The Effect of Foreign Currency Denominated Sovereign
Debt on the Value at Risk of Debt Sustainability
A Stochastic Approach to Emerging Market Economies
Name: Joeri Schouten
Student number: 6163319
Date: 04-01-2015
MSc Economics, specialization Monetary Policy & Banking
Faculty of Economics and Business, University of Amsterdam
First supervisor: Prof. dr. C. van Ewijk
2 Abstract
This study contributes to existing literature by quantifying the sovereign debt sustainability risks associated with foreign currency denominated public debt in Emerging Market economies, using a stochastic Value at Risk (VaR) approach. The results show that for all countries in the sample foreign currency denominated sovereign debt leads to an increase in the VaR of debt sustainability, when comparing it to a scenario for which a country has no foreign currency debt. However, the size of the increase in the Value at Risk differs
substantially across countries. A decomposition of the increase in the VaR into an exchange rate and an interest rate effect, shows that foreign currency public debt has a marginal stabilizing effect via the interest rate channel but a substantial destabilizing effect via the exchange rate channel.
I would like to thank Casper van Ewijk (University of Amsterdam) for providing generous overall feedback, Jasper Lukkezen (CPB) for providing me feedback on the technical modelling and the structure of the Thesis and Hugo Rojas-Romagosa (CPB) for providing feedback on the context and relevance of the thesis.
3 1 Introduction
Foreign currency denominated sovereign debt exposes sovereign debt sustainability to risks of (unexpected) negative exchange rate shocks, resulting in an exogenous debt increase (Debrun, Celasun and Ostry, 2008). If for example the domestic currency depreciates with 50 percent, the domestic currency value of the share of debt denominated in foreign currency doubles
.
Even though there is a lot of literature available explaining what type of risks are associated with issuing foreign currency public debt, there is no literature that relates the size of the share of foreign currency denominated sovereign debt to the Value at Risk (VaR) of public debt sustainability.The relevance of foreign currency denominated debt was first addressed by
Eichengreen and Hausmann (1999). They found that external borrowing in foreign currency was a major reason for the severity of the East Asian and Latin American crises of the 1990s. These findings are supported by Reinhard (2002). She finds that currency composition of sovereign debt is especially relevant in Emerging Markets, since a vast majority of 84 percent of defaults on sovereign debt payments in Emerging Market
economies is associated with a sharp exchange rate devaluation. The main reason that a large part of Emerging Market economy’s debt is denominated in foreign currency is that many countries have adopted either partial or full dollarization of their domestic currency, especially in Latin America (Szpunar and Głogowski, 2012). In case of full dollarization, total government debt is denominated in USD which eliminates exchange rate risk. In case of partial dollarization however, the share of domestic government debt denominated in foreign currency is exposed to exchange rate risk.
Even though exchange rate depreciations have a destabilizing effect on debt
sustainability, literature also provides evidence that foreign currency public debt can have a stabilizing effect on debt sustainability via the interest rate channel. This is because both the real value and the volatility of real interest payments on public debt denominated in foreign currency are expected to be lower than on domestic currency denominated debt in Emerging Market economies (Eichengreen and Hausmann, 2010) (Dell’Ariccia, Laeven and Marquez, 2010) (Szpunar and Głogowski, 2012). This is likely to be explained by the fact that if foreign investors invest in foreign currency denominated bonds of Emerging Market economies which are subjected to a higher inflation than the domestic inflation of the country of the investor. Therefore the investors avoid the time inconsistency problem (Calvo, 2001) and demand a lower nominal interest rate. So there is a clear trade-off when a government decides to issue foreign currency debt: debt sustainability benefits from lower interest rate (volatility) but suffers from exchange rate depreciation risk.
4 A common way to assess the degree of risk exposure faced by future public debt paths is the stochastic Value at Risk approach, also known as the “fan chart approach”, which was first introduced by the IMF (Debrun, Celasun and Ostry, 2006), later extended by Budina and Van Wijnbergen (2008) and recently applied to nine OECD countries by
Lukkezen and Rojas-Romagosa (2014). All papers show how fan charts and probabilistic VaR analysis can be used to guide policymakers in making assessments of the risks associated with future debt sustainability. Lukkezen and Rojas-Romagosa (2014) define a Sustainability Early Warning Indicator (SEI), which measures the difference between the projected median (50 percentile) scenario and the most risky (97,5 percentile) scenario. If for example the SEI is 20%, this indicates that the future debt path is at risk to deviate 20% from its base projection. The SEI I use as the measure of the Value at Risk of debt sustainability. I however extend the model by Lukkezen and Rojas-Romagosa (2014) with a foreign currency debt component and I apply the model to the Emerging Market economies India, Indonesia, the Philippines, Mexico and Peru. In order assess the effect of foreign currency denominated debt on the SEI, I simulate the SEI for the different countries including and excluding foreign currency debt. In this way it can be assessed to what extend foreign currency debt has a stabilizing or destabilizing effect on the SEI. Furthermore, I perform a scenario analysis to estimate the trade-off between the exchange rate effect and foreign currency debt interest rate effect separately.
A country can only face exchange rate depreciations if it is involved in a floating exchange rate regime (or in a managed float with no explicit target), or if it faces a break in an exchange rate peg break (Calvo and Mishkin, 2003). Using a stochastic model the real exchange rate is subjected to shocks which are normally distributed. This assumption makes that the model is not suited for analyzing exchange rate shocks which are caused by a sudden stop in capital inflows or the break of an exchange rate peg (Calvo, Izquierdo and Talvi, 2003). Both events are typically associated with substantial real exchange rate appreciations in a very short period of time, which are likely to be too large to be simulated by the model. Both cases could therefore result in an unsustainable debt path much quicker than the stochastic model would predict.
The results show that for all the countries in the sample, foreign currency
denominated public debt leads to an increase in the Value at Risk of debt sustainability, when comparing it to a scenario in which a country has no foreign currency debt. However, the size of the increase in the Value at Risk differs substantially across countries.
Furthermore I find that foreign currency public debt has a stabilizing effect via the interest rate channel, and a destabilizing effect via the exchange rate appreciation channel.
The structure of this thesis will be as follows. First I will give an overview of existing literature, explaining the reasons why governments issue foreign debt, what type of risk to
5 debt sustainability are associated with doing so and why this is especially relevant for
Emerging Market economies. Then the methodology behind the stochastic VaR approach and scenario analysis will be explained. Next an explanation for the country selection is given and a description of the data of key variables used for the fan chart analysis will be provided. After that the results from the simulations and related risk indicators will be outlined. Finally, a conclusion is formulated and suggestions for future research are given.
2 Literature review
To explain the context in which this study takes place, I first outline the concept and relevance of debt sustainability. Then I summarize the reasons why sovereigns would engage in issuing foreign currency denominated debt. Finally I explain how foreign currency denominated debt can expose sovereign debt sustainability to risks, and how this led to severe economic crisis in Emerging Market economies in the past.
2.1 Debt sustainability
An economy is said to have achieved fiscal sustainability when the ratio of public sector debt to GDP is stationary (i.e. constant over time), and consistent with the overall demand for government securities in both domestic and foreign markets (Bohn, 1998). Whether a country has achieved fiscal sustainability, depend on the future values of
economic growth, interest rates (both on foreign and domestic currency debt), exchange rate changes and the primary government deficit. The primary deficit is the government deficit including seigniorage income but without interest payments (Park, 2012). Seigniorage income are revenues from issuing money and form a substantial source of government income especially in Emerging Market economies (Lange and Sauer, 2005) (Budina and Van Wijnbergen, 2008).
In the public policy debate, a high and growing Debt/GDP ratio is widely viewed as worrisome (Bohn, 2005). Belhocine and Dell’Erba (2013) give an explanation for this phenomenon by showing that when a country’s Debt/GDP ratio increases continuously, the country is at risk to face increasing interest rate spreads. Belhocine and Dell’Erba (2013) estimate the relation between the level of the primary surplus at which debt sustainability is achieved, and the actual primary surplus. Their results show that there is clear relation between the primary surplus (at which debt sustainability is achieved) and Emerging Market bond spreads, with an elasticity of about 25 basis points for each one percentage point departure of the primary balance from its debt stabilizing level. Eventually, the path of debt accumulation might lead to debt overhang problems, reducing domestic investments and economic growth substantially (Yue, 2009). As the interest burden keeps increasing, at some
6 point it may exceed the cost of borrowing. In this case, it will be no longer in the best interest of the sovereign in case to fulfill its debt payments which might lead to a government
deciding to default on its payments (Calvo, 1988).
2.2 Foreign currency denominated debt and original sin
The most common reason why certain governments engage in issuing foreign currency denominated debt, is when they are simply not able to attract foreign investors willing to invest in bonds denominated in domestic currency, and the demand of domestic investors is not sufficient to meet total government lending requirements. This situation of financial markets incompleteness is referred to as original sin and was first addressed by Eichengreen, Hausmann and Panizza (2002). Original sin is a reason why a large number of Emerging Markets have at least 30 percent of the total outstanding debt denominated in foreign currency (Elliot et al., 2000). In case of Latin American countries public liabilities are usually denominated in US dollars, which is commonly referred to as liability dollarization (Hausmann, 2003). The literature does however not provide a consensus about the reasons behind original sin, because it cannot be fully explained by specific institutional and
macroeconomic factors. It does however highlight a path dependency in foreign currency which is primarily caused by past policy mistakes resulting in high inflation, sharp currency devaluations, and defaults on domestic currency debt (Hausmann, 2003). This dependency path was also addressed by Bordo, Meissner, Stuckler (2009), who show that foreign currency denominated debt is only a partial explanation of the risks of financial crisis.
High inflation resulting in sharp currency devaluations, form a large downside risk to the real value of the domestic investment, like government bonds, to foreign investors. When government bonds are denominated in foreign currency however, the exchange rate risk is migrated from the foreign investor to the government. This exchange rate risk elimination makes that there is more demand amongst foreign investors for foreign currency
denominated bonds terms (Chamon and Hausmann, 2000).
2.3 Debt sustainability risks associated with foreign currency denominated debt
The most substantial risk to which debt denominated in foreign currency is exposed is exchange rate risk. If a country’s debt is denominated in foreign currency, for example US dollars, its capacity to pay will be related not to the Debt/GDP value at constant local currency units, but to the Debt/GDP value in foreign currency terms (Chamon and
Hausmann, 2000). When a country thus faces severe exchange rate depreciations, the value of the share of public debt denominated in foreign currency will increase exogenously. This is
7 a serious threat to debt sustainability, because the increase in value of the share of foreign currency sovereign needs to be offset by an increase in the primary surplus. If a government is unable to offset the increase in value of the share of public debt denominated in foreign currency subsequently, it is at risk to face an unsustainable expected future debt path. Furthermore, countries which public debt is composed out of a large share of foreign currency are expected to face a higher exchange rate volatility (Chamon and Hausmann, 2002). The difference in exchange rate volatility between different debt structures was tested in later work by Chamon and Hausmann (2004). The results show that the volatility of
changes in real US dollar denominated GDP is almost three times higher than in domestic currency for developing countries. Exchange rate risk could however be overcome if a country is able to hedge its foreign currency risk against exchange rate volatility, assuming that the country can borrow abroad in its own currency (Eichengreen and Hausmann 1999). The problem is however that a country whose external liabilities are necessarily denominated in foreign exchange is by definition unable to hedge. This leads to a currency mismatch (domestic currency debt which has to be paid back in dollars) or maturity mismatch (long-term debt will be financed with short-(long-term loans) (Chamon and Hausmann, 2002).
Foreign currency denominated debt seems to have a stabilizing effect on debt sustainability via the interest rate channel. Eichengreen and Hausmann (2010) find that issuing foreign currency denominated debt reduces the volatility of total interest payments, which reduces the Value at Risk to which debt sustainability is exposed. Dell’Ariccia, Laeven and Marquez (2010) and Eichengreen and Hausmann (2010) furthermore conclude that liability dollarization (issuing public debt in dollars) typically results in a reduction of total interest payments. These findings are supported by the work of Szpunar and Głogowski (2012), who find that local currency interest rates in Central and Eastern European Countries were consistently higher than in developed economies during the mid-2000s.
The reduction in interest payments is explained by the difference in characteristics between domestic and foreign currency denominated debt. Domestic currency debt is subjected to domestic inflation, which is relatively high especially in Emerging Market economies that follow a policy of stabilizing the real value of their public debt by executing highly expansionary monetary policy. This phenomenon of inflating debt was characterized as debt liquidation by Reinhart and Sbrancia (2011). In case of a domestic investor investing in domestic currency denominated public debt, the real value of the investment is subjected to domestic inflation. High inflation would therefore deteriorate the real value of the
investment. In case of a foreign investor investing in domestic currency debt the real value of the investment is subjected to the inflation of the country of the investor, so the real value of this investment is not deteriorated by domestic inflation. However, expansionary monetary policy not only lead to an increase in inflation but also to a depreciation of the domestic
8 currency. This results in a lower value of the investment in foreign currency terms. When debt is however denominated in foreign currency, depreciating exchange rates, as a result from the inconsistency problem (Calvo, 2001), will be avoided. This might result in that investors from a low inflation country will demand a lower nominal interest rate than domestic investors. In case of a fully dollarized economy, the time inconsistency problem is completely avoided since then a government is not able to create inflation. In this case both the
domestic and foreign nominal value is subjected to foreign inflation. This disability of a government to liquidize its debt would be an incentive for a government to apply a more sustainable fiscal policy, since it has no longer the ability to reduce the real value its debt.
There is a substantial amount of literature that provides evidence that there is an relation between foreign currency denominated sovereign debt and economic growth. However, the literature is ambiguous about what this relation actually is. Dell’Ariccia, Laeven and Marquez (2010) show that liability dollarization has been associated with faster credit and economic growth. For a sample of several Eastern European countries they for example find that credit growth was the fastest in countries that had a larger share of credit
denominated in foreign currency. Furthermore, Eichengreen and Hausmann (1999) find that a largeshare of foreign currency sovereign debt attenuate the severity of business cycles, and increases the degree of financial stability systems in emerging markets in times of low exchange rate volatility. This is in line with reasoning that when a country its Value at Risk of debt sustainability is subjected to low exchange rate volatility, the total Value at Risk could decrease due to the benefit of a lower interest rate (volatility).
Calvo and Reinhart (2000) however address the negative effect of a large share of foreign currency denominated public debt on economic growth. They question how the differences between developed and emerging markets in access to international capital markets influence the outcomes of a currency crisis. Their results show that in the case of a sharp depreciation of an Emerging Market economy currency, the adjustments in the current and capital account are far more acute and abrupt than for developed economies. Hence, currency crises often become credit crises as sovereign credit ratings often collapse following the currency collapse and access to international credit is lost. This is also due to the volatility and co-movement of interest rates and exchange rates which makes debt service volatile and anti-cyclical, increasing drastically at the worst times, thus increasing the probability that a country will not be able to make good on its debt service commitments (Eichengreen and Hausmann 1999). That this might eventually lead to severe output crisis, is supported by recent work of Bordo, Meissner and Stuckler (2009). They find that financial crises are particularly driven by exposure to foreign currency, resulting in significant and permanent output losses.
9 The effect of foreign currency public debt on debt sustainability thus depends on whether a county faces (sequential) currency depreciations, the difference between domestic and foreign currency denominated debt interest rates. In case there is no currency crisis, foreign currency debt can have a positive effect on debt sustainability through lower interest rate (volatility). However, in case of a substantial currency depreciation a large share of public debt amplifies the currency crisis resulting in an exogenous increase in public debt and contraction in economic growth.
Eichengreen and Hausmann (1999) provide an overview of how foreign currency debt problems ware related to economic crisis in Emerging Market economies. They conclude that external borrowing in foreign currency was a major reason for the severity of financial crises, based on an analysis of the East Asian and Latin American crisis. In several South East Asian countries borrowers seemed to ignore or discount the possibility that the value of their foreign currency liabilities might increase due to a sharp depreciation. Repeatedly after 1990 in Mexico, Thailand, South Korea, Indonesia, Malaysia, Russia and Argentina governments failed to uphold (crawling) pegs. This led to severe balance sheet problems resulting in credit crunches, output losses and an unsustainable level of public debt (Bordo, Meissner and Stuckler, 2009).
3 Methodology
The fan chart analysis approach is based on four components. The first component is a standard public debt accounting framework. This framework is used to project the effect of a (fixed) regime with respect to four key economic variables (economic growth, domestic and foreign currency debt interest rates and exchange rate appreciation) into the future. The second component is the fiscal response function (FRF), which estimates the change of a governments’ primary surplus in response to a change in total Debt/GDP. The third is a is a Vector Auto Regression (VAR) model which is used to estimate the relation between the current value of the key variables and their lagged values. Finally, a Monte Carlo simulation is performed which simulates the possible future values of the key economic variables. These components are then combined to make a projected probability distribution of future debt paths.
3.1 Debt sustainability framework
For the basic debt sustainability framework I follow Bohn (1998), who was the first to describe the (stochastic) behavior of public debt and deficits. However, this framework does not distinguish between domestic and foreign currency debt. I will explain the basic
10 framework by Bohn, and use this framework to include a foreign currency debt component into the flow government budget constraint.
Equation (1) states that initial debt, , plus the discounted value of all noninterest expenditure, , should be equal to the discounted value of all public sector noninterest income, which consist of tax revenue . Equation (1) can be rewritten as the second line in equation (1): initial (net) debt should at most equal the discounted value of the primary or noninterest government surplus.
(1)
In order to get better insight in public debt dynamics over time, equation (1) can be written as a flow budget constraint:
(2)
or equivalently:
Equation (2) states that initial debt plus interest payments plus the primary deficit, equal the new level of debt. This can be rewritten as an expression for initial debt, as is done in the second line of equation (2). Here initial debt equals the discounted sum of the primary surplus plus at the end of period debt. This way of writing the flow budget constraint should also show how discounting takes care of interest payment¹. Furthermore, I make the assumption that seigniorage revenues are a part of the composition of the primary surplus. This is why seigniorage revenues are not included explicitly in equation (1). Substituting equation (2) repeatedly into itself, for T starting at 0, yields:
(3)
Equation (3) thus states that the discounted value of the future primary surpluses plus the discounted value of future debt is equal to the current debt level. In case of debt
sustainability, the current debt level is stationary.
To capture the long run attitude of a government towards debt sustainability, I estimate the fiscal response function (FRF). The FRF is a country specific behavioral
equation for the primary surplus-to-GDP ratio, which indicates how the government’s primary
=
11 surplus responds to debt accumulation given a structure of shocks occurring in the
background. The FRF forms a shortcut to modeling the dynamic properties of fiscal policymaking (Bohn, 1998).
(4)
where is the fiscal reaction coefficient. I choose estimate the model twice, once including only the first lag coefficient and one including both the first and second lag reaction
coefficient. The reason for this is that it could be that for Emerging Market economies, governments might take more time to adjusts its primary surplus as a response to an increase Debt/GDP ratio. The null hypothesis for (4) is that the fiscal reaction coefficient is positive. Since the primary surplus not only depends on the debt level but also on other (cyclical) factors a set of control variables, , is included in the model. The first control
variables in this model are a dummy variable indicating whether a country is at war or not, to control for military expenses. In case a country has not been involved in any war during the time horizon used the estimation, this variable is omitted. The second variable controls for business cycle fluctuations. The values for this variable are computed by subtracting the trend in real GDP, as estimated by the Hodrick–Prescott filter ( ), from the actual real GDP in each year. When a government adjusts its primary surplus in such a way that the discounted value of current and future income plus initial wealth should at least be equal to the discounted value of all current and future noninterest expenditure, this eventually leads to the government solvency condition which states that fiscal position does not lead to an future explosive debt path:
3.2 Projected probability distribution
Based on the framework by Bohn, I augment the flow budget constraint by including foreign currency debt component. The government flow budget constraint (equation 5) implies that current year debt is divided into three parts. The first part refers the domestic currency debt and is equal to last year debt multiplied by 1 plus the real interest rate on government debt in this year , minus the rate of real economic growth this year, , times the share of domestic currency denominated debt,
.
The second part refers to debt denominated in foreign currency. Total foreign currency debt is multiplied by 1 plus the interest rate on foreign currency debt, for which I use the real interest rate on newly issued external debt as a proxy, minus the real growth rate and minus the effective exchange rate appreciation , times the share of foreign currency debt, . Finally, the primary government surplus, is subtracted from the debt accumulation identity.12
(5)
In this flow budget constraint I assume that is a policy instrument which remains constant at its 2010 values during the projected period. This is a very important assumption since it implies that during the projected period the government is able to rebalance its debt, if for example the domestic currency value of the share of foreign currency debt increases disproportionally due to exchange rate deprecations. The real, adjusted for inflation, value of the variables in are used in the model since the real value of public debt is of interest. In the flow budget constraint, I choose to use the exchange rate appreciation variable specifically, instead of using exchange rate corrected values of and
.
The reason for this is that exchange rate change is the most important variable in the analysis so I want to state its role in the flow budget constraint very clearly. For domestic currency interest rates I use the total government interest payments and for foreign currency interest rates I use the interest rate on newly issued external debt.In order to determine the historical stochastic properties of the key economic variables and the estimates of , a Vector Auto Regression (VAR) model is estimated:
(6)
Where , is a vector of coefficients and is a well behaved error term: . In this model j = 2 is the number of lags, based on the Bayesian Information Criterion (BIC). The VAR model is tested for unit roots using an Augmented Dickey Fuller (ADF) test. The variance-covariance matrix of residuals characterizes the joint statistical properties of the contemporaneous, non-fiscal disturbances affecting debt dynamics. More specifically, for simulation a sequence of random factors , such that where and Z is such that ( denotes the Cholesky decomposition of the covariance matrix, ). Using the VAR I am able to generate forecasts of which are consistent with the Monte Carlo simulated shocks. These shocks are thus simulated by multiplying a random number (with mean = 0 and variance = 1), with the Cholesky decomposition the corresponding variance-covariance matrix. As shocks occur each period, the VAR produces joint dynamic responses of all elements in . Using this model enables me to make projections of future values of interest rates, growth and
exchange rates. These projections are made from 2011 until 2021 and are repeated 10.000 times. Then the different projected values are ordered ascending and the values at 2,5% (250) and 97,5% (9.750) are used for the respectively minimum and maximum values for the probability of the future debt path, based on the assumption that the future debt paths are
13 normally distributed. A predominant assumption which has a large implication on the
probability simulation is that historical volatility will be stationary during the projected period.
3.3 Risk measurement and scenario analysis
Lukkezen and Rojas-Romagosa (2014) use the fan chart approach to construct a
Sustainability Early Warning Indicator (SEI), which indicates whether a country is likely to be at risk to face a debt crisis in the near future. Their SEI measures the upside deviation from the base projection (median) in the probability distribution after ten years. The reason for choosing only the upside deviation as a measure of risk instead of the total, upside and downside, deviation is that the threat to government debt sustainability is only captured by the upside risk. In case of a budget constraint including a foreign currency debt component, the SEI is defined as follows:
(7)
where is the 97,5 percentile of the simulated probability distribution with a share of foreign currency debt equal to
.
The value ofis the 97,5% percentile of the probability distribution without foreign currency debt and serves as a benchmark.
The SEI without foreign currency debt is defined as follows:
(8)
Because in this case the share of foreign currency debt is zero (
), and no longer apply. For this study,
will primarily serve as a benchmark.
The difference between these indicators is defined as the Foreign Currency Debt Effect and measures the change in the SEI as a percentage Debt/GDP as a result of foreign currency debt:
(9)
If the FCDE is positive foreign currency debt has a destabilizing effect on debt sustainability, i.e. the Value at Risk of debt sustainability increases, and vice versa
The literature suggests that foreign currency debt has a stabilizing effect on the Value at Risk of debt sustainability via a lower foreign currency denominated debt interest rate (volatility), which I define as the interest rate effect. I simulate this effect by assuming that the real exchange rate appreciation is zero, which is the same as assuming a fixed exchange
14 rate regime. By eliminating the exchange rate effect, the interest rate effect is all that
remains. The difference between this scenario and the zero foreign currency debt benchmark I define as the Interest Rate Effect (IRE):
(10)
The IRE can be further decomposed by simulation the IRE using both actual foreign currency debt interest rate volatility and domestic currency debt interest volatility. In this case, in can be determined if foreign currency debt interest rates have a stabilizing effect on the VaR of debt sustainability because of its presumed lower volatility.
An real exchange rate which is volatile and likely to depreciate in the future will have a destabilizing effect on debt sustainability. I simulate this effect by assuming that the foreign currency real interest rate is equal to the domestic currency real interest rate. The difference between this scenario and the zero foreign currency debt benchmark I define as the
Exchange Rate Effect (ERE):
(11)
In this case, the interest rate effect is eliminated and the exchange rate effect is the only remaining effect. Whether foreign currency denominated public debt is associated with an increase in the Value at Risk of debt sustainability thus depends on whether a positive interest rate effect (due to lower foreign currency debt interest rates) offsets a negative exchange rate effect (due to exchange rate depreciations). A key assumption is that the different scenarios are exogenous, meaning that for the different scenarios the VAR estimates and simulated volatilities are identical.
There is however a reason why the effect of exchange rate shocks on the exchange rate risk indicator might be biased by performing this type of analysis. What is implicitly assumed but not specifically mentioned by Budina and Van Wijnbergen (2008), Debrun, Celasun and Ostry (2008) and Lukkezen and Rojas-Romagosa (2014), is that total public debt is exposed to interest rate, exchange rate of growth shocks on a yearly basis. This would imply that a government refinances its total debt on an yearly basis. This is however most likely not the case, since a large part of public debt is normally financed with maturities that exceed one year. So debt securities that have a maturity longer than one year, thus are only subjected to shocks in the period in which refinancing takes place and not in all the previous periods. Since data on the maturity structure of foreign currency denominated debt is however very limited, I decide to follow Budina and Van Wijnbergen (2008), Debrun, Celasun and Ostry (2008) and Lukkezen and Rojas-Romagosa (2014) and assume that total
15 public debt is refinanced on a yearly basis.
4 Data description
In this section I will explain what of data requirements are related to the stochastic Value at Risk approach. Furthermore I will motivate why I decided to use interest rates on external debt as a proxy for foreign currency debt interest rates and I will provide a country specific exchange rate risk vulnerability assessment.
4.1 Country selection
In order to perform a complete analysis there are several data requirements. First a country should have issued foreign currency public debt, and recent data on the share of a country’s foreign currency Debt/GDP ratio should be available. Then data on public Debt/ GDP and the primary surplus should be available for at least 40 years to estimate a country’s long run attitude towards debt sustainability via the fiscal response function (FRF). For estimating the FRF I use a unique dataset by Mauro et al., who gathered fiscal data for their paper by “A Modern History of Fiscal Prudence and Profligacy”. Finally, data on economic growth, interest rates, exchange rates need and inflation need to be available for the last 15 year in order to estimate the VAR coefficient. Next to that, the selected countries should have adopted a floating exchange rate regime or a managed float regime without an explicit target, based on the IMF exchange rate regime classification during the 1995-2010 period. The Emerging Market countries Mexico, Peru, Indonesia, India and the Philippines fit these requirements and are therefore part of the analysis.
4.2 Estimation period
For estimating the vector autoregressive model estimates I use the 1995-2010 period. Budina and Van Wijnbergen (2008) use 14 years of data and Lukkezen and
Rojas-Romagosa (2014) use 25 years of data. The most important assumption for projecting the VAR model into the future is that the historical dynamics of the period on which the VAR model is estimate prolongs in the future. Between 1985 and 1995 the countries in the sample have been faced with very high inflation rates, ranging from 23% to a period of hyperinflation in Peru eventually resulting in a 321% inflation rate. The very high inflation (volatility) has a severe impact on the real value of economic growth, interest rates and exchange rates. Furthermore, since 1995 all central banks of the countries in the sample have adopted inflation rate targeting which substantially reduced the inflation rates in the succeeding years
16 (figure 1) and are likely to remain in the future. Because of this, I choose the 1995-2010 period to estimate the VAR model parameters.
4.3 Domestic and foreign currency debt interest rates
A key assumption in equation (6) is that interest rates on domestic and foreign currency denominated debt are not equal. It is necessary to make this assumption because the literature provides evidence that interest rates on foreign currency denominated debt are lower than on domestic currency denominated debt. Probably because of the lack of long time series data on both foreign currency debt interest payments and the size of foreign currency denominated debt, Budina and Van Wijnbergen (2008) and Debrun, Celasun and Ostry (2008) used interest rates of 10 year US bills as a proxy. The choice of using US T-bills as a proxy only omits the fact that a US T-T-bills rate does not reflect the (default) risk premium associated with Emerging Economies borrowing conditions. In this case an Emerging Economy is likely to benefit from issuing foreign currency debt disproportionally, since the T-bill interest rate and volatility is likely to be substantially smaller than domestic interest rate. A better proxy could be the interest rate on external debt, since this interest rate is more likely to reflect country specific risk premium. Since the risk premium and the
volatility of external debt interest rates are smaller than for interest rates on domestic currency debt (Behandari et al., 1990), I assess to what extent the characteristics of the external debt interest rate fit characteristic of foreign currency debt interest rates. I do this by computing the average values of nominal and real interest rates the corresponding volatilities over the 1995-2010 period for both domestic and external interest rates, which is the time horizon used to estimate the VAR process.
0% 5% 10% 15% 20% 25% 30% 35% 40% 1996 1998 2000 2002 2004 2006 2008 2010
Philippenes Indonesia Mexico Peru India
Inflation
Figure 1: Inflation rate for the Philippines, Indonesia, India, Mexico and Peru during the 1995-2010 period (source: World Bank)
17 Based on the literature findings, foreign currency interest payments should be less volatile and nominal interest rates should be lower. Table 1 shows the values for the
domestic debt interest rate, table 2 shows the values for the external debt interest rate. When comparing the two tables, it can be concluded that for all countries except Peru the interest rate on external debt is lower than interest rates on domestic debt. Furthermore, the nominal interest rate of the US T-bills only exceeds the external debt interest rate of India, which indicated that for the rest of the countries the interest rate on external debt is in between the interest rate on domestic debt and the US Treasury interest rate. For volatility the findings are more ambiguous. For two countries the volatility on external debt is lower, for two
countries it is higher and for one country is it similar. Based on the finding that for three of the four countries external debt interest rates seem to include a risk premium over the risk free US Treasury interest rate, I decide use the interest rate on external debt as proxy for foreign the foreign currency denominated debt interest rate.
4.4 Dynamic efficiency
Intuitively, debt grows at the rate of interest. If GDP growth is above the rate of interest with probability one, the economy is dynamic inefficient and the intertemporal budget
constraint is irrelevant. But if there is a positive probability that future interest rates are above the growth rate, a zero primary surplus is not sustainable (Bohn, 1998). In order to assess historical dynamic efficiency for the five countries, I compute the average values of the growth rate, interest rate and exchange rate appreciation. The flow budget constraint consist of both a domestic and foreign currency part. The first row of table 2 shows the difference between the domestic currency debt interest rate and the growth rate, which Table 1: Domestic currency debt: average values for nominal, nominal interest rate, real interest rate and its volatility during the 1995-2010 period
Table 2: Foreign currency debt: average values for nominal, nominal interest rate, real interest rate and its volatility during the 1995-2010 period
18 assess dynamic efficiency for domestic currency debt. The third row shows the difference between the foreign currency debt interest rate, the growth rate and the exchange rate appreciation. Both domestic and foreign currency debt is dynamic efficient since all
differentials are positive. In case of foreign currency debt, the dynamical efficiency is mainly a result of substantial exchange rate depreciations (negative appreciations). What table 3 also shows is that due to the exchange rate appreciation, foreign currency debt increases the need for governments to run a primary surplus in order to avoid an explosive debt path. This is especially the case of Indonesia and Mexico.
4.5 Exchange rate risk exposure
High exchange rate volatility and a large share of foreign currency denominated debt expose debt sustainability of the share of foreign currency debt, and therefore total debt sustainability, to large exchange rate volatility risk. Based on an analysis of the size of the share of foreign currency debt and the averages and volatility levels of the key variables during the 1995-2010 period, I make an assessment of what countries are likely to be exposed to exchange rate risks.
Table 4 shows that the Philippines has a share of 97,18 percent of its total public debt denominated in foreign currency. This means that almost all of its public debt has been exposed to the exchange rate volatility, which is 9,29%. Next to that, the Philippines experienced an average exchange rate depreciation of 8% over the 1995-2010 period, resulting in a exogenous debt increase. With 6,90% Peru is subjected to a more stable exchange rate volatility but a rather high exposure of 54,93% share of foreign currency denominated debt. Just as the Philippines, Peru faces an average exchange rate
depreciation (4,11%). Mexico has with 18,12% the smallest share of foreign currency debt. This share has however been exposed a large exchange rate volatility of 12,80%. Just as Peru and the Philippines, the share of foreign currency debt has on average also been subjected to exchange rate depreciation (13%). Indonesia’s exchange rate is was subjected to high volatility of 10,16% and a large average depreciation of 11,96% per year. Combined with the fact that almost half of its debt is denominated in foreign currency results in a high exchange rate risk exposure. India is less vulnerable to exchange rate fluctuations, since its average depreciation is 8,16%, the exchange rate volatility is relatively low (6,53%) and the
19 share of public debt which is subjected to exchange rate fluctuations is also relatively low at 27,67%
5 Results
First I give an overview of the results from estimating the Fiscal Response Function. Then I use these estimation results to construct a fan chart, in which two scenarios of foreign currency denominated debt are compared. Finally, I simulate the Exchange Rate Effect and the Interest Rate Effect.
5.1 Fiscal Response
Table 4 displays the results from estimating the fiscal response function. It shows that the Latin American (Mexico and Peru) countries intend to stabilize their Debt/GDP ratio, since their coefficient of the lagged Debt/GDP ratio is positive.
Comparing table 5 and 6 shows that adding a second lag of Debt/GDP to the FRF does not yield any better results in terms of statistical significance. Only in case of Indonesia the two lag fiscal reaction coefficient is significant, but only 10% and the coefficient of marginal of size. Therefore, the one Debt/GDP lag model is used for estimating the primary surplus.
Table 5 shows that the fiscal response for India and Indonesia are positive, of substantial magnitude, 0,15, and highly significant. Peru also has a positive fiscal reaction coefficient, which is significant at 10%. An outcome which contradicts with the null
hypothesis, which indicates a positive coefficient, is that the fiscal reaction coefficients are negative for both Indonesia and the Philippines. In case of Indonesia, the reaction
Table 4: Share of public debt denominated in foreign currency to total debt (2010), average exchange rate appreciation and its yearly volatility (1995-2010)
Table 5: One lag fiscal response coefficient estimate (1970-2010)
*** = significant at 1%, ** = significant at 5%, *significant at 10%
20 coefficients is of marginal size and not significant, but for the Philippines the reaction
coefficient is of substantial size and significant at 1%. The high fiscal reaction coefficients for Mexico is not in line with my expectations, since the during the period used for the estimation Mexico has (almost) defaulted once (Gruben and Welch, 1996). Based on this assessment, I also estimate the FRF on the 1985-2010 period, instead of the 1970-2010 period. The results from this estimation do not fit the null hypothesis better (appendix, table 1). Therefore I decide to use the results from the estimation based on the 1970-2010 period.
Based on the outcomes of the FRF estimation, I want to determine at what debt levels the countries with a positive fiscal reaction coefficient run a primary surplus. This is of
interest because previous to this threshold Debt/GDP ratio, a country is expected to run a primary deficit due to the negative However, from this point on a government is expected to start running primary surpluses in order to reduce the increase in Debt/GDP. Using equation (9) I calculate these debt levels :
(12)
where is the initial primary surplus as estimated by the FRF and
is the (reaction) coefficient which estimates the effect of the level of the lagged Debt/GDP ratio on the size of the primary surplus.
Mexico starts running primary surpluses at on 20% Debt/GDP, which is primarily due to its large fiscal response coefficient (0,14). Based on table 4 it was expected that Mexico would have to have a large fiscal response, since it should compensate for a large growth-interest rate differential in order to avoid an unsustainable debt path. Peru also runs an initial primary deficit, but at a threshold value of 44% its starts running surpluses because of its positive fiscal reaction coefficient. Although India has a high fiscal response coefficient (0,15), it only starts running primary surpluses at 56% Debt/GDP ratio. This is because of its large initial primary deficit of 8,6%. For the Philippines the situation is different because of its negative fiscal reaction coefficient ( = -0,07). Initially, the Philippines run a primary surplus but at a Debt/GDP level of 28% this turns into a primary deficit which amplifies the increase in Debt/GDP ratio instead of reducing it. For Indonesia, both the primary surplus and the fiscal reaction coefficient are negative. This means that Indonesia always runs a fiscal deficit, and the deficit increases as Debt/GDP ratio increases. This might lead to future sovereign debt sustainability problems, since table 4 shows that especially the high average yearly increase of the real value of the foreign currency denominated debt should be offset by an increase in fiscal surplus.
21 5.2 Fan charts
To construct the fan charts I first estimate the Vector Autoregressive (VAR) models for all variables (economic growth, domestic and foreign interest rate and exchange rate appreciation) and test for a unit root using the Augmented Dickey-Fuller test. Table 7 shows that at 10% significance all unit root hypothesis can be rejected. This means that in the case of a shock the process will return to its steady state.
In order to estimate the Foreign Currency Debt Effect (9) I simulate a stochastic debt path from 2011 until 2021 for two scenarios for each country. In the first scenario the country maintains its 2010 share of debt denominated in foreign currency. In the second scenario, the country has no foreign currency debt at all. By subtracting the Debt/GDP ratio
corresponding to the 97,5 percentile from the second scenario in 2021 off the Debt/GDP ratio corresponding to the 97,5 of the first scenario in 2021, I compute how much the upside risk of the future debt paths is influenced by the share of foreign currency risk. For this analysis, I use the actual levels of total Debt/GDP ratio in 2010 as a start off point. In the charts two individual graphs are plotted. The light gray chart is the scenario for which a government has no foreign currency debt. The dark area grey chart is the scenario for which the 2010 share of foreign currency debt to total debt remains during the projected period. The dark grey area, at the upside of the probability distribution, thus indicates the additional risk due to foreign currency debt.
Figure1: Value at Risk scenario analysis with FCD/GDP = actual 2010 level (%), and FCD/GDP = 0%
Fan chart 1.2: Mexico Fan chart 1.1: India
Table 7: ADF test for unit root (with constant, yearly data)
*** = significant at 1%, ** = significant at 5%, *significant at 10%
22 The first column in table 8 shows the SEI values in case the corresponding government has Foreign Currency Debt/GDP ratio of zero (light grey part in the fan chart). The second column shows the SEI values for the different countries for the scenario in which the percentage of Foreign Currency Debt/GDP is equal to its 2010 value (dark grey part in the fan charts). The difference between these values is defined as the Foreign Currency Debt Effect (equation 9) and is displayed in the third column. It is would be expected that a country which has a large share of its sovereign debt denominated in foreign currency, it is more likely to be faced with a high FCDE. Therefore, I correct the FCDE for the total percentage of foreign currency debt to assess how many percentage the SEI increase with a one percent increase in the share of foreign currency debt.
Table 8: Upper 97,5 percentile Debt/GDP values for both scenarios Fan chart 1.3: Peru
Fan chart 1.5: Philippines
23 5.3 Country analysis
The country that is least exposed to debt sustainability risks associated with foreign currency debt is India, with an Debt/GDP ratio increase as measured by the FCDE of only 6,65%. This is most likely to be the results of a substantial fiscal response (0,15), a relatively low share of foreign currency denominated debt (27%) and a marginal domestic growth-interest rate differential (0,12%). For both the scenario including and excluding foreign currency debt, the 97,5 percentile of the future debt path is stable. When the FCDE is corrected for the share of foreign currency debt/GDP, India also shows the lowest risk increase per percent of foreign currency debt.
The analysis for Mexico shows a similar development as the analysis for India. A big difference is however that the total Value at Risk, so the total 95% bandwidth of the
distribution as plotted in the graph, is much larger for Mexico for both scenarios. This is mainly the result of the higher volatility of the key economic variables. The FCDE of Mexico is however relatively low at 12%, and the foreign currency debt does not lead to an
unsustainable debt path. Just as in the case of India, this is due to a low share of foreign currency Debt/GDP (19%) and a high fiscal response of 0,16. This high fiscal response results in that the Mexican government is expected to run fiscal surpluses at Debt/GDP ratios beyond 20%, which enables the government to largely offset the debt increase due to a high growth-interest rate and exchange rate differential.
In the case of Peru, both scenarios result in a unsustainable debt path. The share of foreign currency debt however contributes substantially the increase of the upper bound of the distribution, with a FCDE of 18%. The fiscal response of 0,07, which results in primary surpluses after a threshold value of 40% Debt/GDP, is not sufficient to reduce the increase in the Debt/GDP ratio in such a way that the future debt path become stable. An explanation for this could be that Peru has a relatively large share (55%) of its sovereign debt denominated debt in foreign currency, which makes total debt more to be at risk to increase exogenously due to exchange rate depreciations. This is confirmed by the size of the corrected FCDE, which is similar for Mexico and Peru. The large exchange rate risk exposure and the higher interest payment on foreign currency debt thus result in a higher FCDE.
For Indonesia the upper bound of the future debt paths are unsustainable for both scenarios. However, chart 1.4 shows that the share of foreign currency debt contributes substantially to the size of the VaR of debt sustainability, resulting in a FCDE of 44%. This is due to the fact that the foreign currency debt part is very likely to be subjected to be exposed to severe exchange rate depreciations. This also becomes clear from the results of the corrected FCDE, which indicates that for every percent of foreign currency debt the FCDE increases with 1,7%, which is much higher than for the other countries. The Indonesian
24 government is also not to be expected to increase the sustainability of future sovereign debt by running primary surpluses, since the fiscal reaction coefficient is negative.
The Philippines has the highest percentage of foreign currency debt of total debt (97%) which results in a FCDE of 78%. What sets the situation of the Philippines apart from the other countries is that as a result of the share of foreign currency debt the future debt path become unsustainable, and even explosive. The explosiveness of the future debt path cannot be explained by unit roots in the VAR process, which are all rejected (table 5), but is likely to be caused by the large negative fiscal response of -0,07. In the case of a Debt/GDP ratio increases exogenously because of exchange rate depreciations, this increase is only amplified by the government instead of offset, as a result from the expected increase in the primary deficit. The large share of foreign currency debt of total sovereign debt and the pro-cyclical fiscal policy therefore result in a highly unstable future debt path.
5.4 Exchange rate and interest rate effect
When a country issues foreign currency debt, this part of total debt is subjected to foreign currency interest rates (for which I use external debt interest rates as a proxy) and exchange rate fluctuation, which are respectively defined as the interest rate and exchange rate effect. In order to quantify these effects, I run two additional scenarios; one to calculate the exchange rate exposure effect and one to estimate the interest rate effect. Then I benchmark these scenarios to the scenario for which there is no foreign currency debt, in order to decompose the two effects. This analysis yields results which are cannot be graphed in such a way that the difference between the two scenarios can be displayed in one graph, because the results from different scenarios overlap. Therefore, I plot the scenarios in two separate graphs. Then I compare the upper 97,5 percentile Debt/GDP ratios for the different scenarios, for which the results are outlined in table 9. I finally decompose the interest rate volatility effect, because literature indicates that the foreign currency debt interest rates should be less volatile which should have a stabilizing effect on the VaR of debt
25
Figure 2: Value at Risk scenario analysis with fixed exchange rates and equal and domestic currency debt interest rates
Fan chart 2.1: India Interest Rate Effect (e=0) Fan chart 2.2: India Exchange Rate Effect (rf=rd)
Fan chart 2.3: Mexico Interest Rate Effect (e=0) Fan chart 2.4: Mexico Exchange Rate Effect (rf=rd)
26 The results are in line with the literature findings. For all countries foreign currency debt interest rates are estimated to be lower which results in a stabilizing on the Value at Risk of debt sustainability. However, the degree to which this is the case again differs across countries. Debt sustainability in the Philippines would for example benefit greatly from eliminating exchange rate risk. However, it also benefits substantially from the lower interest rates it pays on its foreign currency debt. This is also, but to a more subtle extent the case for Indonesia. For Peru, Mexico and India the exchange rate and interest rate effect are much smaller.
Fan chart 2.7: Indonesia Interest Rate Effect (e=0)
Fan chart 2.8: Indonesia Exchange Rate Effect (rf=rd) Fan chart 2.10: Philippines Exchange Rate Effect (rf=rd)
Fan chart 2.9: Philippines Interest Rate Effect (e=0)
27 As expected, the exchange rate exposure has a destabilizing effect on the Value at Risk debt sustainability. The size of this negative effect is not offset by a destabilizing interest rate effect. This means that an increase of the share of foreign currency denominated public debt always leads to an increase of the risk to which debt sustainability is exposed, which was also concluded in the first fan chart analysis (figure 1).
The exchange rate volatlity effects indicates the increase in the VaR due to foreign currency debt interest rate volatility, relative to domestic currency interest rate volatility. According to the findings from the literature, this effect should be negative, which means that foreign currency debt interest rates should be less volatile than domestic currency debt interest rates. Based on the simulation of the exchange rate volatilty effect, this however turns out not to be the case. When comparing the domestic currency and foreign currency debt interest rate volatilty, it turns out that the volatility of foreign currency debt interest rates have a destabilizing effect on the VaR of debt sustainability. This effect is however, from an risk perspective, rather marginal.
6 Conclusion
This study contributes to existing literature by quantifying the sovereign debt sustainability risks associated with foreign currency denominated public debt in Emerging Market
economies, using a stochastic Value at Risk approach. The effect of issuing foreign currency debt on public debt sustainability is relevant since it exposes public debt sustainability to risks of (unexpected) negative exchange rate shocks, resulting in an exogenous debt increase which might eventually lead to a sovereign debt crisis. This was a major reason for the severity of the East Asian and Latin American crises of the 1990s (Eichengreen and Hausmann, 1999).
Exchange rate depreciation and exchange rate volatility has a destabilizing effect on debt sustainability. However, Eichengreen and Hausmann (2010) find that interest payments on foreign currency denominated debt reduces total interest payments and interest payment volatility, which has a stabilizing effect on debt sustainability. So there is a clear trade-off: when a government decides to issue foreign currency debt, debt sustainability benefits from a lower interest rate but suffers from exchange rate depreciation (volatility).
The Value at Risk of debt sustainability is defined as the projected Debt/GDP ratio that corresponds to the upper 97,5 percentile of the probability distribution after a 10 year projection. To assess the effect of foreign currency debt on the Value at Risk of debt
sustainability using a stochastic Value at Risk framework, I first simulate two scenarios. For the first scenario I use the 2010 share of foreign currency debt to total debt and project the Value at Risk for debt sustainability. For this scenario, I assume that the interest rate on
28 foreign currency denominated debt equals interest rates on external debt. For the second scenario, I assume a country only has domestic currency debt. When comparing results, it shows that for all of the four countries in the sample foreign currency debt leads to an increase in debt the Value at Risk of sustainability. However, the size differs substantially across countries. The increase in Value at Risk is very considerable for the Philippines (78% risk increase), Indonesia (44% risk increase), Peru (18% risk increase) and Mexico (12% risk increase) but rather marginal for India (7% risk increase).
To simulate the effect of exchange rates and foreign currency debt interest rates on debt sustainability, I run two addition scenarios for all countries. To estimate the interest rate effect I assume that the real exchange rate appreciation is zero, which is the same as
assuming a fixed exchange rate regime. I estimate the exchange rate effect by assuming that the foreign currency interest rate equals the domestic currency interest rate. Then I
benchmark both scenarios to the scenario for which there is no foreign currency debt. This comparison shows that for all countries the interest rate effect is negative, which indicates that the interest rate of foreign currency debt has a stabilizing effect on debt sustainability. The exchange rate effect is positive for all countries, which indicates that exchange rate exposure has a destabilizing effect on debt sustainability. The destabilizing exchange rate effect dominates however dominates the stabilizing interest rate effect.
Even though results show that foreign currency denominated debt has a destabilizing effect on future debt sustainability, countries remain issuing foreign currency debt. The main explanations for this phenomenon is that either a country is unable to fully finance its debt in domestic currency debt, or that it values a lower interest rate on foreign currency debt higher than lower debt sustainability risk.
7 Discussion and suggestions for future research
The stochastic Value at Risk approach proved its ability of assessing the increase of VaR of debt sustainability as a result of foreign currency denominated debt. There are however also some very important drawbacks regarding the stochastic approach which can be improved upon in future research. Furthermore, I propose a way of introducing the central bank to the analysis and I propose a way of extending the framework for which this type of research is relevant.
A major drawback is the unrealistic assumption that total debt, consisting of domestic and foreign currency denominated debt, is subjected to exchange rate and interest rate shocks. This assumption implies that total debt is completely composed out of loans with a one year terms, that refinanced on a yearly basis. This is not realistic since public debt is usually composed out of debt which’ maturity exceeds one year. Taking maturity structure
29 into account using the stochastic Value at Risk method would therefore yield a more realistic analysis on how debt sustainability is affected by exchange rate and interest rate shocks. Data on maturity structure is however very limited, especially for foreign currency debt.
Another drawback is that the flow government budget constraint, which is used to predict future paths of public debt, does not take into account the interaction between the key economic variables in the model. There might however be a reason to take these interaction effect into account. For example, Dell’Erba, Hausmann and Panizza (2008) find that in emerging market countries there is a significant correlation between spreads and debt levels. So an increase in the total Debt/GDP ratio is expected to result in a higher interest rate. Incorporating the relation between the different variables in the flow budget constraints would therefore yield more realistic future debt paths. Carrera and Vergara (2012) do take these interaction effects into account in their attempt of valuing public debt, so this paper could provide a framework for taking interaction effects into account in the flow government budget constraint.
The prediction of the future debt paths relies heavily on the ability of the flow
government budget constraint to predict the evolution of Debt/GDP ratios over time correctly. However, it turns out that actual historical Debt/GDP ratio’s differ (substantially) from
Debt/GDP ratios as predicted by the flow budget constraint. In this case future Debt/GDP ratios are subjected to both uncertainty of the future values of the key economic variables, and to uncertainty of ability of the flow government budget to predict the Debt/GDP ratios correctly. This uncertainty can be taken into account by computing an error term, reflecting the average historical difference between the actual and predicted Debt/GDP ratios, which can be added to the government flow budget constraint. However, since historical data on foreign currency denominated public is limited, it is not possible to compute this error term for countries which engage in issuing foreign currency debt for a long term.
The Value at Risk of debt sustainability is estimated using the real variables, since the real value of future public debt is of interest. However, I show that inflation can heavily affect the volatility of the real variables, resulting in larger size of the simulated errors. Since especially in Advanced Economies inflation is considered to be a policy instrument controlled by the central bank (Calvo, 1978) (Lydland and Prescott ,1977) (Barro and Gordon,1983) (Svensson, 1999), the model could be improved upon by replacing the real variables by nominal variables and a inflation or central bank reaction function. This central bank /
inflation reaction function should, just as the fiscal response function, depend partially on the Debt/GDP ratio. In this case the VAR model is estimated using the nominal values of the variables, which are likely to be less volatile than the real value variables and yield therefore less volatile simulated errors. This will eventually lead to a more precise estimation of future
30 debt paths. After estimating the VAR functions for the nominal variables, these can be
adjusted for inflation by the inflation response function.
The stochastic Value at Risk approach is not suited for analyzing the risk of sudden capital outflows or a break in an exchange rate peg. However, building further on the work of Lukkezen and Rojas-Romagosa (2014), it would be worth investigating if this approach could be used for constructing an early warning indicator for capital flights or the break in an
exchange rate peg. With respect to the break in an exchange rate peg, the work of Lukkezen and Rojas-Romagosa could be combined with the work of Husain, Mody and Rogoff (2004) who asses the durability and performance of exchange rate regimes. The probability of the break of the peg could then be estimated by a probability model, which could replace the real exchange rate appreciation variable in the government budget constraint.
References
Belhocine, Nazim, and Mr Salvatore Dell'Erba. 2013 “The impact of debt sustainability and the level of debt on Emerging Markets spreads”. No. 13-93. International Monetary Fund. Bhandari, Jagdeep S., Ul Haque, Nadeem and Turnovsky, Stephen J. 1990. "Growth, external debt, and sovereign risk in a small open economy." Staff Papers-International Monetary Fund : 388-417
Bohn, Henning. 1998. “The Behavior of U.S. Debt and Deficits.” Quarterly Journal of Economics 113(3):949–63.
Bordo, Michael D., Meissner Christopher M., Stuckler, David. 2010. “Foreign Currency Debt, Financial Crisis and Economic Growth: A Long Run View. Journal of International Money and Finance, 29(4), 642-665.
Budina, Nina, and Van Wijnbergen, Sweder. 2009. Quantitative Approaches to Fiscal Sustainability Analysis: A Case Study of Turkey since the Crisis of 2001. The World Bank Economic Review 23.1 119-140.
Caballero, Ricaro J. and Krishnamurty, A. 2000. “Dollarization of Liabllities: Under insurance and Domestic Financial Underdevelopment”. No. w7792. National bureau of economic research.
Calvo, Guillermo A. 2009. “Capital Markets and the Exchange Rate, with Special Reference to the Dollarization Debate in Latin America”. Journal of Money, Credit and Banking: 312-334.
Calvo, Guillermo A. and Mishkin, Frederic S. 2003. “The Mirage of Exchange Rate Regimes for Emerging Market Countries”. Journal of Economic Perspectives, v17(4,Fall), 99-118 Calvo, Guillermo A., Izquierdo ,Alejandro and Talvi, Ernesto. 2003. “Sudden stops, the real exchange rate, and fiscal sustainability: Argentina's lessons”. No. w9828. National Bureau of Economic Research.
