USER MANuAL work smarter, not harder

financial modelling

SMART

Build Better Models

w w w. n a v i g a t o r p f . c o m / s m a r t

Introduction
Financial modelling is an unregulated, non-institutionalised professional activity with diverse end-user requirements. To account for the needs of all users and to sustain best practice modelling standards, Navigator Project Finance (Navigator) has developed its own best practice nancial modelling methodology SMART. The SMART nancial modelling methodology has been developed from the extensive experience of Navigators founding principals and their belief that a professional model is characterised by transparency, exibility and presentation. The SMART nancial modelling methodology is practical, rather than theoretical or too prescriptive, and is designed to accommodate the needs of a wide range of users from model developers, managers, debt nanciers, equity investors and model auditors to the CFO. SMART is a set of best practice modelling guidelines that emphasises transparency, exibility and presentation, and introduces rules that underpin how Excel should be used to achieve this. When you model the SMART way, these guidelines guarantee the development of robust and exible nancial models that achieve the design objectives as well as the objectives of all users. This practice lowers model risk and increases condence in the model, as well as condence in the model developers and management team. The SMART methodology will ensure you build nancial models according to best practice guidelines and develop models that are transparent, user friendly and exible. This is what Navigator strives for and why we decided to name our methodology SMART it is just that! This is also why we want to share it with other model developers. The SMART methodology works it really is smart and it will increase your modelling productivity, save you time and money, and help you build better models.

Here is your guide to the SMART methodology its time to work SMARTer!

2 | 20

Table of Contents
Notes from the SMART developers User condence, Model risk Transparency 04 05

1 2 3

The principles of transparency Separate inputs, calculations and outputs Concise, simple formula Minimise embedded logic Minimise the use of named ranges Use control accounts Consistently add cells No circular references, minimise use of macros Flexibility The principles of exibility No hard coding Control scenarios with a dedicated manager Consistent formula per row Build in dynamic timing Build in spare lines Use ex cells to vary whilst preserving Base Case Presentation The principles of presentation Visually identify inputs State dimensions and labels concisely Develop customised style guides Worksheets labelled, grouped, colour-coded and ordered Include integrity checks Use conditional formatting Use data grouping and a clear heading hierarchy Make good use of screen with freeze panes Use line summaries Summarise output using plots

06 06 06 07 07 08 09 09

11 11 12 12 13 13 14

15 15 15 16 16 16 17 17 18 18 19
3 | 20

Notes from the SMART developers

The SMART methodology has been developed and rened by Navigators founding principals and has been used to construct hundreds of professional nancial models across a wide range of industries, globally. It is also the foundation of the nancial modelling courses within Navigators training portfolio. SMART recognises that there is no one size ts all solution to the nancial modelling challenge, but if followed, the SMART guidelines, alongside a handful of rules, will result in a nancial model that encourages condence and efciency across all situations and users. Our nancial modelling experience has illustrated the importance of focusing on ways to: Increase user condence Lower model risk Efciently use the model at all stages of its lifecycle This can be achieved by tackling three key areas areas which we have made the foundation of the SMART methodology:

Transparency

Flexibility

Presentation

In this user manual we walk you through the principles of transparency, exibility and presentation and how you can apply this to your nancial models to increase user condence, lower model risk and increase your modelling productivity. We want to help you work SMARTer, not harder, and build better models.

Nick Crawley,
Managing Director Navigator

4 | 20

User Condence
Financial models and spreadsheets are constructed to calculate numbers and summarise data. A SMART model increases user condence in the model output and ensures that it is mechanically correct. To increase user condence, keep in mind User condence results when a model is easily and quickly understood Multiple end users mean the model must satisfy the lowest common denominator of Excel skills A well presented and consistently formatted model allows for easy navigation and updating Logical and concise formulas are quick to understand and audit Models and projects evolve which means assumption cells should be clearly identied Dynamic scenarios will allow the user to test a variety of model parameters

Model risk
Model risk is a very real concept when evaluating a quantitative structure and needs to be managed through the consistent application of a modelling methodology and an overall risk awareness. To minimise model risk keep in mind Unless a model has undergone a stringent design and peer review process, there will be errors It is usually quicker to rebuild than to check or adapt a model with no best practice form Complex mathematical solutions might result in a correct answer but how will the user know? Complexity is expensive to audit whereas transparency through using many rows is free

5 | 20

Transparency
Transparency is one of the three key principles for stress-free modelling. A transparent nancial model is clear and easy to understand and use; it enables updates and reviews to be made efciently and cost effectively. By following the principles of transparency during formula construction, the model will evolve in a way that inspires user condence and ensures changes can be performed efciently and reliably as the commercial framework evolves.

The principles of transparency

Separate inputs, calculations and outputs Concise, simple formula Minimise embedded logic Minimise use of named ranges Use control accounts Consistently add cells No circular references, minimise use of macros

Separate inputs, calculations and outputs

Inputs, calculations and outputs are the fundamental building blocks of any nancial model and should be presented on separate worksheets and differentiated by both location and formatting. Retrospective application is cumbersome and inefcient, so it is important that this rule is followed from the start of the model construction process. Input worksheets contain the hard-coded assumption cells that a user can update Calculation worksheets contain the logic of the model Output worksheets collate the key outputs to meet the users information requirements

Use separate worksheets for Inputs, Calculations and Outputs. Differentiate your worksheets by using different colours: Red tabs Outputs; Blue tabs Inputs; Grey tabs Calculations.

Concise, simple formula

Important sense checks to apply when constructing formula: 1. Can I do this calculation in my head? 2. The rule of thumb i.e. is the formula longer than your thumb when held up to the screen? 3. Refer to formula precedents that are close to the new formula. Formulas in a SMART compliant model satisfy these rules by being short and simple. More complex arguments are dismantled into several shorter sequences, each on their own line. Precedents are physically located on the same screen, eliminating the need to scroll. When F2 is pressed, the user will see the precedents driving the result of the new formula. Example: =SUM (OFFSET (J1, , , 1, MIN (COLUMN () COLUMN ($I), 12)) This formula is easier to understand and check if it is broken down into two components: Row 1 = MIN (COLUMN () COLUMN ($I), 12) Row 2 = SUM (OFFSET (J1, , , 1, the result of row 1)

6 | 20

Transparency
Minimise embedded logic
Errors are commonly found when logical functions such as IF, AND and OR are embedded within each other. Often these can be usefully combined, such as: IF (AND (A=True, B=False) , 1, 0) However, for anything more complex it is best practice to break the logic down into one or more lines. This risk is especially evident when embedding multiple IF statements, so dont if you cant visualise or solve it in your head, dont build a formula to do it. This might sound extreme but the principle of simplicity really works. Example: =IF(Q267=0,0,IF(AND(Inputs!$F$263>Q391,Q391>0,Q521>0),ROUND(MIN(Q245+Q247+Q248,Q52 1),1),IF(AND(Inputs!$F$264>Q429,Q429>0,Q521>0),ROUND(MIN(Q245+Q247+Q248,Q521),1),IF(AN D(Inputs!$F$265>Q445,Q445>0,Q521>0),ROUND(MIN(Q245+Q247+Q248,Q521),1),0))))
Before: The above formula has many parts and embedded IF statements..

After: A formula that is broken down and displayed on several rows is easy to understand and transparent. J67 = IF(AND(J119>0,$E67>J119),1,0)*J16 J68 = IF(AND(J144>0,$E68>J144),1,0)*J16 J69 = IF(AND(J155>0,$E69>J155),1,0)*J16 J70 = MAX(J67:J69)

Minimise the use of named ranges

Applying a name to a single cell or a range can improve readability of a formula; this is especially useful when applied to constants such as conversion factors. However, it is easy to overuse named ranges, which can make interrogating formula cumbersome. Examples of instances to apply named ranges: Numbers used for conversions These include multiples of 10 and converting technical units. = J12 * J23 / 1000000 = J12 * J23 / Million

Using named ranges for conversions such as weight conversions reduces the risk of incorrectly applying the constant and validates the formula.

7 | 20

Transparency
Calendar periods replace with short, descriptive named ranges = J24/365 = J24/Days_in_yr = J55/4 = J55/Qtrs_in_yr

Use a named range for the days in year constant to make the formula above converting an annual rate into a per period rate transparent.

Names that are likely to be often updated and prone to spelling errors: month names project or asset names

Use control accounts

To facilitate the construction of nancial statements and analysis, it is best practice to use control accounts to illustrate the balance at the beginning of a period, movements within it, and the closing balance at the end of the period. This method also forces the model developer to consider what is being added and subtracted during the period and is particularly useful for: debt facilities and repayments stock/inventory balances tax creditor and losses working capital.

Before: Without control accounts and proper formatting it is difcult to identify the outstanding balance of interest.

After: When using control accounts and proper formatting, the interest expense from an accounting, cashow and balance sheet perspective is easily identied. Tracing precedents illustrates the ow of calculations between each period.

8 | 20

Transparency
Consistently add cells
Consistently add cells and line totals by displaying negative quantities or movements as negative values added to positive line items, rather than mixing the use of addition and subtraction in a single formula. For graphing purposes a separate data area may need to be set up which reverses any negative quantities for consistent plotting.

Before: Mixing additions and subtractions increases the model risk and makes it harder for users to follow model calculations.

After: Consistently adding cells and displaying negative numbers in brackets makes it easier to check model calculations which in turn decreases model risk.

No circular references, minimise use of macros

Circular references are generally caused by incorrect coding rather than the requirement for true circular logic. It is important to differentiate between a circular reference, which is a computational alert within Excel, and circular logic. For example, circular logic occurs when: A=B+C C=A/B Circular logic, although not desirable, is acceptable when it is managed appropriately using a transparent, documented and user-controlled macro.

9 | 20

Transparency
When modelling a concept that results in a circular reference, it is essential that the model developer is able to step away from Excel and rationalise with pen and paper why the state has been encountered. This solves most instances of circular references.

The above example illustrates a circular reference triggered by interest incorrectly calculated on the debt closing balance. Excel cannot calculate when it is in circular reference mode.

By isolating circular logic, a developer will be alerted if an additional circular reference is incorporated accidentally. The user will be able to readily know how many circular processes are included in the model and the status of their convergence. There are occasions where macros cannot be avoided, such as when this is requested by the end-user or in forward looking controls where absolute accuracy is required. However, it is worth bearing in mind that many lending institutions will not accept models which run macros that have an impact on any element of the nancial operations within the model; the reason being that macros hide their design, unlike on-page formula, and can lead to instability in a model if badly designed. Hence, always avoid non-essential macros and, where a good proxy is available as an alternative to a macro, use it.

10 | 20

Flexibility
The primary purpose of most nancial models is to analyse the impact of changes to underlying drivers as commercial structures vary from a dened Base Case. A exible model can accommodate different scenarios quickly and reliably as it can be updated without the need to undertake signicant structural changes. Flexibility therefore needs to be incorporated from the outset in a way that doesnt permanently affect the Base Case.

The principles of exibility

No hard coding Control scenarios with a dedicated manager Consistent formula per row Build in dynamic timing Build in spare lines Use ex cells to vary whilst preserving Base Case

No hard coding
In a well-designed model, the end-user can control all sensitivities and therefore perform analysis from one or a couple of clearly dened inputs pages. If a user hard-codes in a calculation worksheet then this functionality is lost and as a result the user has opened himself/herself up to a wide range of potential future errors, especially if that itinerant piece of hard coding is not clearly dened. The following hard-coded example is difcult to update and run a simple 10% sensitivity on: Revenue 2011 = 25000 * 65.10 Revenue 2012 = 25500 * 64.50 In this example it is difcult, when going back to edit, to remember and understand what the initial formula was meant to represent. This makes the formula inexible and the user will have to be very careful when changing components of the formula to ensure the order of magnitude remains correct. An example of a formula with hard coding that will be queried by an auditor: Cost 2011 = (2051 1688) * 0.88 + 34110/4 A third party looking at the formula above will nd it difcult to check that it is correct, which means that each gure has to be queried and rechecked.

No hard coding means the end-user can control all sensitivities and assumptions by changing a few assumptions, rather than updating each calculation cell.

11 | 20

Flexibility
Control scenarios with a dedicated manager
A common use of a model is to run what-if scenarios by changing certain assumptions using ex cells. The ex cells are controlled using a well-laid-out scenario manager. By incorporating a ex cell in the model, the inputs can be adjusted by the desired percentage without having to change the Base Case inputs. For example, a user may wish to test what happens to the Net Present Value (NPV) when capex increases by 5%.

A well-laid-out scenario manager increases usability and is set up with pre-dened scenarios to quickly adjust assumptions.

The results of each scenario can be presented in dynamic tables where different scenarios can be analysed and compared.

Consistent formula per row

Worksheets can generally be categorised into two types time-dependent and time-independent. For time-dependent worksheets it is crucial that there is only one formula assigned to each row. When this rule is followed, to change or check an entire row, only one formula needs to be edited or reviewed and then copied across/down. There are no exceptions to this rule.

Formulas are copied across and down which enables quick auditing as only the unique formulas need to be checked.

12 | 20

Flexibility

Above is the result of running maps of the formulas in a well-laid-out worksheet. Unique formulas and numeric inputs should be minimised and checked.

Build in dynamic timing

Building dynamic timing architecture is good modelling practice and enables robust exibility. The use of Excel dating functions such as EOMONTH allows a simple structure to be built that, with the use of conditional formatting, provides a clear indicator of key project periods such as planning, construction and operations. The dynamic timing structure can be complemented by binary timing ags which further identify what period of a project is underway at any given time. Binary ags allow the developer to create a formula which is consistent across a row, with relevant components, activated in the correct periods in association with the ags.

Dynamic timing illustrates each project phase using conditionally formatted binary ags.

Build in spare lines

There is always the opportunity for error when a late change needs to be made, and that adjustment has consequences for a number of items in different tabs in the model. For example, during soft completion stage, additional lines of capital expenditure may need to be included. Without best practice modelling this usually means the insertion of new rows on as many tabs as those new capital expenditure items need to be recognised. This could easily be in four tabs Inputs, Construction, Integrated Financial Statements and Annual Financial Statements if not more, making it a very risky last minute change. The SMART methodology attempts to anticipate change and the inclusion of spare lines within Inputs for named lines circumnavigates this problem, minimises model error and increases exibility for the user.

13 | 20

Flexibility

Include spare lines within Inputs to increase exibility and minimise model errors by avoiding the need to add new rows.

Use ex cells to vary whilst preserving Base Case

Variations to the Base Case should be applied using clearly identied ex cells. This method allows, for example, a 10% movement in prices to be applied whilst still preserving the original price assumptions in the model.

Use clearly identied ex cells to run sensitivities to the Base Case.

14 | 20

Presentation
Succinct and consistent presentation of a model instills condence as it signals that care has been taken during model development. The presentation and style of a model also assist users not familiar with the model to understand and operate it effectively. This in turn decreases model risk.

The principles of presentation

Visually identify inputs State dimensions and labels concisely Develop customised style guides Worksheets labelled, grouped, colour-coded and ordered Include integrity checks Use conditional formatting Use data grouping and a clear heading hierarchy Make good use of the screen with freeze panes Use line summaries Summarise output using plots

Visually identify inputs

Ensure input cells are formatted consistently so they are easily identied. This practice makes the model user friendly and will prevent users from overwriting calculation cells.

Clearly identied input cells allow cells to be changed with condence and automatically updated in the calculation worksheets.

State dimensions and labels concisely

Assumptions and calculations should always have dimensions (units) stated clearly and labels entered concisely to ensure the model is auditable. This practice will help the user know which form of input number to enter and avoid converting numbers incorrectly. For example, stating USDM instead of USD allows the user to audit the model by checking that the calculations are being converted to millions.

Units entered in a dedicated column clearly show what dimensions the assumptions are to be entered in, which avoids confusion and errors.

15 | 20

Presentation
Develop customised style guides
To understand what a cell does, it is important to develop a style guide of pre-arranged formatting in the Cell Styles manager. This style guide should be applied by the developer from the start of the model build process. By applying a consistent set of styles, a new user immediately understands the function of different cells, for example, which cells are assumption cells and which cells are key section headers. Creating styles for a workbook ahead of the model development can take time, but the benets are many: it increases model usability and transferability, it reduces errors and makes the model a more marketable and professional document.

Create custom styles in the Cell Styles manager to increase model usability and transferability.

Worksheets labelled, grouped, colour-coded and ordered

To enhance model usability and presentation impact, be careful about the formatting of worksheet labels and order them logically.

Careful formatting of worksheet labels increases usability and presentation impact.

Include integrity checks

Integrity, or functional, checks are paramount in best practice modelling as they allow the user to keep an eye on the robustness of the model architecture and formulas. For example, a classic integrity check is to check whether the balance sheet balances. To automate and make such integrity checks available for the user builds user condence in the model and highlights concerns immediately for resolution. SMART also promotes signal checks as best practice modelling. Signal checks are differentiated from integrity checks in that they are user dened and not designed to test the robustness of the model, but instead highlight limits breached. A classic signal check would be a negative cash balance.

Collate checks into one central area to immediately alert the user when there is an integrity or signal failure.

16 | 20

Presentation
Use conditional formatting
Conditional formatting is used to clearly communicate messages within a model and to illustrate when a particular event or phase is active or inactive. Conditional formatting can also improve the user friendliness of a model by greying out input cells or areas that are irrelevant to the chosen scenario.

Use conditional formatting to communicate key messages and events within a model. Each phase of the project is formatted with different colours to increase visual communication.

Use data grouping and a clear heading hierarchy

Best practice model development includes the development of a well-dened hierarchy of headings which indicates key sections and subsections. When developers apply a clear hierarchy, key headers (i.e. top headers within a hierarchy) can be grouped using the Excel Grouping functionality which ensures that all tabs represent a clear and functional summary of their contents. It is important not to be over-zealous with grouping or heading hierarchies. If avoidable, try not to develop a heading structure with more than four levels or to group within groups this can dilute the clarity that subdividing provides by creating a structure that is too deep.

Use a well-dened hierarchy of headings to indicate key sections and subsections. When the worksheet is grouped, it reads like a table of contents to improve navigation.

17 | 20

Presentation
Make good use of screen with freeze panes
Freeze panes is a common but key Excel functionality and should be used to increase usability. Positioned correctly, the freeze panes tool will allow the user to always see the key titles for each row on the left hand side of the sheet and the important dating and titles elements that, in best practice, should be arranged at the top of any given tab.

Freeze panes allow columns to the right and rows on the bottom of the screen to be viewed with the dates and labels. Use freeze panes to increase usability.

Use line summaries

Where integrity and signal checks provide an automated cross-checking service for the developer or user, the application of line summaries for each row that can be totalled is a way of sense checking the calculations. Line summaries also offer an overview of the magnitudes of operations and nancing within a project and allow the user to communicate this. For example, a line summary for a row that calculates principal repayments on debt is useful as it will indicate abnormalities to the user and encourage further investigation. If the total is below the amount of debt drawn, this indicates that the calculating formula is not acting properly or that the debt is not being amortised before the end of a project.

Comparing before and after. Without any formatting, it is difcult to have a clear idea of what the key numbers in the summary are. After formatting is applied using line summaries and styles, the output is well presented and easy to interpret.

18 | 20

Presentation
Summarise output using plots
Plots (charts or graphs) should communicate the analytical objective of the model. If a model aims to communicate a message regarding, for example, returns, nancing or KPIs, a plot may represent the best tool available for the purpose.

Column charts are useful to break down gures into their key components for interpretation.

Combination charts are useful for relaying a lot of information in one place for comparison and analysis.

Tornado charts are useful to identify the impact of key sensitivities on the project.

19 | 20

www.navigatorpf.com/smart

SMART Resources Tutorials E-books Focus papers Financial modelling blog

work smarter, not harder

Sydney +61 2 9229 7400 | Singapore +65 6440 8803 | London +44 20 3402 6034