Graduate Course in Economic Development & History

The Evolution of Income Concentration in Japan, 1886-2005:Evidence from Income Tax Statistics

November 24, 2009

Chiaki Moriguchi, Hitotsubashi University

2

Today's Talk

♦ Top Incomes in Japan from a Comparative HistoricalPerspective:• When did Japan become an egalitarian society?• “Kuznets process” or “historical accident”?• Was Japan’s experience unique?• Empirical analysis of the determinants of top income shares.

♦ Findings from Moriguchi & Saez (2007), Atkinson & Piketty(2007), and Moriguchi & Saez (work in progress) .

3

Income Inequality in Japan: What We Know

♦ Japan widely perceived as "egalitarian” society.OECD income distribution studies:– Sawyer (1976)– Atkinson et al. (1995)

♦ Recent concerns over rising income inequality:– Bubble economy in the 1980s and its collapse.– Ageing workforce, non-standard workers, and youth

unemployment since the 1980s.– Corporate governance reform and performance pay in the 1990s.

Tachibanaki (2001): ”Japan as egalitarian society is a myth." Ohtake (2005): "Increase in inequality due to demographic change,not due to structural change."

4

Income Inequality in OECD Countries in the late 1980s

5

Income Inequality in Japan (all households), 1895-2003

6

Income Inequality in Japan: What We Don't Know

♦ No data between 1940 and 1955.

♦ Pre-1940 data only for selected years.

♦ Cannot compare the levels of pre-1940 series and post-1955 seriesdue to data discontinuity.

♦ Japan became an egalitarian society by 1970 — but how?

♦ No consensus over the recent rise in inequality.

♦ Little knowledge on high income groups.

7

High Income Studies using Income Tax Statistics

♦ Pioneered by Piketty (2003) and Piketty & Saez (2003).

♦ Construct long-run, homogenous, and continuous top income sharesseries using income tax statistics.

♦ Precise data on the level and composition of the high-end of incomedistribution (not available in household surveys).

♦ Document the evolution of income concentration over many decades.

♦ Facilitate international comparison of the experience of other OECDcountries — France, U.K., Sweden, U.S., Canada, Australia etc.

8

Studying Japan

♦ Comprehensive income tax system introduced in 1887.Cf: U.K. in 1908, France in 1910, U.S. in 1913.

♦ Income tax statistics published every year since 1887 to date.

♦ Because Japan's modern economic growth started circa 1886, the dataspan the entire process of industrialization.

♦ Investigate the relationships between economic growth and incomeinequality (see Kuznets (1955); Deininger & Squire (1998); Forbes(2000); Barro (2000); Banerjee & Duflo (2003)).

9

GDP per Capita in Japan and the U.S., 1790-2005

10

Main Findings

♦ Income concentration in Japan extremely high before WWII, declinedsuddenly during WWII, and remained low.

♦ Top capital income collapsed during WWII and never recovered.

♦ Wage income concentration also declined sharply during WWII.

♦ Top wealth fell dramatically during WWII and U.S. Occupation.

♦ Postwar institutional reforms prevented re-concentration of incomeduring the high growth period.

♦ Small but steady increase in top income shares in the last 10 years.

11

Methodology

♦ Income before tax (wages & bonuses; business, farm & self-employed incomes;interest, dividends & rents; but excluding capital gains).

♦ Top income groups (1%, 0.1%, 0.01%) defined relative to total number of adults.

♦ Top 1% income share= incomes accrued to top 1% income earners / total personal income.

♦ Precise data on the level and composition of the high-end of income distributioncompared to sample surveys.

♦ Due to high level of tax exemption, less than 5% of adults filed income taxreturns before 1947.

♦ "Reported income" subject to tax evasion & avoidance.

♦ Due to many tax reforms and revisions, need careful adjustments to createhomogenous series.

12

Major Adjustments

♦ Dividends, bonus, and part of interest not included in the reported income in1899-1920.[ Use corporate income tax statistics to estimate missing components. ]

♦ Introduction of “withholding system (gensen choshu seido)” for wage incomeearners circa 1950: their distributions reported separately.[ Merge self-assessed income tax statistics and wage income tax statistics. ]

♦ Realized capital gains included in the reported income only after 1947.[ Re-estimate top income shares without capital gains using composition data. ]

♦ After WWII, part of dividends and interest can be "taxed separately (sentakubunri kazei)” and missing from the reported income.[ Use savings survey to impute missing income components. ]

13

Top Income Groups in Japan in 2005

Exchange rate in 2005: US$1.00 = 110 yen or AU$1.00 = 85 yen.

14

Top 1% and Next 4% Income Shares in Japan, 1886-2005

15

Decomposition of Top 1% Income Share in Japan, 1886-2005

16

Top 0.1% Income Share with Capital Gains in Japan

17

Top 0.1% Income Shares in France, U.K., and U.S., 1900-2005

18

Composition of Top 1% Income in Japan, 1886-2002

19

Comparing with NSFIE Estimate, 1999

20

Top 0.01% vs. Top 1-0.5% Estate Sizes in Japan, 1905-2005

21

Top Estate Compositions in Japan: 1935, 1950, and 1987

22

Top 5% Wage Income Share in Japan and the U.S.,1929-2002

Note: Wage income is a sum of wages, salaries, bonuses, and pecuniary benefits.

23

Top 1% Wage Income Share in Japan and the U.S.,1929-2005

Note: Inclusion of non-cash compensation would increase Japanese share in 1990 at most by 32% and U.S. share by 10%.

24

Decomposition of Top 1% Wage Income Share, 1951-2005

25

Historical Explanations (1): High Income Concentration in 1886-1938

♦ Top 1% income group receiving 18% of total income in 1900-38.

♦ Top 0.1% income group receiving 8% of total income in 1900-38.

♦ Capital income component roughly 40% of top 1% income before 1938.

Reasons (Yazawa 1992; Okazaki 2000):

1) Concentration of land ownership to “absentee landlords.”

2) Large shareholders receiving high dividends.

3) Large year-end bonuses paid to CEOs.

4) Primogeniture with low estate tax rates.

5) Low marginal income tax rates.

26

Historical Explanations (2): Income De-concentration during WWII (1938-45)

♦ Top 1% income share fell from 20% to 7% in 1938-45.

♦ Top 0.1% income share fell from 9% to 2% in 1938-45.

♦ Collapse of capital income component in top 1% income in 1938-45.

♦ Top 1% "wage income" share fell from 8% to 3% in 1935-44.

Reasons (Nishida 2003; Okazaki 2000; Yazawa & Minami 1993) :

• 1938 General Mobilization Act (regulation on land rents, dividends,

wages, and bonuses in 1939-45)

• Increases in income tax rates in 1938-45.

• Wartime inflation in 1938-45.

• War destruction in 1944-45.

27

Historical Explanations (3):Impact of U.S. Occupational Reforms (1945-52) Reconsidered

♦ Occupational policies: Land reform in 1947-50, Zaibatsu dissolution in 1946-48, Property tax in 1946-51.

with Hyperinflation in 1944-48.

Considered to be major causes of income equalization (Minami 1995).

Then why no impact on top income shares?

Explanations:

• Discontinuity in data? Tax reforms in 1940, 1947, 1950, but unlikely toproduce large enough bias.

• Occupational reforms largely continuation of wartime policies (Dore1985; Gordon 1985; Okazaki 2000; Moriguchi 2000).

• Played important role in (not income but) wealth redistribution.

28

Historical Explanations (4):Why Did Not Top Income Shares Recover after WWII?

♦ Top 1% income share remains stable at 8% in 1950-2002.

♦ Top 0.1% income shares remains at 2% in 1950-2002.

♦ Top 1% "wage income" share remains at 7% in 1970-2002.

Reasons: Changes in institutional infrastructure after WWII:• Primogeniture abolished in 1947,• Progressive estate tax since 1950,

• Progressive individual & corporate income tax since 1950,

• Restrictive land and house lease laws since 1941,

• Tax-exempted savings for the middle class since 1963,

• Change in corporate governance (bank finance, cross shareholding),

• Change in human resource management (enterprise unions, internalpromotion, joint consultation).

29

Top Marginal Tax Rates in Japan, 1886-2005

30

Determinants of Top Income Share, 1886-2005

31

Determinants of Top Wage Income Share, 1951-2005

32

Japan's Experience in Comparative Perspectives

♦ Top income shares series constructed by the similar methods:• France (Piketty 2003)

• U.S. (Piketty and Saez 2003)• Canada (Saez and Veall 2005)• U.K. & Australia (Atkinson and Leigh 2004)• Netherlands (Atkinson and Salverda 2005)

• Switzerland (Dell, Piketty, and Saez 2006)• Sweden (Roine and Waldenstrom 2006)

33

Top 0.1% Income Shares in Anglo-Saxon Countries, 1900-2005

34

Top 0.1% Income Shares in Netherlands, Sweden, andSwitzerland, 1900-2005

35

Is Japan's Experience Unique?

♦ In most OECD countries, income concentration was once extremelyhigh, and then fell dramatically during 1910-1945.

♦ Decline in top income shares during WWII seen in all participatingcountries (including winners), but was most dramatic in Japan.

♦ Top income shares have increased sharply since 1980 in Anglo-Saxoncountries (but remained low elsewhere). Rise in executivecompensation in the U.S. considered to be major driving force.

36

Main Results & Conclusions

♦ Income concentration in Japan was high throughout the 1886-1938 period,then declined dramatically during WWII.

♦ Collapse of top capital income during WWII was the primary reason forincome de-concentration. Wage income concentration also fell sharplyfrom 1935 to 1945.

♦ Postwar institutional reforms made the one-time income de-concentration“irreversible.” Both capital income and wage income became much moreequally distributed after WWII.

♦ Japan achieved two "economic miracles" under very different institutionalsettings. No simple relationships between inequality and growth.

♦ The process of income de-concentration driven mainly by large historicalshocks (wars, depressions) that triggered government interventions.Important policy implications for developing countries.