Investment and the weighted average cost of capital: new micro evidence for France

Size: px

Start display at page:

Download "Investment and the weighted average cost of capital: new micro evidence for France"

Stella Cecilia Barnett
5 years ago
Views:

1 Investment and the weighted average cost of capital: new micro evidence for France J. Carluccio 1 C. Mazet-Sonilhac 1 J.S. Mésonnier 1 1 Banque de France Very Preliminary. Please do not circulate. This version: February 5, 2018 Opinions expressed are those of the authors and do not necessarily reflect the views of the Banque de France or the Eurosystem

2 Motivation Persistent slow growth environment in developed economies post 2008, notably in the euro area. Corporate investment deficit often seen as a major cause. low investment rate in France until 2014, recent upsurge, but maybe partly fueled by temporary accelerated depreciation scheme ( ) Reasons? expected demand (and heightened uncertainty), competition, short-termism... What about financial factors? Too high cost of capital and hurdle rates in spite of lowered cost of debt (very accomodative MP)? cf. Villeroy de Galhau report (2016)

3 Subdued corporate investment in France since 2008? 110 Two metrics of the gross investment rate (2008=100) GI/VA in 2016 = 24% GI/FA in 2016 = 11,7% GI/VA GI/FA Note. Source: INSEE. GI = Gross fixed capital formation of NFCs. VA = Gross VA of NFCs. FA = Net fixed capital of NFCs.

4 Does the high CoE explain subdued investment? Note. Cost of equity for large listed French NFCs as in Mazet-Sonilhac and Mésonnier (2016). Effect of share buybacks not included. Average corporate bond rate from Gilchrist and Mojon (2016). Interest rate on corporate loans: interest rate on new MFI loans to French NFCs above 1 M euros.

5 Main findings We document the link between corporate investment and Tobin s Q components (among which the real WACC) using French micro data for non-financial groups over NB: Listed + Non-listed, Large + Small corporates Main findings: Strong negative correlation of firms investment with WACC. No clear role for lower competition in contrast to the US case Lower coefficient after 2008: margin constraints may dominate post-crisis? WACC more important for SMEs, less so for larger firms (in particular post-2008). At the sector-level, some role of the WACC in driving investment in real estate and equipement, not in intellectual property. Some negative role of uncertainty for investment in Eq. and IP.

6 Related Literature WACC (and cost of equity) matter for investment when properly measured: Franck & Shen (2016), Melolinna, Miller & Tatomir (2017) Investment-less growth in the US since 2000s reflects short-termism of some II and lower competition: Philippon & Gutierrez (2017) Share buybacks explain disconnect between Q, net funding and investment in the US since late 90s: Lee, Shin, Stulz (2016) High COE and low interest rates, a cause of carry trade and share buybacks: Blundell-Wignall & Roulet (2013), Acharya & Plantin (2017) Net lending (saving investment) by NFCs in US and Germany, but not in France: Gruber & Kamin (2015)

7 Empirical model Regressions motivated by a simple Q-model of investment, following Franck & Shen (JFE, 2016). Firm chooses I to max NPV, constrained by K accumulation and K-adjustment cost It K t = 1 φ + qt φ Using symplifying assumptions (Abel & Blanchard, 1986): I it K it 1 = β 0 + β 1 Profits it K it 1 + β 2 (L it (1 τ it ) CoD it + (1 L it ) CoE it ) }{{} WACC it + θx it + δ t + γ f + ɛ it with leverage L it = Gross Debt / [Gross Debt + Equity].

8 Data Consolidated BS data for some 7,510 French corporate groups (hereafter: firms) from BdF s FIBEN Groupes database over the period Stock market information for 600 listed corporates (Datastream, BdF-DE) = sector-specific measure of COE (DS industries matched into NAF-2 A38 sectors) Sector-level (A38) information on fixed capital formation and consumption by asset type (real estate, equipment, intellectual property), from INSEE Cleaning 2,482 corporate groups over ( 200 are listed). Average cross-section: 1, 100 corporate groups (Smaller: 600 / Larger: 1400). Average # consecutive years per group: 5,7. Cleaning details

9 Variable construction Firm-level measure of the WACC: Sector-level (NAF-2/A88) measures of nominal COE using DDM H-model for listed groups in each sector. COE details Firm-level apparent cost of debt, leverage and tax rate. Deflator of COE, COD, WACC: core HICP. Firm controls: size, depreciation rate, investment into financial assets, M&A dummy. Sector-level controls (A38): Herfindahl index of firms sales (using FIBEN database for legal entities) Competition Uncertainty (sector-year dispersion in ROA) Uncertainty Sector-level measures of investment by asset type: Investment and depreciation rates in intangibles (IP) vs equipment (EQ) vs real estate (RE)

10 Micro Data: descriptive statistics Variable N mean sd p10 p25 p50 p75 p90 Investment rate (I/FA 1) 13, Net Investment rate (NI/FA 1) 13, Depreciation rate δ (DEP/FA 1) 13, Cash-flows to Fixed Assets (OIBDP/FA 1) 13, Nominal Cost of Debt 13, Nominal Cost of Equity 13, WACC 13, Real Cost of Debt (CoD) 13, Real Cost of Equity (CoE) 13, (1 - τ) * CoD 13, Real WACC 13, Tax rate τ (Apparent, sector level) 13, Leverage (Debt / Equity + Debt) 13, Firm size (TA) 13, ,451 8,674,926 19,412 30,430 56, , ,606 Firm age 13, Investment in financial assets (IFI/FA 1) 13, Note. Sample of 2,482 corporate groups (firms) over Gross investment: acquisitions minus cession of fixed assets, net investment: gross investment minus depreciation and amortization, Cash-flows: Operating Income Before Depreciation (EBITDA), the (firm-specific) implied cost of debt: interest payments normalized by gross longterm debt and leverage: D/(D+E). CoE: computed at the sector level for 49 sectors with a DDM H-Model. Nominal returns are deflated using the french core HICP

11 Distribution of estimated WACC Figure 1 : WACC at sector-level (Real vs Nominal)

12 Real WACC, CoE and CoD in our sample Figure 2 : Median per year of real WACC, COE and COD (firm-level)

13 Sample validation: micro vs macro Figure 3 : Gross investment measured by macro and micro data

14 Is the Q-model of I vindicated? A preliminary test based on listed firms I/FA Q I/FA (1) (2) (3) (4) (5) (6) Tobin s Q (0.016) (0.016) Cash-flows (OIBDP/FA) (0.041) (0.033) (0.015) (0.015) Real WACC (0.874) (0.830) (0.168) (0.216) Depreciation rate (0.068) (0.069) (0.057) (0.058) Firm FE Yes Yes Yes Yes Yes Yes Year FE No Yes No Yes No Yes Firm controls Yes Yes No No Yes Yes Observations 1,475 1,475 1,475 1,475 1,475 1,475 R-Square Note. OLS regressions. Sample period: Unbalanced panel of listed corporate groups only. Tobin s Q is [Total assets + market value of equity - book value of equity] / total assets. Firm controls: depreciation over lagged fixed assets, log size, a dummy for M&A activity and purchases of financial assets. Standard errors in parentheses are clustered at the firm level.

15 Investment and the WACC: baseline results Investment rate (1) (2) (3) (4) (5) (6) (7) Cash-flows (OIBDP/FA) (0.007) (0.007) (0.008) (0.008) (0.008) (0.008) (0.008) Real WACC (0.084) (0.103) (0.103) Real CoD * (1 - τ) (0.054) (0.053) Real Cost of Equity (0.052) (0.051) Leverage (0.015) (0.015) Firm FE Yes Yes Yes Yes Yes Yes Yes Year FE No Yes Yes Yes Yes Yes Yes Firm controls No No Yes Yes Yes Yes Yes Observations 13,814 13,814 13,814 13,814 13,814 13,814 13,814 R-Square Note. OLS regressions. Sample period: Unbalanced panel of all firms in our sample. Dependent variable: gross fixed capital investment rate (I /FA 1 ). Firm controls: depreciation over lagged fixed assets, log size, a dummy for M&A activity and purchases of financial assets. Sector controls: competition and uncertainty. Standard errors in parentheses are clustered at the firm level.

16 Do competition and uncertainty matter? Investment rate (1) (2) (3) (4) (5) Cash-flows (OIBDP/FA) (0.008) (0.008) (0.008) (0.008) (0.008) Real WACC (0.103) (0.085) (0.085) (0.085) (0.103) Concentration (0.122) (0.122) (0.119) Uncertainty (0.015) (0.015) (0.015) Firm FE Yes Yes Yes Yes Yes Year FE Yes No No No Yes Firm controls Yes Yes Yes Yes Yes Observations 13,814 13,814 13,812 13,812 13,812 R-Square Note. OLS regressions. Sample period: Unbalanced panel of all firms in our sample. Dependent variable: gross fixed capital investment rate (I /FA 1 ). Firm controls: depreciation over lagged fixed assets, log size, a dummy for M&A activity and purchases of financial assets. Sector controls: competition and uncertainty. Standard errors in parentheses are clustered at the firm level.

17 Cost of capital matters less post-2008 (and profits/demand more)... Investment rate Cash-flows (OIBDP/FA) (0.008) (0.014) (0.012) (0.009) Post-2008 x cash-flows (0.008) Real WACC (0.103) (0.195) (0.133) (0.171) Post-2008 x real WACC (0.175) Firm FE Yes Yes Yes Yes Year FE Yes Yes Yes Yes Firm controls Yes Yes Yes Yes Sector controls Yes Yes Yes Yes Observations 13,812 4,551 8,950 13,812 R-Square Note. OLS regressions. Sample period: Unbalanced panel of all firms in our sample. Dependent variable: gross fixed capital investment rate (I /FA 1 ). SMEs: staff < 250 and sales < 50 M euros (or assets < 43 M euros). Mid-Size: staff < 5,000 and sales < 1.5 bn euros (or assets < 2 bn euros). Large firms: above these thresholds. Firm controls: depreciation over lagged fixed assets, log size and a dummy for M&A activity and purchases of financial assets. Sector controls: competition and uncertainty. Standard errors in parentheses are clustered at the firm level.

18 Coefficient break larger for large firms Small Mid - Large Cash-flows (OIBDP/FA) (0.013) (0.028) (0.016) (0.017) (0.010) (0.015) (0.014) (0.010) Post-2008 x cash-flows (0.017) (0.010) Real WACC (0.209) (0.407) (0.253) (0.369) (0.112) (0.184) (0.141) (0.180) Post-2008 x real WACC (0.365) (0.187) Firm FE Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Firm and sector controls Yes Yes Yes Yes Yes Yes Yes Yes Observations 5,353 1,578 3,610 5,353 8,255 3,196 5,373 8,255 R-Square Note. OLS regressions. Sample period: Unbalanced panel of all firms in our sample. Dependent variable: gross fixed capital investment rate (I /FA 1 ). SMEs: staff < 250 and sales < 50 M euros (or assets < 43 M euros). Mid-Size: staff < 5,000 and sales < 1.5 bn euros (or assets < 2 bn euros). Large firms: above these thresholds. Firm controls: depreciation over lagged fixed assets, log size and a dummy for M&A activity and purchases of financial assets. Sector controls: competition and uncertainty. Standard errors in parentheses are clustered at the firm level.

19 Cost of capital and investment by asset-type: sector-level regressions Total Capex Capex by asset type InvTot / FA InvTot / FA InvEQ / EQ InvRE / RE InvIP / IP Cash Flows (Sector, OIBDP/FA) (0.014) (0.013) (0.031) (0.014) (0.017) Real WACC (0.064) (0.064) (0.103) (0.062) (0.063) Depreciation rate (0.147) (0.143) (0.018) (0.375) (0.042) Uncertainty (0.005) (0.009) (0.006) (0.004) Concentration (0.018) (0.045) (0.018) (0.047) Inv. in RE (0.164) (0.040) Inv. in PI (0.108) (0.052) Inv. in Eq (0.053) (0.038) Sector FE Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Observations R-Square Note. OLS regressions over , NACE rev. 2 industries (A38). Sector-level variables from National Accounts, except WACC (aggregated as sector-level from firm-level data). Standard errors in parentheses are clustered at the sector level. EQ stands for investment into machinery and equipment, IP stands for investment in intellectual property (including R&D), RE stands for investment into real estate assets.

20 Conclusion We document negative role of WACC for corporate investment in France since 2003 for a large sample of mostly unlisted corporations. No clear role for competition earlier findings for the US Some role for uncertainty regarding investment in IP/Eq. at sector-level Post 2008: weakened role of cost of capital Relatively high cost of capital has weighed down on investment into fixed assets for SMEs, but less so for large, listed corporations. Increased role for low margins and cash flows during the post-2008 slump? Caveat: mere correlations, no causal claims here. Further research: collect genuine info on hurdle rates and premia? Indeed, the latter may be large (5 8%) (US: Jagannathan et al., JFE 2016, UK: Melolinna et al., BoE 2017).

21 Appendix

22 Data cleaning Drop farm, mining, financial services, real estate Drop years before 2003: not enough groups in cross-section Drop observations w/ missing or negative values of FA, Inv, TA, Equity Drop observations in top 1% / bottom 1% of investment rate and CoE. Drop firms w/ investment rates (corrected for year fixed effects) below 10th percentile. Impose at least 3 consecutive years 2,482 corporate groups over ( 200 are listed). Average cross-section: 1, 100 corporate groups (Smaller: 600 / Larger: 1400). Average # consecutive years per group: 5,7. Cleaning details Back

23 Data: Cost of Equity We construct (NAF-2/A38) sector-level measures of the (real) cost of equity to compute the (real) WACC (deflator: core HICP). Nominal COE estimated based on a standard Dividend Discount Model (H-Model, Fuller and Shia, 1983): CoE t = D t P t 1 ((1 + g LT ) + H 2 ((g ST ) g LT )) + g LT DY : 2-digit sector average of in-sample listed firms DY (A38) g LT : nominal GDP growth, 10 years forecast (source: SPF) g ST : Datastream sector-level expected profit growth, 3-5 years forecast (source: I/B/E/S), mapped into NACE-2 sectors. H : 8 years Back

24 Cost of equity Figure 4 : Cost of equity at sector-level (Real vs Nominal) Back

25 Cost of Debt Figure 5 : Cost of debt at sector-level (Real vs Nominal) Back

26 Competition Figure 6 : Firm concentration at A38 sector-level (Sector Herfinhdal) Back

27 Uncertainty Figure 7 : Uncertainty at A38 sector-level Back

Investment and the weighted average cost of capital: new firm-level evidence for France

Investment and the weighted average cost of capital: new firm-level evidence for France J. Carluccio 1 C. Mazet-Sonilhac 1,2 J.S. Mésonnier 1 1 Banque de France 2 Sciences Po Paris Work in progress. This