Type I and Type II errors in hiring and continuation

Type I and Type II errors in hiring and continuation

Published September 28, 2026

Type I and Type II errors in hiring and continuation

On this page

Learning Outcome Statement

Contrast Type I and Type II errors in manager hiring and continuation decisions.

On the exam: The vignette gives a hiring or continuation decision with noisy performance and process evidence. It asks you to contrast the applicable error types in a constructed response or selected-response question. Read the action first, then compare it with the manager's hidden skill state, because the same noisy performance evidence can lead to either a correct decision or an error.

You are advising an investment committee that must decide whether to hire a manager or keep allocating capital to one. The evidence is noisy, so true skill is not visible when the committee acts. A positive decision means hire or continue, while a negative decision means reject or terminate. A threshold is the amount of evidence the committee requires before making a positive decision.

Decision stageType I errorType II error
HiringHire an unskilled managerReject a skilled manager
ContinuationContinue an unskilled managerTerminate a skilled manager
More permissive thresholdType I risk generally risesType II risk generally falls
Stricter thresholdType I risk generally fallsType II risk generally rises

Hiring and continuation use the same error logic, but the action changes with the stage. A more permissive threshold accepts weaker evidence for a positive decision, so Type II risk generally falls while Type I risk rises.

Use $D^+$ for the positive action of hiring or continuing. Use $D^-$ for the negative action of rejecting or terminating. The hidden state is skilled, $S$, or unskilled, $U$.

$$ \begin{aligned} \text{Type I error} &= \text{false positive} = D^+ \text{ with } U \ \text{Type II error} &= \text{false negative} = D^- \text{ with } S \end{aligned} $$

The Type I line pairs $D^+$ with $U$. The positive action treats the manager as skilled, while $U$ makes that judgment false. The Type II line pairs $D^-$ with $S$, so the rejection or termination is false. Other action-state pairs are correct decisions. Use the action branch before the state branch.

Worked example

You are deciding whether to continue a manager. Recent performance looks weak, but process evidence supports skill, and the case later reveals that the manager is skilled. You recommend termination. The question asks you to identify the error and contrast it with continuing the same manager.

GivenCase fact
Decision stageContinuation
Action takenTerminate manager
Performance evidenceWeak recent performance
Process evidenceSupports skill
Skill stateSkilled manager

The table pins down the case before you classify it: continuation is the stage, termination is the negative action, and the hidden state is skilled. Weak recent performance is evidence the committee considered, while process evidence points toward skill; neither observation changes the error label by itself. Follow the negative action branch first, then check the true state. This order matters because a skilled manager can still be terminated after weak results, making the decision a false negative even when the evidence looked discouraging. The flowchart lays out those two steps for either decision.

Loading diagram…
  1. Classify the stage and action. Identify this as a continuation decision with termination, because the action branch records whether you reject or accept the manager. The result is a negative decision, $D^-$.
  2. Classify the state. Identify the manager as skilled, because the case gives the true skill state rather than asking you to classify recent performance. The result is $S$.
  3. Pair the labels. Combine $D^-$ with $S$, because the error label depends on both the decision and the true state. The result is Type II, a false negative.
  4. Run the contrast. Replace termination with continuation while keeping the same skilled manager, because the comparison must change the action but hold the state fixed. The result is $D^+$ with $S$, a correct positive rather than an error.

The termination is a Type II error because you rejected a skilled manager, and weak performance did not determine the label.

Changed condition

ElementBeforeAfter
ActionTerminate managerContinue manager
Skill stateSkilled managerSkilled manager
ClassificationType II errorCorrect positive

Your prediction before inspecting the classification row should be that the error disappears. Only the action moves from negative to positive, while the skill state stays skilled. The pair moves from Type II to correct positive because the positive action matches the skilled state. Before reading on, write the action and the unobserved skill state for a new case before naming its error.

Common error

You may label any bad realized outcome a Type I error and any good realized outcome a Type II error. That feels right because you can see performance while the manager’s skill stays hidden, but the committee’s decision still matters: did it hire or continue the manager, or reject or terminate the manager? The missing action-state pair exposes the mistake. Type I requires the committee to take a positive action with an unskilled manager, while Type II requires it to take a negative action with a skilled manager.

Check yourself

Practice

Which statement correctly contrasts the effect of a more permissive decision threshold?

Revision snapshot

$$ \begin{aligned} \text{Type I error} &= D^+ \text{ with } U \ \text{Type II error} &= D^- \text{ with } S \end{aligned} $$

RuleStatement
Type I errorHire or continue an unskilled manager, false positive.
Type II errorReject or terminate a skilled manager, false negative.
ClassificationState action and unobserved skill state, not performance.
ThresholdMore permissive lowers Type II risk, raises Type I risk.

On the exam, identify hire or continue versus reject or terminate, then identify the unobserved skill state. Hiring and continuation use the same error logic, but the decision action differs. Noisy performance is evidence, not the error label.

Unlock the full lesson and continue reading

Signup to continue reading this lesson and unlimited access questions, flashcards, AI notes, and more

Scan to download app

Scan to download
UNLOCK FREE ACCESS

Practice Questions: Type I and Type II errors in hiring and continuation

  1. 1. The Halcyon Pension Plan is selecting an active domestic equity manager to complement a low-cost passive core. The plan will allocate USD 500 million immediately. Its investment committee seeks downside protection and has established two decisive quantitative constraints: the selected manager must have a down-market capture ratio below 85%, and total strategy assets after Halcyon's funding must not exceed the manager's estimated capacity. The committee also prefers repeatable performance, stable investment personnel, independent asset verification, and incentives based on longer-term risk-adjusted results. It will not select a manager with a material unresolved operational weakness, even if the manager has attractive investment performance. Exhibit 1 summarizes quantitative information calculated using the same benchmark and measurement periods for all candidates. Capacity estimates refer to the maximum strategy assets that each manager believes can be invested without materially changing implementation or increasing market-impact costs. Exhibit 2 summarizes on-site due diligence. The committee recognizes that historical performance may be less relevant when the personnel or discretionary process responsible for that performance is no longer present. It also distinguishes investment due diligence from operational due diligence. Investment due diligence focuses on philosophy, process, people, portfolio construction, performance sources, and capacity. Operational due diligence examines areas such as valuation, custody, reconciliation, trade allocation, cybersecurity, business continuity, and segregation of duties. The plan's governance policy states that a manager's compensation should encourage persistent risk-adjusted performance and should not primarily reward asset gathering or short-term gross returns. The policy does not require the manager to have the highest historical excess return. Instead, the committee must evaluate whether past results are consistent with the stated process, whether the process is likely to remain intact, and whether the manager can implement the mandate after receiving the planned allocation. After reviewing the candidates, the committee asks its analyst to identify the manager that satisfies the quantitative constraints, assess whether any personnel change undermines track-record relevance, identify the most important operational remediation, compare the proposed compensation arrangements, classify due-diligence findings as operational or investment-related, and confirm consistency of compensation practices with governance policy.

    • 2. The Helix Foundation is selecting an active domestic equity manager for a long-term allocation. Its investment committee requires positive benchmark-relative performance but will not appoint a manager whose organization or operating controls create an unacceptable probability of loss. The mandate will represent a meaningful allocation, so the committee also considers whether strategy assets are approaching the level at which trading costs, liquidity constraints, or reduced opportunity breadth could impair future results. The search consultant has completed an initial quantitative review of Alder Partners and Birch Capital. Exhibit 1 reports annualized figures calculated over the same seven-year period using the same investable benchmark and factor model. Estimated alpha is the intercept from the consultant's multifactor regression, and residual risk is the annualized standard deviation of regression residuals. The consultant notes that a manager whose active return is positive in only a minority of rolling three-year periods provides less persuasive evidence of repeatable skill than one whose active return is positive across most periods. The committee does not treat quantitative rankings as sufficient for appointment. Exhibit 2 summarizes on-site operational due-diligence findings. Committee policy requires independent reconciliation, controlled valuation overrides, tested business-continuity procedures, and sufficient depth of investment decision-making. A candidate with material deficiencies may remain under consideration, but appointment must be deferred until the deficiencies are remediated and independently confirmed. Lower fees do not compensate for an operating structure that fails these minimum standards. The consultant also analyzes capacity. Birch Capital is close to its internally estimated maximum strategy capacity, whereas Alder Partners retains more room for asset growth. Both estimates were prepared using expected market liquidity, position-size limits, and trading-cost assumptions. The committee recognizes that a manager approaching capacity may have difficulty implementing its historical process even if its past return and volatility remain attractive. Alder proposes the fee arrangement in Exhibit 3. The performance fee is assessed only on gross benchmark-relative return above the stated hurdle, and the base fee is charged regardless of performance. The committee asks the consultant to estimate expected active return after both fees using the expected gross active return shown in the exhibits.

      • 3. The Meridian Foundation is searching for an active equity manager. The allocation will be evaluated as a component of a diversified total portfolio, but the manager must satisfy a clearly defined mandate. The investment policy requires a global quality-growth strategy, a long-term orientation, and expected tracking error no greater than 6%. The foundation is willing to accept moderate active risk but does not want a concentrated portfolio whose results are dominated by a small number of holdings. The consultant initially reviews the three candidates in Exhibit 1. Capture ratios were calculated using the same monthly return observations and the same global equity benchmark. An up-market capture ratio above 100% indicates that a manager gained more than the benchmark during positive benchmark months. A down-market capture ratio below 100% indicates that the manager lost less than the benchmark during negative benchmark months. The consultant notes that capture ratios are useful diagnostics but should not override mandate fit, organizational quality, or operational findings. The foundation then conducts operational due diligence. It reviews custody, administration, valuation, cash controls, business continuity, trade-error procedures, cybersecurity, and the independence of service providers. Exhibit 2 summarizes one finding for each manager. No fraud has been alleged, but the committee treats conflicts involving asset valuation and investor reporting as potentially material even when historical returns appear satisfactory. Cedar remains under consideration after the investment and operational reviews. Its proposed fee agreement is described in Exhibit 3. All fee percentages are applied to beginning-of-year assets, and the benchmark and portfolio have no external cash flows during the year. The foundation asks the consultant to calculate the return above the benchmark after all stated fees. Taxes, transaction costs, and other expenses are excluded from this particular comparison.

        • 4. A fund earns 2.00% gross each quarter and incurs a 0.25% quarterly management fee. If each quarterly net return is 1.75%, the calculated annual net return is closest to:

          • A. 7.00%
          • B. 7.19%
          • C. 8.00%
          Show answer

          Answer: B. 7.19%

          7.19% - The quarterly net return is 1.75% after the 0.25% management fee. - Geometrically linking four quarters gives \((1.0175)^4-1=7.1859\%\), which rounds to 7.19%. 7.00% - Adding the four net quarterly returns gives \(4\times1.75\%=\)7.00%. - That arithmetic sum ignores the compounding required to calculate the annual net return. 8.00% - Adding the four gross quarterly returns gives \(4\times2.00\%=\)8.00%. - This omits the quarterly management fee, so it cannot be the annual net return.

        • 5. The Ashford Foundation is searching for an active US equity manager for a USD 300 million allocation. The foundation maintains separate policy allocations to small-cap value and mid-cap core equities. It accepts tracking error when consistent with a manager's stated process, but it requires evidence that reported results are representative, repeatable, and generated by the team that will manage Ashford's assets. The investment staff shortlists Northbank Equity and Summit Partners. Exhibit 1 summarizes quantitative and organizational information. Returns are net of management fees and measured over identical periods against the same broad equity benchmark. The staff recognizes that the benchmark is useful for an initial comparison but will evaluate each manager against the style exposure for which that manager is being considered. Exhibit 2 summarizes on-site due diligence. The foundation's due-diligence policy requires assessment of performance-record construction, personnel stability, investment-process consistency, capacity, trading controls, compliance oversight, and business continuity. A manager is not selected solely because it has the highest historical return. Northbank explains that its concentrated strategy has historically benefited from building meaningful positions in less-liquid small-cap issuers. The firm estimates that execution costs and market impact will increase materially if strategy assets exceed USD 5.0 billion. Ashford's proposed allocation would be invested immediately and would not replace an existing client account. Northbank has recently discussed closing the strategy but has not adopted a formal closing policy. The foundation is also concerned with record representativeness. An account's termination does not automatically make its prior results irrelevant, particularly when the account was managed using the strategy under evaluation. Similarly, a long strategy history is less persuasive when the current decision makers were not responsible for much of it. The committee therefore plans to distinguish a strong numerical record from evidence that the record is attributable to the present team and can be reproduced at the strategy's current asset level. The committee also intends to quantify, rather than merely describe, how close a prospective allocation would bring each candidate to its estimated capacity limit before finalizing any hiring recommendation.

          • 6. A portfolio returned 10%, its minimum acceptable return was 5%, its risk-free rate was 4%, its standard deviation was 4%, and its downside deviation was 6%. The calculated value of the Sharpe ratio is closest to:

            • A. 1.00
            • B. 1.25
            • C. 1.50
            Show answer

            Answer: C. 1.50

            1.50 - The Sharpe ratio uses the return above the risk-free rate divided by standard deviation. - Here, (10% − 4%) / 4% = 1.50. 1.00 - Using downside deviation instead of standard deviation gives (10% − 4%) / 6% = 1.00. - Downside deviation is used for the Sortino ratio, not the Sharpe ratio. 1.25 - Subtracting the minimum acceptable return instead of the risk-free rate gives (10% − 5%) / 4% = 1.25. - The Sharpe ratio requires the 4% risk-free rate, not the 5% minimum acceptable return.

          • 7. A consultant's hedge fund database removes liquidated funds but retains the full records of surviving funds. Its reported median performance is most likely:

            • A. Downward biased by backfill bias
            • B. Unbiased because reporting is voluntary
            • C. Upward biased by survivorship bias
            Show answer

            Answer: C. Upward biased by survivorship bias

            Upward biased by survivorship bias - Removing liquidated funds while retaining the full records of surviving funds creates survivorship bias because the database excludes funds that did not survive. - If liquidated funds generally performed worse, excluding them makes the reported median performance most likely too high. Downward biased by backfill bias - The question describes the removal of liquidated funds, not backfill bias, which arises when a fund adds returns from before it entered the database. - Backfilling favorable earlier returns generally biases reported performance upward, not downward. Unbiased because reporting is voluntary - Voluntary reporting does not undo the exclusion of liquidated funds, so the database does not represent all funds. - Managers’ decisions to report can also create self-selection bias, rather than ensure an unbiased median.

          • 8. The Northstar Foundation is selecting an active global equity manager for a USD 200 million allocation. The allocation is intended to reduce the foundation's losses during broad equity-market declines while generating positive long-term active return. The investment policy statement requires an expected tracking error no greater than 6%, and the selected manager must have at least USD 200 million of available strategy capacity. The foundation prefers stable investment teams and effective operational controls because it has limited internal resources for manager oversight. The investment committee identifies three finalists. Exhibit 1 summarizes quantitative and qualitative due-diligence findings. Expected alpha estimates are forward-looking and are based on the foundation's capital market and manager-skill assumptions. Maximum drawdown and capture ratios were calculated over identical market cycles using the same global equity benchmark. The committee treats a lower down-market capture ratio as preferable when other considerations are equal. One committee member proposes selecting the manager with the highest expected gross alpha. The chief investment officer responds that manager selection must also consider mandate compliance, downside behavior, capacity, organizational stability, and whether the manager's investment process can reasonably be expected to continue. The chief investment officer will not waive the tracking-error limit. The committee separately reviews the effect of fees on Brook's expected value added. The fee information in Exhibit 2 relates to a different Brook strategy under consideration for a smaller account. Expected gross alpha is stated after expected trading costs but before investment management fees. There is no performance fee or applicable tax adjustment. Northbank Partners, another manager with attractive historical returns, is reviewed as part of the foundation's operational due diligence. Exhibit 3 describes weaknesses identified during an on-site visit. Northbank has not experienced a reported client loss, and its executives state that revised controls will be installed within six months. The foundation's operational policy, however, requires effective segregation of client assets, independent custody, and separation between trade authorization, cash movement, and account reconciliation before an allocation is funded. The committee agrees to document its rationale for each finalist so that the board can later review the decision.

            • 9. A fund returned 9%, its minimum acceptable return was 4%, and downside deviation was 5%. The calculated value of the Sortino ratio is closest to:

              • A. 0.80
              • B. 1.00
              • C. 1.20
              Show answer

              Answer: B. 1.00

              1.00 - The Sortino ratio is the fund’s return above its minimum acceptable return, divided by downside deviation. - Using the stated figures, (9% − 4%) / 5% = 1.00. 0.80 - 0.80 results from (9% − 5%) / 5%. - That calculation mistakenly uses the 5% downside deviation as the minimum acceptable return, which is 4%. 1.20 - 1.20 results from (9% − 3%) / 5%. - That substitutes an unstated 3% threshold for the fund’s stated 4% minimum acceptable return.

            Flashcards: Type I and Type II errors in hiring and continuation

            1. 1. Type II error

              Show answer

              Null hypothesis (H0) is that there is no value added. The null hypothesis is not rejected when in fact there was value added.

            2. 2. Manager Continuation Policy as a Filter

              Show answer

              - MCP can be used as a statistical filter to remove -ve value add managers and retain +ve value add managers - Null = manager has no investment skill - Alternative = Managers are not zero value add managers - Type I error = rejecting the null when it is true - Type II error = not rejecting the null when it is false horn = H0 Reject Null

            3. 3. Manager Continuation Policy

              Show answer

              - Hiring and firing managers involves significant costs in terms of money and time - Replace managers only when justified (not based on short term underperformance) - Develop formal policies that are applied consistently - Combine portfolio performance and other measures - Continually monitor and review

            4. 4. Style Drift

              Show answer

              Style Drift is when a portfolio manager strays from his original, stated objective. There are two main reasons why this can be problematic for an investor: 1. The investor will not receive the desired style exposure 2. If the manager starts drifting they may be moving into areas outside their expertise.

            5. 5. Selecting a Fixed Income Manager

              Show answer

              Criteria that should be utilized: - Style analysis - not only style but the risk exposure due to style too - Selection bets - Investment process - what type of research, who makes decissions etc - Alpha correlation - Alphas should be diversified across managers Should not use past performance as an indicator of future success

            6. 6. Hedge Fund Relationships

              Show answer

              - Funds with quarterly lock-up have higher returns than monthly - Funds often dissolved during severe downturns as unlikely to ever earn incentive fees - Large funds underperform small funds - Funds with larger asset base attract talent - Younger funds tend to outperform older funds

            7. 7. Manager Questionnaires

              Show answer

              There are five sections of a questionnaire: 1. Staff and organizational structure 2. Investment philosophy and procedures 3. Resources and how research is conducted 4. Performance 5. Fee Schedule

            8. 8. Framework for Allocating to Managers

              Show answer

              Maximize utility through manager selection using the following equation: Next, given their utility function, the investor needs to investigate the performance characteristics of available managers. An efficient frontier is useful here but plotting active return against active risk for all combinations of available managers. Investors are usually more risk averse when facing active risk than they are when dealing with total for due to: 1. Must believe that managers can beat the market 2. Risk averse as they often have to justify to a superior 3. Less diversification

            9. 9. Selecting Equity Managers

              Show answer

              - Past performance - not a guarantee of future performance - Manager questionnaires - Screens potential managers - Fee Schedules - Ad volorem (aum based), performance based

            10. 10. Type I error

              Show answer

              Null hypothesis (H0) is that there is no value added. The null hypothesis is rejected when in fact there was no value added.

            Everything you need for CFA prep

            Get full Oncourse access with lessons, practice questions, flashcards and AI study tools.

            GET STARTED FOR FREE
            75% OFFLimited time offer
            GET 75% OFF