This study aims to examine the determinants and within-report consistency of appraisal adjustment coefficients in residential land valuation and develops a regression-based screening framework for quality-control review.
Using 168 paired subject–comparable records from 56 valuation reports in Ho Chi Minh City, we estimated ordinary least squares regression with report-clustered standard errors and within-report fixed-effects analysis. Residual-based screening thresholds classified reports into four quality-control tiers.
Location-type mismatch and area dissimilarity are the primary drivers of adjustment magnitude, with each one-tier location difference corresponding to a 4.6-percentage-point change. Within-report analysis supports broad proportionality between adjustments and composite dissimilarity, and the framework flags 25 per cent of reports for review. Leave-one-report-out validation showed bounded out-of-sample value (RMSE 0.079, with about 82 per cent of predictions within ±0.10).
The sample is from a single valuation firm in one former district; multi-firm replication is needed. The adjustment coefficient is an overall ratio, not a decomposition of individual attribute-level adjustments.
The screening framework offers valuation firms a first-stage internal review tool. The identified determinants provide an empirical reference point for regulators implementing Vietnam's 2024 Valuation Standards.
This study provides rare paired subject–comparable evidence from an emerging housing market, introduces a within-report proportionality test exploiting the nested structure of multi-comparable reports and translates findings into a practical quality-control tool for residential valuation practice.
