ريزش مورد انتظاردر بورس اوراق بهادار تهران ( رويكرد نيمه پارامتريك پويا)

Expected Shortfall in Tehran Stock Exchange (Dynamic Semi-Parametric Approach)

ﺟﻪ ﺑﻪ ﭼﺎﻟﺶﻫﺎي ﻣﻮﺟﻮد در ارﺗﺒﺎط ﺑﺎ ﺑﺮآورد و ﭘﯿﺶﺑﯿﻨﯽ ﻣﻌﯿـﺎر رﯾـﺰش ﻣـﻮرد اﻧﺘﻈـﺎر ) ES( ﺑـﻪ ﺻﻮرت ﭘﻮﯾﺎ و ﺑﺎ روﯾﮑﺮد ﻧﯿﻤﻪ ﭘﺎراﻣﺘﺮﯾﮏ، در اﯾﻦ ﭘﮋوﻫﺶ، ﺑﺎ اراﺋـﻪ ﭼـﺎرﭼﻮب ﮐﻠـﯽ، ﺑـﻪ ﻣﻌﺮﻓـﯽ و ارزﯾﺎﺑﯽ ﻋﻤﻠﮑﺮد ﻣﺪلﻫﺎي ﻧﯿﻤﻪ ﭘﺎراﻣﺘﺮﯾـﮏ ﭘﻮﯾـﺎ درﭘـﯿﺶ ﺑﯿﻨـﯽ ﻣﻌﯿـﺎر رﯾـﺰش ﻣـﻮرد اﻧﺘﻈـﺎر ) ES( در -1387/09/14 ﺎ دادهﻫـﺎي دوره زﻣـﺎﻧﯽ ﺷـﻮد. در اﯾـﻦ راﺳـﺘ ﺑﻮرس اوراق ﺑﻬﺎدار ﺗﻬﺮان ﭘﺮداﺧﺘـﻪ ﻣـﯽ 1399/06/05 ﻣﻮرد اﺳﺘﻔﺎده ﻗﺮار ﻣﯽﮔﯿﺮﻧﺪ و ﺑﺎ اﺳﺘﻔﺎده از روﯾﮑﺮد رﺗﺒﻪﺑﻨﺪي اﺗﻮرﮔﺮﺳﯿﻮ ﺗﻌﻤـﯿﻢ ﯾﺎﻓﺘـﻪ )GAS(، ﻣــﺪلﻫــﺎي ﭘﻮﯾــﺎي و ﻧﯿﻤــﻪ ﭘﺎراﻣﺘﺮﯾــﮏ GARCH-FZ ،GAS-1F ،GAS-2F وhybrid GAS/GARCH ﺑﻪ ﻣﻨﻈﻮر ﺑﺮآورد ﻣﻌﯿﺎر رﯾﺰش ﻣﻮرد اﻧﺘﻈﺎر )ES( ﻣﻌﺮﻓﯽ و در ﭘﯿﺶﺑﯿﻨـﯽ اﯾـﻦ ﻣﻌﯿـﺎر در ﺑﻮرس اوراق ﺑﻬﺎدار ﺗﻬﺮان ﻣﻮرد ارزﯾﺎﺑﯽ ﻗﺮار ﻣﯽﮔﯿﺮﻧﺪ. در اداﻣﻪ، ﮐﺎراﯾﯽ اﯾﻦ ﻣﺪلﻫﺎ ﺑﺎ ﻣﺪلﻫﺎي ﺳﻨﺘﯽ در اﯾﻦ ﺣﻮزه از ﺟﻤﻠﻪ ﻣﺪلﻫﺎي ﮔﺎرچ و ﻣﺪلﻫﺎي ﭘﻨﺠـﺮه ﻏﻠﺘـﺎن ﺑـﺮ اﺳـﺎس آزﻣـﻮن ﻫـﺎي ﭘـﺲ آزﻣﺎﯾﯽ ﻣﻘﺎﯾﺴﻪ ﻣﯽﺷﻮﻧﺪ. ﻧﺘﺎﯾﺞ اﯾﻦ ﻣﻄﺎﻟﻌﻪ ﺣﺎﮐﯽ از ﻋﻤﻠﮑﺮد ﺑﻬﺘﺮ ﻣﺪلﻫﺎي ﻧﯿﻤﻪ ﭘﺎراﻣﺘﺮﯾﮏ ﭘﻮﯾـﺎ در ﭘﯿﺶﺑﯿﻨﯽ ﺑﺮون ﻧﻤﻮﻧﻪاي ﻣﻌﯿﺎر رﯾﺰش ﻣﻮرد اﻧﺘﻈﺎر )ES( ﻧﺴﺒﺖ ﺑﻪ ﻣﺪلﻫﺎي رﻗﯿﺐ اﺳﺖ ﻋﻼوه ﺑﺮ اﯾـﻦ ﻣﺪل GAS-1F در ﭘﯿﺶﺑﯿﻨﯽ ﺑﺮون ﻧﻤﻮﻧﻪاي ﺑﻬﺘﺮﯾﻦ ﻋﻤﻠﮑﺮد را در ﺑﯿﻦ ﻣﺪلﻫﺎي ﭘﻮﯾﺎ ﻧﺸﺎن دادهاﺳﺖ.

Considering the challenges related to estimating and forecasting the expected Shortfall dynamically and with a semi-parametric approach, in this study, providing a general framework, dynamic semi-parametric models in forecasting Expected Shortfall in Tehran Stock Exchange be introduced and evaluated. In this regard, the data of the period 2008.12.04-2020.08.26 and Generalized Autoregressive Score (GAS) approach are used to introducing dynamic semi-parametric models (GAS-2F, GAS-1F, GARCH-FZ and hybrid). Then expected Shortfall (ES) in Tehran Stock Exchange be estimated and forecasting performance of these models are compared with traditional models in this field, including GARCH models and rolling window models based on backtesting their results. The results of this study indicate better performance of dynamic semi-parametric models in forecasting the expected Shortfall (ES) than competing models. In addition, the GAS-1F model has shown the best performance among all models.