Blaming suicide on NASA and divorce on margarine: the hazard of using cointegration to derive inference on spurious correlation

Contrary to common belief, cointegration testing may not distinguish between spurious relations and genuine ones. It is demonstrated that highly correlated series appear as cointegrated, even though common sense tells us that the underlying relation does not make sense. Empirical testing using simulated data, data from daily life and historical data on interest rates shows that cointegration may fail not only to detect spurious correlation but also to capture cointegration in a genuine relation. Cointegration testing to reveal spurious correlation can only be used in conjunction with theory, common sense and intuition.     

Are the effects of fiscal policy really nonlinear? A note

Based on a panel data model, Giavazzi, Jappelli and Pagano (2000, European Economic Review 44, 1259–1289) recently found evidence that national saving in OECD countries responds nonlinearly to fiscal policy, the nonlinearity being associated with the size and persistence of the fiscal impulse. The existence of this nonlinearity would have important policy implications because it implies that the short-run costs of a fiscal consolidation program are lower the larger and more persistent it is. However, this paper shows that their finding is not robust, one reason being the inadequacy of the slope homogeneity assumption implicit in their panel model and another reason being their reliance on an implausibly large number of episodes of large and persistent fiscal impulses.     

Modelling the effect of the geographical environment on Islamic banking performance: A panel quantile regression analysis

While studies have focused on Islamic banking, research on the effect of the geographical environment on Islamic banks is scarce. We investigate this issue by using daily data on 12 Islamic banks in four regions (Africa, Asia, Europe, and the United States) from July 2007 to April 2016. We apply different methodological approaches (principal component analysis, panel data tests, and quantile regression). First, the principal component analysis shows that the performance of Islamic banks varies among regions. Second, the linear panel regression highlights that the geographical environment positively and significantly affects Islamic banking, suggesting the importance of externality effects. Finally, the environmental effect seems to vary with quantiles (positive effect for the lowest quantile versus negative effect for the highest quantile). This quantile specification points to nonlinearity in the environment–Islamic bank performance relationship, reflecting a time-varying discipline imposed by the Sharia board (Islamic Law). This finding helps better explain the main difference between Islamic banks in the East (Africa and Asia) and those in the West (Europe and the United States) and also enables investors to adjust their portfolio choices when considering the products of Islamic banks according to regional specificities.     

A Poisson ridge regression estimator

The standard statistical method for analyzing count data is the Poisson regression model, which is usually estimated using maximum likelihood (ML) method. The ML method is very sensitive to multicollinearity. Therefore, we present a new Poisson ridge regression estimator (PRR) as a remedy to the problem of instability of the traditional ML method. To investigate the performance of the PRR and the traditional ML approaches for estimating the parameters of the Poisson regression model, we calculate the mean squared error (MSE) using Monte Carlo simulations. The result from the simulation study shows that the PRR method outperforms the traditional ML estimator in all of the different situations evaluated in this paper.     

Model selection for nonlinear time series

We investigate the finite-sample performance of model selection criteria for local linear regression by simulation. Similarly to linear regression, the penalization term depends on the number of parameters of the model. In the context of nonparametric regression, we use a suitable quantity to account for the Equivalent Number of Parameters as previously suggested in the literature. We consider the following criteria: Rice T, FPE, AIC, Corrected AIC and GCV. To make results comparable with other data-driven selection criteria we consider also Leave-Out CV. We show that the properties of the penalization schemes are very different for some linear and nonlinear models. Finally, we set up a goodness-of-fit test for linearity based on bootstrap methods. The test has correct size and very high power against the alternatives investigated. Application of the methods proposed to macroeconomic and financial time series shows that there is evidence of nonlinearity.First version received: September 2002/Final version received : October 2003I would like to thank Cees Diks, Cars Hommes and an anonymous referee for useful comments that significantly improved the paper.     

Spurious regressions between stationary generalized long memory processes

This paper shows that a spurious regression can occur between two stationary generalized fractional processes, as long as their generalized fractional differencing parameters sum up to a value greater than 0.5 and their spectral densities have poles at the same location. This theoretical finding is supported by simulations.     

The sensitivity of the RESET tests to disturbance autocorrelation in regression analysis

This paper investigates the sensitivity of the RESET tests, proposed by Ramsey (1969) and modified by Thursby and Schmidt (1977), to disturbance autocorrelation in regression analysis. Porter and Kashyap (1984) show that RESET is not robust to autocorrelated disturbance when there is a highly autocorrelated regressor in the model. We show that RESET is sensitive to disturbance autocorrelation even when the regressors are not autocorrelated. We explain the findings of Thursby (1979) and Porter and Kashyap (1984) as well as our result by showing that a spurious correlation between the regressor and the disturbance is responsible for the serious size distortion of the RESET tests. First version received: June 1999/Final version received: November 2000     

A Bayesian note on competing correlation structures in the dynamic linear regression model

A Bayesian posterior odds approach is used to distinguish between different error correlation structures in dynamic linear regression models. Recent classical results are provided with a Bayesian interpretation, and a small empirical example illustrates the approach.     

On the nonlinear relationship between inflation and economic growth

Using a large panel data set from both developed and developing economies and employing the PSTR and dynamic GMM techniques, this study highlights two aspects of the inflation–growth relationship. First, it analyzes the nonlinearity of the relationship and identifies several thresholds for the global sample and for various income-specific sub-samples. Second, it identifies some country-based macroeconomic features that influence this nonlinearity. Our empirical results substantiate both views and validate the fact that inflation–growth nonlinearity is sensitive to a country׳s level of financial development, capital accumulation, trade openness and government expenditures. Moreover, these country-specific characteristics result in some marked differences in this nonlinear relationship.     

Does the Reserve Bank of Australia follow a forward-looking nonlinear monetary policy rule?

ABSTRACT

The conventional monetary policy rule describes a simple linear relationship between the domestic interest rate, inflation rate and output gap. An important extension to this rule is to incorporate the forward-looking behaviour of central banks, where it is assumed that they target an expected level of inflation instead of its current realised value. Using quarterly observations for the period 1993:1-2018:2, this paper investigates whether the conduct of monetary policy in Australia can be described by a forward-looking linear monetary policy rule, or by a nonlinear forward-looking monetary policy rule. In particular, the nonlinear forward-looking monetary policy rule is analysed in a regime-switching framework using a smooth logistic transition regression model. While the results show that the conventional forward-looking linear monetary policy rule describes the application of monetary policy in Australia reasonably well, the interest rate setting behaviour of the RBA is best described by a nonlinear forward-looking monetary policy rule.     

A note on income tax evasion,labor supply,and nonlinear tax schedules

The existing literature on income tax evasion lays claim to two important implications: one is that higher tax rates induce greater income declarations and the other is that increases in gross income induce a fall in the fraction of income declared. The purpose of this paper is to examine the robustness of these results with respect to three modifications of the standard model. The first involves relaxing the assumption of linear income tax schedules. The second modification is to consider an alternative form for the penalty function. Thirdly, we consider the consequences of the tax payer making a joint hours of work-income declaration decision.     

A quantile regression approach for estimating panel data models using instrumental variables

We introduce a quantile regression approach to panel data models with endogenous variables and individual effects correlated with the independent variables. We find newly developed quantile regression methods can be easily adapted to estimate this class of models efficiently.     

Quantile regression for duration data: A reappraisal of the Pennsylvania Reemployment Bonus Experiments

We argue that quantile regression methods can play a constructive role in the analysis of duration (survival) data offering a more flexible, more complete analysis than is typically available with more conventional methods. We illustrate the approach with a reanalysis of the data from the Pennsylvania Reemployment Bonus Experiments. These experiments, conducted in 1988–89, were designed to test the efficacy of cash bonuses paid for early reemployment in shortening the length of insured unemployment spells     

Revisiting growth empirics based on IV panel quantile regression

We analyse the well-known issue of economic growth convergence using quantile regression. Most previous studies have used a least squares (LS) method or variation, which focuses on the issue only at the mean of the growth rate. Therefore, such results cannot provide a satisfactory answer to what can happen if the growth rate is far from the conditional mean level. For example, we consider the following question: do we still have economic growth convergence or is the convergence speed changed in a low growth period such as the ‘Great Recession,’ that started in 2008? We propose using instrumental variable panel quantile regression to answer this question. Our empirical findings demonstrate that economic growth convergence occurs at all quantiles over the entire conditional distribution, but that the convergence speed does depend on quantiles; the convergence speed is much higher when the GDP growth rate is at either high or low quantiles.     

A note on S in a spatially correlated error components regression model for panel data

The ordinary least squares based estimator of the disturbance variance in a panel regression model with spatially correlated error component is shown to be asymptotically unbiased and weakly consistent without any restrictions on the regressor matrix.     

A new Cramer-Von Misses cointegration test with application to environmental Kuznets curve

This article introduces a new Cramer-Von Misses (CVM) cointegration test robust to nonlinearities. We characterize nonlinear cointegration in terms of a nonlinear moving-average filter (high pass filter) of a matrix based on permutation matrices on the discrepancy of empirical distributions. A Cramer-Von Misses (CVM) test statistic is proposed for testing the null hypothesis of two independent random walks against a broad range of cointegrating alternatives with monotonic nonlinearities and level shifts in the cointegration relationship. We derive the asymptotic distribution of this induced-order Cramer-Von Misses (CVM) cointegration test. This new non-parametric test statistic has two important properties: the invariance to monotonic transformations of the series and the robustness for the presence of several parameter shifts or structural changes. We analyse the small sample properties of this test by Monte Carlo simulations and evaluate the power of the test. Finally, this CVM test is applied to the analysis of long run environmental Kuznets curve which relates economic growth and pollution. In particular, we consider a nonlinear cointegration between gross domestic product (GDP) and CO2 emissions. Our new CVM test is able to find evidence of cointegration while classical single equation cointegration tests are not.     

Comparing linear and nonlinear forecasts for stock returns

We compare the out-of-sample performance of monthly returns forecasts for two indices, namely the Dow Jones (DJ) and the Financial Times (FT) indices. A linear and a nonlinear artificial neural network (ANN) model are used to generate the out-of-sample competing forecasts for monthly returns. Stationary transformations of dividends and trading volume are considered as fundamental explanatory variables in the linear model and the input variables in the ANN model. The comparison of out-of-sample forecasts is done on the basis of forecast accuracy, using the Diebold and Mariano test [J. Bus. Econ. Stat. 13 (1995) 253.], and forecast encompassing, using the Clements and Hendry approach [J. Forecast. 5 (1998) 559.]. The results suggest that the out-of-sample ANN forecasts are significantly more accurate than linear forecasts of both indices. Furthermore, the ANN forecasts can explain the forecast errors of the linear model for both indices, while the linear model cannot explain the forecast errors of the ANN in either of the two indices. Overall, the results indicate that the inclusion of nonlinear terms in the relation between stock returns and fundamentals is important in out-of-sample forecasting. This conclusion is consistent with the view that the relation between stock returns and fundamentals is nonlinear.