CONFIDENTIAL Public Disclosure Authorized

INTERNATIONAL BANK FOR RECONSTRUCTION AND DEVELOPMENT

INTERNATIONAL DEVELOPMENT ASSOCIATION

PROJECT PERFORMANCE AUDIT REPORT Public Disclosure Authorized

TANZANIA REVISED FIRST HIGHWAY PROJECT (Credit 115-TA)

June 26, 1975 Public Disclosure Authorized

Operations Evaluation Department Public Disclosure Authorized

PROJECT PERFORMANCE AUDIT REPORT

TANZANIA REVISED FIRST HIGHWAY PROJECT (CREDIT 115-TA)

PREFACE

This report reviews the implementation and achievements of the Tanzania Revised First Highway Project, supported by IDA Credit 115-TA signed in March 1968 and closed in July 1973, The credit provided supplementary financing for the Tanzania First Highway Project, supported by IDA Credit 48-TA, signed in February 1964.

This performance audit is based on discussions with IBRD/IDA staff, study of correspondence and reports from the IBRD/IDA files, and a brief mission to Tanzania in November 1974 to collect information and discuss the project with country officials concerned. The valuable assistance of those officials is gratefully acknowledged.

Note: Currency Equivalent (Tanzania Shilling):

1964-74 -- US$ 1.00 = T Sh 7.14

PROJECT PERFORMANCE AUDIT REPORT

TANZANIA REVISED FIRST HIGHWAY PROJECT (Credit 115-TA)

TABLE OF CONTENTS

Page No.

Summary i-ii

Background 1

Project Implementation 3

Project Impact 5

Road Construction 5 Traffic Growth 7 Detailed Engineering 8 Staffing and Training 9

Conclusion 11

List of Tables

1. Description of Original and Revised Projects 2. Completion Dates and Estimated and Actual Costs for the Project Components 3. Estimated and Actual Disbursement Schedule 4. Central Government Expenditures on Highway Construction and Maintenance 5. Traffic and Unit Vehicle Cost Savings on the Project Roads 6. Estimated Traffic Growth Rates on the Project Roads 7. Vehicle Fleet, 1962-72 8. Staff Authorized and Employed in Roads and Aerodromes Division, Ministry of Communications and Works

Annex

1. Details about the Project Roads

Map

1. Tanzania Highway System

SUMMARY

IDA made a US$ 14 million credit (48-TA) to Tanzania in February 1964 to finance the foreign exchange cost of a First Highway Project which included detailed engineering and construction of eight road sections. Shortly thereafter, most expatriate staff in the Roads and Aerodromes Division (RAD) of the Ministry of Communications and Works (Comworks), which had been charged with project implementation, left the country. Through their exodus, RAD's professional manpower had been depleted and project implementation seriously delayed. In addition, the work quantities of the project had been underestimated by RAD, based on preliminary engineering, and prices had escalated; therefore, the cost of the project had increased considerably by 1966.

Because of the delays in implementation and increased cost, the Ministry of Finance asked IDA to defer the credit's scheduled closing date and to supplement the credit to meet the increased cost. IDA reappraised and revised the project, and provided a US$3 million supplementary credit (115-TA) in March 1968 to enable achievement of a Revised First Highway Project.

Compared with the original project, the revised project was to: (a) retain detailed engineering and construction of 533 mi of six of the eight road sections; (b) extend detailed engineering of the section of the Tanzam Highway between Morogoro and Iringa from 104 mi to 142 mi, but omit construction; (c) include detailed engineering of feeder roads (110 mi of four secondary roads and 98 mi of tertiary roads) on the Geita Peninsula as well as a five-year program of staffing and training for RAD; and (d) omit construction of the Musoma-North Mara road section.

Construction of the project roads, about half done by the time of appraisal of Credit 115-TA, was completed satisfactorily by 1970, and the detailed engineering work was carried out as envisaged. Engineering of the Morogoro-Iringa section of the Tanzam Highway provided the detailed design for its construction, subsequently implemented under the Second Highway Project (Loan 586-TA and Credit 142-TA of 1969). Engineering of the Geita Roads, after further revision, became the basis for their upgrading, now being implemented under the Third Highway Project (Credit 265-TA of 1971). The staffing and training program was carried out to a very limited extent because of slowness in recruiting expatriates, difficulties with training by them and absolute shortage of African staff. The closing date for the credit was post- poned several times due to an outstanding contractor's claim, and the final disbursement was made on July 25, 1973, two years behind schedule. 'About US$ 16.6 million was disbursed from the two credits together, and about US$ 440,000 of Credit 115-TA was cancelled at closing. The actual cost of the revised project was US$ 22.2 million, or 3% less than estimated.

Taking all aspects into account including the increase in fuel prices over the last years which adds about one percentage point to the rate of return, the rather shaky information available suggests that the overall economic return to the project probably approximates the 11% estimated in the 1968 appraisal report. Three of the roads indicate clearly lower rates - 11 - of return at audit than at appraisal and one a substantially higher return. Special allowance in the calculations for truck overloading, which appears to be only somewhat less serious a problem on these roads than on the Tan- zam highway, indicates that the effect of related additional vehicle operating cost savings will be offset or more than offset by the consequent damage to the roads and shortening of their lives.

Traffic overestimation was the main factor in the shortfall from projected economic returns. The primary reasons for the traffic overestimation were IDA's appraisal assumption that decreased vehicle operating costs would be enough to induce substantial generated traffic and omission of an assessment of complementary investment in the regions surrounding the roads. Unstable political conditions as well as low growth of the national economy and the vehicle fleet contributed to the overestimation, but they could not have been anticipated at appraisal.

Also in retrospect the staffing and training program was not for- mulated as carefully as it should have been to fulfil IDA's objective of eas- ing RAD's transition to Africanization. It appears to have been too dependent on one man, the then Permanent Secretary of Comworks, who was transferred shortly after Credit 115-TA was signed. To get the roads built it had been essential to hand over more responsibility for detailed engineering and construction supervision to consultants, as was done in accordance with IDA recommendations, and some RAD Tanzanian staff have been assigned to consultants for useful training, but they did not find this as satisfactory as would have been an earlier transition to carrying fuller responsibility for individual works. The sudden departure of Tanzanian engineers of Asian origin in 1971-72 contributed substantially to staffing problems, but could not have been anticipated at appraisal or reappraisal.

The fact that three of the road improvements, accounting for about 40% of construction investment under the project, show less than 10% economic return on current estimates, combined with the limited manner in which the project has contributed to the development of African capabilities - above all because of the sheer shortage of African secondary school and engineering graduates - suggests that the original project, feared at the time to be too big but proceeded with as a special support tothe newly independent country, should have been cut back in thel967 revision even more than it was. The considerable difficulties of doing this soundly at that time are well illustrated by the fact that the Dar-es-Salaam-Kibiti road which was considered of most dubious economic validity at the time of the revision has turned out to be the one with the highest economic return according to current estimates - mainly due to substantial development of traffic, especially agricultural, affected by the proximity of the large Dar-es-Salaam market - while almost none of the generated traffic projected for other roads appears in fact to have materialized. It may be that the only course which would have provided more assured success was the one suggested by an Executive Director at the discussion of the original 1964 project, namely an integrated approach covering agricultural as well as transportation investments and policies, in unified or separate projects, such as has come to be applied to some extent in con- nection with the project road in the southern region, where the Bank approved support for a cashewnut development project in 1974. PROJECT PERFORMANCE AUDIT REPORT

TANZANIA REVISED FIRST HIGHWAY PROJECT (Credit 115-TA)

BACKGROUND

IDA became involved in Tanzania's road program after the Govern- ment's 1961 decision to further develop main roads and to construct secondary and tertiary (feeder)roads. The existing network consisted of low-standard, low-cost facilities built before independence to carry light traffic between widely dispersed population centers. Main roads were upgraded to fully engineered standards through reconstruction and realignment when traffic reached appropriate levels.

The first highway project developed by the Government for IDA's consideration, on the basis of preliminary discussions in 1961-63, included detailed engineering and construction of eight road sections covering 714 miles (see Table 1). The total cost of the works was estimated at US$ 18.8 million. Considerable doubts within the Bank as to whether such a large project was not beyond the capabilities of the Tanzanian Government to exe- cute in a reasonable period were eventually overruled, apparently largely due to the urgings of Tanzania's expatriate officials and the Bank's desire to demonstrate confidence in the newly independent country. Contrasting the pessimism of the Bank's economic report on the prospects for Tanzanian agriculture with the apparent optimism in the appraisal report's traffic projections, some of the Executive Directors also raised questions whether the project should not be cut back substantially or at least undertaken in the form of an integrated program with supporting agricultural investments and policies, but some assurance was provided that the areas to be traversed did have agricultural potential and programs, and a US$ 14 million credit, to finance the foreign cost of the US$ 18.8 million project, was approved in February 1964, with a closing date in 1967.

In the event, implementation of this project (Credit 48-TA) was seriously delayed by various factors: long negotiations to engage engineering consultants; slow construction due to defective pavement design and unexpectedly bad soil conditions; and, above all, a more rapid than expected exodus of expatriate staff from the Roads ar%d Aerodromes Division (RAD) of the Ministry of Communications and Works (Comworks), responsible for project execution. Before independence in 1961, few RAD top or middle management positions had been open to Tanzanian citizens. Further, few Africans had received the education and training necessary to fill those posts. l/ But by 1965 RAD had lost almost all of its senior engineers, including most of those who had prepared and negotiated the credit. Their replacements inade- quately understood,the conditions of the IDA credit. RAD's work capacity was further impaired by its slow decision-making process regarding,project implementation, which was mainly due to the involvement of several layers of bureaucracy. These problems were eased by the hiring of consultants for

1/ Tanzania's first manpower survey in 1962 revealed that non-Africans occupied all jobs requiring a university degree. -2-

most of the project works, as IDA had insisted, and this approach was eventually extended to two further roads, leaving RAD responsible for detailed engineering and construction supervision only on the one road that it was building by force account under the project.

In 1966, a review of completed detailed engineering and contrac- tors' tenders indicated that the project cost would exceed 1963 estimates by 40%, primarily because RAD had substantially underestimated work quantities and costs, based on preliminary engineering. IDA had recognized the underestimation of costs and, at appraisal, had persuaded RAD to increase its estimates somewha;, but the extent of the underestimation had not been fully appreciated.- Rising prices, aggravated by continuous delays in execution, also had contributed to the increase in costs.

Because of the delays in implementation and cost increases, the Ministry of Finance requested that the scheduled closing date of the credit be deferred and that the credit be supplemented to meet the increased cost. As a result the project was reappraised and revised. Credit negotiations for the Revised First Highway Project were held in 1967. Standard covenants and supplemental agreements for highway projects were included in the documents. In addition, covenants were included whereby all,roads were to be adequately maintained, a senior RAD engineer was to be appointed to oversee project execution, and a staffing and training program was to be carried out by recruitment of foreign engineers.

The estimated cost of the revised project was US$ 22.8 million, including contingencies, or US$ 4 million more than the original project. A US$ 3 million credit (115-TA) was provided in March 1968 to finance 75% of the increased cost. The credit became effective in May 1968, and was scheduled to be closed in December 1971.

Compared with the original project, the revised project (see Table 1) was to: (a) retain detailed engineering and construction of 533 mi on six of the eight road sections; (b) extend detailed engineering of the section of the Tanzam Highway between Morogoro and Iringa from 104 mi to 142 mi, but omit construction; (c) include detailed engineering of feeder roads (110 mi of four secondary roads and 98 mi of tertiary roads) on the Geita Peninsula as well as a five-year program of staffing and training for RAD; and (d) omit construction of the Musoma-Nordi Mara road section.

Of the two roads dropped from the original project, Morogoro- Iringa was the most important. On this road, traffic had grown more than expected and redesign to higher standards was necessary.' Furthermore, traffic had increased considerably because of transit between Zambia and Mombasa Port in Kenya. Consequently, construction of the entire Tanzam

1/ IBRD/IDA now normally require loans and credits for highway projects to be based on completed detailed engineering since preliminary engineering has not proven a satisfactory basis for reliable cost estimates, as illustrated in this project. -3-

Highway between Dar-es-Salaam and Zambia was required. The revised project provided for detailed engineering of a 104 mi section of the Tanzam Highway, previously completed under the original project, to be upgraded to standards consistent with plans for the entire highway and for detailed engineering to be carried out on an additional 38 mi. Construction was deferred, however, until the Second Highway Project (Credit 142-TA and Loan 586-TA of 1969). On the Musoma-North Mara road, traffic had grown less than expected and revised projections indicated that improvement was not economically justified.!1

The Government had financed the preparation of feasibility studies for the Geita roads by the consultants William Halcrow and Partners (UK), Scott, Wilson, Kirkpatrick & Partners (UK), and the Economist Intelligence Unit (UK). Based on these studies and an IBRD/IDA economic mission in 1966, a scheme had been recommended for development of feeder roads in the Geita Peninsula. The revised project provided for detailed engineering of 110 mi of four secondary roads and 98 mi of tertiary roads.

Also, the IBRD/IDA 1966 economic mission had emphasized the need to strengthen RAD staff as quickly as possible to overcome shortages resulting from the exodus of expatriates. Therefore, a staffing and training program, conceived essentially by the very dynamic man holding the position of Permanent Secretary of Comworks at the time, had been included in the revised project. Under the program, a team of experts was to be appointed to RAD executive positions and was to train citizen counterparts designated by the Government. The experts' duties were to become increasingly advisory as the counterparts gained experience and took over responsibilities. IDA was to pay their travel costs and necessary supplements to Tanzanian salaries out of the credit.

Finally, the IBRD/IDA 1966 economic mission had recommended that annual road construction and maintenance expenditures should be about US$ 20 million for 1967 and 1968.

PROJECT IMPLEMENTATION

The actual cost of the revised project was US$ 22.2 million, or 3% less than estimated at appraisal. About half of the construction of the project roads had been completed at appraisal of the revised project, and the balance was completed satisfactorily by 1970 as scheduled (see Table 2). The closing date for the credit was deferred several times due to an outstanding contractor's claim, and the final disbursement was on July 25, 1973, two years behind schedule (see Table 3). About US$ 16.6 million was disbursed out of the two credits together, and about US$ 440,000 of Credit 115-TA was cancelled at closing. Most of the cancelled amount is attributable to unspent funds under the staffing program (salary supplements).

1/ The estimated economic rate of return was 7% based upon an expected ADT of 96 vehicles after road opening. Recent development in the area has been slow, as indicated by the current low ADT of 60 vehicles. -4-

Implementation of the road sections is discussed in Annex 1. Devia- tions from the estimated cost and completion date took place on several sections. As to cost, both overruns and underruns occurred. Overruns resulted on the Kolondoto-Bariadi, Geita Peninsula, Chalinze-Segera, and Mingoyo- Masasi roads. The largest increase, 33%, was on the Kolondoto-Bariadi road construction due to design changes, primarily construction of a bridge not previously included in the project. An overrun of 28% occurred on the Geita road engineering, 3% on the Chalinze-Segera road construction, and 2% on the Mingoyo-Masasi road construction. The reasons for these overruns are not known. Bituminous paving of the Mingoyo-Masasi road was to be carried out under the Third Highway Project,iCredit 265-TA of 1971), in which a substantial cost overrun has occurred.- Underruns also occurred, of 16% on the Mkumbura-Kisangiro road construction, 11% on the Dar-es-Salaam-Kibiti road construction, 6% on the Morogoro-Mahenge road engineering, and 2% on the Uyole-Itungi Port road construction. As to the completion date, the Uyole- Itungi road was delayed one year due to construction difficulties arising from the contractor's inexperience in working under local weather conditions and in dealing with the unexpected soiL conditions.

As regards the staffing and training program, with a view to reducing program costs and for socio-political reasons, the Government concentrated on recruitment of engineers from Asian countries, where salaries were com- parable to Tanzanian levels. In 1969 Comworks recruited 15 engineers from India, which continues to be the major source of RAD expatriate professional staff. Only in 1970, after two years of delay and with the project largely completed, the Government presented proposals to use credit funds to assist in financing four of eight planned professional posts. Ultimately, only salaries for two posts were augmented with IDA funds and only US$ 30,000 was spent of the US$ 450,000 allocated for staffing and training. The staffing and training program is discussed further under "Project Impact."

The Government did not fully comply with all covenants and supplementary agreements in the credit documents. In regard to the covenant that the entire road system would be adequately maintained, the Government has increased expenditures considerably (see Table 4), but maintenance remains inadequate in most areas. The recommendation of the IBRD/IDA 1966 economic mission for annual road construction and maintenance expenditures of about US$ 20 million in 1967 and 1968 was nearly reached in both years, and expenditures rose substantially further in following years. Road maintenance is expected to improve considerably when the program in the Highway Maintenance Project (Credit 507-TA of 1974) is implemented. Regarding the covenant that RAD appoint a senior engineer to oversee project execution, RAD delayed in

1/ The total cost of the Third Highway Project had been expected to be US$ 5.6 million, with a foreign exchange cost of US$ 3.6-million for which the credit had been provided. The total cost has increased considerably, however, to US$ 15.6 million with a US$ 10.9 million foreign exchange cost. The difference between the credit amount and increased foreign cost is being financed through a US$ 4.8 million loan from the African Develop- ment Bank, US$ 1.9 million remaining out of the Bank loan for the Second Highway Project, and Government funds. -5- making this appointment and IDA eventually stopped requesting compliance because RAD was implementing only IDA-financed projects and separate supervision of them was not necessary.

PROJECT IMPACT

Road Construction

The overall economic rate of return for the six roads under the revised project now is estimated at between 9% and 11%, based on recent traffic information (see Tables 5 and 6) which is of uncertain accuracy!' but the only thing available, and actual costs (see Table 2). This does not differ greatly from the overall 11% return estimated at appraisal. If the rate of return calculations were revised to reflect current etroleum prices, the estimated return would rise to between 10% and 127

The actual construction cost of the project roads as well as t appraisal and audit estimates of their rates of return are listed below:-

Actual Construction Rate of Return Ref. No. Cost Appraisal Audit (See Map) Road Section (US$ million) Estimate Estimates ------%------1 Chalinze-Segera 2.89 10 7 to 8

2 Mkumbura-Kisangiro 4.47 15 10 to 11

3 Mingoyo-Masasi 1.80 13 6 to 7

4 Uyole-Itungi Port 5.02 13 11 to 12

5 Kolondoto-Bariadi 4.24 9 -9 to +8a/

6 Dar-es-Salaam-Kibiti 3.04 8 12 to 15 21.46 a/ Traffic information is inconsistent, from a low ADT of 60 vehicles in 1973 to a high ADT of 208 vehicles in 1971

1/ Available traffic infomation does not provide consistent annual and seasonal sampling for the project roads constructed. The economic rates of return were calculated for both a high and a low estimated traffic growth rate between the date of road opening and 1973. Hence,the rates of return shown represent a range including both high and low estimates.

2/ Based on a 100%1 increase in petroleum prices, which yield a 5% increase in user benefits. 3/ Benefits are based on savings in vehicle operating costs, excluding the effect of recent petroleum price increases. Actual construction and supervision costs only were included in the calculations. Maintenance costs were excluded. - 6 -

The rates of return forthe.Chalinze-Segera and Mkumbura-Kisangiro roads are three to five percentage points lower than the appraisal estimate. On the Chalinze-Segera road, the differential between the rates of return resulted because the 3% cost overrun was not compensated by a -significant traffic increase over the appraisal estimate. In addition, the use of estimated actual annual traffic growth rates of 3.5% at audit compared with estimates of 5 to 6% at appraisal contributed to the lower rate of return for this road as well as the Mkumbura-Kisangiro road. On the Mkumbura- Kisangiro road, a differential between audit and appraisal rates of return resulted despite a 16% cost underrun because actual traffic was 28% less than the appraisal estimate.

The audit estimate of the rate of return on the Mingoyo-Masasi road is between 6 and 7%, as compared with the appraisal estimate of 13%. Traffic on the road has apparently been even lower than in 1967,.reportedly due to military restrictions, and this explains the variance between the audit and appraisal rates of return.

The 11 to 12% audit estimate of the rate of return for the Uyole- Itungi Port road is only slightly lower than the appraisal estimate of 13% although traffic did not grow as expected. The reason was the change of traffic composition -- trucks represented 74% of total traffic at audit versus 57% at appraisal.

According to varying data available, the rate of return on the Kolondoto-Bariadi road can only be estimated at between -9% and +8%, compared with +9% expected at appraisal. While traffic information for this road is conflicting, traffic appears to be lower than expected. The differential between the audit and appraisal rates of return is because the 33% cost increase was not compensated by a significant traffic increase over the appraisal estimate.

The audit estimate of the return for the Dar-es-Salaam-Kibiti road, between 12 and 15%, is considerably higher than the appraisal estimate of 8%. At audit, this road produced the highest return of the six project roads whereas at appraisal it was expected to produce the lowest return. This situa- tion resulted because the actual cost of road construction was 11% less than estimated at appraisal and actual traffic was 30% more than expected. The higher than expected traffic increase is attributable to-agricultural development in the region of the road, spurred by proximity to Dar-es-Salaam.

The above-cited estimates of actual rate of return make no allowance either for possible highway maintenance savings (which were included in benefits at appraisal) or for the effects of vehicle overloading, which is believed to be widespread throughout the country (although it is of course highest on the Tanzamw3ighway not included in this project). There is no satisfactory data on these aspects, but an analysis on the basis of the most -reasonable assump- tions possible suggests that allowance for them would not in fact make much difference to the figures presented. The Chalinze-Segera road for which the return cited above is 7-8% may be taken as an example. Inclusion in the benefits of the highway maintenance savings assumed at appraisal would raise -7-

the return to 8-9%, while a 30% upward adjustment of actual truck traffic figures (to allow generously for transport operating cost savings on the extra freight carried by estimated overloading of about this amount) would increase the return further, to some 10-11%. On the other hand overloading causes severe damage to highways and seriously shortens their lives. Assumption of a 15-year life, instead of the 20-year life used in the basic calculation (see Table 6), reduces this return again to 8-9%, while a 13- year life implies a 7-8% return, equal to the basic estimate shown in the table above. Mainly because the effects of vehicle overloading on highway life are so severe it would appear that adequate allowance for this phenomenon would, if anything, reduce the basic estimates of return given above rather than raise them.

Traffic Growth

Comparison of actual traff c with appraisal projections (see Table 5) indicates that generated traffic- after improvement was estimated accu- rately for the Chalinze-Segera road2/, was underestimated for Dar-es-Salaam- Kibiti road, .and was overestimated for the Mkumbura-Kisangiro, Mingoyo- Masasi, Uyole-Itungi Port (the three cases where it was expected in particularly substantial quantity) and Kolondoto-Bariadi roads. Factors contributing to overestimation of traffic were: high projections of generated traffic, lack of complementary investment in development of the regions surrounding the roads, unstable political conditions, low growth of the national economy and the vehicle fleet.

The first factor involved in overestimation appears to have been the relatively simple assumption made at reappraisal that a decrease in vehicle operating costs would be enough to induce substantial generated traffic. The original 1963 appraisal mission had included an agricultural specialist who had been pessimistic about the possibilities of the roads hav- ing any direct automatic impact in inducing agricultural development, and so no significant such benefits were projected. It is not clear whether the views of the Bank's agricultural experts were obtained on the substantially changed projections made at reappraisal.

Lack of complementary investment in development of the regions surrounding the roads, the source of generated traffic, was the second factor which contributed to overestimation. For instance, in the case of the Mingoyo- Masasi road, the potential of the surrounding region for agricultural development, particularly cashewnut production, was considered at reappraisal to lead to a considerable amount of generated traffic for the road. But investment in complementary development of the road and region was not assessed. Subsequently, it became evident that cashewnut production was not being stimulated by the

1/ Generated traffic, which is difficult to project, is defined here as that portion of traffic increase which would not occur if the road were not improved, i.e., the projected difference in traffic with and without the road improvements. 2/ Traffic projections are not substantiated in the appraisal report, therefore no comparison can be made between expected and actual conditions intervening in the growth of traffic. -8-

Government, and that, therefore, the expected generated traffic was not developing. Recently, however, the potential for increased cashewnut production has been the focus of a major Government program receiving Bank support under the Cashewnut Development Project (Loan 1014-TA of 1974). If improvement of the road and development of cashewnut production in the region had been complementary, the investment in the Mingoyo-Masasi road could have resulted in a higher economic return.

Another factor involved in overestimation of traffic was unstable political conditions. Relationships with bordering countries deteriorated during project implementation, resulting in discontinuation of transport on Malawi Lake and military restrictions on vehicle travel in many southern areas, where Mozambique freedom fighters were active. These problems hindered economic development in general and lowered traffic growth in particular on the Mingoyo-Masasi and Uyole-Itungi Port roads. Furthermore, economic integration with Kenya did not proceed as planned. Hence, trade between Kenya and Tanzania did not increase as much as expected, and traffic growth was reduced, particularly on the Mkumbura-Kisangiro road. On the Chalinze-Segera road. actual and estimated traffic are similar, but actual traffic would be lower if the road were not used for Zambian transit traffic. The impact of all these political events on traffic would have been difficult to predict at the time of appraisal.

The last factor which played a part in overestimation of traffic was the growth of the national economy and of the vehicle fleet at lower rates than expected at appraisal. GDP actually increased 4.4% per year at constant prices between 1964 and 1968, compared with the target growth of 5.9% envisioned in the First Five Year Plan (1964-69). Traffic growth rates on most project roads, as indicated in a recent studyL/, are 3 to 4% per year about equal to the actual growth of the agriculture sector between 1964 and 1969. At appraisal, traffic estimates had been based on fragmentary data, which indicated traffic growth rates of 7 to 8% per year. These rates were much higher than expected GDP growth rates of 5.9% per year.

Expansion of the vehicle fleet and trucking services was limited by Government import restrictions and transport regulations after the project had been designed. Consequently, the annual growth rate of the total vehicle fleet was only 3.5% per year between 1970 and 1972, less than half the growth rate between 1962 and 1970 (see Table 7).

Detailed Engineering

Detailed engineering for the Morogoro-Iringa road and the Geita roads was completed under the Revised First Highway Project. Construction followed for the Morogoro-Iringa road under the Second Highway Project (Loan 586/Cr. 142 of February 1969). The engineering standards for this road appear to be adequate and the actual construction cost is about equal to the estimate.

Feasibility studies for the Geita roads, prepared in 1966, concluded that construction was justified. Under the studies, design standards provided for all-weather roads 18 ft wide and for roads near Lake Victoria in

1/ Lyons and Associates, Economic and Engineering St4dy, Tanzania Highway Maintenance and Organization, 1973. -9 -

the northern part of the district to be passable during flooding. Specific- ally, this meant that extensive embankments would be constructed so that the roads would withstand a rise of the lake 10 ft above the 1964 peak level. Detailed engineering, prepared in 1968 and 1969, indicated that construction costs would be considerably higher than previously expected,so that priority ratings had to be established and design standards to be revised.

A research unit from the University of Dar-es-Salaam and the Planning Division of Comworks reviewed the detailed engineering for the Geita roads in 1970. It found that construction would be justified if design standards were reduced, e.g., the road width reduced to 12 ft, the embankment heights decreased, and some existing roads upgraded through minor construction, betterment, and improved maintenance in lieu of new construction. Therefore, the detailed engineering was revised and the upgrading was financed under the Third Highway Project (Credit 265-TA of 1971), currently being implemented. The revised detailed engineering is now being modified to reflect current costs and traffic conditions. In addition, the Govern- ment is considering whether to widen some roads to 18 ft, the original width used in the feasibility studies., because of traffic increases and soil conditions. The decision to include detailed engineering in the 1967 project seems to have been wise, even though the work had subsequently to be revised, because little or no feeder road construction experience had previously been available in the Geita Peninsula.

Staffing and Training

Although all funds provided under the revised project for RAD staffing and training were not used, a program eventually was implemented. However, recruitment of expatriate staff was delayed and training was minimal. Various factors contributed to recruitment delays. Instead of using the credit funds, the Government appears to have searched for technical assistance grants from various countries to support expatriate technical staff. Some technical assistance was obtained, but it did little to resolve staff shortages. Also, the Government recruited staff through established procedures, which were encumbered by bureaucratic delays; e.g., some applications were under consideration for 12 to 18 months. In addition, the major recruitment sources were considered to be the United Kingdom and northern Europe. But in the late 1960s, when colonial staff was leaving a number of African countries and when ample work was available elsewhere, many Europeans viewed employment in Tanzania as less attractive.

IDA noted the delays and lack of implementation of the program. It persisted in ascertaining the reasons and in bringing its findings to the attention of the Government through numerous personal interviews during supervision missions, letters, and cables. However, personnel changes in the Government in 1967 reportedly brought in individuals who were not sym- pathetic to the idea of paying staff salaries with credit funds; even if real costs to Tanzania were closer, the visible budgetary costs of employing people under the Bank's scheme were far higher than for bilateral technical assistance personnel -- partly because IDA only paid a salary supplement and partly because engineers under the IDA scheme were fully subject to the - 10 -

rather high Tanzanian tax levels.-/

In 1969, the Government became aware that staff vacancies were not being filled through technical assistance grants and established procedures. Therefore, it decided to recruit staff from Asian countries, where salary standards were lower than in Tanzania and economic conditions closer to those in Tanzania. A special mission went to India and recruited 15 engineers.

At this time, the staffing and training progran could have been modified to ensure that the objectives were fulfilled. For example, the use of credit funds could have been modified to allow financing of a portion of the foreign exchange costs involved in obtaining technical personnel, thus facilitating recruitment from all countries. Instead, the use of credit funds remained limited to augmenting salaries above Tanzanian levels to attract competent staff. While such assistance was necessary then to attract European and American technical personnel, it was not necessary to attract Asian technical personnel. The Government understood this difference, while IDA did not recognize it. Asian technical personnel were recruited more rapidly than Europeans and Americans. Also their salary levels were lower than those of Europeans and Americans, thus lowering overall salary costs for expatriate staff.

Staff shortages prevailed throughout the Revised First Highway Project despite the Government's belief, as indicated in the appraisal report, that by 1973 the supply of professionals who were Tanzanian citizens would be sufficient to provide trainees in all RAD functions. Table 8 shows that RAD's senior staffing wa2/substantially below approved establishment throughout the period 1964-74.- But it was able to build up its engineering staff by recruiting a number of recent college graduates in the 1968-71 period and training them for various positions. In 1971 and 1972, however, 15 of the 18 citizen engineers in the transport sector left because they perceived a lack of advancement opportunities within RAD for Tanzanians of Asian origin. Their departure resulted in a shortage of citizen engineers. Moreover the Government Establishment Branch's allotment of newly graduating engineers to RAD, which had been five annually between 1968 and 1971, decreased to three engineers in 1972, four in 1973, and one in 1974. Of these eight engineers, only four have remained at RAD.

A number of inspectors left RAD between 1965 and 1970, and inspectors were in critically short supply through 1971. From 1963 to 1972, the number of non-citizen inspectors decreased while the number of Tanzanian citizen inspectors increased.

1/ The tax problem appears to have been eliminated by a Government order of May 1974 greatly reducing the tax burden on expatriates hired for IDA- assisted projects.

2/ While the shortage of engineers appears to haxe increased in 1974, part of the reduction in staff authorized and employed between 1972 and 1974 reflects the partial decentralization of RAD into regional offices. - 11 -

The staff problem was aggravated by factors out of the purview of the proposed staffing and training efforts. Comworks had no systematic means of selecting staff for appropriate training or for evaluating their qualifications for upgrading; the engineers' effectiveness was limited by the lack of intermediate staff; and, most important, engineering salaries in RAD were not made competitive with salaries of private organizations and new para-state organizations.

Training under the project by expatriates, including those paid with credit funds, did not materialize as envisaged, principally because sufficient qualified Tanzanian counterpart staff were not available. In addition, much of the training effort between 1969 and 1971 was lost to RAD when Tanzanian engineers of Asian origin left. Further, the expatriates, who had been selected primarily for their technical expertise, encountered cultural and language barriers and were not successful in some of their training attempts. The two expatriates who were paid with credit funds were unsuccessful: one, who was technically qualified,did not have a qualified counterpart to train and the other did not have adequate training expertise. RAD takes the view that their two-year term was too short for them to con- tribute much, given the inevitable learning period, and that they were probably of too high a caliber.

Although the number of trained roads staff still is limited, the problem may be abating somewhat. This is because key positions are now held by citizen engineers and experts have been provided by the Canadian Inter- national Development Agency to help staff headquarters and field offices. In the future, graduates of the new engineering university in Dar-es-Salaam are expected to be available for the RAD. Three more years must pass, however, before the first class graduates. Those graduates will be in demand by many Government and private organizations and if the RAD is to succeed in recruiting them, it-will haveto offer competitive working conditions, particularly salary and advancement opportunities.

Training experience under the Revised First Highway Project provided the basis for subsequent training programs. Under the Second and Third High- way Projects, Tanzanian citizens were trained through on-the-job positions which were specified in construction and consulting contracts and through a highway construction and maintenance training facility which was established in Morogoro. Sufficient numbers of qualified citizens still are not available, however, and this deficiency is affecting highway projects supported with IBRD/IDA funds. Manpower requirements for highway projects in future years are currently being studied, and preliminary indications suggest that they will continue-to be larger than availabilities through 1978.

CONCLUSION

The Revised First Highway Project launched Tanzania on a new phase of road development. Through the project, IDA helped the country to improve the quality of its roads, including progressive upgrading of main roads from tracks to fully engineered facilities. IDA also encouraged the use of sound engineering data and economic criteria in project preparation, construction, and supervision as well as in preparation of the Government's 1969 Development Plan. Finally, IDA helped the country focus on its need for adequate road staff on immediate and long-term bases. - 12 -

However, available data indicate substantial discrepancies between the economic returns estimated at appraisal and at audit, because traffic was underestimated on one of the project roads (Dar-es-Salaam-Kibiti) and overestimated on four of the roads (Mkumbura-Kisangiro, Mingoyo-Masasi, Uyole-Itungi Port, and Kilondoto-Bariadi). Overestimation occurred primarily because of the appraisal assumption that decreased vehicle operating costs would be enough to induce substantial generated traffic as well as omission of an assessment of complementary development in the regions surrounding the roads, which were expected to generate traffic. Other factors were involved in the traffic overestimation -- unstable political conditions as well as low growth of the national economy and the vehicle fleet -- but they could not have been anticipated at appraisal.

IDA's objective in the staffing and training program was to ease RAD's transition to Africanization. This objective was not achieved, however, mainly because of the extreme shortage of Africans with the necessary basic qualifications. A small manpower analysis at the time of appraisal/reappraisal, or by special assistance of the Bank's education specialists, might have helped to bring out the extent of this absolute shortage, but it probably would not have been able to foresee the sudden departure of the Asians in 1971-72, and moreover a specifically phased program was not necessary to the concept of the Permanent Secretary responsible, who foresaw recruiting as many Tanzanian citizens as possible and holding the expatriates until they were trained. With his departure soon after the credit was signed, neither expatriates nor Africans were recruited in quite the way planned. The scheme appears to have been too dependent on the key man and perhaps not carefully enough prepared in respect of the tax angle for the foreign employees and RAD personnel policies. Yet the more basic problem was sheer shortage of African secondary school and university graduates for meeting all Tanzania's development needs.

Tanzanian authorities have expressed the view that IDA could have ensured that such local personnel as were available were given more responsibilities, without necessarily trying to accomplish whole works unaided. The Bank did leave rather open scope for this under the original project, as mentioned, but foreign consultants had to be brought in to take over the work on two out of the three roads involved. African engineers have sometimes been assigned as understudies to the consultants, but they appear to have sometimes found this an unsatisfactory arrangement and preferred, rather, that there should be pilot projects using Tanzanians at most levels.

A more gradual approach, perhaps stage construction or a reduction in the number of roads, might have allowed local personnel to better develop and use their capabilities and should have reduced the need for relatively costly outside consultants. Scope for greater use of stage construction would have been relatively limited in this case since most of the improvements were in fact from earth to gravel or gravel to paved, although the Mingoyo- Masasi and Kolondoto-Bariadi roads might in fact have been improved simply to 'better earth' at that time rather than to gravel. Reduction of the number of roads in the project could have been a sound approach insofar as economic returns on several of them are now estimated to be quite low (three roads, accounting for 42% of total construction investment under the project, - 13 - with returns fairly clearly below 10%). However, if this approach had been adopted, the Dar-es-Salaam-Kibiti road probably would have been the first to be deleted because it had the lowest economic rate of return at appraisal and was the one about which the 1968 appraisal report showed the most doubt. But this road, benefitting from the stimulating influence of the national capital on regional activity, now shows the highest return of the six project roads. Hence, while it is clear that the original 1964 project was too large, from both economic and staff development viewpoints, and that the 1968 revision suffered the same faults in lesser degree, it is difficult to see how the latter could have been better reduced at reappraisal in 1967 given the apparent paucity of data available at the time on complementary investment plans in the areas surrounding the roads. The course recommended by the Executive Director commenting in 1964, of thorough conjoint effort on road improvement and agricultural development, in unified or separate projects of limited size, might have been the only one to give more assured success.

Table 1

PROJECT PERFORMANCE AUDIT REPORT TANZANIA REVISED FIRST HIGHWAY PROJECT (CREDIT 115-TA)

Description of Original and Revised Projects Original Revised Ref. No. (see Man) and ComRonent TrPe of Work Project A. Detailed Engineering and Construction

1. Cha.inse-Segera Bituminous paving and 108 mi 108 mi widening of bridges

2. Mkumbara-Kisangiro Realignment, reconstruction, and bituminous paving 95 mi 95 mi

3. Mingoyo-Masasi Reconstruction and gravel surfacing 77 mi 77 mi

4. Uyole-Itungi Port Reconstruction and bituminous paving 71 mi 75 mi

5. Kolondoto-Bariadi Reconstruction and gravel surfacing 100 mi 100 mi

6. Dar-es-Salaam-Kibiti Reconstruction and bituminous paving 78 mi 78 mi

7. Musoma-North Mara Reconstruction and gravel surfacing 75 mi Omitted

8. Morogoro-Iringa Reconstruction, and bituminous paving 110 mi Construction omitted Total 74 m B. Detailed Engineering

1. Morogoro-Iringa section of the Tanzam Highway 104 mi 142 mi

2. Geita Peninsula Secondary Not in- and Tertiary Roads cluded 208 mi

C. Staffing and Training Program Not in- 8 expatriate cluded staff PROJECT PERFORMANCE AUDIT REPORT

TANZANIA REVISED FIRST HIGHWAY PROJECT (CREDIT 115-TA)

Completion Dates and Estimated and Actual Costs for the Project Components

Cost estimate during 1967 reappraisal original Actual cost as Ref. No. Completion Dates (inclilding contract a percentage of (see Map) Component Forecast Actual contingencies) amount Actual Cost estimated cost ------US$ million------

A. Detailed Engineering and Construction

1. Chalinze-Segera Dec. 1968 March 1969 2.81 2.08 2.89 103

2, Mkumbara-Kisangiro Sept. 1968 Oct. 1968 5.35 5.16 4.47 84

3. Mingoyo-Masasi June 1968 Oct. 1968 1.76 n.a. 1.80 102

4. Uyole-Itungi Port Dec. 1968 Jan. 1970 5.12 4.72 5.02 98

5. Kolondoto-Bariadi June 1971 Sept. 1970 3.20 4.03 4.24 133

6. Dar-es-Salaam-Kibiti July 1968 Aug. 1968 3.43 3.21 3.04 89

7. Morogoro-Iringa n.a. n.a. 0.50 n.a. 0.47 94

8. Geita Peninsula n.a. n,a,. 0.18 n.a. 0.23 128

B. Staffing and. Training

Staff salaries and incidental costs n.a. n.a. 0.45 n.a. 0.03 7

Total 22.80 22.19 97

IDA participation (75%) 17.00 16.56 97 IN Table 3

PROJECT PERFORMANCE AUDIT REPORT

TANZANIA REVISED FIRST HIGHWAY PROJECT (CREDIT 115-TA)

Estimated and Actual Disbursement Schedule

Actual as a Fiscal Cumulative Percent of Year Estimate Actual Estimated ------US$ million------

1964-67 7.0 6.6 94

1968 13.0 10.5 81

1969 15.0 14.7 98

1970 16.2 15.7 97

1971 17.0 15.8 93

1972 17.0 16.3 96

1973 17.0 16.6 97

1974 17.0 16.6 97 Table b

PROJECT PERFORMANCE AUDIT REPORT

TANZANIA REVISED FIRST HIGHWAY PROJECT (CREDIT 115-TA)

Central Government Expenditures on Highway Construction and Maintenance

Fiscal Year Construction/1 Maintenance Total

------T Sh million------

1962 24.o 30.0 5L.o

1963 20.0 28.0 48.0

196 20.0 30.0 50.0

1965 2h.0 28.0 52.0

1966 29.1 36.5 65.6

1967 73.4 51.3 124.7

1968 67.6 53.1 120.7 1969 178.7 67.L 246.1

1970 143.6 66.6 210.2

1971 127.8 50.8 178.6

1972 135.3 72.8 208.1

/1 Including foreign aid.

Source: ComTaorks, December 1973. PROJECT PERFORMANCE AUDIT REPORT

TANZANIAREVISED FIRST HIGIHAY PROJECT (CREDIT 115-TA)

Traffic and Unit Vehicle Cost Savings on the Project Roads

Appraisal Estimate LI Actual ADT in 1970 Savings per Normal Traffic Generated Traffic Total Traffic ADT in 1973 Vehicle Mile Ref. No. ADr in 1967 Without Improvement With Improvement With Improvement With Improvement (USQ) AUT in 1973 (see Ma) Road Section Rea- Light Total Reavy Light Total Heavy Light Total Heavy Lht Total Heavy Light Total Heavy kight Heavy Light Total

1. Chalinze-Segera 88 56 144 104 66 170 10 7 17 114 73 187 136 87 223 4.0 1.5 119 106 225

2. Mkumbura-Kisangiro 92 92 184 114 112 226 50 50 100 157 155 312 187 185 372 12.8 3.2 92 133 225

3. Mingoyo-Masesi 134 55 189 160 65 225 124 52 176 284 117 401 448 184 632 8.0 1.5 85 55 140

4. Uyole-Itungi Port 80 53 133 97 63 160 58 38 96 155 101 256 185 120 305 16.0 4.0 130 45 175

5. Kolondoto-Bariadi 74 49 123 86 56 142 22 14 36 108 70 178 170 110 280 8.0 1.5 36 24 60

6. Dar-ea-Salaam-Kibiti 60 26 86 69 29 98 31 13 44 100 42 142 113 47 160 12.8 2.5 114 86 200

Li IBRD/IDA, Appraisal of Revised Highway Project. Tanzania, 1968, Report TA-600s.

/2 Ministry of Comunications and Works, Tanzania, October 1974.

L3 This figure is based on tra Vic data, however, a November 1973 study by Hargisson Associates, Oldeani-Mwaza Feasibility Study, reporte traffic of about 208 vebicles per day in 1971. Tnble 6

PROJECT PERFORMANCE AUDIT REPORT

TANZANIA REVISED FIRST HIGHWAY PROJECT (CREDIT 115-TA)

Estimated Traffic Growth Rates on the Project Roads (% per year)

1973 Through End of Economic Ref. No. / Economic Life Life in (see Map) Road Section Alt. ALI Alt. B1-7 of the Project- Years

1. Chalinze-Segera 7.7 3.5 3.5 20

2. Mkumbura-Kisangiro 3.6 3.5 3.5 20

3. Ningoyo-Masasi 1.7 3.5 3.5 15

4. Uyole-Itungi Port 4.7 3.0 3.0 20

5. Kolondoto-Bariadi /3 /3 3.0 15

6. Dar-es-Salaam-Kibiti 15.1 3.5 3.5 20

/1 The figures presented for Alt. A correspond to the growth rate between the appraisal estimate of ADT for 1967 and the actual ADT reported by Com,works in 1973 (Table 5).

/2, The figures presented for Alt. B are taken from unpublished notes of the Eastern Africa Regional Office, Projects Department, Revised Traffic Estimates for Tanzania Roads- 1974.

Z2 Inconsistencies in ADTs Comworks reports 60 ADT in 1973 representing a -11.3% growth rate between 1967 and 1973; and Margisson Associates' Oldeani-Mwaza FeasibilitZ Study indicates a 208 ADT in 1971 representing a +11.1% growth rate between 1967 and 1973.

14 The figures presented are taken from Lyons Associates, Economic and Engineering Study, Tanzania Highway Maintenance and Organization, 1973. DROJECT DERFORMANCE AUDIT REDORT

TANZANIA REVISED FIRST HIGHWAY OROJECT (CREDIT 115-TA)

Vehicle Fleet, 1962-72

Typg of Vehicle 1962 1963 1964 1965 1966 1967 196( 1969 1970 1971 1972 Gw

A. Privatelrpowmed Cars 19,048 21,419 22,854 22,985 25,877 28,748 26,809 29,783 33,078 33,261 34,1W1 8.0 6.3 1.6

Light commercial vehicles 7,755 8,o46 8,454 8,557 9,884 11,213 10,648 12,205 14,171 14,280 14,784 6.3 9.4 2.1

Trucks 6,386 6,779 7,301 7,215 9,004 10,461 10,137 11,578 13,570 13,821 14,849 9.0 10.8 4*6

Bases 1,526 1,687 1,887 2,041 2,354 2,613 2,473 2,969 3,486 3,521 3,816 11.4 10.3 4.6

Motorcycles 4,495 4,783 5,221 5,415 6,449 7,293 6,970 7,989 9,306 9,396 10,072 9.4 9.6 4.0

other/2 2,316 3 3,803 3.971 4,557 5,79 4.434 5,939 7.096 7,151 7,921 18.4 11.7 5.4

Subtotal 41,526 45,815 49,520 50,184 58,125 66,120 61,471 70,463 80,707 81,430 85,583 7.6 8.5 3.0 /3 B. Government-oned 3,623 3,859 4.391 4,238 5,320 6 4,003 4,872 7,008 7,855 8.454 10.1 7.1 9.8

Total Vehicle Fleet 45.49 49,674 53 911 54.22 63,445 72,654 7 75335 87 715 89A285 2!!03 8.9 8.4 3.5

/1 Some discrepancies are evident in the figures abomn for this year. No explanation could be obt7ained. ftmbulances, bulldosers, cranes, fire engines, graders, tractors, and trailers. All types of vehicles.

fources: 1962-68: Tanzania, The Annual Economic Survey 1971. 1969-72: Central Registty of Motor Vehicles, December 1973. lable 8 PROJECT PERFOR4kNCE AUDIT REPORT

TANZANIA REVISED-FIRST HIGHWAY PROJECT (CREDIT 115-TA)

Staff Authorized and Employed in Roads and Aerodromes Division, Ministry of Communications and Works /1

Employed as Employed a Percentage Authorized Citizen Non-Citiien Total of Authorized 1963

Senior Engineers 13 1 11 12 92 Engineers 38 0 25 25 66 Inspectors 64 13 50 63 98 1964

Senior Engineers 8 0 5 5 63 Engineers 38 1 30 31 82 Inspectors 64 16 42 58 91 1965

Senior Engineers 8 0 4 50 Engineers 38 0 27 27 71 Inspectors 79 16 41 57 72

196&

Senior Engineers 9 0 2 2 22 Engineers 67 11 42 Sh 81 Inspectors 59 18 20 38 64 1970

Senior Engineers 12 0 4 4 33 Engineers 68 16 15 31 46 Inspectors 5h 26 10 36 67 1972

Senior Engineers 9 2 2 4 44 Engineers 68 23 25 8 70 Inspectors 76 38 7 45 59 197U

Senior Engineers 10 4 2 6 60 Engineers 53 71][ 8 15 28 Inspectors 60 n.a. n.a. n.a. n.a.

/1 Includes road technical staff employed at headquarters. Excludes road technical staff employed at regional offices, some of whom were formerly employed at headquarters, as well as aerodrome and non-technical staff. /2 The reduction in citizen staff between 1972 and 197U reflects staff appointments to regional offices in accordance with the national decentralization policy. In addition, during 1973, 10 engineers (4 citizens and 6 non-citizens) left Comiorks. ANNEX Page 1

PROJECT PERFORMANCE AUDIT REPORT

TANZANIA REVISED FIRST HIGHWAY PROJECT

Details about the Project Roads

1. Chalinze-Segera Road (108 mi). The project provided for detailed engineering, construction, and bituminous paving of this road, which forms part of the main road system radiating from the capital, Dar-es-Salaam (see map). RAD prepared the pavement design and supervised construction. Ingra (Yugoslavia) was awarded the construction contract in July 1965. From the beginning, the contractor's organization was weak, and work progressed slowly. Also, RAD's supervision was poor due to in- sufficient staff on site and lack of assistance from headquarters.

The first 30 mi of the road were constructed on a lime-stabilized base and paved with bitumen. Soon after construction, the first 10 mi failed and had to be repaired. Later, the remaining 20 mi deteriorated badly. At IDA's request, the consultants Scott, Wilson, Kirkpatrick & Partners (UK) were engaged in November 1966 to review pavement design and supervise construction. The consultants decided that, where necessary, stabilization by lime should be replaced by cement. The consultants also instructed the contractor to improve equipment and work methods and exercised tight control over execution. As a result, the new sections were of better quality than the previously constructed sections. Construction was completed in March 1969, slightly behind the scheduled completion date of December 1968.

Improvement cost US$ 2.89 million, representing a 3% overrun. The total cost per mile (US$ 27,000) was relatively low since the project was mainly for paving of an existing engineered gravel standard road. -

Although the road has been maintained, the first 30 mi and particularly, the first 15 mi, deteriorated considerably; the first twelve miles were repaved (two inches of Asphaltic Concrete) in February 1975. The deterioration of the road appears to have been due to three factors: (a) vehicle overloading, (b) subgrade deterioration, and (c) narrow verge. Vehicle overloading, a widespread problem throughout Tanzania, can be partially attributed to Zambia transit traffic, originating from Mombasa Port (this traffic will be reduced after the on-going expansion of the Dar-es-Salaam port facilities is completed). Subgrade problems ate evident, but the precise reasons for them have not been determined. Hence, they cannot be clearly related to deficiencies in design, construction, supervision, or maintenance. Before this project, the road had been improved from an earth track to an engineered gravel standard road. Hence, the original base and verge width were not part of the project. The narrow verge width has contributed to deterioration, and in some areas, lack of verge support has caused cracks along the edge of the road. - 2 -

2. Mkumbura-Kisangiro Road (95 mi). Under the project, detailed engineering, realignment, and bituminous paving were included for this road, which also forms part of the main road system continuing northeast from the Chalinze-Segera Road towards Arusha or Nairobi. The consultants Scott, Wilson, Kirkpatrick & Partners (UK) carried out detailed engineering and supervised construction by Stirling Astaldi, Ltd. (UK-Italy). The road was completed satisfactorily in October 1968, almost on schedule. The total cost was US$ 4.47 million (US$ 47,000 per mi), or 16% below the estimate. Little maintenance has been required on the road, and it is presently in good condition.

3. Mingoyo-Masasi Road (77 mi). The project provided for detailed engineering and reconstruction to gravel standard of a section of the southern route extending from the Indian Ocean to Lake Malawi. Both components were carried out by RAD. Initially, progress was slow. Field staff were not supported in project implementation, supplies and drawings were not provided promptly, and labor relations were not smooth. Construc- tion was eventually completed in 1968. The total cost was US$ 1.8 million (US$ 23,000 per mi), about 2% over the estimate.

Maintenance has been inadequate,and the road:has become badly corrugated. The crown has settled and some regravelling has been required. A small amount of hand maintenance has been carried out, but it has been largely ineffective.

A feasibility study for bituminous paving of the road has been carried out by the United Research, Inc. (USA) under USAID auspices. Paving is being financed by IDA under the Third Highway Project (Credit 265-TA), the African Development Bank, and the Government.

4. Uyole-Itungi Port Road (75 mi). Under the project, detailed engineering, reconstruction, and bituminous paving were included for the road from Uyole to the north end of Lake Malawi. At the Government'.s request, IDA also agreed to include improvement and bituminous paving of two substandard road sections of 2 mi each. Detailed engineering and construction supervision were completed by Italconsult (Italy). Ph. Holzmann (Germany) constructed the road, which was completed in 1970, one year behind schedule.

Execution suffered from several setbacks. Tenders were opened in January 1966, but the award was delayed until May and the order to proceed until August. Thus the 1966 dry season was practically lost for construction. At the outset, the contractor underestimated the difficulties of working in the rainy season, and did not schedule works according to seasonal variations. Work started on the most difficult section. which crosses a rugged mountain range with one of the wettest climates in Tanzania, when the rains were the heaviest. The contractor later rescheduled the works, expect- ing to benefit from forecast seasonal variations. But he could not fully resume work since the weather deviated from normal in 1967.

The contractor increased plant capacity and personnel and intro- duced shift work to cope with the added burden of difficult earthworks. Construction revealed bad soil conditions in the mountainous region, which -3- made the work unusually difficult. Soils consisted of soft volcanic rocks and decomposed debris soaked with water, even at the end of the dry season. These conditions required extraordinary efforts by the contractor to excavate and build.

The contractor felt these efforts were not reflected in the design and tender documents, and could not have reasonably been anticipated by him when bidding. He considered the additional costs as beyond normal contrac- tual risks and claimed compensation. The consultants rejected the claim, and the contractor requested arbitration. The Government asked the consultants for a detailed analysis of a reasonable claim, based on conditions during the works. In a May 1970 letter, the consultants estimated that compensation of about US$ 600,000 might be awarded to the contractor. This was unacceptable to the contractor. Arbitration was finally concluded in 1973, after many delays, and the same amount estimated by the consultants was awarded. These delays were the major cause of deferring the closing date from December 1971 to July 1973. IDA's involvement in the development and resolution of the claim was minimal, not more than urging the Government to resolve the matter.

The contractor's original estimate (before submitting his claim) was US$ 4.24 million, which appears to have been low. His lack of local experience (this contract was his first job in Tanzania) and intense interest in obtaining the contract perhaps contributed to the low initial estimate. At the time of contract award, both the Government and IDA thought that the bid was somewhat low. But the bid was accepted because the contractor had been prequalified and his low bid should have been realistic. Subsequent additional payment claims brought the actual cost almost up to estimates.

The cost of improving this section was US$ 5.02 million, or 2% below estimates. The total cost per mile was US$ 67,000, representing the most expensive component in this project.

5. Kolondoto-Bariadi Road (100 mi). The project provided for detailed engineering and reconstruction to gravel standard of this feeder road in the vicinity of Lake Victoria. Detailed engineering was prepared by the consultants Ingeroute (France), who also supervised construction. A change in the original design, approved by IDA, added the construction of a bridge over the draft at Bariadi. Engineering and construction were satisfactorily completed ahead of schedule in September 1970 on both the road and the bridge.

The actual cost of the road, including the bridge, was US$ 4.24 million (US$ 42,000 per mile), or 33% over the estimate. The increase was the result of design changes, primarily the bridge construction. The actual cost was only 5% over contract tender, which included all design changes.

6. Dar-es-Salaam-Kibiti Road (78 mi). The project covered detailed engineering, reconstruction and bituminous paving of this road -4-

south of the capital, serving an area of established agriculture. Detailed engineering was completed by RAD, which also supervised the construction undertaken by Ingra (Yugoslavia). Earthwork was subcontracted to Stirling Astaldi. The road was satisfactorily completed by 1968, approximately on schedule. The actual cost was US$ 3.04 million (US$ 39,000 per mile), 11% below estimated cost.

The first half of this road near Dar-es-Salaam has deteriorated and required considerable patching. Conditions which contributed to deterioration are not precisely known, but subgrade problems appear to have occurred. These problems may have been compounded by vehicle overloading. プ