## _wp1308

## Source details

**Canonical URL:** [_wp1308](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp1308.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp1308.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp1308.pdf.json)

---

### Introduction and motivation
- Research question: What determines currency returns when investors globally reduce risk exposure simultaneously (risk-off episodes)?
- Risk-off definition and rationale:
  - VIX: measure of U.S. equity market volatility derived from options prices on the S&P 500.
  - Advantages: measurable at high frequency and in real time; no immediate link with FX markets; risk-off episodes derived from the VIX correspond to priors about disruptive market events in the last two decades.
- Empirical observation:
  - Japanese yen, Swiss franc, and U.S. dollar tend to appreciate against other G-10 and emerging market currencies at horizons from 1-week to 12-weeks after the start of a risk-off episode.
  - Emerging markets (EM) currencies vis-à-vis the USD exhibit a high degree of correlation during and across risk-off episodes.

### Identification and chronology of risk-off episodes (methodology)
- Operational rule:
  - Start of a risk-off episode = when the VIX is 10 percentage points higher than its 60-day backward-looking moving average (MA).
- Implementation notes:
  - VIX chosen over alternatives (AUDJPY, option-implied volatility on major currency pairs, cross-sectional volatility of exchange rates) for conformity with known disruptive events, high-frequency availability, and independence from currency-market-derived measures.
  - Under this rule there were 11 distinct risk-off episodes since 1992. Initial dates listed:
    - #1 29 October 1997 — Escalation of Asian crisis
    - #2 4 August 1998 — Concerns on Russian economy
    - #3 12 October 2000 — Fear of slowing US economy
    - #4 17 September 2001 — 9/11 Attacks
    - #5 10 July 2002 — Fear of slowing US economy
    - #6 10 August 2007 — BNP Paribas halts withdrawals from three money market mutual funds
    - #7 12 November 2007 — Disruptions in USD money markets
    - #8 17 September 2008 — Lehman failure
    - #9 6 May 2010 — Greek crisis
    - #10 16 March 2011 — Uncertainty over impact of Japan‟s March 11 Earthquake
    - #11 4 August 2011 — Confrontation over US debt ceiling and deterioration of crisis in euro area
  - Dating notes: the two 2007 episodes are treated as a single episode starting in August 2007; the algorithm picks 17 September 2001 for the 9/11 episode due to market closures.

### VIX behavior and episode summary statistics
- The fall 2008 VIX spike dwarfs others, but recent episodes still show substantial increases relative to baseline.
- Trends:
  - Risk-off events have happened more often since 2007.
- Statistics for the 12-week horizon after episode start (sample over the 8 episodes reported in Table 2):
  - Average level of the VIX 12 weeks after the start is on average only 2.4 percentage points below the level that triggered the episode.
  - Standard deviation of the change in the VIX for the same 12-week horizon is 9.8 percentage points.

### Recurring FX patterns during risk-off episodes (impulse responses and VAR results)
- Estimation approach:
  - Bivariate VAR estimated for each country on weekly data (first week of 2000 till June 2012) with 12 lags, linear and quadratic trend; variables: risk-off dummy and s = weekly average of bilateral nominal exchange rate relative to the U.S. dollar.
  - Identification: immediate effect of risk-off dummy on exchange rates with no feedback from exchange rates to the dummy in the first week; no lagged variable affects the risk-off dummy.
- Key empirical impulse-response findings (percent changes and standard errors as reported):
  - Safe-haven appreciations:
    - Japan cumulative appreciation after 24 weeks: 2.8 percent.
    - Japan impulse-response row (weeks 0,1,3,6,12,24): 0.75 [0.37], 1.72 [0.59], 1.61 [0.90], 2.26 [1.21], 2.84 [1.71], 2.78 [1.62]
    - Switzerland row (weeks 0,1,3,6,12,24): 1.01 [0.42], 2.17 [0.69], -0.13 [1.07], -1.13 [1.47], 1.29 [2.04], 1.34 [2.08]
  - Large on-impact or near-impact depreciations (one week after start):
    - Brazil: -2.62 [0.95] at week 1; cumulative patterns include -1.52 [0.60] at week 0 and -6.84 [3.50] at week 24.
    - South Africa: -2.31 [1.07] at week 1; -7.35 [3.29] at week 24.
    - Australia: -1.94 [0.81] at week 1; -1.33 [2.78] at week 24.
  - Additional country excerpts (selected):
    - Korea: -1.45 [0.68] at week 1; -6.67 [1.95] at week 24.
    - Ukraine: -0.07 [0.36] at week 1; -7.64 [2.54] at week 24.
  - Horizon-specific patterns:
    - Many advanced-economy currencies that depreciate (e.g., Australia, New Zealand, Sweden) show mean reversion over weeks.
    - Persistent EM depreciations: after 12 weeks, cumulative risk-off depreciation of Brazilian real, Korean won, South African rand, and Ukrainian hryvnia is more than 5 percent.
  - Non-significant pooled responses (no significant effect found in pooled VAR): Argentina, China, Czech Republic, Denmark, Egypt, euro area, Israel, Morocco, Romania, Thailand, Venezuela.
- Cross-sectional correlation:
  - Positive correlation between USD spot returns of different currencies across episodes, especially at a short horizon (three weeks).
  - At twelve weeks, patterns are harder to discern; March 2011 Japan episode is significantly negatively correlated with other episodes due to intervention.

### Predictors and mechanisms explaining cross-sectional currency behavior
- Two non-exclusive mechanisms:
  - Investors reassess the riskiness of each currency in light of new information.
  - Repricing of risky assets in general (change in price of risk).
- Country-level variables available prior to episodes (used in cross-sectional regressions):
  - A. Policy interest rate before the episode.
  - B. External sector: lagged current account balance; net foreign asset position; recent foreign portfolio inflows (4 quarters MA, percent of GDP).
  - C. Exchange rate misalignment: three CGER measures — Macroeconomic Balance (MB), Equilibrium Real Exchange Rate (ERER), External Sustainability (ES).
  - D. Currency behavior prior to episode: beta with respect to the VIX (52 weekly observations prior); beta with respect to AUDJPY.
  - E. Option market pricing: implied volatility of at-the-money 3-month options; risk reversals versus U.S. dollar (25-delta, 3-month maturity).
  - F. Liquidity and openness: bid-ask spread (median in month before episode); Chinn-Ito index of capital liberalization (updated to 2010).
- First-order cross-sectional findings:
  - Higher policy interest rates, higher pre-episode betas with VIX and AUDJPY, and higher option-implied costs are associated with larger depreciations at outbreak.
  - Less liquid currencies (higher bid-ask spreads) and restrictions on international capital flows tend to weaken during risk-off episodes.
  - Overvaluation in MB and ES frameworks is associated with larger risk-off depreciations; ERER misalignment less so.
- Representative numeric panel means (Table 4 panel mean values):
  - Emerging market dummy: 0.70
  - Policy Rate (Interest rate before the episode): 6.61
  - Current account balance ( % GDP): 0.02
  - Net foreign assets ( % GDP): -0.11
  - Foreign portfolio inflows ( % GDP), 4 quarters MA: 0.10
  - MB misalignment: -1.04
  - ERER misalignment: -1.46
  - ES misalignment: -3.31
  - Beta with the VIX: 0.07
  - Beta with AUD/JPY: 0.16
  - Implied volatility of at-the-money option: 11.29
  - Risk reversal versus USD: 1.15
  - Bid-ask spread, in bps: 12.43
  - Chinn-Ito measure of capital liberalization: 0.95

### Multivariate regressions and horizon dependence
- 1-week horizon (multivariate):
  - High-yield currencies and those with larger implied volatility tend to depreciate the most one week into the risk-off episode.
  - No significant difference between EM and advanced economies when controlling for macroeconomic and market pricing variables.
- 12-week horizon (multivariate):
  - Adjusted R-squared rises "from about 0.4 to about 0.6" when extending horizon to 12 weeks.
  - Interest rates matter less at the 12-week horizon.
  - Market-pricing variables such as beta with the AUDJPY and implied volatility gain importance.
  - No coefficient significantly different from zero on the beta with the VIX in multivariate regressions.
- Pre- vs post-2007 changes (sample breakdown: pre-2007 = 3 episodes; post-2007 = 5 episodes; total 8 episodes):
  - Panel A (macroeconomic variables): effect of the interest rate before the episode became "substantially larger after 2007." Adjusted R-squared increases "from 0.40 in the first week ... to 0.60 at the 12-week horizon."
  - Panel B (market pricing variables):
    - Beta with the AUDJPY significant only in post-2007 sample and predicts larger risk-off depreciations for longer horizons.
    - Beta with the VIX shows a large positive coefficient pre-2007 and a large negative interaction X After 2007 for several horizons (example coefficients from Table 6: Beta with the VIX = 11.080, 12.970, 12.920, 7.239; X After 2007 = -10.520, -23.270, -25.650, -24.850).
    - Implied volatility coefficients and interactions reported in Table 6 (Implied volatility = 0.063, 0.111, 0.072, 0.068; X After 2007 = -0.019, -0.023, 0.144, -0.078).

### Changes in currency sensitivity and betas (conclusion highlights)
- Beta with AUDJPY: reported value of 0.60 for the 12-week sample.
- Since 2007:
  - High-yield currencies have become more sensitive to risk-off episodes; policy rate importance increased.
  - The sign on the beta with the VIX flipped: prior to 2007 higher VIX betas tended to predict depreciation; since 2007 higher VIX betas have tended to strengthen as the risk-off episode progresses.
  - Implication: potentially large positive expected returns from investing in currencies with large VIX betas early in a risk-off episode.
- Correlations and medians:
  - Currency betas with the AUDJPY and VIX are positively correlated.
  - Median beta with the AUDJPY among emerging markets has trended up; median beta with the VIX has not trended up.
  - Betas are estimated using 52 weekly observations; Figure measures use past 3 months; sample window January 1, 2000 to June 4, 2012 (01Jan2000 to 04Jun2012).

### Policy implications and monitoring focus
- Monitoring priorities for assessing currency vulnerability in risk-off episodes:
  - Policy interest rates prior to episodes (high-yield status), especially for horizons beyond one week.
  - Market-pricing indicators: implied volatility, risk reversals, betas with AUDJPY and VIX (with attention to pre/post-2007 dynamics).
  - Balance of payments positions: current account balances and net foreign asset positions.
  - Recent foreign portfolio inflows (four-quarter moving average) as a vulnerability for 6-week+ horizons.
  - Measures of exchange rate misalignment (MB and ES) as signals of overvaluation correlated with larger depreciations.
  - Liquidity and openness: monitor bid-ask spreads for rising medium-term vulnerability (12-week horizon); greater capital liberalization (Chinn-Ito) associated with reduced propensity for FX weakness.
- Risks highlighted:
  - For safe-haven currency countries: rapid appreciation can lead to loss of export competitiveness and deflation risks.
  - For countries with risky currencies: risk-off episodes may cause FX weakness, possible inflation consequences, and disruption where high degrees of "dollarization" of liabilities exist.

*Source: _wp1308 - IMF working paper excerpt (figures, tables, appendix, and conclusion as supplied).*

### References ................................................................

### _wp1308 - References ................................................................ 

### Introduction: scope and motivation
- Research question: What determines currency returns when investors globally reduce risk exposure simultaneously (risk-off episodes)?
- Definition and rationale for using the VIX:
  - VIX: measure of U.S. equity market volatility derived from options prices on the S&P 500.
  - Advantages cited:
    - Measurable at high frequency and in real time.
    - No immediate link with FX markets (derived from S&P 500 options).
    - Risk-off episodes derived from the VIX correspond to priors about disruptive market events in the last two decades.
- Empirical observation:
  - The Japanese yen, Swiss franc, and U.S. dollar (USD) tend to appreciate against other G-10 and emerging market currencies at horizons from 1-week to 12-weeks after the start of a risk-off episode (Figure 1).
  - Emerging markets (EM) currencies vis-à-vis the USD exhibit a high degree of correlation during and across risk-off episodes.

### Identifying risk-off episodes (methodology)
- Operational definition used in the paper:
  - Start of a risk-off episode = when the VIX is 10 percentage points higher than its 60-day backward-looking moving average (MA).
- Implementation notes:
  - The VIX was preferred after experiments with other rules (AUDJPY, option-implied volatility on major currency pairs, cross-sectional volatility of exchange rates) because: conformity to known disruptive events, high-frequency availability, and independence from currency-market-derived measures.
  - Under this VIX-based definition, there were 11 distinct risk-off episodes since 1992 (text reference to Figure).

### Recurrent patterns in FX returns during risk-off episodes (findings)
- Cross-sectional and time-series characterization:
  - Recurrent patterns of FX returns in risk-off episodes can be identified by:
    - Simple correlations: the cross section of spot returns against the USD is positively correlated across episodes.
    - Vector Auto Regressions (VARs): used to identify these patterns (paper applies VARs).
  - FX behavior highlighted:
    - Japanese yen, Swiss franc, and USD appreciate versus other currencies on average during risk-off episodes (Figure 1).
    - Emerging market currency returns are highly correlated across episodes.
- Market measures and timing:
  - Horizons shown include 1-week through 12-weeks after episode start (Figure 1).
  - “Others” in Figure 1: equally-weighted average USD return on a list of specified countries' currencies (full list presented in source).

### Predictors and mechanisms explaining cross-sectional currency behavior
- Two non-exclusive mechanisms discussed:
  - Investors reassess the riskiness of each currency in light of new information.
  - Repricing of risky assets in general (change in price of risk).
- Fundamentals and market characteristics associated with smaller risk-off depreciations or larger appreciations:
  - Lower policy interest rates.
  - Stronger current account balances.
  - Stronger net foreign asset or reserve asset positions.
  - Deviations of actual exchange rates from equilibrium levels: overvaluation in two of the three IMF CGER models increases vulnerability.
  - Market liquidity and controls:
    - Less liquid currencies (measured by bid-ask spreads) and restrictions on international capital flows tend to weaken during risk-off episodes.
- Market prices and positioning prior to episodes:
  - Higher beta of a currency’s returns with respect to the VIX (estimated from weekly currency returns in the 52 weeks before the episode) → larger depreciations at outbreak.
  - Higher beta with respect to AUDJPY → greater propensity to sell-offs (AUDJPY betas capture how a currency trades relative to the AUDJPY pair popular in carry trades).
  - Option-implied information: more expensive at-the-money options (higher option prices / higher willingness to pay for insurance) → less resilience during risk-off episodes.

### Changes over time
- Frequency and severity:
  - Risk-off episodes have become more frequent and severe since the beginning of the Global Financial Crisis in the summer of 2007, after a few years of relatively calm markets.
- Stability of fundamentals’ role:
  - No substantial difference over time in the role of fundamentals associated with the balance of payments or international investment position.
- Evolving predictors:
  - A currency’s yield and sensitivity to global risks have become better predictors of risk-off depreciations during more recent risk-off episodes.

### Relation to existing literature (selected citations from source)
- Brunnermeier et al. (2008): link crash risk of carry trade strategies to funding constraints of leveraged investors.
- Ranaldo and Söderlind (2010): study safe-haven properties of major advanced-economy currencies; find Japanese yen and Swiss franc display safe-haven characteristics.
- Grisse and Nitschka (2012): examine safe-haven characteristics of the Swiss franc.
- Habib and Stracca (2012): find net foreign asset position is a robust indicator of safe-haven status.
- Nozaki (2010) and Jordà and Taylor (2012): explore fundamental-based currency strategies exploiting recurrent cross-sectional patterns.
- Melvin and Taylor (2008): conditioning carry trade on a global financial stress index would have been a winning strategy during the Global Financial Crisis.

*Source: _wp1308 - References ................................................................*

### 2. The rectangular shades in grey indicate periods of 12 weeks since the beginning of risk-off

### 2. The rectangular shades in grey indicate periods of 12 weeks since the beginning of risk-off episodes

### Identification and chronology of risk-off episodes
- Risk-off episodes are identified by a rule based on spikes in the VIX; focus is on the 12 weeks subsequent to the beginning of each episode.
- The initial dates of the risk-off episodes (Table 1):
  - #1 29 October 1997 — Escalation of Asian crisis
  - #2 4 August 1998 — Concerns on Russian economy
  - #3 12 October 2000 — Fear of slowing US economy
  - #4 17 September 2001 — 9/11 Attacks
  - #5 10 July 2002 — Fear of slowing US economy
  - #6 10 August 2007 — BNP Paribas halts withdrawals from three money market mutual funds
  - #7 12 November 2007 — Disruptions in USD money markets
  - #8 17 September 2008 — Lehman failure
  - #9 6 May 2010 — Greek crisis
  - #10 16 March 2011 — Uncertainty over impact of Japan‟s March 11 Earthquake
  - #11 4 August 2011 — Confrontation over US debt ceiling and deterioration of crisis in euro area
- Notes on dating:
  - The two risk-off episodes in the second half of 2007 are treated as a single episode starting in August 2007.
  - The algorithm chooses September 17 2001 as the beginning of the 9/11 risk-off episode because US equity markets were closed and the VIX series was held at its 09/10 value until 9/16.

### VIX behavior and summary statistics during risk-off episodes
- The spike in the VIX in fall 2008 dwarfs other spikes, but more recent episodes still show substantial increases in volatility relative to baseline.
- Risk-off events have happened more often since 2007.
- For the 12-week horizon after the start of a risk-off episode:
  - The average level of the VIX 12 weeks after the start is on average only 2.4 percentage points below the level that triggered the episode.
  - The standard deviation of the change in the VIX for the same 12-week horizon is 9.8 percentage points.
  - The sample for this calculation is over the 8 episodes reported in Table 2.

### Why risk-off episodes have become more frequent
- Latent risks in the G-3 economies (notably sovereign or private sector balance sheets in Europe and the United States) are a main driver of higher frequency since 2007; exception: Japan earthquake in March 2011.
- Observers argue for a world with multiple equilibria where policy responses can lead to virtuous or vicious cycles (cited: IMF, 2011b).
- Increased integration of global financial markets leads to more rapid re-pricing of global risks across markets. Taylor (2012) noted unprecedented expansion of external balance sheets over the last two decades.
- After retrenchment following Lehman, growth in international portfolio flows resumed (Figure 3).
- Larger stock of portfolio liabilities worldwide implies:
  - Sharp increase in local currency exposure by foreign investors.
  - Even without pullbacks, hedging of local currency exposure can lead to FX weakness.
  - Periods of strong inflows tend to appreciate currencies, potentially making undervaluations less evident and increasing vulnerability when risk-off hits.

### Recurring features of exchange rate behavior during risk-off episodes
- General cross-sectional patterns:
  - Positive correlation between USD spot returns of different currencies across episodes, especially at a short horizon (three weeks).
  - At twelve weeks, patterns are harder to discern and no episode stands out.
  - The March 2011 Japan earthquake episode is significantly negatively correlated with other episodes; intervention on March 18 reversed initial sell-off.
- VAR approach:
  - A bivariate VAR is estimated for each country on weekly data (first week of 2000 till June 2012) with 12 lags, including a linear and a quadratic trend. Variables: risk-off dummy and s = weekly average of bilateral nominal exchange rate relative to the U.S. dollar.
  - Identification: ordering implies immediate effect of risk-off dummy on exchange rates with no feedback from exchange rates to the dummy in the first week; coefficients constrained so no lagged variable affects the risk-off dummy.
  - Significant immediate and cumulative effects are found across currencies.

### Key empirical findings from impulse response functions (selected exact numeric results)
- Safe-haven appreciations:
  - Japanese yen and Swiss franc appreciate relatively to the U.S. dollar at the beginning of a risk-off episode.
  - Cumulative Japanese yen appreciation after 24 weeks is 2.8 percent.
  - Swiss franc appreciation tends to revert between the first and third week; Japanese yen appreciation appears more persistent.
- Large on-impact or near-impact depreciations (one week after start):
  - Brazilian real and South African rand depreciate on average by more than 2 percent one week after the start of the risk-off episode.
  - Australian dollar depreciates by 1.9 percent one week after the start.
- Currencies depreciating three to six weeks into episodes:
  - Canadian dollar, Singapore dollar, Chilean peso, Indonesian rupiah, Peruvian sol, Swedish krona, Turkish lira, United Kingdom pound.
- Persistent EM depreciations:
  - After 12 weeks, cumulative risk-off depreciation of Brazilian real, Korean won, South African rand, and Ukrainian hryvnia is more than 5 percent.
- Heterogeneity and mean reversion:
  - Advanced-economy currencies that depreciate (e.g., Australia, New Zealand, Sweden) show mean reversion over weeks.
  - EM currency weakness is more persistent for Brazil, Chile, Mexico, Philippines, and South Africa, with little mean reversion over two quarters following onset.
- Currencies with no statistically significant pooled response (no significant effect found in pooled VAR): Argentina, China, Czech Republic, Denmark, Egypt, euro area, Israel, Morocco, Romania, Thailand, Venezuela.
- Example impulse response numeric excerpts (weeks 0,1,3,6,12,24 with standard errors in brackets) — selected rows from Table 3:
  - Japan: 0.75 [0.37], 1.72 [0.59], 1.61 [0.90], 2.26 [1.21], 2.84 [1.71], 2.78 [1.62]
  - Switzerland: 1.01 [0.42], 2.17 [0.69], -0.13 [1.07], -1.13 [1.47], 1.29 [2.04], 1.34 [2.08]
  - Australia: -0.53 [0.50], -1.94 [0.81], -4.85 [1.24], -4.60 [1.73], -1.84 [2.60], -1.33 [2.78]
  - Brazil: -1.52 [0.60], -2.62 [0.95], -6.51 [1.47], -5.69 [2.08], -7.27 [3.24], -6.84 [3.50]
  - South Africa: -1.06 [0.69], -2.31 [1.07], -4.39 [1.63], -7.47 [2.28], -7.45 [3.27], -7.35 [3.29]
  - Korea: -0.53 [0.42], -1.45 [0.68], -4.27 [1.04], -5.30 [1.40], -6.98 [2.02], -6.67 [1.95]
  - Ukraine: -0.02 [0.16], -0.07 [0.36], -0.21 [0.63], -1.70 [1.11], -5.70 [2.00], -7.64 [2.54]
  - Note: numbers above are the impulse responses (percent change) and standard errors exactly as reported in Table 3.

### Cross-sectional explanatory framework (variables available prior to episodes)
- A. Policy interest rates
  - Policy interest rate before the episode: high interest rate currencies are vulnerable due to carry trade and portfolio flows.
- B. External sector
  - Lagged current account balance: larger deficits expected to correlate with larger depreciations.
  - Net foreign asset position: larger positive positions (or reserves) reduce risk of sell-offs; reserves relative to short-term liabilities aid defense.
  - Recent foreign portfolio inflows (4 quarters MA, percent of GDP): captures vulnerability following sharp increases in debt or equity flows.
- C. Exchange rate misalignment
  - Three CGER measures: Macroeconomic Balance (MB), Equilibrium Real Exchange Rate (ERER), External Sustainability (ES).
- D. Currency behavior prior to the episode
  - Beta with respect to VIX: sensitivity to fluctuations in global risk appetite (estimated over 52 weekly observations prior to episode).
  - Beta with respect to AUDJPY: measures position correlation with highly liquid AUDJPY; positive beta means currency tends to depreciate when AUDJPY is weaker.
- E. Cost of buying an option and tail risk insurance
  - Implied volatility of at-the-money options with a 3-month maturity: forward-looking cost to hedge USD moves.
  - Risk reversals versus U.S. dollar (25-delta, 3-month maturity): higher values imply investors pay more to insure against USD weakness of the local currency; expected to be positively correlated with currency weakness during risk-off.
- F. Liquidity conditions
  - Bid-ask spread (median in month before episode): captures ease of unwinding positions; effect on exposure is empirical.
  - Capital liberalization and controls (Chinn and Ito measures updated to 2010): coded so more positive means freer movements of capital flows.
- G. Simple regressions approach
  - Cross-sectional regressions of depreciation against the USD since episode onset on each country-specific variable Xi, including a constant and separate dummy for each risk-off episode; horizons of 1, 3, 6, and 12 weeks.
  - First-order finding: EM currencies do not differ from advanced economies at onset, but at twelve weeks EM currencies tend to weaken significantly (consistent with VAR impulse responses).

*Source: Bloomberg and authors' calculations, as reported in the IMF working paper excerpt.*

### section III. Currencies of advanced economies such as the Australian and New Zealand

### section III. Currencies of advanced economies such as the Australian and New Zealand

### Immediate and medium-term currency responses
- Advanced-economy currencies such as the Australian and New Zealand dollars "tend to depreciate on impact but also rebound sharply."
- Emerging market (EM) currencies such as the Brazilian real or the Mexican peso "face more persistent depreciations on impact."
- High-yielding currencies "tend to depreciate more in the six weeks following the onset of a risk-off episode," linked to disruptions in carry trade positioning by global investors.

### Balance of payments, external positions, and misalignment
- Higher current account balances (3) and stronger net foreign asset positions (4) are correlated with appreciation over all horizons.
- Countries with stronger foreign portfolio inflows (5) in the run-up to a risk-off episode are more vulnerable to FX weakness for horizons of 6 weeks or longer (but this result is not statistically significant for the 12 weeks horizon).
- In a smaller sample, measures indicating overvaluation (CGER variables) are also related to FX weakness during risk-off episodes.
- MB misalignment (6) and External Sustainability ES (8) frameworks show a link between overvaluation and risk-off depreciations, whereas the Equilibrium Real Exchange Rate (ERER, 7) approach does not.
- The MB and ES measures have a correlation coefficient of 85% and a strong negative correlation with the current account balance, suggesting current account dynamics drive MB and ES results.

### Sensitivity to global risks and FX-specific market pricing
- Currency sensitivity to global risks:
  - Beta with the VIX (9) has a "strong unconditional correlation with risk-off depreciations on impact." Betas are computed using currency returns prior to the risk-off episode.
  - Currency beta with AUDJPY (10) is positively correlated with risk-off depreciations at all horizons.
- Option prices:
  - Implied volatility (11) and risk reversals (12) are positively correlated with risk-off depreciations. These variables track the cost to insure against currency moves from current levels.
  - The report notes that "option prices are not a good predictor of currency moves in tranquil times" and that "implied volatilities underprice tail risks" (citing Englander (2012) as an example).
- Liquidity and openness:
  - Bid-ask spreads (13) are not associated with risk-off depreciations on impact but tend to be positively correlated with risk-off depreciations over 12 weeks.
  - Currencies of countries with less restricted capital flows (Chinn-Ito index of capital liberalization, 14) are less prone to FX weakness during risk-off episodes.
  - No significant effect found for depth of domestic bond market (trading volumes from the Emerging Markets Trade Association) on FX performance in risk-off episodes.

### Key empirical magnitudes from regressions (select findings and model performance)
- Simple-regression observations and sample:
  - Sample includes Argentina, Australia, Brazil, Canada, Chile, China, P.R.: Mainland, Colombia, Czech Republic, Denmark, Euro Area, Hungary, India, Indonesia, Israel, Japan, Republic of Korea, Malaysia, Mexico, New Zealand, Norway, Peru, Philippines, Poland, Romania, Russian Federation, Singapore, South Africa, Sweden, Switzerland, Taiwan, Thailand, Turkey, Ukraine, and United Kingdom.
  - Mean [St. Dev] and 13612 N. Obs. appear in Table 4 (table displays estimates of equation (3) of section IV(a); standard errors robust with clustering at the currency level).
- Representative numeric indicators from Table 4 (selected entries preserved exactly as in source):
  - Emerging market dummy: 0.70
  - Policy Rate (2) Interest rate before the episode: 6.61
  - Current account balance (3) (% GDP): 0.02 (panel mean)
  - Net foreign assets (4) (% GDP): -0.11 (panel mean)
  - Foreign portfolio inflows (5) (% GDP), 4 quarters MA: 0.10 (panel mean)
  - MB misalignment (6): -1.04 (panel mean)
  - ERER misalignment (7): -1.46 (panel mean)
  - ES misalignment (8): -3.31 (panel mean)
  - Beta with the VIX (9): 0.07 (panel mean)
  - Beta with AUD/JPY (10): 0.16 (panel mean)
  - Implied volatility of at-the-money option (11): 11.29 (panel mean)
  - Risk reversal versus USD (12): 1.15 (panel mean)
  - Bid-ask spread, in bps (13): 12.43 (panel mean)
  - Chinn-Ito measure of capital liberalization (14): 0.95 (panel mean)
- Multivariate regressions (Table 5) — 1-week horizon:
  - High-yield currencies and those with larger implied volatility tend to depreciate the most one week into the risk-off episode.
  - No difference between EM and advanced economies when controlling for macroeconomic and market pricing variables.
  - Currencies of countries running large current account surpluses "appear to depreciate less," but not statistically significant.
- Multivariate regressions — 12-week horizon:
  - Adjusted R-squared rises "from about 0.4 to about 0.6" when extending horizon to 12 weeks.
  - Interest rates matter less at the 12-week horizon, while market pricing variables such as the beta with the AUDJPY exchange rate and implied volatility gain in relative importance.
  - No coefficient significantly different from zero on the beta with the VIX in the multivariate regressions.

### Have currency risk factors changed since 2007? (pre- vs post-2007 evidence)
- Sample framing:
  - Pre-2007 period includes 3 risk-off episodes (October 2000, September 2001 and July 2002).
  - Post-2007 period includes 5 risk-off episodes (August 2007 subprime crisis, Lehman‟s bankruptcy, Greek crisis in May 2010, Japan's March 2011 earthquake, and U.S. debt ceiling confrontation), for a total of 8 risk-off episodes.
- Panel A (macroeconomic variables) findings:
  - For 1, 3, 6, and 12-week horizons, the effect of the interest rate before the episode became "substantially larger after 2007."
  - Since the global financial crisis, high-yield currencies have depreciated more sharply during risk-off episodes, with the effect larger for longer horizons.
  - Macroeconomic variables explain an increasing proportion of risk-off depreciations when the horizon lengthens: adjusted R-squared increases "from 0.40 in the first week of the risk-off episodes to 0.60 at the 12-week horizon."
- Panel B (market pricing variables) findings:
  - The beta with the AUDJPY exchange rate is significant only in the post-2007 sample and predicts larger risk-off depreciations the longer the horizon.
  - The beta with the VIX shows a large positive coefficient pre-2007 and a large negative interaction X After 2007 for several horizons (example coefficients preserved in Table 6: Beta with the VIX = 11.080, 12.970, 12.920, 7.239; X After 2007 = -10.520, -23.270, -25.650, -24.850).
  - Implied volatility coefficients and interactions are reported in Table 6 (Implied volatility = 0.063, 0.111, 0.072, 0.068; X After 2007 = -0.019, -0.023, 0.144, -0.078).
- Interpretation:
  - Since the global financial crisis, cuts in G-3 policy rates prompted many investors to increase exposure to high-yield currencies; this may have increased sensitivity of high-yield currencies to risk-off episodes.
  - The increased importance of the policy rate post-2007 could reflect investor decisions to hedge or scale back positions when the VIX spikes and the legacy of widespread losses on leveraged carry trades in 2008.
  - The currency return‟s beta with the AUDJPY cross is a measurable indicator of how much a currency trades as a carry play.

### Analytical implications and monitoring focus (from the source analysis)
- When assessing currency vulnerability in risk-off episodes, prioritize:
  - Policy interest rates prior to episodes (high-yield status), especially for horizons beyond one week.
  - Market-pricing indicators: implied volatility, risk reversals, and currency betas with AUDJPY and the VIX (with attention to pre/post-2007 dynamics).
  - Balance of payments positions: current account balances and net foreign asset positions (higher values correlate with appreciation).
  - Recent foreign portfolio inflows (four-quarter moving average) as a vulnerability for 6-week+ horizons.
  - Measures of exchange rate misalignment (MB and ES) as signaling overvaluation correlated with larger depreciations during risk-off episodes.
- Liquidity and openness considerations:
  - Monitor bid-ask spreads for signals of rising medium-term vulnerability (12-week horizon).
  - Greater capital liberalization (Chinn-Ito) is associated with reduced propensity for FX weakness during risk-off episodes.

*Source: _wp1308 - section III. Currencies of advanced economies such as the Australian and New Zealand*

### 0.60 for the 12-week sample. Similarly to the effect of the policy interest rate, the increasing

### VI. CONCLUSION AND POLICY IMPLICATIONS

### Changes in currency sensitivity and betas
- Beta with AUDJPY: reported value of 0.60 for the 12-week sample.
- Increasing importance of the beta with the AUDJPY may reflect the growing importance of the carry trade as a factor driving currency behavior during risk-off episodes.
- AUDJPY betas in the sample of EM countries have increased in recent years.
- Change in signal of the beta with VIX post 2007:
  - Prior to 2007: currencies with a higher beta with the VIX tended to depreciate relative to the USD during risk-off episodes, other things equal.
  - Since 2007: the sign has flipped; currencies with higher VIX betas have strengthened as the risk-off episode progresses.
  - Implication: potentially large positive expected returns from investing in currencies with large VIX betas early in a risk-off episode.
- Currency betas with the AUDJPY and VIX are positively correlated (Table A1 in the appendix).
- Median beta patterns:
  - Median beta with the AUDJPY among emerging markets has trended up.
  - Median beta with the VIX has not trended up (Figure 4).
- Spot returns on select currencies have been behaving more like returns on the AUDJPY currency pair rather than returns on the VIX.

### Cross-sectional behavior of exchange rates during risk-off episodes
- Definition: risk-off episodes are identified as spikes in the VIX, the option-implied volatility of equity prices.
- Recurring patterns documented:
  - Right after onset of risk-off episode, high yield currencies and those with higher betas to the VIX or the AUDJPY tend to have larger depreciations relative to the U.S. dollar.
  - Stronger current account balances and net foreign asset positions are associated with smaller risk-off depreciations or larger appreciations.
  - Overvaluation in two out of three CGER models is associated with larger risk-off depreciations.
  - While global factors have become more important (increased frequency of risk-off episodes), country-specific factors remain essential in determining currency movements within an episode.
- Safe-haven currencies:
  - Swiss franc and Japanese yen tend to appreciate relative to the U.S. dollar during risk-off episodes.
  - Because most other currencies depreciate, the Swiss franc and Japanese yen appreciate even more on a trade-weighted basis.
- Currencies that tend to depreciate during risk-off episodes (for at least some horizon window) include: Australia, Brazil, Canada, Chile, Colombia, India, Indonesia, Korea, Malaysia, Mexico, New Zealand, Peru, Philippines, Singapore, South Africa, Sweden, Turkey, Ukraine, and United Kingdom.

### Policy implications and risks
- Usefulness for policymakers:
  - Identifying currencies most likely to appreciate or depreciate in future risk-off episodes informs monetary policy and asset management decisions.
- Risks for safe-haven currency countries:
  - Rapid currency appreciation can lead to loss of export competitiveness and deflation risks (Swiss case referenced).
- Risks for countries with currencies perceived as risky:
  - Risk-off episodes may cause FX weakness with possible consequences for inflation.
  - Potential disruption if there is a high degree of “dollarization” of liabilities.

### Data, figures, and measurement notes
- Betas estimation note: Betas are estimated coefficients in a regression of a currency’s weekly USD returns on the weekly return to the VIX or AUDJPY exchange rate using a sample of 52 weekly observations.
- Figure measurement notes:
  - Figure 5 note: Betas are median values of past 3 months.
  - Appendix Figure A1 notes: Exchange rates are expressed as log deviations from their average for 01Jan2000 to 04Jun2012 (continued figure notes use January 1, 2000 to June 4, 2012).
- Appendix Table A1: standard deviations on the diagonal; bold cells indicate t-test rejects null that a single correlation coefficient equals zero at the 95 percent level of confidence.
- Selected numeric/time references preserved from source:
  - 0.60
  - 12-week sample
  - post 2007
  - 52 weekly observations
  - past 3 months
  - 01Jan2000 to 04Jun2012
  - January 1, 2000 to June 4, 2012

*Source: IMF working paper excerpt (figures, appendix, and conclusion provided in the supplied content).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp1308.pdf_