Project description Lars Olert 1996-02-16
translated into English
October 1998
Macro economic investigation and simulation.
This document has not been revised since the first Swedish issue. It has only been translated into English. See also "Final notes" at the end.
To starting page

Goal and problems

The project consists of two parts, presentation of historical data and simulation of economic systems.

1. Collection and presentation of historical data. The historical data shall be included in a diagram for economic circular flows. That ensures that the data are complete and consistent. Also data that are outside of the economics flows shall be collected. That data may include population size and composition, unemployment rate, inflation, carbondioxide in the atmosphere and so on.

The historical data shall be add to form balances of supply and consumption etc. It shall be possible to present the data as time histories or pie or bar diagrams. Different variables shall be plotted together, e.g. unemployment versus inflation, budget deficit versus political party in goverment, production versus employment rate etc.

2. The simulation of economic courses or events shall be based on a model of the economic system. The model of the economic system is built as economic flows that connect the economic actors of the system.

The economic actors are households, private capital owners, private enterprises, banks and insurance companies, the central bank, the state, cities and municipalities, unions and cooperative organizations. The country is surrounded by an "economic environment" of commodities from abroad (trade), currency (exchange rates), investors and banks. Note that it is necessary to distinguish between different kinds of households. Unemployed, low paid, well paid and investors cannot be treated the same way. They act under different constraints and have different goals.

Flows of goods, services, payments, taxes, transfers etc. are economic flows.

The economic model describes every actor in the system. The submodels can be founded in hypotheses from the text-books of economics, e.g. that the households maximize their utility, that the companies maximize their profit etc. Also hypothetical functions as marginal cost as a function of production volume or a more detailed model of the inner functioning of a company. Every model has to include a number of constants (parameters) that only can be determined from observations (statistics = historical data).

I intend to look through a number of test-books in economic science and compare diagrams and the agreement between them and historical data. The purpose is to compare how the proposed behavior agrees with reality.

The simulations are aimed to calculate macro economic behavior based on micro economic models. The question is: do the micro economic models imply the macro economic behavior proposed in the text-books.

Some statements to be investigated are:

Statements about equilibrium models as Aggregate Expenditure, AE, as a function of the Real National Income. Is equilibrium an attainable condition? If it is possible to reach equilibrium, how will the route be from one state of equilibrium to a different state of equilibrium? Do the oscillations hide all other effects? When do we get periodic or chaotic behavior? Are reversible routes possible?

Statements about the dependence between unemployment, level of wages and salaries and inflation. I do not believe that lower wages and salaries in general give higher employment. I do not believe that the companies employ more workforce that can be put into production, no matter the wage level.

The question about competition and a free market. How does competition influence the formation of new companies and the elimination of old companies? Is a free market with many companies stable or will the elimination of smaller companies lead to oligopoly or monopoly?

The question about dynamic effects. Do lower taxes promote more new companies and increased employment? Does lower unemployment compensation give higher employment because more people will apply for jobs? Is it possible to use education to eliminate the unemployment? What effect do tax reforms and shorter working hours have?

The question about taxation. Does increased progressivity make that people and companies move abroad? Do raised taxes give more or less tax revenues? What is the optimal taxation level?

The question about economic growth. Can economic growth give less unemployment and better public (state and municipalities) finances? I believe the competition promotes higher productivity that in turn leads to increased production and lower employment level.

How is demand in the economy created? I believe that demand only increases is people get more money left to spend. It is not possible to extrapolate a trend by e.g. 2%/year. Finally there will be a saturation of demand, partly because our physical needs are satisfied, partly because we do not have the income to buy all the new products. The limited space and scarce raw material supplies also set a limit.

It will probably not be possible to make long-term economic forecasts by a simulation. The economy will probably show a chaotic behavior. There are economic cycles, but they never repeat in the same way. It is more fruitful to study the properties of the system, as stability, average values, amplitude of the fluctuations, losers and winners etc.


The simulation method is based upon the calculation of flows during a certain period of time, then the accounts at the end of the time period can be computed. The flows during the next period depend upon the accounts at the beginning of that time period and the expectations about the future. The expectations are due to historical experience and economic policy.

It is necessary to distinguish between state variables of the system and outside factors that act on the system. State variables are e.g. fixed capital (buildings and machinery), assets and debts. Outside factors are decisions in the parliament about taxes or decisions of the central bank about the interest rate. Only in the case of outside factors, it is possible to talk about a cause to a change in the system. The outside factors cannot be modeled, they are decided arbitrarily by politicians and big economic actors.

Submodels for actors can be computed by regression analysis. Neuron networks can also model dependencies that are difficult to analyze. The issue of expectations can be treated by assuming that a fraction of the population follows the trend, an other fraction does not take any actions and the remaining fraction speculates that the trend will be reversed. It turns out afterwards which fraction had the greatest success.

The price formation process can be studied by various models of the market:

- The retail business market. A large number of retailers (shops, mail-order businesses) or monopoly enterprises (railway, mail, air lines) offer their goods and services to a price that is set in advance. The buyer has the choice to by the commodity to the requested price, to buy from a different supplier, buy a different commodity or not to buy anything at all. The supplier get a feedback from the market through the volumes he/she succeeds to sell. The choice of the buyer depends upon the price, the usefulness of the commodity and the cost (effort) to find alternatives on the market.

- The market for durable goods, e.g. refrigerators and cars. In this case buyer and seller can bargain about the price or discounts. The seller get direct information about how much the customer is willing to pay for a certain commodity. The sold volume is a kind of feedback.

- Wholesale dealers and importers. In this case, buyer and supplier bargain about both price and volume.

- Industrial investments an construction. Negotiations between buyer and seller about price, volume and properties (design, quality) of the product.

- Raw materials and sub contractors. Negotiations between buyer and seller about price, volume and quality of the product.

The development of products is done in different ways, depending upon who is the customer. Products for the consumer market (the majority of the population) are developed by "trial and error". The producer will find out afterwards which products can be sold and to what volumes. If the customers consist of a small number of companies or institutions, then the design of the product is agreed upon when the order is made and through the "technical development".

Size and income distribution among economic actors (enterprises, institutions) can be studied by self-copying agents. The copying process occurs with a certain degree of random behavior, which can simulate unexpected development in the system, e.g. how to avoid or counteract to a political intention. By studying a number of individuals with different behavior, instead of an aggregate (only average value and volume), it will be possible to study how size and income distributions develop by time.

The size and composition of the population can be modeled by birth rates, immigration, emigration and mortality rate for different ages.

Final product

The project shall result in computer programs that can be used to analyze propositions from politicians and economists in relation to historical data or to results from simulations.

The computer programs can be designed as a professional version, a student version and a game version. All three versions can be distributed by Internet. They can be used freely or further developed by those interested.

The programs shall be accompanied by a book that can be used for macro economic studies. A basic course in macro economy can use the analyzing and simulation facilities of the program. A course at the engineering faculty may be aimed at mathematical analysis and system design. The book may have the following subtitles:

Final notes:

Since this project description was written in 1996, I have made some new findings. The description does not propose static models. Static models have been found to give some realistic results and can also help to understand the working of the economy. Transitions can be analyzed be taking the difference between two static cases. Chapter 12 and appendix A2 explain better how dynamic models can be built.