The Nature of Japan's Comparative Advantage, 1965-80 Kellman1984

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

World Development, Vol. 12, No. 4, pp. 433-438, 1984. 0305%750X/84 $3.00 + 0.

00
Printed in Great Britain. 0 1984 Pergamon Press Ltd.

The Nature of Japan’s Comparative Advantage,


196580

MITCHELL KELLMAN
City College of New York
and
DANIEL LANDAU
University of Connecticut, Waterbury

Summary-This paper compares the dynamic factor proportions theory to the product cycle
theory in explaining Japan’s comparative advantage for the 1965-80 period. Three tests were
used: (1) domestic Japanese relative prices of more and less competitive export commodities;
(2) trends in capital-labour ratios, skill ratios, R&D intensity and product turnover for Japan’s
manufactured exports: (3) price elasticities over time of Japan’s exports. The results suggest
that: (1) the product cycle must be considered along with the factor proportions theory for the
pre-oil shock period; (2) the product cycle theory is superior for the post-oil shock period.

1. INTRODUCTION cal manufactured export bundle will become


more physical and/or human capital-intensive.
Japan is a major factor in world trade. Since Second, export success (increasing market
growth of Japan’s economy has been closely shares) will tend to be associated with lower
related to the growth of its foreign trade, under- costs of production and therefore lower prices.
standing the sources of its comparative advan- The product cycle approach implies the follow-
tage is both of interest in itself and helpful in ing two propositions. First, if a country is a
the understanding of Japanese economic growth. leader in the product cycle its export bundle
In an important paper, Peter Heller (1976) will be relatively human capital-intensive,
looked at the relationship between changes in R&D-intensive, and (in the Finger version)
Japan’s factor endowment in the 1956-69 exhibit higher product turnover. If a country
period and changes in its comparative advantage. is moving towards leadership in the product
He found that along with the increase in physi- cycle, then its exports should show increasing
cal and human capital per worker Japan’s human capital and R&D intensities plus increas-
exports became more capital- and skill-intensive. ing product turnover. Second, export success
These results indicate that a dynamic version need not be associated with low prices, but it
of the classic factor proportion theory could will be associated with falling prices over time
explain Japan’s comparative advantage.’ (relative to the general price level).2
Our paper extends Heller’s study in two ways: One might suppose that falling prices also
(1) it examines the 1965-80 period emphasiz- characterize the dynamic factor proportions
ing the post-oil price shock years; (2) it counter- theory. However, this is not strictly true. Even
poses an alternative theory, the product cycle if the price of an exportable were falling, it
(see Vernon, 1966; Hirsch, 1967; Finger, 1975). would not tend to capture international market
The two explanations of comparative advan- shares unless the price was lower than like
tage can be tested by comparing how well their commodities. Hence the essential criteria in a
respective implications fit the behaviour of factor proportions world is a relatively low
Japanese exports over the period studied. The rather than a falling price. In any case, it is not
dynamic factor proportions theory implies the true that an increased factor availability neces-
following two propositions. First, as physical sarily implies a falling price of the product that
and/or human capital endowments increase uses it intensively. The classic Rybschinsky
(relative to those of trading partners), the typi- Theorem is a clear illustration of this position;

433
434 WORLD DEVELOPMENT

it shows that increased factor availability may The second test utilized a sample of 93
lead to increased production (and exports) manufactured commodities constituting roughly
without changes in relative prices. 97% of Japan’s manufactured exports in value
The rest of the paper presents the results of terms.6 Four test statistics were examined: the
three tests of how well these four propositions capital-labour ratio (dollars per worker), the
fit Japan’s manufactured exports from the skill ratio (share of professional and technical
mid-1960s to 1980 - especially after 1973. workers in total employment), R&D intensity
(share of R&D costs in total costs) and the
Finger product turnover index (seven digit items
2. TESTS AND RESULTS appearing or disappearing from three digit SITC
groups). The four statistics for each of the 93
The first test was done on a sample of 45 commodity groups are based on US measures,
Japanese manufactured export commodities a common procedure (see Balassa, 1979;Finger,
which constitute roughly 80% of her manufac- 1975).’ While the exact degree of embodied
tured exports in value terms.3 The 45 com- R&D will naturally differ somewhat in each
modities were measured by the gain in Japan’s country, the relative commodity rankings will
market share. Market share was defined as be similar, i.e. aircraft is always ranked higher
Japan’s exports relative to total OECD exports than textiles. The value of each statistic for a
excluding Japan. The 10 most competitive - given year is an export share weighted average
largest gainers in market share - and the 10 least of the values for each of the 93 commodity
competitive - largest losers in market share - categories. The ratio of the values of the four
were then selected for two time periods 1962- test statistics for Japan’s manufactured exports
73 and 1973-77.4 to their values for the OECD manufactured
The average purchasing power parity prices exports is plotted in Figure 1.
of the two groups were then calculated using The two explanations were tested for the
the data from the World Bank-UN Income pre-oil price shock period 1965573 and the
Comparison Project (Second Stage) as reported post-oil shock period 1973-80. Japan’s rate
in Kravis et al. (1978).’ Between these domestic
prices and trade prices there will be a mark-up
reflecting a complex series of taxes, tariffs, sub-
sidies, etc. at various levels. However, barring
tariff structures which systematically discrimi-
nate against that commodity composition ex-
ported by Japan, on average over the wide range
of commodities, we would expect those goods
which are lower priced in Japan (relative to world
prices) to be lower priced in importers’ markets.
Thus the factor proportions theory suggests the
average (weighted average) price of the more
competitive group should be lower than that of
the less competitive group.
The data are in Table 1. They indicate that
the more competitive group had higher prices
inside Japan. This test is hardly conclusive, but
it does cast some doubt on the factor propor-
tions explanation of Japan’s comparative
advantage during these periods.

Table 1. Relative prices more versus less competitive


manufactured exports of Japan : two periods
(constant 1973 Yen per $1

1962-73 1973-77

More competitive 292.2 330.7

Less competitive 243.2 261.9

Sources: See text


JAPAN’S COMPARATIVE ADVANTAGE 435

of investment in physical and human capital stics. In this case the rising trend for Japan
has exceeded the average for the OECD for the alone is not indicative of a change in compara-
whole 1965-80 period.8 Therefore, both her tive advantage.
capital-labour ratio and her skill ratio were The results are in Table 2. They indicate
rising relative to the OECD. Given this, the that both for Japan’s exports and Japan’s
factor proportions theory implies rising capital- exports relative to OECD exports, all test stati-
labour and skill ratios in Japan’s exports. Over stics exhibit statistically significant rising trends
the same period Japan was gaining a reputation for the 1965-73 period. This implies both
as a technological and product innovation theories have a place in explaining Japan’s
leader. Thus the product cycle theories imply comparative advantage for this period.
increasing R&D intensity (Vernon, 1966; For the 1973-80 period only the product
Hirsch, 1967) or increasing product turnover cycle variables, R&D intensity and product
(Finger, 1975) and a rising skill ratio (all three) turnover index have statistically significant
because skilled labour is necessary both for positive trends. The skill ratio has an insignifi-
R&D and to produce new products. cant positive trend while the capital-labour
The implications of the theories were tested ratio has an insignificant negative trend. These
by their time trends over the two periods. The results suggest the product cycle theory is the
time trends were found by regressing the log of appropriate one for explaining Japan’s compara-
each test statistic on time. The regressions were tive advantage for the post-oil shock period to
run both for Japan’s manufactured exports and date.
Japan’s manufactured exports relative to OEFD The third test looks at the implication of
manufactured exports excluding Japan. This the product cycle theory that the prices of new
additional test allows for the possibility that products -- early in the product cycle - will be
manufactured exports in general may have falling over time. Harry Bowen (1982) estimated
exhibited a rising trend in one of the test stati- the elasticities over time for all three digit

Table 2. Trends in test statistics, 1965-73, 1973-80

Statistic 1965-73 1973-80


Trend RZ Trend RZ

Partel A: Japan’s exports (to world)


Capital-labour 0.010 0.642 -0.0008 0.002
ratio (3.54) (0.12)

Skill ratio 0.018 0.675 0.005 0.064


(3.81) (0.64)

R&D intensity 0.034 0.876 0.024 0.584


(7.03) (2.90)

Product turnover 0.040 0.945 0.028 0.921


index (10.99) (8.34)

Panel B: ,?atio Japan’s to OECD


exports (to world)
CapitalLlabour 0.013 0.693 -0.003 0.079
ratio (3.93) (0.72)

Skill ratio 0.011 0.840 0.0015 0.135


(6.05) (0.97)

R&D intensity 0.027 0.891 0.014 0.439


(7.57) (2.17)

Product turnover 0.029 0.924 0.019 0.870


index (9.24) (6.35)

Source: See text. Trend estimated from log X = a+bt+u where X is the statistic
and t is time. The t-statistics are in parentheses under the coefficients.
436 WORLD DEVELOPMENT

SITC commodities. He assumed equation (1) The fact that the rank correlations were more
for the time path of the price of good i where negative for Japan than the Federal Republic
ai and bi are parameters to be estimated for good of Germany, the United States, France or the
i and t is time. Then the elasticity of the price United Kingdom suggests the product cycle
of commodity i with respect to time is ei given theory is relevant to explaining Japan’s com-
by equation (2). The elasticities were used - parative advantage in 1975. However, the cor-
rather than time trends as in Table 2 - for two relations are not highly significant, therefore
reasons: (1) Bowen had estimated them, (2) they are only suggestive evidence.
the non-linear relation used to estimate the
time elasticities - equation (1) - is more appro-
priate than a linear trend for price changes over 3. CONCLUSIONS
time. This paper has used three tests to compare
the dynamic factor proportions theory to the
Pit = ai + hi/t (1) product cycle theory in explaining Japan’s
comparative advantage for the 1965-80 period.
ei = -bil(t x ai + bi). (2) The three tests were domestic Japanese relative
prices of more and less competitive export
The smaller the arithmetic value of an elasti- commodities, trends in capital-labour ratios,
city ei the more clearly can product i be identi- skill ratios, R&D intensity and product turn-
fied as a product cycle leader. The rank correla- over for Japan’s manufactured export bundle,
tion between the eis and value of net exports and price elasticities over time of Japan exports.
was calculated for 79 products (Bowen’s set The three tests are consistent with the conclu-
except for non-ferrous metals) for Japan and sions that: (1) the product cycle must be con-
four other developed countries for the year sidered along with the factor proportions theory
1975.9 The same rank correlation was also in explaining Japan’s comparative advantage in
calculated for the change in net exports between, the pre-oil price shock years; (2) the product
1970 and 1975. The results are in Table 3. The cycle theory is superior in explaining her com-
negative correlation coefficient for Japan means parative advantage in the post-oil shock years.
the commodities with both larger export values These conclusions must of course be considered
and larger increases in export values - 1970- suggested not ‘proven’ and obviously there are
75 ~ tended to have arithmetically smaller no implications about the comparative advan-
eis. This would fit the product cycle theory. tage of other countries.

Table 3. Rank correlations between Bowen’s price elasticities over


time and export volumes and changes in export volumes for Japan
and ,four other developed countries

Japan FRG us France UK

Exports 1975 -0.19 -0.07 0.00 0.06 0.10

Changes in exports
1970-75 -0.15 -0.08 -0.07 0.09 0.07

Sources: See text.

NOTES

1. Balassa has given a general statement of the dynamic 3. The commodities are defined at the three digit
version of the classic Hecksher-Ohlin factor pro- level classification. A list is available from the authors.
portions theory and tested it for a broad sample of
countries in Balassa (1979). 4. The division at 1973 is motivated both by the oil
price shock in that year and also the purchasing power
2. The product cycle theory was pioneered by Vernon parity prices from Kravis et al. (1978) are for 1973.
(1966) and Hirsch (1967). Finger’s approach is found
in l;inger (1975). A recent discussion of the product 5. The consumption categories of the ICP had to be
cycle theory is found in Soete (1981). There is also a translated into the SITC commodity categories. The
demand side approach to comparative advantage, pairing used is found in Table Al.
Linder (1961), which we have not tested.
JAPAN’S COMPARATIVE ADVANTAGE 431

6. This is the 93 commodity sample used by Branson trade bundles of Japan and Japan/OECD for the years
and Monoyios (1977). 1965-80 are found in Table A2.

7. The sources of the values for the four test statistics 8. See World Tables, Comparative Economic Data,
by commodity groups are as follows: capital-labour Table 2, and Social Indicators, Table 4.
ratio, Branson and Monoyios, Appendix A; skill ratio,
Bowen: R&D intensity, Kelly; and product turnover 9. Bowen’s time elasticities of prices differ little in
index, Finger. The trade shares are from OECD (tapes) numerical value between 1958 and 1970, so using his
trade by commodities, Series C. The values for the estimates in 1975 should not be a problem.

REFERENCES

Balassa, B., ‘The changing pattern of comparative The Technology Factor in International Trade
advantage in manufactured goods’, Review ofEco- (NBER, 1970).
nomics and Statistics (May 1979), pp. 259-266. Kelly, R., ‘The impact of technological innovation
Bowen, Harry, Paper presented at the Western Eco- on international trade patterns’, Staff Economic
nomic Association meetings (San Francisco, 1982). Report ER-24 (Washington, D.C.: US Department
Branson, W. and N. Monoyios, ‘Factor inputs in U.S. of Commerce, December, 1977).
trade’, Journal of International Economics (1977), Kravis, I., A. Heston and R. Summers, International
pp. 111-131. Comparisons of Real Product and Purchasing
Finger, J., ‘A new view of the product cycle theory’, Power (Baltimore: Johns Hopkins University Press,
Weltwirtschaftliches Archiv, No. 1 (1975), pp. 1978).
79-99. Linder, S., An Essay on Trade and Transformation
Heller, P., ‘Factor endowment change and compara- (Stockholm: Almquist & Wiksell, 1961).
tive advantage: the case of Japan, 195661969’, Soete, Luc, ‘A general test of technological gap trade
Review of Economics and Statistics (August 1976) theory’, Weltwirtschaftliches Archiv, Vol. 117,
pp. 283-292. No. 4 (1981) pp. 6388659.
Hirsch, Seev, Location of Industry and International Tatemoto, M., and S. Ichimura, ‘Factor proportions
Competitiveness (Oxford University Press, 1967). and foreign trade: the case of Japan’, Review of
Hirsch, Seev and Ruth Arad, ‘Determination of trade Economics and Statistics (November 1959), pp.
flows and choice of trade partners: reconciling the 442-446.
Heckscher-Ohlin and Burenstam-Linder models Vernon, Raymond, ‘International investment and
of international trade’, Weltwirtschaftliches Archiv, international trade in the product life cycle’,
Vol. 117 (1981) p. 276-297. Quarterly Journal of Economics. Vol. S (1966).
Hufbauer, C., ‘The impact of national characteristics World Bank, World Tables, 2nd Edn (Baltimore:
and technology on the commodity composition Johns Hopkins University Press, 1980).
of trade in manufactured goods’. in R. Vernon,

APPENDIX
Table Al. Concordance from ICP categories to SITCgroups, 45 commodities

SITC ICP
No. Description NO. Description
521 Mineral tar and crude chemicals 06.230 Gasoline, oil, grease
541 Medical and pharmaceutical products 05.110-120 Drugs and medical supplies
652 Woven cotton fabrics ’ 02-110 Clothing materials
696 Cutlery 04.400 Household untensils
712 Agricultural machinery 15.220 Other agricultural machinery
714 Office machines 15.300 Office machinery
715 Metal working machinery 15.400 Metal working machinery
718 Machines for special industries 15.600 Special industrial machinery
719 Machines and appliances NES 15.700 General industrial machinery
(not electric)
122 Electric power machinery 16.100 Electrical transmission
equipment
123 Equipment for distributing electricity 12.200 Transport and utility lines
724 Telecommunications apparatus 16.200, Communications equipment
07.110 Communications equipment
438 WORLDDEVELOPMENT

Table Al. (continued)

SITC ICP
NO. Description No. Description
725 Domestic electrical equipment 04.310 Domestic appliances
04.350 Domestic appliances
726 Medical and radiological 05.200 Therapeutic equipment
electric apparatus
129 Other electrical machinery 16.300 Other electrical equipment
731 Railway vehicles 14.110 Locomotives
132 Road motor vehicles 14.200- Automobiles
300 Automobiles
06.110 Automobiles
133 Other road vehicles 06.330 Miscellaneous transport
134 Aircraft 14.400 Aircraft
735 Ships and boats 14.500 Ships and boats
821 Furniture 14.110 Furniture
17.110 Furniture
831 Travel goods, handbags, etc. 08.220 Other personal care goods
841 Clothing (except fur) 02.121-3 Clothing
03.131-2 Clothing
02.150 Clothing
851 Footware 02.211- Footware
13 Footware
861 Scientific, medical, measuring 16.400 Instruments
apparatus
862 Photographic supplies 07.130 Other recreational equipment
891 Musical instruments 07,.110 Phonographs, radios, televisions
894 Toys and sporting goods 07.120 Recreational equipment

Table A2. Capital-labour ratio, skill ratio, R&D intensity and product turnover
index, Japan’s manufactured exports plus the ratio of the statistics for Japan’s
manufactured exports to the statistic for OECD manufactured exports, 1965-80

Capital-labour Product turnover


ratio Skill ratio R&D intensity index
Year Japan Ratio Japan Ratio Japan Ratio Japan Ratio

1965 $11700 84.0 8.4% 87.8 2.25% 76.5 7.51 82.6


1966 11440 81.9 8.6 88.8 2.49 83.0 7.59 82.1
1967 11180 80.1 8.8 90.0 2.60 84.7 7.83 83.6
1968 11560 82.2 8.8 89.0 2.14 87.7 8.31 87.2
1969 11730 85.3 8.8 88.8 2.88 92.2 8.76 90.6
1970 12130 81.6 8.9 91.0 2.91 92.5 9.22 95.2
1971 12370 90.5 9.3 93.2 2.87 90.8 9.79 99.2
1972 12180 89.4 9.5 95.4 3.03 96.5 9.92 100.4
1973 12170 89.0 9.6 96.4 3.03 96.9 9.71 98.4
1974 13720 95.1 9.5 96.3 2.66 97.7 9.62 98.0
1975 13320 95.6 9.8 97.1 2.72 89.0 9.79 100.4
1976 12710 91.8 9.9 97.2 3.03 96.2 10.00 100.5
1977 12690 92.5 10.0 98.4 3.08 97.8 10.30 lU1.4
1978 12830 91.8 10.0 98.9 3.22 100.0 10.70 104.0
1979 13160 92.2 9.8 96.2 3.23 99.9 11.30 108.4
1980 12690 89.3 9.8 97.3 3.26 99.0 11.80 112.6

Sources: See text especially note 6. Capital-labour ratio in thousands of dollars


per worker for Japan. Product turnover index in units for Japan. Skill ratio and
R&D intensity for Japan and all ratios of Japan to OECD are in per cent.

You might also like