Comments on R&D Satellite Account: Strengths and proposals for improvement Bureau of Economic Analysis Conference Measurement Issues and the R&D Satellite Account Methodology December 13, 2006 Carol Corrado, Federal Reserve Board
Focus of Remarks 2006/7 R&D Satellite Account 2012 NIPAs Future R&D (R&D treated as Satellite Account Investment)(eg., RoRs, spillovers)
Conceptual issues Ownership and valuation issues. Sectoral (vs. industry) issues. Defining the scope of R&D investment. –Currently is NSF’s surveyed R&D. –Should other investments in new product/process development also be included? YES! (but subject for another day)
The ownership of R&D investment Funder vs. Performer? Tricky….But I think BEA has this right. Implications: –Treat publicly funded R&D like a bridge or other infrastructure. –Treat private R&D like a machine-tool or software. Public/ private R&D have a different “character” (basic vs. applied or up vs. down stream) –Public R&D is long-lived/valuable. –Within NIPAs => Depreciation rates for private vs public R&D should be different (eg. via “character”).
Public knowledge stock More important than size?
Measurement of R&D investment (current dollar value) BEA is using cost. –The BEA document states (p23), "...there is currently no standard measure of R&D output..." in defense of using costs. –Measuring R&D investment as the cost of R&D is consistent with the NIPAs. How would you measure machine-tool investment? By the cost of the machine-tool. …. software investment, etc. This approach doesn't depend upon whether the firm produces the R&D itself or purchases it on the market. BEA has chosen no gestation lag for R&D.
Price indexes for R&D Economic theory suggests firms will invest in an asset up to the point that cost = PDV of the stream of income expected to be produced by the asset. –Suggests an overall business product price is a reasonable proxy price index for private R&D. –As the marketplace for intellectual property develops, prices of existing IP assets will be able to be recorded (and perhaps used to infer prices for “new” private R&D). To maintain consistency with the treatment of bridges, an input-cost approach can be used to deflate publicly R&D.
Issues needing further study NSF time series is affected by its exclusion of R&D by small firms (unincorporated business). –Potentially a big deal. Internationalization of R&D. –US R&D by foreign-owned MNCs is US investment (and part of the US productive capital stock). –Vice versa for US MNC R&D spending abroad— potentially large to the extent “D” locates near market. Suggests consolidated I&P account for MNC R&D would be of considerable interest –what impact on saving? –FOFAs and balance sheets vs. production accounts.
VC Investments as proxy…
Time-series/Quarterly estimates Time-series integrity of annual series –Documentation on treatment of breaks in NSF data, eg –Dispose the so-called R&D/software overlap NSF instructs respondents to exclude own-account software. But, BEA may be counting software R&D as own-account software. Historical quarterly estimates –How well does (VC-adjusted) NSF spending relate to employment and wages of scientists and engineers? Going forward: Need appropriate instrument for NIPAs –Redesign of NSF survey underway (included a lot of outreach— great effort!) –Quarterly indicator survey (Census)? Does BEA have opinion?
Summary Keep decisions consistent with NIPAs (or change NIPAs) –Measure R&D investment as cost. –Use price index for business product (or product of 5-10 basic technology groups) to deflate private R&D. –Differentiate R&D depreciation by character to highlight key distinction between private and public R&D. Need to consider: –Role of venture capital (corporate/noncorporate) –Integration of BEA MNC data (domestic/r-o-w) Watch the IP marketplace as it develops (price data) Continue to work with the NSF and Census so that survey addresses the needs of the NIPAs. Why wait until 2012?