Hedging a Duration-Heavy AFS Book When the Curve Steepens
The treasury of a Karachi bank holds a PKR 48bn available-for-sale portfolio of fixed-rate Pakistan Investment Bonds with modified duration 5.2 years, plus PKR 20bn held-to-maturity. Equity is PKR 16bn. A 100bp parallel rise would cost roughly PKR 2.5bn, about 15% of equity - a number the board has not seen. The SBP is signalling a hold, but the 3-year curve has steepened 70bp in two months and the bank's economists are split. At ALCO, three options are tabled: (a) sell 40% of the AFS book and park it in 3-month T-bills yielding 10.9%; (b) pay fixed on a PKR 15bn three-year swap at 11.4% while the 3-year PIB yields 10.9%; (c) reclassify the AFS book to held-to-maturity so mark-to-market stops hitting equity. Carry on the AFS book is 11.2%; funding cost is 10.4%.
1. Quantify the risk reduction and the annual cost of options (a) and (b). Which would you recommend?
Read the worked answer
Current DV01 = 48bn x 5.2 x 0.0001 = PKR 24.96m per basis point; the 100bp loss of about PKR 2.5bn reconciles.
Option (a): sell 40% (PKR 19.2bn). Remaining duration exposure = 28.8bn x 5.2 = 149.8bn, plus 19.2bn x 0.25 = 4.8bn, total 154.6bn, implying DV01 of PKR 15.46m - a 38% reduction. Annual carry give-up = (11.2% - 10.9%) x 19.2bn = PKR 57.6m, plus any realised loss on sale if the bonds are above par. Risk reduction per rupee of carry sacrificed is strong.
Option (b): a pay-fixed swap has negative duration approximately equal to the fixed leg, about 2.7 years for a 3-year par swap. DV01 offset = 15bn x 2.7 x 0.0001 = PKR 4.05m, leaving PKR 20.9m - only a 16% reduction. Annual cost = 50bp swap spread x 15bn = PKR 75m, and the bank takes counterparty and basis risk.
Recommendation: (a) is roughly 2.4 times more effective per unit of carry and does not add derivative complexity; (b) preserves carry but is a partial hedge at the wrong point of the curve. A blend - trim AFS and add a smaller, longer-dated hedge - is defensible if the bank wants to keep yield. Note both are parallel-shift fixes; the 70bp steepening means the 5.2-year duration is not the right single statistic.
2. Assess option (c) under IFRS 9 and from a capital and governance standpoint.
Read the worked answer
Reclassification under IFRS 9 is permitted only when the entity's business model for managing the financial assets changes, and that change must be demonstrable to an auditor and a regulator; it is expected to be rare. A risk-driven decision to stop marking a portfolio is not a business model change. The IAS 39 'tainting' concept is gone, so the old fear of selling AFS assets is also no longer the constraint - but the IFRS 9 test is stricter in substance.
Even if a reclassification were accepted, it would not remove economic risk. Duration stays at 5.2 years; a 100bp rise still reduces the economic value of equity by about PKR 2.5bn. It merely moves the loss from fair value through OCI or P&L to an unrealised erosion recognised only on sale or via interest income. The SBP, auditors and the board would reasonably read the timing - after a 70bp steepening and with the board unaware of the 15%-of-equity sensitivity - as accounting arbitrage. That is a conduct and governance problem, not just an accounting one.
Better path: disclose the sensitivity, define an EVE and NII limit, and choose the hedge or the sale on economic grounds. If any accounting action is taken, it should be hedge accounting under IFRS 9 with documented hedge relationships, not reclassification.
3. The CFO says duration tells us everything we need. Where is that wrong, and what should the board see instead?
Read the worked answer
Duration is a first-order, parallel-shift measure. It assumes all yields move by the same amount, ignores convexity, and says nothing about the shape of the curve. Here the 3-year has moved 70bp while the long end may have moved differently; the portfolio's 5.2-year duration does not tell the board whether the risk sits at 3, 5 or 10 years, nor how a steepener versus a parallel rise would affect it. It also ignores the basis between PIB yields, KIBOR-linked funding and deposit repricing, and it is silent on liquidity - how quickly PKR 48bn can be sold without moving the market.
Replace the single number with a dashboard: EVE sensitivity per 100bp in PKR and as a percentage of CET1; 12-month NII sensitivity including deposit betas; a scenario table (parallel +100/+200bp, bear steepener, bull flattener, 2023-style rate shock); hedge ratio and residual DV01 by tenor bucket; limit utilisation; and a liquidity overlay showing days to liquidate by issue. Add a VaR and expected-shortfall view, but label it as a short-horizon trading measure, not a substitute for EVE.
The governance point: the board is being asked to approve a hedge without ever seeing that a parallel 100bp move costs 15% of equity. That disclosure gap is itself the finding.