Skip to main content
main-content

Inhaltsverzeichnis

Frontmatter

Problems and Results

1. Theoretical Background

Abstract
In most of the markets in advanced economies information on exchange-offers, i.e. prices, quantities and qualities, is not communicated instantaneously and without cost. Exceptions are highly organized markets, e.g. the stock-exchange, broker and auction markets. Thus, market participants must frequently search for relevant information. Then they must decide not only on the acceptance, rejection, or modification of offers, but also on how much information they want for decision making purposes. What the information behavior of the individuals looks like and how it influences the processes in such markets has not yet been satisfactorily explained in economics.
Ulrich Witt, Joachim Perske

2. Some Applications and Results

Abstract
The purpose of SMS is twofold: to support research both in the theory of the market process and in the theory of individual behavio’r where information is imperfect or absent. This dual purpose is achieved by explicitly elaborating and integrating theoretical notions about both
  • the dynamics of simultaneous random contacting in search markets with imperfectly informed individuals on both sides of the market and
  • the dynamics of individual learning and adapting behavior starting with a substantial lack of information.
Ulrich Witt, Joachim Perske

The Economic Model of Individual Behavior and the Market Interaction

3. Restrictions on the Firm’s Behavior

Abstract
SMS is a program package written in the higher programming language FORTRAN. More precisely, the lists printed in the text use an ASCII-FORTRAN version (UNIVAC level 6 R 1). The program is constructed for batch processing and includes approximately 3,400 cards. It is organized in two self-contained main programs in order to lower the core-storage requirements of the program package. The two main programs are called MONTE and HAUPT2. Since the latter is only used for statistical purposes, in Part II we will be solely concerned with MONTE and, above all, with its most important subroutine HAUPT, which contains the economic constructs of SMS. HAUPT itself consists of several subroutines, which in turn may call further subroutines; see fig. 3.1.1.
Ulrich Witt, Joachim Perske

4. Simulating Interaction in the Market

Abstract
Following the discussion in ch. 1.2 concerning a market, in which imperfectly informed, searching consumers and firms act, it can be said that search expenditure, allocation of transactions, and individual experience in any period seem to depend on who contacts whom and in what sequence. Staying with the random interpretation as far as the last question is concerned, the outcome of the contact sequence can be described by random variables, e.g. the individual number of successful and unsuccessful contacts between firms and consumers, the quantity demanded and so on. In the absence of further knowledge, an attempt could be made to deduce probability distributions for these random variables, based simply on the principle of insufficient reason; any possible contact sequence being equally likely. This idea was suggested in ch. 1.2. But this strategy results in very complicated combinatorial constructions, except in some simple cases, or even breaks down altogether.
Ulrich Witt, Joachim Perske

5. Simulating the Behavioral Response

Abstract
Once the initial offer of a firm K has been determined by the subroutine ANGEB, the individual response to observed demand data is simulated by the REA..subroutines. This is done sequentially, what means, that an activity output vector is assigned by these subroutines to each information input vector in discrete time. Each of these consists of a sequence of statements which represent controversial theoretical models of the learning and adapting behavior of an imperfectly informed firm. These models mostly use identical information. This includes state variables indicating
  • the environmental feed-back, i.e. the number of contacts with consumers KKAB(K), the quantity sold YV(K), the unsatisfied demand LEER(K), the contacts without selling success OERF(K), and the inventory YLAG(K); current values of these variables are determined in the preceding iteration of the subroutines simulating the market interaction;
  • the relative profitability of the performance, i.e. the earned rent on capital RENR(K), its change since the last period DREN(K), its average value since market entry AVREN(K), and its accumulated value expressed as an index related to initial capital SPSREN(K); current values of these variables are calculated on the basis of the actual cost and demand conditions in the special data processing subroutine AUSWER.
Ulrich Witt, Joachim Perske

6. Applications of SMS in Laboratory Experimentation and Classroom-Teaching: The Gaming Version

Abstract
The gaming version of SMS is a special type of simulation in which human participants act as decision makers in place of the REA..subroutines within the, otherwise unchanged, system being simulated. This construction allows SMS to be utilized as a highly motivating tool in classroom-teaching of the theory of imperfect information and learning, monopolistic competition, market processes and other related topics. Moreover, the gaming version can be used for experimentation purposes. As already mentioned, the learning and adapting behavior of the test persons takes place under conditions identical to those used in exploring the various experimentally programmed behavioral models in simulation. Hence observed time series in each case are directly comparable. The synthetically generated performance can be confronted with that of the human participants so as to make empirical evaluations possible.
Ulrich Witt, Joachim Perske

Program Organization and Input-Output-Facilities: The Technical Part

7. The Controlling Program Sections

Abstract
The FORTRAN program package SMS is designed for batch processing, in the current version it includes approx. 3400 cards. As already mentioned, it is written in ASCII-FORTRAN, UNIVAC level 6 R 1, and was at last implemented on a UNIVAC 1100/82 computer. Since originally it was planned for a computer with a much smaller core storage capacity, we divided the program into two parts to be processed sequentially: the main program MONTE, which monitors the computer operation in the generation of raw data over a sequence of Monte Carlo iterations, and the main program HAUPT2, which processes the raw data into statistics and graphical output. In this way, we reduced storage requirements to approx. 34 K words for MONTE and approx. 45 K words for HAUPT2. Thus, to be implemented, SMS requires a maximum of only 45 K words core storage capacity. For data transfer from the main program MONTE to HAUPT2 a data file is used, which may be temporary. The program’s flow in the total program package is shown in fig. 7.1.1.
Ulrich Witt, Joachim Perske

8. Input and Control Facilities

Abstract
To start a SMS-job in batch processing on a computer, the program package must be implemented and a specification of control and other parameters and initial values must be delivered.If a great deal of input data differing in function and type are needed, it is desirable, for convenient experimentation, to construct the input modalities in such a way that data of the same kind merge in blocks which can be exchanged and combined without punching all the data input again. For this reason we have decided to split up data input needed for a SMS-run into several blocks and to assign each of the blocks to a special format card type.
Ulrich Witt, Joachim Perske

9. Outputfacilities

Abstract
One of the first problems that occurs when output facilities are developed for a program package like SMS is identifying output uniquely. In experimental programming a great deal of test output is generated, which is often difficult to associate with a particular state of the program by the run or experiment identification alone. To avoid confusion, the output must be made distinguishable, independently of experiment identification which, for convenience, will often be maintained. The easiest way to do this, isto mark each kind of output with the actual date and clock time. In SMS this is achieved by using the subroutine DATUHR. It generates a heading line as in fig. 9.1.1 for each kind of printer output, containing experiment identification and actual clock time, identical within each Monte Carlo run.
Ulrich Witt, Joachim Perske

Appendices

10. Annotated Name List

Without Abstract
Ulrich Witt, Joachim Perske

11. Cross-Reference Listing

Without Abstract
Ulrich Witt, Joachim Perske

Backmatter

Weitere Informationen