Artificial neural network regression models in a panel setting: Predicting economic growth

Economic time series often feature non-linear structures such as non-linear time trends, non-linear autoregressive effects, and non-linear interaction effects. In this paper, it is shown that artificial neural network regression models are suitable tools for the analysis of economic panel data because they allow for a compromise between the ability to model these features and the model size. As model specification is a concern in artificial neural network models, previous approaches are discussed critically. It is shown that the growth rates of the gross domestic product of 24 industrialized economies in the period 1992–2016 follow a non-linear time trend which cannot be explained by autoregressive features or polynomial time variables. The unrestricted functional form of the time trend in the artificial neural network model is also the main reason for the superior statistical performance compared to conventional panel models. This is confirmed by out-of-sample predictions for 2017.     

A new methodology for the logistic analysis of evolutionary S-shaped processes: Application to historical time series and forecasting

A new multi-logistic methodology to analyze long range time series of evolutionary S-shaped processes is presented. It conceptually innovates over the traditional logistic approach. The ansatz includes computing the residuals to an optimized multi-logistic trend curve least squares fitted to the time-series data. The elements of the residuals series are checked for autocorrelations and once detected the residuals series is further analyzed to search for eventual presence of underlying periodic structures using a truncated Fourier sine series. The method foundations ensures both a universal applicability and a capacity to disclose the existence of active clocks that can be possibly traced to the driving motors of the evolutionary character of the time series, due to the responsiveness of corresponding process to the development of economic cycles. On associating these two views, it is found that the methodology has a strong potential to improve the quality of short-term forecasts. These findings have been put to test through applications of the methodology to studying the time evolution of two commodities of strong economic and social importance (corn and steel) and good results were consistently obtained for both the analytical and forecasting aspects.     

Economic forecasting with evolved confidence indicators

We present a machine-learning method for sentiment indicators construction that allows an automated variable selection procedure. By means of genetic programming, we generate country-specific business and consumer confidence indicators for thirteen European economies. The algorithm finds non-linear combinations of qualitative survey expectations that yield estimates of the expected rate of economic growth. Firms’ production expectations and consumers’ expectations to spend on home improvements are the most frequently selected variables – both lagged and contemporaneous. To assess the performance of the proposed approach, we have designed an out-of-sample iterative predictive experiment. We found that forecasts generated with the evolved indicators outperform those obtained with time series models. These results show the potential of the methodology as a predictive tool. Furthermore, the proposed indicators are easy to implement and help to monitor the evolution of the economy, both from demand and supply sides.     

Unit Root Investigation of Greek Real Money Supply and GDP

In this article, the time series of Greek real GDP and real money supply are investigated for the presence of a unit root, allowing for maximum two breaks which take place at an unknown point in time. This methodology is preferred to conventional Dickey & Fuller tests because the covered time horizon, namely from 1858 to 1938, is characterized by a number of very important events, the nature of which is either economic or historical. In addition, time series stationarity is checked through a Kwiatkowski, Phillips, Schmidt, and Shin (KPSS) test.University of Macedonia—Greece. The article was presented at the Fifty-Ninth International Atlantic Economic Conference, London, England, March 9–13, 2005     

The intraday directional predictability of large Australian stocks: A cross-quantilogram analysis

This study investigates the directional predictability of overnight periods for intraday returns of large Australian stocks. The intraday reactions to overnight developments are studied using cross-quantilograms, a new, flexible methodology that facilitates detailed insights into the quantile dependence between two time series. The results provide evidence for the existence of intraday reversals after overnight periods that carry very bad news, whereas the picture of the short-term reactions to very positive overnight returns is mixed. The observed rebounds concern extreme quantiles and occur with a short delay during the first part of the trading day. The study also shows that continuation and reversal effects are not mutually exclusive. The economic significance of the identified patterns is illustrated by analysing the performance of a simple contrarian strategy.     

Estimation of the business cycle: A modified Hodrick-Prescott filter

Hodrick-Prescott (HP) filtering of (most often, seasonally adjusted) quarterly series is analysed. Some of the criticism to the filter are adressed. It is seen that, while filtering strongly affects autocorrelations, it has little effect on crosscorrelations. It is argued that the criticism that HP filtering induces a spurious cycle in the series is unwarranted. The filter, however, presents two serious drawbacks: First, poor performance at the end periods, due to the size of the revisions in preliminary estimators, and, second, the amount of noise in the cyclical signal, which seriously disturbs its interpretation. We show how the addition of two model-based features (in particular, applying the filter to the series extended with proper ARIMA forecasts and backcasts, and using as input to the filter the trend-cycle component instead of the seasonally adjusted series) can considerably improve the filter performance. Throughout the discussion, we use a computationally and analytically convenient alternative derivation of the HP filter, and illustrate the results with an example consisting of 4 Spanish economic indicators.     

Economy through a lens: Distortions of policy coverage in UK national newspapers

This paper shows that communication of economic news varies across newspapers in the United Kingdom. We develop new time series of economic news tonality using a unique dataset of policy influenced articles published in major UK newspapers. We show that the volume and tonality of news respond to current economic conditions. For example, the nature of news changes around events of economic uncertainty such as the global financial crisis and the post-EU referendum periods. We also provide illustrative evidence that communication differs across newspaper formats. Tabloids, as opposed to quality newspapers, tend to express news more negatively, and mostly report policy-related news during periods of economic stress. The integral importance of these results is illustrated by news reaction curves showing a strong positive relationship mostly lasting three months between consumer sentiments and news.     

Winner's curse without overbidding

We report the results of a series of second-price auction experiments where each bidder's signal is given by a normally distributed value plus a normally distributed error. While bidders’ values differ in one treatment they are the same in another, which allows for a direct test of the “winner's curse” irrespective of confounding factors. Bidders may also fall prey to a “news curse” when they do not sufficiently take into account that signals and errors are correlated. We find that the effects of the winner's curse are mitigated by a news curse and loss or risk aversion.     

Does consumer confidence affect durable goods spending during bad and good economic times equally?

Recent econometric analysis shows consumer confidence innovations have long lasting effects on economic activities like consumption. Using US data, we show this conclusion is more nuanced when considering an economy that has different potential states. We investigate regime-switching models which use the National Bureau of Economic Research US business cycle expansion and contraction data to create an indicator series that distinguishes bad and good economic times and use this series to investigate impulse responses and variance decompositions. We show the connection between consumer confidence to some types of consumer purchases is important during good economic times, but is relatively unimportant during bad economic times. We also use this type of model to investigate the connection between news and consumer confidence and this connection is also shown to be state dependent. In the context of the animal spirits versus news debate, our findings show that during economic expansions, consumer confidence shocks likely reflect news, while during economic contractions, consumer confidence shocks are consistent with animal spirits. These findings also have important implications for recent policy debates which consider whether confidence boosting policies, like raising inflation expectations on big-ticket items such as automobiles or business equipment, would lead to a faster recovery.     

Global Distortions to Agricultural Markets: Indicators of Trade and Welfare Impacts, 1960 to 2007

Despite recent reforms, world agricultural markets remain highly distorted by government policies. Traditional indicators of those price distortions such as producer and consumer support estimates (PSEs and CSEs) can be poor guides to the policies' economic effects. Recent theoretical literature provides scalar index numbers of trade‐ and welfare‐reducing effects of price and trade policies which this paper builds on to develop more‐satisfactory indexes that can be generated using no more than the data used to generate PSEs and CSEs. We then exploit a new Agricultural Distortion database to provide time‐series estimates of index numbers for 75 developing and high‐income countries over the past half‐century.     

Linear and non-linear price decentralization

Compendious and thorough solutions to the existence of a linear price equilibrium problem, the second welfare theorem, and the limit theorem on the core are provided for exchange economies whose consumption sets are the positive cone of arbitrary ordered Fréchet spaces—dispensing entirely with the assumption that the vector ordering of the commodity space is a lattice. The motivation comes from economic applications showing the need to bring within the scope of equilibrium theory vector orderings that are not lattices, which arise in the typical model of portfolio trading with missing options. The assumptions are on the primitives of the model. They are bounds on the marginals of non-linear prices and for ω-proper economies they are both sufficient and necessary.     

News and monetary shocks at a high frequency: A simple approach

We develop a simple approach to identify economic news and monetary shocks at a high frequency. The approach is used to examine financial market developments in the United States following the Federal Reserve’s May 22, 2013 taper talk suggesting that it would begin winding down its quantitative easing program. Our findings show that the sharp rise in 10-year Treasury bond yields immediately after the taper talk was largely due to monetary shocks, with positive economic news becoming increasingly important in subsequent months.     

Sign and phase asymmetry: News, economic activity and the stock market

In this paper, we present a general model of the joint data generating process underlying economic activity and stock market returns allowing for complex nonlinear feedbacks and interdependencies between the conditional means and conditional volatilities of the variables. We propose statistics that capture the long and short run responses of the system to the arrival of news, conditioning on the sign and time of arrival of the news. The model is applied to US data. We find that there are significant differences between the short and long run responses of economic activity and stock returns to the arrival of news. Moreover, for certain classifications of news, the respective responses of economic activity and stock returns vary according to the nature of the news and the phase of the business cycle at which the news arrives.     

Is there any common knowledge news in the Euro/Dollar market?

This paper analyzes the effect of nine categories of news announcements on the quoting activity of individual foreign exchange (FX) dealers on the Euro/Dollar exchange rate from May to October 2001. We use the double autoregressive conditional Poisson model (DACP), which is designed for time series of count data, which can be both under- or overdispersed. We find that dealers' quoting activity reacts differently to the same announcements, some increasing their activity, whilst others decrease it in response to the same news. Based on the taxonomy of Evans [Evans, M. (2002), Fx trading and exchange rate dynamics. Journal of Finance 57(6), 2405–2447.], we classify our news categories in two groups: common knowledge (CK) and non-common knowledge (NCK) news, according to their effects on quoting activity and price changes. Finally we show that scheduled news announcements are NCK news, and there is no evidence for the existence of CK news amongst our announcements, which means that dealers hardly get a consensus in interpreting the news content.     

Electoral systems and the effects of political events on the stock market: The Belgian case

Efficient stock markets react to news. News about future economic policies can be derived from political events such as elections, the formation of new governments, changes in the composition of governments, etc. However, the news content of these events depends on the electoral system. In the American electoral system, characterized as it is by majority representation and single–party governments, elections generate news to the extent that the results are unexpected. In countries with proportional representation, governments are frequently multi–party coalitions whose composition is difficult to predict from the election results. These results therefore contain much less information about future policies. Our results, obtained for the Brussels stock market, support this distinction. Furthermore, the ideological composition of the government also matters; these effects support a rational partisan approach.     

ASYMMETRIC INFORMATION IMPACTS: EVIDENCE FROM THE AUSTRALIAN TREASURY-BOND FUTURES MARKET

Abstract.  The effect of information flows on the return volatility of Australian 3-year Treasury bond futures is examined using linear and non-linear GARCH models. Results show significant asymmetric information effects, where bad news has a greater impact on volatility than good news and a non-linear Threshold ARCH(1,1) in mean model provides the most accurate estimation of return volatility. Diagnostic tests confirm this finding and out of sample forecasting error statistics verify that the Threshold ARCH(1,1) in mean model yields the lowest forecasting error. The Threshold ARCH(1,1)-M model is best at capturing the asymmetric information impact on the Australian three-year T-Bond futures return volatility.     

我国实际产出序列非对称性和非线性特征的统计检验——基于时域形变模型的检验分析

实际产出序列的非对称性和非线性是经济周期波动的重要动态特征。本文利用时间序列的时域形变检验方法分析我国实际产出序列。研究发现:我国实际产出序列存在一定程度的非对称性和非线性;我国实际产出序列的扩张过程无论是在幅度方面还是持续性方面均强于紧缩过程。这意味着现阶段我国经济增长具有高位持续的趋势,本轮经济周期将形成高位底部并具有拖长的倾向。     

Testing the stationarity of economic time series: further Monte Carlo evidence

Testing the stationarity of economic time series has become a central issue in empirical economics. This paper evaluates, via Monte Carlo simulation, the empirical power and size of the augmented Dickey-Fuller test for a unit root (ADF test), the most widely used in empirical works, and of the test recently proposed by Kwiatkowski et al. (1992; KPSS test), where the null hypothesis is one of stationarity. The evidence confirms that both procedures suffer from very low power and, more so in the case of the KPSS test, large size distortions, especially in samples of the sizes usually available in practical applications. Moreover, their performance is highly sensitive to the true generating process, as well as to the way one parameterizes each test. It is shown, however, that a combined ADF-KPSS procedure would significantly reduce the number of erroneous conclusions, although at the cost of producing a fairly large number of inconclusive answers.     

Instrumental variables and wavelet decompositions

The application of wavelet analysis provides an orthogonal decomposition of a time series by time scale, thereby facilitating the decomposition of a data series into the sum of a structural component and a random error component. The structural components revealed by the wavelet analysis yield nearly ideal instrumental variables for variables observed with error and for co-endogenous variables in simultaneous equation models. Wavelets also provide an efficient way to explore the path of the structural component of the series to be analyzed and can be used to detect some specification errors. The methodology described in this paper is applied to the errors in variables problem and simultaneous equations case using some simulation exercises and to the analysis of a version of the Phillips curve with interesting results.     

地方经济增长目标与要素市场扭曲*

本文采用中国2005—2017年省级政府工作报告中的经济增长目标数据，实证检验了地方经济增长目标对要素市场扭曲的影响。研究发现：（1）地方经济增长目标对要素市场具有显著的扭曲效应。通过干预信贷资源配置、扩大土地出让和压低劳动力工资等方式扭曲要素市场是地方政府实现经济增长目标的主要方式。（2）地方经济增长目标对要素市场扭曲的影响存在区域异质性。经济增长目标对要素市场的扭曲效应在中西部地区和保增长压力较大地区更为明显。（3）地方经济增长目标对要素市场扭曲的影响存在基于制度环境的门槛效应。随着金融制度、法律制度和产权制度的改善，经济增长目标对要素市场的扭曲效应将会减弱。本文为政府目标管理和要素市场扭曲等领域的研究提供了新视角，为相关政策的制定提供了重要的现实依据。