This study has a number of limitations. The sample is comprised of only relatively
larger firms from certain industries and this limits the generalizability of the findings.
Smaller firms and those from excluded industries may have other reasons to choose to
19
disclose environmental information (although prior studies suggest disclosure by these
groups of firms is very limited). Further, we rely upon TRI data to assess pollution
performance. The information in this dataset is self-reported by affected companies, and
the data are only sporadically inspected. Although TRI has been used in many studies in
accounting and economics its reliability is only as good as the inputs. Finally, although
prior studies support that environmental capital spending is a potentially relevant piece of
information (Clarkson et al. 2004; Johnston, 2005), our investigation does not examine
other types of environmental information disclosure. Whether these other information
items serve a signaling function is not addressed by our results.