Title: InterAmerican Development Bank
1Inter-American Development Bank
FINANCE DEPARTMENT
Treasury Risk Management and Middle Office Unit
Review of an Approach for Analyzing Latin
AmericanFixed Income Securities
LAC Debt Group Workshop
Washington, DC
October 26-27, 2006
2Agenda
- Background of the Project Study
- Factors Contributing to an Attractive Investment
Climate in Latin America - Historical Performance of Latin American Debt
Investments - A Modified Framework for Relative Value and Yield
Curve Analysis - Summary of Findings
3Background of the Project Study
PROJECT OBJECTIVES AND DELIVERABLES
- In June 2006, IDB Senior Management requested a
study to review a framework for analyzing Latin
American debt securities. - The review had two objectives
- to evaluate models for relative value and
portfolio analysis and determine their potential
application to capital market and portfolio
management operations of the Bank and, - to apply these models in an analysis of the risks
and returns of Latin American debt securities. - Three analytical tools were developed to carry
out the research study - Portfolio Optimization Tool for efficient
frontier analysis - Yield Decomposition Model for relative value
analysis and yield curve development - Technical Model for facilitating Foreign Exchange
(FX) forecasts
4Factors Contributing to an Attractive Investment
Climate in Latin America
STRONG ECONOMIC FUNDAMENTALS IN LATIN AMERICA
- A confluence of positive factors has led to an
attractive investment climate in Latin America. - Real GDP growth in the region is expected to
continue at a robust 4 to 5, driven by high
commodity prices, increased exports, and the
global economic expansion. - Current account surpluses and foreign direct
investment have contributed to rising
international reserves, supporting a
strengthening of regional exchange rates versus
the US dollar. - Strong economic growth has been coupled with
lower inflation, and a decline in interest rates.
5Factors Contributing to an Attractive Investment
Climate in Latin America
AN IMPROVED ENVIRONMENT FOR INVESTMENT
- Although the region may be vulnerable to a
slowdown in world economic growth and lower
commodity prices, recent policy reforms have
improved the resilience of country economies. - Many countries have reduced budget deficits and
public debt relative to GDP, and have adopted
flexible exchange rates to limit vulnerability to
external events. - Some countries such as Brazil, Chile, Colombia,
Mexico, and Peru have introduced inflation
targeting, reduced exposure to external debt, and
increased debt issuance in local currencies. - To further develop capital markets, measures such
as lowering withholding taxes, eliminating
capital controls, and issuing local currency debt
with greater transparency, were adopted. - Recent structural reforms, sound macroeconomic
policies, and improved fiscal management, if
sustained, provide a favorable environment for
long-term growth and investment in the region.
6Historical Performance of Debt Investments in
Latin America
HIGH RETURNS AND VOLATILITY
- Despite high volatility from a series of
financial and economic crises over the past 9
years, average returns for Latin American debt
had outpaced those of traditional asset classes. - Strong fundamentals over the past 3 years had led
to impressive gains in the bonds of individual
countries, contributing to higher long-term
average annualized returns.
7Historical Performance of Debt Investments in
Latin America
LOW CORRELATIONS
- Low correlations between Latin American bond
indices and traditional asset classes highlight
the potential diversification benefits of adding
Latin bonds to a typical global investment
portfolio. - Latin American local currency versus
USD-denominated debt also exhibited low
correlations.
8Historical Performance of Debt Investments in
Latin America
HISTORICAL EFFICIENT FRONTIER ANALYSIS
- A series of portfolio optimization experiments
were run using data from 06/1997-06/2006 to
determine if the addition of Latin American bonds
would have added value to various global
investment portfolios. - Five portfolios or funds were created and
analyzed using the indices shown below.
9Historical Performance of Debt Investments in
Latin America
GLOBAL INVESTMENT FUND (GIF)
- If Latin bonds were added to the GIF and held
over the past 9 years, higher returns per unit of
risk would have been realized, and the GIF Base
Efficient Frontier would have expanded
moderately. - Allocating just 1.5 to Latin bonds would have
increased returns by 0.14, accompanied by a
negligible increase in volatility of 0.02 the
Sharpe Ratio would have increased from 0.38 to
0.40. - As more Latin bonds were added, returns increased
more relative to risk, so that at the maximum
risk portfolio with 23 Latin bonds, a 2.05
increase in returns would have led to a 1.86
increase in risk.
10Historical Performance of Debt Investments in
Latin America
GLOBAL FIXED INCOME FUND (GFIF)
- Including 15 Latin bonds to the GFIF over the
past 9 years would have resulted in higher
returns and lower risk, and a much improved
efficient frontier. - Returns would have increased by 0.69, and risk
would have actually been lower by 0.34,
resulting in a significant increase in the Sharpe
Ratio from 0.52 to 0.70. - The slope of the efficient frontier would have
been steeper, and at the maximum risk portfolio
(30 Latin bonds), a 2.43 increase in returns
would have entailed only a 0.94 increase in
volatility.
11Historical Performance of Debt Investments in
Latin America
US FIXED INCOME FUND (UFIF)
- The addition of Latin bonds to the UFIF over the
past 9 years would have significantly expanded
the UFIF Base Efficient Frontier, resulting in
higher returns and lower risk at the minimum risk
portfolio. - Allocating 10 to Latin bonds would have
increased returns by 0.23, and lowered risk by
0.29, leading to an increase in the Sharpe Ratio
from 0.69 to 0.84. - Sharpe Ratios would have increased further
through the frontier up to a peak of 0.95, and
then declined to 0.90 at the maximum risk
portfolio.
12Historical Performance of Debt Investments in
Latin America
GLOBAL EMERGING MARKET BOND FUND (GEMF)
- A GEMF with Latin bonds held over the past 9
years would have had high returns with high
volatilities, and would have only been
appropriate for investors with high risk
tolerances. - The minimum risk portfolio, created by allocating
65 emerging market bonds, would have been a high
risk, high return portfolio with returns being
higher by 3.23 but with risk also going up by
4.04. - Sharpe Ratios would have declined throughout the
frontier, so that, at the maximum risk portfolio,
5.76 higher returns would have come at a cost of
8.83 volatility.
13Historical Performance of Debt Investments in
Latin America
LATIN AMERICAN EMERGING MARKET BOND FUND (LAEMF)
- A Latin bond fund operating over the past 9 years
would have had a high risk, high return profile
appropriate only for investors with high risk
tolerances. - Allocating 55 Latin bonds to the minimum risk
portfolio would have increased returns by 2.83,
while risk would have increased by 2.41 versus
the LAEMF Base portfolio. - Sharpe Ratios would have declined throughout the
frontier, although their relatively high levels
show that the performance of the LAEMF would have
surpassed that of the GEMF on a risk-adjusted
basis.
14Historical Performance of Debt Investments in
Latin America
IF 20/20 HINDSIGHT LOOKED GOOD, HOW WOULD ONE
PROCEED?
- Cautiously. Our review of the historical
performance of Latin American debt investments
only provides insights into what happened in the
past, not what may happen in the future. - Professional investors typically complement what
they have studied about the past with what they
are assuming about the future, through the use of
predictive models. - Key players in emerging markets develop
forward-looking models and invest in an
infrastructure of dedicated personnel, models,
and systems due to potential gains from their
use. - Typically, macro-level input is taken from teams
of chief investment officers, portfolio
strategists, and economists, and combined with
micro-level input from analysts and traders. - For instance, models to forecast expected returns
could take into account that credit spreads have
tightened considerably to historically low
levels, so further capital gains may be limited. - The impact of structural changes in the
macro-economic environment and how these may
affect assumptions on volatilities and
correlations may also be considered.
15Local Currency Yield Decomposition Framework
YIELD DECOMPOSITION FRAMEWORK
- Deutsche Bank (DB) gave us a glimpse of some of
the sophisticated relative value methodologies
and tools that they use for emerging market
research, including a methodology for decomposing
local yields. - We then developed a modified framework for
analyzing local currency yields, which includes
our own Technical Model for FX forecasting, and
two models which apply DBs methodologies. - The US Treasury Yield, US Libor Swap Spread, and
USD External Sovereign Debt Spread are derived
from the market.
Yield Decomposition Framework for Offshore Local
Bond and Swap Yields
16Local Currency Yield Decomposition Framework
USES OF THE YIELD DECOMPOSITION FRAMEWORK
- Relative Value Analysis - A Yield Decomposition
Model (YDM) that brings together market as well
as modeled components of yield could be used to
analyze relative value in local currency debt
instruments. - Expected FX depreciation, the largest component
of local yields, is modeled versus the market. - Our Technical Model for expected FX depreciation
translates forecasts from various sources into
their views on real exchange rates, and shows how
they compare with historical equilibrium levels. - Yield Curve Development - A framework for
analyzing yields would be one quantitative
approach for creating or extending local currency
yield curves in illiquid markets. - Modeling of FX depreciation allows valuation of
long-dated swaps, and facilitates extension of
the curve. - If the effect of Negative Sovereign Credit
Correction is properly modeled, an offshore
government curve can be derived from the offshore
swap curve. - The negative sovereign credit spread adjustment
accounts for lower loss given default for
positive correlations between the sovereign
credit spread and FX rates.
17Local Currency Yield Decomposition Framework
EXCHANGE RATE PROCESS
- The exchange rate process is modeled via the real
exchange rate (RER). - RER scenarios are converted into nominal FX
scenarios by incorporating inflation projections
(i.e, from inflation-linked bonds). - The following equations define the relationships
between RER, nominal FX, and inflation
18Local Currency Yield Decomposition Framework
IDB MODEL OF REAL EXCHANGE RATE DYNAMICS
- We modeled the dynamics of the RER process,
capturing the stationary, cyclical nature of RER
movements, and the effect of its long-term
equilibrium, through the following mean-reverting
process
where X cumulative Log(RER) from a base date of
January 1999, X8 is the long-term equilibrium
RER, b the speed of mean reversion, Zt is a
deterministic function for short-term effects,
and s volatility in RER returns.
- After a few transformations, the solution comes
to
- The solution to the RER stochastic differential
equation allows X to be derived analytically. The
models parameters are calibrated to an RER
forecast, which in turn is derived from a given
FX forecast.
19Local Currency Yield Decomposition Framework
RELATIVE VALUE ANALYSIS
- The following example compares BRL/USD FX
forecasts from the market (NDFs) versus
Commerzbank.
- Less expected depreciation from Commerzbank
implies 236 to 580 bps of relative value. - Our RER model translates Commerzbanks forecast
into an expected change in RERs from the current
base level of -0.25, to short term and long-term
equilibrium RERs of -0.32 and -0.02,
respectively. - The NDF market projects short-term and long-term
equilibrium RERs to be -0.08 and 0.08,
respectively.
20Local Currency Yield Decomposition Framework
EXAMPLE OF SWAP CURVE EXTENSION
- To model an extension of the BRL Swap Curve
beyond the 5-year point, we begin by projecting
RER to its long-term equilibrium point of 0.08
implied by the market. - The implied FX yield component is derived from
RERs and market inflation differentials as shown
below.
- Market risk premium is the other yield component
that has to be modeled beyond the 5-year point,
and we applied DBs methodology to calculate this
premium and obtain the full swap curve.
21Local Currency Yield Decomposition Framework
EXAMPLE OF OFFSHORE GOVERNMENT CURVE
- To build an offshore government bond curve for
Brazil, we first derive its market yield
components a) US Treasuries and b) external debt
spreads from credit default swaps. - Expected FX depreciation is derived as explained
in the previous example for the offshore swap
curve. - Market risk premium is implied from the other
components, and derived from the 10-year point,
which was the market yield on 6/30/06. - The negative sovereign credit spread adjustment
is modeled using DBs methodology.
22Local Currency Yield Decomposition Framework
SCENARIO ANALYSIS
- Running various scenarios could provide clues on
what may be the sources of discrepancies in
pricing between the offshore swap market and
offshore government bond market (i.e., liquidity
conditions, supply/demand). - It may be worthwhile studying changes in yield
component differentials between the two markets
over time.
23Summary of Key Findings
FINDINGS ON THE RESEARCH STUDY
- After a comprehensive analysis, we may conclude
that Latin American debt securities warrant
further study as a potential value-adding asset
class in global investment portfolios. - Strong economic fundamentals, continued
structural reforms and sound monetary and fiscal
policies in Latin America have created a
favorable climate for investment. - Although considered a high risk, high return
asset, low correlations of Latin bonds with
traditional asset classes provide diversification
benefits that may improve the risk-return profile
of portfolios. - We have demonstrated that, had Latin American
bonds been part of portfolios with various
investment strategies over the past 9 years, the
portfolio efficient frontiers would have
improved. - These results are based on historical data and
only provide insights into the value potential of
Latin bonds. Forward-looking relative value
models will be required to complement the
analysis.
24Summary of Key Findings
FINDINGS ON THE REVIEW OF METHODOLOGIES AND MODELS
- Regarding our review of relative value and
portfolio methodologies and models, we have
identified a number of applications that are
within the scope of current operations. - The Portfolio Optimization Tool has its obvious
uses in analyses required for establishing
treasury risk policies, and for the prudent
management of risk in the Banks investment
portfolio. - The YDM and the Technical FX forecasting models
could be applied in marking-to-model local
currency financing structures with illiquid or
incomplete market yield curves. - These models will also be used in developing a
methodology for managing model risk in illiquid
markets, and in enhancing the Derivative Credit
Risk Management (DCRM) system. - Beyond these applications, state-of-the-art
relative value and portfolio models could support
an initiative to develop Latin American capital
markets through a Fund similar to the Asian Bond
Fund. - The launch of a similar initiative for Latin
America would require best practice emerging
market research and investment management
functions.
25Conditions and Disclaimer This information has
been prepared solely for informational purposes
and is not an offer to buy or sell or a
solicitation of an offer to buy or sell any
security or instrument or to participate in any
trading strategy. No representation or warranty
can be given with respect to the accuracy,
timeliness or completeness of the information, or
that any future offer of securities will conform
to any potential terms mentioned in this
information. Because this information is
provided solely for general informational
purposes and may or may not, at the time of its
receipt, remain accurate, timely and complete,
the IADB disclaims any and all liability relating
to this information, including without limitation
any express or implied representations or
warranties for, statements contained in, and
omissions from, this information. This document
contains or makes reference to various dates, but
makes no representation for subsequent events.
Past performance is not necessarily indicative of
future results.