The financial and economic crisis has stressed the need for a better understanding of destabilizing effects, in particular arising from the financial sector, and how these may spill over to economic activity. In turn, weakening economic conditions are likely to feed back to the financial sector, giving rise to a so-called adverse feedback loop. So far lacking is work on non-linear linkages and asymmetric dynamics as they may unfold over time. For instance, the magnitude to which financial sector instabilities affect economic activity could depend on the actual state of the financial sector which is more severe at some downswings than others. Thus, non-linearities might not matter much for a long time period but then may generate a rare and large event. In our empirical work we analyze the feedback mechanisms between economic downturns and financial stress for selected euro area countries as they may appear at some time periods. The hitherto theoretical and empirical findings suggest the need for an empirical approach that can accommodate varying dynamic patterns across alternative states of the economy. We propose a non-linear multivariate Vector Smooth Transition Autoregressive (VSTAR) model framework which has not been used empirically so far. In contrast to previous studies it is able to capture smooth regime changes which we expect in the financial market-output nexus. Our contribution is threefold: first, we use financial condition indices which are more comprehensive and put a stronger focus on the banking sector. Second, we apply a non-linear VSTAR model which has not been used before. We think it is more appropriate to model the relationship between output and the financial sector since it allows for smooth regime changes. Third, we comprehensively investigate the (potentially changing) dynamics between the financial sector and the real economy over time. In most countries, a shock in the financial market leads to a long-lasting negative response in economic activity which is regime-dependent. Yet, the negative response is not as pronounced as it is in some other studies. This may first result from regime changes taking place rather smooth than abrupt. Shocks in the presence of smooth regime-changing may be dampened so that the negative effect on economic activity is weaker. Second, the outcomes hinge centrally on the sample period. We show that dynamics in the financial market-macroeconomy link vary over time in the euro area countries. Linearity cannot be rejected for some euro area countries over time questioning non-linearities in the financial sector-output nexus as unambiguous feature. Our results point towards increasing importance of non-linearities with the financial crisis break out. Even if linearity is rejected, the negative output effect typically observed is not always present. In particular, it is not continually found before the Lehman collapse, although we are in a model-defined high stress regime. This suggests that events leading to a strong economic breakdown are rather related to a financial cycle which has low frequency and hence, occur rarely.

Keywords

Vector STAR, financial stress, financial cycle, real economy, regime-switching, euro area