Resumen de The macro financial transmission of shocks
The eagerness to better understand macro-financial linkages has been unprecedented in recent years. The Global Financial Crisis showed the increasing interplay between the real and financial sector and the synchronization of financial and capital markets. More importantly, it highlighted the academic gaps that should be filled to have a comprehensive vision of the transmission of shocks in this increasingly integrated world. Scholars were also confronted to an always challenging task: changing the usual framework of analysis. This thesis contributes to this task by empirically investigating different potential channels of transmission of shocks. A better understanding of the empirical effects of these shocks can be a valuable source of information for policy makers, especially in periods of heightened volatility such as the global financial and the current Covid-19 crises.
A relevant cross-country transmission of real and financial shocks is via gross capital flows. Cross border capital flows play an increasingly important role in the global economy. While capital flows can bring potential benefits to the financial system, they also carry risks due to their volatility, size and the channels of contagion that they create. To prevent the materialization of these risks, a better understanding of capital flows is imperative. While there is extensive empirical research on net capital flows, it is only recently that a new strand of literature has paved the way for research on gross capital flows. Following Forbes et al (2012) and Broner el at (2013), we differentiate between the behavior of domestic agents investing abroad (gross capital outflows) and foreign agents investing in the reporting country (gross capital inflows). As gross financial flows have substantially increased in the last decade and domestic agents and foreign agents can be motivated by different factors, this distinction is all the more relevant. On the other hand, a country's economic performance is highly influenced by its institutional and political environment. Building on these ground-breaking papers, we contribute to the literature by showing the role played by institutional factors in determining gross capital flows in the long term but also during periods of financial stress.
The objective of the first chapter is to assess the role of institutions as a driver of global capital flows over the last twenty years, highlighting the significant changes in the behavior of domestic and foreign investors during periods of financial stress. The focus is on two questions. How relevant are institutions to explain the pattern and the dynamics of flows in the long term? And, in periods of financial stress, do markets discriminate among economies according to their institutions? To do so, we build a database comprising of quarterly data for 56 countries, differentiating between high-income and low and middle-income states, over the period 1996-2012, and use fixed-effects and IV estimates.
Our main findings are the following. Firstly, employing fixed effect models, our results clearly suggest that institutional quality is an important driver of gross capital flows, mainly driven by the dynamics of FDI and portfolio flows. Countries with better quality public services tend to attract more investment and, in high-income countries, create an adequate environment to boost economic activity and investment abroad. Indeed, IV estimates suggest the existence of a causal link for a sub-sample of 25 countries. Secondly, Government Effectiveness and Regulatory Quality seem to be the most important determinants of capital flows. Finally, institutional quality turns out to be also relevant during periods of financial stress. Domestic investors tend to retrench more capital flows if they live in countries with a sound institutional framework, which compensates for the negative effects of declining capital inflows. Therefore, good institutions incentivize the build-up of external savings, by promoting larger outflows, in normal times in high-income countries. And, then, they also facilitate the repatriation of such assets during crises. Moreover, an improvement in institutional quality can soften instability in financial markets through a reduction in capital flows' volatility and therefore can diminish the risk of contagion and financial crises in the host country.
A well-known and exhaustively analyzed transmission mechanism is monetary policy. Central banks of major economies enacted a wide array of unconventional policy measures (UMPs) to influence monetary and financial conditions. These ``nonstandard'' measures go far beyond conventional monetary policy and include balance sheet policies, forward guidance and even negative interest rates. After more than a decade of UMP policies, their role on financial markets is still not well understood. Understanding the impact of UMPs is essential to identify the influence of central banks on financial markets, especially during distressed periods such as the past financial crisis or the recent Covid-19 pandemic, where dysfunctional financial markets were undermining the traditional transmission channel of monetary policy. Chapter 2 and 3 analyze the impact of unconventional monetary policies (UMPs) on the probabilities of extremely adverse macro-financial events. While chapter 2 assesses the effectiveness of UMPs on affecting tail risk perceptions, chapter 3 analyses its cross-border effects.
To do so, we use two main sources of information: UMP announcements by central banks and data from financial markets. On the one hand, a comprehensive daily database tracking the UMP actions of different central banks has been built for our research. This dataset comprises more than 200 events for the four major central banks (the FED, ECB, BoE and BOJ), and it distinguishes among types of measures: liquidity, forward guidance and asset purchases (both sterilized and unsterilized). In our view, each of these groups has a different goal, and their effect on financial variables and market expectations deserves a particular analysis. We also track contractionary announcements that include both “tightening measures” and the tapering of asset purchases. On the other hand, the information about RNDs is extracted from equity options written on the most liquid and representative stock indexes of each economic region: S&P500 (US), the EuroStoxx50 (Eurozone), FTSE100 (UK) and Nikkei225 (Japan).
Chapter 2 evaluates the impact of diverse UMP actions at modifying the perception of extreme market movements, or tail risks. In particular, we exploit the available heterogeneity of UMP measures from four major central banks -- the Federal Reserve (Fed), European Central Bank (ECB), Bank of England (BOE) and Bank of Japan (BoJ) -- to discern their impact on the risk-neutral probability of market crashes. These anticipated probabilities of extreme events are measured through the information contained in the risk-neutral densities (RNDs) of option prices from the main stock market indexes. Contrary to the real-world probability densities, which refer to the dynamics of actual prices based on historical data, option prices are forward looking because their payoffs depend on future states of the underlying asset.
More specifically, in this chapter we first use an event-study analysis to assess the impact of UMP announcements on the changes in markets perceptions. Then, to explore the dynamic impact of UMP on tail risk, we propose a Bayesian SVAR identified using sign restrictions.
The main contribution of Chapter 2 is threefold. First, we dig deeper into the variety of UMP measures implemented by an international sample of leading central banks. Our empirical findings suggest that expansionary UMP announcements mitigate the probability of (expected) sharp market declines for various thresholds of a given loss and across different horizons in the four areas analyzed. Specifically, UMP measures reduce the perception of extreme events by 9 percent in the U.S, 11 percent in the euro area and in the UK, and 7 percent in Japan. An interesting heterogeneity is found when comparing the measures. The most effective measures seem to be the announcement of forward guidance, particularly in the US, the euro area and Japan (where UMPs reduce tail risk perceptions by around 20%) and liquidity announcements. These results highlight the relevance of communication and credibility of central banks as a monetary policy tool. According to our results, unsterilized measures are also more effective than sterilized interventions. Indeed, announcement of asset purchases reduce tail risks by 9% in the euro area. To the best of our knowledge, this is the first attempt at comparing the effectiveness of the heterogeneous measures of the four major central banks at reducing extreme market movement.
Linked to the previous point, we also explore the impact of the reversal of UMP actions on tail risk. Contractionary announcements, which include tightening measures or the announcements of the exit strategy lead to an increment of the probability of (future) extreme market crashes. This result is robust for the US, Europe and UK economies, exhibiting a considerable higher impact for the Euro and UK areas. Interestingly, when analyzing the effect of tapering announcements on the perception of extreme market movements, their effects seem to be beneficial at reducing tail risks. Our estimations show that the tapering announcements significantly diminish the tail risk for the European case, even exhibiting a higher impact than UMP expansionary measures. However, this result is not statistically significant for the US economy.
The second contribution concerns to the current debate about the impact of UMPs in the financial and real economy. Interested in the persistence over time of UMPs in reducing tail risks, we carry out a structural vector-autoregressive (SVAR) analysis. The SVAR confirms the main result of this article: expansionary UMP shocks reduce tail risk perceptions in the four areas under study, leading to a decrease of approximately one or two percentage points. However, the UMP shock has only a temporary effect on tail risk perceptions, fading out after a year. As an additional output of the model, the macroeconomic impact of an expansionary UMP is found to be positive. All these results are corroborated with an exhaustive battery of robustness checks that corroborate our main findings.
Finally, pushing a step forward the analysis, we investigate the nature of UMP transmission into financial markets. Previous results show that UMPs change the perception of investors on the probability of future market crashes. In this way, we wonder if UMPs could be modifying investors' aggregate risk aversion. To answer this question, we employ a novel technique for obtaining risk aversion estimates from option prices. Then, we run several SVAR estimations to explore the response of aggregate risk aversion to an expansionary UMP shock. The results confirm a contemporaneous and positive reduction in the risk aversion which persists up to ten months after the shock.
Chapter 3 explores the potential risk spillovers of UMP shocks across advanced economies, an area where what we know is less obvious. We provide evidence that contractionary UMPs of foreign central banks increase the expectations of overseas investors of a market crash in their national stock markets. These results question the evidence in Obstfeld and Rogoff (2002), who argue that cross-border spillovers were likely to be negligible and therefore should not be considered. As it is also shown in the chapter, the adverse effects of UMPs are attributable to the nature of the measure, and not to a country effect. In other words, our results do not show a hegemonic role of the Fed different from other central banks on increasing the uncertainty of foreign financial markets.
The main contribution of Chapter 3 is twofold. Beyond establishing the side effects of UMPs in the expectations of a market crash in international markets, we also show how the transmission channel of these risk spillovers works. Using the decomposition of UMP announcements into monetary policy and central bank information shocks provided by Jarocinsky (2020), we employ the local projections approach of Jordà (2005) to test the relevance of central bank communication in driving monetary spillovers. The results confirm the prevalence of an information channel, rather than a pure monetary policy channel, in the transmission of cross-border effects. Empirical evidence reported for the Federal Reserve (Fed) and the European Central Bank (ECB) suggests that positive information shocks -e.g., an expansionary announcement- have a positive and significant impact on cross-border tail risks. However, contractionary monetary announcements are not statistically significant. These results suggest that participants in foreign equity markets are particularly affected by the economic outlook of the US and euro area, which are two relevant regions from a financial and economic point of view.
In addition to this previous exercise, we investigate whether the genesis of these cross-border disturbances are due to changes in the aggregate uncertainty of international markets or changes in the risk aversion of their investors -or both-. To this end, we build measures of risk aversion based on the information embedded in option prices. Our results suggest that positive information shocks affect the level of foreign aggregate uncertainty but not the risk aversion of foreign investors.
The second contribution of Chapter 3 is related to the importance of communication for defining the exit strategy of UMPs. We build a comprehensive database tracking the UMP actions of different central banks that permits to disentangle the cross-border effects due to contractionary and tapering measures. The empirical findings show that contractionary UMP tend to increase tail risk perceptions of foreign stock market investors, while tapering announcements either reduce them, such as in the case of the ECB or do not affect them (the Fed). This result might suggest the use of clearer communication in the exit strategy of central banks, reducing somewhat the negative spillovers of these announcements.
Finally, the last chapter focuses on oil prices, which are a simultaneous and common shock to all countries and sectors, with relevant financial implications as a result of the financialization of commodity markets. The surge of shale oil producers brought about a change in the structure of the oil market. In this chapter, our objective is to answer the following two questions. First, how can we account for OPEC production strategies in empirical models in the oil market? Second, what is the impact of the strategic interaction between OPEC and other producers on the price of oil? In Chapter 4, we contribute to the literature by drawing a direct link between strategic interactions in the oil market and its implications for structural empirical models of the global oil market. We start by presenting a simple theoretical framework that shows how strategic interactions can affect the oil market equilibrium. We consider a standard dominant firm-competitive fringe model for the crude oil. This environment is populated by fringe firms (shale producers for instance) that are atomistic, and one large player/producer (namely OPEC) that can exert market power and willingly influence the oil market equilibrium. In our setup, fringe firms take as given market conditions and supply as much as their capacity allows them to produce. OPEC, on the other hand, takes as given the production of the fringe firms and acts as a monopolist on the residual demand, strategically adjusting its production. However, we depart from standard frameworks since in our model, depending on market conditions, OPEC can follow three strategies. Specifically, it can operate as a monopolist on the residual demand curve, it can move its production in tandem with the production of the fringe firms, in an attempt to keep market shares stable. Alternatively, it can modify the production in the opposite direction to non-OPEC producers, this time in an attempt to stabilize prices.
Then, we set up a Structural Vector Autoregressive Model (SVAR) of the global oil market in which we jointly model oil production, global economic activity, oil inventories and the price of oil. We modify the model in two directions to take into account the implications of our theoretical setup. First, we model OPEC and non-OPEC production separately. Second, we allow for distinct reactions of OPEC following a supply shock from non-OPEC countries. In the first case, we let OPEC production move in the same direction as that of non-OPEC production. Consistently with our theoretical model, we define this shock as a Market Share Targeting shock. In the second case, we let OPEC production move in the opposite direction to that of non-OPEC production, therefore stabilizing prices. We call this second shock a Price Targeting shock. To the best of our knowledge, no previous work has so far quantified the implications for price developments of different OPEC's production strategies.
The main results of our empirical analysis are the following. First, Price Targeting shocks account for around 10 percent of the overall oil price forecast error variance on all horizons. They absorb around half of the forecast error variance that is left unexplained in a smaller model in which OPEC and non-OPEC production are not distinguished and only aggregate supply shocks are considered. Second, OPEC targeted prices, rather than market shares, in the 2000s, somewhat contributed to booming oil prices in the run-up to the crisis. Third, we can characterize different supply elasticities for OPEC and non-OPEC producers. We also find the OPEC elasticity of supply three times as high as that of non-OPEC producers.
