Advertisement

Numerical and Computational Strategies for Solving Seemingly Unrelated Regression Models

  • Paolo Foschi
  • Lucien Garin
  • Erricos J. Kontoghiorghes
Part of the Applied Optimization book series (APOP, volume 74)

Abstract

Computationally efficient and numerically stable methods for solving Seemingly Unrelated Regression (SUR) models are proposed. The iterative feasible generalized least squares estimator of SUR módels where the regression equations have common exogenous variables is derived. At each iteration an estimator of the SUR model is obtained from the solution of a generalized linear least squares problem. The proposed methods, which have as a basic tool the generalized QR decomposition (GQRD), are also found to be efficient in the general case where the number of linear independent regressors is smaller than the number of observations.

Parallel strategies based on compound disjoint Givens rotations are designed for computing the main two factorizations that are used in the GQRD. The first factorization requires the triangularization of a set of upper-triangular after deleting columns. The second factorization is equivalent in updating a lower-triangular matrix with a matrix having a block lower-triangular structure. Theoretical measures of complexity and examples are used for comparing and investigating the various parallel strategies.

Keywords

SUR models Matrix factorizations Kronecker products Parallel algorithms 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. [Anderson et al., 1992]
    Anderson, E., Bai, Z., and Dongarra, J. J. (1992). Generalized QR factorization and its applications. Linear Algebra and its Applications, 162: 243–271.CrossRefGoogle Scholar
  2. [Andrews and Kane, 1970]
    Andrews, H. C. and Kane, J. (1970). Kronecker matrices, computer implementation, and generalized spectra. Journal of the ACM, 17 (2): 260–268.CrossRefGoogle Scholar
  3. [Belsely et al., 2002]
    Belsely, D. A., Foschi, P., and Kontoghiorghes, E. J. (2002). Numerical estimation of seemingly unrelated regression models. Computational Economics. (To be submitted for publication).Google Scholar
  4. [Blackford et al., 1997]
    Blackford, L. S., Choi, J., Cleary, A., D’Azevedo, E., Demmel, J., Dhillon, I., Dongarra, J., Hammarling, S., Henry, G., Petitet, A., Stanley, K., Walker, D., and Whaley, R. (1997). ScaLAPACK Users’ Guide. SIAM, Philadelphia.CrossRefGoogle Scholar
  5. [Cosnard and Daoudi, 1994]
    Cosnard, M. and Daoudi, M. (1994). Optimal algorithms for parallel Givens factorization on a coarse—grained PRAM. Journal of the ACM, 41 (2): 399–421.CrossRefGoogle Scholar
  6. [Cosnard et al., 1986]
    Cosnard, M., Muller, J.-M., and Robert, Y (1986). Parallel QR decomposition of a rectangular matrix. Numerische Mathematik, 48: 239–249.CrossRefGoogle Scholar
  7. [Dhrymes, 1994]
    Dhrymes, P. J. (1994). Topics in Advanced Econometrics, volume Vol. 2: Linear and Nonlinear Simultaneous Equations. Springer—Verlag, New York.CrossRefGoogle Scholar
  8. [Foschi and Kontoghiorghes, 2001]
    Foschi, P. and Kontoghiorghes, E. J. (2001). Estimation of VAR(p) models: computational aspects. Computational Economics. (In press).Google Scholar
  9. [Foschi and Kontoghiorghes, 2002]
    Foschi, P. and Kontoghiorghes, E. J. (2002). Estimation of seemingly unrelated regression models with unequal size of observations: computational aspects. Computational Statistics and Data Analysis. (forthcoming).Google Scholar
  10. [Garin, 1999]
    Garin, L. (1999). The QR decomposition of trapezoidal matrices after deleting columns. Diploma Thesis, Institut d’Informatique, Université de Neuchâtel, Switzerland.Google Scholar
  11. [Gatu and Kontoghiorghes, 2002]
    Gatu, C. and Kontoghiorghes, E. J. (2002). Parallel algorithms for computing all possible subset regression models using the QR decomposition. Parallel Computing. (forthcoming).Google Scholar
  12. [Graham, 1986]
    Graham, A. (1986). Kronecker products and matrix calculus: with applications. Ellis Horwood Series in Mathematics and its Applications. Chichester: Ellis Horweod Limited, Publishers; New York: Halsted Press, a division of John Wiley & Sons.Google Scholar
  13. [Kontoghiorghes, 1995]
    Kontoghiorghes, E. J. (1995). New parallel strategies for block updating the QR decomposition. Parallel Algorithms and Applications, 5 (42): 229–239.CrossRefGoogle Scholar
  14. [Kontoghiorghes, 1999]
    Kontoghiorghes, E. J. (1999). Parallel strategies for computing the orthogonal factorizations used in the estimation of econometric models. Algorithmica, 25: 58–74.CrossRefGoogle Scholar
  15. [Kontoghiorghes, 2000a]
    Kontoghiorghes, E. J. (2000a). Parallel Algorithms for Linear Models: Numerical Methods and Estimation Problems, volume 15 of Advances in Computational Economics. Kluwer Academic Publishers, Boston, MA.Google Scholar
  16. [Kontoghiorghes, 2000b]
    Kontoghiorghes, E. J. (2000b). Parallel Givens sequences for solving the general linear model on a EREW PRAM. Parallel Algorithms and Applications, 15 (1–2): 57–75.CrossRefGoogle Scholar
  17. [Kontoghiorghes, 2000e]
    Kontoghiorghes, E. J. (2000c). Parallel strategies for solving SURE models with variance inequalities and positivity of correlations constraints. Computational Economics, 15 (42): 89–106.CrossRefGoogle Scholar
  18. [Kontoghiorghes and Clarke, 1993a]
    Kontoghiorghes, E. J. and Clarke, M. R. B. (1993a). Parallel reorthogonalization of the QR decomposition after deleting columns. Parallel Computing, 19 (6): 703–707.CrossRefGoogle Scholar
  19. [Kontoghiorghes and Clarke, 1993b]
    Kontoghiorghes, E. J. and Clarke, M. R. B. (1993b). Solving the updated and downdated ordinary linear model on massively parallel SIMD systems. Parallel Algorithms and Applications, 1 (2): 243–252.CrossRefGoogle Scholar
  20. [Kontoghiorghes and Clarke, 1993e]
    Kontoghiorghes, E. J. and Clarke, M. R. B. (1993c). Stable parallel algorithms for computing and updating the QR decomposition. In Proceedings of the IEEE TENCON’93, pages 656–659, Beijing. International Academic Publishers.Google Scholar
  21. [Kontoghiorghes and Clarke, 1995a]
    Kontoghiorghes, E. J. and Clarke, M. R. B. (1995a). An alternative approach for the numerical solution of seemingly unrelated regression equations models. Computational Statistics & Data Analysis, 19 (4): 369–377.CrossRefGoogle Scholar
  22. [Kontoghiorghes and Clarke, 1995b]
    Kontoghiorghes, E. J. and Clarke, M. R. B. (1995b). Solving the general linear model on a SIMD array processor. Computers and Artificial Intelligence, 14 (4): 353–370.Google Scholar
  23. [Kontoghiorghes et al., 1996]
    Kontoghiorghes, E. J., Clint, M., and Dinenis, E. (1996). Parallel strategies for estimating the parameters of a modified regression model on a SIMD array processor. In Prat, A., editor, COMPSTAT,Proceedings in Computational Statistics, pages 319–324. Physical Verlag.Google Scholar
  24. [Kontoghiorghes and Dinenis, 1996]
    Kontoghiorghes, E. J. and Dinenis, E. (1996). Solving triangular seemingly unrelated regression equations models on massively parallel systems. In Gilli, M., editor, Computational Economic Systems: Models, Methods & Econometrics,volume 5 of Advances in Computational Economics,pages 191–201. Kluwer Academic Publishers.Google Scholar
  25. [Kontoghiorghes and Dinenis, 1997]
    Kontoghiorghes, E. J. and Dinenis, E. (1997). Computing 3SLS solutions of simultaneous equation models with a possible singular variance—covariance matrix. Computational Economics, 10: 231–250.CrossRefGoogle Scholar
  26. [Kourouklis and Paige, 1981]
    Kourouklis, S. and Paige, C. C. (1981). A constrained least squares approach to the general Gauss—Markov linear model. Journal of the American Statistical Association, 76 (375): 620–625.CrossRefGoogle Scholar
  27. [Modi and Clarke, 1984]
    Modi, J. J. and Clarke, M. R. B. (1984). An alternative Givens ordering. Numerische Mathematik, 43: 83–90.CrossRefGoogle Scholar
  28. [Paige, 1978]
    Paige, C. C. (1978). Numerically stable computations for general univariate linear models. Communications on Statistical and Simulation Computation, 7 (5): 437–453.CrossRefGoogle Scholar
  29. [Paige, 1979]
    Paige, C. C. (1979). Fast numerically stable computations for generalized linear least squares problems. SLIM Journal on Numerical Analysis, 16 (1): 165–171.CrossRefGoogle Scholar
  30. [Paige, 1990]
    Paige, C. C. (1990). Some aspects of generalized QR factorizations. In Cox, M. G. and Hammarling, S. J., editors, Reliable Numerical Computation, pages 71–91. Clarendon Press, Oxford, UK.Google Scholar
  31. [Regalia and Mitra, 1989]
    Regalia, P. A. and Mitra, S. K. (1989). Kronecker products, unitary matrices and signal processing applications. SIAM Review, 31 (4): 586–613.CrossRefGoogle Scholar
  32. [Sameh and Kuck, 1978]
    Sameh, A. H. and Kuck, D. J. (1978). On stable parallel linear system solvers. Journal of the ACM, 25 (1): 81–91.CrossRefGoogle Scholar
  33. [Srivastava and Dwivedi, 1979]
    Srivastava, V. K. and Dwivedi, T. D. (1979). Estimation of seemingly unrelated regression equations Models: a brief survey. Journal of Econometrics, 10: 15–32.CrossRefGoogle Scholar
  34. [Srivastava and Giles, 1987]
    Srivastava, V. K. and Giles, D. E. A. (1987). Seemingly Unrelated Regression Equations Models: Estimation and Inference (Statistics: Textbooks and Monographs), volume 80. Marcel Dekker, Inc.Google Scholar
  35. [Zellner, 1962]
    Zellner, A. (1962). An efficient method of estimating seemingly unrelated regression equations and tests for aggregation bias. Journal of the American Statistical Association, 57: 348–368.CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media Dordrecht 2002

Authors and Affiliations

  • Paolo Foschi
    • 1
  • Lucien Garin
    • 1
  • Erricos J. Kontoghiorghes
    • 1
  1. 1.Institut d’informatiqueUniversité de NeuchâtelNeuchâtelSwitzerland

Personalised recommendations