Improving GDP Measurement: A Measurement-Error Perspective
Policy officials at the Federal Reserve Board and elsewhere depend on having current information about national output. The Bureau of Economic Analysis (BEA) regularly reports two measures of Gross Domestic Product (GDP), one calculated by adding up total domestic expenditures on final goods and services and a second calculated by adding up the incomes realized by those involved in the production of that final output. In principle, the two measures should agree, but in practice they frequently diverge.
The measure traditionally featured in BEA releases has been the expenditure-side measure. Because the traditional expenditure-side measure suffers from significant measurement error, it could provide a misleading signal about the state of the economy. Aruoba et al adopt a measurement error approach to construct a measure of national output that uses both expenditure-side and income-side information. In their calculations, they apply statistical methods developed to extract an underlying signal from multiple noisy measures of a phenomenon of interest. They show that a combination of expenditure- and income-side information can produce more accurate GDP estimates than expenditure-side information alone. Further, the method used to calculate GDP makes a material difference to the conclusions drawn from the data. For example the GDP series Aruoba and his coauthors extract displays twice as much persistence than the expenditure-side GDP, which suggests that the former is more predictable than the latter.
Because the traditional expenditure-side measure suffers from significant measurement error, it could provide a misleading signal about when interest rates should be raised or lowered. Measures of the sort proposed by Aruoba et al can provide a substantially more informative signal.
Updates: Since November 2013, the Federal Reserve Bank of Philadelphia has been producing a measure called GDPplus based on the methods proposed by Aruoba et al that offers analysts an alternative to official GDP statistics. More recently, in July 2015, BEA began to produce a measure of GDP that is calculated as an average of the expenditure-side and income-side measures, a big step in the direction Aruoba and his colleagues have advocated.
For historical values of the GDPplus measure click here.
Aruoba, S. B., Diebold, F. X., Nalewaik, J., Schorfheide, F., & Song, D. (2016). Improving GDP measurement: A measurement-error perspective. Journal of Econometrics, 191(2), 384-397.