Do Client Characteristics Really Drive Big N Quality Differentiation?

While the literature generally concludes that Big N auditors provide higher audit quality than nonBig N auditors, an unresolved question is whether this Big N effect is driven by self-selection. In particular, a recent high profile study suggests that Propensity Score Matching (PSM) on a set of commonly examined client characteristics causes Big N quality differentiation to disappear (Lawrence, Minutti-Meza, and Zhang, 2011, hereafter LMZ). We conjecture, however, that this finding may be affected by PSM’s inherent sensitivity to its design choices. To investigate, we examine 3,000 PSM models, each employing a random combination of three basic design choices. We find that the results are sensitive to the design choices, with the majority of the models finding a Big N effect. Expanding our analysis to a more comprehensive set of audit quality proxies and using a more recent time period, we continue to find that most PSM models support a Big N effect. We also employ Coarsened Exact Matching (CEM), a relatively new matching technique that overcomes some of PSM’s limitations. CEM also finds a Big N effect, but with a smaller variation in the magnitude of the effect, and better covariate balance, suggesting CEM is less sensitive than PSM to its design choices and results in better matching. Overall, our evidence suggests that client characteristics do not drive Big N quality differentiation. Acknowledgments: This study has benefited from helpful comments by workshop participants at Boston College, the Chinese University of Hong Kong and the University of Southern California.