What Is Model Sensitivity in Sensitivity Analysis?

What Is Model Sensitivity in Sensitivity Analysis?

8m Read

Anyone looking to leverage financial modeling at their company should understand sensitivity analysis. Also known as “what if” analysis, this modeling technique allows you to preview a wide range of outcomes under different circumstances.

Naturally, a key component of this type of analysis is model sensitivity. How drastically can one change affect your entire business? How will this one input (independent variable) impact your outputs (dependent variables)?

In this post, we’ll take an in-depth look at model sensitivity so you can better apply sensitivity analysis to your own budgeting and business planning.

What is Sensitivity Analysis?

In short, sensitivity analysis looks at how small changes to inputs affect outcomes. For example, a consumer goods company might look at how different prices for a specific product (an independent variable) might affect sales of that product (a dependent variable).

This is accomplished through stress testing, which runs a model through unlikely and drastic situations in order to determine whether a company can survive worst-case scenarios. What if sales dropped by half the following year? Or what if supply chains were disrupted, dramatically limiting the production of goods?

Analysts look at a variety of changes one by one—called a one-at-a-time (OAT) analysis—to better understand how each independent variable impacts not just one outcome but all possible outcomes. The degree to which a dependent variable is affected by a change in an independent variable is called its sensitivity.

The main goal of sensitivity analysis is to help companies better understand which variables most impact their business, as well as the relationships between different variables. After all, in the real world, it’s unlikely that only one variable will change at any given time.

Sensitivity analysis is used by a wide variety of industries but is most often associated with banks and other financial institutions. Investors use sensitivity analysis to discover the “right” prices for stocks (a dependent variable affected by earnings) and bonds (which depend on interest rates).

man computer model sensitivity

Sensitivity Analysis in Practice

Let’s go back to our e-commerce consumer goods company and take a closer look. Last year, they sold 10,000 units of their flagship Product A at $25 per unit, which resulted in $250,000 in revenue.

The company’s financial analyst was tasked with determining the sensitivity of sales to website traffic. The analyst organized previous sales history in the following manner:

Year

Website Visitors

YoY Traffic Change

Unit Sales

Revenue

YoY Revenue Change

2017

40,900

N/A

9,050

$226,250

N/A

2018

45,000

10%

9,500

$237,500

5%

2019

50,000

11%

10,000

$250,000

5%

We can see that a 10 percent increase in website traffic led to a 5 percent increase in sales. The analyst could conclude that revenue is fairly sensitive to website traffic.

In order to test this assumption, the financial analyst looks at how pricing changes have affected sales of Product B (the price of Product A has stayed the same for several years). The sales history of Product B looks like this:

Year

Price

YoY Price Change

Unit Sales

Revenue

YoY Revenue Change

2017

$16

N/A

7,500

$120,500

N/A

2018

$18

12.5%

6,500

$117,000

-3%

2019

$20

11%

5,300

$106,000

-9%

From this data, we can see that both two-dollar price hikes led to revenue losses. This was most noticeable in 2019 when sales of Product B dropped 9 percent. For Product B, it’s easy to see that revenue is very sensitive to price changes.

This is just one example of why sensitivity analysis can be so useful. Often, the relationships between variables are unclear until you run the numbers.

For our e-commerce consumer goods company, a reasonable assumption might be that continued growth in website traffic will yield more revenue growth for Product A. At the same time, they may want to consider lowering the price of Product B to recoup lost revenue.

Sensitivity Analysis vs. Scenario Analysis

You may be thinking, “Sensitivity analysis sounds a lot like scenario analysis. Aren’t they pretty much the same thing?”

Well… not exactly.

Sensitivity analysis looks at how one variable can impact a set of outcomes. Scenario analysis, on the other hand, looks at how different permutations and combinations of both variables and changes to those variables can impact a set of outcomes.

Going back to our e-commerce example, the company’s financial analyst might want to examine a pandemic-related scenario and how revenue might be affected. 

Variables to consider include:

  • Possible delays in obtaining materials for Product A production, which might affect invoice schedules and cash flow.
  • Delivery delays, given the additional strain increased e-commerce orders may place on carriers.
  • Shifts in consumer spending behavior, which means that revisiting a lower price for Product B might not necessarily lead to higher revenues.

For example, when setting up a pandemic scenario, the analyst might assume a 30-day delay in shipments is the most likely one.

directions for model sensitivity

Limitations of Sensitivity Analysis

While sensitivity analysis can be a powerful tool, it does have its drawbacks. Let’s return to our e-commerce consumer goods company. Our analyst already knows that a change in the price of Product B can affect sales.

But what if pricing isn’t the only independent variable that impacts Product B sales?

Upon further investigation, the analyst discovers that a new competitor entered the space in 2018. Through an aggressive marketing campaign and similar pricing (units were sold for $17 each), the analyst can reasonably conclude that this new market entrant had an impact on sales of Product B.

But is the analyst certain he has accounted for every variable that is affecting Product B sales? Because if he isn’t, then he would be overestimating the impact of the competitor’s product sales since he has no way of determining how many units were sold.

While sensitivity analysis is an excellent tool for examining variables one by one in hypothetical scenarios, it can become unwieldy and time-consuming each time a new variable is thrown into the mix. Variables don’t tend to wait around and change one by one just for our convenience.

Common Sensitivity Analysis Tools

The most popular tool by far for conducting sensitivity analysis and building financial models remains Excel. Spreadsheet-based solutions have been around since the 1970s, and are so ubiquitous that nearly all analysts are very familiar with how to use them.

Spreadsheets also require a lot of repetitive, manual entry, leaving little room for error. What’s more, they’re static, two-dimensional visualizations. There’s no easy way to create a multidimensional analysis in a two-dimensional spreadsheet, which is why analysts often build new spreadsheets from scratch each time stakeholders come up with a new set of questions.

This is why organizations should consider using financial modeling tools built specifically for sensitivity analysis and financial modeling. These advanced sensitivity analysis solutions allow analysts to run multiple scenarios and intuitively see the relationships between different variables.

Not only does this help analysts uncover key drivers faster, but it also lets them stress test their models across far more scenarios in a shorter period of time compared to spreadsheets.

Synario, for example, features patented Multiverse Modeling™ software. This technology allows users to run an unlimited number of scenarios in a single model, with minimal manual entry and no need for new spreadsheets.

We built Synario from the ground up as an all-in-one, out-of-the-box financial modeling solution for analysts and decision-makers tired of wrestling with Excel and other spreadsheet-based solutions.

If you find yourself spending hours each quarter double- and triple-checking every cell in your spreadsheets before meticulously building PowerPoint presentations from scratch, only to discover that your numbers are already outdated, you’re not alone.

Maybe it’s time to say goodbye to Excel for good.