Uses and requirements for various equation systems. *(p, d, q) denotes an ARMA model for the dth differences of a variable that has AR polynomial of degree p and MA polynomial of degree y. Estimated final equations for consumption. "(p, d, q) denotes an ARMA model for the ath differences of a variable that has AR polynomial of degree p and MA polynomial of degree g. Estimated transfer functions for consumption. Results of large sample likelihood ratio tests applied to transfer functions of Haavelmo’s model. To explore this last point more systematically, some alternative transfer function models, formulated without taking into account results of final equation analyses, have been estimated with results shown in tables 7 and 8. For compari- son, results with models implied by the final equation analyses are also presented. A quick look at the residual sum of squares (RSS) indicates that for consump- tion, alternative model M.; yields about a 12 per cent reduction in RSS relative to M, while M, yields about a 20 per cent reduction. For income, Ms yields about a 3 per cent reduction in RSS relative to M, while M, provides the lowest RSS, about 6 per cent lower than for M,. For nested models, a large sample likelihood ratio test procedure has been employed to compare alternative formulations with results reported in table 9. Final equation and transfer function models for dynamized Haavelmo model.”
