Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Multiple index models Ludˇek Benada Department of Finance, office - 402 e-mail: benada@econ.muni.cz Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Content 1 Multi-factor models 2 Arbitrage Pricing Theory - APT 3 Estimation of parameters Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Content 1 Multi-factor models 2 Arbitrage Pricing Theory - APT 3 Estimation of parameters Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Multi-factor models Extend the single model Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Multi-factor models Extend the single model Captures the correlation structure of non-market influences (Economic factors, structural groups, etc.) Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Multi-factor models Extend the single model Captures the correlation structure of non-market influences (Economic factors, structural groups, etc.) Benjamin King demonstrated the influence of industry on stock prices (1966) Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Multi-factor models Extend the single model Captures the correlation structure of non-market influences (Economic factors, structural groups, etc.) Benjamin King demonstrated the influence of industry on stock prices (1966) → Two models have been proposed: Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Multi-factor models Extend the single model Captures the correlation structure of non-market influences (Economic factors, structural groups, etc.) Benjamin King demonstrated the influence of industry on stock prices (1966) → Two models have been proposed: Multi index model Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Multi-factor models Extend the single model Captures the correlation structure of non-market influences (Economic factors, structural groups, etc.) Benjamin King demonstrated the influence of industry on stock prices (1966) → Two models have been proposed: Multi index model Industrial index model Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models One-factor model: Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models One-factor model: ri = ai + bi ∗ F + ei Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models One-factor model: ri = ai + bi ∗ F + ei Where, Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models One-factor model: ri = ai + bi ∗ F + ei Where, bi . . . sensitivity of asset to the factor F Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models One-factor model: ri = ai + bi ∗ F + ei Where, bi . . . sensitivity of asset to the factor F and bi = σF,ri σ2 F Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models One-factor model: ri = ai + bi ∗ F + ei Where, bi . . . sensitivity of asset to the factor F and bi = σF,ri σ2 F Also we can assume: Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models One-factor model: ri = ai + bi ∗ F + ei Where, bi . . . sensitivity of asset to the factor F and bi = σF,ri σ2 F Also we can assume: ¯ri = ai + bi ∗ ¯F Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models Two-factors model: Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models Two-factors model: ri = ai + bi1 ∗ F1 + bi2 ∗ F2 + ei Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models Two-factors model: ri = ai + bi1 ∗ F1 + bi2 ∗ F2 + ei Where, Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models Two-factors model: ri = ai + bi1 ∗ F1 + bi2 ∗ F2 + ei Where, bi1 . . . sensitivity of asset to the factor F1 bi2 . . . sensitivity of asset to the factor F2 Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models Two-factors model: ri = ai + bi1 ∗ F1 + bi2 ∗ F2 + ei Where, bi1 . . . sensitivity of asset to the factor F1 bi2 . . . sensitivity of asset to the factor F2 and bi1 = σF1,ri σ2 F1 , bi2 = σF2,ri σ2 F2 Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models Two-factors model: ri = ai + bi1 ∗ F1 + bi2 ∗ F2 + ei Where, bi1 . . . sensitivity of asset to the factor F1 bi2 . . . sensitivity of asset to the factor F2 and bi1 = σF1,ri σ2 F1 , bi2 = σF2,ri σ2 F2 Also we can assume: Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models Two-factors model: ri = ai + bi1 ∗ F1 + bi2 ∗ F2 + ei Where, bi1 . . . sensitivity of asset to the factor F1 bi2 . . . sensitivity of asset to the factor F2 and bi1 = σF1,ri σ2 F1 , bi2 = σF2,ri σ2 F2 Also we can assume: ¯ri = ai + bi1 ∗ ¯F1 + bi2 ∗ ¯F2 Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models One-factor model: Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models One-factor model: σ2 i = b2 i ∗ σ2 F + σ2 ei Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models One-factor model: σ2 i = b2 i ∗ σ2 F + σ2 ei and, σi,j = bi ∗ bj ∗ σ2 F Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models One-factor model: σ2 i = b2 i ∗ σ2 F + σ2 ei and, σi,j = bi ∗ bj ∗ σ2 F Two-factors model: Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models One-factor model: σ2 i = b2 i ∗ σ2 F + σ2 ei and, σi,j = bi ∗ bj ∗ σ2 F Two-factors model: σ2 i = b2 i1 ∗ σF 2 1 + b2 i2 ∗ σF 2 2 + σ2 ei Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models One-factor model: σ2 i = b2 i ∗ σ2 F + σ2 ei and, σi,j = bi ∗ bj ∗ σ2 F Two-factors model: σ2 i = b2 i1 ∗ σF 2 1 + b2 i2 ∗ σF 2 2 + σ2 ei / + 2 ∗ bi1 ∗ bi2 ∗ σF1,F2 Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models One-factor model: σ2 i = b2 i ∗ σ2 F + σ2 ei and, σi,j = bi ∗ bj ∗ σ2 F Two-factors model: σ2 i = b2 i1 ∗ σF 2 1 + b2 i2 ∗ σF 2 2 + σ2 ei / + 2 ∗ bi1 ∗ bi2 ∗ σF1,F2 and, σi,j = bi1∗bj1∗σ2 F1 +bi2∗bj2∗σ2 F2 Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models One-factor model: σ2 i = b2 i ∗ σ2 F + σ2 ei and, σi,j = bi ∗ bj ∗ σ2 F Two-factors model: σ2 i = b2 i1 ∗ σF 2 1 + b2 i2 ∗ σF 2 2 + σ2 ei / + 2 ∗ bi1 ∗ bi2 ∗ σF1,F2 and, σi,j = bi1∗bj1∗σ2 F1 +bi2∗bj2∗σ2 F2 / + (bi1 ∗ bj2 + bi2 ∗ bj1) ∗ σF1,F2 Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models Multi-factors model: Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models Multi-factors model: ri = ai + n k=1 bik ∗ Fk + ei Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models Multi-factors model: ri = ai + n k=1 bik ∗ Fk + ei Thus, for portfolio, Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models Multi-factors model: ri = ai + n k=1 bik ∗ Fk + ei Thus, for portfolio, Rp = n i=1 wi ∗ (ai + bi1 ∗ F1 + bi2 ∗ F2 + · · · + bik ∗ Fk + ei ) Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models Multi-factors model: ri = ai + n k=1 bik ∗ Fk + ei Thus, for portfolio, Rp = n i=1 wi ∗ (ai + bi1 ∗ F1 + bi2 ∗ F2 + · · · + bik ∗ Fk + ei ) and: Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Equation of models Multi-factors model: ri = ai + n k=1 bik ∗ Fk + ei Thus, for portfolio, Rp = n i=1 wi ∗ (ai + bi1 ∗ F1 + bi2 ∗ F2 + · · · + bik ∗ Fk + ei ) and: σ2 p = n i=1 n j=1 wi ∗ wj ∗ σi,j Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Factor vs. Non-factor risk One-factor model: Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Factor vs. Non-factor risk One-factor model: σ2 p = n i=1 w2 i (b2 i ∗ σ2 F + σ2 ei ) Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Factor vs. Non-factor risk One-factor model: σ2 p = n i=1 w2 i (b2 i ∗ σ2 F + σ2 ei ) Factor (systematic) risk Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Factor vs. Non-factor risk One-factor model: σ2 p = n i=1 w2 i (b2 i ∗ σ2 F + σ2 ei ) Factor (systematic) risk b2 p ∗ σ2 F = ( n i=1 wi ∗ bi )2 ∗ σ2 F Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Factor vs. Non-factor risk One-factor model: σ2 p = n i=1 w2 i (b2 i ∗ σ2 F + σ2 ei ) Factor (systematic) risk b2 p ∗ σ2 F = ( n i=1 wi ∗ bi )2 ∗ σ2 F Non-factor (unsystematic) risk Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Factor vs. Non-factor risk One-factor model: σ2 p = n i=1 w2 i (b2 i ∗ σ2 F + σ2 ei ) Factor (systematic) risk b2 p ∗ σ2 F = ( n i=1 wi ∗ bi )2 ∗ σ2 F Non-factor (unsystematic) risk σ2 ep = n i=1 w2 i ∗ σ2 ei Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Content 1 Multi-factor models 2 Arbitrage Pricing Theory - APT 3 Estimation of parameters Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Model APT This is a special case of a multi-factors model Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Model APT This is a special case of a multi-factors model In general: Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Model APT This is a special case of a multi-factors model In general: σri ,rj = cov ai + n k=1 bik ∗ Fk + ei ; aj + n m=1 bjm ∗ Fm + ej Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Model APT This is a special case of a multi-factors model In general: σri ,rj = cov ai + n k=1 bik ∗ Fk + ei ; aj + n m=1 bjm ∗ Fm + ej Assumptions for ATP: E(ei ) = 0, i = 1, 2, 3, . . . , n Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Model APT This is a special case of a multi-factors model In general: σri ,rj = cov ai + n k=1 bik ∗ Fk + ei ; aj + n m=1 bjm ∗ Fm + ej Assumptions for ATP: E(ei ) = 0, i = 1, 2, 3, . . . , n σFk ,Fm = 0, m̸= k, m,k=1, 2, 3, . . . , K Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Model APT This is a special case of a multi-factors model In general: σri ,rj = cov ai + n k=1 bik ∗ Fk + ei ; aj + n m=1 bjm ∗ Fm + ej Assumptions for ATP: E(ei ) = 0, i = 1, 2, 3, . . . , n σFk ,Fm = 0, m̸= k, m,k=1, 2, 3, . . . , K σei ,ej = 0, i̸= j, i,j=1, 2, 3, . . . , n Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Model APT This is a special case of a multi-factors model In general: σri ,rj = cov ai + n k=1 bik ∗ Fk + ei ; aj + n m=1 bjm ∗ Fm + ej Assumptions for ATP: E(ei ) = 0, i = 1, 2, 3, . . . , n σFk ,Fm = 0, m̸= k, m,k=1, 2, 3, . . . , K σei ,ej = 0, i̸= j, i,j=1, 2, 3, . . . , n σei ,Fk = 0, i = 1, 2, 3, . . . , n k = 1, 2, 3, . . . , K Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Merging CAPM & APT Beta coefficients and factor weights (sensitivity) Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Merging CAPM & APT Beta coefficients and factor weights (sensitivity) σri ,rm = σF1,rm ∗ bi1 + σF2,rm ∗ bi2 + σei ,rm Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Merging CAPM & APT Beta coefficients and factor weights (sensitivity) σri ,rm = σF1,rm ∗ bi1 + σF2,rm ∗ bi2 + σei ,rm ... Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Merging CAPM & APT Beta coefficients and factor weights (sensitivity) σri ,rm = σF1,rm ∗ bi1 + σF2,rm ∗ bi2 + σei ,rm ... Beta of factors: Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Merging CAPM & APT Beta coefficients and factor weights (sensitivity) σri ,rm = σF1,rm ∗ bi1 + σF2,rm ∗ bi2 + σei ,rm ... Beta of factors: βF1 = σF1,rm σ2 m , . . . Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Merging CAPM & APT Beta coefficients and factor weights (sensitivity) σri ,rm = σF1,rm ∗ bi1 + σF2,rm ∗ bi2 + σei ,rm ... Beta of factors: βF1 = σF1,rm σ2 m , . . . Thus, Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Merging CAPM & APT Beta coefficients and factor weights (sensitivity) σri ,rm = σF1,rm ∗ bi1 + σF2,rm ∗ bi2 + σei ,rm ... Beta of factors: βF1 = σF1,rm σ2 m , . . . Thus, βi = βF1 ∗ bi1 + βF2 ∗ bi2 + . . . Multi-factor models Arbitrage Pricing Theory - APT Estimation of parameters Content 1 Multi-factor models 2 Arbitrage Pricing Theory - APT 3 Estimation of parameters