A consolidated filing of corporate income tax may induce firms to manipulate ownership interests in subsidiaries but no study has systematically examined such behavioral responses. This paper examines empirically inclusions/exclusions of subsidiaries to/from consolidation groups in a quasi-experiment that utilizes the Japanese tax reform of 2002. The identification of tax effects is based on a difference-in-difference strategy that exploits disincentives to consolidate subsidiaries with losses carried forward. The data consists of 37,000-40,000 subsidiary-time observations spanning biennially over 1988-2006. The result shows that losses carried forward significantly reduced the propensity to include subsidiaries to consolidation groups. No evidence on tax-motivated exclusion is found. This result suggests that the forced consolidation regime is preferable.