Analysis of risk (probabilistic models and decision trees)

Welcome to the World of Risk and Uncertainty!

Hello there! Welcome to one of the most practical and interesting parts of the P2 Advanced Management Accounting syllabus. In earlier studies, you might have looked at numbers as if they were set in stone. But in the real business world, we rarely know exactly what will happen tomorrow. Will sales be high? Will material prices skyrocket?

In this chapter, we are going to learn how to stop "guessing" and start using logical, mathematical tools to manage the unknown. We will explore probabilistic models and decision trees to help managers make the best possible choices even when the future is "fuzzy." Don't worry if you aren't a math whiz—we will break this down step-by-step!

1. Risk vs. Uncertainty: What’s the Difference?

Before we dive into the calculations, we need to understand two key terms that CIMA loves to test. They might sound the same, but in management accounting, they are very different:

Risk: This is when we don't know the exact outcome, but we do have enough past data or experience to assign probabilities (like a 60% chance of success). Example: A casino knows the "risk" because they know the mathematical odds of the house winning.

Uncertainty: This is when the future is a complete "black box." We have no past data and cannot assign reliable probabilities. Example: Launching a brand-new, revolutionary technology that has never existed before.

Quick Review: The Difference

• Risk: Probabilities are known. We use Expected Values and Decision Trees here.
• Uncertainty: Probabilities are NOT known. We use tools like Maximax, Maximin, and Minimax Regret (which you may see in other chapters).

Key Takeaway: We use probabilistic models when we can assign a percentage chance to different outcomes.

2. Expected Values (EV)

The Expected Value (EV) is a weighted average of all possible outcomes. It tells us: "If we made this same decision 1,000 times, what would the average result be?"

The Formula

The formula for EV is: \( EV = \sum px \)

Where:
p = the probability of an outcome
x = the value (profit or cost) of 그 outcome
\(\sum\) = the sum of all these calculations

Real-World Example: The Coffee Shop

Imagine you run a coffee shop. You are deciding whether to stock a special seasonal cake.

• If demand is High (40% chance), you make $500 profit.
\n
• If demand is Low (60% chance), you make $100 profit.

Calculation:
\( (0.40 \times \$500) + (0.60 \times \$100) \)
\( \$200 + \$60 = \$260 \)

The Expected Value is $260.

The "Catch" with Expected Values

Don't worry if this seems strange: In the example above, you will never actually make $260. You will either make $500 or $100. The EV is just a long-term average. This is a common point of confusion for students!

Pros and Cons of EV

\n
• Pro: It reduces complex data into a single, easy-to-compare number.
\n
• Pro: It is great for repetitive decisions.
\n
• Con: It ignores the risk appetite of the manager (it assumes they are "risk-neutral").
\n
• Con: The average might never actually happen (as seen in our coffee shop).
\n

Key Takeaway: Use EV to find the "mathematically best" option over the long term, but remember it doesn't show the "spread" of risk.

3. Sensitivity Analysis

Sensitivity analysis is basically asking "What if?". It looks at how much a specific variable (like sales volume or material cost) can change before our decision becomes a bad one.

How to calculate it

The simplest way to calculate sensitivity for a variable is:
\n\( \text{Sensitivity \%} = \left( \frac{\text{Net Profit (or Margin of Safety)}}{\text{Present Value of the variable being tested}} \right) \times 100 \)

The Rule of Thumb: The lower the percentage, the more sensitive (risky) that variable is. If a 2% drop in sales makes your project lose money, that's a very "sensitive" and risky project!

Common Mistake to Avoid

Students often think a high sensitivity percentage is bad. It’s the opposite! A 50% sensitivity means the variable has to change by a huge amount before you're in trouble. That’s a "safe" cushion.

Key Takeaway: Sensitivity analysis helps identify which variables are the most critical to watch.

4. Decision Trees

A Decision Tree is a map of a multi-stage decision. It’s perfect when one choice leads to another event, which leads to another choice.

The Symbols You Need to Know

\n
• Square (\(\square\)): A Decision Point. This is where the manager chooses what to do.
\n
• Circle (\(\bigcirc\)): An Outcome Point (or Chance Node). This is where "fate" or the market decides what happens (probabilities go here).
\n

The "Roll-Back" Method (Step-by-Step)

To solve a decision tree, we work backwards (from right to left). This is often called "pruning" the tree.

\n
1. Draw the tree: Start from the left with the first decision and branch out to the right.
\n
2. Add values: Put the profits/costs at the end of the branches (the "leaves").
\n
3. Calculate EVs at circles: Work from right to left. At every circle, calculate the EV.
\n
4. Make choices at squares: At every square, compare the options and pick the one with the highest EV (or lowest cost). Cross out the "losing" branches.
\n
5. Final Result: The value at the very first square is your best expected result.
\n

Did you know?

Decision trees are widely used in medical diagnoses and legal cases, not just management accounting! They help people stay logical when emotions (or big dollar signs) might cloud their judgment.

Key Takeaway: Always "roll back" from right to left. Use EVs at circles and choose the best path at squares.

5. The Value of Information

Sometimes, we can pay for a market research report or a specialist's opinion to reduce our uncertainty. But how much should we pay for it?

Expected Value of Perfect Information (EVPI)

Imagine you have a perfect crystal ball. EVPI is the maximum amount you would be willing to pay for that crystal ball.

The Formula:
\n\( \text{EVPI} = \text{EV with Perfect Information} - \text{EV without Information} \)

Step-by-Step Calculation:

\n
1. Calculate the EV of the best decision without any extra info (we did this in Section 2).
\n
2. Calculate the EV with perfect info: Assume you know for sure the market will be "High"—what's the best profit? Then assume you know for sure it's "Low"—what's the best profit? Weight these "best case" results by their original probabilities.
\n
3. The difference is your EVPI.
\n

Expected Value of Imperfect Information (EVII)

In the real world, market research is rarely 100% perfect. This is EVII. It uses the same logic as EVPI, but the probabilities are adjusted (using something called Bayes' Theorem, though CIMA P2 usually focuses more on the application of the value rather than complex Bayesian calculations).

Important Rule: You should never pay more for info than the value it adds. If the EVPI is $5,000 and the researcher wants $6,000, say "No thanks!"

Key Takeaway: Information value = (New EV) minus (Old EV). It's the "pay-off" for being better informed.

Final Encouragement

Don't worry if Decision Trees or EVPI feel a bit like a puzzle at first. The key is to practice drawing the trees and labeling them clearly. Once you get the "Roll-Back" logic down, you'll find these are some of the most "bankable" marks in the P2 exam!

Quick Review Box:
- Risk = Probabilities known.
- EV = \( \sum px \).
- Sensitivity = Margin / Variable.
- Square = Decision; Circle = Probability.
- Information Value = (Profit with info) - (Profit without info).