IV Case Studies in Cost Sharing: Tanzania

A. Introduction
B. Structure of the Education System
C. The Economy, Public Finance and the Education Sector
D. Trends in Recurrent Education Expenditures in Tanzania
E. Distribution of the Benefits from Public Expenditure
F. Household Expenditures and Cost Sharing in Education
G. Conclusions

A. Introduction

The case of Tanzania provides a good contrast with that of Ghana, although both countries are dependent on foreign aid: aid dependency in Tanzania may even be stronger than in Ghana.¹¹¹ An important difference between the countries is their size: the population of Tanzania is about 28 million compared with Ghana's 16 million, and Tanzania is four times larger than Ghana. The provision of education services is strongly affected by factors such as population density and communications distances. There are also important differences between anglophone West Africa and East Africa in their historical experiences.

[¹¹¹ See Doriye J. & M. Wuyts, Aid, Adjustment and Sustainable Recovery, Institute of Social Studies, The Hague, March 1992, for a powerful analysis of Tanzania's aid dependency and the ineffectiveness of aid. Also Agrawal N. & al, Structural Adjustment, Economic Performance, and Aid Dependency in Tanzania, Policy Research Working Paper WPS 1204, World Bank, October 1993: this paper suggests that adjustment had been successful in Tanzania but accepts that the effectiveness of aid had been very low in spite of its large volume.]

Another common feature is the influence (or attempted influence) of lenders and donors on education policy. To some, Tanzania is seen as a political economy laboratory, and over the recent years the fervency of donors and lenders in promoting (on paper at least) a privatised education system matches the same fervency of 15 years previously in support of a somewhat different type of system. In many respects the experimentation of the World Bank and the donors who support the Bank's approaches is a classic example of the neo-liberal package described by Colclough.¹¹²

[¹¹² Colclough C. & J. Manor, States or Markets? Neo-liberalism and the Development Policy Debate, Clarendon Press, 1991, pp 197-213.]

In both countries lenders and donors have made strong efforts to drive policy, though to my knowledge there are no recent examples in Ghana of Bank authorship of government documents as was the case for the Tanzanian social sector policy. Ghana, however, implemented in the 1980s many of the measures that have thus far eluded Tanzania: for example, a student loans scheme, a textbook fund, and elimination of government expenditures on boarding costs and school food. The effect of those measures has not been fully evaluated, but may not have had the intended results, and it may in the longer run be to Tanzania's benefit that many of the BWI policies have been resisted in spite of the heavy Bretton Woods pressures.

The main difference between the countries in the context of cost sharing is the rapid rise of private secondary schools in Tanzania as a result of liberalisation policies and a history of minimal access to secondary education. Private schooling is the most expensive form of cost sharing for individuals, and it properly forms a significant part of analysis in a study such as this one when it becomes a major plank in government policy. In Tanzania, as elsewhere, the World Bank has vigorously advocated the expansion of private secondary schools as well as fee increases in government schools on the basis of very little evidence and a good deal of dogma, and has also constructed arguments for cost reductions in the state system on the basis of its perceptions of private school performance.¹¹³ The Bretton Woods emphasis on cost cutting and public expenditure reductions had a cruel effect on education expenditures and Tanzania is one of the lowest spenders on education, expressed as a percentage of GDP.

[¹¹³ For a similar criticism related to the health sector, see Mills A, Improving the Efficiency of Public Sector Health Services in Developing Countries: Bureaucratic vs Market Approaches, Health Policy Unit, London School of Hygiene and Tropical Medicine, p 7: 'In the two, recent reviews of developing country health policies (World Bank 1993), it is remarkable how many reforms are proposed, but how little detailed evidence can be put forward on the impact of past reforms in terms of quantitative measures of efficiency or equity.']

This case study is based on work undertaken in Tanzania between 1992 and 1996, including a specially prepared survey which took place in August 1994.¹¹⁴ Other sources of information on households and schooling are taken from two large scale surveys, one in 1991 by Cornell University and the Economic Research Bureau (ERB) of the University of Dar es Salaam and one by the World Bank in 1993/94.

[¹¹⁴ The survey work for this chapter was undertaken with the assistance of TADREG, and researchers included Brian Cooksey, Stella Bendera, Abel Ishumi, P. Lipembe, George Malekela, Emmanuel Nkumbi, Alice Rugumyamheto, Joseph Rugumyamheto, Fred Sichyiza, John Sivalon, and Fidelis Wamara who entered the raw data. I was also assisted by E. E. Moshi, who sadly died before the work was completed: he was a fine man and a great Tanzanian educator. Lucy Stevens and Samer Al-Samarrai assisted with data analysis.]

B. Structure of the Education System

Structure and Curriculum

Tanzania, unlike Ghana, has not changed its education structure over recent years to expand the basic education cycle. Apart from pre-school education, which is not publicly provided, though there is a policy to do so¹¹⁵, there are 7 years of primary education (up to Standard 7, when there is an examination), followed by four years to Ordinary Level (O Level, Form 4). Those who are permitted to proceed beyond O level take a 2 year Advanced Level (A Level, Form 6) course, followed by university.

[¹¹⁵ Tanzania Education and Training Policy, MOE, March 1995, p 10.]

As with Ghana, there has been some concern about the size of the curriculum. The primary curriculum was to have been reduced from 13 to 7 subjects, but the reform was postponed. However, unlike Kenya, not all subjects are examined, and the Standard 7 examination covers only Maths, English and a General paper. Three quarters of the teachers in my survey agreed that secondary curriculum reform was needed.

Enrolments and Staffing

Until the liberalisation and economic reform programmes education policy emphasised primary education, and entry to post-primary education was restricted. Table 23 shows the trends.

The table shows that apparent enrolment ratios were rising slowly for primary education although the enrolment of children of school age only was constant. While primary enrolments rose by 18 per cent between 1989 and 1995, secondary enrolments rose by 48 per cent, with over half accounted for by private schools. The proportion of private enrolments fell for the last three years or so of the period as enrolments grew in the state schools.

Table 23: Enrolments in Tanzanian Primary and Secondary Education

	1990/1	1991/2	1992/3	1993/4	1994/5	1995/6	1996/7
Primary Enrolments (1)(3)	3,379,000	3,507,394	3,599,580	3,732,493	3,793,201	3,872,473	3,942,888
Apparent Enrolment Ratio (2)(3)	73.5	74.4	74.2	74.9	77.9	77.6
Net Enrolment Ratio (2)(3)	54.2	53.8	54.2	53.7	55.2	55.4
Primary Teachers (1)(3)	96,850	98,714	101,306	101,816	103,900	105,280	110,200
PTR	34.9	35.5	35.5	36.7	36.5	36.8	35.8
Secondary Enrolments (1)(2)(3)	145,242	166,812	175,776	180,899	186,246	196,375	199,093
of which private (per cent)	57.4	55.7	55.2	58.2	55.2	53.1
Secondary Teachers (2)(1)(3)	6,930	8,649	8,926	9,568	10,928	11,158	10,908
Apparent Enrolment Ratio						5.1
PTR	21.0	19.3	19.7	18.9	17.0	17.6	18.2

Notes and Sources: From MOE statistics. (1) Basic Education Statistics in Tanzania (BEST): Regional, 1994, MOE March 1996. (2) BEST 1989-1993, MOE June 1994. Most Tanzanian education statistics differ between sources. The population base for the AER is extrapolated from the 1988 census. (3) Basic Education Statistics in Tanzania (BEST), 1995, & 1996, MOE 1996 & 1997

As with Ghana, survey data provide alternative enrolment ratios, and reveal enrolment ratios higher than those given in the normal education statistics which use school registrations and population extrapolations. These are shown in Table 24. The data show that girls' enrolments are relatively high at most income levels, at both primary and secondary school. Secondary enrolment ratios are very low indeed: the high proportion of private enrolments is placed into context, because while it is to be expected that a section of the population will have sufficient income to purchase secondary education, it is highly unlikely that a figure of 50 per cent would be maintained as enrolments rise. A further notable conclusion from the table it is evident difficulty faced by urban children in gaining access to primary education, particularly in Dar es Salaam.

Table 24: Primary and Secondary Enrolment Ratios: Tanzania 1993

	Lowest Quintile		2nd Quintile		3rd Quintile		4th Quintile		Top Quintile		Average
	Male	Female	Male	Female	Male	Female	Male	Female	Male	Female	Male	Female
Apparent Enrolment Ratios
Primary School
Dar es Salaam	n/a	n/a	64	60	79	66	78	80	74	72	75	73
Other Urban	79	94	88	91	90	103	92	86	88	92	89	93
Rural	77	74	75	73	76	96	82	75	90	87	79	80
All Tanzania	77	76	77	76	79	97	85	79	87	86	81	82
Secondary School
Dar es Salaam	n/a	n/a	n/a	16	10	3	14	21	23	21	18	19
Other Urban	1	2	8	13	6	10	17	22	31	28	15	17
Rural	3	2	6	3	14	5	12	6	15	7	8	4
All Tanzania	3	2	7	5	11	6	14	12	23	17	11	8
Net Enrolment Ratios
Primary School
Dar es Salaam	n/a	n/a	53	41	47	46	49	63	51	53	50	54
Other Urban	54	60	57	70	59	68	62	68	69	68	62	8
Rural	50	52	52	53	53	62	51	54	66	65	53	56
All Tanzania	50	53	53	56	54	63	54	59	65	65	55	59
Secondary School
Dar es Salaam	n/a	n/a	n/a	16	7	0	9	19	18	19	13	17
Other Urban	1	2	8	12	5	6	11	16	22	24	11	14
Rural	2	2	3	3	9	4	9	6	7	5	5	3
All Tanzania	2	2	4	5	8	4	9	10	15	14	7	7

Notes and Sources: From Social Sector Household Survey. Note that the sample sizes for secondary students were low, to the point of insignificance in some cases, and margin of error in others.

Like Ghana, the distribution of the PTR in Tanzania between regions varies considerably, as does the number of pupils per stream. The average number of primary teachers per stream (class), however, is usually close to 1, with some exceptions, and the national average is 1. The primary PTRs are also considerably higher than Ghana, while the apparent primary enrolment ratio is similar, implying there will be considerable difficulty in raising the PTR without raising enrolments. Where there is a significant mismatch between the two ratios there exists an excess or deficit of teachers against the norms of one teacher per stream.

It can be seen in Table 25 that the variation around the average is mainly the result of higher levels of staffing per stream in the urban areas and in some cases a higher density of enrolment, resulting in larger average stream sizes. While class and school sizes are larger in urban centres, the PTR is smaller, reflecting underutilised teachers. The PTR exceeds class sizes: to have lower class sizes would require an even greater imbalance between the PTR and stream size (increases in the aggregate pupil teacher ratio do not necessarily achieve larger classes; neither is the opposite automatically true). The efficient disposition of teachers is a critical issue in considering cost sharing, because where the disposition is not efficient parents are made to pay more than they otherwise would.

Table 25: Primary Enrolments and Teachers by Region: Tanzania 1994

Region	Enrolments	Teachers	PTR	Streams	Teachers/Stream	Pupils/Stream
Arusha	234,518	5,557	42.2	6,039	0.9	38.8
DSM	231,437	5,575	41.5	5,304	1.1	43.6
Dodoma	184,671	5,486	33.7	4,874	1.1	37.9
Iringa	232,985	5,992	38.9	6,987	0.9	33.3
Kagera	201,363	6,092	33.1	5,259	1.2	38.3
Kigoma	134,932	3,784	35.7	3,620	1.0	37.3
Kilimanjaro	242,493	8,022	30.2	6,700	1.2	36.2
Lindi	85,001	2,782	30.6	2,722	1.0	31.2
Mara	193,559	5,302	36.5	5,591	0.9	34.6
Mbeya	274,229	7,805	35.1	7,932	1.0	34.6
Morogoro	196,735	5,614	35.0	5,543	1.0	35.5
Mtwara	129,996	4,004	32.5	4,346	0.9	29.9
Mwanza	312,340	7,458	41.9	8,161	0.9	38.3
Pwani	99,039	2,620	37.8	3,180	0.8	31.1
Rukwa	113,934	3,138	36.3	3,198	1.0	35.6
Ruvuma	138,426	4,736	29.2	4,064	1.2	34.1
Shinyanga	284,399	5,700	49.9	8,391	0.7	33.9
Singida	133,833	3,556	37.6	3,575	1.0	37.4
Tabora	151,058	4,393	34.4	4,351	1.0	34.7
Tanga	218,793	6,284	34.8	6,106	1.0	35.8
National	3,793,201	103,900	36.5	105,943	1.0	35.8

Source: Estimated from BEST 1994

Higher Education. The Tanzanian higher education system is fragmented with institutions which duplicate each other's activities. The higher education system is characterised by low student teacher ratios, small institutions, and low capacity utilisation.¹¹⁶ As in many centrally planned economies, sectoral ministries and agencies established their own training institutions: this has the effect of making proper analysis very difficult. The dominant institution is the University of Dar es Salaam with some 3,000 students, with an average student/staff ratio of 4.6. There has been considerable discussion about reforming the sector.

[¹¹⁶ For some detail see Omari I, P. N. Materu & T. Mteleka, Rationalisation of the Tertiary Education and Training Sector in Tanzania, Ministry of Science, Technology and Higher Education, Draft, March 1996.]

In conclusion, apparent primary enrolment ratios may have risen slightly, while the net enrolment ratios are stagnant. The secondary enrolment ratio is low, and there is less likelihood of poorer families being able to place their children in secondary school. Those enrolment trends provide important background to any analysis of cost sharing.

C. The Economy, Public Finance and the Education Sector

As with Ghana, the overall economy and public finances have had a very significant impact on the education sector and the discretionary resources available. Weaknesses in public finance management reduce the feasibility of making budgetary reallocations.

Figure 5: Shares of Government Expenditure on Education in GDP and Total Budget, Tanzania 1990-1996

Notes and Sources: As previous tables.

National Income and Public Expenditures

Figure 5 depicts the trends over recent years. The left hand axis of the Figure 5 measures public expenditures and the right hand axis measures national output. The four lines plot trends in GDP, total government expenditure, discretionary government expenditure (after items such as pensions and debt payment are subtracted), and education sectoral expenditure. The shaded area in the chart illustrates the gap between total government expenditure and total government discretionary expenditures.

The chart shows that there have been fluctuations in the level of non-discretionary expenditures, but some of these fluctuations may be accounted for by poor data, and some by deferral of debt payments: it can be seen that the budgeted 1996 amount is returning to the more normal level of about one third of the budget. In other words, one third of all government recurrent expenditure is not allocated to expenditures on services and other items, but to debt payment.

Table 26: Shares of Government Expenditure on Education in GDP and Total Budget, Tanzania 1990-1996 (Actual and estimated expenditures in million shillings at current and constant 1994 prices)

Current Prices		1990/91	1991/92	1992/93	1993/94	1994/95	1995/96
GDP		935,074	1,130,596	1,267,432	1,635,470	2,284,600	2,998,272
Total Government Recurrent Expenditure		161,224	199,670	250,942	263,226	386,573	470,014
Government Exp/GDP		17.2%	17.7%	19.8%	16.1%	16.9%	15.7%
CFS (1)		49,817	59,165	68,166	33,826	109,376	160,690
CFS % recurrent expenditure		30.9%	29.6%	27.2%	12.9%	28.3%	34.2%
Discretionary Recurrent Expenditure(2)		111,407	140,505	182,776	229,400	277,197	328,805
%GDP		11.9%	12.4%	14.4%	14.0%	12.1%	11.0%
% Total		69.1%	70.4%	72.8%	87.1%	71.7%	70.0%
Education Recurrent (see notes)		21,880	27,587	33,055	46,782	78,351	78,587
	as % GDP	2.3%	2.4%	2.6%	2.9%	3.4%	2.6%
	as % discretionary recurrent exp	19.6%	19.6%	18.1%	20.4%	28.3%	23.9%
	as % total recurrent exp	13.6%	13.8%	13.2%	17.8%	20.3%	16.7%
1994 Constant Prices
GDP		1,753,264	1,729,093	1,602,451	1,635,470	1,746,280	1,823,471
Total Government Recurrent Expenditure		416,004	422,115	429,610	352,708	386,573	373,905
Discretionary Recurrent Expenditure (2)		287,462	297,037	312,911	307,383	277,197	261,570
Education Recurrent		56,457	58,321	56,590	62,685	78,351	62,517
Memo item: deflator (3)		100	123	148	188	245	308

Notes and Sources: From 1996 Public Expenditure Review (draft), which was undertaken by World Bank staff and consultants. I have chosen to use the draft 1996 PER data to provide an up-to-date a picture as possible, and to make the figures in the table consistent with each other. However, the data are very misleading, in that the 'actual expenditure' figures are drawn from the Estimates Books and not the appropriation accounts. The 1993 and 1995 data are particularly suspect. Education expenditure in 1993 was actually TSh 40 million. The 1995 data may contain expected counterpart funds which were neither transferred to the Treasury nor spent, but for some reason remained in the accounts. (1) CFS = Consolidated Fund Service, which is mainly made up of debt costs. (2) Total recurrent less CFS. (3) Deflator from IMF tables.

The data, explained in the notes to Table 26, are also not reliable: they are drawn from the 1996 PER undertaken by World Bank staff. Nevertheless, apart from the 1994/95 estimated expenditures which are plainly wrong, it does appear that Tanzania has been under spending on education, with total recurrent expenditures hovering at about 2.5 per cent of GDP, and 20 per cent of the discretionary budget, which is itself under 20 per cent of GDP. As in the case of Ghana, it must be concluded that there is a strong case for increased education expenditures.¹¹⁷

[¹¹⁷ Cf the World Bank's Social Sector Review, para 36 (in the draft): '...spending is already high for such a poor country' (a curious sentiment when it is intended as a justification for higher household contributions!).]

D. Trends in Recurrent Education Expenditures in Tanzania

Total Expenditure Trends

Table 27 sets out recent trends of public expenditure on education in Tanzania by sub-sector, in both current prices and constant prices (1994 equivalent shillings). The trends are shown graphically in Figure 6. There appears to have been a steady increase in total primary expenditure in real terms over the period, while secondary expenditures, having risen, are on a falling trend.

Table 27: Trends in Government Education Expenditures, Tanzania 1990-1996 (Actual expenditures in '000,000 shillings at constant 1994 prices)

Current prices	1990/91	1990/91	1991/92	1992/93	1994/93	1994/95
Primary	10,284	13,104	16,693	23,859	49,174	51,602
Secondary	4,376	4,828	4,793	6,783	7,533	6,608
Teacher Education	1,750	1,931	1,653	2,105	2,013	1,458
Higher & Technical	3,282	5,242	6,611	9,824	15,922	16,836
Other	2,188	2,483	3,306	4,210	4,524	2,596
Total	21,880	27,587	33,055	46,782	79,166	79,100
Constant 1994 prices
Primary	19,283	20,041	21,105	23,859	37,587	31,383
Secondary	8,205	7,383	6,060	6,783	5,758	4,019
Teacher Education	3,282	2,953	2,090	2,105	1,539	887
Higher & Technical	6,154	8,016	8,358	9,824	12,170	10,239
Other	4,103	3,797	4,179	4,210	3,458	1,579
Total	41,026	42,191	41,793	46,782	60,512	48,107

Notes and Sources: Ministry of Finance. Higher education does not include interest rate subsidies for the student loans scheme. The data are not consistent with the PER.

Figure 6: Trends in Government Education Expenditures, Tanzania 1990-1996 (Actual expenditures in '000,000 shillings at constant 1994 prices)

On the other hand, higher and technical education expenditures rose as the post secondary sector faced the need for rationalisation. In real terms total expenditures on education rose by about 20 per cent between 1989/90 and 1993/94. They appear to have risen by a further third in 1994/95, but those data require further investigation, particularly in view of the 1995/96 expenditure. The total real increase over the six years of the period was under 10 per cent.

Taking into account the enrolment trends, crude average expenditures for primary and secondary education can be derived. These are shown in Table 28 and shown in Figure 7.

Table 28: Average Expenditures per Student, Tanzania Tanzanian Shillings

	1990/91	1991/92	1992/93	1993/94	1994/95	1995/96
Current Prices
Primary	3,044	3,736	4,637	6,392	12,964	13,325
Secondary	30,129	28,941	27,268	37,498	40,447	33,650
1994 Prices
Primary	5,707	5,714	5,863	6,392	9,909	8,104
Secondary	56,492	44,261	34,475	37,498	30,916	20,465

Notes and Sources: Budget and enrolment data as previous tables

Secondary average expenditures fell sharply, even though the pupil teacher ratio also fell, reflecting the increase in enrolments as well as measures to reduce expenditure on items such as food. Primary averages were maintained. Over the period, the ratio of secondary average expenditures to primary average expenditures per pupil fell from 8 to 3.6.

Figure 7: Per student Government Expenditures for School Education, Tanzania 1990-1996 (Actual & Budgeted TSh at constant 1994 prices)

In conclusion, there were strong pressures on education expenditures which resulted in falling secondary education expenditures and slowly rising primary expenditures. Education took a lower proportion of national income compared to other countries. Public sector reforms should release considerably more resources into the discretionary budget, and this weakens the case for higher dependence of the sector on cost sharing.

E. Distribution of the Benefits from Public Expenditure

From the 1993/94 household survey it is possible to compute a Lorenz curve for the distribution of public expenditures on education between consumption quintiles. While the sample size for primary children was large (over 4,300), there were less than 400 secondary children in the sample and only 5 government tertiary students. These samples of course represent the low enrolments at these levels, but also reduce the reliability of the benefit incidence tables. The Lorenz curve is shown in Figure 8.

Figure 8: Benefit Incidence of Public Expenditures on Education in Tanzania, 1992/93

Notes & Sources: Computed from Grosh M. and L. Forgy, Incidence of Selected Social Services in Tanzania, mimeo, April 1994, and Tanzania: Social Sector Review, draft April 1995, table S.6, p xxvi. The sample sizes for post primary education were small.

Primary education expenditures are mildly progressive, while secondary expenditures are skewed in favour of the higher income groups. University education is shown to be entirely benefiting the higher income group, but this may reflect the sample size. Compared with Ghana, the distribution of the benefits of public expenditure on education in Tanzania seems to be less progressive. While the history of the country would suggest that the incidence of primary education expenditures was progressive, the fall in enrolments which occurred during the 1980s and the failure of poorer people to go to school, particularly in urban areas, must be influencing the pattern. In many respects the incidence of secondary education benefits is to be expected, and, indeed, given the absolute low enrolments, it indicates that the distribution of secondary opportunities is not unacceptably regressive. How far this has changed over recent years is unknown, and the effect on the poor of phasing out catering subsidies remains to be seen.