Benchmark - Brazilian Case

Flaviaf

Description

A BENCHMARK FOR PUBLIC DEBT: 1THE BRAZILIAN CASE NATIONAL TREASURY OF BRAZIL PUBLIC DEBT STRATEGIC PLANNING DEPARTMENT Technical staff: 2Rodrigo Silveira Veiga Cabral 3Mariana de Lourdes Moreira Lopes 4William Baghdassarian 5Luiz Fernando Alves 6Pedro Ivo Ferreira de Souza Junior 7Antonio Tiago Loureiro A. dos Santos Draft Version Brasília - DF, Brazil October 2008 1 We are thankful to Otavio Ladeira de Medeiros, Anderson Caputo Silva and all the staff from the Public Debt Secretariat for his contributions and acknowledge the technical support of all staff from the Public Debt Strategic Planning Department, especially from Lena Oliveira de Carvalho and Braulio Santiago Cerqueira. The process and tools described are result of the work and intellectual contribution of a broad group of Treasury analysts. 2 Doctor in economics by Universidade de Brasíia. Subdirector of the Brazilian Public Debt Strategic Planning Department. 3 Doctoring in economics at Universidade de Brasília. Brazilian National Treasury Analyst. 4 Doctoring in finance at University of Reading . Brazilian National Treasury Analyst. 5 Master in economics by Universidade Federal de Minas Gerais. Head of the Risk Management Unit of the Brazilian Public Debt Department. 6 Executive MBA at Fundação Getúlio Vargas. Brazilian National Treasury Analyst. 7 Master in economics at Universidade de São ...

Informations

• Publié par: Mephar
• Nombre de lectures: 35
• Langue: English

Extrait

     

         

A BENCHMARK FOR PUBLIC DEBT: THE BRAZILIAN CASE 1                 NATIONAL TREASURY OF BRAZIL   PUBLIC DEBT STRATEGIC PLANNING DEPARTMENT     Technical staff: Rodrigo Silveira Veiga Cabral 2   Mariana de Lourdes Moreira Lopes 3   William Baghdassarian 4   Luiz Fernando Alves 5   Pedro Ivo Ferreira de Souza Junior 6   Antonio Tiago Loureiro A. dos Santos 7  

Draft Version

Brasília - DF, Brazil October 2008  

                                                  1  We are thankful to Otavio Ladeira de Medeiros, Anderson Caputo Silva and all the staff from the Public Debt Secretariat for his contributions and acknowledge the technical support of all staff from the Public Debt Strategic Planning Department, especially from Lena Oliveira de Carvalho and Braulio Santiago Cerqueira. The process and tools described are result of the work and intellectual contribution of a broad group of Treasury analysts.   2 Doctor in economics by Universidade de Brasíia. Subdirector of the Brazilian Public Debt Strategic Planning Department. 3 Doctoring in economics at Universidade de Brasília. Brazilian National Treasury Analyst. 4 Doctoring in finance at University of Reading . Brazilian National Treasury Analyst. 5 Universidade Federal de Minas Gerais. Head of the Risk Management UnitMaster in economics by of the Brazilian Public Debt Department. 6 Executive MBA at Fundação Getúlio Vargas. Brazilian National Treasury Analyst. 7 Master in economics at Universidade de São Paulo. Deputy Head of the Research and Development Unit of the Brazilian Public Debt Department.

ABSTRACT

  A benchmark for public debt: the Brazilian case Brazilian National Treasury Public Debt Strategic Planning           Public debt management has gained prominence over the last years. The pursuit of an adequate public debt composition is critical to the objective of minimizing long term financing costs, while ensuring the maintenance of prudent risk levels. In this sense, we devise a benchmark for the public debt, defined as a long term, optimal debt structure, to serve as a guide for short and medium term borrowing strategies. The proposed benchmark builds on stochastic finance methods to construct an efficient frontier for public debt. Such a frontier indicates which debt compositions are efficient from a cost/risk trade-off standpoint. In short, this paper aims at formulating a consistent theoretical public debt benchmark model, thus creating the possibility of choosing an optimal debt composition to serve as a guide for public debt management decisions. For those interested in the history of debt management practices in Brazil, we present in the appendix an account of its institutional evolution and tools used, of which the benchmark is the latest addition.     Keywords: Public debt management, Simulation models, Benchmark, Efficient Frontier JEL Classification: H63, E17, E44, E47, E63, G11, G28            

 

 2

1. Introduction     In the recent past, international markets have experienced several episodes of financial turmoil. Those episodes have proved particularly damaging for the emerging markets government finances, since the sharp increase observed in both interest and exchange rates have had disastrous effects on the public debt. As a consequence, debt sustainability and risk management have become key policy issues, given the instability that has been cogent to the international financial system. In fact, over the years, the subject ofpublic debthas received increasing attention due to the perverse effects of a high and volatile debt on economic activity.   From a theoretical perspective, the starting point of public debt management is the Ricardian Equivalence proposition 8 . A major implication of this proposition is that, under certain conditions, an active public debt management would be irrelevant 9 . However, the assumptions underlying this result are quite restrictive. Once a more realistic setting is adopted, a proper management of public debt becomes of critical importance, in terms of improving the government financing conditions and the operation of the local financial markets. Moreover, tax smoothing, credibility and policy signaling issues are also involved.   Last decade has witnessed extensive progress in the discussion regarding public debt management. This discussion was led by those countries with larger experience in the subject, with support from the World Bank and the International Monetary Fund 10 . From the variety of available strategic planning and risk management instruments, one which gained prominence was the adoption of a benchmark. A benchmark for public debt is understood as an optimal long term debt structure, which serves as a reference for short and medium term borrowing strategies.   Several countries, most notably Portugal, Sweden and Denmark, have developed and implemented a benchmark model for the public debt. Its practical implementation requires, firstly, approval by the superior instance of debt management, usually the                                                   8 Barro (1974, 1989) 9 Those conditions will be discussed next. 10  See the IMF/World Bank jointly published document “Guidelines for Public Debt Management” (2001 and 2003)

 

3

Ministry of Finance or an executive committee. It is then published and must, accordingly, be pursued by the competent agency.   In Brazil, the National Treasury, within the Ministry of Finance structure, is in charge of public debt management 11 . The National Treasury has undertaken great efforts over the years to integrate the group of countries most advanced in public debt management. An important step in this sense is the present formulation of a benchmark for the Brazilian public debt.   This article sets forth a benchmark model based on methods of stochastic finance and backed by efficient portfolio theory. It is structured as follows: after this introduction, section 2 briefly comments on the theoretical importance of public debt management and on the international practice, with a stress on benchmark design and implementation. In section 3 we outline our own model, which is based on stochastic finance and efficient portfolio theory. In section 4 we present the results, while section 5 concludes. An appendix presents a brief history of debt management in Brazil, in which issues related to institutional design, analytical tools and policy making are emphasized. Its aim is to illustrate that the benchmark model is, in fact, the latest chapter of an ongoing history.       2. Literature and International Experience     2.1. Public Debt Management     According to the IMF/World Bank jointly published “Guidelines for Public Debt Management”, public debt management is “the processof establishing and executing a strategy for managing the government’s debt in order to raise the required amount of funding, achieve its risk and cost objectives, and to meet any other sovereign debt management goals the government may have set”. Pubilc debt management differs

                                                  11 In many countries, as is the case of Brazil, public debt management is in charge of the Ministry of Finance. In other cases, there is an independent debt management office. Lastly, in other rather unusual cases, debt management is carried within the structure of the Central Bank.

 

 4

from fiscal policy insofar as the fiscal result affects thelevelof public debt, while debt management aims at altering itscomposition, so as to reduce debt volatility.   The importance of public debt management can be best understood against the theoretical background provided by the Ricardian Equivalence proposition. This approach begins with the observation that government expenditures must be financed either by taxes or by debt. More specifically, the intertemporal budget constraint faced by the government requires equality between the present value of expenditures and the present value of government (tax or debt) revenues. Given a fixed path of public spending, a current tax reduction generates deficit and debt, which must eventually be backed by a future tax increase, so as to keep the present value of revenues constant. Hence, the choice of any possible debt and tax combination is immaterial, given fixed funding needs. In other words, debt and taxation are equivalent from an intertemporal 2 viewpoint 1 .   As it turns out, the Ricardian Equivalence can be viewed as a debt-neutrality proposition. Its result depends, however, upon the following three hypotheses: (1) infinite planning horizon; (2) complete markets and (3) non-distortionary taxes 13 . However, those hypotheses are quite restrictive and, more importantly, do not seem to represent adequately the real-world environment faced by the policy maker. Once those hypotheses are relaxed and a more realistic setting is adopted, real effects of public debt become apparent, thus attracting the attention of both specialists and the fiscal authority to the themes of the adequate level and composition of public debt.   An active public debt management aims, precisely, at altering the public debt profile. As such, it is important for a number of reasons. The foremost objectives of public debt management in Brazil are minimizing long-term financing costs while ensuring the maintenance of prudent risk levels, as well as contributing to the smooth operation of the public bonds market 14 . Let us comment briefly on that. In fact, if there were perfect information and arbitrage, the goal of minimizing financing costs would be of

                                                  12 Missale (1999) develops an extension of the Ricardian equivalence to the composition of public debt. 13 Barro (1974, 1989). 14 Annual Borrowing Plan (2007).

 

5

15 no use, for all debt instruments would be equivalent to the public in terms of return . Since, as argued, this is not the case, there is room not only for working on the debt profile but for enhancing the market as well, by means of better liquidity and information conditions. Improving the secondary market, developing the term structure of interest rates, enlarging and diversifying the base of investors are all goals of public debt management in Brazil.    As one can see, the theoretical literature on the importance of debt management does not reduce to the discussion of the Ricardian Equivalence hypotheses, though. There are insights concerning time consistency and credibility of macroeconomic policies, policy signaling, optimal taxation, and real effects of sovereign defaults, among others 16 .   Credibility and signaling issues refer, for instance, to the use of debt instruments to transmit commitment with respect to the objectives of economic policy. An example of that would be the issuance of inflation-linked bonds under an inflation targeting regime, or else, the use of FX-linked bonds under a fixed exchange-rate regime. In fact, an efficient debt management may also aid in stabilizing, if not reducing, the debt to GDP ratio, insofar as it aims at reducing debt volatility and vulnerability to shocks in market variables such as the exchange rate or the overnight interest rate.   Additionally, an active public debt management contributes to budget volatility minimization and tax smoothing purposes. A less volatile debt is much less bound to impose severe shocks on the government budget or on tax collection necessities. More precisely, the literature on optimal taxation suggests the government should structure its debt so as to smooth taxes over time and under different states of nature 17 . Another approach suggests the optimal choice of debt instruments depends on the structure of the economy and on the nature of the shocks that affect the debt.   As regards empirical studies on the default costs, they emphasize the real costs in terms of produce and job losses that result from the economic contractions which                                                   15 Goldfajn and de Paula (1999). 16   Ibid. 17 See, for example, Bohn (1994) and Barro (1999)

 

6

typically occur after a default. In that sense, the government should have low tolerance to risky debt structures and the main objective of the debt manager would be to obtain a portfolio structure that minimizes default risk 18 .   Therefore, given the main objective of minimizing long-term borrowing costs while ensuring prudent risk levels, the National Treasury has been developing and refining instruments with the intent of improving public debt management. Since debt management entails essentially working on its composition in terms of maturity and type of return 19 , it is natural to ask which should be the ideal debt structure to be pursued as a long term goal. The herewith proposed benchmark model aims at offering a consistent answer to this question.   2.2. Some methodological issues   The benchmark model proposed on this paper is based on stochastic finance and efficient portfolio theory. However, there are some restrictions one should bear in mind when applying traditional financial analysis instruments to government policy. Broadly speaking, the government may have objectives more complex than reducing costs subject to prudent risk levels. Moreover, the evolution of the government cash- flow and indicators of budget impact may have implications on the choice of debt portfolio. Finally, given the nature of public debt, the government has a strong influence over bond prices and, as a consequence, over the cost and risk of its borrowing strategies. As a result, those peculiarities may lead the fiscal authority to pursue a composition not on the efficient frontier.   Secondly, it is important to comment on which would be the proper debt concept to work with. The National Treasury, in fact, has direct control over the Federal Public Debt only, which comprises all debt in bonds issued onshore and offshore by the National Treasury 20 , in addition to all Federal Government contractual debt 21 . However, the most widely used indicator is the Net Public Sector Debt to GDP ratio                                                   18 See Dooley (1998) or Sachs, Velasco e Tornell (1996) 19 The return of National Treasury bonds may be of four different basic types: fixed rate; floating rate (overnight interest rate indexed); inflation-linked; and foreign currency linked. 20  Central Bank had in the past the ability to issue its own debt. However, the 2001 Fiscal The Responsibility Law forbade it to do so. 21 Presently all contractual debt is external debt.

 

7

(NPSD/GDP). This the most comprehensive debt concept, for it comprises all liabilities net of liquid assets from the whole Public Sector, hereby defined as the Federal Government (including the Social Security Institute) plus the Central Bank, local government (states and municipalities) and public enterprises.   Our benchmark model considers the NPSD/GDP as the relevant concept. This view is backed by intertemporal government budget constraint analysis, for all of the government’s assets and liabilities must be taken into account in order to evaluate the sustainability of the fiscal position 22 . In fact, most economic analysts and market participants consider it to be the relevant indicator of debt sustainability, in which they are supported by international organisms such as the IMF and rating agencies such as Standard and Poor’s 23 .    Finally, a comment on the steady state premise is due. Our benchmark model presupposes the economy is already at steady state, which means that all variables are at their long run, equilibrium values. Indeed, this should arguably be the proper setting in a discussion about an ideal, long-term optimum debt structure to be pursued. Hence, our reference steady state scenario comprises the following characteristics: stability in the overall economic environment, reduced fiscal vulnerability, investment grade rating, lower domestic interest rates, inflation under control and sustained economic growth. This scenario is to be attained possibly in the next few years, in such a manner that our analysis should be interpreted to proceed from that point of time onwards.                                                         22 In addition to that, one should observe that minimizing the Federal Public Debt to GDP ratio does not imply minimizing the NPSD/GDP ratio, mainly because the former does not consider the government asset structure. Nevertheless, our results do not change considerably when the Federal Public Debt is used, given the strong effect of the GDP on both measures. 23 Standard and Poor’s Net Debt concept is not identical to the NPSD, however. Other rating agencies   such as Moody’s and Fitch use an alternative indicator, the general government debt to GDP. This is the indicator countries most frequently publish, so it has the advantage of making international comparisons easier. We understand, however, this is not the best indicator of fiscal sustainability, for two basic reasons: (i) it does not take into account the government’s assets; and (ii) it does consider the Treasury bonds used by the Central Bank for monetary policy, which we do not think to be appropriate, for in this case the Treasury’s liability is an asset held by Central Bank, what cancels out from the standpoint of the public sector as a whole. Furthermore, any debt which matures while in possession of the Central Bank is automatically replaced by the Treasury according to market rates, so as to recompose the portfolio used for open market operations.  

 

8

2.3. The international experience     According to the “Guidelines” (2001), the benchmarkcan be a powerful managerial tool as it represents the debt structure that the government would like to have, based on its risk preferences and expected costs. As such, the benchmark is conceived to serve as a guide regarding debt issuances and risk management.   Many countries use or are developing methodologies for the determination of an optimum debt structure. In this section, we briefly sketch the experience of some of those countries with respect to the subject.   One of the first countries to adopt a benchmark was Portugal (Matos 2001 and Granger 1999). The benchmark model developed by their Institute of Public Credit Administration adopts a pure liability management outlook, since assets are not taken into account.   The benchmark is derived from a cash flow simulation model, which impacts on budget volatility and, hence, restricts the policy maker’s degrees of freedom. The model has three basic inputs: i) statistical simulation of interest rates; ii) different borrowing strategies, which comprehend predefined rollover and management needs; and iii) deterministic scenarios for other macroeconomic variables.     In Sweden (Bergström and Holmlung 2000), debt management is in charge of the Swedish National Debt Office. They model five macroeconomic variables: inflation, GDP, short and long term interest rates and the real exchange rate. Different compositions are evaluated in view of borrowing needs. They separate scenario risk from time series risk, and cost is measured in nominal terms and as a percentage of GDP, always based on cash flows.   In Ireland debt management is carried by the National Treasury Management Agency. They use the net present value of debt as cost measure, and the fiscal volatility as risk measure. Moreover, they defend the benchmark should not change much over time, as it reflects the structural conditions of the economy as well as the ultimate objectives of fiscal policy.   

 

9

Denmark is one of the most advanced countries regarding public debt management. However, differently from the recommended practice, the Danish agency is located within the Central Bank structure. They do not have a benchmark composition but, rather, work with medium term duration targets. Given the current portfolio and the duration and cost-at-risk (CaR) 24  they arrive, by trial and error (no objectives, optimization), at a long-term structure. The Danish agency emphasizes they have already used complex models in the past and, now, value simplicity and transparency.   Other countries, such as England and Belgium, for instance, do not determine benchmarks. However, those countries also make use of simulation models and indicators such as CaR, CFaR (cash flow-at-risk) and Bar (Budget-at-Risk) for risk management. England also stresses the importance of maintaining simple and tractable models.     Evidently, it has not been our purpose to fully discuss the experience of all countries here, but to present a short account of some important features. More details about the international experience can be found in the Guidelines (2003) and in Nars (1997).   It should be instructive to point out some of the weaknesses of the models aforementioned, several of which were discussed by their own authors, indeed. The Portuguese model, for instance, has a high degree of dependence on the path of economic variables from the present to the hypothetical steady state. Furthermore, the steady-state hypothesis is of a constant nominal debt value, which does not seem very intuitive. Finally, it lacks guarantee of macroeconomic consistence, since the interest rate is generated by a stochastic model while the other variables depend on deterministic scenarios.       Regarding the Swedish model, there does not seem to be a theoretical justification for all the stochastic processes to be auto-regressive of order one. The association of those processes with a Taylor rule evidences the fact the model is a type of mixed stochastic finance and macro-structural model. Moreover, there is no connection between the short and long-term interest rate generation processes. Additionally,                                                   24  Danish debt management has introduced this measure, which is a value-at-risk (VaR) version The applied to debt. More on that in Danmarks Nationalbank (2001).

 

10

contrary to what most debt managers now believe, the model indicates that inflation- linked bonds are not good financing instruments, and that could reflect problems in the modeling process.   Denmark’s model considers domestic debt only. In fact, only by coincidence would a benchmark exclusively focused on domestic debt be optimum for debt management as a whole. Besides, the modeling aims at establishing debt duration goals, which may not be very intuitive and is bound to conflict with other objectives. Finally, Denmark also mixes stochastic with deterministic scenarios and it is not clear how consistence would be attained.   The model that we propose in the next section of this paper tries to work on the solution of some of those problems, even though, obviously, it must incur in several other simplifications and is subject to its own weaknesses, as we shall discuss further on.       3. Modeling the benchmark: a stochastic finance approach 25          The general idea of the model is quite simple. Broadly speaking, we use simulation methods in order to derive an efficient frontier for the public debt. A Federal Public Debt (DPF) composition is efficient when its associated cost is the lowest given any chosen risk level – or, alternatively, when risk islowest for any given cost level. The efficient frontier is defined as the set of all such compositions, thus reflecting the cost/risk trade-off faced by the debt manager, since, at the frontier, it is only possible to improve one variable at the cost worsening of the other.

Let us describe briefly how the model works. We firstly fix a given debt profile, based on a chosen composition of representative DPF bonds which differ in terms of both return type and maturity. Bond prices evolve according to stochastic processes which drive key macroeconomic variables (interest rates, exchange rates and inflation). Despite being stochastic, the equations are correlated to ensure macroeconomic consistence. Under the hypothesis that the chosen debt composition will be                                                   25 See Cabral (2004)  

 

11