The Global Rise of Asset Prices and the Decline of the Labor Share

Transcription

1 The Global Rise of Asset Prices and the Decline of the Labor Share Ignacio Gonzalez 1 Pedro Trivín 2 1 American University 2 Universitat de Girona Bank of Spain Madrid, 11 th April 2018 Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

2 Motivation Labor share Decoupling of { } Equity wealth puzzles standard theory Corporate investment Although wealth-income ratios the capital rate of return (Stiglitz, 2014) Potential drivers: Capital taxes, Market power, Short-termism Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

3 Motivation: Fact#1 The global decline of the labor income share since the 1980s has sparked new interest in the functional distribution of income. Figure 1: Labor income share, LIS (pp.) Notes: Own calculations obtained as year fixed effects from a GDP weighted regression including country fixed effects to control for the entry and exit of countries throughout the sample. Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

4 Motivation: Fact#1 Figure 2: Trends in country labor income share, LIS Trends (%) TUR POL COL IRL LUX HUN CHN GRC NOR FIN ITA ZAF IND DEU NZL BEL SWE BRA ISR VEN NLD CAN AUT AUS USA FRA GBR JPN DNK CHE PHL ESP HKG CHL MEX MAR KOR PRT PER LKA THA Notes: Own calculations obtained from ln (X t ) = α 0 + α 1 t + ɛ t, where X represents the labor share, t is a linear trend, and epsilon is a classic disturbance term. The vertical axis show α 1 in %. Dark bars indicate that α 1 is significant at 5% level. The sample covers 41 countries (915 observations). Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

5 Motivation Attempts to explain this trend: Globalization: Esby et al. (2013) Institutional framework: Kristal (2010) Raurich et al. (2012) Structural causes: Bentolila and Saint-Paul (2003) Piketty and Zucman (2014) Karabarbounis and Neiman (2014) Recent related literature: Eggertsson et al. (2018); Brun and Gonzalez (2017); Autor et al. (2017); Barkai (2017); Koh et al. (2016). Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

6 Motivation Attempts to explain this trend: The role of the capital-output ratio (k) Bentolila and Saint-Paul (2003) CES technology lis = f( k y ) Piketty and Zucman (2014) s > g k y lis if σ > 1 Karabarbounis and Neiman (2014) rp k y lis if σ > 1 Koh et al (2016) k IP P k y lis if σ > 1 Secular increases in the capital-output ratio are the main cause of the long-run labor share decline. Graph This requires σ > 1, a value which has seldom been found in the literature (Chirinko and Mallick, 2014). Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

7 Our Contribution Finance and assets prices have developed since 1980s (Philippon, 2012; Greenwod and Scharfstein, 2013). The accumulation view has ignored the role of asset prices (although Piketty and Zucman (2014) document that part of the rise of wealth ratios comes from changes in asset prices). In this paper we explore a new mechanism that connects the rise of stock prices with the decline of the labor share (through lower capital-deepening). Panel Time Series. We find evidence for several mechanisms that operate through the same channel: i) the rise of monopoly mark-ups, ii) the decline of dividend income taxes and iii) the rise of corporate short-termism. Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

8 Outline 1 Facts 2 Theoretical Framework 3 Data 4 Empirical Methodology 5 Results 6 Conclusions Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

9 Fact#2 Steady increase in financial wealth with respect to productive capital. Figure 3: Tobin s Q, Q (Index) Notes: Own calculations obtained as year fixed effects from a GDP weighted regression including country fixed effects to control for the entry and exit of countries throughout the sample. Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

10 Fact#2 Figure 4: Trends in countries Tobin s Q, Tobin s Q Trends (%) PHL PER CHLHUN HKG IND GRC JPN THA KOR LKA VEN ISR TUR ZAF PRT DEU AUT ITA MEX COL BEL SWE CAN GBR IRL ESP LUX AUS FRA NOR CHE USA NZL POL FIN NLD DNK MAR BRA CHN Notes: Own calculations obtained from ln (X t ) = α 0 + α 1 t + ɛ t, where X represents the Tobin s Q, t is a linear trend, and epsilon is a classic disturbance term. The vertical axis show α 1 in %. Dark bars indicate that α 1 is significant at 5% level. The sample covers 41 countries (915 observations). Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

11 Tobin s Q and LIS Figure 5: Labor income share against Tobin s Q Demeaned LIS (in logs) Demeaned Tobin s Q (in logs) Notes: Own calculation based on a sample of 41 countries and 911 observations. Variables are demeaned to control for country fixed-effects. Correlation coefficient= Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

12 Theoretical Framework: Households Representative household accumulates financial assets. Direct utility from the ownership of wealth (ps = a ) like in Carroll (1998), Piketty (2011) and Saez and Stantcheva (2017). U(a) = max c,a u(c) + h(a) + βu(a ) s.t. c + a = w + (1 + r)a, (1) where a = ps and 1 + r = (1 τ)d+p p 1 The demand of assets a(r) is an increasing function. Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

13 Theoretical Framework: Firms CES Technology y = [φk ( σ 1 σ ) + (1 φ)l ( σ 1 )] σ σ 1 σ, (2) Monopolistic competitive firms (elasticity for each variety is ξ) that maximize their market value, accumulate physical capital and distribute dividends to households. Symmetric FOC wrt to k F k (k, l) = ( ξ ) (δ + r), (3) ξ 1 Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

14 Market Tobin s Q (at the steady state): Q(r) = (1 τ) (1 + F (k(r), l) ) ξrk(r) Equity Wealth (at the steady state): p(r) = Q(r)k(r) = (1 τ) (k(r) + F (k(r), l) ) ξr Market clearing a(r ) = p( r ) Q( r τ, ξ ) k( r ) Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

15 Asset Prices and Productive Capital Figure 6: Market for capital 1.09 Capital Market with Dividend Tax (q<1) a(r) Equity Return 1.06 k'(1-q) 1.05 p'(r) k(r)=p(r)=k'(r) Capital, Assets Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

16 Asset Prices and Productive Capital Figure 7: Market for capital 1.09 Capital Market with Dividend Tax (q<1) a(r) Equity Return 1.06 a(r) with standard Euler equation 1.05 p'(r) k(r)=p(r)=k'(r) Capital, Assets Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

17 Asset Prices and Productive Capital Figure 8: Market for capital 1.09 Capital Market with Monopoly Power (q>1) 1.08 k'(r) p'(r) 1.07 a(r) Equity Return k(r)=p(r) Capital, Assets Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

18 Impact on the Labor Share Assuming a CES production function, we have: ( k ) σ 1 σ LIS = 1 φ y From our model, equilibrium relation between Tobin s Q and LIS: ( k LIS (Q) (Q) = 1 φ y (Q) Therefore: because: LIS Q k y k ) σ 1 σ = ( LIS k y where )( k y k LIS K Y > 0 and k(q) Q < 0 > 0 if σ < 1; (4) )( ) k(q) < 0 (5) Q Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

19 Data 41 countries, Sample Tobin s Q: Worldscope Database. Doidge et al. (2013) methodology. Labor income share: Extended Penn World Table 4.0. No adjustment for mixed rents, no distinction of the corporate sector. Correlation between 0.87 and 0.96 with Karabarbounis and Neiman (2014). Relative prices: Extension of Karabarbounis and Neiman (2014) database. Penn World Table 7.1 and BEA. Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

20 Empirical Implementation We assume a general multiplicative form between our variables of interest: ( k ) ( k ) α, k LIS = g = a and y y y = f(q, RP ) = Qψ 1 RP ψ 2 (6) We use these two forms to obtain an estimable equation of the labor share in terms of Q and RP : or in logs: ( k ) LIS = g = g(f(q, RP )) = a(q ψ 1 RP ψ 2 ) α (7) y lis it = β 0 + β 1 q it + β 2 rp it + Ω it (8) Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

21 Empirical Methodology Macroeconomics panel data makes difficult the use of traditional panel data techniques: Small N compared to T. Parameter heterogeneity. Cross-section dependence. Pesaran (2004) CD test Nonstationary data. Pesaran (2007) CIPS test New Panel Time Series Techniques based on Common factor models: y it = β i x it + u it, u it = ϕ i f t + ψ i + ε it, (9) x it = δ i f t + π i + e it, f t = τ + φf t 1 + ω t, (10) where (f t ) represents unobserved time-variant heterogeneity and raises endogeneity problems which make difficult the estimation of β i. Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

22 Empirical Methodology Common Correlated Effect estimators: Observed regressors are augmented with cross-sectional averages of the dependent variable and the individual-specific regressors (Pesaran, 2006). Intuition lis it = β i0 + β i1 q it + β i2 rp it + β i3 lis it + β i4 q it + β i5 rp i t + Ω it Our reference results are obtained from an ECM version using the Chudik and Pesaran (2015) Dynamic Common Correlated Effects Mean Group Estimator: lis it = β i0 + β i1 lis i,t 1 + β i2 q i,t 1 + β i3 rp i,t 1 + β i4 q it + β i5 rp it + β i6 lis t + β i7 lis t 1 + β i8 q t 1 + β i9 rp t 1 + β i10 q t + β i11 rp t p p p + β i12 lis t p + β i13 q t p + β i14 rp t p + Ω it, l=1 l=1 l=1 Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

23 Table 1: Results: Error Correction Model [1] [2] [3] [4] [5] [6] [7] 2FE CCEP MG CMG CMGt CMGt1 CMGt2 lis t (0.026)*** (0.049)*** (0.034)*** (0.053)*** (0.061)*** (0.085)*** (0.125)*** q t (0.013) (0.015) (0.014)** (0.018)** (0.026)** (0.028)*** (0.033)* rp t (0.024) (0.040) (0.070) (0.091)* (0.115) (0.166) (0.186) q (0.014)** (0.015)** (0.009)*** (0.012)*** (0.017)*** (0.019)*** (0.018)*** rp (0.050)*** (0.056)*** (0.065) (0.108) (0.099) (0.107) (0.095) t (0.001) (0.002) (0.003) (0.004) Constant (0.018)*** (0.033)*** (0.050)*** (0.084)*** (0.089)*** (0.124)*** Number of id Observations R-squared RMSE Trends lr-q se-q lr-rp se-rp CD test Abs Corr Int I(0) I(0) I(0) I(0) I(0) I(0) I(0) Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

24 Table 1: Results: Error Correction Model [1] [2] [3] [4] [5] [6] [7] 2FE CCEP MG CMG CMGt CMGt1 CMGt2 lis t (0.026)*** (0.049)*** (0.034)*** (0.053)*** (0.061)*** (0.085)*** (0.125)*** q t (0.013) (0.015) (0.014)** (0.018)** (0.026)** (0.028)*** (0.033)* rp t (0.024) (0.040) (0.070) (0.091)* (0.115) (0.166) (0.186) q (0.014)** (0.015)** (0.009)*** (0.012)*** (0.017)*** (0.019)*** (0.018)*** rp (0.050)*** (0.056)*** (0.065) (0.108) (0.099) (0.107) (0.095) t (0.001) (0.002) (0.003) (0.004) Constant (0.018)*** (0.033)*** (0.050)*** (0.084)*** (0.089)*** (0.124)*** Number of id Observations R-squared RMSE Trends lr-q se-q lr-rp se-rp CD test Abs Corr Int I(0) I(0) I(0) I(0) I(0) I(0) I(0) Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

25 Quantification Figure 9: Labor income share and Tobin s Q, LIS (pp.) Q (Index) (a) Labor Income Share (b) Tobin s Q Tobin s Q has increased (52%) from 1.15 (1980) to 1.68 (2007). LIS has evolved from 57% (1980) to 52% (2007) (-8.9%). Tobin s Q can explain between 41% and 57%. Robustness Weak Exogeneity Test Intangible Assets Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

26 Beyond the Q: What is Behind? Dividend Income Tax Rate Market Power: The Industry Concentration Rate Corporate Governance Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

27 Dividend Income Tax Rate (I) Data (Max ): Dividend Income Tax Rate: OECD Tax Database Capital-Output Ratio: AMECO Tobin s Q: Worldscope Figure 10: Country-specific Trends: Dividend Tax Rate JPN Dividend Income Tax Rate Trends (pp.) ITA NLD PRT SWE NZL MEX USA ESP SVK AUT TUR NOR SVN AUS CHL CAN CHE DNK CZE GBR FRA DEU BEL LUX EST POL IRL HUN ISR LVA KOR GRC ISL FIN Notes: Own calculations obtained from T AX t = α 0 + α 1 t + ɛ t, where T ax is the dividend tax rate, t is a linear trend, and epsilon is a classic disturbance term. The vertical axis show α 1 in %. Dark bars indicate that α 1 is significant at 5% level. Each regression only includes countries which have at least 10 observations for the period Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

28 Dividend Income Tax Rate: U.S. Figure 11: U.S. Dividend Tax Rate U.S. Dividend Tax Rate (%) Notes: Net top statutory rate to be paid at the shareholder level. This rate takes into account all types of reliefs and gross-up provisions at the shareholder level. Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

29 Dividend Income Tax Rate (II) Figure 12: Tobins Q, Capital-Output Ratios and Dividend Income Tax Rates Tobin s Q Dividend Income Tax Rate (%) GRC CZE DNK CHL CAN FRA GBR BEL USA FIN MEX ESP AUS CHE NLD NOR SWE ITA JPN IRL AUT NZL POL HUN DEU PRT ISR TUR KOR ISL Capital Output ratio Dividend Income Tax Rate (%) IRL SVN ESP POL HUN LVA PRT MEX AUT ITA NZL SWE DEU FIN BEL AUS JPN FRA USA CAN NLD GBR DNK CZE CHE ISL GRC NOR (a) Tobins Q (b) Capital-Output ratio Notes: Own calculations obtained from ln (X t ) = α 0 + α 1 T AX t + ɛ t, where X represents the Tobin s Q or the capital-output ratio, TAX stands for the dividend income tax rate, and ɛ is a classic disturbance term. The vertical axis shows the coefficient α 1 in %. Dark bars indicate that α 1 is significant at 5% level. Each graph shows countries for which we have at least 10 observations for the period under analysis (Max ). Luxembourg is excluded from the graph due to be a clear outlier. Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

30 Dividend Income Tax Rate (III) Figure 13: Tobin s Q, Capital-Output Ratios and Dividend Income Tax Rates ISL Tobin s Q Dividend Income Tax Rate (%) IRL PRT HUN POL ESP DEU AUTNZL ITA JPN SWE NLD AUS MEX FINUSA BEL GBR FRA CAN DNK CZE CHE GRC NOR Capital Output ratio Dividend Income Tax Rate (%) Notes: Own calculations obtained from ln (X t ) = α 0 +α 1 T AX t +ɛ t, where X represents the Tobin s Q and the capital-output ratio in the vertical and the horizontal axis respectively. T AX is the dividend income tax rate, and ɛ is a classic disturbance term. Both axis show the coefficient α 1 in %. Both equations are constraint to have the same number of observations (Max ). The scatter plot is obtained after excluding outliers. An outlier is defined as an observation with a weight of 0 after using the rreg command in STATA. Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

31 Market Power: The Industry Concentration Rate (I) Data (U.S. Industry Data, ): Market Power: U.S. Economic Census Capital-Output Ratio: NBER-CES Manufacturing Industry Tobin s Q: Worldscope Table 2: Tobin s Q, Capital-Output Ratio and Industry Concentration [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] Dependent variable: q Dependent variable: ky Con (0.078) (0.083) (0.068)** Con (0.109) (0.118) (0.087)* Con (0.126)** (0.134)** (0.097) Con (0.157)** (0.174)** (0.094) Constant (0.031)*** (0.030)*** (0.030)*** (0.031)*** (0.027)*** (0.027)*** (0.027)*** (0.028)*** (0.014)*** (0.014)*** (0.014)*** (0.015)*** R-squared Observations SIC SIC Sectors Sector FE YES YES YES YES NO NO NO NO NO NO NO NO SIC2 FE NO NO NO NO YES YES YES YES YES YES YES YES TIME FE YES YES YES YES YES YES YES YES YES YES YES YES Notes: Robust standard errors clustered at 2-digit SIC level in parenthesis. * significant at 10%; ** significant at 5%; *** significant at 1%. SIC4 and SIC2 indicate the number of groups included in the regressions classified at the 4 and 2-digit SIC level. Sectors indicates the number of groups included using the broader sector definition. Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

32 Market Power: The Industry Concentration Rate (II) Figure 14: Tobins Q, Capital-Output Ratios and Industry Concentration 33 Tobin s Q Concentration 20 (%) Capital Output ratio Concentration 20 (%) Notes: Own calculations obtained from ln (X it ) = α 0 + α 1 ln (ConY it ) + ɛ it,, where X represents the Tobin s Q and the capital-output ratio in the vertical and the horizontal axis respectively. Con20 is the share of sales of the 20 largest companies in the industry, and ɛ is a classic disturbance term. Both axis show the coefficient α 1 in %. Both equations are constraint to have the same number of observations. The scatter plot is obtained after excluding outliers. An outlier is defined as an observation with a weight of 0 after using the rreg command in STATA. Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

33 Market Power: International Markups (I) We follow Rotemberg and Woodford (1993); Baso and Fernald (2002); Fernald and Neiman (2011); and Karabarbounis and Neiman (2014) The markup (µ) is defined as: The capital share is: µ = S K = RK Y 1 1 S Π = = ( ζx Y 1 S L + S K (11) )( 1 β 1 + δ δ ), (12) where R is the capital rental rate, K is the stock of capital, and Y is the gross domestic product. The first term in the right hand side is composed by product of the relative price of investment (ζ) and the gross capital formation (X). Data: National Accounts. Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

34 Market Power: International Markups (II) Table 3: Markups: International Comparison ( ) [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] 2FE MG CMG CMGt 2FE MG CMG CMGt 2FE MG CMG CMGt Dependent variable: q Dependent variable: ky Dependent variable: lis Xt (0.059)*** (0.049)*** (0.069)*** (0.070)*** (0.014)** (0.034)*** (0.037)** (0.040)*** (0.046)*** (0.044)*** (0.054)*** (0.067)*** markupt (0.052)** (0.135)*** (0.203)** (0.238)** (0.033)* (0.040)** (0.027)*** (0.030)* (0.050)** (0.045)*** (0.034)*** (0.056)*** markupt (0.106) (0.222) (0.229) (0.272) (0.045)*** (0.042)*** (0.028)*** (0.031)*** (0.070)** (0.061) (0.069)* (0.071)* t (0.002) (0.002) (0.000) (0.001) (0.000) (0.001) Constant (0.047)*** (0.167)*** (0.277) (0.431) (0.030)*** (0.043)*** (0.034) (0.044)*** (0.028) (0.053) (0.033) (0.086)* Number of id Observations R-squared RMSE lr-markup se-markup Trend CD test Abs Corr Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

35 Corporate Governance (I) Data (U.S. Firm Level and Cross-Country, ): Corporate Governance: Asset4 ESG Database Investment: Worldscope Capital-Output Ratio: AMECO Tobin s Q: Worldscope Figure 15: Tobin s Q, Investment and Corporate Governance (U.S.) Tobin s Q Corporate Governance (%) Investment (%) Tobin s Q Investment Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

36 Table 4: Tobin s Q, Investment and Corporate Governance (U.S.) Panel A [1] [2] [3] [4] [5] [6] [7] Dependent variable: q GOV t (0.044)*** (0.047)*** (0.048)*** (0.051)*** (0.036)** (0.045)*** (0.040)*** Constant (0.029)*** (0.029)*** (0.032)*** (0.034)*** (0.024)*** (0.031)*** (0.026)*** R-squared Panel B [1] [2] [3] [4] [5] [6] [7] Dependent variable: INV GOV t (0.019)** (0.019)** (0.019)** (0.018)** (0.018)** (0.017)** (0.019)*** Constant (0.013)*** (0.014)*** (0.013)*** (0.012)*** (0.012)*** (0.011)*** (0.012)*** R-squared Observations Firms SIC SIC SIC SIC2 FE YES YES NO NO NO NO NO SIC3 FE NO NO NO NO NO YES NO SIC4 FE NO NO NO NO YES NO YES Time FE NO YES NO NO NO NO NO SIC2*Time NO NO YES NO NO YES YES SIC3*Time NO NO NO YES NO NO NO Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

37 Corporate Governance (III) Figure 16: Tobin s Q, Capital-Output ratio and Corporate Governance (I) Tobin s Q JPN GRC AUT POL DNK DEU ITA CHE NZL NOR FIN LUX SWE IRL ESP NLD AUS FRA BEL PRT USA GBR CAN Capital Output ratio JPN GRC AUT POL DEU DNK ITA SWE ESP CHE PRT FRA FIN NLD BEL NZLIRL AUS LUX NOR USA GBR CAN Corporate Governance (%) (a) Tobin s Q Corporate Governance (%) (b) Capital-Output ratio Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

38 Corporate Governance (IV) Figure 17: Tobin s Q, Capital-Output ratio and Corporate Governance (II) Tobin s Q Corporate Governance (%) CAN DEU CHE BEL FIN USA SWE JPN NLD NZL AUT PRT GBR DNK FRA IRL ITA ESP NOR Capital Output ratio Corporate Governance (%) GRC Notes: Own calculations obtained from ln (X t ) = α 0 + α 1 GOV t + ɛ t, where X represents the Tobin s Q and the capital-output ratio in the vertical and the horizontal axis respectively. GOV is the corporate governance index, and ɛ is a classic disturbance term. Both axis show the coefficient α 1 in %. Both equations are constraint to have the same number of observations. Each regression only includes countries which have at least 10 observations for the period The scatter plot is obtained after excluding outliers. An outlier is defined as an observation with a weight of 0 after using the rreg command in STATA. Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

39 Conclusions We find that the increase in Tobin s Q can explain between 41% and 57% of the labor income share decline. Our model emphasises the role of asset prices in explaining the decline of the labor share within a standard capital-output framework. Relative prices of investment are not relevant. Indeed, our model suggests that the problem is not too much physical capital, but the increase of financial wealth with respect to productive capital. Compatible with standard values of σ (survey Chirinko, 2008). Policies aiming at reversing the trend should target incentives on corporate investment, even if this is at the expense of equity valuation and equity returns. Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

40 Additional Materials Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

41 Capital-Output ratio Figure A.1: Capital-Output ratio Canada France Germany Capital output ratio Japan United Kingdom United States Source: Ameco back Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

42 Table A.1: Selected economies and sample period id Country Sample period id Country Sample period 1 Australia** Luxembourg* Austria** Mexico** Belgium** Morocco Brazil* Netherlands** Canada** New Zealand** Chile* Norway** China Peru Colombia Philippines** Denmark** Poland Finland** Portugal** France** South Africa** Germany** Spain** Greece** Sri Lanka Hong Kong** Sweden** Hungary Switzerland** India* Thailand Ireland** Turkey Israel 1993, UK** Italy** US** Japan** Venezuela Korea** back Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

43 Pesaran (2004) CD test Table A.2: Cross-section dependence tests a) Levels: b) Diff: Variable lis q rp Variable lis q rp CD-test CD-test p-value p-value corr corr abs(corr) abs(corr) c) Het. AR(2) d) Het. AR(2) CCE Variable lis q rp Variable lis q rp CD-test CD-test p-value p-value corr corr abs(corr) abs(corr) Notes: CD-test shows the Pesaran (2004) cross-section dependence statistic, which follows a N(0, 1) distribution. H0 = cross-section independence. corr, and abs(corr) report, respectively, the average and average absolute correlation coefficients across the N(N 1) set of correlations. back Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

44 Pesaran (2007) CIPS Unit Root Test Table A.3: Unit root tests a) Pesaran (2007) CIPS test: Constant Lags lis (p) q (p) rp (p) b) Pesaran (2007) CIPS test: Constant and deterministic trend Lags lis (p) q (p) rp (p) Notes: Pesaran (2007) CIPS test values are obtained from the standardised Z-tbar statistic. H0 = nonstationarity. Lags indicates the number of lags included in the ADF regression. back Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

45 Intuition y it = β i x it + φ i f t + ψ i + ε it, (A.1) y t = βx t + φf t + ψ, ε t 0 as N (A.2) f t = φ 1 (y t ψ βx t ) (A.3) Substitution for f t in equation (A.1): y it = β i x it + φ i φ 1 (y t ψ βx t ) + ε it, y it = β i x it + Π 1i y t + Π 2i x t + Π 3i + ε it (A.4) (A.5) back Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

46 Table A.4: Error Correction Model: WID Q [1] [2] [3] [4] [5] [6] [7] 2FE CCEP MG CMG CMGt CMGt1 CMGt2 lis t (0.047)*** (0.068)*** (0.111)*** (0.112)*** (0.155)*** (0.181)*** (0.337)*** q t (0.009) (0.012) (0.036) (0.007) (0.029)* (0.038)* (0.078)* rp t (0.032) (0.055) (0.108) (0.093) (0.054) (0.124) (0.461) q (0.018)** (0.022)** (0.039) (0.018)** (0.010)*** (0.020)** (0.036)** rp (0.076) (0.080) (0.104) (0.077) (0.075) (0.054)*** (0.297) t (0.001) (0.003) (0.004) (0.004) Constant (0.034)* (0.082)*** (0.129) (0.130) (0.179) (0.253) Number of id Observations R-squared RMSE Trends lr-q se-q lr-rp se-rp CD test Abs Corr Int I(0) I(0) I(0) I(0) I(0) I(0) I(0) back Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

47 Weak Exogeneity Test We estimate an informal causality test based on the Granger Representation Theorem (GRT), which states that cointegrated series can be represented in the form of an ECM, which in our case is: K K lis it = α 1i + λ 11 û i,t j + φ 11ij lis i,t j + φ 12ij q i,t j j=1 j=1 j=1 K K q it = α 2i + λ 21 û i,t j + φ 21ij lis i,t j + φ 22ij q i,t j j=1 K K rp it = α 3i + λ 31 û i,t j + φ 31ij lis i,t j + φ 32ij q i,t j j=1 j=1 K j=1 K j=1 K j=1 φ 13ij rp i,t j + ɛ 1it, φ 23ij rp i,t j + ɛ 2it, φ 33ij rp i,t j + ɛ 3it, where û it = lis it ˆβ 1i q it + ˆβ 2i rp it is the disequilibrium term. Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

48 Table A.5: Weak exogeneity test no CA CA Model lis q rp lis q rp MG CMG CMGt CMG1 CMGt1 CMG2 CMGt2 Avg. λ ρ * Avg. λ ρ Avg. λ ρ * Avg. λ ρ Avg. λ ρ Avg. λ ρ * * 0.56 Avg. λ ρ * 0.44 Notes: Avg. λ shows the robust mean coefficient for the disequilibrium term on the ECM. Asterisks highlight cases which do not support a causality relationship for our analysis. back Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

49 Intangible Assets Koh et al. (2016) claim that IPP products can explain 100% of the U.S. labor share decline. IPP assets can bias the Tobin s Q if there are measurement problems (but also through equilibrium mechanisms Gutierrez and Philippon, 2016). Tobin s Q raise could be accounting for the increasing importance of IPP products. Figure A.2: Intangible Assets Intensity ( ) Intangible Assets/Total Assets (%) IPP IPP (goodwill excluded) Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

50 Intangible Assets Figure A.3: U.S. Tobin s Q with and without IPP Tobin s Q Q Q without IPP Notes: Market value weighted average Q. 4-digit SIC industry fixed effects. back Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

51 Corporate Governance (IV): Figure A.4: Tobin s Q, Capital-Output ratio and Corporate Governance: Tobin s Q Corporate Governance (%) JPN AUT DEU FIN SWE CHE GRC BEL AUS GBR NLD USA CAN DNK FRA IRL PRT Capital Output ratio Corporate Governance (%) ITA NOR ESP Notes: Own calculations obtained from ln (X t ) = α 0 + α 1 GOV t + ɛ t, where X represents the Tobin s Q and the capital-output ratio in the vertical and the horizontal axis respectively. GOV is the corporate governance index, and ɛ is a classic disturbance term. Both axis show the coefficient α 1 in %. Both equations are constraint to have the same number of observations. The scatter plot is obtained after excluding outliers. An outlier is defined as an observation with a weight of 0 after using the rreg command in STATA. Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51

52 Asset Prices and Productive Capital Figure A.5: Market for capital 1.09 Capital Market with q<1 and two different prices Pk p(r) p'(r) k(r) k'(r) a(r) Equity Return Capital, Assets Gonzalez and Trivín Asset Prices and the Labor Share 11 th April / 51