Understanding machine learning-based forecasting methods: A decomposition framework and research opportunities

Machine learning (ML) methods are gaining popularity in the forecasting field, as they have shown strong empirical performance in the recent M4 and M5 competitions, as well as in several Kaggle competitions. However, understanding why and how these methods work well for forecasting is still at a very early stage, partly due to their complexity. In this paper, I present a framework for regression-based ML that provides researchers with a common language and abstraction to aid in their study. To demonstrate the utility of the framework, I show how it can be used to map and compare ML methods used in the M5 Uncertainty competition. I then describe how the framework can be used together with ablation testing to systematically study their performance. Lastly, I use the framework to provide an overview of the solution space in regression-based ML forecasting, identifying areas for further research.     

企业投资效率演进与中国总产出波动

本文将企业投资效率的变动过程引入包含名义和实际刚性等多种结构特征的新凯恩斯主义宏观模型,并用贝叶斯极大似然方法估计了模型的结构参数。边际数据密度比较结果显示,刚性价格-工资模型能够较好的解释实际数据。同时,我们还就企业投资效率冲击对实际总产出波动的解释力和历史贡献进行了评估,波动性分解的结果说明投资效率冲击对产出波动具有30%以上的解释力,是所有冲击中贡献最大的。历史分解和反事实模拟的结果显示投资效率是动态演化的,在本届政府的两个任期内,投资效率呈现总体改善趋势,相对于十五计划时期,十一五计划时期的投资效率更稳定、效率水平也更高。投资效率改善对2006年后产出增长有显著拉动作用,需要强调的是,如果没有投资效率改善所形成的产出拉动,2008年金融危机所导致的经济衰退要比实际情形严重很多倍。     

Gender Employment Disparities,Financialization, and Profitability Dynamics on the Eve of Italy's Post-2008 Crisis

This study explores aggregate profitability in Italy from 1994 to 2008 in its connection with structural change and gender employment disparities. Using decomposition analysis, the study finds that aggregate profit rate declined, but the profit share did not. Male variables – such as earnings, output, employment, and working hours – tended to have more weight than female ones in explaining aggregate outcomes. Structural change also played a major role, as the economy specialized in sectors with falling real wages and wage shares, especially the financial sector and construction. Further falls in the wage share and widening wage gaps may not guarantee a rise in profitability. This result entails that “neo-mercantilist” approaches to solve the crisis might only prepare the next crisis, while a coordinated expansion of demand could be more successful. Moreover, gender issues should not be neglected in terms of favoring women's employment and entrepreneurship.     

持久收入假说与我国城镇居民消费——基于省级面板数据的实证分析

运用基于持久收入和暂时收入的两种分解方法的省际面板模型检验持久收入假说的三个假设.结果发现,我国城镇居民消费与持久收入和暂时收入均存在显著的相关关系,暂时收入的边际消费倾向不为零;就诸省份而言,暂时收入的边际消费倾向不同.进一步探讨了持久收入假说不能完全解释我国城镇居民消费行为的原因."λ假说"检验表明,城镇居民消费对当期收入过度敏感,其呈现理性化趋势.     

宏观经济运行质量评价指标的选择方法

筛选指标的传统方法或者是定性分析或者是单一的定量分析。本文收集了37个反映宏观经济运行质量的指标信息,依据相关系数及其聚类分析,将相关程度不同的指标聚为不同的组或类,然后在相关系数高度相关的指标组进行因果关系分析,选取那些作为原因的变量或指标。利用相关时间序列建模方法中的向量自回归模型,及其脉冲响应函数和方差分解分析,进一步区分那些影响其他变量的重要变量。最后得到一组既能反映现象变化,又是主要或重要变量的14个指标。这一指标筛选方法不同于传统方法,它充分利用了指标之间的关联信息、因果信息和动态相依信息,具有综合性与实用性。     

Wage differences between Austrian men and women: <Emphasis Type="Italic">semper idem?</Emphasis>

In most OECD countries, the wage gap between men and women has narrowed during the past two decades. Developments of the last 20 years, e.g., increased labour market attachment of women or the introduction of equal pay laws, may have reduced the gender wage gap. We investigate the extent, persistence, and socio-economic determinants of the gender wage gap in Austria, for the years 1983 and 1997. Using wage decomposition techniques, we find that the average gender wage gap was almost as high in 1997 as it was in 1983. Not accounting for differences, the gender wage gap dropped from 25.5 to 23.3% of men’s wages. Taking observable differences between men and women into account, we estimate that the mean gender wage gap that cannot be explained, i.e., discrimination against women, dropped from 17 to 14% of men’s wages. A decomposition of the gender wage gap over time indicates that both returns to human capital and less discrimination were responsible for the narrowing of the gender wage gap.

城镇劳动力市场雇佣关系的演化及影响因素

本文利用跨度为二十年(1988年到2007年)的微观家计数据,分析了我国城镇地区雇佣关系在经济转型时期从单一固定工向多元灵活雇佣关系转变的过程。将我国城镇地区的雇佣关系分成四种类型,探讨了雇佣关系构成从1988年到2007年的变化及其影响因素,以及各种因素的相对重要性。分析发现,从1988年开始,我国的雇佣关系构成中灵活雇佣比重开始逐步上升,到2002年,短期雇佣、无合同雇佣等灵活雇佣形式的比重达到一个较高水平,进入2007年,稳定雇佣关系的比重略有回升。扩展的Oaxaca-Blinder分解所得结果表明,从1995年到2002年,劳动力市场灵活性增强主要受到转型过程中制度变迁等不可观测因素的影响;进入2007年,稳定的雇佣关系比重略有回升,放松管制和市场深化的效应逐渐释放殆尽,特征变量对提高雇佣稳定性的作用开始显现。     

Internet Job Search and Labor Market Outcome

This study examines the effect of the use of internet job search (IJS) on individual wages. The data utilized in the study are obtained from the Internet and Computer Use Dictionary in the Current Population Survey of September 2001 and October 2003. An Oaxaca Decomposition is used to examine the extent to which wage differences are influenced by IJS. The results show that accessibility to the internet is a crucial factor in the decision of an individual to utilize the internet for job search activities. However, no clear evidence exists that IJS increases individual wages. Additionally, the study also demonstrates that IJS is subject to self-selection bias. Thus, failure to control for self-selection bias results in a very serious bias in estimation.     

我国城镇正规就业与非正规就业工资差异的实证研究——基于分位数回归与分解的发现

本文利用2009年CHNS数据对我国城镇正规就业与非正规就业的工资差异进行实证研究,分位数回归与分解的结果表明:正规就业与非正规就业的教育回报率的差异,随工资分布由低端到高端呈现先升后降趋势,经验—工资线在正规就业与非正规就业明显不同,前者为单调递增的线性关系,后者为倒"U"形曲线关系;正规就业与非正规就业的工资差异主要是由中低端的工资差异引起的;在工资分布中低端和歧视等非市场因素是工资差异的主要原因,而在工资分布高端,工资差异主要来自于教育和经验等个人禀赋差异。