Finance Research Applications

Introduction and Context

Finance research applies economic reasoning, accounting information, statistical analysis, and financial models to decisions involving value, risk, return, funding, and markets. In an accounting and finance dissertation, the researcher is expected not merely to describe financial events but to formulate a defensible question, identify appropriate evidence, apply a suitable analytical method, and explain the implications of the findings.

Finance research is applied across several connected domains:

• Valuation: estimating the economic value of companies, securities, projects, and intangible assets.
• Investment analysis: evaluating return, diversification, portfolio performance, and investor behaviour.
• Corporate finance: studying capital structure, dividends, mergers, investment decisions, and financial distress.
• Capital markets: examining prices, liquidity, information efficiency, and market reactions.
• Banking: analysing credit creation, profitability, regulation, liquidity, and systemic risk.
• Fintech: investigating digital payments, platforms, cryptocurrencies, algorithmic advice, and financial inclusion.
• Risk management: measuring and controlling market, credit, liquidity, operational, and cyber risks.

A strong project links a research problem to a theory, a measurable variable, a credible dataset, and a decision criterion. The central discipline is alignment: the research question, literature review, methodology, model, evidence, and conclusion must support one another.

Learning Objectives

By the end of this chapter, students should be able to:

1. Formulate applied research questions in major areas of finance.
2. Select and justify valuation, investment, corporate finance, market, banking, fintech, and risk models.
3. Calculate and interpret important financial measures.
4. Distinguish assumptions, evidence, correlation, causation, and decision relevance.
5. Design a research project using market data and appropriate statistical methods.
6. Evaluate ethical, regulatory, technological, and practical limitations.
7. Communicate results in a rigorous dissertation or professional report.

1. From a Finance Problem to a Researchable Question

A finance problem becomes researchable when it identifies a population, variables, period, relationship, and decision context. “Does fintech improve banking?” is too broad. A more useful question is: “Did the adoption of mobile banking reduce operating costs and increase customer retention in listed commercial banks between 2018 and 2024?”

A research question should normally specify:

• the unit of analysis, such as a firm, bank, security, investor, transaction, or country;
• the dependent variable, such as share return, default probability, or firm value;
• one or more explanatory variables, such as leverage, interest rates, or digital adoption;
• the time horizon and sampling criteria;
• the relevant theory and expected mechanism; and
• the practical decision to which the result contributes.

A conceptual model translates theory into testable relationships. For example, agency theory predicts that ownership structure and governance affect managerial decisions; the trade-off theory predicts that firms balance the tax benefits of debt against expected distress costs; and modern portfolio theory explains how diversification can reduce unsystematic risk.

A hypothesis should be directional where theory supports direction. For example: “Holding size, profitability, and industry constant, higher leverage is associated with greater expected financial distress.” A hypothesis is not a conclusion; it is a proposition subjected to evidence.

2. Valuation Applications

2.1 Discounted Cash Flow Valuation

The discounted cash flow (DCF) approach values an asset as the present value of expected future cash flows. For a finite stream:

Here, V₀ is value today, CF_t is cash flow in period t, r is the discount rate per period, and T is the final period. Cash flows and the discount rate must use consistent timing and currency units. The model assumes that expected cash flows and the risk-adjusted discount rate are sufficiently estimated.

For a firm, free cash flow to the firm (FCFF) is commonly discounted at the weighted average cost of capital (WACC):

where EBIT is earnings before interest and tax, T is the corporate tax rate, D&A is depreciation and amortisation, Capex is capital expenditure, and ∆NWC is the increase in net working capital. This formulation is appropriate when cash flows are available to both debt and equity providers.

The WACC is:

where E and D are market values of equity and interest-bearing debt, R_e is the cost of equity, and R_d is the pre-tax cost of debt. The tax adjustment assumes interest is tax-deductible and that the firm can use the tax shield.

A constant-growth terminal value is:

where g is the perpetual growth rate and must be less than WACC. The terminal value is then discounted back to the valuation date. Because terminal value often forms a large proportion of enterprise value, sensitivity analysis is essential.

2.2 Relative Valuation

Relative valuation compares a company with similar companies using multiples such as price-to-earnings (P/E), enterprise-value-to-EBITDA (EV/EBITDA), or price-to-book (P/B). A peer multiple is meaningful only when firms have comparable growth, risk, accounting policies, capital structure, and profitability. A low multiple may indicate undervaluation, but it may also reflect weak future prospects or higher risk.

A robust dissertation should triangulate methods rather than treat one estimate as truth. DCF provides an intrinsic-value framework; multiples provide market-based evidence; asset-based valuation may be useful for asset-intensive or distressed firms.

3. Investment and Portfolio Applications

The holding-period return is:

where P₀ is initial price, P₁ is ending price, and D₁ is income received during the period. If returns are measured over different periods, annualisation must be performed carefully. For an investment held for n years, the compound annual growth rate is:

Expected portfolio return is the weighted average of asset expected returns:

where w_i is the portfolio weight and the weights normally sum to one. Portfolio variance depends on individual variances and covariances:

Diversification reduces risk when asset returns are not perfectly positively correlated. It does not eliminate systematic risk arising from market-wide factors.

The Capital Asset Pricing Model (CAPM) states:

where R_f is the risk-free rate, E(R_m) is expected market return, and β_i measures the asset’s sensitivity to market movements. The model assumes, among other conditions, frictionless markets, homogeneous expectations, and investors who care about mean and variance. In empirical research, CAPM is a benchmark rather than an unquestionable description of reality.

The Sharpe ratio evaluates excess return per unit of total volatility:

It is most informative when returns, the risk-free rate, and volatility are measured over consistent periods. It can be misleading for non-normal returns or illiquid assets.

4. Corporate Finance Research

Corporate finance research investigates how firms invest, finance operations, distribute cash, and manage governance. Common research topics include:

• whether leverage affects firm value or profitability;
• whether dividends signal future performance;
• how working-capital efficiency affects returns;
• whether governance reduces agency costs;
• how mergers affect shareholders and employees; and
• which factors predict financial distress.

Investment decisions can be evaluated using net present value (NPV):

I₀ is the initial investment. Under conventional assumptions, a positive NPV indicates that the project is expected to add value at the selected discount rate. The internal rate of return (IRR) is the discount rate that makes NPV equal to zero. IRR can be problematic where cash-flow signs change more than once or where projects differ in scale; NPV is generally the stronger value criterion.

Working-capital studies often examine the cash conversion cycle:

Days inventory outstanding (DIO), days sales outstanding (DSO), and days payable outstanding (DPO) measure the operating timing of inventory, receivables, and payables. A shorter cycle may release cash, but excessively strict credit or low inventory can reduce sales and resilience.

A dissertation should distinguish accounting profitability from value creation. Return on equity, for example, can rise because of genuine operational improvement or simply because leverage increases the equity base’s exposure to risk.

5. Capital Markets and Event Studies

Capital-market research commonly tests whether prices respond to information. An event study estimates abnormal returns around an event such as an earnings announcement, dividend change, acquisition, regulatory decision, or cyber incident.

A market model may be written as:

where R_it is security i’s return at time t, R_mt is market return, and ε_it is the disturbance term. Abnormal return is:

Cumulative abnormal return over an event window is:

The researcher must define an estimation window separately from the event window, avoid overlapping events where possible, control for confounding announcements, and test statistical and economic significance. A statistically significant result may be too small to trade after transaction costs.

Market efficiency is not binary in practical research. Evidence may differ by market, asset class, information type, trading costs, and investor group. Survivorship bias, look-ahead bias, data-snooping, and incorrect benchmark selection can create apparently successful strategies that fail out of sample.

6. Banking, Fintech, and Financial Inclusion

Banks transform maturities, pool risks, screen borrowers, create payment mechanisms, and transmit monetary policy. Research measures banking performance through profitability, asset quality, liquidity, capital adequacy, and efficiency. Important variables include net interest margin, non-performing loans, loan growth, capital ratios, and liquidity ratios.

Banking research must consider endogeneity. For example, profitable banks may invest more in technology, while technology may also increase profitability. A simple correlation cannot establish the direction of causality. Panel-data methods, natural experiments, instrumental variables, difference-in-differences, or carefully designed surveys may improve identification, though each has assumptions.

Fintech research includes mobile money, open banking, peer-to-peer lending, robo-advice, digital identity, blockchain, and artificial intelligence. Relevant questions include whether digital credit expands access, whether algorithms discriminate, whether platform concentration creates systemic dependence, and how consumers understand automated financial advice.

Useful fintech measures include adoption rate, transaction success rate, fraud incidence, complaint frequency, cost per transaction, and customer retention. Researchers should not treat app downloads as equivalent to active financial inclusion. Access, affordability, regular usage, consumer protection, and outcomes must be distinguished.

7. Risk Management and Quantitative Evidence

Financial risk is the possibility that outcomes differ adversely from expectations. Major categories include market, credit, liquidity, operational, legal, reputational, cyber, and model risk.

Value at Risk (VaR) estimates a loss threshold over a specified horizon and confidence level. If daily loss is L, a 95% one-day VaR is the loss level exceeded in approximately 5% of comparable outcomes under the chosen model. VaR does not describe losses beyond the threshold. Expected shortfall (ES) estimates the average loss conditional on losses exceeding VaR and is therefore informative about the tail.

Under a simple normal-return approximation, a one-period VaR can be represented as:

where V is portfolio value, σ is return volatility, μ is expected return, and z_α is the relevant standard-normal quantile. This formula assumes a specified normal distribution, stable parameters, linear exposure, and a clearly defined sign convention. It is inappropriate to present the result without reporting the horizon, confidence level, currency, and method.

Credit risk research may estimate probability of default (PD), loss given default (LGD), and exposure at default (EAD). Expected credit loss is often expressed conceptually as:

The simple expression is appropriate as a basic one-period approximation. Practical regulatory and accounting models may require discounting, multiple scenarios, time horizons, correlations, and forward-looking information.

Stress testing asks how a portfolio or institution performs under adverse but plausible scenarios, such as a recession, interest-rate shock, currency depreciation, or cyberattack. Scenario analysis is not a forecast; it is a resilience tool.

8. Research Design, Data, and Model Evaluation

A finance dissertation should proceed through a transparent research design:

1. Define the problem and decision context.
2. Review theory and prior empirical findings.
3. State variables and hypotheses.
4. Identify the population, sample, period, and data sources.
5. Clean, document, and validate the data.
6. Select a model consistent with the question and measurement level.
7. Test assumptions and robustness.
8. Interpret economic as well as statistical significance.
9. Discuss limitations, ethics, and implications.

Descriptive statistics should precede modelling. Researchers should report means, medians, dispersion, missing values, outliers, and distributions. Financial returns are often non-normal and may display volatility clustering. Accounting data may contain structural breaks caused by changes in standards, mergers, inflation, or currency regimes.

Regression analysis can estimate relationships such as:

Here, y_i is the dependent variable, x_ki are explanatory variables, β_k are parameters, and ε_i is the error term. Interpretation requires attention to omitted variables, reverse causality, measurement error, multicollinearity, heteroskedasticity, autocorrelation, and sample selection.

The coefficient of determination is:

where SSE is the sum of squared residuals and SST is total sum of squares. A high R² does not prove causation or indicate that a model is economically useful. Out-of-sample prediction, robustness tests, alternative specifications, and economic effect sizes are essential.

Worked Example: A Small DCF Application

An illustrative company is expected to generate FCFF of £120,000 next year, growing at 4% annually forever. Its WACC is 10%. The constant-growth model gives:

Therefore:

The result assumes perpetual growth, a stable WACC, and cash flows available to all capital providers. If the growth rate rises to 5%, value becomes £2,400,000, showing substantial sensitivity. The researcher should therefore report a sensitivity table rather than a single unsupported point estimate. The model also demonstrates why terminal assumptions require particular scrutiny: a small change in WACC − g materially changes valuation.

Applied Case: Digital Lending and Credit Risk

This is an illustrative case. A mid-sized bank is considering a mobile lending platform for small businesses. Management expects lower processing costs and greater financial inclusion, but the board is concerned about fraud, default, data privacy, and model bias.

Background and Problem

The bank has three years of historical loan data, including borrower characteristics, repayment outcomes, application channel, loan size, and interest rate. However, digital applicants may differ systematically from branch applicants. A simple comparison of default rates could therefore be misleading.

Analysis

The research team defines two outcomes: probability of default within twelve months and cost per approved loan. It compares digital and branch applications while controlling for borrower risk, sector, loan size, and regional conditions. It conducts sensitivity tests using alternative default definitions and examines whether approval rates differ across protected or vulnerable groups.

The team also performs stress tests under higher unemployment and falling business revenues. It reviews whether the training data contain historical discrimination and whether automated decisions can be explained to applicants.

Approach and Solution

The bank adopts a staged pilot rather than an immediate full launch. Human review is retained for borderline cases; credit limits are initially conservative; performance is monitored by cohort; and an independent validation team reviews the model. Customers receive clear explanations, complaint channels, and consent information. The board receives regular reporting on default, approval fairness, fraud, operational outages, and data incidents.

Outcome

The pilot reduces processing time and cost, but default rates rise among thin-file borrowers. The bank therefore adds alternative data cautiously, improves financial-literacy support, and limits automated lending until validation demonstrates acceptable performance. The result is not a simple “technology succeeds” conclusion; it is a conditional finding balancing efficiency, inclusion, risk, and governance.

Learning Points

1. Financial innovation must be evaluated using both benefits and harms.
2. Selection bias can invalidate naïve comparisons.
3. Model performance and model fairness require separate tests.
4. Pilot designs, monitoring, and human accountability reduce implementation risk.
5. A research conclusion should identify conditions under which a decision is appropriate.

Ethical, Regulatory, and Professional Considerations

Finance research uses sensitive information and can influence investment, credit, employment, and public policy. Researchers must obtain permissions, protect personal data, anonymise records where possible, and avoid re-identification. They should disclose conflicts of interest, funding sources, data exclusions, transformations, and unsuccessful tests.

Ethical research does not mean hiding inconvenient results. Selective reporting, p-hacking, fabricated observations, plagiarism, and undisclosed changes to hypotheses undermine professional credibility. Reproducible files, version control, a data dictionary, and an audit trail improve trust.

Regulatory expectations vary by jurisdiction, but common principles include consumer protection, fair lending, market-abuse prevention, prudential resilience, anti-money-laundering controls, and responsible use of personal data. Artificial intelligence introduces additional concerns: opacity, unstable models, biased training data, adversarial manipulation, automation complacency, and unclear accountability. Human oversight should be meaningful rather than merely ceremonial.

Challenges and Limitations

Finance research faces several persistent difficulties:

• Data quality: prices may be missing, restated, stale, or affected by survivorship bias.
• Endogeneity: explanatory variables may be jointly determined with outcomes.
• Model risk: assumptions may fail during crises or structural change.
• Non-stationarity: relationships observed in one period may not persist.
• Measurement problems: constructs such as governance, innovation, and risk appetite are difficult to observe directly.
• External validity: results from one market, sector, or crisis may not generalise.
• Economic significance: statistical significance may not justify a costly decision.
• Forecast uncertainty: scenarios and valuations are conditional estimates, not facts.

Best practice is not to eliminate uncertainty but to expose it. Researchers should use robustness checks, alternative definitions, sensitivity analysis, out-of-sample testing, triangulation, and candid limitation statements.

Contemporary Developments and Employability

Contemporary finance increasingly combines accounting records with high-frequency market data, satellite or transaction data, natural-language disclosures, and machine-learning methods. These tools can improve prediction but may reduce interpretability and amplify hidden data errors. Machine learning should therefore be compared with transparent benchmarks, validated out of sample, and evaluated for fairness and stability.

Graduates who can connect financial theory with evidence are valuable in investment analysis, banking, risk, audit, valuation, treasury, consulting, regulation, and financial technology. Employers expect competence in spreadsheet modelling, databases, statistical software, data visualisation, report writing, professional judgement, and ethical communication. The most employable researcher can explain not only what a model predicts, but why the result is credible, what could invalidate it, and what decision should follow.

Chapter Summary

Finance research applies theories, models, market data, accounting information, and decision criteria to valuation, investment, corporate finance, capital markets, banking, fintech, and risk management. DCF and relative valuation provide complementary approaches to estimating value. Portfolio analysis highlights the relationship between return, volatility, covariance, and diversification. Corporate finance research examines investment, financing, governance, dividends, and working capital. Event studies assess market reactions, while banking and fintech research must address causality, inclusion, privacy, and model risk. Risk research uses measures such as VaR, expected shortfall, PD, LGD, and stress testing, but every measure depends on assumptions.

A credible dissertation aligns its research question, theory, variables, data, model, and decision context. It reports economic as well as statistical significance, tests robustness, discloses limitations, and treats ethical and regulatory responsibilities as part of research quality. Financial models are useful decision aids, not substitutes for judgement.

Key Terms

Abnormal return: The difference between an actual return and a model-based expected return.

Capital Asset Pricing Model: A model relating expected return to systematic market risk.

Cash conversion cycle: The time between paying suppliers and collecting cash from customers.

Discounted cash flow: Valuation based on the present value of expected future cash flows.

Economic significance: The practical magnitude and decision relevance of a result.

Expected shortfall: Average loss conditional on losses exceeding a specified threshold.

FCFF: Cash flow available to debt and equity providers after operating and investment needs.

Fintech: Technology-enabled innovation in financial products, services, and processes.

Hedging: Taking an offsetting position to reduce exposure to a risk.

Market efficiency: The extent to which prices incorporate available information.

Net present value: Present value of benefits minus the initial investment and other costs.

Portfolio diversification: Combining assets to reduce avoidable, asset-specific risk.

Probability of default: The estimated likelihood that a borrower will fail to meet obligations.

Relative valuation: Valuation using prices or multiples of comparable assets or firms.

Research design: The structured plan linking a question, evidence, method, and inference.

Systematic risk: Market-wide risk that cannot be eliminated through diversification.

Terminal value: Estimated value of cash flows beyond an explicit forecast period.

Value at Risk: A specified loss threshold over a stated horizon and confidence level.

WACC: Weighted average required return of debt and equity capital.

Review Questions

5 Short Answer Questions

1. What is the difference between FCFF and free cash flow to equity?
2. Why does diversification reduce unsystematic risk?
3. What is an event window in event-study research?
4. Define probability of default and loss given default.
5. Why must a valuation include sensitivity analysis?

5 Long Answer Questions

1. Explain the DCF valuation process, including FCFF, WACC, terminal value, assumptions, and limitations.
2. Compare CAPM with portfolio theory as frameworks for investment research.
3. Discuss how endogeneity can affect a study of fintech adoption and bank profitability.
4. Explain the design of an event study and identify major threats to valid inference.
5. Evaluate the ethical and regulatory issues involved in AI-based credit scoring.

3 Analytical / Application Questions

1. A project requires £500,000 immediately and produces £180,000 annually for four years. Explain what additional information is required before calculating NPV and deciding whether to invest.
2. Design a research project testing whether working-capital efficiency affects profitability in listed manufacturing firms. Specify the dependent variable, explanatory variables, controls, data period, and likely limitations.
3. A bank reports a high Sharpe ratio for a fund investing in illiquid assets. Analyse why the ratio may overstate performance and propose robustness checks.

Multiple Choice Questions with Answers

Question 1

Which valuation method discounts expected future cash flows using a risk-adjusted discount rate?

A. Discounted cash flow valuation

B. Book-value accounting

C. Horizontal analysis

D. Inventory turnover analysis

Correct Answer: A. Discounted cash flow valuation

Question 2

In the CAPM, beta primarily measures an asset’s sensitivity to:

A. Inflation

B. Market movements

C. Accounting profit

D. Short-term liquidity

Correct Answer: B. Market movements

Question 3

What does a positive NPV generally indicate under the stated discount rate?

A. The project destroys value

B. The project adds expected value

C. The project has zero risk

D. The project must have the highest IRR

Correct Answer: B. The project adds expected value

Question 4

Which component is subtracted in the cash conversion cycle formula?

A. Days inventory outstanding

B. Days sales outstanding

C. Days payable outstanding

D. Gross margin

Correct Answer: C. Days payable outstanding

Question 5

Abnormal return is calculated as actual return minus:

A. Book value

B. Expected model-based return

C. Dividend yield only

D. Transaction cost

Correct Answer: B. Expected model-based return

Question 6

Which risk cannot generally be eliminated through diversification?

A. Firm-specific risk

B. Operational error in one firm

C. Systematic market risk

D. Supplier risk in one firm

Correct Answer: C. Systematic market risk

Question 7

In the simple expected credit loss approximation, which three variables are multiplied?

A. PD, LGD, and EAD

B. ROE, beta, and WACC

C. VaR, volatility, and duration

D. DIO, DSO, and DPO

Correct Answer: A. PD, LGD, and EAD

Question 8

Why can a high R² be insufficient evidence that a finance model is useful?

A. R² is always negative

B. It proves causation too strongly

C. It does not establish causation or economic significance

D. It can only be calculated for portfolios

Correct Answer: C. It does not establish causation or economic significance

Question 9

Which issue occurs when an explanatory variable is jointly determined with the outcome?

A. Endogeneity

B. Diversification

C. Annualisation

D. Amortisation

Correct Answer: A. Endogeneity

Question 10

What is a principal limitation of VaR?

A. It cannot specify a confidence level

B. It does not describe the average size of losses beyond the threshold

C. It always assumes zero volatility

D. It measures only accounting profit

Correct Answer: B. It does not describe the average size of losses beyond the threshold