tax returns and Federal Government budget data are complete enumerations, and only farm proprietors' income and State and local government budget data are regularly adjusted to complete enumerations. For most of the remaining components of GDI, the annual source data are tabulations of samples of income tax returns.

But that's 1997, surely they've gotten better since then, right? Fixler and Nalewaik [2006] say in fact no. Neither measure is perfect and both reveal news about the real economy or "true GDP", which is of course unobserved but, they argue, more closely approximated as one gets more and more revisions of the data. �If BEA still relies on samples of tax returns for parts of GDI, it's never going to be completely right. �If we really thought the income measurement was superior, we'd report it more. �

Another reason to question GDI is that to get the real measure you have to compute a GDI deflator, and one doesn't exist. �You can find them for some of the components, or by using a sales-based approach like that in monthly GDP. � (That measure, by the way, peaks in June 2008. �Not sayin', just sayin'.) �So GDI, and the real monthly series, get a jerry-rigged price index to put them in "real" terms. �I wonder how long the BCDC argued over this. �

This isn't to argue that the NBER is wrong about there being a recession -- I have said as much here many times. �I quibble over the date, though, in no small part because of what Wesbury and Perry have said. �Your garden variety recession in the US is ten to twelve months; your nastier ones post-WW2 run about a half-year longer. �Only one of the contractions since 1900 lasts more than two years, and as you might guess, it's the one we put the Great tag on. �

So if we are going to insist the recession started in 12/07, you get to a recession of a very different nature than those garden varieties if the recession lasts into the third quarter of 2009. �And if it does turn out to be a more normal recession, then passing a huge stimulus package in January which pumps money in at the very end of it creates more instability than stability. �If the recession had started later, the argument for stimulus would be stronger. �How sure are we that this is the Really Really Big Recession of 2008? �Exit quote:

But how do you keep a faith in reason from developing into hubris? How can any economist today argue for say, a stimulus package, with any confidence? Or a further lowering of interest rates by the Fed? I can see making it as a suggestion. But how would you argue for it with passion? Doesn't the current situation and the inability of macroeconomists to predict it (or to have any certainty about whether we are going to have a mild recession or a serious Depression) suggest some humility?
Yes, I'd think so.