24. Corporate Governance

Corporate governance = the field of mechanisms regulating manager behavior and ensuring capital suppliers are reasonably repaid; the relevant parties are the board, managers, and shareholders, and governance specifies each party's responsibilities/rights, sets objectives, and delineates how to achieve them. Failure of the efficient-market hypothesis: product-market competition might punish/replace inefficient management and solve governance, but frictions prevent sufficient corrections. For example, Grossman-Hart (1980): a raider can add value by taking over, but shareholders are so dispersed and free-ride, making it extremely costly for the raider to implement a takeover and weakening the incentive to attack a target. Another: management entrenchment (setting the charter/bylaws so board elections are infrequent or votes hard to pass, making outside takeovers harder). Managers entrench because they enjoy tremendous private benefits from control (Jensen-Meckling 1976 perks; Aghion-Bolton 1992 private control benefit), but these are hard to measure empirically.

24.1.1 Comparing shares with differential voting rights: Zingales (1994). Uses the price difference between voting and nonvoting shares in Italy to show the private benefit of control is large. Institutional: since 1974 Italian public companies can issue saving shares (nonvoting) in addition to common stock (voting), with a promised minimum dividend of 5% of par value, a guaranteed dividend + 2% of par if common shares are paid, seniority in liquidation, but no voting rights. Data: 64 (1987)/86 (1990) Milan-listed companies with both share types, 1987–1990. The voting premium of firm \(i\) at \(t\), \(VP_i^t=\frac{P_{v,t}^i-P_{nv,t}^i}{P_{nv,t}^i}\). Relative Sharpley value RSV: the Sharpley value

$$\varphi_i(v)=\sum_{\mathcal S\subseteq N\setminus\{i\}}\frac{|\mathcal S|!\,(n-|\mathcal S|-1)!}{n!}\big(v(\mathcal S\cup\{i\})-v(\mathcal S)\big)$$

interpreted by Zingales as the probability a small shareholder's votes become pivotal in a random coalition; RSV = the Sharpley value / the fraction of votes held by small shareholders, with the normalized \(RSV/\pi\) (\(\pi\) the fraction of voting shares outstanding) measuring outside shareholders' control power in a contested acquisition. The privilege \(e_t^i\) = the additional dividend distribution of nonvoting shares. Main regression

$$VP_i^t=\alpha+\beta\left(\frac{RSV}{\pi}\right)_t^i+\gamma\frac{e_t^i}{P_{nv,t}^i}+u_t^i \tag{24.1}$$

Results (Figure 24.1 summary: mean voting premium 81.5%; Figure 24.2: higher control rights (higher \(RSV/\pi\)) → higher voting premium): shares with the possibility of pivotal control rights (voting shares) earn higher returns than those without (nonvoting) → parties with higher control rights enjoy private benefits.

Zingales (1994) is constrained by few countries having dual shares, making international comparison hard; Dyck-Zingales (2004) use another method for international data. The controlling block = the block of shares dominating ownership. The price paid per share of a controlling block includes the future-cash-flow claim (same as non-controlling shares) + the private control benefit; the post-transaction market price incorporates only the future-cash-flow claim; the difference reflects the buyer's private control benefit. Let \(P\) be the total price of a controlling-block transaction, \(B_s/B_b\) the private benefit extracted by the seller/buyer, \(Y_s/Y_b\) the total common-security benefit generated by the seller/buyer (from firm performance, shared by all shareholders), \(\alpha\) the controlling block's fraction of total shares (cash-flow rights), \(\lambda\in[0,1]\) the controlling shareholder's (seller's) bargaining power:

$$P=\lambda(B_b+\alpha Y_b)+(1-\lambda)(B_s+\alpha Y_s) \tag{24.2}$$

(\(\lambda=1\) full seller power → price ≈ the buyer's willingness to pay; \(\lambda=0\) → ≈ the seller's reservation price). Subtracting the post-transaction market price of the common benefit in the \(\alpha\) fraction, \(\alpha Y_b\), gives the control premium

$$\hat B=P-\alpha Y_b=\lambda B_b+(1-\lambda)B_s-\alpha(1-\lambda)(Y_b-Y_s) \tag{24.3}$$

\(\hat B\) = a combination of the private control benefits of both buyer and seller. Data: 393 controlling-block transactions in 39 countries 1990–2000 (>10% stock, the buyer moving from below 20% to over 20%). Results: the value of control (control premium \(\hat B\)) averages 14% of the transaction equity value (min −4% Japan, max 65% Brazil); higher where investors are less protected; private benefits have a huge effect on financial development (a 1-std increase in \(\hat B\) → a 67% reduction in external market cap/GNP, an 11% reduction in equity held by non-controlling shareholders, a 36% increase in privatized companies sold via private negotiations). Countries with lower \(\hat B\): legal mechanisms (better accounting standards, minority-shareholder legal protection, law enforcement) + extra-legal mechanisms (higher press diffusion, tax compliance, product-market competition).

Exploit the time variation of state anti-takeover laws as a source of corporate-governance variation to study takeover threat's effect on firm performance. Institutional: the federal Williams Act (1968, protecting target shareholders in tender offers); states extended it via own statutes in the 1970s (first generation); in 1982 the Supreme Court deemed the first generation unconstitutional, the second wave was upheld in 1987 → triggering a stricter third wave. The strictest are business combination laws (this study's focus): they impose a 3–5 year moratorium on specified transactions between a target and a raider holding a threshold percentage, unless the board votes in favor before the raider acts. Once passed, managers of in-state firms get a veto over takeovers and can greatly influence the board not to approve selling assets → hard for raiders (highly leveraged, planning to sell part of the target's assets to finance). Data: 1976–1995; the Census LRD plant-level data (US manufacturing) + COMPUSTAT firm-level financials. Exogeneity: the state law applies to all firms incorporated in the state (regardless of plant location), so it's reasonably exogenous to plants. Main regression

$$y_{jklt}=\alpha_t+\alpha_j+\boldsymbol\gamma'\mathbf X_{jklt}+\delta BC_{kt}+\varepsilon_{jklt} \tag{24.4}$$

\(j\) firm, \(k\) state of incorporation, \(l\) state of location, \(t\) time; \(\alpha_t\) year FE, \(\alpha_j\) firm FE; \(BC_{kt}=1\) if state \(k\) passed an anti-takeover law at \(t\); \(y\) can be log wage, TFP (a percentile from the residual of \(\ln(\text{output}_i)=\alpha+\ln(\text{wage}_i)+\beta\ln(\text{capital}_i)+\gamma\ln(\text{material}_i)+\epsilon_i\) (24.5)), or return on capital. Results: Figure 24.3 (blue-collar wages) quiet life (low takeover threat, anti-takeover law passed) → overpaying workers; Figure 24.4 (productivity TFP, profitability return on capital) → lower productivity and profitability.

Study the consequences of hedge-fund activism: a +7% abnormal return around announcement for the target, higher payout, better operating performance, higher CEO turnover after activism → hedge funds monitor firms as informed shareholders. Institutional: a proxy contest is an unfriendly fight for corporate control (outside shareholders are unsatisfied, typically aiming to change directors/managers); activists persuade shareholders to proxy for them. Incumbent self-protection: staggering boards (different election years), restrictive by-laws, poison pills. Hedge funds = new corporate activists, high leverage, aiming to sell inefficient units or fire ill-performing management; targeting value firms (low market-to-book), low payout, high CEO pay relative to comparables. Data: 13D filings 2001–2006 (LiveEdgar; §13(d) of the 1934 Act mandates: an investor acquiring >5% of a class of securities intending to affect management must file with the SEC within 10 days); 11,602 filers manually classified (banks/brokerage/regular firms/foreign institutions/individuals/insurance/pension/trusts/misc, and most importantly hedge funds/PE/VC). Results: typical objectives (Figure 24.5), typical tactics (Figures 24.6, 24.7); Figure 24.8 (returns and turnover around the 13D filing day) higher turnover and positive abnormal return near the filing (±20-day window mean 8.4%); abnormal returns differ by activism type (Figure 24.9: hedge funds intending to sell the target's assets induce the highest announcement-window abnormal return). Variables: \(\ln(MV)\) log market cap, LTDebt long-term debt/(debt + equity market value), Pre-13F dummy (prior 13F revealed >1% or 1 million USD), Avg Pre-Returns the same fund's prior-event mean abnormal return.

Gabaix-Landier (2008, §3.5.2): in equilibrium a larger company hires a more talented manager and pays a higher salary, an efficient matching. The following papers give empirical evidence on managerial compensation related to talent-size matching.

24.3.1 Frydman-Saks (2010) empirically document: the correlation between US managerial pay and firm growth is weak 1946–1975 but much stronger 1975–2005. Data: 1936–2005, the 3 highest-paid officers in the largest 50 firms (101 firms total); 1936–1991 a panel from SEC-mandated 10K + proxy statements, 1992–2005 the S&P Executive Compensation database. Results: managerial total compensation didn't increase much in the 1940s–1970s but increased dramatically after the 1970s; long-term pay (stocks/options) became increasingly important (Figure 24.10); Figure 24.11 regressions: after firm FE and time trend, the firm-size–pay correlation is weak 1946–75 but strong 1975–2005 (consistent with Gabaix-Landier 2008).

24.3.2 Kaplan-Rauh (2013): more severe US income inequality is more likely a superstar effect (increased scale + skill-biased technological change make superstars in many occupations, not just corporate managers, earn increasingly more). CEO pay (Figure 24.12, S&P 500 CEO 1993–2011 mean/median, 2010 USD, ExecuComp); top sports-player pay (Figure 24.13, top-25 baseball/basketball/football, USA Today). Combining: managers' and sports stars' pay rise together significantly in the same period → it's less likely that something special about CEO pay (greater CEO power or social-norm change) drives the super-high salary.

24.3.3 Shue-Townsend (2017) (overlaps the §17.3.4 example): study executive option grants' effect on risk-taking. Difficulty: omitted variables affect both granting and risk-taking; overcome — compensation has fixed cycles, with the instrument a dummy \(I_{ijt}^{\text{PredictedFY}}\) (the predicted first year of each cycle from the last cycle's length, not co-moving with omitted variables → exclusion; pay is adjusted in the first year → relevance). First stage \(\Delta O_{ijt}=\beta_0+\beta_1 I_{ijt}^{\text{PredictedFY}}+\gamma_t+\psi_j+\mu_{ijt}\) (24.6), second stage \(\Delta y_{ijt}=\delta_0+\delta_1\Delta\hat O_{ijt}+\gamma_t+\psi_j+e_{ijt}\) (24.7). Results: more option grants → higher stock volatility (Figure 24.14 first stage ΔLog B-S/Delta/Vega, Figure 24.15 second stage volatility); the mechanism (Figure 24.16), running 2SLS with leverage/investment as the dependent — higher option value → significantly higher leverage, no significant change in investment → option compensation makes the stock more volatile by raising the firm's leverage.

Use Denmark data to study the impact of family or external CEO appointment on firm performance. The special feature: a plausibly exogenous variation — the gender of the CEO's firstborn child — for identification. Data: 5,334 successions 1994–2002 in Danish limited-liability (public/private) firms, with financial info + personal/family info on departing/incoming CEOs (KOB financial-management, the Danish Civil Registration System, a Danish Commerce and Companies Agency match). The family-CEO effect is ambiguous ex ante: positive (a family CEO gets higher non-monetary reward from firm success, or has better firm-specific knowledge and stakeholder trust, or focuses more on the long run); negative (conflicts between family and company interests, family CEOs selected from a smaller talent pool). Main regression

$$y_i=\alpha_1+\boldsymbol\beta'\mathbf X_i+\gamma\cdot\text{famCEO}_i+\varepsilon_i \tag{24.8}$$

\(y_i\) firm \(i\)'s performance, famCEO\(_i=1\) if the incoming CEO is related by blood or marriage to the departing CEO. OLS is problematic (family-CEO appointment is not exogenous, related to performance factors unobserved in \(\varepsilon\)). 2SLS: the instrument genderfirst\(_i=1\) if the departing CEO's firstborn is male (exclusion: firm performance is orthogonal to the firstborn's gender; relevance: a CEO may count on a son to take over, getting disappointed when the firstborn isn't male and looking for an outside professional CEO, Figure 24.17). First stage

$$\text{famCEO}_i=\alpha_1+\mathbf b'\mathbf X_i+c\cdot\text{genderfirst}_i+e_i \tag{24.9}$$

second stage \(y_i=\alpha_1+\boldsymbol\beta'\mathbf X_i+\gamma\cdot\widehat{\text{famCEO}}_i+\varepsilon_i\) (24.10), reduced form \(y_i=\xi_1+\boldsymbol\psi'\mathbf X_i+\phi\cdot\text{genderfirst}_i+\epsilon_i\) (24.11). Results: the first stage is significant (Figure 24.17, a male firstborn increases family CEO; OROA = operating return on assets); the reduced form (Figure 24.18, profitability = industry- and performance-adjusted OROA); combining the first stage and reduced form gives the IV estimate — appointing a family CEO reduces profitability by 6%–8%; OLS vs IV-2SLS (Figure 24.19): IV-2SLS indicates a stronger negative effect of a family CEO than OLS.