An application of the stone model to the Turkish financial accounts

The study presented in this paper consists of the application of two models of financial accounting to Turkish data. Detailed financial accounts are available for 1950–63 [Yaser, 1967], accounts for 1964–67 are incomplete [Aysan, 1967 and State Institute of Statistics, 1968], and none exist for 1968–70. Even if detailed accounts could be prepared somewhat more rapidly they would still not be up to date; a considerable lag in preparation of the basic data, e.g. company balance sheets, etc., exists in Turkey, as in most underdeveloped countries. For this reason, methods of estimating financial accounts with the use of models requiring limited and speedily available current data have been explored.

The aims of the analysis are estimation and prediction. Explanation of the changes in the financial variables has not been attempted here. The bivariate least‐squares regressions run to estimate linear time trends in all the financial proportions used in the models are not explanatory. Durbin‐Watson tests suggest that other factors were significant over the period.¹ This lends support to the conclusion that even for a financial system such as Turkey's which might on a priori grounds be thought particularly well suited to analysis by the Stone model, the non‐substitutability hypothesis embodied in it must be rejected.

The reasons for believing that the Turkish financial structure might lend itself well to analysis by the Stone model are outlined below. They appear so convincing that the negative results of the model's application are surprising. Indeed, they raise a number of interesting questions about financial structures of underdeveloped countries which are beyond the scope of this paper.