Can one fruitfully model markets whose structures are changing rapidly? CEPR Research Fellow Michael Artis chaired a one-day workshop on Financial Modelling at the Centre on 21st May which addressed this question. It was attended by eighteen people interested in this area - academic economists, as well as those from the Bank of England, H.M. Treasury, the NIESR and the LBS modelling teams. Full-length papers were offered by David Vines and Martin Weale, George MacKenzie, Giles Keating, James Davidson and Christopher Green, together with shorter presentations by David Currie and Peter Spencer.

In the first paper, co-authored with James Meade and titled, 'Financial Policy and the Exchange Rate: A Model for Use in the Investigation of New Keynesian Policies', Martin Weale and Research Fellow David Vines outlined a simple model of the capital markets. Its purpose is to examine the inter- relationships between interest rate policy and exchange rate determination, thus permitting consideration of 'New Keynesian' policies. The model is constructed so as to be embedded eventually within the NIESR forecasting model.

'New Keynesian' policies comprise two distinct elements: the use of monetary and fiscal policy to target nominal GDP on a steady growth path, avoiding undue disequilibrium in the balance of payments or the government's financing requirements; and the reform of labour market institutions to make wage-bargaining reflect the employment consequences of wage settlements. Previous explorations of these policies had treated the exchange rate as an instrument which the government could vary. The model discussed by Vines and Weale linked the exchange rate to interest rates, thus making it endogenous.

Their model aroused much comment: on its high degree of aggregation - the model distinguished only between 'home' and 'foreign' assets; on its treatment of net rather than gross asset demands; and on its assumption of imperfect substitutability between foreign and domestic assets. Charles Goodhart and Peter Spencer also stressed the difficulty of modelling financial markets in the light of the substantial structural changes that have taken place in recent years.

Has the rapid growth in international banking and the size of the Eurocurrency market had any impact on the effectiveness of domestic monetary policy? George MacKenzie, in a paper co- authored with Stephen Thomas, argued that monetary policy, through its effect on non-monetary financial institutions and on the Eurocurrency system, could bring about variations in the velocity of money circulation which might offset the intended effect of the policy. To investigate this, MacKenzie and Thomas had developed a small-scale econometric model of the United Kingdom. The model emphasized the degree to which UK residents would shift their assets between foreign currency and sterling in response to changes in the rates of return on each. The results they obtained seemed to confirm the 'offsetting' hypothesis.

Much of the discussion that followed, however, was sceptical of the interpretation placed on the results and the two strong empirical propositions put forward by MacKenzie and Thomas in the paper. They claimed that an increase in the Treasury bill rate would increase GDP, while an increase in the Euro-dollar rate would have the opposite effect. These results were sharply criticised. Peter Spencer argued that Mackenzie and Thomas should not have interpreted their estimation results as evidence of substitutability of domestic and foreign currency - over the estimation period such substitution had been severely inhibited by foreign exchange controls! Others felt that the authors should examine the possibility that the perverse response of output to higher domestic interest rates might be the result of misspecification within the model structure. Their assumption that both the Treasury bill rate and the exchange rate should be treated as exogenous also seemed unjustifiable.

Giles Keating reported on recent work at the LBS on a highly disaggregated structural model of UK financial flows, embedded within their national income model. The financial model is very large, involving nine separate sectors and thirteen financial assets and liabilities. One of its distinctive features is the recognition of rationing by commercial banks and building societies in the markets for loans, together with the consequent spillover effects. Keating concluded with a discussion of some preliminary 'consistent-expectations' simulation results conducted on the LBS model augmented by the new monetary sector.

Several participants questioned the precise rationing device within the model. How were the periods of loan rationing identified? In the case of house loans by building societies, evidence for rationing is found in the length of mortgage queues; rationing of commercial bank loans is handled judgementally. What happened when rationing was relaxed in the model? Keating responded that, subject to transaction costs, relaxation of rationing would result in a return to portfolio equilibrium within the model.

'Error correction mechanisms' have recently enjoyed some success in econometric models of consumption and money demand. James Davidson used this approach in 'Money Disequilibrium: An Approach to Modelling Monetary Phenomena in the UK'. The paper is based on a specification of the money supply process which treats money as the residually-determined buffer stock of the financial system. Davidson models this process using a system of 'error correction' adjustment equations. In this framework, he argues that an exogenous shock causing an excess demand for money will stimulate either a direct offsetting adjustment of the money stock or an indirect offsetting effect that operates through 'desired velocity'.

In the discussion, some concern was expressed over the general methodology underlying Davidson's model specification. Peter Spencer was uneasy about the narrow specification of the buffer stock within the model. He argued vigorously that a financial disequilibrium would be reflected generally in asset behaviour, not just in that of money. It was agreed that this was a testable proposition, not a foregone conclusion in the Davidson model. Costs of transactions between various assets would determine whether the identification of disequilibria with money alone represented too great a simplification.

The final full-length paper was 'A Survey of the Demand and Supply of Bank Credit', presented by Christopher Green. The paper surveyed the theoretical issues underlying bank lending equations and compared the results of estimated equations with theoretical predictions. In his presentation, Green began by dividing the individual's decision to undertake a bank loan into a 'top-level' borrowing decision (to incur a debt) and a 'second- round' financing decision - the choice of a particular debt instrument. Using this somewhat artificial division he examined the predictions of the Fisherian and Wicksellian models of borrowing, which differ completely in regard to the determinants of borrowing and the effect of a change in interest rates. Moreover, general economic theory provided no guidance as to which of the two models is the more appropriate for empirical work. Similar conflicts were also shown to exist in the theory of corporate finance.

Green also discussed the importance of the market clearing mechanism for the interpretation of estimated equations. He noted that the usual procedure adopted in empirical work assumes that bank lending is demand determined or incorporates rationing through dummy variables. This involves a potential misspecification of the model structure.

The two shorter contributions involved a discussion of research undertaken by the NIESR and H.M. Treasury. David Currie outlined work, jointly undertaken with Gerald Kennally, to respecify the monetary sector of the NIESR forecasting model and to reconsider the role of monetary variables within the model. The methodology treats money as a disequilibrium buffer stock and emphasizes the role of transaction costs in portfolio management. With such costs, the individual must decide whether increases in wealth are permanent or transitory. If the latter, the increased wealth will be held in those liquid assets which allow the least costly realization.

Currie and Kennally use this approach in modelling corporate sector behaviour, where the buffer stock is taken to be corporate liquidity. They had implemented a similar approach for personal sector behaviour which suggested that financial disequilibrium should have spillover effects on personal consumption behaviour. Not only the level but also the composition of personal net worth would therefore be important.
Peter Spencer outlined the salient features of the financial sector of the H.M. Treasury forecasting model. An earlier version of the model took a 'general equilibrium' approach like Keating's LBS model, involving a substantial degree of disaggregation both by sector and asset. Recent policy changes, together with the effects of innovation and structural change in the financial sector, had led the model builders to simpler 'reduced form' approaches, with special emphasis being given to the form of the demand for money equation, bank lending and the modelling of expectations. The structural equations of the larger model were not expected to prove robust in the face of changes in financial market structures. Barry Johnston's development at the Treasury of indicators of financial innovation reflected the concern for these problems. The demand for money function incorporates a disequilibrium (wealth) term, which responds to some of the thinking reported to the workshop. Spencer argued that the effort involved in modelling expectations and making them consistent in simulation has been found to be rewarding, in the sense that model properties were significantly changed by the handling of expectations variables.

Did the workshop suggest new avenues of research? In summing up, Michael Artis noted that the coverage of the modelling strategies presented to the workshop had been extremely broad, ranging from the small-scale models of monetary disequilibrium to the very large-scale, disaggregated, portfolio models of the type discussed by Keating. Those professionally concerned with financial modelling appeared to judge that the pace of innovation and policy change made the payoff from ever-larger and more complex 'general equilibrium' models uncertain (though Keating's paper showed that such an approach was still alive and well at the LBS).

In the discussion of future research interests, two further areas were suggested. Marcus Miller noted that none of the discussion had focussed on the microeconomic aspects of financial modelling and was surprised that no attention had been given to the role of information technology in financial markets. It was suggested that this partly reflected the lack of available data at the level of individual financial institutions. Discussants were rather pessimistic regarding the likelihood of commercial banks releasing such micro data.

An alternative theme was suggested by David Vines who noted that there had been little discussion of the conduct of monetary policy. He suggested that researchers might usefully investigate the implications of the monetary authorities' adopting either a foreign exchange target or a monetary aggregate target in financial or foreign exchange markets that were subject to dramatic structural change or rapid financial innovation.