Understanding the Downward Trend in Labor Income Shares Mai Dao, Mitali Das (team lead), Zsoka Koczan and Weicheng Lian, 1 with contributions from Jihad Dagher and support from Ben Hilgenstock and Hao Jiang
The global labor share of income has been on a downward trend Evolution of the Labor Share of Income (Percent) 56 55 54 53 52 51 5 49 48 47 46 45 AEs EMDEs (right scale) 44 33 197 75 8 85 9 95 2 5 1 14 45 44 43 42 41 4 39 38 37 36 35 34
Gini coefficient Gini coefficient, annual deviation from country mean Declining labor shares are associated with rising inequality. Labor Shares and Income Inequality 65 55 Levels Net/disposable Gross y = 34.397***x + 62.53 R 2 =.149 12 1 8 6 Within-Country Changes (Annual) y =.38***x R 2 =.8 45 4 2 35 y = 38.319***x + 5.459 R 2 =.135 2 25 4 6 15..1.2.3.4.5.6.7 Labor share 8 8 6 4 2 2 4 6 8 1 Labor share, annual deviation from country mean (percentage points)
Central Questions How widespread is the decline in the labor share of income? To what extent have trends in labor income shares differed across countries, industries and skill groups? What are the key drivers of the labor share of income and through which mechanisms do they operate? Do the drivers vary between advanced economies (AEs) and emerging market and developing economies (EMDEs), industries, and skill groups? How have technological advancement and global integration affected labor shares? What has been the role of exposures to routinization and participation in global value chains (GVCs) in declining labor shares?
How Widespread is the Decline in the Labor Share of Income?
Heterogeneity across countries and sectors Estimated Trends in Labor Shares by Country and Sector 8 6 Estimated Trends in Labor Shares by Country, 1991 214 (Percentage points per 1 years) 4 2 2 4 6 CHN COL HKG IRL POL AUT ISR NOR NLD CAN CHL JPN KOR ITA HUN CHE BEL CZE FRA PER NZL UKR ARE BRA MYS KAZ IDN VEN ARG DEU ZAF ROU TUR PRT USA FIN IRN AUS SGP MEX MAR ESP EGY SWE DNK THA GBR RUS GRC PHL 4 3 Estimated Trends in Labor Shares by Sector, 1998 214 (Percentage points per 1 years) 2 1 1 2 3 4 Manufacturing Transportation Mining Health services Trade Construction Financial services Utilities Real estate Agriculture Accommodation
and across skill levels, with a most pronounced decline among the medium-skilled. Labor Share Evolutions and Labor Force Composition by Skill Level (Percent) 34 32 3 28 26 24 22 2 Labor Share Evolution in AEs High Middle Low (right scale) 2 18 16 14 12 1 18 4 1995 2 5 9 8 6 3 28 26 24 22 2 18 16 Labor Share Evolution in EMDEs High (right scale) Middle Low 2 18 16 14 12 1 14 4 1995 2 5 9 8 6 Composition of Hours Worked in AEs Composition of Hours Worked in EMDEs High Middle Low 1 9 8 7 6 5 4 3 2 1 1995 2 5 9 1995 2 5 9 1 9 8 7 6 5 4 3 2 1
What are the Key Drivers of the Labor share of Income?
A key parameter that influences labor shares is the elasticity of substitution between capital and labor. Labor share LS = wage employment wage employment+ rental rate capital stock + profits Cobb-Douglass: EoS=1 => stable LS CES: EOS (, ) Tasks with EoS>1 automated; technological advances cost of capital LS Tasks with EoS<1 offshored LS
Changes in labor shares could be driven by: Technology relative price of investment goods and cost of automating routine tasks substitution of labor by capital LS if EoS>1, i.e. larger effect if more automatable/routinizeable Global integration Trade in final goods: specialization in comp. adv. sectors LS in AEs, LS in EMs Participation in global value chains: LS in AEs and EMs Financial integration Higher capital mobility LS in AEs (lower labor bargaining power); Lower cost of capital LS in EMs. Policy and Institutions corporate taxation after-tax return on capital LS; unionization bargaining power of labor LS; if labor and product market (de)-regulation producer rents and markups LS; Measurement Issues (self-employment; depreciation of capital)
and will depend on exposures to routinization. Routine occupations: easily codified or programmed so they can be executed by a machine; easily automated (substitutable by ICT capital). Higher initial exposure of economy/industry to routinizable jobs greater substitution of labor for capital as the price of capital goods declines larger decline in labor share. Construct sectoral and economy-wide exposures to routinization: occupations are scored on how routine they are (Autor and Dorn 213) routinization exposures are employment-weighted scores: SECTORAL: RTI sct = σ N o=1 ω osct RTI o ECONOMY:RTI ct = σ N o=1 ω oct RTI o ω osct : employment share of occupation o, sector s, country c at time t ω oct employment share of occupation o, country c at time t; RTI o : routinization score
The relative price of investment declined more in AEs than in EMDEs Trends in Potential Drivers of Labor Shares: Relative Price of Investment (Percent change relative to 199) 2 2 AEs EMDEs 4 6 8 1 12 14 1992 96 2 4 8 12 14
Labor share trend (percent)..and countries with higher initial routine exposures experienced larger declines in labor shares. Trends in Potential Drivers of Labor Shares: Initial Routine Exposure and Changes in the Labor Share 6 5 3 AEs EMDEs 2 2 3 5 6 1..75.5.25..25.5 Initial routine exposure (index)
GVC participation increased; corporate income taxes and union density rates fell. Trends in Potential Drivers of Labor Shares: Global Value Chain Participation, Corporate Tax Rates, and Union Density Rates 6 55 5 Global Value Chain Participation (Percent of total exports) AEs EMDEs 65 55 Corporate Income Tax and Union Density Rates (Percent) Corporate income tax rate, AEs Corporate income tax rate, EMDEs Union density rate, AEs Union density rate, EMDEs 5 45 4 35 45 3 45 25 4 35 2 15 35 199 94 98 22 6 1 13 25 1 197 8 9 2 1 15
Adjustments for self-employment and depreciation affect labor share evolutions. Evolution of the Adjusted Labor Share of Income (Percent) Unadjusted labor share Self-employment adjusted labor share Self-employment and depreciation adjusted labor share 8 Advanced Economies Emerging Market and Developing Economies 1 75 9 7 8 65 7 6 6 55 5 5 4 45 198 85 9 95 2 5 1 14 3 1994 2 5 1 14
What are the key drivers of the labor share of income?
Empirical Analysis Shift-Share Analysis Econometric Analysis Analysis of Longrun Trends in Aggregate Labor Shares Analysis of Long-run Trends in Sectoral Labor Shares Analysis of Longrun Trends in Labor Shares by Skill Groups 17
Within Within Changes in labor shares are mostly due to changes within sectors. Shift-Share Analysis 15 1 Labor Share Trends, Within versus Total, One-Digit ISIC (Percentage points per 1 years, 199 214) 1 8 6 Labor Share Trends, Within versus Total, Two-Digit ISIC (Percentage points per 1 years, 1992 27) 5 4 2 5 1 China AEs EMDEs 2 4 6 8 15 15 1 5 5 1 15 Total 1 1 8 6 4 2 2 4 6 8 1 Total
Empirical Analysis Shift-Share Analysis Econometric Analysis Analysis of Long- Run Trends in Aggregate Labor Shares Analysis of Long- Run Trends in Sectoral Labor Shares Analysis of Long- Run Trends in Labor Shares by Skill Groups 19
Analysis of long changes in aggregate labor shares We estimate variants of the baseline regression equation: LS c = α + β 2 PI c + [β 3 RTI,c + β 4 RTI,c PI c ]+β 1 Gc + β 5 Polc + ε c Technology where (hat) variables are long-run annualized changes during 1992-214 at the country level. (see similar approach by KN, 213; Elsby, 213; Acemoglu and Restrepo, 217) PI is the relative price of investment goods, RTI the initial exposure to routine-biased technological change. Expect β 2, β 4 >, β 3 < G subsumes variables measuring evolution of global integration: trade in final goods, participation in global value chains, and financial integration (sum of external assets and liabilities net of reserves). Pol summarizes policy/institutional factors: Unionization, Corporate taxation, EPL, PMR.
Technology is the key driver of labor shares in AEs, GVC participation in EMs. Aggregate Results: Contributions to Aggregate Labor Share Changes, 1993 214 (Deviation from regression constant) 6 4 2 2 4 6 Technology Financial integration Unexplained GVC participation Policy/institutions Actual change A decline in the relative price of investment of 15 pct over 199-214 implies.4 ppt fall in LS in countries with low exposure, 1.7 ppt fall in countries with high exposure to routine-biased technological change. (Median was -3 ppt among countries with declining LS) An increase in intermediate trade by 4 pct of GDP (median increase in sample) implies a fall in LS of 1.6 ppt. Effect of corporate tax and financial integration small in comparison (or offsetting). 8 AEs EMs EMs exluding China
Empirical Analysis Shift-Share Analysis Econometric Analysis Analysis of Longrun Trends in Aggregate Labor Shares Analysis of Longrun Trends in Sectoral Labor Shares Analysis of Longrun Trends in Labor Shares by Skill Groups 22
Mining Financial services Utilities Manufacturing Transportation Agriculture Real estate Construction Trade Accommodation Mining Construction Financial services Utilities Transportation Manufacturing Trade Agriculture Accommodation Real estate Agriculture Mining Manufacturing Utilities Construction Trade Accommodation Transportation Financial services Real estate Aggregate evolutions conceal heterogeneity across sectors. Heterogeneity across Sectors and Countries 4 3 2 1 Changes in Labor Shares, AEs Changes in Labor Shares, EMDEs 8 4 4 1 8 6 Share of Countries Where Labor Shares Decreased (Percent) AEs EMDEs 1 2 3 8 12 4 2 4 16
Analysis of long changes in sectoral labor shares Empirical strategy similar to analysis of aggregate labor shares. Link long changes in labor shares to the long changes in relative price of investment goods (PI); changes in measures of global integration (G) including final goods trade, intermediate trade, and cross-border capital mobility, and initial exposures to routinization (RTI), including in its interaction with the relative price of investment goods. Variants of the following cross-sectional regressions (c=1,.,c, s=1,.,s) are estimated. LS cs = β 1 PI cs + [β 2 RTI,cs + β 3 RTI,cs PI cs ] +β 4 Gcs + γ FE c + γ 1 FE s + ε cs
GVC participation is associated with declines in labor shares only in tradable sectors. Sectoral Results, Advanced Economies: Contributions to Sectoral Labor Share Changes 12 1 8 6 4 2 2 4 Technology Country FEs Unexplained GVC participation Sector FEs Actual change Median decline in relative price of investment -> decline in LS observed in country-sector at the 25 th percentile of routinization. Effect of a decline in relative price of investment is double that at the 75 th percentile of routinization. Move from 25 th to 75 th percentile of routinization distribution roughly matches observed difference. 6 8 1 AEs Tradables sectors AEs Nontradables sectors
Empirical Analysis Shift-Share Analysis Econometric Analysis Analysis of Longrun Trends in Aggregate Labor Shares Analysis of Longrun Trends in Sectoral Labor Shares Analysis of Longrun Trends in Labor Shares by Skill Groups 26
Technological progress and GVC participation hollow out the medium-skilled. Contributions to Aggregate Labor Share Change by Skill, 1995 29 12 1 8 6 4 2 2 4 6 Technology Financial integration Actual change Global value chain participation Skill supply and other composition shifts 8 High skill Middle skill Low skill Middle-skill AEs
Conclusions The decline in the labor share of income in advanced and emerging market economies conceals heterogeneity across countries, sectors and skill levels, with the medium-skilled seeing the sharpest declines. Overall, the declines in the labor share of income are mostly due to within-sector declines not reallocation across sectors. In AEs, technological advancement has been the key driver of the evolution of labor shares. In EMs, participation in GVCs played a larger role, though could reflect benign changes. Technological advancement and GVC participation affected labor shares largely through their impact on middle-skilled labor.
Policy Implications Policies should depend on country circumstances: level of development, extent of decline in labor shares, relative importance of underlying drivers, and existing social safety nets. In AEs: help workers cope with disruptions, including through skill upgrading and facilitating transitions, long-term investment in education, longer-term redistributive measures in line with social contract. In EMs: Decline in labor share by itself may not call for policy intervention, but gains from growth should be shared more broadly. Challenges similar to those in AEs could arise as automation progresses promote skill deepening to prepare for further structural transformation.