Prediction of the Indian summer monsoon using a stacked autoencoder and ensemble regression model

The study of climatic variables that govern the Indian summer monsoon has been widely explored. In this work, we use a non-linear deep learning-based feature reduction scheme for the discovery of skilful predictors for monsoon rainfall with climatic variables from various regions of the globe. We use a stacked autoencoder network along with two advanced machine learning techniques to forecast the Indian summer monsoon. We show that the predictors such as the sea surface temperature and zonal wind can predict the Indian summer monsoon one month ahead, whereas the sea level pressure can predict ten months before the season. Further, we also show that the predictors derived from a combination of climatic variables can outperform the predictors derived from an individual variable. The stacked autoencoder model with combined predictors of sea surface temperature and sea level pressure can predict the monsoon (June-September) two months ahead with a 2.8% error. The accuracy of the identified predictors is found to be superior to the state-of-the-art predictions of the Indian monsoon.     

Public-private sector wage differentials in Germany: Evidence from quantile regression

This paper measures and decomposes the differences in earnings distributions between public sector and private sector employees in Germany for the years 1984–2001. Oaxaca decomposition results suggest that conditional wages are higher in the public sector for women but lower for men. Using the quantile regression decomposition technique proposed by Machado and Mata (2004), we find that the conditional distribution of wages is more compressed in the public sector. At the low end of wages, differences in characteristics explain less than the raw wage gap when it is the opposite at high wages. Separate analyses by work experience and educational groups reveal that the most experienced employees and those with basic schooling do best in the public sector. All these results are stable over the 80s and 90s.I thank Michael Lechner and the Centre for European Economic Research (ZEW), Mannheim, for letting me work with the full sample of the German Socio-Economic Panel (GSOEP). I am grateful to an anonymous referee and particularly to the editor, Bernd Fitzenberger, for providing me detail comments and suggestions that have significantly improved the paper. I have also benefited from discussions with Michael Lechner and Ruth Miquel.First version received: April 2002/Final version received: June 2004     

Does Gibrat's Law hold among young,small firms?

According to Gibrat's Law of Proportionate Effect, the growth rate of a given firm is independent of its size at the beginning of the examined period. Aimed at extending this line of investigation, the present paper uses quantile regression techniques to test whether Gibrat's Law holds for new entrants in a given industry: that is for new small firms in the early stage of their life cycle. The main finding is that for some selected industries in Italian manufacturing Gibrat's Law fails to hold in the years immediately following start-up, when smaller firms have to rush in order to achieve a size large enough to enhance their likelihood of survival. Conversely, in subsequent years the patterns of growth of new smaller firms do not differ significantly from those of larger entrants, and the Law therefore cannot be rejected.JEL Classification: L11, L60Previous versions of this paper were presented at the 27th Annual EARIE Conference (Lausanne, 7-10 September 2000) and at seminars held between 2000 and 2003 at the Economics Department of Harvard University, the Catholic University of Milan, the University of Ferrara, the University of Bologna, the Bank of Italy, and Athens University of Economics & Business. We would like to thank Carlo Bianchi, Giuseppe Colangelo, Giovanni Dosi, Steven Klepper (Editor), Stephen Martin, Ariel Pakes, Aman Ullah and, in particular, Helen Louri and two anonymous referees for their valuable comments on earlier drafts. Financial support from MIUR (Year 2000; protocol #MM13038538_001; project leader: E. Santarelli) is gratefully acknowledged.Correspondence to: E. Santarelly     

Distributional assumptions and a test of the dual labor market hypothesis

Recent application of the switching regression model to allocate workers into the primary and secondary labor markets is considered to be the best solution to the classification problem of the empirical testing of the dual labor market theory. In such models, normality of the error terms is assumed. This paper adopts the switching regression model to test the dual labor market theory by assuming different distributions of the error terms. The test results strongly support the dual labor market theory regardless of the assumption one makes about the error terms. However, the results indicate that different distribution can lead to different percentage distributions of workers in the two segments. In particular, the normal distribution generates more workers in the primary segment than the non-normal distributions. Therefore, care must be taken not to generalize the type of industries or occupations that fall under the primary and secondary segments. First version received: October 2000/Final version received: March 2002 RID="*" ID="*"  I would like to thank Kevin Lang, Robert Marshall, and two anonymous referees for their helpful comments. I am grateful for comments received from the session participants of the Western Economic Association International Conference, San Francisco, CA, June 28–July 2, 1996, and the Midwest Economic Association Conference, Kansas City, 1997. I thank George Bonney, the Chief Statistician of Fox Chase Cancer Center, Philadelphia for his comments. Any remaining errors are my responsibility. I gratefully acknowledge financial support from Penn State Research and Development Grant, 1995.     

Asymptotic behaviour of regression pre-test estimators with minimal Bayes risk

Pre-test estimators (PTE) are considered which are optimal under a Bayes risk among PTE with general measurable sets as “regions of significance” for the test statistic t associated with the estimate of a given regression coefficient. Asymptotic and some finite sample results are stated and numerical experiments are commented on.     

Simulation driven inferences for multiply imputed longitudinal datasets*

In this article, we demonstrate by simulations that rich imputation models for incomplete longitudinal datasets produce more calibrated estimates in terms of reduced bias and higher coverage rates without duly deflating the efficiency. We argue that the use of supplementary variables that are thought to be potential causes or correlates of missingness or outcomes in the imputation process may lead to better inferential results in comparison to simpler imputation models. The liberal use of these variables is recommended as opposed to the conservative strategy.     

基于高频数据的深圳基金市场日内周期性研究

本文利用深市基金指数高频数据,采用Anderson和Bollerslev(1997)提出的弹性傅立叶回归(FlexibleFourierFormregression,即FFF回归)方法首次对深市基金市场进行了日内周期性的研究。通过对高频收益的定性分析,发现基金市场具有同股票市场相似的周期性,并对这一周期性进行了初步的理论解释。通过FFF方法,将该周期因子进行滤波处理以后,基金指数高频绝对收益不再具有明显周期性。FFF回归能较好地确定日内周期因子。     

上市公司审计收费影响因素的实证研究

本文利用我国深沪两市A股上市公司2003年年报中披露的数据,对可能影响我国上市公司年度审计费用的主要因素,运用多元线性回归进行实证分析。研究结果发现,上市公司资产总额、纳入合并报表的控股子公司数目、其他应收款占总资产的比例、公司和事务所是否位于经济发达地区和董事会规模与年度审计费用显著正相关,而独立董事人数与年度审计费用显著负相关;应收账款占总资产的比例、存货占总资产的比例、审计意见、当年和前一年度的盈利状况等因素对年度审计费用的影响并不显著。因此,我国上市公司审计费用主要由审计的产品费用所决定,从而会计师事务所的审计收费主要反映了现实的审计成本,而审计风险等潜在风险成本并没有在审计收费中体现出来。     

Three Corrected Score Tests for Generalized Linear Models with Dispersion Covariates

We develop three corrected score tests for generalized linear models with dispersion covariates, thus generalizing the results of Cordeiro , Ferrari and Paula (1993) and Cribari-Neto and Ferrari (1995) . We present, in matrix notation, general formulae for the coefficients which define the corrected statistics. The formulae only require simple operations on matrices and can be used to obtain analytically closed-form corrections for score test statistics in a variety of special generalized linear models with dispersion covariates. They also have advantages for numerical purposes since our formulae are readily computable using a language supporting numerical linear algebra. Two examples, namely, iid sampling without covariates on the mean or dispersion parameter oand one-way classification models, are given. We also present some simulations where the three corrected tests perform better than the usual score test, the likelihood ratio test and its Bartlett corrected version. Finally, we present a numerical example for a data set discussed by Simonoff and Tsai (1994) .     

Do investors in Green Bond market pay a premium? Global evidence

We examine the pricing difference of Green Bonds (GB) and conventional bonds (CBs) in capital markets worldwide. Credit spread is used to observe whether investors would like to pay a premium for GBs over CBs. This study uses panel data regression with hybrid model to analyse daily observations over the period 2016 to 2017. We employ Option-Adjusted spread (OAS) to measure the credit spreads of bonds while controlling for bond specific, macroeconomic and global factors that influence the spread. With the hybrid model used in the panel data analysis, we were able to capture the fixed-effects of variables in a random effect model. We find that GBs are traded at a premium of 63 basis points (BPS), compared with a comparable corporate bond issue. We find that the green label provides issuers an incentive to raise funds through issuing GBs while providing investors an opportunity to diversify their investments returns. Our findings provide several implications to the major stakeholders driving the GB market to scale up the market to finance the required level of global green investment needs. We stress an urgent need to support the growth of the GB market to achieve sustainable development through mitigating climate change challenges.

Abbreviation GB: Green Bond; CB: Conventional Bond; YS: Yield Spread; BPS: Basis Points; OAS: Option-Adjusted Spread; PCSE: Panels Corrected Standard Errors; CPI: Consumer Price Index; GBPs: Green Bond Principles; CBS: Climate Bond Standard