## _wp15263

## Source details

**Canonical URL:** [_wp15263](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15263.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15263.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15263.pdf.json)

---

### Introduction and framing
- Foreign investors held about a trillion dollars of emerging market (EM) government debt as of end-2014.
- 80 percent of these investments were intermediated through global asset managers (Arslanalp and Tsuda 2014).
- Two investor types distinguished in EM local-currency bond markets:
  - Benchmark-driven investors: invest through a fund that tracks or closely follows a flagship benchmark index (usually the J.P. Morgan Government Bond Index-Emerging Markets (GBI-EM)).
  - Unconstrained investors: can invest without being restricted by index considerations.
- Benchmark-driven category can include “passive,” “closet-index,” or “weakly active” funds, and can represent retail or institutional investors depending on mandates.

### Key findings on flows and investor behavior
- Total foreign holdings of EM local-currency government bonds:
  - $200 billion at end-2007.
  - $600 billion by end-2014.
- Foreign holdings rose similarly across countries in the J.P. Morgan GBI-EM universe, with foreign share increasing steadily from 2007 until the May 2013 taper tantrum.
- Three distinct phases of foreign flows (Arslanalp and Tsuda, 2014):
  - Pre-crisis: 2006–07 (moderate differentiation among countries).
  - Crisis: 2008–09 (sharpened differentiation).
  - Post-crisis: 2010–12 (foreign flows positive almost everywhere).
- Drivers of post-crisis inflows:
  - Improving EM fundamentals; several EMs reached or regained investment grade during 2010–12 (Colombia, Indonesia, Latvia, Romania, Uruguay).
  - Rising popularity of index-funds and benchmark use (Raddatz et al. 2015), contributing to synchronized portfolio flows across diverse countries.
- Evidence on benchmark influence:
  - Raddatz et al. (2015): benchmarks explain, on average, around 70 percent of country allocations after controlling for industry, macroeconomic, and country-specific effects.
  - Miyajima and Shim (2014): median “active share” is 17 percent, implying more than 80 percent overlap in country weights between fund portfolios and benchmarks; nearly 70 percent of actively managed EM bond funds are “closet index” or “weakly active.”

### Size and timing of benchmark-driven investors (aggregate estimates)
- Combined evidence (surveys, event studies, empirical estimates):
  - Benchmark-driven investors represented $200–$250 billion at end-2014.
  - This corresponds to more than a third of total foreign holdings of $600 billion.
  - Peak benchmark-driven investor base: $300 billion just before the May 2013 taper tantrum.
- Conceptual implications:
  - Benchmark-driven investors can be a stable funding source because index-tracking funds must allocate to countries included in the index.
  - They can also introduce correlated behavior among asset managers, increasing sensitivity of countries to global risk sentiment and correlated portfolio flows independent of country fundamentals.

### Box 1 — Emerging Market Bond Indices: major features
- J.P. Morgan GBI-EM (launched 2005):
  - Three versions: GBI-EM Broad, GBI-EM Global, and GBI-EM; each has a diversified overlay.
  - Diversified version places a 10 percent cap for each country.
  - GBI-EM Global Diversified accounted for around 90 percent of all AUM benchmarked to the GBI-EM suite as of end-2014.
  - Main entry requirement: market accessibility; no minimum-rating requirements or explicit market size limits.
  - Treasury bills and inflation-indexed bonds not eligible (only fixed-rate nominal bonds).
  - As of end-2014, 16 countries included: Brazil, Chile, Colombia, Hungary, Indonesia, Malaysia, Mexico, Nigeria, Peru, Philippines, Poland, Romania, Russia, South Africa, Thailand, and Turkey.
- Barclays Emerging Markets Local Currency Government Index (launched 2008):
  - Entry requirements: minimum market capitalization of $5 billion and market accessibility.
  - Nominal bonds and bills eligible; inflation-indexed bonds not eligible.
  - Overlaps largely with J.P. Morgan GBI-EM Global; since March 2013 includes the Czech Republic, Israel, and Korea.
  - Miyajima and Shin (2014): J.P. Morgan and Barclays EM bond indices have more than an 80 percent overlap in country weights.
- Citi Emerging Markets Government Bond Index (EMGBI, launched 2013):
  - Entry requirements: minimum market capitalization of US$10 billion and market accessibility.
  - Treasury bills and inflation-indexed bonds not eligible.
  - As of end-2014 includes all GBI-EM Global countries except Nigeria and Romania.
- Regional indices exist (HSBC ALBI, Markit iBoxx ABF).
- Major global bond indices:
  - Barclays Global Aggregate Index (Global AGG): market capitalization of $43 trillion as of end-2014; Treasury sector market capitalization $23 trillion at end-2014; main entry: investment-grade status and freely traded convertible currency; eight EMs included as of end-2014 with country weights (Chile 0.01 percent, Malaysia 0.2 percent, Mexico 0.4 percent, Poland 0.2 percent, Russia 0.1 percent, South Africa 0.2 percent, Thailand 0.2 percent, Turkey 0.2 percent).
  - Citibank World Government Bond Index (Citi WGBI): market capitalization of $20 trillion as of end-2014; main entry: minimum market capitalization of $50 billion, domestic long-term credit rating of A-/A3 by S&P/Moody's; four EMs included as of end-2014 with country weights (Malaysia 0.4 percent, Mexico 0.8 percent, Poland 0.5 percent, South Africa 0.4 percent).
- Weight of EMs in major global bond indices: less than 1 percent at the country level and around 2 percent at the aggregate level (among global bond indices).

### Event studies: estimating benchmark-driven investor base (2014)
- Aggregate result from three event studies: benchmark-driven investor pool range $170–$270 billion as of 2014 (average around $227 billion), in line with J.P. Morgan survey ($221 billion).
- Identification approach:
  - Use episodes where a country’s weight in the J.P. Morgan GBI-EM index rises sharply due to an increase in the nominal stock of “index-eligible” securities (inclusion events).
  - Treat inclusions as exogenous supply shocks that should trigger automatic demand by benchmark-driven investors but not by unconstrained investors.
  - Decompose change in country weight into buy-and-hold (valuation) and exogenous components; only the exogenous component should trigger benchmark-driven flows.
- Estimation equation variables:
  - B_t = benchmark-driven investor base at time t in U.S. dollars.
  - f_it = net foreign purchase of country i's bonds between time t and t+1 in U.S. dollars.
  - w_it = weight of country i in the benchmark at time t.
  - R_it and R_bt = total gross returns of country i's bonds and the benchmark, respectively, from t to t+1.
- Three 2014 episodes used: Colombia (April–September 2014), Peru (November–December 2014), Romania (April–May 2014).

- Colombia (March–September 2014):
  - Inclusion: five additional Colombian bonds included phased until September 2014.
  - Index-eligible debt stock increased by 150 percent.
  - Country weight rose by 4.45 percentage points from 3.24 percent at end-February to 7.69 percent at end-September.
  - Buy-and-hold component: 0.10 percentage points.
  - Exogenous component (a): 4.35 percentage points.
  - Observed net foreign purchases (b): $7.36 billion.
  - Estimated benchmark-driven investor base (c = b / a): 169 (billion U.S. dollars).

- Peru (November–December 2014):
  - Inclusion: Peru 5.7 percent August 2020 bond (7.4 billion nuevo soles, $2.5 billion) included at end-November 2014.
  - Index-eligible debt stock increased by 17 percent.
  - Country weight rose by 0.39 percentage points from 1.47 percent at end-October to 1.86 percent at end-December.
  - Buy-and-hold component: 0.09 percentage points.
  - Exogenous component (a): 0.30 percentage points.
  - Observed net foreign purchases (b): $0.72 billion.
  - Estimated benchmark-driven investor base (c = b / a): 238 (billion U.S. dollars).

- Romania (April–June 2014):
  - Inclusion: two new Romanian bonds added end-April and end-May 2014.
  - Index-eligible debt stock increased by 45 percent (11.5 billion lei, $3.5 billion).
  - Country weight rose from 1.47 percent at end-Mar 2014 to 2.03 percent at end-Jun 2014 (change: 0.56 percent).
    - Buy-and-hold component: 0.03 percent.
    - Exogenous component (a): 0.53 percent.
  - Net foreign purchases during Apr–Jun 2014 (b): $1.44 billion.
  - Estimated benchmark-driven investor base (c = b / a): 273 (billion U.S. dollars).
  - Text summary: “0.53 percentage points of the rise in the country weight was exogenous (i.e. due to the inclusion of the new bonds) ... Net foreign purchases of Romanian local government bonds were $1.44 billion during April-June, the highest level in 2014. Assuming unconstrained investors did not alter their positions due to the index change, this would imply a benchmark-driven investor base of around $270 billion at the time of the event (Table 4).”

### Empirical estimates: benchmark-driven vs. unconstrained investors
- Method:
  - Balston and Melin (2013) decomposition using monthly foreign holdings.
  - Two pools: benchmark-driven (restricted to index-eligible fixed-rate nominal bonds, 16 index countries, 10 percent country limit) and unconstrained (invests by market capitalization across broader securities and countries).
  - Estimation via constrained least squares (CLS) with at + bt equaling total foreign holdings each month.
  - Sample period: January 2010 to June 2015.
  - Sample includes 18 countries (13 in GBI-EM Global Diversified, 5 off-benchmark: China, the Czech Republic, India, Israel, and Korea).
- Key empirical estimates:
  - Pool of benchmark-driven investors around $240 billion at end-2014.
  - J.P. Morgan survey: $221 billion.
  - Average estimate in event studies: $227 billion.
  - Peak benchmark-driven investor base: $300 billion just before the May 2013 taper tantrum.
  - Total foreign holdings: $600 billion.
- Data notes:
  - Total outstanding and foreign holdings from national data sources (Arslanalp and Tsuda, 2014).
  - For China and India, outstanding amounts reflect the “investable” portion given foreign investment quotas.
  - EPFR Global data show assets under management by EM local currency bond funds; empirical estimates are somewhat higher because they include separate managed accounts and other vehicles.

### Cross-country patterns and dynamics
- Temporal and cross-country variation:
  - At peak (just before taper tantrum), benchmark-driven investors accounted for close to one-half of total foreign holdings in EMs.
  - By end-2014 benchmark-driven share fell; unconstrained investors gained relative importance afterwards.
  - For some countries (Colombia, Peru, Romania) benchmark-driven investors still represent the bulk of the foreign investor base as of end-2014.
- Possible explanations for country differences:
  - Supply-side: larger, more liquid government bond markets attract unconstrained investors (examples: Mexico, Brazil, Poland); smaller markets may attract investors mainly after index inclusion.
  - Demand-side: countries exposed to outflows from unconstrained investors due to country-specific shocks may have a higher presence of benchmark-driven investors.
- Empirical relationships:
  - Share of benchmark-driven investors is inversely related to market size (market capitalization).
  - Benchmark-driven outflows persisted for about one year after the May 2013 taper tantrum, suggesting correlated portfolio flows across benchmarked countries.

### Key implications
- First:
  - Benchmarks link countries in the same portfolio and can trigger correlated portfolio flows, connecting countries beyond fundamentals.
  - Benchmark-driven investor outflows after May 2013 lasted about one year.
  - Policy implication: liquidity buffers can be useful even for countries with strong fundamentals, especially where benchmark-driven investors have a large presence.
- Second:
  - As the share of benchmark-driven investors has fallen since the taper tantrum, it is increasingly important for EMs to maintain strong domestic policies.
  - Unconstrained investors have greater latitude to switch across countries based on risk-return, increasing the importance of domestic fundamentals.
- Third:
  - Major EM local currency bond indices do not require a minimum credit rating for inclusion.
  - Ratings-sensitive investors may be limited within the benchmark-driven base; unconstrained investors could be more sensitive to credit-rating changes.
  - Policy implication: safeguarding investment grade status may be important, especially for countries with a large unconstrained investor base.
- Fourth:
  - For countries not yet included in EM local currency bond indices, the estimate of the benchmark-driven investor base provides a basis to assess potential capital inflows upon index inclusion.

### Conclusion
- Convergent evidence across methods:
  - Various approaches suggest about $200–$250 billion of foreign investments in EM local currency government bonds were benchmark-driven as of end-2014, representing more than one third of total foreign holdings ($600 billion).
  - Benchmark-driven prominence grew until a peak before the May 2013 taper tantrum (nearly one-half of total foreign holdings).
- Behavioral insight:
  - The taper tantrum marked a shift: exchange rate volatility became more important for benchmark-driven investors after May 2013, prompting a move from broad asset allocation to country selection.
  - EM investors have become more differentiating post-taper tantrum.
- Ongoing role:
  - Despite a decline from peak, benchmark-driven investors remain large enough to play an important role in driving portfolio flows to emerging markets, particularly in countries where they have a large presence.

*Source: IMF staff compilation from the content unit titled "Box 1. Emerging Market Bond Indices" and associated sections in the supplied PDF (_wp15263).*

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

### _wp15263 - References

### Introduction and framing
- Foreign investors held about a trillion dollars of emerging market (EM) government debt as of end-2014.
- Most of these investments—80 percent of the total—were intermediated through global asset managers (Arslanalp and Tsuda 2014).
- The paper distinguishes two investor types in EM local-currency bond markets:
  - Benchmark-driven investors: invest through a fund that tracks or closely follows a flagship benchmark index (usually the J.P. Morgan Government Bond Index-Emerging Markets (GBI-EM)).
  - Unconstrained investors: can invest without being restricted by index considerations.
- The benchmark-driven category can include “passive,” “closet-index,” or “weakly active” funds, and can represent retail or institutional investors depending on mandates.

### Key findings on flows and investor behavior
- Total foreign holdings of EM local-currency government bonds:
  - $200 billion at end-2007.
  - $600 billion by end-2014.
- Foreign holdings rose similarly across countries in the J.P. Morgan GBI-EM universe, with foreign share increasing steadily from 2007 until the May 2013 taper tantrum.
- Three distinct phases of foreign flows (Arslanalp and Tsuda, 2014):
  - Pre-crisis: 2006–07 (moderate differentiation among countries).
  - Crisis: 2008–09 (sharpened differentiation).
  - Post-crisis: 2010–12 (foreign flows positive almost everywhere).
- Drivers of post-crisis inflows include:
  - Improving EM fundamentals; several EMs reached or regained investment grade during 2010–12 (Colombia, Indonesia, Latvia, Romania, Uruguay).
  - Rising popularity of index-funds and benchmark use (Raddatz et al. 2015), contributing to synchronized portfolio flows across diverse countries.
- Evidence on benchmark influence:
  - Raddatz et al. (2015): benchmarks explain, on average, around 70 percent of country allocations after controlling for industry, macroeconomic, and country-specific effects.
  - Miyajima and Shim (2014): among active EM local-currency bond funds tracked by EPFR Global, median “active share” is 17 percent, implying more than 80 percent overlap in country weights between fund portfolios and benchmarks; nearly 70 percent of actively managed EM bond funds are “closet index” or “weakly active.”

### Size and timing of benchmark-driven investors (aggregate estimates)
- Combining surveys, event studies, and empirical estimates, the paper finds:
  - Benchmark-driven investors represented a significant portion of EM local bond investors at end-2014: $200–$250 billion.
  - This corresponds to more than a third of total foreign holdings of $600 billion.
  - The prominence of these investors grew rapidly until it reached a peak before the May 2013 taper tantrum.
- Conceptual implications:
  - Benchmark-driven investors can be a stable funding source because index-tracking funds must allocate to countries included in the index.
  - They can also introduce correlated behavior among asset managers, increasing sensitivity of countries to global risk sentiment and correlated portfolio flows independent of country fundamentals.

### Estimating benchmark use — surveys and empirical approaches
- Two broad approaches discussed for estimating benchmark-driven investor pools:
  - Event studies.
  - Empirical estimation.
- J.P. Morgan survey results (used to gauge assets under management tracking GBI-EM):
  - As of end-2014, around $221 billion of assets were managed against the GBI-EM suite of indices.
  - $195 billion were managed against the GBI-EM Global Diversified index alone.
  - Assets tracked by the GBI-EM indices increased tenfold from $21 billion at end-2007 to $221 billion at end-2014.
  - J.P. Morgan (2014) notes double-digit growth rates every year during 2007–2013, with the slowing pace of growth starting from 2013 likely explained by outflows from EM-dedicated funds during the May 2013 taper tantrum.

*Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15263.pdf*

### Box 1. Emerging Market Bond Indices

### Box 1. Emerging Market Bond Indices

### Major EM bond indices
- The J.P. Morgan Government Bond Index–Emerging Markets (GBI-EM) index was launched in 2005.
  - There are three versions: GBI-EM Broad, GBI-EM Global, and GBI-EM; each version has a diversified overlay.
  - The diversified version places a 10 percent cap for each country.
  - According to J.P. Morgan, the GBI-EM Global Diversified is the most popular among the six versions—accounting for around 90 percent of all AUM benchmarked to the GB-EM suite of indices as of end-2014.
  - Main entry requirement: market accessibility. No minimum-rating requirements or explicit market size limits.
  - Treasury bills and inflation-indexed bonds are not eligible (only fixed-rate nominal bonds).
  - As of end-2014, 16 countries were included in the GBI-EM Global index: Brazil, Chile, Colombia, Hungary, Indonesia, Malaysia, Mexico, Nigeria, Peru, Philippines, Poland, Romania, Russia, South Africa, Thailand, and Turkey.

- The Barclays Emerging Markets Local Currency Government Index was launched in 2008.
  - Main entry requirements: a minimum market capitalization of $5 billion and market accessibility.
  - No minimum-rating requirements.
  - Nominal bonds and bills are eligible; inflation-indexed bonds are not eligible.
  - The index overlaps largely with the J.P. Morgan GBI-EM Global index, covering the same countries and (since March 2013) three more: the Czech Republic, Israel, and Korea.
  - Miyajima and Shin (2014) show that comparable EM bond indices provided by J.P. Morgan and Barclays have more than an 80 percent overlap in country weights.

- The Citi Emerging Markets Government Bond Index (EMGBI) was launched in 2013.
  - Main entry requirements: a minimum market capitalization of US$10 billion and market accessibility.
  - No minimum rating requirements.
  - Treasury bills and inflation-indexed bonds are not eligible.
  - As of end-2014, the index includes all the GBI-EM Global countries, except for Nigeria and Romania.

- There are also regional EM bond indices focused primarily on Asia, such as the HSBC Asian Local Bond Index (ALBI) and the Markit iBoxx Asian Bond Fund (ABF) Index.

### Major global bond indices
- The Barclays Global Aggregate Index (Global AGG)
  - Tracks fixed-rate investment-grade bonds of both developed and emerging markets.
  - Market capitalization of $43 trillion as of end-2014.
  - Created in 1992; historical data available from January 1987.
  - Treasury sector tracks central government bonds issued by 37 countries in 24 currency markets, representing a total market capitalization of $23 trillion at end-2014.
  - Main entry requirements: countries must have investment-grade status based on the middle rating of Fitch, Moody’s, and S&P (or the lower rating when not all ratings are available), and “a freely traded, convertible currency with a liquid forward market that allows investors to hedge their currency exposure.”
  - As of end-2014, eight EMs have sovereign bonds included in the Barclays Aggregate index (country weights shown in parentheses): Chile (0.01 percent), Malaysia (0.2 percent), Mexico (0.4 percent), Poland (0.2 percent), Russia (0.1 percent), South Africa (0.2 percent), Thailand (0.2 percent), and Turkey (0.2 percent).

- The Citibank World Government Bond Index (Citi WGBI)
  - Tracks fixed-rate investment-grade sovereign bonds of both developed and emerging markets.
  - Market capitalization of $20 trillion as of end-2014.
  - Created in 1986; historical data available from December 1984.
  - Main entry requirements: minimum market capitalization of $50 billion, a domestic long-term credit rating of A-/A3 by S&P/Moody's, and no barriers to entry as reflected in policies that “actively encourage foreign investor participation.”
  - As of end-2014, four EMs are included in the WGBI (country weights shown in parentheses): Malaysia (0.4 percent), Mexico (0.8 percent), Poland (0.5 percent), and South Africa (0.4 percent).

- Weight of EMs in major global bond indices remains small: less than 1 percent at the country level and around 2 percent at the aggregate level (among global bond indices).

### Event studies: estimating benchmark-driven investor base (2014)
- Overall result from three event studies: the pool of benchmark-driven investors falls within the range of $170–$270 billion as of 2014 (or around $227 billion if taking the average of the three studies), broadly in line with the results of the J.P. Morgan survey ($221 billion).
- Identification approach:
  - Examine episodes where a country’s weight in the J.P. Morgan GBI-EM index rises sharply due to an increase in the nominal stock of its “index-eligible” securities (security inclusion events for countries already in the index).
  - Treat such inclusions as exogenous supply shocks that should trigger automatic demand by benchmark-driven investors but not by unconstrained investors.
  - Separate change in country weight (w_it) into:
    - Buy-and-hold component: w_it (R_it/R_bt) – w_it (valuation effects from returns R_it and R_bt).
    - Exogenous component: captures changes in the stock of index-eligible debt.
  - Only the exogenous component should trigger capital flows by benchmark-driven investors.
- Estimation equation components:
  - B_t = benchmark-driven investor base at time t in U.S. dollars.
  - f_it = net foreign purchase of country i's bonds between time t and t+1 in U.S. dollars.
  - w_it = weight of country i in the benchmark at time t (J.P. Morgan GBI-EM Global Diversified used as representative benchmark).
  - R_it and R_bt = total gross returns of country i's bonds and the benchmark, respectively, from t to t+1.
- Three identified 2014 episodes used for estimation: Colombia (April–September 2014), Peru (November–December 2014), and Romania (April–May 2014).
- Note on valuation effects: Example—Russia’s weight in the GBI-EM (Global Diversified) fell from 10 percent at end-March 2014 to 4.7 percent at end-March 2015, mainly due to the sharp depreciation of the ruble.

#### 1. Colombia (March–September 2014)
- J.P. Morgan announced in March 2014 the inclusion of five additional Colombian bonds (maturing in 2016, 2028, 2022, 2024, and 2028) into the GBI-EM Global Diversified index in a phased manner until September 2014.
- Inclusion effects:
  - Increased Colombia’s index-eligible debt stock (in nominal terms) by 150 percent.
  - Raised Colombia’s weight in the index by 4.45 percentage points from 3.24 percent at end-February to 7.69 percent at end-September.
  - Buy-and-hold component: 0.10 percentage points.
  - Exogenous component (a): 4.35 percentage points.
- Observed net foreign purchases during March–September 2014 (b): $7.36 billion.
- Estimated benchmark-driven investor base (c = b / a): 169 (billion U.S. dollars).
- Conclusion: Net foreign purchases of $7.36 billion imply a benchmark-driven investor base of around $170 billion at the time of the event.

#### 2. Peru (November–December 2014)
- Inclusion: a new Peruvian government bond (Peru 5.7 percent August 2020) of 7.4 billion nuevo soles ($2.5 billion) was included in the index at end-November 2014 (announcement earlier that month).
- Inclusion effects:
  - Increased Peru’s index-eligible debt stock (in nominal terms) by 17 percent.
  - Raised Peru’s weight in the J.P. Morgan index by 0.39 percentage points from 1.47 percent at end-October to 1.86 percent at end-December.
  - Buy-and-hold component: 0.09 percentage points.
  - Exogenous component (a): 0.30 percentage points.
- Observed net foreign purchases during November–December 2014 (b): $0.72 billion.
- Estimated benchmark-driven investor base (c = b / a): 238 (billion U.S. dollars).
- Conclusion: Net foreign purchases of $0.72 billion imply a benchmark-driven investor base of around $240 billion at the time of the event.

#### 3. Romania (April–May 2014)
- Inclusion: two new Romanian bonds (Romania 5.6 percent November 2018, and Romania 5.95 percent June 2021) were included in the index at end-April and end-May 2014.
- Inclusion effects:
  - Increased Romania’s index-eligible debt stock (in nominal terms) by 45 percent (11.5 billion lei, $3.5 billion).
  - Raised Romania’s weight in the J.P. Morgan index by 0.56 percentage points from 1.47 percent at end-March to 2.03 percent at end-June.
- (Event-study calculations for Romania continue beyond the provided excerpt.)

*Source: IMF staff compilation from the content unit titled "Box 1. Emerging Market Bond Indices" in the supplied PDF.*

### 0.53 percentage points of the rise in the country weight was exogenous (i.e. due to the inclusion

### _wp15263 - 0.53 percentage points of the rise in the country weight was exogenous (i.e. due to the inclusion

### Romania event study: index inclusion and foreign flows
- Change in Romania country weight in J.P. Morgan GBI-EM Global Diversified:
  - Country weight, end-Mar 2014: 1.47 percent
  - Country weight, end-Jun 2014: 2.03 percent
  - Change in country weight from Mar to Jun 2014: 0.56 percent
    - Buy-and-hold component: 0.03 percent
    - Exogenous component (a): 0.53 percent
- Net foreign purchases during Apr–Jun 2014 (b): $1.44 billion
- Estimated benchmark-driven investor base (c = b / a): 273 (billion U.S. dollars)
- Text summary: “0.53 percentage points of the rise in the country weight was exogenous (i.e. due to the inclusion of the new bonds), while the rest was due to valuation effects. To maintain the same position on Romania, a benchmark-driven investor would have to allocate more capital to the country beyond the valuation gains. Net foreign purchases of Romanian local government bonds were $1.44 billion during April-June, the highest level in 2014. Assuming unconstrained investors did not alter their positions due to the index change, this would imply a benchmark-driven investor base of around $270 billion at the time of the event (Table 4).”

### Empirical estimates: benchmark-driven vs. unconstrained investors
- Method:
  - Approach: Balston and Melin (2013) decomposition using monthly foreign holdings.
  - Two pools: benchmark-driven (restricted to index-eligible fixed-rate nominal bonds, 16 index countries, 10 percent country limit) and unconstrained (invests by market capitalization across a broader set of securities and countries).
  - Estimation via constrained least squares (CLS) with at + bt equaling total foreign holdings each month.
  - Sample period: January 2010 to June 2015.
  - Sample includes 18 countries (13 in the GBI-EM Global Diversified index, 5 off-benchmark: China, the Czech Republic, India, Israel, and Korea).
- Key empirical estimates:
  - Pool of benchmark-driven investors around $240 billion at end-2014.
  - J.P. Morgan survey: $221 billion.
  - Average estimate in event studies: $227 billion.
  - Peak benchmark-driven investor base: $300 billion just before the May 2013 taper tantrum.
  - Total foreign holdings (context): $600 billion (paper notes benchmark-driven investors represent more than one third of total foreign holdings).
- Data notes:
  - Total outstanding and foreign holdings from national data sources (Arslanalp and Tsuda, 2014).
  - For China and India, outstanding amounts reflect the “investable” portion given foreign investment quotas.
  - EPFR Global data show assets under management by EM local currency bond funds (mutual funds and ETFs); empirical estimates are somewhat higher because they include other vehicles (separate managed accounts, etc.).

### Cross-country patterns and dynamics
- Temporal and cross-country variation:
  - At peak (just before taper tantrum), benchmark-driven investors accounted for close to one-half of total foreign holdings in EMs.
  - By end-2014 benchmark-driven share fell; unconstrained investors gained relative importance afterwards.
  - For some countries (Colombia, Peru, Romania) benchmark-driven investors still represent the bulk of the foreign investor base as of end-2014.
- Possible explanations for country differences:
  - Supply-side: larger, more liquid government bond markets attract unconstrained investors (examples: Mexico, Brazil, Poland); smaller markets may attract investors mainly after index inclusion.
  - Demand-side: countries exposed to outflows from unconstrained investors due to country-specific shocks may have a higher presence of benchmark-driven investors.
- Empirical relationships:
  - Share of benchmark-driven investors is inversely related to market size (market capitalization).
  - Benchmark-driven outflows persisted for about one year after the May 2013 taper tantrum, suggesting correlated portfolio flows across benchmarked countries.

### Key implications (four points)
- First:
  - Benchmarks link countries in the same portfolio and can trigger correlated portfolio flows, connecting countries beyond fundamentals.
  - Benchmark-driven investor outflows after May 2013 lasted about one year.
  - Policy implication: liquidity buffers can be useful even for countries with strong fundamentals, especially where benchmark-driven investors have a large presence.
- Second:
  - As the share of benchmark-driven investors has fallen since the taper tantrum, it is increasingly important for EMs to maintain strong domestic policies.
  - Unconstrained investors have greater latitude to switch across countries based on risk-return, increasing the importance of domestic fundamentals.
- Third:
  - Major EM local currency bond indices (including the GBI-EM) do not require a minimum credit rating for inclusion.
  - Ratings-sensitive investors may be limited within the benchmark-driven base; unconstrained investors could be more sensitive to credit-rating changes.
  - Policy implication: safeguarding investment grade status may be important, especially for countries with a large unconstrained investor base.
- Fourth:
  - For countries not yet included in EM local currency bond indices, the estimate of the benchmark-driven investor base provides a basis to assess potential capital inflows upon index inclusion.

### Conclusion
- Convergent evidence across methods:
  - Various approaches suggest about $200–$250 billion of foreign investments in EM local currency government bonds were benchmark-driven as of end-2014, representing more than one third of total foreign holdings ($600 billion).
  - Benchmark-driven prominence grew until a peak before the May 2013 taper tantrum (nearly one-half of total foreign holdings).
- Behavioral insight:
  - The taper tantrum marked a shift: exchange rate volatility became more important for benchmark-driven investors after May 2013, prompting a move from broad asset allocation to country selection.
  - EM investors have become more differentiating post-taper tantrum.
- Ongoing role:
  - Despite a decline from peak, benchmark-driven investors remain large enough to play an important role in driving portfolio flows to emerging markets, particularly in countries where they have a large presence.

*Source: Authors' calculations and empirical results as presented in the supplied IMF working paper excerpt.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15263.pdf_