This Selected Background Issues paper examines different aspects of the social security system in Spain. The paper describes the role of this system in the process of consolidating the government finances in the short- and medium-term, and to present possible measures to address the policy challenges of the longer term. An analysis of the finances of the social security system is presented. The paper provides a description of the main features of the pension system and projections of its financial situation in the long term.


This Selected Background Issues paper examines different aspects of the social security system in Spain. The paper describes the role of this system in the process of consolidating the government finances in the short- and medium-term, and to present possible measures to address the policy challenges of the longer term. An analysis of the finances of the social security system is presented. The paper provides a description of the main features of the pension system and projections of its financial situation in the long term.

III. Inflation Persistence in Spain 1/

1. Introduction

Spain has experienced higher inflation than most other industrial countries since the 1970s. Inflation performance improved significantly in the mid-1980s, both in absolute and in relative terms. Since then, however, there has been little further progress, despite Spain’s decisions, first, to participate in the exchange rate mechanism of the European Monetary System and, subsequently, to seek to converge toward the inflation rates of the best-performing EU member countries. Even in the wake of the severe recession of 1992–93, with unemployment around 24 percent of the labor force, the rate of increase in consumer prices has moderated only slightly, from over 6.0 percent a year in 1989–92 to 4.7 percent in 1994.

This paper investigates the reasons behind the persistence of inflation in Spain. Section 2 summarizes inflation performance over the past 15 years. Section 3 places it in the context of changes in monetary, fiscal and incomes policies and the behavior of aggregate demand. Section 4 provides information on costs, focusing particularly on the impacts of wage behavior. Section 5 discusses sectoral differences in price behavior and the influences of product market rigidities. Concluding remarks are presented in section 6. An annex presents a model of inflation dynamics with the objective of testing for the degree to which the change in the exchange commitment when Spain joined the ERM was accompanied by changes in the persistence of inflation.

2. Empirical evidence on price performance

a. Aggregate price performance

Spain’s inflation performance since 1980 is summarized in Chart 1. We can distinguish four main periods. There was a gradual deceleration of inflation in 1980–84 which, nevertheless, was still in the two-digit range by the end of the period. From end-1984 until mid-1988 inflation decelerated further to under 5 percent per annum, with a momentary interruption caused by the introduction of the VAT as Spain joined the EC in 1986. Inflation rebounded again to over 6.0 percent per annum in 1989–90, in the face of a widening of the fiscal deficit and an overheating of aggregate demand, and has only recently fallen below 5.0 percent.

Chart III-1
Chart III-1

SPAIN: CPI Inflation and Underlying Inflation

Citation: IMF Staff Country Reports 1995, 039; 10.5089/9781451811988.002.A003

1/ The rates of growth are from a quarter on same quarter in previous year.

Chart 1 also provides a measure of underlying inflation, which removes from the overall CPI those components which have a more erratic behavior, unprocessed foodstuffs and energy products. Since the mid-1980s the underlying inflation rate has generally been higher than the rate of increase in the overall CPI. This has reflected both the weakness of world oil prices, Which have held down the overall inflation rate, and the higher weight given to the price of services in the index of underlying inflation (see below).

As shown in Table 1, Spain’s inflation rate has been consistently higher than the average of other EU member countries, and in particular the three members with the best inflation performance. While this differential has been reduced in recent years, it remains around 3.0 percent. In contrast, the inflation differential with OECD member countries has widened.

Table 1.

Spain: Inflation Differentials

(In percent per annum)

article image
Source: Ministerio de Economia y Nacienda.

Excluding Turkey.

b. Sectoral price performance

One of the distinctive features of the Spanish inflationary process is its dual character, whereby the prices of nontradable goods have been growing much faster than the prices of tradables. This phenomenon is common to a number of European countries and has a direct bearing on the evolution of the real exchange rate, i.e., the relative price of nontradable and tradable goods. In this section we describe these developments by looking at domestic relative prices of services and industrial goods (a proxy for the distinction between tradables and nontradables). 1/

Chart 2 depicts alternative measures of the evolution of inflation differentials in the two sectors. Panel A shows the ratio of the value-added deflator in manufacturing sector relative to that of the services sector, whereas panel B depicts the evolution of the industrial (nonconstruction) and of the services components of the CPI. Both charts clearly show a steady decline in the relative price of industry with respect to services.

Chart III-2
Chart III-2

SPAIN: Components of CPI and Sectoral Deflators

(In percent)

Citation: IMF Staff Country Reports 1995, 039; 10.5089/9781451811988.002.A003

1/ The rates of growth are from a quarter score quarter never used.

The phenomenon of a higher inflation bias in the service sector in Spain is not new. In the period between 1965 and 1980, the rate of change in the prices of services exceeded that of industrial goods by around 3.5 percent per year, due to the lower productivity of the service sector in a setting of uniform wage growth. 1/ Between 1980 and 1985, before Spain’s accession to the EC, however, service goods prices grew at the same rate as those of overall production, and 0.3 points above the overall consumer price index. After Spain’s accession to the EC, the duality of inflation in Spain re-emerges, becoming marked after Spain joined the ERM, reflecting the adoption of an anti-inflation strategy centered on an exchange rate anchor and entailing a real effective appreciation of the peseta. Thus, from 1990 to 1992 while industrial prices grew by 1.7 percent, service prices measured by the sector’s implicit deflator, grew by 8.8 percent, a differential of 7.1 percentage points. Looking at CPI components, the prices of services increased at an average annual rate of 9.9 percent for services compared to 4.6 percent for food and industrial prices), while in 1993 this process slowed somewhat. 2/

The high growth rate of services prices relative to that of goods has been a characteristic of most services and not of specific sub-sectors, as illustrated in Table 2. Nevertheless, the relatively higher growth of health, education, household services, urban transportation and tourism/hotel costs is noteworthy. This group of services is particularly sheltered from competition.

Table 2.

Spain: Rates of Increase in Service Prices

(Percent per year)

article image
Source: Instituto Nacional de Estadistica.

3. Demand factors and financial policies

In the 1970s Spain had to contend with the combined economic implications of the restoration of democracy and of two oil shocks. The resulting economic imbalances culminated in a very high rate of inflation. In the wake of the first democratic election in 1977, a broad social pact was reached—The Moncloa Agreements-paving the way for the implementation of an adjustment program. The agreements were signed by the Government and the opposition parties, with the implicit consent of the socialist (UGT) and Communist (CCOO) trade unions and the newly formed employers’ association (CEOE). Two of their measures were fundamental for the control of inflation. First, by abolishing indexation to past inflation and shifting to a forward-looking pattern of wage determination, it introduced wage moderation. Second, there was an acceptance of the need for non-accommodating financial policies, including a significant reduction in the growth of liquidity and a sharp increase in interest rates, so as to reduce inflation. These measures were accompanied by the creation of a new institutional framework for industrial relations (“the workers’ statute”) in which the principal trade unions and employer associations were recognized as the main actors. (de La Dehesa 1994 and Maravall 1993). 1/

The combination of financial, structural and incomes policies, brought price and wage inflation down from 15.6 percent and 17.3 percent in 1980, respectively, to 8.8 percent and 8.7 percent in 1985, helped by the deceleration in the prices of imported goods, particularly oil. The average growth of GDP over the 1980–1985 period was 1.4 percent, and the unemployment rate jumped from 11.5 percent to 21.6 percent.

The second half of the 1980s was characterized by rapid expansion of domestic demand, coinciding with Spain’s accession to the European Community, the liberalization of the exchange and trade system, and favorable international conditions. Real domestic demand grew at an average annual rate of 6.6 percent in 1986–90, led by an average growth of 11.9 percent in real fixed investment. Employment increased on average by 3.0 percent over this period and unemployment rate fell from 21.5 to 16.3 percent of the labor force. This buoyant performance of economic activity was accompanied by further progress in disinflation, particularly up to 1988, due to the maintenance of tight financial policies, wage moderation, and a favorable international environment. After having increased sharply in the early 1980s, the fiscal deficit was reduced from 6.9 percent of GDP in 1985 to 2.8 percent in 1989. A strict monetary policy led to high interest rates and an appreciation of the nominal exchange rate, which in turn contributed to the cheapening of imports, at a time when international commodity prices were weak in U.S. dollar terms.

Subsequently, however, the fiscal deficit widened again, reaching about 5.0 percent of GDP in 1991–92. 2/ This expansionary fiscal stance complicated the conduct of monetary policy. Spain joined the European Exchange Mechanism (ERM) in 1989 as a way to lock in the disinflation gains made in the preceding years, and as a mechanism to foster nominal convergence. The Spanish authorities believed that entry into the ERM and the announcement of a commitment to a quasi-fixed exchange rate was the appropriate way to attain the discipline and credibility viewed as necessary to set agent’s expectations formation in line with inflation in the best-performing member countries. To help contain inflationary pressures, interest rates were raised sharply, which contributed to a systematic overshooting of broad money and credit aggregates. With the peseta at the top of its ERM band, in an attempt to keep money and credit under control, the authorities resorted to temporary direct restrictions on domestic credit to the private sector and a compulsory nonearning deposit on foreign borrowing. These were largely ineffective, and were removed by 1991 and early 1992, respectively.

Table 3.

Spain: Selected Economic and Financial Indicators

(Percent per year, unless otherwise indicated)

article image
Source: Bank of Spain.

The fiscal impulse is defined as the change during a given year in the discretionary component of the government balance, expressed as a share of GDP. For details see Appendix II in SM/94/10.

December on December, average of daily balances.

December to December.

Relative to industrial countries.

Based on CPI, relative to industrial countries.

Bank of Spain monetary operations rate.

Three-year government bonds.

Real domestic demand started to weaken in late 1990 and 1991, while the loss in external competitiveness fostered a rapid penetration of imports. When the ERM crisis erupted in September 1992, the authorities temporarily resisted the exchange market pressures by raising real short-term interest rates to the 7–9 percent range, which helped to push the economy into recession. Subsequently, the central rate of the peseta was devalued three times between September 1992 and May 1993, by a total of 20 percent.

Throughout this period of weakening of activity and, later, recession, real wages continued to increase significantly. The lack of wage flexibility and high financial costs contributed to a marked squeeze in profit margins followed by bankruptcies and labor shedding. In 1993, employment declined by 4.3 percent, resulting in an increase in productivity and a moderation of the growth of unit labor costs, but also in a sharp rise in the unemployment rate. Partly because of labor and good market rigidities, but also because of the depreciations, inflation continued to react slowly to the growing gaps in labor and product markets, remaining unchanged at 5.9 percent in 1992 and declining to 4.6 percent in 1993.

The question that arises in trying to understand Spain’s disinflation experience, particularly after joining the ERM, is how was it possible that wages and consumer prices could continue to increase above the European average despite nonaccommodating monetary policy. The adjustment process expected to take place under these circumstances assumed that over time the real appreciation of the exchange rate would have led to higher unemployment, slower wage growth, and lower inflation. While the increase in unemployment clearly obtained, the equilibrating responses were subdued and extremely slow to take place. It is therefore important to look for some systematic factors that might have mitigated the effect of those equilibrating forces.

In the annex to this paper a simple two-sector, open-economy, model is developed to help investigate the role of the exchange regime and of wage indexation in the process of inflation persistence. The model suggests that dynamics of domestic inflation can be written as the following first order difference equation:


where inflation in the current period πt is a function of lagged domestic inflation πt-1, lagged foreign inflation πt1*, a measure of demand pressure proxied by the output gap (YtFYt1). and relative productivity growth in the tradables and nontradables sector (aaNT). In this formulation, the coefficient ξ1 (itself a function of, inter alia, policy parameters and of the degree of indexation in the economy) provides a measure of the degree of inflation persistence. The model shows that if the authorities follow an exchange rate rule that accommodates the differential between domestic and foreign inflation, the inflationary process will exhibit persistence. If, on the other hand, an exchange rate commitment, such as the one that characterizes the ERM, is adopted and perceived as credible by the public, price setting behavior and wage contracts will reflect the fall in inflationary expectations in the economy, generating a change in regime. Thus, if the nominal exchange rate anchor policy is credible, we would empirically observe a structural break in the dynamic properties of inflation. The model shows, however, that if the labor market is characterized by indexation, inflation will display some persistence even in the case in which the authorities choose not to follow accommodative monetary and exchange rate policies. The results obtained in the empirical investigation, however, indicate that there was no structural break in the dynamics of inflation after Spain joined the ERM, suggesting that adopting an exchange rate anchor was not sufficient to lower inflation expectations and thus reduce persistence. It is conceivable that this was due to the fact that the exchange rate commitment was not viewed as part of a credible policy package, both because the fiscal stance was not perceived as consistent with the exchange rate peg, and because inertia-generating features of the wage setting framework in Spain remained unchanged.

4. Supply side factors

a. Cost developments

This section examines the evolution of labor costs, import costs, and corporate margins as determinants of inflation in Spain. The analysis is undertaken by looking at cost and price indicators based on the final demand deflator, which permit an evaluation of the supply-side factors that bear on the development of the prices paid by domestic agents in purchasing goods and services produced both domestically and abroad.

The final demand deflator can be decomposed into a cost component (which, in turn, can be broken down into unit labor costs and import costs), a unit margin component, and net taxes, as follows:




where, FDD the final demand deflator, w is compensation per worker, y is real GDP, N is total employment, fd is real final demand, Pm the import deflator, M real imports. τ is the unit margin, T are taxes net of subsidies, E is the gross operating surplus and A is the number of wage earners. Equation (2) shows that the unit cost involved in satisfying final demand can be regarded as a weighted average of the unit labor costs per unit of GDP, (w/(y/N), and of the import deflator, Pm, with the weights y/fd and M/fd, which can be regarded as technical coefficients. 1/

Table 4 provides the rate of change of the final demand deflator and of the costs indicator, from 1980 to 1993, as well as the contributions of total costs (labor, imports and taxes) and margins to change in the deflator. The behavior of the domestic demand deflator paralleled that of the CPI in 1980-93.

Table 4.

Spain: Composition of th e Final Demand Deflator

article image
Sources: Bank of Spain and staff calculations.

Per unit of final demand.

Per unit of output.

Up to 1984 the rate of inflation (measured by the final demand deflator) remained in the two digit range, although showing significant deceleration. Unit labor costs were the main contributor to the growth of the cost and price indicators, growing at above 10 percent a year. The first panel of Chart 3 shows the behavior of unit labor costs (per unit of GDP). The second and third panels show the evolution of productivity and compensation per employee, respectively. The contribution of import costs was also significant, particularly in the early years of the decade, reflecting mainly the delayed impact of the second oil shock (Chart 4) and the fact that the authorities allowed for a systematic depreciation of the nominal effective exchange rate in an attempt to maintain the competitiveness of the economy that was being eroded by the price-wage spiral. Net indirect taxes per unit of final demand added substantially to cost and price pressures in the early 1980s, growing at an average 20 percent a year. 2/ Finally, margins contributed on average one fifth of the growth of the final demand deflator during this period.

Chart III-3
Chart III-3

SPAIN: Growth Rate of Unit Labor Costs and Components

(In percent)

Citation: IMF Staff Country Reports 1995, 039; 10.5089/9781451811988.002.A003

Chart III-4
Chart III-4

SPAIN: Import Deflator and Nominal Effective Exchange Rate

(In percent)

Citation: IMF Staff Country Reports 1995, 039; 10.5089/9781451811988.002.A003

Starting in 1985 there was a notable further deceleration in the rate of inflation in Spain, with the rate of change of the final demand deflator reaching a low of 4.8 percent in 1988, as wage moderation coincided with falling import prices, giving rise to a virtuous cycle. Net indirect taxes increased rapidly up to 1986 as Spain brought its indirect taxes up to EC standards, but after this their contribution to the growth of the final demand deflator was small. Throughout this period, there was an increase in the weight of the gross operating surplus and a slower reduction in unit margins.

The period of inflation deceleration was interrupted in the last years of the decade due in large part to an acceleration of wages. Prior to 1990, changes in the composition of employment, such as the large increase in temporary employment, helped moderate the behavior of wages per employee, 1/ and the substitution effect toward imported goods that resulted from rising import penetration, helped contain labor costs per unit of final demand. From 1990 to 1992, however, compensation per employee grew on average by 8.8 percent. This was only partly offset by an increase in productivity per employee, resulting from the strong reduction in employment, particularly in 1991 and 1992, and the increase in ULC jumped to 6.8 percent, on average, during this period. Only in 1993 was there significant moderation in the growth rate of compensation per employee, an effect that was again reinforced by the strong increase in labor productivity that resulted from labor shedding.

The second half of the 1980s was characterized by a large increase in real imports, as a result of Spain’s accession to the EC. This coincided with a period of very favorable international price developments and exchange rate appreciation, with the resulting decline in the import deflator tending to dampen the impact of rising labor costs on prices. The implication, however, was that goods produced with domestic labor were loosing ground to goods with foreign labor, the other face of which was the rapid increase in the unemployment rate. The devaluations of the peseta in 1992 and 1993 have since begun to reverse the process of substitution of imports for domestically produced goods and services, and the dampening effect of the import deflator on the cost and price indicators. Other particularly noteworthy factors in recent inflation performance were the strong recovery in the gross operating surplus in 1993, and a 9.7 percent fall in indirect taxes net of subsidies per unit of final demand in that year, due to the combined effect of the recession and the reduction of tariff barriers.

b. Wage pressure characteristics of Spain’s labor market

The previous section showed that by the late 1980s wages resumed their fast growth path and that in the early 1990s, even in face of the highest unemployment rate in the EU, wages continued growing rapidly, being a driving force behind inflation. 1/ While the objective of this section is not a detailed analysis of the labor market, 2/ it is interesting to examine the features of the market that might help explain the persistence of the rise in nominal wages.

(1) Degree of bargaining centralization

The great majority of Spanish workers have their pay and conditions determined by collectively negotiated agreements. Official statistics place the share of employees covered by collective agreements at 83.1 percent in 1991, 3/ despite the low level of union affiliation (around 10 percent). This is because collective agreements apply to all employees in a sector—even those not represented by unions—and the terms of industry-wide agreements have constituted binding floors for all firms in the sector. 4/

Bargaining is conducted at various levels: firm, industry, region and nationwide, though predominantly at the regional (provincial) level. This structure entails the existence of a large number of agreements and can be characterized as an intermediary degree of centralization. Studies have shown such bargaining to be more conducive to high wage growth than either centralized or decentralized systems. This is because a highly centralized system generates better coordination and allows unions to take into account the inflationary impact of wage increases, reducing their wage claims. In the other extreme where bargaining mainly takes place at the firm level, excessive wage demands might lead the firm to a delicate financial situation, threatening its level of employment. Awareness of this risk leads unions to relatively lower wage growth rates. In the intermediary position, unions do not have incentives to internalize the economy-wide impacts of wage increases on inflation and on employment, nor do they take into account the particular circumstances of specific firms. 5/ Coricelli (1990) provides evidence that centralization during the period of broad [incomes policies helped the disinflation process up to 1987, suggesting that lower centralization since then might be considered a structural change. Jimeno (1992) provides empirical evidence on the high wage proclivity of the intermediate system that prevails in Spain.

(2) The impact of temporary contracts

The second important feature of the market is its segmentation between workers with temporary and permanent contracts. Temporary contracts were introduced in late 1984 as a way of reducing the rigidities that derived the high severance pay associated with permanent contracts. Up to 1993, these contracts could be used for any activity (temporary or not) and could be signed for short periods and could be renewed for up to three years, 1/ and bear low dismissal costs. Since their introduction they have been used to such an extent that presently more than 30 percent of all employees hold a temporary job.

The introduction of temporary contracts had pervasive wage effects. Bentolila and Dolado (1994) note that up to the mid-to late 1980s, wage drift, i.e., the difference between the rate of growth of wage rates agreed in collective agreements and actual average earnings growth, used to be about 2 percentage points, but that after 1987 it was abnormally low, even negative, rising again since 1990. In the latter period, however, bargained real wages started to rise, in spite of the recession. The authors reconcile these two trends by building an argument that extends the insider-outsider view of the wage bargaining process, by identifying insiders with workers holding permanent contracts and outsiders with those holding temporary contracts. They argue that since workers on temporary contracts receive lower wages than those on permanent contracts, 2/ as the proportion of temporary workers in total employment increased, average wages increased by less than usual, reducing wage drift, explaining the first empirical regularity. However, as the proportion of temporary workers on total employment stabilized, the bargaining behavior of the permanent workers changed. This is because temporary workers are cheaper to dismiss and are, therefore, the first to be fired when labor shedding is needed. The fact that temporary employees will bear the negative employment consequences of high wage demands, leads insiders to bargain for higher wages than they would otherwise, driving wage settlements upward as the proportion of outsiders in total employment increases. This could help explain part of the wage increase in the early 1990s. The authors test their model by running wage equations that include, in addition to the standard variables, the proportion of temporary workers in the firm. They find that this variable exerts a strong upward pressure on wages, concluding that a large part of the rise in the wages of permanent workers in the late 1980s and early 1990s is due to the increase in temporary employment.

(3) Indexation clauses

Expected inflation is the main cause of agreed wage increases in the annual collective agreements, with productivity clauses playing a very limited role. 1/ While formal indexation clauses are not imposed by law, a recent study of collective agreements in the Autonomous Community of Madrid for the period 1991–1992 shows that the tendency is to relate wage increases to inflation. Indeed, some 53 percent of the agreements pacted for the whole Autonomous Region of Madrid explicitly related wage increases to inflation, with around 36 percent of the total linking wages to past inflation. Of the sectoral agreements that explicit inflation clauses (around 52 percent of the total), 56.3 percent are backward looking, in that they ask for compensation for lost purchasing power, and the remainder are forward looking. It is estimated that around 30 percent of all sectoral agreements are indexed to past inflation. Finally, for agreements reached at the firm level, around 40 percent of the agreements are explicitly related to inflation, with 9 percent based on past inflation and the reminder on projected inflation. In the agreements at the firm level, however, it is common to find specific numerical targets for wage increases, without explicit mention of inflation.

5. Explaining the inflation differential between services and industry

What explains the differential between inflation in the manufacturing and in the services sectors? The so-called dependent economy model offers some guidance in the identification of the causes of this phenomenon. 2/ According to the model the differential in inflation in the nontradables (services) and tradables (industrial) sectors is determined by two main groups of factors: aggregate and sectoral demand changes and relative productivity gains in the tradables and nontradable sectors.

a. Aggregate and sectoral demand shifts

An overall increase in demand that creates inflationary tensions would tend to increase the inflation differential because the prices of tradable goods are constrained by foreign competition while those of the nontradables are not. It has also been suggested that the demand for services tends to have a large income elasticity, growing faster than aggregate demand and output. Moreover, factors which tend to change the composition of aggregate demand from traded to nontraded goods would tend to lead to an increased differential. Two important cases are those of capital inflows and government spending. Capital inflows generally increase the relative price on nontradables by increasing demand, although the effect depends on the use the capital is put to. Government spending tends to be heavily concentrated in nontraded goods and, thus, tends to increase the inflation differential. 1/

b. Sectoral differences in cost and productivity growth

Services tend to be labor-intensive productive activities, thus having a higher participation of wages in value added than industry. This, in conjunction with the fact that less international competition makes it easier to translate wage increases into prices in the services sector, contributes to make inflation in the services sector more responsive to wage inflation, than that in the industrial sector. A factor that limits the impact of cost increases on prices is productivity growth. It is often argued that productivity growth is inherently faster in the production of manufacturing goods than in that of services, because the degree of factor-substitutability is lower in the services sector, where labor is difficult to replace, and because the exposure to trade flows facilitates the absorption of more modern technology in the industrial sector. A faster pace of productivity growth in the industrial sector may also tend to reduce its inflation rate relative to that of services.

What is the relative significance of each of these determinants in the Spanish case, particularly since the mid-1980s when the inflation differential started to widen again?

As discussed in section 3, domestic demand was particularly strong in the late 1980s, generating a strong increase in the demand for services by households and by enterprises. It seems to be the case that the inflation differential is positively correlated with aggregate demand pressures (Chart 5). When, as between 1986 and 1990, aggregate demand grows at fast rates, the differential widens, when aggregate demand decelerates, the inflation differential also falls, although with a certain lag. In terms of sectoral demand shifts, the main source of demand increase in the 1980s was fixed investment, partially through foreign direct investment, although both private and government consumption also grew very fast. While it is true that government consumption in Spain has increased its share of GDP substantially since the late 1970s, most of this increase occurred before 1986.

Chart III-5
Chart III-5

SPAIN: Inflation Differential and Aggregate Demand Growth

Citation: IMF Staff Country Reports 1995, 039; 10.5089/9781451811988.002.A003

Is it possible to explain sectoral inflation differential by divergences in wage behavior and labor costs? Table 5 summarizes sectoral developments in wages and unit labor costs during the period in study. The evolution of wages was quite similar in the industry and construction sectors, with wages growth rates decelerating until 1988 or 1989, bouncing back significantly in 1990 and 1991, and showing some moderation again in 1992 and 1993. In the service sector, on the other hand, wage increases started to accelerate in 1987, slowing after 1991. More broadly, throughout the period from 1980 to 1993 wage increases have not been significantly different among sectors.

Table 5.

Spain: Sectoral Labor Cost Developments

(Annual average percentage change)

article image
Sources: IMF and Fund staff calculations.

Without construction.

Unit labor costs show a much greater dispersion, since productivity growth was higher in the industrial sector. Large productivity gains in the first half of the 1980s contained unit labor costs in industry. After accession to the EC, however, the rate of growth of productivity growth has moderated, reaching negative levels in 1990. In the recession years of the early 1990s, productivity again increased more rapidly due to labor shedding. In the construction and service sectors, productivity gains were subdued or negative throughout the period.

It is, therefore, apparent that in spite of the relatively homogeneous behavior of wages, discrepancies in relative productivity make labor costs an important explanation for the differential behavior of inflation. Indeed, for the services sector to pay similar wages while productivity grows more slowly than in industry, its prices most rise faster. This is made possible by the fact that the services sector has a greater ease in translating cost increases into final prices, due to the protection it enjoys from competition. This protection is of two types. First, because services are in general nontradables, they are inherently sheltered from international competition. Second, the sector is also protected at the national level by a multitude of regulations and barriers to entry that limit competition. These structural features of the market for services lead to more inertia in their inflationary process, particularly at times of disinflation.

6. Conclusions

Several conclusions can be drawn from this review of Spain’s disinflation process. The significant progress obtained in the mid-1980s was due to a combination of a tightening of financial policies, magnified by an incomes policy that greatly reduced the incidence of backward-looking indexation. While incomes policy probably reduced the cost of the disinflation, the process was not without sacrifices.

The nonaccommodating exchange rate policy followed by the Bank of Spain after the accession to the ERM was intended to lock in the disinflation gains made in the preceding years, and as a mechanism to foster nominal convergence, by attaining the discipline and credibility viewed as necessary to bring agents’ expectations in line with inflation in the core of the ENS. The concurrent relaxation of fiscal policy and the acceleration of wage settlements, however, highlight the limits of the policy. In fact, the effects of discipline and anti-inflationary reputation, which can be enhanced by the adoption of a pegged or quasi-fixed exchange rate system, can only occur if exchange rate stability is perceived by economic agents as sustainable in the medium run, and this, in turn, is only possible if economic fundamentals are compatible with convergence toward the prevailing inflation rate in the main trading partner economies.

In this sense the experience of Spain shows that a commitment to exchange rate stability does not, by itself, provide a mechanism of sufficient discipline. By directly affecting only the tradable’ sector of the economy, it is likely to be costly in terms of output and employment. This is particularly important in the case of Spain, in view of the rigidities that characterize the nontradables sector and the labor market. Indeed, inflation expectations of wage earners have apparently remained high and wage demands have proved to be relatively insensitive to the unemployment rate. Moreover, once economic agents are aware of the restrictions under which monetary policy operates in a system of fixed exchange rates, its impact on price expectations is conditioned by the action of the other instruments of economic policy, particularly the budget.

While restrictive monetary policy remains an important instrument to foster convergence, particularly while fiscal policy is relatively unambitious, it cannot be the only instrument. If convergence is to be achieved within the time frame set forth in the authorities’ revised Convergence Plan, other policies will have to act consistently. In particular, while fiscal consolidation is at present an objective in itself, fiscal restraint is greatly needed to support the disinflation process, particularly in the present circumstances where inflation expectations appear to be extremely persistent.

Moreover, wages have to be more responsive to the conditions in the labor market. In this respect it is particularly important to reform the system of collective bargaining by reducing the number of levels involved and to push forward with reforms that aim at reducing firing costs. By reducing the protection extended to ‘insiders” it should be possible to increase the responsiveness of wages to economic conditions. Similarly, there is a continuing need for structural policies, to enhance competition and productivity growth, particularly in the services sector, which is characterized by excessive market power and pervasive regulations.

ANNEX A Model of Inflation Persistence

1. Introduction

In this annex a model of inflation in an open economy that produces tradables and nontradable goods is presented, which attempts to incorporate some stylized facts of the Spanish inflationary process. 1/ The model illustrates the fact that if exchange rate management is such that inflation differentials vis-à-vis trade partners is accommodated by nominal exchange rate depreciation, the inflationary process will tend to be more persistent. This is so because agents are aware of the policy rules and, thus, understand that when policies are accommodating they can be less concerned about the unemployment consequences of wage increases. As a result wages and prices will adjust sluggishly to shocks. The same outcome obtains if an anti-inflationary package is not credible and the Government is expected to renege in its preannounced commitments. A corollary is that inflation will tend to be less persistent under credible fixed exchange rate regimes than under a managed floating regime. The model shows, however, that if the labor market is characterized by indexation or by other inertia-generating institutional features, inflation will display some persistence even in the case in which the authorities choose not to follow accommodative monetary and exchange rate policies. The model also incorporates the possibility that the inflationary process be influenced by the different rates of productivity growth in the tradables and nontradables sector.

2. A model of inflation persistence

Consider an economy that produces two types of goods: tradables and nontradables. Tradable prices are assumed to be linked to international prices while nontradable prices are determined by domestic conditions. The model is given in the following equations:


Equation (1) states that the domestic rate of inflation is a weighted average of tradables and nontradables inflation, as denoted by the superscripts T and MT, respectively. Equation (2) states that the law of one price holds ex-ante and that the change in the domestic price of tradables is equal to the expected rate of change in the exchange rate plus the expected rate of world inflation. We assume firms in the nontradables sector are monopolistic competitive price setters and, thus, in equation (3) nontradable prices are determined by a markup over unit-labor costs. In this formulation, the rate of inflation in nontradables equals the rate of change in wages in the sector (wNT), net of productivity increases (aNT). Equation (4) describes the determination of nominal wages in the nontradables sector. A “spillover effect” is assumed, by which nominal wage growth is the same in the two sectors, and is based on average productivity growth (a) 1/ and on the assumption that nominal wage growth responds positively to demand pressures, as measured by the deviations of the log of domestic output (Y) from the log of its potential level (YF). In addition, wage growth is assumed to depend on lagged inflation πt-1 up to k periods, and on the expected rate of inflation πt. The parameter γk in the equation measures the degree of indexation in the economy. Finally, Equation (5) is a specification of aggregate demand growth as a function of the rate of growth of the real money supply (m-π)t and of real exchange rate depreciation, (etπt+πt*),ωt is a real product demand shock, such as a change in fiscal policy and the parameters Ψ and β are positive.

The model is completed by two equations representing the policy rules followed by the authorities. In equation (6) the authorities adjust the nominal exchange rate so as to accommodate the differential between domestic and foreign inflation in the previous period, according to the coefficient φ. In equation (7) the authorities set the money supply so as to compensate for previous period inflation, where θ is the accommodation coefficient. Alternative exchange regimes can be characterized by different values of the accommodation parameter φ. A fixed exchange rate system corresponds to φ=0, whereas a PPP rule for exchange rates to φ=1. A managed floating system corresponds to a 0<φ<1. 2/


The model can be solved in order to find an expression for the dynamics of inflation. In order to simplify the discussion we assume initially that wages are adjusted according to inflation in the last period only (k=1) and that the wage adjustment rule is a strict weighted average of past inflation and expected inflation, i.e., σ=1-γ. Moreover, under rational expectations, we can write Et(πt)=πt and we can rewrite (4) as:


Assuming, in addition, that Et1(πt*)=πt1* we can write, after some algebra:




The coefficient ξ1 provides a measure of the degree of inflation persistence. If ξ1=0, then last period’s inflation has no autonomous impact on the present inflation rate. As ξ1 approaches 1, the degree of persistence increases. A coefficient ξ1=1 means that, unless the other determinants of inflation operate to offset it, inflation this period will reproduce last period’s rate. An analysis of the coefficients ξi reveals that they are made up of combinations of the policy parameters φ, θ and γ, of the elasticities β and Ψ, as well as of the share of tradable goods α. Clearly, for a given set of “structural” parameters β, Ψ and α, changes in the policy parameters φ, θ and γ will change the coefficients ξi in equation (8).

Suppose, for example, that the authorities follow a policy of full accommodation such that φ= θ =1. In this case ξ1=1 and there is full persistence, irrespective of the degree of wage indexation. 1/ In this case, inflation in the current period will be at least equal to inflation in the previous period except for any combination of the following:

(1) Productivity in the nontradables sector is growing faster than in the tradables sector;

(2) World inflation has decreased from the last to the present period;

(3) There is slack in the economy (YtF>Yt1);

(4) The economy experiences a positive supply shock.

If, on the other hand, the authorities follow a nonaccommodative policy stance, such that φ = θ = 0, then ξ1 becomes:


and the degree of persistence becomes a function of the structural parameters β, Ψ and α and of the degree of indexation. If there is no indexation of wages, i.e., γ=0, ξ1 is also equal to zero and the inflationary process does not display persistence. Instead it becomes a function of current world inflation, and of the growth rates of potential output and of relative productivity. As the share of tradables in total production increases, (α→1), domestic inflation tends to collapse to world inflation.

If there is full indexation, i.e., if γ=1, then:


and the degree of persistence will depend on how open the economy is, i.e., on the share of tradables (α), on the responsiveness of nominal wages to the output gap in the nontradables sector (δ) and on the parameters of the aggregate demand function. Suppose, for illustration purposes, that the economy produces only nontradables (α = 0). In this case:


and the degree of persistence will depend on how responsive nominal wages are to the output gap. If responsiveness is limited (i.e., as δ0) persistence will increase (ξ11). It can also be easily shown that the degree of persistence falls as the share of tradable goods increases and that, for a given degree of indexation, the greater the responsiveness of aggregate demand to changes in the policy variables, i.e., to a reduction in the real money supply or in the rate of exchange depreciation, the smaller the degree of persistence will be.

3. Econometric investigation and results

The model presented in the previous section suggests that if the adoption of an exchange rate commitment, such as the one that characterizes the ERM, is credible, price setting behavior and wage contracts will reflect the fall in inflationary expectations in the economy, generating a change in regime that will be reflected in a structural break in the dynamics of inflation described by equation 8. In our particular case, this structural break should take place either at the time Spain joined the ERM or at the moment it became clear that the exchange rate policy had changed, although this will be tested for. 1/ Thereafter, the coefficient of lagged inflation in equation 8 should decline, reflecting the fall in the degree of inflation persistence in the economy. This result, however, depends on two basic conditions: first, that the fundamentals are consistent with a non-inflationary path and, second, that the announcement is credible in the sense that it reduces inflationary expectations and is reflected in newly set prices and contracts. If that is not the case, the degree of inflationary persistence will not be reduced and the coefficient β1 in equation 9 below will not fall.

In order to investigate empirically the response of inflation persistence to Spain’s joining the ERM, we initially estimate equations of the following type using quarterly data:


Tests for parameter stability are than run in order to investigate if and when a structural break might have occurred. In addition to the usual Chow break point test for parameter stability, we run a one-step forecast F test in order to identify a structural break in the regression and run a recursive coefficient estimation, which allows us to trace the evolution of the coefficients as more and more of the sample data are used in the estimation. When applied to the coefficient βl this test provides a picture of way persistence evolved over time.

Equation (9) was estimated using OLS, with an heteroskedasticity consistent covariance matrix. The resulting regression is: 2/ 3/


Adjusted R2=0.968; S.E. of Regression = 0.007; F=477.99

As an initial investigation about the stability of the estimated parameters, we looked at the plot of each estimated coefficient against time. Such plot is obtained by estimating the model over ever increasing sub-samples, until the estimating period is the complete sample, and for each regression plotting the value of the coefficient against the latest period in the sample used for the estimation of that regression. Chart 6 shows the plots obtained for the coefficients. It suggests that the coefficient is not stable over time, increasing significantly after 1986, and stabilizing after 1988. 1/

Chart III-6:
Chart III-6:

Recursive Coefficient Estimates and Two Standard Deviation Bands

Citation: IMF Staff Country Reports 1995, 039; 10.5089/9781451811988.002.A003

As further way of investigating this issue a one-step forecast F test was run. At each recursion in the previous exercise, an OLS residual for the one-step ahead forecast was calculated. If the coefficients are constant over the whole estimating period they should all be of the same order of magnitude. If there is a structural break, however, when we reach the observation that precedes it in the recursive estimation, the one-step ahead forecast is likely to appear abnormal when compared with the other residuals. We can plot such residuals against time and check for abnormal behavior. This is done in Chart 7 which also includes error bands of +/- two standard errors around zero. Values of one-step residuals which lie outside these bands are indicative of exceptional values, which suggest a structural break. Such is the case of the observations dated 1984.3 and 1986.4. On the other hand, a CUSUM test based on the plot of the sum of the recursive residuals (Chart 7) suggests parameter stability. Therefore, Chow stability tests were run for these periods. While the Chow F statistics obtained for 1984.3 (0.92) clearly suggests that we can not reject the null hypothesis of coefficient stability, the result for 1986.4 (2.46) suggests a break at the 10 percent confidence level. More importantly, tests for the quarters around the time Spain joined the ERM, show no evidence of a structural break.

Chart III-7:
Chart III-7:

Recursive Least Squares Tests

Citation: IMF Staff Country Reports 1995, 039; 10.5089/9781451811988.002.A003

Next, an equation of the form was estimated:


Adjusted R2=0.967; S.E. of Regression = 0.007; F=377.80

where the variable D is a dummy variable that takes the value of one for the period after the hypothesized structural break (1988, Q3) and zero otherwise. If joining the ERM is effective and credible, the estimated coefficient of β5 should be significantly negative, indicating that the policy successfully reduced the degree of inertia in the system. In the limiting case of immediately and totally adjusting expectations and prices, β5 would totally offset β1, i.e., (β1 = -β5), and persistence would vanish.

The following equation obtained:


Note that although the coefficient of D turns out to be negative, its absolute size is very small. While the t statistic suggests that one cannot reject the hypothesis that it is equal to zero, it should be looked at with caution given the non-stationary character of the series. Indeed, a unit-root test on the residuals of equation (12) suggests that it is a cointegrating relationship.

There is little evidence of a downward shift in inflation persistence and a change in labor and product market behavior under the ERN. It seems, therefore, that Spain’s commitment to a fixed exchange rate system under the system’s rules was not sufficient to lower inflation expectations and thus reduce persistence. It is conceivable that in view of the structure of Spain’s labor market and given the worsening of the fiscal stance, agents did not perceive the exchange commitment as part of a sufficiently credible policy package, and thus did not change their behavior.

Data Definitions

The data for this study are taken from the Bank of Spain tapes, and runs from 1978.1 to 1994.1. Rates of inflation are measured as the log difference between the consumer price index and its fourth lag. The rate of world inflation is that for the OECD. The demand pressure variable is built by deducting potential output from actual output. Potential output is calculated using a production function approach on annual data, and then linearly interpolating the series to derive the quarterly data. Relative productivity growth is the log difference of the ratio of productivity (defined as value-added divided by employment) in the total economy and in the nontradables sector, where the later is defined to be the services sector plus construction.


  • Alogoskoufis, G. (1990) “Exchange Rate Regimes and the Persistence of Inflation”, CEPR Discussion Paper No. 390, March.

  • Alvarez, L. J., Jareno, J. and Sebastian, M. (1993) “Salarios Públicos, Salarios Privados e Inflación Dual”, in Papeles de Economia Española, 56, pp. 5877, Madrid.

    • Search Google Scholar
    • Export Citation
  • Bank of Spain, Annual Report (Madrid), Various Years.

  • Bank of Spain, Economic Bulletin (Madrid), Various Issues.

  • Bank of Spain, Statistical Bulletin (Madrid), Various Issues.

  • Bentolila, S. and O. Blanchard (1990) “Spanish Unemployment”, Economic Policy, No. 10, P.23381, May.

  • Chadha, B., Masson, P. and Guy Meredith:Modals of Inflation and the Costs of Disinflation”, IMF Staff Papers, Vol. 39, No. 2, June 1992, pages 395431.

    • Search Google Scholar
    • Export Citation
  • Calmfors, L. and J. Driffil (1988), “Bargaining Structure, Corporatism and Macroeconomic Performance”, Economic Policy, P.1361, April.

    • Search Google Scholar
    • Export Citation
  • Confederación Empresarial de Madrid - (CEOE 1994)Los Convenios Colectivos en la Comunidad Autonómica de Madrid: Estructura y Contenido”. March.

    • Search Google Scholar
    • Export Citation
  • Coricelli, F. (1990) “Industrial Relations and Macroeconomic Performance: An Application to Spain”, IMF Working Paper No.93, October.

    • Search Google Scholar
    • Export Citation
  • de La Dehesa, G. (1994) “Spain” in The Political Economy of Policy Reform, J. Williamson, Ed., Institute for International Economics, Washington.

    • Search Google Scholar
    • Export Citation
  • Dolado, J. and S. Bentolila (1993) “Who are the Insiders? Wage setting in Spanish Manufacturing Firms”, CEPR Discussion Paper No. 754.

    • Search Google Scholar
    • Export Citation
  • Dolado, J. and S. Bentolila (1994) “Labour Flexibility and Wages: Lessons from Spain”, Economic Policy, No. 18, p. 5399, April.

  • Dornbusch, R. (1980) “Open Economy Macroeconomics”, Basic Books, New York.

  • Drapper, M. (1993) “Indiciación Salarial y Empleo: Un Análisis Desagregado pars el Caso Español”, Moneda Y Credito, No. 197, p.129165.

    • Search Google Scholar
    • Export Citation
  • Edwards, S. (1992) “Exchange Rates as Nominal Anchors”, NBER Working Paper No. 4246, December.

  • FEDEA (1992) E1 Tipo de cambio de el Mercado Unico Y la Unión Monetaria Europea, Juan Pérez-Campanero, Coordinator, Madrid 1992.

  • FEDEA (1992) España Ante la Union Economica y Monetaria en Europa, Juan Pérez-Campanero, Coordinator, Madrid 1992.

  • Franks J. (1994) “Explaining Unemployment in Spain: Structural Change, Cyclical Fluctuations, and Labor Market Rigidities,” IMF Working Paper WP/94/102, September.

    • Search Google Scholar
    • Export Citation
  • Froot, K. and K. Rogoff (1991) “The EMS, the EMU, and the Transition to a Coomon CurrencyNUR Working Paper No. 3684, April.

  • Galy, Michel and Gonzalo Pastor and Thierry Pujol (1993) “Spain: Converging with the European Community”, IMF Occasional Paper no. 101, February.

    • Search Google Scholar
    • Export Citation
  • Jimeno, J. (1993) “The degree of Centralization of Collective Bargaining, the Inflation-Unemployment Trade-off and Macroeconomic Efficiency Revisited. FEDEA, Docu, mento de Trabajo 9202, January.

    • Search Google Scholar
    • Export Citation
  • Jimeno, J. and L. Toharia (1992) “El Mercado de Trabajo Español en el Proceso de Convergencia Hacia la Unión Económica y Monetaria EuropeaPapeles de Economia Española.

    • Search Google Scholar
    • Export Citation
  • Jimeno, J. and L. Toharia (1993) “The Effects of Fixed-term Employment on Wages: Theory and Evidence from Spain. Investigaciones Económicas.

    • Search Google Scholar
    • Export Citation
  • Lopez-Claros, Augusto (1988) “The Search for Efficiency in the Adjustment Process: Spain in the 1980’s”, IMF Occasional Paper no. 57, February.

    • Search Google Scholar
    • Export Citation
  • Malo de Molina, J.L. and E. Ortega (1985) “Estructuras de Ponderación y de Precios Relativos entre los Deflactores de la Contabilidad Nacional”, Banco de España, Research Memoranda, No. 8505.

    • Search Google Scholar
    • Export Citation
  • Maravall, J. (1993) “Politics and Policy: Economic Reforms in Southern Europe” in Pereira, L. et all, Eds., Economic Reforms in New Democracies, Cambridge University Press.

    • Search Google Scholar
    • Export Citation
  • Milner, S and D. Metcalf (1994) “Spanish Pay Setting Institutions and Performance Outcomes”, Banco de España, Documento de Estudios No. 9420.

    • Search Google Scholar
    • Export Citation
  • Vinals, J. (1992) “La Economia Espanola Ante el Mercado Unico: Las claves del proceso de integración en la Comunidad Europee”, in La Economia Española Ante el Mercado Unico Europeo, Alianza Editorial.

    • Search Google Scholar
    • Export Citation

Prepared by D. Gleizer.


Because of data availability, the focus is in the industrial (tradable) sector versus services (nontradable) sector.


The prices of food and industrial goods incorporate to some extent the impact of high service prices, such as transportation and distribution. The different behavior of the consumer prices of industrial goods and the producer prices of such goods suggests high margins of distribution and commercialization. Indeed, from 1986 to 1993 the former grew by an accumulated 37.3 percent whereas the latter by 17.4 percent.


Between 1979 and 1985 several pacts were signed between the principal labor unions and the employers’ confederation and, at times, the Government, involving trade-offs between wage moderation and improvements in working conditions and unemployment benefits. The main economic feature of this process of Concertacion Social was the shift to a forward-looking indexation of wages to inflation. The two major problems of the pacts were their maintenance of the status quo in the labor market, with all the rigidities inherited from the Franco period, and their contribution to a gradual increase in the fiscal deficit due to large increases in social expenditure.


Estimates of fiscal impulses show that the increase in the general government budgetary deficit over the period 1988–91 was essentially due to a stimulative fiscal stance. See Appendix II in SM/94/10 (1/12/94) for details.


The unit margin is defined as residual after the above mentioned costs are deducted from the final demand deflator. Therefore, it incorporates not only unit profits, but also the costs of capital and other factors of production not included in the cost indicator. Also, in the definition of unit labor costs, w is compensation per salaried worker and N is total employment. Therefore, w.N is total remuneration in the economy, including non-salaried workers. Thus, in order to define the unit margin r in the formulation above it is necessary to deduct the compensation imputed to non-salaried workers from the gross operating surplus. This is done by assuming that compensation per worker is the same for salaried and non-salaried workers. The methodology follows that presented in Economic Bulletin, Bank of Spain, December 1990, pg. 67–72.


The direct inflationary impact of net indirect tax increases is, of course, temporary. Nevertheless, the existence of indexation practices and the impact on inflation expectations might cause a one-time tax increase to have inflationary repercussions over several periods.


See section 4.b below.


A recent econometric investigation of the interaction between wages and prices in the Spanish economy, using a multivariate VAR approach, shows that wages are the fundamental explanatory component of prices, but that the opposite is not true. In particular, nominal wages in the private sector behave as an exogenous variable, determining the time profile of price increases in both the industrial and the services sectors (see, Alvarez et al, 1993).


For an extensive analysis of Spain’s labor market and unemployment see Franks (1994).


Estimates by Jimeno and Toharia (1993) suggest a figure close to 75 percent, the difference arising mainly from possible double counting across different bargaining levels.


This last aspect was recently changed. See Chapter II for details.


Four and a half years since 1994, but no longer freely available. For details see Chapter II.


See Jimeno and Toharia (1993) who estimate the wage premium to be between 8 percent and 11 percent.


The dependent economy model analyses the determinants of the ratio of price of nontradables (PN) and the prices of tradables (PT), i.e., it investigates the determinants of the real exchange rate. See, for example, Dornbusch 1980. By looking at the question in terms of rates of change, we can apply the framework to study the differentials in inflation in the two sectors.


The model is akin to Alogoskoufis (1990), but distinguishes between tradable and nontradable sectors and has a different formulation of wage dynamics. Edwards (1992) presents a similar formulation.


This assumption is common in the so-called Scandinavian models of inflation, where it is justified on the basis of either national negotiations or as a result of a “solidarity principle” in the wage policies of labor unions.


Note that in a fixed exchange rate system the exchange and monetary reaction functions are not independent, unless international reserve flows are completely sterilized.


In this type of system, however, the economy typically learns how to live with relative high inflation, and indexation tends to become generalized.


Spain officially joined the ERM on June, 1989. The peseta, however, had been shadowing the system for quite some time.


The figures in parentheses are t statistics. Given that many economic time series are non-stationary, standard and augmented Dickey-Fuller tests were conducted to test for the orders of integration of the series. Since all series were found to be of the same order of integration, a Dickey-Fuller test on the residuals of equation (9) was run to test for cointegration. The unit-root τμ statistic of -6.11 obtained on this stationarity allows us to reject the null hypothesis of nonstationarity, which indicates that the variables do, indeed, cointegrate, confirming the existence of a long run relationship between the variables.


Similar results were obtained when capacity utilization was used as the proxy for demand pressures.


The peak obtained in 1986 is presumably reflecting the introduction of the VAT, a supposition that is consistent with the subsequent fall of the coefficient.

Spain: Selected Background Issues
Author: International Monetary Fund