CFA Level II Exam · Extensions of Multiple Regression

Dummy Variables in Multiple Regression for CFA Level II

Updated 7 October 2026 · Fact-checked

A dummy variable is a binary (0 or 1) independent variable that marks a qualitative feature, such as a quarter or a group. With n categories you use n − 1 dummies. An intercept dummy shifts the intercept; a slope dummy (an interaction) changes the slope.

Understand Dummy Variables in Multiple Regression

Some drivers of returns are not numbers. Examples are the quarter of the year, whether a firm pays a dividend, or whether a bond is investment grade. A dummy variable turns these into numbers. It equals 1 when the feature is present and 0 when it is not.

If a feature has n categories, you use n − 1 dummies. The category left out is the base (reference) category. If you include all n dummies along with the intercept, the dummies add up to 1 in every row. That is perfect multicollinearity, often called the dummy variable trap, and the regression cannot be estimated.

An intercept dummy changes the intercept only. In Y = b0 + b1D + b2X, the line for D = 1 is parallel to the line for D = 0. The coefficient b1 is the average difference in Y between the category and the base category, holding X constant. The intercept b0 is the expected Y for the base category when X = 0.

A slope dummy (interaction term) lets the slope differ by group. In Y = b0 + b1X + b2(D × X), the slope is b1 for the base group and b1 + b2 for the D = 1 group. You can use both together: Y = b0 + b1D + b2X + b3(D × X). Then both intercept and slope differ.

You test a dummy coefficient with a t-test, like any other coefficient. The null is that the coefficient is zero, meaning no difference from the base category. To test several dummies together, use an F-test.

Key formulas to remember

Number of dummies: Dummies needed = n − 1, for n mutually exclusive categories; Use n − 1 when the model has an intercept. The omitted category is the base.
Intercept dummy model: Y = b0 + b1D + b2X + ε; Base group intercept = b0. D = 1 group intercept = b0 + b1. Slope is b2 for both.
Slope dummy (interaction) model: Y = b0 + b1X + b2(D × X) + ε; Base slope = b1. D = 1 slope = b1 + b2. Intercept is b0 for both.
Combined model: Y = b0 + b1D + b2X + b3(D × X) + ε; D = 1 line: intercept b0 + b1, slope b2 + b3.
Test of a dummy coefficient: t = (b̂ − 0) ÷ s(b̂), with n − k − 1 degrees of freedom; A significant coefficient means that category differs from the base category.

How to solve Dummy Variables in Multiple Regression questions

Use this routine for any item set on dummy variables. Read the vignette for the coding before touching the numbers.

1Identify the categories and which one is the base. It is the one with no dummy in the regression output.
2Write down the coding: which value of each dummy is 1 and which is 0.
3Check the count. A model with an intercept and n categories should have n − 1 dummies. All n means the trap.
4Decide the type of dummy. A dummy alone is an intercept dummy. A dummy multiplied by X is a slope dummy.
5Build the fitted equation for the group asked about by setting the dummies to 0 or 1 and adding the right coefficients.
6For interpretation, state each dummy coefficient as a difference from the base category, holding other variables constant.
7For significance, compare the t-statistic with the critical value, or use the p-value given. Use an F-test for several dummies at once.
8Compute the forecast or difference, and check that the sign and size make sense.

Quickest way: Plug in 0 and 1

When to use it: Use it when the vignette gives a regression table and asks for a forecast or the difference between groups.

Circle the omitted category. Its dummies are all 0, so only b0 and the X terms remain.
For another group, add its dummy coefficient to the intercept. Add any interaction coefficient to the slope.
The difference between two groups is just the dummy coefficient (and the interaction coefficient × X if there is one).
For significance, compute coefficient ÷ standard error and compare it with about 2 for large samples, or use the critical value given.

Common mistakes in Dummy Variables in Multiple Regression

Including a dummy for every category along with the intercept.
It feels natural to give each category its own variable.
Fix: Use n − 1 dummies. The base category is absorbed in the intercept. Otherwise you get perfect multicollinearity.
Reading a dummy coefficient as an absolute level.
Students forget what it is measured against.
Fix: Always say the coefficient is the difference from the base category, holding other variables constant.
Treating the intercept as the average of all groups.
Students ignore the coding.
Fix: The intercept is the expected Y for the base category when all X variables are 0.
Confusing an intercept dummy with a slope dummy.
Both use a 0/1 variable.
Fix: If the dummy is multiplied by X, it changes the slope. If it stands alone, it shifts the intercept only.
Using the interaction coefficient alone as the slope for the group.
Students forget the slope is a sum.
Fix: The group slope is the base slope plus the interaction coefficient.
Judging a dummy by its size instead of its t-statistic.
A large coefficient looks important.
Fix: Test it: t = coefficient ÷ standard error. A large coefficient with a large standard error may be insignificant.

Worked examples

Example 1

An analyst regresses quarterly return on a fund (Y, in %) on market return X (in %) and three quarter dummies Q1, Q2, Q3 (each 1 in that quarter, else 0). Estimates: intercept 0.40, Q1 = 1.20, Q2 = −0.30, Q3 = 0.50, X = 0.90. Standard error of Q1 is 0.50. (1) Which quarter is the base? (2) What is the expected fund return in Q1 if the market returns 4%? (3) Is the Q1 coefficient significant at the 5% level if the critical t-value is 2.00?

Show the solution

(1) Four quarters need three dummies. Q4 has no dummy, so Q4 is the base.
(2) Q1 = 1, other dummies 0. Y = 0.40 + 1.20 + 0.90 × 4 = 0.40 + 1.20 + 3.60 = 5.20.
(3) t = 1.20 ÷ 0.50 = 2.40. This exceeds 2.00, so reject the null that the coefficient is zero.

Answer: (1) Q4. (2) 5.20%. (3) Yes, t = 2.40 > 2.00, so Q1 return differs significantly from Q4, holding market return constant.

Example 2

A model for a stock's monthly excess return is Y = 0.5 + 1.0D + 0.8X + 0.4(D × X), where X is market excess return and D = 1 for months after a regulation change, else 0. (1) What is the slope before the change? (2) What is the slope after? (3) What is the expected excess return after the change when X = 3%?

Show the solution

(1) D = 0, so the interaction drops out. Slope = 0.8.
(2) D = 1. Slope = 0.8 + 0.4 = 1.2.
(3) Intercept after = 0.5 + 1.0 = 1.5. Y = 1.5 + 1.2 × 3 = 1.5 + 3.6 = 5.1.

Answer: (1) 0.8. (2) 1.2. (3) 5.1%.

Exam tips

Find the coding in the vignette first. A dummy that is 1 for the base group flips every interpretation.
Count dummies against categories. A model with all n dummies and an intercept signals a trap question.
When asked for a forecast, write the full equation for the group, then plug in X. Do not skip the interaction term.
For significance, the question usually gives the t-statistic, p-value or critical value. Use it directly rather than recomputing from scratch.
Remember the wording: a dummy coefficient is a difference from the base category, holding other independent variables constant.

Dummy Variables in Multiple Regression in other exams

The same ground in other exams, if you are preparing for more than one or want another angle on it.

FRM Part IDummy Variables and Regression Interpretation

Dummy Variables in Multiple Regression: frequently asked questions

What is the dummy variable trap in the CFA Level II exam?

It occurs when you include a dummy for every category and also an intercept. The dummies sum to 1 in every observation, which gives perfect multicollinearity. The fix is to use n − 1 dummies.

How do I interpret a dummy variable coefficient?

It is the average difference in the dependent variable between that category and the base category, holding the other independent variables constant. A coefficient of 1.2 on Q1 means Q1 is 1.2 units higher than the omitted quarter.

What is the difference between an intercept dummy and a slope dummy?

An intercept dummy shifts the regression line up or down without changing its slope. A slope dummy is the dummy multiplied by X, and it changes the slope for that group. You can use both in one model.

How do I test if a dummy variable is significant?

Use a t-test of the coefficient against zero: t = coefficient ÷ standard error. If the absolute t exceeds the critical value, or the p-value is below the significance level, the category differs from the base. For several dummies together, use an F-test.