مجلة الاقتصاد والمالية
Volume 12, Numéro 2, Pages 250-266
2026-06-01
Authors : Maghnia Houari .
This research evaluates the predictive efficacy of the "Five Cs" of Credit model —comprising Character, Capacity, Capital, Collateral, and Surrounding Conditions—within the credit decision-making processes of Algerian commercial banks. Using a validated 40-item Likert-scale instrument, primary data were synthesized from 30 credit officers across three banking institutions in Saïda Province. While the constructs exhibited robust internal reliability (Cronbach's α = 0.799–0.910), preliminary diagnostics revealed severe multicollinearity among predictors, with inter-construct correlations peaking at r = 0.848. Consequently, this study bypasses traditional OLS limitations by employing Sparse Partial Least Squares (sPLS) regression to ensure analytical precision and resolve sign-reversal artifacts. The resulting two-component sPLS model achieved an R²of 0.397 and a cross-validated Q² of 0.113 (via Leave-One-Out), confirming the model's out-of-sample predictive power. Variable Importance in Projection (VIP) analysis identified Surrounding Conditions (VIP = 1.288) and Capacity (VIP = 1.069) as the primary drivers of credit quality, both significantly exceeding the 1.0 threshold. Conversely, Character emerged as the least influential factor (VIP = 0.688), a finding that likely underscores the structural limitations of credit bureau infrastructure in the Algerian context. These results highlight the necessity of context-specific credit risk governance and offer a refined methodological roadmap for assessing creditworthiness in emerging markets.
5Cs Framework, Creditworthiness, Credit Risk, Sparse PLS, VIP Analysis, Algerian Banking Sector
بوسالم أحلام
.
عابد يوسف
.
ص 117-132.
Yahia Zeghoudi
.
pages 74-88.
Khellil Khaled
.
Loucif Kamilia
.
pages 13-32.