Offshoring, Sourcing Substitution Bias and the Measurement of US Import Prices, Real GDP and Productivity (PDF)

The decade ending in 2007 was a period of rapid sourcing substitution for manufactured goods consumed in the US. Imports were substituted for local sourcing, and patterns of supply for imports changed to give a large role to new producers in emerging economies. The change in the price paid by the buyer of an item who substitutes an import for local sourcing is out of scope for the US import price index, and the price change for an imported item when a new supplier in a different country is substituted for an existing one is also likely to be excluded from the index calculation. Sourcing substitution bias can arise in measures of change in import prices, real GPD and productivity if these excluded price changes are systematically different from other price changes. To determine bounds for how large sourcing substitution bias could be, we analyze product-level data on changes in import sourcing patterns between 1997 and 2007. Next, we identify products in the US industry accounts that are used for household consumption and that are supplied by imports. We aggregate CPIs, and combinations of MPI and PPIs that cover these products up to the product group level using weights that reflect household consumption patterns. With some adjustments, the gap between the growth rate of the product group index containing MPIs and the growth rate of a corresponding product group index constructed from CPIs can be used to estimate sourcing substitution bias. For the nondurable goods, which were not subject to much sourcing substitution, the gap is near zero. Apparel and textile products, which were subject to considerable offshoring, have an adjusted growth rate gap of 0.6 percent per year. Durable goods have an adjusted gap of 1.2 percent per year, but the upper bound calculation suggests that some of this gap comes from effects other than sourcing substitution. During the period examined, sourcing substitution bias may have accounted for a tenth of the reported multifactor productivity growth of the US private business sector.

Marshall Reinsdorf and Robert Yuskavage