When Bits Break Recourse: Counterfactual-Faithful Quantization
For practitioners deploying quantized models in high-stakes domains, this work reveals that accuracy alone is insufficient to guarantee recourse reliability and provides a method to preserve it.
Quantization can silently break algorithmic recourse (actionable changes that flip decisions) even when accuracy is preserved. The authors propose Counterfactual-Faithful Quantization (CFQ), which maintains accuracy while significantly improving recourse stability metrics (Validity Drop and Counterfactual Recourse Gap) across bit budgets on Adult, German Credit, and COMPAS datasets.
Quantization can preserve predictive accuracy under low-bit deployment while silently breaking algorithmic recourse: an actionable change that flips a decision before quantization may fail after quantization, or become substantially more costly. We formalize counterfactual sensitivity under quantization through validity, cost, and direction stability, and introduce two metrics: Validity Drop (VD) and Counterfactual Recourse Gap (CRG) that reveal recourse failures invisible to accuracy. We propose Counterfactual-Faithful Quantization (CFQ), which trains quantizer parameters and mixed-precision bit allocation to preserve counterfactual behavior by enforcing the target outcome at teacher recourse points under a global bit budget. A margin-based analysis gives a sufficient condition for recourse transfer under bounded quantization perturbations. Experiments on Adult, German Credit, and COMPAS show that accuracy-matched baselines can significantly degrade recourse stability, while CFQ maintains accuracy and substantially improves VD and CRG across bit budgets.