Pakistan’s economy experienced many ups and downs during the last four decades. These structural shifts (asymmetries) cannot be detected via linear econometric models. This paper employs the Markov Regime-Switching vector autoregression (MS-VAR) model with time-varying transition probabilities to identify the high and low growth regimes. After establishing structural shifts in the data, next, we estimate the linear VAR model in each regime to test the effects of fiscal shocks on output, and we also test the twin deficit hypothesis as well as the crowding-out investment effect. Different specifications of MS-VAR models with Constant Transition Probability (CTP) and time-varying transition probability (TVTP) were tested, among which the best fit model with four regimes is chosen for analysis. The four regimes identified are the low growth regimes from 1973 to 1979 and from 1989 to 1999 and the high growth regimes from 1980 to 1988 and from 2000 to 2010. The results from the subsample analysis show that the response of output to positive spending shock is increasing in high growth regimes and decreasing in low growth regimes. Similarly, a tax shock has a statistically insignificant impact on output except for the last regime where a tax shock is positively associated with output growth. An expansionary fiscal policy crowds-out private investment in low growth regimes (i.e. in first and third regimes) while a positive effect on private investment is observed during high growth regimes (second and fourth regimes). Lastly, twin deficit is observed in all regimes.