PRI Discussion Paper Series (No.21A-4)

The Decline in the Labor Share: Evidence from Japanese Manufacturers’ Panel Data

Senior Visiting Scholar, Policy Research Institute, Ministry of Finance/

Senior Lecturer, Aichi-Gakuin University

Koyo Miyoshi

May 2021

Research Department Policy Research Institute, MOF

3-1-1 Kasumigaseki, Chiyoda-ku, Tokyo 100-8940, Japan

TEL 03-3581-4111

The views expressed in this paper are those of the

authors and not those of the Ministry of Finance or

the Policy Research Institute.

The Decline in the Labor Share: Evidence fromJapanese Manufacturers’ Panel Data

Koyo Miyoshi ∗†‡

Abstract

This paper replicates Kehrig and Vincent (2021) using Japanese data andtests whether the overall labor share decline is led by an increase in low-labor-share firms. The results can be summarized as follows. The overall labor sharein Japan declined until around 2006, but then rose during the recession in 2008and then leveled off. The results of this paper are similar to the results shownin Autor et al. (2020) and Kehrig and Vincent (2018) for the United States inthe following points. First, although the labor share of median firms did not risewhile the overall labor share was declining, the rate of decline in the labor shareof the median firm was slower than the overall rate of decline. Second, the value-added share of firms with a low labor share increased while their salary share didnot increase when the overall labor share declined. Third, entry and exit are notimportant to the decline in overall labor share, as in the United States.

The results of this paper differ from Autor et al. (2020) and Kehrig andVincent (2018) in the following points. First, the role of firms with an extremelylow labor share, say under decile, which is a good explanation of the changein labor share in the United States, is limited in Japan. Second, the changein actual labor share is very similar to

∑ωi,initialλit, the product of the initial

∗Address for Correspondence: Koyo Miyoshi, Senior Visiting Scholar, Policy Research Institute,Ministry of Finance/ Aichi-Gakuin University, 12 Araike, Iwasaki-cho Nisshin, Aichi, 470-0195, JapanTEL: +81-561-73-1111 FAX: +81-561-73-9305 E-mail: kmiyo@dpc.agu.ac.jp.

†The author gratefully acknowledges research support, especially regarding the supply of dataprovided by the Japanese Ministry of Economy, Trade and Industry used in this study.

‡The author is grateful to Masahiro Abe, Kiseung Kim, Kozo Kiyota, Naoto Jinji, Ryohei Hayashi,Katushiko Hori, Tsukasa Matsuura, Ryosuke Nakamura, Shinpei Sano, and Yasuyuki Sugiyama fortheir extremely helpful and constructive comments on an earlier draft, and also wishes to thankseminar participants at Chuo University, Tokai University, Fukuoka University, Ministry of Finance, theTokyo Labor Economics Workshop, Nagoya International Economics Study Group, Kansai EconomicsWorkshop, AASLE 2018 Conference at Seoul, Ryukyu Economics Workshop, KEBA Fall InternationalConference 2018, JEA Meeting 2019 at Musashi university, JEW at the University of Hawaii and theResearch Center for Advanced Policy Studies at Kyoto University. This report is based on the personalopinions of the author and does not indicate the official opinions of the institutions to which the authorbelongs.

1

value-added share and labor share at the time, which suggests that firms with alarge value-added share are lowering their labor share in Japan.

Keywords: Labor share

JEL Classification Number: F16 D33

2

1 Introduction

The labor share, once said to be very stable (Keynes (1939) and Solow (1958)), has

been declining globally at least for the last 20 years. (Blanchard (1997), Karabarbounis

and Neiman (2014), and Dao et al. (2017)). In the U.S., the overall labor-share decline

is led by some firms (Autor et al. (2020) and Kehrig and Vincent (2018)). The labor

share of a typical (median) firm rose during 1967–2012, while the overall labor share

in manufacturing declined. Kehrig and Vincent (2018) and Autor et al. (2020) show

that the share of value-added of firms with a low labor share has been increasing,

contributing to the decline of the overall labor share in the United States. However, it

is not yet clear whether this will be true in other countries where there are not as many

monopolies as in the United States.

The aim of this paper is to verify that this tendency can be seen in Japan, which

is significantly different from the United States, in the following two ways. First,

although Autor et al. (2020) show that sample firms with a high mark-up and/or

low fixed overhead labor will be low-labor-share firms, Adachi and Saito (2020) show

that Japanese firms show a smaller increase in markups relative to the U.S. until 2006.

Second, Acemoglu and Restrepo (2020) show that robot adoption at the industry level

is associated with low labor share and employment in the United States. However,

Adachi et al. (2020) show that in Japan robots and labor are complementary until

2017.

Therefore, we replicate Autor et al. (2020) and Kehrig and Vincent (2021) using

3

the Basic Survey of Japanese Business Structure and Activities (BSJBSA). First, we

compare the overall labor share with the labor share of the median firm in Japanese

manufacturing and see whether the overall labor share is declining and whether the

labor share of the median firm is increasing as in the United States. Second, to see

how the entry and exit of firms affect changes in labor share, we compare overall labor

share, calculating the labor share using a balanced panel. Third, to investigate how the

proportion of low-labor-share firms contributed to the overall decline in labor share,

we decompose the change in overall labor share to the following four, (1) within-firm

components, (2) a between-firms (reallocation) component, (3) an exiting component,

and (4) an entrant component. Fourth, to see the role of low-labor-share firms on

the overall decline in labor share, we compare actual labor share with the labor share

calculated on the sample of firms excluding the lowest decile. Fifth, to investigate

whether this change was caused by a decrease in payroll or an increase in value-added,

we calculate payroll and value-added share of each labor-share decile by year. Sixth,

to investigate whether the firms with a low labor share initially increased the value-

added share or whether the firms with a large value-added share initially decreased

the labor share, we calculate the following two hypothetical labor shares,∑

ωi,initialλit

and∑

ωitλi,initial, and compare them with actual aggregate labor share (where ω is the

value-added share and λ is the labor share).

The results can be summarized as follows. The overall labor share in Japan also

declined until around 2006, but then rose during the recession in 2008 and then leveled

4

off. The results of this paper are similar to the results for the United States shown

in Autor et al. (2020) and Kehrig and Vincent (2018) in the following points. First,

although the labor share of median firms did not rise during the period when the overall

labor share was declining, the rate of decline in the labor share of the median firm was

slower than the overall rate of decline. Second, the value-added share of firms with

a low labor share increased while the salary share of those did not increase when the

overall labor share declined. Third, entry and exit are not important for the decline in

overall labor share, as in the United States.

The results of this paper differ from Autor et al. (2020) and Kehrig and Vincent

(2018) in the following points. First, the role of firms with an extremely low labor

share, say under decile, which is a good explanation of the change in labor share in the

United States, is limited in Japan. Second, the change in actual labor share is very

similar to∑

ωi,initialλit, the product of initial value-added share and labor share at the

time, which suggests that firms with a large value-added share are lowering their labor

share in Japan.

2 Empirical Analysis

This paper uses the Basic Survey of Japanese Business Structure and Activities 1995–

2016. This survey, which is compiled by Japan’s Ministry of Economy, Trade and

Industry, is conducted every year. The Basic Survey of Japanese Business Structure

and Activities, Japanese METI defined value-added as follows. Value-added = Operat-

5

ing profit + Total payroll + Depreciation expenses + Welfare expenses + Real estate

and movables property rental + Taxes and public imposition. It should be noted that

this value-added is gross value-added and does not include dividends from overseas.

Although this survey contains firm-level information on imports and imports from af-

filiates, several points should be noted. First, this survey does not contain information

on the self-employed. So this paper does not analyze the labor share of gross domestic

product but merely the payroll share as in many studies, such as Kehrig and Vincent

(2021). Second, although the role of small-to-medium enterprises in Japan is important,

manufacturing firms with less than 20 employees produce 10% of total value-added and

hire 25% of total employees, firms with less than 50 employees or capital less than

30 million yen are not included in the survey. However, these statistics cover a wide

range of firms compared with other Japanese statistics such as the Financial Statements

Statistics of Corporations (Ministry of Finance). Third, salary information for on-call

workers, who have increased recently in Japan, is not available. Fourth, the definition

of payroll was changed in 2007. Until 2006, “payroll” included severance pay while

the employer’s share of social security payment was excluded. After 2007, severance

pay is excluded from payroll and the employer’s share of social security payment and

severance pay is included as “welfare payment.” Therefore, the labor share, which is

defined aspayroll

value-added, before 2006 cannot be compared with that after 2006. So, we

divide our sample period into pre-2006 and post-2006.

First, we compare the overall labor share with the labor share of the median firm

6

in Japanese manufacturing as in Kehrig and Vincent (2021). The results are shown

in Figure 1. The left panel shows the results from the 1996–2006 sample and the

right panel shows the results from the 2007–2016 sample. The green dashed line shows

the labor share of the firm at the median. The changes in overall labor share are

depicted as the blue dashed line. It should be noted that the trend in the labor share

of the gross domestic product may be different from that of manufacturing because

Japanese manufacturers hire fewer females and more standard workers compared with

other industries. Therefore, we include the labor share calculated from the National

Accounts. The pink dashed line shows the labor share calculated using System of

National Account. The red line shows the labor share of the firm at the 10% decile.

Figure 1 can be summarized as follows. First, the overall labor share is always below the

median and shows a similar change in lower, say, 10%, decile of labor share as in Kehrig

and Vincent (2018). Second, in the left panel, which shows the result for 1996–2006,

the overall labor share decreases faster than the median firm’s labor share. Unlike the

result for the United States shown in Kehrig and Vincent (2018), the labor share of

the median firm is not increasing. Third, in the right panel, which shows the result for

2007–2016, the overall labor share increases faster than the median firm’s labor share

from 2007 to 2009. In Figure 1, we can see there are differences in changes in the labor

share between the overall and the median firm in Japan.

To investigate the role of entry and exit of firms on this difference in changes in the

labor share between overall and the median firm, we compare the labor share overall

7

with the labor share calculated using our balanced panel. The results are shown in

Figure 2. In Figure 2, the blue dotted line shows the change in labor share calculated

using our balanced panel and the red line shows the change in the labor share calcu-

lated using the whole sample. it can be seen that these two changes are very similar,

suggesting that the role of entry and exit is very limited, similar to the result for the

United States shown in Kehrig and Vincent (2018).

Figure 1 suggests the difference in changes in the labor share between overall and

the median firm may be caused either by low-labor-share firms increasing their value-

added share or large value-added-share firms decreasing their labor share. To examine

the role of value-added reallocation between firms on the change in the overall labor

share, this paper follows Autor et al. (2020) and decomposes the changes in the overall

labor share.

The decomposition method is as follows. Overall labor share can be written as∑λit =

∑ωitλit, where λit is the labor share of individual firm i at time t, defined

as λit =salaryit

value-additand ωit is the value-add share of firm i at time t, defined as

ωit =value-addit∑value-addit

. Then, the change in labor share can be decomposed as follows.

∆λ = ∆λs +∆(∑

(ωi − ωs)(λi − λs))s

+ωX,1( ¯λs,1 − ¯λX,1) + ωE,2( ¯λE,2 − ¯λs,2), (1)

where subscript s means firms extant both at t = 1 and t = 2 in an industry, E means

Entrants and X means eXiters. We call (1) ∆λs the within part, (2) ∆(∑

(ωi− ωs)(λi−

λs))s the between part, (3) ωE,2( ¯λE,2 − ¯λs,2) the entrant part, and (4) ωX,1( ¯λs,1 − ¯λX,1)

8

the exiter part. Autor et al. (2020) adopt this decomposition using U.S. data and

demonstrate that between-components dominate in almost all industries. The decom-

position results imply that models that assume a representative firm, as Karabarbounis

and Neiman (2014), may not be suitable. Bockerman and Maliranta (2012) adopt a

similar decomposition method using Finnish manufacturing plant-level data and also

demonstrate that between-components dominate. This paper adopts this decomposi-

tion and compares the results with Bockerman and Maliranta (2012) and Autor et al.

(2020).

The decomposition results are shown in Figure 3. During 1996–2006, when the

Japanese overall labor share was declining, the second part, called the “between” part,

is largest as in Autor et al. (2020) and Bockerman and Maliranta (2012). This means

not all firms uniformly decrease their labor share during this period when the labor

share was decreasing. By contrast, during the period 2007–2016, when the overall

labor share was increasing, the “within part” dominates. The exit of high-labor-share

firms and the entry of low-labor-share firms may cancel each other out over two periods.

To see if this change in overall labor share is a change in payroll or a change in

value-added, we calculate the payroll and value-added shares of each labor-share decile

by year as in Kehrig and Vincent (2018). The results are shown in Figures 4–7. Figures

4 and 5 show the cumulative value-added share for each labor-share decile. Figures 6

and 7 show the cumulative payroll share for each labor-share decile. Figures 4 and 6

show the difference between 1996 (left panel) and 2006 (right panel) and Figures 5 and

9

7 show the difference between 2007 (left panel) and 2016 (right panel). Although the

payroll share of low-labor-share firms, shown in Figures 6 and 7, seems not to change

during this period, the value-added share of low-labor-share firms, shown in Figure 4,

increased between 1996 and 2006, as in the United States Kehrig and Vincent (2018).

That suggests that the decline of labor share is caused by value-added, especially in low-

labor-share firms, although labor input and/or price of labor seem relatively constant.

To see how large is the role of low-labor-share firms overall to the decline in labor

share, we compare actual labor share with the labor share calculated from the sample

consisting of firms above the lowest decile. The results are shown in Figure 8. The blue

dotted line shows the change in labor share calculated using the sample excluding the

lowest decile, the red line shows the change in labor share calculated using the whole

sample. Unlike the result for the United States shown in Kehrig and Vincent (2018),

the two moves are very similar. That suggests that the role of low-labor-share firms is

not as strong as for the United States.

To investigate whether the company with a low labor share initially increased its

value-added share, or the company with a large share of value-added decreased its

labor share, we calculate the following two hypothetical labor shares,∑

ωi,initialλit and∑ωitλi,initial with the actual overall labor share

∑ωitλit (where ω is the value-added

share and λ is the labor share) as in Kehrig and Vincent (2018). The first is the product

of the initial value-added share and the labor share at each point in time and is called

the “Big player” hypothetical change as in Kehrig and Vincent (2018) to take account

10

of the possibility that large firms can lower their labor share more than their smaller

peers. The second is the product of the initial labor share and the value-added share

at each point in time and is called the “Superstar” hypothetical change to take account

of the possibility that the low-labor-share firms at the initial point increase their value-

added share. The results are shown in Figure 9. Both the left panel and right panel

show that the “Big player” hypothetical change is quite similar to the actual change.

This is in stark contrast to the results for the United States in Kehrig and Vincent

(2018). This suggests that the decline in overall labor share in Japan is due to initially

large value-added share firms lowering their labor share.

3 Results and Conclusion

The results can be summarized as follows. The overall labor share in Japan declined

until around 2006, but then rose during the recession in 2008 and then leveled off.

Not all firms uniformly decreased their labor share during this first period. Japanese

manufacturers are similar in the following points to U.S. manufacturers. First, the

Japanese decline in overall labor share seems to be due to the contribution of some low-

labor-share firms. Although the labor share of median firms does not rise during the

period when the overall labor share was declining, the rate of decline in the labor share

of the median firm was slower than the overall rate of decline. The value-added share of

firms with a low labor share increased while their salary share did not increase when the

overall labor share declined. Second, entry and exit are not important for the decline in

11

overall labor share, as in the United States. The differences causing the changes in the

overall labor share between the United States and Japan are as follows. First, the role

of firms with extremely low labor share, say within the first decile, is limited in Japan.

Second, the change in actual labor share is very similar to∑

ωi,initialλit, the product of

initial value-added share and labor share at the time, suggests that firms with a high

value-added share are lowering their labor share in Japan.

The reason only some firms were contributing to the decline in overall Japanese

labor share until 2006 is a direction for further research. Although Autor et al. (2020)

suggest that firms with high markups and/or low fixed overhead labor costs will be low-

labor-share firms, Adachi and Saito (2020) show there are smaller increases in markups

relative to the U.S. until 2006. Acemoglu and Restrepo (2020) show that robot adoption

at the industry level is associated with low labor share and employment in the United

States. However, Adachi et al. (2020) show that robots and labor are complementary

until 2017 in Japan.

Offshoring is another possible factor. If firms outsource their high labor-share activ-

ities to low-wage countries, the domestic labor share will decrease, even if markups are

unchanged. Although Kiyota and Maruyama (2018) find that offshoring was associated

with increasing demand for highly skilled labor in Japan, Adachi and Saito (2020) show

that over half the decline in the overall decline in labor share in Japan can be explained

by foreign factor augmentation. Therefore, offshoring may be the reason why only some

firms were contributing to the decline in Japanese overall labor share.

12

References

Acemoglu, Daron and Pascual Restrepo (2020) “Robots and Jobs: Evidence from US

Labor Markets,” Journal of Political Economy, Vol. 128, No. 6, pp. 2188–2244.

Adachi, Daisuke and Yukiko Saito (2020) “Multinational Production and Labor

Share,”Technical Report 20-E-012, Research Institute of Economy, Trade and In-

dustry.

, Daiji Kawaguchi, and Yukiko Saito (2020) “Robots and Employment: Ev-

idence from Japan, 1978-2017,”Technical report, Research Institute of Economy,

Trade and Industry.

Autor, David, David Dorn, Lawrence F Katz, Christina Patterson, and John Van

Reenen (2020) “The Fall of the Labor Share and the Rise of Superstar Firms,” Quar-

terly Journal of Economics, Vol. 135, No. 2, pp. 645–709.

Blanchard, Oliver (1997) “The Medium Run,” Brookings Papers on Economic Activity,

Vol. 2, pp. 89–158.

Bockerman, Petri and Mika Maliranta (2012) “Globalization, creative destruction, and

labour share change: evidence on the determinants and mechanisms from longitudinal

plant-level data,” Oxford Economic Papers, Vol. 64, No. 2, pp. 259–280.

Dao, Mai Chi, Mitali Das, Zsoka Koczan, and Weicheng Lian (2017) “Why Is La-

13

bor Receiving a Smaller Share of Global Income? Theory and Empirical Evi-

dence,”Technical report, International Monetary Fund.

Karabarbounis, L. and B. Neiman (2014) “The Global Decline of the Labor Share,”

The Quarterly Journal of Economics, Vol. 129, No. 1, pp. 61–103, feb, DOI: 10.

1093/qje/qjt032.

Kehrig, Matthias and Nicolas Vincent (2018) “Growing Productivity without Growing

Wages: The Micro-Level Anatomy of the Aggregate Labor Share Decline,”Technical

report, Economic Research Initiative at Duke (ERID).

and (2021) “The Micro-level Anatomy of the Aggregate Labor Share

Decline,” Quarterly Journal of Economics.

Keynes, John Maynard (1939) “Relative Movements of Real Wages and Output,” Eco-

nomic Journal, Vol. 49, No. 193, pp. 34–51.

Kiyota, Kozo and Sawako Maruyama (2018) “On the demand for female workers in

Japan: the role of ICT and offshoring,” Asian Economic Papers, Vol. 17, No. 2, pp.

25–46.

Solow, Robert M. (1958) “A skeptical note on the constancy of relative shares,” Amer-

ican Economic Review, Vol. 48, No. 4, pp. 618–631.

14

Figure

1:Overallan

dFirm-Level

Lab

orShares

intheJap

aneseMan

ufacturingSector

0.5

0.6

0.7

0.8

0.9

1998

2001

2004

year

labor_share

lege

nd 10%

90%

aggr

egat

e

med

ian

sna

0.4

0.5

0.6

0.7

0.8

2008

2010

2012

2014

2016

year

labor_share

lege

nd 10%

90%

aggr

egat

e

med

ian

sna

Notes:

(1)Sou

rces:Basic

Survey

ofJap

aneseBusinessStructure

andActivitiesbyMETI(1995–2016).

(2)Theleftpan

elshow

stheresultsfrom

the1996–2006samplean

dtherigh

tpan

elshow

stheresultsfrom

the2007–2016

sample.

15

Figure

2:OverallvsSurvivors:

1996–2006

0.51

0.54

0.57

1998

2001

2004

year

labor_share

lege

nd actu

al

Bal

ance

d pa

nel

0.48

0.50

0.52

0.54

2008

2010

2012

2014

2016

year

labor_share

lege

nd actu

al

Bal

ance

d pa

nel

Notes:

(1)Sou

rces:Basic

Survey

ofJap

aneseBusinessStructure

andActivitiesbyMETI(1996–2016).

(2)Theleftpan

elshow

stheresultsfrom

the1996–2006samplean

dtherigh

tpan

elshow

stheresultsfrom

the2007–2016

sample.

(3)Thebluedottedlineshow

sthelabor

sharecalculatedusingthebalan

cedpan

el.

16

Figure

3:TheDecom

positionResults

−0.

1

0.0

0.1

with

inbe

twee

nex

iter

entr

ant

diffe

renc

ena

mes

value

−0.

1

0.0

0.1

with

inbe

twee

nex

iter

entr

ant

diffe

renc

ena

mes

value

Notes:

(1)Sou

rces:Basic

Survey

ofJap

aneseBusinessStructure

andActivitiesbyMETI(1995–2016).

(2)Theleftpan

elshow

stheresultsfrom

the1996–2006samplean

dtherigh

tpan

elshow

stheresultsfrom

the2007–2016

sample.

17

Figure

4:Value-ad

ded

shareof

each

labor-sharedecile:

1996

vs2006

0.00

3935

8624

1606

245

0.00

8589

4394

0314

529

0.02

0921

6822

5928

9

0.03

3812

7510

5980

04

0.06

0802

8994

3211

07

0.08

8700

3664

9060

87

0.12

5143

0474

6151

4

0.16

8262

2993

4015

7

0.23

1424

4301

7827

2

0.25

8407

2219

5904

1

~01

0%~

020%

~03

0%~

040%

~05

0%~

060%

~07

0%~

080%

~09

0%~

100%

labo

r_sh

are

share_of_value_added

0.00

9355

8204

6638

58

0.00

9928

0746

4044

067

0.04

1970

0081

5963

74

0.04

9074

2185

1455

02

0.10

4851

6106

5125

1

0.10

8652

2024

2738

0.11

9407

7560

6372

6

0.15

4894

2620

7459

9

0.18

3494

2680

1215

2

0.21

8371

7789

8987

8

~01

0%~

020%

~03

0%~

040%

~05

0%~

060%

~07

0%~

080%

~09

0%~

100%

labo

r_sh

are

share_of_value_added

Notes:

(1)Sou

rces:Basic

Survey

ofJap

aneseBusinessStructure

andActivitiesbyMETI(1995–2016).

(2)Theleft

pan

elshow

stheresultsfrom

the1996

sample

andtherigh

tpan

elshow

stheresultsfrom

the2006

sample.

(3)Eachbar

show

sthevalue-ad

ded

shareof

each

labor-sharedecile.

18

Figure

5:Value-ad

ded

shareof

each

labor-sharedecile:

2007

vs2016

0.00

4430

5508

7695

927

0.01

2979

9264

1040

44

0.01

4110

5253

3949

72

0.04

5934

0756

7265

56

0.06

4192

6178

2104

19

0.09

4160

1233

4370

91

0.15

7627

6534

6927

2

0.17

7177

9782

1032

7

0.18

8304

1581

6988

7

0.24

1082

3906

8624

7

~01

0%~

020%

~03

0%~

040%

~05

0%~

060%

~07

0%~

080%

~09

0%~

100%

labo

r_sh

are

share_of_value_added

0.00

3854

9489

8305

89

0.00

4181

9234

3356

447

0.01

6737

7265

3329

49

0.01

9006

0870

9667

18

0.06

7498

1755

7104

44

0.07

5463

7756

1529

62

0.15

5124

6581

6002

2

0.19

8351

9444

9154

1

0.21

4751

7947

9269

9

0.24

5028

9653

2280

7

~01

0%~

020%

~03

0%~

040%

~05

0%~

060%

~07

0%~

080%

~09

0%~

100%

labo

r_sh

are

share_of_value_added

Notes:

(1)Sou

rces:Basic

Survey

ofJap

aneseBusinessStructure

andActivitiesbyMETI(1995–2016).

(2)Theleft

pan

elshow

stheresultsfrom

the2007

sample

andtherigh

tpan

elshow

stheresultsfrom

the2016

sample.

(3)Eachbar

show

sthevalue-ad

ded

shareof

each

labor-sharedecile.

19

Figure

6:Payrollshareof

each

labor-sharedecile:

1996

vs2006

0.00

0560

3770

6585

8976

0.00

2384

5966

3689

925

0.01

4855

5912

2535

03

0.03

4334

6824

7794

69

0.05

4367

5637

6100

56

0.08

9140

5236

2140

83

0.13

3126

0812

4689

0.16

2013

5848

6227

4

0.24

7766

4860

0225

0.26

1450

5131

0011

6

~01

0%~

020%

~03

0%~

040%

~05

0%~

060%

~07

0%~

080%

~09

0%~

100%

labo

r_sh

are

share_of_payroll

0.00

1286

5744

0330

028

0.01

5040

4720

7050

69

0.01

7136

4597

1206

79

0.05

2164

8784

4504

2

0.06

9029

3034

7076

82

0.08

3136

3004

0010

4

0.13

6260

3256

0914

5

0.15

8604

9554

1906

1

0.23

2013

8236

5827

7

0.23

5326

9068

1172

7

~01

0%~

020%

~03

0%~

040%

~05

0%~

060%

~07

0%~

080%

~09

0%~

100%

labo

r_sh

are

share_of_payroll

Notes:

(1)Sou

rces:Basic

Survey

ofJap

aneseBusinessStructure

andActivitiesbyMETI(1996–2016).

(2)Theleft

pan

elshow

stheresultsfrom

the2007

sample

andtherigh

tpan

elshow

stheresultsfrom

the2016

sample.

(3)Eachbar

show

sthepayrollshareof

each

labor-sharedecile.

20

Figure

7:Payrollshareof

each

labor-sharedecile:

2007

vs2016

0.00

1610

7962

6004

334

0.00

8686

1044

0242

092

0.01

5530

9225

8559

57

0.02

4775

0371

5992

23

0.04

9964

8120

3343

18

0.09

9324

5252

2266

17

0.13

2761

9343

4497

3

0.17

9032

1666

4198

3

0.21

3368

1166

6754

7

0.27

4945

5846

8142

2

~01

0%~

020%

~03

0%~

040%

~05

0%~

060%

~07

0%~

080%

~09

0%~

100%

labo

r_sh

are

share_of_payroll

0.00

0582

0812

5358

7544

0.00

6022

7905

3968

806

0.00

7599

7536

8175

33

0.02

7597

7949

2039

8

0.03

2608

0759

5211

18

0.10

6505

5531

0076

9

0.10

8688

7276

1192

2

0.17

6227

9039

1764

0.26

4075

0224

0895

7

0.27

0092

2966

1317

3

~01

0%~

020%

~03

0%~

040%

~05

0%~

060%

~07

0%~

080%

~09

0%~

100%

labo

r_sh

are

share_of_payroll

Notes:

(1)Sou

rces:Basic

Survey

ofJap

aneseBusinessStructure

andActivitiesbyMETI(1995–2016).

(2)Theleft

pan

elshow

stheresultsfrom

the2007

sample

andtherigh

tpan

elshow

stheresultsfrom

the2016

sample.

(3)Eachbar

show

sthepayrollshareof

each

labor-sharedecile.

21

Figure

8:Overalllabor

sharevslabor

shareof

firm

swithou

tlowestdecile

0.55

0.60

0.65

1998

2001

2004

year

labor_share

lege

nd actu

al

Non

−LL

0.50

0.55

0.60

2008

2010

2012

2014

2016

year

labor_share

lege

nd actu

al

Non

−LL

Notes:

(1)Sou

rces:Basic

Survey

ofJap

aneseBusinessStructure

andActivitiesbyMETI(1996–2016).

(2)Theleftpan

elshow

stheresultsfrom

the1996–2006samplean

dtherigh

tpan

elshow

stheresultsfrom

the2007–2016

sample.

(3)Thebluedottedlineshow

sthelabor

sharecalculatedusingthesample

withou

tthelowestdecile.

22

Figure

9:“B

igplayer”

vs“S

uperstar”

0.51

0.54

0.57

0.60

1998

2001

2004

year

labor_share

lege

nd Bal

ance

d pa

nel

Big

_pla

yer

Sup

erst

ar

0.48

0.50

0.52

0.54

2008

2010

2012

2014

2016

year

labor_share

lege

nd Bal

ance

d pa

nel

Big

play

er

Sup

erst

ar

Notes:

(1)Sou

rces:Basic

Survey

ofJap

aneseBusinessStructure

andActivitiesbyMETI(1995–2016).

(2)Theleft

pan

elshow

stheresultsfrom

the2007

sample

andtherigh

tpan

elshow

stheresultsfrom

the2016

sample.

(3)Bluedashed

lineshow

sthehypothetical

labor

sharecalculatedas

∑ ω itλ

i,in

itial.

Thereddottedlineshow

sthe

hypothetical

labor

sharecalculatedas

∑ ω i,initialλ

it.(w

hereωisvalue-ad

ded

sharean

dλislabor

share).

23