Transcript Document

Effects of Terms of Trade & Tariff Changes on
the Measurement of U.S. Productivity Growth
by Robert C. Feenstra, Univ. of California-Davis & NBER,
Marshall B. Reinsdorf, Bureau of Economic Analysis, Mathew J.
Slaughter, Tuck School of Business, Dartmouth College &
NBER, and Michael Harper, Bureau of Labor Statistics, all USA
Barbara M. Fraumeni
Muskie School of Public Service, University of Southern Maine
& the National Bureau of Economic Research (NBER), USA
IARIW, Session 3
Joensuu, Finland
August 22, 2006
Muskie School of Public Service
Ph.D. Program in Public Policy
MISMEASUREMENT OF
 Import & export prices
 The impact of tariff reductions &
currency devaluations
 The import effect on input and
output prices
LEADS TO INCORRECT ESTIMATES
OF PRODUCTIVITY CHANGE
Muskie School of Public Service
Ph.D. Program in Public Policy
Second Half of the 90’s
 New economy
 Globalization of IT industry in the US
 Key IT industries: Computers, peripherals,
semiconductors & telecommunications
equipment, are running trade deficits
 Import & export share about 15% of the
import & export totals for the U.S.
 Tariff reductions (to zero) on a wide range of IT
products
 Currency depreciation after Asian crisis
Muskie School of Public Service
Ph.D. Program in Public Policy
Import & Export Prices Indexes
 BLS constructs Laspeyres indexes
 This paper’s Tornqvist index lies below
Laspeyres
 Expected for net of tariff import index
 For exports, due to substitution along
foreign demand curves
 For IT industries, differences between these
two indexes show up in the 2nd half of the 90’s
Muskie School of Public Service
Ph.D. Program in Public Policy
Import Price Regressions
 Independent variables in fixed effect equation
 Indicator of tariff reduction (ITA) stages
 Tornqvist index of real exchange rates with lags
 Export or domestic price index
 3 variations in industry coverage depending upon %
covered by ITA
 100%
 50-99%
 1-49%
 2 control group regressions
 For manufacturing industries only
 For agricultural industries only when domestic
price is included
Muskie School of Public Service
Ph.D. Program in Public Policy
Regression Results
 In the version with domestic prices
(particularly with 100% ITA coverage), the passthrough coefficient indicates a highly
magnified impact on lowering import prices
 Exchange rate and export or domestic price
coefficients also significant in all versions
 Simulated prices, with no tariff or exchange
rate reductions, show that the ITA effect is
about = to the Laspeyres vs.Tornqvist formula
bias
Muskie School of Public Service
Ph.D. Program in Public Policy
Turning to Productivity Measurement
 Define a revenue function with
intermediate inputs gross of tariffs, but
industry gross output net of tariffs
 Authors show that eliminating tariffs with
INCORRECT/CURRENT measurement
 Increases US GDP rate of growth
 Increases MFP rate of growth
 See Figure 7
Muskie School of Public Service
Ph.D. Program in Public Policy
Figure 7
 Initially at y0 (tangency A0 and PPF, including
tariff on good 1)
 Nominal GDP at B0 at y0 excluding tariffs with
initial welfare at Q0
 After tariff elimination producers line A1 has
same slope as B0 (tangency at y1), thus welfare
increases to Q1
 But there is no shift on PPF, therefore no
change in MFP
 GDP measures the slope of the indifference
curve, which should not be used to estimate
productivity change
Muskie School of Public Service
Ph.D. Program in Public Policy
MFP Is Conflating True Productivity Changes
with International Trade Changes
 Figure 7 result
 Also existing international trade measures are biased
 BLS use of Laspeyres export & import price
measures
 Upward bias on MFP growth 1996-9
 .1 to .12 depending upon variant (BLS)
 Very small effect with NIPA price indexes because
upper level substitution bias is avoided
Muskie School of Public Service
Ph.D. Program in Public Policy
Effect of Measuring Imports Gross of
Tariffs
 Import share too large
 Falling tariffs reduce import prices
 On net, from tariff effect, MFP
overestimated by the same as the
international trade effect without tariffs:
.12
Muskie School of Public Service
Ph.D. Program in Public Policy
Capital Input Effects
 A corrected capital stock estimated with
Tornqvist capital good import prices would
have a higher growth rate than one with a
Laspeyres index
 Accordingly, MFP is too high by .038% per year
for 1995-2002
 Shortcut method of treating investment as an
intermediate input would bias MFP upwards by
.07% per year
Muskie School of Public Service
Ph.D. Program in Public Policy
Conclusions
 Don’t ignore international trade effects
 Both terms of trade (prices) and tariffs are
important
 Biases come from both the output and input
side
 Use of Laspeyre indexes creates biases in
estimates
Muskie School of Public Service
Ph.D. Program in Public Policy
Comments
 Paper is chock full of useful and thoughtprovoking results
 Difficult to find much to criticize
 Only very minor issues – a better summary &
 Argument for domestic price in import price
regressions seem stronger than for an
export price
 Globalization: Is 2000 a poor ending year to
use?
 Not true for a K stock aggregate that the K
stock rates of growth and the CSF growth
rates will be the same as weights will differ