مطالعه مقايسه‌اي الگوهاي پيش‌بيني ورشكستگي و ارائه الگوي بهينه براي محيط اقتصادي ايران

A Comparative Study of Bankruptcy Prediction Models and Presenting an Optimized Model for Iran's Economic Environment

هدف: پيش‌بيني ورشكستگي شركت‌ها يكي از اساسي‌ترين فعاليت‌ها در بررسي ريسك و عدم قطعيت شركت‌ها محسوب مي‌شود. از اينرو، معرفي الگو‌هاي مناسب با دقت بالا براي پيش‌بيني ورشكستگي در بسياري از فرايندهاي تصميم‌گيري از اهميت اساسي برخوردار است. هدف از مطالعه حاضر، بررسي دو الگوي معروف چاوا و جارو و كمپبل و همكاران براي پيش‌بيني ورشكستگي شركت‌ها در محيط اقتصادي ايران و معرفي يك الگوي بهينه است. روش: دوره زماني مورد مطالعه 13 سال (از سال 1384 الي سال 1397) و تعداد شركت‌هاي نمونه 188 شركت و 2444 سال- شركت هستند. با استفاده از رگرسيون لجستيك براي داده‌هاي تابلويي، نسبت به يافتن ضرايب متغيرهاي الگو‌هاي مورد بررسي متناسب با بازار سرمايه ايران اقدام گرديد. الگو‌هاي ارائه شده با استفاده از رگرسيون لجستيك اثرات ثابت اجرا شده و پس از رسم منحني ROC براي هر دو مدل، بهترين الگو بر حسب بيشترين AUC انتخاب شد. يافته‌ها: از بين متغيرهاي مدل چاوا و جارو (2004)، تنها سه متغير نسبت كل بدهي‌ها به كل دارايي‌ها (TLTA)، سود خالص تقسيم بر كل دارايي‌ها (NITA) و نوسانات بازده قيمت سهام (SIGMA) در سطح اطمينان 95 تأثير معنا‌داري بر احتمال ورشكستگي شركت‌ها داشته‌اند. هم‌چنين، تنها پنج متغير كل بدهي‌ها تقسيم بر ارزش بازاري كل دارايي‌ها (TLMTA)، سود خالص تقسيم بر ارزش بازاري كل دارايي‌ها (NIMTA)، وجه نقد و دارايي‌هاي آني تقسيم بر ارزش بازاري كل دارايي‌ها (CASHMTA)، نوسانات بازده قيمت سهام (SIGMA) و ارزش دفتري حقوق صاحبان سهام بر ارزش بازار سهام شركت (RSIZE) در سطح اطمينان 95 تأثير معنا‌داري بر احتمال ورشكستگي شركت‌ها داشته‌اند. نتيجه‌گيري: نتايج نشان داد هر دو الگوي چاوا و جارو (2004) و كمپبل و همكاران (2008) از قدرت مناسب و بسيار بالا براي پيش‌بيني ورشكستگي در محيط اقتصادي ايران برخوردار بودند ولي الگوي چاوا و جارو با AUC برابر 965/0 به عنوان الگوي برتر در پيش‌بيني ورشكستگي شركت‌ها براي محيط اقتصادي ايران معرفي شد.

Objective: Predicting corporate bankruptcy is one of the most important activities in auditing risk and uncertainty of companies. Therefore, introducing appropriate models with high accuracy to predict bankruptcy is essential in many decision-making processes. The purpose of this study is to introduce an appropriate and superior model for predicting corporate bankruptcy in the Iranian economic environment. Chava and Jarrow (2004) and Campbell et al. (2008) have been introduced as hybrid models that consider accounting and market information together. In this study, we intend to use logistic regression and accuracy testing to create a better model. Also, for the first time in Iran, the market value of balance sheet items has been used as a suitable alternative to some balance sheet variables and market variables. Methods: The study period is 13 years (from 2005 to 2019) and the number of sample companies is 188 companies and 2444 years - companies. The data required for this study, which consisted of accounting-based and market-based and combined data, were extracted from financial statements and accompanying notes of sample companies and stock exchange softwares. Using logistic regression, the coefficients of the variables of the mentioned models were found. Results: The proposed models were performed using conditional fixed effect logistic regression and the best model was selected using the ROC curve. The results showed that both Chava & Jarrow (2004), Campbell et al (2008). Models have a suitable and very high power to predict bankruptcy in Iran's economic environment. But Chava & Jarrow model with 96.5% accuracy was introduced as the top model in predicting corporate bankruptcy for Iran's economic environment. Among the variables of the Chava & jarrow model, only three variables Included ratio of total debt-to-assets (TLTA), ratio of net income to total assets (NITA) and Stock returns fluctuations (SIGMA) at 95 confidence level, had a significant effect on corporate bankruptcy. an‎d the other two independent variables of this model did not have a significant effect on the probability of bankruptcy. Also, Among the variables of the Campbell et al model, only five variables Included the ratio of total liabilities to total market value of assets (TLMTA), the ratio of net income to total market value of assets (NIMTA), the ratio of cash and instant assets to total market value of assets (CASHMTA), stock price volatility (SIGMA) ) an‎d the ratio of book value of equity to market value of the company (RSIZE) at the 95 confidence level had a significant effect on the probability of bankruptcy of companies. an‎d the other three independent variables, the difference between the company's stock return and market return (EXRET), the ratio of the company's stock market value to the book value of the company's stock (MB) and the logarithm of the stock price (PRICE) had no significant effect on the probability of bankruptcy. Conclusion: Among the variables that were significant in the model, the ratio of net profit to market value of assets (NIMTA) was the most effective variable. Also, according to the regression coefficients of the variables, it is concluded that bankruptcy is inversely related to the ratio of net income to market value of assets(NIMTA), and the ratio of net income to book value of assets(NITA), and the ratio of cash and instant assets to market value of assets(CASHMTA). Bankruptcy is also directly related to the ratio of total liabilities to the book value of assets (TLTA) and Stock returns fluctuations (SIGMA). In other words, Companies whose stock return fluctuations are not in good shape and have a lot of debts more likely to go bankrupt.