طراحي و تبيين مدل قيمت گذاري داراييهاي سرمايه اي روش چند نمايشي كسري با استفاده از گشتاور مرتبه بالا در بورس اوراق بهادار تهران

Design and explanation of fractional multiresoloution capital asset pricing model with higher co-moments on Tehran stock exchange

اين تحقيق به دنبال ارائه يك مدل مناسب قيمت گذاري و رفع بعضي از اشكالات رايج در مدل هاي قبلي مي باشد. به همين دليل در ابتدا 82 شركت نمونه تحقيق، به پرتفوي هايي براساس دو گشتاور مرتبه سوم و چهارم داده ها يعني همچولگي و هم كشيدگي مرتب مي شوند. سپس در هركدام از پرتفوي ها سه مدل CAPM، فاما و فرنچ و كارهارت مورد آزمون قرار گرفتند. نتايج نشان داد كه تخمين مدلها براساس طبقه بندي همچولگي و هم كشيدگي باعث بهبود معني داري مدلهاي فوق مي شود اما عدم معني داري عرض از مبدا مابه التفاوت دو پرتفوي اول و سوم نشان داد كه صرف همچولگي و هم كشيدگي قيمتگذاري نمي شود كه اين ممكن بود به دليل يكسان فرض كردن افق سرمايه گذاري افراد باشد به همين خاطر با كمك تجزيه و تحليل موجك پرتفوي هاي سه گانه در دو سطح 1(2تا 4 ماهه) و سطح 2(4تا 8 ماهه) مورد بررسي قرار گرفت. نتايج به طور جالبي نشان از بهبود ضريب تعيين مدلها داشت. از طرفي وجود حافظه بلندمدت در داده ها سبب مي شود استفاده از آنها در تخمين با مشكل مواجه مي سازد. براي رفع اين مشكل از مدلهاي قيمت گذاري كسري استفاده كرديم. نتايج مدلهاي الگو براي پرتفوي هايP1،P2 و P3 نشان داد كه ضريب تعيين نسبت به زمانيكه از داده هاي معمولي استفاده مي كنيم بهبود چشم گيري پيدا كرده است اما به غير از CAPM براي P1 و P2 در ساير مدلها صرف ريسك قيمتگذاري نشده است. مقايسه مدلهاي كسري با مدلهاي برآمده از تجزيه و تحليل موجك نشان از برتري محسوس روش تجزيه و تحليل موجك دارد.

The purpose of this study is to provide an appropriate pricing model and also addressing the common deficiencies in other valuation models. To this goal, a sample of 82 companies is selected. (Financial companies, companies with fiscal year different from end of 29th of isfand, companies which their stocks were not traded in past 3 month and those which listed after 1389 are excluded). Then portfolios based on the third and fourth moments are formed by the sampled companies. We used three pricing models including CAPM, Fama- French 3 factor model and Carhart Model for each portfolio. Results show that estimating these models accounting third and fourth moment lead to increase of significance of these models. But the insignificance of the interception of difference between two portfolios showed that third and fourth moment premium are not priced which can be due to assuming homogeneous investment horizon for investors. For this, using wavelet analysis, we analyzed three portfolios with in two horizon levels including level 1 (2 to 4 month) and level 2 (4 to 8 month). Results interestingly showed the improvement in significance and also in R^2. Using GPH estimator, we showed that there is long term memory in studied time series. In order to making these series stationary, we used recent method by calculating fractional data and again we estimated the models. Results showed that models estimation significantly improved for all three portfolios. But except CAPM for the two first portfolios, in other models the risk premium is not priced. Comparing fractional models with those obtained by wavelet models strongly indicates the superiority of the latter. Comparing all models, we concluded that best model for pricing, based on the third moment, is fama- French 3 factor model in level 2 daubeechies4 and by regarding the fourth moment, CAPM is optimal one for the level 2 symlts8.