Sunteți pe pagina 1din 6

Capital Budgeting Simulation

Using Excel: Enhancing the


Discussion of Risk in Managerial
Accounting Classes
By Sylwia Gornik-Tomaszewski, DBA, CMA, CFM

EXECUTIVE SUMMARY any leading managerial accounting textbooks


An Excel-based capital budgeting
simulation model that contains a
degree of randomness and uncertainty
can be used to expand classroom
discussions of risk in the context of
capital investment analysis. This adds
M provide only limited coverage of risk in the
context of capital budgeting, often referred
to as capital investment analysis. Capital
budgeting chapters usually focus on widely
used decision models, such as net present value (NPV), inter-
nal rate of return (IRR), accounting rate of return (ARR),
profitability index (PI), and the payback period (PB), and are
a level of detail that is often over- often enhanced with analyzing tax implications in capital bud-
looked by most managerial accounting geting decisions.1 But only the deterministic versions of these
textbooks. models, in which no randomness is involved, are covered.
Cash flows are forecasted as single figures, and their uncer-
tainty is ignored. The risk associated with an investment proj-
ect is expressed in the selected discount rate (required rate of
return). The lesson being told, therefore, is that the greater
the risk that is associated with an investment, the greater the
return that is required.
In practice, however, several different techniques are used
to deal with the uncertainty of investment projects. Firms
might combine NPV with PB when analyzing the total risk of
a project. They also might use sensitivity analysis, scenario
analysis, risk-adjusted discount-rate approach, or simulation.

M A N A G E M E N T A C C O U N T I N G Q U A R T E R LY 12 SUMMER 2014, VOL. 15, NO. 4


These techniques are applied following the assumption rate. Also needed is the firms required rate of return to
that it is relevant to consider a projects total risk when discount the after-tax net cash flow to present value.
evaluating the project and when the returns from the Rows 5 to 15 of the spreadsheet contain assumed data
project are positively correlated with the returns from on price per unit (p), number of units sold (q), unit pro-
the firm as a whole.2 duction cost (c), unit selling cost (s), annual depreciation
Although the leading textbooks do not delve into (D), the firms marginal tax rate (T), and the firms re-
these methods for dealing with uncertainty in invest- quired rate of return (k).
ments, the techniques can be explained easily in the To build uncertainty into the model, assume that the
classroom, especially when supplemented with spread- first three variablesp, q, and care random (stochas-
sheet applications. An educational model developed in tic) in nature. This requires an estimation of the proba-
Excel can provide a simple simulation of the NPV of an bility distribution for each of these random variables.
investment project whose outcome is uncertain. The Although a discrete distribution could be used with a
model is easy to use, requiring only a basic Excel pack- finite, relatively small number of possible values with
age and not the @Risk add-in. known probabilities, it is more convenient to use con-
tinuous distributions in a simulation. The RAND func-
Basic Simulation Model tion in Excel generates various continuous distributions
A simulation model is a computer model that imitates a commonly used in applications. Assume that the three
real-life situation. When applicable in capital budgeting, random variablesp, q, and care normally distributed.
the simulation approach generally is more feasible for Cells C14:E15 contain the mean () and standard devi-
analyzing large projects because the technique requires ation () for each of the three random variables. As you
estimates to be made of the probability distribution of can see in Figure 1, the price per unit (p) is normally
each cash flow element. distributed with = $10 and = $2; the number of
Simulation uses random numbers to drive the model- units sold (q) is normally distributed with = 2,000 and
ing process. All spreadsheet packages are capable of = 300; and the unit production cost (c) is normally dis-
generating random numbers between 0 and 1. In Excel, tributed with = $2 and = $0.25. These small num-
random numbers are generated by entering the formula bers are selected on purpose to make the model easy to
=RAND() in a cell. The random numbers are uniformly understand. Users of the model can use their own
distributedthat is, any number between 0 and 1 has prices, costs, quantities, and determine distributions of
the same chance of occurrence. Also, different random these numbers. It is suggested, however, that certain
numbers generated in one spreadsheet are probabilisti- basic rules should apply, such as assuming p>c and
cally independent, meaning the random value in one <, to avoid complications that obscure the purpose
cell does not affect random values in other cells.3 and focus of this exercise.
It is critical to understand that flux is an important With those assumptions, applying the Excel function
characteristic of simulation. We will have to get accus- =NORMINV(RAND(),,) will generate random val-
tomed to constantly changing numbers because all the ues for p, q, and c based on normal distribution. For ex-
cells containing the RAND function will change each ample, the formula =NORMINV(RAND(),C14,C15) in
time we press the recalculate key (F9) or do anything to cell C5 generates the random price per unit (p) from a
affect calculation. normal distribution. The values for q and c are gener-
Figure 1 presents a simple capital budgeting model ated in a similar manner from their respective means
involving the introduction of a new product. We will and standard deviations.
need to determine annual after-tax cash flows generated Now that the data inputs are defined, the next step is
by the new product. Therefore, the first step in devel- to build the NPV model, seen in cells A17:E26 of
oping the model is to enter the relevant data needed to Figure 1. The model assumes that (1) the initial invest-
compute these cash flows. This includes data to deter- ment necessary to introduce the new product is $10,000
mine revenues, costs, depreciation, and marginal tax (cell C20), (2) the new product will have a life of five

M A N A G E M E N T A C C O U N T I N G Q U A R T E R LY 13 SUMMER 2014, VOL. 15, NO. 4


Figure 1: Capital Budgeting with Simulation

A B C D E F G H
1
2 CAPITAL BUDGETING WITH SIMULATION
3 Data:
4
5 NCF p 10.16 price per unit
6 q 2,292 number of units sold
7 c 2.04 unit production cost (excluding depreciation)
8 s 1 unit selling cost
9 D 2,000 annual depreciation
10 T 40% the firm's marginal tax rate
11 k 10% required rate of return
12
13 p q c
14 Mean: 10 2,000 2
15 StDev: 2 300 0.25
16
17 The NPV Determination:
18 A f e r - ta x
19 Item Year (t) Amount T ax E ffec t NCF t
20 Investment now (10,000) 1 (10,000)
21 1 14,311 60% 10,587
22 2 14,311 60% 10,587
23 3 14,311 60% 10,587
24 4 14,311 60% 10,587
25 5 14,311 60% 10,587
26 NPV $30,132
27
28 Simulation:
29
30 Iteration NPV Frequency table for 100 iterations
31 $30,132.07 N PV Frequency
32 1 $30,450.64 0 0
33 2 $20,031.96 10,000 8
34 3 $33,898.81 20,000 30
35 4 $14,301.81 30,000 30
36 5 $34,231.82 40,000 27
37 6 $19,130.41 50,000 4
38 7 $31,494.15 >50,000 1
39 8 $28,536.55 100
40 9 $38,796.87
41 10 $30,653.14
131 100 $15,188.38
132 Mean $24,329.62
133 StDev $10,808.35
134 Min $1,537.28

years (rows 21-25), and (3) no additional investments determining the desirability of an investment proposal.
will be required during the product lifespan. These as- Depreciation is not a cash flow. Because it affects the
sumptions make the annual after-tax cash flows gener- taxes that must be paid, however, it has an indirect ef-
ated from the sale of the new product identical. They fect on the companys cash flows. The depreciation de-
are computed using the formula NCFt = [q(p) q(c+s) duction shields the revenues from taxation and thereby
D](1T) + D; where NCFt is the net cash flow in period reduces the amount of taxes that the company must
t. Unless a company is a tax-exempt organization, only pay. The reduction in tax payments made possible by
the after-tax amount (column E) should be used when the depreciation tax shield is equal to D T.

M A N A G E M E N T A C C O U N T I N G Q U A R T E R LY 14 SUMMER 2014, VOL. 15, NO. 4


In the Excel model, the after-tax net cash flows are mate of an expected NPV of the project, as well as its
computed in three steps: risk. With this information, it is possible to compute the
(1) The formula in cell C21 is =$C$6*$C$5- probability of achieving an NPV that is greater or less
$C$6*($C$7+$C$8)-$C$9. than any particular value. The calculation is z = (x
(2) Cell D21 contains =1-$C$10, which is the tax Expected NPV)/Standard Deviation, where z is the
effect. number of standard deviations and x is the particular
(3) Cell E21 finishes the calculation by multiplying value we want to check.
the annual cash flows by the tax effect and then For example, if we wanted to compute the probability
adding depreciation. The formula is of achieving an NPV less than or equal to $0 using the
=(C21*D21)+$C$9. results in Figure 1, the calculation would be z = ($0
The formulas used in C21:E21 are copied in rows $24,329.62)/$10,808.35. The result is -2.25 standard de-
22-25. The NPV amount is computed in cell E26 using viations. By referring to a normal distribution table, we
=NPV(C11,E21:E25)+E20. Net cash flows are dis- can determine that the probability of a value less than -
counted at the 10% required rate of return (C11) except 2.25 standard deviations from the mean is 1.22%. Thus,
for the initial investment (E20), which is incurred now, there is a 1.22% chance that the actual NPV will be
at t=0. negative for this project.

Replications of Simulation Frequency Table and Histogram


The calculations to this point show only one possible A histogram, a graphical representation of the data dis-
scenario for the five-year horizon. The $30,132 NPV in tribution, can help us visualize the results from the
cell E26 indicates that this independent project is model and facilitate further analysis. A histogram is a
acceptablethe NPV represents a $30,000 gain (the re- summary graph showing a count of the data points
turn from the project is greater than the cost of capital, falling in various ranges. Consequently, it is a rough ap-
that is the return available by investing the capital else- proximation of the frequency distribution of the data.
where). The project is expected to add value to the Users of the model may then examine the histogram for
firm, and it should improve shareholders wealth, but the general shape of the frequency distribution and its
this might be a result of a lucky set of random numbers. symmetry. This may give them a better understanding
The next step is to check this by replicating the basic of the data and help them form expectations regarding
simulation 100 times. the probability of various outcomes. For instance, in our
To do this, a data table is created in the range example, the users may gain a better understanding of
A30:B131. Column A (A32:A131) labels each iteration, how probable it is that the project would yield a nega-
1 through 100. To begin with the existing NPV calcula- tive NPV and, therefore, should not be accepted.
tion, the formula =E26 is entered into cell B31. Next, Although a histogram option is available in the
highlight cells A31:B131. From the Data tab in the Analysis Tools (Data Analysis) in Excel that provides a
Excel ribbon, click on the What-If Analysis dropdown one-step method for generating a histogram, that option
menu and select Data Table. This will bring up the does not retain a link to the data that was used to gen-
Data Table dialog box. Leave the Row Input Cell field erate it. That is, the histogram will remain static and
blank. For the Column Input Cell, select any blank cell will not change anytime the spreadsheet recalculates.
in the worksheet (such as C32). Click OK. This opera- The following method does not have this drawback.
tion tricks Excel into repeating the NPV model calcula- Create a frequency table of the data. This will pro-
tions 100 times, each time using new random numbers. vide data ranges for our histogram and retain a link to
Below the data table, formulas have been added to the data. In Excel, this can be done using the FRE-
show the mean, standard deviation, and the minimum QUENCY formula, which calculates how often certain
and maximum values of the 100 iterations (cells values occur within a specified range and then returns a
A132:B135). These simulation results provide an esti- vertical array of numbers. The syntax for the formula is

M A N A G E M E N T A C C O U N T I N G Q U A R T E R LY 15 SUMMER 2014, VOL. 15, NO. 4


Figure 2: Histogram of NPV After 100 Replications

Histogram of NPV

40

35

30

25
Frequency

20
Frequency

15

10

0
0 10,000 20,000 30,000 40,000 50,000 >50,000
NPV

=FREQUENCY(data_array,bins_array), where data_ and $10,000, how many are between $10,000 and
array is the data set that contains the values for which $20,000, and so on up to values greater than $50,000.
we want to count frequencies and bins_array is an array Cells E32:E37 contain the upper limit for each interval,
containing the intervals we want to use to group the so they represent the bins_array. Note that cell E38 lists
values. >$50,000. This will be the label for the extra element in
Because FREQUENCY returns an array, it must be the returned array.
entered as an array formula. That requires selecting a Next we need to select the range G32:G38. Type in
range of adjacent cells into which we want the returned =FREQUENCY(B32:B131,E32:E37) and then press
distribution to appear. Furthermore, the number of ele- CTRL+SHIFT+ENTER, which enters it as an array
ments in the returned array will be one greater than the formula. If the formula is not entered as an array for-
number of elements in bins_array. The extra element in mula, there will be only one result in cell G32. If
the returned array returns the count of any values above entered correctly, results will appear in each of the cells
the highest interval.4 selected.
The frequency table appears in cells E30:G39 in With this frequency table formed, it is now easy to
Figure 1. For this example, it shows how many NPV create a histogram. Select the frequency table cells
observations are negative, how many are between $0 (E32:G38) and create a column chart (see Figure 2).

M A N A G E M E N T A C C O U N T I N G Q U A R T E R LY 16 SUMMER 2014, VOL. 15, NO. 4


The categories from column E of the frequency table in tional managerial accounting curriculum.
Figure 1 serve as the labels for the horizontal axis in
Figure 2. Notice that this method of generating a his- Sylwia Gornik-Tomaszewski, DBA, CMA, CFM, is an as-
togram retains the link to the data. Any changes in the sociate professor in the Department of Accounting and
data will also change the chart. To freeze the results, se- Taxation of The Peter J. Tobin College of Business at St.
lect and copy the range of cells, then use Paste Values Johns University in Queens, N.Y. You can contact Sylwia at
to replace the formulas with the actual numbers. (718) 990-2499 or gornikts@stjohns.edu.

Use in the Classroom Additional References


This model was used in several graduate managerial ac- Enterprise Risk Management Initiative Staff, Impact
counting classes. It was usually discussed at the end of of Risk Management Failures on the Financial
a lecture on capital investment decisions. After a short Crisis, http://erm.ncsu.edu/library/article/financial-
introduction that described the model and its purpose, crisis-failures#.Us9gbNJDvNk, January 3, 2011.
the model was built in real-time, and simulations were Michelle M. Harner, Ignoring the Writing on the Wall:
run. Students also watched the construction of the fre- The Role of Enterprise Risk Management in the
quency table and the histogram. They were usually Economic Crisis, Journal of Business and Technology
very interested in the topic and fascinated by the ever- Law, October 15, 2009, p. 45.
changing simulation output. The spreadsheet was later Jennifer Saranow Schultz, An Education in Risk
posted online so students could further explore the in- Management Can Offer a Leg Up, The New York
tricacies of the model on their own. Times, August 20, 2009, p. F2.
To reinforce students learning and their interest in
simulation, a homework project was assigned that re- Endnotes
quired students to develop a different model. Students 1 For example: Charles T. Horngren, Srikant M. Datar, and
Madhav T. Rajan, Cost Accounting, 15th edition, Prentice Hall,
should be made aware that when they try to build the 2014; Eric Noreen, Peter C. Brewer, and Ray H Garrison,
model using their spreadsheets, each result will un- Managerial Accounting for Managers, third edition, McGraw-Hill,
2014; Carl S. Warren, James M. Reeve, and Jonathan Duchac,
doubtedly be different. No two answers will ever be Managerial Accounting, 12th edition, Cengage Learning, 2014;
exactly the same. This fact should stimulate critical and Susan V. Crosson and Belverd E. Needles, Managerial
thinking and creativity. Accounting, 10th edition, Cengage Learning, 2014.
2 R. Charles Moyer, James R. McGuigan, Ramesh P. Rao, and
William J. Kretlow, Contemporary Financial Management, 12th
Further Uses of the Model edition, Cengage Learning, 2012.
3 Wayne L. Winston and S. Christian Albright, Practical
The simple capital budgeting example presented in Management Science: Spreadsheet Modeling and Applications, fourth
Figure 1 can be extended in a variety of ways. For ex- edition, Cengage Learning, 2011.
4 Ibid.
ample, more of the input variables could be made ran-
5 For example, users may use the RAND function to generate
dom, or the probability distribution for individual vari- random numbers from such distributions as: (1) uniform
ables used in the simulation may assume a shape other distribution between a and b (the Excel formula is =a + (b-
a)*RAND(), where the known values are substituted for a and
than normal distribution.5 In addition, the NPV model b); and (2) exponential distribution with mean m (the Excel
can be iterated hundreds or even thousands of times, function is =-m*LN(Rand()), where the known value is
substituted for m).
and simulation may enhance capital budgeting analysis
based on other techniques and decision rules.
The main objective of using this model is to demon-
strate to students that a project contains risk whenever
more than one return is possible, and they need to un-
derstand that risk is deeply inherent to capital invest-
ment decisions. Enhancing capital budgeting analysis
with risk consideration should add value to the tradi-

M A N A G E M E N T A C C O U N T I N G Q U A R T E R LY 17 SUMMER 2014, VOL. 15, NO. 4

S-ar putea să vă placă și