SIMULATING MARKET AND FIRM LEVEL DEMAND - A ROBUST DEMAND SYSTEM Developments in Business Simulation & Experiential Exercises, Volume 10, 1983 101 SIMULATING MARKET AND FIRM LEVEL DEMAND - A ROBUST DEMAND SYSTEM Steven C. Gold, Rochester Institute of Technology Thomas F. Pray, Rochester Institute of Technology ABSTRACT The paper presents an approach to modeling and simulating demand which is based on contemporary economic and marketing theory and employs stable mathematical Functions. The paper reviews same of the pitfalls of Functions have been used by simulation designers to model both the industry and firm level demand in an-going computerized business simulations, A robust and ideal demand function (system) is presented, consisting of a series of mathematical equations which embody the following concepts and points: multiplicative industry and firm demand functions with variable elasticities, exponential smoothing on demand variables, a current period stack-out reallocation algorithm, a system of checks an faulty decision inputs, and other marketing and economic concepts such as diminishing returns and market share considerations. The paper concludes with a numerical example demonstrating the flexibility of the ideal system. INTRODUCTION During the past two (2) years, ABSEL conferences have started to deal with design issues for simulation games. Both Gaasen’s [1] and Pray and Golds [2] papers have addressed the need, by ABSEL members, to be more open about the design and the internal workings of simulations Still numerous questions by both new designers and users of computerized business games are often raised. Some typical questions include: How are the production processes modeled? How does one mathematically specify the market demand curve? Users of existing game often want to modify the simulation to eliminate the conventional wisdom that often occurs after several semesters of play. Sometimes this may be accomplished easily through the use of control cards and variable parameters. Often, however, thy modifications require alterations to the program itself, Thus an appreciation for the internal workings of the simulation is needed, This understanding can help answer user-based questions such as Why was there such instability of price in this simulation ? or Why did all the firms spend so much on marketing and/or research and development? Business and management simulations, in particular, are modeled to represent the “real world" firm and market environments. Students are supposed to gain insights into the workings of the “real world’ by participating in the simulation, As a result it is necessary far the functions and algorithms contained within the simulation to be at least consistent with, the economic, managerial, and financial relationships found in the business world, Although these underlying principles and theories explaining the ‘real world” phenomenon are well-known, the task of precisely modeling and quantifying these relationships in a simulation are not straight forward, Understanding the pros an cons of different functional forms can vastly facilitate the modeling process and ensure realistic simulation results, A proper appreciation by designers and users of the different modeling approaches can also prevent games from yielding unreasonable results (i.e. blowing up!). This paper presents and analyzes an effective method for modeling and simulating demand, often the key comment of on-going business games. PURPOSE The purpose of this paper is threefold: (i) to review the problems associated with market and firm level demand models currently used in eight business games [[l],[2],[3],[4],[5],[6],[7],and [9]], (ii) to present a system of equations which embody a number of key theoretical properties and practical issues, including 1) a multiplicative (power) industry demand function which incorporates the principle that the marginal impact of any variable, say advertising, on the total demand is not constant but is dependent on the level of other independent variables, such as price. 2) having variable elasticities For one or more of the independent variables, 3) permitting the introduction of diminishing returns on any of the independent variables, 4) eliminating the impact of irrational or faulty decision inputs on total market demand determination. 5) utilizing exponential smoothing to recognize intertemporal movements in the decision variables, 6) having an intrinsic stockout adjustment routine which reallocates, in the same decision period, excessive stockouts to other firms in the industry. 7) employing a multiplicative firm level share equation which has variable price elasticity of demand. (iii) to demonstrate, with a numerical example, the power and flexibility of the demand system. COMMON PITFALLS OF SIMULATING DEMAND In a paper presented by Pray and Gold [2] at the 1982 ABSEL conference in Phoenix, Arizona, the authors investigated the underlying demand functions in eight contemporary simulations. Their analysis demonstrated that a number of the eight simulations investigated were unstable and that extreme decision values could induce unrealistic results, They also found that a number of the simulations inadequately incorporated diminishing returns, thus explaining why certain games induce such non-price competition as continually increasing expenditures on market related decisions One game failed to adequately differentiate the firm from the industry level demand, which in turn could cause price instability. (The opposite of the well-known Sweezy Kinked Demand theory for Oligopoly). They concluded the following: Developments in Business Simulation & Experiential Exercises, Volume 10, 1983 102 It was found that many different modeling forms and techniques were employed by the simulation designers, Some of the simulations incorporated lagged stockouts in the demand function, while others introduced uncertainty, either in actual demand, ar in stockout returns, Most of the simulations incorporated an intertemporal movement in demand analysis. The majority of these used exponential smoothing with larger values of the smoothing coefficient being applied to the price and marketing variables. As noted, certain of the demand functions were somewhat unstable and yielded unrealistic results, if left unconstrained, The designers, in most cases, imposed constraints on the decision variables to prevent discrepancies between theory and simulation play. The following points summarize the key advantages and disadvantages to the three forms used o Linear Demand Model permits variable elasticities. However, the impact of the marginal change in an independent variable is not related to the level of the other independent variables, Tentative elasticity analysis suggests the functional form is sensitive and the elasticities may vary rapidly, Input constraints should be imposed to insure realistic results, o Non Linear Model- permits variable elasticities. Tentative analysis suggests it is difficult to separate out the impact of an individual decision on the demand. Highly unstable and constraints on the decision variables are needed, o Multiplicative Demand Function- maintains a constant elasticity over the range of decision values, The impact of the marginal change in an independent variable is related to the level of the other independent variables. Appears to be stable at the industry level, However, at the firm level care must be taken to avoid “zero” level decisions. For stability of price (i.e. Sweezy Kink Demand Theory) care should taken to insure that the firm level price elasticity be larger than the industry level, The “Inverse Kink” found in one simulation will probably induce instability in prices. The elasticities of marketing and R&D variables measure the degree of diminishing returns. They in turn suggests the relative importance of those variables in the decision process, The lack of adequate diminishing returns at the firm level even with substantial diminishing returns at the industry level, induces non price competition and may cause excessive expenditures on that decision variable.” This conclusion was the force behind this current paper and the development a system of equations which may be employed to model both industry and firm level demand under many different conditions (and with a variety of decision variables) and a system which will remain stable and does not have the shortcomings found in the other games. A SUGGESTED SYSTEM FOR MODELING DEMAND The system recommended for modeling demand is composed of four critical components (i) conventional sample mean calculations for the independent and dependent variables, with the exception of the price variable, The harmonic mean should be employed to calculate the average market price; (ii) exponential smoothing on all demand variables to capture the intertemporal effects; (iii) a generalized multiplicative market demand function which allows for variable elasticities and (v) a multiplicative firm- level demand function which has variable firm level elasticities and is constrained by the total market demand, The suggested system is presented in Table 1-1, and consist of eight (8) equations. Each of the principle components is described in detail below. The Harmonic Mean The harmonic mean computes that average market price by weighting low prices relatively more than high prices This property is desirable since low priced products (firms) generate higher quantities demanded than high priced firms, The formula used to calculate the harmonic mean is: where n = the number of firms in the industry pi= the price of Firm i P= the average price for the industry Developments in Business Simulation & Experiential Exercises, Volume 10, 1983 103 A simple example will illustrate the effect of using the harmonic mean. Suppose that the market consists of a duapoly situation with firm * 1 charging $10 and firm * 2 charging $20. The $15 conventional mean, implicitly assigning equal weights to each item, would overstate the “true” average price since the lower price would actually induce more sales than the larger price. The harmonic mean calculation yields an average price of $13.33, which would more closely reflect the actual weighted average for the industry. Exponential Smoothing The demand for a product depends not only on the current values of the independent demand variables, but also on their historical values, For instance, both current and past expenditures on advertising impact the sales potential of a firm. Exponential smoothing is a convenient technique allowing the simulation designers to specify the rate and the importance of history on current demand, The conventional formulas are presented below with an example: where: P = exponentially smoothed harmonic price M = exponentially smoothed marketing expenditures R = exponentially smoothed R&D expenditures An "o" subscript indicates a period-aid smoothed value An "n" subscript indicates the mast current mean value The values of a, b, and c (the exponential smoothing coefficients) determine the impact of historical data on the current demand, Larger values for these coefficients put mare weight on current data. For example, suppose that the value far "b" was .75. This means that a weight Df .75 is assigned to the current average for marketing expenditures and the historical data is assigned the residual .25 weight in an exponentially declining fashion. Larger values for the coefficients are equivalent to a smaller number of terms in a moving average. For instance if “a’ is .5, this will have the about thy same effect as a moving average with four terms in it.1 A value of .05 is roughly equivalent to a thirty-nine period moving average. Both theory and evidence indicate that large values for a’ and ‘b’ would be desirable, while a smaller smoothing value would be appropriate for variables such as research and development. Market Demand A three variable market demand function which thaws for both multiplicative demand properties and variable elasticities is specified below:2 1 When the average age of data is used as a criteria, the value of the smoothing coefficient may be equated to the number of terms in the moving average by: N = (2-a]/a where N is the number of periods in the moving average and “a” is the exponential smoothing coefficient. 2 This model may be easily generalized to the n-variable case. The values assigned to the parameters (g1,g2,..,g7) depend on the designer’s specification concerning the elasticities (sensitivities) assonated with the demand. The determination of the values of the seven (7) parameters is discussed in detail in the forthcoming ILLUSTRATIVE EXAMPLE section of the paper. Firm Demand There are four basic components of the firm demand function: (i) the weighting function, (ii) the market share equation, (iii) the stockout routine, and (iv) the quantity equation. Each of theses are described below: The weighting function: The weighting function determines the magnitude of the value which is used to calculate the market share of the firm as a function of total market demand. The weighting function is a variable elasticity multiplicative function which is similar to the market demand equation previously specified. The weighting function suggested is as follows: The purpose of parameters k1, k4, and k7 is to prevent the weight from equaling zero if a firm enters a zero decision For one of the demand variables, The magnitude of these parameters (k1, k4, and k7) may be arbitrarily set at any small magnitude relative to the respective demand variables (pi, mi, or ri). The Share Equation: The share equation is the weight of the firm (i.e. eqn. (6)) divided by the sum of the weights for all firms in the market, The share equation is as follows: The Stockout Routine If a firm behaves in an irrational fashion, causing them to receive an inordinate amount of the industry demand, and they are unable to supply the required goods, the stockouts (unsatisfied demand) are redistributed in the sane period to the other firms in the industry via the farces of supply and demand. Two criteria must be satisfied before a stockout condition is declared the firm’s share, based on equation (7), is “too large” for the market structure and the number of firms. (The “too large” is defined by equation (8)), (ii) the firm cannot satisfy the demand. If criteria are meet then the stockouts are redistributed to the other firms, Developments in Business Simulation & Experiential Exercises, Volume 10, 1983 104 In reference to the first criteria, the maximum firm share in a given market is determined by using a quality control p-chart The essence of the control chart is based on a firm's share being within 3 standard deviations of the expected share, Equation (3) demonstrates the upper control limit In reference to the second criteria while smax indicates that the share is suspect the routine then ascertain whether or not the firm could supply the desired quantity. This may be accomplished by comparing the firm’s demand potential, from equation (9), with their total finished goods available. IF the share is beyond the smax and the firm is unable to satisfy the demand, the routine reallocates the excessive demand to the other firms by normalizing equation (7), after removing the unrealistic firm’s share. The Quantity Equation The quantity demanded of firm i is equal to the market share of firm i multiplied by the total market demand from equation (5), The firm demand equation is as follows (9) qi = s iQ The benefit of this approach is that it restricts the sum of the individual firm demands to equal the market demand determined by equation (5), As noted, if excessive stockouts occur to one or more firms, the stockout routine reallocates the unsatisfied demand to the other firms. This prevents the industry from being distorted by bad decisions or errors in data entry. Deriving the elasticities The elasticity assonated with each demand variable is derived by the applying the conventional formula The market and firm level elasticity equations for the demand Function previously mentions are given below The firm elasticity is derived for the weighting function (equation (7)). Price Elasticities (10) EP = g2 +g3P (1 + lmP) Market Level: (11) Epi = k2 + k3[pi +ki][1 + lm(pi + ki)] Firm Level: Market and firm price elasticities, in this demand system increase with increases in price, since it is assumed that all parameters are positive (i.e. all g~>O and k)O). Additionally the rate of increase of the price elasticity with respect to increases in price level is also non-linear, Furthermore, unlike the linear system; observed in [2] the price elasticity is independent of the other demand variables in the system which enhances the stability of the system. Marketing Elasticities Market and firm level marketing elasticities decrease with increases in marketing expenditures since g5 and k6>0. As in the case of price elasticity, the mtkpt elasticity relationship is non-linear and independent of other demand variables. R&D Elasticities (15) ER = g6 – g7R (1 + lnR) Market Level: (16) Eri = k8 – k9 [ri+k7][1+ln(ri+k7)] Firm Level: Since g7 and k9 are both assumed to be greater thin zero the research And development elasticities have the urns properties as the marketing elasticities, Parameter Determination Salving the parameters in the demand system to obtain the desired elasticities simply involves the following three flip procedure: U) Select the starting value for each demand variable and th. corresponding elasticity value, (ii) Select the second value (data paint) for each demand variable and corresponding elasticity value. (iii) Substitute the elected values into the elasticity formulas and solve the simultaneous equations to caiwlats the required parameter values, To assist in understanding how the system works a simple example is presented, ILLUSTRATIVE EXAMPLE An example of how to determine the parameter values far the market demand function is presented to illustrate the general procedure, and to demonstrate the properties of the function. It is assumed, for simplicity, the desired demand function consists of only two independent variables, say price (P) and marketing expenditure (H), Furthermore, the designer has specified apriori the following values for the demand variables and the corresponding elasticities: Developments in Business Simulation & Experiential Exercises, Volume 10, 1983 105 Substituting the values for price and price elasticity into the market level price elasticity formula (equation 10), the following two equations are obtains: Solving these two equations simultaneously, the values of and g3 are: Repeating this procedure and substituting the values for the marketing expenditures and marketing elasticities into the market level elasticity formula t equation 12) yields the following two equations: Solving equations (13) and (19) simultaneously, the values for and 95 are Consequently, the illustrative market demand function in this example is: The parameter g1 is a scaling factor and may be arbitrarily assigned a value, It in no way impacts on the elasticities. For this problem was assigned the value of 2.34 x 10 SIMULATING THE MARKET DEMAND FUNCTION The market demand Function derived in the above example will be simulated to illustrate the impact of price and marketing expenditures on the sales potential, More specifically, the simulation considers two cases: (1) the impact of variations in price on quantity demanded, holding marketing expenditures fixed at a starting value of $50,000; and (2) the impact of variations in marketing expenditures on quantity demanded, holding the price variable Fixed at the starting value of $10.00. The results of the simulation are reported in Table; 1-2 and 1-3. As noted in Table 1-2, the price elasticity of demand increases slowly and steadily with increases in the average market price, The price elasticity is initially inelastic with a value of .5 C in absolute terms) at a starting price of $10.00. It increases to a unitary elastic value when price reaches $20.00, as specified apriori. The marketing elasticity of demand, noted in Table 1-3, decreases relatively quickly with increases in the average marketing expenditure for the industry. As noted, marketing expenditures are highly Elastic at the starting point of $50,000, and decrease to unity at an expenditure of $150,000. These values are consistent with those specified in the illustrative example. It is interesting to note that the advertising elasticity in this example, turns negative at an expenditure level of $210,000. This indicates, that after some point ($210,000) increases in marketing will actually hurt market demand because of oversaturation. This demand system is sufficiently flexible to permit or not permit turning points (i.e. inflection points) to occur in the function. These turning points can be readily specified by the simulation designer by following the three step procedure outlined earlier, Developments in Business Simulation & Experiential Exercises, Volume 10, 1983 106 SUMMARY AND CONCLUSION This paper represents an ongoing attempt to encourage open discussion concerning the design and development of computerized business simulations, Unfortunately, there appears to be a reluctance, even by ABSEL members, to discuss internal modeling components. In all the past ABSEL conferences less than 20 professional papers have dealt with design issues. To encourage open discussion about modeling, the paper developed a mathematical model of a demand system which may be used in designing a computerized business game, The demand system has been shown to possess a number of desirable properties: • A harmonic mean to more effectively approximate average market price. A conventional sample mean has been shown to overstate the true average price. • Exponential smoothing to capture intertemporal effects. The simulation designer, by specifying the smoothing coefficients can easily control the rate of history in the simulation, • Multiplicative market and firm level demand functions which permit variable elasticities, diminishing C or increasing) returns and is stable. The parameters of the demand system have been shown to be easily’ solved for once the desired elasticities are specified by the designer. • A stockout routine which redistributes unsatisfied demand within the same period of simulation play. This prevents unrealistic market or Industry results from occurring, if a firm (or set of firms) make economically irrational set of decisions. REFERENCES [1] Brooks, LeRoy D. Instructor’s Manual-THE FINANCIAL MANAGEMENT DECISION GAME (Richard D. Irwin Inc, Homewood Ill.) 1975. [2] Darden William R., William Lucas, THE DECISION MAKING GAME- AN INTEGRATED OPERATIONS MANAGEMENT SIMULATION, (Appleton-Century-Crofts, Mew York), 1969. [3] Edge, Alfred G., Bernard Keys, William E. Reymus. Instructor’s Manual--- THE MULTINATIONAL GAME (Business Publications Inc.~ Dallas), 1980. [4] Gold~ Steven C, Thomas F, Pray, and Terry Dennis, Instructor’s Manual-- MICROSIM-A MICROECONOMICS SIMULATION, Macmillan, New York) Forthcoming. [5] Henshaw, Richard C. and James Jackson, THE EXECUTIVE GAME (Richard Irwin Inc., Homewood Ill.), 1972 [6] Pray, Thomas F., Daniel Strang Instructor's Manual-- DECIDE, (Random House Inc., New York) 1981, [7] Pray, Thomas F,, Daniel Strang, Instructor’s Manual-- PRODUCE- OPERATIONS MANAGEMENT SIMULATION, (Random House, New York) Forthcoming, [8] Smith, W. Nye, Elmer Estey, Ellsworth Vines, INTEGRATED SIMULATION. (South Western Publishing, Cincinnati) 1974. [9] Pray, T. and Golth S. “Inside The Black Box An Analysis of Underlying Demand Functions in Contemporary Business Simulations," DEVELOPMENTS IN BUSINESS SIMULATION AND EXPERIENTIAL EXERCISES, Vol. 9, 1982, pp. 110-115. [10] Goosen, K. A Generalized Algorithm For Designing And Developing Business Simulations," DEVELOPMENTS IN BUSINESS SIMULATION AND EXPERIENTIAL EXERCISES, Vol. 8,1981. Table of Contents Volume 10, 1983 Is the Computerized Business Simulation Relevant? Business Professionals Play a Student Game A Methodology For Assessing the Internal Validity of Business Simulations A Longitudinal Study of the External Validity of a Business Management Game Policy Analysis and Decision: A corporate Relocation Simulation Exercise Effective Listening: An Exercise in Managerial Communication The Symbol Exercise: An Initial Group Activity Concept Based Simulations Institutional Users of Experiential Learning Packages: A Preliminary View from Publishers' Adoption Lists How to Use Business Games in the Business Policy Course: The Students' Perspective Professors' Ratings of Business Policy Learning Methods Moot Trial: An Exercise in Trial Procedure and Evidence The Advertising Agency Game: An Experiential learning Exercise The Use of Videotaped Cases in Teaching Information Acquisition and Decision-Making Skills Experiencing Information Processing Strategies as a Means to Explore Decision Making Importance Ratings and Operations Data as Predictors of Business Game Performance Determinants of Performance in Computer Simulations Predicting Business Game Performance form Perceptions of Manager Information and actions Business Consulting: A Practicum for Undergraduate Internship as a Contingency Based Experiential Learning Program for More Effective Organizational Socialization: A Conceptual Framework MANSYM III Decision Support System Demonstration Learning the Concept of market Value Through Simulation Conflict Management for Economic Developers Development of Data Analysis Units Designed to Enhance Reasoning and Knowledge Transfer in the College Level Course BOSS: A Behavioral-Quantitative, Computer-Supported Game Entrepreneurial Potential: An Experiential Exercise in Self Analysis and Group Assessment Development of Strategists: Simulated Cases BANKRUPT: A Deceptively Simple Business Strategy Game Simulating Market and Firm Level Demand - A Robust Demand System COMPSIM A Computer Center Management Simulation Hiving Model: Assessing Management Skill Awareness The Johari Window, A Reconceptualization Role-Playing Based on Video-Tape Scenarios: An Application of Modeling to Building Supervisory Skills How to Internationalize Your Curriculum A Computerized Model of Human Behavior in a Total-Firm Management Simulation the Worksheet Approach for Simulation Game Strategy Analysis Teaching Competitive Bidding Using a DSS Generator Do We Learn from Experience? The Use of Theory Power for Increased Research Momentum in Business Simulation and Experiential Exercises Research Report on Programmatic Research on Perceived Learning Barriers with Simulation and Experiential learning An Empirical Examination of Conflict - and Nonconflict - Oriented Problem-Solving Technologies Management Curriculum: 1982 Humor as a Management Tool: Use in Formal Game Presentations Student Behavioral Change Through Teacher behavioral Change