Why ARC selection matters more than most institutions realise
The difference between a well-matched ARC and a poorly matched one for a given portfolio can be 8 to 14 percentage points of recovery rate on the charged-off book value. An ARC that specialises in secured SME debt in UAE, with a field team experienced in agricultural equipment hypothecation and strong relationships with local sub-registrars, will recover significantly more from a portfolio of distressed UAE SME loans than an ARC whose expertise is unsecured personal finance / Murabahas in metropolitan cities. The portfolio match is not a preference — it is the primary driver of recovery yield.
Most institutions enroll ARCs through a periodic tender process — issuing an RFP, evaluating bids, enrolling three to five ARCs for a one-year term, and then assigning portfolios to whichever ARC has the lowest current load. The Post Charge-Off Recovery AI replaces this with a continuous evaluation framework: seven criteria, updated monthly as each ARC's performance data comes in, producing a dynamic ranking that reflects current capability rather than capability at the time of the last tender.
The 7 ARC evaluation criteria
| Criterion | Weight | What Is Measured | Data Source |
|---|---|---|---|
| Historical recovery rate — matched portfolio type | 30% | Actual cash recovery as % of assigned book value for portfolios of the same type (secured SME, unsecured personal, LAP, etc.) in the same geography, over the last 24 months. | Performance data from the institution's own prior assignments + CBUAE / SAMA CRILC data where available + ARC self-disclosure |
| CBUAE / SAMA registration and regulatory standing | 20% | Active CBUAE / SAMA Certificate of Registration as a Securitisation Company / Reconstruction Company under mortgage / security enforcement law Section 3. No CBUAE / SAMA enforcement actions or show-cause notices in the last 3 years. CAR above 15%. | CBUAE / SAMA CRILC registry + ARC's most recent quarterly compliance report |
| Geographic and segment specialisation match | 20% | Proportion of the ARC's active portfolio that matches the institution's proposed assignment — by geography (UAE, Saudi Arabia), product type (SME, home finance / Ijara, personal finance / Murabaha), and borrower profile (salaried, SE proprietor, agri). | ARC's portfolio disclosure + field team geography mapping |
| Net asset value and financial capacity | 15% | ARC's net owned funds relative to the proposed assignment value. An ARC that cannot financially absorb the assignment value (even on deferred consideration terms) creates completion risk. Minimum NOF for any assignment: 15% of assignment face value. | ARC's audited balance sheet + latest CAR report |
| Legal and recovery track record | 10% | ARC's track record in mortgage / security enforcement law enforcement, enforcement court proceedings, and insolvency court insolvency proceedings — specifically its ability to convert legal process into cash recovery within 24 months of assignment. Proportion of assigned accounts resolved (any outcome) vs lingering. | ARC self-disclosure + court records + IBBI proceedings data |
| Technology and data management capability | 3% | ARC's ability to receive portfolio data in structured format, provide monthly recovery reporting in the institution's required format, and integrate with the institution's Recovery AI data feed for real-time tracking. Manual-only ARCs create reconciliation complexity. | Technical assessment + integration test |
| Conduct and FPC compliance record | 2% | Complaints filed against the ARC with the CBUAE / SAMA, consumer forums, or the institution's own grievance system. Any ARC with more than 3 upheld FPC complaints in the last 12 months is excluded regardless of financial score. | Grievance AI database + CBUAE / SAMA complaint registry + consumer forum records |
The ARC evaluation scorecard: three ARCs for a UAE SME portfolio
The assignment structure: what the Recovery AI recommends alongside the ARC
The ARC recommendation includes not just which ARC but what assignment structure. The institution has three assignment options: outright sale (institution sells the portfolio to the ARC for an upfront cash consideration — certain recovery, lower yield); recovery-sharing structure (ARC issues recovery interests representing a share of future recoveries — the institution holds the recovery interests and recovers as and when the ARC collects — variable yield, potentially higher); or hybrid (a partial upfront consideration plus recovery interest assignment for the balance). The Post Charge-Off Recovery AI models the expected recovery yield under each structure for this specific portfolio, based on the ARC's historical performance curve and the portfolio's account characteristics, and recommends the structure that maximises the institution's expected recovery in net present value terms.
For the UAE SME batch, the Recovery AI recommends the recovery-sharing structure with ARC-Alpha: an recovery interest assignment at 18% of book value (AED8.7 million face value of recovery interests), with an expected recovery of 28–32% of book value (AED13.5–15.5 million) over 36 months. This is compared against the outright sale alternative — ARC-Beta offered an outright purchase at 12% of book (AED5.8 million cash) — and the recovery-sharing structure with ARC-Alpha is projected to return AED7.9 million more in NPV terms, primarily because of ARC-Alpha's superior performance record with matched portfolios.
The ARC whose bid was best eighteen months ago is not necessarily the ARC whose performance is best today
ARC performance changes. A specialist team leaves. A geography-specific regulatory change affects enforcement timelines. A competitor ARC acquires a portfolio that competes with the institution's accounts for the same field recovery resources. The institution that locked its enrollment panel at the last tender and has not updated its performance data since then may be assigning its best portfolios to an ARC whose capability has materially deteriorated since it won the bid. The Post Charge-Off Recovery AI evaluates ARC performance monthly against actual recovery data — not against the credentials submitted in an RFP response — and updates the assignment recommendation for each portfolio batch accordingly.