5. Concluding remarks
The present study aims to employ the GM in an accounting field and forecast audit reports for top 50 companies listed on the TSE during 2011-2016. In this regard, the information of financial statements is analysed by using the GRA. Specifically, we used some financial ratios (liquidity, leverage, market, activity and profitability ratios) for this purpose and indicated and calculated their weights by using the Shannon entropy technique. After examining the assumptions of standard linear regression models and conducting the specification tests in panel data models, our findings provide empirical support for our hypotheses. In other words, the GM, the NGBM and the NNGBM deliver a suitable performance for forecasting audit reports. However, we rank the estimated regression models by using the MSE. The paper’s findings provide some evidence indicating that the NNGBM delivers the best forecasting performance as compared to the other models including the GM and the NGBM. This finding is consistent with Chun et al. (2010), Wang (2013) and Zhang et al. (2014). Furthermore, our results are consistent with Wang and Hsu (2008), Huang and Jane (2009) and Mohammadi and ZeinodinZade (2011) in terms of forecasting precision of GMs in comparison with other methods. Using the MSE, we also examine the level of influence that each financial ratio has on audit report. In this case, the CR yields the highest possible impact on audit report whereas the WCT indicates the lowest influence. Consistent with Spathis (2003), Gaganis et al. (2005, 2007), Sajjadi et al. (2007), Amini et al. (2011) and Jamei et al. (2013), audit reports are most influenced by the CR, DER and EPS, respectively.