Redesigning mandatory pre‑departure training in South–South corridors can cut brokerage rents and raise migrants' employability by delivering earlier, decentralized, and TVET‑aligned learning with portable credentials. Generative AI can serve as low‑cost, multilingual learning support, but must be limited to auditable, assistive roles to avoid becoming a hidden gatekeeper.

Main Finding

Mandatory pre-departure training in South–South labour corridors (examined via the Myanmar–Malaysia corridor) is a highly implementable, cross-level lever for improving regularity and rights-protecting mobility in contexts with limited enforcement and coordination capacity. By redesigning training along four axes—standards, timing, delivery architecture, and recognition/portability—governance can reduce information asymmetries, lower dependence on brokers, and better connect migration to labour‑market value without waiting for slower reforms in permit/enforcement regimes. Generative AI can play a limited, auditable role as multilingual, low‑bandwidth learning support but must be governed to avoid digital gatekeeping.

Key Points

• Problem diagnosis

  • Current mandatory pre-departure training is typically delivered late, generically, and with weak assessment, limiting its capacity to change recruitment choices or support migrants after arrival.
  • Corridor governance is fragmented, with uneven implementation capacity across sending and receiving actors.

• Conceptual reframing

  • Treat training as migration‑governance infrastructure that can function simultaneously as:
  • A capability intervention (actionable navigation, contract comprehension, safe help‑seeking).
  • A labour‑market signal when aligned with TVET and human‑capital planning.
  • A potential gatekeeping node if access, assessment, and accountability are weak.

• Three testable propositions

• Earlier, decentralised access to training reduces information asymmetry and dependence on intermediaries.
• TVET alignment and portable skills recognition (functional, employer‑usable verification such as micro‑credentials) let training convert into labour‑market value and mobility options.

• Rights‑based effectiveness requires measurable capability outcomes and institutional follow‑through (beyond information transfer).

• Skills recognition

  • Emphasis on functional, employer‑usable verification and portability (e.g., micro‑credentials, QA/transparency instruments), not formal legal harmonisation.

• Role of AI

  • Proposed role: bounded inclusion infrastructure — multilingual, low‑bandwidth learning support.
  • Strict exclusions: not to be used for eligibility screening, risk scoring, or automated decision‑making.
  • Governance requirements: content validation, transparency, data minimisation, and human accountability to prevent digital gatekeeping.

• Practical outputs

  • Sequenced policy toolkit specifying roles across corridor actors.
  • Empirical agenda for testing propositions.
  • Transferable instrument layer drawn from EU qualification/QA/translation tools for recognition/portability.

Data & Methods

• Approach: conceptual analysis anchored in the Myanmar–Malaysia corridor.
• Methods used:
  • Structured desk review of policy, program, and governance materials.
  • Corridor process mapping to identify timing, actors, and touchpoints.
  • Governance gap analysis to reveal capacity and accountability shortfalls.
• Nature of evidence: policy‑oriented, qualitative and analytical; produces testable propositions and an empirical agenda but does not report causal estimates from new field data.

Implications for AI Economics

• Economic roles and potential impacts

  • Transaction costs and intermediation: earlier, decentralized training (with digital support) could reduce search frictions and brokerage rents by improving migrants’ information and bargaining capacity.
  • Human‑capital returns and signalling: TVET-aligned training with portable, employer‑recognised credentials can change how employers value pre‑departure training—potentially raising match quality, wage outcomes, and mobility options.
  • Labour‑market matching and flows: better‑designed training could affect the composition and quality of migrant flows, with implications for wages, remittances, and destination labour supply elasticity.
  • Distributional effects: access to digital learning and credential portability could unevenly benefit those with connectivity or prior skills; risks of digital divides should be measured.

• Role for generative AI (economically scoped)

  • Value: reduce language barriers, lower per‑trainee delivery costs, provide low‑bandwidth personalized learning support, and scale multilingual outreach.
  • Risks: if misgoverned, AI could become a hidden gatekeeper (through opaque content, biased guidance, or data misuse) that increases exclusion or information asymmetry.
  • Governance for economic legitimacy: require content validation (accuracy, context relevance), transparency of model use, data minimisation, human oversight, and auditable logs—so economic benefits do not come at the cost of hidden biases or exclusion.

• Research and evaluation agenda (suggested empirical strategies)

  • Randomized or quasi‑experimental evaluations comparing:
    • Early vs late timing and decentralized vs centralized delivery of training on broker use, contract outcomes, and wage/placement outcomes.
    • TVET‑aligned + portable micro‑credentials vs information‑only training on employer hiring, wage premia, and mobility choices.
    • AI‑assisted multilingual modules vs standard modules on learning outcomes, time‑to‑placement, and cost‑effectiveness.
  • Data linkage: connect training completion records, credential verification logs, and destination employment outcomes (surveys, administrative data) to estimate returns.
  • Market signalling tests: employer surveys and hiring experiments to gauge how portable credentials affect hiring and wage offers.
  • Cost‑benefit and distributional analyses: assess program unit costs, reductions in brokerage fees, impacts across gender/skill subgroups, and digital access barriers.

• Policy design recommendations relevant to AI economics

  • Prioritize low‑cost, scalable interventions that improve information and verifiable competence (micro‑credentials, QA, translation/clarity instruments).
  • Build interoperability and transparency into credential systems to enable market validation and reduce asymmetric information.
  • Limit automated use of AI to assistive learning; prohibit AI for screening/decision‑making unless strong, audited governance and proportionality are established.
  • Monitor market effects (brokerage, wage dynamics) and include safeguards to prevent credential inflation or unintended labour‑market segmentation.

Overall, the paper positions training governance—and carefully governed, assistive AI—as pragmatic, implementable tools with measurable economic effects on South–South migration markets. It calls for empirical testing of the supply‑side and signalling mechanisms to quantify impacts on brokerage, wages, mobility options, and distributional outcomes.