With the passage of time, business operations around the world have become complex because of the dynamic nature of the industries in which the firms operate. The main reason for that is the increase in the level of uncertainty of the factors that play an important role in the business processes. For almost any decision to be taken, the management committee in any firm is now faced with a) an increased number of alternatives, b) massive change in the number and nature of goals, criteria or evaluations and c) increased complexity of the decision-making environment. Business decisions involve choosing between alternative courses of action and developing formal plans for future action.
Deciding between alternatives, such as competing projects, has always been of immense importance to organizations and has attracted much interest in the field of decision making. Sufficient research exists to show that often "good" decisions are not made – good in the sense of following robust, systematic processes rather than good in the sense of beneficial decision outcomes. The reason may lie partially in the unrecognised effects of heuristics or rules of thumb that managers rely on while making decisions. By their nature, heuristics may be largely or completely unseen, occurring without the individual decision-maker recognizing their presence or impact. In general, the human brain is limited and cannot handle the necessary manipulation to solve even a relatively simple decision problem. Given a decision problem, managers often frame a simplified scenario and then choose an alternative using heuristics or rules of thumb. Unfortunately, these heuristics often manifest themselves as subtle but powerful biases in decision making.
Even today, Fortune 1000 companies typically base their business decisions on a limited number of factors. There are indeed many internal and external factors that can affect and influence these business decisions. These factors can have a tremendous impact on crucial issues such as corporate profitability, the ability to deliver new products to market more quickly, decisions regarding entry or exit mode in any industry and, ultimately, whatever business decisions are made. One manifestation of this increased complexity is that more balanced approaches have to be introduced where even the non-financial factors have to be offset against the more traditional financial measures. A recent report from Goldman Sachs states that companies can expect to reap up to 15 percent costs savings when incorporating decision optimization applications into their business processes.
Decision Analysis – A Potential Solution
Decision analysis provides tools for quantitatively analyzing decisions with uncertainty and/or multiple conflicting objectives. These tools provide a systematic quantitative approach to making better decisions and can be especially useful when there is limited directly relevant data and the decision- making process calls for expert judgement.
Research in the field of decision analysis has led to the application of advanced analytical tools to help in the decision-making process. Such a tool is multicriteria decision making (MCDM) that identifies the alternatives to and attributes of a decision problem and helps decision- makers reach a conclusion through the formulation and solution of a mathematical model. Techniques such as MCDM have become important because they not only take into account quantifiable or financial figures but also include qualitative variables in the model.
During the 1990s, some decision analysis methods moved from the research stage to applications or became more widely recognized. Work also continued during the 1990s on well-established decision analysis methods to better understand their characteristics and to make them applicable to a broader range of decisions. Based on MCDM techniques, various decision support systems (DSS) have been developed to help in managerial decision making.
Decision-makers must reach a mental state in which they are confident and prepared to take action. To reach this state, a DSS may be employed to gain insights into the problem and explore a number of options. Decision-makers employ the DSS as a tool to help in decision making but rarely will treat the DSS as something that tells them the course of action to take. The results of any analysis using the DSS must agree with the users’ own intuitive feel for the problem.
Figure 1: The Four Stages of Decision Making
Making decisions consist of several different activities. Figure 1 describes decision making as a four-stage process: intelligence, design, choice and implementation. Intelligence consists of identifying and understanding the problem occurring in the organization. During solution design, possible resolutions to the problem are developed. DSS systems are ideal in this stage of decision making because they operate on simple models, can be developed quickly and can be operated with limited data. Choice consists of selecting from solution alternatives. Here the decision- maker might need a system to develop more extensive data on a variety of alternatives. During solution implementation, when the decision is put into effect, managers can use a reporting system that delivers routine reports on the progress of a specific solution. Decisions are often arrived at after a series of iterations and evaluations at each stage in the process. The decision-maker often must loop back through one or more of the stages before completing the process.
Computer support is required to facilitate the use of advanced decision- making techniques (e.g., MCDM). Two most important MCDM techniques include the simple multiattribute rating technique (SMART) and the analytic hierarchy process (AHP). It must be noted in this respect that SMART or AHP are not analytical software but analytical techniques. The techniques are implemented in a variety of commercial software packages.
The Concept of Uncertainty and Its Representation
Almost all information is subject to uncertainty. Uncertainty may arise from inaccurate or incomplete information, from linguistic imprecision and from disagreement between information sources. The degree of uncertainty is also a variable factor. The representation of uncertainty is intrinsic to the representation of information, which is its dual opposite.
Decisions often involve uncertainty about the external world as well as conflict regarding one's own preferences. The decision-making process often begins at the information- gathering stage and proceeds through likelihood estimation and deliberation, until the final act of choosing. In some decision contexts, the availability of the chosen option is essentially certain. Other decisions are made under uncertainty. These decisions can be risky, where the probabilities of the outcomes are known, or they can be ambiguous, as are most real-world decisions, in that their precise likelihood is not known and needs to be judged subjectively by the decision-maker. When making decisions under uncertainty, a person has to consider both the desirability of the potential outcomes and their probability of occurrences. Indeed, part of the study of decision making concerns the manner in which these factors are combined.
Flaws of Traditional Spreadsheets and Estimation Processes
Initially, for major decision-making problems, only Excel spreadsheets were in use. Tools such as estimation were given much importance in determining particular issues. One of the problems associated with traditional spreadsheet models is that for variables that are uncertain, managers are forced to supply a single, best-guess value. Of course, this value can be changed and results can be reviewed. But this manual process of what-if analysis becomes a tedious job with multiple uncertain variables. Instead, most managers opt for best case, worst case and most likely case evaluations, which still lack any sense of probability of occurrence.
Estimation provides a value that is as close as possible to the actual (unknown) value, based on some definition of goodness or quality including unbiasedness, minimal quadratic error and usage of linear combination of data.
Monte Carlo Simulation – Applications
Spreadsheet or estimation does not have any answer when the variables in a business model exceed their threshold value. Because of complex real-life business scenarios, it is very difficult to assign specific values to variables in a model and most of them will have a tendency to cross their threshold values. The reason behind this is the increase in the degree of uncertainty of the variables. The value of a variable in a business model is likely to have more uncertainty because its dependence on information is ever increasing. And as previously stated, almost all information is a question of uncertainty.
Monte Carlo simulation is a technique that uses random number generation to simulate reality. The benefits of a simulation modeling approach are: 1) an understanding of the probability of specific outcomes, 2) the ability to pinpoint and test the driving variables within a model, 3) a far more flexible model, and 4) clear summary charts and reports.
With the help of such techniques, managers may have the ability to replace each uncertain variable with a probability distribution, a function that represents a range of values and the likelihood of occurrence over that the range. Monte Carlo simulation uses these distributions, referred to as assumptions, to automate the complex what-if process and generate realistic random values. Instead of modeling a single uncertain outcome, managers can quickly generate thousands of possible scenarios, view the result statistics, and evaluate risks. The ability to quantify risks is indeed a crucial tool for a successful negotiation.
Figure 2: Probability Distribution Function
The selection of the probability distribution functions (commonly called the pdf) in a Monte Carlo Simulation is extremely important. The crucial factor behind is that the pdfs that are used in any simulation must reflect reality. Figure 2 shows a typical pdf for a variable X. The horizontal axis is the value of the variable X and the vertical axis is the frequency of X. It can be seen that the highest frequency occurs for a single value of X. This value of X for which the frequency is largest is defined as the most likely value of X. The minimum and maximum values of X (XMin and XMax) are at the end-points of the curve. In most real-world systems of measured values, the XMin and XMax values also will have minimum frequencies.
Major pdfs that can be used in different situations include Gaussian, Poisson, Flat and Triangular. The choice of the pdf depends on the industry characteristics and the nature of the variable. For example, a Poisson distribution is used for the failure of manufacturing or process components. This pdf serves as a useful probability model where the only input required is the average rate of occurrences per unit time or space. The variance (a measure of dispersion) is always equal to the mean (a measure of central tendency) of a Poisson distribution.
Computer Support for Advanced Decision Analysis
These advanced mathematical techniques form the basis of decision analysis, and computer support is essential to make the process easy to use and readily applicable by managers in different organizations. These software packages use MCDM techniques and Monte Carlo simulation to build business models and help managers make effective decisions by providing the available information in a more quantitative way. The availability of powerful computing capabilities has allowed broader use of quantitative decision analysis methods in intensive decision-making contexts.
Application of Decision Analysis in Different Industries
Banking: These days major international banks use decision analysis as a framework for making cost-effective risk management decisions. They develop and extensively apply a risk management methodology that uses tools such as decision trees, judgmental probabilities and simulation to analyze operational risks in probabilistic fashion.
Manufacturing: Recent trends in the manufacturing sector show widespread applications of decision analysis tools for issues such as new product development (NPD), market entry and exit strategy and product layer introduction. The management committee in leading firms within the industry use cognitive models to identify the factors that are important for them to make decisions on issues such as NPD. Extensive market research forms the basis of such models and decision analysis software is used to build business models and desired results are obtained.
Consulting: Management consultancy has been at the core of application of decision analysis tools for strategy formulation. Consultants create, analyze, choose and implement business strategies by incorporating decision analysis tools and creating structured interactions between the management decision board and the strategy development team.
Medical: Decision analysis tools (e.g., multiattribute utility analysis and decision trees) are now being extensively used in medical decision making. Issues such as effect of time pressure on decision making is a major issue in this context as medical doctors in emergency departments are under tremendous time pressure. Much research is being carried out in the field to come up with readily available decision support. As an initial step toward building up processes for structured decision problems in this context, medical doctors are being trained by professional decision analysts to equip them with tools and knowledge of decision analysis that may help them make better judgements.
Advanced decision analysis tools are used to distribute the uncertainty in a business scenario and thereby quantify it through the use of probability distribution functions. In other words, applications of different decision-making tools are steps taken up by managers to clearly identify and analyze the different attributes and alternatives to a specific problem. In absence of such tools, managers build up mental models and most often human brains are not capable enough to consider and judge all the alternatives. Instead, they create a simpler version of the original problem inside their brain and choose an alternative which may not be the best one. The different decision-making models help the decision-makers look at the problem from a more explicit and analytical viewpoint. The tools cannot solve a decision problem but can definitely help the decision-makers ask the right questions.
Further Applications of Decision Analysis
Decision analysis can be applied in any field of management/finance/economics which needs to deal with multiple factors and consequently come to a conclusion regarding a decision (among a certain number of alternatives) by assigning weights or by prioritising judgements. A macroeconomic topic of foreign direct investment (FDI) involves consideration of factors such as social systems, time, uncertainty, goals and constraints while deciding upon a particular project. Making a major decision among such complicated alternatives and attributes is a difficult task. In such complex situations, structured approaches of decision analysis become crucial for businesses.
Riddhi Dutta worked in manufacturing industry in operations after graduating with a degree in Economics. He earned his MBA degree in Finance in September 2002 at Leeds University Business School, U.K. As an intern in the IT/IS Department of Leeds University, he designed and constructed a decision support system for the evaluation and prioritization of IS projects. Dutta’s career and research interests involve applications of decision analysis tools within different industrial sectors. You can contact him at firstname.lastname@example.org.