## APPENDIX I. EPFR GLOBAL MUTUAL FUND FLOWS DATA

## Source details

**Canonical URL:** [APPENDIX I. EPFR GLOBAL MUTUAL FUND FLOWS DATA](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15277.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15277.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15277.pdf.json)

---

### Data scope and coverage
- Monthly data on roughly 1,100 equity funds, from February 1996 to September 2013.
- Monthly data on 200 bond funds, from November 2003 to September 2013.
- EPFR Global covers about 11,000 equity funds and about 4,500 bond funds, with a combined US$22 trillion in total assets as of the end of 2013.
- Mutual funds intermediary size reaching about USD 30 trillion in 2013 (Investment Company Institute, ICI, 2014).
- Pension funds and insurance companies with about USD 50 trillion assets globally (IMF, 2014a).
- EPFR Global reports coverage of more than 95 percent of EM-focused bond and equity funds; share of U.S. investment in EMs covered by EPFR Global is around 58 percent for equities, and more than 42 percent for bonds as of the end of 2012.
- Sample counts and frequency:
  - Bond funds: around 74,000 fund-country-month observations from November 2003 to September 2013.
  - Equity funds: around 470,000 fund-country-month observations from February 1996 to September 2013.
- Sample exclusions: offshore market economies, economies extremely dependent on oil production, certain euro area countries often classified as EMs, and economies without sufficient data.

### Stylized facts on portfolio flows to emerging markets (EMs)
- Relative growth of bond vs equity flows:
  - Between 2003 and 2013, on average annual bond flows to EMs amounted to 12.5 percent of AUM of mutual funds included in the sample, compared to 5.1 percent for equity flows.
- Structural shifts in intermediaries and instruments:
  - Increased intermediation through open-end funds, passive ETFs, crossover (global) funds, and local-currency bond funds.
  - Decline in the role of closed-end funds relative to open-end funds; rapid growth of ETFs and passive/index funds.
  - Growth of global funds relative to dedicated EM regional funds; global EM funds have grown more rapidly than regional EM funds.
  - Local-currency bond funds expanded more rapidly than hard-currency bond funds.
  - Percentage of defined contribution plans’ assets in mutual funds in the United States rose from 20 percent in 1993 to 60 percent in 2013.
  - Since 2009, flows to EMs from U.S. domiciled funds have grown relative to those from offshore funds.
- Differential sensitivity to global shocks:
  - Bond outflows during the first six months after September 2008 were 25.5 percent of total AUM, while equity outflows were 5.1 percent.
  - Fund types more sensitive to global financial conditions: active funds (excluding ETFs), open-end funds, EM regional funds, U.S. domiciled funds, and local currency bond funds.
  - Fund types less sensitive: passive ETFs, closed-end funds, global or global EM funds, offshore funds, and hard currency bond funds.
- Institutional vs retail resilience:
  - During some distress episodes, equity flows from institutional investors were more resilient than those from retail investors; similar for bond funds during most crisis episodes except the global financial crisis.

### Research questions and approach
- Investigate how different types of mutual funds active in EMs react to global financial shocks.
- Identify fund types more likely to engage in return chasing (momentum trading).
- Examine drivers of differences, including role of flows from end investors versus portfolio managers.
- Use micro-level behavior analysis rather than aggregated capital-flow analysis.

### Key findings (five highlighted results)
- Bond funds are substantially more sensitive to global financial shocks and engage more strongly in momentum trading (sensitivity to past returns) than equity funds.
- Newer local-currency bond funds do not behave very differently from foreign-currency bond funds.
- Flows from retail-oriented funds, open-end funds, and dedicated EM funds are more sensitive to global financial shocks. Sensitivity also differs by fund size, domicile, and style.
- Momentum trading is more prevalent among funds that display a high sensitivity to global financial conditions.
- Differences in volatility of flows from end investors play a key role in explaining the observed patterns across fund types.

### Empirical structure and baseline regression results
- Estimation approach:
  - Fund-level panel regressions run separately for bond and equity funds.
  - Baseline model includes a global factor (Globalt), lagged country return residual (Returni,j,t-1 adjusted for Global), ICRG first difference, and real interest rate differential (RID).
  - Global factors used: VIX (primary), TED spread, MOVE, volatility of Eurodollar futures (robustness).
- Key baseline regression results (VIX as global factor; Table 5):
  - Bond (Full): VIX coefficient -0.150*** (p-value 0.000); Return coefficient 0.037*** (p-value 0.004); ICRG 0.515*** (p-value 0.002); RID -0.015 (p-value 0.790). Observations 73,089; R-squared 0.040.
  - Equity (Full): VIX coefficient -0.048*** (p-value 0.000); Return coefficient 0.008*** (p-value 0.000); ICRG 0.248*** (p-value 0.002); RID 0.007 (p-value 0.649). Observations 453,420; R-squared 0.032.
  - Equity (Common): VIX coefficient -0.052*** (p-value 0.001); Return 0.008*** (p-value 0.000); ICRG 0.288*** (p-value 0.003); RID 0.014 (p-value 0.747). Observations 280,038; R-squared 0.039.
- Economic magnitudes highlighted:
  - An increase in the VIX similar to that of the last four months of 2008—roughly 50 percent—leads monthly bond flows to drop by 7.5 percent of AUM.
  - Average monthly bond flows between 2003 and 2013 were 1 percent of AUM.
- Robustness:
  - Results robust to alternative global factors (TED spread, MOVE, volatility of Eurodollar futures) and to inclusion of a measure of financial openness.

### Heterogeneity across fund characteristics and interactions
- Specification augmentation:
  - Interaction regressions examine how sensitivity to global factor varies by fund characteristics (size, end-investor type, redemption structure, active/passive/ETF, domicile, geographic focus, currency focus).
  - Multiple-characteristic models estimated to control for correlated characteristics; reference group for multiple-characteristic regressions is medium-sized open-end funds (excluding exchange traded funds) sold to retail investors, actively managed, domiciled in the U.S., and investing in emerging markets globally (in hard currency in case of bond).
- Main findings on sensitivity to global risk:
  - Institutional-only funds: country flows less sensitive to changes in global risk than retail-oriented funds.
  - Closed-end funds: flows less sensitive than open-end funds.
  - Bond funds: active non-ETFs more sensitive to global shocks than passive ETFs; for equity funds no strong evidence that ETFs are more reactive than other fund types.
  - Global (crossover) funds: more stable sources of capital flows than dedicated EM funds; result holds after controlling for fund size.
  - Local-currency bond funds: do not seem more sensitive to global shocks than hard-currency bond funds.
  - Small funds: bond and equity fund flows more sensitive to the VIX than larger funds.
  - Domicile effects: funds domiciled in the United Kingdom and the United States are less sensitive to changes in global factors than offshore funds and other European funds.
- Selected interaction coefficients from robustness tables (numeric values preserved):
  - Table 1 (TED Spread): Small bond funds: -1.234* (Single); Small bond funds Multiple: -2.244*** and -2.195*** in related columns. Large bond funds: 1.508** (Single). Close-end funds (bond): 3.293*** (Single). Offshore domicile (bond) in Multiple: -2.224***. UK domicile (bond) Single: 2.735***. Global focus (bond) Single: 2.908***. Hard currency (bond) Multiple: -1.376***. Local currency (bond) Single: -1.233*.
  - Table 2 (MOVE): Small bond funds Single: -0.016. Large bond funds Single: 0.023**. Institutional only (bond) Multiple: 0.032***. Open-end bond Single: -0.056***. Offshore domicile (bond) Single: -0.045***. Global focus (bond) Single: 0.043***. Hard currency (bond) Single: -0.021*.
  - Table 3 (Volatility of Eurodollar Futures): Small bond funds Single: -0.861. Large bond funds Single: 1.106*. Institutional only (bond) Single: 0.853. Open-end bond Single: -2.433***. Offshore domicile (bond) Single: -1.485**. Global focus (bond) Single: 1.988**. Hard currency (bond) Single: -1.691**.
  - Table 4 (VIX with interactions): Small bond funds Single: -0.085** (first column) and -0.054** (second column). Open-end bond: -0.184*** and -0.041**. US domicile (bond): 0.213*** and 0.028**. Offshore (bond): -0.182*** and -0.038***. Global focus (bond): 0.155*** and 0.025. Unassigned currency (bond): 0.164***.

### Return chasing (momentum) by fund type
- Identification: fund-characteristic interaction with lagged index return used to detect return chasing.
- Principal results:
  - Return chasing more prevalent among fund types that are relatively more sensitive to the VIX:
    - Open-end funds show more return chasing than closed-end funds.
    - EM regional funds show more return chasing than global/crossover funds.
    - Offshore funds and continental European funds engage more in momentum trading than U.K. funds.
  - Exceptions among bond funds:
    - Momentum trading appears more prevalent in bond funds with institutional investors’ money relative to retail funds.
- Specific reported coefficients (examples):
  - Bond, Open-end interaction with lagged return: 0.039***.
  - Equity, Open-end interaction with lagged return: 0.006***.
  - Bond, Institutional-only interaction with lagged return: 0.023**.

### Roles of end investors versus fund managers (variance decomposition)
- Decomposition formula:
  - Flowi,j,t = Flowi,t (ultimate investor flows allocated proportionally) + (Flowi,j,t - Flowi,t) (manager allocation changes).
- Stability measures (selected entries from Table 9):
  - Global bond funds: standard deviation 5.0 (percent); share of variance attributable to end-investor flows 17.1 percent.
  - Global EM and EM regional bond funds: standard deviations over 7 percent and share of variance greater than for global bond funds.
  - Close-end funds (bond): standard deviation 1.3 percent; share of variance explained by end-investor flows 6.4 percent.
  - By size (bond): Small standard deviation 8.2; Medium 7.4; Large 4.9.
  - By domicile (bond): US standard deviation 5.1; Offshore 8.2.
  - By other structure (equity, passive ETF): standard deviation 8.7; share of variance 49.2 percent.
- Interpretation:
  - Funds with more stable ultimate investor flows are generally less sensitive to global financial conditions and display weaker return chasing.
  - Lower sensitivity of crossover/global funds partly attributable to lower fluctuations of flows from ultimate investors.

### Methodology and data construction details
- Flow-to-AUM ratio: w_{i,j,t} is allocation weight at end of month t; A_{i,t} is AUM at end of month t; r_{j,t} is index return from t-1 to t. Estimated flow normalized by average of assets allocated to the country at the beginning and end of the month.
- Indexes used:
  - MSCI for equity funds.
  - GBI-EM for local currency bond funds (returns adjusted using bilateral exchange rates to obtain U.S. dollars-denominated returns).
  - EMBI Global for other bond funds.
- Data cleaning rules:
  - Discard samples when (1) internal inconsistency for AUM > 0.1 percent for the same fund at the same period in different parts of the database; and (2) computed fund flow per AUM > 100% in absolute value.
- Global risk measures:
  - VIX: implied volatility index for the S&P 500 (end of month).
  - TED spread: three-month Eurodollar deposit rate minus the three-month Treasury bill rate (end of month).
  - MOVE: average implied volatility across options on U.S. Treasuries (end of month).
  - Volatility of Eurodollar futures: realized volatility from daily changes in Thomson Reuter’s futures continuous series index of the 9th three-month futures in the corresponding month and annualized by multiplying with the square root of 250.
- Return: country index return calculated as excess dollar returns of relevant indices over one month U.S. Eurodollar deposit rate; one-month-lagged return used and orthogonalized relative to the VIX or relevant global factor.
- Country risk and interest differential:
  - ICRG: composite country risk rating based on 22 variables; higher rating means lower risk.
  - RID: one-month-lagged real interest rate differential = local short term interest rate (around one month) minus U.S. short term interest rate minus the difference in the one-year-ahead forecasts of CPI inflation from Consensus Forecasts.

### Robustness tests
- Alternative global factors and interaction specifications examined:
  - TED Spread, MOVE, Volatility of Eurodollar Futures, and VIX with interactions.
- Tables presented report coefficients on interactions between global factors and fund characteristic dummies with significance denoted by ***, **, and * for 1, 5, and 10 percent levels; inference based on Driscoll and Kraay (1998) robust standard errors.

### Conclusions and policy-relevant implications
- Overall findings:
  - The “maturing” of the EM asset class has coincided with capital flows that are more volatile, more sensitive to global financial conditions, and more procyclical, despite stronger macroeconomic fundamentals in EMs.
  - Bond funds have risen in importance in intermediation to EMs and are relatively more sensitive to global factors and more engaged in momentum trading than equity funds.
  - No support found for the view that the growth of crossover funds, ETFs, and local currency bond funds per se has increased sensitivity of EM portfolio flows to global financial conditions; crossover funds appear relatively stable, partly due to more stable end-investor flows.
  - Domicile matters: funds domiciled in the United Kingdom and the United States are less sensitive to changes in the VIX than funds domiciled elsewhere.
- Country-level implication and recommendation:
  - From a recipient country perspective, knowing and diversifying the investor base (including end investors) is important.
  - Understanding and diversifying the investor base, possibly by working closely with asset managers, can help EM countries mitigate the volatility risk associated with international capital flows while reaping benefits from financial globalization.

*Source: APPENDIX I. EPFR GLOBAL MUTUAL FUND FLOWS DATA, _wp15277*

### APPENDIX I. EPFR GLOBAL MUTUAL FUND FLOWS DATA ..................................................... 26

### APPENDIX I. EPFR GLOBAL MUTUAL FUND FLOWS DATA

### Data scope and coverage
- Monthly data on roughly 1,100 equity funds, from February 1996 to September 2013.
- Monthly data on 200 bond funds, from November 2003 to September 2013.
- Mutual funds intermediary size reaching about USD 30 trillion in 2013 (Investment Company Institute, ICI, 2014).
- Pension funds and insurance companies with about USD 50 trillion assets globally (IMF, 2014a).

### Changes in the investor landscape and market structure
- EM financial markets have become deeper and more interconnected since the 1990s; role of bond flows has grown strongly with development of local-currency bond markets.
- Mix of global portfolio investors shifted toward mutual funds and ETFs; rising share of EM investment is channeled through ETFs and crossover investors (portfolios including both DMs and EMs).
- Open-end mutual funds allow end investors to redeem quickly, which can force funds to sell underlying assets; ETFs can issue/withdraw shares via authorized participants (APs), which can cushion but not eliminate transmission of end-investor sales to underlying assets.

### Research questions and approach
- Investigate how different types of mutual funds active in EMs react to global financial shocks.
- Identify fund types more likely to engage in return chasing (momentum trading).
- Examine drivers of differences, including role of flows from end investors versus portfolio managers.
- Use micro-level behavior analysis rather than aggregated capital-flow analysis.

### Key findings (five highlighted results)
- Bond funds are substantially more sensitive to global financial shocks and engage more strongly in momentum trading (sensitivity to past returns) than equity funds.
- Newer local-currency bond funds do not behave very differently from foreign-currency bond funds.
- Flows from retail-oriented funds, open-end funds, and dedicated EM funds are more sensitive to global financial shocks. Sensitivity also differs by fund size, domicile, and style.
- Momentum trading is more prevalent among funds that display a high sensitivity to global financial conditions.
- Differences in volatility of flows from end investors play a key role in explaining the observed patterns across fund types.

### Implications and interpretation
- The rising share of bond flows in recent years may have increased total portfolio flows to EMs’ sensitivity to global financial conditions and made those flows more procyclical.
- Distinctive mandates, constraints, EM expertise, and incentives across fund types (dedicated EM funds, crossover funds, institutional investors) drive heterogeneous behavior during global shocks.
- Institutional investors (including pension funds, insurance companies, sovereign wealth funds, but excluding mutual funds or ETFs) are generally less subject to rapid redemptions but often constrained to investment-grade assets and buy-and-hold strategies.
- Open-end mutual funds’ liability structure makes ultimate end-investor behavior a significant driver of flow volatility into EMs.

### Empirical structure (overview of appended materials)
- Appendix components include: A. Portfolio Flows to Each Country from Each Fund; B. Explanatory Variables; C. Definition of Mutual Fund Characteristics.
- Tables inventory includes observation counts by investment destination economy, fund characteristics, domicile, correlation coefficients across alternative global factors, base estimation and robustness results, momentum/return-chasing measures, and variance decomposition of end-investor flows.
- Figures inventory includes cumulative bond and equity flows to EMs, cumulative flows by fund type (end investors), and global factors.

*Source: APPENDIX I. EPFR GLOBAL MUTUAL FUND FLOWS DATA, _wp15277*

### Section IV concludes.

### _wp15277 - Section IV concludes.

### Data description and sample coverage
- Mutual fund data source: EPFR Global, with a coverage of US$22 trillion in total assets as of the end of 2013.
- Data frequency and content: monthly fund-level data including assets under management (AUM), inflows and outflows (redemptions), and asset allocation by country; flows to each country estimated adjusting for valuation effects.
- Sample scope and counts:
  - Bond funds: around 74,000 fund-country-month observations from November 2003 to September 2013.
  - Equity funds: around 470,000 fund-country-month observations from February 1996 to September 2013.
- Sample exclusions: offshore market economies, economies extremely dependent on oil production, certain euro area countries often classified as EMs (e.g., Estonia), and economies without sufficient data.

### Stylized facts on portfolio flows to emerging markets (EMs)
- Relative growth of bond vs equity flows:
  - Between 2003 and 2013, on average annual bond flows to EMs amounted to 12.5 percent of AUM of mutual funds included in the sample, compared to 5.1 percent for equity flows.
- Structural shifts in intermediaries and instruments:
  - Increased intermediation through open-end funds, passive ETFs, crossover (global) funds, and local-currency bond funds.
  - Decline in the role of closed-end funds relative to open-end funds; rapid growth of ETFs and passive/index funds.
  - Growth of global funds relative to dedicated EM regional funds; global EM funds have grown more rapidly than regional EM funds.
  - Local-currency bond funds expanded more rapidly than hard-currency bond funds.
  - Mutual funds historically sold mainly to retail investors; institutional investors have been purchasing an increasing number of mutual fund shares (example: percentage of defined contribution plans’ assets in mutual funds in the United States rose from 20 percent in 1993 to 60 percent in 2013).
  - Since 2009, flows to EMs from U.S. domiciled funds have grown relative to those from offshore funds.
- Differential sensitivity to global shocks:
  - Bond flows more volatile than equity flows: bond outflows during the first six months after September 2008 were 25.5 percent of total AUM, while equity outflows were 5.1 percent.
  - Fund types more sensitive to global financial conditions: active funds (excluding ETFs), open-end funds, EM regional funds, U.S. domiciled funds, and local currency bond funds.
  - Fund types less sensitive: passive ETFs, closed-end funds, global or global EM funds, offshore funds, and hard currency bond funds.
- Institutional vs retail resilience:
  - During some distress episodes, equity flows from institutional investors were more resilient than those from retail investors; similar for bond funds during most crisis episodes except the global financial crisis.

### Regressions: bond vs equity fund behavior and global factors
- Estimation approach:
  - Fund-level panel regressions run separately for bond and equity funds.
  - Baseline model includes a global factor (Globalt), lagged country return residual (Returni,j,t-1 adjusted for Global), ICRG first difference, and real interest rate differential (RID).
  - Global factors used: VIX (primary), TED spread, MOVE, volatility of Eurodollar futures (robustness).
- Key baseline regression results (VIX as global factor; Table 5):
  - Bond (Full): VIX coefficient -0.150*** (p-value 0.000); Return coefficient 0.037*** (p-value 0.004); ICRG 0.515*** (p-value 0.002); RID -0.015 (p-value 0.790). Observations 73,089; R-squared 0.040.
  - Equity (Full): VIX coefficient -0.048*** (p-value 0.000); Return coefficient 0.008*** (p-value 0.000); ICRG 0.248*** (p-value 0.002); RID 0.007 (p-value 0.649). Observations 453,420; R-squared 0.032.
  - Equity (Common): VIX coefficient -0.052*** (p-value 0.001); Return 0.008*** (p-value 0.000); ICRG 0.288*** (p-value 0.003); RID 0.014 (p-value 0.747). Observations 280,038; R-squared 0.039.
- Economic magnitudes highlighted:
  - An increase in the VIX similar to that of the last four months of 2008—roughly 50 percent—leads monthly bond flows to drop by 7.5 percent of AUM.
  - Average monthly bond flows between 2003 and 2013 were 1 percent of AUM.
- Robustness:
  - Results robust to alternative global factors (TED spread, MOVE, volatility of Eurodollar futures) and to inclusion of a measure of financial openness.

### Heterogeneity across fund characteristics and interactions
- Specification augmentation:
  - Interaction regressions examine how sensitivity to global factor varies by fund characteristics (size, end-investor type, redemption structure, active/passive/ETF, domicile, geographic focus, currency focus).
  - Multiple-characteristic models estimated to control for correlated characteristics; reference groups detailed for equity and bond models.
- Main findings on sensitivity to global risk (summary of Table 7 results):
  - Institutional-only funds: country flows less sensitive to changes in global risk than retail-oriented funds.
  - Closed-end funds: flows less sensitive than open-end funds.
  - Bond funds: active non-ETFs more sensitive to global shocks than passive ETFs; for equity funds no strong evidence that ETFs are more reactive than other fund types.
  - Global (crossover) funds: more stable sources of capital flows than dedicated EM funds; result holds after controlling for fund size.
  - Local-currency bond funds: do not seem more sensitive to global shocks than hard-currency bond funds.
  - Small funds: bond and equity fund flows more sensitive to the VIX than larger funds.
  - Domicile effects: funds domiciled in the United Kingdom and the United States are less sensitive to changes in global factors than offshore funds and other European funds.
- Interaction-augmented models (Model (4)) that allow characteristic-specific reactions to all explanatory variables produce broadly comparable results.

### Return chasing (momentum) by fund type
- Identification: Fund-characteristic interaction with lagged index return used to detect return chasing.
- Principal results (Table 8 summary):
  - Return chasing more prevalent among fund types that are relatively more sensitive to the VIX:
    - Open-end funds show more return chasing than closed-end funds.
    - EM regional funds show more return chasing than global/crossover funds.
    - Offshore funds and continental European funds engage more in momentum trading than U.K. funds.
  - Exceptions among bond funds:
    - Momentum trading appears more prevalent in bond funds with institutional investors’ money relative to retail funds (contrast with some prior findings).
  - Specific coefficients (examples):
    - Bond, Open-end interaction with lagged return: 0.039*** (significant).
    - Equity, Open-end interaction with lagged return: 0.006*** (significant).
    - Bond, Institutional-only interaction with lagged return: 0.023** (significant).

### Roles of end investors vs fund managers (variance decomposition)
- Decomposition: Flowi,j,t = Flowi,t (ultimate investor flows allocated proportionally) + (Flowi,j,t - Flowi,t) (manager allocation changes).
- Stability measures (Table 9 highlights):
  - Standard deviation of ratio of end-investor flow to fund AUM and share of total variance attributable to end-investor flows calculated by fund type.
  - Global bond funds: standard deviation 5.0 (percent); share of variance attributable to end-investor flows 17.1 percent.
  - Global EM and EM regional bond funds: standard deviations over 7 percent and share of variance greater than for global bond funds.
  - Close-end funds show much lower standard deviation (bond: 1.3 percent) and lower share of variance explained by end-investor flows (bond close-end share 6.4 percent).
  - Examples (selected entries):
    - By size (bond): Small standard deviation 8.2; Medium 7.4; Large 4.9.
    - By domicile (bond): US standard deviation 5.1; Offshore 8.2.
    - By other structure (equity, passive ETF): standard deviation 8.7; share of variance 49.2 percent.
- Interpretation:
  - Funds with more stable ultimate investor flows are generally less sensitive to global financial conditions and display weaker return chasing.
  - Lower sensitivity of crossover/global funds partly attributable to lower fluctuations of flows from ultimate investors.

### Conclusion and policy-relevant implications
- Overall findings:
  - The “maturing” of the EM asset class has coincided with capital flows that are more volatile, more sensitive to global financial conditions, and more procyclical, despite stronger macroeconomic fundamentals in EMs.
  - Bond funds have risen in importance in intermediation to EMs and are relatively more sensitive to global factors and more engaged in momentum trading than equity funds.
  - No support found for the view that the growth of crossover funds, ETFs, and local currency bond funds per se has increased sensitivity of EM portfolio flows to global financial conditions; crossover funds appear relatively stable, partly due to more stable end-investor flows.
  - Domicile matters: funds domiciled in the United Kingdom and the United States are less sensitive to changes in the VIX than funds domiciled elsewhere.
- Country-level implication and recommendation:
  - From a recipient country perspective, knowing and diversifying the investor base (including end investors) is important.
  - Understanding and diversifying the investor base, possibly by working closely with asset managers, can help EM countries mitigate the volatility risk associated with international capital flows while reaping benefits from financial globalization.

*Source: _wp15277 - Section IV concludes.*

### APPENDIX I. EPFR GLOBAL MUTUAL FUND FLOWS DATA

### APPENDIX I. EPFR GLOBAL MUTUAL FUND FLOWS DATA

### Data coverage and scope
- EPFR Global covers about 11,000 equity funds and about 4,500 bond funds, with a combined US$22 trillion in total assets as of the end of 2013.
- According to EPFR Global, its data track more than 95 percent of EM-focused bond and equity funds.
- The database is composed mainly of mutual funds and ETFs, with a limited number of hedge funds.
- The share of U.S. investment in EMs covered by EPFR Global is around 58 percent for equities, and more than 42 percent for bonds as of the end of 2012.
- Coverage limitations:
  - The database covers primarily mutual funds and does not cover all investment fund flows nor capital flows intermediated directly by institutional investors.
  - Such coverage gaps may not materially affect the analysis because the focus is on differences across different types of mutual funds to understand aggregate dynamics.
- Reporting frequency and basis:
  - EPFR Global provides high-frequency reporting (monthly for funds’ asset allocation by country; daily and weekly for selected indicators from sub-samples of funds).
  - EPFR Global records transactions on a nationality basis; Balance of Payments (BoP) and CPIS report on a residency basis.
  - The CPIS can misclassify some bond fund flows as equity flows (Felettigh and Monti, 2008).

### A. Portfolio flows to each country from each fund — methodology
- Flow-to-AUM ratio of fund i for country j at month t is calculated using:
  - Definitions: w_{i,j,t} is the allocation weight at the end of month t; A_{i,t} is the AUM at the end of month t; r_{j,t} is the index return from t-1 to t.
  - The estimated flow to a country from a fund in a month is normalized by the average of the assets allocated to the country at the beginning and end of the month.
  - Asset returns are approximated by country index returns, following Gelos and Wei (2005).
- Indexes used:
  - MSCI for equity funds.
  - GBI-EM for local currency bond funds (since GBI-EM is a local currency index, returns are adjusted using bilateral exchange rates to obtain U.S. dollars-denominated returns).
  - EMBI Global for other bond funds.
- Data cleaning rules:
  - Discard samples when (1) there is a large (more than 0.1 percent) internal inconsistency for AUM of the same fund at the same period in different parts of the database; and (2) computed fund flow per AUM is greater than 100% in absolute value.

### B. Explanatory variables
- Global risk measures:
  - VIX: implied volatility index for the S&P 500 as reported by the Chicago Board Options Exchange (end of month data used).
  - TED spread: defined as the three-month Eurodollar deposit rate minus the three-month Treasury bill rate (end of month data used).
  - MOVE: average implied volatility across a wide range of outstanding options on U.S. Treasuries (end of month data used).
  - Volatility of Eurodollar futures: realized volatility calculated using daily changes in the Thomson Reuter’s futures continuous series index of the 9th three-month futures in the corresponding month and annualized by multiplying with the square root of 250.
- Return:
  - Country index return calculated as excess dollar returns of relevant indices over one month U.S. Eurodollar deposit rate.
  - MSCI used for equity funds; U.S. dollars-denominated GBI-EM for local currency bond funds; EMBI Global for other bond funds.
  - One-month-lagged return used to mitigate endogeneity; three-month moving averages also taken.
  - Individual country index return is orthogonalized relative to the VIX or the relevant global factor to reduce collinearity with global factors.
- Country risk and interest differential:
  - ICRG: composite country risk rating based on 22 variables (political, financial, economic); higher rating means lower risk.
  - RID: one-month-lagged real interest rate differential = local short term interest rate (around one month) minus U.S. short term interest rate minus the difference in the one-year-ahead forecasts of CPI inflation from Consensus Forecasts.

### C. Definition of mutual fund characteristics (classification)
- Fund size:
  - Large and small funds defined as those above the 80th and below 20th percentiles of AUM in each month, respectively; remainder labeled medium funds.
- End investor type:
  - EPFR provides share-class data identifying shares targeted to retail or institutional investors.
  - Funds sold only to institutional investors or sold also to retail investors are identified.
  - When ultimate investor type is not fully reported, a fund is identified as a retail fund when some of its shares are known to be sold to retail investors.
  - Unidentified observations are limited to around 7 percent of bond funds and around 14 percent of equity funds.
- Open-end vs closed-end:
  - Open-end funds allow flexible additions/redemptions; closed-end funds do not.
- Investment style and ETF status:
  - Four combined characteristics: active, non-ETF; active, ETF; passive, non-ETF; passive ETF.
  - Most ETFs are passively managed; most non-ETFs are actively managed.
- Fund domicile (five regions):
  - The United States; offshore market economies; the United Kingdom; other European countries; and others.
  - Fund domicile roughly corresponds to residence of ultimate mutual fund shareholders; offshore domiciles may weaken that correspondence.
- Geographic focus of fund investment destination:
  - Four groups: global funds; global EM funds; EM regional funds; advanced economy (AE) regional funds.
  - Example: global funds are categorized as “Global" or "Global ex-US" by EPFR Global.
- Currency focus (bond funds only):
  - Five groups: hard currency; local currency; blend currency; unassigned; other currency focus.

### Robustness tests — sensitivity of bond and equity flows to global factors by fund types (summary of presented tables)
- Tables reported:
  - Table 1: TED Spread as the global factor — estimated coefficients on interactions between the TED spread and fund characteristic dummies (single and multiple specifications). Significance indicated by ***, **, and * for 1, 5, and 10 percent levels; inference based on Driscoll and Kraay (1998) robust standard errors.
  - Table 2: MOVE Index as the global factor — same structure and inference note as Table 1.
  - Table 3: Volatility of Eurodollar Futures as the global factor — same structure and inference note as Table 1.
  - Table 4: VIX as the global factor with interactions with other factors — reports estimated coefficients on interactions between the VIX and one fund characteristic dummy in a model including additional interaction terms; same significance and inference notes.
- Reference group for multiple-characteristic regressions (model (3) or analogous) is explicitly stated:
  - Medium-sized open-end funds (excluding exchange traded funds) sold to retail investors, actively managed, domiciled in the U.S., and investing in emerging markets globally (in hard currency in case of bond).
- Select illustrative coefficient values as reported in the tables (preserving numeric values exactly as presented):
  - Table 1 (TED Spread): Small bond funds: -1.234* (Single); Small bond funds Multiple: -2.244*** and -2.195*** in related columns. Large bond funds: 1.508** (Single). Close-end funds (bond): 3.293*** (Single). Offshore domicile (bond) in Multiple: -2.224***. UK domicile (bond) Single: 2.735***. Global focus (bond) Single: 2.908***. Hard currency (bond) Multiple: -1.376***. Local currency (bond) Single: -1.233*.
  - Table 2 (MOVE): Small bond funds Single: -0.016. Large bond funds Single: 0.023**. Institutional only (bond) Multiple: 0.032***. Open-end bond Single: -0.056***. Offshore domicile (bond) Single: -0.045***. Global focus (bond) Single: 0.043***. Hard currency (bond) Single: -0.021*.
  - Table 3 (Volatility of Eurodollar Futures): Small bond funds Single: -0.861. Large bond funds Single: 1.106*. Institutional only (bond) Single: 0.853. Open-end bond Single: -2.433***. Offshore domicile (bond) Single: -1.485**. Global focus (bond) Single: 1.988**. Hard currency (bond) Single: -1.691**.
  - Table 4 (VIX with interactions): Small bond funds Single: -0.085** (first column) and -0.054** (second column). Open-end bond: -0.184*** and -0.041**. US domicile (bond): 0.213*** and 0.028**. Offshore (bond): -0.182*** and -0.038***. Global focus (bond): 0.155*** and 0.025. Unassigned currency (bond): 0.164***.
- Note: The tables present many additional coefficients across fund characteristics and for both bond and equity flows; statistical significance is denoted by ***, **, and * for 1, 5, and 10 percent levels, respectively.

*Source: _wp15277 - APPENDIX I. EPFR GLOBAL MUTUAL FUND FLOWS DATA*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15277.pdf_
