# Block 027 — Pre-read teach (2026-08-11, taught by mentor in lieu of user reading) **Mode:** Pre-emptive teach-first substitution (per macro-mentor skill `references/teach-pre-reads.md`). User opted to have the substrate taught by the mentor rather than reading the three pre-reads themselves. **Sources taught:** 1. Menkhoff, Sarno, Schmeling, Schrimpf (2012) "Carry Trades and Global FX Volatility", JoF 67(2), 681-718. Open-access preprint: 2. BIS Bulletin #90 — Aquilina, Lombardi, Schrimpf, Sushko (Aug 2024) "The market turbulence and carry trade unwind of August 2024". 3. BIS Bulletin #124 — "Monetary policy transmission to exchange rates: the role of currency carry trades". --- ## Source 1 — Menkhoff et al. (2012) **Layer 1 — load-bearing identity.** Carry trade = borrow low-rate funding currency, invest in high-rate target currency. Empirical puzzle: why is this profitable? Paper answer: risk premium — high-yielders co-move **negatively** with global FX volatility, low-yielders co-move **positively** (act as hedge). Excess return on carry = compensation for bearing volatility risk. Core finding is a number: **a single proxy for global FX volatility explains >90% of the cross-sectional return spread across five carry-sorted portfolios.** **Layer 2 — methodology.** Carry portfolios = quintile-sort currencies by forward-implied rate differential. Five portfolios from lowest- to highest-carry. Returns measured as excess return vs. USD funding. Global FX volatility proxy = principal-component-style aggregate of cross-sectional realized FX volatilities. Outperforms VIX / FX VIX analogs / individual currency vols. Liquidity risk matters but is dominated by vol risk. **Layer 3 — trade construction.** - Sizing = carry-to-risk ratio (carry ÷ option-implied vol). High = fat edge but vol-regime-sensitive. Low = thin edge, vol compression already priced. - Direction of risk is asymmetric in vol. High-vol regimes hit target currencies; low-vol regimes help them. Carry edge has hidden **short-vol option** character. - Time-series signal: excess returns track **unexpected** global FX vol shocks (residual), not expected levels. Edge doesn't bleed slowly with rising realized vol; it blows up when vol shocks. Substrate for Target 3 asymmetry. --- ## Source 2 — BIS Bulletin #90 (Aug 2024 yen carry unwind) **Layer 1 — load-bearing identity.** Aug 5, 2024 event = *positioning unwind*, not a funding crisis. Trigger = weak US NFP print on Aug 2. Amplifier = leveraged carry positioning concentrated in yen. Size estimate: **¥40 trillion ($250B) total carry exposure** going in (lower bound). Net short yen futures at historical peak (~¥2T / $14B). **Layer 2 — methodology.** - Cross-sectional size triangulation: 3 independent measurements — (a) CFTC net short futures (~$14B), (b) OTC forward back-of-envelope (~$160B upper bound), (c) BIS Global Liquidity Indicators on yen-denominated cross-border bank claims ($500B+ including Cayman SPVs). Bracket the $250B middle estimate. - "Amplifier" framing: bulletin explicitly distinguishes *trigger* (NFP print) from *amplifier* (deleveraging + margin calls + thin August markets). Trigger was small; amplifier was the story. Same structure as Block 015's signal-layer diagnostic. - Speed of recovery: "Markets then stabilised quickly." Two-week half-life. Not a 2008/2020-style funding crisis. **Layer 3 — trade construction.** - Trigger ≠ amplifier; amplifier is where money is lost. A 1% NFP surprise = 5-10% yen move if positioning amplifier is loaded. - Carry-to-risk peaked Q1 2024 before BoJ tightening → leading indicator for unwind risk. - Time-to-resolution asymmetry: positioning unwinds in 5-10 trading days; funding crises take weeks-to-months. Distinguishes first-day trade response from multi-week re-build. --- ## Source 3 — BIS Bulletin #124 (state-dependent MP transmission) **Layer 1 — load-bearing identity.** Exchange rate response to MP is **state-dependent on carry-trade positioning.** When speculators net short the funding currency, +25bp tightening shock moves CHF/USD ~4% and JPY/USD ~10%. When speculators NOT net short, same shock = statistically insignificant FX move. Quantified form of the Bulletin #90 mechanism. **Layer 2 — methodology.** - Local projections (Jordà 2005) for impulse response at horizons h=0,...,20 days. - Conditioning variable: net non-commercial short futures on CME (CHF, JPY). Incomplete but reliable, cross-validated against carry-to-risk. - Sample split: "carry-trade periods" (net shorts above median) vs. "low/no short" periods. +25bp effect shifts from 0% (no significance) to 4% CHF / 10% JPY. **Layer 3 — trade construction.** - Asymmetry = tightening only (Graph 3.B is load-bearing). Tightening shocks drive carry unwinds and FX amplification; **easing shocks do not**. Deleveraging pressure requires positive funding-rate shock. - State-dependence = the diagnosis. Pull CFTC positioning on CHF/JPY; condition expected FX vol on positioning state; resize exposure accordingly. - PM implication: funding CB tightening + heavy positioning = FX vol prints > non-positioning model predicts. Block 027 Target 4 substrate. --- ## Synthesis — how the three sources unlock Block 027 main material Through-line: **the carry trade edge is vol-risk premium (Source 1), the unwind amplifier is positioning-fueled deleveraging (Sources 2+3), and the asymmetry is vol-shock-driven in time and tightening-shock-driven in policy direction.** Mapping onto the four block targets: - **Target 1 (Carry trade construction):** Sources 1 + 3 — high-rate target funded by low-rate funding, sized by carry-to-risk ratio, conditioned on funding-CB policy posture. - **Target 2 (Volatility drag):** Source 1's math — carry < risk-adjusted carry due to convexity of vol. Breakeven vol regime = realized vol = carry/risk-adjusted edge. Above breakeven, carry bleeds. - **Target 3 (Carry unwind asymmetry):** Sources 2 + 3 — positioning unwinds faster than funding unwinds; tightening amplifies, easing doesn't; time-to-resolution = days (positioning) vs weeks (funding). - **Target 4 (PM implications):** Source 3's state-dependence → size positions based on conditioning variable (CME net non-commercial shorts, Aug-2024-style preconditions), not unconditional. Funding-currency selection = pick the funding CB where positioning is *least* crowded, all else equal. --- ## Pre-block probe queue (4 items, to be fired at next session open) 1. Menkhoff et al.'s load-bearing number — what % of cross-sectional carry-spread returns does the global FX volatility proxy explain? (>90% — should recall cleanly.) 2. BIS Bulletin #90's middle ballpark size estimate of yen carry exposure going into Aug 2024? (~$250B / ¥40T.) 3. BIS Bulletin #124's +25bp effect on CHF/USD vs JPY/USD when speculators are net short the funding currency? (CHF ~4%, JPY ~10%.) 4. The asymmetry — does monetary policy tightening or easing drive carry unwind + FX amplification? (Tightening only.) --- ## Skip list (mirrors block 027 plan) - Detailed regression specifications in Menkhoff et al. (Tables 4-7) — not load-bearing. - Bulletin #90 Graph 2-3 details on VIX dynamics — only carry-relevant facts (amplifier ≠ trigger) taught. - Bulletin #124's local projection bootstrap mechanics — finding is what Block 027 needs. - Liquidity vs vol decomposition in Menkhoff (Section IV) — second-order; "vol dominates" is captured in short-vol-option comment under Source 1 Layer 3. --- ## Anki cards pushed 15 atomic Basic cards pushed to `Macro Study::Phase 1 — Components::FX` on 2026-08-11 (note IDs 1786448537065–1786448537420). Tagged `block-027`. Card 13 is the user-flagged one — non-commercial CME net short futures as the BIS #124 risk proxy. ## Substitution flag for next session verification gate At next session open, the pre-read verification gate asks **"did you do the read OR get the teach?"** rather than **"did you do the read?"** — the gate adapts to substrate source (teach vs read) per the pre-emptive teach-first rule.