References

• , historical project pipelines, and political visibility rather than systematic analysis of need and impact. Collier and Hoeffler (2004) demonstrated econometrically that infrastructure investment — particularly roads — is among the most effective mechanisms for post-conflict economic recovery, with each 10% increase in road network passability associated with a 2.3% reduction in the probability of conflict recurrence. This finding underscores the strategic significance of rigorous rehabilitation prioritization as a tool not only for economic development but for conflict prevention.
• Specifically, in South Sudan, Christensen and Harild (2009) documented that road inaccessibility was the single most cited constraint to return by internally displaced persons (IDPs) in camp surveys across Jonglei and Upper Nile States, with 74% of respondents indicating they would return to their home counties if road access were restored. These findings highlight the critical humanitarian dimension of road rehabilitation prioritization in the South Sudan context. However, the authors noted the absence of a formal prioritization framework, with rehabilitation decisions at the time being driven largely by road length, donor interest, and political lobbying rather than multi-dimensional need assessments.
• The AHP of Saaty (1980) structures a decision problem into a hierarchy of goal, criteria, sub-criteria, and alternatives. Relative importance weights are elicited through n(n−1)/2 pairwise comparisons on a 1–9 ratio scale, and the resulting comparison matrix is checked for consistency using the Consistency Ratio (CR), which should not exceed 0.10 for the results to be considered acceptably consistent. Global priority weights are derived by geometric mean aggregation of individual expert judgements in group AHP applications (Dyer and Forman, 1992).
• TOPSIS (Hwang and Yoon, 1981) identifies the positive ideal solution (PIS) as the alternative that maximises benefit criteria and minimises cost criteria, and the negative ideal solution (NIS) as its opposite. Each alternative is then ranked by its closeness coefficient C_i, defined as the ratio of its distance from the NIS to the sum of its distances from both ideal solutions, with higher C_i indicating greater preference. The method accommodates both quantitative and normalised qualitative criteria and is computationally efficient for moderate numbers of alternatives and criteria, making it well-suited to the scale of the present application.
• South Sudan's classified road network comprises approximately 16,500 km of primary (national), secondary, and tertiary roads, managed respectively by the Ministry of Roads and Bridges, ten State highway authorities, and county governments. The network is structured around two north–south spines (the eastern N-8 Juba–Malakal corridor and the western Juba–Wau–Aweil corridor) and several east–west links connecting the oil-producing Unity and Upper Nile States. The majority of the network is unpaved, with surface types ranging from graded earth (45%) to gravel (38%) to bituminous surface treatment (13%) and asphalt concrete (4%) (MoRB, 2022). The road condition index (RCI), measured on a 0–100 scale consistent with the Road Note 9 methodology (TRL, 2004), averages 31 across the classified network, with 68% of sampled sections recording RCI below 40 (poor) and 29% below 20 (very poor / impassable).
• Twelve candidate road segments were selected for analysis in consultation with the Ministry of Roads and Bridges and the United Nations Mission in South Sudan (UNMISS) Civil Affairs Division, based on the following criteria: (i) inclusion in the South Sudan National Transport Master Plan 2023–2035 (MoRB, 2023) as a priority rehabilitation project; (ii) availability of sufficient condition and traffic data for quantitative assessment; and (iii) geographic spread across all major administrative regions. Table 1 provides a summary of the twelve segments, their total lengths, and key characteristics.
• MoRB road surveys 2022–23
• Population within 10 km buffer of road segment (WorldPop 2023)
• A structured questionnaire presenting 21 pairwise comparisons [n(n−1)/2 = 7×6/2 = 21] on Saaty's 1–9 integer scale was administered to 34 subject matter experts comprising 14 senior civil/transport engineers from MoRB and consulting firms, 8 humanitarian logistics specialists from UN agencies and NGOs, 7 academic researchers in civil engineering and economics, and 5 government planning officials. Individual expert matrices were aggregated using the geometric mean method recommended by Dyer and Forman (1992):
• RI = average Random Index for n=7 criteria = 1.32 (Saaty, 1980)
• This is equivalent to the classic 0–1 Knapsack Problem and was solved using the branch-and-bound algorithm implemented in Python's PuLP library (Mitchell et al., 2011). The estimated rehabilitation budget requirements B_i for each segment were derived from preliminary engineering cost estimates prepared by MoRB (2023), adjusted for South Sudan unit cost factors from the UNOPS infrastructure cost database, and are presented alongside the ILP solution in Table 4.
• Table 4 presents the ILP solution for budget levels of USD 50 million, USD 75 million, and USD 100 million, identifying the selected segments and the corresponding total TOPSIS benefit score. At the USD 75 million budget level — the baseline scenario consistent with projected donor funding under the South Sudan Transitional Development Assistance Framework 2025–2028 — the optimal portfolio comprises segments RS-01, RS-03, RS-04, RS-05, RS-09, and RS-12, with a total rehabilitation cost of USD 73.8 million and a combined benefit score of 4.411.
• The exclusion of RS-02 (Bor–Malakal) from the USD 75 million optimal portfolio despite its second-highest TOPSIS score illustrates the important distinction between TOPSIS ranking and portfolio optimisation. RS-02's estimated rehabilitation cost of USD 31.2 million is disproportionate relative to its TOPSIS score when compared to the alternative of funding RS-04, RS-05, and RS-09 simultaneously — three segments with a combined cost of USD 27.8 million and a combined benefit score of 2.044 versus RS-02's single contribution of 0.793. This result highlights the value of combining TOPSIS ranking with ILP optimisation rather than simply funding segments in rank order, a point corroborated by Farhan and Murray (2008) in the Sub-Saharan context.
• A limitation of the present study is its reliance on a single round of expert elicitation for AHP weighting. While the CR of 0.043 confirms adequate consistency in the aggregated matrix, the weights reflect expert judgement as of early 2024 and may not capture evolving political priorities, new security dynamics, or shifts in donor funding commitments. An adaptive version of the framework — with annual weight updates and data refreshes — is recommended for operational implementation. Additionally, the decision matrix was constructed at segment level, with single representative values for criteria such as AADT and population served; spatial heterogeneity within long segments (e.g., RS-03 at 630 km) could be better captured by segmental disaggregation, which is recommended for the detailed appraisal stage following initial prioritization.