• business cycle
• financial markets
• fiscal multipliers
• Macroeconomic dynamics
• regime switching models
• technology shocks
• TVAR
• vector autoregressions

Almost a decade after the beginning of the Great Recession advanced economies are desperately searching for sustained growth prospects. Although expressing the most brilliant performance among western countries, even the US economic record is far from being fully satisfactorily, and the secular stagnation debate also entered the presidential campaign.
Hysteresis, persistent slumps, debt overhang, balance sheet recessions and global savings glut are, together with the above mentioned secular stagnation hypothesis, competing (and often complementary) diagnoses to tackle with the issues of declining US productivity growth, boom and bust cycles and slow recoveries.
Financial markets are key for explaining the Great Recession and the subsequent disappointing recovery. If the role of financial markets is a distinct feature of all post mid 80s recessions, dramas in the wake of deep financial crises involve dramatic deleveraging, high fiscal costs and long lasting and painful recessions.
In this respect, coupling existing empirical evidence and new developments in macroeconomic theory with earlier insights, the macroeconomics of boom and bust cycles and persistent slumps seems to be sensibly captured by regime switching models.
The present dissertation explores some of the patterns characterizing the US post WWII macroeconomic dynamics employing regime switching econometric techniques, with a focus upon the linkages between financial markets and real economy. This task is accomplished throughout a broad review of the literature and three different applications.
In Chapter 2 we review the different research avenues which have inspired this work. In particular, the chapter is divided into three parts, so as to discuss the main contributions about i) the finance-growth nexus in the long-run; ii) the action of financial markets in shaping short- to medium-term fluctuations in the macroeconomy; iii) the policy implications stemming from the complex trade-offs generated by such relations. The aim of the chapter is to give an insight about how complex and deeply intertwined are the mechanisms at work when financial markets kick in. This is done by highlighting how the distinction between long-run and shorter horizons gets blurred once financial development does not induce financial deepening. Thus, although existing theories and empirics do not support the old idea that, to say it with Joan Robinson, ``where enterprise leads finance follows''; the finance and growth nexus is at the same time characterized by a bright side -- finance spurs growth -- and a dark side -- financial crises dampen growth -- as well. The policy implications of this line of reasoning are demanding both in order to prevent financial crises ex ante and to clean the ensuing mess ex post.
Our journey goes on with Chapter 3 in which we investigate the responses of GDP and employment with respect to technology shocks in a multi-regime framework. Although this application may look far with respect to the focus of our broad research project, it substantially contributes to it in, at least, two fundamental ways, as it will be clear in a short while.
We estimate over several time spans (the longest being 1957:1-2011:4) different threshold vector autoregressions (TVAR) in which the threshold variable is constituted by GDP annualized growth and we assess the paths of generalized impulse response functions in order to judge whether tech shocks differently affect hours and output according to the state of the economy. Finally, consistently with the prevailing literature, we assess whether the relation is time dependent and investigate some plausible explanations, among which the role of financial markets is stressed.
In a nutshell, our results are the following. First, technology shocks while spurring GDP growth, display a negative effect on hours worked at least on impact, independently of the state of the economy. Moreover, within the main sample the effects on hours are persistently abundantly negative in low growth periods whereas they do not turn out to be significantly different from zero in good times in the long-run. In this respect our work generalizes the existing literature: technology shocks spur economic growth without impinging employment in good times, whereas in recessions the effect on employment is strongly negative and that on GDP is negligible. We also find that the dynamics of hours worked in the aftermath of a technology shock is mostly due to the response of the number of employees rather than hours per worker. Adjustments via the extensive margin appears then more relevant than those occurring along the intensive margin.
However, the effects on aggregate demand depend on the sampled period, with a higher impact during the Great Moderation and the Great Recession that followed it. Moreover, the impact of technology shocks in good or bad times depends on the chosen sample. Indeed, tech shocks exert a stronger negative effect on employment prior the mid 80s in low growth periods, whereas the opposite holds with the beginning of the Great Moderation and it is preserved with the Great Recession: the long-run response of hours worked turns positive in recessions. Thus, our analysis generalizes the existing evidence about the milder effects of tech shocks on employment during the Great Moderation. First, we show that such result is due to the higher effects exerted in low growth periods. Moreover, we demonstrate that the Great Recession doesn't appear to change the results. We also show that the observed switch is more due to the extensive margins (increased number of employees) rather than to the intensive margin (employees working harder). Finally, as to the transmission channels at work, although we are not provided with a definitive answer regarding the reasons for such a regime shift, we document several changes in the responses of different relevant economic time series which are consistent with some of the proposed explanations. In particular, we stress the role of investment, credit markets and stabilization policies.
What about the contribution of chapter with respect the main focus of our analysis here? The work explores some of the most studied relations in describing macroeconomic dynamics -- those between technology, output and employment -- and assess how they have changed over time with special regard to macroeconomic regimes and transformations of the structure of the economy. As to the latter, financial markets emerge as a key actor in determining a broad shift in the transmission mechanisms of technology shocks. These two features are enlightening with respect to the analysis that follows. First, Chapter 3 shows that we cannot sensibly analyze post WWII US economy without distinguishing the pre Volcker era with respect to the post mid 80s period. Second, it also shows that financial markets may have played a pivotal role in shaping this transformation.
In the light of this findings, the applications which follow characterize the linkages between financial markets and real economy by concentrating upon the post mid 80s period; and extend the main ideas explored in Chapter 3 to consider their contribution as a further source of macroeconomic regime switching.
In Chapter 4 we investigate financial fluctuations vis-à-vis the business cycle within the US economy. In particular, we propose a set of intertwined exercises whose main goals consist in assessing i) the behavior of several items within the balance sheets of different financial institutions and credit spreads with respect to GDP so as to uncover their properties over the business cycle; ii) the role of financial markets in shaping business cycle fluctuations throughout the dynamic response of the economy to financial shocks. In this respect, our contribution is twofold. On the one hand, we offer a Stock and Watson (1999)-like set of stylized facts about the relation between financial and business cycle fluctuations. In particular, we look at how macro-financial aggregates, the items within the balance sheets of different financial as well as nonfinancial institutions and credit spreads do interact with output at business cycle frequencies, throughout the exploitation of cross-correlations and a battery of Granger causality as well as of Johansen cointegration tests. On the other hand, we explore the parameter space of several univariate and multivariate models in search of the effects of financial frictions on economic performances, looking at the impulse response functions generated by a great deal of vector autoregressions and paying attention to potential nonlinearities stemming from conditions of financial markets. This latter exercise is performed throughout the estimation of different threshold autoregressions (TAR) involving relevant economic indicators (e.g., GDP; the unemployment rate; private aggregate demand components). According to our results, financial markets play a pivotal role in shaping GDP dynamics and its components in the US. More precisely, whereas most of the examined credit aggregates and financial items tend to move with lags with respect to the business cycle acting mostly as amplifiers, credit spreads display a counter-cylical dynamics and tend to lead fluctuations demonstrating a fairly high predictive content. Moreover, many items within the balance sheets of financial businesses do show a good predictive content and, in general, financial aggregates demonstrate a far more nervous behavior than GDP, whose fluctuations are generally characterized by a lower dispersion. Furthermore, it is not the case that financial institutions are all alike, as the dynamics of the items within the balance sheets of depository institutions displays different patterns with respect to that of investment banks and other institutions involved in shadow banking activity. As to the role of financial markets either in directly shocking the economy or in amplifying other shocks, our results support the key insights from standard macroeconomic models with financial frictions. In particular, shocks to credit spreads severely hit real economy whereas, throughout both the spreads and other relevant credit aggregates, negative shocks are dramatically amplified by financial markets. Moreover, in line with recent research, long-run effects due a deceleration of technological change are also uncovered. Furthermore, in the aftermath of financial shocks, financial as well as nonfinancial institutions actively adjust their balance-sheets and ``flight-to-quality'' and ``flight-to-liquidity'' emerge. Finally, financial markets seem to constitute a relevant source of nonlinearities within the economic system. More precisely, during “tight” credit periods, output and most of its main components demonstrate a higher persistence and therefore a lower attitude to reabsorb negative shocks.
Finally, in Chapter 5 we estimate a Threshold Vector Autoregression model to study how the effects of fiscal policy can be amplified or dampened according the state of credit markets. More precisely, we conjecture that fiscal policies should be more successful in stimulating output in regimes where the financial accelerator leads to “tight” credit conditions, which increase the difficulties of firms to finance their investment and production activities forcing them to curb their employment. We estimate a two credit-market regime TVAR model in first differences on US quarterly data for the period 1984-2010. In our analyses, we proxy non-linearities resulting from credit conditions using as threshold variable the spread between the BAA-rated corporate bond yield and the 10-year treasury constant maturity rate. We find that the responses of output to fiscal policies significantly change according to the state of credit markets. Whenever the economy is in the “tight” credit regime, the GIRFs display a strong and persistent reaction of output to fiscal policy shocks. On the contrary, the response of GDP to fiscal policies is much milder when the economy experiences ``normal'' credit conditions. The different patterns exhibited by the GIRFs in the two credit regimes are reinforced by the computation of fiscal multipliers. When firms face increasing financing costs, multipliers are much higher than one at different time horizons. Conversely, the multipliers are much weaker --- usually lower than one --- when the external finance premium is reducing. A battery of t-tests applied on the bootstrapped distributions of GIRFs confirm that fiscal multipliers are significantly different in the two credit regimes. We test the robustness of our results to four potential issues concerning: i) the specification of the model (first differences vs. levels); ii) the presence of expectations about fiscal policies not already absorbed by the model (i.e., fiscal foresight); iii) the adoption of a different threshold variables; iv) alternative measures of output, fiscal and monetary variables and different sample periods. In all cases, we find that our main results robustly hold.