Journal Issue
Share
Article

Portugal: Selected Issues

Author(s):
International Monetary Fund
Published Date:
December 1999
Share
  • ShareShare
Show Summary Details

I. The Portuguese Education System: Performance and Reform Agenda1

A. Introduction

13. It is widely accepted that the accumulation of human capital plays a vital role in the process of economic development (e.g., Rebelo, 1991). Given the positive externalities associated with many kinds of education spending, and credit market failures that can lead to underinvestment in human capital by households, there is a clear rationale for public sector involvement in the education sector. What is less clear, however, is the precise role of the public sector, be it through direct provision of educational services, or the financing of private sector provision. A key issue in this regard is the appropriate level of public expenditure, and whether this public expenditure is efficient, that is, is producing educational outputs at the lowest possible cost.

14. These issues are of particular concern to Portugal as it continues its drive to bridge the gap in living standards with the rest of the EU. With a large share of its population having failed to complete secondary education, there is a dearth of workers qualified for intermediate, technical-level positions, which constitutes a major stumbling block to higher labor productivity and economic growth.2 And despite recent improvements, rates of educational attainment for the present school-age population still lag behind much of the EU and OECD, implying little relief from these labor bottlenecks in the near term. At the same time, public education spending, at nearly 5¾ percent of GDP, is well above the OECD average. Given the need to contain the growth of public expenditure under the Stability Program, an improvement in the efficiency of educational spending could help the government improve the output of the education system without raising spending.3

15. This chapter attempts to shed light on these issues by examining the performance of the Portuguese education system and various options that could be explored for improving its efficiency and performance. The chapter is organized as follows. First, an overview of the education system and educational attainment levels over the longer term is presented. Second, public expenditures on education are described. Third, educational performance for the most recent periods is assessed. Fourth, the interface between education and labor market outcomes is addressed. Fifth, the efficiency of spending in Portugal relative to other countries is assessed with the help of a nonparametric production frontier technique called Free Disposable Hull (FDH) analysis. Sixth, recent reform efforts and proposals of the government are discussed. Finally, a discussion of possible directions for future reform and additional areas for further research concludes the chapter.

B. The Portuguese Education System: An Overview

Organization of the system

16. The public system covers the lion’s share of the student population in Portugal (Tables 1 and 2), although the private sector maintains an important presence in tertiary education. Primary or “basic” schooling is comprised of three cycles, with the first cycle covering years one through four; the second, years five and six; and the third, seven through nine. Secondary school covers the years 10 through 12. The minimum years of required schooling was raised from six years to nine years in 1986, which resulted in a significant rise in the number of students at the third cycle of primary education. Primary education follows a common curriculum in all schools determined by the central government. At the secondary level, curricula differ, depending on whether general or technical studies are pursued. All students, however, must pass the same final examination to complete the twelfth grade.

Table 1.Portugal: Student Enrollment at Preschool and Primary Levels, 1985/86–1995/96
PrimaryPrimaryPrimaryPrimary
Preschool1st cycle 1/2nd cycle3rd cycletotal
1985/86817,544388,994375,1621,581,700
1986/87 2/784,264395,064372,3911,551,719
1987/88 2/140,246738,734394,536395,6901,528,960
1988/89 2/154,357717,924372,450429,4221,519,796
1989/90 2/160,129668,033378,531444,6261,491,190
1990/91 2/170,052626,340356,420458,1381,440,898
1991/92 2/176,822613,578354,631496,2461,464,455
1992/93179,135572,762339,244488,9481,400,954
1993/94183,298544,445343,437500,3531,388,235
1994/95185,088539,717321,492506,4741,367,683
1995/96513,671315,209471,8161,300,696
Memorandum items:(In percent)
Public share (1995/96) 3/91.491.891.1
Private share (1995/96) 3/8.68.28.9
Source: Ministry of Education (1998a, 1998d).

Figures refer to continental Portugal only (and thus exclude the autonomous regions of Azores and Madeira).

Preschool figures exclude Azores.

Figures for 1st cycle based on figures that include Azores and Madeira.

Source: Ministry of Education (1998a, 1998d).

Figures refer to continental Portugal only (and thus exclude the autonomous regions of Azores and Madeira).

Preschool figures exclude Azores.

Figures for 1st cycle based on figures that include Azores and Madeira.

Table 2.Portugal: Student Enrollment at Secondary and Tertiary Levels, 1970/71–1997/98(In thousands)
SecondarySecondarySecondary
general studiestechnicaltotalTertiary
1970/7159,683
1975/7686,665
1980/8195,001
1985/86 1/204,46317,488221,951107,789
1986/87219,03123,690242,721
1987/88248,37027,826276,196
1988/89244,21831,741275,959134,162
1989/90272,50936,242308,751155,032
1990/91300,04846,214346,262186,800
1991/92339,78061,483401,263218,300
1992/93347,52268,339415,861246,100
1993/94338,77299,509438,281270,000
1994/95339,074118,120457,194290,400
1995/96 2/336,577119,829456,406313,500
1996/97 2/316,519110,865427,384334,100
1997/98 2/295,208118,718413,926344,900
Memorandum items:(In percent)
Public share (1995/96)87.263.8
Private share (1995/96)12.836.2
Sources: Ministry of Education (1998a); unpublished Ministry of Education data; and Barreto et al. (1996).Note: Data for years before 1985/86 were drawn from Bareto et al. (1996), while data from 1985/86 through 1994/95 were taken from Ministry of Education (1998a).Figures from 1995/96 onward were drawn from the Ministry of Education (1998c, 1998d).Unpublished data provided by the authorities were used for tertiary education estimates for 1990/91 to the present.

Figure for tertiary refers to school year 1984/85.

Figures refer to continental Portugal alone (and thus exclude the autonomous regions of Azores and Madeira) for secondary education.

Sources: Ministry of Education (1998a); unpublished Ministry of Education data; and Barreto et al. (1996).Note: Data for years before 1985/86 were drawn from Bareto et al. (1996), while data from 1985/86 through 1994/95 were taken from Ministry of Education (1998a).Figures from 1995/96 onward were drawn from the Ministry of Education (1998c, 1998d).Unpublished data provided by the authorities were used for tertiary education estimates for 1990/91 to the present.

Figure for tertiary refers to school year 1984/85.

Figures refer to continental Portugal alone (and thus exclude the autonomous regions of Azores and Madeira) for secondary education.

17. The central government finances the vast majority of pubic spending on education, and is responsible for paying teachers’ salaries at the primary and secondary level. The central government also pays for other current expenditures and capital outlays for the second and third cycles of primary education and secondary education. Municipal governments are responsible for transportation outlays for grades one through six, and for capital expenditures for grades one through four. With the exception of spending at the tertiary level, education expenditures in the autonomous regions of Azores and Madeira are financed through their coparticipation in national tax revenues and general transfers from the central government. The public sector finances the provision of education in a number of private sector institutions at the preschool through secondary level, and in 1998 these outlays reached about Esc 60 billion (0.3 percent of GDP).

18. The management of the public schools at the primary and secondary level is centralized in some ways but decentralized in others. Some of the more centralized aspects are the common curricula followed by all schools, and the central government’s role in hiring all teachers and paying all salaries, which follow a uniform pay schedule based on seniority. One of the more decentralized elements of school management is the fact that individual school directors are not appointed by the central or local government, but are elected by teachers.4 Another aspect of management that is decentralized is the method of student evaluation, which varies from school to school (Alaiz and others, 1997).

19. The public universities are autonomous units that receive funding from the central government, and their expenditures are included in the Ministry of Education budget. These transfers are determined by formulas based on the number of students. Some university expenditures are financed by tuition fees. These fees are very modest, as they presently are equal to just one monthly minimum wage (about US$327) per annum. These fees were expected to generate about Esc 12 billion (0.1 percent of GDP) for academic year 1998/99, or less than 5 percent of the Ministry of Education’s outlays for higher education.5

School enrollments and attainment indicators: a long-run view

20. Universal coverage at the primary level was achieved in the early 1970s, with gross enrollment rates exceeding 100 percent of the school-age population.6 Nevertheless, total enrollment in primary education has declined in absolute terms since the mid-1980s, owing to the reduction in the school-age population. At the secondary level, the number of students has climbed steadily, as a result of the extension of the mandatory age of school attendance to 14 in 1987 and success in reducing school dropout rates. The student population at the tertiary level has risen dramatically in the 1990s, more than doubling between 1989/90 and 1997/98.

21. Despite the surge in secondary school attendance, gross enrollment rates continued to be well below the rest of the OECD in the 1970s and 1980s. Throughout those decades gross enrollment rates remained below 60 percent. The high dropout rate reflected difficulties in advancing in school, including at the primary level. These difficulties are also reflected in the figures on repetition rates at the primary level, which hovered near 25 percent in the early 1970s, before declining to below 15 percent by 1990.7 These repetition rates far exceeded those in other OECD countries; a sample of 10 OECD countries in the UNESCO database indicate an average rate of under 3 percent in 1990.8 Low gross enrollment rates and high dropout rates are mirrored in the low rate of educational attainment of the population, especially for older generations. Table 3 indicates that in 1996 only 20 percent of the population aged 25–64 had completed upper secondary education (12 years of schooling), compared with 64 percent for other OECD countries.9 The share of the population that had completed tertiary education was 7 percent, compared with 13 percent for the OECD. As late as 1970 the illiteracy rate exceeded a third of the population, and in that year only 1½ percent of the population had received a tertiary degree.10

Table 3.Percentage of the Population that has Attained a Specific Level of Education(By age group, 1996)
At least upper secondary educationAt least university-level education
Age 25–64Age 25–34Age 35–44Age 45–54Age 55–64Age 25–64Age 25–34Age 35–44Age 45–54Age 55–64
Portugal203224159711964
Other OECD Countries6475685945131515138
Australia5762605446151618148
Austria718275675367754
Belgium5370584731111411106
Canada76858173561720181711
Czech Republic8492878471101112108
Denmark66747065501516171611
Finland6783766040121313127
France6074645638101210105
Germany8186858171131316149
Greece4466523622121614116
Hungary6380756228131415159
Ireland506654383011141196
Italy3852463117881185
Korea6188634125193018117
Netherlands63726657472325252116
New Zealand6065645649111413106
Norway8291877862161917148
Poland 1/7488826847101010128
Spain3050342011131915106
Sweden74878070531311151610
Switzerland808782787110111096
United Kingdom7687817160131515128
United States86878886772626262820
Source: Modified version of Table A1.2a from OECD (1998).

1995 data.

Source: Modified version of Table A1.2a from OECD (1998).

1995 data.

22. Education attainment levels of more recent generations are closer to those of the rest of the OECD, but significant disparities continue to exist. Of the population aged 25–34, only 32 percent had completed secondary education in 1996, compared with 75 percent for the rest of the OECD. Female educational attainment in recent years has outstripped that of males in Portugal, with 36 percent of women aged 25–34 having completed secondary school in 1996. Data from the national employment survey (Inquérito ao Emprego) for 1997 indicate a further strengthening of attainment levels for younger cohorts, with 46 percent of the population aged 20–24 having completed at least secondary education.

C. Education Expenditure

23. Portugal’s expenditure on education, as a share of GDP, is high relative to the OECD average (Table 4).11 At 5.7 percent of GDP, spending for the public sector alone is also higher than in most OECD countries.12 Outlays are especially high for primary and secondary education. Portugal’s high expenditures at the primary level do not appear to be linked to an unfavorable demographic profile, as the share of the population between 5 and 14 years old (12 percent in 1996) was actually one percentage point lower than that in comparator OECD countries. Nevertheless, the percentage of the population attending primary education is more than 20 percent above the European average, due to high repetition rates.13 At the secondary level, Portugal has a relatively large school-age population, as roughly 8 percent of the population is aged 15–19, compared with 7 percent for other OECD countries. Tertiary spending is slightly above the OECD average.14

Table 4.Educational Expenditure from Public and Private Sources for Educational Institutionsas a percentage of GDP by level of education (1995)
Primary and secondary educationTertiary educationAll levels of education
AllPrimarySecondaryAllNon-university-combined (includingAll levels of
Universitylevelpreprimary andeducation combined
undistributed)(public sector only) 1/
Portugal 2/4.52.12.41.56.45.7
Other OECD countries or regions (mean)3.81.42.31.30.21.15.75.0
Australia3.71.62.11.80.31.55.64.5
Austria3.91.22.71.00.10.95.55.3
Belgium (Flemish Community)5.0
Canada4.32.50.91.57.05.8
Czech Republic3.90.92.91.00.11.05.74.8
Denmark4.31.72.61.37.16.5
Finland4.21.82.41.70.31.36.66.6
France4.41.23.21.16.35.8
Germany3.81.10.01.05.84.5
Greece 3/2.81.31.50.80.20.73.73.7
Hungary3.61.12.51.0n.a.1.05.54.9
Iceland3.60.70.00.65.24.5
Ireland3.41.32.11.35.34.7
Italy3.21.12.10.80.00.84.74.5
Japan3.11.31.71.00.10.94.73.6
Korea3.81.72.21.90.41.56.23.6
Luxembourg4.3
Netherlands3.21.22.01.3n.a.1.34.94.6
New Zealand5.3
Norway6.8
Poland5.2
Spain4.0132.71.1n.a.1.15.74.8
Sweden4.52.02.51.76.76.6
Switzerland5.5
United Kingdom1.04.6
United States3.91.82.02.40.42.06.75.0
Sources: Modified version of Tables B1.1a and B1. 1d from OECD (1998); for Portugal, Fund staff estimates based on data provided by the authorities.

Excludes public subsidies to households and students. These averaged 0.13 of GDP for countries reporting these data.

Author’s estimate based on 1996 data. Total spending includes outlays of local governments and independent regions of 0.5 percent of GDP. Figures include private sector spending, which are based on Fund staff estimates using public sector costs per student and private enrollment shares.

Public sector only.

Sources: Modified version of Tables B1.1a and B1. 1d from OECD (1998); for Portugal, Fund staff estimates based on data provided by the authorities.

Excludes public subsidies to households and students. These averaged 0.13 of GDP for countries reporting these data.

Author’s estimate based on 1996 data. Total spending includes outlays of local governments and independent regions of 0.5 percent of GDP. Figures include private sector spending, which are based on Fund staff estimates using public sector costs per student and private enrollment shares.

Public sector only.

24. Measured in terms of expenditure per student in purchasing-parity adjusted dollars, Portugal’s spending is modest at the primary and secondary levels (Table 5). Spending at the tertiary level is also lower than the OECD average, although it exceeds that of some of its richer neighbors (Italy and Spain).

Table 5.Expenditure per Student (U.S. Dollars Converted Using PPPs) on Public and Private Institutions by Level of Education (Based on full-time equivalents, 1995)
TertiaryAll levels of
EarlyNon-University-education
childhoodPrimarySecondaryAlluniversitylevelcombined
Portugal1/2/2,6993,0696,073
Other OECD countries or regions (mean)3,3373,6734,7408,3576,8759,0134,861
Australia3,1214,89910,5907,69911,572
Austria 3/4,9075,5727,1187,94312,8347,6876,763
Belgium (Flemish Community) 4/2,3913,2705,7706,0434,694
Canada5,37811,47110,43412,2176,717
Czech Republic2,0521,9992,8206,7952,5027,6562,885
Denmark4,9645,7136,2478,1575,968
Finland5,9014,2534,9467,3156,9337,4125,323
France3,2423,3796,1826,5695,001
Germany 3/5,2773,3616,2548,8976,8179,0015,972
Greece 4/1,9502,7161,7503,1691,991
Hungary 3/1,3651,5321,5914,792n.a.4,7921,782
Ireland2,1082,1443,3957,2493,272
Italy 3/3,3164,6735,3485,0136,7054,9325,157
Japan2,4764,0654,4658,7686,4099,3374,991
Korea1,4502,1352,3325,2033,9805,7332,829
Netherlands3,0213,1914,3519,026n.a.9,0264,397
New Zealand2,2622,6384,1208,73710,0188,3804,099
Norway 3/9,6476,360
Spain2,5162,6283,4554,9443,9734,9663,374
Sweden3,2875,1895,64313,1685,993
Switzerland 3/2,4365,8937,60115,6858,22618,3657,241
United Kingdom 4/5,0493,3284,2467,2254,222
United States5,3716,81216,2627,97319,9657,905
Sources: Modified version of Table B4.1 from OECD (1998); for Portugal, Fund staff estimates based on data provided by the authorities. Tertiary spending figures for Portugal are from OECD (1998).

Public institutions, continental Portugal, based on 1996 data. Tertiary data refer to 1995.

Staff estimates, except for tertiary spending.

Public institutions.

Public and government-dependent private institutions.

Sources: Modified version of Table B4.1 from OECD (1998); for Portugal, Fund staff estimates based on data provided by the authorities. Tertiary spending figures for Portugal are from OECD (1998).

Public institutions, continental Portugal, based on 1996 data. Tertiary data refer to 1995.

Staff estimates, except for tertiary spending.

Public institutions.

Public and government-dependent private institutions.

25. Another angle for assessing educational outlays is expenditure per student as a fraction of GDP per capita (Table 6). From this vantage point Portuguese outlays per student are above the OECD average, except for secondary education. While the margin is not exceptionally large (15 percent) at the primary level, it is nonetheless surprising in light of the low level of expenditure for nonwage inputs.15 As in other OECD countries, tertiary spending per student in Portugal is more than twice as high as per student outlays at the primary level.

Table 6.Expenditure Per Student as a Share of GDP per Capita, 1995
Early

childhood
PrimarySecondaryTertiaryAll levels of

education

combined
AllNon-

University
university-

level
Portugal1/2/232649
Other OECD countries or regions (mean)18202746365026
Australia1625543959
Austria 3/24273539623733
Belgium (Flemish Community) 4/1116272922
Canada265255505832
Czech Republic20192766247428
Denmark2327293828
Finland33242841394130
France1617313325
Germany 3/211643334429
Greece 4/171622142616
Hungary 3/20222370n.a.7026
Iceland
Ireland1212204219
Italy 3/17242726342526
Japan11192040294323
Korea12171942324623
Netherlands15162245n.a.4522
New Zealand13162452594924
Norway 3/4228
Spain18182435283524
Sweden1828307032
Switzerland 3/10243063337429
United Kingdom 4/2819244024
United States202661307530
Sources: Modified version of Table B4.3 from OECD (1998); for primary and secondary education in Portugal, Fund staff estimates based on data provided by the authorities. Tertiary spending figures for Portugal are from OECD (1998).

Public institutions, continental Portugal, based on 1996 data. Tertiary data refer to 1995.

Fund staff estimates, except for tertiary spending.

Public institutions.

Public and government-dependent private institutions.

Sources: Modified version of Table B4.3 from OECD (1998); for primary and secondary education in Portugal, Fund staff estimates based on data provided by the authorities. Tertiary spending figures for Portugal are from OECD (1998).

Public institutions, continental Portugal, based on 1996 data. Tertiary data refer to 1995.

Fund staff estimates, except for tertiary spending.

Public institutions.

Public and government-dependent private institutions.

26. What are the reasons behind Portugal’s above-average outlays per student at the primary level? One reason appears to be the low student/teacher ratio (Table 7). Portugal’s student/teacher ratio is well below the mean of its OECD counterparts at both the primary and secondary levels. This is in part attributable to the low student/teacher ratios in rural areas and the fact that the primary school network—which was designed before World War II—has not yet been reconfigured, resulting in an extraordinarily high number of schools with just a smattering of students. In 1997/98, there were over 5,300 schools with less than 30 students, and over 600 schools with fewer than five students. There are over 8,000 primary school covering the first four grades, with an average student body of just 50. These numbers can be put in perspective by noting that the total number of public educational institutions (excluding the tertiary level) in Portugal is only 10,700, and that the average secondary school has more than a thousand students.

Table 7.Ratio of Students to Teaching Staff by Level of Education, 1996(Calculations based on full-time equivalents)
EarlyLowerUpperAllNon-
childhoodPrimarysecondarysecondarysecondaryuniversityUniversity-All tertiary
educationeducationeducationeducationeducationtertiaryleveleducation
Portugal1/2/12.611.29.518.5
Other OECD Countries or regions (mean)17.317.914.613.714.513.117.116.2
Australia18.115.4
Austria18.912.79.28.58.914.5
Canada21.517.020.019.519.712.816.414.6
Czech Republic11.920.413.011.712.39.011.711.2
Denmark13.111.210.112.111.0
Finland11.916.812.4
France24.619.513.317.217.1
Germany23.720.916.013.115.012.312.512.5
Greece14.915.011.411.311.323.023.923.6
Hungary11.712.29.511.310.4n.a.9.99.9
Iceland 2/4.517.6
Ireland24.122.615.812.221.616.7
Italy13.911.210.89.810.27.629.025.7
Japan17.819.716.215.615.910.813.512.4
Korea24.931.225.523.124.3
Netherlands20.020.018.6n.a.18.718.7
New Zealand6.022.018.114.116.111.616.114.9
Spain19.418.017.814.215.112.317.617.4
Sweden20.212.712.215.213.7
Switzerland 2/18.315.913.010.212.321.2
United Kingdom19.121.316.015.315.616.7
United States21.916.917.514.716.119.414.115.4
Sources: Modified version of Table 7.1 from OECD (1998); for primary and secondary education in Portugal, data provided by the Ministry of Education.

Preliminary data for 1997/98 for public schools in continental Portugal only. Primary school estimates based on an unweighted average of the first and second cycles of primary education. University data refer to 1996.

Public institutions only.

Sources: Modified version of Table 7.1 from OECD (1998); for primary and secondary education in Portugal, data provided by the Ministry of Education.

Preliminary data for 1997/98 for public schools in continental Portugal only. Primary school estimates based on an unweighted average of the first and second cycles of primary education. University data refer to 1996.

Public institutions only.

27. The central government’s preponderant role in financing education expenditures has contributed to the resistance of local governments to the closure of these small schools. Given that all costs are borne by the central government in maintaining these schools and paying wages, high costs per student have not been born by local governments or their citizens, and hence the incentive to consolidate these schools has been muted. Increased private sector costs associated with school consolidation (travel time to and from school), as well as the loss of prestige for the community when a school closes, also explain the dearth of local support for school consolidation. In spite of these obstacles, however, the government’s reform efforts include the planned consolidation of these small schools, with a view to improving the quality of education for the affected students (see Section G).

28. Relatively high teacher salaries also contribute to Portugal’s high spending per student. At the primary level, beginning salaries, as a multiple of per-capita income, are about 20 percent above the average for other OECD countries (Table 8). This discrepancy increases dramatically for more experienced teachers: for those with 15 years of experience, Portuguese primary school teachers earn a third more. A similar situation prevails at the secondary level, with differentials for experienced teachers at the lower secondary and uppersecondary level (general programs) of 27 and 17 percent, respectively. Higher salaries in Portugal do not coincide with longer teaching hours relative to other countries. At the primary level, teaching hours are about the same, but for lower secondary, general secondary, and vocational secondary schools, the number of hours taught are between 8 and 13 percent lower than the rest of the OECD. Teaching hours in public schools declined between 1990 and 1996, falling by some 10–12 percent at the secondary level, compared with a 1–2 percent decline in other OECD countries. 1617

Table 8.Annual Statutory Teachers’ Salaries in Public Institutions at the Primary Level of Education, in Equivalent U.S. Dollars Converted Using PPPs (1996)
Starting

salary

/minimum

training
Salary after

15 years’

experience

/minimum

training
Salary at top

of scale

/minimum

training
Ratio of

starting

salary to per

capita GDP
Ratio of

salary after

15 years’

experience to

per capita

GDP
Ratio of

salary after

15 years’

experience to

starting

salary
Salary after

15 years’

experience

per teaching

hour
Salary after

15 years’

experience

per student

enrolled
Portugal16,28324,50142,3031.21.91.531
Other OECD countries and regions19,43226,56332,1341.01.41.4311,511
Australia (New South Wales)19,16634,8970.91.71.81,931
Austria19,50825,00539,3230.91.21.3371,970
Belgium19,92427,05532,1940.91.21.431
Czech Republic6,3918,2799,9100.60.81.31.3405
Denmark23,26928,38829,0861.01.31.238
Finland17,66423,38424,0570.91.21.3
France19,47426,29836,4090.91.31.4291,346
Germany28,38435,88538,7031.33.71.3461,720
Greece13,94117,15620,6991.11.31.2221,147
Hungary3,5334,7896,1840.50.71.49394
Ireland22,68135,06141,4951.21.81.5381,550
Italy17,72521,39225,9410.91.11.2291,913
Korea23,67542,31167,3531.73.11.81,357
Netherlands23,32128,42434,9471.11.41.229
New Zealand15,26722,82122,8210.91.31.5281,039
Norway17,32821,12721,4160.70.91.230
Spain24,54428,78336,8501.61.91.2321,599
Sweden16,24620,8150.81.11.3331,635
Switzerland32,50843,46750,0481.31.71.3502,733
United Kingdom19,43429,94829,9481.01.61.538
United States24,09032,53340,3980.91.21.4341,924
Source: Modified version of Table E1.1a from OECD (1998).
Source: Modified version of Table E1.1a from OECD (1998).

29. Measured in terms of purchasing-parity adjusted dollars, teacher salaries in Portugal are below the OECD average. Nevertheless, it should be noted that salaries for experienced teachers exceed those in a number of countries with higher per-capita income (Italy, Spain, Norway, and Sweden). Furthermore, salary per teaching hour for experienced teachers is at the OECD average at the primary level and just 5–10 percent below at the secondary level. Pensions for teachers (in terms of replacement rates) are also relatively generous in Portugal (European Commission, 1996).

30. To some extent Portugal’s high salaries relative to per capita income are not surprising, as cross-country comparisons reveal that this measure of teacher compensation tends to fall as income levels rise (OECD, 1996b). However, Portuguese compensation levels are still some 22 percent above the level predicted from a simple ordinary least squares regression of compensation levels as a function of GDP per capita.18 Furthermore, Portugal’s high salaries in relation to per-capita income are associated with low student/teacher ratios; comparing across countries, more generous compensation is usually accompanied by a higher teaching burden. For example, in a sample of countries excluding Portugal, salaries for experienced teachers at the primary level in relation to GDP per capita are positively correlated with student/teacher ratios (r = .75), and also at the lower secondary level (correlation coefficient of r equal to .76). They are also positively correlated at the upper secondary level (r = .61), although this result is heavily influenced by Korea’s high student/teacher ratio; excluding Korea, the relationship becomes insignificant (r = .17).

31. Portugal’s high spending per student relative to GDP per person appears to be linked to the high level of outlays for staff compensation, rather than large expenditures for capital and other current inputs. Current outlays absorb, on average, a higher share of the budget than in other OECD countries (Table 9).1920 Measured in terms of purchasing-parity adjusted dollars, outlays for nonwage spending are small, and are lower than in every country but Greece. The low level of these expenditures over the years may have contributed to deficiencies in physical infrastructure. Maintenance spending, for example, declined in real terms from 1986 through the first half of the 1990s, and in 1995 these outlays equaled just 8 percent of nonwage current spending (Pinto, Barros, and Lopes, 1998). The modest state of the physical infrastructure and facilities in the mid-1990s is also illustrated by the dearth of computers available to students; in 1995, there were 137 students per computer at the eighth grade level, for example, compared with an average of 34 in other OECD countries.21 Studies on the state of school infrastructure and other problem areas in Portugal relative to other OECD countries, conducted in the early 1990s, are also suggestive of infrastructure deficiencies.2223 Furthermore, a survey of teachers in 1995/96 at the primary level indicated that only 23 percent of teachers in grades one through four thought there was a sufficient level of teaching materials (material didáctico), while only about half felt the level of these inputs was sufficient for grades five through nine (Alaiz and others, 1997). Furthermore, less than half of teachers surveyed indicated sufficient physical space at their schools. More recent data on the composition of expenditures indicate no increase in the share of nonwage spending in current outlays for primary and secondary schools between 1996 and 1998.24 In sum, there is evidence to suggest that the wage bill may have squeezed other inputs (such as maintenance) that may potentially have high productivity.

Table 9.Educational Expenditure on Primary and Secondary Education by Resource Category for Public and Private Institutions (1995)
Percentage of total expenditurePercentage of

current expenditure
Average compensation per student

(In equivalent U.S. dollars)
CurrentCapitalCompensation

of all Staff
Other Current

Expenditure
All StaffOther Current

Expenditure
CurrentCapitalTotal
Portugal 1/9289462,5011632,6632282,891
Other OECD countries and regions91981203,1608583,9163224,239
Australia92879212,8497413,5893333,922
Austria9197624
Belgium (Flemish Community) 2/86143,9886734,661104,671
Canada96481194,2771,0125,2891965,485
Czech Republic 3/871360401,3308772,2073272,534
Denmark95580204,5661,1685,7332956,028
Finland 2/93772283,0851,2284,3133104,623
France91979213,6179754,5924495,041
Germany 3/92876243,2621,0574,3193714,690
Greece 3/86149731,658571,7152801,995
Hungary 3/93775251,0963741,4701021,572
Iceland88127129
Ireland 3/96489112,3912882,6791232,802
Italy 3/96489114,3805324,9121875,099
Japan851587133,1824793,6616214,282
Korea 3/80201001,8101,8104392,248
Luxembourg928973
Netherlands96478222,8697923,6611533,814
Norway 3/881282184,2209005,1206905,810
Spain95584162,5024862,9881603,148
Sweden56443,0352,3945,4307,824
Switzerland 3/891186145,1748586,0327716,803
United Kingdom 2/95570302,5221,0923,6141963,809
United States 3/91980204,5541,1685,7225596,281
Sources: Modified version of Table B5.la from OECD (1998); for Portugal, Fund staff calculations based on data provided by the authorities.

Based on 1996 expenditure data deflated to 1995 prices and converted to dollars using ppp exchange rates for 1995. Figures on the composition of spending derived on the basis of expenditures of the Ministry of Education for public institutions. Total spending per student calculated for continental Portugal only. Spending figures include local government outlays of 0.1 percent of GDP, but exclude outlays for unallocated spending of the Ministry of Education, which totaled 0.1 percent of GDP for all levels of education.

Public and government-dependent private institutions.

Public institutions.

Sources: Modified version of Table B5.la from OECD (1998); for Portugal, Fund staff calculations based on data provided by the authorities.

Based on 1996 expenditure data deflated to 1995 prices and converted to dollars using ppp exchange rates for 1995. Figures on the composition of spending derived on the basis of expenditures of the Ministry of Education for public institutions. Total spending per student calculated for continental Portugal only. Spending figures include local government outlays of 0.1 percent of GDP, but exclude outlays for unallocated spending of the Ministry of Education, which totaled 0.1 percent of GDP for all levels of education.

Public and government-dependent private institutions.

Public institutions.

D. Recent Improvements in Educational Performance: Convergence to OECD Averages?

Enrollment, attainment, and graduation rates

Primary and secondary education

32. Educational attainment rates for past generations provide a misleadingly dour view of educational performance for the current cohort of Portuguese students, given the large gains in enrollment and graduation rates realized in recent years. Figures on gross enrollment rates (Table 10) indicate a groundswell in the percentage of the population attending school, with rates rising at the third cycle of basic education and secondary education from 74 percent at the beginning of the decade to almost 112 percent by 1995/96. Due to high failure rates and dropout rates throughout the secondary school years (Tables 11 and 12), however, the percentage of the population that graduates from secondary school on time (at 18 years of age) is considerably lower. Failure rates are especially acute in the twelfth grade, with only 56 percent of students passing the grade in 1995/96. The high failure rate has led to a high repetition rate (Table 13), with a large share of the student body well beyond normal school age.25

Table 10.Portugal: Gross Enrollment Rates, 1989/90–1995/96
YearGradesGradesTertiary
1–67–12
1989/90126.574.219.5
1990/91124.480.622.8
1991/92128.191.627.0
1992/93126.295.029.9
1993/94128.0102.332.9
1994/95129.5109.235.4
1995/96127.9111.737.9
Source: Ministry of Education.
Source: Ministry of Education.
Table 11.Percentage of Students Successfully Completing the School Year in Continental Portugal, 1994/95 and 1995/96 1/
1994/951995/96
Basic education87.286.5
Grade 1100.0100.0
Grade 284.884.1
Grade 391.792.1
Grade 485.086.0
Grade 587.785.3
Grade 688.488.5
Grade 780.679.0
Grade 883.482.0
Grade 986.485.0
Secondary education71.363.6
General studies70.264.9
Year 1071.063.7
Year 1190.682.3
Year 1260.456.0
Vocational studies75.759.2
Year 1074.251.4
Year 1189.475.6
Year 1256.256.7
Source: Data provided by the Ministry of Education.

Defined as the percentage of students entering the school year that successfully pass or conclude that grade.

Source: Data provided by the Ministry of Education.

Defined as the percentage of students entering the school year that successfully pass or conclude that grade.

Table 12.Dropout Rates by Grade in Continental Portugal 1/
1981 2/1990 2/1995
Basic education
Grade 100
Grade 212
Grade 300
Grade 4840
Grade 51172
Grade 624163
Grade 71496
Grade 81185
Grade 9979
Secondary education
Year 103817
Year 1114
Year 1217
Sources: OECD (1995a), for 1981 and 1990; and Ministry of Education for 1995.

The dropout rate is measured as the percentage of students who are enrolled in a given school year that are not enrolled in the following year.

Covers public schools only.

Sources: OECD (1995a), for 1981 and 1990; and Ministry of Education for 1995.

The dropout rate is measured as the percentage of students who are enrolled in a given school year that are not enrolled in the following year.

Covers public schools only.

Table 13.Repetition Rates by Grade in Continental Portugal 1/
1981 2/1990 2/1994/951995/96
Basic education810
Grade 10000
Grade 241311414
Grade 30068
Grade 427191116
Grade 5201358
Grade 6171078
Grade 72518712
Grade 82418910
Grade 92415710
Secondary education1519
General studies 3/
Year 1011231014
Year 11312629
Year 12 4/15332428
Vocational studies
Year 101619
Year 11314
Year 1233
Sources: OECD (1995a) for 1981 and 1990; and Ministry of Education for other years.

The repetition rate is measured as the percentage of students who are enrolled in the same grade over two consecutive school years.

Covers public schools only.

Includes all secondary students for 1981 and 1990.

Refers to rate for all 12th grade students, including those in vocational courses.

Sources: OECD (1995a) for 1981 and 1990; and Ministry of Education for other years.

The repetition rate is measured as the percentage of students who are enrolled in the same grade over two consecutive school years.

Covers public schools only.

Includes all secondary students for 1981 and 1990.

Refers to rate for all 12th grade students, including those in vocational courses.

33. The extent to which Portugal lags behind the rest of the OECD in completing secondary education at a normal age is suggested by the data on net enrollment rates (gross enrollments corrected for overaged students) by age (see OECD (1998), Table C1.3, for data on countries besides Portugal). According to data from the Portuguese Ministry of Education, only 31 percent of 17 year-olds in continental Portugal were enrolled in twelfth grade or tertiary studies for the beginning of the academic year 1995/96, compared with an average of 85 percent at grade level in other OECD countries. The 31 percent figure overstates to some degree the eventual percentage that successfully completes high school, given the high failure rate in that grade. In fact, only 15 percent of all 18 year olds graduated from secondary school in that year.

34. Figures on net enrollment rates in Portugal require a fairly nuanced interpretation, as low net enrollment does not necessarily translate into failure to eventually complete secondary education. Through the repetition of courses and extension of education beyond the normal school age, approximately 56 percent of students eventually complete secondary education.26 This figure is nevertheless well below the 87 percent of the population in other OECD countries that finishes high school by a normal graduation age (Table 14).

Table 14.Ratio of Upper Secondary Graduates to Population at Typical Age of Graduation (times 100), First Educational Programs, 1996
TotalGeneralVocational and

apprenticeship
M + WMenWomenM + WMenWomenM + WMenWomen
Portugal 1/564865453853111012
Other OECD countries868388403746484944
Austria868884151318717666
Belgium (Flemish Community) 2/100100100343039827490
Canada737077
Czech Republic83818511914717270
Denmark817687463855353832
Finland 2/9893100484057505347
France858586342940515646
Germany868686252229616458
Greece807586544663262923
Hungary8625183359
Ireland797583777282222
Italy797682191622595959
Japan 2/9996100736976262726
Korea919191545750373341
Netherlands813348
New Zealand938699635967302733
Norway 2/100100100494356689045
Poland942569
Spain73658144272529
Sweden818082272134545948
Switzerland818676201823616853
United States726976
Sources: OECD (1998), Table C2.3 for all countries except Portugal. Portuguese data are from the Ministry of Education.

Refers to the number of persons graduating in 1996 divided by the number of 18 year olds in continental Portugal. Thus, it includes overage students, whereas the figures for other countries do not.

Total does not equal sum of general and vocational studies, due to problems of double counting. In these cases, the total has been set to a maximum of 100.

Sources: OECD (1998), Table C2.3 for all countries except Portugal. Portuguese data are from the Ministry of Education.

Refers to the number of persons graduating in 1996 divided by the number of 18 year olds in continental Portugal. Thus, it includes overage students, whereas the figures for other countries do not.

Total does not equal sum of general and vocational studies, due to problems of double counting. In these cases, the total has been set to a maximum of 100.

35. Portugal’s high repetition rates continue to signal inefficiencies in the educational system, given the high public and private costs of repeating the same course of education.27 From the standpoint of the public finances, the costs of the high repetition rate can be assessed by quantifying the savings that would occur if the gross enrollment rates in primary and secondary education fell, for example, by 10 percentage points. Under this scenario, outlays would fall by 0.5 percentage point of GDP.28

36. Partly as a consequence of Portugal’s high repetition rate, a significant share of the student population consists of overaged students (including adults) in special repeater courses (ensino recorrente and ensino nocturno) in attempts to complete either primary or secondary education. In 1995/96, these students accounted for some 4 percent of the total student population at these levels in Portugal.29 These intensive courses follow the traditional curriculum in Portugal, which is designed to prepare students for university studies. The success rate in these courses has been very low, however. For ensino recorrente—which covers the bulk of these students—the percentage of students in continental Portugal successfully completing the coursework necessary for completion of fourth grade was only 9 percent, and for completion of sixth grade, 28 percent (Ministry of Education, 1998a). A recent study of success rates for ensino recorrente for completion of ninth grade and secondary education also pointed to success rates below 10 percent (Contreras, 1999). The experience of schools using alternative curricula however, has been more favorable (see Contreras, 1999).

37. Educational performance shows some variation across regions in Portugal (Tables 1517). Net enrollment rates at the secondary level vary with per-capita income levels, and are about ten percentage points above the continental average in the Lisboa and Vale do Tejo region, while they are below average in the north and central regions. The passing rates of enrolled students are very similar, with the exception of Algarve. The percentage of above-age students (a reflection of repetition rates in earlier years) varies somewhat more dramatically, but generally reinforces the view that the performance of Portugal’s poorer regions are not the cause of its relative standing vis-à-vis other OECD countries.

Table 15.Net Enrollment Rates in Continental Portugal, 1994/95
Basic educationSecondary education
Grades 1–4Grades 5–6Grades 7–9(average)
Continent109.784.978.551.2
North108.782.274.442.5
Central110.282.376.349.7
Lisbon and Tejo Valley109.989.584.761.5
Alentejo106.983.576.953.7
Algarve121.891.284.759.1
Table 16.Conclusion Rates in Continental Portugal, 1995/96
4th Grade6th Grade9th Grade12th Grade
GeneralVocationalVia
studiesstudiesensino
Continent86.088.585.064.556.748.8
North84.789.586.165.358.752.4
Central86.989.584.461.850.946.3
Lisbon and Tejo Valley87.186.984.265.657.447.6
Alentejo87.586.885.664.865.753.2
Algarve84.787.783.858.749.944.2
Table 17.Portugal: Students Above the Normal Age in Grades 7–9, by Region, 1994/95(Percentage of students by number of years behind grade)
0 Year1 Year2 Years3 or More
North57.324.012.26.4
Central50.824.814.59.9
Lisbon and Tejo Valley57.022.013.08.0
Alentejo53.122.914.49.6
Algarve49.622.615.112.7

38. Differences in expenditure levels—as proxied by student/teacher ratios—do not appear to be correlated with differences in conclusion rates by region (Table 18). Interestingly, student/teacher ratios are the highest in continental Portugal in its wealthiest region (Lisboa and Vale do Tejo) for the first cycle of primary education.30 This may reflect the absence of the extremely small schools that account for the bulk of these institutions in the less populous regions of Portugal. At the secondary level, student/teacher ratios show little regional variation, and in all regions there are fewer students per teacher than the prevailing average for other OECD countries.

Table 18.Portugal: Student/Teacher Ratios by Region, 1997/98 1/
Basic educationSecondary schoolPer-capita GDP
Grades 1–4Grades 5–6Grades 7–9Grades 10–12in US$, 1996
Portugal14.510,865
Continental Portugal14.58.811.29.5
North13.99.811.89.69,573
Central13.28.19.99.19,161
Lisbon and Tejo Valley16.58.311.49.413,823
Alentejo13.58.311.39.98,918
Algarve16.48.711.79.910,846
Azores16.97,733
Madeira13.68,066
Sources: Ministry of Education; and National Statistics Office (INE).

Preliminary estimates.

Sources: Ministry of Education; and National Statistics Office (INE).

Preliminary estimates.

39. Striking differences in performance are observed between public and private schools in Portugal. Table 19 indicates that conclusion rates are systematically higher in private schools across all grades. At grades 4, 6, and 9, private schools’ success rates average over 94 percent, some 8 percentage points higher than in public schools. At the secondary level, performance differences are narrower for general courses and negligible for vocational studies, but very wide (over 20 percent) for college preparation students studying via ensino.

Table 19.Conclusion Rates in Continental Portugal for Private and Public Schools, 1995/96
4th Grade6th Grade9th Grade12th Grade
GeneralVocationalVia
studiesstudiesensino
Total86.088.585.064.556.748.8
Men83.785.583.361.351.445.8
Women88.691.886.866.661.251.7
Public85.288.084.164.056.646.6
Men82.985.082.161.051.543.4
Women87.991.486.066.060.949.8
Private95.493.193.869.557.466.9
Men93.390.591.764.649.565.1
Women97.795.996.073.265.268.8
Source: Translation of Table 2.1, Ministry of Education (1998d).
Source: Translation of Table 2.1, Ministry of Education (1998d).

40. These higher conclusion rates enjoyed by private schools are associated with lower expenditures, as proxied by student/teacher ratios. For the first cycle of primary education (grades one through four), student/teacher ratios in public schools averaged about 12.8 in 1995/96, compared with 18.3 in private schools. For other levels of primary education and secondary education combined, student/teacher ratios in that year for public and private schools averaged 11.2 and 14.7, respectively.31 These figures suggest that the private schools may be more efficient than their public sector counterparts.

41. In sum, Portugal continues to lag behind the rest of the OECD in educational attainment at the primary and secondary level. Nevertheless, impressive strides have been made in recent years to improve performance, especially at the primary level, where dropout rates and repetition have fallen sharply in response to the government’s reform efforts (see Section G below). One especially telling sign is the rapid rise in the percentage of students finishing the obligatory nine years of schooling, which rose from 63 percent in 1991/92 to 89 percent in 1994/95 in continental Portugal. Progress has been more uneven at the secondary level, however; while repetition rates fell between 1990 and 1995, they rebounded in 1996, and dropout rates at grade 10 show no systematic trend toward improvement. The increased success of students at lower levels is reflected in the Ministry of Education’s projections regarding school completion (Table 20); by 2000/01, it is expected that 100 percent of all students will finish ninth grade, and 66 percent twelfth grade. Looking further ahead, by 2005 it is projected that 54 percent of the population aged 20–34 in continental Portugal will have completed secondary education (Ministry of Education, 1998b).

Table 20.Actual and Projected Educational Attainment Rates, Continental Portugal, 1990/91–2000/01 1/
Year6th Grade9th Grade11th Grade12th GradeTertiary
1990/91825847
1995/969885645635
2000/01100100726640
Source: Ministry of Education (1998c).

Figures refer to the number of actual or projected graduates divided by the number of students beginning school n years earlier. For example, the 66 percent figure for 2000/01 implies that the number of new high school graduates in that year will be 66 percent of the student population that started school 12 years earlier.

Source: Ministry of Education (1998c).

Figures refer to the number of actual or projected graduates divided by the number of students beginning school n years earlier. For example, the 66 percent figure for 2000/01 implies that the number of new high school graduates in that year will be 66 percent of the student population that started school 12 years earlier.

Tertiary education

42. In contrast to the rather slow movement toward the OECD norm on secondary educational attainment, in recent years the share of the near school-age population attending and completing tertiary education appears to have converged to OECD averages. The number of university students nearly doubled between 1990/91 and 1997/98, despite a decline in the population aged 18–22 (Table 21). Total enrollment as a share of the population aged 18–22 rose from 23 percent at the turn of the decade to about 43 percent in 1997/98, and the share of 22 year olds with a tertiary degree soared from under 12 percent in 1990/91 to 25 percent in 1996/97.32 Almost half of all tertiary students are older than 22, and hence net enrollment rates for 18–22 year olds lag behind OECD averages by a substantial margin; however, when measured in terms of a wider age group (17–34 year olds), net enrollment in Portugal is just one year behind the average for other OECD countries. In light of the recent strong gains in enrollment, it appears safe to infer that this gap has now been closed, and that participation rates in higher education are now comparable to the rest of the OECD.

Table 21.Portugal: Net Enrollment Rate and Share of the Population Completing Tertiary Education, 1990/91–1997/98
1990/911991/921992/931993/941994/951995/961996/971997/98
Number of students (thousands)186.8218.3246.1270.0290.4313.5334.1344.9
Number of new graduates (thousands)18.721.427.532.736.439.242.6
Total Portuguese
population ages 18–22 (thousands)816.2819.2826.5841.3848.5842.3827.8807.2
Total Portuguese population
age 22 (thousands)161.4161.8159.4163.2167.7166.5170.5173.7
Enrollment as share of population 19–2222.926.729.832.134.237.240.442.7
Net enrollment rate, ages 17–3410.5
Tertiary graduates as share of
population age 2211.613.317.220.021.723.525.0
Memorandum item:
Net enrollment rate, ages 17–34, other OECD countries11.5
Source: Ministry of Education, except for net enrollment rate, which is drawn from OECD (1998), Table C3.3.
Source: Ministry of Education, except for net enrollment rate, which is drawn from OECD (1998), Table C3.3.

43. Despite the rapid ascent of enrollment rates, higher education is still plagued by a number of problems, including high repetition rates. For example, of the 51 percent of students that advanced onward from the conclusion of secondary education into tertiary studies in 1996, repetition rates in the first year ranged from 19 to 24 percent (Ministry of Education, 1999). Dropout rates are also large; for 1993, the OECD estimates a figure of 50 percent, compared with the average in other OECD countries of 31 percent (OECD, 1998, Table C4.1). Another problem identified by the Ministry of Education is the gap between chosen fields of study and job openings; in particular, there appears to be an excessive number of students with humanities degrees (Ministry of Education, 1999). OECD data for 1996 (OECD, 1998, Table C4.4) also support the view that the percentage of students specializing in the humanities is out of step with the rest of the OECD: in that year 48 percent of Portuguese students specialized in the humanities or general studies, 11 percentage points higher than the average for other OECD countries. Correspondingly, the share of Portuguese students in the natural sciences was some five percentage points below average.33

Performance on international examinations

44. One drawback to the use of educational attainment rates as a measure of performance is that they say little about what students learn, given differences in curricula and standards across countries. Thus, it is useful to supplement these measures of performance with other yardsticks that assess basic abilities in mathematics, science, and reading.

45. Table 22 provides data on the performance of Portugal relative to other selected countries on international examinations conducted during the 1990s for eight-grade students. Portuguese scores on reading are slightly above the average of other countries. Math and science scores, however, are substantially below par. The math score was some 14 percent below the average for other OECD countries, and the lowest among the group of OECD countries covered in the 1995 examinations. Performance on science was marginally better, but still some 10 percent below average. Similar lags in mathematics performance are evidenced in the results of the fourth-grade examinations, where scores were some 15 percent below the mean.

Table 22.Performance on International Examinations
8th Grade8th Grade8th Grade4th GradePPP per capita
Mathematics 1/Science 1/Reading 2/Mathematics 1/(US$)
(1995)(1995)(1991)(1995)(1995)
Portugal45448050034011,968
Other OECD countries or regions (mean)52753149840217,956
Australia 3/53054540819,943
Austria 3/53955842120,211
Belgium (Flemish Community)565550
Belgium (French Community)526471446
Canada 4/52753149439521,733
Czech Republic5645744289,145
Denmark 3/50247820,659
Finland54517,776
France53849820,896
Germany 3/5/50953150019,394
Greece 3/4844974823568,950
Hungary 3/5375544106,341
Iceland48749451433818,729
Ireland52753848441216,061
Italy48819,808
Japan60557145721,461
Korea60756547111,829
Netherlands 3/54156048643819,621
New Zealand50852552836216,974
Norway50352748936521,596
Spain48751745614,496
Sweden51953552919,313
Switzerland54552251523,612
UK (England)506552376
UK (Scotland) 3/499517383
United States50053451440726,479
Sources: OECD (1998); TIMSS International Study Center; and OECD (1997).

Third International Mathematics and Science Study (TIMSS).

International Association for the Evaluation of Educational Achievement (IEA) Reading Literacy Study for 13–14 year olds.

Countries that did not satisfy one or more guidelines for sample participation rates, age/grade specifications, specifications, or classroom sampling procedures under the TIMSS.

Reading score refers to British Columbia only.

Reading score represents an average of the former Democratic Republic of Germany and the former Federal Republic of Germany. The scores of the two regions varied by three points.

Sources: OECD (1998); TIMSS International Study Center; and OECD (1997).

Third International Mathematics and Science Study (TIMSS).

International Association for the Evaluation of Educational Achievement (IEA) Reading Literacy Study for 13–14 year olds.

Countries that did not satisfy one or more guidelines for sample participation rates, age/grade specifications, specifications, or classroom sampling procedures under the TIMSS.

Reading score refers to British Columbia only.

Reading score represents an average of the former Democratic Republic of Germany and the former Federal Republic of Germany. The scores of the two regions varied by three points.

46. To only a small extent does the discrepancy in performance reflect the impact of Japan and Korea in boosting up the overall average; removing these countries from the sample reduces the average math score by only 1 percent (from 527 to 520). It is also interesting to note that test scores do not appear to be a simple function of per-capita income. One factor that does appear to be correlated with test scores is the repetition rate: countries with a high share of above-age students score lower on eight-grade math examinations (OECD, 1997). Thus, it appears that countries that force students to repeat grades do not enjoy improved academic performance. This result has potentially important implications for Portugal, given its high repetition rate relative to the rest of the OECD.

E. Education and Labor Market Outcomes in Portugal

47. The importance of improving educational performance in Portugal is underscored by the high rates of return for completing secondary and tertiary education. The high rate of return for completing secondary education can be observed by examining the penalty for not doing so, as measured by the earnings of those without their secondary school diplomas relative to those finishing high school. Table 23 reveals that the earnings of nongraduates are only 64 percent of those enjoyed by their counterparts finishing school, the lowest share amongst the OECD. More formal studies of rates of return to one additional year of education, controlling for worker experience and age, also indicate high returns relative to many OECD countries (OECD, 1995a). The high disparity in earnings reflects Portugal’s relatively flexible labor markets and the small share of the population completing high school, as indicated earlier in Table 3. In a similar vein, the premium on university education is also the highest in the OECD; in light of these high rates of return to schooling, some have called into question the low degree of cost recovery in public education at the tertiary level (Machado and Mata, 1998).

Table 23.Relative Earnings of 25–64 Year-olds with Income from Employment (Upper secondary education= 100) by level of educational attainment and gender (1996)
Below upper secondaryUniversity-level
Yeareducationeducation
M+WMenWomenM+WMenWomen
Portugal1996646264184182175
Other OECD countries808178162161156
Australia19958910587142161139
Canada1996878776161152172
Czech Republic1996677275161155149
Denmark1996848687134138132
Finland1995939193185187173
France1996828579178185167
Germany1996768282158152151
Hungary1996727968169189150
Ireland1994857762183171187
Italy1995767376156173129
Netherlands1995868777137135143
New Zealand1996827885176171148
Norway1996858781142143146
Spain1995786276153145147
Sweden1996908889153158144
Switzerland1996718075161146161
United Kingdom1996747969181161190
United States1996676464183183175
Source: OECD (1998), Table F7.1.
Source: OECD (1998), Table F7.1.

48. Recent studies indicate an increase in the returns to education since the 1980s (Machado and Mata, 1998), despite the fact that levels of education have risen dramatically. This is consistent with the observed increase in the inequality of earnings (OECD, 1995a, 1996a). Figures on income differentials by education level for more recent generations (those aged 30–44) also support this view, as they indicate an even wider earnings gap (41 percent) for those not finishing secondary school than for the adult population as a whole. In sum, it appears that the increase in demand for skilled labor has outstripped the increase in supply, which may be a harbinger of increased levels of inequality. In this context, it is not surprising that recent studies indicate that aggregate income inequality in Portugal is the highest in the EU.34

F. The Efficiency of Education Expenditure in Portugal: Insights from FDH Analysis

An overview of FDH analysis

49. One method for assessing the efficiency of an educational system is to assess how well it converts spending inputs into educational “outputs.” In this section of the chapter, such an analysis is conducted with the help of a nonparametric technique for production frontier estimation called Free Disposable Hull (FDH) analysis. The analysis involves a comparison of spending levels across different countries of the OECD and different measures of output or educational performance.35

50. FDH analysis, like other production frontier techniques, provides a framework for ranking the efficiency of producers through comparison of their performance with a production frontier reflecting “best practices.”36 The first step in FDH analysis is to establish a production frontier that shows, for each level of input use, the highest level of output that can be observed among the producers in the sample. Once this production frontier is established, a ranking of inefficient producers (those that produce less output than possible with a given level of input use) can be determined. The advantage of FDH analysis relative to other production frontier techniques is its parsimonious approach to the construction of the production frontier. The only assumption is that inputs and outputs can be freely disposed of; this implies that for the same production technology, a continuous production frontier can be established that maps any given input level with the highest possible level of output.

51. FDH analysis is particularly attractive for the task at hand, where the “producer” in question is the government, and the output is educational services. In particular, the fact that FDH makes no assumptions about the shape of the production frontier is appealing, as past research on educational production functions provides no conclusive evidence on the appropriate shape of the frontier.37 Under a parametric approach, in contrast, a functional form must be assumed, and parameters chosen that best fit the data. In light of the uncertain relationship between educational inputs and outputs, a parametric estimate based on a small number of observations (such as the sample of countries examined in this chapter) could be well off the mark. FDH is also preferable to other nonparametric approaches in the present context, such as Data Envelopment Analysis (DEA), which assumes a convex production frontier.38 The parsimonious approach of FDH is not without drawbacks; as a nonparametric approach, the production frontier is more heavily influenced by (and indeed determined by) outliers, and thus is more vulnerable to measurement error than a parametric technique. Furthermore, the absence of any restrictions on the shape of the production frontier (such as convexity) means that a smaller number of observations can be identified as inefficient than with DEA, reducing the ability of FDH to sort out efficient and inefficient producers.

52. The approach of FDH in measuring efficiency can be better understood with the help of Figure 1, which is reproduced from Gupta, Honjo, and Verhoeven (1997). Let us assume the simple case of one input (X) and one output (Y). Suppose there are four firms (A, B, C, and D). The first step of FDH analysis is to construct the production frontier on the basis of the most efficient producers. In the context of Figure 1, those firms are A, C, and D. A firm is efficient (that is, on the production frontier) if there are no other firms associated with the same (or greater) level of output at the same (or lower) level of input use. For example, firm C is efficient, even though firm D produces a higher level of output. Firm A is also efficient; although C and D produce more output than A, they use more input. Firm B, however, is clearly inefficient, as firm A is able to produce more output with an even smaller level of input. By assuming the free disposal of inputs and outputs, a continuous production frontier can be formed by connecting a line through the points representing the most efficient producers. FDH provides a framework for calculating both input efficiency and output efficiency scores that vary from a minimum of 0 for producers on the horizontal axis beyond X(A) to a maximum of 1 for firms on the production frontier. The input efficiency score indicates how much less input could be used by the inefficient firm to produce the same or a greater level of output. In Figure 1, firm B’s input efficiency score is given by X(A)/X(B). The output efficiency score reveals the other side of the coin, assessing how much more output could be produced with the same (or an even lower) level of input use. Firm B’s output efficiency score is given by Y(B)/Y(A). FDH analysis can also be applied in the case of multiple inputs and outputs to compute input and output efficiency scores, although their computation is slightly more involved.39

Figure 1.Free Disposable Hull (FDH) Production Possibility Frontier

Source: Reproduced from Gupta, Sanjeev, Keiko Honjo, and Marijn Verhoeven, 1997, “The Efficiency of Government Expenditure: Experiences from Africa,” Working Paper 97/153 (Washington: International Monetary Fund).

53. This example also underscores the weakness of FDH analysis, which ranks all three observations (A, C, and D) as equally efficient (with scores of 1.0). If a functional form had been assumed, then efficiency could have been addressed by comparing the observed mapping of inputs and outputs with that predicted from the production function. In FDH analysis, however, these observations are all ranked equally. In the case of multiple inputs and outputs the information content of FDH is even more limited, as producers can only be identified as inefficient if they produce less of all of the outputs and use the same or more of at least one of the inputs.

Educational inputs and outputs for FDH analysis

54. The choice of educational inputs and outputs merits discussion. With respect to inputs, one alternative is to measure spending in purchasing-parity adjusted (PPP) dollars. A drawback of this measure is that it will be highly correlated with country income levels. That is, in richer countries, education spending tends to be higher, as teacher salaries tend to move upward with rising income levels. Higher salaries do not necessarily correspond to a higher quality of inputs from one country to another, as teachers across many OECD countries have similar levels of education and training. An alternative input measure that overcomes these difficulties is spending per student as a share of per-capita GDP, or the share of educational expenditure to GDP adjusted for cross-country differences in the school-age population. It should be noted that the FDH analysis could potentially yield very different results, depending on the choice of inputs; in this context, it is worthwhile to note that Portuguese spending per student is quite modest in PPP terms, although it is relatively high when viewed from the standpoint of GDP per capita. To assess the sensitivity of the results to these assumptions, the results are computed using both input measures.

55. Regarding output measures, an important consideration is the desirability of selecting variables that measure the performance of the present system, rather than the accumulated performance of the system over years past. This is an important consideration for the case of Portugal, where educational attainment rates for recent generations are far above those in earlier years. In this light, a good measure of the performance of the education system is the percentage of the population that completes secondary education at a normal graduation age, as indicated in Table 14, or educational attainment rates for a cohort slightly older than the normal age for school completion (Table 3). The latter figure is especially fitting in the case of Portugal, as it allows for the possibility that low net enrollment rates at the secondary level are attributable to relatively high standards for graduation. We also use test scores on international examinations at the eight grade as an output indicator. An especially attractive feature of this measure is that it can be seen as a barometer that assesses the quality of education across countries, rather than a simple quantity measure such as school completion rates.40

Empirical results

56. We start by presenting the simple case of one input and one output in Table 24. The input is measured as total primary and secondary expenditure per student in purchasing-parity adjusted dollars (Table 9). The output measure is the percentage of the population finishing secondary school at the normal graduation age (see Table 14). 41 This measure of output provides a good yardstick for an analysis of efficiency, since countries that take additional time and expenditure to finish secondary school (e.g., due to high repetition rates) are penalized in the resulting efficiency scores. This measure is also to be preferred to enrollment rate figures, since these do not indicate whether students actually complete their intended course of study.

Table 24.Efficiency Scores: Expenditure per Student in Purchasing-Power Adjusted Dollars and Secondary Graduates to Population at Typical Graduation Age
(Input: Spending per student in purchasing-power adjusted U.S. dollars at primary and secondary level; output indicator: ratio of secondary graduates to population at typical graduation age)
Input efficiencyOutput efficiency
ScoreRankScoreRank
Portugal0.54490.62020
Austria0.401120.9456
Canada0.287160.74018
Czech Republic0.62170.9108
Denmark0.261170.81314
Finland0.92650.9925
France0.312140.86412
Germany0.335130.87010
Greece0.78860.9287
Hungary1.00011.0001
Ireland0.56180.86811
Italy0.308150.79817
Japan1.00031.0003
Korea1.00021.0002
Netherlands0.412110.8929
Norway1.00041.0004
Spain0.500100.80216
Sweden0.201200.81313
Switzerland0.231190.80915
United States0.250180.72519
Source: Fund staff calculations.
Source: Fund staff calculations.

57. The results indicate that Portugal is about in the middle of the pack when efficiency is gauged from the input side. Among the 20 countries assessed in Table 24, Portugal is ranked ninth with respect to input efficiency. Nevertheless, its score of .54 denotes that at least the same level of output could be achieved with 54 percent of the present level of input (spending per student), demonstrating ample scope for improving efficiency. Among the most efficient countries are Japan, Korea, Hungary, and Norway, where all four countries are efficient (a score equal to 1.0) and located on the production possibilities frontier.42

58. One notable pattern indicated in Table 24 is that high income countries (e.g., the United States and Sweden) tend to have low efficiency scores. This reflects the observations made in the section above, where it was noted that the higher teacher salaries found in higher-income countries make it difficult to avoid relatively elevated spending per student. For this reason it is also worthwhile to assess output efficiency, which provides less of a penalty for advanced countries that must pay high salaries to their educational staff. In general, the output efficiency scores reveal much less inefficiency (i.e., higher scores) than the input efficiency measures, and present some interesting differences in country rankings. Most notable among these is Portugal, which falls to the bottom of the ranking in terms of efficiency. The output efficiency score of .62 indicates that the secondary school completion rate is only 62 percent of what it could be if spending was efficient. This implies that if the production of educational services was efficient in Portugal, present levels of spending would allow a graduation rate of close to 91 percent, as opposed to 56 percent. This suggests that the gap in educational attainment in Portugal may not be due to inadequate levels of spending, but to how efficiently that spending is translated into educational output.

59. Given the focus here on differences in secondary school completion, a closer examination of the linkage between secondary school spending per se and graduation rates is warranted. Table 25 provides an evaluation of the case of one output (graduation rates, as before) and one input, where the input is defined as secondary school spending per student as a share of GDP per capita. The advantage of this measure is that it quantifies spending per student relative to prevailing income levels in the country, and hence incorporates the fact that higher income countries tend to pay more generous teacher salaries. The results are similar to those in Table 24, with a correlation coefficient of .7 between common observations for input efficiency, and over .9 for the output efficiency scores. Both Japan and Korea are among the most efficient countries; within Europe, expenditure is also relatively efficient in Greece and the Netherlands, reflecting low levels of spending per student. The high efficiency score of Belgium (Flemish Community), on the other hand, mirrors the universal completion of secondary school by normal graduation age. Portugal’s ranking among the least efficient reflects both its high spending per student and its low graduation rate; as noted earlier, this high spending is due to both high salaries in relation to GDP per capita and low student/teacher ratios.43

Table 25.Efficiency Scores: Education Spending Per Student as Share of GDP Per Capita (Secondary Level) and Secondary Graduates to Population at Typical Graduation Age
(Input: Spending per student at the secondary level as a share of GDP per capita; output indicator: ratio of secondary graduates to population at typical graduation age)
Input efficiencyOutput efficiency
ScoreRankScoreRank
Portugal0.613150.57120
Austria0.540190.8609
Belgium (Flemish)1.00031.0003
Canada0.310200.73119
Czech Republic0.681100.83811
Denmark0.640120.81313
Finland0.73890.9805
France0.600170.85410
Greece1.00041.0004
Hungary0.80180.8747
Ireland0.81270.8688
Italy0.583180.78816
Japan1.00011.0001
Korea1.00021.0002
Netherlands0.85150.82212
New Zealand0.83960.9376
Spain0.664110.73917
Sweden0.619140.81314
Switzerland0.612160.80915
United States0.628130.73418
Source: Fund staff calculations.
Source: Fund staff calculations.

60. A drawback of these assessments of efficiency based on spending per student is that they fail to provide any penalty for countries with high repetition rates and the concomitantly large share of the population in school beyond the normal age of graduation. To address this shortcoming, efficiency was also measured in terms of the nexus between graduation rates at a normal age and primary and secondary expenditure to GDP, adjusted for population structure.44 The results in Table 26 reveal that Portugal falls to last in the assessment of efficiency under this yardstick, reflecting the fiscal burden of high repetition rates.

Table 26.Efficiency Scores: Education Spending to GDP and Educational Attainment Levels
(Input: Educational expenditure to GDP at the primary and secondary, level, adjusted for population structure; output indicator: ratio of secondary graduates to population at typical graduation age)
Input efficiencyOutput efficiency
ScoreRankScoreRank
Portugal0.579170.57517
Austria0.721100.8779
Canada0.598160.80314
Czech Republic0.87870.9106
Denmark0.638140.83012
Finland1.00021.0002
France0.73280.87110
Germany0.703110.8778
Greece1.00031.0003
Hungary0.89660.9495
Ireland1.00031.0003
Italy0.651130.86611
Korea1.00011.0001
Netherlands0.90350.8927
Spain0.622150.80215
Sweden0.654120.82913
United States0.72390.76416
Source: Fund staff calculations.
Source: Fund staff calculations.

61. Portugal fares only slightly better when educational efficiency is assessed in terms of the production of both secondary and tertiary graduates. Table 27 presents the results for the one input, two output case, where the input is total education spending to GDP, adjusted for the share of the population under age 29, and the two separate outputs are (i) graduation rates at the normal age for secondary education and (ii) graduation rates at the normal age for tertiary education. Portugal’s input efficiency score of 0.62 suggests that the same number of secondary and tertiary graduates could be produced using 38 percent less spending.45

Table 27.Efficiency Scores: Education Spending to GDP and Ratio of Secondary and Tertiary Graduates to Population at Typical Graduation Age
(Input: Educational expenditure to GDP, adjusted for population structure; output indicators: ratio of graduates to population at typical graduation age, secondary school, and ratio of graduates to population at typical graduation age, tertiary)
Input efficiencyOutput efficiency
ScoreRankIndependently

efficient 1/
ScoreRankIndependently

efficient 1/
Portugal0.623170.57117
Austria0.83090.8709
Canada1.000+1.000+
Czech Republic0.97660.83811
Denmark0.633160.82312
Finland0.758110.9926
France0.826100.86410
Germany0.749120.8708
Greece1.00041.0005
Hungary0.95880.8747
Ireland1.00031.0003
Italy0.676150.79815
Japan1.00011.0001
Korea1.00021.0002
Netherlands0.96470.82214
Spain0.681140.73916
Sweden0.654130.82213
United States1.00041.0004
Source: Fund staff calculations.

Refers to observations that do not dominate any others, and are not dominated by any other observation.

Source: Fund staff calculations.

Refers to observations that do not dominate any others, and are not dominated by any other observation.

62. Portugal’s ranking is also low (but not the lowest of the OECD countries) when educational output is measured in terms of performance on international examinations (Table 28). In Case (A) in Table 28, input efficiency scores are provided for the one-input, two-output case, where the two outputs refer to math and science scores for eight-grade students (see Table 22). Portugal is ranked thirteenth out of the 18 countries assessed, with Korea, Japan, Ireland, and the Czech Republic among the most efficient. Surprisingly, the rankings are very similar to those presented in Table 26, where efficiency was assessed in terms of the linkage between spending and completion of secondary school; the correlation coefficient between the input efficiency measures is .88. This suggests that there is no apparent tradeoff between achieving high rates of school completion and providing quality education (as proxied by high scores in international examinations) in a cost-effective manner.

Table 28.Efficiency Scores: Education Spending to GDP At Primary and Secondary Level and Achievement in Eighth-Grade International Achievement Examinations
(Input: Educational expenditure to GDP at the primary and secondary level, adjusted for student enrollment as a share of the population; output indicators: (A) scores on Eighth-grade math and science exams; and (B) reading examinations)
(A)(B)
Math and ScienceReading
Input efficiencyInput efficiency
ScoreRankScoreRank
Portugal0.648130.7109
Australia0.8928
Austria0.56417
Canada0.632140.69710
Czech Republic1.0004
Denmark0.57316
Finland1.0002
France0.70812
Germany0.598150.66012
Greece0.98350.9834
Hungary0.7809
Iceland0.90271.0001
Ireland1.00031.0003
Italy0.67311
Japan1.0002
Korea1.0001
Netherlands0.93360.9185
Spain0.737100.7378
Sweden0.556180.8936
United States0.711110.7787
Source: Fund staff calculations.
Source: Fund staff calculations.

63. In Case B, where the output is reading scores, Portugal’s efficiency is higher (.710) than when efficiency is assessed in terms of math and science achievement (.648). This is due to Portugal’s above-average scores for reading, compared to its subpar scores on international math and science examinations. Among the most efficient countries in this regard are Finland, Iceland, and Ireland.

64. What country characteristics or factors are generally associated with high efficiency in the provision of education? Some insights into this question can be gained from Table 29, where correlation coefficients between various input measures and input efficiency scores are presented. The results should be interpreted with caution, given the small number of observations available for some variables and the fact that correlations at best are only suggestive of causality, as other factors are not controlled for; furthermore, there may be some nonlinearities in the relations between the relevant variables. Keeping these caveats in mind, it appears that countries with high student/teacher ratios at the primary level are relatively efficient, regardless of whether educational output is measured in terms of graduation rates or scores on math and science examinations. This relationship is more tenuous at the secondary level, as the correlation coefficient for one of the measures is low (r =.09). Teacher salaries show a positive but low correlation with efficiency scores, but this relationship evaporates when Korea is excluded from the sample. Higher nonteacher spending is not associated with efficiency, given the negative and sizeable correlation between nonstaff current outlays in purchasing-parity adjusted dollars and efficiency. There also is no hard rule of thumb regarding the composition of spending, as there is no consistent relationship between the share of spending absorbed by personnel costs and efficiency.

Table 29.Correlations Between Educational Inputs and Efficiency Measures
Input efficiency:Input efficiency:
Secondary graduationMath and Science 2/
Rates 1/
Student/teacher ratio, primary0.490.54
Student/teacher ratio, second. 3/0.090.31
Total spending per student, ppp dollars-0.59-0.69
Nonstaff current outlays, ppp dollars-0.27-0.58
Share of personnel costs
in total spending0.030.24
Teacher salary to GDP per capita 4/0.180.25
Memorandum items:
Correlations with other country
characteristics
GDP per capita-0.46-0.40
Private share of education provision 5/-0.240.24
Repeater rate, secondary-0.61-0.32
Source: Fund staff calculations.

Based on input efficiency score in Table 26.

Based on input efficiency score in Table 28.

Lower secondary school (grades 7–9).

Refers to teachers with 15 years of experience.

Based on direct private sector payments to institutions. Includes tertiary level.

Source: Fund staff calculations.

Based on input efficiency score in Table 26.

Based on input efficiency score in Table 28.

Lower secondary school (grades 7–9).

Refers to teachers with 15 years of experience.

Based on direct private sector payments to institutions. Includes tertiary level.

65. One notable result from Table 29 is the deleterious effect of repetition rates on efficiency, not only in terms of the ability of countries to translate educational spending into graduates at a normal age (an almost tautological result), but also to produce high scores in math and science achievement tests. The latter results confirm the aforementioned findings of the OECD (1997), which indicated that high repetition rates tend to weaken, rather than strengthen student performance.

G. Education Reform in Portugal

66. A number of initiatives have been taken in the 1990s to address the problems of high repetition and dropout rates and increase educational achievement. One of the cornerstones of the government’s strategy to enhance student achievement has been the expansion of the preschool system. Preschool education is widely believed to have a salutary effect on student achievement, and it has been suggested that Portugal’s relatively low participation in preschool education could be contributing to high dropout rates (OECD, 1995a). Aided by a substantial investment program supported by the European Community, participation in preschool education rose from 34 percent in 1988 to 64 percent in 1997/98.46

67. The government has also introduced special programs to reduce school dropout rates, including the Program for Education Success (PIPSE) during 1988–92 and the Program of Education for All (PEPT) during 1991–98. In addition, school attendance of lower-income groups has been promoted through the Guaranteed Minimum Income Program introduced in 1997, which requires school attendance for school-age children of recipient families. More recently, a concerted effort has been made to target poorer regions with high dropout rates through the Territories of Priority Educational Intervention (TEIP) program covering schools at the basic level (grades one through nine). The TEIP attempts to coordinate public services targeted to the poor by grouping together schools in poorer regions with a common set of problems and better coordinating existing social services. Forty-seven TEIPs were in place in 1998/99, covering 330 schools.

68. The government has also stepped up its program of school consolidation at the first cycle of primary education. The Ministry of Education plans to consolidate 1,300 schools over the next few years, with 120 being consolidated in academic year 1998/99. The grouping of schools is expected to enhance both the quality of education and reduce expenditures over the longer term. The quality of education will be improved by placing students in class settings of more appropriate sizes and in improved facilities; cost savings will come by a reduction in outlays for substitute teachers and maintenance costs for older, underutilized buildings. Reduced spending on substitute teachers will occur in line with the decline in the number of schools with only one teacher, which must use substitutes more often than their larger counterparts. As a first step for improving the efficiency of primary and secondary spending more generally, the government has also improved the dissemination of information on expenditures by school to school directors.

69. The government’s efforts to boost performance at the secondary level have centered on expanding the educational options for students at this level, with a view to increasing the share of students in vocational studies. Among the most important of the government’s efforts was the creation of separate professional schools at the secondary level (Escolas Professionais) in 1989, with the intent of attracting students to vocational training. Between 1989 and 1993 the number of such schools rose from 50 to 168, but has remained roughly constant since that year. Approximately seven percent of all secondary school students were enrolled in these professional schools in 1997/98. In order to provide better guidance to students in their choice of studies, counseling services were also expanded. Coverage is still modest, however, as only about 10 percent of all students had access to these services by school year 1998/99. Further improvements in student guidance are expected through the AZIMUTE program launched in 1998, which seeks, among other objectives, to create an Internet database on public educational opportunities for secondary students.

70. A wider gamut of technology and vocational courses has also been offered through the regular secondary school system. This has coincided with a rise in the share of secondary students enrolled in these courses from 13 percent of all students in 1991/92 to 29 percent in 1997/98.47 Apprenticeship programs through the Institute for Employment and Professional Training (IEFP) were also expanded sharply in the 1990s. In 1998, some 16,000 students participated in apprenticeship programs. Apprenticeships increased rapidly in 1998 (by 27 percent), exceeding the Employment Action Plan target of 20 percent. The total number of apprenticeship opportunities remains modest, however, in relation to the total student population at lower and upper secondary school (less than 5 percent), and the shortfall in training opportunities is estimated at about 40 percent. Vocational training was also strengthened in 1996 with the creation of special one-year programs for students completing the ninth grade; by 1998/99, enrollment had reached approximately 1,200 students. Adult education was also expanded in 1998 with the initiation of the ENDURANCE program, which seeks to promote life-long learning.

71. Despite the gains of recent years, the share of students in vocational studies remains significantly below the OECD average, as well as the percentage of the population graduating with a secondary degree in a vocational field. More rapid progress on these fronts has been hampered by the low success rates of students enrolled in these classes, which are actually lower than those in the more academically oriented studies at the secondary level. Thus, the available evidence indicates that a reallocation of students from general studies into vocational studies, while potentially helpful, would not be a panacea for boosting aggregate completion rates.

72. To enhance the success rate of repeater courses (ensino recorrente), the government has also introduced alternative curricula in some schools that are designed to better meet the needs of older students returning to the classroom. These alternative curricula are still not available at the secondary level, however.

73. A more general reform of the curriculum is slated for completion in 1999. The new curriculum will provide more flexibility for schools to tailor their course offerings to the needs and preferences of their students by concentrating on a core set of subjects, leaving more time than at present up to the discretion of individual school administrators. In addition, secondary students specializing in technical studies (including those at professional schools) will begin taking different twelfth grade exit examinations than those in general studies, beginning in academic year 2000/01. This will allow a greater share of classroom time to be devoted to technical material, and hence improve the effectiveness of these programs. In addition, the Ministry of Education has proposed a reform of the secondary school curriculum—to be put in place for school year 2001/02—that would allow more flexibility and reduce the required amount of mathematics for those students concentrating on the humanities.48

74. School attendance rates of vulnerable groups at the secondary level are also expected to rise on account of a new merit scholarship program for low-income families. Introduced in school year 1998/99, these scholarships provide cash of 100,000 escudos per annum (US$555) in three payments. Between 3,600 to 5,000 students are projected to receive this benefit this school year.

75. A critical component of the government’s strategy at both the primary and secondary level is improved monitoring and evaluation of educational performance. Research efforts of the Ministry of Education have been stepped up through the creation of a research department within the ministry in 1997. In 1998, the Permanent Observatory for Secondary Education was created, with a view to disseminating information on best practices in secondary education; looking forward, the government has proposed the creation of a similar program for primary education to commence in the academic year 1999/2000, with a view to conducting further research on the factors behind high dropout rates at the primary level. The government will also finish by end-1999 the design of nationwide standardized tests at the fourth, sixth, and ninth grades, which will provide better data on academic performance across schools and regions.

76. The government is also promoting an increase in school autonomy as a means to improve academic outcomes. The reorganization of school administration proposed in 1998 envisages that schools will receive greater autonomy in resource management (including personnel) in return for agreeing to meet specific targets or quantitative goals formalized through “contracts” between the school and the Ministry of Education. By end-1999, it is envisaged that the government will agree upon the quantitative targets that could provide the basis for such contracts, which could be put in place by academic year 2000/01. It is unclear how many schools will move to the more autonomous form of management, however, given that participation will be voluntary.

77. In sum, the government’s reforms planned in the near term are likely to have a salutary effect on school achievement, and hence offer the possibility of improved efficiency. The effects on expenditures are likely to be limited, however, given the absence of any planned reduction in the number of regular teaching staff.

H. Possible Directions for Further Reform and Research

78. Given the sizable disparity in performance that still exists between Portugal and the rest of the OECD, and some concern about the pace of expected improvement, the examination of possible reforms to further strengthen the performance of the educational system merits attention. Many of these reforms could be incorporated into the authorities’ framework for educational reform delineated above. Among those that could be considered are the following:

  • Set explicit and appropriately defined targets for performance by school. Educational performance could be improved by setting explicit targets on key quantitative measures of achievement or performance, such as success rates (percentage of students successfully completing the grade), repetition rates, and scores on national examinations. The establishment of quantitative targets could be woven into the school contracts envisaged under the government’s planned increase in school autonomy. The new achievement examinations being implemented at various grade levels also provide an important opportunity for tracking and assessing performance by school. In this context, additional modifications to the curriculum could be required to strengthen academic performance and facilitate meeting these targets.

  • Design an appropriate management and incentive system to facilitate achievement of educational goals. Imbuing the public education system with a more goal-oriented focus would require a substantial change in the incentive and evaluation system for educational personnel. At present, school directors are chosen largely on the basis of the preferences of their fellow teachers. Increasing the accountability of school directors would appear to be a necessary first step in providing the basis for improving school performance. In a similar vein, the system of teacher compensation could warrant review, as salary increments are based on length of service, rather than any measure of merit. In tandem with the tenure system, the present system provides little incentive or reward for improved performance.49 Drawing on the Dutch experience, public school performance could also be enhanced by wide publicity and dissemination of evaluations of school performance (for example, average scores on the forthcoming national tests at various grades).50 This information would allow the public to evaluate the comparative performance of their schools and allow greater public pressure to bear on those schools with sub-par academic outcomes.

  • Improve the flexibility of personnel management. The quality of education could be enhanced by providing greater flexibility in managing educational personnel. At present, teachers compete for school openings at a national level, with preference given to those with the most experience. This results in a systematic movement of teachers from less desirable areas to developed urban centers, and a high degree of turnover of both teachers and school directors relative to other OECD countries—to the detriment of the quality of education (Clímaco, 1997). The quality of education could be improved if greater discretion was given to the Ministry of Education to allocate teachers by region or school, especially in light of the desire to concentrate resources in areas where dropout rates are high and educational attainment levels are low. The present system of teacher tenure could also be reviewed, in particular the practice of granting tenure to teachers in a particular school. This implies that schools that lose student population over time may have excessively high teacher/student ratios and high costs per student, while other schools are relatively understaffed. Improved flexibility of staffing could limit the need for new hires and allow some reduction in public outlays, without any adverse effect on educational output. Consideration could be given to hiring more teachers on a contractual basis, without the obligation of providing tenure.

    Increased flexibility of resource management is also important in light of uncertainties over the projected demand for secondary education. The Ministry of Education does not foresee a decrease in the student population, owing to rising enrollment rates. If enrollment rates do not climb as expected, however, it is likely that the projected decline in the population of 15–19 year olds (almost a 30 percent drop from end-1995 to 2005) could lead to a much smaller student body and a reduced demand for teachers. Levels of teacher compensation, which exceed those in other OECD countries relative to GDP per capita, could also be reviewed, especially in light of generous pension benefits and the modest level of working hours.

  • Establish minimum student/teacher ratios across schools and minimum school sizes. The Ministry of Education’s plan to consolidate primary schools at the first cycle provides a useful first step for a more economical allocation of educational resources, and the planned consolidation should move forward as soon as possible, with a view to merging all small schools where cost savings can be achieved. At the same time, there is a need to examine student/teacher ratios at all levels of education, and consideration should be given to establishing minimum student/teacher ratios and minimum school sizes.

  • Examine the composition of expenditures. Given the low level of nonwage current outlays, and the low level of spending for maintenance indicated in 1995 data, there is an urgent need to assess the adequacy of present allocations for nonwage inputs. As such, the possibilities of reallocating spending to achieve improved performance should be contemplated. Further, as noted by Pinto, Barros, and Lopes (1998), an improved database is needed to provide information on spending per student and a more detailed composition of expenditure per student to form the basis for improved management of educational expenditures.

  • Reassess the level of cost recovery in public universities. Given the high subsidy element involved in public education, as well as the high rates of return to university education, a more targeted approach to this subsidy may be warranted. This could be achieved, for example, by an increase in tuition fees, matched by an expansion of scholarships to poor students.

In many respects the conclusions of this study are tentative and further analysis will be needed. Further research could usefully focus on the following topics:

  • The causes of Portugal’s high repetition and failure rate. In particular, it would be useful to assess whether promotion standards are unduly high relative to other countries; whether assessment and testing methods are responsible for high failure rates; and whether assessment based on national examinations would provide a more uniform and accurate assessment of students. An item of special concern is the high failure rate at the twelfth grade, which may indicate that graduation standards are not consistent with the learning goals established at earlier grades.

  • The comparative performance of public and private schools. As suggested by Pinto, Barros, and Lopes (1998), additional research is warranted on the relative efficiency and performance of private schools in Portugal. The preliminary evidence gathered in this chapter suggests that private schools may be more efficient than their public sector counterparts, as they achieve higher success rates with higher student/teacher ratios. Additional work is needed to address whether these differences simply reflect the more favorable socio-economic status of students in private schools, rather than greater efficiency per se.

References

Prepared by Benedict Clements.

See OECD (1995a) for evidence from firm survey data regarding the adverse impact of shortages of educated labor on labor productivity.

Continued improvements in job training are also critical for increasing labor productivity. An examination of this topic is outside the scope of the present chapter.

The Ministry of Education provides guidelines on the qualifications and experience necessary to qualify as a candidate. It should be noted also that since 1998, a representative of the students’ parents also has a vote in selecting the school director.

Calculation based on Ministry of Education transfers in the 1999 budget to universities, polytechnical institutions, and scholarships, and for living expenses at universities (Acção Social).

Gross enrollment rates can exceed 100 percent when students from older age groups retake courses with a younger cohort.

Figures derived from the UNESCO educational database.

Belgium, Denmark, Western Germany, former Czechoslovakia, France, Hungary, Italy, Norway, Poland, and Sweden.

Throughout this chapter, references to the OECD average indicates the unweighted average for OECD countries (excluding Mexico and Turkey) for which data are available.

See the historical tables assembled in Barreto and others (1996).

Figures on primary and secondary education spending for 1996 are based on data provided by the Ministry of Education, and include outlays of the regional and local governments of 0.5 percent of GDP. Figures for public spending are based on budgetary allocations, rather than actual budgetary execution. Spending for the private sector is estimated based on enrollment shares (approximately 10 percent in primary and secondary school and 36 percent at the tertiary level), netting out the financial support these institutions receive from the public sector. GDP figures are taken from the Bank of Portugal. More recent expenditure figures for Ministry of Education outlays alone indicate an increase in this spending from 5.3 percent of GDP in 1996 to 5.5 percent of GDP in 1998.

The Ministry of Finance also provides an estimate of public education expenditure, on a national accounts basis, which includes some spending on training activities. Net of transfers (primarily pensions to retired teachers) estimated by the author on the basis of 1995 data, these figures indicate projected spending of approximately 7 percent of GDP in 1999.

UNESCO data indicate an average gross enrollment rate at the primary level of about 105 in Europe in 1995; the comparable figure for Portugal for 1995/96 was 128.

The expenditure figure for Portugal includes a staff estimate for spending in the private universities.

This issue is discussed further below. See Table 9 for comparative data on educational expenditure by economic category.

See OECD (1998), Table E3.1

The number of teaching hours is an imperfect measure of teacher effort, given that nonteaching duties may vary across countries. Some notion of the extent of nonteaching duties can be culled from information on the length of the school year. According to the European Commission (1996), the length of the school year is 175 days, compared with an average of 188 in 13 other EU countries (figures based on 1992/93 data).

See OECD (1996b), Figure 4.3. This figure refers to predicted salary for 1994 data, when the average salary of primary school teachers with 15 years of experience was 2.0 times per capita income. Given the decline in this ratio between 1994 and 1996, this discrepancy may have subsequently narrowed.

These figures should be interpreted with caution, as at least in one country (Korea) it appears that some nonwage current spending is classified under capital spending.

Data from other years show roughly a similar pattern with respect to the share of capital expenditures in total spending.

See OECD (1998), Table E6.1. Based on the number of schools without any computers, however, Portugal was near the mean of 13 percent. Furthermore, it should be noted that in recent years the availability of computers has risen dramatically, and by 1998 the ratio of students to computers rose to 35 (ratio calculated excluding the first cycle of primary education).

As part of the International Assessment of Educational Progress, schools were asked whether or not they had one or more of the following problems: overcrowded classrooms; inadequate facilities and maintenance; shortages of textbooks and other educational materials; student absenteeism; lack of discipline; and vandalism of school property. Fifty-six percent of Portuguese schools reported one or more problems, compared with 23 percent of schools in 11 OECD countries (based on the unweighted average across these countries). These figures are only suggestive of infrastructure problems, as they include problem areas not pertaining to infrastructure per se. See http://www.nces.ed.gov/internat/index.html.

For an examination of the difficult conditions at a secondary school in one of the poorer suburbs of Lisbon (Ameixoeira), see Santos Silva (1999). Among the difficulties he cites are the deterioration in the prefabricated housing structure, which was designed in the 1980s and originally intended for just temporary use, and the lack of space for sports or recreational activity. According to one school official, the school is inadequately heated in the winter months, providing a poor environment for learning.

Figures based on the Ministry of Education budget for public schools only.

For purposes of comparison, UNESCO data indicate that the repetition rate at the secondary level for a group of 12 other OECD countries, based on the latest available data over the 1987–95 period, was 5 percent, compared with a weighted average in grades 10 through 12 of 19 percent in Portugal in 1995/96.

This figure is an approximation for the number of students currently finishing secondary school, and may overstate current performance; recall that labor force survey data from 1997 indicate that 46 percent of the population aged 20–24 had finished secondary school. It is based on the total number of students graduating in 1995/1996 divided by the number of 18 year olds in the population. As such, it includes overage students in the nominator. A similar measure used by the Portuguese Ministry of Education is the rate of schooling, which is used for projections of future educational attainment by grade level. This ratio is defined as the number of graduates in grade n divided by the number of students entering the school system n years earlier.

The repetition rate should be evaluated with due caution as an indicator of educational inefficiency; reductions in the repetition rate achieved through a lowering of educational standards would not imply any improvement in efficiency.

This assessment overstates the savings that could be achieved in the short run, given the high share of fixed costs in total education spending (including for salaries of tenured teachers).

Figure based on total student population from Ministry of Education in continental Portugal of 1,712,275 students and 70,896 students at the primary and secondary levels in ensino recorrente and ensino nocturno (1998a). The figure does not include the over 62,000 students attempting to finish secondary school via ensino, of which 96 percent are over the normal school age.

Student/teacher ratios are above the average of continental Portugal in the Azores, but below average in Madeira.

Fund staff calculations based on data from Ministry of Education (1998d) for 1995/96. It should be noted that the dominant role of the public sector in serving rural areas—where student/teacher ratios tend to be lower—has the effect of lowering the country-wide student/teacher ratio relative to the private sector. Nevertheless, it should be noted that 1995/96 data for primary education (1st cycle) reveal that student/teacher ratios were still lower in public than private schools across all regions in continental Portugal.

Figures on net enrollment (which exclude students over the ages of 22) are lower (see OECD (1998)).

See also “Da Escola para o Trabalho,” Expresso, April 2, 1999, regarding the need to strengthen the ties between the workplace and universities.

A recent study by Eurostat (1998) of 13 member countries, based on 1994 data, revealed a Gini coefficent of 0.37 for Portugal, compared with .31 for the 12 other EU countries included in the study.

Recent examples of studies applying this technique to measure the efficiency of government spending are Vanden Eeckaut, Tulkens, and Jamar (1993), who assess the efficiency of Belgian municipalities; Fakan and Crombrugghe (1997), who rank the efficiency of OECD and transition economies’ government spending in terms of its ability to provide a diverse set of outputs (inter alia, a low infant mortality rate, life expectancy, the number of telephone lines); and Gupta, Honjo, and Verhoeven (1997), who examine the efficiency of government spending on health and education in developing countries.

The FDH technique was first introduced in Deprins, Simar, and Tulkens (1984). The present exposition of the FDH technique draws heavily from Gupta, Honjo, and Verhoeven (1997).

See Harbison and Hanushek (1992) for a review of educational production function studies.

See Gupta, Honjo, and Verhoeven (1997) and Tulkens and Vanden Eeckaut (1995) for further elaboration on the differences between FDH and DEA.

See Gupta, Honjo, and Verhoeven (1997) for an exposition of FDH analysis using multiple inputs and outputs.

Educational attainment rates are nevertheless an important measure of educational output and the production of human capital. The completion of secondary education is strongly correlated with reading and numerical literacy, a more direct measure of worker skills (OECD, 1997). Furthermore, more educated workers tend to receive greater amounts of job-related adult education (OECD, 1995b, 1997), implying that initial levels of educational attainment are strongly correlated with levels of human capital throughout the life cycle.

Note that in the case of Portugal, as in Table 14, we include those students that finish secondary school above the normal age. In this sense, the results overstate the efficiency of the Portuguese education system relative to other OECD countries.

Following Gupta, Honjo, and Verhoeven (1997), countries with a score of 1.0 are ranked according to the number of countries they “dominate.” The number of countries dominated refers to the number of countries which produce less output with the same or a greater level of input. The rationale for this procedure is that the efficiency score of a country dominating a large number of other countries is likely to be more robust to small variations in the sample than one dominating a smaller number of countries.

Fairly similar results (not reported here) were obtained when educational output was measured in terms of the percentage of the population aged 25–34 having completed secondary education, and input measured as primary and secondary expenditure to GDP (adjusted for student enrollment to population rates).

The adjustment for population structure is meant to incorporate the fact that countries with relatively younger populations (e.g., Ireland) can be expected to spend a higher share of GDP on education than countries where a larger share of the population is not of school age (e.g., Germany). Using data from the OECD (1998), this adjustment is made on the basis of the share of the population under 19 years of age.

It should be noted that Canada is not included in the ranking in Table 27 because it is “independently efficient,” that is, it does not dominate any other observation, but is not dominated by any other observation. This reflects a high level of input matched by a high level of output.

Figures for 1996 from the OECD (1998) indicate that schooling rates for 2–4 year olds (at 33 percent) were below the average for other OECD countries (42 percent).

This figure includes students in professional schools; see Ministry of Education (1998a), page 23.

On the importance of incentives in achieving improved educational performance, see Hanushek and others (1994).

Parents are also free to choose among different schools for their children in the Netherlands. In this context, the wide dissemination of information on school performance is seen as a critical element in ensuring that competition leads to improved educational efficiency (see Ritzen, 1999).

Other Resources Citing This Publication