about BMAN30071
1 · Exam mapofficial facts
“Assessment Method Exam (100%) in January 2027 examination period” (BMAN30071 course outline 2026/27) · “Students should be capable of completing this open book exam without the use of a calculator.” (BMAN30071 exam paper 2025/26)
ONLINE OPEN-BOOK: online, unseen and open book. 48-hour completion window; answer two of three questions, 50 marks each, combined 2,000-word limit.
| What the final rewards | Visible working |
|---|---|
| Regression evidence | DV, base, units, combined slope |
| Integrated argument | mechanism → model → finding |
| Evaluation | proxy/design-specific consequence |
2 · Question decodercourse FAQ
- Define the claim: valuation level, price revision, recognition timing or disclosure effect?
- Map the equation: dependent variable, regressor, benchmark/base category, sign prediction.
- Read evidence: direction and economic magnitude; then precision; then model fit.
- Evaluate the path: timing, proxy noise, omitted traits, selection or forecast structure.
Answer architecture
| Paragraph job | Minimum content |
|---|---|
| 1 · mechanism | why the accounting signal should reach value or return |
| 2 · specification | Y, main X, base group, predicted sign |
| 3 · result | combined coefficient; magnitude and precision separately |
| 4 · evaluation | one model-specific threat and how it changes inference |
| 5 · judgement | answer the wording; keep association inside design limits |
Questions can integrate several teaching sessions. Link neighbouring mechanisms explicitly: information timing → disclosure anticipation; recognition → conservatism; book value → adaptation or abnormal earnings; narrative construction → investor response.
3 · Price relevanceC1
Return model: price revision ↔ news during a period
Price model: equity-value level ↔ accounting variables at a date
| Question | Use | Do not claim |
|---|---|---|
| Value at date? | Price specification | level = unexpected news |
| Revision in interval? | Return specification | raw change = surprise |
| Equal surprises? | compare persistence, risk, prior information | equal price response |
Earnings can grow yet disappoint: the year-on-year change and the surprise can have opposite signs. A more persistent innovation revises a longer expected stream and can justify a larger response, subject to discounting, risk and other information.
Earnings-response coefficient = estimated return change for a defined unit of earnings news. Translate its units; association is not causation.
Persistence decision
| Surprise story | Future revision | Expected response |
|---|---|---|
| temporary | few periods affected | relatively weaker |
| persistent | longer earnings stream revised | relatively stronger |
| already anticipated | little genuinely new information | weaker at announcement |
A complete comparison holds the current surprise constant, varies the persistence or prior-information assumption, traces the altered future stream, then qualifies the response for discounting and risk. Do not jump directly from a reported earnings level to return.
Safe synthesis: price models explain value at a date; return models explain revisions during an interval using a defensible news proxy.
3b · Fit is not usefulnessC1 FAQ
A low R² says the fitted model associates earnings with a limited share of return variation in that sample. Timely non-earnings information, noisy surprise proxies and heterogeneous persistence can weaken contemporaneous fit without proving earnings irrelevant.
4 · Regression toolkitC2
OLS chooses α̂, β̂ to minimise Σeᵢ²
R² = variation in Y associated with the fitted model / sample variation in Y
t = estimated coefficient / standard error
Table reading order
- Dependent variable and event/window; sample and scaling.
- Base group; set each dummy to 0 and 1.
- Direction + magnitude in stated units; combine interactions.
- Precision: t-statistic/test evidence is distinct from economic size.
- Explanatory power: R² is fit, not the coefficient or causal strength.
- Specific limitation: name the noisy proxy/path and predicted consequence.
D=0: intercept α; slope β
D=1: intercept α+γ; slope β+δ
δ tests the slope difference; the full D=1 slope may require a joint test.
| Noise enters | Lecture consequence |
|---|---|
| Dependent variable Y | larger residual variation; weaker fit |
| Independent variable X | slope may attenuate toward zero under maintained conditions |
Never collapse: statistical precision ≠ economic importance ≠ explanatory power. A small coefficient can matter if the unit change is large; a precise coefficient can still be economically slight.
Six-sentence table translation
- “The dependent variable is … measured over …”.
- “The omitted/base category is …”.
- “A one-unit increase in X is associated with β units of Y for the base group.”
- “For D=1, the slope is β+δ; δ is the group difference.”
- “The reported test gives evidence about …, while R² describes sample fit.”
- “Interpretation is limited by [named proxy/design path], which may [specific consequence].”
5 · Prices lead earningsC3
Prices need not know the future perfectly. Alternative information can update expectations before accounting verification recognises the consequence. Earlier return then contains news without later earnings; later earnings contains recognition after the main price move.
| Empirical remedy | Gain | Cost |
|---|---|---|
| Longer return window | captures earlier price response | admits unrelated firm/market news |
| Future earnings variables | represents delayed accounting consequences | future variables/proxies remain noisy |
6 · Disclosure anticipationC4
Low disclosure slope = β
High disclosure slope = β+δ
A positive δ is consistent with stronger anticipation only if it is estimated precisely and the design is credible. It does not prove disclosure caused the full association: disclosure groups may differ in predictability, industry, size or incentives.
Controls have jobs: future returns address price-related measurement effects in future earnings; earnings yield, growth and firm traits address specified alternative paths. Explain each from the design, never as generic “controls”.
Joint-evidence test
Show why later earnings can represent consequences priced today; show that disclosure changes that slope; then ask whether disclosure merely sorts firms by underlying predictability. A positive, precise δ is consistent with stronger anticipation, while systematic group differences still threaten a causal claim.
7 · Earnings conservatismC5
D=0 good-news slope = β
D=1 bad-news slope = β+δ
Prediction: δ>0 under the stated negative-return dummy convention.
| Correct reading | Trap |
|---|---|
| δ is the incremental bad-news sensitivity | δ is the full bad-news slope |
| earnings responds to return as news proxy | return responds to earnings |
| larger magnitude reflected more promptly | bad news raises income |
| timeliness = when recognised | persistence = how long it lasts |
Verification can delay favourable developments while losses enter sooner. The result is a recognition-timing claim, not a claim that losses are beneficial or permanent.
Evaluation points
- Is return a clean proxy for economic news?
- Is the dummy coding stated rather than assumed?
- Do scaling and cross-firm recognition differences change the comparison?
- Does test evidence support the slope difference and/or the combined slope?
Reconstruction grid
| Set | Equation piece | Meaning |
|---|---|---|
| D=0 | α + βR | good-news base relation |
| D=1 | (α+γ) + (β+δ)R | bad-news intercept and slope |
| δ>0 | bad slope − good slope | predicted asymmetric timeliness |
Keep sign and magnitude distinct. Negative returns proxy adverse news, so earnings generally moves downward; the steeper branch means the magnitude is reflected more strongly. A positive δ supports asymmetry only under the stated coding and with adequate precision.
Essay bridge: conservatism helps explain why current earnings can lag some value-relevant developments and why equal reported earnings may have different implications when recognition histories differ. Keep this connection distinct from the persistence of the recognised component.
8 · Adapt or continue?C6
| Profitability region | Relative anchor | Logic |
|---|---|---|
| low E/BV | book/adaptation value ↑ | resources poorly used now |
| switching | both matter | options are close |
| high E/BV | earnings/recursion value ↑ | continuation produces more |
Adaptation includes restructuring, sale, takeover or internal redeployment—not only liquidation. Recursion capitalises sustainable earnings from the existing business technology. Scale earnings by book value before ranking operating success.
Proxy limits: recorded equity may differ from recoverable resources; current earnings may not be sustainable; discrete profitability groups approximate a continuous switching process.
9 · Abnormal-earnings valueC7
Normal earningsₜ = r × BVₜ₋₁
Abnormal earningsₜ = Eₜ − rBVₜ₋₁
Vₜ = BVₜ + PV(expected future abnormal earnings)
Use opening book value for the capital charge. Dividends reduce closing book value through clean surplus; they are not deducted again from earnings. Zero abnormal earnings means the firm earned exactly the required return, not zero accounting profit.
Information dynamics forecast persistence and other information. Separate the clean-surplus identity from linear forecasting assumptions: structural change or conservative recognition may weaken the dynamics without erasing the book-value anchor.