*6.7. Model Development*

Consistent with prior research (Amran and Haniffa 2011; Amran et al. 2008), we use multiple regressions to assess the variability of the extent of financial risk disclosure and its association with firm characteristics. This statistical method is widely used in previous studies. Thus, the following regression model is developed in testing our hypotheses:

> FRDIi = b0 + b1FS + b2ROA + b3FL + b4LIQD + b5CEM + b6FP + b7JUTE +b8PHARM + b9TAN + b10AT + *i*

where

FRDI = financial risk disclosure index; FS = firm size; ROA = return on assets; FL = corporate financial leverage; LIQD = liquidity; CEM = cement industry = 1 if the company is in the cement industrial sector, otherwise 0; FP = fuel and power industry = 1 if the company is in the fuel and power industrial sector, otherwise 0; JUTE = jute industry = 1 if the company is in the jute industrial sector, otherwise 0; PHARM = pharmaceuticals and chemical industry = 1 if the company is in the pharmaceuticals and chemical industrial sector, otherwise 0; TAN = tannery industry = 1 if the company is in the tannery industrial sector, otherwise 0; AT = auditor type = 1 if audit firms link with Big 4, otherwise 0; and *i* = error terms. To reduce the impact of outliers, we winsorize all continuous variables at 1 percent and 99 percent.
