## onlineannexes11 - annex explains how these adjustments were estimated for the 2020 external sector assessments

## Source details

**Canonical URL:** [onlineannexes11 - annex explains how these adjustments were estimated for the 2020 external sector assessments](https://www.imf.org/-/media/files/publications/esr/2021/english/onlineannexes11.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/esr/2021/english/onlineannexes11.pdf.md)
- [Structured JSON version](/-/media/files/publications/esr/2021/english/onlineannexes11.pdf.json)

---

### Overview and purpose
- Four cross-cutting COVID-19 related adjustors applied for the 2020 External Sector Assessments:
  - Travel services balance (including tourism) due to restrictions on international travel.
  - Oil balances reflecting the sharp impact on oil prices and oil volumes traded.
  - Trade in COVID-19–related medical products.
  - Shift in household consumption composition from services to durables and other consumer goods.
- Additional country-specific adjustors related to income, gold, and remittances applied where relevant.

### Method for Oil and Travel adjustors
- Estimation framework (annual data 1986–2019 for EBA sample):
  - y_{i,t} – y_{i,t-1} = α_i + λ_t + β1 x_{i,t} + β2 z_{i,t} + γ Controls_{i,t} + ε_{i,t}
    - y_{i,t} is the EBA cyclically adjusted current account balance (percent of GDP).
    - x_{i,t} change in travel services balance (percent of GDP) based on WEO historical data.
    - z_{i,t} change in oil balance (percent of GDP) based on WEO historical data.
    - Controls: three lags of the change in the current account, the output gap, and the change in the terms of trade.
    - α_i and λ_t are economy and time fixed effects; ε_{i,t} is the residual.
- Estimation approach:
  - Two-stage least squares (2SLS) used to address endogeneity; generalized method of moments (GMM) also used for robustness.
- Key regression summary (Online Annex Table 1.1.1):
  - Change in Travel Balance/GDP: point estimates 0.80***, 0.79***, 0.73***, 0.73***. Observations 1,248; 1,247; 1,246; 1,246. R-squared 0.23; 0.23; 0.22; 0.22 across columns.
  - Change in Oil Balance/GDP: point estimates 0.91***, 0.92**, 0.89***, 0.94***.
  - Note: ***, **, * indicate statistical significance at the 1, 5, and 10 level, respectively.
- Interpreted impacts:
  - "the impact effect of a 1 percent of GDP rise in the travel services balance on the current account is about 0.75 percent of GDP, while the effect of a rise in the oil balance is somewhat higher (0.90 percent of GDP)."
  - Effect being less than 1 reflects partial adjustment through domestic demand and substitution effects.
- Construction of 2020 adjustors:
  - Travel adjustor = estimated relationship × projected transitory COVID-19 impact on travel services balance in 2020 (baseline: impact fades over medium term; modified where IMF staff projections assume more persistence).
  - Oil adjustor = estimated relationship × temporary component of unexpected change in oil balances in 2020, where unexpected change = revision relative to pre-pandemic (January 2020 WEO Update) IMF staff forecasts; temporary component = (change in oil balance projection for 2020) − (change in oil balance projection for 2025).

### Trade in medical products adjustor
- Data and coverage:
  - Uses UN Comtrade export and import data for a WTO (2020) list of COVID-19–related medical products covering pharmaceuticals, medical supplies, medical equipment, and personal protective equipment.
  - Values computed at Harmonized System six-digit subheading level for 92 subheadings.
- Adjustment for foreign value added:
  - Export data adjusted to subtract foreign value added using OECD TiVA data. The subtracted foreign intermediate goods are reallocated to imports for each economy.
  - These intermediate imports were allocated to exporters based on total goods trade export shares for 2020 from the IMF WEO database (shares computed to add to 100 percent for the sample).
  - Reallocation has only a modest influence on results.
- Outcome:
  - Calculated change in net exports in 2020 compared with 2019 for economies with necessary Comtrade data.
  - Estimated positive net effect on the current account particularly large for medical goods exporters such as China and Malaysia.

### Household consumption composition shift adjustor
- Rationale:
  - Focus on composition shift (not overall level), since level effects are reflected in standard EBA cyclical adjustment.
- Estimation method:
  - Compare:
    1. Counterfactual level of durables, nondurables, and services consumption in 2020:Q2–Q4 based on 2019:Q2–Q4 shares and 2020 total private consumption evolution.
    2. Actual level of durables, nondurables, and services consumption in 2020.
    3. Import content of durables, nondurables, and services consumption.
  - Import content:
    - For the United States: based on Hale and others (Federal Reserve Bank of San Francisco staff) 2019 study.
    - For other economies: scaled by percentage of foreign value added in domestic demand (OECD TiVA) relative to the United States.
  - Data availability:
    - Quarterly data available for 14 advanced economies and 7 emerging market and developing economies.
    - For missing economies: used advanced economy and emerging market and developing economy averages respectively.
- Accounting for cross-adjustor overlap:
  - Excluded foreign travel from consumption of services to avoid double-counting with travel adjustor.
  - Excluded fuel and pharmaceutical/medical products from nondurables to avoid double-counting with oil and medical goods adjustors.
- Outcome:
  - Largest net positive and negative estimated current account impacts in US dollars are for China (largest positive) and the United States (largest negative), respectively.

### Other idiosyncratic COVID-19 factors
- Income balance:
  - Economies with large FDI liabilities experienced increases in income and current account balances due to lower dividend payments to foreign investors (examples: Australia, Poland).
- Gold balance:
  - Increased global demand for gold led to temporary increases in gold exports for gold producers (examples: South Africa, Thailand) and temporary current account decreases for gold importers (example: Switzerland).
- Remittances:
  - Fluctuations deemed important for Mexico and, to a lesser extent, Malaysia.
- Overall magnitude:
  - These additional idiosyncratic factors were relatively small as a share of world GDP.

### Country-level COVID-19 adjustors (summary examples)
- Adjustors to the current account (equal to estimated impact but with opposite sign) produce highly asymmetric effects across economies.
- Examples highlighted:
  - Large negative travel-related current account impacts for tourism-intensive economies: Spain, Thailand, Turkey.
  - Large negative oil-related current account impacts for oil exporters: Russia, Saudi Arabia.
  - Medical goods exporters with large positive current account effects: China, Malaysia.

### Online Annex 1.2 — The COVID-19 crisis and “downhill” flow of capital
- Background:
  - Standard models: capital flows from capital-abundant rich economies to capital-scarce poorer economies.
  - Post-global financial crisis decade (2010–19): observed “downhill” capital flows from richer to poorer economies.
- Empirical evidence 2010–19:
  - Relationship between current account balance and (log) per capita income (lagged) during 2010–19: a doubling in per capita income associated with a 1.02 percentage point of GDP rise in the current account balance.
  - Online Annex Table 1.2.1, column (1): Log PPP GDP per capita coefficient = 1.018*** (robust standard error (0.369)). Time fixed effects: Yes. Observations 1883.
- COVID-era change in downhill flows (2020–25):
  - The COVID-19 crisis appears to have slowed downhill flows: poorer economies saw, on average, positive revisions to current account balances compared with pre-pandemic forecasts; richer economies saw unexpected downward revisions.
  - Forecast errors for the current account balance in percent of GDP in 2020 (compared with January 2020 WEO forecast) are negatively correlated with initial (2019) log PPP GDP per capita across 192 economies.
  - Online Annex Table 1.2.1, column (2): Log PPP GDP per capita coefficient = -1.050*** (robust standard error (0.204)). Observations 192.
  - Interpretation: the estimated slope coefficient (−1.05) implies that a doubling in income per capita is associated with a 1.05 percentage point of GDP reduction in the current account balance compared with pre-pandemic forecasts—contrasting with the previous decade’s positive association.
  - Excluding China and the United States decreases the coefficient modestly (in absolute terms) to −0.99, remaining statistically significant at the 1 percent level.
  - Repeating analysis for forecasts for 2021 yields a slope coefficient of −0.80, and the slope flattens progressively toward zero for subsequent years—suggesting slowdown in expected downhill flows is persistent but expected to fade over time.
- Drivers of the reversal:
  - Negative coefficient on current account revisions mainly reflects larger declines in government saving in richer economies, outweighing rises in private saving.
  - Decomposition (Online Annex Table 1.2.1, columns (3)–(4)):
    - Private Saving-Investment (forecast errors) vs. Log PPP GDP per capita: coefficient 2.104*** (robust standard error (0.624)). Observations 147.
    - Public Saving-Investment (forecast errors) vs. Log PPP GDP per capita: coefficient −3.055*** (robust standard error (0.596)). Observations 150.
  - Interpretation: public saving–investment revisions are almost entirely driven by lower public saving in richer countries; private saving–investment revisions reflect mainly higher private saving in richer economies and, to a lesser extent, declines in private investment.
- Net implication:
  - Relative to previous forecasts, patterns imply an uphill flow of capital from poorer to richer economies—highlighting uneven impact and policy responses during the pandemic across country income groups.

### Income Levels and Current Account Forecast Errors, 2020 (figure notes)
- Online Annex Figure 1.2.1 plots forecast errors (outcomes minus January 2020 WEO Update forecast) in percent of GDP against Log PPP GDP per capita in 2019.
- Variables shown (each panel uses bubble sizes proportional to US dollar GDP):
  - 1. Private Saving Forecast Errors (Percent of GDP)
  - 2. Private Investment Forecast Errors (Percent of GDP)
  - 3. Public Saving Forecast Errors (Percent of GDP)
  - 4. Public Investment Forecast Errors (Percent of GDP)
- Axis and labeling details preserved:
  - Horizontal axis: "Log PPP GDP per capita in 2019" with tick cluster "6 8 10 12" represented as "681012" in the figure labels.
  - Vertical axis range: "-20 -10 0 10 20" (Percent of GDP).
- Note: PPP = purchasing power parity.

### Online Annex 1.3. Computing Valuation Effects
- Accounting identity (preserved notation):
  - Changes in net foreign asset (NFA) positions between period t – 1 and t are equal to
    - NFA_{t} − NFA_{t−1} = CA_{t} + VAL^{FX}_{t} + VAL^{P}_{t}   (1.3.1)
    - where CA_{t} denotes the current account, VAL^{FX}_{t} are currency-induced valuation effects, and VAL^{P}_{t} are valuation effects due to changes in asset prices.
  - Given that it is possible to calculate VAL^{FX}_{t}, VAL^{P}_{t} is obtained as a residual from the accounting identity (1.3.1).
- Decomposition of currency-induced valuation effects:
  - VAL^{FX}_{t} = EQ^{FX}_{VAL} + D^{FX}_{VAL} + OT^{FX}_{VAL}
    - EQ^{FX}_{VAL — currency-induced valuation effects from external equity positions,
    - D^{FX}_{VAL — currency-induced valuation effects from external debt positions,
    - OT^{FX}_{VAL — other positions, which include foreign exchange reserves.
  - Equity positions include direct investment equity and portfolio equity. Debt positions include portfolio debt, direct investment debt, and other investment.
- Calculation of currency-induced valuation effects (formula preserved):
  - VAL^{FX}_{t} = Σ_{c} [ 0.5% × (Δ% F I^{c}_{t} ) × (A^{c}_{t−1} − L^{c}_{t−1}) ]  (presentation preserved in source notation)
    - where c denotes the asset class and includes equity, debt, or other positions, %ΔF I^{c}_{t} is the percentage change in the net financial exchange rate index in period t for asset class c, and A^{c}_{t−1} (1 − L^{c}_{t−1}) denote assets (liabilities) of asset class c.
- Financial exchange rate index definition (formula preserved):
  - F I^{c}_{t} = Σ_{j} ω^{c}_{j,t−1} × (1 + %ΔE_{j,t})
    - where %ΔE_{j,t} is the percentage change in the bilateral end-of-period nominal exchange rate between the currency of a given country and the foreign currency j between t – 1 and t, ω^{c}_{j,t−1} is the net financial weight between a given country and currency j in period t – 1 for asset class c.
- Net financial weight calculation (preserved expression and definitions):
  - ω^{c}_{j,t−1} = (A^{c}_{j,t−1}/A^{c}_{t−1}) − (L^{c}_{j,t−1}/L^{c}_{t−1})
    - where A^{c}_{j,t−1} (L^{c}_{j,t−1}) are the proportion of assets (liabilities) of asset class c denominated in foreign currency j,
    - s^{c}_{A,t−1} = A^{c}_{t−1}/(A^{c}_{t−1} + L^{c}_{t−1}), and s^{c}_{L,t−1} = L^{c}_{t−1}/(A^{c}_{t−1} + L^{c}_{t−1}) (expressions preserved from source).
- Author and footnotes:
  - The author of this annex is Luciana Juvenal.
  - Footnote: Therefore, any errors or discrepancies are included in VAL^{P}_{t}.
  - Footnote: Equity positions include direct investment equity and portfolio equity. Debt positions include portfolio debt, direct investment debt, and other investment.

*Source: onlineannexes11 - annex explains how these adjustments were estimated for the 2020 external sector assessments (International Monetary Fund, 2021).*

### annex explains how these adjustments were estimated for the 2020 external sector assessments

### onlineannexes11 - annex explains how these adjustments were estimated for the 2020 external sector assessments

### Overview and purpose
- Explains four main COVID-19 related adjustors applied across economies in an evenhanded and multilaterally consistent way for the 2020 External Sector Assessments (reported in Chapter 3 and summarized in Chapter 1).
- Four cross-cutting adjustors:
  - Travel services balance (including tourism) due to restrictions on international travel.
  - Oil balances reflecting the sharp impact on oil prices and oil volumes traded.
  - Trade in COVID-19–related medical products.
  - Shift in household consumption composition from services to durables and other consumer goods.
- Additional country-specific adjustors related to income, gold, and remittances were also applied where relevant.

### Method for Oil and Travel adjustors
- Estimated the historical relationship between the EBA cyclically adjusted current account balance and travel services and oil balances after controlling for relative output gap and terms of trade.
- Estimated equation (annual data 1986–2019 for EBA sample):
  - y_{i,t} – y_{i,t-1} = α_i + λ_t + β1 x_{i,t} + β2 z_{i,t} + γ Controls_{i,t} + ε_{i,t}
    - y_{i,t} is the EBA cyclically adjusted current account balance (percent of GDP).
    - x_{i,t} change in travel services balance (percent of GDP) based on WEO historical data.
    - z_{i,t} change in oil balance (percent of GDP) based on WEO historical data.
    - Controls include three lags of the change in the current account, the output gap, and the change in the terms of trade.
    - α_i and λ_t are economy and time fixed effects; ε_{i,t} is the residual.
- Estimation approach:
  - Two-stage least squares (2SLS) used to address endogeneity; generalized method of moments (GMM) also used for robustness.
  - Online Annex Table 1.1.1 (summary):
    - Change in Travel Balance/GDP: point estimates 0.80***, 0.79***, 0.73***, 0.73*** (robust standard errors in parentheses). Observations 1,248; 1,247; 1,246; 1,246. R-squared 0.23; 0.23; 0.22; 0.22 across columns. Regressions include economy and year FE; some specifications include Output Gap and TOT.
    - Change in Oil Balance/GDP: point estimates 0.91***, 0.92**, 0.89***, 0.94*** (robust standard errors in parentheses).
    - Note: ***, **, * indicate statistical significance at the 1, 5, and 10 level, respectively.
- Interpreted impacts:
  - Textual summary of averaged results: "the impact effect of a 1 percent of GDP rise in the travel services balance on the current account is about 0.75 percent of GDP, while the effect of a rise in the oil balance is somewhat higher (0.90 percent of GDP)."
  - Effect being less than 1 reflects partial adjustment through domestic demand and substitution effects.
- Construction of 2020 adjustors:
  - Travel adjustor = estimated relationship (equation) × projected transitory COVID-19 impact on travel services balance in 2020 (baseline: impact fades over medium term; modified where IMF staff projections assume more persistence).
  - Oil adjustor = estimated relationship × temporary component of unexpected change in oil balances in 2020, where unexpected change = revision relative to pre-pandemic (January 2020 WEO Update) IMF staff forecasts; temporary component = (change in oil balance projection for 2020) − (change in oil balance projection for 2025).

### Trade in medical products adjustor
- Data and coverage:
  - Uses UN Comtrade export and import data for a WTO (2020) list of COVID-19–related medical products covering pharmaceuticals, medical supplies, medical equipment, and personal protective equipment.
  - Values computed at Harmonized System six-digit subheading level for 92 subheadings.
- Adjustment for foreign value added:
  - Export data adjusted to subtract foreign value added using OECD TiVA data. The subtracted foreign intermediate goods are reallocated to imports for each economy.
  - These intermediate imports were allocated to exporters based on total goods trade export shares for 2020 from the IMF WEO database (shares computed to add to 100 percent for the sample).
  - Reallocation has only a modest influence on results.
- Outcome:
  - Calculated change in net exports in 2020 compared with 2019 for economies with necessary Comtrade data.
  - Estimated positive net effect on the current account particularly large for medical goods exporters such as China and Malaysia.

### Household consumption composition shift adjustor
- Rationale:
  - Pandemic shifted household consumption composition away from services toward durables and other consumer goods; focus is on composition shift (not overall level) because level effects are reflected in standard EBA cyclical adjustment.
- Estimation method:
  - Compare:
    1. Counterfactual level of durables, nondurables, and services consumption in 2020:Q2–Q4 based on 2019:Q2–Q4 shares and 2020 total private consumption evolution.
    2. Actual level of durables, nondurables, and services consumption in 2020.
    3. Import content of durables, nondurables, and services consumption.
  - Import content:
    - For the United States: based on Hale and others (Federal Reserve Bank of San Francisco staff) 2019 study.
    - For other economies: scaled by percentage of foreign value added in domestic demand (OECD TiVA) relative to the United States.
  - Data availability:
    - Quarterly data available for 14 advanced economies and 7 emerging market and developing economies.
    - For missing economies: used advanced economy and emerging market and developing economy averages respectively.
- Accounting for cross-adjustor overlap:
  - Excluded foreign travel from consumption of services to avoid double-counting with travel adjustor.
  - Excluded fuel and pharmaceutical/medical products from nondurables to avoid double-counting with oil and medical goods adjustors.
- Outcome:
  - Largest net positive and negative estimated current account impacts in US dollars are for China (largest positive) and the United States (largest negative), respectively.

### Other idiosyncratic COVID-19 factors
- Income balance:
  - Economies with large FDI liabilities experienced increases in income and current account balances due to lower dividend payments to foreign investors (examples: Australia, Poland).
- Gold balance:
  - Increased global demand for gold led to temporary increases in gold exports for gold producers (examples: South Africa, Thailand) and temporary current account decreases for gold importers (example: Switzerland).
- Remittances:
  - Fluctuations deemed important for Mexico and, to a lesser extent, Malaysia.
- Overall magnitude:
  - These additional idiosyncratic factors were relatively small as a share of world GDP.

### Country-level COVID-19 adjustors (summary examples)
- Adjustors to the current account (equal to estimated impact but with opposite sign) produce highly asymmetric effects across economies.
- Examples highlighted:
  - Large negative travel-related current account impacts for tourism-intensive economies: Spain, Thailand, Turkey.
  - Large negative oil-related current account impacts for oil exporters: Russia, Saudi Arabia.
  - Medical goods exporters with large positive current account effects: China, Malaysia.
- (Online Annex Table 1.1.2 and Figures 1.1.1–1.1.2 present country-level percent of GDP and US$ results.)

### Online Annex 1.2 — The COVID-19 crisis and “downhill” flow of capital
- Background:
  - Standard models: capital flows from capital-abundant rich economies to capital-scarce poorer economies; richer economies run current account surpluses and lend to poorer economies.
  - Post-global financial crisis decade (2010–19): observed “downhill” capital flows from richer to poorer economies.
- Empirical evidence 2010–19:
  - Relationship between current account balance and (log) per capita income (lagged) during 2010–19: a doubling in per capita income associated with a 1.02 percentage point of GDP rise in the current account balance.
  - Online Annex Table 1.2.1, column (1): Log PPP GDP per capita coefficient = 1.018*** (robust standard error (0.369)). Time fixed effects: Yes. Observations 1883.
- COVID-era change in downhill flows (2020–25):
  - The COVID-19 crisis appears to have slowed downhill flows: poorer economies saw, on average, positive revisions to current account balances compared with pre-pandemic forecasts; richer economies saw unexpected downward revisions.
  - Forecast errors for the current account balance in percent of GDP in 2020 (compared with January 2020 WEO forecast) are negatively correlated with initial (2019) log PPP GDP per capita across 192 economies.
  - Online Annex Table 1.2.1, column (2): Log PPP GDP per capita coefficient = -1.050*** (robust standard error (0.204)). Observations 192.
  - Interpretation: the estimated slope coefficient (−1.05) implies that a doubling in income per capita is associated with a 1.05 percentage point of GDP reduction in the current account balance compared with pre-pandemic forecasts—contrasting with the previous decade’s positive association.
  - Excluding China and the United States decreases the coefficient modestly (in absolute terms) to −0.99, remaining statistically significant at the 1 percent level.
  - Repeating analysis for forecasts for 2021 yields a slope coefficient of −0.80, and the slope flattens progressively toward zero for subsequent years—suggesting slowdown in expected downhill flows is persistent but expected to fade over time.
- Drivers of the reversal:
  - Negative coefficient on current account revisions mainly reflects larger declines in government saving in richer economies, outweighing rises in private saving.
  - Decomposition (Online Annex Table 1.2.1, columns (3)–(4)):
    - Private Saving-Investment (forecast errors) vs. Log PPP GDP per capita: coefficient 2.104*** (robust standard error (0.624)). Observations 147.
    - Public Saving-Investment (forecast errors) vs. Log PPP GDP per capita: coefficient −3.055*** (robust standard error (0.596)). Observations 150.
  - Interpretation: public saving–investment revisions are almost entirely driven by lower public saving in richer countries; private saving–investment revisions reflect mainly higher private saving in richer economies and, to a lesser extent, declines in private investment.
- Net implication:
  - Relative to previous forecasts, patterns imply an uphill flow of capital from poorer to richer economies—highlighting uneven impact and policy responses during the pandemic across country income groups.

*Source: onlineannexes11 - annex explains how these adjustments were estimated for the 2020 external sector assessments (International Monetary Fund, 2021).*

### CHAPTER 1  EXTERNAL POSITIONS AND POLICIES

### CHAPTER 1  EXTERNAL POSITIONS AND POLICIES

### Income Levels and Current Account Forecast Errors, 2020
- Online Annex Figure 1.2.1 plots forecast errors (outcomes minus January 2020 WEO Update forecast) in percent of GDP against Log PPP GDP per capita in 2019.
- Variables shown (each panel uses bubble sizes proportional to US dollar GDP):
  - 1. Private Saving Forecast Errors (Percent of GDP)
  - 2. Private Investment Forecast Errors (Percent of GDP)
  - 3. Public Saving Forecast Errors (Percent of GDP)
  - 4. Public Investment Forecast Errors (Percent of GDP)
- Axis and labeling details preserved from source:
  - Horizontal axis: "Log PPP GDP per capita in 2019" with tick cluster "6 8 10 12" represented as "681012" in the figure labels.
  - Vertical axis range: "-20 -10 0 10 20" (Percent of GDP).
- Note: PPP = purchasing power parity.
- Sources: IMF, International Financial Statistics; IMF, World Economic Outlook (WEO); and IMF staff calculations.

### Online Annex 1.3. Computing Valuation Effects
- Purpose:
  - Expresses the role of valuation effects in the dynamics of the external position using an accounting framework.
- Accounting identity (preserved notation and equation numbering):
  - Changes in net foreign asset (NFA) positions between period t – 1 and t are equal to
    - 1,,tttFX tP t
      NFA  NFA  CA VAL   VAL
      
         
      ,        (1.3.1)
    - where t
      CA
      denotes the current account, ,FX t
      VAL
      are currency-induced valuation effects, and
      ,P t
      VAL
      are valuation effects due to changes in asset prices.
  - Given that it is possible to calculate ,FX t
    VAL
    , ,P t
    VAL
    is obtained as a residual from the accounting identity (1.3.1).
- Decomposition of currency-induced valuation effects:
  - ,FX
    VAL
    can be further decomposed into:
    - EQ
      FX
      VAL — currency-induced valuation effects from external equity positions,
    - D
      FX
      VAL — currency-induced valuation effects from external debt positions,
    - OT
      FX
      VAL — other positions, which include foreign exchange reserves.
  - Equity positions include direct investment equity and portfolio equity. Debt positions include portfolio debt, direct investment debt, and other investment.
- Calculation of currency-induced valuation effects (formula preserved):
  - Following Lane and Shambaugh (2010), the currency-induced valuation effects are calculated as
    - ,
      ,1   1
      %  (    )
      cF c   cc
      FX tttt
      VALI  A  L
      
        
    - where c denotes the asset class and includes equity, debt, or other positions, ,%
      F c
      t
      I
      is the percentage change in the net financial exchange rate index in period t for asset class c, and 1
      c
      t
      A
      
      (1
      c
      t
      L
      
      ) denote assets (liabilities) of asset class c.
- Financial exchange rate index definition (formula preserved):
  - The financial exchange rate is calculated as
    - ,,,
      1, 1,
      (1    %  )
      F cF cF c
      ttj tj t
      I   IE
      
      
      
    - where ,%
      j t
      E
      is the percentage change in the bilateral end-of-period nominal exchange rate between the currency of a given country and the foreign currency j between t – 1 and t, ,
      , 1
      F c
      j t
      
      
      is the net financial weight between a given country and currency j in period t – 1 for asset class c.
- Net financial weight calculation (preserved expression and definitions):
  - This is calculated as
    - .,   ,    ,   ,
      , 1   , 1  1   , 1  1
      F cA c  A cL c  L c
      j tj t   tj t   t
      ss
            
      
    - where ,,
      , 1
      A c
      j t
      
      
      (
      ,
      , 1
      L c
      j t
      
      
      ) are the proportion of assets (liabilities) of asset class c denominated in foreign currency j,
      ,1
      1
      1   1
      c
      A c
      t
      t
      cc
      tt
      A
      s
      A  L
      
      
      
      
      
      , and
      ,1
      1
      1   1
      c
      L c
      t
      t
      cc
      tt
      L
      s
      A  L
      
      
      
      
      
- Author and footnotes:
  - The author of this annex is Luciana Juvenal.
  - Footnote: Therefore, any errors or discrepancies are included in P
    VAL.
  - Footnote: Equity positions include direct investment equity and portfolio equity. Debt positions include portfolio debt, direct investment debt, and other investment.

### References (as listed in source)
- Alfaro, Laura, Sebnem Kalemli-Ozcan, and Vadym Volosovych. 2014. “Sovereigns, Upstream Capital Flows and Global Imbalances.” Journal of the European Economic Association 12 (5): 1240–89.
- Boz, Emine, Luis Cubeddu, and Maurice Obstfeld. 2017. “Revisiting the Paradox of Capital.” Vox: CEPR’s Policy Portal, March 9.
- Cubeddu, Luis, Signe Krogstrup, Gustavo Adler, Pau Rabanal, Mai Chi Dao, Swarnali Ahmed Hannan, Luciana Juvenal, and others. 2019. “The External Balance Assessment Methodology: 2018 Update.” IMF Working Paper 19/65, International Monetary Fund, Washington, DC.
- Gourinchas, Pierre-Olivier, and Olivier Jeanne. 2013. “Capital Flows to Developing Countries:  The Allocation Puzzle.” Review of Economic Studies 80 (4): 1484–515.
- Hale, Galina, Bart Hobijn, Fernanda Nechio, and Doris Wilson. 2019. “How Much Do We Spend on Imports?” FRBSF Economic Letter, Federal Reserve Bank of San Francisco, San Francisco, CA.
- Lane, Philip R., and Jay C. Shambaugh. 2010. “Financial Exchange Rates and International Currency Exposures.” American Economic Review 100:518–40.
- McQuade, Peter, and Martin Schmitz. 2017. “The Great Moderation in International Capital Flows: A Global Phenomenon?” Journal of International Money and Finance 73 (PA): 188–212.
- Obstfeld, Maurice. 2021. “Two Challenges from Globalization.” Keynote Speech at the Federal Reserve Bank of San Francisco 2019 Asia Economic Policy Conference. Journal of International Money and Finance (April).
- World Trade Organization (WTO). 2020. “Trade in Medical Goods in the Context of Tackling COVID-19: Developments in the First Half of 2020.” Information Note, Geneva.

*Source: onlineannexes11 - CHAPTER 1  EXTERNAL POSITIONS AND POLICIES*

---


_Source: https://www.imf.org/-/media/files/publications/esr/2021/english/onlineannexes11.pdf_