Investment-Bank Economics, Performance and Strategy

Introduction

Investment banking is often described through transactions—mergers, equity offerings, bond issues, derivatives, and market-making—but its economic substance is broader. An investment bank converts information, relationships, balance-sheet capacity, technology, and risk-bearing ability into financial services. Its performance therefore cannot be judged by revenue alone. A business may generate substantial fees while consuming excessive capital, exposing the firm to volatile risks, or failing to earn an adequate return for shareholders.

This chapter examines the economics and strategy of investment banking. It links revenue pools to costs, capital consumption, return on equity (ROE), and risk-adjusted performance. It also considers competitive positioning, talent, culture, regulation, technology, and strategic choices across market cycles. The central managerial question is not simply “How much revenue did the bank produce?” but “What risk-adjusted return did it generate, with what resources, and how sustainably?”

Learning Objectives

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

1. Explain the principal revenue pools of investment banks.
2. Distinguish gross revenue, net revenue, contribution profit, and economic profit.
3. Analyse the relationship among costs, balance-sheet usage, capital, and ROE.
4. Apply risk-adjusted performance measures to investment-banking businesses.
5. Evaluate competitive positioning, talent, culture, and operating models.
6. Assess strategic choices across expansion, contraction, diversification, and technology investment.
7. Identify governance, conduct, regulatory, and ethical risks in performance management.

1. The Economic Architecture of an Investment Bank

An investment bank generally combines several businesses:

• Advisory: mergers and acquisitions (M&A), restructuring, strategic advice, and fairness opinions.
• Origination: underwriting and distribution of equity and debt securities.
• Sales and trading: client execution, market-making, financing, derivatives, and securities lending.
• Research and data: investment research, market intelligence, analytics, and proprietary information products.
• Asset and wealth-related activities: private banking, asset management, and related distribution, where applicable.
• Technology and infrastructure services: electronic execution, analytics, workflow platforms, and post-trade services.

The businesses differ materially in revenue timing, scalability, risk, capital intensity, and talent requirements. Advisory revenue is usually transaction-based and comparatively asset-light, but cyclical. Market-making revenue may be recurring in the sense that clients trade continuously, but it requires inventory, funding, systems, and risk management. Underwriting combines fee income with market and commitment risk. A sound strategy therefore evaluates businesses as economic portfolios rather than as a single undifferentiated bank.

A simplified income identity is:

Net revenue should be interpreted carefully. Depending on accounting policy and business model, it may include trading gains and losses, interest income and expense, underwriting fees, advisory fees, commissions, and fair-value changes. Some businesses report revenue gross of client-related expenses, while others report net amounts. Comparisons require consistent definitions.

The fee-pool concept

A fee pool is the total amount of fees available in a market or product category during a period. For example, the global M&A fee pool depends on transaction value, fee rates, sector composition, geography, and the proportion of transactions requiring investment-bank advice. An individual bank’s addressable fee pool is smaller than the global pool because of its sector coverage, geography, client access, product capability, and conflicts policy.

A useful decomposition is:

The realisation rate captures the portion of nominal opportunity that becomes recognized revenue after fee discounts, syndication, refunds, failed transactions, revenue sharing, and other adjustments. This relationship is a managerial framework rather than a financial-reporting identity. It is most useful for forecasting advisory and underwriting revenues; it is less directly applicable to trading, where revenue depends on spreads, inventory returns, financing, client flow, and market conditions.

Major revenue pools

Revenue quality matters as much as revenue quantity. A recurring electronic-execution stream may be strategically more valuable than a large one-off transaction fee, even if the latter is larger in a single quarter. Conversely, apparently stable revenue may conceal substantial tail risk or dependence on a small number of clients.

2. Cost Economics and Operating Leverage

Investment-bank costs are commonly divided into compensation, non-compensation operating costs, and funding or balance-sheet costs.

Compensation

Compensation includes salaries, bonuses, benefits, deferred awards, and employment-related taxes. It is both a cost and a strategic investment. Senior bankers bring relationships and judgment; traders bring market expertise; engineers build scalable infrastructure; risk and compliance professionals protect the franchise. However, compensation is often semi-variable rather than fully variable. Employment contracts, minimum staffing, deferred compensation, and retention arrangements limit short-term flexibility.

A useful measure is the compensation ratio:

The ratio is dimensionless and is meaningful only when revenue and compensation are measured over the same period and under consistent accounting definitions. A low ratio is not automatically superior: it may indicate productivity, but it may also reflect underinvestment in talent or an unusually strong revenue year.

Non-compensation costs

These include technology, premises, communications, data, professional services, travel, marketing, insurance, legal expenses, and regulatory infrastructure. Technology costs often have a fixed or step-fixed character. Once a platform is built, additional transactions may be processed at low marginal cost; however, cyber controls, data licensing, cloud usage, resilience, and model governance can increase with scale.

Operating leverage

Operating leverage describes how changes in revenue affect operating profit because costs do not move proportionately with revenue. Let revenue be R, variable cost be vR, and fixed cost be F, where 0 ≤ v < 1. Then:

The degree of operating leverage at a given revenue level is:

This formula assumes a linear cost relationship, constant prices and variable-cost rates, and a relevant operating range. A higher fixed-cost burden makes profits more sensitive to revenue changes. Operating leverage magnifies gains in strong markets but magnifies losses in weak markets.

Worked example. Suppose annual revenue is £500 million, variable costs are 40% of revenue, and fixed costs are £200 million. Operating profit is:

(1 − 0.40) × 500 − 200 = £100 million

The degree of operating leverage is:

(500 × 0.60) ÷ 100 = 3.0

Approximately, a 1% change in revenue would produce a 3% change in operating profit, assuming the cost structure remains within the relevant range. This is an approximation, not a forecast of actual earnings, because compensation decisions, risk losses, and business-mix changes may alter costs.

3. Capital, Balance Sheet, and Risk Capacity

Capital is not the same as cash, assets, or funding. A bank may hold assets funded by deposits, secured borrowing, unsecured debt, and equity. Regulatory capital is designed to absorb losses and protect creditors and the financial system. Economic capital is an internal estimate of capital needed for unexpected losses at a selected confidence level and time horizon.

Capital consumption may arise from:

• trading inventories and market-risk exposures;
• counterparty credit exposure from derivatives and financing;
• underwriting commitments and bridge loans;
• settlement and clearing exposures;
• operational, legal, conduct, and model risks;
• leverage and liquidity requirements.

Risk-weighted assets (RWA) are a regulatory construct that applies prescribed or approved risk weights to exposures. A simplified capital ratio is:

The ratio is meaningful only when the numerator and denominator follow the same regulatory framework, scope of consolidation, and reporting date. It is not a complete measure of risk: market liquidity, model error, concentration, and stress losses may not be fully captured.

Capital has an opportunity cost. If one business uses scarce capital, that capital cannot simultaneously support another business. Strategic decisions should therefore compare expected return with capital consumption, liquidity usage, operational complexity, and downside exposure.

Return on equity

ROE relates profit attributable to shareholders to the equity deployed:

Average equity is commonly calculated as opening plus closing equity divided by two, although more frequent averages may be appropriate for volatile balance sheets. ROE is an accounting performance measure; it is not automatically a risk-adjusted return.

A simplified DuPont-style decomposition is:

The first term is profit margin, the second is asset turnover, and the third is an equity multiplier. The decomposition is algebraically valid when the quantities are defined consistently. In investment banking, however, revenue-to-assets can be difficult to interpret because off-balance-sheet exposures, derivatives, and fair-value accounting complicate the asset base.

Worked example. A division earns £180 million of profit and uses average common equity of £1.5 billion:

ROE = 180 ÷ 1,500 = 12%

If the firm’s hurdle rate for that risk profile is 10%, the division appears to create accounting value before considering risk-adjustment, liquidity charges, taxes not included in the measure, and strategic externalities. If its earnings are highly volatile or exposed to severe tail losses, a 12% ROE may not be attractive.

Economic profit and RAROC

Economic profit can be expressed as:

The hurdle rate should reflect the opportunity cost of equity capital and the risk of the activity. With £180 million profit, £1.5 billion capital, and a 10% hurdle rate:

£180m − (£1,500m × 0.10) = £30m

Risk-adjusted return on capital (RAROC) is commonly expressed as:

Risk-adjusted return may subtract expected loss, funding costs, liquidity charges, taxes, and other direct costs from revenue. Definitions vary across institutions. RAROC is therefore a management metric, not a universally standardized accounting number. It is most useful for comparing activities under a common methodology and should be supplemented with stress testing and qualitative judgment.

4. Risk-Adjusted Performance Across Businesses

A credible performance system considers at least five dimensions:

1. Revenue and profit: What was earned after direct and allocated costs?
2. Capital efficiency: How much equity, RWA, liquidity, and balance-sheet capacity was consumed?
3. Risk: What market, credit, operational, legal, conduct, and reputational risks were taken?
4. Durability: Is performance recurring, client-based, and defensible, or dependent on unusual market conditions?
5. Strategic value: Does the business strengthen relationships, data, distribution, or future capabilities?

A performance scorecard can include risk-adjusted profit, ROE, RAROC, stress losses, client retention, control quality, employee turnover, and conduct indicators. Metrics should not be mechanically aggregated without judgment. A trader who generates high current revenue by accumulating hidden tail risk should not be rewarded as a superior performer.

Stress testing

Stress testing examines performance under adverse but plausible conditions. Scenarios may include a sharp rise in interest rates, widening credit spreads, equity-market declines, a collapse in issuance volume, a major counterparty default, or a cyber incident.

A simple scenario change can be represented as:

This is a conceptual equation. It becomes a quantitative model only when the shock, behavioural responses, correlations, and loss assumptions are specified. Scenario analysis is particularly important for businesses whose historical volatility understates potential losses because the sample period lacked a severe crisis.

5. Competitive Positioning and Business Models

Competitive advantage in investment banking rarely depends on one factor. It is usually a system of reinforcing capabilities:

• trusted client relationships;
• sector and product expertise;
• distribution and balance-sheet capacity;
• high-quality research and information;
• execution reliability;
• technology and data;
• global or regional network effects;
• strong risk and compliance infrastructure;
• a credible brand.

Universal banks, boutiques, and specialist platforms

Large universal banks can offer integrated products, global distribution, financing, and cross-border execution. Their disadvantages may include complexity, higher fixed costs, slower decision-making, and regulatory constraints. Independent boutiques may offer senior attention, specialized advice, and fewer conflicts, but generally have less balance-sheet capacity and product breadth. Electronic and specialist platforms can compete through speed, transparency, data, automation, and low marginal cost.

The appropriate strategy depends on client segment and product economics. A bank should not imitate a competitor merely because the competitor is larger. It should identify where it has a defensible right to win and where the required risk-adjusted returns justify investment.

Market share versus economic value

Market share can be a useful indicator of franchise strength, but pursuing share through excessive fee discounting may destroy value. A bank should distinguish:

• headline share: reported volume or league-table position;
• profitable share: volume that earns an adequate risk-adjusted return;
• strategic share: business that deepens valuable relationships or enables future activity.

A low-margin mandate may be rational if it creates substantial future value, but such claims should be tested rather than used to rationalize uneconomic growth.

6. Talent, Culture, and Incentives

Investment banking is a knowledge-intensive business. Human capital is embodied in relationships, judgment, technical skill, credibility, and institutional memory. Talent management therefore affects both revenue and risk.

Important talent dimensions include:

• recruiting and technical training;
• apprenticeship and succession planning;
• diversity of experience and perspective;
• mobility between product, coverage, risk, and technology teams;
• retention of high performers without tolerating misconduct;
• effective collaboration across functions and jurisdictions.

Culture is the set of informal norms that influence how people act when rules are incomplete or supervision is absent. A culture focused only on revenue can produce mis-selling, conflicts of interest, concealment of losses, manipulation of benchmarks, and excessive risk-taking. A healthy culture aligns commercial ambition with client duty, market integrity, challenge, escalation, and accountability.

Compensation design should include deferral, malus, clawback where legally enforceable, risk-adjusted measures, and non-financial conduct criteria. These mechanisms are imperfect. Deferred compensation may encourage retention but can also create perceived unfairness; clawbacks may be difficult to enforce; short-term metrics may still dominate behaviour. Governance must therefore combine incentives with supervision, independent control functions, and credible consequences.

7. Strategic Choices Across the Market Cycle

Investment banking is cyclical. M&A and equity issuance tend to weaken when valuation uncertainty, financing costs, or recession fears rise. Debt markets can be active during refinancing needs but may close during credit stress. Trading may benefit from volatility and client repositioning, although liquidity crises can produce severe losses. The strategic challenge is to avoid building a cost base on the assumption that peak conditions will persist.

Expansion

Expansion may involve hiring teams, entering new geographies, adding products, acquiring technology, or increasing underwriting capacity. It is justified when the addressable opportunity, competitive advantage, risk controls, and expected risk-adjusted return support the investment.

Retrenchment

Retrenchment may involve reducing headcount, exiting products, shrinking inventories, or selling businesses. It can protect capital and improve returns, but poorly managed exits may damage clients, morale, and future franchise value.

Diversification

Diversification can reduce dependence on a single fee pool, but it may increase complexity and dilute management attention. Correlations often rise in crises, so apparent diversification based on normal-period data may fail under stress.

Countercyclical investment

A strong institution may invest selectively during downturns, when talent and assets become available at lower cost. This requires liquidity, capital resilience, and confidence that the firm can sustain investment before revenue recovers. Countercyclical investment is not synonymous with indiscriminate expansion.

8. Technology, Data, and Artificial Intelligence

Technology changes the economics of execution, research, surveillance, risk measurement, and client coverage. Automation can lower marginal processing costs, reduce errors, and improve scalability. However, technology also introduces cyber risk, model risk, vendor dependency, data-quality problems, and supervisory challenges.

Artificial intelligence may support document review, comparable-company analysis, client intelligence, coding, surveillance, and workflow automation. Its use should be governed by:

• approved data sources and privacy controls;
• explainability proportionate to the use case;
• human review for material decisions;
• testing for bias, hallucination, and model drift;
• audit trails and access controls;
• clear accountability for outputs.

AI-generated analysis must not be treated as reliable merely because it is fluent. In investment banking, an erroneous statement about a client, transaction, security, or regulation can create financial, legal, and reputational harm.

9. Regulation, Ethics, and Governance

Investment banks operate under overlapping requirements concerning capital, liquidity, market conduct, conflicts, client suitability, disclosure, anti-money laundering, sanctions, privacy, recordkeeping, and operational resilience. Regulatory compliance is not simply a cost centre. It protects the licence to operate and can be a source of competitive trust.

Common governance risks include:

• conflicts between advisory, underwriting, trading, and research activities;
• information leakage and insider trading;
• inaccurate valuation or disclosure;
• unsuitable products and aggressive sales practices;
• benchmark or market manipulation;
• inadequate segregation of duties;
• pressure to override controls during high-revenue periods.

The three-lines model—business ownership of risk, independent risk and compliance oversight, and internal audit assurance—can clarify responsibilities, although effectiveness depends on authority, expertise, independence, and board engagement. Governance should also address third-party technology, model validation, climate and transition risks where material, and operational resilience.

10. Illustrative Case: Northbridge Investment Bank

This is a fictional case designed for educational purposes.

Background

Northbridge Investment Bank is a mid-sized institution with strong domestic M&A and equity-underwriting franchises. It also operates a rates-trading business and a small electronic execution platform. During a period of low interest rates and high transaction activity, management hired senior bankers, expanded technology capacity, and increased underwriting commitments.

Problem

Market activity subsequently declined. Annual net revenue fell from £1,000 million to £820 million. Compensation and non-compensation costs were initially slow to adjust. The board faced three choices: preserve the expanded platform, reduce costs aggressively, or reallocate capital toward businesses with stronger risk-adjusted returns.

The following simplified data are available:

The operating-profit calculation is Revenue − 0.40(Revenue) − £420m. Prior-year profit is 1,000 − 400 − 420 = £180m. Current-year profit is 820 − 328 − 420 = £72m. ROE falls from 180 ÷ 1,500 = 12% to 72 ÷ 1,500 = 4.8%.

Analysis

The business has high operating leverage because fixed costs are large relative to contribution profit. The decline in revenue of 18% produces a 60% decline in operating profit. The board should not conclude that all expansion was wrong; some technology and relationship investments may have long-term value. It should instead identify which capabilities remain strategically defensible and which consume capital without adequate returns.

Northbridge’s advisory franchise has low direct capital usage and high client retention. Its rates-trading business produces attractive revenue in volatile periods but consumes substantial risk limits and requires strong controls. The electronic platform has high initial technology costs but could become scalable if client adoption reaches a viable level.

Approach and solution

Management develops a two-stage plan:

1. Protect core advisory relationships and retain high-performing teams with strong conduct records.
2. Reduce low-return underwriting commitments and impose transaction-level hurdle rates.
3. Reprice balance-sheet-intensive activities to reflect funding, liquidity, and capital costs.
4. Consolidate duplicated technology infrastructure while preserving resilience and cyber controls.
5. Introduce a scorecard combining economic profit, stress losses, client outcomes, employee conduct, and strategic milestones.
6. Conduct downside scenarios in which revenue falls by a further 15% and volatility rises sharply.
7. Invest selectively in electronic execution only where client demand and unit economics are demonstrated.

Outcome

After two years, Northbridge has a smaller but more resilient platform. Advisory market share is stable, underwriting volume is lower but more profitable, and the electronic business reaches break-even in its targeted client segment. Trading limits are more closely linked to stressed liquidity and counterparty exposures. ROE improves, not because revenue returns to its peak, but because capital and fixed costs are better aligned with the firm’s competitive strengths.

Learning points

The case illustrates five principles. First, revenue growth can destroy value if it is purchased through uneconomic pricing. Second, fixed costs make downturns especially damaging. Third, capital allocation should incorporate stress and liquidity, not only average-period returns. Fourth, technology investments require measurable strategic and economic hypotheses. Fifth, culture and control quality are part of performance, not obstacles separate from it.

11. Limitations of Performance Analysis and Best Practices

Performance metrics can mislead when:

• revenue is allocated inconsistently across products;
• shared technology and control costs are ignored;
• capital models underestimate tail risk;
• current-period profit receives more weight than deferred losses;
• transfer prices do not reflect funding and liquidity scarcity;
• strategic benefits are asserted without evidence;
• peer comparisons ignore business mix and accounting differences.

Best practice is to use a portfolio of measures and examine trends across a cycle. Management should reconcile reported results to management accounts, document allocation methodologies, validate risk models, challenge assumptions, and compare base-case returns with stress outcomes. Boards should ask whether reported performance is repeatable, whether risks are transparent, and whether incentives support prudent client and market behaviour.

Chapter Summary

Investment-bank economics links fee pools, market share, pricing, costs, capital, risk, and strategy. Advisory and underwriting revenues are driven by transaction activity and client franchise, while trading revenues depend on client flow, spreads, volatility, financing, inventory, and risk capacity. Revenue must be evaluated after compensation, technology, funding, liquidity, and control costs.

Operating leverage explains why fixed-cost platforms can produce strong gains in expansionary markets and severe earnings declines in downturns. ROE measures profit relative to equity, but economic profit and RAROC provide more useful perspectives when capital and risk differ across businesses. Capital allocation should include stress testing, liquidity, tail risk, and opportunity cost.

Competitive advantage depends on relationships, expertise, distribution, technology, data, talent, and trust. Sustainable strategy requires disciplined choices about market share, diversification, expansion, retrenchment, and countercyclical investment. Culture, incentives, governance, and regulation are central economic variables because misconduct and uncontrolled risk can destroy franchise value. Technology and AI can improve scalability and insight, but only with strong data, model, cyber, privacy, and human-accountability controls.

Key Terms

Fee pool: Total fees available in a market, product, or transaction category.

Realisation rate: Proportion of nominal business opportunity converted into recognized revenue.

Operating leverage: Sensitivity of operating profit to changes in revenue caused by fixed costs.

Variable cost: Cost that changes approximately with activity or revenue.

Fixed cost: Cost that does not change proportionately with short-term activity.

Capital consumption: Equity, regulatory capital, risk-weighted assets, liquidity, or limits used by a business.

Risk-weighted assets: Regulatory measure that applies risk weights to exposures.

ROE: Profit attributable to common shareholders divided by average common equity.

Economic profit: Profit after deducting the opportunity cost of capital.

Hurdle rate: Minimum required return for a specified risk and capital usage.

RAROC: Risk-adjusted return divided by economic capital.

Expected loss: Average anticipated loss over a defined exposure and time horizon.

Stress testing: Analysis of performance under adverse but plausible scenarios.

Market-making: Providing executable prices and liquidity to clients while managing resulting exposures.

Operating model: Arrangement of people, processes, technology, governance, and resources used to deliver services.

Franchise value: Long-term economic value of client relationships, reputation, capabilities, and market access.

Conduct risk: Risk of harm arising from inappropriate behaviour toward clients, markets, or other stakeholders.

Operating resilience: Ability to continue critical services during disruption.

Countercyclical investment: Investment made during downturns to build capability for future recovery.

Review Questions

Short Answer Questions

1. What is a fee pool, and why is the realisation rate important?
2. Distinguish ROE from RAROC.
3. Identify three sources of capital consumption in investment banking.
4. Why can high market share fail to create economic value?
5. State two governance risks associated with artificial intelligence in investment banking.

Long Answer Questions

1. Explain how revenue pools, operating leverage, and capital consumption jointly determine investment-bank economics.
2. Compare the strategic advantages and disadvantages of a universal bank, an independent boutique, and an electronic specialist.
3. Discuss how compensation and culture can influence both performance and risk.
4. Explain why stress testing should supplement ROE and accounting profit.
5. Evaluate the strategic choices available to an investment bank during a prolonged downturn.

Analytical / Application Questions

1. A division earns £240 million after direct costs and uses £2 billion of average common equity. Calculate its ROE. If the hurdle rate is 9%, calculate its economic profit and interpret the result.
2. An investment bank has revenue of £600 million, variable costs equal to 35% of revenue, and fixed costs of £300 million. Calculate operating profit and explain the likely effect of a 10% revenue decline, assuming the cost structure is unchanged.
3. Design a balanced performance scorecard for a trading and underwriting platform. Justify the financial, risk, client, conduct, talent, and strategic measures you would include.

Multiple Choice Questions with Answers

Question 1

Which factor most directly expands an investment bank’s addressable advisory fee pool?

A. Higher employee turnover

B. Greater relevant transaction activity in its target sectors and geographies

C. Lower regulatory capital

D. A wider internal reporting hierarchy

Correct Answer: B. Greater relevant transaction activity in its target sectors and geographies

Question 2

A bank has revenue of £500 million, variable costs equal to 40% of revenue, and fixed costs of £200 million. What is operating profit?

A. £100 million

B. £200 million

C. £300 million

D. £500 million

Correct Answer: A. £100 million

Question 3

In the simplified operating-leverage example, a degree of operating leverage of 3 implies that, approximately, a 1% revenue change produces what profit change, holding assumptions constant?

A. 0.33%

B. 1%

C. 3%

D. 30%

Correct Answer: C. 3%

Question 4

Which statement best describes economic profit?

A. Revenue minus variable costs only

B. Profit minus the opportunity cost of capital

C. Assets minus liabilities

D. Regulatory capital divided by risk-weighted assets

Correct Answer: B. Profit minus the opportunity cost of capital

Question 5

A division earns £180 million and uses average common equity of £1.5 billion. What is its ROE?

A. 8.3%

B. 10.0%

C. 12.0%

D. 18.0%

Correct Answer: C. 12.0%

Question 6

Which activity is generally most directly associated with inventory, liquidity, and market-risk requirements?

A. Market-making

B. Human-resources administration

C. Corporate premises management

D. Basic archival storage

Correct Answer: A. Market-making

Question 7

Why can a low compensation ratio be misleading?

A. It always proves that controls are weak

B. It may reflect unusually high revenue or underinvestment in talent

C. It cannot be calculated mathematically

D. It measures regulatory capital directly

Correct Answer: B. It may reflect unusually high revenue or underinvestment in talent

Question 8

What is the principal purpose of stress testing?

A. To guarantee a specific future profit

B. To replace all accounting measures

C. To examine performance under adverse but plausible conditions

D. To eliminate the need for risk governance

Correct Answer: C. To examine performance under adverse but plausible conditions

Question 9

Which is a potential advantage of an independent investment-banking boutique?

A. Unlimited balance-sheet capacity

B. Guaranteed global distribution

C. Specialized advice and potentially greater senior attention

D. Absence of all conflicts and risks

Correct Answer: C. Specialized advice and potentially greater senior attention

Question 10

Which control is most appropriate when using AI for material investment-banking analysis?

A. Accepting fluent outputs without verification

B. Removing all human review

C. Using approved data, audit trails, testing, and proportionate human oversight

D. Allowing unrestricted access to confidential client data

Correct Answer: C. Using approved data, audit trails, testing, and proportionate human oversight