Event Studies & Financial Risk Services
How did the market react to an announcement? What return should an asset earn for its risk? How likely is a borrower to default? These are the questions of event studies, asset pricing, and risk modelling. MAS Research delivers all three—measuring abnormal returns, estimating risk–return relationships, and modelling credit and market risk, to the standard finance review demands.
An event study measures the abnormal return around a specific event (an announcement, merger, or policy change) to gauge its market impact, by comparing actual returns to those expected from a benchmark model. Asset-pricing models relate expected return to risk; credit-risk models estimate default probability and loss; and financial-risk modelling quantifies market and tail risk through measures such as Value-at-Risk and Expected Shortfall.
What these methods do
This page covers three closely related pillars of empirical finance that share a common concern: measuring return and risk precisely. Event studies ask how the market reacts to a specific event—an earnings announcement, a merger, a regulatory change, an index inclusion—by measuring the abnormal return around it: the difference between the actual return and the return a benchmark model says would have been expected. Cumulated over an event window, this abnormal return quantifies the event’s market impact, making event studies a workhorse of corporate finance, accounting, and market-efficiency research.
Asset-pricing models address the complementary question of what return an asset should earn for its risk—from the CAPM through multi-factor models (Fama–French three-, five-factor, and extensions) that relate expected returns to systematic risk factors. Credit-risk models estimate the probability of default and expected loss on loans or bonds. And financial-risk modelling quantifies the risk of a position or portfolio through measures such as Value-at-Risk (VaR) and Expected Shortfall (ES). Together they form the measurement backbone of empirical and applied finance.
When to use each
Use an event study when the question is the market impact of a discrete, dated event—does the market value this announcement, and by how much? Use an asset-pricing model when the question is about risk-adjusted returns, risk premia, or whether an anomaly survives standard risk controls—and as the benchmark (“normal return”) model inside an event study itself. Use credit-risk modelling when the question concerns default probability, credit scoring, or loss given default. Use financial-risk modelling when you need to quantify and backtest the market or tail risk of a position, often drawing on volatility models for the time-varying variance that risk measures depend on.
These methods also connect: an event study needs an asset-pricing model to define normal returns; a VaR estimate needs a volatility model; a short-run event impact can be read alongside a difference-in-differences design for a slower-moving regulatory effect. We select and combine them to fit the question and the data.
Four related questions, four methods
| Question | Method | Key output |
|---|---|---|
| How did the market react to an event? | Event study | Abnormal & cumulative abnormal returns (AR/CAR) |
| What return should an asset earn for its risk? | Asset-pricing / factor models | Risk premia, alphas, factor loadings |
| How likely is default, and what is the loss? | Credit-risk models | Probability of default, expected loss |
| How much could this position lose? | Financial-risk modelling | Value-at-Risk, Expected Shortfall |
Getting return and risk measurement right
A credible event study rests on several careful choices. The normal-return model (market model, market-adjusted, or a factor model) and the estimation and event windows must be chosen appropriately—too wide an event window invites contamination from other events, too narrow risks missing the reaction. Inference must handle the well-known pitfalls: event-induced variance (volatility rises around events, so standard tests over-reject), cross-sectional correlation when events cluster in calendar time, and the choice between parametric and non-parametric tests. We use tests robust to these issues rather than a naive t-test, and are careful that a measured abnormal return reflects the event, not a confounding one in the same window.
For asset-pricing models, the standard concerns are the factor specification, the handling of the well-documented errors-in-variables and standard-error problems in two-pass regressions, and honest treatment of anomalies. For credit and financial-risk models, the decisive discipline is validation and backtesting: a VaR or Expected Shortfall model must be backtested against realised outcomes—checking whether the frequency and clustering of exceptions match the model—and credit models assessed out of sample, with attention to the fat tails and non-normality that make tail risk easy to underestimate. Across all of these, we report the specification choices and the validation evidence, because a return or risk number is only as good as the model behind it.
The common failures are predictable—so we guard against them. Event studies need variance- and clustering-robust tests and clean windows free of confounding events; risk models (VaR, ES) must be backtested, not just fitted, with fat tails taken seriously. A number without these checks is not evidence.
Software
We deliver these methods in established, reproducible tools—R (eventstudies, PerformanceAnalytics, rmgarch), Python, and Stata/EViews—covering event-study estimation with robust tests, asset-pricing and factor regressions, credit-risk and VaR/ES modelling, and backtesting, all with versioned code.
How we deliver an event-study or risk analysis
This work sits within our wider financial-econometrics practice—so the normal-return or risk model is appropriate, the tests are robust, and risk measures are backtested.
For an event study, we define the event and windows, choose and estimate the normal-return model, compute abnormal and cumulative abnormal returns, and test them with variance- and clustering-robust methods, checking for confounding events. For asset-pricing work, we specify and estimate the factor model with appropriate inference. For risk, we build the credit or market-risk model, drawing on volatility models where needed, and—crucially—backtest it against realised outcomes.
Reporting sets out the design and model choices, the results (AR/CAR with robust tests, risk premia, default probabilities, or VaR/ES), and the validation and robustness evidence—so the findings can be judged against the standards finance reviewers apply.
You receive the estimates (abnormal returns and CARs with robust significance tests, factor-model results, credit-risk estimates, or VaR/ES figures), the specification and window choices, the backtesting and robustness evidence, clear visualisations, and reproducible analytical code and analysis-ready files (where appropriate and permitted).
Event studies & risk across finance research
Return and risk measurement sits at the centre of empirical and applied finance—so these methods run throughout the field.
Corporate Finance & Accounting
Event studies of announcements, mergers, disclosures, and index inclusions—measuring how markets value corporate events.
Asset Pricing & Markets
Factor models, risk premia, and anomaly testing—whether returns survive standard risk controls.
Banking & Credit
Credit-risk and default modelling, credit scoring, and loss estimation for loans and bonds.
Risk Management
Value-at-Risk, Expected Shortfall, and market-risk measurement, with the backtesting regulators and reviewers expect.
Regulation & Policy
Market reactions to regulatory and policy events, often alongside quasi-experimental designs for slower effects.
Green, Crypto & Emerging Assets
Event studies and risk modelling for ESG events, digital assets, and new markets, with appropriate tail-risk treatment.
Event studies & financial risk: common questions
Measuring a market reaction, a risk premium, or a risk exposure?
Whether it is the abnormal return around an event, a factor-model estimate, or a VaR that must be backtested, we deliver the analysis to finance-journal standards—robust tests, appropriate models, and validation that makes the number trustworthy.