THE TECHNOLOGY FACTOR AND THE EXPORT PERFORMANCE
OF U.S. MANUFACTURINGINDUSTRIES
THOMASC. LOWINGER’
Washington State University
T h e paper examines t w o central theories advanced to
explain the revealed comparative advantage of U.S. industries.
The neo-technological account centers on the process of innovation among industries and i s represented in the regression
analysis by an R b D intensity variable. The neofactor theory
advances both human and “physical” capital as important
variables in determining countries’ comparative advantage.
Foreign protection is postulated to affect the export performance 0-f i1.S. industries. Generally the results suggest that
U.S. revealed comparative advantage is most pronounced in
R b D intensive industries that give the U.S. a temporary
technological lead in world markets.
I. INTRODUCTION
U.S. trade balance experienced a sharp reversal in the 1960’s. From
persistent surpluses up until 1965 to approximate balance in 1968-69 the
U.S. registered a large trade deficit in 1971-72. From 1961-62 to
1971-72 the balance of trade swing amounted to about $9.8 billion.
The observed deterioration in the U.S. trade balance would have been
much greater, but for the large trade surpluses generated by the
technologically intensive industries.
Between 1963 and 1969 the combined trade surplus of five technologically intensive industries rose from $7.7 billion to $1 1.1 billion.’
(Figure 1) During the same time period the trade deficit of all other
manufacturing industries grew from $1.0 billion to $7.5 billion. Thus,
there is a sharp contrast in the performance of net exports (or imports) of
these two groups of industries.’
A number of studies in recent years attempted to establish the
determinants of U.S. comparative advantage in manufactured g o o d s . 3
*Research on this paper was done in part while the author was an International Economist with
the U.S. Treasury Department on leave from Washington State University. The helpful comments
of William Cline, Hang-Sheng Cheng, George Kopits and the referee are gratefully acknowledged.
1. The five technologically intensive industries based on the ratio of R&D expenditure to sales
are: Aircraft (SIC 372). Chemicals (SIC 28), Electrical equipment and communication (SIC 36),
Instruments (SIC 38) and Machinery (SIC 35).
2. The contrast is between a 653 percent increase in the trade deficit of “other” manufacturing
industries compared to a 45 percent rise in the trade surplus of the five technologically intensive
industries, between 1963 and 1969.
3. A useful synopsis of this area of research is provided in Stern (1973).
Economic Inquiry
Vol. XIII, June 1975
22 1
ECONOMIC INQUIRY
222
Figure 1. Trade Balance of Five Technologically Intensive Industries:
and All Other Manufacturing Industries’ Trade Balance, 1963-69
$Bill.
I
7
-1.5
-2.0
-
-2.i.
- - - - \ \
\
\
\
\
\
\
-3.0
\
\
-3.5--
\
\
\
-4.0--
\
\
\
-4.L-5.0-
Trade Balance of All Other
Manufacturing Industries
\
\
\
\
\
-5.5--
\
\
-6.0
\
\
-6.5--
\
\
-7.0
-7.1-
\
\
\
\ - - -
-
*Includes: Aircraft (SIC 372); Chemicals (SIC 28); Electrical equipment and communication
(SIC 36); Instruments (SIC 38) and Machinery (SIC 35).
Source: U S . , Department of Commerce, U.S. Commodity Exports and Imports as Related t o
Output, Washington, D.C. (various years).
LOWINGER: TECHNOLOGY AND EXPORTS
223
Specifically, it has been observed that a significant and positive relationship exists between U.S. export performance and the Research and
Development (R&D)“intensity” of U S . i n d ~ s t r i e s . ~
The chief purpose of this paper is to examine two competing theories
purporting to explain United States manufacturing industries’ export
performance. Although the emphasis is on the technological determinants of U.S. “revealed” comparative advantage, a streamlined version
of the factor proportions account is introduced as well. The technological
explanation of U.S. comparative advantage in manufactured goods
focuses on the variance in the industries’ ability to introduce new goods
and improved (lower cost) processes. Initially these new products are
aimed at the home market, but in time U.S. industries are able to
penetrate foreign markets as well. As a proxy for the industries’
propensity to innovate, this study uses a number of different formulations
of the R&D intensity variable despite certain acknowledged weaknesses
of these measures.
Section I1 takes up the theoretical and empirical issues involved in
testing the effect of the technology factor on the commodity composition
of trade.
Section I11 sets the framework for the multiple regression analysis
to be performed. The results of the analysis are reported and evaluated
in Section IV.
11. THE NEOFACTOR AND NEOTECHNOLOGICALTHEORIES
OF THE COMMODITY COMPOSITION OFTRADE
The human capital approach to determining comparative advantage
did not obviate the Heckscher-Ohlin-Samuelson (H-0-S) theoretical
framework, rather it considerably expanded the scope of its empirical
testing. The so-called neofactor account of trade structure is cast in a
three factor framework (i.e., “physical” capital, human capital and
“raw” labor) in lieu of the two factor version (capital, labor) of the
original Heckscher-Ohlin model.’
The human capital extension of the basic H-0-S model was found
useful in “explaining” the trade pattern of the U.S. and a number of
other countries6 Branson recently concluded that U.S. comparative
4. See, for example, the studies by: Gruber, Mehta and Vernon (1967)and Keesing (1967).
5. Kenen using a three factor framework was able to demonstrate (based on some rough calculations) that when account is taken of capital invested in workers, the well-known “Leontief paradox”
could be disposed of. See Kenen (1965,pp. 456-58).
6. The main difficulty with this approach revolves around the theoretical validity of the various
measures approximating human capital intensity. If one assumes that investment in people is fully
reflected in their earned income, then the value of their capitalized earnings over and above the
earnings of “raw” labor would approximate their human capital intensity. Alternatively, one could
calculate the cost of investment of people in themselves through education, training, etc. Finally,
skill indexes have been used as a proxy for the importance of human capital in certain activity. This
last measure, however, fails to deal adequately with the process of capital accumulation that is
involved in the formation of human capital. For further details, see: Kenen (1970) and Branson and
Junz (1971).
224
ECONOMIC INQUIRY
advantage lies in products that are human capital intensive (based on
capitalized wage differentials); industries’ skill ratios were found to
have an independent influence in explaining U.S. export patterne7These
results may reflect the fact that technologically advanced industries,
that export a large proportion of new and improved products, are
also characterized by a relatively high ratio of skilled personnel (professional, scientific and technical). This naturally leads one to inquire into
the role of technological factors in determining the commodity composition of trade.
The neotechnological theories of trade structure differ from the H-0-S
model by allowing for the conditions of production to vary across
countries (for the same commodity).’ Posner’s (1961) technological gap
account and Vernon’s (1966) “product life cycle” theory challenged
the H-0-S assumption of uniform production functions across countries.
In Posner’s scheme a random innovation (a new product or an improved
process) takes place in one country. Production ensues but no exports
will take place until the foreign country’s demand pattern adjusts
(i.e., overcomes the “natural” opposition to the new product). After
a certain time lag consumption abroad takes place. When foreign
producers perceive a threat to their market position foreign production
commences (this is dubbed the imitation lag). Posner’s hypothesis
though suggestive, does not offer any clues as to the determinants of
differential rates of innovation across countries and the length of lags
in foreign imitation to the perceived threat from abroad.
Vernon’s product life cycle theory to some extent overlaps Posner’s
technological gap explanation of trade but the emphasis is different
and the content richer in detail. Entrepreneurs in advanced countries,
especially the U.S., tend to introduce new products that cater to the
tastes of high-income consumers as well as industrial processes that
favor labor-saving type technology (in view of the high cost of labor).
In the early stages of the product cycle production is characterized by
highly differentiated products and processes so that locational considerations and specialized (i.e., highly skilled) labor is of great importance.
Later on as the demand for the product (both at home and abroad)
expands standardization of the new product becomes possible as the
number of feasible processes declines. The standardization of production
processes and the establishment of wider marketing facilities enable
firms to attain economies of scale due to longer production runs.
However, the process of standardization of production techniques
7. Branson (1971, pp. 754-59) noted that while R&D expenditure has a positive impact on U.S.
export performance, it is only of marginal importance compared to the variables representing
human and physical capital intensity.
8. The proposition that comparative advantage may be determined both by technological and
factor endowment differences between countries has been expounded by Jones (1970).
LOWINGER: TECHNOLOGY AND EXPORTS
225
eventually enables other countries to produce and even export these
goods. In the final states of the cycle, factor price differentials will play
the paramount role in determining the pattern of trade, enabling
countries abundant in labor (or capital) to become net exporters of
products they formerly imported from the innovating country.
The neotechnoiogical explanation of the commodity composition of
trade has been tested in a number of studies that focused primarily on
selected industries or products.
Hufbauer found the technological gap and the economies of scale
theories to be “consistently useful” in explaining export shares in
synthetic materials trade.g Findings by Freeman (1963) based on a
study of the German plastics industry and Hirsch (1965) dealing with
the U.S. electronics industry gave additional support to the neotechnological account of trade.”
Vernon’s concept of the product cycle fits comfortably into a three
factor framework using capital, “raw” labor and highly skilled labor.
The early stages of production of the new commodity require high
concentrations of specialized (skilled) labor that the U S . and few other
advanced industrial countries have a relative abundance of. Hence, the
comparative advantage in the early phases of the product cycle lies with
the United States. Later on, when the technology has become more
standardized and the product can be mass produced, economies of scale
are important and further capital deepening takes place. In the final
stage, the product being now standardized, the cost of unskilled labor
may become the decisive factor in determining the location and pattern
of exports of the now “mature” product.
These industry studies, while informative as such, did not attempt to
set the neotechnological explanation of trade against (in competition
with) the neofactor account in a context of a more generalized set of
empirical tests. Hufbauer (1970) has done this by computing the export
characteristics of twenty-four countries that were designed to reflect
central elements of both the neofactor and neotechnological theories.
’
9. In his empirical tests, though, Hufbauer (1966) was unable to separate the contribution of
the technological gap and the economies of scale to export performance for plastics, synthetic
rubber and man-made fibers.
10. Hirsch (1965)noted that US.competitive strength was most pronounced in industries whose
products were in the early growth phases of the cycle (when mass production economies and
management skills are predominant) as compared to those in the mature phases when labor costs
became the dominant factor.
In a later study Wells (1969) found that U.S. export performance in consumer durables was
generally consistent with Vernon’s model. Vernon’s hypothesis also provides a reasonably good
explanation of the pattern of production and exports of certain petrochemicals with economies of
scale being the central factor in determining their initial composition. See Stobaugh (1972).
11. As a test of the neotechnological account, Hufbauer focussed on economies of scale, product
innovation suggested by the technological gap theory (using “first trade dates”), and product differentiation as a stage in the product life cycle. See Hufbauer (1970, pp. 176-93).
226
ECONOMIC INQUIRY
Hufbauer reached the baffling conclusion that a number of different
characteristics, related both to neofactor and neotechnological factors,
influence the commodity composition of trade. It is evident from
this brief survey that a consensus has yet to emerge as to the validity
of either the neotechnological or neofactor explanations of trade
structure. It is the purpose of this paper to attempt a further clarification
along these lines.
111. EXPORT PERFORMANCE AND INDUSTRIES RESEARCH
AND DEVELOPMENT (R&D) EFFORT
A number of recent studies found the R&D variable to be a potent
factor in “explaining” the observed pattern of U.S. commodity trade.
Gruber, Mehta, and Vernon (1967, p. 30) concluded that: “. , . ; one
derives a picture of high research effort being correlated with industries
that experience substantial trade surpluses.” Similar results were
obtained by Keesing (1967) and subsequently by Baldwin (1971), and
Branson (1971, pp. 756-58). The R&D variables customarily employed
in these studies are: (a) the ratio of scientists and engineers engaged in
R&D to total employment of the industry; or (b) expenditure on R&D as
a ratio of industry’s sales (or value added).
These proxies representing industries’ R&D effort fail to reflect the
stock of “disembodied knowledge” that forms the basis for the generation of new products and/or improved processes. The use of a research
intensity variable in the regression analysis implicitly assumes that the
primary input into the R&D process (scientists and engineers engaged
in R&D) is highly correlated with the end products of that process,
the generation of new products that are thought to be the main source
of U.S. comparative advantage in manufactured products. The present
state of statistical information precludes the construction of economically
meaningful indexes of inventive output generated by the industries’ R&D
expenditure. A study of the Chemical, Drug and Petroleum industries
found a research intensity variable and a measure of research output to
be positively correlated; yet because the conclusions are based on limited
coverage of industries these results must be considered tentative. l 2
One difficulty involved in the approach that seeks to introduce a
measure of R&D effort into the analysis of trade performance is whether
a line can be drawn between the technological and the human capital
aspects of investment in R&D. While it has been established that the
relative abundance of highly skilled labor (especially those engaged
in R&D activities) is related to the competitive export performance
12. The research input variable was approximated by the R&D expenditures to sales ratio, while
research output (productivity) was given by the total number of patents issued per scientist and
engineer (in the previous four years). See Grabowski (1968).
LOWINGER: TECHNOLOGY AND EXPORTS
227
of U.S. industries, the interpretation of these findings remains somewhat
beclouded.
Can it be said that U.S. comparative advantage is determined by
her abundant supply of capital embodied in human beings, or is it that
the highly skilled personnel involved in the R&D process act as an
indispensable input in the production of new and improved products,
which give U.S. a competitive edge in world trade based on a temporary
technological advantage? In the later case, the skill intensity factor acts
as a “stand-in” for the ability of an industry to push the production frontier further out through the generation of new disembodied knowledge.
The second issue concerns the appropriate measurement of export
performance in the context of a study that seeks to “explain” U.S.
industries revealed comparative advantage. This paper employs two
measures of U.S. export performance.
(a) X , - U.S. industries’ export shares in “world” markets during
a recent time period.14
and;
(b) X2 - Changes in U S . industries’ relative export shares over
time.”
13. Kenen (1970) ran a set of regressions using a number of skill indexes and a R&D variable to
explain the trade performance of U S . industries. His results were inconclusive depending largely
on the form of the trade performance variable employed.
14. The first measure of U.S. export performance ( X , ) is defined as:
where:
X,
= Total U.S. exports of industry-K during 1968-1970
?Xk,{
=
Total exports of industry-K by the ith country 1968-70, aggregated
for all group-of-ten countries.
15. Equations (1) and (2) below show the relative share of
(1960-62)and the final period (1968-1970).
(l)
E,, / E,o
TkO
where:
US.exports of industry-K in the initial
(2)
TrO
Ekl
1- E t l
Tkl
‘tl
E and T refers to US.exports and the grou of ten countries exports respectively.
Subscripts k and t denote industry-K anf-al; manufactured goods, respectively,
and subscripts 0 and 1 refer to the time periods.
Finally, X , is obtained by dividing the relative share in 1968-70 into the relative share in the
base period, 1960-62.
5,s
(3)
x,
=
Tkl
Tri
&/%
TkO
TrO
228
ECONOMIC INQUIRY
The first ratio ( X I ) simply measures the ability of U.S. industries to
penetrate world markets in competition with other industrial countries
at a certain point in time. This measure has the advantage that it is not
dependent on the absolute size of the industry in the domestic economy.
An alternative comparative advantage indicator, net trade balance, is
not independent of the industries’ size. The second measure (X,) takes a
dynamic view of export performance by calculating the changes in
export shares of U.S. industries over the 1960-62 to 1968-70 time period.
The changes in export shares of each industry are normalized for overall
performance of U.S. manufactured exports. While these two measures
are not without defects, they do adequately represent the competitive
performance of U.S. exports in world markets at a point in time and also
as it changed over time (U.S. “revealed” comparative advantage.)16
The analysis then proceeds to relate a number of industry characteristics and a measure of foreign protection to the competitive export
performance indexes - X , and X,. The neotechnological theory is
alternately represented by a number of measures of industries’ R&D
effort as well as an economies of scale variable:
(a) RD, - Scientists and engineers engaged in Research and
Development as a percentage of industry’s total employment, average
for 1 9 6 7 - 6 9 . 1 7
and;
(b) The economies of scale variable has been adapted from
Hufbauer’s (1970) paper; it relates productivity changes to increases in
the size of establishments. l 8
16. To some extent, the export share measure (X 1 reflects both the ability of the U.S. to penetrate
international markets and its success in limiting foieign access into the home market. The later effect
is indirect, and therefore of minor importance only.
17. Two additional R&D intensity variables used in the empirical tests are:
RDb - Research and Development expenditures as a percentage of net industry sales, average
for 1967-69.
RD, - Wages and salaries of scientists and engineers performing R&D work as a percentage
of industry’s total payroll, 1967.
Our reservations concerning the use of an index of research input notwithstanding, we have used
the above measures in lieu of an acceptable index of research output. A crude attempt at estimating
innovative output by calculating the (alproximate) number of patents issued in 1969 per R&D
scientist and engineer (1966-67) has yiel ed insignlficant results when related to the three measures
of R&Deffort.
18. Industry’s scale elasticity parameter was calculated by Hufbauer (1970, pp. 178-81) based
on the following equation:
(1) V = K n R
where: V - is the ratio of value-added per worker in a given size plant to the average value
per employee in the industry.
n - is the average number of employees per establishment.
K - a constant.
(I - the scale elasticity parameter.
Hufbauer’s coefficients were converted to our classification scheme by means of industry’s export
shares as weights.
LOWINGER: TECHNOLOGY AND EXPORTS
2 29
The neofactor proportions account is represented on the one hand by
a measure of “physical” capital, and on the other hand, by a variable
that reflects industries’ human-capital intensity. Overall capital intensity
using value added per employee as its indicator has been used as a
proxy for the combined inputs of human and physical capital into the
production of manufactured goods. This statistic first proposed by Lary
(1968, pp. 20-22) reflects the aggregate flow of services rendered by
human and “physical” capital into the manufacturing process. Furthermore, the wage and salary component of value added can be taken to
represent the human capital intensity of an industry while the non-wage
component of value added reflects the extent of “physical” capital
intensity in an inter-industry setting.
Finally, a foreign protection variable is introduced into the empirical
analysis in order to test the proposition that differential rates of foreign
protection may affect the inter-industry pattern of U.S. exports. If a
country wants simply to protect its import competing sector and has no
desire to favor a particular industry (or industries), a uniform tariff
would be called for. Yet, countries quite legitimately may want to attain
a desired degree of self-sufficiency in certain “essential” industries,
specifically those related to national defense, and will use differential
rates of protection in pursuit of that goal.lg
The foreign protection variable is defined as a weighted average of
the group-of-ten countries tariffs (exclusive of the U.S., of course) for our
sample of industries.
IV. SUMMARY AND EVALUATION OF EMPIRICAL RESULTS
The major results of the tests relating U.S. industries’ export performance to industry characteristics that are hypothesized to represent the
central features of the neotechnological and neofactor trade theories are
reported in Tables 1 and 2.
O n the whole, the technological intensity variable in its various
forms turns out to be the single most potent explanatory variable of
U.S. industries’ revealed comparative advantage. In regression equations
using R&D intensity as the sole explanatory variable of U.S. competitive
performance in world markets the coefficient of determination (R2)
ranges from .SO to . 7 3 . Namely, up to 7 3 percent of the total variation
in U.S. industries’ export performance is associated with the differences
19. For national defense and global prestige reasons, European countries and especially member
countries of the EC are likely to favor certain technologically sophisticated industries through
higher tariffs or direct subsidies.
Servan-Schreiber (1969, pp. 110-11) has argued forcefully that the traditional (static) division of
labor will relegate European industries to the role of technological satellites of American firms. To
meet the “American Challenge” he calls for additional government intervention and support in the
areas of electronics, data processing, space research and atomic energy.
XI =
=
=
x,
x,
3
4
5
.054
(8.861)* *
-.037
(.536)
.04 1
(3.625)* *
,058
(10.188)**
.055
,230
(S.SSO)**
-.012
(.166)
(7.538)* *
.054
(6.463)* *
-.070
(1.073)* *
.078
(2.975)* *
RDA
=
U.S. industries export shares of “world” trade, 1968-70.
.212
(1.402)
.436
(3.308)* *
.329
(2.4 17)*
W
T F = Export weighted averages of foreign tariff rate on the industry’s products.
E S = Economies of scale variable, taken from: Hufbauer (1970, pp. 2 12-20).
W = Ratio of wages and salaries of all employees in industry’s value added, 1967-69.
RDA = Scientists and Engineers engaged in R&D as a percentage of total employment, 1967-69.
X2 = Changes in the relative export shares of U.S. industries, 1960-62 to 1968-70
Xl
Definition of variables for Tables 1 , 2 and 3.
Xi =
2
=
x,
1
Eq. Dependent
No.
Variable
+
( 1.5 19)+
.495
ES
.818
.891
.057
.045
.049
.059
,800
.871
.069
S.E.
.731
R2
2.0799
2.1209
2.3468
2.0103
1.8485
D.W.
Summary Statistics
Notes: Numbers in parentheses are “t” coefficients.
- indicates 90 percent level of significance.
* - indicates 95 percent level of significance.
* * - indicates 99 percent level of significance.
-.010
(2.841)* *
-.009
(2.563) *
TF
Independent Variables Coefficients
TABLE 1
Regression Equations Using RD, as Principal Explanatory
Variable of U.S. Export Performance (XI)
0
a.N
c
X2 =
X2 =
x 2 =
X,
2
3
4
5
-.I74
(.315)
,435
(2.114)*
-.718
(1.395)*
-.260
(.500)
1.734
(5.852)**
‘Onst.
.137
(.617)+
1.456
(1.280)
2.553
(2.388)*
3.240
(3.331)**
W
-.055
(3.988)**
NW
-.091
(3.170)**
-.092
(3.553)**
TF
.822
.783
.625
.so0
R2
4.631
,687
(1.887)*
ES
,429
.330
.364
.470
.543
S.E.
2.3358
1.7037
2.1912
2.2499
2.1119
D.W.
Summary Statistics
Trade: OECD, Trade by Comrnoditics, Series B (various years)
Tariffs: The weighted average foreign tariff rates were obtained from a Treasury Department study. The foreign countries included were the Group-of-Ten.
The average foreign protection was computed from post-Kennedy round tariffs. Cline and Hays (1973, Data Appendix).
Research and Development: R&D is defined as: “Basic and applied research in the Sciences and Enginrering, and the design and drvelopmmt of prototypes
and processes.”
National Science Foundation, Research and Development in Industry, 1970, Washington, D.C., 1972.
Other Industry Characteristics: U.S. Department of Commerce, Annual Survey ofManufacturers 1970, Washington, D.C.
US.Burcauof theCrnsus, Census of.Manufactures 1967, Vol. 1. Washington, D.C., 1971.
Sources:
RDA
.262
(3.996)*
.264
(4.658)**
.25 1
(5.647)**
.291
(6.935)**
aFor definitions of variables see TABLE I
=
x2 =
1
Eq. Dependent
No. Variable
Independent Variables Coefficients
TABLE 2
Regression Equations Using RDA as Principal Explanatory
Variable of U.S. Export Performance (X2)a
I
w
N
2
P
v
5
m
5
”
>
E2
$!
4
8
p
2
2
r
0
ECONOMIC INQUIRY
232
in industries’ research intensity as such. 2o
The economies of scale measure that is often associated with the neotechnological explanation of competitive trade performance, turns out
to be only marginally significant in the regression analysis.
In fact, the statistical significance of all the regression coefficients
has been reduced when the economies of scale and R&D variables have
been included in some of the equations. This is not entirely unexpected
in light of the fairly high degree of multicollinearity present in our
sample of industries. This contributes to the overall unreliability of the
estimated regression coefficients in Equation 5 in both Tables 1 and 2.
Specifically, we found that an approximate linear relationship exists
between the R&D and economies of scale variables (Table 3).
TABLE 3
Simple Correlation Coefficients Between Various Industry
Characteristics for the Sample of 16 Industriesa
RDA
RDA
RDB
RDC
TF
W
NW
ES
1.oo
RDB
.95*
RDC
.91*
TF
-.12
W
-.06
1.00
.75*
-.13
.21
1.00
-.lo
1.00
-.40
.11
1.oo
NW
.29
.03
.59*
-.14
-.92*
ES
.75*
.83*
.56*
-.20
.25
1.00
-.02
1.00
aSee Table 1 for definitions of variables.
NW = The non-wage and salary component of industry’s value added, per employee, 1967-69.
RDB = R&D expenditure as a percentage of net industry sales, 1967-69.
RDC = Wages and salaries of scientists and engineers performing R&D work as a percentage of
industry’s total payroll, 1967.
* Indicates coefficients are significant at the 95 percent level.
It has been observed that the products of industries characterized by
the existence of important scale economies when produced by countries
characterized by large markets (relative to the industries’ minimum
efficient scale) give those countries a definite competitive edge in world
20. Additional regression equations using RDb and RDc as the main explanatory variables are
available from the author on request. The statistical results are generally similar to those reported
in Tables 1 and 2.
LOWINGER: TECHNOLOGY AND EXPORTS
233
markets. However, the ability of firms to develop new products and
improved processes may be constrained due to the existence of scale
effects in the research effort itself. Presumably, certain minimum
economic size of investment in R&D is required in order to produce
results that can be profitably marketed. Economies of scale may be
present in R&D because of the lumpiness of investment required in
carrying out significant innovations and the added advantages of using
highly specialized personnel available to large firmse2’
Furthermore, small firms are unlikely to generate a significant number
of innovations because they do not have a large enough volume of
production to be able to utilize the innovations profitably. Available
data support the notion that R&D effort is highly concentrated among
“large” and “very large” firms. While the average R&D expenditure of
small and medium firms (fewer than 5,000 employees) was less than
$200,000 per company, the very large firms (employing 10,000 or more
workers) spent in 1967 $46 million per company.22 Further evidence
is presented in Table 4, where the concentration of R&D effort and sales
is found to be considerably higher among industries that we characterized as research intensive, compared to all R&D performing industries.
In an important study of R&D expenditure in three selected industries,
Mansfield (1964, p. 336) concluded that when the size of firms is held
constant “. . . , the number of significant inventions carried out by a
firm seems to be strongly influenced by the size of its R&D expenditures.” Furthermore, the size of firms as such does not guarantee a
greater productivity of the firms’ R&D program. 23 Thus, Mansfield’s
overall conclusion is that the scale of the R b D program is an important
determinant of inventive output, but not the size of firms per se.
The neofactor proportions account of trade receives some qualified
support in the empirical analysis. The measure of human-capital
intensity appears to be positively related to U.S. industries’ export
performance. Furthermore, the test of the “basic” H-0-S model yields
by now the familiar results that the “physical” capital intensity of
industries has a negative coefficient when regressed against the revealed
comparative advantage indexes.24 These findings are generally in
agreement with previously published studies, beginning with Leontief’s
21. Freeman and Young (1965) observed that in international comparisons the large size of U.S.
firms compared to their European competitors gives the U.S. an edge whenever the development
work is on a very large scale involving sizeable R&D expenditure as the case may be in computers,
aircrafts and certain chemical plant work.
22. In 1969, the top 40 companies (ranked in terms of the size of their total R&D expenditure) of
all R&D performing companies, accounted for 25 percent of net sales; 27 percent of employment;
and 68 percent of all R&D funds. SeeNSF (1972, pp. 46-7).
23. However, it should be born in mind that Mansfield’s sample of industries only includes “large”
and “medium” size firms.
24. These results are also available on request from the author.
ECONOMIC INQUIRY
234
TABLE 4
Distribution of R&D Funds and Net Sales
Based on Company Size, 1970
Industry
Aircraft
Electrical Equipment
Drugs
Scientific Instruments
Industrial Chemicals
Machinery
All Industries
Sic. No.
Percent
Percent of Sales
of All R&D
of All RW-Performing
Funds’
Companies*
Top 8
Companies
Top 20
Companies
Top 8
Companies
Top 20
Companies
372,19
87
98
48
92
36
283
71
61
84
91
51
50
65
88
38
77
87
50
72
281-82
73
91
54
84
35
70
82
33
48
32
55
9
16
* The companies are ranked by size of R&D program (based on total R&D funds) in 1970.
Source: NSF, Research and Development in Industry, 1970, Washington, D.C., April 1972, p. 48.
(1953) well-known study and culminating in a recent paper by Baldwin
( 1 9 7 1 , ~132-38).
~.
The inter-industry structure of foreign tariffs has acted as a significant impediment to U.S. exports directed to the markets of the major
industrial nations. For reasons that we suggested (and perhaps others)
advanced industrial nations tend to protect their technologically
intensive industries through differential tariffs (and non-tariff barriers
as well) and thereby blunt the “inherent” comparative advantage of
U.S. firms in international markets.
V. CONCLUDING REMARKS
U.S. competitive performance in international trade is largely
determined by the country’s ability to invest a comparatively high
proportion of its resources in the development of new products and
improved processes. A high rate of generation of new knowledge whether
embodied (in capital) or disembodied is typical of a high income country
such as the U.S. and may be thought to be the mainstay of its comparative advantage in international trade. Because of its high income and
large size the U S . is typically able to affect a high rate of capital
accumulation and innovation relative to less affluent countries (clearly
LOWINGER: TECHNOLOGY AND EXPORTS
235
it is a matter of degree, at least among the most developed nations).
This view accords with our findings that U S . comparative advantage is
most pronounced in industries that are R&D intensive, giving the U.S.
a temporary technological lead in exports of their products. U.S.
competitive position in world markets is further enhanced through the
effective use of its relatively abundant factor - highly skilled labor.
REFERENCES
1 . Baldwin, R. E., “Determinants of the Commodity Structure of U.S. Trade,” American Economic
Review, March 1971,61, 126-46.
2. Branson, W H., “U.S. Comparative Advantage: Some Further Results,” Brookings Papers on
Economic Activity, 3: 1971, 754-59.
3.
and Junz, H. B., “Trends in U.S. Trade and Comparative Advantage,” Brookings
Papers on Economic Activity, 2:1971, 285-338.
4. Cline, W. R., and Hays, L., “Competitiveness Rankings and Disaggregated Industrial Trade
Liberalization Effects,” US. Department of Treasury, January 1973.
5. Freeman, C., “The Plastics Industry: A Comparative Study of Research and Innovation,”
National Institute Economic Review, November 1963.22-62.
6. ___,
and Young, A., The Research and Development Effort in Western Europe, North
America and the Soviet Union, Paris, 1965.
7. Grabowski. H. S . . “The Determinants of Industrial Research and Develoument: A Studv of the
Chemical, Drug, and Petroleum Industries,” Journal of Political Econoky, MarchiApril 1968,
76, 292-306.
8. Gruber, W. H., Mehta, D., and Vernon, R., “The R&D Factor in International Trade and International Investment of United States Industries,” Journal of Political Economy, February 1967,
75, 20-37.
9. Hirsch, S . , “The United States Electronics Industry in International Trade,” in The Product
Life Cycleand International Trade, ed., L. T. Wells, Jr., Boston, 1972,39-52.
10. Hufbauer, G. C., Synthetic Materials and the Theory of International Trade, Cambridge, 1966.
1 1 . ~,
“The Impact of National Characteristics and Technology on the Commodity Composition of Trade in Manufactured Goods,” in The Technology Factor in International Trade,
ed., R. Vernon, New York, 1970.
12. Jones, R. W., “The Role of Technology in the Theory of International Trade,” in The Technology
Factor in International Trade, ed., R. Vernon, New York, 1970.
13. Keesing, D., “The Impact of Research and Development on United States Trade,” Journal of
Political Economy, February 1967, 75,38-45.
~
14. Kenen, P. B., “Nature, Capital. and Trade,” Journal of Political Economy, October 1965,
73,437-60.
15. ___,
“Skills, Human Capital, and Comparative Advantage,” in Education, Income, and
Human Capital, ed., W. L. Hansen, New York 1970.
16. Lary, H. B., Imports of Manufacturersfrom Less Developed Countries, New York, 1968.
17. Leontief, W., “Domestic Production and Foreign Trade: The American Capital Position Reexamined,” Proceedings ofthe American Philosophical Society, September 1953,97, 332-49.
236
ECONOMIC INQUIRY
18. Mansfield. E., “Industrial Research and Development Expenditures: Determinants. Prospects.
and Relation to Size of Firm and Inventive O;tputs,” Jdurnal of Political Economy, Lugust
1964, 72, 319-40.
19. Posner, M. V., “International Trade and Technjcal Change,” Oxford Economic Papers, October
1961, 13, 323-41.
21. Servan-Schreiber, J. J.. The American Challenge, New York, 1969.
21. Stern, R. M., “Testing Trade Theories,” Paper Presented at the Conference on Research in
International Trade and Finance, March 30-3 1 , 1973, International Finance Section, Department of Economics, Princeton University.
22. Stobaugh, R. B., “The Neotechnology Account of International Trade: The Case of Petrochemicals,” in The Product Life Cycle and International Trade, ed., L. T. Wells, Jr., Boston,
1972.83-105,
23. Vernon, R., “International Investment and International Trade in the Product Cycle,” Quarterly
Journal of Economics, May 1966, 80. 190-207.
24. Wells, L. T., Jr.. “Test of a Product Cycle Model of International Trade: U.S. Exports of
Consumer Durables,” Quarterly journal ofEconomics, February 1969.83, 152-62.
25. National Science Foundation, Research and Deoelopmat in Industry, 1970, Washington, D.C.,
April 1972.
© Copyright 2026 Paperzz