On predicting the semiconductor industry cycle: a Bayesian model averaging approach

This study considers the model uncertainty and utilizes the Bayesian model averaging (BMA) approach to identify useful predictors of the semiconductor industry cycle from a list of 70 potential predictors. The posterior inclusion probabilities, posterior means, and posterior standard deviations over the period of 1995:05–2012:10 are estimated and consequently used to identify the main determinants of the industry cycle. It is found that the Philadelphia Semiconductor Index and total inventories in various downstream industries have important roles in signaling the industry growth. The results from an out-of-sample forecasting exercise also reveal the predictive potential and usefulness of BMA for the long-term prediction.     

Forecasting house prices using dynamic model averaging approach: Evidence from China

Forecasting house price has been of great interests for macroeconomists, policy makers and investors in recent years. To improve the forecasting accuracy, this paper introduces a dynamic model averaging (DMA) method to forecast the growth rate of house prices in 30 major Chinese cities. The advantage of DMA is that this method allows both the sets of predictors (forecasting models) as well as their coefficients to change over time. Both recursive and rolling forecasting modes are applied to compare the performance of DMA with other traditional forecasting models. Furthermore, a model confidence set (MCS) test is used to statistically evaluate the forecasting efficiency of different models. The empirical results reveal that DMA generally outperforms other models, such as Bayesian model averaging (BMA), information-theoretic model averaging (ITMA) and equal-weighted averaging (EW), in both recursive and rolling forecasting modes. In addition, in recent years it is found that the Google search index, instead of fundamental macroeconomic or monetary indicators, has developed greater predictive power for house price in China.     

Growth,convergence and public investment. A Bayesian model averaging approach

The aim of this study is twofold. First, the determinants of economic growth are studied among a wide set of potential variables for the Spanish provinces (NUTS3). Among others, various types of private, public and human capital in the group of growth factors are included. Also, it is analysed whether Spanish provinces have converged in economic terms in recent decades. The second objective is to obtain cross-section and panel data parameter estimates that are robust to model specification. For this purpose, a Bayesian Model Averaging (BMA) approach is used. Bayesian methodology constructs parameter estimates as a weighted average of linear regression estimates for every possible combination of included variables. The weight of each regression estimate is given by the posterior probability of each model.     

Research & development and growth: A Bayesian model averaging analysis

We examine the effect of research and development (R&D) on long-term economic growth using the Bayesian model averaging (BMA) to deal rigorously with model uncertainty. Previous empirical studies, which applied BMA, investigated the effect of dozens of regressors on long-term growth, but they did not examine the effect of R&D due to data unavailability. We extend these studies by proposing to capture the investment in R&D by the number of Nobel prizes in science. Using our indicator, the estimates show that R&D exerts a positive effect on long-term growth. This result is robust to many different parameter and model prior structures as well as to alternative definitions of R&D indicator.     

Model uncertainty and asset return predictability: an application of Bayesian model averaging

We investigate model uncertainty associated with predictive regressions employed in asset return forecasting research. We use simple combination and Bayesian model averaging (BMA) techniques to compare the performance of these forecasting approaches in short-vs. long-run horizons of S&P500 monthly excess returns. Simple averaging involves an equally-weighted averaging of the forecasts from alternative combinations of factors used in the predictive regressions, whereas BMA involves computing the predictive probability that each model is the true model and uses these predictive probabilities as weights in combing the forecasts from different models. From a given set of multiple factors, we evaluate all possible pricing models to the extent, which they describe the data as dictated by the posterior model probabilities. We find that, while simple averaging compares quite favorably to forecasts derived from a random walk model with drift (using a 10-year out-of-sample iterative period), BMA outperforms simple averaging in longer compared to shorter forecast horizons. Moreover, we find further evidence of the latter when the predictive Bayesian model includes shorter, rather than longer lags of the predictive factors. An interesting outcome of this study tends to illustrate the power of BMA in suppressing model uncertainty through model as well as parameter shrinkage, especially when applied to longer predictive horizons.     

Forecasting European economic policy uncertainty

Forecasting the economic policy uncertainty in Europe is of paramount importance given the ongoing sovereign debt crisis. This paper evaluates monthly economic policy uncertainty index forecasts and examines whether ultra‐high frequency information from asset market volatilities and global economic uncertainty can improve the forecasts relatively to the no‐change forecast. The results show that the global economic policy uncertainty provides the highest predictive gains, followed by the European and US stock market realized volatilities. In addition, the European stock market implied volatility index is shown to be an important predictor of the economic policy uncertainty.     

Forecasting economic magnitudes with neural network models

This paper is an introduction to artificial neural networks as a statistical and econometric tool. The fundamentals of the theory are presented and two applications illustrate the power of artificial neural networks in predicting results.An earlier version of this paper was incorrectly printed in the May 1999 issue ofIAER.     

Forecasting economic magnitudes with neural network models

This paper is an introduction to artificial neural networks as a statistical and econometric tool. The fundamentals of the theory are presented and two applications illustrate the power of artificial neural networks in predicting results.     

Bayesian averaging of classical estimates in forecasting macroeconomic indicators with application of business survey data

In this paper, we develop a methodology for forecasting key macroeconomic indicators, based on business survey data. We estimate a large set of models, using an autoregressive specification, with regressors selected from business and household survey data. Our methodology is based on the Bayesian averaging of classical estimates method. Additionally, we examine the impact of deterministic and stochastic seasonality of the business survey time series on the outcome of the forecasting process. We propose an intuitive procedure for incorporating both types of seasonality into the forecasting process. After estimating the specified models, we check the accuracy of the forecasts.     

Credit growth,monetary policy and economic activity in a three-regime TVAR model

We employ a Threshold Vector Autoregression (TVAR) methodology in order to examine the nonlinear nature of the interactions among credit market conditions, monetary policy and economic activity. We depart from the existing literature on the subject along two dimensions. First, we focus on a model in which the relevant threshold variable describes the state of economic activity rather than credit market conditions. Second, in contrast to the existing TVAR literature, which concentrates exclusively on single-threshold models, we allow for the presence of a second threshold, which is overwhelmingly supported by all relevant statistical tests. Our results indicate that the dynamics of the interactions among credit market conditions, monetary policy and economic activity change considerably as the economy moves from one phase of the business cycle to another and that single-threshold TVAR models are too restrictive to fully capture the nonlinear nature of those interactions. The impact of most shocks tends to be largest during periods of subpar economic activity and smallest during times of moderate economic growth. By contrast, credit risk shocks have the largest impact when output growth is considerably above its long-term trend.     

TFP growth and its determinants: a model averaging approach

Total Factor Productivity (TFP) accounts for a sizable proportion of the income differences across countries. Two challenges remain to researchers aiming to explain these differences: on the one hand, TFP growth is hard to measure empirically; on the other hand, model uncertainty hampers consensus on its key determinants. This paper combines a non-parametric measure of TFP growth with Bayesian model averaging techniques in order to address both issues. Our empirical findings suggest that the most robust TFP growth determinants are time-invariant unobserved heterogeneity and trade openness. We also investigate the main determinants of two TFP components: efficiency change (i.e., catching up) and technological progress.     

Unanticipated inflation and economic activity

This paper presents a new empirical test of the expectations-adjusted supply fucntion, obtaining the market's inflationary expectations from the market interest rate. The results support the existence of a positive relationship between unanticipated inflation and the level of economic activity, while anticipated inflation seems to have no effect.     

Budget deficits and economic activity

Using annual data of the Greek economy, this paper explores the relationship between budget deficits and real output. The ultimate purpose is to empirically evaluate the validity of the Keynesian proposition and the Ricardian equivalence hypothesis. The econometric methodology is based on error-correction modeling, Granger bivariate and trivariate causality, and Hendry's general-to-specific technique. The ECM results are consistent with the Keynesian proposition, suggesting a significant and positive relationship between budget deficits and real GNP.     

Dimension reduction and model averaging for estimation of artists' age-valuation profiles

In hedonic regression models of the valuation of works of art, the age or period at which an artist produces a particular work is often found to have highly significant predictive value. Most existing results are based on regressions that pool many painters. Although the uniqueness of artists' career paths makes it interesting to estimate such regressions for individual artists, sample sizes are often inadequate for a model that would also include the large number of other relevant variables. We address this problem of inadequate degrees of freedom in individual artist regressions by using two statistical methods (model averaging and dimension reduction) to incorporate information from a potentially large number of predictor variables, despite relatively small samples. We find that individual age-valuation profiles can differ substantially from general pooled profiles, suggesting that methods that are more responsive to the unique features of individual artists may provide better predictions of art valuations at auction, and may be useful more generally in hedonic valuation problems.     

Forecasting technology diffusion with the Richards model

The Richards model has a shape parameter m that allows it to fit any sigmoidal curve. This article demonstrates the ability of a modified Richards model to fit a variety of technology diffusion curvilinear data that would otherwise be fit by Bass, Gompertz, Logistic, and other models. The performance of the Richards model in forecasting was examined by analyzing fragments of data computed from the model itself, where the fragments simulated either an entire diffusion curve but with sparse data points, or only the initial trajectory of a diffusion curve but with dense data points. It was determined that accurate parameter estimates could be obtained when the data was sparse but traced out the curve at least up to the third inflection point (concave down), and when the data was dense and traced out the curve up to the first inflection point (concave up). Rogers' Innovation I, II and III are discussed in the context of the Richards model. Since m is scale independent, the model allows for a typology of diffusion curves and may provide an alternative to Rogers' typology.     

Federal lending and aggregate economic activity

Attention to federal activity in credit markets is typically focused on the government's role as a borrower. In contrast, scant attention is paid to its equally large and dominant role as a lender. This paper evaluates the aggregate impact of federal lending activity within the framework of a vector autoregressive representation of the US macroeconomy. The empirical regularities uncovered suggest that aggregate federal lending activity does not have a net positive impact on output.     

First announcements and real economic activity

The recent literature suggests that first announcements of real output growth in the US have predictive power for the future course of the economy while the actual value of output growth does not. We show that this need not point to a behavioural relationship, whereby agents respond to perceptions instead of the truth, but may instead simply be a by-product of the data revision process. The revisions to the initial estimates which define the final values of the observations are shown to be key in determining any relationship between first announcements and the future course of the economy.