You are on page 1of 4

DISTRIBUTIONS, SIMULATION, AND EXCEL FUNCTIONS What Is A Distribution?

A distribution is a mathematical function showing the location of the probability for all events that can occur. Table 1 shows two kinds of distributions for a random variable X. Probabilities must sum to 1.0000 (for a discrete variable) or the area under the function must be 1.0000 (for a continuous variable). Table 1. Discrete and Continuous Distributions
Type of Distribution Discrete variable Characteristics X has countable, distinct values (e.g., X = 13). X is measured on a continuous scale (e.g., X = 84.37). Example The historical billing error rate is 3%. 300 invoices are mailed. How many will contain errors? This variable is an integer that follows a binomial distribution. To monitor quality on a daily basis, a hospital emergency facility records the percentage of patients treated within 30 minutes. This ratio variable follows a normal distribution.

Continuous variable

Finding Probabilities In the old days, we found probabilities from an equation or a table. Now we rely on spreadsheet functions. Table 2 shows some examples. In Excel, use Insert | Function | Statistical to see a menu of distributions that are available, as illustrated in Figure 1. Table 2. Examples of Probabilities Using Excel Functions
Distribution Normal Example The PalmPuppy has a 14 hour battery life with a standard deviation of 3 hours. Find the probability that the battery will last 11 hours or less. 150 employee travel expense claims are processed. The historical rate of overstated expenses is 10%. Find the probability of 7 or fewer over-claims. Excel Function
=NORMDIST(11,14,3,1)

= 0.1587 or about 16%. Note: The 1 requests a cumulative probability.


=BINOMDIST(7,150,0.10,1)

Binomial

=0.2026 or about 20%. Note: The 1 requests a cumulative probability.

Figure 1. Inserting an Excel Function

Distributions, Excel Functions, and Simulation - Updated June 11, 2003

Page 1 of 4

When you click the name of the distribution, a screen like Figure 2 will guide you through the process of pasting the desired function in the cell. In this illustration, we are inserting a binomial function in cell B2. Cell A2 (whose value is 0 in this illustration) is the X value for the binomial distribution, n = 4 is the number of trials, p = 0.2 is the probability of success, and 0 indicates that we do not want a cumulative distribution. The resulting probability (0.4096) is shown in the lower right of this preview screen.

Figure 2. Binomial Function in Excel When you click OK the formula =BINOMDIST(A2,4,0.2,0) will appear in cell B2 (the designated cell) as shown in Figure 3. The value of this function is displayed in cell B2, and its formula is shown in the edit window at the top of the screen. Once you learn the syntax of the function, it may be quicker to type the formula directly into the cell (be careful to start with = to let Excel know that you are entering a formula). If you type a function correctly, it will be capitalized; otherwise, the cell will display #NAME?

Figure 3. Cell Formula for a Binomial Probability Monte Carlo Simulation Excel isnt optimized for statistics, so other vendors have created add-ins offering more features (e.g., Crystal Ball, @Risk). They not only calculate probabilities, but also permit Monte Carlo simulation to draw repeated samples from a distribution. Using simulation, a deterministic value (e.g., a collection rate) can become a stochastic variable. We can then study the impact of changes in the variable on the rest of the spreadsheet. Monte Carlo simulation requires hundreds or thousands of iterations. Each sample yields one possible outcome for the variable(s) of interest. By studying the distributions of results, we can see the range of possible outcomes and the most likely results.

Distributions, Excel Functions, and Simulation - Updated June 11, 2003

Page 2 of 4

Advantages of Simulation Simulation has many advantages for risk assessment. Among them are: (a) Simulation forces us to state our assumptions clearly (b) Simulation helps us visualize the implications of our assumptions (c) Simulation reveals the potential variation in the variables (d) Simulation quantifies risk (the probability of a given event). Useful Distributions Here are several probability distributions that are of interest because they correspond to the way managers often think and are available in most simulation software. Table 3. Some Useful Distributions
Distribution Normal Illustration Characteristics The familiar bell-shaped curve. Symmetric, with a peak in the middle and gradually tapering tails. Pro: Familiar, well-known. Con: Extreme outcomes possible. Truncated normal Same as normal but with limits to prevent extreme cases from arising. Pro: No wild outcomes. Con: More complicated.

Triangular

Has a central peak and clearlydefined end points (lowest, most likely, highest). Can be skewed. Pro: Easy to understand. Con: No extremes can occur.

General
0.40 Pr o bability 0.30 0.20 0.10 0.00

Revenue from Asse t Sale

Define any k categories and make sure the probabilities sum to 1. Pro: Easy to understand. Con: Need to create categories.

50

100

200 Re ve nue

500

1000

The corresponding @Risk functions are shown in Table 3. When a cell contains one of these functions, the mean of the distribution is initially displayed (i.e., the expected value). Each time the simulation is executed, the cell will be updated to show a random value drawn from the specified distribution.

Distributions, Excel Functions, and Simulation - Updated June 11, 2003

Page 3 of 4

Table 3. Some Common @Risk Distributions


@Risk Function Normal Truncated normal Example
=RiskNormal(47,2) =RiskTnormal(47,2,43,51)

Triangular General

=RiskTriang(3,8,14) =RiskGeneral(A2:B6)

Interpretation Normal with mean = 47 and standard deviation = 2. Normal with mean = 47 and standard deviation = 2. Lowest allowable value is 43 and highest allowable value is 51 (arbitrarily set at 2). Lowest value is 3, most likely value is 8, highest value is 14. Any distribution you want (in this case X values in column A and P(X) in column B).

Triangular Distribution In a triangular distribution, you choose the minimum, best guess (mode), and maximum. For example,RiskTriang(50,60,100) is right-skewed while RiskTriang(50,90,100) is leftskewed. Managers often visualize unknown variables this way (e.g., raw materials prices, rates of return, or market shares). Sketch a picture to clarify your intended distribution. The outcome cannot lie beyond the limits you specify. Normal Distribution When you specify a normal distribution, there are no limits on X, but the event probabilities are known: 99.71% of the area is within 3 (the three-sigma range) 95.44% of the area is within 2 (the two-sigma range) 68.26% of the area is within 1 (the one-sigma range) For example, when you specify = 50 and = 2, you are saying that the effective range of your data will be 44 to 56 (i.e., 3). Outliers beyond 3 are rare but possible.

Distributions, Excel Functions, and Simulation - Updated June 11, 2003

Page 4 of 4

You might also like