Finance and labor: an overview 1 Marco Pagano University of Naples Federico II, CSEF, EIEF and CEPR F IR N C o r po r ate F i n an ce Wo r k s ho p S ydne y, 14 O ct obe r 20 13 Expanding research field 2 Today I will focus only on three areas related to corporate finance: 1. Does corporate ownership (family vs. non-family) affect firm-level insurance provision to workers? 2. Does capital structure (leverage) affect the relative bargaining power of capital and labor? 3. What is workers’ role in corporate governance, and how does it affect firm performance? But there are many other fast-developing areas: effect of finance on employment and job reallocation, effect of employment laws and trade unions on innovation and risk taking, etc. 3 Issue no. 1: corporate ownership and employment risk Which firms are better at insuring employees? 4 In principle, large firms with good access to financial markets should be better at it But if insurance is given via implicit long-term contracts, these firms also face greater commitment problems (“breach of trust”): managerial turnover hostile takeovers Family firms can better commit to such contracts with employees: no danger of hostile takeovers family “name” is on the line: reputation! Symmetric weakness: they often have no easy access to financial markets to unload risk The verdict is up to the evidence… French and U.S. evidence 5 In France, the pro seems to prevail on the con, at least for listed family firms in the late 1990s: heir-managed firms pay lower average wages and earn larger profits, and their employment is less sensitive to industry sales shocks (Sraer and Thesmar, 2007) family-promoted CEOs are associated with lower job turnover and less wage renegotiation (Bach and Serrano-Velarde, 2010) family firms have fewer layoffs, sanctions and disputes ending in court (Müller and Philippon, 2007; Waxin, 2009) In U.S. listed companies, the evidence is more limited: family management: downsizing is less likely, but more severe family owners: large job cuts (> 6%) are less likely (Block, 2008). Worldwide evidence 6 Ellul, Pagano and Schivardi (2013) use data from 41 countries to investigate if employment and wages respond to sales shocks differently between o firms with different characteristics, e.g. family vs. non-family o in countries with different social arrangements, e.g. with high vs. low public provision of job security Also look at interaction between the two: is the insurance role of family firms less prominent where there is more job security? Distinguish between different types of shocks to sales: industry- vs. firm-level negative vs. positive transitory vs. persistent Firm-level international data 7 Financial and accounting data from 41 countries for the period 1988- 2011 obtained from Worldscope and Osiris Use firms with employment data for at least 5 years: this screen reduces the number of firms to 6,298, i.e. 89,815 firm-year observations Wage data is only available for 2,485 firms Ownership data from Ellul, Pagano and Panunzi (2010) identifying a family as the firm’s ultimate blockholder Dependent variable in the employment insurance regressions: log change of total employment Two different dependent variables in the wage regressions : log change of real wage to test for wage insurance log of average wage to test whether insurance is priced by wages Country-level data 8 Country-level worker protection provided by social security system: gross replacement rates: unemployment benefits as fraction of last wage, time-varying Country-level worker protection provided by the market: labor market tightness: reciprocal of ratio of long-term to total unemployed Measure of financial development (stock market cap. to GDP) Employment regression 9 The specification of the employment growth regression is: ηijct = growth rate in the employment of firm i in year t εijct = measure of the shock: either the unexpected change in the sales of firm i or the change in the sales industry j (ex-firm i) in year t Fit = family-firm dummy – Family if a family blockholder has at least 20% of cash flow rights Sct = replacement rate (measure of the effectiveness of the public employment insurance system) in country c and year t Xijct = vector of company-specific variables μcj = country-industry effect μt = year effect Employment insurance: industry shocks 10 Δ Industry Sales Family Firms Δ Industry Sales Family Firms Δ Industry Sales Unemployment Security Δ Industry Sales Family Firms Unemployment Security Δ Industry Sales Family Firms Labor Market Tightness Control Variables Fixed Effects Year Dummies R2 (1) (2) (3) (4) 0.1083** (2.58) 0.0253 (1.27) 0.0906** (2.27) 0.0174 (1.21) 0.0722** (2.10) 0.0101 (1.07) 0.0863** (2.39) -0.0991*** (-2.81) -0.0898** (-2.49) 0.0314 (1.46) 0.1928** (2.10) -0.0659** (-2.20) 0.0415 (1.44) 0.1399* (1.80) -0.0750** (-2.40) 0.0259 (1.24) 0.1604* (1.88) - 0.0049 (1.28) Yes Yes Yes Yes CountryIndustry Yes 0.45 CountryIndustry Yes 0.49 CountryIndustry Yes 0.50 Firm Yes 0.56 Insurance provided by family firms and social security 11 Vertical axis: country-level estimate of employment insurance provided by family firms Wage growth regression 12 The specification of the wage growth regression is: wijct = growth rate of the average real wage of firm i in year t εijct = measure of the shock (either to the sales of firm i or of its industry j in year t) Fit = family-firm dummy variable (equal to 1 for family firms, and 0 otherwise) Sct = replacement rate (measure of the effectiveness of the public employment insurance system) in country c and year t Xijct = vector of company-specific variables μcj = country-industry effect μt = year effect Wage insurance: industry shocks 13 Δ Industry Sales Family Firms (1) (2) (3) (4) 0.0426*** (3.12) 0.0391*** (2.82) 0.0340** (2.53) 0.0427** (2.65) -0.0104* (-1.90) -0.0095* (-1.70) -0.0051 (-1.52) - 0.0182** (2.61) 0.0109* (1.92) 0.0233** (2.35) -0.0186* (-1.70) -0.0212 (1.57) 0.0580* (1.74) 0.0662 (1.50) Δ Industry Sales Family Firms Δ Industry Sales Unemployment Security Δ Industry Sales Family Firms Unemployment Security Firm Control Variables Fixed Effects Year Dummies R2 Yes Yes Yes Yes CountryIndustry Yes CountryIndustry Yes CountryIndustry Yes Firm 0.10 0.12 0.12 0.12 Yes Is employment insurance priced? 14 (1) (2) (3) (4) -0.0921** (-2.45) 0.0047** (2.18) -0.0541** (-2.28) 0.0049** (2.05) -0.0380** (-2.04) 0.0041* (1.89) 0.0058** (2.28) 0.0091 (1.01) No 0.0087 (0.93) Yes 0.0030 (0.92) 0.0072 (0.85) Yes 0.0170 (1.34) Yes CountryIndustry CountryIndustry CountryIndustry Firm Year Dummies Yes Yes Yes Yes R2 0.08 0.09 0.11 0.14 Family Firms Unemployment Security Family Firms Financial Development Family Firms Unemployment Security Firm Control Variables Fixed Effects Is employment insurance priced? (2) 15 Firms that provide less employment insurance pay higher real wages Vertical axis: firm-level wage net of industry, country and time effects Horizontal axis: firm-level estimate of sales shocks “pass-through” (inversely related to employment insurance) 16 Issue no. 2: capital structure and wage bargaining Theoretical literature 17 By taking more debt, a company reduces “surplus on the bargaining table”: Baldwin (1983), Bronars and Deere (1991), Matsa (2010) Perotti and Spier (1993) add incentive problem on shareholders’ side: debt not only reduces surplus, but also creates debt overhang shareholders have less incentive to invest wage concessions when profits are small Dasgupta and Sengupta (1993) add incentive problem also on workers’ side: bankruptcy risk reduces workers’ effort (or investment in firm-specific human capital) too much debt not optimal Evidence on debt and wages 18 Hanka (1998): U.S. firms with more debt pay lower wages and fund their pension plans less generously, controlling for performance Benmelech, Bergman and Enriquez (2009): airlines in distress obtain wage concessions from workers with underfunded pension plans; effect is weaker for workers with greater pension insurance Myers and Saretto (2010): in wage negotiations, unions are more likely to strike and “win” if firm debt has decreased in previous years; when firms win, they do not increase debt further Evidence on debt and workers’ insurance 19 Hypothesis: when workers are protected by larger unemployment benefits, firm choose higher leverage Agrawal and Matsa (2010) find precisely this using U.S. data from 1950 to 2008: increases in U.S. state unemployment insurance (UI) benefit entitlements are associated with increases in firm leverage. Doubling UI 4.1 percentage points increase in debt/assets ratio Impact stronger for firms where workers face greater unemployment risk (layoff separation rates), that are more likely to fire workers in adversity (low operating cash flow, no dividend) and have greater labor intensity Evidence on debt and unions 20 Matsa (2010): in the U.S., collective bargaining coverage and pro-union changes in state labor laws increase firm leverage (except in industries with low union presence). The same is found for Sweden by Cronqvist et al. (2009) Bronars and Deere (1991): workers less likely to join a union if debt leaves less surplus on the bargaining table. In the U.S. leverage is higher in firms facing greater threat of unionization (but possible endogeneity bias) Simintzi, Vig and Volpin (2009): in firm-level data from 21 countries, greater employment protection (EPL union power) leads to lower debt in countries with weak creditor rights (with likely renegotiation in bankruptcy) International evidence on debt and EPL Effect of Employment Protection on Corporate Debt 21 0.06 NZ, UK 0.04 0.02 NL, DE, DK, AT, AU JP 0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 ES -0.02 SE -0.04 IE -0.06 NO, IT, BE FI CA US, CH, PT, GR -0.08 FR Creditor Rights -0.1 Source: estimates of Simintzi, Vig and Volpin (2009), Table VII, Column 3 Assessing existing evidence 22 U.S. evidence: in line with strategic view of corporate debt: leverage reduces wages, and is more intensively used when unions are stronger International evidence: strategic use of leverage (more debt where unions are stronger) in countries with pro-liquidation bankruptcy law not in those with pro-renegotiation bankruptcy law Workers’ protection in bankruptcy 23 Existing research neglects that there is much cross-country (and some time-series) variation in worker protection in bankruptcy: workers’ seniority in bankruptcy law government guarantees, in two varieties: guarantee funds for wages, severance pay and pensions unemployment insurance benefit Andrew Ellul and I have collected data on these items via questionnaires to Lex Mundi law firms (for OECD countries) information drawn from the web (for non-OECD countries) We investigate how leverage correlates with workers’ protection: same spirit as Agrawal and Matsa (2010) for U.S. Measuring worker seniority in bankruptcy 24 Worker seniority in liquidation differs across countries. Andrew Ellul and I looked at the rank of workers’ claims relative to the following claim classes: secured debt (e.g. real estate mortgage loans) expenses of the bankruptcy procedure post-petition credit extended to debtor unpaid taxes unsecured debt Define workers’ seniority from 0 to 4, so that 0 = they are treated as unsecured creditors, 4 = they are the most senior 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 Austria Finland Germany Ireland Slovak republic Australia Denmark Luxembourg Netherlands New Zealand Sweden Switzerland Turkey USA Belgium Canada Greece Hong Kong Iceland Italy Japan Korea Malaysia Singapore South Africa Thailand UK Spain Argentina Norway Poland Brazil Czech Republic France Hungary India Indonesia Mexico Salary priority around the world 25 Salary priority Public guarantees around the world 26 Unpaid wages, retirement allowance and pensions are paid upfront by a government fund that acquires the same seniority as the workers, in all EU countries since the 1980s, based on Council Directive 80/987/EEC (limits vary: e.g. 3 months’ pay in Germany, 5 in Hungary) Australia since 2001, with annual income cap of A$108,300 in 2010 Canada since July 2008, Hong Kong (both with caps) Japan: up 80% of unpaid wages and severance pay In the U.S., pensions (only) are guaranteed by the PBGC, up to $51,750 per year/employee in 2008 No guarantee fund in: Argentina, Brazil, Malaysia, Mexico, New Zealand, South Africa Unemployment benefits around the world 27 60 50 1971 40 1981 30 1991 2001 20 2007 Denmark Portugal Belgium France Ireland Spain Finland Netherlands Norway Switzerland Sweden Italy Austria New Zealand Germany Australia USA Greece Canada Japan 0 UK 10 Average unemployment benefit replacement rates for 2 earnings levels, 3 family situations and 3 unemployment durations, sorted by 2007 level. Source: OECD. Merge these indicators with company data 28 We started exploring how firm-level leverage correlates with these measures of workers’ protection in bankruptcy Our initial data set of company-level financial and accounting information: 11,290 firms from 38 countries, 1990-2008, drawn from Worldscope and Osiris We only use firms for which we can find accounting data for at least 5 years: this reduces the number of firms to 7,588 94,056 firm-year observations Standard errors clustered at the country-industry level Leverage and workers’ protection in bankruptcy 29 Dep. Variable: Book Leverage (1) (2) (3) (4) (5) 0.0138*** (2.98) - - - - Salary Priority x Firm Employees (Scaled by Assets) - 0.0592** (2.09) - - 0.0418* (1.89) Unemployment Insurance - - 0.0218* (1.74) - - Unemployment Insurance x Firm Employees (Scaled by Assets) - - - 0.0914 (1.60) 0.0816 (1.49) Control Variables Yes Yes Yes Yes Yes Firm Fixed Effects No Yes No Yes Yes Industry and Time Fixed Effects Yes Yes Yes Yes Yes 94,056 94,056 94,056 94,056 94,056 Salary Priority No. of observations 30 Issue no. 3: labor and corporate governance Employees’ ownership and control rights 31 31 percent of U.S. workers invest in their company. Employee Stock Ownership Plans (ESOPs) have become common: 1,601 plans in 1974, 11,500 in 2000 5.3 million workers in 1980, 7.2 million in 1995 Co-determination is mandated by law in some countries: Germany: 1/2 of seats on supervisory board in companies with more than 2,000 employees, 1/3 in listed companies with 500-2,000 employees also in other EU countries workers have some control rights (see graph) 2 Australia Austria Belgium Canada Czech Republic Denmark Finland France Germany Greece Hungary Ireland Italy Japan Korea Mexico Netherlands New Zealand Norway Poland Portugal Slovak Republic Spain Sweden Switzerland Turkey United Kingdom United States Mandatory control rights to workers 32 Work councils mandated by law Employees appoint some board members 1 0 Impact on corporate governance 33 Worker control rights may affect corporate governance, e.g. shift balance between shareholders and management If there is separation between ownership and control, e.g. with dispersed share-ownership, it is a 3-players game: shareholders • industrial relations conflict over division of surplus • agency conflict: workers shirk, underinvest in firm-specific human capital workers owner-manager agency conflict: manager steals or under-monitors workers (if he owns a small equity stake) manager • conflict if manager owns a large equity stake • congruence if manager has small equity stake Managers’ “easy life” and “natural alliance” with workers 34 Pagano and Volpin (2005): If managers have a small equity stake (conflict with shareholders), they wish to overpay workers to buy an “easy life”: less monitoring, better industrial relations They will also want to deter takeover raiders who would replace them So their interest is aligned to that of workers: pay generous wages to buy an easy life deter control changes that would breach implicit contract “natural alliance” between workers and managers in firms with owner-manager conflict (“bad governance”) Evidence on “easy life” and “natural alliance” 35 Swedish managers go for the “easy life” (Cronqvist et al., 2009): wages are inversely associated with the manager’s equity stake (after wage bargaining became decentralized) U.S. anti-takeover state laws in the 1980s were associated with wage increases (Bertrand and Mullainathan, 1999) ESOPs correlate with wage increases (Kim and Ouimet, 2009) less frequent takeovers (Chaplinsky and Niehaus, 1994; Beatty, 1995) Successful raiders cut wages: transfer from workers to shareholders accounts for 10% of hostile takeover premium in 18 subsequent years (Rosett, 1990) hostile takeovers lower union wage premium (Becker, 1995) Incentive effects of workers’ ownership/control 36 So workers’ ownership and control tends to increase their “share of the pie” But they may also change the “size of the pie” by affecting workers’ effort and human capital investment productivity workers’ cooperation in labor relations strike frequency corporate strategic choices company growth, profitability, ... Most studies find that after ESOPs firms experience productivity increases (e.g. Jones and Kato, 1995; Beatty, 1995) positive stock price reactions (e.g., Chaplinsky and Niehaus, 1994) Incentive effects (2) 37 Kim and Ouimet (2009) show that sign depends on stake: Tobin’s Q and profits increase for ESOPs < 5% are unaffected for ESOPs > 5% Same for German codetermination (Fauver & Fuerst, 2006): it raises Tobin’s Q (in manufacturing and transportation) and dividend payout, only if workers have at most 1/3 of seats Greater cooperation in labor relations: ESOPs reduce strike incidence in labor disputes (Cramton, Mehran and Tracy, 2008) in France, employee board representation reduces the incidence of strikes and individual labor disputes (Waxin, 2009) Conclusion 38 The interface between labor and corporate finance is capable of giving exciting insights, as it goes beyond the interplay between financial claimholders Lively research area – yet, still much we don’t know Part of the problem is practical: need to merge databases of worker-level or plant-level data with standard financial databases = lots of hard work! Part of the problem is that “silo-busting is exceptional in academia – one is expected to specialize: there is a lot of turf warfare” (Jared Diamond, author of Guns, Germs and Steel, on yesterday’s FT, p. 13)