To measure timing, we follow prior literature and define a variable name, LAG, which is the number of days between the date of the ARPR and the end of the fiscal year (Bamber and Cheon 1998; Sengupta 2004).
Our first model investigates the effect of proprietary costs on the decision to issue an ARPR. The proxy to measure proprietary information costs (GROWTH) is calculated as the market value scaled by total equity in year t-1.
Once management has decided to issue an ARPR the decision in relation to the date of disclosure must be addressed.
We first explore company growth opportunities affecting the release of an ARPR. The dependent variable is dichotomous (issuers/ non-issuers of ARPRs) and we perform logistic regression.
As in prior research (Guillamon-Saorin and Sousa 2010), we observe a significant positive association between the decision to issue an ARPR and COUNTRY.
For the ARPRs included in this study, the mean of the number of days between the year-end and disclosure of an ARPR is 59 days, with a range that varies between a minimum of 3 days and a maximum of 145 days.
The results indicate that UK companies delay the release of the ARPR longer than the Spanish ones, which contradicts our hypothesis (H2).
5.3 Additional Tests on the Decision to Issue an ARPR
Results not tabulated remain unchanged for the rest of the factors and BETA does not show any statistical association with the decision to issue an ARPR.
Considering that our dependent variable (LAG), which measures the number of days between the year-end and the date of the ARPR, is left truncated, we rerun models (2) (2a) and (2b) using a limited dependent variables approach as a sensitivity test (Maddala 1991; SAS Institute 2004; Schleicher and Walker 2010; Tobin 1958).
Endogeneity might exist in the decision concerning the date of disclosure of the ARPR in relation to whether or not to issue the ARPR at all.