On this page
What this lesson covers
Regression transformations can make a nonlinear relationship linear in transformed variables, changing how coefficients and predictions are interpreted.
The lesson distinguishes log-lin, lin-log and log-log models and interprets their slopes. It converts a log forecast back to the original scale and checks model choice against residuals and financial reasoning.
How do you apply log-lin, lin-log and log-log models?
Log Y for a log-lin model, log X for a lin-log model and log both variables for a log-log model to match the relationship being estimated.
Log-lin: ln(Y) = intercept + slope × X + error
- Positive dependent variable
- Untransformed independent variable
Lin-log: Y = intercept + slope × ln(X) + error
- Positive independent variable
- Untransformed dependent variable
Log-log: ln(Y) = intercept + slope × ln(X) + error
- Positive variables
- Elasticity of Y with respect to X
For example, in the lesson's log-lin P/E model, the fitted log of P/E is 2.2 + 0.07 × forecast growth, with growth measured in percentage points. For growth of 15%, the fitted log value is 2.2 + 0.07 × 15 = 3.25; taking the antilog gives a predicted P/E of exp(3.25) ≈ 25.8.
Common mistakes
- Reporting a fitted log value as the predicted level confuses a forecast of the log with a forecast of the level itself.
- Comparing R-squared values directly across models with different dependent variables compares measures that describe different things.
- Choosing a functional form only for its statistical fit can yield an economically meaningless model.
Try a practice question
Choose an answer, then check it to read the explanation for every option.
Continue studying in Quizara
In Quizara the lesson comes with more practice questions on the topic, explanations for every option and an AI tutor for your follow-up questions.
Written by the Quizara Editorial Team under our editorial standards.