RDP 7902: Financial Modelling in Australia 4. What has been Learnt?

The following points highlight the characteristics of the Australian financial sector studies. They discuss some of the lessons which have been learnt from them.

i) with the exception of RBA76 the full model econometric studies express the financial interactions of the household-corporate group in terms of a simple modified stock adjustment model. The holdings of the asset with the lowest cost of adjustment are determined as the residual asset in the formation table identity. This choice of the asset with lowest adjustment costs as the residual asset (which is money or some component of money) is supported by some empirical evidence although no strong theoretical support has been provided. However, the modelling of money holdings in this way, provides a more convincing explanation of the large fluctuations in money during the 1970's than the single equation studies which treat money as demand determined.

ii) the specifications of the sectoral studies of the non-bank groups are much more general than the simple stock adjustment model. Unlike the full model studies, these relationships may be efficiently estimated although they suffer badly from multicol-linearity and low degrees of freedom. The evidence from these studies for the asset with the lowest cost of adjustment being the residual asset is not conclusive. Because of the statistical and data problems with the sectoral studies care must be exercised in interpreting their results. Nevertheless, there is support from some of these studies for the integration of the financial asset, real asset, and expenditure decisions through both relative prices and quantities.

iii) RBA76 is the only full model study which uses a more general adjustment process than simple modified stock adjustment. In this model, evidence is found for the importance of the spillover effect of monetary disequilibrium in the expenditure and asset holdings decisions. This evidence is consistent with the general approach of some of the sectoral studies, although no strong theoretical case has been established for the unique importance of monetary disequilibrium.

iv) the role for money directly affecting prices has been outlined and tested in a full model context in RBA76. The evidence here is a major potential addition to our knowledge of the effect of the financial sector on the dynamics of the Australian economy.

v) the simulation properties of the full economy models with reasonably complete financial sectors illustrate that their short-run properties may be quite similar although their long-run properties differ markedly.

vi) simulation experiments with RBA1/74 suggest that the links between wealth and real activity are not strong.

vii) simulation experiments with RBA76 illustrate an important link between money and prices. They also suggest that money can be controlled within 2–4 quarters by appropriate variation in interest rates.

viii) estimates of interest rate and income elasticities are surprisingly unstable. Income elasticities for government securities appear to be falling as the data covers more of the 1970's, whilst interest rate elasticities for both government securities and fixed deposits have risen.

ix) estimates suggest that adjustment speeds in financial markets are quite slow.

x) there has been some progress in explaining short term market determined interest rates through the supply reactions of the trading banks, whilst bond rates have been modelled in terms of policy reaction functions. As additional observations from the 1970's, when market forces have played a more important role in determining interest rates than the 1960's, become available, further success in explaining short term interest rates can be expected.

xi) there is possibly conflicting evidence regarding the influence of the international sector on Australian financial markets. The only full model study which directly addresses the issue finds quite weak offsetting effects from international capital flows on the domestic monetary policy for given exchange rate expectations. However, a stimulating reduced form study finds evidence to the contrary. Perhaps the most important influence which has not yet been adequately explained by the model-builders is the explanation of exchange rate expectations and their role in domestic financial markets. (As noted, the role of expectations may provide the eventual reconciliation of these two studies).

xii) although studies have shown the Australian capital markets to be inefficient only one study has identified a satisfactory influence for interest rate expectations on portfolio allocations.

xiii) no attention has been given to the disaggregation of the corporate-household sector in the full model studies. This will be necessary before adequate explanation of the role of the equity market is possible. However, before this can be done there are considerable data problems to be overcome.

xiv) the equilibrium studies of the household/corporate groups suffer from high autocorrelation. One possible explanation for this is that the appropriate way of explaining portfolio decisions is with disequilibrium models.

xv) the full model studies all treat liquid assets as the banks' residual asset. This choice is supported by the simulation evidence of a sectoral study of the savings banks.

xvi) studies which include data from the 1970's find bank advances to be essentially supply determined, whilst those of the 1960's suggest that trading bank advances are essentially demand determined. Given the special overdraft arrangements of the Australian trading banks the best approach to explaining bank advances is probably that which models the specific interaction of supply and demand.