 Good morning, everyone. Welcome to this 12 ECB forecasting and conference of forecasting techniques. I'm really honored to open the conference. I'm very happy to see that this can happen, at least in an hybrid form. So it's great to see you back live. This edition is a title, Forecasting at Risks, which brings together experts from academia and policy institutions to exchange ideas on some of the main current challenges faced by forecasters, including the modeling of economic dynamics and the assessment of risks after crisis or extreme events. Why this title? The 9th conference was on forecast uncertainty. The 11th conference was on forecasting in abnormal times. I'm sure the organizers had quite a hard time to find a good title after the last one because the world has certainly not got closer to normal since then. On the contrary, we currently face instability on various nature. Forecasting has become even more challenging. Major shocks with specific dynamics have occurred one after the other, during which the ECB also went through a large process of its review, its strategy. This was based on the main assumptions that low inflation was there to stay. This picture is obviously reversed. Now, one thing I want you to bring your attention to is that every time a crisis arises, models and forecasting tools come under big scrutiny, certainly and understandably from society by the society at large, but sometimes perhaps more surprisingly also by policy makers themselves. To cope with this eye uncertainty, some people, including policy makers, may believe that we can leave models aside and just trust our own judgment, our instinct. That is the best way to navigate through a storm. It is indeed true that calibrated or estimated on historical regularities, models have a hard time to reconcile those regularities with extraordinary developments, possible structural breaks, unprecedented socioeconomic or geopolitical events. For these events, the past, although similar, might not be useful to inform the current models or the available models under new circumstances. But let's be bold and clear and loud. Models, although with recommendations, are essential to discipline our judgment, no matter how good this judgment is, and to produce any credible macroeconomic forecast. In special times, of course, we need to adapt the way we use models. We might need to rapidly adapt models themselves, and we cannot rely only on baseline or point forecast, but we need to produce risk indicators around the projection of baseline and to quantify the uncertainty around this baseline, with alternative scenarios. A comprehensive characterization and proper assessment of macroeconomic risks is important for a timely and effective conduct of monetary policy. In fact, if you remember during the COVID pandemic, instead of publishing the usual point forecast, the ECB has published scenarios under possible alternative assumptions in what if type of experiments, and these experiments, although designed by wise judgment, only models are able to quantify accurately. So make no mistakes. Models and tools that allow us to give such a comprehensive picture around baseline projections are more important than ever, especially during crisis time to guide our judgment or instinct to quantify uncertainty and to attach risks to the projections. And therefore, this conference, like the one this one, hosted at the ECB is much welcome because it is timely, relevant, and very important for policymakers. Now, the econometric methodology is developing rapidly around this concept. Large data set, the adaptation of ideas for machine learning provide great stimulus for that. Several new ideas have come up to cope with the large shocks and with time variation at the same time to explore this large data set. One development is precisely this at-risk literature, namely the idea that the economy behaves differently after extreme events, or at the tail of the distribution. So economic variables are subject to extreme events that are poorly captured by, say, Gaussianity or linearity. For instance, when macroeconomic models typically assume that the economic disturbances follow a normal distribution, they systematically underestimate the frequency of large economic downwards. Several factors could explain deviations from this normal distribution. This might include financial frictions, sectoral shocks, zero lower bound or nominal interest rate, SPACs in uncertainty, national disaster, government policy themselves, so many things. This is extremely important for policy design. The ECB is taking up this seriously and now includes various chapters on risks into the governing council briefing documents that are based on quantile regressions and related risk analysis tools. Thanks also to the work of a recent expert group on macro at-risk, EGMA is called, by the working group on econometric modeling, the WGM, and the working group on forecasting. These are two substructures of the Monetary Policy Committee, which had the goal to develop a set of tools and approaches to perform and analyze macroeconomic risk assessment in the course of the policy conduct at ECB. The report of EGMA, I think, is being finalized and hopefully can soon become publicly available. I see from the list of contributors that some authors of papers used in that report are actually here at the conference, of course. So cross-checking the results of that paper and many papers presented at this conference, we know, for instance, that the empirical findings in this literature often point to asymmetric effects of the risk factors, notably financial risk factors, on GDP growth. The predictive power of different financial indices depends on the forecast horizon and country characteristics. For instance, asset prices are informative for a shorter time horizon while sovereign spreads would be more significant for the group of emerging market economies. Shocks to monetary policy, credit, and productivity affect disproportionately the lower GDP quantites, with unexpected monetary policy tightening found to increase the probability of very low growth in the short term. Now, in contrast to the abundance and clarity of GDP growth results, somewhat scant research on inflationary risks provides mixed evidence on the ability of, for instance, the baseline quantile regression approach to underpin asymmetric responses to risk factors. Employing a Phillips curve specification also results in a mixed picture on the detection of asymmetries for inflation and all these methods have been found to perform well in out-of-sample analysis forecast. However, like the empirical findings for GDP, there is considerable heterogeneity across regressors, estimation samples, and country groups in the way that inflation quantiles react to business cycle movements, financial contributions, and other global risk factors. Now, the novel contributions included in the program of this conference will very much enhance, say, the EGMA report on these topics. Just glancing at the conference program, a non-exhaustive list of novel applications includes, for instance, Bayesian non-parametric, Bayesian neural network analysis, non-linear dynamic fatal models, quantile regression, time-varying parameters, or time-varying volatility, forecasting macro-tailed risks in real-time, textual analysis, conditional forecast in large models with hard and soft constraints, variable selections, and, of course, forecasting inflation and forecast accuracy, which is part of our main mandate. Now, this list brings me to final words. But before starting the conference, I think it's important to spend the last words on structural issues, because understanding asymmetric risks in macroeconomic variables is an important task in uncertain times, and this conference will certainly shed light on it. But understanding risks that have structural interpretations and are good for building narratives around projections is also an important and perhaps even more challenging task. And I believe that this at-risk literature on time series provides an excellent starting point to understand the determinants of asymmetries, but to understand structural sources of tailed risks policy makers need to use model that allow to disentangle causal relationships while handling non-linearities at the same time. And this is a big issue. So the literature has provided different ways of incorporating non-linearities in current mainstream structural models, such as DSG models that can generate tailed risks. But there are big computational challenges when it comes to include non-linearities in structural models, especially if you want to include also heterogeneous agents, for instance. Many of the current algorithms to solve and estimate non-linear DSG models can be applied to small models, but small models, although they provide sound theoretical guidelines, might not be of practical use at central banks for policy analysis where typically medium and large-scale models are used. Therefore, I think it's still difficult to conclude that DSG models can be a perfect solution to understand the macroeconomic tailed risks until researchers bring new and more efficient algorithms that make use of better computational power. I hope this conference can also help in this direction with new computational methods, especially with the idea of suggesting valuable satellite models given that it is tricky for structural models to incorporate all different sources of risks in one single model. Now, without further ado, let's start the conference. I think it is an amazing opportunity to learn about all this directly from invited speakers and scholars, also with a view to cross-check the tools that we have at the ECB and to incorporate these tools with the results of these new frontier studies. It's a privilege to be able to actually rely on such an extraordinary library of methods and data, some of which are already being included in our toolkit, some others will certainly soon be. So thanks a lot to the organizer for who made this terrific event possible, and possible at least in an hybrid form. I'm looking forward to learn a lot from this conference. I wish you all an enjoyable and very fruitful experience. Thank you.