Journal Issue


International Monetary Fund. Fiscal Affairs Dept.
Published Date:
January 2013
  • ShareShare
Show Summary Details

I. Rethinking the State—Implications for Expenditure Reform

1. This report discusses both general considerations for reforming government spending and specific issues related to Portugal. The current section (Section I) reflects on general options and approaches to efficiency-enhancing expenditure reform, based on the long-standing debate on the appropriate size and functions of the state and the reform experiences of other countries. Section II introduces the specific macroeconomic context and background for reforms in Portugal. The remainder of the report (Sections IIIVII) sets out a possible menu of reform options, using an analytical approach that benchmarks Portugal relative to other European countries in terms of spending and spending outputs/outcomes in various sectors. More specifically, Section III focuses on issues related to government wages and employment; Section IV looks at pensions; Section V discusses non-pension social spending; Section VI reviews various (mostly non-wage) issues in the education sector; and Section VII does the same for the health sector.

A. Benchmarking the Size and Functions of the State in Portugal

2. The economic literature offers sharply different views on the desirable size of the state. The contrasting theoretical views may be illustrated by looking at two “classics”: Richard Musgrave’s Theory of Public Finance and James Buchanan’s and Gordon Tullock’s The Calculus of Consent.1 Musgrave’s book, which provided the first comprehensive and rigorous treatment of public finance, embodied an activist and positive perspective on the role of the public sector in the economy. Musgrave defined three major roles for government: (i) the provision of public goods and other measures to correct for “market failure” in the allocation of resources; (ii) the redistribution of income to achieve an equitable distribution of societal output among households; and (iii) the use of policies to attain high levels of employment with reasonable price stability. In contrast, Buchanan’s and Tullock’s book, explored the potentially deleterious effects of majoritarian politics on the economy and society. They argued that special interest groups and coalitions push government to institute programs that promote their own interests at the expense of society and the economy at large. The result is a powerful tendency toward public sector overexpansion, with rising tax rates accompanying the increasing levels of transfers and benefits. For this reason, Buchanan and Tullock counseled governments to adopt a set of “rules” or a constitution that effectively constrains public sector expansion. Overall, Musgrave on the one hand and Buchanan and Tullock on the other offer sharply different views of the state: an activist and interventionist state that corrects market failures and redistributes income versus a state that has to be tamed to prevent harm to its citizens. The true optimum is likely somewhere in between.

3. Ultimately, however, the size of the state is a matter of political choice, and European voters have generally opted for relatively bigger governments. Government spending (and taxation) in Europe have traditionally been higher (relative to GDP) than in other advanced economies (Figure 1.1). Part of that difference is due to higher than average spending on social protection (pensions and other social benefits) to households. Another key factor that explains the bigger governments in Europe is the strong emphasis on public provision of social services, in particular education and health. The “European Social Model” with its emphasis on social solidarity and the public provision of key services remains a distinguishing element of European economies.

Figure 1.1.Total Government Spending and Primary Spending, 2000 and 2010

(In percent of GDP)

Source: IMF staff calculations based on OECD data.

4. Big governments have usually been linked to lower growth. Research using historical data shows a clear association between the size of government and low growth in advanced economies. For example, in a recent survey Bergh and Henrekson2 conclude that, in wealthy countries, there is a negative correlation between government size and economic growth—if government size were to increase by 10 percentage points, annual growth rates decrease by 0.5 to 1 percent. The overall findings are also supported by other research, e.g., Afonso and Furceri,3 who show that both size and volatility of government revenue and spending are detrimental to growth in OECD and EU countries. Their research also shows that composition effects matter; on the expenditure side, for example, subsidies and government consumption had the most negative impact on growth.4

5. However, while there are thresholds above which more government will do little for growth, government intervention should not only be seen in quantitative terms. Tanzi and Schuknecht have argued that, relative to 1960, when public spending was on average below 30 percent of GDP, they “could not find much evidence that the large growth of government spending over the post-1960 years contributed much to the further achievement of identifiable social and economic objectives.”5 Still, some countries with big governments have enjoyed above-average growth, and some countries with small governments have had stagnant economies. The Scandinavian welfare states, for example, have achieved fairly steady growth over the last decade despite having large governments. However, these nations compensate for high taxes through market- and business-friendly policies in other areas. Clearly, the impact of the state on the economy depends not only on size, but also on the marginal impact of taxes and spending on private economic decisions and on the quality of public services that the state provides. Hence, government policy should not primarily aim for a smaller state per se, but rather for an efficient and effective state, that complements and enables private economic activity. Still, achieving a more efficient and effective state is likely to result in a smaller state, even when this is not the primary goal.

6. Another way of looking at this issue is to consider that there is not necessarily a tradeoff between equity and efficiency when there is waste in government spending. Sapir6 illustrates the idea that more equity tends to be ‘bought’ at some cost to efficiency. However, Sapir’s analysis also shows that countries of Southern Europe (including Portugal and Spain) do badly on both equity and efficiency, while those in Northern Europe scored high in both. These findings are supported by more detailed analyses on the efficiency of producing outputs in different European Economies. There are many such studies, most of which look at specific sectors. For example, for education spending, Mandl, Dierx, and Ilzkovitz,7 find that “Finland, Austria and Portugal spend roughly the same share of GDP on education, but the performance in PISA is very different.” Similarly, Afonso and St. Aubyn8 show that, of the 17 countries in their sample and by some measures of efficiency (output efficiency) “Portugal is the least efficient country. Resources employed by the Portuguese in the education sector yield a PISA result 15.6 percent lower than the one under efficient conditions.” Likewise, in analyzing tertiary education in Europe, St. Aubyn and others 9 find that “Portugal appears in our analysis as a poor performer, both when we consider only research outputs and only teaching outputs.” The various studies, for the education sector and for other sectors, suggest that, in general, Portugal could improve the efficiency of government spending without hurting equity.

B. Rethinking the State to Support Exit from the Crisis

7. An efficient state is essential to lay the groundwork for robust economic growth and sustainable public finances. Portugal confronts the challenge of reducing government budget deficits and public debt while laying the groundwork for sustained and equitable economic growth over the longer term. A broad body of research suggests that expenditure reforms are a crucial ingredient for successful and growth-friendly fiscal adjustments in advanced economies.10 Reforms to government spending on wages and social transfers are especially important for unlocking the economy’s growth potential through their positive effects on investor confidence, private employment, and economic competitiveness.

8. Altering the composition and priorities of government spending can help to remove obstacles to growth, increase equity, and help to exit from the crisis. The need to bring expenditures in line with a sustainable resource envelope should be seen as a challenge to maintain the promises of the European welfare state through better expenditure efficiency. This, however, requires both a new approach to existing problems, and a proactive attitude toward looming future problems. “Smart” public expenditure reforms, i.e., targeted reforms with a clear rationale (rather than across-the-board austerity), are likely to help convince investors about the long term sustainability of public finances and can play an important role in restarting growth. Thus, expenditure reforms must be guided by three principles:

  • Improving efficiency. Improving efficiency is often a two-step process. First, priorities need to be defined and spending allocated to priority activities, while low priority spending should be cut. Second, the efficiency of priority spending can be improved by cutting the input costs of producing given outputs. For example, if it is decided that a given priority output should be produced with lower labor costs, the government can reduce the number of employees and/or remuneration. Efficiency gains can also be achieved by reducing distortions on private economic decisions. For instance, social transfers can be redesigned to reward work and savings.

  • Improving equity. Government spending promotes equity goals. These can be combating inequality between high and low income and the reduction of poverty, for example by better targeting of social spending and reinforcing the social safety net for the poor. Equity between the young and the old (intergenerational equity), requires attention to age-based spending, like the terms of social insurance and pensions.

  • Safeguarding consensus. Potentially, many of the reforms to be considered could promote efficiency and equity goals simultaneously. For example, a growth-inducing retrenchment may have losers in the short term, who would, however, gain in the medium and long term. Making a convincing case for this, places demands on good governance and requires building trust between the government and wider society.

C. Efficiency and Effectiveness of the State

9. An efficient and effective state enables and empowers its citizens to handle the demands of the global economy. In many countries, the state has moved away from being a provider of services (or the sole provider of services) and toward being a setter and enforcer of service standards, while service provision itself is handled by the private sector. Seeing the state as an activator or enabler has important implications in many areas. Taking education as an example, and notwithstanding recent reforms, the Portuguese state still attempts to do (almost) everything: it provides education, sets standards, evaluates (its own) performance, and enforces standards. Yet, the state has been falling behind in providing quality education: of the 50 top schools, 44 are private, 4 are charter schools, and only 2 are public schools. Refocusing the state on setting education standards and evaluating providers of education could promote better equipped and higher-achieving students, which is the ultimate yardstick of success. The state should thus seek options for promoting greater effectiveness while still paying for and ensuring equal access to education. The existence of true competition from alternative providers means that even when the state provides services directly, these services will need to face a “market test” of approval by beneficiaries who have alternatives available.

10. Public sector pay and employment policies need to emphasize competitiveness and providing value for money to the population. A modern enabling state needs to be on par with the private sector in the way it operates—it cannot be seen as sheltering privileges for itself, either in the form of employment conditions or remuneration. International experience is not encouraging: on aggregate, public sector jobs pay too much.11 The reform of public sector pay and employment can boost economic growth by helping reduce private sector labor costs.12 Public sector employment may be reduced in some areas where it seems too high to produce required outputs. The public sector pay premium should be reduced, particularly for jobs that do not require advanced skills, and options should be considered for rewarding the acquisition of new skills and the achievement of good results and outcomes.

11. The state needs to move from passive to active service delivery. There must be greater use of preventive interventions that attack root cause of indentified issues, as well as greater coordination across different interventions and a move away from passive income transfers. Social policies need to be designed with an eye not only to addressing immediate needs of citizens but also support longer-term capacity building. For example, unemployment benefits need to link automatically to retraining and job search support to facilitate finding employment. Child benefits need to be designed to enable, rather than hold back, mothers that wish to return to work and increase their incomes. Long-term care needs to be available in the home so that the elderly can remain independent.

12. Achieving greater efficiency and effectiveness requires the state to look at the impact of all interventions on households and businesses. The provision of social protection still lacks strategic consistency. What matters is the effect of the system as a whole, not the performance of each of its parts taken separately. This has application at the level of system planning and strategic clarity, but also at the level of the individual: fragmented systems frequently interact to produce unwanted effects, such as ‘dependency traps,’ where benefits or regulations penalize attempts by beneficiaries to leave the trap.

D. Equity and Social Cohesion

13. Portugal’s social protection system could do better in mitigating inequalities. The operation of the contributory social protection system reflects the logic of insiders and outsiders and serves to reinforce the gap between rich and poor. In contrast to many other OECD and EU countries, Portugal’s social transfers provide more benefits to upper income groups than to lower income groups, aggravating inequality.13 Particularly in times of fiscal distress and growing concerns about social cohesion, a regressive social protection system looks less and less sustainable both economically and conceptually.

14. Government spending must be focused where it is most needed and where it will have most results. Effective states set clear priorities and pursue them with focused interventions. Compared to the time when welfare states were founded, the overall level of and prosperity of society is much greater. At the same time, inclusion problems faced by the poor are often more intractable. Limited resources will go a longer way if they prioritize help to the bottom of the income distribution, while the remainder of the population, far from being abandoned, is being provided with the tools to help themselves. In this way, an intervention of a given size is magnified and targeted.

15. Successful targeting of state interventions requires improvements in public sector governance. For focused interventions to work, the needs of individuals and families must be assessed transparently and simply. A good social policy requires effective information processing and efficient interactions with individuals and businesses. For example, taxes need to be payable and benefits receivable electronically. Public services need to be subjected to transparent performance monitoring, using outcome-based indicators and league tables that compare performance.

16. The enabling state must also pay more attention to the needs of the young. The existing Portuguese welfare system emphasizes life-long accrual of age-related entitlements. The priority given to older people reflected an expectation of sustained growth, high employment, and limited migration, which have now been disproved. The problems faced by the young—both in entering the labor force and in facing greater income and employment insecurity once they get there—imply a growing distance between those entering the labor market and the more mature population. The issue of intergenerational equity is already important but is likely to increase dramatically in the next decade. The state should thus place greater emphasis on interventions affecting young people’s earning capacity (e.g., specific education interventions), and should temper age-related social protection demands and spending, most notably by promoting active ageing.

E. Consensus and the Sustainability of Reform

17. To be growth enabling, reforms have to be built on consensus. This requires clear ideas of the objectives of social policy, an appreciation of the available options, and an understanding of the crucial distinction between ends and means. Facing issues openly and frankly is the best means of basing reforms on the widest possible consensus in society.

18. An open debate in society enables better reforms to be selected. Recent research shows that, among different types of expenditure retrenchment, the first best option may often be not politically feasible. Frequently, the expenditure items easiest to eliminate in political terms are those that have the greatest long-term beneficial impact, such as investment in human capital. Where this is the case, an open discussion on the tradeoffs implicit in the available policy choices may help to enlarge the effective choice set. A consensus in society as well as the existence of trust could allow, in many cases, a superior package of reforms to be chosen. Trust could enable states to tackle the issue of rights acquired under arrangements that have proven unsustainable, support reform implementation, and enable benefits of reforms to be reaped sooner.

19. The kind of social spending reforms considered in this report can help to foster dialogue and support decision making. A reduction of public spending is clearly needed, but it needs to be achieved with a view to safeguarding social cohesion. An active dialogue with the broader public should highlight potential benefits and the need for various reforms. Building consensus, offering effective social safety nets, and introducing clear procedures (e.g., means testing) can help to alter the perceived balance of costs and benefits of the spending reforms that are needed, and promote both higher economic growth and a more equitable distribution of income.

II. Macroeconomic Context and Background

20. Following a decade of loose fiscal policies, strong adjustment efforts have been underway in 2011–12. By 2010, Portugal’s fiscal position had become unsustainable, with a general government deficit of 10 percent of GDP in 2010 and expenditures at over 51 percent of GDP. Since then, the overall deficit was cut in half and expenditures were reduced to about 45½ percent of GDP.

21. The government plans for further fiscal adjustment during 2013–14, which is needed to return public finances to sustainability. Building on achievements made in 2011–12, the 2013 budget further advances Portugal toward fiscal sustainability. However, also reflecting the partial reinstatement of the 13th and 14th monthly payments to civil servants and pensioners, about 80 percent of the planned adjustment for 2013 will fall on the revenue side of the budget. Government spending, at almost 47 percent of GDP, will still remain high. For 2014, the governments seeks to refocus fiscal adjustment to the expenditure side, where a recent benchmarking exercise14 suggests that the state remains overextended with respect to comparator countries, with particularly high spending on social protection, education, health, and security—which, taken together, account for about 2/3 of total government outlays. To achieve fiscal sustainability, the government sees a need to reduce spending by about €4 billion (about 2.4 percent of GDP) by 2014; about €0.8 billion of these should be put in place already in 2013.

22. The government seeks to ensure that the needed adjustment of spending is carried out within the current constitutional framework, and goes hand-in-hand with increasing efficiency, enhancing equity, and safeguarding social cohesion. The government has called for a review of public spending and for a broader rethinking of the role of the state to guide the review. In this context, it is seeking to identify what functions the state should no longer be involved in, what functions should it continue to be involved in, and how should it be involved in these, given the current constitutional framework and the constraints on affordability. The rethinking (refundação) that has been announced is both useful and timely: while fiscal adjustment is inevitable, it has to be done with a view to improving equity and efficiency, maintaining social cohesion, and avoiding that measures are taken solely for their ease of implementation. The mission team very much shares the governments concerns. As discussed in Section I, the announced rethinking will need to involve moving toward a state that is characterized by both higher equity and higher efficiency, a state that will enable its citizens to participate and compete in a globalized economy, as well as a state that is caring, more flexible, and fully attuned to the changing needs of the population.

23. This report provides some initial inputs into the government’s spending review and its rethinking of the role of the state. Taking into account the experiences of other European countries, the report provides a menu of potential reform options for selected spending areas that are geared toward strengthening equity and efficiency, and some general thoughts on reforming the state. While leaving aside the question of the exact magnitude of expenditure savings that the government should target, the analysis presented in this report offers a menu of options for achieving expenditure savings in key spending areas. The report also indicates areas, particularly in the area of (non-pension) social spending, where the balance of reforms should be more focused on efficiency gains rather than indentifying expenditure savings per se.

III. Government Wages and Employment

A. Background

24. The government wage bill remains relatively large. Over 600,000 people (12.4 percent of the employed workforce) work in the general government sector. Their compensation, at about 10 percent of GDP in 2012 (projected), represents almost one fourth of total primary spending. At this level, the government wage bill is near the EU average, following a steady decline since 2009 (when it stood at 12.7 percent of GDP) that reflects various measures (Figure 3.1). However, these numbers do not take in consideration employees of hospitals in the state-owned enterprise (SOE) sector, which are classified as outside of the general government for statistical purposes,15 even though the health care system is mostly financed from budgetary resources. Including the wage bill of SOE hospitals (near 1½ percent of GDP) results in a much higher underlying wage bill than what appears in the general government accounts.

Figure 3.1.Total Compensation of Public Employees as a Percentage of GDP

Source: OECD, Eurostat, and IMF calculations.

25. The government workforce is concentrated in education, security, and health care, and also comprises a large share of workers with secondary education or less. Employment in these three key areas accounts for nearly 83 percent of all government workers. Across all areas, a large part (about 40 percent) of the workforce handles mostly administrative or operational tasks; typically these are mid-career workers (average age is 48 years), and the overwhelming majority (90 percent) has at most secondary education qualifications. The central government accounts for 75 percent of general government workers; regional and local governments employ the remaining 25 percent.

26. The compensation structure includes multiple contractual arrangements and the extensive use of wage supplements. While recent reforms have simplified the contracting of government workers, there still are three main contractual arrangements: 15 percent have a “permanent contract” (a fully tenured nomeação contract), 65 percent have an “open-ended contract” (a tenured funções públicas contract, which allows dismissal under some conditions16), and 15 percent have an “individual contract” (without tenure and with the possibility of dismissal). In addition, major complexities in the wage structures result from wage supplements, which comprise benefits that apply to all workers (e.g., food, function, productivity) and benefits that apply only to some sectors or occupations (e.g., risk, overtime, responsibility). About 15 percent of total compensation is in the form of such supplements and there is a high degree of wage dispersion across the government workforce (Figure 3.2).

Figure 3.2.Relative Weight of Base and Supplemental Salaries by Function, 2011

(Percent of total pay)

Source: Government of Portugal (SIEP, Oct-12) and IMF calculations.

27. Recent efforts have sought to contain the wage bill, but are being partially reversed, in part due to constitutional constraints. Since 2010, the government has significantly reduced the wage bill through both employment reductions and wage cuts.

  • Employment: The number of public employees was reduced by 3.2 percent in 2011, and a further reduction by 1.4 percent was implemented in the first semester of 2012.17 This was done mainly by attrition and reducing the number of workers on individual contracts.

  • Wages: In nominal terms, government wages were cut on average by 5 percent in 2011. The suspension of the 13th and 14th monthly salaries in 2012 resulted, on average, in an additional 12 percent reduction. Moreover, promotions, performance bonuses, and mobility-related salary changes have been frozen since 2011—a measure that is expected to be in place until at least 2014. However, due to a 2012 Constitutional Court ruling, the 2013 draft budget had to reverse parts of these cuts, reinstating one monthly wage to all public employees. This is projected to increase the government wage bill to 10.4 percent of GDP in 2013. This occurs notwithstanding efforts to reduce by 2 percent the number of employees across the general government, cut overtime compensation by 50 percent, and align sick leave rules to those prevailing in the private sector.

B. Key Issues

28. Public sector reform is overdue. While the wage bill has moved closer to the European average due to temporary cuts (and not considering health care workers in SOE hospitals), Portugal is still an above-average spender if controlled for income per capita (Figure 3.3). Savings achieved to date have helped fiscal consolidation, but have not sufficiently addressed structural reform needs. For example, merit increases cannot continue to remain frozen if performance and efficiency are to be increased. Instead, wage bill reductions should be the result of a clear strategy targeted at reducing wasteful spending in specific areas and aimed at developing a more efficient and effective public sector.

Figure 3.3.Compensation per Employee Divided by GNP Per Capita in 2011

Source: European Commission.

29. Such a reform needs to address key weaknesses, while striking a balance between fiscal consolidation and the need for a more modern and better qualified civil service. It should aim at reducing the number of employees in a targeted manner (i.e., where over-employment exists), further reducing the high public sector wage premium (the difference between public and private pay18), and generating administrative savings through public sector rationalization. This can be done without reducing outputs if overall efficiency is increased. Any reform should be based on measures of a permanent nature, and should therefore be consistent with the existing constitutional constraints. The following paragraphs describe key components of such a strategy.

30. Excess employment, which is concentrated in the education sector and the security forces, is a key concern. While the overall size of general government employment (in percent of the total population) is in line with the EU average, some areas are clearly overstaffed. For example, public employment in education and the security forces amount to around 2.5 and 1 percent of the total population respectively, which is between 5 and 20 percent above the European average in those categories.19 These two areas account for more than 60 percent of the government workforce, significantly above EU comparators.

  • The Ministry of Education (MEC) employs close to 230,000 workers (of which close to 160,000 are teachers and university professors). These numbers are high relative to other countries, particularly given a decreasing the number of students due to demographic trends. According to recent statistics, the ratio of students per teacher is much lower than in most other countries, and, without additional reforms, likely to drop further due to demographic trends (see Section VI). In addition, salaries in education are above the OECD average, particularly for the high end of the compensation scale, with a premium of 25 percent relative to the OECD average pay for primary school teachers, 11 percent for secondary school teachers, and 15 percent for university professors. Portugal was the country with the fourth-largest increase in teachers’ salaries among OECD countries between 2005 and 2010 (Figure 2.4).20

  • The security forces employ nearly 100,000 individuals. Of these, 37,000 fall under the Ministry of Defense and 53,000 under the Ministry of Internal Administration. In addition, there are 1,700 employed in municipal police force, 1,800 in the scientific police force, and 4,000 prison guards.21 Altogether, the security forces represent about 17 percent of government employment, and the density of police forces (470 per 100,000 inhabitants) is also among the highest in Europe (Figure 3.5). Spending for these functions is above EU average, mostly due to the number of active forces, rather than excessive pay. However, excess employment is a concern not only for active forces but also for the reserve forces, due to their impact on the pension system.

Figure 3.4.Growth in Teacher Salaries, 2005–2010

Source: OECD, Education at a Glance (2012).

Figure 3.5.Police Personnel per 100,000 Inhabitants

Source: Eurostat, and IMF calculations.

31. Another key concern is the extensive use of overtime, particularly in the health sector. The relatively high salaries in the health sector reflect overtime compensation (for work beyond the 35 hour public sector work week). In 2011, Portugal had the lowest average regular working hours per year among OECD countries (Figure 3.6), and government employees were paid 60 million hours of overtime. During the last decade, the health sector alone accounted for 35–45 percent of total annual overtime, and overtime pay represents over 1/3 of the total salary of doctors and explains the relative high levels of compensation (Figure 3.7).22 To address these issues, the Ministry of Health has proposed to increase the work week to 40 hours (equal to the private sector) and change work arrangements at hospitals. While this change is limited in scope, and only applies to part of the public sector, it sends a clear signal for improving equity vis-à-vis the private sector.

Figure 3.6.Average Regular Working Hours per Year, 2011

(Hours per year)

Source: OECD and IMF calculations.

Figure 3.7.Compensation of Doctors and Nurses, 2010

Source: OECD and IMF calculations.

32. The fragmentation of services, overlaps of responsibilities, and duplication of tasks across government agencies also contribute to inefficiencies. Portugal should be able to realize savings by making greater use of shared services and coordinated decisions (until now, human resources decisions have been fully delegated to the Ministries, Figure 3.823). Some initiatives are being taken in this respect, but a comprehensive strategy of shared services should be put in place. This would increase synergies, increase economies of scale, and avoid duplication across entities. For example, each ministry has its own payroll system, international relations department, and even security personnel. Even though some of these units may be small and very specialized, an assessment of potential gains from reducing fragmentation could be undertaken across the public sector. Also, some of the less specific back-office functions could be outsourced to the private sector.

Figure 3.8.Delegation of Key HRM Decisions in Central Government, 2010

Source: OECD, and IMF calculations.

33. The relatively flat wage structure is costly and impairs talent attraction. The public wage premium is among the highest in Europe even after recent wage cuts and also after controlling for several factors associated with earnings levels, like educational attainment, labor market experience or managerial responsibilities. Most of this disparity is explained by the relatively high pay provided to workers with lower qualifications. In addition, the wage grid is relatively flat and depends mostly on years of experience rather than performance. Hence, private-sector opportunities (with lower entry salaries but steeper increases for performance than in the public sector) remain more attractive to highly-trained and motivated individuals.

34. Recent measures have focused on incremental savings without addressing structural weaknesses. Pay reductions have focused on high earners which have further flattened the public-sector wage structure. This policy has tried to preserve the incomes at the lower end of the pay scale for equity purposes, but it should be noted that these public-sector employees are not at risk of poverty. Moreover, cuts in public sector wages have failed to reduce the public-private sector compensation gap, which is still above EU levels (Figure 3.9). In addition, the sole reliance on attrition to reduce government employment by 2 percent may not adequately address excess employment in certain areas while adversely affecting the provision of public services in others. Also, the existing mobility scheme is currently not being used to reassign personnel across the government.

Figure 3.9.Ratio of Public to Private Compensation of Employee

Source: European Commission.

C. Reform Options

35. Public sector compensation reform is urgent, and should help modernize and build a more efficient, nimble, and sustainable public administration. This could include three main elements: (i) reforming public pay rates—including basic salaries bonuses, working hours and overtime compensations, with a view of encouraging meritocracy; (ii) targeting a permanent reduction in the number of employees while opening space to attract skilled younger workers; and (iii) increasing shared services and technology in order to limit duplications and overlaps. In carrying out these various reforms, it will be important to achieve permanent changes. For example, reductions of employment in one part of the public sector (e.g., the central government) should not lead to an increase of employment in another part of the public sector (e.g., state-owned enterprises).

36. Reforming public pay rates is a key element of compensation reform. The reform should increase the steepness of the wage scale, linking it to performance rather than years of experience. Specifically, the reform could consider:

  • Adjusting the overall pay scale.24 A revised pay scale could support a streamlined career path and allow for a permanent reduction on the pay level. To avoid the temporary nature of past measures (the 5 percent reduction introduced in 2011 and the suspension of the 14th month pay), starting in 2014, a permanent cut in base salaries could be introduced across the board, that would aim (at a minimum) to attain the same savings generated by the wage cuts of the 2013 budget. For example, reducing the base salary by 3–7 percent could save €325–760 million annually. This would also help to reduce the public wage premium that currently exists, particularly in the lower pay grades, where the private sector tends to adjust faster to the crisis.

  • Implementing additional reductions in wage supplements. Supplements should complement the basic wage only in specific circumstances, such as overtime above 40 hours a week (see below), night-time work, or to compensate for flexible working hours and additional responsibilities. Currently these supplements vary widely across functions and ministries, but on average represent about 14 percent of the monthly pay (1/3 of which comes from overtime). They should not be transformed into a “regular pay” to increase basic wages (as is currently the case), and their value should be also contained. For example, the introduction of an additional reduction between 20 to 30 percent in supplements that are not related to overtime could save €200–300 million per year.

  • Increasing working hours and further limiting overtime pay. The standard government work week amounts to 35 hours, and generates additional spending through overtime pay. Conceptually, there is no reason for government employees to have a shorter work week than much of the private sector (40 hours). Hence, the government work week should be brought in line with private-sector work week, with proper rotating schedules (mainly in health and the security forces). Measures to increase the work week have already been taken in the health sector (Law 62/79), and supplementary pay was reduced by the new labor code in August 2012, but overtime is still remunerated with a 25 percent premium in the first hour and up to 50 percent from the third hour onwards. Simplifying this system further across the board, e.g., by using a flat overtime premium of 15 percent per additional overtime hour, could generate substantial savings. In addition, measures could be taken to address work-time related issues in the education sector. For instance, defining a lesson in primary and secondary education as closer to 60 minutes (instead of the current 45 minutes) could help reduce the relevant wage bill by up to 20 percent. These measures could generate substantial savings: increasing the work week to 40 hours could yield savings of about €150 million; together with the measures related to class time for teachers and flat overtime, total savings could reach up to €300 million per year.

37. Targeted reductions should focus on specific pockets of excessive employment in the public sector, but savings would largely depend on how these are implemented. There are basically two options: targeted reductions of employment or across-the-board reductions by further attrition (Box 1 summarizes international experiences with downsizing of personnel). A targeted approach should clearly identify appropriate goals for workforce reduction across different areas of government. This would imply an assessment of the appropriate level of employment to deliver certain public services. An alternative would be to set a global target number, ideally adapted to capture gaps and surpluses of personnel in different areas (e.g., education and security) and functions (e.g., administrative personnel with low qualifications). Savings from reducing the workforce by between 10 to 20 percent could amount to around €795-2,700 million (0.5–1.6 percent of GDP), distributed across the three different groups highlighted above. Once the target is set, it is crucial to put in place a mechanism for departure with the appropriate incentives and aiming at minimizing costs of adjustment. Several options can be explored:

  • Enhanced use of the special mobility pool (SMP). A more active use of the SMP could yield greater savings over the medium-term without additional up-front costs.25 Usage of the SMP has been low in the past due to inappropriate incentives for both managers and employees.26 A more active use of the SMP generates up-front savings (as the government only pays half of the wage) and can encourage meritocracy if those with below-average performance are targeted for the SMP. A clear mandatory target could be set in order to put a certain number of employees into the SMP, taking into account the rules that currently apply to different contractual arrangements. Participation in the SMP should be temporary, before dismissal from the public sector or re-absorption. The legal act on the SMP could be revised to simplify existing procedures, reassess the maximum duration (limiting it to up to two years), and revise the wage replacement for those in the SMP to decline over time.27 Also, people in the SMP could be given incentives to take up private-sector employment and receipt of SMP pay could be made conditional on participation in training.28

  • Voluntary departures. Providing positive incentives for separation through severance packages (as in Canada in the 1990s, or more recently in Estonia and Latvia) increases up-front costs but can also raise the medium-term savings. The key to success is to provide attractive incentives to separate while retaining a veto option for the government to prevent indispensable and/or high performing employees from leaving the civil service. Voluntary departures are usually the least adversarial option. However, these schemes tend to be expensive in the short term and may not be effective in reducing significantly the workforce if the reductions are not targeted appropriately. Given tight budgetary conditions, this option seems more feasible in the future, for example, after some of the up-front savings through more active use of the SMP have been materialized and fiscal adjustment is on track.

  • A two-stage approach. Naturally, some combinations of the above-mentioned options are possible, and even a different sequencing. Financing permitting, in the first phase, and for a limited amount of time, public workers could be offered a severance package for voluntary exits. In a second phase, dismissals would take place to meet the pre-established target based on objective criteria.29

  • Further reductions by attrition. Untargeted reductions through attrition are not the best option, but, since sizeable targeted reductions could be difficult to implement in the short run, further reductions by attrition could be considered. While this option would not imply upfront costs, it would yield the lowest level of savings over the medium-term. For example, reducing employment by 2 percent would correspond to an entry-to-exit ratio of 1:3.30 To go further, Portugal could introduce a 1:5 replacement ratio (as in Greece) or a complete temporary freeze (as in Spain).

38. Reducing overlap and fragmentation will require additional reorganization of the central public administration. Small savings but greater synergies and efficiencies can be generated by reducing fragmentation of services across different ministries (e.g., unifying payroll systems, international relations departments, and general secretaries from different ministries) or of functions (e.g., security forces where different organizations exist in parallel,31 or health where different medical services coexist for different groups). In addition, once the current Plan for the Reduction and Improvement of Central Administration (PREMAC) has been fully implemented, there are still areas in the central administration that could be considered for further streamlining:

  • Auditing services (inspecções-gerais), could be merged to create a single national auditing office, with major gains for the efficiency and effectiveness of the current auditing functions with the financial perspective included.

  • The concentration of services in the area of economic activities’ inspection could also add savings and efficiency gains; for example, the fusion of inspection services dispersed throughout ministries (e.g., economy, agriculture, culture).

  • The financial autonomy of public institutes could be addressed (including of the public institutes with special status) by effectively applying the legal rules that determine the maintenance of the financial autonomy, namely, the rule related with the 2/3 of own revenues.

  • In some situations, the allocation of identical competencies throughout a wide number of services, for example, at the level of management of EU funds, brings about somewhat dysfunctional arrangements. The number of services dedicated exclusively or mainly to this competence can be rationalized, with a positive impact at the level of the management of funds and the interaction with the funded entities.

Box 1.Approaches to Downsizing Government Employment: International Experience

A wide range of approaches have been used by advanced economies to reduce public sector employment. These included automatic salary cuts imposed on ministries (Australia, Denmark, Finland, New Zealand and Sweden), employment cuts mandated after spending reviews (Australia, Canada, France, Japan, Netherlands, and the United Kingdom), early retirement schemes (Canada), and job cuts and redundancy programs (Australia, Canada, Finland, Netherlands, Sweden, Switzerland, and the United Kingdom). The impact of these programs can be significant (for example, the reduction of public workforce in the United Kingdom, Canada, and Latvia went from 2 to 7 percent of the total public employment).

The experience of different countries suggests that it is important to avoid an across-the-board approach, which results in less durable reductions as rehiring is often needed (Haltiwanger and Singh, 1999). Downsizing that targets specific workers is likely to be more successful in achieving permanent reductions in employment. Achieving these targets, however, often requires the use of many different kinds of instruments that complement severance payments, such as training programs. This approach, however, has bigger fiscal costs, in the short run, than an untargeted across-the-board cut of employment. Country experience suggests that voluntary departure schemes have not been very effective, as they usually fall short of employment reduction targets, and suffer from adverse selection problems.

Sources: OECD, 2011, “Getting it right: restructuring the government workforce,” in Public Servants as Partners for Growth: Toward a Stronger, Leaner, and More Equitable Workforce (Paris: OECD Publishing); R. Holzman and others, 2011, “Severance Pay Programs around the World: History, Rationale, Status, and Reforms (Washington: World Bank); J. Haltiwanger and M. Singh, 1999, “Cross-Country Evidence on Public Sector Retrenchment,” 1999, The World Bank Economic Review, Vol. 13, No. 1.

IV. Pensions

A. Background

39. This section discusses issues related to the equity and efficiency of the public pension system from a macro-fiscal perspective. Portugal currently spends about 14½ percent of GDP in all of its public pension programs.32 Among the advanced economies, this is one of the largest shares of GDP devoted to pension spending. A key question this section tries to address is whether Portugal could achieve better outcomes in terms of equity (e.g., addressing poverty among the elderly) and efficiency (e.g., promoting participation in the formal employment even at older ages) at lower levels of public spending.

40. The public pension system has complex administrative and benefit structures. The system has been unified under the General Contributory Regime (GCR) for all new entrants to the labor force after 2006. Still, for those who entered the labor force prior to 2006, including the majority of the workforce and nearly all pensioners, the system remains fragmented. Some 4 million workers and 3 million pensioners are covered by GCR, and 0.5 million workers and 0.6 million retirees are covered by Caixa Geral de Aposentações (CGA, the scheme for public employees). These systems provide old-age, disability, and survivors’ benefits. In addition, the retirement income system includes complements (minimum pensions) for those who qualify for low pensions, and non-contributory benefits (means-tested social pensions) for those who do not meet the minimum contribution requirements.

41. Public spending on pensions increased rapidly since 2000. Over 2000–2012, public spending on pensions (as a share of GDP) increased from 9 percent to 14½ percent (Figure 4.1). Most of this increase happened already before the crisis. Population ageing only explains about 30 percent (1½ percentage points of GDP) of the increase in spending during the overall period. The remainder is explained by other factors:

Figure 4.1.Public Spending Increase, 2000–2012

Source: IMF staff calculations, OECD.

  • About 45 percent (2½ percentage points of GDP) is due to increases in average pensions relative to GDP per capita. For example, over 2000–2012, average annual old-age pensions under the GCR increased by 75 percent (from €3,130 in 2000 to €5,515 in 2012) compared to an increase in GDP per worker of 40 percent over the same period.33 Part of this increase reflects efforts to protect the most vulnerable during the early 2000s: the minimum pension increased from €2,300 per year in 2000 to over €3,000 per year in 2005.34 In addition, new programs were introduced, including a pension complement for the low-income elderly in 2006 (Complemento Solidário para Idosos), at a cost of about 0.2 percent of GDP.

  • About 10 percent (½ percentage point of GDP) is due to increases in the number of retirees beyond what is implied by ageing. This reflects in part the impact of past reforms, which often led to a wave of new (early-retirement) pension claims by workers, given the tradition of protecting pension rights once claimed. In addition, both the CGA and the GCR have been merging some smaller pension schemes from SOEs and private funds into their systems.35

  • About 15 percent (1 percentage point of GDP) corresponds to macroeconomic factors. For example, employment dropped from about 72 percent of the population 15–64 in 2000 to nearly 65 percent of the population 15–64 in 2012.36

42. The various past pension reforms are projected to generate fiscal savings, but these have been backloaded. In the last two decades, several reforms have made the pension system more affordable (by reducing accrual rates and increasing the years used to estimate pensionable contributions); increased retirement ages (by increasing CGA retirement ages to those of the GCR); equalized benefits across all workers (i.e., for men and women, and the private sector and civil service); increased incentives to participate (by increasing accrual rates by years of contributions, raising penalties for early retirement, and rewarding delayed retirement); helped to fight poverty (by introducing targeted complements); and adapted the system to changing demographics (by introducing automatic adjustment factor linked to increasing longevity).37 However, the reforms have also included extensive grandfathering rules, which protect current retirees while placing the adjustment burden on their children and grandchildren. This has backloaded potential fiscal saving: the reforms are projected to stabilize spending only after 2020 and at relatively high levels.38 While some of the most recent measures have tried to address the length of the transition—e.g., by bringing forward to 2013 some of the initial grandfathering of the CGA retirement rules—further efforts would be needed to reign in the growth of pension spending over the next decade.

B. Key Issues

43. Compared to other countries, Portugal has relatively high average pensions (GCR and CGA), particularly when taking into account its GDP per capita. A useful yardstick for comparing pensions across countries is to look at the economic replacement rate, i.e., the ratio of spending per individual age 65 and older to GDP per worker. The numerator takes into account ageing (e.g., older populations decrease its value, other things equal), the size of benefits per pensioner (e.g., higher benefits per pensioner increase its value, other things equal), and the extent to which early retirement is used (e.g., providing pensions to those younger than 65 increases its value, other things equal). The denominator takes into account the resources available in each country (e.g., countries with higher GDP per capita might be able to afford higher benefits per capita). Using this yardstick suggests that the pension system (GCR and CGA), with an economic replacement rate of currently 37 percent, is substantially more costly from a fiscal perspective than it was in the past (the value for 2000 was 27 percent) and that pensions are relatively higher than in most OECD countries, including Germany, Ireland, Italy, Spain, and Sweden (Figure 4.2). For illustrative purposes only: reducing pensions by about 20 percent—which would generate savings of nearly €4.5 billion (2¾ percent of GDP)—would still result in an economic replacement rate above the level that prevails in countries with similar levels of per-capita GDP and above where Portugal itself was in 2000.

Figure 4.2.Public Pension System Benefit Rates

Source: IMF staff calculations, OECD.

Note: Data for 2012 (Portugal) and 2010 (all other countries). The economic replacement rate is defined as pension spending per individual age 65 and older as a percent of GDP per worker.

44. However, Portugal’s public pension system does not protect adequately against old-age poverty and remains inequitable. While Portugal has one of the highest pension spending ratios in the EU, it also has a high share of population that is at risk-of-poverty in old age (defined as the share of the over-65 population with incomes below 60 percent of median household income).39 Hence, old-age spending is inefficient: each percentage point of pension spending reduces old-age poverty by only 4.3 percentage points, well below the European average of 6.8 percent. High pension spending and high elderly poverty are reflective of an inequitable system, where 40 percent of old age pension spending is received by the top quintile in the income distribution (Table 5.2). This suggests that there may be room to reevaluate benefits without compromising equity goals.

Table 5.1.Non-Pension Social Benefits (Overview)
Type of

Who Is Eligible?Responsible AgencySource of Finance
Social insuranceMaternity/Paternity cash benefitsContributing employers, employees and self employedMinistry of Solidarity and Social Security, Directorate General for Social Security; CGA for programs for public sector employeesCompulsory social insurance scheme (individual and employer contributions & transfers from the general budget)
Long-term care
Death grant
Accident (work)Compulsory private insurance for employees and self-employed personsFinanced by the employers; self-financed for self-employed
Social insurance: Labor market programsUnemployment benefitContributing employers, employees and self employedMinistry of Solidarity and Social Security, Directorate General for Social SecurityCompulsory social insurance scheme (individual and employer contributions & transfers from the general budget)
Social unemployment benefitQuasi non-contributive. For unemployed with lower qualifying periods than required for unemployment benefit or to extend the duration of benefits for unemployment benefit recipients who still remain unemployed.
Social assistanceEducationMeans-tested grant for tertiary level students for living expenses and tuitionMinistry of Education and ScienceFinanced by transfers from the general budget
Housing benefitNon-contributive. Means tested to households below a per-person eligibility threshold based on income; Means-testing is less strict than RSI.Municipal governmentsMunicipal budgets
Child benefitsNon-contributive. Means-tested to families below a certain reference income; threshold is higher than that for RSI benefits.Ministry of Solidarity and Social Security, Directorate General for Social SecurityFinanced by transfers from the general budget
Minimum guaranteed income (social insertion income, RSI)Non-contributive. Means tested to households below a per-person eligibility threshold based on income.Ministry of Solidarity and Social Security, Directorate General for Social Security
Funeral allowancePaid to a person who paid the funeral costs of any member of his family or of any other person residing in Portugal and with no right to a death grantMinistry of Solidarity and Social Security, Directorate General for Social Security
Long-term care cash benefitNon-contributory means-testedSocial Security system and SNS
Source: Information provided to the mission.
Source: Information provided to the mission.
Table 5.2.Distributional Impact of Social Programs(Targeting Accuracy by Population Quintile and Poverty Status)
Quintiles of welfare aggregate

(income), net of all social

assistance transfers

All social protection13.217.616.418.933.813.386.7
All social insurance10.516.815.819.337.610.689.4
Old-age benefits (gross)9.315.414.719.041.69.390.7
Sickness Benefits (gross)
Disability benefits (gross)20.927.422.518.810.421.278.8
Survivor benefit13.319.920.120.925.813.386.7
All labor market programs17.525.625.319.711.917.782.3
Unemployment benefits (gross)17.525.625.319.711.917.782.3
All social assistance43.519.914.413.38.943.656.4
Education-related allowances (gross)
Housing allowances (gross)7.224.715.
Family/Child-related allowances (gross)30.424.
Social Exclusion not elsewhere classified (gross)
Source: World Bank staff calculations based on EU-SILC 2010, Eurostat.Notes: Each benefit’s incidence is the transfer amount received by the group as a percent of total transfers received by the population. Specifically, the benefit incidence is: (Sum of all transfers received by all individuals in the group)/(Sum of all transfers received by all individuals in the population). Aggregated transfer amounts are estimated using household size-weighted expansion factors. Social transfer data in the EU SILC does not include exemptions to medical fees and pharmaceutical copayments, or means-tested benefits for transport and energy.
Source: World Bank staff calculations based on EU-SILC 2010, Eurostat.Notes: Each benefit’s incidence is the transfer amount received by the group as a percent of total transfers received by the population. Specifically, the benefit incidence is: (Sum of all transfers received by all individuals in the group)/(Sum of all transfers received by all individuals in the population). Aggregated transfer amounts are estimated using household size-weighted expansion factors. Social transfer data in the EU SILC does not include exemptions to medical fees and pharmaceutical copayments, or means-tested benefits for transport and energy.

45. Like in many other countries, the pension system faces significant challenges in the decade ahead. The pension reforms carried out to date have imposed much of the adjustment burden on the children and grandchildren of current pensioners (i.e., future pensioners) by backloading the impact of the reforms. As a result, Portugal has both one of the largest projected increases in pension spending over 2010–2020 and one of the smallest projected increases in pension spending over 2020–2060 (Figure 4.3). For example, in terms of benefit reductions, the reforms puts only about a third of the adjustment on those who receive pensions over the next two decades, compared to an average of about half of the adjustment burden in the EU27 (Figure 4.4).40 In addition, benefit eligibility is rapidly curtailed for younger workers: about 80 percent of the adjustment in eligibility occurs within the next two decades compared to about 60 percent in the EU27.

Figure 4.3.Projected Pension Spending Increases, 2010–2020 and 2020–2060

(Percentage points of GDP)

Source: EC 2012 Ageing Report.

Figure 4.4.Projected Reductions in Benefit Ratios and Pension Eligibility (2010–2050) and Projected Reductions Achieved During 2010–2030

(In percent)

Source: IMF staff calculations, EC 2012 Ageing Report.

Figure 4.5.Public Pension Spending and Employment Rates of Older Individuals, 2010

Source: IMF staff calculations, OECD.

Note: Size of bubbles is proportional to ageing (old-age dependency ratios).

46. Implicitly, recent reforms place an additional burden on the current generation workers by requiring them to generate the fiscal space through higher tax payments. While this is transitory, the key question is whether Portugal can afford a long transition toward the more affordable pension system envisaged by the reforms. Put differently, do current retirees want their children and grandchildren to pay both higher taxes today and receive lower pensions when they retire?

47. Recent reforms are likely to contain pension spending in the long run, but spending will continue to increase for the next 10 years still; the system remains highly inequitable; and incentives to retire early remain in place. More specifically:

  • The reforms enacted so far do not contain pension spending in the near term. Public pension spending increased by 0.6 percent of GDP during 2010–12, and is projected to increase by nearly an additional 0.5 percentage points of GDP in 2013, in part due to reinstating the 13th monthly pension which has been offset somewhat by progressive reductions in high pensions.41 As a result of these various changes, pension spending has already increased beyond what was projected for 2010–2020 (by 1 percentage point of GDP).

  • The pension system is not equitable. Workers in the civil service and workers in the private sector receive vastly different pensions. Civil servants, who account for about 15 percent of all retirees, receive 35 percent of all pension spending. The average old-age pension in the CGA (€16,052 per year) is nearly three times higher than the average old-age pension in the GCR (€5,515 per year). The differences in average pensions significantly exceed the differences in average earnings (€1,800 vs. €700 per month). This suggests that civil service pensions carry a premium of about 15 percent relative to private sector pensions.42 This is further exacerbated by considering that civil servants work fewer hours per week and, in many cases (e.g., for the military, diplomats, judges, and justice officers), have a benefit formula that counts more than one year of contributions for each year of work. Supplemental benefits also exist for some groups which receive relatively high pensions (military forces and retirees from SOEs). In addition, in some cases the accumulation of different pensions can lead to relatively high total benefits. For example, pensioners who receive benefits from their own earnings records are also eligible to receive survivor pensions.

  • The reforms have put most of the burden of the adjustment on future generations of retirees. For example, the 2007 pension reform introduced a sustainability factor equal to the ratio of life expectancy at age 65 in a given year and life expectancy at age 65 in 2006. The intuition of this factor was that each generation of new retirees would receive benefits adjusted to reflect the increasing number of years over which they were expected to receive benefits—partially isolating younger workers from paying higher taxes to finance longevity increases. Today, this factor implies a reduction of 3.92 percent on benefits, reflecting an increase in the remaining life expectancy at 65 from 18 years in 2006 to 18.6 years in 2011. However, this adjustment applies only to those retiring after 2007; those who received pensions before that (or were grandfathered) are exempt from this adjustment. This is a particular concern for CGA pensioners, most of which retired at a lower retirement age and without incurring the adjustment factor.

  • The pension system does not foster formal labor force participation. High labor force participation rates help keep pension systems sustainable. The current pension system, however, provides adverse incentives for labor force participation (and increases labor market informality). For example, minimum pensions increase in steps with years of contributions: with a minimum contribution history of 15 years, they increase at 20 years of contributions, and then again at 30 years of contributions. Hence, low income workers, who likely receive a minimum pension, do not gain from contributing in years 16–19 and 21–29. In addition, minimum pensions can introduce incentives to underreport, particularly for those with low incomes. For example, with a minimum pension at about 45 percent of the minimum wage, and assuming an annual accrual rate of 2 percent, workers earning 1 to 1.5 times the minimum wage will receive the same benefit after 15 years of contributions (i.e., 1.5*2%*15=45%). In addition, the current system continues to offer various pathways to early withdrawal from the labor force with a pension, particularly for CGA workers. For example, for the military and police, a “reserve” mechanism is used as an early retirement scheme.43 In the GCR, there continues to be a bridge between unemployment and retirement, which can introduce adverse job search incentives for those who lose their jobs at older ages (¶58).44

48. Overall, the pension system does not deliver “social insurance” in the traditional sense. The GCR remains roughly a flat-rate system—about 90 percent of the pensioners receive the minimum. This makes the GCR similar to social assistance. By contrast, CGA pensions are relatively high—average public pensions are nearly 100 percent of average public wages, reflecting mostly benefits granted under the old system—and are provided only to a small share of the population. This makes the CGA more similar to a private defined benefit system rather than social insurance, although in an unsustainable way.

C. Reform Options

49. The current expenditure review offers an opportunity to take up again the dialogue on pension reform. The fiscal adjustment currently underway presents an opportunity to take stock on the status and direction of public pension policy. Building upon the principles from past pension reforms, lasting reform needs to focus on restoring fiscal sustainability, increasing equity (within the workforce and pensioners and across generations), minimizing the incentives for informality and inactivity, and retaining the ground gained on protecting the most vulnerable.45

50. There are three possible lines of action. The first is to continue with incremental adjustments to pension payments with a view to delivering short-term savings. This might be an effective option to generate savings in the short term without opening the discussion on pensions. As an illustration, a temporary 10 percent across-the-board reduction for all pensions would generate nearly €2¼ billion in savings. However, this approach will not address the underlying equity and incentive problems of the pension system. In addition, since pension spending is projected to increase at least until 2020, this approach might require further rounds of measures over the next decade, increasing the risk of reform fatigue.

  • If such a marginal approach were considered, progressive reductions might be preferable as means to protect the most vulnerable. But the thresholds under which pension reductions are minimized should be appropriately designed to avoid further weakening the link between contributions and benefits. One way to address this is to incorporate a flat cut for all pensions above the minimum. For example, about a 15 percent reduction in pensions above the minimum pension would deliver €1,500 million in savings.

  • The 13th and 14th pension payments could be permanently restored but converted into bonuses that depend on GDP growth. For example, these bonuses could be paid only when GDP growth exceeds some threshold (say, 3 percent in nominal terms).46 To protect the vulnerable, benefits could be maintained for those with pensions under certain value. Savings could be about €1,000 million on years when GDP grows under the selected threshold while protecting pensioners with benefits under the IAS. Savings would decline to about €500 million if the bonuses are maintained for those with pensions under €1,000 per month.

51. The second option is to curtail benefits for future pensioners by shortening the transition period of the reform. This means addressing grandfathering for those who are not yet retired, and tightening some of the pension eligibility rules, particularly for those who would receive pensions from the CGA. This approach can include several measures:

  • Increasing retirement ages. This will entail centering the benefit formula on age 66 instead of age 65. This could generate about €400–600 million per year in savings.

  • Retirement formulas should be equalized for all, adopting the formula prevalent in the private sector, including for those who joined CGA prior to 1993.

  • Eligibility conditions could be equalized for all who retire starting in 2014—including those in the CGA. This means setting retirement age at 65 with no exceptions other than disability; early retirement could be restricted to those with 40 years of contribution.

  • The age at which the minimum pension and other non-contributory old-age benefits are triggered should rise in line with increases in life expectancy—this is an important measure, particularly since most pensioners in the GCR receive the minimum pension and are thus unaffected by the sustainability factor.

  • For the CGA, additional measures could include eliminating all contribution extra credits, including for service before 2006, for all special regimes—for example, for military and police one year counts for more than one year of contributions; eliminating all special provisions—such as the one that increases judges pensions in line with judges wages.

  • Other measures could include equalizing the minimum pensions for GCR and CGA so as to increase equity; increasing the reward for additional years of contributions for low income workers by widening the range of minimum pensions by years of contribution; equalizing the assessment and definition of disability between CGA and the GCR (currently CGA disability pensions are awarded to those unable to perform their current jobs and are never audited after first awarded).

  • Also, minimum pensions could be reviewed to reduce adverse labor market participation incentives. In particular, the minimum pension could be set to increase in strict proportion to the number of years of contributions. This would eliminate the current steps increases (at 20 and 30 years of contributions) which create incentives for informality.

52. The third option is a more radical reform, including to existing pension rights, in order to address inequities and improve incentives. The continued protection of existing pension rights is akin to continued protection of inherent inequities of the system. Also, the international experience suggests that adjustments to existing pension rights could be useful to restore sustainability. For example, in Sweden initial benefits are adjusted to account for improvements in life expectancy—similar to the mechanism in place in Portugal for new pensioners. But then, Swedish pensions are further adjusted to ensure the sustainability of the system, including through nominal reductions. For example, reflecting the crisis, pensions paid were reduced in Sweden by 3 percent in 2010 and by an additional 4.6 percent in 2011.47

  • One option for generating immediate savings and increasing equity is to apply a “sustainability” adjustment to pensioners. For example, applying an immediate reduction based on the improvement in life between 2000 and 2007. This could entail a reduction to current pensioners, including many who received pensions before the sustainability factor was into place. This would be equivalent to an average reduction of about 4 percent in pension spending on those pensions (with savings of about €800 million). To safeguard against poverty, an option could be to apply a sustainability factor only to the portion of pensions above the IAS. This option would reduce spending by up to €500 million.

  • A more structural reform could be considered along the lines of the Swedish pension system. This would include estimating the actuarial balance for the overall pension system and adjusting benefits to reflect adverse developments to this balance.

  • Additional reductions could be applied to CGA pensioners who retired usually with an accrual rate of 2.5–3 percent, much higher than the 2 percent prevailing in the private sector. To restore equity, this would require a reduction of about 20 percent in the CGA pensions (relative to 2012). This could generate savings of nearly €600 million annually.

  • Also, it could be considered to impose a pension cap, say at about 12 times the IAS for all pensions, consistent with the view of the 2007 law to limit accruals to that level for past pensions. This could generate savings of about €200 million annually.

  • Similarly, supplementary pension funds could be phased out. This could be done by first raising the retirement age for those funds to age 70 by 2015, and then offer to those who accrued benefits a lump sum payment equal to a fraction of the net present value of accrued benefits. Another option is to convert them into notional defined contribution plans with adjustment factors that rapidly generate solvency.

  • Finally, the administration structures of the GCR and CGA could be merged. Although this would generate few savings, it would be an important step towards harmonizing the two systems. This merger could improve monitoring and oversight of the pension systems and eliminate duplicative administrative tasks. It would also contribute to the consolidation of national pension policy under one agency.

V. Non-Pension Social Benefits

A. Background

53. Non-pension social benefits include different programs and account for a small and fairly stable share of the total spending envelope.48 These benefits include a relatively large number of programs, comprising contributory (e.g., unemployment benefits) and non-contributory (e.g., minimum guaranteed income) benefits. Spending for these benefits has remained pretty flat over the last decade, and is projected to reach 2.9 percent of GDP in 2012. Also in light of the increasing social needs arising from the crisis, reforms of these various benefits should not necessarily focus on obtaining large spending reductions, but on redeploying budgetary resources to achieve gains in efficiency and equity.

54. Unemployment and family benefits represent a large share of non-pension social benefits. Unemployment benefits currently accounted for 1.5 percent of GDP (in 2012), reflecting the increase in the unemployment rate due to the crisis. Family benefits are another important component of non-pension benefits and include cash benefits for mothers (income-tested), parental leave, and supplementary benefits for children (which are income-tested since 2003 and depend on the age of children). These larger programs coexist with other small benefits directed at social insurance contributors, such as those for disabled children, funeral and death grants, and prenatal child benefits. The main program against social exclusion is the minimum guaranteed income (rendimento social de inserção, RSI), a last-resort social assistance program with job search requirements, that provides benefits equal to the difference between the claimant’s income and a fixed minimum income level, taking into account family composition. The remainder of non-pension social benefits largely consists of various sickness and disability programs.

55. Non-pension social spending is not particularly high compared with EU peers, but some programs are poorly targeted and have relatively high benefits. For example, Portugal’s spending on unemployment benefits is below the EU1549 average, despite relatively high unemployment rates. This mostly reflects a narrow coverage, even though benefits are relatively high for those who qualify. Also, spending on benefits for families and children is smaller than in most EU countries (i.e., only above Greece, Italy, and the Netherlands). Similarly, poverty-focused programs are relatively small in budgetary terms compared to the EU15. About 0.7 percent of GDP is devoted to active labor market programs. Spending on active labor market programs (ALMPs) per unemployed worker is higher than in Spain or Greece, but significantly below the EU15 average (Figure 5.1).

Figure 5.1.Comparison of Spending on Active Labor Market Programs

(Per unemployed person)

Source: Mission calculations based on Eurostat data.

56. There is room for improving the design of many benefits, and reforms should mainly aim at strengthening social protection in a cost-efficient way. For instance, social assistance benefits, which are not linked to social insurance contributions, do not go only to those with low income: in fact, the available data show that 56.4 percent was spent on people other than the poorest 20 percent of the population in 2009 (Table 5.1). In particular, housing allowances and child/family-related benefits, though income tested, could be better targeted. At the same time, there is a need to consider options for strengthening social protection arrangements, particularly in light of the current economic crisis. This may not necessarily result in budgetary savings.

B. Key Issues

57. Overall social protection spending (including pensions) is disproportionately directed at the better off and elderly, and the design of some programs is sub-optimal based on efficiency considerations and labor-market incentives. In 2009, the top quintile of Portuguese income earners received 33.8 percent of total social protection spending, while the bottom quintile received only 13.2 percent (based on the latest available EU-SILC 2010 data).50 In contrast, in the EU15, the highest income quintile averaged 24.8 percent of social protection spending, while the bottom income quintile received 16.3 percent. In Portugal, the larger share of spending received by the better off reflects the targeting of social benefits to those with a social insurance contribution history (e.g., pensioners), which goes hand-in-hand with having had permanent and better-paid jobs. Means-testing applied to less than a third of non-retirement benefits in 2012; the criteria used range from the relatively generous ones for family allowances to the tighter requirements of the minimum income program. In general, there is room for better targeting of family/child, housing, and education-related allowances. Similarly, the job-search and activation incentives are weak or even somewhat negative for some of the programs.

58. Unemployment benefits do not promote activation, and leave important equity concerns. The unemployment benefit system has strict eligibility rules, excluding younger job seekers. For those who qualify, the duration of benefits is long (Figure 5.2) and the benefit amount relatively high. In a recent EU benchmarking exercise, Portugal stands out in terms of the duration of benefits compared to the EU average.51 The reform of unemployment benefits implemented in 2012 introduces some corrections, such as extending benefits to workers with shorter contribution histories and reducing benefits by 10 percent after six months, yet leaves the general character of the system intact. However, the transition of beneficiaries to the new system will be slow; for some of the older unemployed the reform will begin to have an effect only after 2015. This postpones the advantages of the reform in terms of efficiency and equity at a time when these are most needed. Key issues related to unemployment benefits are as follows:

Figure 5.2.Unemployment Benefit Duration, 2010

(Number of months)

Source: Joint European Commission-OECD project, using OECD Tax-Benefits model (See

Note: The figure shows unemployment insurance duration if the recipient has a contribution history 22 years. Portugal implemented a reform limiting duration in 2012, but it currently applies to very few beneficiaries.

  • Benefits are relatively high and available for a fairly long period. Even under the new system—which applies to very few beneficiaries—the cut in the duration of the unemployment benefit for those aged 40 and over is offset by the provision of unemployment assistance (subsídio social de desemprego) for a subsequent period. Also, even after the reduction of the maximum benefit amount (to €1,048), it remains more than double the minimum wage (€485), which itself is high relative to the average wage. This creates a strong disincentive for beneficiaries to go back to work.

  • Coverage is low for the young and high for older workers. Unemployment benefits favor the older and the better off. Tight entitlement conditions still exclude many. Unemployment benefit coverage in 2011 was 8.5 percent for those under age 25, 30 percent for 25–34 year olds, and 71.4 percent for those over 45.52

  • Specific design features reduce work incentives for the unemployed, particularly for older beneficiaries. The system’s design pushes younger workers out and keeps older ones in, and for some of the latter it may operate as a path to early retirement. After drawing on unemployment benefits for up to a maximum of 26 months, older workers can—depending on their contribution history—get subsequent unemployment social benefits for another 26 months. In contrast, workers under the age of 30 can be entitled to as little as five months. Benefits remain almost unchanged for long unemployment spells, and job search requirements are lax. The system also makes it risky for beneficiaries to accept a job if they are not sure they can keep it. Given that a long contribution history translates to higher benefits, accepting a job offer means having to build up again your entitlement history. Moral hazard in favor of remaining unemployed is especially strong for those currently unemployed, as the old (more generous) rules apply as long as they remain unemployed.53 While stricter eligibility criteria would keep fiscal costs down, this would lead to even higher exclusion rates for younger workers.

  • A recovery in labor demand is critical for reducing unemployment. However, once the economy recovers, it will be important to strengthen incentives for workers to search for jobs. Long-term unemployment is significantly above the EU average (Figure 5.3), particularly among older workers. Therefore, the framework for unemployment benefits could be further examined to ensure that it encourages beneficiaries to take jobs once labor market conditions ease.

Figure 5.3.Long-term Unemployment and Short-term Unemployment

(In percent of labor force)

Source: Mission calculations based on Eurostat data.

59. The connection between unemployment benefits and employment policy—particularly ALMPs—could be further strengthened. While Portugal maintains strict eligibility requirements for unemployment benefits, in practice, monitoring and sanctions (e.g., for failure to participate in activation measures) appear far less rigorous. The emphasis on training measures has increased since 2007, with their budget share rising from 37 to 57 percent (Figure 5.4). However, international evidence suggests that job search initiatives are more effective than training measures. Even though the success rate of activation policies is affected by the ongoing crisis (given the compression of the labor market), resources could be reinforced in the area of employment services.

Figure 5.4.Portugal: Spending on Labor Market Policies

Source: Mission calculations based on Eurostat data.

60. For family/child benefits, and notwithstanding the recent introduction of means testing, over two-thirds of the benefits go to the non-poor. Family benefits have been reduced and eligibility criteria tightened; the benefit now goes to families with incomes under a certain threshold (currently €8,803.62 annually). As the eligibility threshold is currently higher than the minimum wage, further targeting could be justified. However, the government currently has no plans to change the eligibility criteria, also given low birth rates and poverty reduction goals. Nevertheless, while the program does increase the incomes of qualifying families, international evidence is rather discouraging on the role of cash benefits in improving fertility (Box 2).

61. Portugal’s minimum guaranteed income scheme (RSI) was designed as a last resort social safety net. The RSI is a social exclusion/insertion benefit that provides cash transfers (minimum guaranteed income) and support for social/professional integration.54 Even though the program seems well targeted, it may contribute to welfare dependency and be subject to abuse. Its main characteristics are as follows:

  • Targeting. Social exclusion/insertion benefits are targeted to the poor. Available data suggest that 89 percent of social exclusion benefits go to the poorest 20 percent of the population (Figure 5.5). The RSI covers 3.6 percent of the population (379,000 beneficiaries as of September 2012) and largely benefits families (70 percent of recipients): about 40 percent of the minimum income beneficiaries are children, overwhelmingly from the lowest income group.

  • Welfare dependency. There is a concern that the RSI may contribute to welfare dependency among program recipients when compared with the potential gains from work. In Portugal, the poorest 20 percent rely much more on social exclusion benefits than is the case on average in the EU. These benefits make up almost 40 percent of the disposable incomes of the poor. While this could indicate that benefits are high, it could also mean that income-generating capacity is low. A non-working household with two adults and three or more children would get a minimum income benefit equivalent to the minimum wage of €485 per month. There are many such households amongst the beneficiaries. Further allowances, such as housing benefits from local authorities and health care coverage, are often added to the RSI benefit package. Hence, the system may provide a disincentive for those of working age to take a job. Given that beneficiaries on the whole are detached from the labor market, it is necessary to conduct detailed profiling analysis of these recipients to make more definitive conclusions regarding the effects of the RSI on welfare dependency and the changes needed to create greater incentives for certain groups of beneficiaries to work. The adoption of a global cap on non-pension benefits could help to avoid this problem.

Figure 5.5.Portugal: Targeting Accuracy of Social Exclusion Benefits, 2009

Source: World Bank staff calculations based on EU-SILC 2010.

Note: Population-weighted average.

62. Many of the smaller non-pension social benefits remain fragmented, overlapping, and duplicative. There are large number of sub-programs in the social welfare budget, which add to administrative costs and complicate monitoring and evaluation efforts. There is significant scope for simplification and harmonization of criteria. For example, there are 93 different sickness benefits classified in 39 different types related to such aspects as length of payment, industries (textiles, energy), and type of disease. The fragmentation and complexity of the social benefit system could lead to error and fraud and increases administration costs.

C. Reform Options

63. There is some room for fine-tuning non-pension social benefits to promote effectiveness and equity. Improving coverage for the unemployed, while enhancing incentives for those who get unemployment benefits to take up work, would reduce poverty and use limited public resources more effectively. Increasing targeting of social assistance is a further option. Recommendations for making spending more effective include options for further targeting and tightening of benefit regimes for the long-term unemployed to improve effectiveness and expand coverage would yield up to €665 million in gross budgetary savings. However, a large part of these potential savings should be used to strengthen the overall social safety net. The overall thrust of reforming these programs should be to ensure that public money is spent efficiently, i.e., that social benefits go to the right people and provide the correct incentives for people to work, rather than to merely generate budgetary savings.

  • For unemployment benefits, limit duration and introduce a larger and more staggered reduction of benefit amounts. While the government has taken steps in this direction, program design could be improved by expanding coverage and reducing work disincentives (e.g., by strengthening the link between unemployment benefits and activation incentives). Also, bringing the duration of unemployment benefits for those with longer work histories in line with the EU average, would make the unemployed graduate more rapidly into means-tested social assistance benefits, thus increasing the incentives to look for work. For illustration, if all those who are unemployed for more than 10 months were to be moved from the unemployment benefit to a flat benefit equal to the average unemployment assistance benefit (subsídio social de desemprego), the annual savings would be about €300-600 million, while work incentives would be improved. This would impact more than half of current unemployment benefit recipients. Such a system is similar to the flat-benefit job-seeker allowances in the United Kingdom and Ireland. As an option, some of the savings could be used to increase coverage of the young unemployed. Moreover, any new programs would have to be strongly linked to activation policies.

  • Expand means-testing for family allowances. An option is to eliminate eligibility for family allowances to those in the third income group (for those earning €5,869 and over). This would result in annual savings of approximately minimum €89 million (a reduction in 280,000 benefits of €26.54 per month). Also, the provision of benefits for adult children of 19–24 years who are in full-time education could be reconsidered: this group is already benefiting from government funding of tertiary education. Eliminating the benefit for those aged 19–24 could generate savings of about €10 million.

  • Improve targeting of certain benefits and eliminate other benefits. Some programs that serve similar purposes should be reformed. For example, the death grant appears to duplicate the function of the funeral allowance and the remaining dependents are already protected by the survivors benefit. Eliminating the death grant for social insurance contributors and public sector employees would yield additional annual savings (€101.4 million for social insurance beneficiaries and €14.4 million from the CGA). Given the number and fragmentation of benefits, further such examples could be identified and the specific programs consolidated. In the longer-term, the authorities need to investigate each cash and in-kind benefit in terms of effectiveness and targeting and eliminate/target spending, including those given out by local governments. For example, the mean-testing for housing benefits given at the local level should be brought in line with the minimum income program (RSI).

  • Consider a cap on overall or individual social benefits. The government is planning to put in place a system to monitor the total benefits received by households. This would include the range of benefits being provided by the central government (such as RSI and family benefits) and by local governments (such as housing support and other discretionary allowances). In order to ensure that benefit levels do not generate a disincentive to work, the benefit receipt information can provide the basis for placing a total benefit cap for working-age households or to cap individual benefits. For example, if a large number of children or high housing costs are deemed to lead to too high a benefit entitlement, the associated family or housing benefits could be capped. Any policy change directed at capping benefits would have to incorporate into its design possible behavioral consequences (i.e., moral hazard), such as couples reacting to benefit design by living apart.

  • Improve the control of the minimum guaranteed income. While Portugal’s program is considered well-targeted, there may be scope to improve system’s control in order to fight abuse (e.g. if individuals with undeclared incomes benefit from the program).

  • Consolidate the large number of fragmented, duplicate and overlapping subprograms. Small benefits that lack a clear rationale could be eliminated or absorbed into larger programs.

  • Integrate the administration of benefits, ALMPs, and employment services. It will be important to implement the public employment services reform announced in February 2012. These measures would make more effective job search assistance and help to enhance effectiveness by evaluating program performance. Full and coordinated implementation will be critical.

Box 2.Do Family Cash Benefits Increase Fertility?

Though family cash benefits are a feature of many systems, empirical studies on the impact of policies that seek to increase fertility have found little, if any, impact. The evidence on “pronatalist” interventions is that financial measures that attempt to encourage parents to have more children—ranging from birth bonuses and tax breaks for children to more generous allowances to higher-parity births—have individually little or no impact.1 They may induce parents that were in any case going to have a child to have a child earlier. Examining two countries with relatively high fertility rates, Sweden and France, confirms that cash benefits, even rather generous ones, play a limited role. Sweden has shaped interventions to meet the needs of individual and dual breadwinner families, leading some to conclude that the ‘persistent focus on gender equality is a better strategy’ so that ‘women don’t see childbearing as reducing their freedom.’2 In contrast, France has historically stressed male breadwinner models. Even so, since the 1990s an extensive policy of promoting crèche facilities (often means-tested) existed alongside more traditional benefits as part of a constellation of 28 separate interventions.3 Though empirical evidence finds some impact of these policies, it is small and of temporary duration. Therefore, if Portugal wants to engage in policies to raise fertility, it would be well advised to look in directions other than cash benefits to families.

1/ A. Gauthier and D. Philipov, 2008, “Can Policies Enhance Fertility in Europe?” Vienna Yearbook of Population Research 2008, pp. 1–16.2/ See, for example, G. Andersson, 2008, “A Review of Policies and Practices Related to the ‘highest-low’ Fertility of Sweden.” Vienna Yearbook of Population Research 2008, pp. 89–102.3/ A. Pailhé, C. Rossier, and L. Toulemon., 2008, “French Family Policy: Long Tradition and Diversified Measures,” Vienna Yearbook of Population Research 2008, pp. 149–64.

VI. Education Spending

A. Background

64. Portugal’s education system has long underperformed relative to its EU peers, but recent measures have begun to address this issue. At 6.2 percent of GDP (in 2010), education spending was relatively high, including relative to per-capita income. Data for 2006-10 show spending per student as a share of per-capita GDP to be 16 percent above the EU15 average (Figure 6.1), with the wage bill as the main cost driver. However, as many teachers carry a reduced (or zero) workload of teaching assignments, actual class sizes were close to EU and OECD averages. In 2010, education outcomes still were below the EU15 average: for example, while PISA assessment scores have increased sharply, and, in 2010, exceeded those of several EU countries with a higher per-capita income (Figure 6.2), they were still below the EU average. Also, class repetition rates remained high in secondary education. Equity in the allocation of resources is also a concern, with large differences in resources per student across schools, largely reflecting rigidities in teacher pay. Over the last two years, the Portuguese government has been implementing significant reforms focused on strengthening teaching in core subjects, raising curriculum standards (metas curriculares), and improving teacher and student assessments. In addition, to improve cost efficiency and reduce disparities in input indicators relative to other EU countries, the government has also implemented a 10-percentage point reduction in the number of educators.

Figure 6.1.Expenditure Per Student as a Share of GDP Per Capita, 2006–2010

(In percent, period average)

Source: Mission calculations based on Eurostat data.

Figure 6.2.Quality of Education as Measured by PISA, 2009

(PISA score)

Source: Mission calculations based on OECD data.

65. The education system is somewhat more fragmented than in other EU countries, with public institutions covering the vast majority of students. The education system has four cycles rather than the usual three: lower and upper primary, and lower and upper secondary. Public primary and secondary education are financed largely at the central government level, including for the most important component, teacher salaries. Private schools and charter schools now cover about 22 percent of pre-primary, basic, and secondary level students, up from 16 percent in 2000. Charter schools (i.e., schools that are publicly funded but privately operated) are subject to the same educational standards and curricula as public schools. In some cases the whole school is run under a charter agreement while in other cases it is only some classes. As of 2012, there were 81 licensed charter schools covering almost 46,000 students. Their per-student cost is lower, and, on average, they produce better student test scores.

66. There have been significant shifts in the composition of demand for education services. Despite a shrinking student-age population, the number of students in the education system has expanded due to increased enrollments at the pre-primary and secondary levels (Figure 6.3).55 Only for primary schools there has been a drop in the student population, with enrollment falling by 6 percent during 2000–10, reflecting declining birth rates that will continue to affect the education system in the next two decades. Enrollment at the primary level is projected to shrink by 13 percent over 2010–20, and by 9 percent during 2020–30. Smaller primary school age cohorts will impact on secondary enrollment a decade later, with student numbers projected to fall by 10 percent over 2020–30, notwithstanding the 2011 increase in compulsory schooling from grade 9 to grade 12 (i.e., up to 18 years of age).

Figure 6.3.Evolution of the Student Population in Portugal, 2001–2010

(Thousands of students by level of education)

Source: Mission calculations based on Eurostat data.

67. Falling student numbers will necessitate major adjustments to the school network on top of the rationalization that has already begun. For rural areas, the need for adjustment is most severe. While Portugal has a legacy of small primary schools across the country, the adverse overall demographics coupled with out-migration from rural areas has meant that hundreds of such facilities no longer have enough students to justify their existence. Already, the government successfully closed 536 rural school facilities since 2010 and transferred their students to new “school clusters” that are being created to consolidate education services in non-urban areas (including, for example, by sharing teaching as well as administrative/oversight staff); there were 150 new clusters as of end-2012.

68. Meeting an increasing demand for higher education is a challenge, also given the limited scope for student fees. In 2010, 15 percent of the student population was in the tertiary education system, as the number of tertiary level students grew to about 384,000 currently (including private institutions) from less than 200,000 in 1990.56 The growing demand for university degrees was met by both new private and public universities. This has added pressures on public finances, also given that cost recovery in the public university system is limited by the constitutional stipulation that tertiary education should become progressively free of charge. The nominal value of tuition fees was frozen from 1941 until 1993, when, following a Constitutional Court ruling, these were allowed to increase to take into account inflation. As of 2012/13, the maximum undergraduate fee for public universities was €1,037 per year, but some universities chose to charge less. University attendance in Portugal is still highly correlated with family income and private returns to university education are among the highest in the OECD,57 making the impact of tertiary education spending highly regressive (i.e., spending on tertiary education largely benefits the better-off).58 Private institutions are an alternative, but they charge substantially more and require generally lower academic standards for attracting and retaining students, sometimes to the extent of compromising the credibility of their degrees.

69. Public school teachers have remained a relatively privileged group within society in general and within the civil service in particular. The compensation of teacher and other staff comprises about 70 percent of education spending. In 2012, out of 230,000 professional staff in public education, 160,000 are teachers (of which 117,000 are tenured teachers, including over 12,000 in regional or local administrations) (Table 6.1). These permanent teachers earn more than other senior civil servants at the top of their pay scales, and work fewer teaching hours (by earning schedule reduction entitlements). To date, their seniority privileges have not been curbed, and, although there are possibilities for teacher mobility, there is no mechanism to forcefully redeploy permanent teachers from overstaffed schools to schools with shortages of teachers.

Table 6.1.Selected Education Indicators, 2010–13
Year endBudget
(In Millions of Euros)
(Number, End of School Year)
Education employment (by occupation)249,490242,259229,492215,444
Primary education37,06935,83432,95829,933
Secondary education107,546104,17695,98787,003
Higher education24,76825,08724,99424,941
Special education5,2525,5245,5455,550
Primary education10,6419,9989,5879,400
Secondary education44,13942,15941,49340,750
Higher education16,02615,58215,49214,500
Special education0000
In central and regional administrations4,0493,8993,4363,367
Education employment (by work status)249,490242,259229,492215,444
Permanent staff123,427120,355116,768114,441
Fixed term contractuals50,69549,95342,40332,673
Permanent staff49,93948,17948,17747,492
Fixed term contractuals5,3024,0072,5421,875
Number of students systemwide1,881,5051,844,317
Other operational indicators(Units)
Number of “Zero Schedule” education professio4732/985
Number of schools systemwide6,0915,5825,4304,970
University fee revenues (millions of Euros)239245316
Sources: Ministry of Finance, and Ministry of Education and Science.Note: These data reflect national information sources and classifications, and are not necessarily consistent with comparative OECD and Eurostat data and classifications that are used elsewhere in this section. Main conclusions, however, are not significantly affected by these differences.

As of end-2012, the full-time equivalent (that takes into account partial schedules) was 139,633 positions, of which 19,665 were in higher education.

As of end-2012, the number of “zero schedule” professionals was 756.

Sources: Ministry of Finance, and Ministry of Education and Science.Note: These data reflect national information sources and classifications, and are not necessarily consistent with comparative OECD and Eurostat data and classifications that are used elsewhere in this section. Main conclusions, however, are not significantly affected by these differences.

As of end-2012, the full-time equivalent (that takes into account partial schedules) was 139,633 positions, of which 19,665 were in higher education.

As of end-2012, the number of “zero schedule” professionals was 756.

B. Key Issues

70. Portugal’s education system remains overstaffed and relatively inefficient by international standards. As of September 2012, the MEC employed one out of 25 working-age Portuguese as teachers. With 1.5 million (non-tertiary) students in the system, there are 8 students per educator (including teachers performing non-teaching functions). The result is that the cost of paying teachers dominates education spending to a larger degree than in other OECD countries (Table 6.2). Even a mildly ambitious education sector reform that would bring student–teacher ratios closer to the prevailing EU averages for primary and secondary education (Table 6.3) would imply that 50–60,000 staff (teachers and non-teachers) would have to be cut.59

Table 6.2.Decomposition of Education Expenditure, 2009
Primary educationSecondary education
Composition of recurrent costsComposition of recurrent costs
CurrentCapitalTeacherOther staffOtherCurrentCapitalTeacherOther staffOther
Czech Rep.88.111.946.319.534.290.59.547.313.938.8
U. K.87.112.946.830.123.188.411.657.113.329.6
Source: OECD, 2012, Education at Glance, OECD Publishing.
Source: OECD, 2012, Education at Glance, OECD Publishing.
Table 6.3.Student to Teacher Ratios in Selected European Countries, 2010(Calculations Based on Full-Time Teaching Staff Equivalents)
Education Level
Czech Republic18.711.720.0
The Netherlands15.716.514.7
United Kingdom19.816.018.5
OECD average15.813.815.5
EU-21 average14.312.315.8
Source: OECD, 2012, Education at a Glance 2012.
Source: OECD, 2012, Education at a Glance 2012.

71. Prevailing rigidities in the education system further aggravate the overstaffing problem. There is limited autonomy at the school level; notably, schools do not control the hiring of teachers—their largest expense. Teachers can apply to move schools every four years, and compete for available slots on the basis of seniority. Under the system, it is not the more qualified teachers who get redeployed to a position of their choice, but the more senior ones. As a result, schools have little control over their budget or their faculty: schools in desirable locations are forced to absorb more senior teachers with higher salaries and lower teaching-hour requirements.

72. To date, downsizing measures have not targeted the lowest performing or most highly-paid teachers. Portugal’s public sector employment guarantee precludes the laying off of the tenured teachers, and virtually all the 6,500 reduction in permanent staff over the past two years resulted from attrition, with tenured teachers taking retirement. With some exceptions of voluntary separations, the remaining departures (8,300) were fixed-term contract teachers who did not have their contracts renewed.

73. The current school financing model generates wide disparities of per-student spending and of the resources available across schools. Two separate studies—one by Portugal’s Tribunal de Contas60 and one by a working group commissioned by the MEC to examine per-class costs61—noted large disparities in spending per student across the country for the 2009/10 school year. The studies found that schools with higher per-student costs do not necessarily perform better academically. In fact, some schools managed to score well in academic tests although they spent less and served a student population that faced more adverse socioeconomic conditions (Figure 6.4). Costs per student tend to be higher in non-urban areas, where smaller student populations prevent economies of scale. However, in urban areas, teachers tend to be more senior and better-paid, and the schools better equipped.

Figure 6.4.Student Performance, Socioeconomic Conditions, and Per-Student Spending in Selected Schools, 2010/2011

Source: World Bank, based on MEC data.

Note: The socioeconomic background of students at a school is measured by an index that combines data on the share of financial aid and the level of schooling attained by students’ mothers. Only schools with an average student age of 14.5 years are displayed. The vertical and horizontal lines indicate national averages weighted by enrollment.

74. There is evidence that per-student costs are lower in charter schools compared to public schools. Charter schools receive a fixed amount from the state of €85,000 per class (for grades 5 to 12) to administer the same classes and curriculum as public schools. They hire their own teachers and do not charge fees, except for extra-curricular classes and activities. Where they coexist with public schools, they are often the first choice for parents. The Tribunal de Contas study found that charter school costs were lower by about €400 per student than for regular public schools, while the subsequent MEC working group study estimated the difference at only €50 per student after adjusting for expenditure cuts that have been made since 2009/10. While the charter school program is under revision until the end of the current school year,62 the government maintains its intention to promote charter schools as well as freedom of choice between public schools and charter schools.

C. Reform Options

75. A main challenge for Portugal’s education system is to enhance outcomes while reducing costs, i.e., to do more with less. To this end, the government will have to make difficult choices based on cost-efficiency and effectiveness considerations.

76. Continuing the current approach for attrition and school closures would only meet in part the immediate fiscal consolidation goals. With the 2012/13 school year underway, major policy changes can only result in savings after July 2013. As in 2011/12, the default strategy of lowering the number of teachers and administrative staff by a further 14,000 could generate another €300 million (0.2 percent of GDP) in wage savings in 2014, and some productivity gains to the extent that outgoing teachers are replaced with teachers who can deliver more teaching hours. Moreover, any salary savings from attrition would need to take into account additional government expenditures on retirement pensions and unemployment support for departing teachers and other staff.

77. Achieving larger savings would require policy options that are geared toward making the education system more flexible and limiting the state’s role as a supplier of education services. Under this approach, the state’s role would shift more toward setting standards and monitoring their compliance,63 and to pursuing cost recovery in a more equitable way. Key recommendations would be as follows:

  • Implement a simple formula-based funding framework that allows money to follow the student. Portugal could opt for a more challenging reorganization of the education system by applying a per-student financing formula, which has the advantage of enhancing equity, reducing inefficiency, and allowing the system to adapt to changing demographics. This would require giving autonomy to schools to hire teachers, and would need to be accompanied by a closer monitoring of school performance and schools being held accountable for education outcomes. If, under the new system, costs were benchmarked to the charter-school limit of €85,288 per class, per-student cost would fall by at least €400 based on the higher gross public-school costs.64 Assuming savings of €400 per student, and further assuming this number to be the same for the primary and secondary education systems, would result in total savings (before separation payments) of as much as €580 million (0.3 percent of GDP) if applied to the total population of 1.5 million primary and secondary education students. The change would inevitably imply a reduction in teacher numbers and salaries down to the levels in effect at charter schools, but would achieve a redistribution of financing from better-off to poorer schools.

  • Apply the special mobility scheme for surplus teachers. Unlike the current voluntary mobility options for teachers, a compulsory mobility scheme should be created for teachers deemed excessive, triggering an immediate reduction in their remuneration levels. Non-teaching employees of the MEC could also be placed in the general civil service mobility scheme and contribute further to savings. This would not only serve to reduce costs and redeploy teachers, but also as a performance incentive for teachers who want to remain outside the mobility lists. It could be designed to allow permanent teachers to move to charter schools without losing their permanent employment guarantee. The potential savings of placing as many as 30–50 thousand staff on mobility could be about €430–710 million, and more in subsequent years if at least some teachers were to depart the civil service permanently (see Section III on wages and employment).

  • Increase student fees for tertiary education. Higher university fees, taking into account the cost of supplying tertiary education and the market value of the degrees offered, would aid cost recovery and reduce the extent of redistribution to the better-off. In 2012, the public university system spent about €1.6 billion, of which about €1.0 billion was financed from the education budget, €0.3 billion from enrollment fees, and the remainder from other sources. It seems sensible for the public tertiary education to contribute to the ongoing adjustments in the education system, including through further increases in tuition fees that could help to achieve significant and lasting budgetary savings. However, a stronger emphasis on cost recovery should not come at the expense of access to tertiary education, and may require support for low-income students.

VII. Health

A. Background

78. Among advanced economies, Portugal has experienced one of the largest increases in public health spending over the last three decades.65 While Portugal was spending less than 4 percent of GDP on health care in the early 1980s, which was then at the low end of the advanced economies, by 2008 this had grown to about 7 percent of GDP, about the average for advanced economies. While much of the increase occurred in the 1980s and 1990s, public spending on health care (including capital formation) continued to grow through 2010 (e.g., from 6.2 percent of GDP in 2000 to 7 percent of GDP in 2010).66

79. The increases in spending over the last decades have gone hand-in-hand with large improvements in population health status. In 2008, the World Health Organization (WHO) classified Portugal among the top five countries in terms of having made progress in reducing child mortality rates.67 This is reflective of the broad and significant progress in a range of standard population health status indicators, all of which have all shown large improvements over the last 30 years.68 Today, Portugal’s population health status indicators are not very different from other advanced economies. This is a major achievement, and has raised public expectations of high-quality care through the National Health Service (SNS).

80. Yet, the health care system remains fragmented, with three main coexisting and overlapping systems. Portugal’s public health care system is heavily regulated, with very limited patient choice among providers, and gate-keeping functions play an important role.69 The SNS constitutes the largest part of the health care system, and is a universal and largely tax-financed system. The SNS is complemented by special public and private insurance schemes for certain professions (e.g., “health subsystems” for civil servants, the armed forces, and the police) that account for about 20–25 percent of the population, and by voluntary private health insurance.70

81. Also, the progress in population health status has been achieved with an expensive input mix that is heavily tilted towards doctors. At end-2011, there were 25,000 doctors, 39,000 nurses, and 57,000 administrative and support staff according to data by the Ministry of Health. According to OECD data for 201071 (which also counts support staff with medical training), Portugal employs more doctors and fewer nurses: there were 3.8 medical doctors and 5.7 nurses per 1,000 people, which compares to OECD averages of 3.1 for doctors and 8.7 for nurses. There are shortages in medical professionals in some regions. While the Ministry of Health can induce redeployment to these areas, an expected surge in health care professionals is expected to relieve remaining shortages over the next years, but could also pose new problems (e.g., unemployed health professionals).

B. Key Issues

82. In the aftermath of the 2008 financial crisis, it became clear that the health system had become unsustainable and was in urgent need of reform. While public health care expenditures were similar to OECD averages, continued excess cost growth, an ageing population, and a stagnant number of younger contributors, were projected to continue to exert very high upward pressure on health care spending in the absence of reform.72 Without reform, public health spending was projected to increase 2.5–4.6 percentage points of GDP during 2010-2030, significantly above the average for advanced economies.73

83. Since 2011, the government has embarked on an ambitious and comprehensive program of health care reforms, aimed at containing spending while safeguarding health outcomes. The government’s efforts have focused on controlling spending by (i) achieving a more rational use of health services; (ii) limiting the cost of pharmaceuticals to bring, by 2013, spending more in line with the European average of 1 percent of GDP; and (iii) reducing hospital operating costs.74 These consolidation efforts have already reduced public health care spending to about 6½ percent of GDP currently through a number of well-considered measures that will generate further savings in the years ahead.75

84. While ongoing reforms are potentially far-reaching and need to be sustained, additional reforms are needed. Reflecting the trade-offs between competing reform objectives—such as achieving continued improvements in health outcomes and controlling health costs—health care reforms are necessarily complex. In addition, health reforms have to be implemented at a time when large fiscal adjustments are needed to put public finances on a sustainable footing. While one could be led to think that, at about 6 ½ percent of GDP, options for further spending reductions may be somewhat limited, there are significant inefficiencies that need to be addressed to meet the challenges of continued cost and ageing pressures over the next decades, and to ensure equitable access to health care. This is not only the case in Portugal, but also in other advanced economies, with research showing that exploiting potential efficiency gains could result in savings that average 2 percent of GDP in advanced economies by 2017.76 These indications suggest that further reform is warranted.

C. Reform Options

85. Reforms should be targeted at key areas of inefficiencies, where further cost savings could be generated while ensuring high service standards and equitable access.

In particular, this would involve the following:

  • Reducing the over-reliance on medical doctors relative to nurses. With an input mix that relies heavily on medical doctors rather than nurses, many routine tasks that are handled by nurses in other countries are tended to by doctors in the case of Portugal. This generates an input mix that seems unnecessarily costly for Portuguese taxpayers.

  • Limiting the routine use of overtime, particularly for doctors, that currently boosts remuneration. Overtime compensation has been used overtly to boost salaries, especially for doctors. As overtime pay is decided at hospital level, it distorts incentives and creates inequities across the system. The government is trying to reduce the cost of overtime. It recently signed an agreement with the doctors’ association that adds more normal (non-overtime) hours and more patients to family doctors. In addition, the government intends to align overtime pay in the health sector with overtime pay in the overall public sector. While these are encouraging steps, they are unlikely to prove sufficient to bring the remuneration of doctors fully in line with other advanced economies.

  • Reducing the reliance on expensive primary care facilities for cases that could be handled by tertiary care facilities. There is evidence that expensive hospital beds are routinely used for long-term stays of geriatric patients who could instead be handled by more cost-effective tertiary-care facilities. There is a clear difference between private cost and cost to the public: it is cheaper for an elderly patient to stay in a hospital rather than a tertiary care facility, but more expensive to the public. With a growing number of elderly citizens, having an adequate supply of tertiary care facilities becomes a growing concern that needs to be taken up. This issue warrants further study to ensure that incentives are compatible with efficiency considerations, and patients are channeled to the most adequate and cost-effective care available.

  • Reducing the reliance on expensive emergency care for non-emergency situations. Access to emergency care is readily available throughout most of the country, but there are indications that emergency care is frequently used for non-emergency situations. Accordingly, 25 percent of all emergency room visits reflect non-emergency situations that could be handled more cost-effectively by other parts of the public health network, e.g., by primary care practitioners who cover almost the full population.

  • Addressing inefficiencies that result from the fragmentation of the public health care system. The public health system remains overly fragmented through the various health subsystems for civil servants (ADSE), the armed forces, the police, specific state enterprises, and several other groups. Having different subsystems for producing a single output (population health status) necessarily creates inefficiencies, even when different population groups have different needs. As part of the current fiscal adjustment, some consolidation efforts are underway,77 and budget allocations are being capped. While plans exist for increasing the ties between the SNS and the health subsystems, it would seem legitimate to ask whether such plans go far enough in terms of generating efficiency gains. For example, while the armed forces have distinct needs for medical services and facilities, the armed forces and police facilities also provide services that are no different from services that are provided to the general population (e.g., routine exams, and services for family members). While opening the facilities of the subsystems to the general population (e.g., by providing services to the SNS) is one option, further gains could be reaped from fully integrating into the SNS the various health subsystems. This could be achieved over time, and could even involve maintaining an armed forces branch of the SNS, to ensure that the specialized health care needs of the armed forces are being met, including for peak demands. More limited gains could be achieved by shifting family members of armed forces employees to the SNS.

  • Controlling inefficiencies that result from overconsumption of health care services. Taxas moderadoras are fees charged to less than 50 percent of the SNS users (over 50 percent of the population is exempt).78 They are fairly small and recover less than 2 percent of the health care spending. They start at €5 per medical consultation, come to €20 per emergency room visit, and amount to a maximum of €50 for specific diagnostic tests. Taxas moderadoras can be higher for pharmaceuticals (up to 90 percent of the price for certain pharmaceuticals, depending on patient status).79 While the taxas moderadoras have been increased sharply since 2010, they are expected to generate no more that €160 million in revenues in 2012.80 A past Constitutional Court ruling established 1/3 as the maximum cost recovery threshold, a level much higher than the taxas moderadoras that are currently being charged. More cost-sharing would help to address concerns about overconsumption while observing the Constitutional constraints. This could involve better targeting (or means testing) of exemptions, introducing higher charges for nonessential medical procedures, and further increases in the taxas moderadoras in line with safeguarding universal access to health care services.

86. Controlling health care costs without curtailing access or affecting equity can only be achieved by making health care more efficient. With public finances under stress, there is a need to achieve efficiency savings and to rethink the overall system in order to safeguard the main pillars of public health care system, and to continue making progress in improving the health status of the population. While the reforms need to be constrained by fiscal affordability, they have to be guided by considerations of equity and efficiency. The options discussed above could generate both budgetary savings and enhance equity and efficiency. They should be complemented with a number of other interventions. Options for further reform could include, for example, defining more clearly the scope and priorities of publicly funded health benefit packages (i.e., setting clear supply constraints that take into account fiscal affordability); and broadening the role of the private sector, including in health care provision and insurance (including for meeting demands that go beyond the health care benefits provided by the public sector).81

Appendix 1. European Experiences with Integrating Security Forces82

Austria. All Austrian law enforcement agencies were merged into the Federal Police, or Bundespolizei, in 2005. Before the reform, the Austrian police consisted of three forces: the gendarmerie (Bundesgendarmerie), which was responsible for about two thirds of the population and approximately 98 percent of the Austrian territory, the federal safety guard (Bundessicherheitswachekorps), and the judicial police (Kriminalbeamtenkorps). The new federal police force comprises a uniformed generalist police force and specialized bodies in certain areas. The main objectives of the 2005 reform were to (1) increase efficiency of crime prevention and repression, (2) maximize the field presence, and (3) streamline administrative processes, flatten hierarchical levels, and avoid of parallel structures. In the reform process, employee satisfaction was taken into account. Several public debates were held across the country in presence of policemen, politicians, trade unionists, and the Interior Minister. The fusion of the different police forces occurred at all the levels of the hierarchy.

Belgium. The Belgian police used to be fragmented, comprising the gendarmerie (with military status), the local police, and the judicial police. In addition, there were specialized police forces, principally the railroad police and the military police (marine and air force). As a result, the Belgian police system appeared inefficient and insufficiently coordinated. In the early 1990s, the “Pentecôte” plan contributed to demilitarizing the gendarmerie. Previously, under the supervision of the Defense Ministry, with co-responsibility in the Ministries of the Interior and Justice, the gendarmerie was now placed under the sole supervision of the Interior Ministry. Also, it was changed into a classic police force and its recruitment statutes and training were modified. In 1998, with the “Octopus” agreement, the Belgium authorities decided to create a fully integrated police, merging all general and specialized police forces. This integrated police has two levels: the federal level and the local level. The gendarmerie was abolished and its officers were integrated into the new police force. The reform built upon the principles of the “community-oriented policing”: the decentralization (i.e., responsibilities at the lowest possible level of the organization), deconcentration (i.e., spreading of police forces across the territory), and de-specialization (i.e., police forces have broad-based skills to avoid the need for intervention by other services).

Greece. In 1984, the Hellenic Police (Elliniki Astynomia) was created as a result of merging the gendarmerie (Chorofylaki) and the Urban Police Forces (Astynomia Poleon). The aim of abolishing the gendarmerie force was to demilitarize the police system. The police includes central and regional services. The policy force is under the supervision of the Ministery for Citizen Protection and consists of police officers, border guards and special police guards.

Luxembourg. On January 1, 2000, Luxembourg unified its police forces into the Grand Ducal Police (Police Grand-Ducale, PGD), which resulted from a merger of the Gendarmerie and the National Police. The objective was to rationalize means and enhance outcomes. The PGD is under the supervision of Interior Ministry but operates under the ultimate control of the Grand Duke. The PGD is responsible for ensuring Luxembourg’s internal security, maintaining law and order, border control and enforcing all laws and national decrees. It is also responsible for assisting the military in its internal operations.

Richard A. Musgrave, 1959, The Theory of Public Finance, New York: McGraw Hill; and James M. Buchanan and Gordon Tullock, Gordon, 1962, The Calculus of Consent: Logical Foundations of Constitutional Democracy, University of Michigan Press.

See, for example, A. Bergh and M. Henrekson, 2011, “Government Size and Growth: A Survey and Interpretation of the Evidence,” IFN Working Paper No. 858, Stockholm/Sweden; available at

A. Afonso and D. Furceri, 2008, “Government Size, Composition, Volatility and Economic Growth” ECB Working Paper, No. 849 (January 2008), European Central Bank ( Similar results can also be found in A. Afonso and J. Tovar Jalles, 2011, “Economic Performance and Government Size,” ECB Working Paper, No. 1399 (November 2011), European Central Bank (

For revenues, Afonso and Furceri argue that indirect taxes and social contributions are more detrimental to growth than direct taxes. However, their analysis is not in line with theoretical considerations and other empirical studies, which suggest that consumption taxes, recurrent property taxes, and environmental taxes are least detrimental to growth. Hence, a shift away from income (in particular, corporate income) taxation to consumption taxes would be growth-enhancing. See OECD, 2010, Tax Policy Reform and Economic Growth, OECD Publishing, ( Also see D. Prammer, 2011, “Quality of Taxation and the Crisis: Tax Shifts from a Growth Perspective,” Working Paper No. 29, European Commission (DG TAXUD) (, as well as European Commission, 2012, “Tax reforms in EU Member States 2012—Tax Policy Challenges for Economic Growth and Fiscal Sustainability,” European Economy, 6/2012 ( The main take-away from this discussion is that composition effects matter.

See V. Tanzi and L. Schuknecht, 2000, Public Spending in the 20th Century, Cambridge University Press. Also see V. Tanzi and L. Schuknecht, 2003, “Public Finances and Economic Growth in European Countries,” in Fostering Economic Growth in Europe, conference volume of the 31st Economics Conference of the Östereichische Nationalbank, Vienna, 2003, 178–96, available at Today it is certainly more difficult (compared to over 50 years ago) to achieve a spending level of 30 percent of GDP, given higher living standards (and expectations), longer life expectancies, and much older population structures.

A. Sapir, 2006, “Globalization and the Reform of European Social Models,” Journal of Common Market Studies, Vol. 44, No. 2, pp. 369–90 (

U. Mandl, A. Dierx, and F.Ilzkovitz, 2008, “The Effectiveness and Efficiency of Public Spending,” European Economy, Economic Papers No. 301, February 2008 (

A.. Afonso and M. St. Aubyn, 2005, “Non-parametric Approaches to Educational and Health Expenditures Efficiency in OECD Countries”, Journal of Applied Economics, Vol. VIII(2), November, pp.227–46 ( More recent research would support similar conclusions. See, for example, the information on Portugal contained in the data set that is presented in F. Grigoli, 2012, “Public Expenditure in the Slovak Republic: Composition and Technical Efficiency,” IMF Working Paper, WP/12/173 (July 2012), IMF/Washington (

M. St. Aubyn, Á. Pina, F. Garcia, and J. Pais, 2009, “Study on the Efficiency and Effectiveness of Public Spending on Tertiary Education,” European Economy, Economic Papers No. 390, November 2009 (

See, for example, IMF, 2010, From Stimulus to Consolidation, Departmental Paper (Washington: IMF), ( Also see, Margit Molnar, 2012, Fiscal Consolidation: Part 5. What Factors Determine the Success of Consolidation Efforts? OECD Working Paper No. 936 (

See Raffaela Giordano and others, 2011, “The Public Sector Pay Gap in a Selection of Euro Area Countries,” ECB Working Paper, No.1406. Available at

See Silvia Ardagna, 2004, “Fiscal Stabilizations: When Do They Work and Why,” European Economic Review, Vol. 48 (5), pp.1047–74. Available at

A. Lemgruber and M. Soto, 2012, “Growth Friendly, Equitable, and Sustainable Fiscal Reform in Portugal” Selected Issues Paper, prepared for the IMF’s 2012 Article IV Consultation with Portugal (forthcoming).

The benchmarking exercise, which the government carried out earlier this year, analyzes the development of different government expenditures items over time, and compares Portuguese spending patterns and spending outcomes with those of EU comparator countries. The mission team broadly agrees with the analysis.

When public hospitals were transformed into corporations and reclassified in the national statistics (as part of the public enterprise sector), the number of SOE hospital employees increased from 55,000 in 2006 to 96,000 in 2010, and their total compensation increased from 1.7 billion (1 percent of GDP) in 2006 to 3 billion (1.7 percent of GDP) in 2010. This reclassification lowered the government wage bill (in the budget) accordingly. For 2012, the total number of SOE hospital employees was estimated at about 90,000.

Around 90 percent of those with a funções públicas contract received this contract before 2009 and still have full tenure due to a safeguard clause agreed during the reform.

These numbers refer to different universes, given that a broad-coverage statistical database on general government public employment has been launched in 2012. Data on 2011 and earlier years have a narrower coverage (central administration only).

The public wage premium is high in nominal terms and after controlling for skill differences. This premium increases further when controlling for the number of hours worked and holiday entitlements.

For a detailed account on how to evaluate the public wage bill see B. Clements and others, 2010, “Evaluating Government Employment and Compensation,” IMF Technical Notes and Manuals, IMF (Washington DC).

R. Giordano, and others, 2011, “The Public Sector Pay in a Selection of Countries,” ECB Working Paper, No. 1406. These figures do not take into account the latest measures taken by the authorities in 2011–2012, i.e., the average reduction of 5 percent on public-sector salaries and the elimination of the 14th monthly pay.

Data are based on information available at Discussions during the mission suggested that there are additional security forces (e.g., courts) that are not included in these data.

After controlling for purchasing power, the remuneration packages for doctors are above those in Germany, Norway, and Italy.

The Index of HRM Delegation to the Ministries has been built using data from OECD (Government at a Glance, 2011). The index ranges from 1 to 9 and larger numbers imply that all human resources decisions are delegated to the ministries or department units. The dimensions that this index takes into account are the following: wage setting decisions, flexibility of working conditions, allocations of budgets, performance appraisals, performance bonuses, number of posts within units and decisions to hire new employees.

The term “overall pay scale” refers to Tabela Remuneratória Unica.

The SMP is intended for people who are redundant (i.e., in surplus) as a result of reorganizations within the public administration. It pays full salary for the first 2 months, 66 percent of the salary the next 10 months, and 50 percent from one year onwards. In general terms, under the existing rules, the majority of workers in public administration (comprising “nominees” and “ex-nominees”) can stay in the SMP (earning 50 percent of their previous salary and full social security contributions) until they retire.

There was a significant use of the special mobility pool arising from the former streamlining program (PRACE-2008), but from that time onwards, the mobility pool has been rarely used. The number of workers in the mobility scheme as of June 30th, 2012 was 1,144 (based on SIEP data).

Part of the difficulty of using the special mobility pool has been attributed to the cumbersome evaluation model that it entails. In order to speed up the process, one possibility could be to introduce a system of online national exams for the public workforce that generates objective criteria to assign workers to the mobility pool.

For example, the SMP could incorporate retraining activities that retrain participants in new competency areas. To provide incentives (including for voluntary participation in the SMP), this could be combined with an offer to take back those who meet certain qualifications in new positions and at a pay premium. SMP training could be combined with other active labor market tool s (e.g., based on the job rotation experiences of the Nordic countries). Also, a limited number of unemployment insurance recipients could be offered to participate in SMP training activities to provide them with relevant training and experiences that could help them to compete for a future job in the government.

The causes for dismissals of public sector workers would have to be modified, especially for those with contratos em funções públicas.

As reference, in 2011-12, gross attrition has been above 3 percent.

See Appendix 1 for European experiences with consolidating security forces.

As defined here, the term “pensions” includes both contributory and noncontributory benefits for old age, disability, and survivors.

The ongoing pension system maturation might also have put pressure on average benefits in the GCR. For example, new GCR retirees have longer contribution histories—the average number of contribution years taken into account for old-age pension calculations increased from 26 years in 2002 to nearly 30 years in 2010. However, this likely played only a small role, since most GCR pensioners receive minimum pensions and the benefit formulas have changed to account for full contribution careers.

After 2006 the minimum pension has been linked to the Social Support Index (IAS), which has grown in line with GDP per worker.

However, these mergers have largely been neutral for the state budget from an actuarial perspective, as the assets transferred to the treasury roughly equaled the net present value of all future benefits to be paid.

Pension spending to GDP can be decomposed into four factors as follows: spending/GDP = (population 65+/population 15–64) * (spending/pensioners)/(GDP/worker) * (pensioners/population 65+) * (population 15–64/workers). The last component helps to capture macroeconomic effects.

For more information, see Decreto-Lei 277/1993, Decreto-Lei 329/1993, Lei 32/2002, Decreto-Lei 35/2002, Lei 60/2005, Lei 4/2007, Decreto-Lei 187/2007, and Lei 52/2007.

Portugal has one of the lowest projected increases in public pension spending over 2010–2060. See the 2012 Ageing Report of the European Commission (Table 2, Page 38 of

In advanced Europe, only Belgium, Greece, Ireland, Italy, Slovenia, Spain, and the United Kingdom have higher at-risk-of-poverty rates for the over-65 population.

The 2012 Ageing Report (, which Figure 4.5 draws upon, does not yet fully take into account some of the most recent measures which have reduced benefits for current retirees with relative high pensions and thereby somewhat reduced the future adjustment burden.

These projections were finished by end-2011 and assumed the continuation of the 13th and 14th payments. Thus, assuming that only the 14th payment is eliminated would result in an even lower projected increase over 2010–2020 than originally projected. Note also that earlier projections have proven optimistic. For example, the key study underlying the 2002 pension reform projected spending in the GCR to increase by just 1 percentage point of GDP over 2000–2010, about half the actual increase. See Table 10 of “A sustentabilidade financeira do sistema de solidariedade e segurança social: relatório final,” available at

Average pensions are calculated using the total amount of old-age spending projected for 2013 (€6,773 million for CGA million and €11,330 for GCR) and the number of pensioners (393,000 and 2,014,000 respectively). Pensionable wages are from Table 10 of the Ageing Report Country Fiche for Portugal.

Recent reforms have curtailed the use of the reserve system for younger ages, but the retirement remains a possibility at age 60 for the military and police forces.

Experiences of European countries (e.g., France, Greece, Italy) also suggest that relatively high pension spending may lower the incentives to participate in the labor market (Figure 4.5).

Note that these objectives can easily be at odds with each other, and therefore require carefully planned and thought-out measures to avoid moral hazard. For example, focusing on benefit reductions for high-end pensions could potentially erode the link between contributions and benefits and increase incentives for informality.

Hungary replaced the 13th pension with a bonus that depends on GDP growth (

In Sweden, pensions are normally indexed to the rate of growth of earnings minus 1.6 percent. However, when the balance ratio (a measure of actuarial assets to actuarial liabilities) is less than one, pensions are indexed to the growth of earnings minus 1.6 percent minus (1-two year lag of the balance ratio). For example, the balance ratio was 0.9826 in 2008 and the growth rate of wages was 0.3 percent. As a result, pension benefits were reduced by 3 percent in 2010 (0.3-1.6-1.74). See

An overview of non-pension social benefits is given in Table 5.1

The EU15 grouping includes Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain, Sweden, and the United Kingdom.

See Table 5.2. This refers to the EU’s “Statistics on Income, Social Inclusion and Living Conditions.” For more information, refer to Eurostat at

K. Stovicek and A. Turrini. 2012, “Benchmarking Unemployment Benefit Systems.” European Economy, Economic Papers 454, Directorate-General for Economic and Financial Affairs, European Commission, May. (

OECD, 2012, OECD Economic Surveys: Portugal 2012, OECD Publishing (

Current workers will be entitled to the benefits under the old rules for their first post-reform unemployment spell. Hence, to be moved to the new system, a worker who becomes newly unemployed or is already unemployed would need to find a new job and then lose it again. This increases risk aversion against accepting a new job in the first place, unless it comes with reasonable job security.

The higher pre-primary enrollment reflects an increased supply of such facilities in response to demand (e.g., from families with two income earners). The higher secondary enrollment largely reflects a new high-school equivalency program (Novas Oportunidades) that was introduced in 2005. Novas Oportunidades widens options in vocational education, provides double certification, and (at the secondary level) is geared toward reducing the number of drop-outs.

Participation in tertiary education has increased significantly. Currently, total enrollment in tertiary institutions amounts to 62 percent of the 20–24 age cohort, up from 47 percent in 2000.

OECD, 2010, OECD Economic Surveys: Portugal 2010, OECD Publishing. (

See S. da Cruz Martins, R. Mauritti, and A. Firmino da Costa, 2005, Condições Socioeconómicas dos Estudantes do Ensino Superior em Portugal, DGES (Ministry of Science, Technology and Higher Learning), Lisbon (

Calculations take into account the reduction in education sector employment that has already taken place during 2010–12 (Table 6.1) and target average student-teacher ratios that are more in line with other advanced countries (12–15), while also assuming some reductions in non-teacher employment in the education sector.

Tribunal de Contas, 2012, Apuramento do Custo Médio Por Aluno, Relatório N.° 31/2012 (Proc.° n.° 39/2011), (

P. Roseta (Presidente), A. Egídio dos Reis, C. S. Sarrico, L. Carvalho (Grupo de Trabalho Para o Apuramento do Custo Real Dos Alunos do Ensino Público por Ano de Escolaridade), 2012, Estimativas do Custo por Turma do Ensino Básico (2° e 3° ciclos) e Secundário: Relatório Final. August 31, 2012.

An example of such a system is the Netherlands, where education policy is determined centrally, but the administration and management of schools is decentralized. Schools have the freedom to manage financial resources and personnel policy. Families are free to choose the school their child attends. It is easy for new providers to enter the market. The result is a substantial degree of competition in the system. The Dutch school choice system not only performs well academically, but also yields these results at a relatively low cost. See

The €400 figure is flagged in the study of the Tribunal de Contas. The MEC working group estimated per student savings of only €50, but excluded overhead expenditures; if included, much higher savings would result.

See D. Coady and K. Kashiwase, “Public Health Care Spending: Past Trends,” in B. Clements, D. Coady, and S. Gupta (editors), 2012, The Economics of Public Health Care Reform in Advanced and Emerging Economies, International Monetary Fund (Washington DC), pp. 23–36 (

In addition, Portugal has above-average private health expenditures, which pushed total health expenditures up to 10.7 percent of GDP in 2010, more than 1 percentage point of GDP above the OECD average. Overall, the government accounts for only 2/3 of all health care expenditures, which compares to an average of over 75 percent in OECD and EU-27 countries.

World Health Organization, ACS, 2008, The World Health Report 2008:Primary Health Care – Now More than Ever. Geneva, WHO, available at

See, for example, World Health Organization, 2010, Portugal Health System Performance Assessment (, or OECD, 2012, OECD Health Data 2012 (

See I. Joumard, C. André and C. Nicq, 2010, “Health Care Systems: Efficiency and Institutions,” Economics Department Working Papers No. 769, OECD (Paris) (

For a detailed review, see, P. Barros, S. Machado, and J. Simões, 2011, “Portugal: Health System Review,” Health Systems in Transition, 13(4), pp. 1–156. (

See M. Soto, B. Shang, and D. Coady, 2012, “New Projections of Public Health Spending, 2010-50,” in Benedict Clements et al., The Economics of Public Health Care Reform in Advanced and Emerging Economies, International Monetary Fund (Washington DC), pp. 37–52 (

The EC 2012 Ageing Report projects an increase in public health spending of 0.7 percentage points of GDP over 2015–2030. This lower increase relative to the IMF estimates reflects an assumption that technology does not increase costs, which would be a sharp break from past trends. Nevertheless, even under these EC estimates Portugal is projected to experience one of the largest increases in health spending in the EU27 over this period, only surpassed by Austria, Malta, and the Slovak Republic.

In addition, the government has focused on eliminating arrears in the health system. A summary of ongoing and planned health care reforms is contained in the government’s reform agenda as set out in its Letters of Intent to the European Commission and the European Central Bank. See, for example, “European Economy: The Economic Adjustment Programme for Portugal, Fifth Review—Summer 2012,” European Commission Occasional Papers 117 (October 2012), available at, and also earlier Letters of Intent (e.g.,

This has included, for example, higher user fees (taxas moderadoras) for the SNS; changes in the pricing and reimbursement of pharmaceuticals; changes in the prescription system (such as compulsory electronic prescriptions) and the monitoring of prescriptions; changes in distribution mark-ups of pharmaceuticals for retailers and wholesalers; reductions in budget allocations for the health subsystems which, together with a review of the scope of health benefits, intend to make these systems self-financed by 2016; a centralization of purchasing and procurement; changes in primary care services and coordination to prevent unnecessary visits to specialists and emergency rooms; and various measures to reduce operational expenditures in hospitals. The various changes have been implemented carefully so far, including by safeguarding access for the poor. See for example,, regarding exemptions from the taxas moderadoras.

I. Joumard, 2011, “Cross-Country Comparison of Health Care System Efficiency,” Presentation at FAD/EUO Conference on “Public Health Care Reforms: Challenges and Lessons for Advanced and Emerging Europe,” available at

For example, the separate hospitals for the different branches of the security forces are slated for consolidation into one single military hospital (currently under construction) that is to open in 2014.

A total of 5.2 million users are exempt on the basis of “insufficient resources” alone. Also exempt are specific categories of patients, including all children up to 12 years of age (regardless of the income of their parents), blood donors, firefighters, and pregnant women. See, ACCS, Revisão de Categorias de Isenção e Atualizacão das Taxas Moderadoras, available at

There are four levels of taxas moderadoras for pharmaceuticals: level A (90 percent); level B (69 percent); level C (37 percent); and level D (15 percent). Most medicines are included in levels C and D.

Portugal’s Constitution calls for affordable public health care, and increasing taxas moderadoras generates resistance from entrenched interests. As a result, cost recovery from patients, even those who can afford to pay more, is small and partial. An increase of €50 million in user fees was targeted for 2012 but will not be achieved. For 2013, the government projects to collect €190 million.

See for example, J. Tyson et al,, 2012, “Containing Public Health Spending: Lessons from Experiences of Advanced Economies,” in Benedict Clements et al., The Economics of Public Health Care Reform in Advanced and Emerging Economies, International Monetary Fund (Washington DC), pp. 101–23. (

Prepared by Jeanne Pavot.

Other Resources Citing This Publication