## _wp15127

## Source details

**Canonical URL:** [_wp15127](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15127.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15127.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15127.pdf.json)

---

### I. Introduction — scope, question, and contribution
- Focus: systematic analysis of the sensitivity of 34 EMs to global push factors using quarterly balance-of-payments (BOP) data for the period 2001Q1-2013Q4.
- Key question: Why do some EMs respond more strongly than others to changes in global (push) factors?
- Main contributions:
  - Use of a latent factor model (in the spirit of Kose et al. (2003)) to extract unobserved common dynamics in gross inflows (total and by component).
  - Disaggregation of gross inflows into FDI, Portfolio Equity, Portfolio Bonds, Other Investment (OI) to Banks, and OI to Non-Banks.
  - Identification that commonality in inflows is sizable but heterogeneous across asset types and countries.
  - Finding that financial market characteristics and composition of foreign investor base (rather than macroeconomic or institutional fundamentals) most robustly explain cross-country heterogeneity in sensitivities to push factors.

### II. Data, definitions, and empirical setup
- Sample and measurement:
  - Data: Gross capital inflows from IMF’s Balance of Payment database; components: FDI, Portfolio Equity, Portfolio Bonds, OI to Banks, OI to Non-Banks.
  - Sample period: 2001Q1-2013Q4 (section later notes sample coverage 2001Q2-2013Q4 and total observations 50 in some regressions).
  - Sample size: 34 emerging markets.
  - All series measured in U.S. dollars and normalized by recipient country GDP (quarterly).
- Key definitions and controls:
  - Capital “inflows” = changes in domestic financial liabilities vis-à-vis non-residents (residence criterion).
  - Push factors: (i) average GDP growth rate in four core economies (U.S., Euro Area, Japan, and U.K.); (ii) US VIX; (iii) changes in the expected U.S. policy rate; (iv) slope of the U.S. yield curve (10 year minus 3 month U.S. government T-bill yields); (v) U.S. real effective exchange rate (REER).
  - Type-specific push variables:
    - Banking inflows: U.S. dealer bank leverage and TED spread.
    - Bond inflows: 10-year U.S. government bond yield; (lagged) return of the EMBI.
    - Equity inflows: (lagged) return in the MSCI emerging market index.
  - Pull controls: index of aggregate commodity prices; (lagged) aggregate real GDP growth in EMs.
- Market-structure and investor-base proxies:
  - New proxy approach: for each EM and asset, compute correlation between BOP inflows and investor-reported inflows (EPFR Global for funds; BIS locational international banking statistics for global banks). High correlation interpreted as sign that those investors account for most movements in BOP-recorded inflows.

### III. Econometric framework and estimation details
- Model specification:
  - Dynamic latent factor model: y_{i,t} = β_{i,W} f_t^W + β_{i,R} f_{j,t}^R + υ_{i,t}, with latent world factor f_t^W and regional factor f_{j,t}^R.
  - Idiosyncratic factors follow AR(p); factors follow AR(q); implementation sets p = 2 and q = 2.
- Identification and priors:
  - Normalize sign for first-country loadings and one country per region; factor variances normalized to 1.
  - Conjugate priors with IG(6,0.001) on innovation variances.
- Estimation:
  - Bayesian MCMC with data augmentation; posterior based on 300,000 MCMC replications after 30,000 burn-in replications.

### IV. Main empirical results — factor estimates and heterogeneity
- Aggregate result:
  - Gross inflows to EMs co-move substantially across countries due to global (push) factors, but magnitudes vary substantially across countries and asset types.
- Asset heterogeneity:
  - Portfolio Equity, Portfolio Bond, and OI to Banks co-move substantially across EMs.
  - FDI and OI to Non-Banks show no identifiable commonality structure in this model.
- Cross-country heterogeneity in factor loadings and variance decomposition:
  - Strong dispersion in β_{i} and θ_{i} (share of variance attributable to common EM dynamic).
  - Examples:
    - Equity flows: a unit standard deviation in the common EM factor causes a unit standard deviation in equity flows to Pakistan and 0.6 standard deviation in equity flows to India; Latvia, Estonia, Belarus show insignificant responses.
    - Bond flows: Indonesia and South Africa bond flows react strongly; China, Colombia, Bulgaria are almost insensitive.
  - Country groupings:
    - “High sensitivity” (sensitive across components): Brazil, Indonesia, Thailand, Turkey, South-Africa.
    - “Asymmetric” (sensitive in one or two components): Pakistan, Philippines, India, Mexico.
    - “Insensitive” (low sensitivity across components): Estonia, Latvia, Chile.
  - Notable statistic: for Pakistan and Philippines, more than half of the variance in equity funding is accounted for by the common EM factor.
- Variance decomposition examples (selected exact shares reported in source):
  - India: All inflows Global 55% Regional 4%.
  - Mainland China: All inflows Global 28% Regional 1%.
  - Indonesia: Portfolio Bond Global 43% Regional 9%; All inflows Global 21% Regional 8%.
  - Latin America averages: All inflows Global 14% Regional 15%.
  - Emerging Europe averages: All inflows Global 10% Regional 43%.
  - Other averages: All inflows Global 24% Regional 29%.

### V. What drives cross-country heterogeneity (regressions on betas)
- Strategy:
  - Regress estimated factor loadings (beta_i^asset) on fundamentals and market-structure variables separately and jointly; confirm with Bayesian Model Averaging (BMA).
  - Cross-sectional sample: 34 countries per asset.
- Key findings (selected exact coefficients and inference as reported):
  - Push vs pull in explaining common EM factors (time-series regressions, Observations = 50):
    - US VIX: consistently negative and significant (examples: Column [1] -0.0313*** (0.0105); Column [5] -0.0516*** (0.00997); Column [13] -0.0397*** (0.0108)).
    - US REER: consistently negative and significant (examples: Column [1] -0.0592*** (0.00995); Column [3] -0.0855*** (0.0188); Column [15] -0.0617*** (0.0182)).
    - Commodityprice_pch: positive and significant across many specifications (example: Column [1] 0.0561*** (0.0119)).
    - Push factors account for large shares of explanatory power: about 75 percent of the overall R2s for Portfolio equity flows; about 65 percent for OI banks; about 60 percent for Portfolio bonds (group-level R2s).
  - Cross-country regressions explaining betas (selected coefficients preserved exactly):
    - Equity and bank flows: higher betas do not robustly coincide with weaker fundamentals (no robust link to lower growth, higher debt, or poor institutions).
    - Bond flows: higher sensitivity associated with lower reserves, higher trade openness, and more flexible FX regimes in some specifications.
    - Market-structure and investor-base variables:
      - Correlation with EPFR/BIS flows: quantitatively large and highly significant predictors.
        - Going from zero to perfect correlation increases predicted response to a shock in the common EM factor by 0.45 for equity (reported in text), 0.24 for bonds, and 0.75 for banks.
      - Equity market liquidity and index inclusion:
        - Turnover Ratio and MSCI Frontier Benchmark dummy associated with higher Equity betas (example: Turnover Ratio Column (2) 0.00176***; MSCI Frontier Benchmark Column (2) 0.228***).
    - R-squared performance:
      - Equity Beta regressions: R-sq 0.244 (col 1), 0.546 (col 2), 0.521 (col 3).
      - Bond Beta regressions: R-sq 0.409 (col 4), 0.288 (col 5), 0.429 (col 6).
      - Bank Beta regressions: R-sq 0.324 (col 7), 0.520 (col 8), 0.530 (col 9).
  - Bayesian Model Averaging (selected entries preserved exactly):
    - FX Regime (Bond): Coef. 0.019; t-Stat 1.34; PIP 0.73.
    - Correlation measures and turnover/benchmark dummies remain among variables with higher post-inclusion probabilities in many specifications.

### VI. Robustness checks and auxiliary evidence
- Betas capture sensitivity beyond the GFC:
  - Countries with higher betas experienced deeper retrenchments during both the GFC and the Taper Tantrum (illustration in Figure 5).
- Model uncertainty:
  - BMA confirms robustness of variables highlighted in cross-sectional regressions (Table 7 results).
- Institutional quality:
  - No robust relationship between correlation proxies (EPFR/BIS correlations) and institutional measures (ICRG Law and Order, Investor Protection) (illustrated in Figure 6).

### VII. Interpretations and policy implications
- Core interpretation:
  - Financial market structure and the composition of the foreign investor base drive cross-country heterogeneity in sensitivities to global push factors more than standard macroeconomic or institutional fundamentals.
  - EMs with deeper, more liquid markets and greater reliance on international mutual funds and global banks are more exposed to inflow reversals when global financial conditions deteriorate and benefit more when conditions improve.
- Policy recommendations:
  - Good macroeconomic fundamentals (higher growth, lower public debt, better institutions) do not guarantee insulation from global financial shocks.
  - EM authorities should collect and use information on their foreign investor base and market structure (including proxies such as correlations with EPFR and BIS series, turnover ratios, index inclusion dummies) to assess sensitivity to push factors.
  - Further research and microdata collection recommended to understand the macroeconomic transmission at country level, including the role of domestic investors absorbing assets during stress episodes and the interaction between flows and asset prices.

### VIII. Selected descriptive statistics (exact values reported in source)
- Fundamentals (Obs = 34 unless stated):
  - Trade Openness: Obs 34; Mean 80.52; Std. Dev. 41.22; Min 23.99; Max 187.30
  - Public debt: Obs 34; Mean 40.69; Std. Dev. 19.08; Min 5.95; Max 79.21
  - Reserves: Obs 34; Mean 19.58; Std. Dev. 9.37; Min 6.96; Max 44.33
  - Exchange Rate Regime: Obs 34; Mean 8.29; Std. Dev. 3.04; Min 2.00; Max 13.00
  - Average Growth: Obs 34; Mean 4.76; Std. Dev. 1.73; Min 0.78; Max 9.94
  - Investor Protection: Obs 34; Mean 8.83; Std. Dev. 1.73; Min 3.69; Max 11.23
  - Rule of Law: Obs 34; Mean 3.59; Std. Dev. 1.01; Min 1.54; Max 5.00
- Equity market measures:
  - Foreign Openness -Equity: Obs 34; Mean 6.88; Std. Dev. 7.01; Min 0.07; Max 24.59
  - Stock Market turnover: Obs 33; Mean 49.59; Std. Dev. 59.50; Min 0.96; Max 226.99
  - Share of Equity Funding from Advanced Economies: Obs 34; Mean 67.19; Std. Dev. 22.72; Min 12.48; Max 95.67
  - BOP Equity correlation with EPFR flows: Obs 34; Mean 0.24; Std. Dev. 0.26; Min -0.41; Max 0.73
- Bond market measures:
  - Foreign Openness -Bond: Obs 34; Mean 10.04; Std. Dev. 6.48; Min 0.12; Max 30.47
  - Share of Bond funding from Advanced Economies: Obs 34; Mean 67.27; Std. Dev. 16.74; Min 23.25; Max 92.09
  - BOP Bond correlation with EPFR flows: Obs 34; Mean 0.26; Std. Dev. 0.26; Min -0.55; Max 0.67
- Banking market measures:
  - Foreign Openness - Other Investment: Obs 34; Mean 36.29; Std. Dev. 19.08; Min 4.32; Max 96.40
  - Private Credit/GDP: Obs 32; Mean 63.39; Std. Dev. 29.11; Min 24.57; Max 138.60
  - BOP OI-Bank correlation with BIS flows: Obs 34; Mean 0.13; Std. Dev. 0.22; Min -0.52; Max 0.59

*Source: _wp15127 (PDF).*

### 1. Sample of Countries ______________________________________________________22

### _wp15127 - 1. Sample of Countries ______________________________________________________22

### Major Sections
- 1. Sample of Countries ______________________________________________________22
- 2.  Variable Definitions, Frequency and Sources __________________________________23
- 3.  Raw Statistics ___________________________________________________________25
- 4.  Variance Decompositions  _________________________________________________26
- 5.  Drivers of the Estimated EM Common Factors _________________________________28
- 6.  Explaining Countries’ Sensitivities to Push Factors  _____________________________29
- 7.  Bayesian Averaging Results  _______________________________________________30

### Figures
- 1.  Inflows to EMs – BOP Raw Data  ___________________________________________31
- 2.  Common EM Factors – Gross vs. Disaggregated Flows __________________________32
- 3.  Estimated Betas _________________________________________________________33
- 4.  Estimated Common Factor in Total Gross Inflows vs. VIX _______________________35
- 5.  The Model vs. the GFC and vs. the Taper Tantrum  _____________________________36
- 6.  Institutional Quality vs. Correlations _________________________________________38

*Source: _wp15127 - 1. Sample of Countries ______________________________________________________22 (PDF).*

### References ________________________________________________________________29

### _wp15127 - References ________________________________________________________________29

### I. Introduction — scope, question, and contribution
- Focus: systematic analysis of the sensitivity of 34 EMs to global push factors using quarterly balance-of-payments (BOP) data for the period 2001-2013.
- Key question: Why do some EMs respond more strongly than others to changes in global (push) factors?
- Main contributions:
  - Use of a latent factor model (in the spirit of Kose et al. (2003)) to extract unobserved common dynamics in gross inflows (total and by component).
  - Disaggregation of gross inflows into FDI, Portfolio Equity, Portfolio Bonds, Other Investment (OI) to Banks, and OI to Non-Banks.
  - Identification that commonality in inflows is sizable but heterogeneous across asset types and countries.
  - Finding that financial market characteristics and composition of foreign investor base (rather than macroeconomic or institutional fundamentals) most robustly explain cross-country heterogeneity in sensitivities to push factors.

### II. Key empirical setup and definitions
- Data and sample:
  - Gross capital inflows from IMF’s Balance of Payment database.
  - Components: FDI, Portfolio Equity, Portfolio Bonds, OI to Banks, OI to Non-Banks.
  - Sample period: 2001Q1-2013Q4.
  - Sample size: 34 emerging markets.
  - All series measured in U.S. dollars and normalized by recipient country GDP (quarterly).
- Terminology:
  - Capital “inflows” = changes in domestic financial liabilities vis-à-vis non-residents (residence criterion).
  - “Outflows” refer to residents’ changes in net investment position abroad (not analyzed here).
- Push factors used (as per literature):
  - (i) average GDP growth rate in four core economies (U.S., Euro Area, Japan, and U.K.)
  - (ii) US VIX
  - (iii) changes in the expected U.S. policy rate
  - (iv) slope of the U.S. yield curve (10 year minus 3 month U.S. government T-bill yields)
  - (v) U.S. real effective exchange rate (REER)
  - Type-specific push variables:
    - Banking inflows: U.S. dealer bank leverage and TED spread
    - Bond inflows: 10-year U.S. government bond yield; (lagged) return of the EMBI
    - Equity inflows: (lagged) return in the MSCI emerging market index
  - Pull controls: index of aggregate commodity prices; (lagged) aggregate real GDP growth in EMs
- Fundamentals and market characteristics considered:
  - Macroeconomic fundamentals: trade openness (% of GDP), public debt (% of GDP), foreign exchange reserves (% of GDP), foreign exchange regime (fixed vs. degree of floating), average real GDP growth rate
  - Institutional quality: ICRG Rule of Law and Investor Protection indexes
  - Financial market dimensions (where possible): (i) degree of foreign openness; (ii) size of the market; (iii) liquidity of the market; (iv) composition of foreign investor base
    - New proxy approach: for each EM and asset, compute correlation between BOP inflows and investor-reported inflows (EPFR Global for funds; BIS locational international banking statistics for global banks). High correlation interpreted as sign that those investors account for most movements in BOP-recorded inflows.

### III. Econometric framework and estimation details
- Model: dynamic latent factor model (Kose, Otrok and Whiteman (2003) framework) specified as
  - y_{i,t} = β_{i,W} f_t^W + β_{i,R} f_{j,t}^R + υ_{i,t}, with latent world (EM) factor f_t^W and regional factor f_{j,t}^R, country-specific loadings β, and idiosyncratic residuals υ_{i,t}.
- Dynamics:
  - Idiosyncratic factors follow AR(p) process; factors follow AR(q) processes.
  - In implementation, set the length of both the idiosyncratic and factor auto-regressive polynomials to 2.
- Identification and priors:
  - Normalize sign: loading on world factor for the first country > 0; loading on regional factor for one country in each region > 0.
  - Normalize scales: each factor variance assumed equal to 1.
  - Conjugate priors used (as in text), including IG(6,0.001) on innovation variances.
- Estimation:
  - Bayesian MCMC with data augmentation.
  - Posterior based on 300,000 MCMC replications after 30,000 burn-in replications.

### IV. Results — factor estimates, heterogeneity across assets and countries
- Aggregate findings:
  - Gross inflows to EMs co-move substantially across countries due to global (push) factors, but the magnitude varies substantially across countries and across asset types.
- Asset heterogeneity:
  - Portfolio Equity flows, Portfolio Bond flows, and OI to Banks co-move substantially across EMs.
  - FDI and OI to Non-Banks do not exhibit identifiable co-movement structure across EMs (model could not identify commonality for these types).
- Cross-country heterogeneity:
  - Factor loadings β_{i} and variance-decomposition θ_{i} (share of variance attributable to common EM dynamic) display strong cross-country dispersion.
  - Examples highlighted:
    - Equity flows: a unit standard deviation in the common EM factor causes a unit standard deviation in equity flows to Pakistan and 0.6 standard deviation in equity flows to India; Latvia, Estonia, Belarus show insignificant responses.
    - Bond flows: Indonesia and South Africa bond flows react strongly; China, Colombia, Bulgaria are almost insensitive.
  - Country groups:
    - “High sensitivity” (sensitive across components): Brazil, Indonesia, Thailand, Turkey, South-Africa.
    - “Asymmetric” (sensitive in one or two components): Pakistan, Philippines, India, Mexico.
    - “Insensitive” (low sensitivity across components): Estonia, Latvia, Chile.
  - Notable statistic: for Pakistan and Philippines, more than half of the variance in equity funding is accounted for by the common EM factor.
- Drivers of heterogeneity:
  - Financial market characteristics and investor-base composition robustly explain sensitivities:
    - Market liquidity and depth increase sensitivity for equity flows (liquid markets more sensitive).
    - Composition of foreign investor base: greater reliance on international funds (mutual funds, ETFs) and global banks correlates with higher sensitivity across bond, equity, and bank flows.
  - Macroeconomic and institutional fundamentals:
    - Generally do not provide insulation from push-factor shocks (consistent with Aizenman et al. (2014) and Eichengreen and Gupta (2014)).
    - Exception: portfolio bond inflows — lower reserves, higher trade openness, and more flexible FX regimes are associated with higher sensitivity to global push factors.
  - Observed proxies vs. latent factors:
    - Typical observed proxies (e.g., VIX) capture only a small fraction of the actual co-movement identified by latent factors; estimated latent factors explain a much greater part of co-movement.

### V. Interpretations and implications
- Financial market structure matters more than “good” macro/ institutional fundamentals for sensitivity to push factors:
  - EMs with deeper, more liquid financial markets and higher exposure to “fickle investors” (international funds, global banks) are more exposed to inflow reversals when global financial conditions deteriorate, and conversely benefit more when conditions improve.
- Relation to existing literature:
  - Confirms broad literature that push factors (U.S. monetary policy, global liquidity, risk aversion) help explain synchronicity of flows, but emphasizes heterogeneity by asset and recipient-country characteristics.
  - Latent-factor approach circumvents the need to select observed global proxies and shows those proxies may miss substantial co-movement.
- Methodological note:
  - Distinguishing gross inflows by asset is crucial: net flows and price dynamics may lead to different conclusions; studying gross flows yields different policy-relevant insights.

### VI. Paper structure and next steps (as stated)
- Section 2: data and methodology (described).
- Section 3: results and relation to literature.
- Section 4: robustness checks.
- Final section: conclusions and outstanding issues for policy and research.

*Source: _wp15127 - References ________________________________________________________________29*

### Section 2. Given the 2001Q2-2013Q4 coverage of our sample, the total number of

### _wp15127 - Section 2. Given the 2001Q2-2013Q4 coverage of our sample, the total number of

### Sample and model setup
- The sample coverage is 2001Q2-2013Q4 and the total number of observations is 50.
- Expected signs of coefficients:
  - Push variables: negative (slower growth in advanced economies, higher VIX, higher U.S. REER, flatter yield curve, higher expected policy rates).
  - Pull variables: positive for commodity prices and EM growth.
  - Asset/type-specific factors: positive for return-chasing variables and bank leverage.
- Model specification notes:
  - Change in expected U.S. policy rates is used instead of the level of a short-term rate.
  - Factor model: EM_t = alpha + beta*Push_t + gamma*Pull_t + specific factors + epsilon_t (notation preserved from source).

### Regression results and explanatory power
- General result: almost all coefficients, when statistically significant, have the expected signs.
- Push variables:
  - VIX and REER are the most robust determinants of commonality in aggregate inflows and across various types of capital inflows (see columns 1, 5, 9, and 13 referenced in the source).
  - The explanatory power of the VIX is largely driven by the global financial crisis (GFC).
  - Slope of the yield curve, GDP growth rate of core countries, and expected policy rates are significant only in some cases.
- Pull variables:
  - Commodity price is the only significant pull variable across various types of capital inflows when pull factors are used alone.
  - EM growth is significant only for OI-to-bank and total inflows (see columns 2, 6, 10, and 14).
- Combined push and pull regressions yield similar results (see columns 3, 7, 11, and 15).
- Relative contribution to R2 (group-level R2s):
  - Push factors account for about 65 percent of the overall R2s for OI banks.
  - Push factors account for about 60 percent of the overall R2s for Portfolio bonds.
  - Push factors account for about 75 percent of the overall R2s for Portfolio equity flows.
- Conclusion: push factors dominate pull factors in explaining common dynamics, though asset-specific variables are also statistically significant with expected signs.

### What drives cross-country heterogeneity in sensitivity to push factors
- Approach:
  - Regress estimated factor loadings (beta_i^asset) for each asset on two sets of variables: fundamental-related and market-structure based variables (see Annexes I and II in the source).
  - Cross-sectional regression estimated separately for each type of flow; sample size is small (34 cross-country observations per asset).
  - Strategy: regress betas on fundamentals and market variables separately, then combine significant variables from each category; confirm stability with Bayesian Model Averaging (BMA).
- Key benchmark findings (Table 6 referenced):
  - Equity and bank flows:
    - Higher betas do not coincide with weaker fundamentals (no robust link to lower growth, higher debt, or poor institutions) (columns 1-3 and 7-9).
  - Bond flows:
    - Countries with higher reserves, higher trade openness, and more flexible foreign exchange regimes are more sensitive to global push factors (columns 4-6).
  - Role of global investors:
    - Proxies for importance of global investors are highly significant and suggest strong quantitative impact:
      - Going from a zero to a perfect correlation increases predicted response to a shock in the common EM factor by 0.45 for equity.
      - The same increase is 0.24 for bonds.
      - The same increase is 0.75 for banks.
    - These are described as large effects given general levels of betas (and as depicted in Figure 3).
  - Equity market features:
    - More liquid equity markets (measured by turnover ratio) and inclusion in the MSCI Frontier index are associated with higher betas.
  - Overall:
    - Most cross-sectional variation in loadings can be explained by market-related variables.
    - Macroeconomic fundamentals (level of reserves, trade openness, type of exchange rate regime) matter mainly for bond flows.
    - No robust evidence that institutional fundamentals (Investment Climate, Rule of Law) or standard macro indicators (public debt, growth) explain differences in sensitivities.

### Robustness checks
- Betas reflect sensitivity beyond the GFC:
  - Countries with higher betas generally experienced deeper retrenchments in flows during both the GFC and the Taper Tantrum (illustrated in Figure 5).
- Model uncertainty addressed with Bayesian Model Averaging (BMA):
  - Significant variables in base regressions remain the most robust in the BMA exercise (Table 7).
- Institutional fundamentals do not drive the importance of global investors:
  - No relation found between the correlation variables (measuring global investor importance) and institutional quality measures (Law and Order, Investor Protection Index from ICRG) (Figure 6).

### Conclusions and policy implications
- Main conclusions:
  - Cross-country differences in EM sensitivities to global push factors are largely a function of market characteristics, especially the nature of the foreign investor base:
    - International mutual funds matter for equity and bond flows.
    - Global banks matter for bank inflows.
  - Macroeconomic fundamentals, notably FX regime type, matter for bond flows.
  - Institutional fundamentals and standard macro performance measures (higher growth, lower debt) do not explain differences in sensitivities to push factors.
- Policy implications and recommendations:
  - Good fundamentals do not guarantee insulation from global financial shocks; sensitivity to push factors depends importantly on the investor base and market structure.
  - EM authorities should collect information on their foreign investor base and the role of large funds or asset managers.
    - Despite limited systematic information on foreign holdings decomposition, measures can be created and used to assess sensitivities.
  - Further research is needed to understand the ultimate macroeconomic impact of push factors at the country level, including:
    - The role of domestic investors absorbing assets from foreigners during stress episodes (example: Malaysia offsets negative inflows via local institutional investors repatriating foreign assets).
    - Discrepancies between how flows and asset prices react to global push factors and the interaction with local institutional setups.

*Source: _wp15127 - Section 2.*

### References

### References

### Reference List
- Alfaro, Laura, Sebnem Kalemli-Ozcan, and Vadim Volosovych, 2008, “Why Doesn't Capital Flow from Rich to Poor Countries? An Empirical Investigation,” Review of Economics and Statistics, Vol. 90, No. 2, pp. 347-68.
- Adrian, Tobias, and Hyun Song Shin, 2010, “Liquidity and Leverage,” Journal of Financial Intermediation, Vol. 19, No. 3, pp. 418-37.
- Adrian, Tobias, and Hyun Song Shin, 2014, “Procyclical Leverage and Value-at-Risk,” Review of Financial Studies, Vol. 27 No. 2, pp. 373-403.
- Aizenman, Joshua, Mahir Binici, and Michael M. Hutchison, 2014, "The Transmission of Federal Reserve Tapering News to Emerging Financial Markets," NBER Working Paper No. 19980.
- Ahmed Shaghil, Coulibaly Brahima, Zlate Andrei, 2014, “International Financial Spillovers to Emerging Market Economies: How Important Are Economic Fundamentals?” Board of Governors of the Federal Reserve System Working Paper D2015-37.
- Arslanalp, Serkan, and Tsuda, Takahiro, 2014, “Tracking Global Demand for Advanced Economy Sovereign Debt. IMF Economic Review, Vol. 62, Issue 3, pp. 430-64.
- Broner, Fernando, Tatiana Didier, Aitor Erce, and Sergio L. Schmukler, 2013, “Gross Capital Flows: Dynamics and Crises,” Journal of Monetary Economics Vol. 60, No. 1, pp. 113-33.
- Bruno, V., and Hyun Song Shin, 2015a, “Cross-Border Banking and Global Liquidity,” Review of Economic Studies, forthcoming.
- Bruno, V., and Hyun Song Shin, 2015b, “Capital Flows and the Risk-Taking Channel of Monetary Policy,” Journal of Monetary Economics, forthcoming.
- Calvo, Guillermo, Leonardo Leiderman, and Carmen Reinhart, 1993, “Capital Inflows and Real Exchange Rate Appreciation in Latin America: The Role of External Factors,” IMF Staff Papers, Vol. 40, No. 1, pp. 108-51.
- Calvo, Guillermo, Leonardo Leiderman, and Carmen Reinhart, 1996, “Inflows of Capital to Developing Countries in the 1990s,” Journal of Economic Perspectives, Vol. 10, No. 2, pp.123-39.
- Cerutti, Eugenio, and Stijn Claessens, 2014, “The Great Cross-Border Bank Deleveraging: Supply Side Characteristics and Intra-Group Frictions,” IMF Working Paper 2014/180.
- Cerutti, Eugenio, Stijn Claessens, and Patrick McGuire, 2012, "Systemic Risks in Global Banking: What Available Data Can Tell Us and What More Data are Needed?" NBER Working Papers 18531, NBER.
- Cerutti, Eugenio, Stijn Claessens, and Lev Ratnovski, 2014, “Global Liquidity and Drivers of Cross-Border Bank Flows,” IMF Working Paper, 2014/69.
- Cetorelli, Nicola, and Linda Goldberg, 2012a, “Liquidity Management of U.S. Global Banks: Internal Capital Markets in the Great Recession,” Journal of International Economics, Vol. 88, pp. 299-311.
- Cetorelli, Nicola, and Linda Goldberg, 2012b, “Banking Globalization and Monetary Transmission,” Journal of Finance, Vol. 67, No. 5, pp. 1811-843.
- Chuhan, Punam, Stijn Claessens, and Nlandu Mamingi, 1998, “Equity and Bond Flows to Asia and Latin America: The Role of Global and Country Factors,” Journal of Development Economics, Vol. 55, pp. 439-63.
- Claessens, Stijn, and Swati Ghosh, 2013, “Capital Flow Volatility and Systemic Risk in Emerging Markets: The Policy Toolkit,” in O. Canuto and S. Ghosh (eds.), Dealing with Challenges of Macro Financial Linkages in Emerging Markets, World Bank, Washington, D.C.
- Claessens, Stijn, and Neeltje Van Horen, 2014, “Foreign Banks: Trends and Impact,” Journal of Money, Credit, and Banking, Vol. 46, No. 1, pp. 295-326.
- Eichengreen, Barry, and Poonam Gupta, 2014, "Tapering Talk: The Impact of Expectations of Reduced Federal Reserve Security Purchases on Emerging Markets," Policy Research Working Paper Series 6754, The World Bank.
- Fernandez, Carmen, Eduardo Ley, and Mark Steel, 2001, “Benchmark Priors for Bayesian Model Averaging,” Journal of Econometrics, Vol. 100, No. 2, pp. 381-427.
- Forbes, Kristin J., and Francis E. Warnock, 2012, “Debt- and Equity-Led Capital Flow Episodes,” NBER Working paper, No. 18329.
- Forbes, Kristin J., 2007. "The Microeconomic Evidence on Capital Controls: No Free Lunch," in Capital Controls and Capital Flows in Emerging Economies: Policies, Practices and Consequences, pp. 171-202, NBER.
- Fratzscher, Marcel, 2011, “Capital flows, Push versus pull factors and the global financial crisis,” ECB Working Paper No 1364.
- Fratzscher, Marcel, Marco Lo Duca, and Roland Straub, 2013, “On the International Spillovers of U.S. Quantitative Easing,” Discussion Papers of DIW Berlin 1304, DIW Berlin, German Institute for Economic Research.
- Ghosh, Atish R., Mahvash Qureshi, Jun Il Kim, and Juan Zalduendo, 2014, "Surges," Journal of International Economics, Vol. 92, No. 2, pp. 266-85.
- IMF, 2013a, “Global Impact and Challenges of Unconventional Monetary Policies, Background paper,” Washington, D.C.: IMF.
- IMF, 2013b, "The Yin and Yang of Capital Flow Management: Balancing Capital Inflows with Capital Outflows," by Jaromir Benes, Jaime Guajardo, Damiano Sandri, and John Simon, World Economic Outlook, Chapter 4, Fall.
- Jotikasthira, Chotibhak, Christian Lundblad, and Tarun Ramadorai, 2012, “Asset Fire Sales and Purchases and the International Transmission of Funding Shocks,” Journal of Finance, Vol. 67, No. 6, pp. 2015–050.
- Koepke, Robin, 2014, “Fed Policy Expectations and Portfolio Flows to Emerging Markets,” IIF Working Paper, May, Washington, D.C.
- Koepke, Robin, 2015, “What Drives Capital Flows to Emerging Markets: A Survey of the Empirical Literature,” IIF Working Paper, April, Washington D.C.
- Koepke, Robin, and Saacha Mohammed, 2014, “Portfolio Flows Tracker FAQ,” IIF Research Note.
- Kose, Ayhan, Christopher Otrok, and Charles H. Whiteman, 2003, “International Business Cycles: World, Region, and Country-Specific Factors,” American Economic Review, Vol. 93, No. 4, pp. 1216-239.
- Magnus, Jan, Powell,O and Prufer, P, 2010, “A Comparison of Two Model Averaging Techniques with an Application to Growth Empirics.” Journal of Econometrics, Vol. 154, No. 2, pp. 139-153.
- Minoiu, Camelia, and Javier Reyes, 2013, “A Network Analysis of Global Banking: 1978-2010,” Journal of Financial Stability, Vol. 9, No. 2, pp. 168-84.
- Milesi-Ferretti, Gian Maria, and Cédric Tille, 2011, "The Great Retrenchment: International Capital Flows during the Global Financial Crisis," Economic Policy, Vol. 26. No. 66, pp. 285-342.
- Prachi, Mishra, Kenji Moriyama, Papa N'Diaye, and Lam Nguyen, 2014, "Impact of Fed Tapering Announcements on Emerging Markets," International Monetary Fund and the Reserve Bank of India. Mimeo.
- Puy, Damien, 2013, "Institutional Investors Flows and the Geography of Contagion," Economics Working Papers ECO2013/06, European University Institute.
- Raddatz, Claudio, and Sergio Schmukler, L., 2012, "On the International Transmission of Shocks: Micro-evidence from Mutual Fund Portfolios," Journal of International Economics, Vol. 88, No. 2, pp. 357-74.
- Rey, Hélène, 2013, “Dilemma not Trilemma: The Global Financial Cycle and Monetary Policy Independence,” in Proceedings of the Federal Reserve Bank of Kansas City Economic Symposium at Jackson Hole.
- Sahay, Ratna, Vivek Arora, Thanos Arvanitis, Hamid Faruqee, Papa N'Diaye, Tommaso Mancini-Griffoli, and an IMF Team, 2014. “Emerging Market Volatility: Lessons from the Taper Tantrum,” IMF Staff Discussion Note SDN 14/09, September.
- Shin, Hyun Song, 2012, “Global Banking Glut and Loan Risk Premium” Mundell‐Fleming Lecture, IMF Economic Review Vol. 60, No. 2, pp. 155‐92.

### Tables and Data (extracted from source)

- Table 1. Sample of Countries — country lists by region:
  - Latin America: Argentina, Brazil, Chile, Colombia, Mexico, Peru, Uruguay, Venezuela, Rep. Bol.
  - Asia: India, China, Mainland, Indonesia, Republic of Korea, Malaysia, Pakistan, Philippines, Thailand
  - Emerging Europe: Belarus, Kazakhstan, Bulgaria, Russian Federation, Ukraine, Czech Republic, Slovak Republic, Estonia, Latvia, Hungary, Lithuania, Croatia, Slovenia, Poland, Romania
  - Other: Turkey, South Africa, Israel

- Table 2. Variable Definitions, Frequency and Sources (selected entries preserved exactly)
  - Capital inflows: gross inflow as % GDP, total and by component; Frequency: Quarterly; Source: IMF Balance of Payment Statistics.
  - Global Bank flows: inflow as % GDP; Frequency: Quarterly; Source: Bank of International Settlements - Locational Statistics.
  - Mutual Fund Flows: inflow as % GDP; Frequency: Quarterly; Source: EPFR.
  - Real GDP Growth: in %, QoQ, un-weighted average of US, Euro Area, Japan, and UK; Frequency: Quarterly; Source: IMF WEO.
  - US VIX: CBOE S&P500 Volatility VIX; Frequency: Quarterly; Source: Datastream.
  - Expected Change in Policy Rate: Difference between Policy Rate and 30 Day Federal Funds 6 Month Futures; Frequency: Quarterly, average of monthly figures; Source: Datastream and Cleveland Fed.
  - US Yield Curve: 10 year/3 month US Treasury yield spread; Frequency: Quarterly; Source: Datastream.
  - US REER: US Real Effective Exchange Rate; Frequency: Quarterly; Source: IMF WEO.
  - Commodity Prices: growth rate, QoQ; Frequency: Quarterly; Source: IMF WEO.
  - US Dealer Bank Leverage: (Equity+Total Liabilities)/Equity; Frequency: Quarterly; Source: US Fund Flows.
  - US TED Spread: 3-month TED spread (LIBOR - Treasury bill); Frequency: Quarterly; Source: Datastream.
  - 10Y Bond Yield: 10 year US Tresuary yield; Frequency: Quarterly; Source: Datastream.
  - MSCI returns: Return in the MSCI EM index; Frequency: Quarterly; Source: Datastream.
  - EMBI returns: Return in EMBI index; Frequency: Quarterly; Source: Datastream.
  - Trade Openness (X+M)/GDP: Average over 2001-2013; Source: World Development Indicators.
  - FX regime*: Index from 1 to 13; Average over 2001-2013; Source: Ilzetzki, Reinhart and Rogoff (2004).
  - Public Debt as % GDP: Average over 2001-2013; Source: World Development Indicators.
  - Reserves as % GDP: Average over 2001-2013; Source: World Development Indicators.
  - Real GDP Growth %, annual: Average over 2001-2013; Source: World Development Indicators.
  - Rule of Law* Index from 1 to 10: Average over 2001-2013; Source: ICRG.
  - Investor Protection* Index from 1 to 10: Average over 2001-2013; Source: ICRG.
  - Foreign Openness (Market Characteristics): Stock of foreign Equity, Bond or Bank claims/GDP; Average over 2001-2013; Source: IIP.
  - Stock Market Capitalization: Stock Market Cap/GDP; Average over 2001-2013; Source: World Bank Financial Development Database.
  - Bond Market Capitalization**: Bond Market Cap/GDP; Average over 2001-2013; Source: World Bank Financial Development Database.
  - Private Credit: Bank Credit to the Private Sector/GDP; Average over 2001-2013; Source: World Bank Financial Development Database.
  - Stock Market Turnover: Sum of all shares traded over the period / Stock market cap.; Average over 2001-2013; Source: World Bank Financial Development Database.
  - Share of Funding coming from Advanced Economies: Sum of Bond (Equity) coming from AEs and Financial centers***/Total Bond (Equity) Funding; Average over 2001-2013; Source: CPIS.
  - MSCI EM: Country listed in the MSCI Emerging index over the sample period; Dummy; Source: Morgan Stanley.
  - MSCI FM: Country listed in the MSCI Frontier Market index over the sample period; Dummy; Source: Morgan Stanley.
  - EMBI EM: Country listed in the EMBI Emerging index over the sample period; Dummy; Source: JP Morgan.
  - Notes:
    - * In the case of ICRG ratings, a higher value of the index indicates better institutions. For the FX regime, a higher value implies a more flexible exchange rate.
    - ** Bond Market Capitalization data are not available for all countries in our sample. When used on the restricted sample however, the bond market capitalization is not found significant.
    - *** See Annex I for the list of source countries.

- Table 3. Raw Statistics (selected rows and exact numeric values preserved)
  - Fundamentals (Obs = 34 unless stated):
    - Trade Openness: Obs 34; Mean 80.52; Std. Dev. 41.22; Min 23.99; Max 187.30
    - Public debt: Obs 34; Mean 40.69; Std. Dev. 19.08; Min 5.95; Max 79.21
    - Reserves: Obs 34; Mean 19.58; Std. Dev. 9.37; Min 6.96; Max 44.33
    - Exchange Rate Regime: Obs 34; Mean 8.29; Std. Dev. 3.04; Min 2.00; Max 13.00
    - Average Growth: Obs 34; Mean 4.76; Std. Dev. 1.73; Min 0.78; Max 9.94
    - Investor Protection: Obs 34; Mean 8.83; Std. Dev. 1.73; Min 3.69; Max 11.23
    - Rule of Law: Obs 34; Mean 3.59; Std. Dev. 1.01; Min 1.54; Max 5.00
  - Equity Market Characteristics:
    - Foreign Openness -Equity: Obs 34; Mean 6.88; Std. Dev. 7.01; Min 0.07; Max 24.59
    - Relative Market Size -Equity: Obs 34; Mean 2.94; Std. Dev. 4.66; Min 0.00; Max 17.36
    - Stock Market Capitalization: Obs 33; Mean 43.19; Std. Dev. 40.52; Min 0.49; Max 190.54
    - MSCI EM Country: Obs 34; Mean 0.59; Std. Dev. 0.50; Min 0.00; Max 1.00
    - MSCI FM Country: Obs 34; Mean 0.29; Std. Dev. 0.46; Min 0.00; Max 1.00
    - Stock Market turnover: Obs 33; Mean 49.59; Std. Dev. 59.50; Min 0.96; Max 226.99
    - Share of Equity Funding from Advanced Economies: Obs 34; Mean 67.19; Std. Dev. 22.72; Min 12.48; Max 95.67
    - BOP Equity correlation with EPFR flows: Obs 34; Mean 0.24; Std. Dev. 0.26; Min -0.41; Max 0.73
  - Bond Market Characteristics:
    - Foreign Openness -Bond: Obs 34; Mean 10.04; Std. Dev. 6.48; Min 0.12; Max 30.47
    - Relative Market Size - Bond: Obs 34; Mean 2.94; Std. Dev. 4.22; Min 0.00; Max 19.27
    - EMBI Country: Obs 34; Mean 0.65; Std. Dev. 0.49; Min 0.00; Max 1.00
    - Share of Bond funding from Advanced Economies: Obs 34; Mean 67.27; Std. Dev. 16.74; Min 23.25; Max 92.09
    - BOP Bond correlation with EPFR flows: Obs 34; Mean 0.26; Std. Dev. 0.26; Min -0.55; Max 0.67
  - Banking Market Characteristics:
    - Foreign Openness - Other Investment: Obs 34; Mean 36.29; Std. Dev. 19.08; Min 4.32; Max 96.40
    - Private Credit/GDP: Obs 32; Mean 63.39; Std. Dev. 29.11; Min 24.57; Max 138.60
    - BOP OI-Bank correlation with BIS flows: Obs 34; Mean 0.13; Std. Dev. 0.22; Min -0.52; Max 0.59

- Table 4. Variance Decompositions Results (selected country-level shares, percentages preserved exactly)
  - Notes: For each country the table reports (mean) share of variance accounted for by common and regional factors for Portfolio Equity, Portfolio Bond, OI Bank, and All inflows.
  - Latin America (selected entries):
    - Argentina: Portfolio Equity Global 25% Regional 7%; Portfolio Bond Global 12% Regional 11%; OI Bank Global 29% Regional 5%; All inflows Global 12% Regional 20%
    - Brazil: Portfolio Equity Global 24% Regional 4%; Portfolio Bond Global 25% Regional 6%; OI Bank Global 25% Regional 11%; All inflows Global 27% Regional 5%
    - Chile: Portfolio Equity Global 2% Regional 10%; Portfolio Bond Global 2% Regional 20%; OI Bank Global 4% Regional 4%; All inflows Global 11% Regional 13%
    - Mexico: Portfolio Equity Global 7% Regional 8%; Portfolio Bond Global 29% Regional 11%; OI Bank Global 8% Regional 32%; All inflows Global 20% Regional 15%
    - Peru: Portfolio Equity Global 3% Regional 5%; Portfolio Bond Global 8% Regional 2%; OI Bank Global 26% Regional 3%; All inflows Global 34% Regional 19%
    - Latin America averages reported: Portfolio Equity Global 8% Regional 8%; Portfolio Bond Global 13% Regional 11%; OI Bank Global 13% Regional 10%; All inflows Global 14% Regional 15%
  - Asia (selected entries and averages):
    - India: Portfolio Equity Global 35% Regional 16%; Portfolio Bond Global 4% Regional 34%; OI Bank Global 7% Regional 2%; All inflows Global 55% Regional 4%
    - China, P.R.: Mainland: Portfolio Equity Global 18% Regional 4%; Portfolio Bond Global 2% Regional 3%; OI Bank Global 35% Regional 7%; All inflows Global 28% Regional 1%
    - Indonesia: Portfolio Equity Global 18% Regional 8%; Portfolio Bond Global 43% Regional 9%; OI Bank Global 23% Regional 23%; All inflows Global 21% Regional 8%
    - Asia averages reported: Portfolio Equity Global 29% Regional 10%; Portfolio Bond Global 15% Regional 11%; OI Bank Global 16% Regional 15%; All inflows Global 34% Regional 12%
  - Emerging Europe (selected entries and averages):
    - Kazakhstan: Portfolio Equity Global 30% Regional 2%; Portfolio Bond Global 20% Regional 11%; OI Bank Global 1% Regional 42%; All inflows Global 4% Regional 34%
    - Bulgaria: Portfolio Equity Global 16% Regional 62%; Portfolio Bond Global 1% Regional 2%; OI Bank Global 5% Regional 38%; All inflows Global 6% Regional 60%
    - Russian Federation: Portfolio Equity Global 8% Regional 2%; Portfolio Bond Global 15% Regional 3%; OI Bank Global 23% Regional 37%; All inflows Global 36% Regional 17%
    - Romania: Portfolio Equity Global 30% Regional 1%; Portfolio Bond Global 11% Regional 10%; OI Bank Global 2% Regional 60%; All inflows Global 8% Regional 69%
    - Emerging Europe averages reported: Portfolio Equity Global 9% Regional 13%; Portfolio Bond Global 11% Regional 9%; OI Bank Global 7% Regional 38%; All inflows Global 10% Regional 43%
  - Other (selected entries and averages):
    - Turkey: Portfolio Equity Global 22% Regional 20%; Portfolio Bond Global 26% Regional 3%; OI Bank Global 37% Regional 6%; All inflows Global 36% Regional 5%
    - South Africa: Portfolio Equity Global 17% Regional 14%; Portfolio Bond Global 30% Regional 22%; OI Bank Global 22% Regional 10%; All inflows Global 41% Regional 18%
    - Israel: Portfolio Equity Global 3% Regional 33%; Portfolio Bond Global 13% Regional 39%; OI Bank Global 2% Regional 53%; All inflows Global 17% Regional 30%
    - Other averages reported: Portfolio Equity Global 14% Regional 22%; Portfolio Bond Global 23% Regional 10%; OI Bank Global 20% Regional 23%; All inflows Global 24% Regional 29%

- Table 5. Finding the Drivers of the Estimated EM Common Factors (selected regression coefficients and statistics preserved exactly)
  - Regression period: 2001Q2-2013Q4. Observations for reported regressions: 50 (for each column).
  - Selected coefficient estimates (exact values and standard errors in parentheses where provided):
    - Core_GDP_Growth:
      - Column [1]: 0.132*** (0.0387)
      - Column [2]: 0.0736* (0.0412)
      - Column [4]: -0.00808 (0.0543)
      - Column [13]: 0.0789* (0.0418)
    - US VIX:
      - Column [1]: -0.0313*** (0.0105)
      - Column [5]: -0.0516*** (0.00997)
      - Column [9]: -0.0356*** (0.0106)
      - Column [13]: -0.0397*** (0.0108)
    - Exp. Change in Policy Rate:
      - Column [11]: -0.401* (0.235)
      - Column [12]: -0.372* (0.214)
      - Column [15]: -0.415* (0.206)
    - US yield_curve:
      - Column [7]: -0.258** (0.110)
      - Column [8]: -0.392** (0.147)
      - Column [11]: -0.268** (0.0849)
    - US REER:
      - Column [1]: -0.0592*** (0.00995)
      - Column [2]: -0.0567*** (0.0115)
      - Column [3]: -0.0855*** (0.0188)
      - Column [4]: -0.0386*** (0.00755)
      - Column [11]: -0.0358** (0.00776)
      - Column [12]: -0.0334*** (0.00940)
      - Column [15]: -0.0617*** (0.0182)
    - Commodityprice_pch:
      - Column [1]: 0.0561*** (0.0119)
      - Column [2]: 0.0314** (0.0119)
      - Column [3]: 0.0282** (0.0124)
      - Column [4]: 0.0451*** (0.0121)
      - Column [11]: 0.0552** (0.0131)
      - Column [12]: 0.0382*** (0.00954)
      - Column [15]: 0.0279** (0.0119)
    - L.RGDP_EM_growth:
      - Column [1]: 0.0988** (0.0439)
      - Column [9]: -0.120** (0.0530)
      - Column [11]: -0.0896* (0.0508)
      - Column [15]: 0.0896* (0.0448)
    - Global_bank_leverage:
      - Column [7]: 0.0682** (0.0335)
      - Column [8]: 0.0569 (0.0516)
    - TED:
      - Column [7]: -0.527** (0.261)
      - Column [8]: -0.778** (0.292)
    - US 10 bond yield:
      - Column [5]: -0.240* (0.135)
    - L.EMBI_growth:
      - Column [9]: 0.0330* (0.0165)
    - L.MSCI_growth:
      - Column [11]: 0.0153** (0.00592)
  - Observations: 50 across reported columns.
  - R-squared (overall) reported for columns:
    - Column [1]: 0.660
    - Column [2]: 0.415
    - Column [3]: 0.710
    - Column [4]: 0.731
    - Column [5]: 0.510
    - Column [6]: 0.238
    - Column [7]: 0.569
    - Column [8]: 0.580
    - Column [9]: 0.396
    - Column [10]: 0.091
    - Column [11]: 0.421
    - Column [12]: 0.436
    - Column [13]: 0.643
    - Column [14]: 0.468
    - Column [15]: 0.733
    - Column [16]: 0.789
  - R-squared (push variables), (pull variables), and (type variables) reported across columns (values preserved exactly in table).

- Table 6. Explaining Countries’ Sensitivities to Push Factors (selected coefficients and R-squared values preserved exactly)
  - Method: regressions of EM sensitivities by flow type on macro, institutional, and market characteristics. Columns (1)-(9) organized by Equity Beta, Bond Beta, Bank Beta with different regressor sets.
  - Selected coefficient estimates (exact values):
    - Trade Openness:
      - Column (2): 0.00179**
      - Column (3): 0.00110**
    - Debt/GDP:
      - Column (4): 0.00163
      - Column (5): 0.00193
      - Column (7): -0.00104
    - Reserves/GDP:
      - Column (2): 0.00190
      - Column (6): -0.00450
      - Column (7): -0.00559***
      - Column (8): 0.00567
    - FX Regime:
      - Column (2): 0.0192
      - Column (3): 0.0315***
      - Column (4): 0.0203**
      - Column (5): 0.0342**
      - Column (6): 0.00766
    - MSCI Frontier Benchmark (dummy):
      - Column (2): 0.228***
      - Column (3): 0.242***
    - Turnover Ratio:
      - Column (2): 0.00176***
      - Column (3): 0.00170***
    - Correlation with EPFR (or BIS) flows:
      - Equity Beta columns: 0.474*** (one specification), 0.454***, 0.260**
      - Bond Beta columns: 0.241**, 0.705***, 0.75***
    - R-squared values:
      - Equity Beta group: R-sq 0.244 (col 1), 0.546 (col 2), 0.521 (col 3)
      - Bond Beta group: R-sq 0.409 (col 4), 0.288 (col 5), 0.429 (col 6)
      - Bank Beta group: R-sq 0.324 (col 7), 0.520 (col 8), 0.530 (col 9)

- Table 7. Bayesian Averaging Results (selected entries preserved exactly)
  - Purpose: Bayesian Model Averaging to test robustness of variables highlighted in Section 3; reports coefficient, t-stat, and Post-Inclusion Probabilities (PIPs) for Equity, Bond, and Bank regressions.
  - Selected reported values (as they appear):
    - Equity- Bayesian Averaging: Trade/GDP Coef. 0.000; t-Stat -0.17; PIP 0.09
    - Bond- Bayesian Averaging: Trade/GDP Coef. 0.000; PIP 0.15
    - Bank- Bayesian Averaging: Trade/GDP Coef. -0.041; t-Stat -0.79; PIP 0.47
    - Debt/GDP: Equity Coef. 0.000; t-Stat 0.04; PIP 0.07
    - Debt/GDP (Bond): Coef. 0.000; PIP 0.24; t-Stat 0.12
    - Debt/GDP (Bank): Coef. -0.024; t-Stat -0.4; PIP 0.21
    - Reserves/GDP: Equity Coef. 0.000; t-Stat -0.02; PIP 0.07
    - Reserves/GDP (Bond): Coef. -0.001; t-Stat -0.41; PIP 0.21
    - Reserves/GDP (Bank): Coef. 0.023; t-Stat 0.25; PIP 0.14
    - FX Regime (Equity): Coef. 0.002; t-Stat 0.29; PIP 0.13
    - FX Regime (Bond): Coef. 0.019; t-Stat 1.34; PIP 0.73
    - FX Regime (Bank): Coef. 0.048; t-Stat 0.19; PIP 0.12
    - Average Growth (Equity): Coef. 0.008; t-Stat 0.43; PIP 0.22
    - Average Growth (Bond): Coef. -0.001; t-Stat -0.14

*References and tabulated data extracted exactly as presented in the source unit._

### 0.1      Average  Gr

### _wp15127 - 0.1      Average  Gr

### Key empirical coefficients and statistics (selected entries from table)
- Average Growth: -0.017  -0.05  0.1
- Investor Protection: -0.003  -0.27  0.13
- Investor Protection (alternate entries): 0.000  -0.05  0.09; -0.129  -0.2  0.12
- Law and Order: -0.009  -0.34  0.17; -0.001  -0.13  0.09; 0.105  0.16
- Foreign Equity Stock/GDP: 0.000  0.17  0.09
- Foreign Bond Stock/GDP: 0.001  0.27  0.14
- Foreign OI stock: -0.004  -0.12  0.1
- Local Equity Size: 0.000  0.17  0.09
- Private credit/GDP: 0.027  0.52  0.29
- Relative Equity Size: 0.000  0.07  0.08
- Relative Market Size: 0.000  0.13
- MSCI Benchmark (dummy): 0.000  -0.01  0.09
- EMBI Benchmark (dummy): 0.007  0.25  0.12
- MSCI Frontier Benchmark (dummy): 0.184  1.48  0.77
- Turnover Ratio: 0.001  1.45  0.77
- Share of Equity Funding from AE: 0.000  0.24  0.11
- Share of Bond Funding from AE: 0.000  0.26  0.13
- Correlation w/ EPFR Equity flows: 0.325  1.39  0.74
- Correlation w/ EPFR Bond flows: 0.102  0.7  0.41
- Correlation w/ BIS flows: 37.545  4.09  0.99

### Figures and model outputs (descriptions and highlights)
- Figure 1: Inflows to EMs – BOP Raw Data – Aggregated for 34 EMs
  - Time series plotted from Mar-01 through Sep-13 with values ranging approximately from -1.5 to 3.5 in the visual scale.
- Figure 2: Common EM Factors – Gross vs. Disaggregated Flows
  - Presents estimated common EM dynamics from the model in Section 2.
  - Series labeled: All inflows, Portfolio Equity, Portfolio Bond, OI-Bank, FDI, OI-NonBank.
- Figure 3: Estimated Betas
  - Separate panels for Equity Flows, Bond Flows, OI to Bank Flows.
  - Note: lower and upper dots report the 5th and 95th percentile of the posterior distribution.
- Figure 4: Estimated Common Factor in Total Gross Inflows vs. VIX
  - Vertical scales include VIX and Common Factor - Total Gross Inflows plotted over Mar-01 to Sep-13.
- Figure 5: The Model vs. the GFC and vs. the Taper Tantrum
  - Left panel: cumulative drop in Equity (Bond or Bank) inflows during the GFC (computed at 2007 Q4 and 2008 Q1) vs corresponding beta coefficient.
  - Right panel: beta coefficient vs cumulative drop during the Taper Tantrum (measured as the minimum value among 2013 Q3 and 2013 Q4 for each country).

### Institutional quality and correlations (Figure 6)
- Institutional proxies used:
  - ICRG Law and Order rating.
  - ICRG Investor Protection (Investment Risk) rating.
- Correlation outcomes reported:
  - EPFR Equity Correlation and EPFR Bond Correlation plotted against Law and Order and Investment Risk indices for countries in sample.
  - BIS Bank Correlation plotted similarly.
- Countries displayed include (sample listing from figures): Turkey, South Africa, Argentina, Brazil, Chile, Colombia, Mexico, Peru, Uruguay, Venezuela, Israel, India, Indonesia, Korea, Malaysia, Pakistan, Philippines, Thailand, Belarus, Kazakhstan, Bulgaria, Russian Federation, China,P.R.: Mainland, Ukraine, Czech Republic, Slovak Republic, Estonia, Latvia, Hungary, Lithuania, Croatia, Slovenia, Poland, Romania.

### Annex I — Market structure variables (definitions and proxies)
- Foreign Openness and Size:
  - Based on stock variables to proxy foreign funding received and market size.
  - Stock of foreign equity (or bond) normalized by (i) GDP of recipient market and (ii) total foreign equity (or bond) into the 34 EMs considered.
- Liquidity measures:
  - Equity markets: turnover ratio = total value of shares traded every year divided by average market capitalization.
  - Bond markets: dummy variable capturing membership of country to EMBI (Bond) and MSCI (Equity) benchmarks, used as proxy for liquidity due to limited trading statistics coverage.
- Composition of the Foreign Investor Base:
  - Share of total stock of foreign equity (or bond) funding coming from advanced economies (includes G10 and select financial/offshore centers).
  - Correlation between BOP recorded flows and EPFR recorded flows for portfolio equity/bond (EPFR captures mutual funds in advanced economies).
  - Correlation between BOP recorded bank flows and BIS recorded flows for bank flows.
  - High correlation interpreted as funds/global banks accounting for most movements in capital inflows for that economy.

### Annex II — Data sources and interpretation (EPFR and BIS)
- EPFR global dataset:
  - As of 2013: collecting from more than 29,000 equity funds and 18,000 fixed-income funds representing US$20 trillion of assets invested in over 80 markets.
  - Widely used as a high-frequency proxy of gross inflows when funds are important in the investor base.
  - Methodological caveats: EPFR covers only a fraction of mutual funds and is based on dealer transactions leading to possible discrepancies with residency-based BOP flows; hence reliance on correlations rather than direct subtraction.
- BIS locational banking statistics (LBS):
  - Tracks cross-border positions of banks domiciled in reporting areas and follows residency principles similar to BOP.
  - LBS used to proxy importance of global banks in driving Other Investment to banks by correlating BIS global banks’ claims against borrower-country banking sectors with BOP Other Investment to banks.
  - High correlation interpreted as larger activity of global banks in that borrower country.
- Rationale for correlations:
  - Correlation between BOP flows and EPFR/BIS series used to approximate the importance of mutual funds and global banks respectively in driving recorded gross inflows to Emerging Markets.

*Source: _wp15127 - 0.1      Average  Gr (PDF content provided).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15127.pdf_