Advanced Corporate Finance
Complete Examinable Revision

Everything examinable across Weeks 1 to 12: theory, derivations, empirical evidence, how they connect, and the critical-evaluation angles the exam rewards. Built from the lecture slides, required readings, and the official exam information document. Tip: press / to search, tick "done" on each week to track progress (saved locally), and use Print / PDF for a paper copy.

Exam Logistics & Strategy

Item	Detail
Structure	Q1-4 cover Weeks 1-6 (50 marks); Q5-8 cover Weeks 7-12 (50 marks). Questions have multiple parts and are not equally weighted.
Scope	All lecture notes and all required readings are examinable. Anything marked "Optional" is not.
Calculator	Casio FX82 (any suffix, non-programmable). The FX-8200 is explicitly NOT permitted.
Answer style	Concise, short answers preferred over rambling. Bullet points are allowed. Label any answers written in the extra blank pages.

How to prepare (their words) For each topic know: (1) the theory, (2) the empirical evidence, and (3) how they link up. For each required paper know the primary motivation, how hypotheses were tested (including how to read the tables), and how the findings relate to other papers. That third point is where marks are won: this document ends each week with explicit cross-links.

Time budget 100 marks across 8 questions. Budget roughly 1 minute per mark and bank the residual for checking. Within each half, expect one heavily theoretical question, one calculation/interpretation question (convertible mechanics, reading a regression table), and applied scenario questions. Answer the parts you know first; partial-credit structure rewards breadth.

Critical habit to practice Past exam questions in the slides (e.g. backdating risk in 2020, merger waves and cash partial-firm deals) are applied: they give you a scenario and ask which theory survives it. Memorising findings is not enough; rehearse the logic of why each test discriminates between theories, and what each test's weaknesses are (statistical power, endogeneity, omitted variables, survivorship).

WEEK 1Capital Structure: Foundations

The benchmark of irrelevance (MM), and the first friction-based theory (trade-off). Also the empirical toolkit the whole course uses.

Modigliani & Miller (1958): the irrelevance propositions

Theory In perfect capital markets (competitive, frictionless, complete; no taxes, no distress costs, no asymmetric information, investment fixed), the financing mix does not affect firm value.

V = D + E = constant, regardless of the D/E mix (Proposition I)

r_E = r_A + (D/E)(r_A − r_D) (Proposition II) Expected return on equity rises linearly with leverage; cheap debt is exactly offset by more expensive equity, so WACC is flat.

The arbitrage proof in one paragraph (be able to reproduce)

Take two firms with identical asset cash flows, one levered (L), one unlevered (U). If V_L > V_U, an investor sells shares in L, borrows on personal account in the same proportion as L's corporate debt ("homemade leverage"), and buys U. The investor replicates the levered equity payoff at lower cost, so arbitrage forces V_L = V_U. The proof only needs that investors can borrow on the same terms as firms and repackage cash flows costlessly. Titman's point (in Myers 2003): even incomplete markets do not rescue relevance if financial institutions can repackage securities competitively.

Critical point the lecturer stresses MM is not a description of reality, it is a diagnostic device: "showing what doesn't matter can also show what may matter." Every later theory in this course is built by relaxing exactly one MM assumption: taxes/distress (trade-off), information (pecking order, Myers-Majluf), agency (Jensen-Meckling, free cash flow), and market efficiency (market timing). An exam answer that names which assumption a theory relaxes signals real understanding.

Static trade-off theory

V_L = V_U + PV(interest tax shields) − PV(costs of financial distress) Optimum where the marginal tax benefit of debt equals the marginal expected distress cost. Modern versions add agency costs to both sides.

The trade-off diagram (reproduce this for any "draw and explain optimal capital structure" question). Firm value rises with debt through tax shields, then bends down as expected distress costs grow convexly.

Component	Content (Myers 2003)
Tax benefits	Interest deductibility at corporate level; Miller (1977) personal-tax offset; DeAngelo & Masulis: non-debt tax shields substitute for debt.
Direct distress costs	Legal and administrative bankruptcy costs; small relative to firm value for large firms.
Indirect distress costs	Creditor-shareholder conflicts: debt overhang / underinvestment (Myers 1977), asset substitution / risk shifting (Jensen & Meckling 1976), plus lost customers, employees, suppliers.
Agency additions	Free cash flow theory (Jensen 1986): debt disciplines managers of mature cash-rich firms. Too much equity can be a cost too.

Evidence for and against For: large safe tangible-asset firms borrow more; firms hit hard by distress are leverage-averse; target-adjustment regressions find mean reversion. Against (the famous one): the most profitable firms borrow least, the opposite of the tax-shield prediction; market-to-book effects on leverage persist for decades (Baker & Wurgler). DeAngelo (2022, Week 12 callback): 60+ years post-MM we still lack a model explaining broad-brush capital structure facts; perhaps imperfect managerial knowledge is the missing ingredient.

Empirical toolkit (recurs all semester)

Fixed effects: the slides' "hours studied vs exam grade" example. Pooled OLS can show a downward slope when, within each student-quality group, the slope is upward. Firm FE absorb time-invariant heterogeneity; identification then comes only from within-firm variation. Knowing what FE do (and what they cannot fix: time-varying omitted variables, reverse causality) is repeatedly examinable.
Event studies: short-window abnormal returns around announcements as the cleanest value-impact measure (used in W4, W6, W10, W11).
Surveys and interviews (Graham-Harvey, Bancel-Mittoo, Dong et al.): direct but vulnerable to managers rationalising or not admitting motives.
Identification concerns: endogeneity, selection, statistical power (the SSM lesson below), instruments and natural experiments.

WEEK 2Information Asymmetry

Relax the symmetric-information assumption of MM. Core paper: Myers & Majluf (1984). Then four empirical papers fight over what actually drives financing: pecking order, surveys, market timing, or persistent firm effects.

Myers & Majluf (1984): the model, step by step

Setup

Three dates. Firm has assets in place worth a (realisation of random variable A) and an investment opportunity with NPV b ≥ 0 requiring investment I, financed partly by slack S; the required issue is E = I − S. Managers know (a, b); investors only know the distributions. Managers act in the interest of old, passive shareholders. P′ is the market value of old shares if the firm issues; P if it does not.

Issue and invest iff E/(P′+E) · (E + S + a + b) ≥ S + a Equivalently: old shareholders' share of the post-issue firm must beat the value of standing pat. Rearranged: issue iff a ≤ P′(1 + b/E) − S. High-a (undervalued) firms refuse to issue.

Why stock price MUST fall on an issue announcement (the proof)

Define region M = set of (a,b) where the firm does not issue. The no-issue decision reveals a > P′(1+b/E) − S ≥ P′ − S (since b/E ≥ 0). So P = Ā(M) + S > P′: not issuing signals good news, issuing signals (relatively) bad news. Hence P′ < P whenever the issue is not a foregone conclusion. If the issue probability is 1, no information, no price drop. Both P and P′ are rational, unbiased prices that already incorporate the firm's decision rule.

Key comparative statics and the six implications (slide-level list)

The asymmetry must concern both assets-in-place and the opportunity; with no asymmetry about a, stock is always issued for positive-NPV projects and investment is efficient.
Firms can pass up positive-NPV projects: real underinvestment, value loss L.
Slack is valuable: build it via retentions, or issue when the information gap is small. Don't pay dividends you must recoup with risky issues.
Safe (default-risk-free) debt avoids the price fall; risky securities fall in between. This generates the pecking order: internal funds, then debt, then equity as last resort. Matches Dann-Mikkelson: negative announcement returns for equity, roughly zero for straight debt.
Mergers: a slack-rich firm buying a slack-poor firm creates value, but negotiation is hard because the slack-poor managers cannot credibly convey their private information; expect tender offers instead.
Price falls on issue even though issuing is in old shareholders' interest at that point.

Critical evaluation The result hinges on two contestable assumptions: managers serve old shareholders only, and those shareholders are passive. With "active" shareholders who rebalance, an MM-style irrelevance returns and only the investment decision conveys news. The model also has no equilibrium issue strategy under permanent asymmetry, and small growth firms empirically issue equity happily with small price drops, which Myers (2003) reconciles by noting the damaging asymmetry attaches to assets in place, not growth options. Jung, Kim & Stulz find low-growth firms issuing equity despite debt capacity: direct violation.

The pecking order ladder. Securities are issued in order of information sensitivity, because the Myers-Majluf discount grows with how exposed the security's value is to the manager's private information. Convertibles (W3) sit between rungs 2 and 3, which is exactly Stein's point.

Myers-Majluf decision regions. For a given issue size E and slack S, the boundary line splits (a, b) space. Firms above/right of the line refuse to issue: good firms pass up positive-NPV projects (the deadweight loss), and the no-issue decision itself signals high a, which is why P > P′. More slack shifts the boundary right and shrinks the loss region.

30-second recall: Week 2 Mechanism: issuing risky securities signals overvaluation, so good firms underinvest and prices fall ~2-3% on equity announcements. First-order fix: slack and the pecking order. Tests: SSM β_PO≈0.75 with the power-simulation twist; Frank-Goyal degrade it; Graham-Harvey say flexibility/ratings; Baker-Wurgler say timing persists; LRZ say a fixed effect beats everyone, including before the IPO.

Shyam-Sunder & Myers (1999): testing pecking order vs trade-off

DEF_t = DIV_t + CAPEX_t + ΔWC_t − CF_t ; ΔD_t = α + β_PO·DEF_t + ε Pecking order prediction: α = 0, β_PO = 1 (every dollar of deficit met with debt). Trade-off prediction: target adjustment ΔD = a + b_TA(D* − D_t-1).

Findings (157 firms, 1971-1989) β_PO ≈ 0.75-0.85, α ≈ 0, R² ≈ 0.68-0.75. Both models "fit" on real data. The clever part is the power simulation: on data simulated under pure pecking-order behaviour, the target-adjustment model still fits (false positive); on data simulated under pure target adjustment, the pecking-order model is rejected (β near 0). Conclusion: the pecking-order test has power, the trade-off test does not, so the pecking order is the better first-order description.

Why the trade-off test is weak (mechanism) Pecking-order behaviour mechanically generates mean-reverting debt ratios when investment is lumpy and cash flow is cyclical (deficit years raise D, surplus years lower it), so mean reversion cannot discriminate. General lesson for the whole course: a significant coefficient supports a theory only against the alternatives it can reject. Limits: small, survivorship-tilted sample of large mature firms; Frank & Goyal show β_PO collapses (to ~0.3 by the 1990s) in larger samples and fails worst for small growth firms, exactly where asymmetric information should be largest; Chirinko & Singha show the test lacks power against other plausible rules; Lemmon-Zender: debt capacity rescues a modified pecking order.

Graham & Harvey (2001): what CFOs say they do

What ranks high	What ranks low (theory casualties)
Financial flexibility and credit ratings for debt policy; EPS dilution and recent stock price appreciation for equity; ~75% use NPV/IRR; CAPM dominant for cost of equity.	Using debt to discipline managers (Jensen): mean 0.33, second lowest. Product-market threats (Brander-Lewis): 0.40. Customer concern about uniqueness (Titman): 1.24. Signalling, asset substitution, personal taxes: weak.

Critical evaluation Moderate support for trade-off (targets exist, loosely) and for informal pecking-order behaviour, but the stated reasons (flexibility, timing windows) do not map cleanly onto Myers-Majluf's mechanism. Survey caveats the paper itself concedes: managers may not admit unflattering motives (discipline by debt), may rationalise, and the response rate (~9%) raises selection concerns. Use surveys as a complement to, not substitute for, archival evidence; Dong et al.'s interviews (Week 3) push the same point further.

Baker & Wurgler (2002): market timing and capital structure

Idea + measure Capital structure is the cumulative outcome of past attempts to time the equity market. Key variable: "external finance weighted-average" market-to-book, M/B_efwa, which weights past M/B by the amounts of external finance raised. Prediction: firms that raised funds when valuations were high stay persistently low-leverage.

Findings M/B_efwa strongly negatively predicts leverage, and the effect persists for a decade or more after IPO; it dominates current M/B and standard trade-off determinants. The leverage change decomposes mostly into net equity issues made in high-valuation periods, not retained earnings or growth.

Critical evaluation Trade-off can only explain persistence with implausibly large adjustment costs; under trade-off, leverage should re-adjust after temporary M/B shocks but does not. But M/B is a noisy bundle: it proxies growth options (distress-cost channel) and misvaluation simultaneously, so interpretation is contested. And Lemmon, Roberts & Zender (next) show much of the persistence pre-dates the IPO, which timing cannot explain.

Lemmon, Roberts & Zender (2008): "Back to the Beginning"

Findings Variance decomposition of leverage: the firm fixed effect (a permanent, unexplained component) explains the majority of variation; the traditional determinants (size, profitability, M/B, tangibility) explain little once FE are included. Leverage portfolios converge slowly but never cross: High stays above Low for 20+ years, and crucially the ordering already exists before the IPO (shown with unexpected-leverage portfolios in IPO event time and UK private-firm data).

Implication, stated bluntly All three workhorse theories (trade-off, pecking order, market timing) are built on time-varying firm characteristics, yet the dominant source of leverage variation is a time-invariant firm effect none of them speaks to. This is the empirical anchor for DeAngelo's "are we there yet?" pessimism. Counterpoint to keep balance: stable leverage could still reflect stable unobserved fundamentals (technology, asset risk) that a richer trade-off model would capture; FE results identify our ignorance, not a specific alternative theory.

Week 2 linking sentence (memorise the structure) Myers-Majluf supplies the mechanism, SSM gives it first-order empirical support but teaches the power lesson, Graham-Harvey shows managers' stated motives only partially match, Baker-Wurgler reinterprets the same M/B facts as timing, and LRZ shows that the bulk of leverage variation is explained by none of the above. A top answer can run that arc in four sentences.

WEEK 3Security Design: Convertible Securities

Why does a security that is "debt plus an equity option" exist at all? Mechanics first (calculations are fair game), then the Big Four theories, then the modern hedge-fund-driven market.

Mechanics and formulas

Term	Definition / formula
Conversion ratio (rate)	Number of shares per bond (e.g. Disney 2003: 34 shares per $1,000 bond).
Conversion price	Principal ÷ conversion ratio. Disney: 1000/34 = $29.41.
Conversion value	Conversion ratio × current stock price. Convert at maturity iff conversion value > principal (at S=$25: 34×25=$850, take the $1,000; at S=$35: $1,190, convert).
Conversion premium	(Conversion price ÷ stock price at issue) − 1. Practice Q: 50.02/36.03 − 1 = 38.8%. Higher premium = more debt-like (lower conversion probability).
Call feature	Firm may redeem at the call price (near par); holders then choose max(conversion value, call price). Call protection = period when calling is prohibited.
Delta	e^−δTN(d₁): sensitivity to the stock; the issue-date conversion probability (>50% = equity-like, <50% = debt-like) is the standard classification (Lewis et al. 1999).

Convertible value vs stock price. The hockey-stick: bond floor protects on the downside, conversion option captures the upside. The gap between fair value and max(floor, conversion value) is option time value. This asymmetry is exactly why convertibles are insensitive to disagreements about firm risk (Brennan-Kraus/Schwartz).

The Big Four rationales (and who tests them)

Theory	Mechanism	Predicted design	Evidence verdict
Green (1984) risk shifting	Conversion option lets bondholders share the upside, killing shareholders' incentive to substitute into riskier assets after issuing debt.	Debt-like	Supported in quantitative work (Lewis et al. 1998, 1999: debt-like issuers have high risk-shifting potential), but systematically rejected by every survey/interview study.
Brennan-Kraus (1987), Brennan-Schwartz (1988) risk uncertainty	If the market overestimates risk, straight debt is undervalued but the conversion option is overvalued; the convertible's value is risk-insensitive, so pricing agreement is easier.	Debt-like (mid)	Mixed; weak in design studies, but Dong et al.'s interviews find it the strongest motive.
Stein (1992) backdoor equity	Firm wants equity but faces Myers-Majluf adverse-selection costs; issues a callable convertible and later forces conversion. Needs (1) meaningful distress costs (so bad firms can't mimic) and (2) call provisions.	Equity-like, callable	Strongest in surveys (Graham-Harvey, Bancel-Mittoo) and in design evidence for equity-like convertibles.
Mayers (1998) sequential financing	Convertible matches staged real options: if the future project is good, call/convert wipes the debt and frees capacity for new financing; if bad, debt stays and disciplines (avoids free-cash-flow waste and re-issue transaction costs).	Debt-like, callable	Mayers: investment and new long-term financing spike after conversion-forcing calls; Chang et al. consistent. Surveys: mixed to none.

The honest overall verdict (Dutordoir et al. 2014) No clear winner. The empirical pattern is messy: results differ by geography (US convertibles are equity-like; European ones debt-like, "sweetened debt"), by method (qualitative work kills Green; quantitative work supports it), and across issuer subsamples. Two interpretations: each theory is incomplete, or issuers are heterogeneous and different firms issue for different reasons. The "free lunch" pitch (cheap coupon plus issuing equity at a premium) is a fallacy: each component is priced fairly; managers in Dong et al. appear aware of true costs.

Call policy: theory vs practice

Textbook (Ingersoll/Brennan-Schwartz): call immediately when conversion value reaches the call price, to cap the value handed to bondholders.
Practice: many firms delay. Candidate reasons (from the OSI practice question): expected dividends (holders convert voluntarily), avoiding a failed call (price drops below trigger during the notice period, leaving the firm to fund a cash redemption: firms wait for a safety cushion, e.g. call at 18.7% above par rather than 12.5%), signalling confidence, or plain distraction (a takeover bid). Note the discipline of the answer: each explanation is tested against the observed delay length and firm history before being accepted or discarded. That style of reasoning is what the exam wants.
Average delays have shrunk since the 1980s, so "theory fails in practice" is overstated once notice periods and failed-call risk are priced in.

Brown, Grundy, Lewis & Verwijmeren (2012): convertibles and hedge funds

Equity rationing / investor demand rationale Modern (post-2000, mostly Rule 144A) convertibles are largely bought by convertible arbitrage hedge funds: buy the convertible, short delta shares. The firm effectively issues equity synthetically: the arbitrageurs' shorting supplies the share sales a seasoned offering would have required, faster and cheaper, for firms rationed out of the SEO market (high seasoned-equity issue costs).

Evidence Issuers sold to hedge funds look like would-be equity issuers with high SEO costs and stock that is attractive to short (liquid, available borrow). Short interest in the issuer spikes at issuance. Institutional ownership rises afterwards, consistent with anticipated, information-free selling pressure. Convertibles placed with hedge funds have short effective lives. This demand-side story also rationalises the long-run underperformance findings (Lewis et al. 2001) that embarrass the classic supply-side theories.

Linking sentence Week 3 is Myers-Majluf applied: Stein is literally "Myers-Majluf plus a call option," and the hedge-fund channel is the market's institutional fix for the same adverse-selection problem. If a question asks "why convertibles rather than equity," answer in adverse-selection language and then critically note the heterogeneity verdict.

WEEK 4Mergers & Acquisitions

Facts (Andrade, Mitchell & Stafford 2001), the merger-wave horse race (Harford 2005), and whether acquirer skill exists (Golubov, Yawson & Zhang 2015).

Stylised facts: Andrade, Mitchell & Stafford (2001)

Merger activity comes in waves (1960s conglomerate, 1980s hostile/bust-up, 1990s stock-financed strategic) and clusters by industry; deregulation is a major driver in the 1990s (deregulated industries account for a large share of deal value).
Announcement returns (3-day): targets +16% (about +24% through completion), acquirers roughly 0 to slightly negative (-0.7%, not reliably ≠ 0), combined +1.8% to +2%: mergers create some value, targets capture it.
Stock-financed deals have systematically worse acquirer returns (consistent with Myers-Majluf: paying with stock signals overvaluation); ~70% of 1990s deals involved stock.
Operating performance: combined abnormal operating margins improve ~1% post-merger, consistent with the positive combined announcement return.
Long-run post-event returns are negative for some samples but the authors are skeptical: long-horizon tests are statistically fragile (benchmark sensitivity), so the short-window event study is the most reliable evidence.

The headline distribution of merger gains. Mergers create modest combined value; targets capture essentially all of it. Through completion, target returns reach ~+24%.

Three US merger waves. Each wave has a distinct character: useful colour for any waves question, and the deregulation point feeds straight into Harford's shock evidence.

Merger waves: behavioral vs neoclassical (Harford 2005)

Behavioral (Shleifer-Vishny, Rhodes-Kropf et al.)Managers use overvalued stock to buy less-overvalued assets. Predictions: waves follow high returns with high dispersion; method of payment is stock; post-wave industry returns poor.

Neoclassical + liquidity (Harford)Industries reallocate assets after economic/regulatory/technological shocks, but only when capital liquidity is high enough to fund the reallocation. Predictions: shocks precede waves; cash and partial-firm (divisional) acquisitions also cluster in waves.

Harford's horse race Logit models predicting industry-wave starts: economic shock measures predict waves only when capital liquidity is high; a deregulation index plus liquidity adds sharply; M/B and behavioral variables lose significance once shocks and liquidity are included. Killer test: cash-financed partial-firm acquisitions move with stock-swap merger activity, which the behavioral story cannot generate (there is no "overvalued currency" motive for paying cash for a division). Aggregate waves are just clustered industry waves.

Known exam question (from slides) "Partial-firm acquisitions paid with cash are more common during merger waves. Which wave explanation does this support?" Answer: the neoclassical/liquidity explanation, because efficient reallocation after shocks implies transactions of any form (whole-firm or divisional, stock or cash), whereas the behavioral story requires stock as currency and whole-firm targets. Then add the critical caveat: this evidence does not prove misvaluation never matters; it shows misvaluation is not the primary cause of clustering.

Why do acquirers earn ~0? Candidate explanations

Competition in the market for corporate control: targets capture synergies via premiums (bidding competition).
Size effect: deal NPV is small relative to acquirer market cap; news is diluted.
Signal contamination: announcement bundles deal news with revelation about the acquirer (e.g. stock payment reveals overvaluation; "running out of internal growth").
Agency/hubris: empire-building or overconfident CEOs overpay (Roll's hubris hypothesis; free cash flow).

Golubov, Yawson & Zhang (2015): "Extraordinary acquirers"

Findings Acquirer fixed effects explain as much (sometimes more) of the variation in acquirer announcement CARs as all major firm- and deal-level characteristics combined. The 25th-to-75th percentile spread in acquirer FE is over 6 percentage points (~$184m for the mean acquirer). Returns are persistent: good acquirers keep making good deals. Established determinants (size, payment method) remain significant within-firm, but a large firm-specific, time-invariant component sits on top.

Critical reading A fixed effect is a measure of our ignorance with a label on it: "skill" is one interpretation (consistent with Kaplan-Schoar PE persistence, dedicated deal teams, Buffett), but FE also absorb any stable firm attribute (industry niche, deal pipeline type, governance). Note the methodological echo of Lemmon-Roberts-Zender in Week 2: in both literatures, fixed effects dwarf the theories' favourite covariates. That parallel is an excellent exam linkage.

WEEK 5Corporate Governance Around the World

Governance as the answer to "why do outside investors ever get their money back?" Cross-country variation in law shapes ownership and financial development.

La Porta, Lopez-de-Silanes, Shleifer & Vishny (1998), "Law and Finance"

Setup Securities are not just cash flows (the MM view); they are bundles of rights, and rights are only worth what the law and its enforcement make them. Laws are transplanted from a few legal families: common law (English) and civil law (French, German, Scandinavian). Legal origin is plausibly exogenous to modern firm outcomes, which is what makes it usable for identification.

Findings across 49 countries

Shareholder rights ("antidirector rights") and creditor rights: common law strongest, French civil law weakest, German/Scandinavian in between.
Enforcement quality and accounting standards: best in Scandinavian/German law countries and rich countries generally; enforcement (unlike rules) rises with income; weakest in French-civil-law countries.
Ownership concentration: extremely high worldwide (top-3 shareholders hold ~half the average large firm) and negatively related to investor protection: concentration is an adaptive substitute for weak law. Other substitutes: mandatory dividends, legal reserve requirements (more common in civil law).
Consequences via companion papers: weak protection means smaller debt and equity markets (LLSV 1997), slower growth in finance-dependent sectors (Rajan-Zingales 1998), and few widely-held firms outside strong-protection countries (most are family or state controlled).

Critical evaluation Causality leans on legal origin being exogenous, but origin travels with colonial history, culture, and politics; "the only truly exogenous variable is legal origin" is the authors' own concession. Coding legal indices involves judgment; later work (Spamann's recoding) weakened some antidirector-rights results. And rich counterexamples exist: France and Belgium are rich despite "bad" law, so weak protection is a friction, not a wall. Still, the substitution result (concentration where law is weak) is robust and ties straight into Week 7's blockholder theory.

Denis & McConnell (2003), International Corporate Governance

Generations of research: first-generation work studied internal mechanisms (boards, ownership, compensation) country by country; second-generation work, post-LLSV, studies how the legal system shapes which mechanisms exist and work.
Mechanisms taxonomy worth reproducing: internal (board, large shareholders, managerial ownership, compensation, capital structure) vs external (takeover market, legal/regulatory system, product market competition).
Cross-country evidence: board effectiveness, blockholder roles, and turnover-performance sensitivity vary with the institutional environment (e.g. Kaplan on Germany/Japan: turnover responds to poor performance there too); controlling shareholders create a different agency problem: controller vs minority expropriation (tunnelling, pyramids, dual-class wedges) rather than manager vs dispersed shareholders.

Linking sentence LLSV explains why the Berle-Means dispersed firm is a US/UK peculiarity; Denis-McConnell maps which governance tools substitute when law is weak; Week 6 then zooms into one internal mechanism (the board), and Week 7 into another (blockholders).

30-second recall: Week 5 LLSV: rights come from law; common law protects most, French civil law least; enforcement rises with income; ownership concentration substitutes for weak protection; weak law means small capital markets. Denis-McConnell: internal vs external mechanisms; with controlling owners the agency problem rotates 90 degrees, from manager-vs-shareholders to controller-vs-minority.

WEEK 6Boards of Directors

Does board structure matter for value, what do directors actually bring, and one striking governance failure: option backdating.

Yermack (1996): small boards, higher value

Design and findings 452 large US firms, 1984-1991 (mean board ~12.25 members). Inverse relation between board size and Tobin's Q, in cross-section and within firm; robust to controls for size, industry, ownership, growth opportunities, governance structure. The relation is steepest moving from small to medium boards (concave). Supporting evidence: profitability/efficiency ratios decline with board size; CEO pay-performance sensitivity and dismissal threat weaken with board size; event study: announcements of board reductions (≥4 members) earn positive abnormal returns, expansions negative.

Mechanism (Lipton-Lorsch, Jensen) Beyond 7-8 members, coordination and free-rider problems dominate any extra monitoring capacity: slower decisions, politeness over candour, easier CEO control.

Critical evaluation Endogeneity is the obvious worry: maybe troubled firms enlarge boards. Yermack tests reverse causality and finds board size insensitive to past performance, but board size is still a choice; optimal size plausibly varies with firm complexity (later work: complex/advice-intensive firms benefit from larger boards), so "small is good" is not a universal prescription. The event-study sample is only ~10 clean events: indicative, not decisive.

Adams, Akyol & Verwijmeren (2018): Director Skill Sets

Findings Exploits Regulation S-K disclosure of each director's skills. Directors are multidimensional (skills come bundled), and the main dimension along which boards differ is skill diversity. Surprising result: boards with more commonality (common ground) in skills are associated with better firm performance, not more diverse ones. Interpretation: shared ground aids communication and coordination; firms face a constrained, multidimensional optimisation that one-attribute regulation (independence quotas, single-trait diversity mandates) can distort.

Critical note Skill disclosures are self-reported marketing as much as measurement, and performance regressions remain associational. The paper's value is the reframing: stop scoring boards on one attribute at a time.

Lie (2005): option backdating

The smoking gun Stock returns are abnormally negative before and abnormally positive after CEO option grant dates, and (the decisive part) the post-grant pattern includes market-wide movements, which no legitimate grant-timing story can produce: executives cannot forecast the market. Inference: grant dates were chosen retrospectively to coincide with low prices (backdating). Heron & Lie (2007): the pattern weakens sharply after SOX required grants to be reported within 2 days: regulation closed the window.

Known exam question (from slides) "Holding regulation fixed, would backdating risk be higher in 2020 (volatile) than in calm years, and how would you test it?" Logic: backdating's payoff rises with volatility (deeper troughs to pick), so incentive is higher in 2020; test by comparing post-grant abnormal (incl. market) return patterns for grants reported late vs on time, across high- and low-volatility periods, or compare V-shaped price patterns around grant dates in 2020 vs calm years. Who loses: shareholders (stealth pay, misreported compensation, tax/accounting violations), and trust in disclosure.

30-second recall: Week 6 Yermack: board size inversely related to Q, robust, plus supportive event study; endogeneity caveats. AAV 2018: directors are skill bundles; common ground beats diversity for performance; one-attribute regulation distorts. Lie: post-grant market-wide returns prove retrospective dating; SOX 2-day reporting killed it.

WEEK 7Blockholders and Large Investors

If law gives small investors weak control, size is the substitute. Modern twist: a handful of institutions now hold decisive stakes in most US firms.

Concepts

Why large investors: concentrated control rights plus a large cash-flow stake make intervention pay; dispersed holders free-ride (Grossman-Hart). Two governance channels: voice (intervention, engagement, voting) and exit (selling, pushing the price down, disciplining managers whose wealth is price-sensitive).
Costs of large investors: under-diversification, potential expropriation of minorities (links to W5), conflicts of interest (institutions manage the firm's pension assets and may vote with management), and herding/short-termism: institutional trading raises volatility (Dennis-Strickland; Bushee-Noe: transient institutions associate with higher volatility, dedicated long-horizon ones with lower).
Stylised facts: ownership of US equities has migrated to institutions; the top-five investors' combined stakes have risen so far that the dispersed-ownership assumption underlying classical theory now fails for most US public firms; how multiple blockholders interact (complementary monitoring vs free-riding) is an open question.

Required papers

Paper	Question	Finding	Critical note
Gompers & Metrick (2001)	Do institutions move prices?	Institutional ownership nearly doubled 1980-1996; institutions prefer large, liquid stocks; demand shifts from institutional growth can explain a large part of the rise in large-stock prices relative to small (compositional demand effect, not necessarily information).	Demand-based price effects challenge frictionless asset pricing; but preferences are estimated from holdings, so "preference" and "information" are hard to separate.
Yan & Zhang (2009)	Are institutions informed?	Split institutions by churn: short-term institutions' trades predict future returns and earnings surprises; long-term institutions' trades do not. Informed trading is concentrated in high-turnover investors.	Cuts against blanket "institutions are smart money" and against blanket "short-termism is bad": short horizons and information production coexist.
Bharath, Jayaraman & Nagar (2013)	Is exit a real governance force?	Uses stock liquidity shocks (decimalisation etc.) as natural experiments: when liquidity rises, exit threats become more credible, and firm value rises more for firms with more blockholders, especially when managerial wealth is tied to the stock price.	Clean identification of the threat channel: governance can work without any visible activism. Caveat: liquidity shocks affect much else (information environment), so the exclusion restriction is arguable.

Linking sentence W5 said concentration substitutes for weak law; W7 shows the mechanisms (voice and exit) and prices the trade-offs. Bharath et al. is the empirical twin of Edmans' exit theory; Yan-Zhang qualifies which institutions carry information.

30-second recall: Week 7 Theory: size substitutes for legal protection; voice and exit channels; costs are under-diversification, conflicts, herding. Gompers-Metrick: institutional demand shifts move relative prices. Yan-Zhang: only short-horizon institutions trade on information. Bharath et al.: liquidity shocks show the exit threat alone raises value where blockholders and price-linked pay coexist.

WEEK 8M&A: Additional Topics

Three modern papers, each attacking a different stage of the deal: the announcement text, the payment choice, and what failure reveals.

Filipovic & Wagner (2025): "The intangibles song"

Findings Text-analyse takeover announcements for intangibles talk (synergy, brand, culture, know-how language). More intangibles talk: −0.53pp three-day acquirer CAR, worse industry-adjusted performance over the next year, more likely cash payment, faster deal completion, managers buy stock. Interpretation: deal-specific managerial overoptimism, not agency: a rational manager uncertain about intangible value should pay with stock (contingent pricing shares overpayment risk with target shareholders); paying cash means the manager perceives little downside.

Dutordoir, Strong & Sun (2022): short-selling potential and payment choice

Mechanism: merger arbitrage On announcement, the target trades at a discount to the offer (the arbitrage spread, compensating for time value and failure risk). Arbitrageurs buy the target; in stock deals they also short the bidder to hedge, and with downward-sloping demand curves this pressures the bidder's price. In cash deals there is no hedge and no pressure (slides: short interest in the acquirer jumps around share offers, flat for cash offers).

Findings Bidders whose stock is easier to short (high short-selling potential) use more cash, but only for public targets (arbitrageurs cannot buy private targets, so no pressure to pre-empt: a built-in placebo). Robust to overvaluation and misvaluation controls. Adds market microstructure to the standard payment-choice list (asymmetric information, taxes, control).

Malmendier, Opp & Saidi (2016): target revaluation after failed takeovers

Findings After bids fail: targets of cash bids stay revalued upward (~15%); targets of stock bids return to pre-announcement values. Interpretation: cash bids reveal positive bidder information about the target's standalone value (you only pay certain cash for value you believe in); stock bids do not carry that signal. Failed deals are the clean laboratory because no synergies are realised, isolating the information content of the bid itself.

Critical thread across the three papers All three exploit the idea that payment method is information, but about different things: bidder overoptimism (cash + intangibles talk), anticipated arbitrage pressure (cash to dodge shorting), and target value revelation (cash bids revalue targets). An exam question that asks "what does cash vs stock tell us?" should be answered with this three-way decomposition plus the classic Myers-Majluf overvaluation signal from Week 4. Note they partly pull in different directions, which is precisely why payment choice remains a live research area: be willing to say the literature is not settled.

WEEK 9Private Firms

Most firms are private, and increasingly large ones choose to stay private. The public/private margin is a trade-off across liquidity, disclosure, agency, and issuance costs.

The landscape (Lowry 2024; Ewens & Farre-Mensa 2022)

The public market's share of the total firm population has always been tiny; the compositional news is that large, well-capitalised firms now elect to remain private.
Non-traditional investors (PE funds, corporations, sovereign wealth, mutual and hedge funds) supplied ~25% of late-stage private capital in 2002 and ~73% by 2019: private-market investors increasingly resemble public-market investors in incentives and benchmarks, eroding the valuation distinction between staying private and listing.

The public vs private trade-off

Dimension	Public	Private
Liquidity	Exchange-traded; broad diversified investor base.	Illiquid: estimated 15-30% valuation discount in acquisition markets; 26-60bp higher loan spreads.
Issuance costs	IPO: ~7% underwriting fee plus underpricing (first-day returns ~15-20% avg); SEO: ~4.5% fee, ~−2% announcement return.	PE fees ~2/20, ongoing and relationship-embedded, hard to compare directly.
Agency	Separation of ownership/control (Jensen-Meckling vertical agency); short-termism risk; but liquid prices aid monitoring (Holmström-Tirole) and enable stock-based pay.	Horizontal agency: controlling vs minority shareholders, especially under weak property rights. Net differential is empirically unresolved.
Disclosure	Mandatory, costly, reveals to competitors.	Opaque: higher cost of debt (Badertscher et al.); disclosure mandates change M&A activity (Ortiz et al.).

Required papers

Paper	Finding	Critical note
Gogineni, Linn & Yadav (2022)	Within private firms, both vertical (owner-manager) and horizontal (majority-minority) agency problems hurt operating performance; performance is best when ownership and control are aligned (e.g. 100% owner-managed), and deteriorates with ownership dispersion and outside managers.	Rare large-sample look inside private firms, but ownership structure is itself chosen, so causality is murky.
Ortiz et al. (2023)	Mandatory financial disclosure by private firms increases M&A activity: disclosure lowers acquirers' information costs of identifying targets.	Nice causal design from regulation thresholds; effect is on deal matching, not necessarily deal quality.
Badertscher et al. (2019)	Private (vs public) ownership raises the cost of public debt: bond investors price the weaker information environment, even holding the security type fixed.	Identification leans on comparing bond issuers; private firms issuing public bonds are a selected group.

Linking sentence This week is the agency (W5-W7) and information (W2) toolkits applied to the listing decision. The Ewens et al. evidence that public-firm agency proxies exceed private comparables is a useful counterweight to the reflex that "public = better governed."

30-second recall: Week 9 Private firms dominate by count and increasingly by quality; non-traditional investors fund ~73% of late-stage capital by 2019. Trade-off: liquidity (15-30% private discount, 26-60bp loan spreads) and issuance costs vs disclosure and horizontal agency. Public-firm agency proxies are not obviously lower. Papers: vertical and horizontal agency hurt performance (GLY); disclosure mandates boost M&A matching (Ortiz); private status raises public-bond costs (Badertscher).

WEEK 10Investment Banks

Banks as reputational intermediaries: do top-tier advisors deliver premium quality for a premium price, and are their valuation methods (comps) strategically bent?

Theory

Reputation equilibrium (Klein-Leffler 1981; Chemmanur-Fulghieri 1994) Quality is unobservable ex ante; reputations reveal it. A premium price, premium quality equilibrium compensates banks for the sunk cost of reputation building and deters fly-by-night behaviour. Banks are sorted into top-tier (bulge bracket) vs non-top-tier via league tables. Banks are both brokers (matching, lowering search costs) and underwriters (absorbing placement and credit risk), so relationship banking concepts carry over.

Evidence

Fees and reputation: the raw correlation between top-tier advisors and fees is negative, but this is a deal-size artifact (top banks do bigger deals; percentage fees fall with size). Controlling for ln(deal value), top-tier advisors earn a robust fee premium, consistent with the reputation equilibrium. Slide-flagged trap: never interpret the raw correlation.
Do they deliver? Top-tier advisors are associated with better acquirer outcomes in public deals (where reputational exposure is greatest), supporting "you get what you pay for" in the visible segment.
Comparable company analysis incentives: banks advising targets may pick high-multiple peers (justify a high price); deal-completion-contingent fees push toward peer sets that make the offer look generous so the deal closes. Evidence indicates strategic peer selection in fairness opinions: comps are an advocacy document, not a neutral valuation.
Relationships and information: takeover likelihood rises when acquirer and target share a bank (relation intensity is a strong predictor), raising conflicts: banks can transmit information across clients.

Required papers

Paper	Question	Finding	Critical note
Golubov, Petmezas & Travlos (2012)	Do top-tier advisors deliver?	Top-tier advisors earn a robust fee premium (masked in raw data by deal size: top banks do bigger deals at lower percentage fees) and deliver higher bidder returns in public acquisitions — the premium-price, premium-quality reputation equilibrium in the data.	Supports reputation theory, but top banks self-select into deals likely to succeed; the quality effect is concentrated in public targets where reputation is most exposed.
Eaton, Guo, Liu & Officer (2022)	Are valuation "comps" neutral?	In fairness opinions, target advisors strategically select higher-valuation comparable firms; deal-completion-contingent fees bias the peer set toward justifying a high offer price. Comparable-company analysis is an advocacy document, not a neutral valuation.	Direct evidence that intermediaries are conflicted certifiers; the strategic-selection read is hard to fully separate from genuine differences in which peers are truly comparable.
Ivashina, Nair, Saunders, Massoud & Stover (2009)	Do banks shape the takeover market?	Bank lending intensity and the size of a bank's client network raise the probability a borrower becomes a target and raise completion rates; the effect is strongest when bidder and target share a bank — banks transmit borrower information to potential acquirers.	Identifies an information-conduit channel (governance via the credit relationship), but it raises client-conflict concerns; identified off unsolicited takeovers 1992–2003.

Critical evaluation The fee-premium evidence supports reputation theory but selection lurks everywhere: top banks choose deals likely to succeed. The comps and shared-bank evidence are reminders that intermediaries are conflicted agents, not neutral certifiers, which connects to Week 6 (governance of agents) and Week 11 (issuance costs).

30-second recall: Week 10 Reputation theory predicts premium price, premium quality. Raw negative fee correlation is a deal-size artifact; controlled, top-tier banks earn a fee premium. Comps are bent by advocacy and completion-contingent fees; shared banks predict takeovers (information conduits). Banks are conflicted certifiers.

WEEK 11Raising Capital

SEOs as the test bed for every capital structure theory at once: pecking order, trade-off, timing, lifecycle, and behavioral anchoring.

Institutional menu and announcement effects

Methods: firm-commitment SEOs, rights offerings, private placements, shelf offerings. US SEO announcement returns ~−2% (the Myers-Majluf prediction in the data).
Shareholder approval matters internationally: rights-offer announcement returns are negative where managers can act unilaterally (Australia −3.53%, Netherlands −2.17%) but positive where shareholder approval is required (Finland +4.29%, Singapore +3.69%); same split for private placements (Sweden +7.27%, India +6.18% with approval vs Netherlands −0.52% without). Governance moderates the adverse-selection discount: approval acts as certification. Strong W5 link.

What drives the SEO decision?

Horse race: fundamentals vs timing vs lifecycle A near-term financing need is the primary driver of SEOs; market-timing variables and corporate lifecycle stage (years listed) are statistically significant but economically secondary, with lifecycle dominating timing across specifications. Timing effects are real but small in absolute terms because issuance is rare. This disciplines Baker-Wurgler (W2): timing exists, but it is not the first-order motive for any given SEO.

Anchoring SEOs The most recent equity offer price serves as a salient reference point: firms are more likely to issue when the price is above it. Instrumented anchoring-driven SEOs raise cash holdings and acquisition activity (+2.1-3.7pp likelihood, +8.7-11.3% spending) and those acquisitions are inefficient; capex, R&D, and employment do not move. So behaviorally-timed issues fund empire-ish deals, not productive investment: a free-cash-flow story (Jensen, W1) born of a behavioral trigger.

Required papers

Paper	Question	Finding	Critical note
DeAngelo, DeAngelo & Stulz (2010)	What really drives an SEO?	A near-term cash need is the first-order motive — most issuers would run low on cash without the SEO. Market-timing and corporate-lifecycle stage are statistically significant but economically secondary, with lifecycle dominating timing.	Disciplines Baker–Wurgler (W2): timing is real but second-order. The "need" is reconstructed ex post from pro-forma cash, so the test is somewhat mechanical.
Dittmar, Duchin & Zhang (2020)	Is SEO timing behavioral?	SEO likelihood jumps discontinuously when the price reaches the most recent equity offer price (a salient anchor). A fuzzy RDD shows anchored SEOs raise cash and fund lower-quality acquisitions (+2.1–3.7pp likelihood, +8.7–11.3% spending); no effect on capex, R&D, or employment.	Clean local-randomization identification of a behavioral trigger; anchoring → free-cash-flow misallocation. External validity is local to firms near the cutoff.
Mota & Siani (2025)	How do firms manage the debt investor base?	Firms trade off minimizing cost of capital against diversifying their investor base when choosing which bonds to issue; investor specialization in bond characteristics lets them shape bondholder composition through issuance. Greater bondholder diversification raises resilience to credit shocks, and firms time the market both to cut costs and to diversify credit supply.	Reframes "raising capital" around the investor-base margin and the rise of non-bank intermediaries; built on new firm–bond matched data, but the relationships are largely associational.

Linking sentence W11 is the adjudication week: the same firm decision (issue equity?) is fought over by W2's four theories, and the verdict is "fundamentals first, lifecycle second, timing third, with behavioral anchoring producing real misallocation." Bring the SEO announcement-return evidence as the standing confirmation of Myers-Majluf.

30-second recall: Week 11 SEO announcements ~−2% (Myers-Majluf in the wild); shareholder approval flips the sign internationally (governance certifies). Drivers: financing need first, lifecycle second, timing third. Anchored SEOs raise cash and fund inefficient acquisitions, not capex/R&D: behavioral trigger, free-cash-flow consequence.

WEEK 12AI and Finance

Two questions: what does AI do to firms and finance jobs, and what can LLMs do for finance research?

Labour demand: ChatGPT's launch (30 Nov 2022) is a structural break in finance-sector job postings; AI-skill demand roughly doubled by 2024 and generative-AI/LLM skills emerged from zero; general AI skills dominate, deep learning stays modest.
Firm value: portfolios sorted on generative-AI exposure tracked together pre-ChatGPT; in the event window the high-exposure ("Artificial") portfolio jumped and the gap never closed, reaching ~8-10% abnormal return by GPT-4's release, with low-exposure ("Human") firms drifting down: a sustained revaluation, not a fad bounce.
Adoption effects: firms investing more in AI grow faster in sales and employment, primarily via product innovation (trademarks, product patents) rather than cost cutting; AI investment and scale reinforce each other, raising industry concentration: winner-take-most, superstar-firm dynamics.
LLMs as research tools: ChatGPT-extracted "expected investment" scores from earnings calls predict future capex beyond Tobin's q and cash flow, move analyst capex forecasts, and beat older NLP (e.g. RoBERTa) at extracting policy-relevant content; extensions to dividends and employment look feasible.

Required papers

Paper	Question	Finding	Critical note
Babina, Fedyk, He & Hodson (2024)	What does AI adoption do to firms?	Firms investing more in AI (measured from employee skills/resumes) grow faster in sales and employment, primarily through product innovation (new products, trademarks, product patents) rather than cost-cutting. AI investment and scale reinforce each other, raising industry concentration: winner-take-most, superstar-firm dynamics.	The AI-skill measure is noisy, and fast-growing firms may simply hire more AI talent (reverse causality); concentration findings reopen the W1/W11 competition questions.
Eisfeldt & Schubert (2024), "AI and Finance"	How does generative AI affect asset values and finance research?	ChatGPT's launch (30 Nov 2022) is a structural break: high generative-AI-exposure ("Artificial") portfolios revalue upward ~8–10% and the gap persists, while finance-sector AI-skill demand jumps. LLMs are also research tools — ChatGPT-extracted "expected investment" from earnings calls predicts capex beyond Tobin's q and beats older NLP.	Event-study revaluations bundle AI cash-flow expectations with discount-rate and attention effects; text-based "exposure" is gameable (managers can sing the AI song, cf. W8 intangibles talk). Treat as early-stage evidence.

Critical evaluation Event-study revaluations bundle expectations of AI cash flows with discount-rate and attention effects; "exposure" measures are text-based and gameable (cheap talk, cf. W8's intangibles talk: managers can sing the AI song too). Concentration findings raise the W11/W1 product-market questions about competition and entry. Treat this week's empirics as early-stage: clean identification is scarce.

30-second recall: Week 12 ChatGPT launch = structural break in finance skill demand; high-AI-exposure firms sustained ~8-10% revaluation; AI adopters grow via product innovation and concentrate industries; LLM-extracted investment scores predict capex beyond q. Caveats: text measures are gameable, identification is young.

All Weeks: Methodology, Results & Conclusions at a Glance

The whole course in one table, grouped by theme. For each week: how it was tested (methodology), the empirical results, the bottom-line conclusion, and where it sits in the broader academic debate. Author names are secondary — what matters is the method, the finding, the takeaway, and the literature it speaks to.

Theme	Week & topic	Methodology	Key results	Conclusion & broader debate
Capital structure & financing	W1 Foundations: MM & trade-off	Theory (MM homemade-leverage arbitrage proof); trade-off diagram; review of cross-sectional leverage and target-adjustment regressions.	In perfect markets financing is irrelevant (V independent of D/E); with taxes vs distress an interior optimum exists. But the most profitable firms borrow least (anti-trade-off) and M/B effects on leverage persist for decades.	MM is a diagnostic benchmark; every later theory relaxes one assumption. 60+ years on, no model explains broad-brush capital-structure facts (DeAngelo's "are we there yet?").
	W2 Information asymmetry	Myers–Majluf signalling model; SSM deficit regression with power simulations; CFO survey (Graham–Harvey); M/B_efwa panel regressions; variance decomposition / firm fixed effects (LRZ).	Issuing risky securities signals overvaluation → ~−2 to −3% equity-announcement returns and a pecking order; β_PO≈0.75 but collapses in larger samples; timing persists a decade; a firm fixed effect dominates all covariates and predates the IPO.	Pecking order is the best first-order description, yet the dominant source of leverage variation is a time-invariant effect none of the three theories explains. Pecking order vs trade-off vs market timing — unresolved.
	W3 Security design: convertibles	Option-pricing mechanics; design-classification studies (delta / conversion probability); surveys & interviews (Dong et al.); hedge-fund short-interest evidence.	Convertibles = debt + equity option; the Big Four rationales (risk-shifting, risk-uncertainty, backdoor equity, sequential financing) each win under some method/geography; modern issues are bought by convertible-arb hedge funds who short delta shares (synthetic equity).	No single rationale wins; results differ by method and region (Dutordoir et al.). Convertibles are a Myers–Majluf adverse-selection fix (Stein = "Myers–Majluf plus a call"); the "free lunch" is a fallacy.
	W11 Raising capital (SEOs)	International announcement-return event studies; horse-race regressions (need vs timing vs lifecycle); fuzzy RDD on the most-recent-offer-price anchor; firm–bond matched-data analysis.	SEO announcements ~−2%; shareholder approval flips the sign internationally; need first, lifecycle second, timing third; price-anchored SEOs fund inefficient acquisitions (+8.7–11.3% spending); bond issuers trade cost of capital vs investor-base diversification.	Issuance is adjudicated by need > lifecycle > timing, with behavioural anchoring causing real misallocation; disciplines Baker–Wurgler. Pecking order vs timing, plus the rise of non-bank intermediation.
Mergers & acquisitions	W4 M&A foundations	Large-sample 3-day-CAR event studies plus operating-performance; logit wave-prediction horse race (Harford); acquirer fixed-effects variance decomposition (GYZ).	Targets +16% (+24% to completion), acquirers ~0/−0.7%, combined +1.8–2%; stock deals worse; waves cluster by industry, driven by shocks + capital liquidity (cash partial-firm deals also cluster — kills the behavioural story); acquirer FE explain as much as all covariates (IQR spread 6pp ≈ $184m), persistent.	Mergers create modest value and targets capture it; clustering is neoclassical, not behavioural; persistent acquirer "skill" (or stable unobservables). Neoclassical vs behavioural waves; skill vs luck.
	W8 M&A: additional topics	Textual analysis of announcement language (intangibles talk); short-selling-potential tests with a public/private placebo; failed-deal event studies.	Intangibles talk → −0.53pp acquirer CAR and more cash; easier-to-short bidders use more cash only for public targets; after deals fail, cash-bid targets stay revalued ~+15% while stock-bid targets revert.	Payment method is information — about bidder overoptimism, anticipated arbitrage pressure, and target standalone value. The signals partly pull apart, so the literature is unsettled.
	W10 Investment banks	Cross-sectional fee & outcome regressions controlling for ln(deal value); analysis of fairness-opinion peer selection; shared-bank takeover-prediction models.	Top-tier advisors earn a fee premium (after size controls) and deliver better public-deal outcomes; fairness-opinion comps are strategically selected; shared banks predict takeovers and higher completion.	Banks are reputational intermediaries and conflicted certifiers: premium-price/premium-quality coexists with valuation advocacy and cross-client information leakage. Reputation theory (Klein–Leffler) vs conflict of interest.
Governance, ownership & control	W5 Governance around the world	Cross-country legal-index construction (49 countries); legal origin as a quasi-exogenous instrument; taxonomy/review (Denis–McConnell).	Common law gives the strongest investor/creditor rights, French civil law the weakest; ownership concentration is high and negatively related to investor protection; weak law → small capital markets.	Securities are bundles of legally-enforced rights; concentration substitutes for weak law. Law-and-finance; legal-origin exogeneity is contested (Spamann's recoding).
	W6 Boards of directors	Tobin's-Q regressions (cross-section & within-firm) plus board reduction/expansion event studies (Yermack); Reg S-K skill-disclosure analysis (AAV); grant-date abnormal-return patterns incl. market moves (Lie).	Board size inversely related to Q (concave), reductions earn positive CARs; skill commonality (not diversity) associates with better performance; post-grant market-wide returns prove retrospective backdating; SOX 2-day reporting killed it.	Small, cohesive boards monitor better, but endogeneity and firm-complexity caveats bite; one-attribute board regulation distorts. Internal-mechanism governance.
	W7 Blockholders & large investors	Demand-system estimation from holdings (Gompers–Metrick); churn-sorted return/earnings predictability (Yan–Zhang); liquidity-shock natural experiments (decimalisation).	Institutional demand shifts move relative prices; only short-horizon institutions' trades are informed; liquidity shocks raise value more where blockholders and price-linked pay coexist (the exit threat).	Large investors govern via voice and exit, and size substitutes for weak law; the exit threat works without visible activism. Berle–Means vs concentrated control; voice-vs-exit (Edmans); common ownership.
Public vs private firms	W9 Private firms	Within-private-firm performance regressions on ownership structure; regulation-threshold disclosure quasi-experiment (Ortiz); public-vs-private public-bond-cost comparison (Badertscher).	Vertical and horizontal agency both depress performance; mandatory disclosure increases M&A matching; private ownership raises public-debt costs (26–60bp), with a 15–30% private valuation discount.	The public/private margin is a trade-off: liquidity and disclosure vs agency and issuance costs. "Public = better governed" is not obvious. Jensen–Meckling agency; the declining-US-listings debate (Doidge–Karolyi–Stulz; Ewens–Farre-Mensa).
AI & finance	W12 AI and finance	Event study around the ChatGPT launch (exposure-sorted portfolios); job-postings text analysis; AI-investment firm-growth regressions; LLM extraction of "expected investment" from earnings calls, validated against realised capex.	Structural break in finance AI-skill demand; high-exposure firms revalue +8–10% and the gap persists; AI adopters grow via product innovation, raising concentration; LLM-extracted scores predict capex beyond Tobin's q.	AI is revaluing firms and reshaping labour and competition, and LLMs are credible research tools — but evidence is early and text-based exposure measures are gameable. Superstar-firm concentration (Autor et al.); text-as-data; event-study identification limits.

How to use this in the exam This is your synthesis scaffold for any cross-topic question: name the methodology (it signals you know how the claim was identified), state the result with a number or the identifying table, give the conclusion, then locate it in the broader debate. The recurring meta-theme across all five themes is that information and incentives — who knows what, who is conflicted, and what a choice signals — drive capital structure, payment methods, governance, issuance, and now AI revaluations alike. The recurring methodological meta-theme is identification: event studies, natural experiments (liquidity shocks, regulatory thresholds, RDD), placebos, and the warning that fixed effects often dwarf the theories' favourite covariates (LRZ in W2, GYZ in W4).

How to Read a Regression Table (explicitly examinable)

The exam info says you must know "how hypotheses were tested (including interpreting the tables)." Here is the checklist, then the course's recurring table patterns.

Unit of observation and sample: firm-year? deal? country? What years, what filters (the SSM survivorship issue starts here).
Dependent variable and its scaling: e.g. ΔD and DEF scaled by total assets; CAR over which window; Tobin's Q.
The coefficient that matters and its predicted value: not just sign and stars. SSM is the canonical case: the test is β_PO = 1, so β = 0.62 with three stars is significant evidence against the strict pecking order (see the Wald statistic row), even though it is significantly positive. Stars test against zero; theories often predict numbers other than zero.
Controls and fixed effects: what does identification come from once firm/industry/year FE are in? (Within-firm variation only.) Watch coefficients that die when FE enter (Harford's M/B; raw top-tier fee effect dying to ln(deal value)).
Standard errors: clustered by firm? Panel data with un-clustered SEs overstate significance (Petersen).
Economic magnitude: translate a coefficient into something real (Golubov: interquartile FE spread = 6pp of CAR = $184m for the mean acquirer; anchored SEOs = +8.7-11.3% acquisition spending).
Endogeneity strategy, if any: IV (Ortiz thresholds, Bharath liquidity shocks, anchoring instruments), natural experiment (Heron-Lie post-SOX), placebo (private targets in Dutordoir et al.), or simulation (SSM power tests). If none, say "associational" and state the most plausible confounder.

Recurring patterns worth recognising on sight Coefficient shrinks toward zero as controls/FE are added: the variable was proxying something else (Harford, fee tables). Coefficient stable across specifications: robustness claim (Yermack). Effect present in treatment subsample, absent in placebo subsample: mechanism evidence (Dutordoir et al.: public vs private targets; Lie: market component). High R² from fixed effects, low from covariates: theories explain little (LRZ, Golubov).

One-Page Formula & Number Sheet

Item	Statement
MM I / II	V = D + E constant; r_E = r_A + (D/E)(r_A − r_D)
Trade-off	V_L = V_U + PV(tax shields) − PV(distress costs)
MM84 issue rule	Issue iff E/(P′+E)·(E+S+a+b) ≥ S+a, i.e. a ≤ P′(1+b/E) − S
SSM regression	ΔD = α + β_PODEF + ε; DEF = DIV + CAPEX + ΔWC − CF; H₀: α=0, β=1; found β≈0.75
Convertible	Conv. price = principal/ratio; conv. value = ratio × S; premium = conv. price/S_issue − 1; delta = e^−δTN(d₁)
Herfindahl	H = Σ s_i² (50 largest); <0.1 competitive; in concentrated industries a merger raising H by >0.01 can raise antitrust concern
Key magnitudes	Target CAR +16% (3-day) / +24% to completion; acquirer ~0; combined +1.8%; SEO announcement ~−2%; IPO fee ~7% + underpricing ~15-20%; SEO fee ~4.5%; PE 2/20; private firm discount 15-30%; loan spread gap 26-60bp; mean board 12.25; Golubov FE IQR >6pp ($184m); Disney convertible: 34 shares, conv. price $29.41

Cross-Topic Synthesis: the threads the exam loves

Thread	Where it appears
Asymmetric information	W2 Myers-Majluf core; W3 convertibles as the design fix; W4/W8 payment method as signal; W9 private-firm opacity and cost of debt; W10 banks as certifiers; W11 SEO discounts and approval effects.
Agency conflicts	W1 trade-off additions (overhang, risk shifting, FCF); W3 Green and Mayers rationales; W4 hubris/empire building; W5-W6 law and boards as constraints; W7 blockholder voice/exit; W9 vertical vs horizontal agency; W11 anchored SEOs funding bad deals.
Fixed effects beat the theories	W2 Lemmon-Roberts-Zender (leverage) and W4 Golubov et al. (acquirer returns): in both, a time-invariant firm effect explains more than the covariates the theories propose. Be ready to say what FE can and cannot tell us.
Statistical power and identification	W2 SSM simulations; W4 Harford's discriminating prediction (cash partial-firm deals); W6 Lie's market-component logic; W7 liquidity natural experiments; W8 failed-deal laboratory; W9 disclosure-threshold designs. The exam rewards explaining why a test discriminates.
Theory vs survey vs archival evidence	Graham-Harvey (W2), Bancel-Mittoo and Dong et al. (W3): managers' stated motives never line up perfectly with archival inference; each method has bias.

Self-Test Question Bank ↑ top

Click a question to reveal a model-answer sketch. These mirror the style of the slide-embedded "potential exam questions."

State MM Proposition I and explain why relaxing exactly one assumption generates each major capital structure theory.

V = D + E constant in perfect markets (arbitrage/homemade leverage proof). Relax taxes/distress: trade-off. Relax symmetric information: pecking order and market timing (mispricing). Relax aligned incentives: agency theories (JM 1976, Jensen 1986). The propositions are a diagnostic baseline, not a description.

In Myers-Majluf, why does the announcement of a (risky) stock issue lower the price, and when would it not?

Managers acting for old shareholders only issue when shares are not too undervalued (a ≤ P′(1+b/E) − S), so issuing reveals a is low: P′ < P. No drop if the issue is a foregone conclusion (probability 1) or if default-risk-free debt is issued (equivalent to having slack, so the only news is a positive-NPV project).

Shyam-Sunder & Myers found support for BOTH models. Why did they still conclude in favour of the pecking order?

Power simulations. On data generated by a pure pecking order, the target-adjustment regression still fit (so its fit on real data is uninformative); on data generated by target adjustment, the pecking-order regression failed (so its fit on real data is informative). A test only supports a theory if it can reject the alternative.

A firm issues a convertible with a very high conversion premium and no call protection. Which issuance theory fits best and why?

High premium = low conversion probability = debt-like design: points to Green (risk shifting) or Mayers (sequential financing, given callability), not Stein (backdoor equity needs equity-like design plus call provisions to force conversion). Mention design-as-evidence logic (Lewis et al.) and the caveat that surveys reject Green.

Cash-financed partial-firm acquisitions cluster during merger waves. Which wave theory survives, and why exactly?

Neoclassical/liquidity (Harford). Behavioral waves require overvalued stock as currency and whole-firm targets; there is no behavioral motive to pay cash for a division. Efficient post-shock reallocation predicts transactions of all forms clustering together, which is what the data show.

Why is the post-grant pattern in MARKET-WIDE returns the decisive evidence for backdating (Lie 2005)?

Opportunistic timing on private information could explain firm-specific run-ups after grants. But executives cannot forecast the aggregate market; positive abnormal market-wide movements after grant dates are only consistent with the date being chosen after the fact. Heron-Lie: pattern fades once 2-day reporting removes the backdating window: regulatory natural experiment confirming the mechanism.

How can blockholders govern without ever intervening?

Exit threat: informed blockholders sell on bad managerial actions, depressing the price; managers with price-linked wealth internalise this ex ante. Bharath et al.: exogenous liquidity increases (which sharpen exit threats) raise value more in firms with more blockholders and price-sensitive managers.

After a failed CASH bid, the target stays revalued; after a failed STOCK bid it does not. Interpret.

Failed bids isolate information content (no synergies realised). Cash commits the bidder to a certain payment, credible only if the bidder believes in the target's standalone value: positive revelation persists. Stock shifts valuation risk to target shareholders, so a stock bid reveals little (or signals bidder overvaluation): no persistent revaluation (Malmendier, Opp & Saidi 2016).

Top-tier M&A advisors show a NEGATIVE raw correlation with advisory fees. Does this refute reputation theory?

No: a size artifact. Top-tier banks work larger deals, and percentage fees decline mechanically in deal size. Controlling for ln(deal value), the top-tier premium is positive and robust: consistent with the Klein-Leffler/Chemmanur-Fulghieri premium-price, premium-quality equilibrium. Always interrogate raw correlations for composition effects.

Why do convertible issuers' stocks attract heavy short selling at issuance, and is that bad news?

Convertible arbitrage: hedge funds buy the bond and short delta shares. The shorting is anticipated and information-free, so price impact is limited; it effectively synthesises an equity issue for firms rationed out of the SEO market (Brown et al. 2012). Institutional ownership actually rises afterwards. Bad news inference from short interest fails when shorting is hedging.

Common-law countries show dispersed ownership; French-civil-law countries show concentration. Causal story and main caveat?

Weak investor protection makes minority stakes expensive to hold (expropriation risk), so concentrated ownership emerges as a substitute governance mechanism: LLSV find concentration negatively related to protection, with legal origin as the plausibly exogenous driver. Caveat: origin correlates with colonial history, culture, politics; coding of rights indices involves judgment; rich civil-law countries show weak law is not an insurmountable bottleneck.

Does Yermack (1996) justify a regulation capping board size at 8?

No. The association is robust but board size is endogenous to firm complexity; the concavity suggests biggest costs in small-to-medium moves, and later evidence finds complex firms benefit from larger advisory boards. Yermack himself shows size is insensitive to past performance, but optimal size heterogeneity means one-size caps would bind exactly where large boards may help. Adams-Akyol-Verwijmeren reinforce: single-attribute regulation distorts a multidimensional optimisation.