Normalizing bid discount rates in construction procurement by package scale and contract duration in Ho Chi Minh city

Abstract

This paper proposes a two-dimensional normalization framework for bid discount rates in civil-works procurement along package scale (S) and contract duration (T). Using a dataset of 201 construction packages in Ho Chi Minh City (2023-2024), we stratify observations by tertiles to form an SXT matrix and compile P25-P50-P75 percentile benchmarks for each cell. The methodology combines robust statistics (median, interquartile range) with the Kruskal-Wallis test and epsilon-squared (ε²) effect size to ensure inferences that are stable under skewness and outliers. Results show that, when each dimension is examined in isolation (S or T), differences in the distribution of discount rates are statistically insignificant with very small effect sizes; by contrast, clear interactions emerge in the S×T space: some configurations (e.g., large packages with moderate duration) exhibit higher discounts, whereas others (e.g., small packages with long duration) display lower discounts. Based on these estimates, we develop S×T reference tables that classify observed discounts into “low-typical-high” states consistent with the package configuration. The resulting percentile benchmarks by scale × duration provide owners with an objective basis for evaluation and anomaly detection in discount levels, while offering contractors a market reference to optimize bidding strategies by package configuration, balance risk-return, and position competitive capability.