A generalization of the Friedlander Algorithm balancing of national accounts matrices
β Scribed by Kasper Bartholdy
- Publisher
- Springer US
- Year
- 1991
- Tongue
- English
- Weight
- 453 KB
- Volume
- 4
- Category
- Article
- ISSN
- 1572-9974
No coin nor oath required. For personal study only.
β¦ Synopsis
Owing to stochastic and systematic errors in test samples, the first estimates of national accounts items rarely satisfy all the well known definitional identities. Stone, Champernowne and Meade presented the initial argument for GLS balancing of such accounting matrices in 1942. The GLS principles were, however, rarely used in practical empirical work until Byron facilitated the computational process by applying the Conjugate Gradient Algorithm in 1978. This paper shows how a slight adjustment of another numerical process, the Friedlander Algorithm, generates an efficient and conceptually simple approach to the computation of a GLS-balanced set of accounts.
π SIMILAR VOLUMES
we propose a "fast" algorithm for the construction of a data-sparse inver'~ of a general Toeplitz matrix. The computational cost for inverting an N Γ N Toeplitz matrix equals the cost of four length-N FFTs plus an O(N)-term. This cost should be compared to the O(Nlog2N) cost of previously published