This paper begins by discussing what ought to be the approach to the problem of forecasting in the medium term. The author emphasizes the advantages of a very detailed breakdown of sectors. He then considers certain econometric problems which arise when forecasting consumption in the medium term. He contrasts the formulation of the consumer sector in CANDIDE with that of two other types of model and compares the results of each.
The author proposes an alternative way of treating consumption in CANDIDE which would take into account different categories of households. Finally, he discusses how to resolve problems of organization of the data, of modelling, and of estimation which may arise when using such an approach.
In the first part, a synoptic review of the residential sectors in the balkanizing field of macro-econometric models is made à la M. Nerlove. The structural characteristics of most canadian and some american models are compared.
Then, a detailed description of two versions (1.0 and 1.1) of the residential sector of CANDIDE is given and a non parametric evaluation presented. Some comparative parametric discussions are also introduced.
The main weakness are stressed and some improvements are suggested.
The residential sector of CANDIDE is generally adequately specified generating appropriate parameters and is relatively accurate in its medium term projections. Nevertheless, the model can be of little value as a policy instrument since it is not regionalized (provincialized) and does not include policy variables in a sufficiently explicit manner.
The purpose of this paper is to comment on the investment equations of the CANDIDE model. The formulation of these equations is based on Jorgenson's theory of investment which is frequently used as a basis for investment expenditures analysis.
In the first two sections the authors endeavour to situate Jorgenson's model with respect to other approaches to investment theory. The last two sections contain, first, an evaluation of the way the Jorgenson's model has been adapted and, secondly, an evaluation of the problem of estimation and stability of the investment model.
The CANDIDE model referring to the traditional stock adjustment model to specify its inventory investment equations, the comment stresses the difficulty of that kind of model to explain and predict correctly the behaviour of inventory holdings. After pointing out the incredible loan adjustment lags usually associated with the stock adjustment model, and also obtained with CANDIDE, the comment draws the attention on the annual data basis of CANDIDE and the fact that the behaviour of inventory holdings is most likely to be better observed with monthly data. Some references to alternative models of inventory holdings and to the temporal decision problem are also given.
The following ill-tempered article provides a review of the foreign sector of CANDIDE; after briefly commenting the disagregation by products and services, the author examines the geographical disagregation of imports and exports into and from Canada between the U.S. and the Rest-of-the-World. He found that the proposed CANDIDE procedure is not after all that much meaningful and does not bring any additional information, while being dangerous as far as the forecasters using this model hope to get good results for imports from and exports to the Rest-of-the-World. Finally, the author gives some hints about what could have been alternatively done about this problem of geographical disagregation.
In this paper we deal with the financial sector of CANDIDE 1.1. We are concerned with the determination of the short-term interest rate, the term structure equations, and the channels through which monetary policy influences the real sector.
The short-term rate is determined by a straightforward application of Keynesian liquidity preference theory. A serious problem arises from the directly estimated reduced form equation, which implies that the demand for high powered money, but not the demand for actual deposits, is a stable function of income and interest rates. The structural equations imply the opposite.
In the term structure equations, allowance is made for the smaller variance of the long-term rates, but insufficient explanation is given for their sharper upward trend. This leads to an overstatement of the significance of the U.S. long-term rate that must perform the explanatory role. Moreover a strong structural hierarchy, by which the long Canada rate wags the industrial rate, is imposed without prior testing.
In CANDIDE two channels of monetary influence are recognized: the costs of capital and the availability of credit. They affect the business fixed investment and housing sectors. The potential of the personal consumption sector is not recognized, the wealth and real balance effects are bypassed, the credit availability proxy is incorrect, the interest rate used in the real sector is nominal rather than real, and the specification of the housing sector is dubious.
The paper is a brief critical evaluation of the treatment of the "supply of labour" in the CANDIDE model. The authors note that the "supply of labour block" does not occur to have retained much of the effort of the CANDIDE developing team. With regards to the specification of the equation of supply of labour, no attempts have been made to integrate new developments in labour economics (a better treatment of the secondary workers supply of labour, a Fisherian utility maximising approach). We note certain econometric difficulties which have not been overcome: The limited dependant variable problem, the lag structure problem…
Finally, though it is encouraging to see attempts in macro-economic models to integrate explicitly a labour demand and supply block with a certain disagregation, important economic problems in the labour market can be dealt only with much more flexible and disagregated models than CANDIDE.