Chartered Financial Analyst (CFA) Practice Exam Level 2

What does a correct specification of an AR(1) model require?

1. Significance of residuals

2. Independence among variables

3. Non-significant residuals

4. Uncorrelated error terms

The correct answer is: Non-significant residuals

A correctly specified AR(1) model requires non-significant residuals in order to ensure that the model captures all relevant information from the data. In an autoregressive model of order 1, the residuals should ideally be white noise, meaning they should exhibit no significant autocorrelation. If the residuals are significant, it indicates that important dynamics in the data may be omitted from the model, suggesting a need for modification or inclusion of additional terms. An AR(1) model's purpose is to predict the current value based on the immediately preceding value, and if the residuals show significant patterns, it may point to an inappropriate model specification or the need for a more complex model. Hence, achieving non-significant residuals is crucial in validating the model's specification, ensuring that it accurately reflects the underlying process generating the data.