Financial Modelling in Uncertain Times

Size: px
Start display at page:

Download "Financial Modelling in Uncertain Times"

Transcription

1 Financial Modelling in Uncertain Times John Tennent 5 December 2012 Objective To explore some practical techniques in Excel that assist in evaluating the robustness of investment evaluations to endure uncertain times All techniques can be applied in Excel 2002 (version 10), through to 2010 (version 14) except Sparklines. Statistical part requires the Analysis Toolpak Add In With everyone having different capability there is no easy place start! Will use Functions and Macros Macros do not work well on Apple! 1

2 John Tennent Corporate Edge John Tennent Chartered Accountant Corporate Edge Economist Books +44 (0) Corporate Edge Business Consultancy Business Training Business Modelling Training Model Creation (0) Belief and Trust What enables you to Believe and Trust an investment evaluation? Charles Blondin walking over Niagara Falls on a tightrope in 1859 Belief = evidence of a convincing concept, capable team and persuasive forecast of results Trust = confidence in judgement through weight of evidence What is an acceptable probability of failure? 2

3 Increasing Confidence Three techniques: Building a dynamic model Dynamic meaning that a change in any one assumption will always generate a valid answer. Multiple scenarios and Monte Carlo analysis Explore 10,000 scenarios and use statistics to understand the probability of failure Turning a model from being deterministic to stochastic Dependency ranking List assumptions in order of their importance and support with evidence of their validity Example Model A simple model to look at Excel principles (not operating complexities) Scenario: A 10 year plan for a new pub restaurant The Pen & Quill 3

4 Fundamental Structure for a Dynamic Model Build the model with maximum flexibility for scenarios Inputs Assumptions Workings Outputs Answers Scenarios What if? Options Alternatives Monte Carlo Values Measures Probabilities Sensitivities Standard Deviations Fundamental Structure for Building a Dynamic Model Build the model to be robust and reliable - Destroyed by hard coded numbers in the Workings or Outputs Inputs Workings Outputs Only Data - Never ever any calculations Only Calculations - Never ever ever ever any numbers Cells are free so keep formulae simple Summarise the Workings for Presentation 4

5 Tour of the Model Structure of three sheets Cell names in formulae Sparklines Scenarios and the use of OFFSET Headings in column A and B for quick movement Unit definition and totals Consistent column letter for each year (year 0 always in column H) Using OFFSET Indirect cell referencing to enable data changes = OFFSET (Anchor cell, Rows Away, Columns Away) 1 2 A B C D E F G =Offset(C2,0,B1) 5

6 Monte Carlo Simulation Instead of an input assumption have a range where the assumption will lie and use random numbers to select actual assumption and thus explore arena of possible outcomes A B C D 1 Max Min Random 2 Customers =Round(MIN(B1:C1)+(RAND()*(MAX(B1:C1)-MIN(B1:C1))),0) BUSINESS NB WITH you CONFIDENCE can use =NORMINV(Random Number, Mean, Standard Deviation). to set up skewed random numbers (evenly around a centre point or biased to one end) Write a Macro to Capture Results Set Excel off to capture a set of Randomly generated results From Control toolbox create a button on the sheet. Double click the button to enter Visual Basic and write the following code: For A = 1 To 500 Sheet4.Range("A" & A + 2) = A Sheet4.Range("B" & A + 2) = Sheet3.Range("NPV") Sheet4.Range("C" & A + 2) = Sheet3.Range("IRR") Sheet4.Range("D" & A + 2) = Sheet3.Range("Payback") Application.Calculate Next NB remember to line up the sheet, range and range names with those in your own model! 6

7 Monte Carlo Statistical Confidence Frequency One standard deviation from the mean = 68.3% confidence NPV value Expected Value is the Likely NPV Mean = Expected NPV 1 SD = 68.3% confidence 2 SD = 95.5% confidence 3 SD = 99.7% confidence Tour of the Model Range of assumptions (narrower the better) Data captured on its own sheet Use =Average and =STDEV to calculate the statistics on the data captured What proportion of results give an unacceptable investment? Use =NormDist to create a graph of results 7

8 Dependency Ranking By how much does an assumption have to change before the result is unacceptable? The smallest changes are the most important and thus these assumptions need extra validation Use Goal Seek to find the value of assumption that gives an NPV of zero The NPV of the model A value e.g. Zero A cell with a number not formulae Set up the Input Sheet for Dependency Ranking Three columns to capture the data (plus two more to calculate the degree of change and to rank them) Column K : Base data for the model Column L : The same as K, but one by one each cell will be stress tested for its limit Column M : Where the stress point is for each assumption A K L M 1 Base Data Testing Data Stress Point 2 Customers This column holds the base data This is the column being run through the model 8

9 Set up the Input Sheet for Dependency Ranking Set Excel off to test each assumption individually From Control toolbox create a button on the sheet. Double click the button to enter Visual Basic and write the following code: For B = 11 To 40 If Range("K" & B) <> "" Then Range("L" & B) = Range("K" & B) Next For B = 11 To 40 If Range("L" & B) <> "" Then Sheet3.Range("NPV").GoalSeek Goal:=0, ChangingCell: =Range("L" & B) Range("M" & B) = Range("L" & B) Range("L" & B) = Range("K" & B) End If Next Dependency Ranking Present the results in the form of a Tornado diagram Excel will do this without the need for any Add In. 9

10 Tour of the Model 5 columns for running the ranking Model is run through the Testing column Use =Rank to automatically order the data Use the graph function to create a Tornado diagram (to get labels on left double click vertical axis select Patterns, tick mark labels and low ) Build your own Best way is to have a go yourself We are happy to share model shown here, just me at: John-tennent@corporateedge.co.uk 10

11 Questions 11