Spain
Unemployment, Debt Management, and Interest Rate Differentials
  • 1 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

The IMF Working Papers series is designed to make IMF staff research available to a wide audience. Almost 300 Working Papers are released each year, covering a wide range of theoretical and analytical topics, including balance of payments, monetary and fiscal issues, global liquidity, and national and international economic developments.

Abstract

The IMF Working Papers series is designed to make IMF staff research available to a wide audience. Almost 300 Working Papers are released each year, covering a wide range of theoretical and analytical topics, including balance of payments, monetary and fiscal issues, global liquidity, and national and international economic developments.

I. Introduction

“The markets are penalizing Spain not only for its high public deficit and debt, and its resistance to reducing the inflation rate, but also because of having lost the political capacity to manage the country’s problems.” (El Pais, Negocios, page 16, March 12, 1995, commenting the developments since December 1994)

National governments may face high financing costs either because they show no intention to correct existing imbalances or, when committed to correct such imbalances, the market doubts their ability to do so. In principle, there is a distinction between the two types of governments, unreliable versus committed. However, in practice, once the policies to manage “the country’s problems” have failed, the true ex-ante government intentions become largely irrelevant. The quote above seems a good example of this identification problem. It provides an epilogue to the several stabilization attempts, aimed at controlling the fiscal imbalance and reducing the inflation rate, which failed in spite of the seemingly sincere commitment of the Spanish authorities to reach by 1997 the targets of the Maastricht Treaty.

This study is about the market’s assessment of that policy commitment as reflected in the interest rate premium on long-term debt instruments. The latter has been among the highest when compared with other, more developed but highly indebted, countries in Europe. The analysis extends from the last quarter of 1989 to the crisis of September 1992, a period when there was a strong commitment to keep the Spanish peseta under the rules of the EMS.

Even though it considers the standard explanations for the interest rate premium, such as the level of outstanding debt and the degree of fiscal imbalance, this paper focuses on two particular aspects of the Spanish economy that may be acting as constraints to policy credibility. These are the high unemployment rate and the lengthening of the maturity of the debt pursued by the Spanish authorities in the recent past. A high unemployment rate detracts from the credibility of a stabilization program as the concomitant additional destruction of jobs would worsen the initial level of unemployment. On the other hand, a deliberate attempt to increase the maturity of peseta-denominated debt may have undesired signaling effects in regard to the inflation target. Though such a policy could strengthen the credibility of a fixed exchange rate, anybody could still wonder what might move a government that purports to be committed to lowering inflation below the prevailing rate, to willingly pay the premium the market demands for longer maturities.

This study investigates the determinants of the servicing cost for the Spanish public debt taking into account the existence of these additional constraints. The findings intend not only to help explain the cause of interest rate differentials in Spain vis-à-vis other European countries, but also to reveal the responsiveness of such premia to policy actions in Spain. If the current level of unemployment is actually acting as a constraint to policy credibility, then the signaling effect of a fiscal correction could be mild unless it is perceived as genuinely permanent. Or worse, the use of monetary policy as disciplinary device could be extremely costly unless accompanied by permanent and transparent fiscal corrections. Similarly, if debt management has an important impact on the market’s perception of policy determination, then it should be an integral feature of any comprehensive stabilization attempt.

The paper is organized as follows. Section II discusses some theoretical aspects of risk premia, distinguishing the concepts of sovereign credit risk and currency risk, and elaborates on the credibility aspect related to this topic. The following section describes the current situation in Spain, including a cross country analysis of interest rate differentials and general economic performance, while section IV provides a summary on recent economic developments on debt management in Spain.

After the background sections, section V outlines an integrated model illustrating the issue of credibility discussed in section II, where special emphasis is placed on distinguishing policy credibility from the credibility of policymakers, following the approach introduced by Drazen and Masson (1994). This model is empirically tested in section VI for the period when Spain adhered fully to the EMS and section VII deals with model limitations and possible extensions to the results presented in section VI. Section VIII concludes.

The evidence presented in sections VI and VII provides support to the view that policy credibility in Spain is not only a question of policymaker’s type, but also of policy viability. In the case of Spain this limitation is mainly given by its massive unemployment. This implies that anti-inflationary attempts could be very costly, even if carried out by the most conservative policymaker. Clearly, it also reinforces the need of acting directly on structural variables and casts doubts on the usefulness of the sole action of monetary or exchange rate policies in helping resolve the fiscal conflict. No conclusive evidence is found in regard to the effect that the increase in debt maturity may have had on the interest rate premium; although it would seem that, in the Spanish case, it may have helped reduce it.

II. Interest Rates Differentials and Risk Premium: The Foundations

Arbitrage conditions in a world with free movements of capital flows and deep international capital markets ensure the inter-link of expected returns on different types of investment opportunities all over the world. In such an environment, differences in borrowing costs facing national governments can be broadly explained by perceived differences in sovereign credit risk and currency risk. Credit risk accounts for the possibility of outright or partial default, the latter including any change in the asset’s expected return not due to currency effects (i.e., measures such as debt rescheduling, changes in tax treatment, or policies that affect returns through Ricardian equivalence in public finance). Currency risk refers to changes in the purchasing power of the relevant currency, in terms of the investor’s consumption basket, during the lifetime of a particular asset. Evidently, expected inflation and exchange rate changes, both interdependent phenomena, constitute the currency risk.

In the light of the situation inherited from the past, the perception of these two types of risks represents the market’s expectation of future policy measures consistent with a trusted debtor. Policymakers’ credibility in this area is not free from the standard restraints in macroeconomics; e.g., the tension between ex ante and ex post optimal policies. For example, the government’s incentive to default on the national debt, a situation similar to the capital-levy problem in taxation policy, is manifested in a temptation to reduce the debt’s real return and results in the well known monetary and fiscal policy inflationary bias. 2/ Unquestionably, any problem of time inconsistency could be efficiently solved by utilizing a commitment technology. However, policymakers rarely possess such technology and a common approach used to deal with this problem is to set incentive constraints that satisfy forward looking private agents with rational expectations. In other words, policymakers make their policy announcements credible by imposing initial conditions that would discourage or prevent them from deviating from what had been announced.

The risk of outright and partial default of the public debt is directly linked with the possibility that the service of the current debt, in the prevailing conditions and in any future state of nature, will not be met. In some circumstances, the government’s incentive to honor completely its liabilities may be weak, as the stock of debt is given at any point in time and the government can take advantage of this fact to maximize revenue and avoid using other distortionary taxes. Therefore debt repudiation would always be a policy alternative for national governments unless they face some enforcement mechanism.

Some authors argue, for instance Grossman and van Huyck (1988), that reputational forces—such as the threat of not having access to the international market after a major default—are likely to limit the debtor country’s incentive not to honor its debt. 3/ Nevertheless, it is reasonable to think that the higher the debt outstanding and the more painful is fulfilling its service, the greater the incentive is to repudiate it. 4/ Thus, a high credit risk country is usually identified with high initial debt, high expected future fiscal deficit, poor collateral, weak enforcement of current regulations (like tax regime) and general lack of constraints on the government’s discretionary power. These would be the indicators that would need to be strengthened by a committed government which wants to reduce the cost of its debt.

The currency component of the risk premium relates to the time inconsistency of monetary and fiscal policy that may engender an inflationary bias. Preannounced inflationary targets may not be credible as policymakers have an ex post incentive to pursue expansionary policies along the short run Phillips curve or to maximize revenue from non distortionary sources—i.e., by reducing the real value of the national debt and the associated future flows of distortionary taxes. 5/

The literature on monetary policy credibility has usually emphasized the role of government’s type (for example, the relative weights it puts on losses from inflation versus the gains from employment). In this approach, introduced into macroeconomics by Backus and Driffill (1985a, 1985b), as government’s preferences are unobservable, the observed policy becomes the main piece of information for private agents to update their expectation of future policies. Therefore, similar to the case of credit risk, the government’s intention to reduce currency risk is realized by signaling and building up reputation.

This reputation building could take different forms, ranging from informal commitments to institutional arrangements. For example, there is a well known advantage in delegating monetary policy to a conservative central banker—someone who puts a higher weight on inflation relative to employment in his loss function than society’s true weights. 6/ Another institutional set up that enhances credibility is a fixed exchange rate in a multilateral agreement, topic that received attention with the pioneer work by Giavazzi and Pagano (1988) about the advantages of the EMS. 7/

Though appropriate, the representation of the so called “credibility game” sketched above suffers an important shortcoming: the assumed independence of government actions to the state of the economy. As it was indicated in the introduction, the concept of credibility is applicable to the policies themselves as well as to the policymaker’s type. This is what, in a recent paper, A. Drazen and P. Masson (1994) do. Their concept of credibility has two components: “the private sector’s assessment of the government’s type, and also, given the type of the policymaker, an assessment of the probability that an optimizing government will actually decide to carry out its announced policies in the face of adverse shocks”. They discuss the issue of policy elections that have persistent effects on the economic environment and show that some policies can constrain the future set of policy tools, and following them may actually harm rather than enhance credibility.

Because of its massive unemployment, the situation in Spain would appear to be better represented by a Drazen-Masson approach, where the credibility of the government’s type could always be jeopardized by the state of the economy. In this context, the authorities’ attempt of an anti-inflationary policy faces the possibility of a perverse result. If this policy is not fully credible, it is likely to worsen the current disequilibrium in the labor market and, given the initial magnitude of such disequilibrium, this outcome would ultimately not be acceptable. This fact is known by the market and this, in turn, reinforces the prior perception that the policy is not credible.

Also related to the particularities of the Spanish case is the role played by debt management in the so called “credibility game”. Although government finance is usually associated with the Modigliani-Miller proposition (the irrelevance of the financing source, in particular, and debt management, in general), it is well known that this theorem breaks down when there are incentive problems between the government and the economic agents. 8/

Debt management can serve as an important incentive device. For example, issuing foreign currency denominated debt or indexed debt could reduce currency risk. Similarly, short term debt might limit the incentive problem, as inflation would translate immediately into interest rates and this would reduce the gains from inflationary surprise, 9/ however short debt maturity might lead to counteracting effects as well. As it is noted in the literature on confidence crisis, 10/ countries with a fixed exchange rate may be exposed to self-fulfilling financial crises that have to be managed either by raising interest rates to defend the currency or accommodating a depreciation of the currency by monetizing some public debt. Clearly in these cases, foreign currency denominated debt will reduce the incentive to devalue, making the exchange rate policy more credible. On the other hand, concentrated debt in the short term would make confidence crises more likely, as it is more costly for the government to handle them within the committed exchange rate.

III. Interest Rates and Other Economic Indicators in Spain

As is shown in Chart 1, debt servicing costs in Spain have been among the highest when compared with other more developed but highly indebted countries in Europe. In a first approach to an explanation, a set of economic indicators are presented in Table 1 for Belgium, Italy, Ireland, Germany and Spain. Germany as a benchmark and Belgium, Italy and Ireland as contrasting cases of highly indebted countries. Chart 2 shows the behavior of the Spanish Peseta.

Chart 1
Chart 1

Selective European Countries Yields on Long-Term Debt, 1989 - 1994

(Differential with respect to German benchmark; in percent)

Citation: IMF Working Papers 1995, 107; 10.5089/9781451946819.001.A001

Sources: IMF: International Financial Statistics; and Central Bank of Spain.
Chart 2
Chart 2

Spain: The Peseta

(January 1989=100) 1/

Citation: IMF Working Papers 1995, 107; 10.5089/9781451946819.001.A001

Sources: Central Bank of Spain and IFS.1/ A decline in the index indicates a depreciation of the Peseta.
Table 1.

Selected European Countries: Comparative Economic Indicators, 1989-94 1/

(In percent of GDP uniess otherwise Indicated)

article image
Source: International Monetary Fund, World Economic Outlook: and Staff estimates.

Figures for 1994 are Fund staff estimates.

In percent, annual average.

Filename: comp_ei.TAB

Date: May 3, 1995

Bearing in mind the discussion presented in the previous section, the findings in Table 1 are somehow disappointing. Although growing, the Spanish government’s outstanding debt is still relatively small. 11/ In terms of GDP, it went from 49 percent in 1989 to around 64 percent in 1994, less than half that of Belgium or Italy and substantially smaller than that of Ireland, which is the only among these countries that has succeeded in reducing its stock of debt. When looking at prospects of the indebtedness situation, the Spanish case does not appear in isolation either. Government deficits are considerable in Belgium, Italy and Spain, but Spain still seems to have more room to manoeuver in consolidating the fiscal imbalance, given its lower revenue ratio. 12/ Additionally, given the higher investment ratio experienced in Spain and its relative economic backwardness, a higher rate of growth in Spain than in the other countries could be expected, which could alleviate the debt burden at a more rapid pace.

Moreover, the Spanish government quality as a debtor, as measured by the usual financial ratings, Standard & Poor and Moody’s does not reveal a considerable credit risk on Spain’s debt. For instance, in October 1994, Moody’s classification of Spanish government bonds was Aaa in pesetas and Aa2 in foreign currency. This is also reflected in the tiny premium paid by Spanish debt denominated in foreign currency, which is around 10 to 20 basis points. So, what is wrong with the Spanish debt?. The next sections attempt to provide an answer to this question.

IV. Recent Developments on Debt Management in Spain

By 1989 the market for public debt in Spain could be deemed to be fully developed. 13/ The market seems transparent and competitive enough to guarantee cost minimization, while adequate volume of trade provides a deep secondary market. Therefore, institutional changes do not appear to offer much potential for reducing debt financing costs. 14/ This notwithstanding, following the institution of restrictions on credit from the Bank of Spain in 1989, the Spanish authorities have pursued a number of initiatives in the area of debt management.

In order to reduce the risk associated with growing financial needs, the Treasury worked to increase the average maturity of the debt. By reducing the volume maturing every year the re-financing risk was narrowed and, supposedly, the swings in total debt service were lowered. Accordingly, the share of medium and long term debt instruments—Government Bonds and Obligations—grew rapidly since 1989, from 27 percent to 54 percent in 1993 (See Table 2, Table 3, and Chart 3). Similarly, there was an increasing use of foreign denominated debt, though the bulk of public debt continued to be denominated in pesetas. The proportion of public debt denominated in foreign currency increased from 3 percent in 1989 to 9 percent in 1993. And finally, a not less consequential feature in the market for public debt, although not directly connected with a specific policy, was the significant increase of non-resident holdings. These holdings, concentrated mainly on long term bonds, increased from 8 percent of total long term debt in 1989, to 41 percent by 1993 (See Table 4).

Chart 3
Chart 3

Spain: The Maturity of Public Sector Liabilities

(In percentage of GDP)

Citation: IMF Working Papers 1995, 107; 10.5089/9781451946819.001.A001

Source: Central Bank of Spain
Table 2.

Spain: Central Government Financial Liabilitties.

(In billions of pesetas)

article image
Source: Central Bank of Spain, Boletin estadistico.Filename: ESPbl218.TABDate: February 24, 1995.
Table 3.

Spain: Central Government Medium and Long Term Debt, Average Maturities and Returns

(In years end percentages)

article image
Source: Central Bank of Spain, Boletin Estadstico.Filename: ESPb2212.TABDate: May 4, 1995.
Table 4.

Spain: Central Government Medium and Long-Term Debt, Balances by Holders

(In billions of pesetas)

article image
Source: Central Bank of Spain, Boletin Estadistico.Filename: ESPb2204.TABDate: February 17, 1995.

Clearly, these developments had a considerable impact on the servicing costs of public debt in Spain. As it was discussed in section II, one could expect to see a reduction in the interest premium for Spain due to the increase in foreign currency denominated debt, and this could have affected the servicing cost differential with other countries. To have an idea of the relative magnitudes, Ireland’s share of foreign currency denominated debt is at present over one third of its total debt; this figure for Belgium is around 15 percent and for Italy around 7 percent. In Italy, the incentive compatibility constraint takes the form of indexation, affecting more than half of its total debt. 15/

Also from the discussion in section II, the increase in debt maturity could have either increased or reduced the risk premium, depending on whether the inflationary bias effect or the confidence crisis effect is the most noticeable. 16/ Notwithstanding the rapid increase in debt maturity in Spain, the share of long term debt in total debt that was reached by 1994 is not different than that existing in Ireland and it is lower than that in Belgium, which has more than 60 percent of debt instruments with maturity higher than a year. 17/

The same ambiguous result could be expected for the increase of nonresident holdings. On the one hand, as non-residents are not taxed in Spain for their receipts on debt, one could expect to see that an increase in nonresident holdings be accompanied by a reduction in the interest rate premium due to two effects: 1) tax exempted foreign investors are willing to take a lower return than domestic investor subjected to taxes, 2) differently from domestic investors, tax exempted foreign investor do not care about the future tax burden either, thus the expected effect due to Ricardian equivalence is also reduced. 18/ On the other hand, the higher number of nonresident holders could increase the volatility of interest rates, perhaps affecting its average value, as the demand for bonds becomes more erratic (or more sensitive to economic and political developments, not only in Spain but elsewhere).

All in all, the simultaneity of the various developments in debt management in Spain observed during the period of analysis make it difficult to pinpoint the net effect coming from each of them. The main fact is that during the exchange based stabilization attempt there was a reduction in the interest rate differential between Spanish Government bonds and German bonds, though it was a common pattern within the EMS. Answering what was due to the exchange commitment, what was the net effect of the increase in debt maturity, or that of the steady increase in non-resident holdings of Spanish debt, is part of the objective below.

V. A Simplified Representation of the Credibility Game

Based on the literature about policy credibility already discussed and the analysis of relevant issues for the case of Spain, this section builds a model to address the question of the Spanish interest rate premium. In view of supporting evidence indicating that pure credit risk is negligible in the case of Spain, the model here only deals with the currency risk component of the risk premium. This assumption is nevertheless tested in the empirical section.

The model in this section stresses the question of conditionality on the state of the economy, which, in the case of Spain, is represented by massive initial unemployment. For that reason, it follows the papers by Drazen and Masson (1994), and Masson (1994), which combine the idea of the credibility of the policymaker with that of the credibility of policies. 19/ Additionally, the model introduces another particular aspect of the Spanish experience, the reputation effect of debt management.

Specifically, in a Barro-Gordon’s type of framework, 20/ it is assumed that the government defends a fixed exchange rate and there are stochastic shocks to unemployment. Before the shocks are realized, private agents incorporate their price expectation in wage contracts and accept fixed nominal returns on government debt denominated in domestic currency. The Government, which decides its policy after the shock is known, has a double incentive to inflate. One is given by the Government’s interest in reducing the high unemployment rate. The other is given by the possibility of reducing the real value of the national debt, which may allow the authorities to avoid using distortionary taxation in the future.

Credibility here is about the Government’s commitment to defend a fixed exchange rate, a policy that Spain maintained from the entry of the peseta in the EMS, June 1989, to the crisis in September 1992. Initially, it is assumed that the exchange rate is the Government’s only policy tool and, therefore, that the market’s credibility on the exchange rate commitment is not affected by fiscal or any other Government policy. Actually, addressing the optimal design of policies would be interesting, but it would be beyond the purpose of this schematic presentation. Instead, and for the sake of model presentation, fiscal policy is assumed exogenous for the government, though the validity of this assumption is tested later in the empirical section using a extended version of the basic model described below. In addition, and for simplicity, it is assumed that the price level equals the exchange rate (i.e., Pt-Pt-1 = et - et-1, where p is the log of the price level and e is the log of the exchange rate).

Although the assumption about fiscal policy is immaterial for the purpose of illustrating the forces discussed above, the implications of such assumption deserve a comment. Undoubtedly, fiscal policy may have important effects on the credibility of the exchange rate policy, and disregarding them can be interpreted in, at least, two ways. It might simply indicate that fiscal—and any other Government action—is made consistent with the exchange rate target, so that all Government’s actions are summarized by the commitment of maintaining the exchange rate. Alternatively, it might imply that fiscal policy does not affect, initially, the credibility of maintaining the prevailing parity. In the latter case, it could be expected that the exchange rate commitment would force any needed adjustment on the fiscal side in the future.

The timing of policies just mentioned is probably where the past affects the credibility of future actions the most. If the situation inherited from the past is that of a significant fiscal imbalance, it could be unreasonable to expect it to be immediately corrected in a way consistent with the exchange rate policy. Besides, it has been argued that the exchange rate commitment usually attempts to help resolving the fiscal struggle, making stabilization programs a two stage approach.21/ In the case of Spain, not much of the fiscal imbalance was resolved during the period of analysis and the market’s perception of such policy shall be revealed in the empirical section below. If fiscal policy was expected to be consistent with the exchange rate policy, then any deviation from that path should have affected the interest premium on Spanish paper. On the other hand, fiscal policy, in the short run, could have been seen as determined by the past and, though apparently inconsistent with the exchange rate policy, it might not have affected the credibility of the Government’s commitment.

Continuing with the model, it is assumed that the situation at the beginning of the program is such that unanticipated inflation reduces unemployment relatively to the natural rate: a Phillips curve is assumed to represent this relationship as in Equation (1). There, μ represents a shock and ur is current unemployment rate measured in deviations from the natural rate.

urt=(a)[(etEt1et)+μt+δurt1](1)

Debt dynamics are represented by equation (2), where m is an index of debt maturity (such as the share of long term maturity on total debt); r is the real interest rate, D is real value of total debt, and G and T reflect government expending and revenues, respectively (the current fiscal balance)—for simplicity, seignorage is assumed to be zero. Equation (2) symbolizes in the simplest way the relationship between debt maturity and the policymakers’ gains from unexpected devaluation. On one hand, it implies that short term debt yields the actual real interest rate as inflation is immediately incorporated in short term rates, while returns on long term debt only incorporate expected inflation—the relevant inflation at the time of the investment. On the other, it includes the likelihood of a currency crisis and the cost of dealing with it as depending on the maturity structure of the debt. Equation (2) reflects the fact that debt concentrated in long maturities leaves more room to affect real returns through unexpected devaluation, providing strong incentive to do so, but, at the same time, it reduces the net gain of a devaluation in case of a confidence crisis, reducing the likelihood of a defensive devaluation. When debt starts concentrating in the short term, the probability of a confidence crisis increases and so does the gains from a devaluation.

Dt=(1+rt)(1(mt1+φmt12)(etEt1et))Dt1+GtTtwhereφispositive(2)

Equation (3) formalizes the government loss function. There k indicates the effect of labor market distortions and λ the distortionary aspect of available taxes. Clearly, the linear presentation of this tax distortion is rather unappealing but it is made for convenience.

Lt=(urt+k)2+θ(Δet)2+λTt(3)

For the moment, m is assumed not to be a policy instrument and Gt is assumed to be constant, the latter given from the past. Simplifying further, rt is also assumed constant and Tt is assumed to be set so that debt remains constant in the absence of unexpected inflation, T=r(1 - (m+φ/m2)(et - Et-1et))Dt-1+G.22/

In order to complete the set up, the shock μ is assumed to be uniformly distributed in the range [-v, v] and the government may choose between a devaluation of magnitude d or keeping to the peg. Besides, it is accepted that there are two types of government, weak and tough, and the market does not know which type has the current government, although it knows the preference of each type (θT>θW). The unique value of the possible devaluation simplifies greatly the presentation without losing much generality.

In this context, both types of government will devalue if the shock is strong enough, as indicated by condition (4) below.

μt>(a+θ)dλr(m+φm2)Dt12akEt1et+et1δurt1(4)

And the probability of devaluation may be represented by equation (5), where π is the probability that the government is weak ρ1 and is the probability that being i the government will devalue.

Et1etet1=[πtρtW+(1πt)ρtT]d(5)

where

ρti=prob{ut>yt[πtρtW+(1πt)ρtT]d+θid2a}

and

yt=adλr(m+φm2)Dt12akδurt1

or

ρtW=vyt2vdθWd/2a2vd+(1πt)(θtWθtT)d2/2a2v(2vd)ρtT=vyt2vdθTd/2a2vd+πt(θtWθtT)d2/2a2v(2vd)

Therefore, the probability of devaluation at each point in time can be represented by ρ in equation (6).

ρt=πtρtW+(1πt)ρtT=vyt2vdθTd/2a2vd+πt(θTθW)d/2a2vd(6)

Finally, it is reasonable to assume that the market updates its probability π following a Bayesian rule, like equation (7)

πt=1ρt1W(1ρt1W)πt1+(1ρt1T)(1πt1)πt1(7)

which after linearizing around steady state values for ur, m, and π, -given the assumptions made- provides the second equation needed for the empirical exercise. Rewriting conveniently, the system of equations to be estimated is the following:

iStNiGtN=dρssT+d(ρssWρssT)πt+γurt1+(χmt1+φ*mt12)Dt1+ϵt(8)
πt=απt1+βurt2+(τmt2+Φmt22)Dt2+ηt(9)

where iSN and iGN are the annual nominal yield of long term bonds for Spain and Germany, respectively, and ρss is the steady state value of ρt (see Appendix for derivation).

The structural system represented by equations (8) and (9) neatly illustrates the determinants of the interest premium. There, as expected, unemployment enters with a positive parameter in the first equation and a negative in the second, showing the double effect of a tough policy in this framework. In the case of the maturity index, the final parameter signs are not clear as they combine in both equations two opposite effects on reputation. On one side, a longer debt maturity makes the gains from inflating higher, increasing the currency risk. On the other, a longer debt maturity makes the fixed exchange rate more credible, reducing the currency risk.

VI. Empirical Assessment

The model developed in the previous section is estimated using monthly data for the period 1989:08 to 1992:08, period of full adherence to the EMS. Two estimating methods were applied: Kalman Filters on the system of equations (8) and (9), and least squares with robust errors (Newey-West type of errors) on a quasi-differentiated equation (8). 23/ The first method is the standard signal extraction technique that provides an efficient estimate of the unobservable series π, but only assures asymptotically the desired properties for the estimates. Bearing this in mind, the second method, robust errors, although not as efficient as Kalman Filters in large samples, provides a useful contrasting evidence for the Kalman Filter estimates.

An extra estimation problem would have arisen if maturity or total debt had been correlated with the error term of the bond yield series. This could be the case if the Government reacted to changes in the interest rate differential by changing the size or composition of its liabilities. In other words, though mt and Dc were assumed not to be policy tools in the presentation section, they, in fact, represent policy options and whether they are predetermined variables in the equation should be checked empirically. 24/ This was done by instrumenting the debt variables with a linear trend and the results were not different from those reported in the tables below. Indeed, the latter confirms what seems to indicate a simple graphical analysis.

The results obtained are presented in Table 5. There, ur is the actual unemployment rate—the controversial issue of the appropriate natural rate of unemployment for Spain is avoided by assuming that it has been constant through the sample and always below the actual rate of unemployment—and m is the share of Government Bonds and Obligations in total debt. In this simple set of equations, the outcome of the two methods of estimation coincide.

Table 5.

Regression Results, Basic Approach

article image
Standard errors in parentheses. Shadow represents significant at 5%.

These results support the view that unemployment acted as a constraint at the time the Spanish government was trying to build credibility on its anti-inflationary type through alignment to the EMS. According to the regression results, people were aware of the limitation imposed to tough policies by the prevailing imbalance in the labor market. On the other hand, the results also reveal that debt management played an important role in the policymakers-agent interaction on policy credibility. The increase in maturity sought by the authorities seems to have resulted in a lower interest rate premium, meaning that the strength in the credibility of the exchange rate policy, as predicted by the confidence crises literature, was more significant than the counter-effecting signal in terms of the inflationary bias hypothesis (in the three year period starting in August 1989, the increase in the interest rate premium due to the inflationary bias imposed by a higher debt maturity is estimated to have been 2.36 percentage points, while the effect coming from a lower probability of confidence crisis reduced it by 2.66 percentage points).

The model presented above helps illustrate the main mechanisms in the “credibility game”, notwithstanding that it is obviously too simplified and that it has serious limitations. The most important limitations are the following:

  • -There is no intertemporal dimension in the “credibility game”, and so no room for strategic considerations.

  • -The only policy considered is exchange rate management. Debt management, and fiscal and monetary polices are not policy choices here. This is a particularly important shortcoming as they are clear policy tools and therefore have an equilibrium meaning not included in the model (i.e. the optimal debt maturity and debt instruments).

  • -As a corollary of the above, the information content of actual policy measures, such as money growth and fiscal deficit, or the deviation produced in expected values of indicators such as inflation or current account balance, are neglected here.

  • -The pure credit risk component of the interest premium has also been neglected here.

In order to overcome partially the limitations pointed out above, a number of extensions are proposed in the next section. Unfortunately these do not solve the problem of not having an intertemporal dimension in the approach, nor do they consider the optimality concept of debt management. Addressing these two limitations would impede having a representation suitable to perform econometric analysis, and, as such, it is beyond the purpose of the paper. Nevertheless, there are two good justifications to overlook the lack of intertemporal dimension and strategic interactions in order to gain simplicity. First, the model here preserves the main effects of a two period model (see Drazen and Masson (1994) for such results), and second, it is still highly questionable whether a stabilization program can be analyzed in a frame of repeated games. 25/

VII. Extensions to the Basic Approach

According to the limitations acknowledged above, equations (8) and (9) are extended in two ways, each one testing a different type of mispecification. In one extension, information on the actual fiscal and monetary policy choice is added to test whether they contain useful information. This includes the Government’s financing needs, the increase in domestic credit, the inflation differential with Germany and the trade balance. The first two variables attempt to measure the possible deviation of actual policy from the policy consistent with the fixed exchange rate. The second two variables also measure such a possible policy inconsistency but looking at resulting state conditions or probable disequilibria.

The other extension tests the relevance of pure credit risk in the risk premium. This is tested by including the stock of debt and the share of nonresident holdings of public debt. The first variable measures directly the benefits of default and the non-resident share holdings attempts to measure indirectly the weakening of the Ricardian equivalence effect. 26/

The empirical results obtained by including the additional information are reported in Tables 6 to 8. In the first two tables the new variables are added one at a time, and Table 8 summarizes the results by showing a nested estimate, both for the working sample and for an extended sample up to September 1994.

Table 6.

Regression Results, Actual Policy Extension

article image
Standard errors in parentheses. Shadow represents significant at 5%.
Table 7.

Regression Results, Credit Risk Extension

article image
Standard errors in parentheses. Shadow represents significant at 5%.
Table 8.

Regression Results, Nested Equation and Two Samples

article image
Standard errors in parentheses. Shadow represents significant at 5%.

The variables in Table 6 representing the information related to actual fiscal and monetary policies are the following: gfnt is the Government’s financing needs, Δct is domestic credit growth, cpi_dt is the difference in consumer price inflation between Spain and Germany, and tbt is the trade balance; all variables are calculated in an annual basis, between t and t-11. The results reported there do reveal that data on actual policy are relevant to explain interest rate premia. However, it does not discredit the argument that unemployment and debt management have been playing an important role in determining such premia, probably indicating that what the actual policy information reveals is the product of its inconsistency with the committed exchange rate. In summary, the significance of unemployment is confirmed while a slightly weaker result is revealed for debt maturity. In addition, two other variables have the correct sign and are significant: greater fiscal financial needs and higher trade deficit increase the interest rate premium.

The inclusion of variables testing the significance of pure credit risk for Spain reveals also neat results. These are presented in Table 7, where Dt is the debt-GDP ratio, and nres_st is the share of national debt held by nonresidents. First, the stock of debt shows to be of significance when explaining the interest rate premium, challenging the hypothesis that pure credit risk is not relevant for Spain. On the other hand, the change in nonresident holdings does seem to affect the interest rate premium in the direction anticipated by a contemporaneous demand improvement and a weakening of the intertemporal restriction (Ricardian equivalence argument). A final implication of including these two variables is that they do not reject the importance of the unemployment rate and debt maturity in explaining the risk premium.

Finally, Table 8 presents the complete equation estimates. There, an estimation for the sample outside the period of EMS is also reported, mainly as a reference. 27/ The main conclusions from this nested equations are that they confirm the importance of unemployment in both periods and corroborates the reputation effect of an increase in debt maturity in the period within the EMS, though its net effect is now positive as the higher maturity seemed to have increased the interest premium.

VIII. Conclusions

This paper presents evidence that policy credibility in Spain, as reflected by the interest rate premium, is not only a question of policymaker’s type but also of policy viability. In the case of Spain, the massive unemployment sets a limit to the private agent’s beliefs about how far a committed government will push its policy, constraining the build up of reputation needed for a successful and painless anti-inflationary policy.

In addition, macroeconomic policies that are perceived as not consistent with the committed target also are found to help in explaining the interest rate premium. This is especially true for an indicator of the Government’s financing needs, which in some way rejects the conclusion reached in Ballabriga and Sebastian (1993), where no interaction between fiscal variables and interest rates is found for the period 1964-91. The results in the present paper also partly conflict with those in Ayuso, Jurado and Restoy (1994), where, using an ad hoc approach they find that the credibility of the peseta during the EMS period is explained by the inflation differential, unemployment, and the current account deficit, while the fiscal deficit and monetary indicators have no explanatory power.

On the other hand, the inclusion of debt management policies among the possible determinants of the interest rate premium attempted in this paper does not lead to a definite conclusion. Using the simplest equation specification, the increase in debt maturity seems to have helped to reduce the risk premium. However, this finding weakens when information about the increase in non-resident holdings of Spanish debt is incorporated.

All of the findings reported above are of importance for macroeconomic policy in Spain. Essentially, they suggest that “two stage” anti-inflationary attempts in Spain could be very costly even if they are carried out by the most conservative of policymakers; for instance, a tight monetary policy at a time when government financial consolidation is undertaken, but the latter can only be perceived by private agents after some time, could be a costly policy. In this regard, the findings here, which seem to be supported by recent developments in Spain, do not underscore brawny optimism about the expected outcome from the recently instituted independence of the Banco de España (Central Bank) unless it is accompanied by a conspicuous fiscal consolidation. Indeed, the results reported above make a stronger than ever case for the need of structural measures in Spain, which do have a permanent effect on inherited disequilibria. Institutional reforms that enhance “fiscal responsibility”, or policies that weaken distortions in the labor market could be good examples of such structural measures. 28/

Appendix: Working Equation

The steady state probability of devaluation for each government type,

ρssW|(πt=1;mt=m0,Dt1=D0)=12+λr(m0+φm02)D0+2ak2a(2vd)θWd2a(2vd)ρssT|(πt=1;mt=m0,Dt1=D0)=12+λr(m0+φm02)D0+2ak2a(2vd)θTd2a(2vd)

are incorporated in equation (6) to obtain the following,

ρt=ρssT+πt(ρssWρssT)+λ((r(mt1+φmt12)Dt1)r(m0+φm02)D0)2a(2vd)+δrut12vd

which, assuming m0-1 leads to the final expressions:

πt=1ρt1W1ρt1πt1
ρt1W=ρssW(1πt1)(ρssWρssT)(d2v)+λr((mt2+φmt22)Dt2(1+φ)D0)2a(2vd)+δrut22vd
πt=1ρssW+(1πt1)(ρssWρssT)d2vλr((mt2+φmt22)Dt2(1+φ)D0)2a(2vd)δrut22vd1ρssT+πt1(ρssWρssT)λr((mt2+φmt22)Dt2(1+φ)D0)2a(2vd)δrut22vdπt1

The latter is the expression to be linearized in order to obtain equation (9), which is presented in its complete form below:

(linearizing around urt-2-0, mt-2-1 and πt-10, and assuming Dt-2-D0)

πt=[π0(1π0)(ρssWρssT)δ2vA2]urt2[π0(1π0)(ρssWρssT)14av(12φ)λrDt2A2](mt2+φmt22){[(1ρssT)(1ρssW+(12π0)(ρssWρssT)d2v)A2]+[(1π0)(1ρssW)(ρssWρssT)+(12π0(1π0)(ρssWρssT)2d2vA2]}πt-1whereA=[1ρssTπ0(ρssWρssT)]

References

  • Alesina, A. and R. Perotti,The Political Economy of Budget Deficits,” IMF Working Paper, 85, 1995.

  • Alesina, A., A. Prati, and G. Tabellini,Public Confidence and Debt Management: A Model and Case-Study of Italy,” in R. Dornbusch and M. Draghi, eds., “Public Debt Management: Theory and History,” CEPR, 1990.

    • Search Google Scholar
    • Export Citation
  • Ayuso, J., M. P. Jurado, and F. Restoy,Credibility Indicators of an Exchange Rate Regime: The Case of the Peseta in the EMS,” Banco de España, Documento de Trabajo 9307, 1993.

    • Search Google Scholar
    • Export Citation
  • Backus, D. and J. Driffill,Inflation and Reputation,” American Economic Review, 75, 1985 (a).

  • Backus, D. and J. Driffill,Rational Expectations and Policy Credibility Following a Change in Regime,” Review of Economic Studies, 52, 1985 (b).

    • Search Google Scholar
    • Export Citation
  • Ballabriga, F. and M. Sebastian,Déficit Público y Tipos de Interés en la Economía Española: Existe Evidencia de Causalidad,” Banco de España, Documento de Trabajo 9220, 1992.

    • Search Google Scholar
    • Export Citation
  • Barro, R. and D. Gordon,Rules, Discretion, and Reputation in a Model of Monetary Policy,” Journal of Monetary Economics, XII, 1983.

    • Search Google Scholar
    • Export Citation
  • Bulow, J. and K. Rogoff,Sovereign Debt: Is to Forgive to Forget,” American Economic Review, 79, 1989. Also in T. Persson and G. Tabellini, 1994.

    • Search Google Scholar
    • Export Citation
  • Calvo, G.,On the Time Consistency of Optimal Policy in Monetary Economics,” Econometrica, 46, 1978. Also in T. Persson and G. Tabellini, 1994.

    • Search Google Scholar
    • Export Citation
  • Calvo, G.,Servicing the Public Debt: The Role of Exectations,” American Economic Review, 78, 1988.

  • Calvo, G. and P. Guidotti,Credibility and Nominal Debt: Exploring the Role of Maturity in Managing Inflation,” IMF Staff Papers, 37, 1990.

    • Search Google Scholar
    • Export Citation
  • Carranza, E. T.,Distribución de las Tenencias de Deuda Pública en Pesetas,” in Información Comercial Española, 731, 1994.

  • Chari, V. and P. Kehoe,Sustainable Plans and Mutual Default,” Review of Economic Studies, 60, 1993.

  • Delso, F. F.,La Financiación en Divisas del Estado,” in Información Comercial Española, 731, 1994.

  • Dornbusch, R.,Credibility and Stabilization,” Quarterly Journal of Economics, 1991.

  • Drazen, A. and P. Masson,Credibility of Policies versus Credibility of Policymakers,” The Quarterly Journal of Economics, 109, 1994.

    • Search Google Scholar
    • Export Citation
  • Eaton, J. and M. Gersowitz,Debt with Potential Repudiation,” Review of Economic Studies, 48, 1981.

  • Fontenay, P., G. M. Milesi-Ferretti, and H. Pill,The Role of Foreign Currency Debt in Public Debt Management,” IMF Working Paper, 21, 1995.

    • Search Google Scholar
    • Export Citation
  • Giavazzi, F. and M. Pagano,The Advantage of Tying One’s Hands: EMS Discipline and Central Bank Credibility,” European Economic Review, 32, 1988. Also in T. Persson and G. Tabellini, 1994.

    • Search Google Scholar
    • Export Citation
  • Giavazzi, F. and M. Pagano,Confident Crises and Public Debt Management,” in R. Dornbush and M. Draghi, eds., “Public Debt Management: Theory and History,” CEPR, 1990.

    • Search Google Scholar
    • Export Citation
  • Guidotti, P. and C. Vegh,Losing Credibility: The Stabilization Blues,” IMF Working Paper, 73, 1992.

  • Grossman, H. and J. Van Huyck,Sovereign Debt as a Contingent Claim: Excusable Default, Repudiation, and Reputation,” American Economic Review, 78, 1988.

    • Search Google Scholar
    • Export Citation
  • Kydland, F. and E. Prescott,Rules Rather than Discretion: The Inconsistency of Optimal Plans,” Journal of Political Economy, 1977, vol. 85, no. 3.

    • Search Google Scholar
    • Export Citation
  • Masson, P.,The Credibility of the United Kingdom’s Commitment to the ERM: Intentions Versus Actions,” IMF Working Paper, 147, 1994.

    • Search Google Scholar
    • Export Citation
  • Missale, A.,Public Debt Management,” Unpublished Ph. D. dissertation, MIT, 1994.

  • Missale, A., and O. Blanchard,The Debt Burden and Debt Maturity,” NBER Working Paper, 3944, 1991.

  • Pascual, J. L.,La Evolución de los Mercados de la Deuda,” in Información Comercial Española, 731, 1994.

  • Persson, T. and G. Tabellini eds., “Monetary and Fiscal Policy. Volume 1: Credibility,” The MIT Press, 1994.

  • Rogoff, K.,The Optimal Degree of Commitment to an Intermediate Monetary Target,” Quarterly Journal of Economics, 100, 1985. Also in T. Persson and G. Tabellini, 1994.

    • Search Google Scholar
    • Export Citation
  • Garcia, G. H.,La Organización del Mercado Español de Deuda Pública,” in Información Comercial Española, 731, 1994.