{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,1]],"date-time":"2026-10-01T18:09:50Z","timestamp":1790878190426,"version":"4.1.0"},"reference-count":25,"publisher":"Public Library of Science (PLoS)","issue":"10","license":[{"start":{"date-parts":[[2026,10,1]],"date-time":"2026-10-01T00:00:00Z","timestamp":1790812800000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["www.plosone.org"],"crossmark-restriction":false},"short-container-title":["PLoS One"],"abstract":"<jats:p>\n                    Synthetic benchmarks can support tax-risk method development when administrative records are inaccessible, but a simulator can also predetermine the patterns later reported as detection findings. We present the Tax Data Generation Model (TDGM), a mechanism-guided simulator that enforces specified income-statement and balance-sheet identities, generates a three-year panel for 10,000 firms, and injects four prespecified risk strategies. This study uses no administrative audit data; its purpose is to test whether the implementation reproduces its design assumptions and to provide a reproducible benchmark. All predictive evaluations used five repeated 70\/30 group splits by firm, so no firm appeared in both training and test data. Observed marginal effects were heterogeneous rather than uniformly small: for all risky observations, Cohen\u2019s\n                    <jats:inline-formula id=\"pone.0353137.e001\">\n                      <jats:alternatives>\n                        <jats:graphic xmlns:xlink=\"http:\/\/www.w3.org\/1999\/xlink\" id=\"pone.0353137.e001g\" mimetype=\"image\" position=\"anchor\" xlink:href=\"info:doi\/10.1371\/journal.pone.0353137.e001\" xlink:type=\"simple\"\/>\n                        <mml:math xmlns:mml=\"http:\/\/www.w3.org\/1998\/Math\/MathML\" display=\"inline\" id=\"M1\">\n                          <mml:mrow>\n                            <mml:mrow>\n                              <mml:mi>d<\/mml:mi>\n                            <\/mml:mrow>\n                          <\/mml:mrow>\n                        <\/mml:math>\n                      <\/jats:alternatives>\n                    <\/jats:inline-formula>\n                    was \u22121.29 for profit margin, \u22121.20 for tax burden, 1.05 for the transfer-pricing indicator, and 0.15 for the income-gap ratio. In a controlled cross-sectional comparison, raw CTGAN failed both row-level accounting identities, whereas deterministic projection restored 100% validity but worsened dependence fidelity. Under the revised classifier specification, a one-hidden-layer neural network with conventional L2 regularization achieved mean ROC-AUC 0.9942 and PR-AUC 0.9710; XGBoost achieved 0.9714 and 0.8975, respectively. Removing the income-gap ratio changed XGBoost ROC-AUC only from 0.9714 to 0.9702, while restricting the model to raw Level 1 variables reduced it to 0.8941. At 1% prevalence, XGBoost ROC-AUC remained 0.9720 but PR-AUC fell to 0.6703, showing why prevalence-sensitive metrics are required. Together, the three contributions are a constraint-by-construction simulator, a reusable audit protocol for mechanism-guided generators, and explicit measurement of the accounting-repair\/statistical-fidelity trade-off. Because selected flows are changed after stock generation, derived ratios can carry design-induced information. These are internal properties of a synthetic data-generating process, not estimates of real tax-evasion behavior or deployable audit performance.\n                  <\/jats:p>","DOI":"10.1371\/journal.pone.0353137","type":"journal-article","created":{"date-parts":[[2026,10,1]],"date-time":"2026-10-01T17:51:26Z","timestamp":1790877086000},"page":"e0353137","update-policy":"https:\/\/doi.org\/10.1371\/journal.pone.corrections_policy","source":"Crossref","is-referenced-by-count":0,"title":["A mechanism-guided simulation framework for synthetic tax-risk data: Constraint validation and reproducible benchmarking"],"prefix":"10.1371","volume":"21","author":[{"ORCID":"https:\/\/orcid.org\/0009-0008-2620-2117","authenticated-orcid":true,"given":"Xiaojing","family":"Fan","sequence":"first","affiliation":[{"name":"University of Shanghai for Science and Technology","place":["Shanghai, China"],"department":["Business School"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"340","published-online":{"date-parts":[[2026,10,1]]},"reference":[{"key":"pone.0353137.ref001","author":"Internal Revenue Service","year":"2024"},{"key":"pone.0353137.ref002","author":"Tax Justice Network","year":"2020"},{"issue":"3","key":"pone.0353137.ref003","doi-asserted-by":"crossref","first-page":"559","DOI":"10.1016\/j.dss.2010.08.006","article-title":"The application of data mining techniques in financial fraud detection: A classification framework and an academic review of literature","volume":"50","author":"EWT Ngai","year":"2011","journal-title":"Decision Support Systems"},{"key":"pone.0353137.ref004","doi-asserted-by":"crossref","first-page":"105847","DOI":"10.1016\/j.jeconom.2024.105847","article-title":"Refining public policies with machine learning: The case of tax auditing","volume":"249","author":"M Battaglini","year":"2025","journal-title":"J Econom"},{"issue":"1","key":"pone.0353137.ref005","doi-asserted-by":"crossref","first-page":"2012002","DOI":"10.1080\/08839514.2021.2012002","article-title":"Improving tax audit efficiency using machine learning: The role of taxpayer\u2019s network data in fraud detection","volume":"36","author":"V Baghdasaryan","year":"2022","journal-title":"Appl Artif Intell"},{"issue":"1","key":"pone.0353137.ref006","doi-asserted-by":"crossref","first-page":"3432","DOI":"10.1038\/s41598-025-33382-z","article-title":"Customs fraud detection using a gradient boosting approach for joint classification and risk estimation","volume":"16","author":"R Alwanin","year":"2025","journal-title":"Sci Rep"},{"issue":"3","key":"pone.0353137.ref007","doi-asserted-by":"crossref","first-page":"101090","DOI":"10.1016\/j.ecosys.2023.101090","article-title":"Using accounting information to predict aggressive tax location decisions by European groups","volume":"47","author":"M Borrotti","year":"2023","journal-title":"Econ Syst"},{"issue":"1","key":"pone.0353137.ref008","first-page":"1","article-title":"Multiple imputation for statistical disclosure limitation","volume":"19","author":"TE Raghunathan","year":"2003","journal-title":"J Off Stat"},{"key":"pone.0353137.ref009","unstructured":"Xu L, Skoularidou M, Cuesta-Infante A, Veeramachaneni K. Modeling tabular data using conditional GAN. In: Advances in Neural Information Processing Systems, 2019. 7335\u201345."},{"issue":"6","key":"pone.0353137.ref010","doi-asserted-by":"crossref","first-page":"7499","DOI":"10.1109\/TNNLS.2022.3229161","article-title":"Deep neural networks and tabular data: A survey","volume":"35","author":"V Borisov","year":"2024","journal-title":"IEEE Trans Neural Netw Learn Syst"},{"key":"pone.0353137.ref011","first-page":"17564","article-title":"TabDDPM: modelling tabular data with diffusion models","volume":"202","author":"A Kotelnikov","year":"2023","journal-title":"Proceedings of Machine Learning Research"},{"issue":"2","key":"pone.0353137.ref012","doi-asserted-by":"crossref","first-page":"1","DOI":"10.15611\/eada.2024.2.01","article-title":"Synthetic financial data: A case study regarding polish limited liability companies data","volume":"28","author":"A Szymura","year":"2024","journal-title":"Econometrics"},{"issue":"3","key":"pone.0353137.ref013","doi-asserted-by":"crossref","first-page":"545","DOI":"10.1162\/rest_a_01101","article-title":"The dynamic effects of tax audits","volume":"105","author":"A Advani","year":"2023","journal-title":"Rev Econ Stat"},{"issue":"3","key":"pone.0353137.ref014","doi-asserted-by":"crossref","first-page":"651","DOI":"10.3982\/ECTA9113","article-title":"Unwilling or unable to cheat? Evidence from a tax audit experiment in Denmark","volume":"79","author":"HJ Kleven","year":"2011","journal-title":"Econometrica"},{"issue":"1","key":"pone.0353137.ref015","first-page":"1","article-title":"Under the radar: The effects of monitoring firms on tax compliance","volume":"10","author":"M Almunia","year":"2018","journal-title":"American Economic Journal: Economic Policy"},{"key":"pone.0353137.ref016","doi-asserted-by":"crossref","first-page":"12","DOI":"10.1016\/j.jclinepi.2019.02.004","article-title":"A systematic review shows no performance benefit of machine learning over logistic regression for clinical prediction models","volume":"110","author":"E Christodoulou","year":"2019","journal-title":"J Clin Epidemiol"},{"key":"pone.0353137.ref017","doi-asserted-by":"crossref","unstructured":"Grinsztajn L, Oyallon E, Varoquaux G. Why Do Tree-Based Models Still Outperform Deep Learning on Typical Tabular Data?. In: Advances in Neural Information Processing Systems 35, 2022. 507\u201320. https:\/\/doi.org\/10.52202\/068431-0037","DOI":"10.52202\/068431-0037"},{"key":"pone.0353137.ref018","doi-asserted-by":"crossref","first-page":"84","DOI":"10.1016\/j.inffus.2021.11.011","article-title":"Tabular data: Deep learning is not all you need","volume":"81","author":"R Shwartz-Ziv","year":"2022","journal-title":"Information Fusion"},{"issue":"3","key":"pone.0353137.ref019","doi-asserted-by":"crossref","first-page":"235","DOI":"10.1214\/ss\/1042727940","article-title":"Statistical fraud detection: A review","volume":"17","author":"RJ Bolton","year":"2002","journal-title":"Stat Sci"},{"key":"pone.0353137.ref020","doi-asserted-by":"crossref","unstructured":"Liu FT, Ting KM, Zhou Z-H. Isolation Forest. In: 2008 Eighth IEEE International Conference on Data Mining, 2008. 413\u201322. https:\/\/doi.org\/10.1109\/icdm.2008.17","DOI":"10.1109\/ICDM.2008.17"},{"issue":"3","key":"pone.0353137.ref021","doi-asserted-by":"crossref","DOI":"10.1145\/1541880.1541882","article-title":"Anomaly detection: A survey","volume":"41","author":"V Chandola","year":"2009","journal-title":"ACM Comput Surv"},{"key":"pone.0353137.ref022","volume-title":"Statistical power analysis for the behavioral sciences","author":"J Cohen","year":"1988","edition":"2"},{"issue":"5536","key":"pone.0353137.ref023","doi-asserted-by":"crossref","first-page":"1818","DOI":"10.1126\/science.1062081","article-title":"Zipf distribution of U.S. firm sizes","volume":"293","author":"RL Axtell","year":"2001","journal-title":"Science"},{"issue":"6","key":"pone.0353137.ref024","doi-asserted-by":"crossref","first-page":"1417","DOI":"10.2307\/1911408","article-title":"Biases in dynamic models with fixed effects","volume":"49","author":"S Nickell","year":"1981","journal-title":"Econometrica"},{"key":"pone.0353137.ref025","unstructured":"Bennett KP, Bredensteiner EJ. Duality and geometry in SVM classifiers. In: Proceedings of the Seventeenth International Conference on Machine Learning, 2000. 57\u201364."}],"container-title":["PLOS One"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0353137","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,1]],"date-time":"2026-10-01T17:51:32Z","timestamp":1790877092000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0353137"}},"subtitle":[],"editor":[{"given":"Ahmed","family":"Eltweri","sequence":"first","affiliation":[{"name":"Liverpool John Moores University","place":["UNITED KINGDOM OF GREAT BRITAIN AND NORTHERN IRELAND"]}],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,10,1]]},"references-count":25,"journal-issue":{"issue":"10","published-online":{"date-parts":[[2026,10,1]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pone.0353137","relation":{},"ISSN":["1932-6203"],"issn-type":[{"value":"1932-6203","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,10,1]]}}}