The conventional framing of the credit underwriting question — how to grow the loan book while preserving asset quality — sets up a tension that is usually overstated. Growth and quality are not in fundamental opposition. What is in opposition, more often, is growth velocity and the discipline of underwriting at the point of origination. Institutions that have struggled to maintain asset quality through growth phases have rarely done so because their underwriting policy was wrong. They have done so because the practical discipline of applying the policy weakened under the volume of decisions the growth phase required.
This is a different problem than the conventional framing suggests. It is also a more tractable one, because the answer is not to lower the growth target. The answer is to build an underwriting framework that scales with growth — where the discipline is embedded in the decisioning chain rather than in the human judgement that growth volume eventually overwhelms.
For institutions across the GCC operating in competitive corporate and SME lending segments, the question of how to do this has become more pointed as the regional growth cycle has accelerated. The institutions that have managed it have tended to have done several specific things differently from those that have not.
Scorecard-driven decisioning — the substantive version
Credit scorecards have been a feature of underwriting practice for decades. What distinguishes scorecard frameworks that genuinely support asset quality from those that exist as policy documentation is the substantive linkage between the scorecard output and the decisioning chain.
A scorecard that produces a score but does not bind the decision is, in operational terms, an input. A scorecard that determines the decision within defined parameters, with overrides governed by explicit authority and reviewed on a defined frequency, is a control. The supervisory and audit examination that institutions face on credit risk increasingly distinguishes between these two — not by reading the policy but by examining the application data.
The substantive test is the override rate. A scorecard that is genuinely binding produces a low override rate, with each override documented, justified, and reviewed. A scorecard whose output is routinely overridden — whether to approve loans the scorecard would decline, or to decline loans the scorecard would approve — is not functioning as a control regardless of how its policy is written. The override pattern itself often surfaces the substantive constraint the institution is operating under: pressure from origination teams to write business that does not meet the scorecard threshold, or risk team caution that does not trust scorecard-approved cases without secondary review.
Both override patterns are diagnostic. The institutions that have built durable underwriting discipline have tended to monitor override rates as a primary metric, with movement in the override rate triggering review of either the scorecard calibration or the operating culture around scorecard use. The metric is more informative than the scorecard accuracy itself, because it captures whether the scorecard is functioning as designed in the institution’s actual operating environment.
Early warning systems — connected, not parallel
Early warning systems have become a standard feature of credit risk frameworks in GCC institutions. The substantive question is whether the EWS is connected to the rest of the credit risk architecture, or whether it sits as a parallel discipline that produces watch lists and reports but does not connect to decisions.
The connections that make an EWS substantive run in three directions. The first is the connection to IFRS 9 staging. A material EWS signal should be a candidate input to the qualitative SICR criteria the institution applies under paragraph 5.5.9 of IFRS 9. The institution that has both an EWS and a SICR framework but no documented linkage between them is operating two parallel monitoring systems, with the potential for inconsistency that any parallel system produces.
The second is the connection to the underwriting decision. EWS signals from existing portfolio behaviour should feed back into the underwriting parameters being applied to new origination in the same segment. If a particular segment is showing rising delinquency in the EWS, the underwriting standards for new exposure to that segment should tighten, with the linkage governed and documented. An EWS that does not feed back into underwriting is monitoring without learning.
The third is the connection to portfolio management. EWS outputs should inform the active management of existing exposures — concentration adjustments, sectoral limits, single-name limits, and where appropriate, exit decisions. The EWS that produces watch lists but no exposure management activity is, in effect, a documentation exercise.
In practice, the institutions that have built these connections have tended to do so by establishing explicit governance forums where EWS outputs are discussed alongside underwriting calibration and portfolio management. The forum is the mechanism that makes the connection operational. Without it, the disciplines tend to drift back into parallel operation.
The Stage 3 boundary — what it actually represents
The Stage 3 boundary in IFRS 9 is, in the standard’s framing, a credit-impaired state requiring lifetime ECL calculated based on actual rather than expected default. The boundary is the point at which the credit has stopped being a forward-looking estimate and has become a current-state recognition.
For underwriting discipline specifically, the Stage 3 boundary matters because the institution’s ability to influence the credit outcome diminishes substantially once the boundary is crossed. The interventions available before Stage 3 — restructuring, additional collateral, covenant negotiations, exposure reduction — are largely unavailable or substantially less effective after. The implication for underwriting is that the work to prevent Stage 3 entries is more leveraged than the work to manage Stage 3 cases after the fact.
The institutions whose Stage 3 rates have remained controlled through growth periods have tended to be those whose Stage 2 management is substantive. Stage 2 is, in the standard’s logic, the warning state. The institution that uses Stage 2 as a portfolio management category — with explicit watch-list management, active credit review, and interventions calibrated to the specific exposure — tends to convert Stage 2 cases back to Stage 1 at higher rates than institutions that treat Stage 2 as a provisioning category only. The difference compounds over time. The cumulative Stage 3 rate is materially affected by what happens at Stage 2.
The underwriting connection is that Stage 2 patterns inform what the underwriting framework needs to detect at origination. The exposures that the institution most often finds itself moving to Stage 2 are the exposures whose underwriting could most usefully have been more selective. The feedback loop from staging back to underwriting is, in our observation, one of the most underused mechanisms in conventional credit risk frameworks.
SME underwriting — where the discipline most often weakens
For SME and mid-corporate lending specifically, the underwriting discipline tends to come under the most pressure for three reasons that are specific to the segment.
The financial information available on SME borrowers is materially less reliable than the information available on corporate borrowers. Audited financials are sometimes available, sometimes not, and the quality of the audit varies substantially. Management accounts are usually the primary input. The institution’s ability to verify the financial picture independently is limited. This is not a reason to avoid the segment — it is a reason to build the underwriting framework around what can be verified rather than around what is asserted.
The growth opportunity in SME lending across the GCC has been substantial, particularly in the UAE and Saudi Arabia as government policy has actively encouraged credit availability to the segment. The pressure on origination volume creates the conditions under which underwriting discipline most often weakens. Institutions that have grown SME books substantially while maintaining asset quality have tended to do so by investing in the underwriting infrastructure — scorecards calibrated to SME behaviour, EWS tuned to SME signals, portfolio management governance specific to the segment — rather than by accepting underwriting compromise as the cost of growth.
The collateral picture in SME lending is structurally different from corporate lending. Personal guarantees, residential property used as collateral for business credit, and unsecured exposure with covenant protection are common. The LGD assumptions used in IFRS 9 modelling for SME segments need to reflect this collateral reality, and the underwriting standards need to incorporate the substantive enforceability of the collateral arrangements rather than the nominal coverage ratio.
A closing observation
Across recent engagements involving credit risk framework review for institutions across the GCC, the most consistent observation has not been that the underwriting policies were inadequate. The policies, in most institutions, were defensible. What more often required attention was the operational reality of how the policies were applied — the override patterns, the connection between EWS and underwriting, the use of Stage 2 as a management category, and the specificity of the SME underwriting framework.
The work to address these is, characteristically, work the institution can do largely with the resources it already has. The constraint is not external expertise. The constraint is the governance attention required to treat underwriting discipline as a continuous operational concern rather than as a policy framework set at annual review. The institutions that maintain asset quality through growth cycles are not, in our observation, the institutions with the most sophisticated underwriting frameworks. They are the institutions with the most disciplined application of the framework they have.