Why Small Scale Macroeconomic Model?

67 Recognizing some deficiencies in the MODBI, we feel that some improvements can be made. First of all, as we have argued earlier, the trend of economic modelling is to move towards the small-scale models. In contrast, MODBI consists of around 200 equations, for which analytical solutions are impossible to obtain 5 . Hence, it seems more productive to develop an SSMM for Indonesia. Secondly, MODBI was built in 1996 which meant the financial crisis as well as recent changes in the Indonesian exchange rate system are not captured. Lastly, the exclusion of a labour sector, e.g. the wage contracting specification, might create further problems in the future as correctly pointed out by the MODBI model-builders. However, in Chapter 4 we argue that wages might not be very important in explaining inflationary pressures in developing countries and especially Indonesia. 3.2 Why Small Scale Macroeconomic Model? Macroeconomic models have usually been developed in a large-scale setting comprising hundreds, or even thousands, of equations. This is meant to capture all possible components in the economy. While this substantial effort should be recognized, some pitfalls are nevertheless inevitable such as the partial failure of forecasts and some theoretical shortcomings of a large macroeconomic model Deaton 1981; Barker 1985; 5 Hence, Bank Indonesia has developed a stochastic dynamic macroeconomic model called General Equilibrium Model of Bank Indonesia GEMBI for policy simulation purposes. GEMBI employs modern stochastic dynamic macroeconomic theory. The model uses some stylized fact parameters to perform calibration and to provide long-term relationships of a number of key macroeconomic variables such as consumption, inflation, exchange rate, etc Warjiyo 2001. The purpose of GEMBI is for policy simulation that can provide the dynamic paths of the mentioned key macroeconomic variables in the medium- to long- term horizon based on simulation of some alternative policy rules. MODBI, SOFIE, and GEMBI complements one another so as to understand the behaviour and future conditions of the Indonesian macroeconomy. Specifically, SOFIE and MODBI are suitable for quarterly and annual analysis as well as forecasts. A combination between SOFIE and GEMBI can be used to provide the expected dynamic paths of key macroeconomic variables based on some policy rules. 68 Lawson 1992; Wallis 1989. Another drawback is mentioned by Holden 1989 who questioned whether policy simulations from large macroeconomic models really give insights into economies in the real world or merely demonstrate the properties and the beliefs of the model and the model-builders respectively. Yet, a macroeconomic model is an indispensable tool for policy analysis and gives valuable insights to the authorities for making decisions that affect the economy. In the early 1990s, small scale macroeconomic model SSMM emerged as an alternative tool for comparing the results obtained from larger macroeconomic models. An SSMM is deliberately kept small with a substantial level of aggregation, thus forcing the model to focus on key issues rather than to look at excessive details in the economy. An SSMM is characterized by a compact system of equations that describe the behaviour of key macroeconomic aggregates such as output GDP, inflation, money demand, exchange rates for a very open economy, interest rates, etc. Although the SSMM is not meant primarily for forecasting, it helps to clarify and understand the developments of key macroeconomic variables, for example those relevant in the process of determining the inflation rate in the economy. This is an advantage of an SSMM which is highly aggregated—it could help to mitigate the problem of poor fit in large macroeconomic models. This might explain the recent trend of employing SSMM in central bank modelling activities to counter-check the results from larger macroeconomic models. The aim of this study is to come up with an SSMM of a forward-looking nature that captures the characteristics of the Indonesian economy and which can be potentially 69 useful for carrying out policy analysis, in particular, analyzing the effects of monetary policy. The goal is to better understand the options available to Indonesia policy-makers to deal with economic shocks. A natural question to be asked is why we choose a small- scale macroeconomic model instead of a larger one in modelling the whole economy. There are several reasons why SSMMs are chosen in this research methodology: • SSMM is typically designed to carry out policy analysis. The simplicity and flexibility of the underlying structure of these models makes them ideally suited for this purpose. Note that Indonesia is a big economy in terms of area and inhabitants and an SSMM, in essence, is a small-scale model, not a model for small economy. SSMMs provide powerful insights for policy analysis too: it is relatively straightforward to trace the transmission of policy actions onto key variables such as output and inflation. • Estimation or calibration of parameters may be more straightforward for smaller models. For example, it is easier to impose some restrictions suggested by economic theory such as the vertical long-run Philips curve which can be explicitly accommodated by SSMMs so that it is possible to examine directly what the model implies with respect to the inflationoutput trade-off. • Because there are fewer parameters and equations in SSMMs than in larger models, it is easier to experiment with alternative assumptions, such as those concerning the 70 formation of expectations by economic agents or the degree of credibility in the central bank. • Exercises with SSMMs can function as a cross-check on results, such as policy simulations from larger and more sophisticated macroeconometric models. Although the aggregation inherent in SSMMs means that they are lacking in details, they are still often structural models. Hence, meaningful economic interpretations that are directly linked to policy options can be derived from SSMMs. • SSMMs are often small enough to allow their key properties to be investigated analytically. Some caveats, however, need to be acknowledged in using SSMMs as well, such as: • The high level of aggregation in SSMMs means that they lack detail about certain macroeconomic variables. For example, SSMMs do not provide a breakdown of aggregate demand into expenditure categories such as consumption, investment, or net exports. One consequence is that it will provide little information about movements in the balance of payments. • Since SSMMs abstract from the detailed and multiple lag dynamics that are typically required to describe quarterly macroeconomic data accurately, they will be less 71 suitable for complex macroeconomic forecasting and simulations as compared to large-scale macroeconometric model. To sum up, an SSMM is particularly useful for carrying out policy simulations and experiments that can provide valuable analytical insights. The SSMM is simple and tractable and can also provide a cross-check on the results of large macroeconometric models. Some other assumptions suggested by economic theory can also be imposed in SSMM in order to investigate the implications for policy. It is hoped that as the research in this area makes progress, more insights and modifications and also additional features can be implemented to improve the SSMM further. This relatively new variant of macroeconomic models can benefit Indonesian authorities by providing better analyses for policymaking and implementation.

3.3 The Variants of SSMM