Breusch-Godfrey and Newey-West Tool

Real Statistics Data Analysis Tool: The Real Statistics Breusch-Godfrey and Newey-West data analysis tool can also be used to test for autocorrelation and to calculate the Newey-West standard errors.

We now show how to use this data analysis tool for Example 1 of Newey-West Standard Errors, whose data is repeated in Figure 1.

Autocorrelation example data

Figure 1 – Regression data

Press Ctrl-m and select the Breusch-Godfrey and Newey-West option from the Reg tab (or from the Regression option when using the original user interface). Now fill in the dialog box that appears as shown in Figure 2.

BG-NW dialog box

Figure 2 – Breusch-Godfrey and Newey-West dialog box

Note that if the Lag Order field were left blank it would default to .75 × n1/3, which for this example is .75 × 201/3 = 2.034, which would round up to 3.

Upon pressing the OK button the output shown in Figure 3 appears.

Data analysis tool output

Figure 3 – Output from the data analysis tool

We see from the left side of the figure that both versions of the Breusch-Godfrey test are significant, indicating that there is autocorrelation. The regression shown on the right side of the figure uses both OLS coefficient standard errors as well as Newey-West standard errors.

Leave a Comment