Factor Intensity Reversals Redux∗ †

‡

Kozo Kiyota

Yoshinori Kurokawa

Keio University and RIETI

University of Tsukuba

First Version: September 15, 2016 This Version: August 7, 2017

Abstract Factor intensity reversal means that an industry is relatively capital intensive compared with other industries within a country/region but relatively labor intensive within another country/region. Though theoretically interesting, factor intensity reversals have been abandoned in empirical studies since the 1960s. The main reason for this unpopularity is that little evidence for factor intensity reversals among countries or regions has been found in previous empirical studies. Newly developed region-level data, however, suggest that the abandonment of factor intensity reversals in the empirical analysis has been premature. Specically, we nd that the degree of the factor intensity reversals is higher than that found in previous studies on average and has increased over the last two decades. It is higher when we use disaggregated industry-level data, weakening a possible criticism that several factor intensity reversals may be a result of the aggregation of industries. Finally, our regression results suggest dierences in exporter concentration across regions in each industry as a possible reason for the observed factor intensity reversals.

Keywords: Factor intensity reversals, Capital, Labor, Region, Japan, Exporters JEL Classications: F11, F14

∗

We are very grateful to Tomohiro Ara, Hiroki Arato, Rudolfs Bems, Ichiro Daitoh, Josh Edering-

ton, Kyoji Fukao, Taiji Furusawa, Makoto Hasegawa, Kaoru Hosono, Toshihiro Ichida, Tomohiko Inui, Jota Ishikawa, Hirokazu Ishise, Arata Ito, Katsuhito Iwai, Hiroyuki Kasahara, Hayato Kato, Fukunari Kimura, David Kuenzel, Bingjing Li, Gang Li, Kiyoshi Matsubara, Daisuke Miyakawa, Koyo Miyoshi, Masayuki Morikawa, Hiroshi Mukunoki, Yasusada Murata, Kentaro Nakajima, Ryo Nakajima, Eiichi Nakazawa, Michio Naoi, Ayako Obashi, Masao Ogaki, Michihiro Ohyama, Raymond Riezman, Hisamitsu Saito, Yukiko Saito, Chisato Shibayama, Daichi Shirai, Yoshimasa Shirai, Yoichi Sugita, Michio Suzuki, Hajime Takatsuka, Miho Takizawa, Jumpei Takubo, Mari Tanaka, Kimiko Terai, Eiichi Tomiura, Akihiko Yanase, Makoto Yano, Miaojie Yu, and Lianming Zhu for their useful comments. We also thank the participants in seminars and conferences at Keio, RIETI, Tokyo Tech, the Japan Society of International Economics KANTO, WITS, Spring 2017 Midwest International Trade Conference, 2017 JEA Spring Meeting, and Ryukyu Economics Workshop. This study was conducted as part of the Project Microeconomic Analysis of Firm Growth undertaken at the Research Institute of Economy, Trade and Industry (RIETI). The study uses the micro data from the questionnaire in the Census of Manufacture, which is conducted by the Ministry of Economy, Trade and Industry (METI). Kiyota gratefully acknowledges the nancial support of Japan Society for the Promotion of Science (JSPS) Grant-in-Aid (JP16H02018). The usual disclaimers apply. † Address: 2-15-45 Mita, Minato-ku, Tokyo 108-8345 Japan; Tel.: +81-3-5427-1597; E-mail address: [email protected]. ‡ Address: 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8571 Japan; Tel.: +81-29-853-7426; E-mail address: [email protected].

1

In connection with the two factor case, I have the impression that the phenomenon of goods that interchange their roles of being more labor intensive is much less important empirically than it is interesting theoretically (Samuelson, 1951, pp. 12122).

1

Introduction

Factor intensity reversal means that a good/industry is relatively capital intensive compared with other goods/industries within a country/region but relatively labor intensive

1

compared with other goods/industries within another country/region.

As quoted in

the opening, Samuelson's (1951) well-known view is that factor intensity reversals are of theoretical interest rather than empirical importance. In fact, little evidence for factor intensity reversals among countries or regions has been found in previous empirical

2 Therefore, factor

studies (e.g., Fuchs, 1963; Leontief, 1964; Ball, 1966; Moroney, 1967).

intensity reversals have been abandoned in empirical analysis since the 1960s. This issue of whether factor intensity reversals exist or not is an important issue particularly in the analysis of the HeckscherOhlin model. This is because all of the major four theorems of the standard HeckscherOhlin model (i.e., the StolperSamuelson theorem, the Rybczynski theorem, the factor price equalization theorem, and the Heckscher Ohlin theorem) assume no factor intensity reversals.

Some trade economists, such as

Bhagwati and Dehejia (1994) and Feenstra (2015), have noted that factor intensity re-

3 So far, however, the empirical studies

versals might not be a mere theoretical curiosum.

on the HeckscherOhlin model have ruled out the possibility of factor intensity reversals

4

because little evidence has been found as mentioned above.

Recently, however, the availability of the data on capital and labor has been improved signicantly compared with that in the 1960s. It is thus worth revisiting the issue of the factor intensity reversal controversy in the 1960s with new data on capital and labor. In fact, newly developed region-level data for the period 19732009 suggest that the abandonment of factor intensity reversals in the empirical analysis has been premature.

1

In this paper, the expression factor intensity reversal refers to a capital intensity reversal and is

distinguished from a skill intensity reversal. Of course, there are factors of production other than capital and labor, such as land, but we focus on capital and labor, as in neoclassical trade models such as the HeckscherOhlin model. We note, however, that even in the case of more than two factors, we can still dene capital intensity as the capital/labor ratio, as in the case of two factors. Moreover, the meaning of factor intensity reversal in such a case remains the same as that in the case of two factors: the ranking of capital intensity among sectors is not the same among countries/regions. For example, see Wong (1990) for factor intensity reversals in the case of multiple factors.

2

At rst sight, Minhas (1962) seemed to show evidence of factor intensity reversals using both para-

metric tests (i.e., estimates of elasticity and distribution parameters of production functions) and nonparametric tests (i.e., examination of the rank correlation of capital intensities). However, his parametric test was criticized by Fuchs (1963) and Leontief (1964) because the test results are sensitive to the ratio of the distribution parameters and the specication of the production function. For example, Fuchs (1963) showed that the estimated elasticities of substitution were less dispersed if the production function includes a dummy variable that allows for the dierences between developed and developing countries. The nonparametric test was also criticized by Ball (1966) and Moroney (1967). For example, Ball (1966) showed that the test results were sensitive to whether or not the agricultural industry is included.

3

Although he does not empirically test the existence of factor intensity reversals, Feenstra (2015)

considers the footwear industry as an example.

The footwear industry is relatively capital intensive

compared with other industries within the United States but relatively labor intensive within Asia. In addition to the footwear industry, Feenstra and Taylor (2014) consider the agriculture sector as an example.

4

For example, Tomiura (2005), Bernard, Redding, Schott, and Simpson (2008), and Bernard, Red-

ding, and Schott (2013) examined whether factor price equalization exists in Japan, the United Kingdom, and the United States, respectively. These studies found that factor price equalization did not exist even within a country. Similarly, Kiyota (2012) conrmed that in Japan, the average manufacturing wage rate in Kanagawa prefecture is almost twice as high as that in Aomori prefecture. However, all of these studies ignored the possibility of factor intensity or skill intensity reversals.

2

Table 1, which shows the capital/labor ratios for the manufacturing industries in 47 Japanese prefectures in 2005, indicates that several factor intensity reversals might

5 Note that a Japanese prefecture corresponds to a

have existed in these industries.

US state. The industries and the prefectures are sorted in order of capital intensity and relative capital abundance, respectively. The color of each cell indicates the capital intensity of a given industry in a given prefecture.

Light gray, gray, dark gray, and

black mean that the capital intensity is in the rst, second, third, and fourth quartiles within each prefecture, respectively.

Here, we emphasize that the color indicates not

absolute capital intensity but relative capital intensity within each prefecture. If there is no factor intensity reversal, cells will become darker from left to right in the same way in all prefectures. That is, the order of industry capital intensities will be the same in all prefectures.

6 As can be seen, however, Table 1 indicates the existence of factor

intensity reversals. For example, transportation machinery was more capital intensive than were pulp and paper in Aichi, where Toyota is located. In contrast, pulp and paper were more capital intensive than was transportation machinery in Hokkaido, where many

7 Similarly, electrical machinery was more

plants of large paper companies are located.

capital intensive than was transportation machinery in Nagasaki, whereas transportation machinery was more capital intensive than was electrical machinery in Kyoto. === Table 1 === This paper thus examines whether or not factor intensity reversals indeed existed, using prefecture-level data in Japan over the period 19732009. An advantage of using Japanese prefecture-level data is that identical technology across prefectures is plausible within a country as compared with the situation across countries. One of the key assumptions in the HeckscherOhlin model is that technology is identical across countries or regions.

8 If one industry in a country/region employs dierent production technology

from an industry in another country/region, it is impossible to classify the industry as the same industry. Indeed, Harrigan (1997) found that technology dierences as well as factor supplies were important determinants of the international specialization of production. Bernstein and Weinstein (2002) pointed out that the use of international data was sometimes subject to problems such as measurement error and government policy. The use of national data can overcome some of these problems. Bernstein and Weinstein (2002) and Kiyota (2012) used Japanese regional data to test the empirical validity of the HeckscherOhlin model. Indeed, Moroney (1967) also used US regional data in 1957

5

In Section 2, we present a more detailed description of the data. Kiyota (2012) also showed a similar

table for Japan in 2000, although his focus was not on factor intensity reversals but on the existence of multiple cones of diversication.

6

See Table A1 in Appendix A for a hypothetical example of no factor intensity reversal. The well-

known theoretical explanation for no factor intensity reversal is that the isoquant curves of two industries have only one intersection and thus only one diversication cone exists. Note, however, that even if the isoquant curves have more than one intersection and thus multiple diversication cones exist, no factor intensity reversal occurs when all prefectures are located in the same cones.

7

It is worth pointing out that as shown in Table 1, the ascending order of capital/labor ratio is

Kyoto, Hokkaido, and Aichi.

However, in Kyoto, as in Aichi, transportation machinery was more

capital intensive than pulp and paper.

This implies that the rankings of capital intensities of these

two industries reverse from Kyoto to Hokkaido and then re-reverse from Hokkaido to Aichi. A possible explanation of it is that the isoquant curves of two industries have more than two intersections, and thus more than two diversication cones exist. Then Kyoto, Hokkaido, and Aichi are located in dierent cones, respectively.

8

By identical technology, we mean that for the same industry, countries/regions have the same

isoquant curve, that is, technological knowhow.

However, if the isoquant curves of two industries

have more than one intersection, then the same industry can be dierent in capital intensity between countries/regions in equilibrium, even though technological knowhow is identical. Bhagwati and Deheja (1994) interpret this dierence as a

de facto

dierence in technology.

3

in examining the existence of factor intensity reversal. Another advantage of the use of Japanese prefecture data is that, as we will see in Section 2, real capital stock and labor inputs data are available at the prefecture-industry level in Japan. While such data are available at the country level, to the best of our knowledge, they are not available at the

9 This study thus focuses on Japan.

state or prefecture level in many countries.

In contrast, there is a disadvantage in that the factors are more mobile in a crossregion analysis than in a cross-country analysis.

One of the key assumptions in the

HeckscherOhlin model is that there is no mobility of factors across countries or regions.

10 Ac-

It is fortunate, however, that domestic migration rate is relatively low in Japan.

cording to the Ministry of Internal Aairs and Communications (2000), the migration

11

rate of manufacturing workers among prefectures was 6.6 percent from 1995 to 2000.

This means that the annual domestic migration rate in Japan was about 1 percent, which is almost the same as the international migration rates of some OECD countries, such

12 This low domestic migration rate thus implies low work-place mobility 13 across prefectures, i.e., low labor mobility in Japan. as Switzerland.

The contribution of our paper is as follows. We revive and add to the factor intensity reversal literature. Recently, some studies such as Kurokawa (2011) and Sampson (2016) have documented empirically that there exist skill intensity reversals: a good/industry is relatively high-skill intensive within a country but relatively low-skill intensive within

14 These studies, however, focused on skill intensity rather than capital

another country. intensity.

Thus, although they provided the fresh viewpoint of a skill division among

labor to the factor intensity reversal literature, these studies did not directly tackle the issue of factor intensity reversals, which were controversial in the 1960s. We now revive this issue with new data on capital and labor. Moreover, while the results by Minhas (1962) have been criticized and rejected, we present strong evidence, weakening the

15

criticisms.

We also perform several robustness checks: 1) the sample includes agriculture and mining industries; 2) 47 prefectures are aggregated into eight regions; 3) the analysis takes human capital into account; 4) we compare dierent years; and 5) the industries are disaggregated at the four-digit level.

In particular, we emphasize that the factor

intensity reversals are stronger when we use the disaggregated industry-level data than when we use the aggregated data. We can thus weaken the possible criticism that several factor intensity reversals may be a result of the aggregation of industries. Finally, we consider possible reasons for the observed factor intensity reversals among prefectures. As a possible reason, our regression results suggest intra-industry hetero-

9

Note that because the US Bureau of Economic Analysis provides data on the net capital stock for

the nation but not for individual states, Garofalo and Yamarik (2002) and Yamarik (2013) estimated state-level capital stock. Their estimates, however, are not at the state-industry level but at the stateaggregate-level.

10

We acknowledge that capital mobility is not low compared with labor mobility in Japan.

Using

Japanese prefecture-level data, however, can still be compatible with the HeckscherOhlin model assuming factor intensity reversals. In such a model, the isoquant curves of two sectors have more than one intersection, and multiple diversication cones exist. countries/regions that are located in dierent cones.

Then the rental/wage ratios dier among

In that case, even if we allow capital mobility

across countries/regions, as long as labor mobility is low, it is possible that the countries/regions remain in dierent cones and thus dierences in the rental/wage ratios remain. In fact, the standard deviation of the capital/labor ratios for 47 prefectures in Japan increased over 19752005.

11 12 13

The migration rate refers to the inows divided by the total labor force in manufacturing. For more details, see OECD (2006, p.32, Chart I.1.). Note that there is evidence that labor mobility across regions is slow and incomplete even in the

United States (e.g., Autor, Dorn, and Hanson, 2016). As noted by Blanchard and Katz (1992), there is also evidence for low labor mobility in Britain, Italy, and Germany.

14 15

Reshef (2007) also seriously considered the possibility of skill intensity reversals. For criticisms of Minhas's (1962) results, see footnote 2.

4

geneity, in particular, dierences in exporter concentration across prefectures in each industry. The rest of this paper is organized as follows.

Section 2 describes the data and

methodology used in this paper. Section 3 presents the results, and Section 4 investigates intra-industry heterogeneity. Section 5 concludes the paper and discusses opportunities for future research.

2

Data and Methodology

2.1 Data We use the Regional-Level Japan Industrial Productivity (R-JIP) Database 2014 for

16 It is a region/prefecture-level version of the Japan

real capital stock and labor data.

Industrial Productivity Database, which has been widely used in several studies (e.g., Dekle et al., 2010; Dekle et al., 2015). It provides us with annual information on capital and labor inputs, as does the National Bureau of Economic Research manufacturing database. One of the notable features of the database is that the information is available at the prefecture-industry level. The data cover 47 prefectures in Japan for the period from 1970 to 2009. Note again that a prefecture in Japan corresponds to a state in the United States.

The data include 13 manufacturing industries and the agriculture

and mining industries.

17 The 47 prefectures are aggregated into eight regions.18

In the R-JIP Database 2014, capital stock is dened as the net real capital stock. The unit of measurement is one million Japanese yen (2000 constant prices). Labor is measured as man-hours (i.e., number of workers times working-hours per worker divided by 1,000).

19 All the inputs are identied at the work-place. Section 3.1 will focus on the

year 2005 as in Table 1. In Section 3.2, we examine the period 19732009. Note that the reason that we use the data from 1973 is that Okinawa prefecture was returned to Japan in 1972.

2.2 Methodology Using the data on capital and labor from the R-JIP Database 2014, we rst calculate

20 We then calculate Spearman's rank

capital intensity by industry and by prefecture. correlations of industry capital intensity,

ρ.

ρ,

for all prefecture pairs, and their mean,

Spearman's rank correlation presents the correlation of rankings of capital intensity

between two dierent prefectures, which is dened as:

ρ=1− where

∑ 6 d2 , n(n2 − 1)

(1)

d is the dierence between the two ranks of each observation, and n is the number 21 It takes values from −1 to 1. The value 1 indicates a perfect agreement

of observations.

among rankings of capital intensity between two prefectures, whereas the value 0 indicates no agreement; and the value words, the smaller the value of

ρ,

−1

indicates a perfect negative association. In other

the stronger the factor intensity reversals between two

prefectures.

16 17 18 19 20

The data are available at http://www.rieti.go.jp/jp/database/R-JIP2014/index.html. See Table A2 in Appendix A for the classications of prefectures and industries. See Table A2 for the region classication. For more detailed explanations on how to measure capital and labor, see Tokui et al. (2013). As we have noted in footnote 1 in Section 1, even if we add factors other than capital and labor to

our analysis, we can similarly dene/calculate capital intensity as the capital/labor ratio, and discuss capital intensity reversals among prefectures.

21

Note that this formula assumes that all ranks are distinct in each prefecture.

5

Following Moroney (1967), we also calculate Kendall's coecient of concordance,

W.

W is another useful statistic for measuring the uniformity of rankings among m (m > 2) sets of rankings. It takes values from 0 (no agreement among ranks) to 1 Kendall's

(perfect agreement). It can easily be calculated by using the following linear relationship with the mean of the Spearman's rank correlations,

ρ=

ρ:22

mW − 1 . m−1

(2)

It should be noted that instead of a parametric approach, this paper takes a nonparametric approachSpearman's rank correlationsto measure the degree of the factor

23 There are two main reasons for this. First, we ensure the compa-

intensity reversals.

rability of our ndings with previous studies by following the nonparametric approach taken by Minhas (1962) and Moroney (1967). Minhas (1962) showed that Spearman's rank correlation of capital intensities for 20 industries between Japan and the United States was 0.730.

Moroney (1967) analyzed factor intensity reversals among regions

in the United States and found higher rank correlations (0.87740.9074) than those of Minhas (1962). As can be seen, as in our paper, Moroney (1967) also analyzed factor intensity reversals at the region level. Japan's prefectures.

While he focused on US regions, we focus on

Second, by taking a nonparametric approach, our results do not

depend on the specication of the production function. Note also that there is no single criterion for the correlations regarding whether factor intensity reversals exist or not. Because the previous studies often referred to the correlations reported by Minhas (1962) (i.e., 0.730) and Moroney (1967) (i.e., 0.8774

24

0.9074), we also consider these values as reference values.

3

Empirical Results

3.1 Evidence for 2005 As we have seen in Section 1, Table 1 indicates that several factor intensity reversals might have existed among the manufacturing industries of the 47 prefectures in Japan in 2005. To determine the degree of the factor intensity reversals, we calculate Spearman's

ρ, for all prefecture pairs. (= 46 + 45 + . . . + 1). We then obtain the rank correlations,

0.645 (the standard deviation is 0.186).

Here, the number of prefecture pairs is 1,081 mean of Spearman's rank correlations,

ρ,

of

This is much lower than the values obtained

by Moroney (1967), who concluded that few factor intensity reversals existed among regions in the US manufacturing industries in 1957: 0.8774 (six regions & 14 industries) and 0.9074 (ve regions & 16 industries). Note that the lower the value of

ρ, the stronger

the factor intensity reversals. This value of 0.645 indicates that several reversals existed in 2005 because the value is even lower than the value obtained by Minhas (1962) (i.e., 0.730), who argued that several

25

reversals existed.

While his results have been criticized and rejected as mentioned

in Section 1, our results can withstand such criticisms.

Thus, we can no longer say

that few factor intensity reversals existed among the manufacturing industries in the 47

22 23

See p. 315 in Agarwal (2007). As a reference, in Appendix B we also show results by a parametric approach (i.e., estimates of

elasticity and distribution parameters of production functions).

24

Minhas (1962) argued that the dierence between unity and .730 is large enough to suggest that

reversals in relative capital intensity do exist (p.148).

25

We also compute the 95 percent condence interval, assuming that

ρ

follows a normal distribution.

The 95 percent condence interval is between 0.634 and 0.656 (the number of observations is 1,081, and the standard error is 0.006), implying that the rank correlation obtained in this study is signicantly lower than that of the previous studies.

6

prefectures of Japan in 2005. We also calculate Kendall's

W

of 0.652, which is also much

lower than the values obtained by Moroney (1967) (i.e., 0.89550.9228). This reconrms our above argument based on

ρ.

One may argue that as shown in Table 1, factor intensity reversals seem to be observed mainly in the four machinery industries (general machinery, electrical machinery, transportation machinery, and precision machinery) and thus our results might be driven by these machinery industries. To address this concern, we aggregate these four machinery industries into one industry (the total number of industries is now 10) and calculate the rank correlations for the 47 prefectures. The mean of Spearman's rank correlations,

ρ, for 2005 increases to 0.753 (the standard deviation is 0.171).

Although higher than the

value before aggregation, it is still comparable to that of Minhas (1962) and lower than the values obtained by Moroney (1967). Our main messages therefore are not changed by the aggregation of the machinery industries.

3.1.1

Inclusion of the agriculture and mining industries

Ball (1966) found that the factor intensity reversals estimates of Minhas (1962) were sensitive to whether or not the agricultural industry is included. Specically, the rank correlation increased considerably if the analysis excluded one agricultural industry (i.e., from 0.732 for all industries to 0.833 for manufacturing). One may thus be concerned that our results are sensitive to the inclusion of agricultural and mining industries. To address this concern, we add the agriculture and mining industries to the previous analysis that focused only on the manufacturing industries in 47 prefectures in 2005. Table 2 shows the capital intensities for all industries including agriculture and mining in 47 prefectures in 2005. The mean of Spearman's rank correlations,

ρ, is now 0.649 (the

standard deviation is 0.171), which is slightly higher than the previous value of 0.645, with the focus only on the manufacturing industries. In other words,

ρ for the case of only

the manufacturing industries is lower than that for the case including the agriculture and mining industries. This lower value of Spearman's rank correlation implies that factor intensity reversals are more prevalent when we focus only on the manufacturing industries than when we include the agriculture and mining industries. === Table 2 === Moreover, Kendall's

W

is now 0.657, which is also slightly higher than the value

of 0.652 when the analysis focuses only on the manufacturing industries.

This again

indicates that focusing only on the manufacturing industries does not show fewer factor intensity reversals. Interestingly, while previous studies such as Ball (1966) showed that the case of only the manufacturing industries showed fewer factor intensity reversals, our Japanese data indicate the opposite pattern.

3.1.2

Aggregation of prefectures

Another concern may be that our results are sensitive to the aggregation of prefectures. For example, Moroney (1967) focused on ve or six aggregated regions in the United States, whereas our study focuses on 47 disaggregated prefectures. The aggregation of the prefectures may aect the results. To address this concern, we repeat the analysis in the previous sections, with eight aggregated regions. Here, the number of region pairs is 28 (=

7 + 6 + . . . + 1).

Tables 3 and 4 present the results for the aggregated eight-region counterparts of Tables 1 and 2, respectively. Tables 3 and 4 indicate that few factor intensity reversals might have existed among the eight regions in 2005. In fact, the mean of Spearman's

7

rank correlations,

ρ,

is now 0.831 (the standard deviation is 0.097) for the case of only

the manufacturing industries; it is 0.844 (the standard deviation is 0.079) for the case including the agriculture and mining industries. These values are higher than those in Minhas (1962) and are close to those in Moroney (1967), although still smaller.

The

results indicate that in 2005, the degree of the factor intensity reversals was lower at the aggregated-eight-region level than at the 47-prefecture level but still not negligibly small. Kendall's

W

also indicates similar patterns. It is 0.853 for the case of only the

manufacturing industries, while it is 0.864 for the case including the agriculture and mining industries. These values are also close to those in Moroney (1967), although still smaller. === Tables 3 & 4 ===

3.1.3

Human capital

One may be also concerned that our results are driven by the dierences in human capital across prefectures because man-hours, which is our measure of labor input, does not take

26 For example, consider the case in which

into account the dierences in human capital.

the capital/labor ratio of an industry is relatively large within prefecture small within prefecture

B.

A but relatively A is

If, however, the skill level of labor used in prefecture

higher and thus the labor productivity is higher, then it is possible that for the industry in prefecture

A, man-hours employed is smaller and thus the capital/labor ratio is larger.

This example implies that the same number of man-hours does not necessarily mean the same level of human capital. Therefore, to take into account crudely the dierences in human capital across prefectures, here we use total wages rather than man-hours as the measure of labor input. This approach is also employed by Hsieh and Klenow (2009) to take into account the dierences in hours worked and human capital. The data on total wages by prefecture and industry are also available in the R-JIP Database.

Both wages and the number

of workers are identied at the work-place. The unit of measurement of total wages is millions of Japanese yen. Table 5 presents the results for the manufacturing industries in the 47 prefectures for 2005. As in Table 1, actual capital intensities are dierent from the pattern presented in Table A1.

This suggests the existence of factor intensity reversals.

Spearman's rank correlations,

ρ,

The mean of

is 0.503 (the standard deviation is 0.231), which is

lower than that in Table 1. These results together suggest that our main results hold

27

even when we take into account the dierences in human capital. === Table 5 ===

3.2 Evidence for 19732009 In Section 3.1, we found that in 2005, factor intensity reversals were less severe among the eight aggregated regions of Japan, focusing only on the manufacturing industries and in the case including the agriculture and mining industries.

However, factor intensity

reversals were prevalent among the 47 prefectures in the same year in both cases, but more existed in the former case. In this section, to determine whether the above results also hold for other years, we construct a table that shows the mean of Spearman's rank

26 27

Although it is dicult to dene quality of labor, here we dene it as equal to human capital. It is worth mentioning that although their focus is not on factor intensity, Bernard et al. (2013)

empirically show that relative factor prices are not equal across US regions even when they take into account the dierences in factor productivity or quality across regions.

8

correlations, Kendall's

ρ,

and its standard deviation for each of the years 19732009 as well as

W.

The analysis consists of 47-prefectures level and eight-aggregated-regions level analyses. In each analysis, we compare two cases: (1) only the manufacturing industries and (2) the manufacturing plus agriculture and mining industries. We rst present the 47-prefectures level analysis. Table 6 shows that

ρ

ranges from

0.603 to 0.750 for the case focusing only on manufacturing and from 0.612 to 0.753 for the case including agriculture and mining, over the period 19732009. shows that Kendall's

W

The table also

ranges from 0.611 to 0.755 for the former case and from 0.620 to

0.759 for the latter case, over the same period. The values of

ρ

and

W

are much smaller

than those obtained by Moroney (1967) and even smaller than those obtained by Minhas (1962). Thus, the results indicate that several factor intensity reversals existed among the 47 prefectures in both cases during 19732009, but more existed in the case of only manufacturing. Moreover, it should be emphasized that the degree of the factor intensity reversals among the 47 prefectures has increased in the last two decades as indicated by both

ρ

and

W,

which have decreased since 1985. === Table 6 ===

We next present the eight-aggregated-regions level analysis. Table 7 shows that both

ρ and W

exceeded 0.8 in both cases during the period 19732009, and they even exceeded

0.9 in some years. These values are close to those obtained by Moroney (1967). Thus, like previous studies, our results indicate that factor intensity reversals were less severe among the eight aggregated regions over 19732009. === Table 7 === It is worth pointing out that as shown in Table 6, the standard deviation of Spearman's rank correlations for all prefecture pairs has also increased recently, as has the mean of

ρ,

both in the case focusing only on manufacturing and in the case including

agriculture and mining. Figure 1 shows the distribution of Spearman's rank correlations for all prefecture pairs for the years 1975, 1985, 1995, and 2005, with the focus on the manufacturing industries. As can be seen, the distribution across prefecture pairs was concentrated from 1975 to 1985, but it was dispersed from 1985 to 2005. This indicates that dierences in the degree of the factor intensity reversals between prefecture pairs have increased recently. Figure 2 shows similar changes in the distribution for the case including the agriculture and mining industries. === Figures 1 & 2 === It is also worth pointing out that though unreported in Table 6, our main results hold for 19732009 even when we take into account the dierences in human capital as in Section 3.1.3. The results show that

ρ

ranges from 0.451 to 0.614 for the manufacturing

industries of the 47 prefectures during 19732009 and that it has shown a decreasing trend during the period.

4

Intra-industry Heterogeneity

4.1 Disaggregation of industries One may also be concerned about intra-industry heterogeneity because our 13 manufacturing industry classications might be too aggregated to address the issue of identical

9

technology across prefectures. For example, transportation machinery includes not only automobiles but also other transportation machines such as trains, ships, and airplanes. If dierent prefectures specialize in dierent products within an industry, it may be nat-

28

ural to nd dierences in factor intensity.

Thus, a possible criticism is that factor

intensity reversals among prefectures may be found because of the aggregation of industries; in other words, if industries are disaggregated, then fewer factor intensity reversals

29

among prefectures may be found.

To address this concern, here we use condential plant-level data from the Census of Manufacture published by the Japanese Ministry of Economy, Trade and Industry. These data come from an annual census that is compulsory for plants with more than three employees.

For plants that have at least 30 workers, it records information on

tangible assets and number of workers. Although it is an annual census, we cannot trace the same industry throughout the period because of the revisions of the industry classications.

In fact, other than the

R-JIP Database there are no time-series data on real capital stock and labor that use the same industry classications over time at the prefecture-industry level. We thus focus on the year 2005. The industry classications are available at the four-digit industry level: 560 manufacturing industries in 2005. Note that the information on tangible assets is not available for plants that have fewer than 30 workers.

As a result, information on

tangible assets and the number of workers is available for 552 manufacturing industries in 2005.

30

Table 8 presents the results. Capital stock is measured as nominal tangible assets measured in millions of yen.

Labor is measured as number of workers.

As in Table

1, the industries and the prefectures are sorted in order of capital intensity and relative capital abundance, respectively. Note that there are 552 manufacturing industries, which prevents us from reporting the name of each industry. === Table 8 === The color of each cell indicates the capital intensity of a given industry in a given prefecture. Note that some prefectures report no production at the four-digit industry level. Accordingly, we now have some white cells, which means no production. As in the previous tables, light gray, gray, dark gray, and black mean that the capital intensity is in the rst, second, third, and fourth quartiles within each prefecture, respectively. Similar to Table 1, several factor intensity reversals exist among the 47 prefectures even if we use plant-level data. We also computed the mean of Spearman's rank correlations,

ρ.

The correlations are calculated using pairwise deletion of observations with

missing values. The mean of the correlations is 0.366 (the standard deviation is 0.108), which is signicantly smaller than that of Moroney (1967) and even smaller than that of Minhas (1962).

31 Our results show that our main results hold even when we use a

disaggregated industry classication. In fact, the degree of the factor intensity reversals

28

In a similar context, using detailed product-level import data for the United States, Schott (2004)

found that US imports were inconsistent with factor-proportion specializations across products but were consistent with such specialization within products.

29

Deardor (1986) emphasizes that factor intensity reversals can be made to appear and disappear,

just by redening goods.

30

Other issues in the use of the Census of Manufacture data are as follows. One is that tangible assets

are reported as a nominal book value rather than a market value.

Another is that if a plant that is

classied in one industry produces multi products across industries, the capital intensity for such a plant might not accurately represent capital intensity for that classied industry.

31

We also compute the 95 percent condence interval, as in the previous section, which is from 0.360

to 0.373 (the number of observations is 1,081, and the standard error is 0.003).

10

is higher when we use the disaggregated industry-level data than when we use the aggregated data. Thus, we weaken any possible criticism that factor intensity reversals may

32

be a result of the aggregation of industries.

We note that as indicated by equation (1) in Section 2.2, Spearman's rank correlation,

ρ,

depends on the number of observations,

n.

Thus, one may be concerned that the

correlation decreases from 0.645 (the benchmark case in Section 3.1) to 0.366 primarily because the number of manufacturing industries, 552.

n,

increases substantially from 13 to

However, equation (1) indicates that the correlation should become larger as

becomes larger, other things unchanged. is not because of the increased

n

n

Therefore, the decreased correlation (0.366)

(552) but because of the increased factor intensity

reversals. One may also be concerned that if the number of plants in each cell in Table 8 is small, the capital intensity will be aected by the small number of large (or small) plants. To address this concern, we exclude cells whose number of observations is fewer than 10. The results indicate that the mean of the correlations is 0.473 (the standard deviation is 0.569).

33 Although the rank correlation is slightly higher, this result is similar to the

result that includes all plants in that the mean correlation is signicantly smaller than

34

that of Moroney (1967) and even smaller than that of Minhas (1962).

Moreover, regardless of the number of plants in each cell in Table 8, the capital intensity may be aected by very large plants.

To address this concern, we exclude

plants that are in the top 1 percent of capital intensity in each cell. The results indicate

35 This result

that the mean of the correlations is 0.343 (the standard deviation is 0.109).

is also similar to the result that includes all plants. Another concern may be that, as we conrmed in Section 3, the correlation increases if prefectures are aggregated at the region level.

To address this concern, we compute the rank correlation, aggregating

47 prefectures to eight regions while using the same detailed four-digit industry level data. The results indicate that the mean of the rank correlation is 0.571 (the standard deviation is 0.047).

36

This is higher than that for the prefecture-level results, but it

is still smaller than that in the previous studies.

In sum, our main results hold even

when we take into account intra-industry heterogeneity (i.e., even when we use detailed industry-level data). Finally, note that some prefectures have industries with zero production at the detailed industry level. Our measure of factor intensity reversals exclude such industries. Specically, suppose that one prefecture produces goods 1, 2, and 3 while another prefecture produces goods 1 and 2 only, factor intensity reversals can be measured using the information on goods 1 and 2. A question therefore is whether or not our results are aected by the exclusion of the industries with zero production.

32

It may be worth mentioning an aggregation problem associated with the so-called lens conditions.

Debaere (2004) showed theoretically that with more disaggregation of sectors, the goods lens becomes even wider, making a violation even less likely.

Thus a possible criticism of empirical studies that

documents the satisfaction of the lens condition is that the lens condition may be satised because of the disaggregation of industries.

Evidence indeed supports this criticism.

Bernard et al.

(2005)

empirically showed that lenses created with more disaggregated data are wider than lenses created with more aggregate data and that the satisfaction of the lens condition is more likely when industries are relatively disaggregated compared with countries or regions.

33

The 95 percent condence interval is from 0.435 to 0.511 (the number of observations is 867, and

the standard error is 0.019).

34

To see whether the capital intensity is correlated with the number of plants, we also calculate the

cross-prefecture correlation between plant numbers and capital intensity for each of 552 industries. We nd that the correlations are around zero: the mean is 0.042 (the standard deviation is 0.147).

35

The 95 percent condence interval is from 0.336 to 0.349 (the number of observations is 1,081, and

the standard error is 0.003).

36

The 95 percent condence interval is from 0.553 to 0.589 (the number of observations is 28, and the

standard error is 0.009).

11

To answer this question, we focus on a set of prefectures and industries with positive production only.

37

We rst select seven prefectures that have more than 300 4-digit

industries with positive production. We then select 154 industries that are commonly observed in these seven prefectures. The result indicates that the mean of the correlation is 0.435 (the standard deviation is 0.050). We can thus conclude that our main results still hold even when we exclude the industries with zero production.

4.2 Capital intensity and exporter concentration Finally, we consider possible reasons for the observed factor intensity reversals among prefectures in Japan. As a possible reason, we suggest intra-industry heterogeneity, in particular, dierences in exporter concentration across prefectures in each industry. This is because as several studies such as Bernard, Jensen, Redding, and Schott (2012) pointed out, exporters are generally more capital-intensive than non-exporters

38 This in turn implies that if the degree

in both developed and developing countries.

of exporter concentration is dierent across prefectures, the capital intensity could be dierent across them even within a narrowly dened industry. To explore this possibility, we run the following simple regression:

Kij = α + βEXP Rij + γXij + µi + µj + εij , Lij where

Kij /Lij

is the capital intensity of industry

i

in prefecture

(3)

j ; EXP Rij is exporter i in prefecture

concentration measured by the ratio of exporters to all plants in industry

j ; Xij

includes additional control variables;

µi

and

µj

are industry and prefecture xed

eects to control for unobserved industry and prefecture heterogeneity, respectively; and

εij

is an error term. The data are obtained from the Census of Manufacture as in the

previous section. Our parameter of interest is tion is a possible reason,

β

β.

If the dierences in the degree of exporter concentra-

will be positive and signicant. As for the control variables,

we use the number of plants and their market shares in industry

i

in prefecture

j

to

control for possible agglomeration and size eects, respectively. As for the market share, we rst compute the output share of each plant, by industry

i,

and then aggregate it to

the industry-prefecture level. This variable becomes large if the plants with large market shares in industry

i

concentrate in prefecture

Table 9 presents the estimation results. signicantly positive coecient for

β.

j. Consistent with our expectation, we nd

As the concentration of exporters becomes higher,

the capital intensity of the prefecture becomes higher. This result is robust even when

39 These results

we use the ratio of exports to total sales instead of the ratio of exporters.

in turn imply that the plant heterogeneity regarding exports is one of the possible reasons why we observe large dierences in capital intensity even within a narrowly dened industry. Moreover, in Section 3.2 we found that the degree of the factor intensity reversals had increased in the last two decades. This suggests the increasing importance of

37

This exercise is similar to a panel-data study that focuses on balanced panel data when the original

data are unbalanced.

38

Note that in contrast, Ma, Tang, and Zhang (2014) found that in China, exporters were less capital-

intensive than non-exporters. They also developed a model that was consistent with this nding. In their model, when a rm in a labor-abundant country becomes a exporter, it specializes in its core products by allocating more resources to produce more labor-intensive goods.

On the ipside, the model also

implies that an exporter produces more capital-intensive goods in a capital-abundant country, which is consistent with the ndings in developed countries.

39

The number of observations is slightly smaller in columns (4)(6) than that in columns (1)(3).

This is because some plants report zero sales. Since this may be due to the small value of sales, we keep these plants in columns (1)(3).

12

plant heterogeneity. While only indicative, these results suggest that the factor intensity reversals across prefectures may be related to the spatial distribution of exporters. === Table 9 === It is worth noting that the spatial distribution of exporters depends upon the denition of the space (e.g., prefecture, region, etc.). We checked the distribution of exporters across eight regions as well as 47 prefectures.

We found that the mean and standard

deviation of the exporter ratios is 0.117 and 0.249 respectively at the prefecture-industry level while 0.122 and 0.208 respectively at the region-industry level. This indicates that the distribution of the exporter ratios becomes narrow when the 47 prefectures are aggregated at the eight region level. This may be a reason why we nd less factor intensity reversals when we aggregate 47 prefectures to eight regions in Section 3.1.2. It is also worth noting that the observed dierences in exporter concentration among prefectures imply dierences in technology among prefectures.

It, however, does not

contradict the HeckscherOhlin model that assumes identical technology (technological knowhow) but allows factor intensity reversals across regions. This is because as noted in footnote 8, the same industry can be dierent in capital intensity between regions in equilibrium, even though technological knowhow is identical. This dierence is interpreted by Bhagwati and Deheja (1994) as a

de facto

dierence in technology.

According to the well-known empirical evidence that exporters generally have higher productivity than nonexporters, the dierences in exporter concentration among prefectures also imply that prefectures have dierent productivity, in particular, prefectures with the higher exporter concentration have higher productivity.

This casts a doubt

on the identical-technology assumption of the HeckscherOhlin model. As empirically shown by Todo (2011), however, it is known that productivity is, though statistically signicant, quantitatively unimportant as a factor that determines whether Japanese rms export or not. Therefore, the higher exporter concentration does not necessarily mean higher productivity in Japan, which can alleviate the above doubt.

5

Conclusion

Based on newly developed Japanese prefecture-level data, we argue that the abandonment of factor intensity reversals in empirical analysis has been premature. Specically, we have found that the degree of factor intensity reversals is higher than that found in previous studies on average. Our empirical results have shown that while factor intensity reversals are less severe at the aggregated eight-region level, they are prevalent at the 47-prefecture level. Furthermore, the degree of the factor intensity reversals has increased in the last two decades. We have also performed several robustness checks: 1) the sample includes agriculture and mining industries, 2) 47 prefectures are aggregated into eight regions, 3) the analysis takes into account human capital, 4) we compare different years, and 5) the industries are disaggregated at the four-digit level. In particular, we have found that the degree of the factor intensity reversals is higher when we use disaggregated industry-level data than when we use aggregated data. Thus we have successfully weakened any possible criticism that factor intensity reversals may be a result of the aggregation of industries.

Finally, as a possible reason for the observed factor

intensity reversals, our regression results have suggested intra-industry heterogeneity, in particular, dierences in exporter concentration across prefectures in each industry. The implications of our study are threefold. First, the standard industry classications may not be appropriate for testing the empirical validity of the HeckscherOhlin model. As was pointed out by Schott (2003) and Kiyota (2012), the standard industry

13

classication groups output loosely, according to the similarity of end use (e.g., electrical machinery, transportation machinery) rather than actual factor use (e.g., capitalintensive goods, labor-intensive goods). However, our results show that the same industry can be relatively capital intensive in one prefecture but relatively labor intensive in another prefecture. This indicates that the standard industry classications may not be able to capture the actual capital intensity dierences among countries or regions. It thus may be important to adapt a theoretically appropriate aggregation method such as the HeckscherOhlin aggregates developed by Schott (2003). Second, it is important for policy makers to understand the intra-industry capitalintensity heterogeneity. A capital-intensive industry in one country or one region may not necessarily be capital intensive in the same industry in another country or region because of the intraindustry capital-intensity heterogeneity. This in turn implies that industryspecic policies may not work eectively because of the intra-industry heterogeneity. Before designing industrial policies, policy makers need to examine the heterogeneity across countries and/or regions. Third, both theoretical and empirical studies on international trade need to place more importance on the possibility of the factor intensity reversals. As long as we rely on the end-use industry classications, factor intensity reversals can be expected to exist. It may be appropriate to relax the assumption of no factor intensity reversals. Several next steps can be taken for future research. First, it is important to further investigate the possible reasons why factor intensity reversals have recently existed and even increased among the 47 prefectures in Japan.

As suggested in Section 4.2, a di-

rection for future research is further investigating dierences in exporter concentration across prefectures in each industry. Second, it is also important to further disaggregate industries because the four-digit level industry classications that we have used in Section 4 may not be disaggregated enough to control for technology dierences. For example, it might be possible that the degree of computerization is dierent across prefectures even

40 Thus the use of more detailed prefecture-industry-level data may

within industries.

weaken the degree of factor intensity reversals, although it may also further strengthen them.

To address this issue, the quality and coverage of the prefecture-industry-level

data must be improved and further disaggregated. Third, another interesting disaggregation might be investigating dierences in capital intensity within prefecture while this paper has focused on those between prefectures. Fourth, this paper has shown clear-cut evidence for the existence of factor intensity reversals among regions.

It can encourage building theoretical models with factor in-

tensity reversals among regions that provide celebrated insights. One possible direction is developing models of production fragmentation between regions. Upon application of Venables' (1999) model, for example, it would be possible to examine whether a reduction in transport costs for intermediate goods can cause vertical or horizontal production

41 We leave to future research this direction of develop-

fragmentation between regions. ment of models.

Appendix A

Data Appendix === Tables A1 & A2 ===

40 41

For example, see Autor et al. (2003) for computerization and job tasks within industries. Using a model with the possibility of factor intensity reversals among countries, Venables (1999)

examines how a fall in transport costs for intermediate goods causes vertical or horizontal production fragmentation among countries.

14

Appendix B

Parametric Results

This paper employed a nonparametric approach to examine the possibility of factor intensity reversals. However, one may be concerned how the results are changed if we employ a parametric approach. To address this concern, following Minhas (1962) and Leontief (1964), we estimate the constant elasticity of substitution (CES) production function, by industry.

V = [aK −γ + bL−γ ]

The CES production function is specied as: value added;

a, b,

γ

and

are the parameters.

capital and labor is given by

−1/γ

, where

V

is

The elasticity of substitution between

σ = 1/(γ + 1).

Suppose that the good and factor markets are perfectly competitive. Let

w

and

r

be

the prices of capital and labor, respectively. From the prot maximization of an industry, we can obtain:

ln Adding the error term

ε,

(w) r

( ) ( ) b K = ln + (γ + 1) ln . a L

the industry subscript i, and the prefecture subscript

ln

wij rij

(

) = ln

bi ai

(

) + (γi + 1) ln

j,

we can

i: )

obtain the following regression equation for each industry

(

(4)

Kij Lij

+ εij .

(5)

Because we assume that the CES production function is the same in all the prefectures,

ai , bi ,

and

γi

carry only the industry but not the prefecture subscript.

Using the R-JIP Database 2014, we estimate this equation for the 13 manufacturing industries in the 47 prefectures. The capital and labor data are the same as those used in Section 2. Wages are obtained from wage bill divided by the man-hours. In the R-JIP database, the price of capital

rij =

1.42

r

is the same across prefectures. We thus normalize that

Figure B1 presents the hypothetical relationship between

w/r

presents the case where there are no factor intensity reversals: good intensive than good

B

in both

α (kA > kB )

and

′ > k ′ ). β (kA B

K/L. Panel (a) A is more capital

and

Panel (a) implies that

the elasticity of substitution between capital and labor is the same across industries. In contrast, panel (b), which has the intersection on the right side, presents the case where there is a factor intensity reversal: good (kA

> kB )

but good

A

A

is more capital intensive than good

is less capital intensive than good

B

in

′ < k ′ ). β (kA B

B

in

α

Panel (b)

implies that the elasticity of substitution is dierent across industries. === Figures B1 & B2 === Table B1 and Figure B2 present the estimation results. Figure B2 shows intersections like panel (b) of Figure B1, indicating the existence of the factor intensity reversals even when we employ a parametric approach.

Table B1 shows that the elasticity of

substitution is dierent across industries.

Note, however, that Table B1 also shows

σ.

This implies that either the CES specication

unusual gures for some of the estimated

or the assumptions of perfectly competitive markets, or both are violated. This result justies our use of a nonparametric approach as a baseline analysis. === Table B1 ===

42

The assumption that wages dier but rentals equal across prefectures is compatible with the dis-

cussion about low labor mobility and (probably) high capital mobility in Japan in Section 2 (footnote 8).

15

References Agarwal, B.L. 2007.

Programmed Statistics: QuestionsAnswers.

New Delhi: New Age

International. Autor, David H., David Dorn, and Gordon H. Hanson.

2016.

The China Shock:

Annual

Learning from Labor-Market Adjustment to Large Changes in Trade,

Review of Economics,

8: 20540.

Autor, David H., Frank Levy, and Richard J. Murnane. 2003. The Skill Content of Recent Technological Change: An Empirical Exploration.

Economics,

Quarterly Journal of

118(4): 1279333.

Ball, David S. 1966. Factor-Intensity Reversals in International Comparison of Factor Costs and Factor Use.

Journal of Political Economy,

74(1): 7780.

Bernard, Andrew B., J. Bradford Jensen, Stephen J. Redding, and Peter K. Schott. 2012. The Empirics of Firm Heterogeneity and International Trade,

view of Economics,

Annual Re-

4: 283313.

Bernard, Andrew B., Stephen J. Redding, Peter K. Schott, and Helen Simpson. 2008. Relative Wage Variation and Industry Location in the United Kingdom.

Bulletin of Economics and Statistics,

Oxford

70(4): 43159.

Bernard, Andrew B., Stephen J. Redding, and Peter K. Schott. 2013. Testing for Factor Price Equality with Unobserved Dierences in Factor Quality or Productivity.

American Economic Journal: Microeconomics,

5(2): 13563.

Bernard, Andrew B., Raymond Robertson, and Peter K. Schott. 2005. A Note on the Empirical Implementation of the Lens Condition. NBER Working Paper Series, No. 11448, National Bureau of Economic Research (NBER). Bernstein, Jerey R. and David E. Weinstein.

2002.

Do Endowments Predict the

Location of Production? Evidence from National and International Data.

of International Economics,

Journal

56(1): 5576.

Bhagwati, Jagdish, and Vivek H. Dehejia.

1994.

Freer Trade and Wages of the

UnskilledIs Marx Striking Again? In Jagdish Bhagwati and Marvin H. Kosters, eds.,

Trade and Wages: Leveling Wages Down?

Washington, D.C.: The American

Enterprise Institute Press, pp. 3675. Blanchard, Olivier J. and Lawrence F. Katz. 1992. Regional Evolutions.

Papers on Economic Activity, Debaere, Peter.

2004.

Brookings

23(1): 176.

Does Lumpiness Matter in an Open Economy?

International Economics with Regional Data.

Studying

Journal of International Economics,

64(2): 485-501. Deardor, Alan V. 1986.

Firless Firwoes:

Theorems of International Trade.

How Preferences Can Interfere with the

Journal of International Economics,

20(12):

13142. Dekle, Robert, Kyoji Fukao and Murat Ungor. 2010. The Japan-US Exchange Rate, Productivity, and the Competitiveness of Japanese Industries. In Koichi Hamada, Anil K Kashyap, and David E. Weinstein, eds.,

Long-term Stagnation.

Japan's Bubble, Deation, and

Cambridge: The MIT Press, pp. 10528.

16

Dekle, Robert, Atsushi Kawakami, Nobuhiro Kiyotaki, and Tsutomu Miyagawa. 2015. Product Dynamics and Aggregate Shocks: Evidence from Japanese Product and Firm Level Data. RIETI Discussion Paper Series, 15-E-137, Research Institute of Economy, Trade and Industry (RIETI).

Advanced International Trade: Theory and Evidence.

Feenstra, Robert C. 2015.

Prince-

ton University Press, 2nd ed. Fennstra, Robert C. and Alan M. Taylor.

2014.

International Economics.

Worth

Publishers, 3rd ed. Fuchs, Victor R. 1963.

and Statistics,

CapitalLabor Substitution: A Note.

Review of Economics

45(4): 43638.

Garofalo, Gasper A. and Steven Yamarik. 2002. Regional Convergence: Evidence from a New State-by-State Capital Stock Series.

Review of Economics and Statistics,

84(2): 31623. Harrigan, James. 1997. Technology, Factor Supplies, and International Specialization: Estimating the Neoclassical Model.

American Economic Review,

87(4): 47594.

Hsieh, Chang-Tai and Peter J. Klenow. 2009. Misallocation and Manufacturing TFP in China and India.

Quarterly Journal of Economics,

Kiyota, Kozo. 2012. A Many-cone World?

124(4): 140348.

Journal of International Economics, 86(2):

34554. Kurokawa, Yoshinori. 2011. Is A Skill Intensity Reversal A Mere Theoretical Curiosum? Evidence from the US and Mexico.

Economics Letters,

112(2): 15154.

Leontief, Wassily. 1964. An International Comparison of Factor Costs and Factor Use: A Review Article.

American Economic Review,

54(4): 33545.

Ma, Yue, Heiwai Tang, and Yigan Zhang. 2014. Factor Intensity, Product Switching, and Productivity:

Economics,

Evidence from Chinese Exporters,

Journal of International

92: 34962.

Minhas, Bagicha S. 1962. The Homohypallagic Production Function, Factor-Intensity Reversals, and the HeckscherOhlin Theorem.

Journal of Political Economy, 70(2):

13856. Ministry of Internal Aairs and Communications (MIC). 2000.

12 Nen (2000 Population Census of Japan),

Tokyo:

Kokusei Chousa: Heisei

Nihon Tokei Kyokai.

(In

Japanese) Moroney, John R. 1967. Test.

The Strong-Factor-Intensity Hypothesis:

Journal of Political Economy,

A Multisectoral

75(3): 24149.

Organization for Economic Corporation and Development (OECD). 2006.

Migration Outlook, 2006,

International

Paris: OECD.

Reshef, Ariell. 2007. HeckscherOhlin and the Global Increase of Skill Premia: Factor Intensity Reversals to the Rescue. Working Paper, New York University. Sampson, Thomas. Inequality.

2016.

Assignment Reversals: Trade, Skill Allocation and Wage

Journal of Economic Theory,

17

163: 365409.

Samuelson, Paul A. 1951. A Comment on Factor Price Equalisation.

nomic Studies,

Review of Eco-

19(2): 12122.

Schott, Peter K. 2003. One Size Fits All? Heckscher-Ohlin Specialization in Global Production.

American Economic Review,

93(3): 686708.

Schott, Peter K. 2004. Across-Product versus Within-Product Specializaion in International Trade.

Quarterly Journal of Economics,

119(2): 64778.

Todo, Yasuyuki. 2011. Quantitative Evaluation of the Determinants of Export and FDI: Firm-level Evidence from Japan.

The World Economy,

34(3): 35581.

Tokui, Joji, Tatsuji Makino, Kyoji Fukao, Tsutomu Miyagawa, Nobuyuki Arai, Sonoe Arai, Tomohiko Inui, Kazuyasu Kawasaki, Naomi Kodama, and Naohiro Noguchi. 2013. Compilation of the Regional-Level Japan Industrial Productivity Database. RIETI Discussion Paper Series, 13-J-037, Research Institute of Economy, Trade and Industry (RIETI). Tomiura, Eiichi.

2005.

Economic Review,

Factor Price Equalization in Japanese Regions.

56(4): 44156.

Venables, Anthony J. 1999. Fragmentation and Multinational Production.

Economic Review,

Japanese European

43(4-6): 93545.

Wong, Kar-yiu. 1990. Factor Intensity Reversal in A Multi-Factor, Two-good Economy.

Journal of Economic Theory,

Yamarik, Steven. tioms.

2013.

51(2): 43442.

State-Level Capital and Investment: Updates and Implica-

Contemporary Economic Policy,

18

31(1): 6272.

Manufacturing

3

Figure 1. Distribution of Rho, Prefecture Level, Manufacturing Figure 1: Distribution of ρ, Prefecture Level,

10

Density

21

Density

32

Figure 1. Distribution of Rho, Prefecture Level, Manufacturing

-.5

0

.5

1

Spearman's rho

0

1975 1995

-.5

0

1985 2005

.5

1

Spearman's rho

Note: Kernel density function. Note: Kernel density function.

1975

1985

Figure b. Distribution of Rho, Prefecture Level, 1995including Agriculture 2005 and Mining

4

Note: Kernel density function. Figure b. Distribution of Rho,ρPrefecture Level, including Agriculture and Mining , Prefecture Level, Including Agriculture

and Mining

-.5

0

.5

1

Spearman's rho 1975 1995

0

1

0

Density 2 1

3

Density 4 2

3

Figure 2: Distribution of

-.5

Note: Kernel density function.

0

1985 2005

.5 Spearman's rho 1975 1995

Note: Kernel density function. Note: Kernel density function.

19

1985 2005

1

Table 1: PrefectureIndustry Capital Intensity, Manufacturing, 2005 dĂďůĞϭ͘WƌĞĨĞĐƚƵƌĞͲ/ŶĚƵƐƚƌLJĂƉŝƚĂů/ŶƚĞŶƐŝƚLJ͕DĂŶƵĨĂĐƚƵƌŝŶŐ͕ϮϬϬϱ /ŶĚƵƐƚƌLJĂǀĞƌĂŐĞĐĂƉŝƚĂůůĂďŽƌƌĂƚŝŽ ϭϬ͘ϱϱ

ůĂ ď Ž ƌ ƌĂ ƚŝ Ž ; ŵ ĂŶ Ƶ ĨĂ Đƚ Ƶ ƌŝ Ŷ Ő ƚŽ ƚĂ ůͿ

 Ŷ Ě Ž ǁ ŵ Ğ Ŷ ƚ͗ Ɖ ƌĞ ĨĞ Đƚ Ƶ ƌĞ Đ ĂƉ ŝƚ Ăů Ͳ

ϯ͘ϴϲ

ϲ͘ϭϯ

D Ğ ƚĂ ůƉ ƌŽ Ě Ƶ Đƚ Ɛ

K ƚŚ Ğ ƌ ŵ ĂŶ Ƶ ĨĂ Đƚ Ƶ ƌŝ Ŷ Ő

ϲ͘Ϯϲ

d Ğ džƚ ŝůĞ Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

ϳ͘ϭϮ

&Ž Ž Ě Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

ϳ͘ϴϯ

 Ğ ƌĂ ŵ ŝĐ ͕Ɛ ƚŽ Ŷ Ğ Ă Ŷ Ě Đ ůĂ LJ Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

ϴ͘ϭϮ

ϭϭ͘ϱϱ

' Ğ Ŷ Ğ ƌĂ ůŵ ĂĐ Ś ŝŶ Ğ ƌLJ

W ƌĞ Đŝ Ɛŝ Ž Ŷ ŵ ĂĐ Ś ŝŶ Ğ ƌLJ

ϭϮ͘ϳϲ

W Ƶ ůƉ Ă Ŷ Ě Ɖ ĂƉ Ğ ƌ

ϭϯ͘Ϭϲ

ϭϯ͘ϭϰ

d ƌĂ Ŷ ƐƉ Ž ƌƚ Ăƚ ŝŽ Ŷ ŵ ĂĐ Ś ŝŶ Ğ ƌLJ

 ůĞ Đƚ ƌŝ ĐĂ ůŵ ĂĐ Ś ŝŶ Ğ ƌLJ

Ϯϱ͘ϱϯ

Ϯϳ͘ϵϵ

ϭϮϯ͘ϰϴ

W ƌŝ ŵ Ăƌ LJ ŵ Ğ ƚĂ ů

 Ś Ğ ŵ ŝĐ Ăů Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

W Ğ ƚƌ Ž ůĞ Ƶ ŵ Ă Ŷ Ě Đ Ž Ăů Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

WƌĞĨĞĐƚƵƌĞ dŽŬLJŽ

ϰ͘ϴϮ

Ϯ͘ϱϬ

ϯ͘ϲϴ

ϭ͘ϯϳ

ϱ͘ϭϱ

ϯ͘ϯϴ

ϰ͘ϭϬ

ϭϬ͘ϱϬ

ϰ͘Ϭϰ

ϵ͘ϳϴ

ϱ͘ϵϭ

ϲ͘ϭϳ

ϲ͘Ϭϰ

ϴ͘ϵϲ

KŬŝŶĂǁĂ

ϲ͘Ϭϵ

Ϯ͘ϴϬ

Ϯ͘ϴϰ

Ϯ͘ϬϬ

ϲ͘ϲϭ

ϰ͘ϯϯ

ϲ͘ϮϮ

Ϭ͘Ϭϴ

ϱ͘ϵϱ

ϭ͘ϯϮ

ϭ͘ϱϭ

ϭϬ͘Ϭϱ

ϱ͘ϯϲ

ϭϰϱ͘ϳϰ

<ŽĐŚŝ

ϲ͘ϯϮ

ϭ͘Ϯϵ

ϯ͘ϱϲ

ϰ͘ϲϮ

ϯ͘ϯϮ

ϭϯ͘ϱϭ

ϲ͘Ϭϳ

ϴ͘Ϭϭ

ϴ͘ϵϲ

ϰ͘ϴϱ

ϭϰ͘ϲϮ

ϵ͘ϱϬ

ϭϮ͘ϯϭ

ϵ͘ϳϮ

'ŝĨƵ

ϳ͘Ϭϰ

ϯ͘ϳϮ

ϲ͘ϮϮ

ϱ͘ϯϴ

ϰ͘ϲϯ

ϱ͘Ϯϲ

ϳ͘ϬϬ

ϭϭ͘ϵϮ

ϭϯ͘ϳϳ

ϵ͘Ϯϵ

ϴ͘ϯϯ

ϵ͘Ϭϰ

Ϯϭ͘ϰϭ

ϲϵ͘ϳϱ

^ŚŝŵĂŶĞ

ϳ͘ϭϲ

ϯ͘ϴϬ

ϱ͘ϭϯ

ϲ͘ϲϲ

ϯ͘ϭϰ

ϱ͘ϱϲ

ϳ͘ϭϰ

ϭϰ͘ϬϮ

ϳ͘Ϯϰ

ϭϬ͘Ϭϲ

ϵ͘ϳϱ

ϭϭ͘ϱϲ

ϴ͘Ϭϲ

Ϯϱ͘ϳϬ

ŬŝƚĂ

ϳ͘ϯϲ

Ϯ͘ϵϰ

ϯ͘ϰϵ

ϭ͘Ϯϲ

ϯ͘ϴϮ

ϰ͘ϰϱ

ϰ͘ϯϳ

ϭϴ͘ϲϮ

ϯϯ͘ϭϭ

ϳ͘ϴϭ

ϭϬ͘ϵϴ

ϭϲ͘ϴϰ

ϯϰ͘ϭϳ

ϴϱ͘ϭϵ

<ĂŐŽƐŚŝŵĂ

ϳ͘ϱϯ

ϯ͘ϰϱ

ϯ͘Ϭϴ

ϯ͘ϭϲ

ϱ͘ϭϭ

ϴ͘ϳϳ

ϱ͘Ϯϲ

ϰ͘ϴϬ

ϭϳ͘ϲϮ

ϯ͘ϯϬ

ϭϯ͘ϴϭ

ϭϮ͘ϰϮ

ϭϲ͘ϱϰ

ϯϮ͘ϳϮ

/ƐŚŝŬĂǁĂ

ϳ͘ϳϰ

Ϯ͘ϱϵ

ϰ͘ϴϱ

ϭϬ͘ϭϮ

ϰ͘ϳϮ

ϯ͘ϴϵ

ϳ͘ϵϱ

ϲ͘ϭϯ

ϯ͘ϳϯ

ϰ͘ϲϴ

ϭϯ͘ϭϱ

ϭϬ͘ϳϲ

ϭϵ͘Ϯϴ

ϲϯ͘ϰϮ

zĂŵĂŐĂƚĂ

ϳ͘ϵϭ

ϯ͘ϲϱ

ϰ͘ϲϯ

Ϯ͘ϵϬ

ϰ͘ϰϯ

ϴ͘ϱϴ

ϱ͘ϴϮ

ϲ͘ϲϬ

ϱ͘Ϭϰ

ϴ͘ϱϳ

ϭϮ͘ϳϴ

ϭϰ͘ϭϴ

Ϯϰ͘Ϭϴ

Ϯϵ͘ϳϴ

KƐĂŬĂ

ϳ͘ϵϳ

ϯ͘ϲϳ

ϱ͘Ϯϯ

ϱ͘Ϭϰ

ϲ͘ϲϰ

ϲ͘Ϭϴ

ϲ͘ϰϬ

ϳ͘ϯϱ

ϲ͘ϲϭ

ϭϬ͘ϴϯ

ϵ͘Ϯϯ

ϭϴ͘ϲϭ

ϭϴ͘ϭϱ

ϭϮϯ͘ϴϲ

dŽƚƚŽƌŝ

ϴ͘Ϭϲ

ϯ͘ϴϴ

ϯ͘ϱϯ

ϯ͘ϭϵ

ϴ͘ϭϯ

ϰ͘Ϯϲ

ϯ͘ϵϱ

ϱ͘Ϯϭ

Ϯϱ͘Ϯϰ

ϯ͘ϯϰ

ϭϬ͘ϭϵ

ϴ͘ϭϭ

ϰ͘ϯϳ

ϯϳ͘ϵϴ

<LJŽƚŽ

ϴ͘ϭϮ

ϯ͘ϭϲ

ϱ͘Ϭϰ

ϰ͘ϭϵ

ϭϬ͘Ϭϲ

ϳ͘ϯϱ

ϱ͘ϲϴ

ϭϭ͘ϰϭ

ϱ͘ϰϵ

ϭϵ͘ϯϲ

ϭϭ͘ϱϮ

ϭϭ͘ϯϰ

ϭϳ͘ϯϰ

Ϯϵ͘ϱϵ

^ĂŝƚĂŵĂ

ϴ͘ϭϱ

ϯ͘ϱϯ

ϳ͘ϭϳ

ϯ͘ϴϭ

ϱ͘ϴϰ

ϴ͘ϯϯ

ϲ͘ϱϯ

ϭϮ͘ϲϳ

ϳ͘ϴϱ

ϭϭ͘Ϯϱ

ϵ͘ϰϱ

ϭϮ͘ϳϮ

ϭϴ͘ϰϲ

ϰϱ͘ϵϭ

/ǁĂƚĞ

ϴ͘ϭϳ

ϯ͘ϭϵ

ϯ͘ϵϮ

Ϯ͘ϳϭ

ϯ͘ϴϵ

ϵ͘Ϭϳ

ϱ͘ϱϳ

ϭϬ͘ϳϵ

ϭϴ͘ϭϲ

ϴ͘ϵϳ

ϭϯ͘ϲϬ

Ϯϴ͘Ϭϴ

ϯϯ͘ϰϯ

ϱϵ͘ϳϴ

<ĂŐĂǁĂ

ϴ͘ϱϱ

ϯ͘ϱϮ

ϲ͘ϭϬ

ϰ͘ϱϬ

ϱ͘ϰϴ

ϱ͘ϱϬ

ϵ͘ϬϬ

ϯ͘ϵϱ

ϭϭ͘Ϭϴ

ϳ͘ϯϴ

ϲ͘ϭϰ

ϮϮ͘ϱϮ

Ϯϯ͘ϱϳ

ϭϰϲ͘ϰϱ

EŝŝŐĂƚĂ

ϴ͘ϲϭ

ϯ͘ϲϭ

ϰ͘ϲϲ

ϰ͘ϱϭ

ϱ͘ϱϲ

ϳ͘ϳϮ

ϲ͘ϰϵ

ϲ͘ϯϵ

ϭϱ͘ϰϭ

ϱ͘ϲϳ

ϭϯ͘ϮϮ

ϭϱ͘ϯϮ

ϱϬ͘ϵϯ

ϭϮϵ͘Ϯϱ

DŝLJĂnjĂŬŝ

ϴ͘ϲϯ

Ϯ͘Ϯϲ

ϱ͘ϱϳ

ϭϮ͘ϴϯ

ϰ͘ϲϬ

ϰ͘ϭϴ

ϯ͘ϳϰ

ϭϳ͘Ϭϵ

Ϯϯ͘ϰϬ

ϰ͘ϴϮ

ϭϭ͘ϭϮ

ϭϮ͘ϵϭ

ϰϬ͘ϴϱ

ϲϱ͘ϲϵ

^ĂŐĂ

ϴ͘ϵϬ

ϴ͘ϱϱ

ϳ͘Ϯϰ

ϯ͘ϭϵ

ϳ͘ϲϴ

Ϯ͘ϳϭ

ϲ͘ϲϴ

ϵ͘Ϯϵ

ϭϮ͘ϯϭ

ϳ͘ϴϯ

ϭϲ͘ϳϳ

ϭϴ͘ϴϳ

ϭϲ͘ϲϬ

ϱϬ͘ϲϯ

EĂƌĂ

ϵ͘Ϭϭ

ϰ͘ϰϬ

ϱ͘ϵϵ

ϰ͘ϴϱ

ϵ͘ϰϲ

ϰ͘ϵϯ

ϭϱ͘ϰϵ

Ϯ͘ϴϰ

ϰ͘ϳϬ

ϵ͘ϭϵ

ϭϲ͘ϭϰ

ϭϭ͘ϵϯ

ϵ͘ϭϬ

Ϯϲ͘ϵϭ

&ƵŬƵŝ

ϵ͘ϭϵ

ϯ͘Ϯϵ

ϱ͘ϭϲ

ϭϬ͘ϭϮ

ϯ͘ϱϵ

ϭϬ͘ϰϬ

ϰ͘ϲϳ

ϱ͘ϭϮ

ϴ͘ϳϳ

ϭϮ͘ϭϲ

ϭϯ͘Ϭϵ

Ϯϰ͘Ϯϴ

Ϯϭ͘ϲϯ

ϱϴ͘ϲϭ

DŝLJĂŐŝ

ϵ͘ϯϮ

ϲ͘Ϭϭ

ϳ͘ϱϵ

ϱ͘ϯϯ

ϲ͘ϯϱ

ϲ͘ϵϯ

ϱ͘ϭϰ

ϱ͘ϳϱ

Ϯϱ͘ϬϬ

ϱ͘ϳϳ

ϭϮ͘ϭϱ

ϮϮ͘ϳϱ

ϭϬ͘ϵϳ

ϭϳϰ͘ϵϴ

zĂŵĂŶĂƐŚŝ

ϵ͘ϱϲ

ϯ͘ϮϬ

ϰ͘Ϭϵ

ϲ͘ϭϬ

ϴ͘ϴϯ

ϱ͘ϯϴ

ϭϬ͘ϯϴ

ϭϰ͘ϲϯ

ϰ͘ϰϰ

ϲ͘ϳϬ

ϭϲ͘ϳϮ

ϳ͘ϭϴ

Ϯϭ͘ϰϴ

ϯ͘ϯϭ

EĂŐĂŶŽ

ϵ͘ϵϴ

ϯ͘ϳϯ

ϱ͘Ϯϱ

ϲ͘ϵϭ

ϳ͘ϲϳ

ϳ͘ϯϴ

ϴ͘ϯϰ

ϭϵ͘ϲϬ

ϱ͘ϵϰ

ϴ͘ϵϴ

ϭϰ͘ϰϭ

ϴ͘Ϯϵ

ϭϰ͘ϮϮ

ϯϵ͘ϳϲ

,ŽŬŬĂŝĚŽ

ϭϬ͘Ϭϭ

ϯ͘ϴϱ

ϱ͘Ϭϲ

ϯ͘ϰϮ

ϲ͘ϱϰ

ϴ͘Ϯϭ

ϲ͘ϯϲ

Ϯ͘ϴϯ

ϯϵ͘ϭϲ

ϭϳ͘ϱϵ

ϭϱ͘Ϯϱ

ϯϯ͘ϵϰ

ϮϮ͘ϵϴ

ϮϮϳ͘ϭϭ

ŽŵŽƌŝ

ϭϬ͘ϭϬ

Ϯ͘ϲϮ

Ϯ͘ϲϮ

ϰ͘ϲϲ

ϰ͘ϵϱ

ϳ͘ϲϬ

ϳ͘ϵϮ

ϯ͘ϵϱ

Ϯϴ͘ϵϬ

Ϯ͘ϴϭ

ϳ͘ϰϬ

ϲϮ͘ϱϬ

ϯϭ͘ϳϱ

ϲϰ͘ϱϮ

'ƵŵŵĂ

ϭϬ͘ϭϮ

ϯ͘ϴϳ

ϳ͘Ϯϰ

ϯ͘ϲϵ

ϭϭ͘ϯϱ

ϳ͘Ϯϲ

ϳ͘ϱϱ

ϳ͘ϮϮ

ϴ͘ϭϲ

ϭϮ͘ϵϮ

ϭϮ͘ϯϳ

ϭϰ͘ϯϬ

ϯϬ͘ϯϬ

ϯϴ͘ϭϲ

^ŚŝnjƵŽŬĂ

ϭϬ͘ϲϱ

ϯ͘ϲϭ

ϳ͘ϰϰ

ϮϮ͘ϭϴ

ϴ͘ϰϵ

ϳ͘ϰϴ

ϲ͘ϰϯ

ϭϯ͘ϱϯ

ϭϴ͘Ϭϲ

ϭϮ͘ϱϳ

ϵ͘ϴϭ

ϭϰ͘ϴϰ

ϯϭ͘ϴϭ

ϯϱ͘ϱϬ

&ƵŬƵŽŬĂ

ϭϬ͘ϲϱ

ϯ͘ϴϳ

ϱ͘ϯϯ

Ϯ͘ϰϱ

ϲ͘ϰϮ

ϴ͘ϯϯ

ϱ͘ϵϰ

ϲ͘ϱϵ

ϱ͘Ϯϯ

ϭϲ͘ϲϲ

ϭϮ͘Ϯϯ

ϯϲ͘ϯϮ

ϯϴ͘ϳϰ

ϵϱ͘ϭϯ

<ƵŵĂŵŽƚŽ

ϭϬ͘ϳϯ

ϰ͘ϰϯ

ϲ͘ϳϭ

ϰ͘Ϯϵ

ϲ͘ϭϲ

ϱ͘ϲϯ

ϳ͘ϯϲ

ϯ͘ϴϯ

Ϯϰ͘ϯϮ

ϳ͘Ϭϲ

Ϯϯ͘ϳϲ

ϭϱ͘ϯϯ

Ϯϯ͘Ϭϯ

ϲϯ͘ϯϰ

&ƵŬƵƐŚŝŵĂ

ϭϬ͘ϴϬ

ϰ͘ϯϵ

ϲ͘Ϯϵ

Ϯ͘ϳϵ

ϴ͘ϯϳ

ϳ͘ϯϯ

ϲ͘ϭϳ

ϵ͘ϲϭ

ϭϯ͘ϳϰ

ϭϬ͘ϯϱ

ϭϱ͘Ϭϳ

ϭϰ͘ϴϱ

ϰϱ͘ϰϲ

ϮϭϬ͘ϲϯ

dŽLJĂŵĂ

ϭϭ͘Ϯϳ

ϱ͘ϯϭ

ϲ͘ϯϯ

ϭϬ͘ϯϱ

ϱ͘ϱϮ

ϳ͘ϭϰ

ϴ͘ϱϲ

ϭϮ͘ϱϮ

ϮϬ͘Ϭϯ

ϴ͘ϭϭ

ϭϳ͘ϭϲ

Ϯϳ͘ϭϬ

Ϯϳ͘ϱϰ

ϭϮϴ͘ϴϳ

dŽŬƵƐŚŝŵĂ

ϭϭ͘ϰϵ

ϯ͘Ϭϵ

ϱ͘ϱϰ

ϲ͘ϰϰ

ϴ͘Ϭϵ

ϯ͘Ϯϯ

ϲ͘ϵϯ

ϭ͘ϯϲ

ϮϮ͘ϳϭ

ϰ͘Ϯϱ

ϭϳ͘ϱϬ

Ϯϯ͘ϱϮ

ϯϭ͘ϴϬ

Ϯϰϭ͘ϱϳ

EĂŐĂƐĂŬŝ

ϭϭ͘ϱϱ

ϭ͘ϴϲ

Ϯ͘ϱϲ

Ϯ͘ϵϵ

ϰ͘ϭϬ

ϯ͘ϯϲ

ϭϰ͘ϲϰ

ϭ͘ϵϵ

Ϯ͘ϭϮ

ϰ͘ϭϴ

ϱϵ͘ϯϵ

ϱ͘ϰϱ

ϮϬ͘ϱϵ

Ϯϭϲ͘Ϯϭ

dŽĐŚŝŐŝ

ϭϭ͘ϲϮ

ϱ͘ϵϳ

ϵ͘Ϯϭ

ϰ͘ϲϬ

ϭϬ͘ϭϰ

ϴ͘ϲϮ

ϭϮ͘Ϯϱ

ϭϰ͘Ϯϱ

ϭϲ͘ϭϰ

ϭϮ͘ϲϭ

ϭϯ͘ϭϴ

ϭϵ͘Ϭϭ

Ϯϱ͘ϳϱ

ϭϵ͘ϵϯ

ŝĐŚŝ

ϭϭ͘ϵϴ

ϯ͘ϱϲ

ϳ͘ϱϴ

ϭϬ͘ϳϭ

ϳ͘ϵϵ

ϳ͘ϳϮ

ϴ͘ϭϬ

ϭϮ͘ϭϭ

ϵ͘Ϯϴ

ϭϲ͘ϵϲ

ϭϭ͘ϯϵ

ϮϮ͘Ϭϰ

Ϯϲ͘ϯϯ

ϭϲϰ͘ϯϵ

,LJŽŐŽ

ϭϮ͘ϳϵ

ϯ͘ϯϰ

ϲ͘ϵϴ

ϲ͘ϭϰ

ϵ͘ϵϯ

ϭϭ͘ϴϴ

ϭϮ͘ϵϭ

ϰ͘ϱϯ

ϭϱ͘ϴϯ

ϭϬ͘Ϭϲ

ϭϰ͘Ϭϴ

ϯϯ͘ϴϱ

Ϯϲ͘Ϭϰ

ϭϮϮ͘ϴϵ

,ŝƌŽƐŚŝŵĂ

ϭϮ͘ϴϲ

ϯ͘ϳϯ

ϲ͘ϯϯ

ϰ͘ϱϲ

ϱ͘ϵϮ

ϰ͘ϳϱ

ϴ͘ϯϳ

ϭϮ͘ϯϳ

ϭϴ͘Ϯϭ

ϭϮ͘ϱϳ

ϯϬ͘ϰϴ

ϯϳ͘ϭϵ

ϯϮ͘ϳϯ

ϮϮ͘Ϭϱ

^ŚŝŐĂ

ϭϮ͘ϵϳ

ϲ͘ϬϬ

ϭϯ͘ϲϴ

ϭϯ͘ϵϴ

ϳ͘ϵϳ

ϭϱ͘ϭϳ

ϵ͘Ϯϰ

ϭϯ͘ϳϯ

ϵ͘ϯϮ

ϭϳ͘ϱϭ

ϭϰ͘ϱϯ

ϭϳ͘ϴϲ

ϭϴ͘ϳϱ

ϲϯ͘ϳϴ

KŬĂLJĂŵĂ

ϭϰ͘ϬϬ

ϯ͘ϳϵ

ϱ͘Ϯϳ

ϳ͘ϰϲ

ϵ͘Ϭϱ

ϳ͘ϮϮ

ϴ͘ϰϰ

ϭϴ͘ϵϲ

ϳ͘Ϯϰ

ϭϬ͘ϵϯ

ϭϭ͘ϯϴ

ϰϱ͘ϱϰ

ϰϱ͘Ϯϴ

ϭϳϵ͘ϯϳ

ŚŝŵĞ

ϭϰ͘Ϭϵ

Ϯ͘ϭϵ

ϯ͘ϵϱ

ϭϳ͘ϳϲ

ϱ͘ϭϵ

ϰ͘ϳϰ

ϳ͘ϯϰ

ϭ͘ϰϭ

ϭϴ͘Ϭϳ

ϴ͘ϳϱ

Ϯϱ͘ϱϯ

ϰϯ͘Ϯϭ

ϱϭ͘ϯϮ

ϭϱϰ͘ϳϰ

/ďĂƌĂŬŝ

ϭϰ͘ϴϳ

ϰ͘ϴϮ

ϭϬ͘ϲϲ

ϰ͘ϵϲ

ϭϭ͘ϯϱ

ϴ͘ϱϲ

ϭϱ͘ϯϯ

ϭϯ͘ϬϬ

ϭϯ͘Ϭϱ

ϲ͘Ϯϯ

ϭϮ͘ϵϭ

ϯϲ͘ϵϳ

ϰϰ͘ϲϬ

ϭϯϵ͘ϯϬ

<ĂŶĂŐĂǁĂ

ϭϰ͘ϵϴ

ϱ͘ϲϰ

ϴ͘ϲϯ

ϰ͘Ϭϳ

ϭϬ͘ϯϰ

ϭϯ͘ϰϱ

ϭϭ͘ϳϵ

ϭϭ͘ϴϮ

ϵ͘ϰϯ

ϭϱ͘ϬϬ

ϭϰ͘Ϭϲ

ϯϲ͘ϭϰ

ϯϴ͘ϰϳ

ϭϱϮ͘ϵϭ

DŝĞ

ϭϱ͘ϰϵ

ϯ͘ϳϴ

ϴ͘ϳϲ

ϭϭ͘ϭϵ

ϳ͘ϯϲ

ϵ͘ϵϲ

ϵ͘Ϭϳ

ϱ͘ϲϭ

ϵ͘ϵϭ

ϭϲ͘Ϭϱ

ϮϬ͘ϳϭ

ϭϲ͘ϰϵ

ϰϴ͘ϵϬ

ϭϵϰ͘Ϯϰ

tĂŬĂLJĂŵĂ

ϭϲ͘ϯϲ

Ϯ͘ϴϬ

ϰ͘Ϭϲ

ϲ͘ϳϴ

ϰ͘ϵϴ

ϱ͘ϭϵ

ϳ͘ϲϮ

ϯϬ͘ϮϬ

ϵ͘ϭϭ

ϱ͘ϲϵ

ϳ͘Ϭϭ

ϱϯ͘ϵϴ

ϯϴ͘ϲϲ

Ϯϯϱ͘Ϭϳ

ŚŝďĂ

ϭϲ͘ϲϭ

ϰ͘ϲϵ

ϳ͘Ϭϯ

Ϯ͘ϭϵ

ϴ͘ϮϮ

ϭϬ͘ϵϯ

ϵ͘Ϭϭ

ϭϭ͘ϲϱ

ϵ͘ϬϬ

ϱ͘ϯϳ

ϭϵ͘ϴϱ

ϯϵ͘ϳϵ

ϱϬ͘ϲϲ

ϭϰϱ͘ϲϳ

KŝƚĂ

ϭϳ͘ϱϴ

Ϯ͘ϲϵ

ϰ͘Ϭϭ

ϯ͘ϳϲ

ϳ͘ϱϬ

ϭϭ͘ϴϱ

ϳ͘ϰϭ

ϭϳ͘ϴϳ

ϮϮ͘ϱϯ

ϰ͘ϴϲ

Ϯϴ͘Ϭϴ

ϲϭ͘ϰϵ

ϲϭ͘ϲϰ

ϭϰϴ͘ϲϱ

zĂŵĂŐƵĐŚŝ

ϮϬ͘ϱϮ

ϯ͘ϴϲ

ϳ͘ϵϲ

ϭϱ͘Ϭϴ

ϱ͘ϴϰ

ϭϰ͘ϳϱ

ϵ͘Ϯϳ

Ϯ͘ϮϮ

Ϯϳ͘ϵϯ

ϭϮ͘ϵϲ

ϮϬ͘Ϯϴ

ϯϬ͘ϴϬ

ϲϭ͘ϴϲ

ϭϰϭ͘Ϯϭ

EŽƚĞƐ͗dŚĞĐŽůŽƌŽĨĞĂĐŚĐĞůůŝŶĚŝĐĂƚĞƐƚŚĞĐĂƉŝƚĂůŝŶƚĞŶƐŝƚLJŽĨĂŐŝǀĞŶŝŶĚƵƐƚƌLJŝŶĂŐŝǀĞŶƉƌĞĨĞĐƚƵƌĞ͘tŚŝƚĞ͕ůŝŐŚƚŐƌĂLJ͕ĚĂƌŬŐƌĂLJ͕ĂŶĚďůĂĐŬ ŵĞĂŶƚŚĂƚƚŚĞĐĂƉŝƚĂůŝŶƚĞŶƐŝƚLJŝƐŝŶƚŚĞĨŝƌƐƚ͕ƐĞĐŽŶĚ͕ƚŚŝƌĚ͕ĂŶĚĨŽƵƌƚŚƋƵĂƌƚŝůĞƐ͕ƌĞƐƉĞĐƚŝǀĞůLJ͘dŚĞŝŶĚƵƐƚƌŝĞƐĂŶĚƚŚĞƉƌĞĨĞĐƚƵƌĞƐĂƌĞƐŽƌƚĞĚ ŝŶŽƌĚĞƌŽĨĐĂƉŝƚĂůŝŶƚĞŶƐŝƚLJĂŶĚƌĞůĂƚŝǀĞĐĂƉŝƚĂůĂďƵŶĚĂŶĐĞ͕ƌĞƐƉĞĐƚŝǀĞůLJ͘ Notes: The color of each cell indicates the capital intensity of a given industry in a given prefecture. ^ŽƵƌĐĞ͗Z/d/;ϮϬϭϰͿZͲ:/WϮϬϭϰ͘ Light gray, gray, dark gray, and black mean that the capital intensity is in the rst, second, third, and

fourth quartiles within each prefecture, respectively. The industries and the prefectures are sorted in order of capital intensity and relative capital abundance, respectively.

Source: RIETI (2014) R-JIP Database 2014.

20

Table 2: PrefectureIndustry Capital Intensity, Including Agriculture and Mining, 2005

dĂďůĞϮ͘WƌĞĨĞĐƚƵƌĞͲ/ŶĚƵƐƚƌLJĂƉŝƚĂů/ŶƚĞŶƐŝƚLJ͕ŝŶĐůƵĚŝŶŐŐƌŝĐƵůƚƵƌĞĂŶĚDŝŶŝŶŐ͕ϮϬϬϱ /ŶĚƵƐƚƌLJĂǀĞƌĂŐĞĐĂƉŝƚĂůůĂďŽƌƌĂƚŝŽ ϭϭ͘ϱϲ

 Ŷ Ě Ž ǁ ŵ Ğ Ŷ ƚ͗ Ɖ ƌĞ ĨĞ Đƚ Ƶ ƌĞ Đ Ă Ɖ ŝƚ Ă ůͲ

ϯ͘ϴϲ

ϲ͘ϭϯ

D Ğ ƚĂ ůƉ ƌŽ Ě Ƶ Đƚ Ɛ

K ƚŚ Ğ ƌ ŵ Ă Ŷ Ƶ ĨĂ Đƚ Ƶ ƌŝ Ŷ Ő

dŽŬLJŽ

ϱ͘ϭϰ

Ϯ͘ϱϬ

ϯ͘ϲϴ

ϭ͘ϯϳ

<ŽĐŚŝ

ϳ͘ϴϬ

ϭ͘Ϯϵ

ϯ͘ϱϲ

ϰ͘ϲϮ

ůĂ ď Ž ƌ ƌĞ ƚŝ Ž ; Ă ůů ƚƌ Ă Ě Ă ď ůĞ ƐͿ

ϲ͘Ϯϲ

d Ğ džƚ ŝůĞ Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

ϳ͘ϭϮ

ϳ͘ϴϯ

ϴ͘ϭϮ

ϭϭ͘ϱϱ

ϭϮ͘ϳϲ

ϭϯ͘Ϭϲ

d ƌĂ Ŷ ƐƉ Ž ƌƚ Ă ƚŝ Ž Ŷ ŵ Ă ĐŚ ŝŶ Ğ ƌLJ

ϭϯ͘ϭϰ

ϭϱ͘ϲϭ

 Őƌ ŝĐ Ƶ ůƚ Ƶ ƌĞ ͕ ĨŽ ƌĞ Ɛƚ ƌLJ Ă Ŷ Ě

ϭϲ͘ϳϮ

Ϯϱ͘ϱϯ

Ϯϳ͘ϵϵ

ϭϮϯ͘ϰϴ

 Ś Ğ ŵ ŝĐ Ă ůƉ ƌŽ Ě Ƶ Đƚ Ɛ

W Ğ ƚƌ Ž ůĞ Ƶ ŵ Ă Ŷ Ě Đ Ž Ă ůƉ ƌŽ Ě Ƶ Đƚ Ɛ

 Ğ ƌĂ ŵ ŝĐ ͕ Ɛƚ Ž Ŷ Ğ Ă Ŷ Ě Đ ůĂ LJ

' Ğ Ŷ Ğ ƌĂ ůŵ Ă ĐŚ ŝŶ Ğ ƌLJ

W ƌĞ Đŝ Ɛŝ Ž Ŷ ŵ Ă ĐŚ ŝŶ Ğ ƌLJ

ϱ͘ϭϱ

ϯ͘ϯϴ

ϰ͘ϭϬ

ϭϬ͘ϱϬ

ϰ͘Ϭϰ

ϵ͘ϳϴ

ϱ͘ϵϭ

ϭϲ͘ϰϬ

ϲ͘ϯϳ

ϲ͘ϭϳ

ϲ͘Ϭϰ

ϴ͘ϵϲ

ϯ͘ϯϮ

ϭϯ͘ϱϭ

ϲ͘Ϭϳ

ϴ͘Ϭϭ

ϴ͘ϵϲ

ϰ͘ϴϱ

ϭϰ͘ϲϮ

ϵ͘ϬϮ

ϵ͘ϬϮ

ϵ͘ϱϬ

ϭϮ͘ϯϭ

ϵ͘ϳϮ

&Ž Ž Ě Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

W Ƶ ůƉ Ă Ŷ Ě Ɖ Ă Ɖ Ğ ƌ

 ůĞ Đƚ ƌŝ ĐĂ ůŵ Ă ĐŚ ŝŶ Ğ ƌLJ

Ĩŝ ƐŚ Ğ ƌŝ Ğ Ɛ

D ŝŶ ŝŶ Ő

W ƌŝ ŵ Ă ƌLJ ŵ Ğ ƚĂ ů

WƌĞĨĞĐƚƵƌĞ

KƐĂŬĂ

ϴ͘ϭϲ

ϯ͘ϲϳ

ϱ͘Ϯϯ

ϱ͘Ϭϰ

ϲ͘ϲϰ

ϲ͘Ϭϴ

ϲ͘ϰϬ

ϳ͘ϯϱ

ϲ͘ϲϭ

ϭϬ͘ϴϯ

ϵ͘Ϯϯ

ϭϰ͘ϰϱ

ϵ͘ϰϰ

ϭϴ͘ϲϭ

ϭϴ͘ϭϱ

ϭϮϯ͘ϴϲ

'ŝĨƵ

ϴ͘Ϯϳ

ϯ͘ϳϮ

ϲ͘ϮϮ

ϱ͘ϯϴ

ϰ͘ϲϯ

ϱ͘Ϯϲ

ϳ͘ϬϬ

ϭϭ͘ϵϮ

ϭϯ͘ϳϳ

ϵ͘Ϯϵ

ϴ͘ϯϯ

ϭϲ͘ϲϮ

Ϯϯ͘ϭϬ

ϵ͘Ϭϰ

Ϯϭ͘ϰϭ

ϲϵ͘ϳϱ

zĂŵĂŶĂƐŚŝ

ϴ͘ϴϰ

ϯ͘ϮϬ

ϰ͘Ϭϵ

ϲ͘ϭϬ

ϴ͘ϴϯ

ϱ͘ϯϴ

ϭϬ͘ϯϴ

ϭϰ͘ϲϯ

ϰ͘ϰϰ

ϲ͘ϳϬ

ϭϲ͘ϳϮ

ϲ͘ϳϭ

ϭϰ͘ϯϲ

ϳ͘ϭϴ

Ϯϭ͘ϰϴ

ϯ͘ϯϭ

^ĂŝƚĂŵĂ

ϴ͘ϴϴ

ϯ͘ϱϯ

ϳ͘ϭϳ

ϯ͘ϴϭ

ϱ͘ϴϰ

ϴ͘ϯϯ

ϲ͘ϱϯ

ϭϮ͘ϲϳ

ϳ͘ϴϱ

ϭϭ͘Ϯϱ

ϵ͘ϰϱ

ϭϰ͘ϴϰ

ϴ͘ϵϰ

ϭϮ͘ϳϮ

ϭϴ͘ϰϲ

ϰϱ͘ϵϭ

<LJŽƚŽ

ϵ͘ϬϬ

ϯ͘ϭϲ

ϱ͘Ϭϰ

ϰ͘ϭϵ

ϭϬ͘Ϭϲ

ϳ͘ϯϱ

ϱ͘ϲϴ

ϭϭ͘ϰϭ

ϱ͘ϰϵ

ϭϵ͘ϯϲ

ϭϭ͘ϱϮ

ϭϰ͘ϵϭ

ϭϮ͘ϱϭ

ϭϭ͘ϯϰ

ϭϳ͘ϯϰ

Ϯϵ͘ϱϵ

^ĂŐĂ

ϵ͘ϱϯ

ϴ͘ϱϱ

ϳ͘Ϯϰ

ϯ͘ϭϵ

ϳ͘ϲϴ

Ϯ͘ϳϭ

ϲ͘ϲϴ

ϵ͘Ϯϵ

ϭϮ͘ϯϭ

ϳ͘ϴϯ

ϭϲ͘ϳϳ

ϭϬ͘ϰϲ

Ϯϱ͘ϱϵ

ϭϴ͘ϴϳ

ϭϲ͘ϲϬ

ϱϬ͘ϲϯ

zĂŵĂŐĂƚĂ

ϵ͘ϲϭ

ϯ͘ϲϱ

ϰ͘ϲϯ

Ϯ͘ϵϬ

ϰ͘ϰϯ

ϴ͘ϱϴ

ϱ͘ϴϮ

ϲ͘ϲϬ

ϱ͘Ϭϰ

ϴ͘ϱϳ

ϭϮ͘ϳϴ

ϭϯ͘ϰϭ

ϭϵ͘Ϭϱ

ϭϰ͘ϭϴ

Ϯϰ͘Ϭϴ

Ϯϵ͘ϳϴ

<ĂŐĂǁĂ

ϵ͘ϴϱ

ϯ͘ϱϮ

ϲ͘ϭϬ

ϰ͘ϱϬ

ϱ͘ϰϴ

ϱ͘ϱϬ

ϵ͘ϬϬ

ϯ͘ϵϱ

ϭϭ͘Ϭϴ

ϳ͘ϯϴ

ϲ͘ϭϰ

ϭϯ͘ϮϮ

ϭϴ͘ϵϲ

ϮϮ͘ϱϮ

Ϯϯ͘ϱϳ

ϭϰϲ͘ϰϱ Ϯϲ͘ϵϭ

EĂƌĂ

ϵ͘ϵϰ

ϰ͘ϰϬ

ϱ͘ϵϵ

ϰ͘ϴϱ

ϵ͘ϰϲ

ϰ͘ϵϯ

ϭϱ͘ϰϵ

Ϯ͘ϴϰ

ϰ͘ϳϬ

ϵ͘ϭϵ

ϭϲ͘ϭϰ

ϭϰ͘ϭϱ

Ϯϲ͘ϭϬ

ϭϭ͘ϵϯ

ϵ͘ϭϬ

^ŚŝŵĂŶĞ

ϭϬ͘ϭϴ

ϯ͘ϴϬ

ϱ͘ϭϯ

ϲ͘ϲϲ

ϯ͘ϭϰ

ϱ͘ϱϲ

ϳ͘ϭϰ

ϭϰ͘ϬϮ

ϳ͘Ϯϰ

ϭϬ͘Ϭϲ

ϵ͘ϳϱ

ϭϱ͘ϭϬ

ϴ͘ϯϲ

ϭϭ͘ϱϲ

ϴ͘Ϭϲ

Ϯϱ͘ϳϬ

/ƐŚŝŬĂǁĂ

ϭϬ͘ϰϱ

Ϯ͘ϱϵ

ϰ͘ϴϱ

ϭϬ͘ϭϮ

ϰ͘ϳϮ

ϯ͘ϴϵ

ϳ͘ϵϱ

ϲ͘ϭϯ

ϯ͘ϳϯ

ϰ͘ϲϴ

ϭϯ͘ϭϱ

Ϯϱ͘ϵϬ

Ϯϳ͘ϴϵ

ϭϬ͘ϳϲ

ϭϵ͘Ϯϴ

ϲϯ͘ϰϮ

<ĂŐŽƐŚŝŵĂ

ϭϬ͘ϱϭ

ϯ͘ϰϱ

ϯ͘Ϭϴ

ϯ͘ϭϲ

ϱ͘ϭϭ

ϴ͘ϳϳ

ϱ͘Ϯϲ

ϰ͘ϴϬ

ϭϳ͘ϲϮ

ϯ͘ϯϬ

ϭϯ͘ϴϭ

ϭϯ͘ϲϳ

ϭϵ͘Ϭϳ

ϭϮ͘ϰϮ

ϭϲ͘ϱϰ

ϯϮ͘ϳϮ

<ƵŵĂŵŽƚŽ

ϭϬ͘ϱϯ

ϰ͘ϰϯ

ϲ͘ϳϭ

ϰ͘Ϯϵ

ϲ͘ϭϲ

ϱ͘ϲϯ

ϳ͘ϯϲ

ϯ͘ϴϯ

Ϯϰ͘ϯϮ

ϳ͘Ϭϲ

Ϯϯ͘ϳϲ

ϭϬ͘ϭϳ

ϭϵ͘ϴϰ

ϭϱ͘ϯϯ

Ϯϯ͘Ϭϯ

ϲϯ͘ϯϰ

DŝLJĂnjĂŬŝ

ϭϬ͘ϱϯ

Ϯ͘Ϯϲ

ϱ͘ϱϳ

ϭϮ͘ϴϯ

ϰ͘ϲϬ

ϰ͘ϭϴ

ϯ͘ϳϰ

ϭϳ͘Ϭϵ

Ϯϯ͘ϰϬ

ϰ͘ϴϮ

ϭϭ͘ϭϮ

ϭϮ͘ϲϲ

ϭϬ͘ϱϭ

ϭϮ͘ϵϭ

ϰϬ͘ϴϱ

ϲϱ͘ϲϵ

EĂŐĂŶŽ

ϭϬ͘ϳϵ

ϯ͘ϳϯ

ϱ͘Ϯϱ

ϲ͘ϵϭ

ϳ͘ϲϳ

ϳ͘ϯϴ

ϴ͘ϯϰ

ϭϵ͘ϲϬ

ϱ͘ϵϰ

ϴ͘ϵϴ

ϭϰ͘ϰϭ

ϭϮ͘ϰϳ

ϭϴ͘ϰϲ

ϴ͘Ϯϵ

ϭϰ͘ϮϮ

ϯϵ͘ϳϲ

ŬŝƚĂ

ϭϬ͘ϵϱ

Ϯ͘ϵϰ

ϯ͘ϰϵ

ϭ͘Ϯϲ

ϯ͘ϴϮ

ϰ͘ϰϱ

ϰ͘ϯϳ

ϭϴ͘ϲϮ

ϯϯ͘ϭϭ

ϳ͘ϴϭ

ϭϬ͘ϵϴ

ϭϲ͘ϱϴ

ϭϲ͘ϰϭ

ϭϲ͘ϴϰ

ϯϰ͘ϭϳ

ϴϱ͘ϭϵ

/ǁĂƚĞ

ϭϭ͘ϬϬ

ϯ͘ϭϵ

ϯ͘ϵϮ

Ϯ͘ϳϭ

ϯ͘ϴϵ

ϵ͘Ϭϳ

ϱ͘ϱϳ

ϭϬ͘ϳϵ

ϭϴ͘ϭϲ

ϴ͘ϵϳ

ϭϯ͘ϲϬ

ϭϱ͘Ϭϭ

ϲ͘ϰϵ

Ϯϴ͘Ϭϴ

ϯϯ͘ϰϯ

ϱϵ͘ϳϴ

&ƵŬƵƐŚŝŵĂ

ϭϭ͘Ϭϵ

ϰ͘ϯϵ

ϲ͘Ϯϵ

Ϯ͘ϳϵ

ϴ͘ϯϳ

ϳ͘ϯϯ

ϲ͘ϭϳ

ϵ͘ϲϭ

ϭϯ͘ϳϰ

ϭϬ͘ϯϱ

ϭϱ͘Ϭϳ

ϭϭ͘ϴϮ

ϭϬ͘ϰϱ

ϭϰ͘ϴϱ

ϰϱ͘ϰϲ

ϮϭϬ͘ϲϯ

EŝŝŐĂƚĂ

ϭϭ͘ϭϴ

ϯ͘ϲϭ

ϰ͘ϲϲ

ϰ͘ϱϭ

ϱ͘ϱϲ

ϳ͘ϳϮ

ϲ͘ϰϵ

ϲ͘ϯϵ

ϭϱ͘ϰϭ

ϱ͘ϲϳ

ϭϯ͘ϮϮ

ϭϴ͘ϰϱ

ϭϰ͘ϰϬ

ϭϱ͘ϯϮ

ϱϬ͘ϵϯ

ϭϮϵ͘Ϯϱ

'ƵŵŵĂ

ϭϭ͘Ϯϱ

ϯ͘ϴϳ

ϳ͘Ϯϰ

ϯ͘ϲϵ

ϭϭ͘ϯϱ

ϳ͘Ϯϲ

ϳ͘ϱϱ

ϳ͘ϮϮ

ϴ͘ϭϲ

ϭϮ͘ϵϮ

ϭϮ͘ϯϳ

ϭϲ͘ϯϬ

ϭϱ͘ϯϵ

ϭϰ͘ϯϬ

ϯϬ͘ϯϬ

ϯϴ͘ϭϲ

^ŚŝnjƵŽŬĂ

ϭϭ͘Ϯϲ

ϯ͘ϲϭ

ϳ͘ϰϰ

ϮϮ͘ϭϴ

ϴ͘ϰϵ

ϳ͘ϰϴ

ϲ͘ϰϯ

ϭϯ͘ϱϯ

ϭϴ͘Ϭϲ

ϭϮ͘ϱϳ

ϵ͘ϴϭ

ϭϱ͘Ϯϴ

ϵ͘ϲϯ

ϭϰ͘ϴϰ

ϯϭ͘ϴϭ

ϯϱ͘ϱϬ

dŽƚƚŽƌŝ

ϭϭ͘ϱϭ

ϯ͘ϴϴ

ϯ͘ϱϯ

ϯ͘ϭϵ

ϴ͘ϭϯ

ϰ͘Ϯϲ

ϯ͘ϵϱ

ϱ͘Ϯϭ

Ϯϱ͘Ϯϰ

ϯ͘ϯϰ

ϭϬ͘ϭϵ

ϭϲ͘ϴϯ

ϯϬ͘ϳϭ

ϴ͘ϭϭ

ϰ͘ϯϳ

ϯϳ͘ϵϴ

ŽŵŽƌŝ

ϭϭ͘ϱϵ

Ϯ͘ϲϮ

Ϯ͘ϲϮ

ϰ͘ϲϲ

ϰ͘ϵϱ

ϳ͘ϲϬ

ϳ͘ϵϮ

ϯ͘ϵϱ

Ϯϴ͘ϵϬ

Ϯ͘ϴϭ

ϳ͘ϰϬ

ϭϮ͘ϳϰ

ϮϬ͘ϱϯ

ϲϮ͘ϱϬ

ϯϭ͘ϳϱ

ϲϰ͘ϱϮ

dŽĐŚŝŐŝ

ϭϮ͘ϭϮ

ϱ͘ϵϳ

ϵ͘Ϯϭ

ϰ͘ϲϬ

ϭϬ͘ϭϰ

ϴ͘ϲϮ

ϭϮ͘Ϯϱ

ϭϰ͘Ϯϱ

ϭϲ͘ϭϰ

ϭϮ͘ϲϭ

ϭϯ͘ϭϴ

ϭϰ͘ϭϮ

ϭϱ͘ϳϲ

ϭϵ͘Ϭϭ

Ϯϱ͘ϳϱ

ϭϵ͘ϵϯ

&ƵŬƵŽŬĂ

ϭϮ͘ϭϮ

ϯ͘ϴϳ

ϱ͘ϯϯ

Ϯ͘ϰϱ

ϲ͘ϰϮ

ϴ͘ϯϯ

ϱ͘ϵϰ

ϲ͘ϱϵ

ϱ͘Ϯϯ

ϭϲ͘ϲϲ

ϭϮ͘Ϯϯ

ϭϳ͘ϯϳ

ϯϲ͘Ϯϭ

ϯϲ͘ϯϮ

ϯϴ͘ϳϰ

ϵϱ͘ϭϯ

ŝĐŚŝ

ϭϮ͘ϭϳ

ϯ͘ϱϲ

ϳ͘ϱϴ

ϭϬ͘ϳϭ

ϳ͘ϵϵ

ϳ͘ϳϮ

ϴ͘ϭϬ

ϭϮ͘ϭϭ

ϵ͘Ϯϴ

ϭϲ͘ϵϲ

ϭϭ͘ϯϵ

ϭϰ͘ϰϴ

ϰ͘ϵϬ

ϮϮ͘Ϭϰ

Ϯϲ͘ϯϯ

ϭϲϰ͘ϯϵ

EĂŐĂƐĂŬŝ

ϭϮ͘ϭϳ

ϭ͘ϴϲ

Ϯ͘ϱϲ

Ϯ͘ϵϵ

ϰ͘ϭϬ

ϯ͘ϯϲ

ϭϰ͘ϲϰ

ϭ͘ϵϵ

Ϯ͘ϭϮ

ϰ͘ϭϴ

ϱϵ͘ϯϵ

ϭϮ͘ϵϳ

Ϯϭ͘ϰϱ

ϱ͘ϰϱ

ϮϬ͘ϱϵ

Ϯϭϲ͘Ϯϭ

&ƵŬƵŝ

ϭϮ͘ϭϵ

ϯ͘Ϯϵ

ϱ͘ϭϲ

ϭϬ͘ϭϮ

ϯ͘ϱϵ

ϭϬ͘ϰϬ

ϰ͘ϲϳ

ϱ͘ϭϮ

ϴ͘ϳϳ

ϭϮ͘ϭϲ

ϭϯ͘Ϭϵ

Ϯϲ͘ϲϱ

ϵ͘ϱϭ

Ϯϰ͘Ϯϴ

Ϯϭ͘ϲϯ

ϱϴ͘ϲϭ

KŬŝŶĂǁĂ

ϭϮ͘ϲϮ

Ϯ͘ϴϬ

Ϯ͘ϴϰ

Ϯ͘ϬϬ

ϲ͘ϲϭ

ϰ͘ϯϯ

ϲ͘ϮϮ

Ϭ͘Ϭϴ

ϱ͘ϵϱ

ϭ͘ϯϮ

ϭ͘ϱϭ

ϭϴ͘ϴϲ

ϭϱ͘ϯϬ

ϭϬ͘Ϭϱ

ϱ͘ϯϲ

ϭϰϱ͘ϳϰ

,ŝƌŽƐŚŝŵĂ

ϭϮ͘ϴϴ

ϯ͘ϳϯ

ϲ͘ϯϯ

ϰ͘ϱϲ

ϱ͘ϵϮ

ϰ͘ϳϱ

ϴ͘ϯϳ

ϭϮ͘ϯϳ

ϭϴ͘Ϯϭ

ϭϮ͘ϱϳ

ϯϬ͘ϰϴ

ϭϮ͘ϵϯ

ϭϳ͘ϱϮ

ϯϳ͘ϭϵ

ϯϮ͘ϳϯ

ϮϮ͘Ϭϱ

DŝLJĂŐŝ

ϭϯ͘Ϭϵ

ϲ͘Ϭϭ

ϳ͘ϱϵ

ϱ͘ϯϯ

ϲ͘ϯϱ

ϲ͘ϵϯ

ϱ͘ϭϰ

ϱ͘ϳϱ

Ϯϱ͘ϬϬ

ϱ͘ϳϳ

ϭϮ͘ϭϱ

ϮϮ͘ϰϭ

ϭϯ͘ϵϳ

ϮϮ͘ϳϱ

ϭϬ͘ϵϳ

ϭϳϰ͘ϵϴ

ŚŝŵĞ

ϭϯ͘ϭϵ

Ϯ͘ϭϵ

ϯ͘ϵϱ

ϭϳ͘ϳϲ

ϱ͘ϭϵ

ϰ͘ϳϰ

ϳ͘ϯϰ

ϭ͘ϰϭ

ϭϴ͘Ϭϳ

ϴ͘ϳϱ

Ϯϱ͘ϱϯ

ϭϭ͘ϱϮ

ϭϬ͘ϯϲ

ϰϯ͘Ϯϭ

ϱϭ͘ϯϮ

ϭϱϰ͘ϳϰ

dŽŬƵƐŚŝŵĂ

ϭϯ͘ϰϰ

ϯ͘Ϭϵ

ϱ͘ϱϰ

ϲ͘ϰϰ

ϴ͘Ϭϵ

ϯ͘Ϯϯ

ϲ͘ϵϯ

ϭ͘ϯϲ

ϮϮ͘ϳϭ

ϰ͘Ϯϱ

ϭϳ͘ϱϬ

ϭϲ͘ϵϮ

ϳ͘ϭϭ

Ϯϯ͘ϱϮ

ϯϭ͘ϴϬ

Ϯϰϭ͘ϱϳ

KŬĂLJĂŵĂ

ϭϯ͘ϲϯ

ϯ͘ϳϵ

ϱ͘Ϯϳ

ϳ͘ϰϲ

ϵ͘Ϭϱ

ϳ͘ϮϮ

ϴ͘ϰϰ

ϭϴ͘ϵϲ

ϳ͘Ϯϰ

ϭϬ͘ϵϯ

ϭϭ͘ϯϴ

ϭϮ͘Ϯϳ

ϭϳ͘ϯϲ

ϰϱ͘ϱϰ

ϰϱ͘Ϯϴ

ϭϳϵ͘ϯϳ

,LJŽŐŽ

ϭϯ͘ϵϳ

ϯ͘ϯϰ

ϲ͘ϵϴ

ϲ͘ϭϰ

ϵ͘ϵϯ

ϭϭ͘ϴϴ

ϭϮ͘ϵϭ

ϰ͘ϱϯ

ϭϱ͘ϴϯ

ϭϬ͘Ϭϲ

ϭϰ͘Ϭϴ

Ϯϭ͘ϲϯ

ϭϬϳ͘ϰϳ

ϯϯ͘ϴϱ

Ϯϲ͘Ϭϰ

ϭϮϮ͘ϴϵ ϭϮϴ͘ϴϳ

dŽLJĂŵĂ

ϭϰ͘ϳϮ

ϱ͘ϯϭ

ϲ͘ϯϯ

ϭϬ͘ϯϱ

ϱ͘ϱϮ

ϳ͘ϭϰ

ϴ͘ϱϲ

ϭϮ͘ϱϮ

ϮϬ͘Ϭϯ

ϴ͘ϭϭ

ϭϳ͘ϭϲ

ϯϳ͘ϴϯ

ϮϬ͘ϰϬ

Ϯϳ͘ϭϬ

Ϯϳ͘ϱϰ

<ĂŶĂŐĂǁĂ

ϭϰ͘ϵϮ

ϱ͘ϲϰ

ϴ͘ϲϯ

ϰ͘Ϭϳ

ϭϬ͘ϯϰ

ϭϯ͘ϰϱ

ϭϭ͘ϳϵ

ϭϭ͘ϴϮ

ϵ͘ϰϯ

ϭϱ͘ϬϬ

ϭϰ͘Ϭϲ

ϭϰ͘Ϯϰ

ϳ͘ϭϵ

ϯϲ͘ϭϰ

ϯϴ͘ϰϳ

ϭϱϮ͘ϵϭ

ŚŝďĂ

ϭϱ͘Ϯϴ

ϰ͘ϲϵ

ϳ͘Ϭϯ

Ϯ͘ϭϵ

ϴ͘ϮϮ

ϭϬ͘ϵϯ

ϵ͘Ϭϭ

ϭϭ͘ϲϱ

ϵ͘ϬϬ

ϱ͘ϯϳ

ϭϵ͘ϴϱ

ϭϭ͘ϭϵ

Ϯϰ͘ϰϱ

ϯϵ͘ϳϵ

ϱϬ͘ϲϲ

ϭϰϱ͘ϲϳ

/ďĂƌĂŬŝ

ϭϱ͘ϰϮ

ϰ͘ϴϮ

ϭϬ͘ϲϲ

ϰ͘ϵϲ

ϭϭ͘ϯϱ

ϴ͘ϱϲ

ϭϱ͘ϯϯ

ϭϯ͘ϬϬ

ϭϯ͘Ϭϱ

ϲ͘Ϯϯ

ϭϮ͘ϵϭ

ϭϳ͘ϭϮ

Ϯϵ͘ϴϬ

ϯϲ͘ϵϳ

ϰϰ͘ϲϬ

ϭϯϵ͘ϯϬ

tĂŬĂLJĂŵĂ

ϭϱ͘ϱϬ

Ϯ͘ϴϬ

ϰ͘Ϭϲ

ϲ͘ϳϴ

ϰ͘ϵϴ

ϱ͘ϭϵ

ϳ͘ϲϮ

ϯϬ͘ϮϬ

ϵ͘ϭϭ

ϱ͘ϲϵ

ϳ͘Ϭϭ

ϭϰ͘Ϭϰ

ϭϰϱ͘ϲϬ

ϱϯ͘ϵϴ

ϯϴ͘ϲϲ

Ϯϯϱ͘Ϭϳ

^ŚŝŐĂ

ϭϱ͘ϴϯ

ϲ͘ϬϬ

ϭϯ͘ϲϴ

ϭϯ͘ϵϴ

ϳ͘ϵϳ

ϭϱ͘ϭϳ

ϵ͘Ϯϰ

ϭϯ͘ϳϯ

ϵ͘ϯϮ

ϭϳ͘ϱϭ

ϭϰ͘ϱϯ

ϯϴ͘ϱϵ

ϭϳ͘ϲϬ

ϭϳ͘ϴϲ

ϭϴ͘ϳϱ

ϲϯ͘ϳϴ

,ŽŬŬĂŝĚŽ

ϭϲ͘ϭϲ

ϯ͘ϴϱ

ϱ͘Ϭϲ

ϯ͘ϰϮ

ϲ͘ϱϰ

ϴ͘Ϯϭ

ϲ͘ϯϲ

Ϯ͘ϴϯ

ϯϵ͘ϭϲ

ϭϳ͘ϱϵ

ϭϱ͘Ϯϱ

Ϯϯ͘ϴϰ

ϭϯ͘ϲϬ

ϯϯ͘ϵϰ

ϮϮ͘ϵϴ

ϮϮϳ͘ϭϭ

KŝƚĂ

ϭϲ͘ϮϮ

Ϯ͘ϲϵ

ϰ͘Ϭϭ

ϯ͘ϳϲ

ϳ͘ϱϬ

ϭϭ͘ϴϱ

ϳ͘ϰϭ

ϭϳ͘ϴϳ

ϮϮ͘ϱϯ

ϰ͘ϴϲ

Ϯϴ͘Ϭϴ

ϭϰ͘Ϭϵ

ϵ͘ϰϯ

ϲϭ͘ϰϵ

ϲϭ͘ϲϰ

ϭϰϴ͘ϲϱ

DŝĞ

ϭϲ͘ϰϭ

ϯ͘ϳϴ

ϴ͘ϳϲ

ϭϭ͘ϭϵ

ϳ͘ϯϲ

ϵ͘ϵϲ

ϵ͘Ϭϳ

ϱ͘ϲϭ

ϵ͘ϵϭ

ϭϲ͘Ϭϱ

ϮϬ͘ϳϭ

Ϯϭ͘ϳϮ

ϭϲ͘ϱϯ

ϭϲ͘ϰϵ

ϰϴ͘ϵϬ

ϭϵϰ͘Ϯϰ

zĂŵĂŐƵĐŚŝ

ϭϳ͘ϴϭ

ϯ͘ϴϲ

ϳ͘ϵϲ

ϭϱ͘Ϭϴ

ϱ͘ϴϰ

ϭϰ͘ϳϱ

ϵ͘Ϯϳ

Ϯ͘ϮϮ

Ϯϳ͘ϵϯ

ϭϮ͘ϵϲ

ϮϬ͘Ϯϴ

ϭϬ͘ϰϯ

ϭϭ͘ϵϲ

ϯϬ͘ϴϬ

ϲϭ͘ϴϲ

ϭϰϭ͘Ϯϭ

EŽƚĞƐĂŶĚƐŽƵƌĐĞ͗^ĞĞdĂďůĞϭ͘

Notes and source: See Table 1.

21

Table 3: RegionIndustry Capital Intensity, Manufacturing, 2005

dĂďůĞϯ͘ZĞŐŝŽŶͲ/ŶĚƵƐƚƌLJĂƉŝƚĂů/ŶƚĞŶƐŝƚLJ͕DĂŶƵĨĂĐƚƵƌŝŶŐ͕ϮϬϬϱ /ŶĚƵƐƚƌLJĂǀĞƌĂŐĞĐĂƉŝƚĂůůĂďŽƌƌĂƚŝŽ ϭϬ͘ϱϱ

ůĂ ď Ž ƌ ƌĞ ƚŝ Ž ; ŵ Ă Ŷ Ƶ ĨĂ Đƚ Ƶ ƌŝ Ŷ Ő ƚŽ ƚĂ ůͿ

 Ŷ Ě Ž ǁ ŵ Ğ Ŷ ƚ͗ Ɖ ƌĞ ĨĞ Đƚ Ƶ ƌĞ Đ Ă Ɖ ŝƚ Ă ůͲ

ϯ͘ϴϲ

ϲ͘ϭϯ

ϲ͘Ϯϲ

K ƚŚ Ğ ƌ ŵ Ă Ŷ Ƶ ĨĂ Đƚ Ƶ ƌŝ Ŷ Ő

D Ğ ƚĂ ůƉ ƌŽ Ě Ƶ Đƚ Ɛ

d Ğ džƚ ŝůĞ Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

ϳ͘ϭϮ

&Ž Ž Ě Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

ϳ͘ϴϯ

 Ğ ƌĂ ŵ ŝĐ ͕Ɛ ƚŽ Ŷ Ğ Ă Ŷ Ě Đ ůĂ LJ Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

ϴ͘ϭϮ

ϭϭ͘ϱϱ

' Ğ Ŷ Ğ ƌĂ ůŵ Ă ĐŚ ŝŶ Ğ ƌLJ

W ƌĞ Đŝ Ɛŝ Ž Ŷ ŵ Ă ĐŚ ŝŶ Ğ ƌLJ

ϭϮ͘ϳϲ

ϭϯ͘Ϭϲ

ϭϯ͘ϭϰ

d ƌĂ Ŷ ƐƉ Ž ƌƚ Ă ƚŝ Ž Ŷ ŵ Ă ĐŚ ŝŶ Ğ ƌLJ

W Ƶ ůƉ Ă Ŷ Ě Ɖ Ă Ɖ Ğ ƌ

Ϯϱ͘ϱϯ

 ůĞ Đƚ ƌŝ ĐĂ ůŵ Ă ĐŚ ŝŶ Ğ ƌLJ

W ƌŝ ŵ Ă ƌLJ ŵ Ğ ƚĂ ů

Ϯϳ͘ϵϵ

ϭϮϯ͘ϰϴ

 Ś Ğ ŵ ŝĐ Ă ůƉ ƌŽ Ě Ƶ Đƚ Ɛ

W Ğ ƚƌ Ž ůĞ Ƶ ŵ Ă Ŷ Ě Đ Ž Ă ůƉ ƌŽ Ě Ƶ Đƚ Ɛ

ϭϯϱ͘ϴϬ

ZĞŐŝŽŶ ϵ͘ϭϳ

ϰ͘ϭϮ

ϱ͘ϯϰ

ϯ͘ϬϬ

ϱ͘ϱϯ

ϳ͘ϰϰ

ϱ͘ϳϬ

ϵ͘ϴϳ

ϮϬ͘Ϭϭ

ϴ͘ϰϭ

ϭϮ͘ϳϲ

Ϯϲ͘Ϯϯ

ϯϰ͘ϴϲ

,ŽŬŬĂŝĚŽ

ϭϬ͘Ϭϭ

ϯ͘ϴϱ

ϱ͘Ϭϲ

ϯ͘ϰϮ

ϲ͘ϱϰ

ϴ͘Ϯϭ

ϲ͘ϯϲ

Ϯ͘ϴϯ

ϯϵ͘ϭϲ

ϭϳ͘ϱϵ

ϭϱ͘Ϯϱ

ϯϯ͘ϵϰ

ϮϮ͘ϵϴ

ϮϮϳ͘ϭϭ

<ĂŶƚŽ

ϭϬ͘ϭϵ

ϰ͘ϭϰ

ϲ͘ϭϬ

Ϯ͘ϵϲ

ϴ͘ϭϴ

ϴ͘ϰϲ

ϴ͘ϴϱ

ϭϭ͘ϳϭ

ϴ͘ϭϬ

ϭϮ͘Ϯϰ

ϭϬ͘ϵϮ

Ϯϱ͘ϯϭ

Ϯϯ͘ϳϮ

ϭϬϬ͘ϰϯ

ŚƵďƵ

ϭϬ͘ϰϮ

ϯ͘ϳϰ

ϲ͘ϱϵ

ϵ͘ϱϰ

ϳ͘ϭϰ

ϲ͘ϴϯ

ϳ͘ϲϳ

ϭϯ͘Ϭϴ

ϭϯ͘ϱϬ

ϭϰ͘ϳϰ

ϭϮ͘ϯϴ

ϭϴ͘ϯϳ

Ϯϵ͘ϭϴ

ϭϭϯ͘ϬϮ

<LJƵƐŚƵ

ϭϬ͘ϲϬ

ϯ͘ϴϯ

ϱ͘ϭϰ

ϰ͘ϯϵ

ϱ͘ϵϯ

ϲ͘ϲϬ

ϳ͘ϭϯ

ϭϮ͘ϳϯ

ϭϮ͘ϵϰ

ϴ͘ϵϴ

ϮϬ͘ϰϯ

ϯϱ͘ϯϲ

ϯϱ͘ϯϱ

ϭϬϵ͘ϵϰ

<ŝŶŬŝ

ϭϬ͘ϳϯ

ϯ͘ϳϯ

ϲ͘ϱϳ

ϱ͘ϵϵ

ϴ͘Ϯϲ

ϭϬ͘ϬϬ

ϴ͘ϲϭ

ϵ͘ϳϱ

ϴ͘ϲϲ

ϭϯ͘ϰϳ

ϭϯ͘ϱϮ

Ϯϱ͘Ϯϱ

Ϯϯ͘ϳϯ

ϭϱϭ͘ϮϬ

^ŚŝŬŽŬƵ

ϭϬ͘ϵϲ

Ϯ͘ϴϮ

ϱ͘Ϭϯ

ϭϬ͘ϲϳ

ϱ͘ϱϮ

ϲ͘ϱϮ

ϳ͘ϰϵ

Ϯ͘ϴϴ

ϭϲ͘ϯϵ

ϳ͘ϲϮ

ϭϳ͘ϰϲ

Ϯϵ͘ϮϮ

ϯϳ͘ϬϬ

ϭϰϱ͘ϲϯ

ŚƵŐŽŬƵ

ϭϯ͘ϳϱ

ϯ͘ϳϴ

ϲ͘ϬϮ

ϲ͘ϴϯ

ϲ͘ϲϭ

ϴ͘ϭϳ

ϴ͘Ϯϰ

ϭϮ͘ϰϵ

ϭϲ͘ϴϬ

ϭϮ͘Ϭϲ

ϭϳ͘ϳϰ

ϯϰ͘ϰϵ

ϰϵ͘ϭϯ

ϭϰϯ͘Ϭϲ

dŽŚŽŬƵ

EŽƚĞƐ͗WƌĞĨĞĐƚƵƌĞƐĂƌĞĂŐŐƌĞŐĂƚĞĚŝŶƚŽĞŝŐŚƚƌĞŐŝŽŶƐ͘&ŽƌŽƚŚĞƌŶŽƚĞƐĂŶĚƐŽƵƌĐĞ͗^ĞĞdĂďůĞϭ͘

Notes: Prefectures are aggregated into eight regions. For other notes and source, see Table 1.

Table 4: RegionIndustry Capital Intensity, Including Agriculture and Mining, 2005

dĂďůĞϰ͘ZĞŐŝŽŶͲ/ŶĚƵƐƚƌLJĂƉŝƚĂů/ŶƚĞŶƐŝƚLJ͕ŝŶĐůƵĚŝŶŐŐƌŝĐƵůƚƵƌĞĂŶĚDŝŶŝŶŐ͕ϮϬϬϱ /ŶĚƵƐƚƌLJĂǀĞƌĂŐĞĐĂƉŝƚĂůůĂďŽƌƌĂƚŝŽ ϭϭ͘ϱϲ

ůĂ ď Ž ƌ ƌĞ ƚŝ Ž ; ŵ Ă Ŷ Ƶ ĨĂ Đƚ Ƶ ƌŝ Ŷ Ő ƚŽ ƚĂ ůͿ

 Ŷ Ě Ž ǁ ŵ Ğ Ŷ ƚ͗ Ɖ ƌĞ ĨĞ Đƚ Ƶ ƌĞ Đ Ă Ɖ ŝƚ Ă ůͲ

ϯ͘ϴϲ

ϲ͘ϭϯ

D Ğ ƚĂ ůƉ ƌŽ Ě Ƶ Đƚ Ɛ

K ƚŚ Ğ ƌ ŵ Ă Ŷ Ƶ ĨĂ Đƚ Ƶ ƌŝ Ŷ Ő

ϲ͘Ϯϲ

d Ğ džƚ ŝůĞ Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

ϳ͘ϭϮ

&Ž Ž Ě Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

ϳ͘ϴϯ

 Ğ ƌĂ ŵ ŝĐ ͕Ɛ ƚŽ Ŷ Ğ Ă Ŷ Ě Đ ůĂ LJ Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

ϴ͘ϭϮ

ϭϭ͘ϱϱ

' Ğ Ŷ Ğ ƌĂ ůŵ Ă ĐŚ ŝŶ Ğ ƌLJ

W ƌĞ Đŝ Ɛŝ Ž Ŷ ŵ Ă ĐŚ ŝŶ Ğ ƌLJ

ϭϮ͘ϳϲ

W Ƶ ůƉ Ă Ŷ Ě Ɖ Ă Ɖ Ğ ƌ

ϭϯ͘Ϭϲ

d ƌĂ Ŷ ƐƉ Ž ƌƚ Ă ƚŝ Ž Ŷ ŵ Ă ĐŚ ŝŶ Ğ ƌLJ

ϭϯ͘ϭϰ

 ůĞ Đƚ ƌŝ ĐĂ ůŵ Ă ĐŚ ŝŶ Ğ ƌLJ

ϭϱ͘ϲϭ

 Őƌ ŝĐ Ƶ ůƚ Ƶ ƌĞ ͕Ĩ Ž ƌĞ Ɛƚ ƌLJ Ă Ŷ Ě Ĩ ŝƐ Ś Ğ ƌŝ Ğ Ɛ

ϭϲ͘ϳϮ

Ϯϱ͘ϱϯ

Ϯϳ͘ϵϵ

ϭϮϯ͘ϰϴ

D ŝŶ ŝŶ Ő

W ƌŝ ŵ Ă ƌLJ ŵ Ğ ƚĂ ů

 Ś Ğ ŵ ŝĐ Ăů Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

W Ğ ƚƌ Ž ůĞ Ƶ ŵ Ă Ŷ Ě Đ Ž Ăů Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

ZĞŐŝŽŶ <ĂŶƚŽ

ϭϬ͘ϳϰ

ϰ͘ϭϰ

ϲ͘ϭϬ

Ϯ͘ϵϲ

ϴ͘ϭϴ

ϴ͘ϰϲ

ϴ͘ϴϱ

ϭϭ͘ϳϭ

ϴ͘ϭϬ

ϭϮ͘Ϯϰ

ϭϬ͘ϵϮ

ϭϰ͘ϲϳ

ϭϯ͘ϵϱ

Ϯϱ͘ϯϭ

Ϯϯ͘ϳϮ

ϭϬϬ͘ϰϯ

dŽŚŽŬƵ

ϭϭ͘Ϯϰ

ϰ͘ϭϮ

ϱ͘ϯϰ

ϯ͘ϬϬ

ϱ͘ϱϯ

ϳ͘ϰϰ

ϱ͘ϳϬ

ϵ͘ϴϳ

ϮϬ͘Ϭϭ

ϴ͘ϰϭ

ϭϮ͘ϳϲ

ϭϰ͘ϵϱ

ϭϰ͘ϭϮ

Ϯϲ͘Ϯϯ

ϯϰ͘ϴϲ

ϭϯϱ͘ϴϬ

ŚƵďƵ

ϭϭ͘ϯϯ

ϯ͘ϳϰ

ϲ͘ϱϵ

ϵ͘ϱϰ

ϳ͘ϭϰ

ϲ͘ϴϯ

ϳ͘ϲϳ

ϭϯ͘Ϭϴ

ϭϯ͘ϱϬ

ϭϰ͘ϳϰ

ϭϮ͘ϯϴ

ϭϲ͘ϯϰ

ϭϰ͘ϲϴ

ϭϴ͘ϯϳ

Ϯϵ͘ϭϴ

ϭϭϯ͘ϬϮ

^ŚŝŬŽŬƵ

ϭϭ͘ϰϰ

Ϯ͘ϴϮ

ϱ͘Ϭϯ

ϭϬ͘ϲϳ

ϱ͘ϱϮ

ϲ͘ϱϮ

ϳ͘ϰϵ

Ϯ͘ϴϴ

ϭϲ͘ϯϵ

ϳ͘ϲϮ

ϭϳ͘ϰϲ

ϭϮ͘Ϯϵ

ϭϭ͘ϮϬ

Ϯϵ͘ϮϮ

ϯϳ͘ϬϬ

ϭϰϱ͘ϲϯ

<LJƵƐŚƵ

ϭϭ͘ϳϬ

ϯ͘ϴϯ

ϱ͘ϭϰ

ϰ͘ϯϵ

ϱ͘ϵϯ

ϲ͘ϲϬ

ϳ͘ϭϯ

ϭϮ͘ϳϯ

ϭϮ͘ϵϰ

ϴ͘ϵϴ

ϮϬ͘ϰϯ

ϭϯ͘ϰϳ

ϭϵ͘ϴϭ

ϯϱ͘ϯϲ

ϯϱ͘ϯϱ

ϭϬϵ͘ϵϰ

<ŝŶŬŝ

ϭϭ͘ϳϰ

ϯ͘ϳϯ

ϲ͘ϱϳ

ϱ͘ϵϵ

ϴ͘Ϯϲ

ϭϬ͘ϬϬ

ϴ͘ϲϭ

ϵ͘ϳϱ

ϴ͘ϲϲ

ϭϯ͘ϰϳ

ϭϯ͘ϱϮ

ϭϵ͘ϳϬ

ϰϮ͘ϴϲ

Ϯϱ͘Ϯϱ

Ϯϯ͘ϳϯ

ϭϱϭ͘ϮϬ

ŚƵŐŽŬƵ

ϭϯ͘ϲϭ

ϯ͘ϳϴ

ϲ͘ϬϮ

ϲ͘ϴϯ

ϲ͘ϲϭ

ϴ͘ϭϳ

ϴ͘Ϯϰ

ϭϮ͘ϰϵ

ϭϲ͘ϴϬ

ϭϮ͘Ϭϲ

ϭϳ͘ϳϰ

ϭϯ͘ϭϯ

ϭϰ͘ϱϳ

ϯϰ͘ϰϵ

ϰϵ͘ϭϯ

ϭϰϯ͘Ϭϲ

,ŽŬŬĂŝĚŽ

ϭϲ͘ϭϲ

ϯ͘ϴϱ

ϱ͘Ϭϲ

ϯ͘ϰϮ

ϲ͘ϱϰ

ϴ͘Ϯϭ

ϲ͘ϯϲ

Ϯ͘ϴϯ

ϯϵ͘ϭϲ

ϭϳ͘ϱϵ

ϭϱ͘Ϯϱ

Ϯϯ͘ϴϰ

ϭϯ͘ϲϬ

ϯϯ͘ϵϰ

ϮϮ͘ϵϴ

ϮϮϳ͘ϭϭ

EŽƚĞƐ͗WƌĞĨĞĐƚƵƌĞƐĂƌĞĂŐŐƌĞŐĂƚĞĚŝŶƚŽĞŝŐŚƚƌĞŐŝŽŶƐ͘&ŽƌŽƚŚĞƌŶŽƚĞƐĂŶĚƐŽƵƌĐĞ͗^ĞĞdĂďůĞϭ͘

Notes: Prefectures are aggregated into eight regions. For other notes and source, see Table 1.

22

Table 5:

PrefectureIndustry Capital Intensity, Alternative Measure of Labor Input,

2005 dĂďůĞϱ͘WƌĞĨĞĐƚƵƌĞͲ/ŶĚƵƐƚƌLJĂƉŝƚĂů/ŶƚĞŶƐŝƚLJ͕DĂŶƵĨĂĐƚƵƌŝŶŐ͕ϮϬϬϱ /ŶĚƵƐƚƌLJĂǀĞƌĂŐĞĐĂƉŝƚĂůůĂďŽƌƌĂƚŝŽ ϯ͘ϳϵ

ůĂ ď Ž ƌ ƌĂ ƚŝ Ž ; ŵ ĂŶ Ƶ ĨĂ Đƚ Ƶ ƌŝ Ŷ Ő ƚŽ ƚĂ ůͿ

 Ŷ Ě Ž ǁ ŵ Ğ Ŷ ƚ͗ Ɖ ƌĞ ĨĞ Đƚ Ƶ ƌĞ Đ ĂƉ ŝƚ Ăů Ͳ

ϭ͘ϱϯ

Ϯ͘ϰϲ

Ϯ͘ϲϱ

D Ğ ƚĂ ůƉ ƌŽ Ě Ƶ Đƚ Ɛ

K ƚŚ Ğ ƌ ŵ ĂŶ Ƶ ĨĂ Đƚ Ƶ ƌŝ Ŷ Ő

ϯ͘ϬϬ

ϯ͘ϲϴ

' Ğ Ŷ Ğ ƌĂ ůŵ ĂĐ Ś ŝŶ Ğ ƌLJ

 Ğ ƌĂ ŵ ŝĐ ͕Ɛ ƚŽ Ŷ Ğ Ă Ŷ Ě Đ ůĂ LJ Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

W ƌĞ Đŝ Ɛŝ Ž Ŷ ŵ ĂĐ Ś ŝŶ Ğ ƌLJ

ϯ͘ϳϯ

d Ğ džƚ ŝůĞ Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

ϯ͘ϴϭ

d ƌĂ Ŷ ƐƉ Ž ƌƚ Ăƚ ŝŽ Ŷ ŵ ĂĐ Ś ŝŶ Ğ ƌLJ

ϯ͘ϵϬ

&Ž Ž Ě Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

ϰ͘Ϯϵ

 ůĞ Đƚ ƌŝ ĐĂ ůŵ ĂĐ Ś ŝŶ Ğ ƌLJ

ϰ͘ϱϳ

W Ƶ ůƉ Ă Ŷ Ě Ɖ ĂƉ Ğ ƌ

ϲ͘ϴϴ

ϲ͘ϵϳ

Ϯϱ͘ϰϭ

W ƌŝ ŵ Ăƌ LJ ŵ Ğ ƚĂ ů

 Ś Ğ ŵ ŝĐ Ăů Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

W Ğ ƚƌ Ž ůĞ Ƶ ŵ Ă Ŷ Ě Đ Ž Ăů Ɖ ƌŽ Ě Ƶ Đƚ Ɛ

dŽŬLJŽ

ϭ͘Ϯϴ

Ϭ͘ϳϯ

ϭ͘ϭϭ

Ϭ͘ϵϵ

Ϭ͘ϵϭ

Ϯ͘ϰϵ

Ϭ͘ϱϴ

Ϯ͘ϭϯ

ϭ͘ϵϱ

ϭ͘ϰϬ

ϭ͘Ϭϰ

ϭ͘Ϯϯ

ϭ͘ϭϱ

ϭ͘ϯϵ

KƐĂŬĂ

Ϯ͘ϰϰ

ϭ͘Ϯϰ

ϭ͘ϴϯ

ϭ͘ϳϵ

ϭ͘ϴϴ

Ϯ͘Ϭϭ

Ϯ͘ϰϱ

Ϯ͘ϳϯ

Ϯ͘ϵϬ

Ϯ͘ϱϮ

ϭ͘ϵϳ

ϰ͘Ϯϵ

ϰ͘Ϭϭ

ϮϮ͘ϮϮ

'ŝĨƵ

ϯ͘Ϭϭ

ϭ͘ϳϭ

Ϯ͘ϵϳ

Ϯ͘ϲϳ

Ϯ͘ϮϮ

ϰ͘ϰϲ

ϯ͘ϱϴ

ϯ͘ϮϬ

Ϯ͘ϳϲ

ϯ͘ϭϭ

ϱ͘ϲϭ

Ϯ͘ϴϱ

ϲ͘ϰϲ

ϭϳ͘ϭϭ

EĂƌĂ

ϯ͘ϭϯ

ϭ͘ϱϵ

Ϯ͘Ϯϱ

ϰ͘ϲϱ

ϭ͘ϲϰ

Ϭ͘ϴϰ

Ϯ͘ϱϰ

Ϯ͘ϰϵ

ϰ͘ϰϰ

ϰ͘ϳϰ

ϭ͘ϱϭ

Ϯ͘ϵϲ

Ϯ͘ϭϲ

ϱ͘ϭϵ

<LJŽƚŽ

ϯ͘ϭϯ

ϭ͘Ϯϳ

Ϯ͘ϭϬ

ϭ͘ϴϵ

Ϯ͘ϳϭ

ϯ͘ϳϯ

Ϯ͘ϰϯ

ϱ͘ϴϭ

ϱ͘Ϯϯ

ϯ͘ϳϱ

ϭ͘ϵϱ

ϯ͘ϭϮ

ϰ͘ϱϲ

ϲ͘ϯϯ

/ƐŚŝŬĂǁĂ

ϯ͘Ϯϴ

ϭ͘ϭϱ

Ϯ͘Ϯϰ

Ϯ͘ϵϯ

ϭ͘ϱϵ

Ϯ͘ϮϮ

ϲ͘ϱϭ

ϭ͘ϱϱ

Ϯ͘ϳϮ

ϰ͘ϳϰ

ϭ͘ϰϳ

ϯ͘Ϯϴ

ϱ͘ϲϮ

ϭϱ͘Ϭϯ

<ŽĐŚŝ

ϯ͘ϰϳ

Ϭ͘ϳϯ

Ϯ͘Ϭϳ

Ϯ͘ϴϯ

ϲ͘ϵϴ

ϯ͘ϲϲ

ϯ͘ϳϱ

Ϯ͘Ϭϰ

Ϯ͘ϰϭ

ϲ͘ϲϲ

ϰ͘ϰϱ

ϯ͘ϲϲ

ϰ͘ϱϯ

Ϯ͘ϵϭ

^ŚŝŵĂŶĞ

ϯ͘ϰϵ

Ϯ͘Ϭϭ

Ϯ͘ϴϭ

ϯ͘ϭϯ

Ϯ͘ϳϬ

ϲ͘Ϭϯ

ϱ͘ϭϬ

ϯ͘ϵϴ

Ϯ͘ϭϱ

ϰ͘ϭϴ

ϯ͘ϯϵ

ϰ͘ϭϵ

Ϯ͘ϳϵ

ϳ͘Ϯϰ

'ƵŵŵĂ

ϯ͘ϱϭ

ϭ͘ϱϮ

Ϯ͘ϵϰ

Ϯ͘ϰϱ

Ϯ͘ϲϮ

Ϯ͘ϯϬ

Ϯ͘Ϭϵ

ϯ͘ϳϵ

ϱ͘ϳϲ

ϯ͘ϵϯ

Ϯ͘ϴϯ

ϯ͘ϴϰ

ϳ͘ϳϴ

ϳ͘ϵϳ

EĂŐĂŶŽ

ϯ͘ϱϱ

ϭ͘ϱϬ

Ϯ͘ϭϴ

Ϯ͘ϳϳ

Ϯ͘ϳϮ

ϲ͘ϯϵ

ϰ͘ϬϬ

Ϯ͘ϲϵ

ϯ͘ϵϴ

ϰ͘ϲϴ

Ϯ͘ϭϭ

Ϯ͘Ϯϴ

ϯ͘ϳϯ

ϴ͘ϰϵ ϯϬ͘ϲϴ

ŝĐŚŝ

ϯ͘ϲϮ

ϭ͘Ϯϱ

Ϯ͘ϳϱ

Ϯ͘ϯϱ

Ϯ͘ϰϵ

ϯ͘ϰϱ

ϱ͘ϰϮ

ϰ͘ϰϰ

ϯ͘ϲϯ

ϯ͘Ϯϯ

Ϯ͘ϴϴ

ϱ͘Ϯϵ

ϲ͘Ϭϱ

^ŚŝnjƵŽŬĂ

ϯ͘ϲϯ

ϭ͘ϰϭ

ϯ͘Ϭϭ

Ϯ͘Ϭϴ

Ϯ͘ϲϴ

ϰ͘Ϯϴ

ϭϮ͘ϰϵ

ϯ͘ϲϲ

ϰ͘Ϯϴ

ϯ͘ϭϬ

ϲ͘Ϯϯ

ϯ͘ϵϲ

ϴ͘ϭϮ

ϳ͘ϯϳ

^ĂŝƚĂŵĂ

ϯ͘ϳϲ

ϭ͘ϴϬ

ϯ͘ϳϴ

Ϯ͘ϳϲ

ϯ͘ϵϬ

ϱ͘Ϯϰ

Ϯ͘ϴϬ

ϰ͘Ϯϴ

ϯ͘ϴϱ

ϯ͘ϵϬ

ϯ͘ϱϰ

ϰ͘ϰϰ

ϲ͘ϭϲ

ϭϮ͘ϰϱ

dŽĐŚŝŐŝ

ϯ͘ϴϱ

Ϯ͘Ϯϯ

ϯ͘ϱϳ

ϯ͘ϴϬ

Ϯ͘ϵϲ

ϰ͘ϯϯ

Ϯ͘ϰϵ

ϯ͘ϱϯ

ϰ͘ϵϭ

ϰ͘ϬϬ

ϱ͘ϯϱ

ϰ͘ϴϴ

ϲ͘ϯϭ

ϯ͘ϵϳ

zĂŵĂŐĂƚĂ

ϯ͘ϴϴ

ϭ͘ϵϯ

Ϯ͘ϱϰ

Ϯ͘ϱϲ

ϰ͘ϭϴ

Ϯ͘ϴϰ

Ϯ͘ϮϮ

ϯ͘ϯϵ

ϯ͘Ϭϰ

ϱ͘ϰϵ

Ϯ͘ϯϲ

ϱ͘ϭϱ

ϴ͘ϯϲ

ϴ͘ϰϭ

EŝŝŐĂƚĂ

ϯ͘ϵϳ

ϭ͘ϳϲ

Ϯ͘ϯϲ

Ϯ͘ϲϯ

ϯ͘ϰϳ

Ϯ͘ϱϰ

ϯ͘ϭϵ

Ϯ͘Ϭϳ

ϯ͘ϱϮ

ϱ͘Ϯϰ

ϲ͘ϲϳ

ϱ͘ϭϯ

ϭϲ͘ϯϭ

ϯϯ͘ϲϰ

<ĂŐĂǁĂ

ϰ͘ϭϬ

ϭ͘ϳϲ

ϯ͘ϭϲ

ϯ͘ϳϯ

Ϯ͘ϱϮ

ϭ͘ϲϬ

ϯ͘Ϯϱ

Ϯ͘ϳϲ

ϯ͘ϱϱ

Ϯ͘ϰϵ

ϰ͘ϵϬ

ϳ͘ϳϮ

ϳ͘ϳϮ

ϯϵ͘ϬϬ

,LJŽŐŽ

ϰ͘ϭϯ

ϭ͘ϮϬ

Ϯ͘ϲϬ

ϯ͘ϴϱ

ϯ͘ϵϯ

ϭ͘ϯϯ

ϯ͘ϮϬ

Ϯ͘ϳϭ

ϰ͘ϲϯ

ϰ͘ϭϭ

ϱ͘Ϭϱ

ϴ͘ϯϱ

ϲ͘ϭϰ

Ϯϯ͘ϱϳ

&ƵŬƵŝ

ϰ͘ϭϵ

ϭ͘ϱϰ

Ϯ͘ϱϭ

ϭ͘ϴϭ

ϰ͘ϰϴ

ϭ͘ϵϱ

ϲ͘ϴϱ

ϰ͘Ϯϲ

Ϯ͘ϭϴ

ϰ͘ϵϳ

ϯ͘ϲϰ

ϳ͘ϴϬ

ϲ͘ϲϰ

ϭϰ͘ϲϮ

<ĂŐŽƐŚŝŵĂ

ϰ͘ϮϬ

ϭ͘ϵϭ

ϭ͘ϳϳ

Ϯ͘ϰϭ

ϰ͘ϰϱ

Ϯ͘ϭϱ

Ϯ͘ϱϮ

ϭ͘ϯϲ

ϯ͘ϲϲ

ϲ͘ϭϴ

ϴ͘ϲϮ

ϰ͘ϳϬ

ϱ͘ϵϵ

ϵ͘ϲϯ

zĂŵĂŶĂƐŚŝ

ϰ͘Ϯϴ

ϭ͘ϱϳ

Ϯ͘Ϭϵ

ϰ͘Ϯϯ

Ϯ͘ϰϯ

ϱ͘ϴϱ

ϰ͘ϯϰ

Ϯ͘ϰϲ

ϱ͘ϲϮ

ϲ͘ϲϲ

ϭ͘ϵϯ

Ϯ͘ϰϮ

ϲ͘ϵϮ

Ϭ͘ϴϳ

dŽƚƚŽƌŝ

ϰ͘Ϯϵ

Ϯ͘ϭϴ

Ϯ͘Ϭϲ

ϭ͘ϴϰ

Ϯ͘ϮϬ

Ϯ͘ϯϴ

Ϯ͘ϲϬ

ϭ͘ϰϬ

ϱ͘ϵϮ

ϰ͘ϲϰ

ϭϮ͘ϱϳ

ϯ͘ϭϯ

ϭ͘ϲϭ

ϭϭ͘ϯϴ

ŬŝƚĂ

ϰ͘ϯϮ

ϭ͘ϴϯ

Ϯ͘Ϯϱ

Ϯ͘Ϯϱ

Ϯ͘ϱϰ

ϵ͘ϰϭ

ϭ͘ϭϰ

ϯ͘ϲϯ

ϯ͘Ϭϳ

ϱ͘ϱϯ

ϭϴ͘Ϯϭ

ϳ͘ϭϳ

ϭϯ͘ϵϭ

Ϯϴ͘ϭϵ

DŝLJĂŐŝ

ϰ͘ϯϰ

Ϯ͘ϵϲ

ϯ͘ϴϳ

Ϯ͘Ϭϵ

ϯ͘ϭϯ

Ϯ͘ϯϬ

ϯ͘ϳϵ

Ϯ͘ϭϮ

ϰ͘Ϭϰ

ϰ͘ϴϰ

ϭϬ͘ϴϴ

ϳ͘ϲϲ

ϯ͘ϱϯ

ϰϱ͘ϴϮ

/ǁĂƚĞ

ϰ͘ϯϳ

ϭ͘ϴϬ

Ϯ͘ϯϬ

Ϯ͘ϲϭ

ϰ͘ϳϮ

ϰ͘ϵϲ

Ϯ͘ϮϮ

ϯ͘ϳϵ

Ϯ͘ϴϱ

ϲ͘Ϯϰ

ϵ͘ϭϬ

ϭϬ͘ϴϵ

ϭϮ͘ϰϬ

ϭϴ͘ϬϮ

&ƵŬƵŽŬĂ

ϰ͘ϰϭ

ϭ͘ϳϮ

Ϯ͘ϰϲ

Ϯ͘ϭϵ

ϯ͘ϰϭ

Ϯ͘ϯϴ

ϭ͘ϱϴ

ϱ͘ϱϰ

ϯ͘ϳϬ

ϰ͘ϰϭ

Ϯ͘Ϭϲ

ϭϭ͘Ϭϴ

ϭϭ͘ϯϬ

ϮϮ͘ϱϲ

&ƵŬƵƐŚŝŵĂ

ϰ͘ϰϯ

Ϯ͘ϬϬ

Ϯ͘ϵϳ

Ϯ͘ϯϯ

ϯ͘Ϭϲ

ϯ͘ϱϲ

ϭ͘ϴϰ

ϯ͘ϱϮ

ϰ͘ϵϰ

ϱ͘ϱϲ

ϱ͘ϱϰ

ϰ͘ϲϯ

ϭϯ͘ϱϲ

ϱϭ͘Ϭϳ

<ƵŵĂŵŽƚŽ

ϰ͘ϱϬ

Ϯ͘ϬϬ

ϯ͘ϭϰ

Ϯ͘ϳϱ

Ϯ͘ϯϯ

ϭ͘ϰϬ

Ϯ͘ϴϬ

Ϯ͘ϯϴ

ϯ͘ϲϬ

ϴ͘ϲϵ

ϵ͘ϳϭ

ϰ͘ϳϰ

ϲ͘ϴϭ

ϭϱ͘ϮϮ

dŽŬƵƐŚŝŵĂ

ϰ͘ϱϭ

ϭ͘Ϯϵ

Ϯ͘ϯϵ

Ϯ͘ϯϵ

ϭ͘Ϯϰ

Ϭ͘ϰϲ

ϯ͘ϴϴ

ϭ͘ϯϮ

ϰ͘ϯϳ

ϱ͘ϵϭ

ϴ͘ϯϴ

ϲ͘ϳϮ

ϴ͘ϲϵ

ϱϯ͘ϲϯ

,ŝƌŽƐŚŝŵĂ

ϰ͘ϱϱ

ϭ͘ϰϵ

Ϯ͘ϲϮ

Ϯ͘ϳϳ

ϭ͘ϳϰ

ϰ͘Ϭϭ

Ϯ͘ϲϯ

ϯ͘ϳϱ

ϯ͘Ϭϲ

ϵ͘ϴϱ

ϲ͘ϰϯ

ϭϬ͘ϭϳ

ϴ͘ϱϱ

ϰ͘ϲϴ

^ĂŐĂ

ϰ͘ϲϬ

ϰ͘ϱϲ

ϰ͘ϬϬ

Ϯ͘ϵϱ

ϭ͘ϯϯ

ϰ͘Ϭϯ

Ϯ͘ϰϲ

ϯ͘ϭϮ

ϱ͘ϯϬ

ϳ͘Ϯϰ

ϱ͘ϴϭ

ϲ͘ϴϵ

ϱ͘ϴϬ

ϭϰ͘ϯϴ

dŽLJĂŵĂ

ϰ͘ϳϭ

Ϯ͘ϰϲ

ϯ͘Ϭϰ

ϯ͘Ϯϵ

ϯ͘Ϭϰ

ϰ͘ϳϮ

ϲ͘ϵϰ

Ϯ͘ϴϭ

ϯ͘ϯϭ

ϲ͘ϰϰ

ϴ͘ϮϮ

ϴ͘ϲϭ

ϴ͘ϯϲ

ϯϭ͘ϴϭ

KŬŝŶĂǁĂ

ϰ͘ϳϯ

Ϯ͘Ϭϭ

Ϯ͘ϭϮ

ϯ͘ϳϭ

Ϯ͘ϴϲ

Ϭ͘Ϭϱ

Ϯ͘Ϭϴ

Ϭ͘ϳϭ

ϲ͘ϭϲ

Ϭ͘ϴϴ

ϯ͘ϳϵ

ϰ͘ϵϲ

Ϯ͘ϱϯ

ϱϱ͘ϴϯ

DŝĞ

ϰ͘ϴϯ

ϭ͘ϯϲ

ϯ͘Ϯϳ

Ϯ͘ϳϭ

ϯ͘ϯϬ

ϭ͘ϲϰ

ϱ͘ϴϮ

ϰ͘ϯϮ

ϯ͘ϰϯ

ϲ͘Ϭϰ

ϯ͘ϭϲ

ϰ͘Ϭϳ

ϭϭ͘ϱϰ

ϯϳ͘Ϯϱ

^ŚŝŐĂ

ϰ͘ϵϱ

Ϯ͘ϱϴ

ϲ͘Ϭϴ

ϯ͘Ϯϵ

ϱ͘ϵϴ

ϰ͘ϳϵ

ϴ͘ϲϴ

ϱ͘ϲϭ

ϰ͘ϰϯ

ϱ͘Ϭϱ

ϯ͘ϱϰ

ϱ͘Ϯϱ

ϱ͘Ϯϳ

ϭϰ͘ϱϴ

DŝLJĂnjĂŬŝ

ϱ͘ϬϮ

ϭ͘ϯϯ

ϯ͘ϯϵ

ϭ͘ϴϮ

Ϯ͘Ϯϱ

ϴ͘ϭϲ

ϭϬ͘ϴϵ

Ϯ͘ϭϮ

ϯ͘ϱϬ

ϱ͘Ϯϵ

ϭϮ͘ϭϳ

ϱ͘ϮϬ

ϭϱ͘ϳϮ

ϮϬ͘ϱϱ

/ďĂƌĂŬŝ

ϱ͘ϭϰ

ϭ͘ϴϲ

ϰ͘Ϯϲ

ϰ͘ϵϬ

ϯ͘Ϭϯ

ϰ͘Ϭϳ

Ϯ͘ϳϲ

ϭ͘ϴϬ

ϱ͘ϲϳ

ϰ͘Ϭϯ

ϰ͘ϰϱ

ϵ͘ϳϳ

ϭϭ͘Ϯϳ

Ϯϴ͘ϲϭ

KŬĂLJĂŵĂ

ϱ͘ϯϮ

ϭ͘ϱϵ

Ϯ͘ϯϬ

Ϯ͘ϵϱ

Ϯ͘ϳϵ

ϲ͘ϰϵ

ϰ͘ϱϰ

ϯ͘ϰϰ

ϰ͘ϵϯ

ϯ͘ϴϴ

Ϯ͘ϳϬ

ϭϯ͘ϭϰ

ϭϮ͘ϰϵ

ϰϬ͘Ϯϯ

EĂŐĂƐĂŬŝ

ϱ͘ϯϯ

Ϭ͘ϵϭ

ϭ͘ϯϬ

ϱ͘ϵϰ

ϭ͘ϱϭ

Ϭ͘ϳϵ

Ϯ͘ϭϭ

ϭ͘ϱϯ

Ϯ͘ϲϬ

Ϯϯ͘ϱϰ

Ϭ͘ϵϮ

ϭ͘ϴϯ

ϲ͘ϲϬ

ϱϲ͘ϯϯ

ŚŝďĂ

ϱ͘ϯϳ

ϭ͘ϲϲ

Ϯ͘ϱϴ

Ϯ͘ϲϱ

ϯ͘ϱϲ

ϯ͘ϯϲ

ϭ͘ϭϯ

ϭ͘ϰϮ

ϯ͘ϳϴ

ϱ͘ϳϭ

Ϯ͘ϴϯ

ϵ͘ϲϴ

ϭϭ͘ϳϴ

Ϯϳ͘ϱϰ

<ĂŶĂŐĂǁĂ

ϱ͘ϰϰ

Ϯ͘ϰϬ

ϯ͘ϴϭ

ϰ͘ϭϲ

ϱ͘Ϯϲ

ϰ͘Ϭϵ

Ϯ͘ϱϬ

ϰ͘ϳϳ

ϱ͘ϳϬ

ϰ͘ϴϱ

ϯ͘ϱϱ

ϭϬ͘ϱϰ

ϭϬ͘ϳϯ

ϯϰ͘ϲϳ

ŚŝŵĞ

ϱ͘ϳϰ

Ϭ͘ϵϲ

ϭ͘ϳϵ

Ϯ͘ϲϱ

ϭ͘ϵϬ

Ϭ͘ϱϬ

ϭϭ͘ϮϬ

Ϯ͘ϴϱ

Ϯ͘ϵϯ

ϵ͘ϬϮ

ϲ͘ϵϴ

ϭϮ͘ϵϭ

ϭϰ͘ϲϳ

ϯϱ͘ϵϱ

ŽŵŽƌŝ

ϲ͘ϭϴ

ϭ͘ϲϯ

ϭ͘ϲϵ

ϰ͘Ϭϴ

ϰ͘ϯϰ

Ϯ͘ϬϬ

ϰ͘ϭϵ

ϭ͘ϯϭ

ϯ͘ϵϵ

ϯ͘ϳϯ

ϭϱ͘ϵϬ

Ϯϲ͘ϲϯ

ϭϮ͘ϵϰ

Ϯϭ͘ϯϲ

,ŽŬŬĂŝĚŽ

ϲ͘ϯϴ

Ϯ͘ϯϲ

ϯ͘Ϯϭ

ϯ͘Ϯϯ

ϰ͘ϲϭ

ϭ͘ϰϭ

ϯ͘ϬϮ

ϴ͘Ϭϰ

ϱ͘ϭϴ

ϳ͘ϱϲ

Ϯϭ͘ϮϬ

ϭϰ͘ϮϮ

ϵ͘Ϯϭ

ϳϯ͘ϵϳ

tĂŬĂLJĂŵĂ

ϲ͘ϱϬ

ϭ͘ϭϵ

ϭ͘ϳϴ

Ϯ͘ϲϴ

Ϯ͘ϬϮ

ϭϬ͘ϰϬ

ϰ͘ϭϱ

ϭ͘ϴϬ

Ϯ͘ϳϯ

Ϯ͘ϰϭ

ϯ͘ϰϭ

ϭϱ͘ϲϳ

ϭϬ͘ϳϯ

ϱϯ͘ϬϮ

zĂŵĂŐƵĐŚŝ

ϲ͘ϵϱ

ϭ͘ϱϭ

ϯ͘Ϯϰ

ϯ͘ϬϮ

ϱ͘ϯϮ

Ϭ͘ϳϭ

ϴ͘ϱϲ

ϯ͘ϴϬ

Ϯ͘ϵϳ

ϲ͘ϰϱ

ϵ͘ϳϬ

ϴ͘Ϯϴ

ϭϱ͘ϵϭ

Ϯϵ͘ϱϮ

KŝƚĂ

ϳ͘Ϯϰ

ϭ͘Ϯϰ

ϭ͘ϵϮ

Ϯ͘ϴϯ

ϱ͘ϬϮ

ϲ͘ϳϬ

Ϯ͘ϱϬ

ϭ͘ϲϳ

ϰ͘ϰϴ

ϭϬ͘ϰϴ

ϵ͘ϭϵ

ϭϵ͘ϰϮ

ϭϴ͘ϲϮ

ϯϲ͘ϰϵ

EŽƚĞƐ͗dŚĞĐŽůŽƌŽĨĞĂĐŚĐĞůůŝŶĚŝĐĂƚĞƐƚŚĞĐĂƉŝƚĂůŝŶƚĞŶƐŝƚLJŽĨĂŐŝǀĞŶŝŶĚƵƐƚƌLJŝŶĂŐŝǀĞŶƉƌĞĨĞĐƚƵƌĞ͘tŚŝƚĞ͕ůŝŐŚƚŐƌĂLJ͕ĚĂƌŬŐƌĂLJ͕ĂŶĚďůĂĐŬ ŵĞĂŶƚŚĂƚƚŚĞĐĂƉŝƚĂůŝŶƚĞŶƐŝƚLJŝƐŝŶƚŚĞĨŝƌƐƚ͕ƐĞĐŽŶĚ͕ƚŚŝƌĚ͕ĂŶĚĨŽƵƌƚŚƋƵĂƌƚŝůĞƐ͕ƌĞƐƉĞĐƚŝǀĞůLJ͘dŚĞŝŶĚƵƐƚƌŝĞƐĂŶĚƚŚĞƉƌĞĨĞĐƚƵƌĞƐĂƌĞƐŽƌƚĞĚ

Notes and source: See Table 1. ŝŶŽƌĚĞƌŽĨĐĂƉŝƚĂůŝŶƚĞŶƐŝƚLJĂŶĚƌĞůĂƚŝǀĞĐĂƉŝƚĂůĂďƵŶĚĂŶĐĞ͕ƌĞƐƉĞĐƚŝǀĞůLJ͘ ^ŽƵƌĐĞ͗Z/d/;ϮϬϭϰͿZͲ:/WϮϬϭϰ͘

23

Table 6: Rank Correlation of Industry Capital Intensities, 19732009: Prefecture-Level Results

Table 5. Rank Correlation of Industry Capital Intensities, 1973-2009: Prefecture Level Results Manufacturing only Mean S.D.

N

Year 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

0.679 0.682 0.706 0.716 0.728 0.734 0.735 0.727 0.731 0.735 0.750 0.727 0.734 0.726 0.720 0.711 0.697 0.709 0.719 0.716 0.705 0.705 0.712 0.712 0.700 0.697 0.693 0.678 0.657 0.654 0.649 0.659 0.645 0.625 0.622 0.603 0.618

0.150 0.146 0.139 0.138 0.136 0.140 0.146 0.142 0.137 0.135 0.127 0.144 0.145 0.141 0.145 0.154 0.168 0.156 0.144 0.147 0.152 0.157 0.155 0.151 0.150 0.151 0.151 0.163 0.181 0.180 0.186 0.181 0.186 0.197 0.189 0.199 0.184

1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081

Kendall's W 0.686 0.689 0.712 0.722 0.734 0.740 0.740 0.733 0.736 0.740 0.755 0.733 0.739 0.732 0.726 0.717 0.704 0.715 0.724 0.722 0.711 0.711 0.718 0.718 0.706 0.703 0.700 0.685 0.664 0.661 0.656 0.666 0.652 0.633 0.630 0.611 0.626

Includes agriculture and mining Mean S.D. N 0.691 0.681 0.697 0.708 0.720 0.731 0.736 0.733 0.740 0.736 0.753 0.734 0.739 0.727 0.725 0.723 0.707 0.709 0.714 0.700 0.685 0.694 0.695 0.692 0.685 0.681 0.677 0.665 0.649 0.649 0.651 0.664 0.649 0.638 0.635 0.612 0.615

0.140 0.143 0.138 0.138 0.131 0.132 0.134 0.127 0.124 0.123 0.113 0.128 0.131 0.130 0.129 0.136 0.149 0.147 0.138 0.140 0.144 0.146 0.147 0.145 0.144 0.143 0.143 0.152 0.163 0.162 0.170 0.165 0.171 0.177 0.174 0.189 0.180

1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081 1,081

Kendall's W 0.697 0.688 0.704 0.715 0.726 0.736 0.741 0.738 0.745 0.741 0.759 0.740 0.745 0.733 0.731 0.729 0.713 0.716 0.720 0.707 0.692 0.700 0.701 0.699 0.692 0.688 0.684 0.673 0.657 0.657 0.658 0.671 0.657 0.645 0.643 0.620 0.624

Notes: Rank correlation of capital intensities is calculated for different prefecture pairs. The number of correlations is 1,081 (= the number of prefecture pairs (46for + 45 + … + 1)). Notes: Rank correlation of capital intensities is calculated dierent prefecture pairs. The number of Source: RISTI (2014) R-JIP2014.

correlations is 1,081 (= the number of prefecture pairs (46 + 45 + ... + 1)).

Source: RIETI (2014) R-JIP Database 2014.

24

Table 7:

Rank Correlation of Industry Capital Intensities, 19732009:

Region-Level

Results

Table 6. Rank Correlation of Industry Capital Intensities, 1973-2009: Region Level Results Manufacturing only Mean S.D.

N

Year 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

0.827 0.823 0.839 0.840 0.853 0.874 0.871 0.867 0.891 0.888 0.887 0.892 0.898 0.890 0.882 0.880 0.884 0.892 0.898 0.905 0.898 0.896 0.890 0.871 0.862 0.860 0.853 0.850 0.847 0.828 0.825 0.827 0.831 0.826 0.827 0.821 0.818

0.083 0.086 0.082 0.084 0.076 0.062 0.064 0.068 0.057 0.067 0.070 0.055 0.059 0.059 0.061 0.056 0.061 0.053 0.055 0.055 0.054 0.059 0.063 0.080 0.081 0.083 0.086 0.086 0.083 0.090 0.102 0.109 0.097 0.099 0.104 0.101 0.094

28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28

Kendall's W 0.848 0.845 0.859 0.860 0.872 0.890 0.887 0.884 0.905 0.902 0.901 0.905 0.911 0.904 0.897 0.895 0.899 0.906 0.910 0.917 0.911 0.909 0.904 0.887 0.880 0.878 0.872 0.869 0.866 0.850 0.847 0.849 0.853 0.848 0.849 0.843 0.840

Includes agriculture and mining Mean S.D. N 0.821 0.818 0.828 0.827 0.850 0.861 0.863 0.861 0.877 0.873 0.881 0.883 0.894 0.890 0.885 0.882 0.886 0.895 0.897 0.899 0.887 0.888 0.886 0.881 0.870 0.855 0.853 0.834 0.832 0.821 0.831 0.832 0.844 0.842 0.817 0.829 0.825

0.084 0.083 0.084 0.083 0.073 0.065 0.067 0.069 0.069 0.075 0.068 0.057 0.057 0.057 0.054 0.056 0.059 0.053 0.054 0.059 0.061 0.061 0.064 0.071 0.074 0.085 0.084 0.085 0.075 0.077 0.082 0.094 0.079 0.084 0.104 0.090 0.086

28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28 28

Kendall's W 0.843 0.841 0.849 0.848 0.869 0.878 0.880 0.878 0.892 0.889 0.896 0.898 0.907 0.904 0.900 0.897 0.900 0.908 0.910 0.912 0.901 0.902 0.900 0.896 0.886 0.873 0.871 0.855 0.853 0.843 0.852 0.853 0.864 0.861 0.840 0.851 0.847

Notes: Rank correlation of capital intensities is calculated for different region pairs. The number of correlations is 28 (= the number of region pairs (7 + 6 + … + 1)).

Notes: Rank correlation of capital intensities is calculated for dierent region pairs. The number of

Source: RIETI (2014) R-JIP2014.

correlations is 28 (= the number of region pairs (7 + 6 + ... + 1)).

Source: RIETI (2014) R-JIP Database 2014.

25

Source: Ministry of Economy, Trade and Industry (2007) Census of Manufacture, 2005.

of capital intensity and relative capital abundance, respectively.

mean that the capital intensity is in the rst, second, third, and fourth quartiles within each prefecture, respectively. The industries and the prefectures are sorted in order

ϴ͘ϵϳϱϭ ϱ͘ϴϱϱϭ ϵ͘ϱϭϰϭ ϲ͘Ϭϴϯϭ ϱ͘ϯϲϯϭ Ϯ͘ϮϯϮϭ ϱ͘ϲϵϭϭ ϴ͘ϳϲϭϭ ϳ͘ϮϬϭϭ ϴ͘ϯϵϬϭ ϳ͘ϭϴϬϭ Ϭ͘ϲϳϵ ϯ͘ϰϲϵ ϴ͘ϴϱϵ ϳ͘ϯϯϵ ϲ͘ϯϭϵ ϴ͘ϴϬϵ ϯ͘ϯϴϴ ϰ͘ϱϱϴ ϯ͘ϵϯϴ ϰ͘ϴϮϴ ϴ͘ϮϮϴ Ϯ͘ϵϭϴ ϭ͘ϱϬϴ Ϭ͘ϱϬϴ ϵ͘Ϭϴϳ ϳ͘ϲϳϳ Ϯ͘ϭϳϳ Ϯ͘ϳϲϳ ϯ͘ϲϰϳ Ϯ͘Ϭϰϳ ϲ͘ϳϮϳ ϯ͘ϲϬϳ ϰ͘ϱϬϳ ϰ͘Ϯϵϲ Ϭ͘ϵϳϲ ϳ͘ϴϲϲ Ϭ͘ϵϱϲ ϯ͘ϭϰϲ ϵ͘ϲϯϲ ϭ͘ϱϯϲ ϳ͘ϴϭϲ ϳ͘ϰϭϲ ϵ͘ϴϵϱ Ϯ͘ϯϴϱ ϯ͘ϱϳϱ ϱ͘Ϭϭϱ ƚŶĞŵǁŽĚŶ

ŝƌŽŵŽ ŝŚĐƵŐĂŵĂz ĂŵĂLJĂŬĂt ĂďŝŚ ĂƚŝK ĞŵŝŚ ŝŬĂƌĂď/ ĞŝD ĂŵŝŚƐƵŬŽd ĂŵĂLJĂŬK ĂŵŝŚƐŽƌŝ, ĂǁĂŐĂŶĂ< ĂŐŝŚ^ ŽŐŽLJ, ŽĚŝĂŬŬŽ, ŝŐŝŚĐŽd ŝŬĂƐĂŐĂE ĂǁĂŶŝŬK ĂŵĂLJŽd ŽƚŽŵĂŵƵ< ŝŚĐŝ ĂŵŝŚƐƵŬƵ& ĂŬŽƵŬƵ& ĂŵŵƵ' ŝŚƐĂŶĂŵĂz ĂŬŽƵnjŝŚ^ ŝŬĂnjĂLJŝD ŝŚĐŽ< ĂƚĂŐŝŝE ĂǁĂŐĂ< ĂƌĂE ŝƵŬƵ& ĂŬĂƐK ĂŵĂƚŝĂ^ ŝŐĂLJŝD ĂŐĂ^ ĂǁĂŬŝŚƐ/ ŽƚŽLJ< ŽŶĂŐĂE ŝƌŽƚƚŽd ƵĨŝ' ĞŶĂŵŝŚ^ ĞƚĂǁ/ ĂŵŝŚƐŽŐĂ< ŽLJŬŽd ĂƚĂŐĂŵĂz ĂƚŝŬ ĞƌƵƚĐĞĨĞƌW

Notes: The color of each cell indicates the capital intensity of a given industry in a given prefecture. White means no production. Light gray, gray, dark gray, and black

͘ϱϬϬϮ͕ƐĞƌƵƚĐĂĨƵŶĂDĨŽƐƵƐŶĞͿϳϬϬϮ;LJƌƚƐƵĚŶ/ĚŶĂĞĚĂƌd͕LJŵŽŶŽĐĨŽLJƌƚƐŝŶŝD͗ĞĐƌƵŽ^ ͘LJůĞǀŝƚĐĞƉƐĞƌ͕ĞĐŶĂĚŶƵďĂůĂƚŝƉĂĐĞǀŝƚĂůĞƌĚŶĂLJƚŝƐŶĞƚŶŝůĂƚŝƉĂĐĨŽƌĞĚƌŽŶŝĚĞƚƌŽƐĞƌĂƐĞƌƵƚĐĞĨĞƌƉĞŚƚĚŶĂƐĞŝƌƚƐƵĚŶŝĞŚd͘LJůĞǀŝƚĐĞƉƐĞƌ͕ƐĞůŝƚƌĂƵƋŚƚƌƵŽĨĚŶĂ͕ĚƌŝŚƚ͕ĚŶŽĐĞƐ ͕ƚƐƌŝĨĞŚƚŶŝƐŝLJƚŝƐŶĞƚŶŝůĂƚŝƉĂĐĞŚƚƚĂŚƚŶĂĞŵŬĐĂůďĚŶĂ͕LJĂƌŐŬƌĂĚ͕LJĂƌŐ͕LJĂƌŐƚŚŐŝ>͘ŶŽŝƚĐƵĚŽƌƉŽŶŶĂĞŵĞƚŝŚt͘ĞƌƵƚĐĞĨĞƌƉŶĞǀŝŐĂŶŝLJƌƚƐƵĚŶŝŶĞǀŝŐĂĨŽLJƚŝƐŶĞƚŶŝůĂƚŝƉĂĐĞŚƚƐĞƚĂĐŝĚŶŝůůĞĐŚĐĂĞĨŽƌŽůŽĐĞŚd͗ƐĞƚŽE

Table 8: Prefecture-Industry Capital Intensity, 4-Digit Industry Level, 2005

ͿůĞǀĞůƚŝŐŝĚͲϰ;ƐĞŝƌƚƐƵĚŶ/

ϱϬϬϮ͕ůĞǀĞ>ƚŝŐŝͲϰ͕LJƚŝƐŶĞƚŶ/ůĂƚŝƉĂLJƌƚƐƵĚŶ/ͲĞƌƵƚĐĞĨĞƌW͘ϳĞůďĂd

26

Table 9: Estimation Results: Capital Intensity and Exporter Concentration

/ŶĚĞƉĞŶĚĞŶƚǀĂƌŝĂďůĞƐ ZĂƚŝŽŽĨĞdžƉŽƌƚĞƌƐ ZĂƚŝŽŽĨĞdžƉŽƌƚƐ

ĞƉĞŶĚĞŶƚǀĂƌŝĂďůĞ͗ĐĂƉŝƚĂůŝŶƚĞŶƐŝƚLJ ;ϭͿ ;ϮͿ ;ϯͿ Ϯϯϴ͘ϲΎΎΎ ϭϰϵ͘ϮΎΎ ϭϰϮ͘ϲΎ ΀ϳϯ͘Ϭ΁ ΀ϳϯ͘ϯ΁ ΀ϳϯ͘ϯ΁

DĂƌŬĞƚƐŚĂƌĞ

ϲϴϳ͘ϬΎΎΎ ΀ϲϯ͘ϲ΁

EƵŵďĞƌŽĨƉůĂŶƚƐ ŽŶƐƚĂŶƚ /ŶĚƵƐƚƌLJĨŝdžĞĚĞĨĨĞĐƚ WƌĞĨĞĐƚƵƌĞĨŝdžĞĚĞĨĨĞĐƚ E ZͲƐƋƵĂƌĞĚ

ϯϴϯ͘ϴΎΎΎ ΀ϱϮ͘ϱ΁ zĞƐ zĞƐ ϵ͕ϵϱϰ Ϭ͘ϱϭ

Ϯϱϱ͘ϳΎΎΎ ΀ϱϯ͘Ϯ΁ zĞƐ zĞƐ ϵ͕ϵϱϰ Ϭ͘ϱϮ

ϳϯϱ͘ϵΎΎΎ ΀ϳϭ͘ϵ΁ Ͳϰ͘ϳΎ ΀Ϯ͘ϲ΁ ϯϬϯ͘ϴΎΎΎ ΀ϱϵ͘Ϭ΁ zĞƐ zĞƐ ϵ͕ϵϱϰ Ϭ͘ϱϯ

;ϰͿ

;ϱͿ

;ϲͿ

ϱϲϯ͘ϳΎΎΎ ΀ϭϱϰ͘ϵ΁

ϯϬϯ͘ϴΎΎ ΀ϭϱϮ͘ϯ΁ ϲϲϳ͘ϲΎΎΎ ΀ϲϯ͘ϳ΁

ϰϬϯ͘ϲΎΎΎ ΀ϱϮ͘ϴ΁ zĞƐ zĞƐ ϵ͕ϯϳϴ Ϭ͘ϱϮ

Ϯϳϳ͘ϴΎΎΎ ΀ϱϯ͘ϳ΁ zĞƐ zĞƐ ϵ͕ϯϳϴ Ϭ͘ϱϯ

Ϯϵϵ͘ϰΎΎ ΀ϭϱϮ͘ϭ΁ ϳϭϴ͘ϳΎΎΎ ΀ϳϮ͘ϯ΁ Ͳϰ͘ϴΎ ΀Ϯ͘ϳ΁ ϯϮϴ͘ϮΎΎΎ ΀ϲϬ͘Ϯ΁ zĞƐ zĞƐ ϵ͕ϯϳϴ Ϭ͘ϱϯ

Notes: OLS estimates are reported. ***, **, * indicate statistically signicant at 1%, 5%, and 10% levels, respectively. Figure in brackets indicates standard error (clustered by industry).

27

Table A1: Hypothetical PrefectureIndustry Capital Intensities

Table A2. Hypothetical Prefecture-Industry Capital Intensity Industry average capital labor ratio 164 100 103 106

112

115

118

121

124

127

130

133

136

Endowment: prefecture capital-labor ratio (manufacturing total)

Food products

Textile products

Pulp and paper

Chemical products

Petroleum and coal products

Ceramic, stone and clay products

Primary metal

Metal products

General machinery

Electrical machinery

Transportation machinery

Precision machinery

Other manufacturing

Prefecture Hokkaido Aomori Iwate Miyagi Akita Yamagata Fukushima Ibaraki Tochigi Gumma Saitama Chiba Tokyo Kanagawa Niigata Toyama Ishikawa Fukui Yamanashi Nagano Gifu Shizuoka Aichi Mie Shiga Kyoto Osaka Hyogo Nara Wakayama Tottori Shimane Okayama Hiroshima Yamaguchi Tokushima Kagawa Ehime Kochi Fukuoka Saga Nagasaki Kumamoto Oita Miyazaki Kagoshima Okinawa

109

118 120 122 124 126 128 130 132 134 136 138 140 142 144 146 148 150 152 154 156 158 160 162 164 166 168 170 172 174 176 178 180 182 184 186 188 190 192 194 196 198 200 202 204 206 208 210

100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100

103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103 103

106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106 106

109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109 109

112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112 112

115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115

118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118 118

121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121

124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124 124

127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127 127

130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130 130

133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133 133

136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136 136

Notes: Hypothetical capitallabor ratio is presented in each cell. Light gray, gray, dark gray, and black mean that the capital intensity is in the rst, second, third, and fourth quartiles within each prefecture, respectively.

28

A2: Prefecture Table A1. PrefectureTable and Industry Classificationand

Panel A. Region and prefecture classification Region Prefecture Region Prefecture ID ID 1 Hokkaido 1 Hokkaido 2 Tohoku 2 Aomori 3 Iwate 4 Miyagi 5 Akita 6 Yamagata 7 Fukushima 3 Kanto 8 Ibaraki 9 Tochigi 10 Gumma 11 Saitama 12 Chiba 13 Tokyo 14 Kanagawa 4 Chubu 15 Niigata 16 Toyama 17 Ishikawa 18 Fukui 19 Yamanashi 20 Nagano 21 Gifu 22 Shizuoka 23 Aichi 5 Kinki 24 Mie 25 Shiga 26 Kyoto 27 Osaka 28 Hyogo 29 Nara 30 Wakayama 6 Chugoku 31 Tottori 32 Shimane 33 Okayama 34 Hiroshima 35 Yamaguchi 7 Shikoku 36 Tokushima 37 Kagawa 38 Ehime 39 Kochi 8 Kyushu 40 Fukuoka 41 Saga 42 Nagasaki 43 Kumamoto 44 Oita 45 Miyazaki 46 Kagoshima 47 Okinawa Source: RIETI (2014) R-JIP2014.

Industry Classication

Panel B. Industry classification Industry Industry ID 1 Agriculture, forestry and fisheries 2 Mining 3 Food products 4 Textile products 5 Pulp and paper 6 Chemical products 7 Petroleum and coal products 8 Ceramic, stone and clay products 9 Primary metal 10 Metal products 11 General machinery 12 Electrical machinery 13 Transportation machinery 14 Precision machinery 15 Other manufacturing

Source: RIETI (2014) R-JIP Database 2014.

29

Figure B1: FIRs: A Theoretical Explanation Panel (a)

Panel (b)

Figure B2: Estimation Results: CES Production Function

1985

.2 −.2

0

−.3 −.2 −.1 −2

−1

0

1

2

−1

0

1

3

2005

.2

.2

.4

.4

.6

.6

1995

2

0

Fitted value of log(w/r)

0

.1

.4

1975

−2

−1

0

1

2

−1

log(K/L)

30

0

1

2

Table B1: Estimation Results: CES Production Function

DĂŶƵĨĂĐƚƵƌŝŶŐ &ŽŽĚƉƌŽĚƵĐƚƐ dĞdžƚŝůĞƉƌŽĚƵĐƚƐ WƵůƉĂŶĚƉĂƉĞƌ ŚĞŵŝĐĂůƉƌŽĚƵĐƚƐ WĞƚƌŽůĞƵŵĂŶĚĐŽĂůƉƌŽĚƵĐƚƐ ĞƌĂŵŝĐ͕ƐƚŽŶĞĂŶĚĐůĂLJƉƌŽĚƵĐƚƐ WƌŝŵĂƌLJŵĞƚĂů DĞƚĂůƉƌŽĚƵĐƚƐ 'ĞŶĞƌĂůŵĂĐŚŝŶĞƌLJ ůĞĐƚƌŝĐĂůŵĂĐŚŝŶĞƌLJ dƌĂŶƐƉŽƌƚĂƚŝŽŶŵĂĐŚŝŶĞƌLJ WƌĞĐŝƐŝŽŶŵĂĐŚŝŶĞƌLJ KƚŚĞƌŵĂŶƵĨĂĐƚƵƌŝŶŐ

ŽŶƐƚĂŶƚ Ϭ͘ϮϭϮΎΎΎ ΀Ϭ͘Ϭϭϯ΁ Ϭ͘ϬϬϵ ΀Ϭ͘ϬϰϮ΁ Ϭ͘ϭϯϬΎΎΎ ΀Ϭ͘Ϭϰϲ΁ Ϭ͘ϰϴϮΎΎΎ ΀Ϭ͘Ϭϱϭ΁ Ϭ͘ϰϰϲΎΎΎ ΀Ϭ͘Ϭϴϴ΁ Ϭ͘ϲϭϴΎΎΎ ΀Ϭ͘Ϭϲϯ΁ Ϭ͘ϯϮϱΎΎΎ ΀Ϭ͘Ϭϲϯ΁ Ϭ͘ϰϳϯΎΎΎ ΀Ϭ͘Ϭϲϴ΁ Ϭ͘ϮϳϰΎΎΎ ΀Ϭ͘Ϭϰϴ΁ Ϭ͘ϮϮϵΎΎΎ ΀Ϭ͘ϬϴϬ΁ Ϭ͘ϯϯϲΎΎΎ ΀Ϭ͘Ϭϳϵ΁ Ϭ͘ϯϬϴΎΎΎ ΀Ϭ͘Ϭϯϲ΁ Ϭ͘ϯϴϲΎΎΎ ΀Ϭ͘ϬϮϴ΁ Ϭ͘ϭϯϮΎΎ ΀Ϭ͘Ϭϱϱ΁

ŐĂŵŵĂнϭ Ϭ͘ϭϵϱΎΎΎ ΀Ϭ͘Ϭϭϭ΁ Ϭ͘ϮϳϱΎΎΎ ΀Ϭ͘Ϭϰϵ΁ Ϭ͘Ϭϳϱ ΀Ϭ͘Ϭϱϲ΁ ͲϬ͘ϬϴϭΎ ΀Ϭ͘Ϭϰϱ΁ Ϭ͘Ϭϱϴ ΀Ϭ͘ϬϲϬ΁ ͲϬ͘ϬϬϭ ΀Ϭ͘ϬϯϮ΁ Ϭ͘Ϭϲϲ ΀Ϭ͘ϬϳϬ΁ Ϭ͘ϬϮϳ ΀Ϭ͘Ϭϱϭ΁ Ϭ͘ϭϮϰ ΀Ϭ͘Ϭϴϭ΁ Ϭ͘ϮϮϲΎΎ ΀Ϭ͘Ϭϴϲ΁ Ϭ͘Ϭϴϴ ΀Ϭ͘Ϭϲϳ΁ Ϭ͘ϭϳϴΎΎΎ ΀Ϭ͘ϬϰϬ΁ Ϭ͘ϬϱϱΎ ΀Ϭ͘ϬϯϬ΁ Ϭ͘ϮϲϱΎΎΎ ΀Ϭ͘Ϭϲϵ΁

E ZͲƐƋƵĂƌĞĚ ϲϭϭ Ϭ͘ϯϱ ϰϳ

Ϭ͘Ϯϰ

ϰϳ

Ϭ͘Ϭϲ

ϰϳ

Ϭ͘Ϭϴ

ϰϳ

Ϭ͘Ϭϰ

ϰϳ

Ϭ͘ϬϬ

ϰϳ

Ϭ͘ϬϮ

ϰϳ

Ϭ͘Ϭϭ

ϰϳ

Ϭ͘Ϭϱ

ϰϳ

Ϭ͘ϭϲ

ϰϳ

Ϭ͘Ϭϲ

ϰϳ

Ϭ͘Ϯϴ

ϰϳ

Ϭ͘Ϭϴ

ϰϳ

Ϭ͘Ϯϲ

ůĂƐƚŝĐŝƚLJ ϱ͘ϭϯϵ ΀Ϭ͘Ϯϵϭ΁ ϯ͘ϲϯϰ ΀Ϭ͘ϲϰϳ΁ ϭϯ͘Ϯϱϲ ΀ϵ͘ϳϱϰ΁ ͲϭϮ͘Ϯϳϲ ΀ϲ͘ϳϴϲ΁ ϭϳ͘ϮϮϯ ΀ϭϳ͘ϳϳϯ΁ Ͳϭ͕ϲϭϴ ΀ϴϰ͕ϳϳϳ΁ ϭϱ͘Ϯϭϳ ΀ϭϲ͘Ϯϭϵ΁ ϯϲ͘ϵϳϲ ΀ϲϵ͘ϭϰϯ΁ ϴ͘Ϭϯϰ ΀ϱ͘Ϯϯϳ΁ ϰ͘ϰϯϰ ΀ϭ͘ϲϵϱ΁ ϭϭ͘ϯϮϳ ΀ϴ͘ϱϱϭ΁ ϱ͘ϲϭϰ ΀ϭ͘Ϯϱϳ΁ ϭϴ͘Ϭϭϴ ΀ϵ͘ϲϰϯ΁ ϯ͘ϳϲϳ ΀Ϭ͘ϵϳϳ΁

Notes: ***, **, * indicate statistically signicant at 1%, 5%, and 10% levels, respectively. Figure in brackets indicates robust standard error. Elasticity is an elasticity of substitution between capial and labor. It is computed from

1/(γ + 1)

and its standard error is obtained from the delta method.

31