The core of this paper is an econometric estimation of the relation between financial stress and a number of macroeconomic variables (consumption, real GDP, investment, unemployment). This estimation is done on Turkish quarterly data for the period 2002-2021 using threshold vector autoregression (i.e., TVAR). The paper observes the non-linear trade-off between financial stress and macroeconomic indicators. The effect of financial stress appears to be adverse when the stress level is already at a higher level. During high stress episodes, any further increase in financial stress drags economic growth down and the effect appears to be prolonged in nature. Consumption and investment growth also moderate due to a higher stress level. Furthermore, the forecast error decomposition indicates sustained contribution of financial stress impeding growth prospects over the forecast horizon. The findings corroborate with the financial friction mechanism. As borrowing constraint tightens during a high stress regime, the effect of financial stress moderates economic activities. Lastly, the paper extends a local projection approach for estimating a threshold VAR model as a robustness check.

Keywords: Financial stress index, threshold VAR model, Markov switching model, local projection, forecast error variance decomposition