OLS with Newey-West and Hansen-Hodrick SE

Computes OLS and reports Robust SE, NW AND HH corrected standard errors.

You are now following this Submission

% PURPOSE: computes OLS and reports Robust SE, and Newey-West and Hansen-Hodrick adjusted heteroscedastic-serial consistent standard errors.

% Inputs:
% y = T x 1 vector, left hand variable data
% X = T x n matrix, right hand variable data
% L = number of lags to include in NW corrected standard errors
% H = number of lags to include in HH corrected standard errors
%
%Note: you must make one column of X a vector of ones if you want a
% constant.
% Output:
% beta = regression coefficients 1 x n vector of coefficients
% R2 = unadjusted
% R2adj = adjusted R2
% X2(Degrees of Freedom) = : Chi-squared statistic for all coefficients
% jointly zero.
% std = corrected standard errors.
% t_ = t-stat for NW and HH
%Note: For chi-square test program checks whether first is a constant and ignores that one for
% test. If there is only one beta the program does not report X^2
% results since t_stat^2= X2.
%Note: program automatically displays outputs in an organized format. If you want
%to disable the automatic display just comment lines 129-136.

Cite As

Sandra (2026). OLS with Newey-West and Hansen-Hodrick SE (https://www.mathworks.com/matlabcentral/fileexchange/43259-ols-with-newey-west-and-hansen-hodrick-se), MATLAB Central File Exchange. Retrieved .

Acknowledgements

Inspired by: Newey-West Standard Errors

General Information

MATLAB Release Compatibility

  • Compatible with any release

Platform Compatibility

  • Windows
  • macOS
  • Linux
Version Published Release Notes Action
1.2.0.0

Now you can get results for Newey West, Hansen Hodrick and Robust SE all at once. If there is only one beta the code does not report a chi-square statistic since it is just the square of the single t_stat. Overall, I cleaned up the code.

1.1.0.0

changed beta from a 1 by n vector to a n x 1

1.0.0.0