Search

Working Paper The main objective of this study is two-fold. First, we aim to detect the underlying existing periodicities in business and consumer survey data. With this objective, we conduct a spectral analysis of all survey indicators. Second, we aim to provide researchers with a filter especially designed for business and consumer survey data that circumvents the a priori assumptions of other filtering methods. To this end, we design a low-pass filter that allows extracting the components with periodicities similar to those that can be found in the dynamics of economic activity. The European Commission (EC) conducts monthly business and consumer tendency surveys in which respondents are asked whether they expect a set of variables to rise, fall or remain unchanged. We apply the Welch method for the detection of periodic components in each of the response options of all monthly survey indicators. This approach allows us to extract the harmonic components that correspond to the cyclic and seasonal patterns of the series. Unlike other methods for spectral density estimation, the Welch algorithm provides smoother estimates of the periodicities. We find remarkable differences between the periodicities detected in the industry survey and the consumer survey. While business survey indicators show a common cyclical component of low frequency that corresponds to about four years, for most consumer survey indicators we do not detect any relevant cyclic components, and the obtained lower frequency periodicities show a very irregular pattern across questions and reply options. Most methods for seasonal adjustment are based on a priori assumptions about the structure of the components and do not depend on the features of the specific series. In order to overcome this limitation, we design a low-pass filter for survey indicators. We opt for a Butterworth filter and apply a zero-phase filtering process to preserve the time alignment of the time series. This procedure allows us to reject the frequency components of the survey indicators that do not have a counterpart in the dynamics of economic activity. We use the filtered series to compute diffusion indexes known as balances, and compare them to the seasonally-adjusted balances published by the EC. Although both series are highly correlated, filtered balances tend to be smoother for the consumer survey indicators.

Working Paper In this study we construct quarterly consumer confidence indicators of unemployment for the euro area using as input the consumer expectations for sixteen socio-demographic groups elicited from the Joint Harmonised EU Consumer Survey. First, we use symbolic regressions to link unemployment rates to qualitative expectations about a wide range of economic variables. By means of genetic programming we obtain the combination of expectations that best tracks the evolution of unemployment for each group of consumers. Second, we test the out-of-sample forecasting performance of the evolved expressions. Third, we use a state-space model with time-varying parameters to identify the main macroeconomic drivers of unemployment confidence and to evaluate whether the strength of the interplay between variables varies across the economic cycle. We analyse the differences across groups, obtaining better forecasts for respondents comprised in the first quartile with regards to the income of the household and respondents with at least secondary education. We also find that the questions regarding expected major purchases over the next 12 months and savings at present are by far, the variables that most frequently appear in the evolved expressions, hinting at their predictive potential to track the evolution of unemployment. For the economically deprived consumers, the confidence indicator seems to evolve independently of the macroeconomy. This finding is rather consistent throughout the economic cycle, with the exception of stock market returns, which governed unemployment confidence in the pre-crisis period.

Working paper In this work we assess the role of data characteristics in the accuracy of machine learning (ML) tourism forecasts from a spatial perspective. First, we apply a seasonal-trend decomposition procedure based on non-parametric regression to isolate the different components of the time series of international tourism demand to all Spanish regions. This approach allows us to compute a set of measures to describe the features of the data. Second, we analyse the performance of several ML models in a recursive multiple-step-ahead forecasting experiment. In a third step, we rank all seventeen regions according to their characteristics and the obtained forecasting performance, and use the rankings as the input for a multivariate analysis to evaluate the interactions between time series features and the accuracy of the predictions. By means of dimensionality reduction techniques we summarise all the information into two components and project all Spanish regions into perceptual maps. We find that entropy and dispersion show a negative relation with accuracy, while the effect of other data characteristics on forecast accuracy is heavily dependent on the forecast horizon.

Working paper In this study we present a geometric approach to proxy economic uncertainty. We design a positional indicator of disagreement among survey-based agents' expectations about the state of the economy. Previous dispersion-based uncertainty indicators derived from business and consumer surveys exclusively make use of the two extreme pieces of information coming from the respondents expecting a variable to rise and to fall. With the aim of also incorporating the information coming from the share of respondents expecting a variable to remain constant, we propose a geometrical framework and use a barycentric coordinate system to generate a measure of disagreement, referred to as a discrepancy indicator. We assess its performance, both empirically and experimentally, by comparing it to the standard deviation of the share of positive and negative responses, which has been used by Bachman et al. (2013) as a proxy for economic uncertainty. When applied in sixteen European countries, we find that both time-varying metrics co-evolve in most countries for expectations about the country's overall economic situation in the present, but not in the future. Additionally, we obtain their simulated sampling distributions and we find that the proposed indicator gravitates uniformly towards the three vertices of the simplex representing the three answering categories, as opposed to the standard deviation, which tends to overestimate the level of uncertainty as a result of ignoring the no-change responses. Consequently, we find evidence that the information coming from agents expecting a variable to remain constant has an effect on the measurement of disagreement.

Working paper In this study we use agents’ expectations about the state of the economy to generate indicators of economic activity in twenty-six European countries grouped in five regions (Western, Eastern, and Southern Europe, and Baltic and Scandinavian countries). We apply a data-driven procedure based on evolutionary computation to transform survey variables in economic growth rates. In a first step, we design five independent experiments to derive the optimal combination of expectations that best replicates the evolution of economic growth in each region by means of genetic programming, limiting the integration schemes to the main mathematical operations. We then rank survey variables according to their performance in tracking economic activity, finding that agents’ “perception about the overall economy compared to last year” is the survey variable with the highest predictive power. In a second step, we assess the out-of-sample forecast accuracy of the evolved indicators. Although we obtain different results across regions, Austria, Slovakia, Portugal, Lithuania and Sweden are the economies of each region that show the best forecast results. We also find evidence that the forecasting performance of the survey-based indicators improves during periods of higher growth.

Working paper This study presents a multiple-input multiple-output (MIMO) approach for multi-step-ahead time series prediction with a Gaussian process regression (GPR) model. We assess the forecasting performance of the GPR model with respect to several neural network architectures. The MIMO setting allows modelling the cross-correlations between all regions simultaneously. We find that the radial basis function (RBF) network outperforms the GPR model, especially for long-term forecast horizons. As the memory of the models increases, the forecasting performance of the GPR improves, suggesting the convenience of designing a model selection criteria in order to estimate the optimal number of lags used for concatenation.

Working paper By means of Self-Organizing Maps we cluster fourteen European countries according to the most suitable way to model their agents’ expectations. Using the financial crisis of 2008 as a benchmark, we distinguish between those countries that show a progressive anticipation of the crisis and those where sudden changes in expectations occur. By mapping the trajectory of economic experts’ expectations prior to the recession we find that when there are brisk changes in expectations before impending shocks, Artificial Neural Networks are more suitable than time series models for modelling expectations. Conversely, in countries where expectations show a smooth transition towards recession, ARIMA models show the best forecasting performance. This result demonstrates the usefulness of clustering techniques for selecting the most appropriate method to model and forecast expectations according to their behaviour.

Working paper This study aims to analyze the effects of data pre-processing on the performance of forecasting based on neural network models. We use three different Artificial Neural Networks techniques to forecast tourist demand: a multi-layer perceptron, a radial basis function and an Elman neural network. The structure of the networks is based on a multiple-input multiple-output setting (i.e. all countries are forecasted simultaneously). We use official statistical data of inbound international tourism demand to Catalonia (Spain) and compare the forecasting accuracy of four processing methods for the input vector of the networks: levels, growth rates, seasonally adjusted levels and seasonally adjusted growth rates. When comparing the forecasting accuracy of the different inputs for each visitor market and for different forecasting horizons, we obtain significantly better forecasts with levels than with growth rates. We also find that seasonally adjusted series significantly improve the forecasting performance of the networks, which hints at the significance of deseasonalizing the time series when using neural networks with forecasting purposes. These results reveal that, when using seasonal data, neural networks performance can be significantly improved by working directly with seasonally adjusted levels.

Working Papers This study attempts to improve the forecasting accuracy of tourism demand by using the existing common trends in tourist arrivals form all visitor markets to a specific destination in a multiple-input multiple-output (MIMO) structure. While most tourism forecasting research focuses on univariate methods, we compare the performance of three different Artificial Neural Networks in a multivariate setting that takes into account the correlations in the evolution of inbound international tourism demand to Catalonia (Spain). We find that the MIMO approach does not outperform the forecasting accuracy of the networks when applied country by country, but it significantly improves the forecasting performance for total tourist arrivals. When comparing the forecast accuracy of the different models, we find that radial basis function networks outperform multilayer-perceptron and Elman networks.

Working Paper This study attempts to assess the forecasting accuracy of Support Vector Regression (SVR) with regard to other Artificial Intelligence techniques based on statistical learning. We use two different neural networks and three SVR models that differ by the type of kernel used. We focus on international tourism demand to all seventeen regions of Spain. The SVR with a Gaussian kernel shows the best forecasting performance. The best predictions are obtained for longer forecast horizons, which suggest the suitability of machine learning techniques for medium and long term forecasting.

Working paper This paper aims to compare the performance of different Artificial Neural Networks techniques for tourist demand forecasting. We test the forecasting accuracy of three different types of architectures: a multi-layer perceptron, a radial basis function and an Elman network. We also evaluate the effect of the memory by repeating the experiment assuming different topologies regarding the number of lags introduced. We used tourist arrivals from all the different countries of origin to Catalonia from 2001 to 2012. We find that multi-layer perceptron and radial basis function models outperform Elman networks, being the radial basis function architecture the one providing the best forecasts when no additional lags are incorporated. These results indicate the potential existence of instabilities when using dynamic networks for forecasting purposes. We also find that for higher memories, the forecasting performance obtained for longer horizons improves, suggesting the importance of increasing the dimensionality for long term forecasting.