## ftdea

## Source details

**Canonical URL:** [ftdea](https://www.imf.org/-/media/files/publications/dp/2019/english/ftdea.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/dp/2019/english/ftdea.pdf.md)
- [Structured JSON version](/-/media/files/publications/dp/2019/english/ftdea.pdf.json)

---

### Executive summary — key messages
- Managing large and volatile capital flows is a central economic challenge in Asian emerging market economies (EMEs).
- Unconventional monetary policies in advanced economies contributed to record liquidity channeled to EMEs in Asia, with large reversals during tighter global financial conditions (for example, the “taper tantrum”).
- Exchange rates can both absorb external shocks and amplify them when domestic financial frictions exist; amplification is more salient with foreign exchange (FX) liabilities and shallow financial markets.
- Policymakers deploy four main tools: monetary policy (interest rate), exchange rate policy (including foreign exchange intervention, FXI), macroprudential measures (MPMs), and capital flow management measures (CFMs).
- Many Asian EMEs operate as de facto “flexible” inflation targeters with quasi-managed floats and pursue multiple objectives (price, growth, financial stability) using multiple instruments.

### Exchange rate effects, micro and macro evidence
- Firm-level evidence (12 Asian economies, 1994–2016):
  - Corporate vulnerability measured by Altman Emerging Market Z’’-score.
  - A 30 percent domestic currency depreciation against the US dollar is associated with a 0.4 decrease in the Z’’-score and results in 7 percent of firms moving into the D-rating bucket.
  - When firms’ share of FX-denominated debt exceeds 20 percent, an exchange rate depreciation is associated with lower firm-level investment.
  - Thresholds: if share of US dollar debt < 10 percent competitiveness channel dominates; between 10 and 20 percent financial channel offsets competitiveness channel; > 20 percent financial channel dominates.
- Macro-level evidence (Local Projection method):
  - A 1 percent real depreciation lowers the investment ratio by 0.6 percent when markets are shallow (credit-to-GDP ratio below 100 percent).
  - Effect on investment not statistically significant in countries with more-developed financial markets.
  - High nominal exchange rate volatility negatively affects investment and growth, especially in less-developed financial markets.
- Growth-at-Risk (GaR) findings:
  - Exchange rate volatility affects the conditional distribution of future GDP growth and growth-at-risk (left-tail growth).
  - In many less-financially developed economies, the exchange rate affects the growth distribution more than domestic financial conditions (FCIs).
  - Near-term easing of financial conditions and lower exchange rate volatility limit near-term downside risks but amplify medium-term downside risks; a sizable decline in the 5th percentile of the growth distribution over the medium term is estimated under favorable near-term shocks.

### Policy toolkit and stylized policy responses in Asia
- FXI:
  - FXI absorbs about 70 percent of net capital inflows on average.
  - Asian EMEs purchase about 75 percent of portfolio debt flows and about 100 percent of portfolio equity flows.
  - FXI is not systematically used to absorb FDI flows.
  - FXI response is stronger to more volatile flow types and where FX liabilities are high; greater financial depth is associated with lower FXI.
- Monetary policy:
  - Policy rates respond to inflation and output gap in a standard Taylor-rule fashion, and also to the US policy rate, the exchange rate, capital flows, and credit growth.
  - Example quantitative point: a net inflow surge of 10 percent of GDP is associated with a 10 basis point monetary tightening (illustrative chart-based result).
  - Exchange rate depreciation tends to induce monetary tightening in countries with shallower financial markets; policy rate tightening in response to depreciation is greater where private credit/GDP is lower.
- Macroprudential measures (MPMs):
  - MPMs tightened in periods of rapid domestic credit growth, real exchange rate appreciation, and positive net capital inflows; less likely to be tightened with a negative output gap.
  - MPMs grouped into five categories: bank capital, bank liquidity, credit demand, credit supply, and foreign-currency exposures; LTVs and reserve requirements have been most popular.
  - MPM tightening probability is positively associated with net capital inflows, lower US policy rates (in some specifications), inflation, and real credit growth.
- Capital flow management measures (CFMs):
  - CFMs used less frequently than MPMs and mainly targeted inflows and housing-related risks (stamp duties, taxes on nonresident property buyers).
  - CFMs often tightened outside of major capital flow surge quarters and frequently used in combination with MPMs.
  - CFMs broad enough to significantly reduce inflows have been used infrequently.

### Quantified policy reaction function results — FXI, monetary policy, and MPMs
- FXI (Annex estimates):
  - Net capital flows (% GDP) coefficients: Asia FE 0.67*** (0.067); Asia IV 0.66*** (0.044); EM FE 0.77*** (0.118); EM IV 0.99*** (0.091).
  - Inflows (Asia EMs) 0.622*** (0.114); Net portfolio (Asia EMs) 0.770*** (0.028); Net equity (Asia EMs) 1.040** (0.291).
  - FX Liabilities (FXL), % GDP: FXL effect (Asia, FXI and FXL) −0.050*** (0.014).
  - Interaction FXL × exchange rate example: −0.013* (0.007) in one specification.
  - Credit (% GDP) positive in some specs: Credit (Asia, FXI and Financial Development) 26.062** (2.301).
- Monetary policy (Annex estimates):
  - US interest rate coefficients: 0.072*** (0.018); 0.078** (0.026) in fixed-effects specifications.
  - GFC dummy coefficients: −0.871*** (0.187); −1.102** (0.335) (examples).
  - CPI coefficients: 0.119** (0.049); 0.200** (0.073).
  - REER often negative and sometimes significant: −0.029* (0.012); −0.027*** (0.008).
  - Net portfolio (Asia EMs) on policy rate: 0.008*** (0.002).
  - Credit (% GDP) negative and significant in some specs: −0.694*** (0.166); −0.852** (0.302).
  - Interaction Δexchange rate # Credit (% GDP): −0.043** (0.015); −0.039* (0.019).
  - Lagged dependent variable large (persistence): e.g., 0.812*** (0.036); 0.741*** (0.045).
- Macroprudential measures (MPM) (Annex estimates, probit):
  - Net capital flows (/GDP) associated with higher MPM activation: 0.022** (0.010) Baseline Asia; 0.038** (0.019) Baseline EMs.
  - External liability flows (/GDP): 0.022** (0.010) in Baseline Asia.
  - US policy rate reported with negative sign in some table layouts (example: −0.105*** (0.038) Baseline AEs).
  - Inflation: 0.092** (0.040) Baseline Asia; 0.062** (0.028) Baseline AEs.
  - Credit growth marginally significant in EMs: 0.034* (0.020) Baseline EMs.
  - Probit IV example: Real GDP growth 0.163*** (0.054).

### Channels and heterogeneity
- Competitiveness channel:
  - Depreciation can make exports cheaper and boost activity but may have asymmetric short-term effects; trade pricing in dominant currencies and global value chains can weaken pass-through to exports.
- Financial channels and balance-sheet effects:
  - Depreciation raises FX debt burdens, tightens domestic financial conditions when foreign investor holdings of local-currency debt are large, and can impair bank balance sheets and lending.
  - The negative impact of depreciation on investment and growth is larger where FX liabilities are high and financial markets are shallow.
- Heterogeneity:
  - Effects vary across firms and countries depending on FX debt share, leverage, Tobin’s Q, cash holdings, and financial development.
  - Annex Table 1.3 shows ΔNominal Exchange Rate coefficients by USD debt proportion partitions with negative and statistically significant coefficients in some partitions (e.g., −0.064** (0.0318) in column (1) reported as 20.064** (0.0318) and similar in columns (2)–(3)) and positive small coefficients in higher USD debt partitions (e.g., 0.011* (0.00504) and 0.024** (0.00841) in columns (5)–(6)).

### Potential costs, trade-offs, and market-development concerns
- Coordination and communication costs increase when internalizing joint calibration of multiple instruments to reach multiple objectives.
- FXI:
  - Theoretical ambiguity: FXI could de-anchor inflation expectations or help achieve the inflation target by adding an instrument.
  - Empirical cross-country evidence: intensity of FXI does not appear to be associated with worse inflation outcomes in the Asian sample (Figure 24).
  - Sterilization costs arise depending on reserves size and the yield gap between foreign and domestic assets; sterilized intervention may prevent domestic interest rates from declining and impede portfolio rebalancing.
- Market development concerns:
  - Extensive use of MPMs, CFMs, and FXI may hamper development of domestic financial markets and FX derivatives needed for hedging.
  - CFMs can reduce cross-border financing that supports liquid domestic markets.
- Trade and competitiveness concerns:
  - FX purchases could be perceived as seeking unfair competitive advantage in export markets, notably if current account surpluses and undervalued exchange rates are present.

### Policy implications and conclusions
- Multiple instruments are used in Asia to manage volatile capital flows and exchange rates; these instruments show complementarities and trade-offs.
- FXI reacts strongly to capital flows, especially volatile flows and where FX exposures are high and financial markets are shallow.
- Monetary policy reflects multiple objectives: inflation stabilization, responses to US interest rates, exchange rate movements, capital flows, and credit growth.
- MPMs respond to domestic macro-financial risks and external influences, and CFMs have primarily targeted housing-related risks.
- Policy frameworks in Asian EMEs often deviate from textbook single-instrument single-objective models; managing flows requires careful calibration across FXI, monetary policy, MPMs, and CFMs while considering market-development and communication costs.

_Italic: Source — Executive Summary, Introduction, Chapters 1 and Annexes (ftdea — "Facing the Tides: Managing Capital Flows in Asia") from the provided IMF PDF content._

### Executive Summary ������������������������������������������������������������������������������������������������������

### Executive Summary

### Introduction
- Managing large and volatile capital flows is a central economic challenge in Asian emerging market economies (EMEs).
- Unconventional monetary policies in advanced economies contributed to record amounts of liquidity that were channeled to EMEs in Asia, with large reversals during tighter global financial conditions (for example, the “taper tantrum”).
- Capital flows can be large compared to the size of domestic financial systems, posing allocation challenges and crisis risks (Gourinchas and Obstfeld 2012).
- Large and volatile capital flows can weaken monetary policy as a stabilization tool and reduce the insulating properties of floating exchange rates (Rey 2015).

### Exchange Rate: Shock Absorber or Amplifier?
- Exchange rate fluctuations can both absorb external shocks and amplify them when domestic financial frictions exist, aggravating corporate vulnerabilities and discouraging investment.
- Amplification is more salient in the presence of foreign exchange (FX) liabilities and in countries with shallow financial markets.
- Empirical evidence (firm-level and macro-level) indicates exchange rate and domestic financial shocks have a significant impact on investment and growth, including higher left-tail risks (Growth-at-Risk).

Key numeric/contextual points from figures and charts:
- Capital flow data span 2000:Q1–2018:Q4 in the analysis of net capital flows to Asian countries.
- Maximum observed net capital flow for Hong Kong SAR is 48 (percent of GDP) and minimum for Malaysia is –37 (percent of GDP) in the charting of quarterly net capital flows (note: these values are presented in the source charts).
- Exchange rate volatility is measured as the coefficient of variation for the daily bilateral exchange rate vis-à-vis the US dollar using three-week rolling windows (2006–19).

### Policy Responses in Asian EMEs
- Policymakers generally employ four main tools: monetary policy (interest rate), exchange rate policy (including foreign exchange intervention, FXI), macroprudential measures (MPMs), and capital flow management measures (CFMs).
- Many Asian EMEs operate as de facto “flexible” inflation targeters with quasi-managed floats and pursue multiple objectives (price, growth, financial stability) using multiple instruments.

Findings on instrument use:
- FXI is used extensively to moderate exchange rate fluctuations in response to volatile capital flows.
  - FXI is more intensively used against more volatile types of flow (for example, portfolio flows), where unhedged balance sheet mismatches are more salient, and where financial markets are shallow.
- Monetary policy reacts to inflation but also to:
  - the US interest rate (consistent with global financial cycle considerations),
  - the exchange rate (particularly where financial markets are relatively less developed),
  - credit growth.
  - This indicates policy interest rate movements reflect multiple objectives.
- Macroprudential and capital flow management measures respond to external, domestic macro, and domestic financial-stability considerations:
  - Drivers include the US policy rate, capital flows, inflation, credit growth, and housing-related risks.
  - Policies (often related to housing) are adjusted amid inflow surges as well as during more normal periods.

Illustrative evidence and chart-based results:
- Foreign exchange intervention absorbs significant amounts of capital flows (Figure references in source).
- Asian monetary policy action often follows the United States (Figure references in source).
- Macroprudential measures were tightened in periods of strong capital inflows.
- CFMs are dominated by inflow measures, often targeting real estate.

### Estimated Policy Reaction Functions
- FXI absorbs a large share of capital flows.
- Monetary policy reaction functions reflect multiple objectives, including inflation, the US policy rate, the exchange rate, and credit growth.
- MPMs respond to domestic macro-financial risks and global factors.
- Housing risks appear to be an important driver of CFM use.
- Specific empirical patterns documented:
  - FXI response is stronger to more volatile flow types and where FX liabilities are high.
  - Monetary tightening tends to follow exchange rate depreciation in shallower financial markets.
  - MPMs have centered on liquidity and credit demand measures.

### Potential Costs of Policy Approaches
- The paper analyzes economic costs stemming from volatile capital flows and exchange rates, and how policy toolkits are deployed in response.
- Volatile flows and exchange rates can lead to:
  - higher leverage and crisis risks,
  - real exchange rate appreciations and deteriorating current account balances during inflow episodes,
  - amplified downturns when inflows reverse,
  - weakened monetary policy effectiveness when portfolio flows influence domestic financial conditions.

### Conclusion
- The data-driven analysis highlights trade-offs and complementarities in the use of FXI, monetary policy, MPMs, and CFMs to manage volatile capital flows and exchange rates in Asian EMEs.
- Findings aim to inform reflections on managing volatile capital flows and exchange rates both in Asian EMEs and more broadly.

*Source: Executive Summary of the IMF paper "Facing the Tides: Managing Capital Flows in Asia" (ftdea - Executive Summary).*

### Introduction

### Introduction

### Purpose and scope
- Aim: shed light on how Asian countries deploy policies to manage external financial shocks.
- Focus: identify common trends in policy responses among Asian EMEs; document stylized facts about the impact of exchange rate fluctuations and changing domestic financial conditions driven by volatile capital flows; describe monetary, exchange rate, macroprudential, and capital flow management policy responses.
- Methods: panel estimation and other techniques to identify common aspects of policy responses.

### Main findings
- Large and volatile capital flows exert significant pressure on exchange rates in Asian EMEs.
- While the exchange rate is normally viewed as an absorber of economic shocks, under certain circumstances it can act as a shock amplifier: exchange rate depreciation can aggravate corporate vulnerabilities and discourage firm-level investment, especially where FX debt is significant and financial markets are shallow.
- Exchange rate depreciation is found to have a significant negative impact on investment and growth at the macro level.
- Faced with a global capital flow cycle, Asian EMEs have deployed multiple policy instruments, often deviating from traditional policy frameworks.
- The management of currency fluctuations using FXI is widespread; FXI is used more heavily when financial markets are shallow and FX liabilities are larger and unhedged, and more intensely for inherently more volatile flows (for example, portfolio flows).
- Monetary policy reacts to standard Taylor rule variables, including inflation and the output gap, and also to external and domestic influences including the US policy rate, capital flows, the exchange rate, and credit growth.
- MPMs (macroprudential measures) are responsive to domestic and external influences including credit growth, inflation, the US policy rate, and capital flows.
- CFMs (capital flow measures) have been adjusted a limited number of times and mostly in response to large increases in domestic house prices.

### Channels: how currency depreciations affect activity
- Competitiveness channel:
  - Depreciation makes exporting goods cheaper, improving international competitiveness and boosting economic activity.
  - Short-term response of trade flows can be asymmetric: depreciation may reduce imports but exert little immediate effect on exports due to trade pricing in dominant currencies.
  - Relationship between exchange rate adjustment and trade flows may be weakened by the buildup of global value chains.
- Financial channels:
  - Financial frictions and balance sheet mismatches are captured via FCIs and balance sheet indicators.
  - Currency mismatches: depreciation increases borrowers’ debt burdens when borrowing in foreign currency.
  - Large foreign investor holdings of local-currency debt mean depreciation may tighten domestic financial conditions, exerting contractionary effects.
  - Aggregate lending channel: weaker bank balance sheets impair lending and reduce firms’ access to finance.
  - Depreciation can negatively affect global value chains (GVCs) via tighter dollar funding and higher financing requirements.

### Micro-level evidence
- Firm-level database: 12 Asian economies, 1994–2016.
- Corporate vulnerability measured by a summary indicator based on profitability, leverage, liquidity, and solvency (proxy for probability of default).
- Key estimates:
  - A 30 percent domestic currency depreciation against the US dollar is associated with 7 percent of firms in the sample moving into the category of high probability of default.
  - The marginal impact of depreciation on corporate vulnerability is stronger when the share of FX debt in firms’ balance sheets is higher.
- Investment effects:
  - Although depreciation is associated with higher firm-level investment on average, for firms with large FX liabilities investment contracts with a depreciation.
  - When firms’ share of FX-denominated debt exceeds 20 percent, an exchange rate depreciation is associated with lower firm-level investment.
  - Effect is more significant in non-exporting firms (no competitiveness benefit or natural hedging).
- Financial development interaction:
  - Degree of financial development determines the extent to which exchange rate volatility is associated with lower investment.
  - In depreciations, higher FX liabilities tend to weaken balance sheets of companies in countries with relatively less-developed financial markets where hedging opportunities are limited, constrained, or costly.
  - No statistically significant impact in countries with more-developed financial markets.

### Macro-level evidence
- Method: Local Projection (LP) method to quantify dynamic effects of exchange rate shocks on investment-to-GDP ratio and growth.
- Key macro estimates:
  - A 1 percent real depreciation lowers the investment ratio by 0.6 percent when markets are shallow (defined as credit-to-GDP ratio below 100 percent).
  - The effect is not statistically significant in countries with more-developed financial markets.
  - Similar pattern for GDP growth: impacts larger in shallow financial markets but more short-lived for GDP than for investment as competitiveness gains accumulate over time.
  - High nominal exchange rate volatility also exerts negative effects on investment and growth, particularly in countries with less-developed financial markets.

### Growth-at-Risk (GaR) findings
- Exchange rate volatility affects the conditional distribution of future GDP growth outcomes and growth-at-risk (left-tail growth).
- Exchange rate fluctuations are an important driver of the growth distribution, controlling for other financial conditions.
- In many less-financially developed economies, the exchange rate affects the growth distribution more than domestic financial conditions (FCIs).
- Global risk-on episodes amplify medium-term left-tail risks in Asia:
  - Near-term easing of financial conditions (domestic loosening or exchange rate appreciation/reduced volatility) helps limit near-term downside risks to growth.
  - However, loose near-term financial conditions and lower exchange rate volatility tend to amplify medium-term downside risks to growth.
  - Estimated effect: a sizable decline in the 5th percentile of the growth distribution over the medium term when subject to favorable near-term external and domestic financial shocks, indicating a significant leftward shift of the left tail.

*Source: ftdea - Introduction (IMF staff estimates and analysis).*

### 1. Shallow Financial Markets

### 1. Shallow Financial Markets

### A bird’s eye view of policy tools and behavior
- Asian EMEs rely substantially on foreign exchange intervention (FXI) when responding to capital flows; countries tend to accumulate foreign reserves when receiving capital inflows and decumulate during outflows.
- Policy interest rates display considerable synchronization across Asian economies, in association with US policy rates.
- Macroprudential policy (MPM) tightened more aggressively and concentrated on liquidity and credit demand instruments during risk-on episodes (pre-GFC and 2010/11); post-GFC easing shifted the pace and toward measures strengthening banking system capitalization.
- Capital flow management measures (CFMs) have been used to a limited extent:
  - CFMs adopted by Asian economies since 2013 can be classified in nine distinct types.
  - CFMs have been used less frequently than MPMs and applied more often to inflows than to outflows.
  - Stamp duties to curb rising property prices have been a favored measure, particularly in advanced economies.
  - CFMs on inflows and outflows have generally been used in a complementary way; easing of CFMs on outflows can be associated with capital account liberalization, while countries tightening inflow CFMs tend to loosen outflow CFMs in tandem, helping mitigate net inflows.

### Policy responses in Asian EMEs (empirical patterns)
- Foreign exchange intervention:
  - Employed in periods of exchange rate appreciation and capital inflows.
  - Statistically significant differences in real exchange rate dynamics and net capital inflows when central banks purchase foreign exchange versus when they do not.
- Monetary policy:
  - Reacts to the output gap: output gaps tend to be slightly more positive when policy is tightened.
  - Responds to inflation and the output gap in standard Taylor-rule fashion, but also to global factors (US policy rate) and capital flows.
- Macroprudential measures:
  - Tightened in periods of more rapid domestic credit growth, real exchange rate appreciation, and positive net capital inflows.
  - Less likely to be tightened when there is a negative output gap.

### Links between instruments and risks (model-free evidence)
- FXI is associated with episodes of reserve accumulation and is responsive to exchange rates and capital flows (real exchange rate, net capital flows, output gap, real credit growth).
- Monetary policy tightening is associated with positive output gaps (Figure 14 evidence).
- MPM tightening is prompted by financial stability and external drivers (Figure 15 evidence): faster credit growth, REER appreciation, and net capital inflows increase the likelihood of tightening.

### Estimated policy reaction functions — key quantitative findings
- Overview:
  - Reaction functions were estimated for policy interest rates, FXI, and MPMs for 13 Asian economies using quarterly data during 2000–18.
  - Regressions control for net capital flows (and components, percent of GDP), global factors (VIX, commodity prices, US policy interest rate), and domestic controls (output gap, inflation, credit growth).
- FXI:
  - FXI absorbs about 70 percent of net capital inflows on average.
  - FXI absorbs a greater proportion of portfolio flows compared to total flows:
    - Asian EMEs purchase about 75 percent of portfolio debt flows.
    - Asian EMEs purchase about 100 percent of portfolio equity flows.
  - FXI is not systematically used to absorb FDI flows.
  - FXI responds more forcefully to outflows where unhedged corporate FX liabilities are relatively large; the interaction term between FX liabilities and depreciation is significantly negative, indicating stronger intervention to dampen depreciation when FX liabilities are higher.
  - Greater financial depth (proxy: private credit-to-GDP ratio) is associated with significantly lower FXI by central banks.
- Monetary policy reaction functions:
  - Inflation significantly influences the policy rate for both Asia as a whole and the EME subsample.
  - The US policy rate is significant in explaining Asian monetary policy responses, consistent with limited monetary independence under a global financial cycle.
  - Monetary policy reacts somewhat to capital flows, especially more volatile types (net portfolio flows, particularly portfolio debt flows into EMEs).
  - Quantified example: a net inflow surge of 10 percent of GDP is associated with a 10 basis point monetary tightening.
  - Exchange rate depreciation tends to induce monetary tightening in countries with shallower financial markets; policy rate tightening in response to depreciation is greater where financial development is lower (private credit/GDP).
  - Policy rates also react to financial-stability considerations, notably rapid credit growth.
- Macroprudential policy reaction functions:
  - MPMs grouped into five categories: bank capital, bank liquidity, credit demand, credit supply, and foreign-currency exposures; LTVs and reserve requirements have been the most popular MPMs in Asian economies.
  - MPM tightening is positively associated with:
    - Net capital inflows (stronger response to nonresident inflows; reversals with resident outflows).
    - Lower US policy rates (countries use MPMs when monetary policy autonomy is constrained).
    - Higher inflation (MPMs help support macroeconomic stabilization).
    - Real credit growth (an acceleration raises probability of MPM tightening, particularly in Asian EMEs).
- CFMs:
  - CFMs are often tightened outside of major capital flow surge episodes (capital flow surge quarters defined as top three deciles of country distribution in 2000–18).
  - CFMs since 2013 have often complemented MPMs and mostly targeted the housing market.
  - CFMs broad enough to significantly reduce inflows have been used infrequently.

*Source: IMF staff estimates, from the chapter "1. Shallow Financial Markets."*

### 1. Main types of MPMs2. Use of MPMs by Type in Asia, 2000–2016

### 1. Main types of MPMs2. Use of MPMs by Type in Asia, 2000–2016

### Use of Macroprudential Measures (MPMs) in Asia
- Source: IMF Macroprudential Measure database; and staff calculations. Note: Asia includes 14 major economies.
- Main instrument labels presented (preserve terminology):
  - Countercyclical + SIFI capital buffers (CCB + SIFI)
  - Leverage Limit (Leverage)
  - Loan loss provisions (LLP)
  - Regulatory capital requirement (Capital)
  - Capital conservation buffer (Conservation)
  - Loan-to-value ratio (LTV)
  - Debt-service-to-income ratio (DSTI)
  - Tax on loans (Tax)
  - Limits on foreign currency lending (LFC)
  - FX open position limits (LFX)
  - Reserve requirement on FX deposits (RR-FCD)
  - Reserve requirement (RR)
  - Liquidity coverage ratio, NSFR (Liquidity)
  - Loan-to-deposit ratio limits (LTD)
  - Credit growth limits (LCG)
  - Loan restrictions (LoanR)
- Observations from figures and notes:
  - Counts presented for use of MPMs by type in Asia, 2000–2016 (no numeric counts reproduced beyond axis labels in source excerpt).
  - Panel distinctions: Asia and Asia EMs.

### Macroprudential Policy Reaction Function (Figure 22)
- Method: Probit coefficient estimates show the marginal probability of a tightening of macroprudential policies associated with each variable.
- Significance notation: *** p < 0.01, ** p < 0.05, * p < 0.1.
- Variables shown affecting probability of MPM tightening:
  - Net capital flows
  - US policy rate
  - Inflation
  - Real credit growth
- Axis range in figure: from –0.2 to 0.2 (coefficients, probability).

### Capital Flow Management Measures (CFMs) and Housing (Figure 23)
- Key findings:
  - CFMs have been frequently used in response to rising housing prices, including through stamp duties and taxes on nonresident property buyers.
  - The measures have been concentrated among AEs in Asia.
  - The cumulative tightening of CFMs is correlated with the average rise in house prices in the countries using them.
  - Tightening of housing-targeted CFMs often involved frequent, but small, policy actions (data count actions without differentiating size or impact).
  - CFMs targeting housing are often used in combination with MPMs.
  - MPMs targeting housing (e.g., LTV and DSTI) constrain financing for house purchases, making them less effective in deterring demand by nonresidents who are less likely to raise financing domestically.
  - Countries that tightened housing-related CFMs did so after already tightening MPMs.
  - In some cases, housing prices continued to rise after credit growth had peaked following MPM tightening, with CFMs continuing to be tightened as the divergence between housing prices and credit growth persisted.
- Figure components:
  - Panel 1: House Prices, MPMs and CFMs (Index, 2012:Q1 = 100; sum). Sample covers Australia, Hong Kong SAR, New Zealand, and Singapore.
  - Panel 2: CFMs and Capital Inflow Surges (Number of measures, 2012–18). Series include Tightening capital inflows and Loosening capital outflows; time axis includes quarters 2012:Q1 through 2018:Q3.
- Data sources: Haver Analytics, CFM Database and IMF staff calculations; IMF 2018 Taxonomy of CFMs and IMF Staff estimates.

### Potential Costs of Policy Approaches
- Coordination and communication costs:
  - Coordinating and communicating policy frameworks that internalize joint calibration of instruments to reach multiple objectives is more challenging than single-objective frameworks.
- FX intervention (FXI) and credibility:
  - Concern: FXI could signal a lack of commitment to the inflation target, potentially de-anchoring inflation expectations (Freedman and Otker-Robe 2010).
  - Theoretical ambiguity: FXI could help achieve the inflation target by adding an instrument (Ostry, Ghosh, and Chamon 2012).
  - Empirical: In Asian economies, intensity of FXI does not appear to be associated with worse inflation outcomes (Ostry and others 2019, Figure 24).
- Sterilization costs:
  - FXI may give rise to sterilization costs depending on the size of accumulated reserves and the yield gap between foreign and domestic assets.
  - Sterilized intervention may prevent domestic interest rates from declining, aggravating the challenge posed by capital inflows by preventing portfolio equilibrium restoration.
- Market development concerns:
  - Extensive use of MPMs, CFMs, and FXI may hamper development and use of domestic financial markets.
  - MPMs can inhibit expansion of financial intermediation that supports economic growth.
  - FXI suppresses FX volatility, which may reduce incentives to hedge foreign-currency risk and make the economy more vulnerable to exchange rate fluctuations (Burnside, Eichenbaum, and Rebelo 2001).
  - CFMs can hamper cross-border financing that supports more robust and liquid domestic financial markets.
  - If CFMs and FXI result in the central bank becoming a dominant player in the foreign exchange market, this can prevent development of FX derivative markets needed to manage exchange rate risk.
- Trade/competitiveness concerns:
  - FX purchases could be interpreted as motivated to gain unfair competitive advantage in export markets, especially in countries with substantial current account surpluses and undervalued exchange rates.

### FX Intervention Intensity and Inflation Outcomes (Figure 24)
- Note: Intervention intensity (standard deviation) is based on foreign exchange reserve flows (proxy for foreign exchange intervention), obtained from the balance of payments statistics, excluding valuation changes. Sample covers 2000:Q1–18:Q4.
- Axis and labels present intervention intensity (standard deviation) against Average inflation (%).
- Country markers listed: Australia, China, Hong Kong SAR, India, Indonesia, Japan, Korea, Malaysia, New Zealand, Philippines, Singapore, Taiwan Province of China, Thailand, Vietnam.
- Conclusion presented: Inflation outcomes not obviously affected by foreign exchange intervention in the sample.

### Conclusion on Policy Reaction Functions and Instruments (Chapters 4–5)
- Managing capital flows to internalize multiple macroeconomic and financial-stability objectives is a difficult challenge for emerging market economies in Asia.
- Capital flows in Asian EMEs can be large and volatile, driven by external developments and investor sentiment in advanced economies.
- Capital flows tend to be expansionary, fueling domestic credit growth and amplifying vulnerabilities, notably from high corporate leverage and unhedged currency exposures.
- Exchange rate can act as an amplifier of shocks in Asian EMEs:
  - Depreciation can increase corporate financial vulnerability where firms have large FX liabilities and limited hedging opportunities, weakening investment and growth.
  - Appreciation can relax firms’ financing constraints, fueling financing booms and increasing vulnerability to shocks.
- Estimated policy reaction functions suggest multiple instruments are used:
  - FXI reacts strongly to capital flows, especially where FX exposures are high and financial markets are shallow.
  - Monetary policy reacts to inflation developments; when US interest rate, capital flows, and exchange rate are included in augmented Taylor rules, these also influence domestic policy interest rates.
  - MPMs respond to domestic macro-financial risks and external influences, including capital flows and US policy rates, and combine with monetary policy to achieve multiple objectives.
  - CFMs have generally been used less frequently in Asian EMEs and mainly against real estate-related risks.
- The multitude of policy responses reflects complexity in managing volatile capital flows; the data-driven analysis aims to inform ongoing policy reflections.

### Annex 1. Corporate Vulnerability in Asian Firms — Methods and Key Findings
- Data and scope:
  - New firm-level data set constructed with comprehensive information on Asian firms’ FX liabilities for 12 Asian economies during 1994–2016.
  - Currency decomposition includes corporate bond issuance and syndicated loans.
  - Inclusion of pre-Asian financial crisis period for benchmark comparison.
  - Annex is based on Jiang and Saadi Sedik (2019).
- Vulnerability metric:
  - Corporate vulnerability defined using the Altman Emerging Market Z’’-score (Z’’-score), an aggregate measure based on profitability, leverage, liquidity, and solvency; lower Z’’-scores imply higher probability of corporate bankruptcy.
  - Altman (2005) correspondence: Z’’-score maps to corporate bond ratings in EMEs.
- Regression specifications:
  - Vulnerability regression (equation (1)):
    - Z_{i,j,t} = α + β1 F_{i,j,t} + β2 M_{j,t} + β3 S_{i,j,t} * ER_{j,t} + β4 S_{i,j,t} * IR_{t} + μ_i + ε_{i,j,t}
    - Definitions:
      - Z_{i,j,t}: firm-level Altman Z’’-score
      - F_{i,j,t}: vector of firm-level variables
      - M_{j,t}: vector of macro variables for country j
      - ER_{j,t}: change in exchange rate at time t (positive = appreciation)
      - IR_{t}: measure of global interest rates (or domestic short-term interest rate)
      - S_{i,j,t}: share of US-dollar denominated debt on firm i’s balance sheet
      - μ_i: company fixed effect
  - Investment regression (equation (2)):
    - Y_{i,j,t} = α + β1 F_{i,j,t} + β2 M_{j,t} + β3 S_{i,j,t} * ER_{j,t} + μ_i + ε_{i,j,t}
    - Y_{i,j,t}: capital expenditure as share of total assets (proxy for firm investment)
- Main empirical results:
  - Exchange rate as shock amplifier:
    - Exchange rate depreciation is associated with a higher probability of default of Asian firms.
    - A 30 percent domestic currency depreciation against the US dollar is associated with a 0.4 decrease in the Z’’-score, corresponding, on average, to a two-notch downgrade in corporate credit rating (for example, from A to BBB+).
    - Such a shock will result in 7 percent of firms in the sample falling into D-rating bucket (or falling into bankruptcy).
    - Results robust to replacing bilateral exchange rate with NEER or REER, and to controlling for VIX.
  - Interaction with FX debt share:
    - The marginal impact of local currency depreciation is stronger when the share of US-dollar denominated debt on firms’ balance sheets is higher.
  - Exchange rate effect on investment:
    - On average, exchange rate depreciation could increase firm-level investment (capturing competitiveness channel).
    - Interaction regressions show the positive impact of depreciation on investment declines as FX liabilities increase; the financial channel offsets competitiveness gains as FX liabilities rise.
  - Thresholds from regressions (Annex Table 1.3 summary):
    - If share of US dollar debt < 10 percent: competitiveness channel dominates.
    - If share of US dollar debt between 10 and 20 percent: financial channel offsets competitiveness channel.
    - If US dollar debt > 20 percent: financial channel dominates.

### Key Regression Table Highlights (Annex Tables 1.1 and 1.2)
- Annex Table 1.1 (Currency Decomposition—Nominal Exchange Rate) — Selected coefficient highlights:
  - Nominal ER coefficient examples: 2.369*** (0.805) in column (1); 1.219** (0.605) in column (3); 2.109*** (0.545) in column (8).
  - U.S. dollar Debt Portion * Nominal ER interaction: 3.949*** (1.471) in column (4); 3.322*** (0.985) in column (5); 3.417** (1.524) in column (6).
  - VIX coefficient examples: –0.021*** (0.0068) in column (1); –0.026** (0.010) in column (3).
  - Observations range reported across columns: 15,614; 15,047; 14,545; 15,614; 14,545; 15,047; 15,047; 15,046; 14,545; 14,545.
  - R-squared values reported: 0.306, 0.304, 0.306, 0.307, 0.307, 0.305, 0.305, 0.305, 0.306, 0.306.
  - Firm FE: Yes across columns.
  - Note: Standard errors in parentheses. *** p , 0.01, ** p , 0.05, * p , 0.1.
- Annex Table 1.2 (Investment—Nominal Exchange Rate) — Selected coefficient highlights:
  - U.S. dollar Debt Portion coefficients: –0.00031 (0.00138) in column (1); –0.0011 (0.00144) in column (3).
  - Nominal Exchange Rate coefficient examples: –0.02 (0.0167) in column (1); –0.0281* (0.0169) in column (2); –0.0520*** (0.0181) in column (5).
  - U.S.D. Debt Portion * Nominal ER interaction: 0.0312** (0.0151) in column (3); 0.0297** (0.0152) in column (4).
  - Leverage coefficient: –0.00526*** (0.00103) in column (4).
  - Tobin’s Q coefficient: 0.0120*** (0.00106) in column (1).
  - Cash coefficient: 0.0672*** (0.0215) in reported column.
  - Observations across columns: 19,960; 19,202; 19,202; 18,395; 17,527.
  - R-squared values: 0.157, 0.166, 0.166, 0.163, 0.163.
  - Firm FE: Yes across columns.
  - Note: Subsample uses 1,420 firms covering tradable and nontradable industries. Investment defined as capital expenditure / lagged total assets; leverage as total debt / lagged total assets; cash as cash holdings / lagged total assets. Tobin’s Q is total market value / total book value lagged by one period. USD debt portion formula provided. Standard errors in parentheses. *** p , 0.01, ** p , 0.05, * p , 0.1.

*Source: IMF staff calculations; content extracted from "FACING THE TIDES: MANAGING CAPITAL FLOWS IN ASIA."*

### Annex Table 1.3. Regressions by USD Debt Proportion

### Annex Table 1.3. Regressions by USD Debt Proportion

### Regression results (Investment as dependent variable)
- Specification: Investment regressed on USD debt portion, ΔNominal Exchange Rate, Leverage, Tobin’s Q, Cash, and constant; firm fixed effects included.
- Sample partitions by USD Debt Proportion:
  - ,51.5%
  - [1.5%, 4.5%]
  - [4.5%, 10.8%]
  - [10.8%, 20.3%]
  - [20.3, 35.9]
  - .35.9%
- Key coefficient estimates by column (coefficient (standard error)):
  - USD debt portion:
    - (1) 1.207 (1.253)
    - (2) 0.159 (0.238)
    - (3) 0.257* (0.134)
    - (4) 0.0328 (0.0957)
    - (5) 20.166 (0.159)
    - (6) 20.640 (0.561)
  - ΔNominal Exchange Rate:
    - (1) 20.064** (0.0318)
    - (2) 20.059** (0.0243)
    - (3) 20.072** (0.0298)
    - (4) 0.041 (0.0396)
    - (5) 0.011* (0.00504)
    - (6) 0.024** (0.00841)
  - Leverage:
    - (1) 20.00440** (0.00171)
    - (2) 20.0063*** (0.00162)
    - (3) 20.0118*** (0.00220)
    - (4) 20.00425** (0.00215)
    - (5) 20.013*** (0.00422)
    - (6) 20.009*** (0.00210)
  - Tobin’s Q:
    - (1) 0.00870*** (0.00141)
    - (2) 0.0316*** (0.00399)
    - (3) 0.0392*** (0.00473)
    - (4) 0.0437*** (0.00545)
    - (5) 0.0476*** (0.00788)
    - (6) 0.0319*** (0.00460)
  - Cash:
    - (1) 0.0744** (0.0332)
    - (2) 0.0586 (0.0442)
    - (3) 0.0834* (0.0455)
    - (4) 0.153*** (0.0483)
    - (5) 0.0754 (0.0842)
    - (6) 0.0562 (0.0602)
  - Constant:
    - (1) 0.0582*** (0.00389)
    - (2) 0.0413*** (0.00882)
    - (3) 0.0335*** (0.0116)
    - (4) 0.0291* (0.0158)
    - (5) 0.00367 (0.0403)
    - (6) 0.0598*** (0.00793)
- Sample sizes and fit:
  - Observations:
    - (1) 10,579
    - (2) 1,381
    - (3) 2,081
    - (4) 1,789
    - (5) 837
    - (6) 1,482
  - R-squared:
    - (1) 0.157
    - (2) 0.707
    - (3) 0.653
    - (4) 0.639
    - (5) 0.681
    - (6) 0.526
- Notes:
  - Firm FE: Yes for all columns.
  - Standard errors in parentheses. *** p , 0.01, ** p , 0.05, * p , 0.1.

### Interpretation and substantive findings
- ΔNominal Exchange Rate:
  - Negative and statistically significant effects on investment in several USD debt proportion partitions (columns (1)–(3): 20.064**, 20.059**, 20.072**), indicating local currency depreciation is associated with lower investment for those partitions.
  - Positive and statistically significant small coefficients in columns (5) and (6) (0.011* and 0.024**), suggesting heterogeneity across USD debt proportion groups.
- Leverage:
  - Consistently negative and statistically significant across all partitions (coefficients range from 20.00425** to 20.013***), indicating higher leverage relates to lower investment.
- Tobin’s Q:
  - Positive and highly significant across all partitions (coefficients range from 0.00870*** to 0.0476***), indicating expected profitability is a robust positive predictor of investment.
- Cash:
  - Positive coefficients with varying significance across partitions (notably 0.153*** in column (4)), suggesting liquidity supports investment but effect varies by USD debt share.
- USD debt portion:
  - Coefficients vary in sign and significance across partitions; only column (3) shows a positive statistically significant coefficient (0.257*), while larger USD debt proportion partitions (columns (5) and (6)) show negative point estimates (20.166 and 20.640) that are not statistically significant at conventional levels in the table.

### Methodology (summary)
- Estimation: Fixed-effects panel regressions of firm-level investment ratio on USD debt portion and controls, partitioned by firms’ USD debt proportion bands.
- Controls include ΔNominal Exchange Rate, leverage, Tobin’s Q, cash, and firm fixed effects.
- Significance levels reported as: *** p , 0.01, ** p , 0.05, * p , 0.1.

### Key implications
- Heterogeneous investment responses to exchange rate movements across firms grouped by USD debt share:
  - Some partitions display negative investment responses to depreciation, consistent with a balance-sheet channel.
  - The varied sign and significance of USD debt portion coefficients across partitions imply the relationship between dollarization of liabilities and investment is non-linear and dependent on the relative share of USD debt.
- Policy-relevant indicators:
  - Leverage and Tobin’s Q are robust predictors of investment and thus important targets for firm-level resilience assessments.
  - Liquidity (cash) matters for investment, particularly for certain USD debt share groups.

*Annex Table 1.3. Regressions by USD Debt Proportion (source content).*

### 1. China2. India

### ftdea - 1. China2. India

### Exchange Rate Volatility and Growth at Risk — Short and Medium Term Effects
- Annex Figure 4.1 and Annex Figure 4.2 present one-year-ahead growth distributions (compound annual growth rate) and changes in the left tail (GAR 5 percent) of the growth distribution in response to shocks to domestic financial conditions and exchange rate volatility.
- Key empirical observations:
  - The exchange rate shock does not shift the distribution as much as a domestic financial conditions shock in many cases, but effects are heterogeneous across countries.
  - For most countries in the sample, the GaR 5 percent indicator (the position of the left tail, 5th percentile, of the growth distribution) declines over the medium term in response to:
    - easing of domestic financial conditions, and
    - lower exchange rate volatility/depreciation pressures.
  - The medium term is defined as the number of quarters ahead (8, 12, or 16) in which the decline in the GAR5% is the largest relative to the near term (four quarters ahead).
- Transmission channels highlighted as explanations for heterogenous effects:
  - Corporate balance sheet and uncertainty channels (countries with high unhedged FX liabilities or higher aversion to exchange rate volatility may see lower investment and growth).
  - Competitiveness channel (a more depreciated and flexible exchange rate may make exports more competitive, potentially shifting growth distribution rightward in some cases, e.g., Malaysia).
  - Capital flows channel (tighter domestic financial conditions can attract capital flows, temporarily boosting growth).
- Policy implication from these patterns:
  - Easing of financial conditions — whether via exchange rate or domestic FCI — may boost short-run growth but can build financial vulnerabilities over the medium term and raise tail risks to growth.

### Data, Samples, and Reaction Function Framework
- Sample and frequency:
  - Data set of 13 Asian economies: Australia, China, Hong Kong SAR, India, Indonesia, Japan, Korea, Malaysia, New Zealand, the Philippines, Singapore, Taiwan Province of China, and Thailand.
  - Quarterly data from the first quarter of 2000 to the fourth quarter of 2018.
- Baseline reaction functions:
  - Equation (1): Y_it = α + β K F_it + Σ_j=1^J δ_j Z_jt + Σ_k=1^K γ_k X_kit + μ_i + ε_it — analyzes response of FXI and monetary policy.
  - Equation (2): Pr(Policy change = 1)_it = F( α + β K F_it + Σ_j=1^J δ_j Z_jt + Σ_k=1^K γ_k X_kit + μ_i + ε_it ) — probit specification for MPMs where policy change is a dummy equal to 1 when policy is tightened.
- Variable groups used (Box 1):
  - KF: net capital flows (percent of GDP) or its components.
  - Z: global factors (VIX, commodity prices, U.S. real interest rates) and a global financial crisis dummy.
  - X: domestic controls (real GDP growth, inflation, real credit growth, real housing price increase, output gap, financial development, foreign currency liabilities).
  - Country fixed effects, seasonal dummies, and instrumentation strategies described (e.g., net flows to other regional countries used as instruments).

### FXI (Foreign Exchange Intervention) Results — Quantitative Findings
- FXI absorbs on average 70 percent of net inflows.
- EM subgroup specifics:
  - EM central banks react stronger to net capital flows; coefficients largest for EM subgroups.
  - EMEs purchase 60 percent of inflows (nonresidents) but sell less than 20 percent of outflows (residents).
- Flow-type sensitivity:
  - EMEs purchase about 75 percent of portfolio debt flows, and nearly 100 percent of equity flows.
  - No significant evidence of FXI purchases for FDI flows (coefficient not significant).
- Interaction with FX-denominated corporate debt:
  - In regressions including an interaction between foreign currency liabilities (FXL) and the exchange rate, the interaction term is negative and statistically significant for the Asia and EMEs excluding China samples — implying countries with high corporate FX debt sell more reserves to stem depreciation.
- Financial depth:
  - Higher financial development (private sector credit scaled to GDP) is associated with lower FXI levels; countries with deeper financial markets rely less on FXI.
- Credit growth:
  - Credit growth is not statistically significant in the FXI regression for the general sample or subgroups — suggesting FXI does not react to higher credit growth in the sample.

### Monetary Policy Reaction Results — Quantitative Findings
- Policy rates respond to inflation and output gap:
  - Policy rates respond to inflation in both advanced economies and EMEs (coefficients statistically significant for the Asia sample and EM sample).
  - The output gap coefficient is statistically significant for the Asia sample but not for the EM sample.
- External influences:
  - The US policy interest rate is statistically significant in explaining policy rates in Asia — indicating restricted monetary policy independence.
  - In Asian EMEs, policy rates also respond to changes in the real effective exchange rate: an appreciation leads to loosening; a depreciation leads to tightening, over and above contemporary inflation effects.
- Capital flow composition:
  - Policy rates react to volatile types of capital flows (nonresident inflows and net portfolio flows, especially net portfolio debt flows), but not to FDI flows.
  - The quantitative response of the policy rate to capital flows is small but statistically present for volatile flows.
- Financial development and depreciation episodes:
  - The marginal impact of exchange rate depreciation on policy rate actions is stronger when countries are less financially developed.
- Credit growth:
  - For EMEs, private sector credit growth has a positive and statistically significant effect on policy rates — suggesting monetary policy is used to respond to financial stability concerns as well as standard mandates.

### Macroprudential Measures (MPMs) Reaction Function — Findings
- MPM measure:
  - Aggregate MPM measure constructed by summing tightening and loosening actions across 17 individual macroprudential instruments and their subcategories (database by Alam and others (2019)).
  - Dependent variable coded as “1” for a tightening action in a quarter and “0” otherwise; probit panel estimator used for a panel of 13 Asian countries (first quarter 2000 to fourth quarter 2016 in MPM regressions).
- Main results on drivers of MPM tightening probability:
  - Capital inflows:
    - A net capital inflow surge raises the probability of an MPM tightening in Asia; this response is significant for EMEs only, not for AEs.
    - Disaggregation shows EMEs significantly tighten MPMs only in response to an increase in external liabilities.
  - Global monetary shocks:
    - MPMs are tightened with an easing of global monetary policy (cuts in the US policy rate) — significant for AEs but not robust across all IV specifications.
  - Global financial crisis:
    - The Global Financial Crisis significantly reduced the probability of an MPM tightening, but only in AEs.
  - Inflation, REER, and growth:
    - Inflationary pressures raise the probability of MPM tightening in Asia — significant for AEs and in some IV specifications for EMEs.
    - MPMs respond to contemporary inflation but not to the REER (in contrast to monetary policy).
    - Evidence of a direct response of MPMs to real GDP growth is weaker.
  - Real credit growth:
    - An acceleration in credit growth raises the probability of MPM tightening in EMEs, although the response is relatively weak and significant only at the 10 percent level in some specifications.
    - A credit gap variable was significant in many specifications, but data limitations prevent widespread reporting.
- Estimation notes:
  - Probit estimates for tightening actions produced stronger and more consistent results than logit or estimations including both tightening and loosening actions (most observed actions in the sample were tightenings).
  - Instrumental Variables regressions were used to correct for potential endogeneity; differences between IV and non-IV estimates are small.

*Italic: Source — Annexes 4–5 (Exchange Rate Volatility and Growth at Risk; Estimating Policy Reaction Functions) from the provided IMF content unit.*

### Annex 5. Estimating Policy Reaction Functions

### Annex 5. Estimating Policy Reaction Functions

### Foreign exchange intervention response (Tables 5.1–5.3)
- Net capital flows (% GDP) are positively and statistically significantly associated with foreign exchange intervention:
  - Asia FE: 0.67*** (0.067)
  - Asia IV: 0.66*** (0.044)
  - Emerging Market Economies FE: 0.77*** (0.118)
  - Emerging Market Economies IV: 0.99*** (0.091)
- Other covariates reported (coefficients and robust standard errors):
  - US interest rate: 0.135 (0.153); 0.465 (3.280); 0.416* (0.170); 0.972 (5.092)
  - VIX: 0.182 (0.395); 0.392 (2.581); 1.155* (0.561); 20.410 (4.112)
  - Commodity prices: 0.418 (0.698); 2.035 (14.767); 0.561 (1.144); 20.432 (22.987)
  - GFC dummy: 0.635 (0.921); 6.346 (7.863); 0.426 (0.898); 10.597 (12.121)
- Sample and fit:
  - Observations: 981 (Asia specifications) and 453 (EM specifications)
  - R-squared (selected): 0.623, 0.571
  - No. of countries: 131, 366 (varies by specification)
- Intervention responses by composition of flows (Annex Table 5.2):
  - Capital flows influence inflows (liabilities) and several flow components differently in Asia and Asia EMs. Selected significant coefficients (coefficient (std. error)):
    - Inflows (Asia): 0.050** (0.022)
    - Inflows (Asia EMs): 0.622*** (0.114)
    - Net portfolio (Asia EMs): 0.770*** (0.028)
    - Net equity (Asia EMs): 1.040** (0.291)
  - US interest rate shows several positive and sometimes significant coefficients across inflows/outflows and components (e.g., 0.355** (0.112) for Inflows Asia EMs).
  - VIX often positive; examples: 1.970*** (0.405) for Inflows Asia EMs.
  - GFC dummy coefficients vary, with some significant negative coefficients for Outflows in Asia (e.g., 23.138* (1.640)).
  - Robust standard errors in parentheses. Significance: *** p , 0.01, ** p , 0.05, * p , 0.1.
- FX liabilities, exchange rate interaction, and credit (Annex Table 5.3):
  - FX Liabilities (FXL), % GDP: negative and significant in several specifications:
    - FXL effect (Asia, FXI and FXL): 20.050*** (0.014)
  - Exchange rate change (increase = depreciation) is associated with reduced intervention in several specs:
    - Example: 20.110** (0.045)
  - Interaction term FXL X exchange rate: 20.013* (0.007) in one specification.
  - Credit (% of GDP) positively associated with FX intervention in some specs:
    - Credit (Asia, FXI and Financial Development): 26.062** (2.301)
  - Observations and R-squared reported per specification (e.g., Observations 969, R-squared 0.638).

### Monetary policy reaction functions (Tables 5.4–5.6)
- Monetary policy response to net capital flows (Annex Table 5.4):
  - Net capital flows, % of GDP: coefficients include 20.002 (0.006), 0.009 (0.008), 0.006 (0.007), 0.031* (0.017)
  - US interest rate significant and positive in fixed-effects specifications:
    - 0.072*** (0.018); 0.078** (0.026)
  - GFC dummy strongly negative and significant:
    - 20.871*** (0.187), 21.102** (0.335)
  - Lagged dependent variable large and significant:
    - 0.812*** (0.036); 0.741*** (0.045)
  - CPI positive and in some specs significant:
    - 0.119** (0.049); 0.200** (0.073)
  - REER often negative and sometimes significant:
    - 20.029* (0.012); 20.027*** (0.008)
  - Observations: 956 (Asia) and 441 (EMs) in specified panels. R-squared: 0.947, 0.938.
- Monetary policy response by composition of flows (Annex Table 5.5):
  - Capital flows coefficients on policy rate changes (selected):
    - Inflows (liabilities) Asia: 20.000 (0.002)
    - Inflows (liabilities) Asia EMs: 0.008* (0.004)
    - Net portfolio (Asia): 0.006** (0.002)
    - Net portfolio (Asia EMs): 0.008*** (0.002)
    - Net debt and Net FDI coefficients reported (some significant in EMs).
  - US interest rate robustly positive and significant across many specifications:
    - e.g., 0.072*** (0.018); 0.079** (0.026)
  - VIX generally negative but not consistently significant (e.g., 20.166 (0.110)).
  - Commodity prices negative and significant in some EMs specifications:
    - 21.372** (0.451); 21.325** (0.491)
  - GFC dummy negative and highly significant across specifications:
    - 20.866*** (0.184); 21.094** (0.329)
  - Lagged dependent variable consistently large and significant:
    - 0.813*** (0.036); 0.742*** (0.045)
  - Output gap positive and significant in some Asia specs:
    - 1.937** (0.745)
  - CPI positive and significant in many specifications:
    - 0.119** (0.050); 0.200** (0.073)
  - REER small negative coefficients (e.g., 20.030* (0.012))
  - Observations and R-squared: e.g., Observations 956, R-squared 0.947.
- Monetary policy, financial development, and credit growth (Annex Table 5.6):
  - Credit (% of GDP) negative and significant in multiple specifications:
    - 20.694*** (0.166); 20.852** (0.302)
  - Interaction exchange rate # Credit (% GDP): 20.043** (0.015); 20.039* (0.019)
  - Net capital flows coefficients near zero in these specs (e.g., 20.003 (0.006), 0.003 (0.007))
  - US interest rate positive and significant in several specs:
    - 0.062*** (0.020); 0.075*** (0.021)
  - GFC dummy negative and significant:
    - 20.805*** (0.227); 20.873*** (0.254)
  - Lagged dependent variable large and significant (e.g., 0.794*** (0.034))
  - CPI positive and significant: 0.120** (0.048)
  - Credit growth (yoy) sometimes positive and significant:
    - 0.020*** (0.001); 0.020*** (0.002)
  - Observations and R-squared: e.g., Observations 882, R-squared 0.947.

### Macroprudential measure (MPM) reaction function estimations (Table 5.7)
- Net capital flows (/GDP) associated with higher probability of MPM activation in some baseline Asia specifications:
  - 0.022** (0.010) in Baseline Asia
  - 0.038** (0.019) in Baseline EMs
  - 0.062** (0.026) in Baseline Asia (alternative column)
- External liability flows (/GDP) and external asset flows have reported coefficients (selected):
  - External liability flows (/GDP): 0.022** (0.010) in Baseline Asia
  - External asset flows (/GDP): −0.022** (0.009) in one reported line (note negative sign in table layout)
- US interest rate (reported sign conventions are negative in table layout) significant in some Baseline AEs and Baseline Asia specifications:
  - e.g., −0.105*** (0.038) in Baseline AEs
- GFC dummy generally strong in some specifications (e.g., −5.305*** (0.576) in Baseline AEs), with variation across regions and probit IV specifications
- Real GDP growth, inflation, credit growth, house price growth enter as controls:
  - Inflation positive and significant in several baseline specifications:
    - 0.092** (0.040) in Baseline Asia
    - 0.062** (0.028) in Baseline AEs
  - Credit growth sometimes positive and marginally significant in EMs:
    - 0.034* (0.020) in Baseline EMs
  - Probit IV specifications: selected significant coefficients (e.g., Real GDP growth 0.163*** (0.054) in Probit IV - Asia)
- Model fit and sample:
  - Observations vary (e.g., 784 for Baseline Asia, 408 for Baseline AEs/EMs)
  - Pseudo R2 examples: 0.176 (Baseline Asia), 0.228 (Baseline AEs), 0.152 (Baseline EMs)
  - Robust standard errors in parentheses. Significance: *** p , 0.01, ** p , 0.05, * p , 0.1.

### Key empirical patterns and implications (synthesis of tabulated results)
- Foreign exchange intervention is strongly responsive to net capital flows (% GDP) across Asia and emerging market economies, with coefficients consistently positive and statistically significant (e.g., 0.67***, 0.66***, 0.77***, 0.99***).
- Composition matters: inflows (liabilities), net portfolio, and net equity flows show stronger associations with intervention than some other flow types in certain specifications (e.g., inflows Asia EMs 0.622*** (0.114); net portfolio Asia EMs 0.770*** (0.028); net equity Asia EMs 1.040** (0.291)).
- Monetary policy reaction functions show:
  - Strong persistence (lagged dependent variable ~0.74–0.83).
  - Positive sensitivity to US interest rates in several fixed-effects specifications (e.g., 0.072*** (0.018)).
  - A consistent negative GFC dummy effect on policy rates (e.g., 20.871*** (0.187)).
  - CPI commonly enters positively and significantly (e.g., 0.119** (0.050)).
- Financial development (credit levels) and interactions with exchange rate movements condition policy responses:
  - Credit (% of GDP) is negatively associated with monetary policy adjustment in several specifications (e.g., 20.694*** (0.166)).
  - Interaction exchange rate # Credit (% GDP) is negative and significant in some specs (e.g., 20.043** (0.015)), suggesting stronger contractionary policy response when exchange-rate depreciation coincides with higher credit.
- Macroprudential policy activation correlates with net capital flow measures and macrofinancial indicators:
  - Net capital flows (/GDP) positively associated with MPM activation in several baseline specifications (e.g., 0.022** (0.010)).
  - Inflation and real GDP growth appear as significant controls in some specifications; credit growth and house price growth have mixed effects.

*Annex 5. Estimating Policy Reaction Functions — tables and coefficients reproduced from the source PDF.*

---


_Source: https://www.imf.org/-/media/files/publications/dp/2019/english/ftdea.pdf_