Structural change estimation in time series regressions with endogenous variables

We propose to apply the group fused Lasso to estimate time series models with endogenous regressors and an unknown number of breaks. It can correctly determine the number of breaks and estimate the break dates asymptotically. Simulations and applications are given.