Declining Labor Shares and the Global Rise of Corporate Savings by Loukas Karabarbounis and Brent Neiman
NBER 23nd East Asian Seminar on Economics discussion by Martin Berka Victoria University and CAMA
Chung-Hua Institution for Economic Research, Academia Sinica and National Taiwan University Taipei, Taiwan, Republic of China June 16, 2012 Martin Berka (Victoria University)
⇓ Labor shares & global ⇑ of corp. savings
June 2012
1 / 21
Contribution Magnum opus offers description and then a unifying theory of global decline in labour income share (1975-2007) and rise in corporate savings, all due to a single shock: decline in investment prices Data: I
I
I
I I
Pool together sectoral datasets for 59 countries: national statistical agencies, UN, OECD, WB, OECD books, UN books, EIU Carefully construct measures of economic activity and savings for 3 sectors: corporate, household and government Focus on corporate: parse out the contributions of labour share, profits, and other payments to capital Numerous robustness checks on data sources (10 pages) Findings: F F F
I
Global corporate labour income shares (in USD) declined by about 5 pp Declining shares correlate with declining PI /PC ⇑ sc /s by around 20pp
dC ) Goal: match decomposition SYC = QYC (1 − sL,C − sK ,C )(1 − Π C
Martin Berka (Victoria University)
⇓ Labor shares & global ⇑ of corp. savings
June 2012
2 / 21
Decomposition No change in the size of corporate sector Large increase in profit shares, but not dividends Thus, global corporate savings share grows
Martin Berka (Victoria University)
⇓ Labor shares & global ⇑ of corp. savings
June 2012
3 / 21
Contribution: Model GE model with neoclassical growth, closed economy, two sectors (H,C), and no uncertainty Household’s utility depends on k (think housing) H/H taxes: τ n , τ d > τ g ⇒ prefer equity buy-backs σ −1
Corporations: single final good Q = zA(αkκσ (k c )
σ −1 σ
σ −1
σ −1
κσ
+ αnκσ n σ ) σ−1 k 0 = (1 − δ)k + x /ξ − Ψ(k, k 0 ). ξ ∼ exogenous relative PI /PC Frictions: I I I I
costly equity issuance (iceberg), high τ d collateral constraint limits debt issuance (1 + r )b ≤ ηk Firm prefers pre-dividend redistribution because of relatively high τ d . But regulatory constraint on pre-dividend redistributions: et ≥ −(e 0 + e 1 k ): ⇑ e 0 ⇒⇑ s c .
Martin Berka (Victoria University)
⇓ Labor shares & global ⇑ of corp. savings
June 2012
4 / 21
Policy function: highly non-linear Desired level of capital but costly accumulation A: issuance costs, B,D: self-finance, C: buy-back (till binds), E: dividends
Martin Berka (Victoria University)
⇓ Labor shares & global ⇑ of corp. savings
June 2012
5 / 21
Modeling results Calib: {αk , σ, e 0 , e 1 } to d /Π = 0.28, d (S C /S )/d (log (ξ )) = −0.46 Steady state matches data averages nicely Scenario: negative price shock (↓ PI /PC by 21%) Look at SS change in 4 static models (CD, CES, with/out imperfections) Baseline (CES, with imperfections) I firm wants more capital as user cost ↓ I w /r ⇑ as k/n ⇑ I ⇑ y and ⇑ Π I Output and wages rise, while ↓ n, wn/Y ⇓ I
In the background, corporate savings rise
Drop in PI /PC causes rise in savings due to desire to accumulate more capital while facing imperfections CES magnifies this mechanism, and allows realistic sL,C movement Adding shocks to β, τ c , τ k sees SS replicate data averages nicely Martin Berka (Victoria University)
⇓ Labor shares & global ⇑ of corp. savings
June 2012
6 / 21
Comments
Wow! Very thorough work on establishing the time-series stylized facts at global level Heavy artillery modeling I I I
rich but realistic micro-foundations highlights the role of non-unitary EOS interaction of shape of production function and imperfections
Martin Berka (Victoria University)
⇓ Labor shares & global ⇑ of corp. savings
June 2012
7 / 21
Comments
Long-run model, with quantities adjusting freely, while ”price” is fixed and exogenous Clarifying intuition in modeling section. I
Basic intuition is simple; what is the value added of complexity? Matching moments quantitatively.
Also look at corporate investment to provide a clearer picture of the last 35 years I Figure 12 shows a highly non-linear (I , K ) relationship Time-series and cross-sections equivalent in the static model: data?
Martin Berka (Victoria University)
⇓ Labor shares & global ⇑ of corp. savings
June 2012
8 / 21
Panel data in Europe: 16 years, 31 countries Unique Eurostat dataset: Prices of investment goods in levels, 1995-2010 I
All of Europe: Belgium, Germany, Greece, Spain, France, Ireland, Italy, Luxembourg, Netherlands, Austria, Portugal, Finland, Sweden, UK, Iceland, Norway, Switzerland, Cyprus, Czech Republic, Estonia, Hungary, Slovenia, Bulgaria, Romania, Turkey
Information on expenditures in every country, year and category Also contains 146 Consumer good categories allowing to construct consumption price levels (details in Berka and Devereux (2011)) Fabricated metal products, except machinery and equipment (IG) Engines and turbines, pumps and compressors (IG) Other general purpose machinery (IG) Agricultural and forestry machinery (IG) Machine tools (IG) Machinery for metallurgy, mining, quarrying and construction (IG) Machinery for food, beverages and tobacco processing (IG) Machinery for textile, apparel and leather production (IG) Other special purpose machinery (IG) Office machinery (IG) Computers and other information processing equipment (IG) Electrical machinery and apparatus (IG) Radio, television and communications equipment and apparatus (IG) Medical, precision and optical instruments, watches and clocks (IG) Other manufactured goods n.e.c. (IG) Motor vehicles, trailers and semi-trailers (IG)
Martin Berka (Victoria University)
Other road transport (IG) Ships, boats, steamers, tugs, floating platforms, rigs (IG) – reference ppp Locomotives and rolling stock (IG) – reference ppp Aircraft, helicopters and other aeronautical equipment (IG) – reference ppp One or two dwelling buildings (IG) Multi-dwelling buildings (IG) Agricultural buildings (IG) Industrial buildings and warehouses (IG) Commercial buildings (IG) Other non-residential buildings (IG) Transport infrastructures (IG) Pipelines, communication and power lines (IG) Other civil engineering works (IG) Products of agriculture, forestry, fisheries and aquaculture (IG) Software (IG) Other products n.e.c. (IG)
⇓ Labor shares & global ⇑ of corp. savings
June 2012
9 / 21
Panel data in Europe: 16 years, 31 countries Clear positive relationship on average – as in K & N Positive relationship between Price of investment and Labour income share in Europe 0.8
0.75
Labour Income Share (OECD)
0.7
0.65
0.6
0.55
0.5
0.45
0.4
0.35 0.5
0.6
Martin Berka (Victoria University)
0.7 0.8 0.9 1 1.1 1.2 Price of investment relative to consumption (base = EU15)
⇓ Labor shares & global ⇑ of corp. savings
1.3
1.4
June 2012
10 / 21
Panel data in Europe: 16 years, 30 countries But the relationship is heterogeneous I I
Some countries have a positive comovement over time Other countries have a negative comovement over time Heterogeneity of the relationship between Price of investment and Labour share in Europe 0.8 BE GER GRE SPA FRA IRE ITA LUX NET AUS POR FIN SWE DEN UK ICE NOR SWI CZE EST HUN POL SVK SVN
0.75
Labour Income Share
0.7
0.65
0.6
0.55
0.5
0.45 0.6
0.7
Martin Berka (Victoria University)
0.8
0.9 1 1.1 1.2 Price of investment relative to consumption (base = EU15)
⇓ Labor shares & global ⇑ of corp. savings
1.3
1.4
June 2012
11 / 21
Time series: positive in average annual growth rates Outliers during GFC: violent price movements fog the view Average growth rates of Labour Income share and of Price of Investment in Europe (30 countries) 2
Mean growth rate of Labour income share
1.5 LSi = −0.06 + 0.08 PI,i + ei, R2=0.02 1
0.5
0
−0.5
−1
−1.5 −2
−1.5
Martin Berka (Victoria University)
−1
−0.5 0 0.5 mean growth rate of [PI/PC relative to EU15]
⇓ Labor shares & global ⇑ of corp. savings
1
1.5
2
June 2012
12 / 21
Time series: positive in median annual growth rates Clear picture: elasticity 0.22, vs. 0.21 in K & N! Median growth rates of Labour Income share and of Price of Investment in Europe (30 countries) 3
2 Median growth rate of Labour income share
LS = −0.3 + 0.22 P + e , R i
I,i
i
=0.02
bar
1
0
−1
−2
−3
−4 −4
−3
Martin Berka (Victoria University)
−2 −1 0 1 median growth rate of [PI/PC relative to EU15]
⇓ Labor shares & global ⇑ of corp. savings
2
3
June 2012
13 / 21
Time series: not unanimously positive within countries Many have time-series corr (PI ,t /PC ,t , LSt ) < 0 Germany
Relative L share
Belgium 1.1 1.08 1.06 1.04 1.02 0.9
1
1.1
Relative L share
1.02 1
1 0.98 1
1.2
0.98 0.9 0.95
1
0.8
1.08
0.9 0.88 0.86 0.84 0.82 1
1
0.85 1
1.2
1.1
1.2
0.9
Portugal
1
1.1
Finland 1.1
1.04
Denmark
0.9
1.02
1.1
1.06
0.95 1 1.051.1
0.95
1.04
1
1.1
1.04
Sweden Relative L share
0.9
1.08
1.06
1.02 0.8
1
Austria
Netherlands
Ireland 1.05
1.06
1.02
Luxembourg
1.04
France 1.08
1.04
Italy
1.06
Spain
Greece 1.06
1.1 1.09 1.08 1.07 1.06
1.05
1
1
0.95
0.95
0.9
0.95 1 1.05 1.1
UK
1.05
0.85 0.9 0.95
Norway
1
Switzerland
1.1
Czech Republic
1.15 1.06
1.1 1.1
1.04
1
1 1
1.1
1.2
1.1
Relative L share
Estonia
1.2
1.3
1.02 0.8
Hungary
1.2
1.2 1.3 1.4
1
0.96 0.94
0.8
1.2
0.7 0.8 0.9
Slovenia 1.2
0.98 1
1
Slovakia
Poland
1
1.2
0.85
0.98
0.8 1
0.9
1
0.85
1.04
0.95
1.02
0.9
1.06
1.05
1.02
1.04
0.95
1.08
0.82
0.95
0.8
0.9
0.78
0.85
0.92 0.6
1.15 1.1
0.76
0.8 0.8
0.85 P
0.9
0.7
I
Martin Berka (Victoria University)
0.8 P I
0.9
0.8 P
I
1
0.7
0.8 0.9 P I
0.75 0.8 0.85 0.9 0.95 P
⇓ Labor shares & global ⇑ of corp. savings
I
June 2012
14 / 21
Cross-section: negative relationship of LS and PI /PC Average levels (over 16 years): elasticity −0.31∗∗∗ Does West (East) Europe have ”high” or ”low” average PI /PC ? Does West (East) Europe have ”high” or ”low” labour income shares?
Martin Berka (Victoria University)
⇓ Labor shares & global ⇑ of corp. savings
June 2012
15 / 21
East Europe: expensive capital East Europe is cheap PIeast < PIwest but east west PCeast << PCwest ⇒ PPCI > PPCI On average 20% higher relative price of investment PI/PC in EU12 countries, relative to EU15
PI/PC in Floating countries, relative to EU15
BE GER 1.2 GRE SPA 1.15 FRA 1.1 IRE 1.05 ITA LUX 1 NET AUS 0.95 POR 0.9 FIN
SWE DEN UK ICE NOR SWI
1.05 1 0.95 0.9 0.85 0.8
0.85 0.75 0.8 1995
2000
2005
2010
1995
PI/PC in East European countries, relative to EU15 CYP CZE EST HUN LAT LIT MAL POL SVK SVN BUL ROM TUR
2000
2005
2010
PI/PC, relative to EU15 1.25
1.4
1.2
1.3
1.15
EZ Float East
1.1 1.2 1.05 1.1 1 1
0.95
0.9 1995
Martin Berka (Victoria University)
0.9 2000
2005
2010
1995
2000
⇓ Labor shares & global ⇑ of corp. savings
2005
2010
June 2012
16 / 21
East Europe: cheap labour LS east is 8pp below LS west (low wages) BE GER 0.75 GRE SPA 0.7 FRA IRE ITA 0.65 LUX NET 0.6 AUS POR FIN 0.55
0.5 1995
CYP CZE EST HUN LAT LIT MAL POL SVK SVN BUL ROM TUR
Labour income share in EU12
Labour income share in Floating countries of Europe 0.75 SWE DEN UK ICE NOR SWI
0.7
0.65
0.6
0.55
2000
2005
2010
0.5 1995
Labour income share in East Europe
2000
2005
2010
Labour Income Share in GDP
1
0.68
0.9
0.66
0.8
EU12 Float East
0.64
0.7 0.62 0.6 0.6
0.5
0.58
0.4
1995
2000
Martin Berka (Victoria University)
2005
2010
1995
2000
⇓ Labor shares & global ⇑ of corp. savings
2005
2010
June 2012
17 / 21
Investment price – labour shares in Europe: Summary Data: significantly negative correlation of PI /PC and Labour income shares in cross-section Consistent with: cheap labour, relatively expensive capital in poorer countries Now let’s think in terms of the model in K & N 1 As a country develops (East ⇒ West), investment price drops c c ξ east > ξ west 2 reallocate from L ⇒ K , ⇑ Y 3 as user cost of capital drops, w /r ⇑ ↑wL↓ 4 ⇓ ls ≡ Y↑ Positive relationship between labour income share and relative investment price in the model
Martin Berka (Victoria University)
⇓ Labor shares & global ⇑ of corp. savings
June 2012
18 / 21
Corporate saving rates in cross-section
Model: corporate saving a result of capital market imperfections Consider a thought experiment: East Europe has larger capital market imperfections than West. Then in the model
Martin Berka (Victoria University)
d (S c /S ) east dξ
>
d (S c /S ) west dξ
⇓ Labor shares & global ⇑ of corp. savings
June 2012
19 / 21
Corporate saving rates in cross-section Don’t have data on trends, but in Figure 3 of K & N we see: I I
High S C /Y : DNK, CHE, NLD, NOR, EST, LUX, JPN, AUT, MEX Some low S C /Y : IND, PRT, ITA, FRA, COL, TWN, UKR, DEU
Do Japan, Denmark or the Netherlands really have much higher capital market frictions then India, Columbia or Ukraine? Is corporate saving a purely supply-side phenomenon? Alternative: drying-out of new I opportunities, liquidity looking for home, some distributed as dividends, some saved. Martin Berka (Victoria University)
⇓ Labor shares & global ⇑ of corp. savings
June 2012
20 / 21
Is elasticity of substitution > 1 important? Hungary and Austria: Both 40% ⇑ TFP between 1995 and 2007 RGDP PC in Austria ⇓ relative to Hungary by almost 30 pp Model (Figure 14): Austria should have rising RGDP gap - either permanently, or temporarily, relative to more imperfect country Imperfections either unimportant, or working in ”wrong” direction CES implication of permanent effects not supported by data
Martin Berka (Victoria University)
⇓ Labor shares & global ⇑ of corp. savings
June 2012
21 / 21