This Selected Issues paper analyzes the sources of Mexico’s economic growth since the 1960s, and compares various decompositions of historical growth into trend and cyclical components. The role of the implied output gaps in the inflation process is assessed. The paper presents medium-term paths for GDP based on alternative productivity growth rates. The paper also describes the significant steps Mexico has taken to strengthen the structure of its public debt in recent years, both in terms of currency composition and maturity.

Abstract

This Selected Issues paper analyzes the sources of Mexico’s economic growth since the 1960s, and compares various decompositions of historical growth into trend and cyclical components. The role of the implied output gaps in the inflation process is assessed. The paper presents medium-term paths for GDP based on alternative productivity growth rates. The paper also describes the significant steps Mexico has taken to strengthen the structure of its public debt in recent years, both in terms of currency composition and maturity.

II. Structure and Cost of Public Debt in Mexico1

Abstract

This paper describes the significant steps Mexico has taken to strengthen the structure of its public debt in recent years, both in terms of currency composition and maturity. In conjunction with this process, financing costs have fallen, as financial stability has led to enhanced creditworthiness. We then seek to explain empirically changes in the structure of debt, based on standard debt management principles, and assess how far this process has taken Mexico relative to debt structures of comparator countries. Finally, the paper looks at upcoming challenges to Mexico’s debt management program, including the evolving status of “off balance sheet” debt, and the possible trade off between lowering financing costs and strengthening the debt structure.

A. Introduction

1. Strengthened debt management in Mexico has had wide-ranging benefits. It has: reduced public sector vulnerabilities; helped fiscal consolidation by providing substantial debt service savings in NPV terms; and promoted the development of domestic financial markets. This chapter reviews these achievements and discusses some of the remaining challenges. First, debt management policies undertaken since 1998 and their effects on the debt structure are described. Second, the role of financial and external conditions in shaping the debt structure is investigated empirically. Third, Mexico’s debt structure is compared to debt structures in other emerging market countries.

2. This chapter extends an earlier staff paper on public debt management (see IMF Country Report 02/238). The earlier paper discussed in detail the public debt strategy followed by Mexico, including the institutional and legal framework as well as regulatory reforms aimed at modernizing the domestic debt market. It also provided a useful description of the specific debt instruments used by the authorities. These issues have therefore not been developed in detail in this chapter. In updating the analysis of debt management policies (section B), this chapter innovates by presenting indicators of debt structure for the consolidated public sector. The empirical investigation and country comparisons are also new.

B. Debt Management Policies 1998–2003

3. There has been a continuous improvement in the structure of public debt since 1998. This has been the result of well-articulated policies, improved market access, and declining interest rates. This section describes the policies and the evolving structure of public debt since 1998.

Background

4. Public debt in Mexico consists of the debt of the budget sector and debt guaranteed by the public sector. The budget sector comprises the federal government, non-financial public enterprises, and other public entities such as the social security institutes. Public debt guaranteed by the public sector (the federal government, congress, or public companies) includes that of the Saving Protection Institute (IPAB), extrabudgetary trust funds, public development banks, and public sector investment projects undertaken by the private sector (PIDIREGAS). IPAB manages the debt resulting from the banking crisis that followed the 1994-95 financial crisis. The government provided an implicit guarantee on most bank liabilities at the time, and bore the bulk of the cost of the resolution of the banking problems. A trust fund for the rescue of toll roads (FARAC) was also established in the aftermath of the financial crisis, with the role of rescuing private toll companies. The debt used to finance PIDIREGAS projects is transferred to the government after completion of the project.2

5. The authorities monitor two definitions of public debt (Figure 1 and Table 1). The traditional definition covers the debt under direct budgetary control and is used for the purpose of budgetary legislation. Specifically, it covers the domestic debt of the federal government and the external debt of the public sector, including public development banks, and a portion of the PIDIREGAS debt (an amount equivalent to amortization obligations for the next two years is recorded in the traditional debt upon completion of individual PIDIREGAS projects). The augmented definition encompasses all government and public-sector guaranteed debt, thus covering the entire nonfinancial public sector as well as the public development banks. The components of the augmented debt are reported on a quarterly basis, for information purposes. Budget laws (appendices) set legal ceilings for most of them on an annual basis.

Figure 1.
Figure 1.

Mexico: Indicators of Net Public Debt, 1998-2004

Citation: IMF Staff Country Reports 2004, 418; 10.5089/9781451825626.002.A002

Source: SHCP
Table 1.

Mexico: Public Sector Gross and Net Debt, 1998–2004

article image
Source: SHCP.

As of end-March 2004; ratios refer to quarterly nominal GDP.

All domestic debt.

Overall strategy

6. Debt policies have aimed at reducing both vulnerabilities and financing costs, while supporting the development of the domestic financial system, as stated in the government’s 2002-06 economic program (PRONAFIDE). There were four main objectives: (i) improving the debt amortization profile; (ii) reducing financing costs; (iii) reducing the vulnerability of public finances to interest rate and exchange rate variations; and (iv) attenuating the adverse effects of variations in international capital availability. Other objectives were also identified. The development of a long-term domestic yield curve was seen as important to enhance financial efficiency and facilitate private sector bond issues. Regarding external debt, the broadening of the investor base, the development of euro and yen yield curves, and liability management operations to reduce interest and exchange rate risks were priorities.

7. The federal government has made great strides in adjusting the structure of its debt in line with these objectives. The strategy has been to substitute domestic for external debt, while developing the domestic debt market. The federal government started in 2001 to finance its entire fiscal deficit in the domestic market. A further step was taken in 2004, as domestic borrowing was used to repay external debt (US$500 million). As a result, the share of external debt in net traditional debt declined from 70 percent in 1998 to 48 percent in 2003. The decline was less pronounced for net augmented debt, from 50 percent to 38 percent during the same period, reflecting the growth of foreign-currency PIDIREGAS debt (Figure 1).

8. Sustained efforts have been made to develop the domestic debt market. Its rapid growth has been supported by the rise in domestic savings.3 Financial regulations have been modernized to facilitate the growth of institutional investors’ domestic assets, while specific debt instruments have been created to fit their portfolio requirements. Initiatives to broaden the investor base include the deepening of the domestic derivative market and the development of a zero-coupon government bond market, in order to attract foreign investors and investors requiring long-duration instruments. Measures have been taken to promote liquidity in the secondary market, by improving regulations (e.g., market-maker programs) and by reopening issues to provide fewer and more liquid benchmark issues. Transparency has also been improved. In january 2004, the federal government for the first time made public its domestic debt strategy for the year, and has since published quantitative auctioning targets for each type of security on a quarterly basis.

Domestic debt

9. Domestic debt management has concentrated on reducing interest rate and refinancing risks. The main vehicle has been the development of long-term fixed-rate instruments and the lengthening of other instruments’ maturities. These policies have resulted in the extension of the average duration of market debt, the development of a domestic yield curve, and the smoothing of future amortizations. Another milestone has been the transformation of IPAB’s debt structure. The concentration of amortization payments in 2005-06 has been partly resolved through large-scale refinancing operations. Legal impediments to completing these efforts have also been resolved, as the government and banks recently reached agreement over disputed FOBAPROA notes.4 This agreement should result in the exchange of the old FOBAPROA notes for new IPAB debt that can be refinanced. Until now, IPAB debt has continued to consist entirely of indexed short-duration instruments, thereby limiting the strengthening of overall public sector debt.

10. The federal government has strengthened its debt structure by using long-term fixed-rate debt. In 1998, the federal debt consisted entirely of short-term debt and debt indexed to short-term rates or inflation. The first bonds with fixed nominal interest rates were introduced in 2000 with tenures of 3 and 5 years. Longer maturities were issued in the following years, with ten-year bonds in 2001, a seven-year bond in 2002, and a 20-year bond in 2003. This has made the composition of the federal government market debt safer, with more than a third in fixed-rate bonds as of March 2004. From the perspective of the broader public sector, progress has been slower. The share of long-term fixed-rate debt in the consolidated domestic debt of the government, IPAB, and FARAC, rose from zero to 16½ percent from 1998 to March 2004 (Table 2 and Figure 2). IPAB debt is still almost entirely indexed to short-term rates. The debt of FARAC, in contrast, has been refinanced using inflation-linked instruments.5 As a result of the use of fixed-rate bonds, the shares of inflation-indexed and short-term indexed debt have declined by 4 and 5½ percentage points since 1998, whereas the share of other debts (essentially loans) has declined by 6 percentage points.

Table 2.

Mexico: Public Sector Domestic Debt by Debtor and Type of Instrument, 1998-2004

article image
Source: SHCP.

As of end-March 2004; ratios refer to quarterly nominal GDP.

Staff estimate for 1998 and 2004.

Figure 2.
Figure 2.

Mexico: Consolidated Government Domestic Debt by Type of Instrument, 1998-2004 1/

(in percent of GDP)

Citation: IMF Staff Country Reports 2004, 418; 10.5089/9781451825626.002.A002

1/ Includes Federal government (budget sector), IPAB/FOBAPROA, and FARAC debt. 2004 refers to first quarter data.

11. The use of longer-duration instruments has lengthened the maturity profile of the federal government domestic debt (Figure 3). The average maturity of federal domestic securities more than doubled between 1998 and 2003, reaching 2.6 years in early 2004. This was achieved initially by relying on debt indexed to short-term rates, and subsequently by using fixed-rate bonds. The duration of bonds indexed to short-term rates or inflation was also extended. The average time interval between rate revisions increased from 34 to 78 days between 1998 and early 2004 for bonds indexed to short-term rates; and from 36 to 63 months for bonds indexed to inflation.6 As a result, the average duration of federal securities has risen faster than their average maturity since 2001 (65 percent versus less than 30 percent). Progress has been slower with respect to other public sector securities (Table 3). While the duration of FARAC securities has stayed in the 10-11 year range, the duration of IPAB securities has improved only marginally, reaching between 30 to 50 days in early 2004.7 The duration of IPAB’s total debt is even lower, at about 17 days, reflecting the more frequent rate revisions of its bank loans (Table 4).

Figure 3.
Figure 3.

Mexico: Maturity of Federal Government Domestic Securities

(in days)

Citation: IMF Staff Country Reports 2004, 418; 10.5089/9781451825626.002.A002

Table 3.

Maturity Structure of Public Sector Domestic Securities

article image
Source: IMF staff estimates.
Table 4.

Mexico: Structure of IPAB Debt, 1999–2004

article image
Source: IPAB.

As of June 2004.

New debt issued since 1999 to refinance the original debt, as a percentage share of the outstanding debt..

Calculated as the ratio of accrued interests to the average of the debt stock at the beginning and the end of the year. Excludes commissions and fees.

12. The domestic amortization profiles of the federal government and IPAB have improved. The share of government domestic debt with remaining maturity of less than a year fell from 62 percent at end-1998 to 34 percent at mid-2004 (Figure 4). In the case of IPAB, amortization humps expected in 2004 and 2005 have been substantially reduced compared to the original profile (Figure 5). This has been achieved through pro-active liability management. As of June 2004, 60 percent of the initial debt had been refinanced (Table 4). The recent agreement between the government and banks to exchange old FOBAPROA notes has created further scope for smoothing IPAB’s amortization profile. These notes represented close to 28 percent of IPAB’s total debt or 3¼ percent of GDP as of end-2003. While the entire stock falls due in 2005–06, the notes could not be refinanced because they were disputed. The exchange is expected to result both in the refinancing and reduction of the IPAB debt. Tentatively, it has been estimated that banks will receive close to half of the outstanding amount of FOBAPROA notes in new IPAB bonds, with the difference reflecting recovered loans, loss sharing covered by banks, and related credits banks will pay. The terms of the exchange are to be finalized following the audits of the banks’ loan portfolios.

Figure 4.
Figure 4.

Mexico: Amortization Profile of Federal Government Domestic Debt, 1998-2004

(in percent of outstanding debt)

Citation: IMF Staff Country Reports 2004, 418; 10.5089/9781451825626.002.A002

Figure 5.
Figure 5.

Mexico: Amortization Profile of IPAB Debt, as of 1999 and 2003

(in billions of pesos)

Citation: IMF Staff Country Reports 2004, 418; 10.5089/9781451825626.002.A002

Source: IPAB.

13. Another beneficial effect has been the establishment of a domestic yield curve with long-term benchmarks (Figure 6). Efforts underway to promote the liquidity of the benchmarks are gradually improving the efficiency of the curve (by reducing the number of outlier issues).

Figure 6.
Figure 6.

Mexico: Government Peso Yield Curve, 2000-04

(In percent)

Citation: IMF Staff Country Reports 2004, 418; 10.5089/9781451825626.002.A002

Source: Bloomberg.

External debt

14. External liability management has been pro-active against the backdrop of declining world interest rates and strong investor demand for emerging market securities. The main feature of liability management has been the substitution of global bonds for restructured debts, allowing reductions both in the stock of external debt and in future debt-servicing costs. New bond issues aimed at improving the liquidity and structure of bonds and diversifying the investor base. Refinancing risks have been kept under control.

15. The significance of capital market debt in public sector external debt has grown rapidly since 1998 (Figure 7). The share of capital market debt, including PIDIREGAS debt,8 in public sector external debt increased from 29 percent at end-1998 to 43 percent at end-March 2004. This reflected the pre-payment of restructured debts and the growth of PIDIREGAS debt. Mexico finished pre-paying its outstanding Brady bonds in July 2003 (for US$3.5 billion). These pre-payments, permitted by the call option embedded in the bonds, resulted in substantial external debt reduction and net present value savings. The pre-payment produced some external debt reduction because the bonds were bought back at a discount and the collateral released was partly used to repay the debt. Similar pre-payments have been initiated for debt that was obtained at less favorable market conditions. For example, two floating-rate notes, which included call options, were bought back in 2002 (for a total close to US$900 million); and several multilateral loans have also been recently pre-paid.

Figure 7.
Figure 7.

Mexico: Public Sector External Debt by Type of Creditor, 1998-2004

in billions of US$

Citation: IMF Staff Country Reports 2004, 418; 10.5089/9781451825626.002.A002

16. New global bonds were issued with special attention given to improving the liquidity and structure of bonds and diversifying the investor base. The liquidity of bond benchmarks has been increased by doing larger issues and reopening them. For instance, ten-year dollar denominated bonds of US$2 billion were placed in 2003, double their size in 1997-99. The structure of bonds has been improved by including Collective Action Clauses (CACs). CACs have been included in all new bonds since March 2003. Financing sources have also been diversified by tapping more in non-U.S. capital markets. A recent example is the sovereign placement of a 20-year, 500 million British pound sterling denominated bond in 2004—the cost of which was lower than a similar placement in dollars.9

17. Refinancing risks have been kept under control. The government has been able to extend the maturity of foreign bond issues. The average maturity of bonds issued in 2001-03 was significantly higher than the maturity of bonds issued in the three previous years (Figure 8). At the same time, the average maturity of the market external debt has continued to decline. This reflects the effect of the pre-payment of the Brady bonds, in addition to the natural downward drift in the maturity of existing debt.10 Nevertheless, refinancing risks have been kept under control, as the government has managed to maintain a favorable amortization profile (Figure 9). As in 1998, the share of short-term debt by remaining maturities represented only 15 percent of the government external debt at the end of 2003.

Figure 8.
Figure 8.

Mexico: Maturity of the Federal Government External Market Debt, 1998-2003

(in years)

Citation: IMF Staff Country Reports 2004, 418; 10.5089/9781451825626.002.A002

Figure 9.
Figure 9.

Mexico: Amortization Profile of Public Sector External Debt 1/, 1998-2004

(in millions of U.S. dollars)

Citation: IMF Staff Country Reports 2004, 418; 10.5089/9781451825626.002.A002

1/ Excluding PIDIREGAS debt.

18. As in the case of domestic debt, external debt management has had a strong focus on improving the efficiency of the yield curve (Figure 10). Benchmarks in the yield curve have been developed through new issues and reopenings. For instance, a 20-year U.S. dollar bond was issued for the first time in 2002. Mexico completed in 2004 an exchange of global bonds to improve the efficiency of the yield curve (the first debt exchange of this type by an emerging market country).11 The exchange was designed to allow investors to trade out bonds that perennially traded above the yield curve into bonds that priced closer to the curve.

Figure 10.
Figure 10.

Mexico: Government U.S. Dollar Yield Curves, 2000-2004

(in percent)

Citation: IMF Staff Country Reports 2004, 418; 10.5089/9781451825626.002.A002

Source: Bloomberg.
Table 5.

Mexico: Public Sector External Debt by Type of Creditor, 1998-2004

article image
Source: SHCP.

As of end-March 2004.

Includes FX stabilization fund and PIDIREGAS debt assumed by the government.

Table 6.

Mexico: Public External Debt by Currency, 1998–2004

article image
Source: SHCP.

As of end-March 2004.

Refers to PIDIREGAS external debt. These are mainly in U.S. dollars.

Debt costs

19. While the strengthening of the domestic debt structure may entail additional interest costs, the evolution of debt costs suggests scope for interest savings on external debt (Figure 11).12 The average cost of domestic debt was higher than the 28-day cetes rate by 2.3 percentage points in 2003. This gap, which has grown, partly reflects the premium paid for strengthening the debt structure, in addition to fees equivalent to 0.5 percentage points in that year. It is expected to grow further as the government expands its use of long-term nominal debt. Higher interest costs could also materialize if IPAB debt is to be strengthened. The cost of IPAB debt is currently lower than the cost of government debt, reflecting its indexation structure. The gap between the average cost of the external debt and the cost of new borrowing was significant in 2003 (estimated at 2.3 percentage points for the whole external debt of the federal government, including bonds and other borrowings, compared to none in the previous year). While fees increased significantly in 2003, this gap mainly reflected the slow response of the debt structure to interest rate changes, being mostly long-term and at fixed-rates (Table 7 shows this for external bonds alone). Interest savings should take place in the future. Liability management operations were substantial in 2003, with new bond issues equivalent to 17½ percent of outstanding external bonds at end-2002, and primarily guided by cost-reducing objectives. The lag and the magnitude of future cost reductions is difficult to forecast, however.

Figure 11.
Figure 11.

Mexico: Financial Cost of Federal Government Debt, 1998-2004 1/

Citation: IMF Staff Country Reports 2004, 418; 10.5089/9781451825626.002.A002

Source: SHCP and staff calculations.1/ Includes commissions and fees. For domestic debt, fees have been stable at around 0.3-0.5 percent since 2001. For external debt, fees have been more volatile. They amounted to 0.8 percent in 2001, 0.2 percent in 2002, and 0.7 percent in 2003.
Table 7.

Mexico: Structure of the Federal Government External Marketable Debt, 1998-2003 1/

article image
Source: SHCP.

All sovereign bonds.

Average coupon of the fixed-rate bonds

C. Determinants of Debt Structure

20. Debt management responds to optimization criteria and outside constraints. This section attempts to identify factors influencing the structure of public debt in Mexico.

Theoretical background

21. The available literature provides two ways to think about a country’s appropriate debt structure. One is the type of debt structure that is optimal for the borrowing country. The other is the perspective of lenders (or investors) and the type of risk sharing that they are willing to take.1

22. The optimality of sovereign debt structures has been linked to the credibility of macro-policies and the robustness of the budget in the borrowing country. The borrowing country can gain or lose in both respects by varying the composition of its debt. On the one hand, anti-inflationary credibility can be improved by increasing reliance on indexed, short-term, or foreign currency debt relative to long-term domestic nominal debt. On the other hand, the vulnerability of the budget to inflationary shocks can be diminished by relying on nominal debt rather than indexed debt when inflationary shocks worsen the fiscal balance. There are other, similar, tradeoffs of relevance for sovereign debt structures. For instance, the use of short-term debt will increase vulnerability to confidence crises compared to long-term debt. Also, the use of short-term debt and foreign-currency debt can constraint monetary and exchange rate policies, respectively. Borrowing countries can be expected to adjust the structure of their debt to optimize this type of trade-off.

23. Attempts at estimating optimal sovereign debt structures have been made using risk management tools. For instance, Giavazzi and Missale (2004) found that the risk-return characteristics of treasury securities in Brazil warranted a greater share of inflation-linked debt and a further reduction in foreign debt. With risk minimization as the main goal, the preferred debt instruments are those with stable returns and good hedging properties for the budget. The optimal debt structure then reflects covariations of each security’s returns with shocks affecting the budget (e.g., output, price, exchange rate, and interest rate shocks in the Brazil study).

24. Attention has also been given to constraints from the lending side. Vulnerable sovereign debt structures have often been associated with the reluctance of investors to lend in domestic currency and at longer terms. Primarily, investors want to protect themselves from uncertainty arising from high and variable inflation. This gives a central role to macroeconomic stability and puts countries with more frequent disturbances at a disadvantage. Nevertheless, demand for foreign-currency and/or indexed debt tends to persist long after disinflation or fiscal adjustment have been achieved. Borensztein and al. (2004) provides several explanations. Anti-inflationary credibility can take a long time to establish, with some countries being unable to demonstrate discipline in the presence of long-term domestic debt. There may be impediments to transiting to new debt instruments, including investors’ familiarity with the current instruments. In addition, foreign investors tend to remain reluctant to hold exchange rate risk.

25. Altogether, these considerations point to the country specificity and the broad range of determinants of sovereign debt structures. Among other things, debt structures will reflect the quality of a country’s macro-policies, the robustness of its budget to inflationary and real shocks, the hedging properties of debt instruments, and investors’ willingness to share risk. The variety of these determinants warrants an empirical investigation.

Empirical approach

26. Our empirical investigation of Mexico’s debt structure focuses on two questions: (i) the significance of credibility and hedging motives, and (ii) the influence of economic shocks. With a view to inferring the effects of debt management improvements, the same estimations are repeated for two time intervals, from 1996 to mid-20042 and from 1999 to mid-2004.

27. The role of credibility and hedging motives is based on Goldfajn (1998). Goldfajn investigates the relationship between the share of nominal debt (respectively the share of foreign-currency debt) with the size of public debt, the variance of inflation (respectively the real exchange rate), and the covariance of inflation and budgetary spending (respectively of the real exchange rate and spending). Consistent with the trade-off between credibility and hedging motives, he finds that the proportion of nominal debt is negatively correlated with the size of public debt and the variance of inflation, and positively correlated with the covariance of inflation and budgetary spending. The average maturity of debt is found to behave in the same way. No significant relationship is found between the proportion of foreign debt and the variance of the real exchange rate and its covariance with spending.

28. The same relationships are estimated for Mexico, using monthly data. The primary balance of the nonfinancial public sector is used as the fiscal variable. Monthly debt statistics of the Ministry of Finance, covering the traditional debt,3 are used to compute debt shares. The debt is divided into three categories: (i) nominal debt (cetes and fixed-rate bonds); (ii) indexed debt (floating-rate and inflation-linked debt); and (iii) foreign currency (external and dollar-linked debt). The size of public debt is measured as a ratio to GDP using quarterly statistics. Variance and covariance time series are computed as in Goldfajn (1998), using VAR estimations for inflation, the real peso-dollar exchange rate, and the primary balance. For each observation, a new VAR is estimated with the previous 60 observations and two lags. Variances and covariances are calculated using the VAR’s residuals.

29. The role of economic shocks is studied focusing on oil prices, U.S. interest rates, and Mexico’s business cycle. Regressions are estimated linking indicators of debt composition with PEMEX’s oil export price, the yield of 3-month U.S. treasury bills in the secondary market, Mexico’s industrial production index, and the variance of these indicators. To avoid problems of nonstationarity, first differences are used for the oil price and the industrial production index. Variance time series are computed as before, using variables in levels.

Results

30. A simple look at the joint evolution of debt composition, maturity, inflation, and the debt ratio suggests a strong influence of macro-fiscal conditions on the structure of debt. The first two charts in Figure 12 show that the transition to a safer debt structure was largely concurrent with disinflation and the reduction of debt levels. The last chart in Figure 10, showing the fall in the volatility of inflation and the real exchange rate around 2000, corroborates this finding.

Figure 12.
Figure 12.

Mexico: Monthly Financial and Debt Variables, 1990 - 2004

Citation: IMF Staff Country Reports 2004, 418; 10.5089/9781451825626.002.A002

Source: SHCP, Central bank of Mexico, and Fund staff estimates.

1996-2004

31. Regression results are broadly consistent with the theory (Box 1). The first set of regressions finds a significant relationship between the composition of public debt and the size of the debt and the volatility of inflation, consistent with credibility considerations. Reliance on indexed and foreign debt tends to increase with the size of debt and the volatility of inflation. In turn, hedging variables are not found to be significant except for one specification (equation 4). Nevertheless, these variables generally appear with the expected signs. In the estimation results, a positive covariance between inflation and the primary balance tends to reduce the share of nominal debt; and a positive covariance between the real exchange rate and the primary balance tends to increase the share of foreign currency debt. Both relations are consistent with the search for debt instruments that smooth budgetary needs. One observation is that these covariance series have not been stable throughout the period, making it difficult to link the choice of debt instrument to the budget’s sensitivity to inflation and real exchange rate shocks.

32. The second set of regressions demonstrates the influence of U.S. interest rates, the oil price, and the level of industrial activity on the structure of debt. These variables, or their volatility, are correlated significantly with parameters of the debt structure. The coefficients have the expected signs—with greater volatility associated with riskier debt structures. The regressions show that an improvement in the external environment, including a reduction in the level and volatility of U.S. interest rates and a higher oil export price, tends to be associated either with a rise in the share of nominal debt or an extension of the average maturity.

33. The latter result has two explanations. First, investors may be more willing to take risks in lending to Mexico when the external environment is more favorable. Second, the cost of improving the debt structure in good times tends to be lower, and the authorities can seize the opportunity to reduce the risk of their debts. The risk-cost trade-off tends not to be a constraint during those times.

Regression Results, 1996-2004

Notations:

m: average term to maturity of federal government securities

SNOM: share of nominal debt in total debt

SFX: share of foreign currency debt in total debt

d: debt ratio

i: 12-month inflation

rer: real exchange rate against the U.S. dollar

pb: primary balance

POIL: oil export price

rUS: U.S. interest rate

y: production index

t: time trend

D(-) indicates the first difference of a variable, using a 12 month lag

Var(-) indicates the variance series of a variable

Cov(-,-) indicates the covariance series between two variables

MA(a, b) indicates a moving average process, with coefficients a and b for the first two lags

Estimation Results:1
m=5.9(33.3)t2.7(2.7)d121.4(5.4)Var(i)+ɛwithɛMA(1.1,1,0.6)(1)
SNOM=0.2(22.9)t0.1(3.1)d7.2(7.2)Var(i)0.3(1.6)Cov(i,pb)(t12)+ɛwithɛMA(1.3,1.1,0.5)(2)
SFX=1.8(47.5)d+10.5(3.6)Var(i)+0.1(5.1)Var(reer)+0.9(1.5)Cov(rer,pd)(t12)+ɛwithɛMA(0.9,0.4)(3)
D(SNOM)=0.1(10.4)t0.1(3.7)d1.8(2.4)Var(i)0.7(5.6)Cov(i,pb)(t12)+ɛwithɛMA(1.3,1.2,1.2,0.2)(4)
SNOM=0.2(27.3)t4.5(4.0)Var(i)0.1(2.5)D(POIL)1.2(7.1)rUS91.4(3.6)Var(rUS)+0.1(2.7)D(y)+ɛ(6)
SNOM=0.2(27.3)t4.5(4.0)Var(i)0.1(2.5)D(POIL)1.2(7.1)rUS91.4(3.6)Var(rUS)+0.1(2.7)D(y)+ɛ(6)withɛMA(1.1,0.9,0.4)
SFX=0.2(6.2)t3.7(3.24)Var(POIL)+3.814.7rUS+285.7(5.6)Var(rUS)+15.6(20.3)Var(y)+ɛwithɛMA(1,0.6)(7)
1/ t-statistics are shown in parentheses. The Durbin Watson statistic is close to 2 for all equations.

1999-2004

34. The results are broadly unchanged by reducing the time interval to 1999-2004 (Box 2). The responsiveness of the debt structure to the size of debt and the economic environment seems to change, although not necessarily in statistically significant terms. On one hand, the oil price and the volatility of inflation seem to lose some significance as determinants of the debt structure. On the other hand, the debt structure responds more to U.S. interest rates and the size of the debt (equation 9 and 13).

Regression Results, 1999-2004

m=6.1(18.8)t4.2(2.3)d103.7(4.7)Var(i)+ɛwithɛMA(1.1,1.3,0.8)(8)
SNOM=0.3(16.4)t0.6(6.4)d6.1(5.8)Var(i)0.1(0.53)Cov(i,pb)(t12)+ɛwithɛMA(0.9,0.9,0.5)(9)
SFX=0.1(2.5)t+1.2(6.2)d+19.3(7.5)Var(i)+0.1(1.8)Var(rer)+0.1(0.2)Cov(rer,pb)(t6)+ɛwithɛMA(1.1,0.6)(10)
D(SNOM)=0.1(5.6)t0.2(3.2)d1.1(0.9)Var(i)0.5(2.8)Cov(i,pb)(t12)+ɛwithɛMA(1.3,1.7,1.2,0.7)(11)
m=6(61.9)t+1.6(2.4)D(POIL)33.1(13.2)rUS+1.6(1.8)D(y)36.5(2.7)Var(y)+ɛwithɛMA(1,0.4)(12)
SNOM=0.2(30.1)t2.7(2.1)Var(i)1.6(9.7)rUS139.2(6.4)Var(rUS)2.5(3.4)Var(y)+ɛwithɛMA(0.9,0.6,0.2)(13)
SFX=0.2(5.9)t4.3(3.5)Var(POIL)+4.0(15.9)rUS+276.9(5.4)Var(rUS)+12.8(7.5)Var(y)+ɛwithɛMA(0.9,0.5)(14)

D. Country Comparisons

35. This section presents international comparisons of government debt structures. Indicators of the size, composition, maturity, and average cost of government debt are assembled for twelve emerging market countries and Canada. These countries have sovereign ratings in the same range as Mexico (e.g., Croatia, Malaysia, South Africa), a higher range (e.g., Korea, Poland), or a lower range (e.g., Brazil, Turkey). Canada serves as a reference for the debt structures of more economically advanced countries.

Coverage issues

36. The focus is limited to central government gross debt. While national authorities document the debt of the wider public sector, they generally do not provide a consolidated analysis of the debt structure. Government assets are not taken into account, as conventions for identifying assets that can be netted out and valuing them remain largely country-specific.36

37. A common definition of central government debt is used. The debt of public financial restructuring agencies is considered as government debt. Countries that experienced banking crises have often established agencies separate from the government to intervene in troubled banks. This has been the case in Colombia, Korea, Malaysia, Mexico, and Thailand. These agencies are typically responsible for recapitalizing or liquidating banks, taking over their non-performing assets, and providing assistance to bank debtors and depositors. These agencies’ debts, guaranteed implicitly or explicitly by the government, represent an additional debt burden and need to be taken into account in international comparisons. Other countries have generally included the debt resulting from financial restructuring in the government debt records (e.g., Turkey and the Philippines) or are in the process of transferring these liabilities to the government (e.g., Thailand).37 In the case of Chile, the debt of the central bank needs to be consolidated with that of the government. The central bank still holds debt that it issued to bail out the financial sector in the 1980s.38 In addition, the central bank has been the main issuer of public domestic debt, with the aim to support financial sector development through establishing benchmark issues. On similar grounds, the debt of extrabudgetary funds (e.g., FARAC in Mexico) needs to be counted as government debt, because their activities are generally budgetary in other countries. In the case of Mexico, the definition is extended to the debt of direct PIDIREGAS projects. By design, this debt is bound to be transferred to the government at the time of completion of the projects.

Debt levels

38. The size of the government debt is an important constraint on the quality of the debt structure. The previous section found a significant relationship between the improvement in the debt structure and debt reduction in Mexico. Country comparisons corroborate this finding. Higher levels of debt are generally associated with greater reliance on foreign currency and domestic indexed debt, and shorter maturities (Figures 13, 14, and 15).

Figure 13.
Figure 13.

International Comparisons: Size of Central Government Debt

(in percent of GDP)

Citation: IMF Staff Country Reports 2004, 418; 10.5089/9781451825626.002.A002

Figure 14.
Figure 14.

International Comparisons: Composition by Type of Instrument

(in percent)

Citation: IMF Staff Country Reports 2004, 418; 10.5089/9781451825626.002.A002

Figure 15.
Figure 15.

International Comparisons: Average Term to Maturity of Domestic Debt

(in years)

Citation: IMF Staff Country Reports 2004, 418; 10.5089/9781451825626.002.A002

39. While the size of Mexico’s government debt is lower than the middle range of the sample, the government’s aim at further debt reduction appears appropriate (Table 8). Mexico’s debt, at 47 percent of GDP at the end of 2003, is 5 percentage points lower than the median value of the fourteen countries compared. This is significantly lower than the level of debt observed in countries which have recently been hit by crises (e.g., Brazil and Turkey). Countries like Hungary, Malaysia, and Poland also combine similar or higher levels of debts with higher sovereign ratings, suggesting that the size of Mexico’s debt does not, in itself, prevent future rating upgrades. However, the government’s aim at continuing to reduce Mexico’s debt-to-GDP ratio is appropriate to bolster resistance to crisis.39 The short average maturity of domestic debt combined with the high share of debt indexed to short-term rate or foreign-currency denominated suggests a lower threshold of sustainable or “tolerable” debt for Mexico than for countries that have relied more on long-term domestic nominal debt.

Table 8.

International Comparisons: Size of Central Government Debt, As of End-2003

article image
Source: National Authorities Official Reports, IFS, and staff calculations.

See Moody’s Statistical Handbook, April 2004. Ratings are for long-term foreign currency borrowing.

40. Table 8 underscores the need to focus on broad measures of the debt. Narrow measures indicate an artificially low debt level for emerging market countries in Asia and Mexico, which manage financial restructuring outside the government sector.

Debt composition

41. The composition of the debt determines its vulnerability to refinancing risks and interest rate and exchange rate risks. Overall risk grows with reliance on foreign currency and short-term domestic instruments as opposed to long-term nominal domestic-currency denominated debt. Domestic debt of longer maturity, but indexed to short-term interest rates, protects the debt against refinancing and exchange rate risks but maintains exposure to the full interest rate risk.

42. Mexico is at an intermediate stage in terms of debt structure risk. Emerging market countries in Asia and Eastern Europe have been able to rely more on domestic-currency nominal debt and less on external debt and indexed domestic debt compared to Mexico. They have benefited from more developed domestic markets and, in some cases, a higher level of domestic savings.

43. More specifically, Mexico’s debt structure stands out in the following respects (Table 9):

Table 9.

International Comparisons: Composition of Government Debt by Type of Instruments, as of end 2003 1/

article image
Source: National Authorities Official Reports, IFS, and IMFstaff calculations.

Uses for some countries (e.g. Korea) a narrower definition of the debt due to data availability.

Computed by assuming that treasury bills are the only short-term domestic debt, when the information was not available.

Computed as the sum of known non-marketable securities and other debt and loans, when the information was not available.

  • Mexico’s share of external debt is in the middle range of the sample. Adjusting for the fact that other countries use foreign exchange indexed domestic debt, Mexico’s share is close to the median value of 33 percent. Canada, Poland, Hungary, and most of the East Asian countries rely substantially more on domestic debt than Mexico.

  • Mexico relies heavily on domestic floating-rate debt. On the positive side, Mexico has eliminated foreign-exchange indexed debt, which continues to be used in Brazil, Croatia, Turkey, and Chile.40 Mexico’s share of inflation-indexed debt is in line with the median value and only marginally higher than in Canada. However, Mexico’s use of debt indexed to short-term interest rates is the largest of the countries in the sample. Combining all types of indexation (to short-term interest rates, inflation, and foreign exchange), only Brazil and Chile have a higher share of indexed debt than Mexico. Other countries, in Eastern Europe and Asia, have been able to use fixed-rate bonds as their main debt instrument. Until recently, Canada had a policy of maintaining the share of fixed-rate debt at around two thirds (it was 63 percent in March 2003).41

  • The use of floating-rate and inflation-indexed debt has enabled Mexico to reduce refinancing risks in line with other countries, based on debt maturity. Mexico’s share of long-term domestic debt reached 60 percent in 2003 (based on original maturity), close to the median value of the sample. 42 It is higher than in Hungary and Poland, but lower than in Malaysia, Thailand or Korea. Again, this reflects the ability of the East Asian countries to issue large amounts of domestic nominal bonds.

  • Mexico’s external debt is more concentrated in securities than the other emerging market countries. Together with Malaysia, the significance of official financing in the external debt is the lowest.

Maturity structure

44. Indicators of average maturity and duration allow an assessment of exposure to refinancing and interest rate risks. With few exceptions (e.g., Turkey, Brazil, South Africa), countries in the sample have taken advantage of the recent period of low interest rates to increase the maturity of their domestic debt. Even Canada, presumably exempt from refinancing risks, has increased slightly the maturity of its debt, and had the second longest maturity of the sample in 2003.

45. Despite its rapid rise, the maturity structure of Mexico’s domestic debt43 is generally lower than in other comparator countries (Table 10). The average maturity of Mexico’s domestic debt was 2½ years at the end of 2003 compared to a median value of almost 4 years. Brazil and Poland have broadly similar maturity structures, based on average maturity and the shares of short-term debt by remaining maturities. Duration indicators, however, indicate that Brazil’s debt is re-priced more frequently than Mexico’s (11 versus 17 months), while Poland’s debt is more robust to interest rate changes. Colombia provides an interesting example of a country with, on the one hand, a higher level of debt and a higher share of foreign-currency debt, and, on the other hand, longer domestic debt maturity than Mexico.

Table 10.

International comparisons: maturity indicators, as of end 2003

article image
Source: National Authorities Official Reports, IFS, and staff calculations.

46. The case of Turkey shows the potential impact of nonmarketable debt on maturity measures. The average maturity of Turkey’s domestic debt was only half a year lower than Mexico’s at end-2003. More detailed data indicate that this is due to the structure of the nonmarketable debt (i.e. the bank restructuring debt). Focusing on marketable debt, Mexico’s average maturity was more than twice as high. The use of non-marketable debt has also contributed to the higher maturity of domestic debt in Korea—where national housing bonds, which are not traded, have maturities up to 20 years whereas the maximum maturity of treasury bonds is 10 years.

47. The maturity structure of Mexico’s public external debt compares favorably to other countries. The average maturity of government external bonds in Mexico (10 years) is longer than in Colombia, Brazil, and South Africa. It is similar to or lower than the maturity of overall external debt in Thailand and Malaysia. In the case of Thailand, however, this reflects reliance on bilateral and multilateral loans, which have very long maturities (around 20 years and 10 years respectively for the consolidated public sector as of March 2004). The maturity of Thailand’s external bond issues (consolidated public sector), at 4.2 years, is less than half the maturity of Mexico’s government external bonds.

Debt costs

48. An important issue is the interest rate premium associated with establishing a safer debt structure. The evolution of domestic debt costs has already shown a widening gap between the average cost of domestic debt and short-term rates, as Mexico strengthens its debt structure. However, in time, the strengthening of the debt structure may help reduce interest costs, as perceptions of country risk diminish and domestic markets become more liquid. Hence the rise in debt costs would be a temporary phenomenon.

49. The situation in other countries suggests that there are little or no long-term costs, in terms of higher interest payments, for improving the debt structure, while there are clear benefits in terms of reducing vulnerability (Table 11). Effective debt costs do not appear to be higher in countries relying less on short-term, floating-rate, or foreign debt than Mexico. These costs, estimated by the ratio of interest payments in 2003 to the average of the debt stock (domestic and external) at the beginning and end of the year, were lower in Poland, Korea, Malaysia, Thailand, and Canada than in Mexico. This finding remains true when considering separately the domestic and external debt and adjusting for inflation (with the caveat of missing data for several countries). Two factors need to be noted. First, lower inflation contributes to reduce debt costs. As anti-inflationary credibility is established, real interest rates tend to decline and a country is able to issue medium to long-term domestic debt at lower interest rates. Poland is a good example of the benefits of disinflation (Figure 16). Second, Mexico’s relatively high external debt costs suggest a longer lag in the transmission of world interest rates than in other countries. Of course, endogeneity is an issue in making these simple comparisons, as countries benefiting from low financing costs for exogenous reasons (e.g. EU accession) are more likely to achieve low-risk debt structures.

Table 11.

International Comparisons: Effective Cost of Central Government Debt, as of 2003 1/

article image
Source: National Authorities Official Reports, IFS, and staff calculations.

Computed as the ratio of annual interest payments to the average of the debt stock outstanding at the beginning and at the end of the year. Includes commission and fees for Mexico. Uses end-of-period exchange rates.

As of 2002.

As of fiscal year 2003/2004.