Statistics-Friendly Confidentiality Protection for Establishment Data, with Applications to the QCEW

2025-09-01

AI Summary

Commercial datasets exhibit severe skewness and contain critical outliers; conventional disclosure control and differentially private mechanisms designed for individual-level data fail to simultaneously ensure confidentiality and statistical utility. Method: We propose a novel privacy framework tailored for statistical usability, grounded in Gaussian differential privacy. Our approach introduces an interpretable noise injection mechanism that rigorously preserves extreme values while enabling high-fidelity reconstruction of confidential microdata from noisy aggregate queries. Contribution/Results: Evaluated on real-world economic census data (e.g., QCEW), our method substantially outperforms existing techniques: it satisfies formal privacy guarantees (e.g., Gaussian DP), maintains high query accuracy, and supports policy-relevant analyses. The framework bridges theoretical rigor with practical deployability, establishing a new paradigm for releasing sensitive microenterprise data.

Technology Category

• Machine Learning: Privacy
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Data Mining & Knowledge Management: Intelligent Query Processing

Application Category

• Security and Privacy: Data transparency and provenance
• Economics, Online Markets and Human Computation: Fairness, privacy, and diversity in economic environments
• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems

Abstract

Confidentiality for business data is an understudied area of disclosure avoidance, where legacy methods struggle to provide acceptable results. Modern formal privacy techniques designed for person-level data do not provide suitable confidentiality/utility trade-offs due to the highly skewed nature of business data and because extreme outlier records are often important contributors to query answers. In this paper, inspired by Gaussian Differential Privacy, we propose a novel confidentiality framework for business data with a focus on interpretability for policy makers. We propose two query-answering mechanisms and analyze new challenges that arise when noisy query answers are converted into confidentiality-preserving microdata. We evaluate our mechanisms on confidential Quarterly Census of Employment and Wages (QCEW) microdata and a public substitute dataset.

Problem

• Protecting confidentiality for highly skewed business data
• Developing interpretable privacy framework for policy makers
• Addressing outlier sensitivity in formal privacy techniques

Innovation

• Gaussian Differential Privacy inspired framework
• Two query-answering mechanisms for business data
• Confidentiality-preserving microdata conversion techniques