V Applications of MULTIMOD

International Monetary Fund
Published Date:
July 1990
  • ShareShare
Show Summary Details

This section gives an overview of the range of applications of MULTIMOD by referring to some of the papers that have used it. As mentioned in the Introduction, the model was designed to analyze alternative policies on the part of industrial countries in the context of the World Economic Outlook. The medium-term issues that it has been used to examine include (1) what might be the consequences of financial tensions in exchange markets that result from continuing external imbalances, and how could policy changes help to mitigate them? (2) what would be the effects of fiscal adjustment in those major industrial countries with large budget deficits? and (3) how can inflationary pressures be counteracted by a combination of monetary and fiscal policy changes? Other scenarios that have been performed have involved the effects of oil price changes and the consequences of increases in financing flows to developing countries.

Going beyond the World Economic Outlook, the model has also been used in the consultations with member countries, and in research papers, to analyze the degree of interdependence among industrial country policies, and the extent to which different policies can lessen unfavorable spillovers. It has also been used to analyze the transmission of industrial country policies to developing countries in Masson and Helliwell (1989), which highlights the key roles of constraints on external financing and of demand for developing country exports.

Turning to the interdependence among industrial country policies, the degree of transmission of policy changes from one country to another was considered in Frenkel, Goldstein, and Masson (1988a) (Tables 921 give results for the current version of the model). How a country can best respond to various shocks hitting its economy was analyzed using MULTIMOD in Frenkel, Goldstein, and Masson (1988b, 1989a). In particular, shocks to such variables as aggregate consumption, investment, exports, and the exchange rate will have different effects depending on the policy regime. Some simple alternatives were considered: targeting a monetary aggregate or nominal GNP, fixed nominal exchange rates, or more complicated rules that use both monetary and fiscal policy to target either the real effective exchange rate or the current balance on the one hand, and some measure of domestic activity on the other. The results showed that the nature of the shock had an important influence on the appropriate response: no simple policy rule did best for all the shocks.

Since the relative importance of the different shocks is therefore a key determinant in evaluating policy rules, a natural extension was to evaluate the performance of rules when shocks to all variables are considered, and when they take on values consistent with their historical distributions. An application of the method of stochastic simulations to compare simple policy rules is contained in Frenkel, Goldstein, and Masson (1989b). Many more simulations are needed for stochastic simulations, because the path expected for future variables has to be revised each successive period in the light of shocks that have already occurred. Moreover, to calculate means and variances to a fair degree of precision, many replications have to be made, each time with a new drawing from the distribution of shocks. Although one needs an explicit objective function to rank the policies, some general conclusions emerged. Fixity of exchange rates did not unambiguously reduce macroeconomic variability in the economy; moreover, without active use of fiscal policy, achieving close control over the real effective exchange rate was impossible. Given this limitation, the results nevertheless suggested that it was better to use the exchange rate as an external target than the current account balance—at least on a year-to-year basis.29

If the choice has been made to target nominal exchange rates (as is true within the European Monetary System), and as a result, monetary policy is constrained, the question arises about the degree to which fiscal policy should be used flexibly. This issue was analyzed in Masson and Melitz (1990), which considered various shocks that France and Germany might face in the context of a closer monetary union between the two countries (and other EMS countries). It was argued that because certain shocks affect the two countries differently, and because of different preferences in the two countries, retaining a degree of fiscal policy flexibility would be desirable—rather than imposing constraints on national fiscal policies.

The model, because it includes expectations that are consistent with its solution values in later periods, is well suited to evaluate the effects of policies that are announced and credible. However, it is quite possible that the private sector does not believe that announced policies will be carried out, either because it believes that the authorities have goals that differ from their announced ones or because it expects that shocks may deflect them from their targets. Even though the model is set up to impose equality between expectations and model solutions, it can also be solved in a way that permits expectations to take into account the possibility of alternative outcomes. Hooper, Symansky, and Tryon (1989) simulate paths for monetary and fiscal variables that differ from those expected, and show how the effects of policies depend on the degree of credibility. Masson and Symansky (1990) endogenize the degree of credibility, making it depend on the distribution of outcomes and a postulated objective function on the part of the authorities: the latter are assumed to abandon the rule if shocks make it optimal for them to do so, and credibility reflects this possibility. More work in this area is planned.

The medium-term horizon for which the model was constructed was pushed back in some analysis of the long-term demographic developments that are projected to occur in industrial countries. In most countries, the first few decades of the next century will likely see a shift to a markedly older population, with fewer people of working age supporting a larger number of retirees. In Masson and Tryon (1990),30 MULTIMOD was simulated to 2040 (though only results to the year 2025 were reported; the later years were needed to minimize the effects of arbitrary terminal conditions). To perform these simulations, some changes (which are included in the description above) were made to the model so that the age structure of the population affected aggregate demand and supply. The simulation results suggested a substantial effect on all countries through higher real interest rates caused by lower saving, and through different current account developments, which derived from differences in the extent of population aging.

Another long-term issue that has been treated using MULTIMOD relates to the sustainability of policies. Chadha and Symansky (1990) use the model to examine how external imbalances might feed back onto interest rates and exchange rates. Unsustainable policies might therefore cause anticipatory reactions in financial markets.

Other issues have been addressed using INTERMOD, a variant of MULTIMOD that embodies both model-consistent and adaptive expectations (see Helliwell and others (1988) and Meredith (1989b)). The adaptive expectations framework allows the analysis of policy actions when expectations adjust gradually to observed changes in economic variables, as opposed to reflecting the full model solution for future values of the endogenous variables. Meredith (1989a) looks at policy innovations in the Canadian context: the results support the theoretical proposition that the effects of fiscal policy on real activity are reduced when expectations are forward looking. The effects of monetary policy, on the other hand, are magnified, as the exchange rate “jumps” by more when agents incorporate the future effects of monetary policy on exchange rates and interest rates. Bryant, Helliwell, and Hooper (1989) look at the predictions of INTERMOD and MULTIMOD, as well as other models, for the effects of U.S. Government spending cuts on domestic and foreign activity. One conclusion based on simulations with INTERMOD is that the real output loss associated with fiscal contraction is substantially reduced when fiscal policy is both preannounced and credible.

This short summary of applications of the model demonstrates its flexibility. New uses have also suggested extensions to the model. Like all tools, it is better for some purposes than others, and one should not make the mistake of expecting it to give sensible answers to all questions. Nevertheless, it has proved a useful framework for addressing a range of empirical macroeconomic questions.

    Other Resources Citing This Publication