ABSTRACT. In this paper we derive a general parametric bootstrapping approach to compute density forecasts for various types of mixed-data sampling (MIDAS) regressions. We consider both classical and unrestricted MIDAS regressions with and without an autoregressive component. First, we compare the forecasting performance of the different MIDAS models in Monte Carlo simulation experiments. We find that the results in terms of point and density forecasts are coherent. Moreover, the results do not clearly indicate a superior performance of one of the models under scrutiny when the persistence of the low frequency variable is low. Some differences are instead more evident when the persistence is high, for which the ARMIDAS and the AR-U-MIDAS produce better forecasts. Second, in an empirical exercise we evaluate density forecasts for quarterly US output growth, exploiting information from typical monthly series. We find that MIDAS models provide accurate and timely density forecasts.

ABSTRACT. We provide a frequentist framework to elicit the forecaster's expected utility by introducing an LM-type test for the null of locality of scoring rule underlining the observed probabilistic forecast. The scoring rule is looked as an observed transition variable in a smooth transition autoregression in order to nest a known framework. The test behaves consistently with the requirement of the theoretical literature. Applications to US Business Cycle, several economic time series and Bank of Norway's Fan Charts reveal that scoring rules affect the dating algorithm of recessions events and the model-based forecast performances in favor of a nonlinear specification, advocating the importance of a correct score selection and that the locality of the scoring rule underlining the estimated predictive density is a strong assumption.

ABSTRACT. In this paper we extend the targeted-regressor approach suggested in Bai and Ng (2008) for variables sampled at the same frequency to mixed-frequency data. Our MIDASSO approach is a combination of the unrestricted MIxed-frequency DAta-Sampling approach (U-MIDAS) (see Foroni et al., 2014; Castle et al., 2009; Bec and Mogliani, 2013), and the LASSO-type penalised regression used in Bai and Ng (2008), called the elastic net (Zou and Hastie, 2005). We illustrate our approach by using empirical example with the Purchasing Managers' Index computed for Switzerland. We address whether the xed weighting scheme of the PMI components is supported by the data. We nd that the relative weights of the PMI components are generally supported by the data, except the fact that one component, found very informative for explaining GDP growth, is currently omitted from the PMI composition. We also compare the out-of-sample forecasting performance of the MIDASSO approach of GDP growth in Switzerland based on pre-selection of predictors with that of the approach without such variable pre-selection. The results of the forecasting exercise conform the results of Bai and Ng (2008), reported for single-frequency data, that targeting most relevant predictors boosts the forecasting performance also when applied to mixed-frequency data.

ABSTRACT. We propose new methods for evaluating predictive densities that focus on the models' actual predictive ability in finite samples. The tests offer a simple way of evaluating the correct specification of predictive densities, either parametric or non-parametric. The results indicate that our tests are well sized and have good power in detecting mis-specification in predictive densities. An empirical application to the Survey of Professional Forecasters and a baseline Dynamic Stochastic General Equilibrium model shows the usefulness of our methodology.

ABSTRACT. We investigate empirically how sellers react to changes in the population of their con- sumers, identifying the effects of demand composition and demand size with limited information on costs. We show how pharmacists in Italy selectively increase the price of some products when they observe in their cities an exogenous influx of parents of newborns, conceivably less elastic buyers as compared with other more experienced and less pressed consumers. Exploiting population based laws that fix the number of pharmacies in a city, we use RDD to measure the effect of competition on sellers’ ability to extract surplus from less elastic buyers.

ABSTRACT. We use a panel dataset from the Dutch Household Survey, covering annually the period 1995-2012, to analyse whether individual financial risk taste changes over time with the background macroeconomic and financial conditions, as well as personal and subjective exposure to portfolio risk. Considering six different self-assessed facets, we find that risk appetite is higher during periods of economic growth and lower during periods of recession. Risk taste is however unrelated to time when it refers to safe investments. Risk appetite is also higher when perceived risk exposure in past investments is generally large or falls from one year to another.

ABSTRACT. We relate two procedures, panel data analyses and a case study methodology, the synthetic control method, to explicitly quantify the economic cost of civil war on the growth rate of GDP and its level. Our 27 case studies suggest that, on average, civil war reduces the GDP level by 9.1% and the growth rate of the GDP by 2.3%. The estimated losses from panel data analyses range from 1.1% to 1.3% of the GDP growth rate while the magnitude of the impact on the GPD level is around 8.6% per year. Therefore, the incidence of internal conflicts has an economically significant one-off negative effect on the GDP level, as well as a negative effect on the growth rate of the GPD, thus suggesting a persistent output loss and a permeant damage to the prospects for economic growth. The models estimated from panel data and the synthetic control are closely related and give very similar results, which are also in line with the recent literature on the economic cost of civil war.

ABSTRACT. In 2000 Italy replaced its traditional system of severance pay for public employees with a new system. Under the old regime, severance pay was proportional to the final salary before retirement; under the new regime it is proportional to lifetime earnings. This reform entails substantial losses for future generations of public employees, in the range of €20,000-30,000, depending on seniority. Using a difference-in-difference framework, we estimate the impact of this unanticipated change in lifetime resources, on the current consumption and wealth accumulation of employees affected by the reform. In line with theoretical simulations, we find that each euro reduction in severance pay reduces the average propensity to consume by 3 cents and increases the wealth-income ratio by 0.32. The response is stronger for younger workers and for households where both spouses are public sector employees.

ABSTRACT. We propose a new VAR identification strategy to study the impact of noise shocks on aggregate activity. We do so exploiting the informational advantage the econometrician has, relative to the economic agent. The latter, who is uncertain about the underlying state of the economy, responds to the noisy early data releases. The former, with the benefit of hindsight, has access to data revisions as well, which can be used to identify noise shocks. By using a VAR we can avoid making very specific assumptions on the process driving data revisions. We rather remain agnostic about it but make our identification strategy robust to whether data revisions are driven by noise or news. Our analysis shows that a surprising report of output growth numbers delivers a persistent and hump-shaped response of real output and unemployment. The responses are qualitatively similar but an order of magnitude smaller than those to a demand shock. Finally, our counterfactual analysis supports the view that it would not be possible to identify noise shocks unless different vintages of data are used.

ABSTRACT. Small-scale dynamic stochastic general equilibrium have been treated as the benchmark of much of the monetary policy literature, given their ability to explain the impact of monetary policy on output, inflation and financial markets. One cause of the empirical failure of New Keynesian models is partially due to the Rational Expectations (RE) paradigm, which entails a tight structure on the dynamics of the system. Under this hypothesis, the agents are assumed to know the data genereting process. In this paper, we propose the econometric analysis of New Keynesian DSGE models under an alternative expectations generating paradigm, which can be regarded as an intermediate position between rational expectations and learning, nameley an adapted version of the "Quasi-Rational" Expectatations (QRE) hypothesis. A frequentist estimations for the U.S. economy are proposed and a comparison between RE and QRE illustrates the advantages of the new approach.

ABSTRACT. This paper proposes a Bayesian, graph-based approach to identification in vector autoregressive (VAR) models. In our Bayesian graphical VAR (BGVAR) model, the contemporaneous and temporal causal structures of the structural VAR model are represented by two different graphs. We also provide an efficient Markov chain Monte Carlo algorithm to estimate jointly the two causal structures and the parameters of the reduced-form VAR model. The BGVAR approach is shown to be quite effective in dealing with model identification and selection in multivariate time series of moderate dimension, as those considered in the economic literature. In the macroeconomic application the BGVAR identifies the relevant structural relationships among 20 US economic variables, thus providing a useful tool for policy analysis. The financial application contributes to the recent econometric literature on financial interconnectedness. The BGVAR approach provides evidence of a strong unidirectional linkage from financial to non-financial super-sectors during the 2007-2009 financial crisis and a strong bidirectional linkage between the two sectors during the 2010-2013 European sovereign debt crisis

ABSTRACT. A growing line of research makes use of structural changes and different volatility regimes found in the data in a constructive manner to improve the identification of structural parameters in Structural Vector Autoregressions (SVARs). A standard assumption made in the literature is that the reduced form unconditional error covariance matrix varies while the structural parameters remain constant. Under this hypothesis, it is possible to identify the SVAR without needing to resort to additional restrictions. With macroeconomic data, the assumption that the transmission mechanism of the shocks does not vary across volatility regimes is debatable. We derive novel necessary and sufficient rank conditions for local identification of SVARs, where both the error covariance matrix and the structural parameters are allowed to change across volatility regimes. Our approach generalizes the existing literature on `identification through changes in volatility' to a broader framework and opens up interesting possibilities for practitioners. An empirical illustration focuses on a small monetary policy SVAR of the U.S. economy and suggests that monetary policy has become more effective at stabilizing the economy since the 1980s.

ABSTRACT. We identify the effects of changes in financial intermediation, referred to as credit shocks here, both from the demand side and the supply side, using a structural GVAR. We gain insights into the contribution of these shocks to the real economy, in particular their impact on lending to NFCs and GDP growth, which we evaluate by means of historical decomposition and FEVD. Furthermore we focus on transmission of shocks in one country to other countries with close financial linkages to understand better the role of contagion during turbulent times.

ABSTRACT. This paper presents a simple theory of state dependence in credit rationing, and provides an empirical test for state dependence in access to credit. We estimate a first-order Markov model with sample selection in the initial condition, that takes into account unobserved heterogeneity and selection bias. The results, based on a representative sample of Italian firms, show that a past credit denial has a negative effect on the outcome of the current loan application and the decision to apply for a new loan. State dependence in access to credit is more relevant for larger firms and after a global liquidity shock

ABSTRACT. We quantify the real effects of the bank lending channel exploiting the dramatic liquidity drought in interbank markets following the 2007 financial crisis as a source of variation in credit supply. Using detailed firm-bank matched Italian data, we find that banks’ pre-crisis exposure to interbank markets negatively predicts subsequent credit supply and has a significant, negative direct impact on firms’ investment. Using exposure as instrument, we show that investment is highly sensitive to bank credit. We also find that credit shocks affect firms' value added, employment and inputs purchases, and that they propagate through firms' trade credit chains.

ABSTRACT. We exploit a natural experiment in Italy to identify the causal effect of fiscal policy on economic activity. A law issued to allocate reconstruction grants following the 2009 ’Aquilano’ earthquake has resulted in a large and unanticipated discontinuity across municipalities with comparable damages. Using diff-in-diff analysis we estimate three things. First, we estimate the ’reconstruction grants multiplier’, that is the elasticity of local economic activity to an exogenous increase in grants. Second, we estimate the ’local spending multiplier’ and the ’local tax multiplier’ according to the composition of the fiscal stimulus at the local level. Finally, we estimate the negative supply shock generated by the seismic event using a quantified measure of damages for 75,424 buildings. In our findings, the multiplicative effects of fiscal policy are below unity, although the ’local tax multiplier’ is well above one. Yet the size of the grants act as a public insurance scheme, preventing a fall in output.

ABSTRACT. This paper investigates the relation between job search behaviour and housing tenure, exploiting exogenous variation in job search requirements for benefit eligibility introduced by the UK Jobseeker's Allowance (JSA). Previous JSA impact evaluations have found that stricter enforcement increased claimant outflows without altering moves into jobs. Job search models suggest that the removal effect should be larger for unemployed with low search intensity or with lower incentives to adjust search. Based on Labour Force Survey data, our impact evaluation shows that many claimants leaving the register kept on searching with higher intensity, revealing preference for independent search. Decomposing the treatment effect by housing tenure, we find that the JSA increased the probability to find a job among mortgagers, induced a significant outflow from claimant status among renters, but not for outright owners and significantly less for mortgagers. This effect can be attributed to an increase in the search effort among owners that is not present for renters. Moreover, among unemployed leaving the claimant pool, private renters decreased their search effort relative to owners. These findings suggest that the JSA was especially successful for unemployed people with higher incentives to find a job. Moreover, homeowners proved to be better aligned to the objectives of the reform by keeping commitment to look for job, either because they adjusted their behaviour to the new requirements or because they continued to search independently.

ABSTRACT. We study the repeal of a regulation that imposed maximum wholesale and retail markups for all but five fresh fruits and vegetables. We compare the prices of products affected by regulation before and after the policy change and use the unregulated products as a control group. We find that abolishing regulation led to a significant decrease in both retail and wholesale prices. However, markup regulation directly affected wholesalers and only indirectly retailers. The results are consistent with markup ceilings providing a focal point for collusion among wholesalers.

ABSTRACT. We analyse the impact of the current financial crisis on the determination of technical efficiency in a sample of Italian small banks, highlighting the interaction of the crisis with different regulatory regimes existing for cooperative banks (CB’s) and other banks. Relatively to the extant literature on bank regulation and efficiency, we innovate by adopting a difference-in-difference approach, and by relying on a novel data-set, where the banks' economic environment is measured at a territorially very disaggregated level (the Sistemi Locali del Lavoro). We find that the crisis has a negative impact on efficiency, more so for CB's. This is to be expected, as the CB's principle of external mutuality and their branching regulations are likely to lock them in less performing areas. In accordance with this prior, the differential impact of the crisis strongly attenuates when we include in the production set some indicators of local environment (GDP per capita). Correspondingly, we find novel evidence in favour of the “bad luck” hypothesis.

ABSTRACT. We derive computationally simple and intuitive score tests of neglected serial correlation in unobserved component univariate models using frequency domain techniques. In some common situations in which the information matrix is singular under the null we derive extremum tests that are asymptotically equivalent to likelihood ratio tests, which become one-sided, and explain how to compute reliable Wald tests. We also explicitly relate the incidence of those problems to the model identification conditions and compare our tests with tests based on the reduced form prediction errors. Our Monte Carlo exercises assess the finite sample reliability and power of our proposed tests.

ABSTRACT. This paper develops an approach to rank testing that nests all existing rank tests and simplifies their asymptotics. The approach is based on the fact that implicit in every rank test there are estimators of the null spaces of the matrix in question. The approach yields many new insights about the behavior of rank testing statistics under the null as well as local and global alternatives in both the standard and the cointegration setting. The approach also suggests many new rank tests based on alternative estimates of the null spaces as well as the new fixed-b theory. A brief Monte Carlo study illustrates the results.

ABSTRACT. The exponential model for the spectrum of a time series is based on the Fourier series expansion of the logarithm of the spectral density. The coefficients of the expansion are the cepstral coefficients and their collection is the cepstrum of the time series. Approximate likelihood inference based on the periodogram leads to a generalised linear model for asymptotically independent exponential data with logarithmic link. The paper introduces the class of generalised linear cepstral models with Box-Cox link, which is based on the truncated Fourier series expansion of the Box-Cox transformation of the spectral density; the coefficients of the expansions can be termed generalised cepstral coefficients and are related to the generalised autocovariances of the series. The link function depends on a power transformation parameter, and encompasses the exponential model. Other important special cases are the inverse link (which leads to modelling the inverse spectrum), and the identity link. One of the merits of this model class is the possibility of nesting alternative spectral estimation methods (autoregressive, exponential, etc.) under the same likelihood-based framework.

ABSTRACT. We consider alternative asymptotics for frequency domain based estimates of the long run variance, in which the bandwidth is kept fixed. For a weakly dependent process, this does not yield a consistent estimate of the long run variance, but the standardized mean has t limit distribution. For given bandwidth, we find that this limit is more precise than the standard normal one. In presence of fractionally integrated data, the limit distribution of the estimate is not standard, and we derive critical values for the standardized mean for various bandwidths. We find that this asymptotics provided a better approximation than with Memory Autocorrelation Consistent (MAC) estimate. In multivariate set up, fixed bandwidth asymptotics may be used to characterize the limit distribution in alternative to standard Narrow Band asymptotics.

ABSTRACT. We develop a class of Poisson autoregressive models with additional covariates (PARX) that can be used to model and forecast time series of counts. We establish the time series properties of the models, including conditions for stationarity and existence of moments. These results are in turn used in the analysis of the asymptotic properties of the maximum-likelihood estimators of the models. The PARX class of models is used to analyse the time serie properties of monthly corporate defaults in the US in the period 1980-2011 using Önancial and economic variables as external covariates. Results show that our model is able to capture the time series dynamics of corporate defaults well, including the well known default counts clustering found in data. An out-of-sample analysis shows that in terms of forecast performance, the inclusion of covariates in the PARX speciÖcation is important. Finally, our empirical analysis allows us to shed some light on the presence of contagion e§ects over time

ABSTRACT. In the context of stock return predictability and optimal portfolio allocation, model combination methods have been shown to often produce improvements upon the performance of even the best individual models entering the combination. We add to this literature by proposing a novel model combination technique that combines the predictive densities of the individual models with combination weights that depend on how the individual models fare relative to the underlying objective function of the investor. Empirically, we find that our novel density combination method produces improvements both in terms of statistical and economic measures of out-of-sample predictability, relative to the best individual models entering the combination as well as a variety of existing model combination techniques. We also explore the importance of having both the regression parameters and the volatility of the return-predicting models change over time, and find that the gains from using our novel model combination method increase significantly when we allow for instabilities in the individual models entering the combination.

ABSTRACT. The paper investigates explanations for forecasting invariance to structural breaks. We isolate possible structural invariance in a simplified dynamic macro model which nevertheless has features in common with standard macro models. We find, as expected, that structural breaks in growth rates and in the means of long-run relationships will always damage some of the variables. But we also find examples of "insulation" from shocks. The results about partial robustness is a property of the economy itself (here represented by the DGP) and not of the forecasts.

ABSTRACT. The purpose of the present paper is to investigate the structure and dynamics of professionals’ forecast of inflation. Recent papers have focused on their forecast errors and how it may be affected by informational rigidities, or inattentiveness. In this paper we extend the existing literature by considering a second form of inattentiveness. While showing that both types of inattentiveness are closely related, we focus on the inattentiveness forecasters face when undertaking multi-period forecast and, thereby, the expected momentum of inflation. Using a number survey-based data for the US and UK, we establish a new structure for the professional’s forecast error with direct implications for the persistence of real effects.

ABSTRACT. We analyze the impact of immigration on voting. Using Italian municipality data and IV estimation strategy, we find that immigration generates a sizable causal increase in votes for the centre-right coalition, which has a political platform less favorable to immigrants. Additional findings are: big cities behave differently, with no impact of immigration on electoral outcomes; gains in votes for the centre-right coalition correspond to loss of votes for the centre-left parties, a decrease in voter turnout, and a rise in protest votes; cultural diversity, competition in the labor market and for public services are the most relevant channels at work.

ABSTRACT. Abstract We extend the univariate results in [Wooldridge, J. M. (2005): “Unobserved heterogeneity and estimation of average partial effects,” in Identification And Inference For Econometric Models: Essays In Honor Of Thomas Rothenberg, ed. by D. W. K. Andrews, and J. H. Stock. Cambridge University Press, New York] to multivariate probit models, proving the following. 1) Average partial effects (APEs) based on joint and marginal response probabilities are consistently estimated by conventional multivariate probit models under general forms of conditionally independent latent heterogeneity (LH). The normalization of choice is not neutral to consistency in models with cross-equation parameter restrictions beyond normalization, such as those implemented by Stata's asmprobit or in the panel probit model: if the normalization is through an error covariance matrix in correlation form, consistency breaks down, unless the LH components are truly homoskedastic. This is substantial since an error covariance matrix in correlation form is the only normalization allowed by Stata's biprobit and mvprobit or Limdep's BIVARIATE PROBIT and MPROBIT. Covariance restrictions beyond normalizations generally conflict with an arbitrary covariance matrix for the LH components. The multinomial probit model with i.i.d. errors, implemented by Stata's mprobit, is a case in point. 2) Conditional independence of the LH components is not generally sufficient for consistent estimation of APEs on conditional probabilities. Consistency is restored by maintaining an additional independence assumption. This holds true whether or not the response variables are used as regressors. 3) The dimensionality benefit observed by [Mullahy, J. (2011): “Marginal effects in multivariate probit and kindred discrete and count outcome models, with applications in health economics,” NBER WP SERIES 17588, NBER] in the estimation of partial effects extends to APEs. We exploit this feature in the design of a simple procedure estimating APEs, which is both faster and more accurate than simulation-based codes, such as Stata's mvprobit and cmp. To demonstrate the finite-sample implications of our results, we carry out extensive Monte Carlo experiments with bivariate and trivariate probit models. Finally, we apply our procedure in (3) to Italian survey data of immigrants in order to estimate the APEs of a trivariate probit model of ethnic identity formation and economic performance.

ABSTRACT. Using Italian data on language standardized tests for different levels of schooling we investigate 1) if the observed gap in educational attainments in first generation immigrants tends to lower the longer their stay in Italy and 2) if younger children tend to catch up faster than their older schoolmates. The analysis confirms the presence of a significant gap between natives and immigrants students in school outcomes for all grades, with first generation immigrants showing the largest gap. Further, the comparison between both first and second generation immigrant students and the results across the different grades suggest that the significant gap observed in the first generation is mainly due to the negative performance of immigrant children newly arrived in Italy and that interventions at younger ages are likely to be more effective. Finally, we find that also the immigrant students’ area of origin play a role in their schooling performance, suggesting that cultural differences affect children from different origins differently. We control for endogeneity concerns using both schools and classroom FE estimators, and results are robust to a specific sub-sample that controls for cheating, different model specifications and the use of math test scores as dependent variable.

ABSTRACT. We use twenty years of Italian administrative panel data, a uniquely rich source of information on internal migration experiences from the poorer South to the wealthy North, to identify the role of unobserved worker characteristics in the selection of migrants and returns to migration. We propose and implement a novel iterative estimation method for a switching regression model with the same worker-specific source of unobserved heterogeneity ("ability") present in the selection and both outcome equations. Estimated returns to ability are lower in the north than in the south of Italy and accordingly migrants tend to be drawn from the lower-end of the ability distribution. Around half the gains to migration are due to higher wages, and the other half due to greater labor market attachment. Differential returns to observable characteristics are far less important. Return migration reinforces the original negative selection of migrants, consistent with migrants facing considerable uncertainty about their income in Northern Italy.

ABSTRACT. Given a scalar random variable $Y$ and a random vector $X$ defined on the same probability space, the conditional distribution of $Y$ given $X$ can be represented by either the conditional distribution function or the conditional quantile function. To these equivalent representations correspond two alternative approaches to estimation. One approach, distributional regression (DR), is based on direct estimation of the conditional distribution function; the other approach, quantile regression (QR), is instead based on direct estimation of the conditional quantile function. Indirect estimates of the conditional quantile function and the conditional distribution function may then be obtained by inverting the direct estimates obtained from either approach. Despite the growing attention to the DR approach, and the vast literature on the QR approach, the link between the two approaches has not been explored in detail. The aim of this paper is to fill-in this gap by providing a better understanding of the relative performance of the two approaches, both asymptotically and in finite samples, under the linear location model and certain types of heteroskedastic location-scale models.

ABSTRACT. We provide a general dynamic model averaging (DMA) approach to sequential estimation of quantile regression models with time-varying parameters. We build a new sequential Markov-chain Monte Carlo (MCMC) algorithm to approximate the posterior distribution of models and parameters. The efficiency and the effectiveness of the proposed DMA approach and the MCMC algorithm is shown through simulation studies and applications to macro-economics and finance.

ABSTRACT. This study attempts to estimate the causal effect of staying longer in the labor force on the distribution of lifetime income and assess its consequences for overall inequality in lifetime income. Results for cross-national setting are estimated through Local Quantile Treatment Effect estimator by Abadie, Angrist and Imbens (2002), and are confronted with the Instru- mental Variables Quantile Regression by Chernozukov and Hansen (2005). Relevant country specific estimates rely on Frandsen, Froelich and Melly (2012) approach. While the results of cross-national setting clearly suggest heterogenous effect across the distribution, negative at the bottom tail, increasing in magnitude across the quantiles, the results of country specific estimates are less readible. The effect of postponing retirement to later ages on the overall inequality is being assessed based on the notion of Stochastic Dominance by Abadie (2002).

ABSTRACT. Quantile aggregation (or 'Vincentization') is a simple and intuitive way of combining probability distributions, originally proposed in Vincent (1912). In certain cases, such as under Gaussianity, the Vincentized distribution belongs to the same family as that of the individual distributions and it can be obtained by averaging the individual parameters. This papers compares the properties of quantile aggregation with those of the forecast combination schemes normally adopted in the econometric forecasting literature, based on linear or logarithmic averages of the individual densities. In general we find that: (i) larger differences among the combination schemes occur when there are biases in the individual forecasts, in which case quantile aggregation seems overall preferable; (ii) the choice of the combination weights is important in determining the performance of the different methods. Monte Carlo simulation experiments with forecasts with partially misspecified time series models indicate that the properties of quantile aggregation are in between those of the linear and the logarithmic pool and that quantile averaging appears particularly useful for combining forecast distributions with large differences in location. An empirical illustration is provided with density forecasts from time series and econometric models for Italian GDP.