The Macsim project: macroeconomic structure
Jean Louis Brillet, Gilbert Cette, Ian Gambini
Introduction
In this short paper we shall present the economic structure of the MacSim model, used by the MacSim package.
A following document will describe the package itself, both on terms of logic and practical use. It will be authored by all the project participants, and mostly written by Gilbert Cette.
It will be available before the actual presentation, which will take place “live” during the MacSim session.
The goal of the MacSim project is to produce an educational software illustrating key concepts and aspects of traditional macroeconomics, focusing on international trade. It is based on a set of simplified representations of a set of countries of the European Union, plus the USA and Japan. To a formalization of the major internal mechanisms, including some financial elements, it adds a treatment of trade in the form of bilateral flows, including the rest of the world.
The simulations will be carried out as shocks, which will focus on economic and fiscal policies, but also financial. On this occasion, the method of determination of some concepts of interest and exchange rates may be chosen by the user.
Such a tool will allow in particular:
•To enhance the reasoning on traditional single-country exercises, by taking into account exchanges between partners, and feedbacks on the local economy.
•To highlight the individual and global consequences of concerted or competitive policies.
•To show how the choicein the financial rules affects the efficiencies of policies.
We will establish the objectives sought, and the way we treat them. We will present the general structure of the national models, the methodology of linkages between models, and the various options available for financial elements. We will go into details for some equations.
The list of eleven countries covered are as follows, in alphabetical order:
Belgium, France, Germany, Great Britain, Italy, Japan, Luxemburg, Netherlands. USA, and a reunion of the 6 remaining countries in the Euro-12 group (Austria, Finland, Greece, Ireland, Portugal, Spain) .
The primary reasons for this choice is the economic importance of the countries. However, to show the impact of country size on their properties and sensitivity to policies, we have also introduced Belgium and Luxemburg.
Startingoptions
Our technical goal was to establish a national system linking models in limited numbers describing the economies of the main industrialized developed countries. This system had to meet three basic principles: technical, gameplay and realism of properties. More specifically, our model had to :
•be sufficiently small and simple enough in structure to be solved quickly, without requiringtoo powerful tools.
•be based on real data and spontaneously follow past trends in variables, providing an acceptable path on future periods.
•present realistic economic properties comparable to operational, larger models.
•differentiate the countries covered, to improve playability.
•but not necessarily be based on full econometric tests.
Indeed, our main objective is to create a simplified multinational model, highlighting a number of mechanisms and concepts. However, we leave it to outside experts to judge the quality of the result, compared to the real operational tools in this field.
How to treat the problem
Let us one by one the objectives mentioned above.
•The size and feasibility did not face too much of a problem. As we will see, our system does not exceed a few hundred equations (currently 1683). Experience shows that professional software solves any model of this size in fractions of seconds per cycle. Our program is dedicated to a single issue: it should go even faster, provided we adopt an efficient algorithm. Size faces the same remark, as a simple calculation shows that the commonly available memory capacity is more than sufficient.
The tests of the complete model have been, carried either on a specialized software (EViews) or through specific programming (we used an algorithm designed by the author), haveproven quite successful in this regard.
•The production of the data set does not pose too many problems either: OECD provides a set of information (Economic Outlook) for each of the national models with the desired structure. The only difficulties concern bilateral trade and financial data. For the former, we had to access a complete structure of the countries concerned, at least for one year close to the simulation periods, if not all. Several sources are available (Eurostat, and CEPII’s Chelem), but problems with nomenclature and base year may occur. As regards financial data, the ones we use for now (short and long rates, exchange rates) are available in the OECD database.
On the future, it seems logical to start the exercises at the current period, 2013. The OECD database provides actually two years of forecast, which we can adjust the beginning of our simulation. In following periods, we will apply a common reasonable rate for the change in assumptions, and check that the results are reasonable too.
Our goal here is not to forecast precisely the world economy, but to provide a realistic basis for the consequences of our economic policy decisions.Indeed, the entire system easily produces reasonable simulations.
To help reach this goal, wesystematically formulated our equations as error correcting, to ensure stability in the medium and long term.
•As for obtaining realistic properties, it is a problem on two levels. Achieving acceptable individual models has not proved too difficult, given the lack of originality of our frameworks, and the fact that we allow ourselves to choose the values of some coefficients in case statistical methods failed. Producing a realistic trading system proved more difficult, as this field is little less trod, and the interpretation of the results may be less obvious. But the highest difficulty was met in making the country sensitivities consistent with each other, so that no team should be at a disadvantage playing a “bad” country. This means that our attention has been focused on this area, both in terms of formulations and properties
The differentiation of countries is first treated independently of any formalization. Structural characteristics, such as size, factor productivity, price competitiveness, will affect the properties of any model (e.g., the larger a country is and the less the stimulus on demand is penalized by imports). Perhaps more importantly, the starting level of economic criteria will modify the judgment on the results (e.g., a given improvement in the trade balance will be more appreciated in case of a large deficit).
These elements are clearly indisputable, but a satisfactory system must go further, presenting differences in formulations. It is obvious that the econometric estimates differentiates the country models, and thus the effectiveness of economic policies (especially when these estimates have been individualized successfully).
It is dangerous, however, to allow such differentiation to favor some countries too, especially if it is not fully confirmed by econometrics. In our context, we have allowed ourselves to modify significant coefficients even while remaining within their confidence interval. In practice, we favored estimating common coefficients, in many cases.
The structure of the work
It is clear that the construction of an acceptable system could not be carried out immediately on the full product. We decomposed the process into these steps:
First phase:
•Construction of a single country model, then a set of countries.
•Establishing links between the country models ("trade block ").
•Creating a minimal (and artificial)structure containing two identical country models and the rest of the world described in an elementary way.
The goal was essentially to test the principles the connection.
Second phase:
•Construction of a second model country (UK).
•Replacement in the previous system andtest of the properties.
This phase controlled the feasibility of a differentiated system, especially the consistency of differences between the coefficientvalues and the properties.
Third phase:
•Construction of a complete system (in fact, ten countries, a rest of the world) and analysis of its properties.
•Introduction of a differentiation in exchange rates and interest rates. Check of the relationship between formulation and properties.
Fourth phase.
•Integration of the model in the software context,
•Study of the gameplay,
•Definition of exercises and a manual.
The theoretical content of this phase is much lower. However, the definition of effective and realistic exercises can lead to review some formulations.
The country models
We limit ourselves here to presenting their common structure.
We observe in this diagram the following behaviors.
behaviors:
Overall: a Keynesian structure, with significant supply side effects.
•Productive capacity: follows a Cobb-Douglas formulation, with a substitution effect depending on the relative cost (including a role of the long term interest rate).
•Both factors adapt to desired values with some inertia, higher for investment. A role of the profits rate is added.
•Value added price (with an accounting effect of intermediate consumption) present a tradeoff between the margin and capacity utilization rates.
•Wages: short term indexation to inflation, long-term adjustment of the margin rate to the rate of unemployment.
•Changes in inventories: based on changes in production.
•Household consumption is based on real income, inflation (real holdings effect), unemployment (security savings), and the short term real interest rate.
•Trade prices: geometric weighting of producer prices of exporters and importing market (margins and competitivenessbehaviors).
•Trading volume: according to demand associated price competitiveness and capabilities.
In the full model, the compatibility of bilateral flows will make identities of import prices equations, as we shall see.
A full document gives details on the economic theory, the formulations and the values of coefficients.
Relating the models
The tradeblock
The principle that we have adopted is as follows:
We begin by establishing consistency between export prices and import prices. For this we start from the estimated export prices:
We now construct a set of aggregate indicators, based on our a priori distribution of a country's exports between markets. The matrix A (components ai,j) contain the share of country j's exports to country i. We will see later how it is obtained. We deduce the matrix B, giving the share of country j imports from country i.
The import price indicator will be obtained as an average of the export prices in the partner countries, taking into account their share in the local market. This gives
All the same, for each importing country we compute an average rate of use of exporters to that country.
Finally, the average production price of export markets (in local currency of the exporting country) is calculated as the average producer price of recipient countries:
Now we can determine the total imports of country i, by slightly modifying the equation of the individual model, to take into account the capacity of exporters:
Thus, as in the usual mono-country exports equation, increased tensions among exporters will result in a decline in exports, through a substitution effect. The coefficient c, lower than unity, is supposed to reflect the larger capacity of this group of countries.
We have to allocate imports between exporters. Again, we take into account the relative competitiveness, and fluctuations in available capacity, compared to the average calculated above. Rather than estimated equation, we adopt the following formula:
which expresses, as the estimated formulation, as compared to a "natural" part of their market, that a country's exports increase with price competitivenessand available capacity, this time in relation to their competitors on the market. As noted above, the matrices ai, j and bi, j (one being deduced from the other) reflect the distribution of bilateral flows associated with identical utilization rates and competitivenessin all countries (which may not be the most natural assumption). If the scalarsa and b are fixed a priori, their value will be chosen in line with the estimates of country models.
It should be noted that this approach guarantees the identity of the sum of individual exports with the overall value, without need for any correction.
The set can be summarized by the following diagram:
In addition to the above changes, we were led to introduce accounting equations describing:
- Exports of country i to j:
Total exports are computed as a sum :
Exports of country i at current prices :
- The global export price of country i
Imports of country i at current prices:
- The global import price of country i
The treatment of the "Rest of World"
Given its size, we considered that the Rest of the World:
•Has sufficient capacity for a change in trade not to affect its utilization rate.
•Sees the level of its production price (but not its trade deflators) insensitive tofluctuations of country domestic prices, whether exporting or importing.
These assumptions may be acceptable for a shock to one of our models, are much less in the case of a global shock.
We will limit ourselves for now, regarding the rest of the world, to describing its bilateral flows with our area, as well as aggregated elements: exports, imports, balance with the area,export-import ratio in value and volume, the terms of trade.
•
As the model countries, imports from our area will be modulated from a m0 base, utilization rates and competitiveness - price. But the rate and output prices in the Rest of the World remain exogenous.
•
The export prices of the rest of the world to our area will be divided between behavior margins and competitiveness, taking the base price exogenous local production
Where ppr is the producer price of the rest of the world (exogenous).
References
JL BRILLET, G. CETTE, R. GAMBINI (2013, forthcoming), Macro-économie internationale : MacSim 2, un logiciel de simulation de l’économie mondiale, (with a CD including a program in both French and English).
P. AUGIER, J.L BRILLET, G. CETTE, R. GAMBINI (2001), Macro-économie européenne : MacSim, un logiciel de simulation, Montchrestien (older version, also with a CD).
J.L BRILLET (2013), Structural macroeconomic modelling with applications under EViews, textbook with files representing a complete modelling toolbox.
This book is available on the company site at:
It is also available for download on scribd as PDF files.
(book: methodology and tools, kmark version) .