Simulation and Reality

5 Relating Emergent Behaviour to Non-Statistical Data:
Modelling the Transition Economy of the Russian Federation

Fig. 4

The percentage monthly growth rate of inter-enterprise debt arrears in the Russian Federation from February, 1992 to April, 1995. Source: Finansy promyshlennosti Rossii, The Monthly Bulletin of the Working Centre on Economic Reform of the Government of the Russian Federation, no.1, June 1995, p.2

The particular period considered was that of the "arrears crisis" of 1992. The time pattern of the growth of inter-enterprise arrears is shown in fig. 4.

Fig. 5

The structure of the model of inflation and inter-enterprise debt arrears for the economy of the Russian Federation.

The evolution of the arrears crisis provides a clear example of how enterprises cope with situations characterised by significant uncertainty. In the Russian case, enterprises were forced to adopt a survival strategy giving priority to existing, recognized constraints. There was no possibility to maximize anything in the framework of those constraints [17]. Indeed, the scale of the accumulated debt and its persistence suggest that enterprises that debt reduction was not a high priority. For one thing, both the liquid assets of enterprises and their debts have been growing simultaneously. The debt became an element of their survival strategy being instrumental in prolonging the existence of a business environment they were accustomed to, i.e., the one governed by soft budget constraints. Because the accumulated bad debt grew out of proportion on the national scale and became commonplace in all industrial sectors, this important business indicator ceased to be seen as a symptom of poor management efficiency.

There were three production sectors: factories producing outputs for both the mining and agricultural sectors, mines producing outputs for the manufacturing sector and farms producing outputs for the agricultural sector itself and for consumption by households. the model cycled over dates and within each date the agents in the economy cycled over communication cycles during which they made offers to buy and agreements to sell products

With this model, it was by no means difficult to generate rapidly growing inter-enterprise debt arrears. In fact, all models had these arrears but the early models had flat price, employment and output series. It was clear that some important aspect of the cognitive processes of enterprise managers was being missed. The mental models of the agents were generated initially by random combinations of actions and implied consequences for the goals of the enterprises which included sales volumes and cash holdings. Because of the extent to which the modelled economy captured the instability of the actual Russian Federation economy, there were no signals from this generate-and-test approach to developing mental models to identify a small set of models which would inform goal-enhancing behaviour. Noise but no signals were being provided by the economy itself.

We then looked for other sources of information which we found in the transactions process itself. The information available to enterprises included the prices it was being charged by its suppliers as well as the supplier's record of filling orders for their outputs. They also knew which of their own customers were paying for the goods they acquired and the prices they charged. A natural hypothesis was that enterprises would imitate the behaviour of the most successful enterprises known to them. The instability and limited information sources and general lack of reliability of information in the economy indicated that the best source of information was observation and the observations were of the suppliers and customers of the enterprise. Paying bills without holdings of cash was not a possibility even if this were a mark of success in customers. But raising prices was always possible and this was the behaviour that emerged. A result of assuming that mental models were informed by direct observation, the simulation model generated a volatile price inflation series with a strong upward trend. It turned out, though we did not have the published inflation figures at the time, that the official price series is indeed marked by the sort of volatility we found. Once again, and acknowledging its defects, the notion of the statistical signature provided a useful target for social simulation modelling.

Generated with CERN WebMaker