–Georgia Border Regional Road (M6 –Bagratashen) Improvement Project (RRP ARM 49244)

ECONOMIC AND FINANCIAL ANALYSIS

A. General

1. The proposed project will rehabilitate and improve 91 kilometers (km) of the two-lane M6 highway in northern Armenia from Vanadzor through Alaverdi to the border with Georgia. The Asian Development Bank (ADB) and the European Investment Bank (EIB) are cofinancing the project. For economic evaluation purposes, the project comprises both sections.

2. The M6 is of import both nationally and locally. This route handles 80% of all Armenia’s imports from Georgia by land and 45% of all Armenia’s road imports (2011–2013 data).1 The M6 is a vital transport link for the settlements that lie along the route. The M6 has an important role in catering to tourism traffic in the region. Improved accessibility from Yerevan and Tbilisi will improve the prospects for tourism development in the region. However, the M6 is ill-equipped to support these functions. The road is hazardous, averaging six fatalities a year during 2012–2014. It runs through the Debed gorge, which is also occupied by the Yerevan–Tbilisi railway line. This has resulted in poor vertical and horizontal alignments. Sight distances are poor, especially through its three tunnels; rock falls are frequent; proper road marking and delineation are lacking; and guardrails are inadequate.

3. In the absence of the project (the without-project scenario), the M6 is expected to remain in poor condition, with maintenance largely confined to sealing cracks and patching potholes. Under the with-project scenario, most of the road will be reconstructed, safety features will be improved, and the tunnels will be brought up to modern standards. The improved pavement will have an initial international roughness index (IRI) of 2.0. Average travel speed on the road is expected to increase from 48 km per hour to 54 km per hour . The selection of the project scenario by EIB followed an analysis of alternative alignments and was the option with the lowest initial cost and highest economic internal rate of return (EIRR).

B. Demand Estimate

4. Estimates of traffic demand used in the economic analysis are based on traffic counts undertaken by a consultant engaged by EIB to undertake a feasibility study. The study undertook four 6-day classified counts on the project road in April 2014 (one 24-hour count plus five 12-hour counts). These counts are consistent with historic data (including those derived from one automatic traffic counter) held by Armenian Road Directorate (ARD) and traffic survey conducted by ARD in 2015. Table 1 shows average traffic flows on each road section, and the fleet composition. In 2014, the weighted annual average daily traffic was 2,840 vehicles per day. Most freight traffic ends in central Armenia and Georgia. Local trips are generated by Vanadzor and Alaverdi, and the local market close to the border. There is also some tourism-related traffic. Hourly traffic data indicate a fairly uniform traffic flow pattern, with two-thirds of daily traffic during daylight hours.

Table 1: Baseline Demand, 2014 ≥4-Axle Rigid AADT Mini- 2-Axle 3-Axle and Truck- Vehicle/ Section Cars Buses Bus Trucks Trucks Trailers day Fleet: 76% 9% 2% 6% 2% 5% 1. Vanadzor Urban–M8 2,253 275 55 172 65 158 2,979

1 European Investment Bank. 2015. Feasibility Study, EIA and Detailed Engineering Design of the Rehabilitation and Improvement of M6 Vanadzor–Georgian Border Interstate Road. Luxembourg. 2

≥4-Axle Rigid AADT Mini- 2-Axle 3-Axle and Truck- Vehicle/ Section Cars Buses Bus Trucks Trucks Trailers day 2. M8–Vahagnadzor 2,253 275 55 172 65 158 2,979 3. Vahagnadzor– 2,253 275 55 172 65 158 2,979 4. Dsegh–Alaverdi 2,289 279 56 175 66 161 3,026 5. Alaverdi– 2,289 279 56 175 66 161 3,026 6. Shnogh–Karkop 1,832 223 45 140 53 129 2,422 7. Karkop–Bagratashen 2,201 268 54 168 64 155 2,910 AADT = Average Annual Daily Traffic . Source: European Investment Bank (EIB) feasibility study (footnote 1).

5. From 2009 to 2014, traffic growth on the M6 averaged 4% per year, but with considerable variability. At least in the short term, this growth rate is expected to continue, driven in part by economic growth of 2%–3% and demand for imports and exports through Bagratashen, which grew 7% annually in tonnage from 2010 to 2014.

6. There are two factors affecting traffic growth in the longer term. First, trade with Iran and Georgia may increase, which would raise transport demand. Second, some traffic may divert to Bagratashen from other border crossing points. Based on the various influences on traffic growth, the growth rate is assumed to drop to 3% from 2022. A growth rate of 3% implies an income elasticity of 0.9 with respect to forecast GDP growth in 2020. This is a conservative assumption, as transport demand is generally expected to display an income elasticity of at least unity. The effect of diversions to Bavra was assessed separately as part of sensitivity testing (para. 22). Significant congestion was not observed under this traffic growth scenario, even at the end of the evaluation period.

C. Economic Costs

7. The economic costs of the project comprise (i) capital investment, which includes civil works, land acquisition, and resettlement, as well as consulting services for construction supervision and social safeguard management; and (ii) road maintenance. Costs related to taxes, duties, and financing charges during implementation have been excluded. Table 2 gives a breakdown of the investment costs. Construction was assumed to take place over a 3-year period from 2017.

8. Financial costs were converted to economic costs in line with ADB guidelines.2 All predicted project costs and benefits are measured in late 2015 and early 2016 economic prices expressed in United States (US) dollar. Traded goods are measured at world prices and non-traded inputs at domestic prices less indirect taxes multiplied by a standard conversion factor (SCF) estimated at 0.98.3 A shadow wage rate factor (SWRF) of 0.84 was estimated and applied to unskilled labor used in road construction. A SWRF of 1.0 was applied to skilled and professional labor. Land and resettlement costs were based on the EIB feasibility study, which also considered compensation for crop yields lost, and have been adjusted by the SCF. A residual value equivalent to 20% of the investment cost—estimated by applying the straight- line depreciation method to individual project items based on assumed lifespans—has been included in the economic analysis.

2 ADB. 1997. Guidelines for the Economic Analysis of Projects. Manila. 3 Using the ADB simplified method based on merchandise imports of $3.892 billion, exports of $1.116 billion, and taxes on trade of $88 million (averages in current US dollar for 2009–2012 from World Bank data). 4 An approximation based on the ratio of rural and urban incomes from all sources for 2014 (Armenia National Statistical Service: Household Income, Expenditures, and Basic Food Consumption). Armenia–Georgia Border Regional Road (M6 Vanadzor–Bagratashen) Improvement Project (RRP ARM 49244)

9. The financial cost of the project at 2015 prices was estimated at $109 million. This was converted to an economic price of $89 million by (i) deducting the value-added tax, (ii) converting to US dollar at the February 2016 official exchange rate of AMD495=$1, (iii) applying the SCF (para. 10), (iv) applying the SWRF to an estimated 5% of construction costs, and (v) doubling physical contingencies from 5% to 10%. The weighted average economic cost per km was $0.93 million. The costs per km in Table 2 were based on earlier estimates for each road section.

Table 2: Investment Costs Economic Cost Section Intervention ($ million/km) 1. Vanadzor Urban–M8 (6.7 km) Mill and Replace 0.33 2. M8–Vahagnadzor (13.5 km) Reconstruction 1.05 3. Vahagnadzor–Dsegh (13.4 km) Reconstruction and Tunnel works 1.42 4. Dsegh–Alaverdi (17.1 km) Reconstruction 1.38 5. Alaverdi–Shnogh (2.9 km) Reconstruction 0.75 6. Shnogh–Karkop (22.7 km) Reconstruction 1.05 7. Karkop–Bagratashen (13.9 km) Mill and replace 0.26 Total 0.98 km = kilometer. Source: Asian Development Bank (ADB) estimates based on total financial cost and section costs from EIB.

10. Identical maintenance regimes were assumed for both the with-project and without- project cases. Routine winter and summer maintenance included pothole patching and crack sealing, as well as snow clearance and drain cleaning, at an average financial cost of about $4,600/km. Periodic maintenance comprised a 40 millimeter overlay at a unit financial cost of $29 per square meter applied from 2025 and triggered by an IRI≥10.

D. Economic Benefits

11. The main quantifiable economic benefits are vehicle operating cost (VOC) savings, travel time savings, and crash cost savings. Analyses were undertaken on the seven homogeneous sections listed in Table 1. Section geometrical and pavement characteristics were taken from the EIB feasibility study (footnote 1). As the scope for generated traffic was considered insignificant, benefits were confined to normal traffic.

12. Vehicle operating cost savings. The shadow price of gasoline was estimated at $0.53 per liter. This reflects the expected average price of crude oil to 2020 (from the January 2016 World Bank Quarterly Commodity Outlook) and the proportion of vehicles using compressed natural gas. The shadow price of diesel was estimated at $0.50 per liter. Other salient vehicle characteristics and costs are shown in &able 3.

Table 3: M6 Vehicle Fleet Item Unit Cars, SUVs Mini- Bus 2-Axle 3-Axle ≥4-Axle Rigid etc. Buses Trucks Trucks and Truck– Trailers Axles No 2 2 2 2 3 5 Km/year km 10,000 80,000 100,000 100,000 100,000 100,000 Service life Years 15 10 10 8 8 8 No. of passengers No 3 12 40 1 0 0 4

Item Unit Cars, SUVs Mini- Bus 2-Axle 3-Axle ≥4-Axle Rigid etc. Buses Trucks Trucks and Truck– Trailers Operating weight tonnes 1.4 3.2 15 13 20 30 Vehicle cost $ 23,000 18,000 43,000 21,000 40,000 90,000 Tire cost $ 125 80 330 170 330 330 Maintenance labor $/h 1.5 1.0 1.5 1.5 1.5 1.5 Crew $/h 0 1.2 1.2 1.2 1.2 2.4 Passenger working time $/h 3 1.5 1.0 1.0 Passenger nonworking time $/h 0.75 0.38 0.25 0.25 % of work-related trips % 75 50 50 50 h = hour, km = kilometer Source: ADB estimates.

13. The fleet average VOC components against IRI are shown in Table 4. The most significant item is capital, i.e., the equivalent annual cost of a vehicle, which is calculated with the Highway Development and Management model (HDM-4) using the “optimal life” approach that links vehicle life to IRI.

Table 4: Vehicle Operating Cost Components ($/vehicle-km) IRI, m/km Fuel Spares Capitala Othersb Total 2.5 0.074 0.054 0.118 0.036 0.28 4.0 0.074 0.061 0.119 0.036 0.29 5.0 0.075 0.066 0.124 0.037 0.30 7.0 0.075 0.075 0.144 0.038 0.33 9.0 0.077 0.084 0.167 0.039 0.37 10.0 0.078 0.089 0.179 0.040 0.39 IRI = international roughness index, m = meter, km = kilometer. Notes: (a) equivalent annual cost of vehicle, computed using the optimal life approach, and (b) includes maintenance, crew costs, tires, lubricating oil, and overhead. Source: Asian Development Bank estimates.

14. VOC and travel time savings across the entire fleet during the evaluation period are summarized in Table 5.

Table 5: Vehicle Operating Cost and Travel Time Savings Without project With project Saving IRI $/veh-km IRI $/veh-km $/veh-km

VOC 8.4a 0.34 4.1a 0.30 0.04 Travel time savings 0.14 0.12 0.02 IRI = international roughness index, km = kilometer. veh = vehicle, VOC = vehicle operating cost a average over evaluation period. Source: Asian Development Bank estimates.

15. Travel time cost savings. Travel time savings have been identified based upon the vehicle speed relationships included in HDM-4. These identify the number of minutes saved for each vehicle trip. These benefits have been monetized by applying values of time estimated for different categories of road users, differentiating between working and nonworking time. The increase in average speed over the evaluation period is modest at just under 6 km/hour, leading to a travel time savings of about 12 minutes. Based on recent Armenian wages, working Armenia–Georgia Border Regional Road (M6 Vanadzor–Bagratashen) Improvement Project (RRP ARM 49244)

time is given a shadow price of between $1/hour (bus passengers) and $3/hour (car passengers). Nonworking time is valued at 25% of working time. As a result, time savings only account for 21% of project benefits. Reducing the inventory costs of goods in transit also has a travel time benefit, but its magnitude is highly uncertain and depends on details concerning, for example, the time sensitivity of goods carried. As such, these benefits are not included in the economic analysis.

16. Crash cost savings. During 2012–2014, an annual average of six deaths and 56 injuries were recorded along the M6.5 The economic evaluation assumed that the number of deaths is halved as a result of safety investments. No reduction in injuries was assumed. The value of a fatality is estimated at 70 times GDP per head in accordance with methodology suggested by the International Road Assessment Programme (iRAP) methodology.6 According to the World Bank, GDP per head in 2014 was $3,870, the value of a fatality is estimated at $0.27 million. GDP per head was assumed to increase at 2% annually at constant prices. Crash cost saving benefits amount to 5% of base case benefits. About 10% of VOC and travel time savings will accrue to the poor, comprising (i) all bus travel time savings ($2.5 million), (ii) 20% of road user cost savings accruing to two-axle trucks ($0.67 million), and (iii) 10% of car road user cost savings ($7.8 million).

E. Results of Economic Analysis

17. An economic assessment of the project was carried out using the standard appraisal methodology. This methodology compares the incremental benefits of reductions in VOCs, travel times, and crash costs resulting from the road rehabilitation with the initial investment costs and changes in operation and maintenance costs over a 23-year appraisal period (3 years implementation and 20 years operation). Table 6 shows the results of the economic analysis expressed in terms of the following key economic indicators: benefit-to-cost ratio, EIRR, and net present value at a 12% discount rate. The results are presented using the world price numeraire. The results indicate that the project is economically viable, with a benefit-to-cost ratio of 1.7, an EIRR of 20.9%, and a net present value of $48 million.

18. Table 6 also shows the stream of costs and benefits. Incremental costs are negative in years when the cost of overlays (considered as an investment cost) and maintenance of the improved road are less than the costs that would have been incurred without the improvements. Main project benefits start in 2019; the small benefits shown in 2018 arise from works on sections 1 and 7, which finish a year earlier than those on other sections.

Table 6: Costs and Benefits Streams (2015 world prices, $ million) Year Incremental Costs Incremental Benefits Total Net Investment Maintenance VOC Time Safety Benefits Benefits 2016 0.0 0.0 0.0 2017 47.2 (0.4) (46.8) 2018 41.4 0.0 2.3 0.4 0.0 2.7 (38.7) 2019 0.0 (0.3) 9.6 2.1 0.9 12.6 12.9 2020 0.0 0.0 11.7 2.8 0.9 15.5 15.5 2021 0.0 (0.4) 14.3 3.8 0.9 19.1 19.4 2022 0.0 0.0 16.8 5.0 1.0 22.7 22.7 2023 0.0 (0.1) 19.7 6.5 1.0 27.2 27.3

5 Lotti. 2015. Feasibility Study, EIA, and Detailed Engineering Design of the Rehabilitation and Improvement of M6 Vanadzor–Georgian Border Interstate Road: Road Safety Audit. Rome. 6 iRAP. 2008. The True Cost of Road Crashes. London. 6

Year Incremental Costs Incremental Benefits Total Net Investment Maintenance VOC Time Safety Benefits Benefits 2024 0.0 0.0 22.6 8.1 1.0 31.7 31.7 2025 (11.8) 0.1 24.9 9.5 1.0 35.3 47.0 2026 (9.9) 0.0 15.8 4.3 1.0 21.2 31.1 2027 (1.8) 0.1 9.2 2.1 1.1 12.4 14.1 2028 0.0 0.0 8.7 1.9 1.1 11.7 11.7 2029 1.9 0.2 9.5 2.2 1.1 12.8 10.6 2030 (3.8) 0.0 11.4 2.8 1.1 15.3 19.1 2031 (6.1) 0.2 9.7 2.5 1.1 13.3 19.1 2032 1.8 0.0 4.3 1.0 1.2 6.5 4.6 2033 1.9 0.2 6.4 1.7 1.2 9.3 7.1 2034 (1.8) 0.0 8.1 2.2 1.2 11.4 13.3 2035 0.0 0.3 6.9 1.8 1.2 10.0 9.7 2036 0.0 0.0 7.6 2.0 1.3 10.9 10.9 2037 1.8 0.2 8.1 2.2 1.3 11.6 9.5 2038 (20.4) 0.0 10.9 3.3 1.3 15.5 35.9 NPV 47.6

EIRR 20.9% BCR 1.7 ( ) = negative. VOC = vehicle operating cost, NPV = net present value, EIRR = economic internal rate of return BCR = benefit-to-cost ratio Source: Asian Development Bank estimates.

19. Sensitivity tests and calculations of switching values were carried out to determine the effect of variations in key input parameters. Table 7 shows switching values of 174% with respect to construction costs and 45% with respect to benefits, meaning that the project would still be economically efficient if construction costs were to rise by 74% or the benefits were to fall to 55% of base case values. Project viability is also robust in the face of traffic diversion to Bavra from 2022 and the absence of crash cost benefits. Similarly, a sensitivity test was conducted on a scenario involving larger traffic disruption than anticipated during the construction years. This involves assigning a disbenefit to each of the construction years equal to half the time saving benefit immediately after opening. Again, the project remains economically viable. In summary, the economic analysis was undertaken in line with ADB guidelines. The project yields an EIRR above the 12% threshold and is therefore considered economically viable.

Table 7: Sensitivity Analysis Case EIRR (%) NPV, $ million Switching Value (%) Base case 20.9 47.6 Cost +20% 17.6 34.8 174 Benefits –20% 17.0 26.6 55 Cost+20% and benefits –20% 14.2 13.7 Traffic diversion to Bavra from 2022a 19.2 36.3 No crash cost benefits 20.0 41.6 No periodic maintenance 24.7 99.1 Traffic disruption during construction 20.4 45.6 EIRR = economic internal rate of return, NPV = net present value. Notes: (a) in line with estimates made as part of the North-South Road Corridor Development Program, growth is reduced to allow for a potential 700 vehicles/day diverted to Bavra from 2022 subject to ramp up over 3 years. Source: ADB estimates. Armenia–Georgia Border Regional Road (M6 Vanadzor–Bagratashen) Improvement Project (RRP ARM 49244)

F. Financial Analysis

20. The project road will not generate revenue from tolls. Therefore, the financial analysis focuses on financial sustainability. The project is expected to reduce M6 periodic and routine maintenance requirements considerably (Table 8). The routine unit maintenance requirements with the project are close to those budgeted by Ministry of Transport and Communications. The with-project periodic maintenance, when averaged over the evaluation period, is considerably greater than the current Ministry of Transport and Communications allocation, but periodic maintenance interventions will not be needed until 2029 at the earliest.

21. Under the Infrastructure Sustainability Support Program, Phase 2,7 ADB is helping the Government of Armenia take proactive steps to increase the budget allocation for road maintenance and increase its effectiveness. Such efforts will help further ensure the availability of funding for road maintenance in the years to come, including to the project road.

Table 8: Project Maintenance Cost Implications ($ 2015 prices) Without Project With Project ($/km/year) ($/km/year) Periodic maintenance 30,000 5,500 Routine maintenance 6,000 4,600 Km = kilometer Source: Asian Development Bank estimates.

7 ADB. 2015. Technical Assistance to Armenia for the Infrastructure Sustainability Support Program, Phase 2. Manila (TA 9065-ARM).