half a border is no border

Problem statement

A land border crossing is two facilities run by two governments, and a truck is not through until both have released it. The CAREC corridor-monitoring programme — which since 2009 has had drivers from freight associations in eleven Central and West Asian countries log every stop on real shipments (2,420 samples in 2023) — keeps finding that the slowest crossings in the region are ones where a single side has been modernised. At Tsiteli Khidi (Georgia)–Krasnyi Most (Azerbaijan), Georgian customs runs risk-based, one-stop clearance in which "drivers spent 5–10 min in clearance," yet the outbound crossing averaged 54.9 hours in 2023 (up from 24.2 in 2022) because "cleared trucks on the Georgia side cannot be released until the Azerbaijani side clear the trucks inside Krasnyi Most," and because drivers must wait there for scarce bilateral road permits to enter Azerbaijan and Kazakhstan. At Torkham, ADB is financing an expansion of the Pakistani side to six lanes while "the Afghanistan side remains limited to one lane," constrained by a single bridge — the report itself warns "the throughput mismatch could lead to sub-optimal performance." At Dostyk and Nur Zholy (Kazakhstan, opposite China) the mismatch runs the other way: "the infrastructure and materials handling equipment are more limited here than in the Chinese BCPs," and inbound crossings averaged 22.2 and 16.5 hours. The unsolved problem is that border-post investment, procedure reform and permit policy are decided nationally, one side at a time, while the crossing's performance is set by the slower of the coupled pair.

Why this matters

The CAREC corridors are the overland link between China, Central Asia's landlocked states, the Caucasus, Pakistan and Europe, and traffic on them surged after February 2022 as shippers diverted around Russia. Region-wide average road border-crossing time went the wrong way in 2023 — from 9.9 to 11.0 hours — driven by exactly these paired crossings, even as border costs fell 37 percent to US$131 and speed-with-delay crawled from 23.4 to 24.9 km/h. Every hour a loaded truck sits at a border is a driver's day, a perishable cargo's shelf life (perishables were 31 percent of sampled shipments) and, at the crossings the report singles out, an invitation to informal payments to jump the queue. Development lenders are financing border-post modernisation across the region; if the binding constraint is the unmodernised partner facility, that money buys new lanes that feed a one-lane bridge.

What’s been tried and why it hasn’t worked

The tried remedies are all single-sided. Georgia adopted risk management, redesigned its layout and processes for a one-stop shop, and clears trucks in minutes — but cannot release them. Pakistan's Regional Improvement in Border Services (RIBS) project at Torkham added parking for heavy vehicles, 24-hour operations (since July 2019) and a planned six-lane outbound facility with a design capacity of 1,200 vehicles a day outbound and 550 inbound, on the Pakistani side only. Kazakhstan's China-facing posts sit opposite Chinese facilities with more handling equipment, so queues form on the Kazakh side. Some mismatches are regulatory rather than physical: Chinese trucks are generally prohibited from operating in Central Asia, so at Khorgos (51.4 hours outbound) goods are unloaded into temporary warehouses to wait for Kazakh trucks; the post-2022 shortage of bilateral road permits leaves Georgian and Turkish drivers idling at Tsiteli Khidi. Others are procedural: at Chaman (48.3 hours) and Torkham the delay is "high incidences of full physical examination and cumbersome paperwork," and along the 55-km approach to Torkham "numerous police checkpoints" where "transport operators paid varied sums to secure advantageous positions in the queue." The monitoring programme can see all of this because it measures the crossing as a whole (queue outside the gate, activities inside, authorisation to leave, on each side); the investment and reform decisions it feeds are still made per side, per agency, and Torkham alone involves at least eight Pakistani agencies (customs, immigration, terminal operator, army, anti-narcotics, three quarantine bodies) before any Afghan counterpart is counted.

What would unlock progress

The reframing is to treat a paired crossing as one coupled queueing system whose throughput is the minimum of the two sides, and to make joint capacity — lanes, staff hours, scanners, permit quotas, hours of operation — the unit of investment appraisal and monitoring. That would need a shared, low-cost measurement of the mismatch (the CPMM data already exist per direction and per side), a way to model where an added lane or a risk-based procedure on one side actually changes end-to-end time, and financing or agreement templates that condition one side's upgrade on a matching commitment from the other, or that fund the poorer side. Adjacent precedents: airport slot coordination and one-stop border posts in East Africa, where a single shared facility removes the mismatch by construction; and industrial-engineering "theory of constraints" scheduling, where the only investment that raises throughput is at the bottleneck.

Entry points for student teams

A team could take the public CPMM data (border-crossing time by BCP, direction and activity) and build a two-sided queueing simulation of one pair — Tsiteli Khidi–Krasnyi Most or Torkham — to show how much end-to-end time falls under (a) the funded one-sided upgrade, (b) a matched upgrade on the other side, and (c) procedural changes alone (mutual recognition of clearance, joint hours, permit-quota changes); this is a proof-of-concept that development banks and CAREC's transport committee could use. A policy team could draft a "paired-BCP appraisal" checklist for lenders that requires the partner side's capacity and permit regime to be documented before a modernisation loan is approved. Relevant skills: operations research/queueing simulation, data analysis, transport economics, public policy.