Functional Form in Finished Goods Inventory Investment

The Maccini and Rossana (1984) model of finished goods investment is tested for functional form and autocorrelation as suggested by Savin and White (1978). The Maccini and Rossana model employs a multivariate flexible accelerator equation with a log-linear function and can be adapted to the Savin-White procedure by including a lagged dependent variable as one of the regressors on which the Box-Cox transformation is applied. The analysis uses data on durable, nondurable, and total manufacturing sectors. The same data were used by Maccini and Rossana in the 1984 analysis. The present analysis indicates that a log-linear specification of inventory adjustment processes would be statistically incorrect for the data in the study. Misspecification of the functional form can lead to errors in conclusions about the extent of serial correlation problems. A linear inventory demand equation would be more appropriate for use with the data in this study.