Risk Appetite Breach in a Korean Balanced Fund
HanRiver Asset Management manages a KRW 2.4 trillion balanced fund for a corporate pension. The board-approved risk appetite statement sets three hard limits: one-year 95% VaR no more than 6.0% of NAV, ex-ante tracking error versus the KOSPI 200 no more than 250 bp, and no single issuer above 4.0% of NAV. In September the CIO buys KRW 180 billion of 10-year BBB+ corporate bonds from a single issuer, funding the purchase by selling KOSPI 200 futures. The monthly risk report shows portfolio VaR at 5.4% of NAV, so the CIO tells the risk committee the fund is inside appetite. The report also shows tracking error at 310 bp and the issuer at 5.1% of NAV. The head of risk objects that the VaR figure was produced by taking a 1-day 95% VaR and multiplying by the square root of 250. The CIO replies that the bond is held for liability matching and therefore should be excluded from the trading-risk limit. The committee must decide whether to ratify the trade, demand remediation, or escalate to the board.
1. Identify each risk-appetite metric that is breached, and assess the CIO's argument that the bond should be excluded from the trading-risk limit.
Read the worked answer
Two hard limits are breached: tracking error of 310 bp exceeds 250 bp, and single-issuer concentration of 5.1% exceeds 4.0%. Portfolio VaR of 5.4% is below 6.0%, but it is not reliable because of the scaling issue, so the CIO cannot use it as evidence of compliance. The liability-matching argument does not exempt the bond: risk appetite is stated at fund level, and a long-dated BBB+ corporate bond carries credit spread, default, and interest-rate risk; excluding it would let the CIO arbitrage the limit. Even if the bond matched a specific liability, the funding trade changed the fund's equity beta and tracking error. The committee should treat the trade as a breach, not ratify it, and require remediation. The CIO's argument also confuses hedging a liability with eliminating risk; the pension liability is not marked through the fund NAV in the same way. Breach escalation is mandatory because two board limits are hard constraints.
2. Recommend the specific governance actions the risk committee should take, assigning responsibilities across the three lines of defense.
Read the worked answer
First line: the CIO and portfolio manager must present a remediation plan within five business days. That plan should cut the issuer to at most 4.0% and reduce tracking error to 250 bp or below, either by selling part of the bond or by adding offsetting index exposure. Second line: the risk function should independently re-estimate VaR without square-root scaling, document the breach, and refuse any limit exclusion. It should report to the board risk committee, not only to the CIO. Third line: internal audit should review why the trade was executed, whether pre-trade limit checks failed, and whether the risk report's methodology was approved. The committee should escalate to the board because this is a hard-limit breach and an attempted override; it should also require a look-back at other liability-matching trades. If the strategy genuinely needs a higher issuer limit, the correct route is a board-approved risk-appetite change, not an ex post exclusion. Compensation linkage should also be considered.
3. Critique the square-root-of-time scaling used to produce the 5.4% VaR figure. State the conditions under which it is valid and the likely direction of bias here.
Read the worked answer
The square-root-of-time rule scales VaR by sqrt(T) only when returns are independent and identically distributed, with constant volatility and no serial correlation. Here the portfolio holds 10-year BBB+ bonds, whose spreads and prices are persistent and exhibit volatility clustering. If returns are positively autocorrelated, variance over 250 days grows faster than linearly, so sqrt(250) scaling understates the one-year VaR. If mean reversion is strong, it could overstate, but credit spread jumps and fat tails make understatement more likely. A one-day VaR also cannot capture the liquidity horizon for corporate bonds. The estimated 5.4% should be treated as a floor, not proof of compliance. Better methods are overlapping one-year historical simulation, Monte Carlo with fat tails, or a liquidity-adjusted stressed VaR. The committee should not accept the figure as evidence that the fund is inside its VaR appetite.