【经济学的机器学习工具】Machine-learning Techniques in Economics New Tools

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

《Machine-learning Techniques in Economics New Tools for Predicting Economic Growth》
In this book, we develop a Machine Learning framework to predict economic growth and the likelihood of recessions. In such a framework, different algorithms are trained to identify an internally validated set of correlates of a particular target within a training sample. These algorithms are then validated in a test sample.
Why does this matter for predicting growth and business cycles, or for predicting other economic phenomena? In the rest of this chapter, we discuss how Machine Learning methodologies are useful to economics in general, and to predicting growth and recessions in particular. In fact, the social sciences are increasingly using these techniques for precisely the reasons we outline. While Machine Learning itself is not a new idea, advances in computing technology combined with a recognition of its applicability to economic questions make it a new tool for economists (Varian 2014). Machine Learning techniques present easily interpretable results particularly helpful to policy makers in ways not possible with the standard sophisticated econometric techniques. Moreover, these methodologies come with powerful validation criteria that give both researchers and policy makers a nuanced sense of confidence in understanding economic phenomenon.
As far as we know, such an undertaking has not been attempted as comprehensively as here. Thus, we present a new path for future researchers interested in using these techniques. Our findings should be interesting to readers who simply want to know the power and limitations of the Machine Learning framework. They should also be useful in that our techniques highlight what we do know about growth and recessions, what we need to know, and how much of this knowledge is dependable.
Our starting point is Xavier Sala-i-Martin’s (1997) paper wherein he summarizes an extensive literature on economic growth by choosing theoretically and empirically ordained covariates of economic growth. He identifies a robust correlation between economic growth and certain variables, and divides these “universal” correlates into nine categories. These categories are as follows:
1.
Geography. For example, absolute latitude (distance from the equator) is negatively correlated with growth, and certain regions, such as sub-Saharan Africa and Latin America underperform, on average.
2.
Political institutions. Measures of institutional quality like strong Rule of Law, Political Rights, and Civil Liberties improve growth, while instability measures like Number of Revolutions and Military Coups and War impede growth.
3.
Religion. Predominantly Confucianist/Buddhist and Muslim countries grow faster, while Protestant and Catholic grow more slowly.
4.
Market distortions and market performance. For example, Real Exchange Rate Distortions and Standard Deviation of the Black Market Premium correlate negatively with growth.
5.
Investment and its composition. Equipment Investment and Non-Equipment Investment are both positively correlated with growth.
6.
Dependence on primary products. Fraction of Primary Products in Total Exports are negatively correlated with growth, while the Fraction of Gross Domestic Product in Mining is positively correlated with growth.
7.
Trade. A country’s Openness to Trade increases growth.
8.
Market orientation. A country’s Degree of Capitalism increases growth.
9.
Colonial History. Former Spanish Colonies grow more slowly.
Sala-i-Martin’s findings are standard in the growth literature. His econometric techniques cull the immense proliferation of explanatory variables into a tractable and parsimonious list. However, there are several problems with his approach that in turn hint at fundamental gaps in our understanding of the economic growth process. The Machine Learning framework can fill precisely these kinds of gaps in evidence.
The findings of the standard econometric techniques deployed by Sala-i-Martin cannot say anything about why certain variables matter, or which matter more than others. For example, if a country’s GDP has a large , it is likely to be a growth laggard, though if it has a high Fraction of GDP in Mining, it is in the high growth category. This sort of contradiction suggests that maybe the Sala-i-Martin list is not parsimonious enough. It is certainly not always amenable to consistent theoretical explanations.
In our treatment, we start with a set of variables and dataset that largely mirrors Sala-i-Martin’s comprehensive list of (what he identifies as) robust correlates of economic growth. Next, we randomly pick a set of countries to divide the data set into a learning sample (70% of the data) and a test sample (30% of the data). We use multiple Machine Learning algorithms to find the algorithm with the best out-of-sample fit. We then identify the variables that contribute the most to this out-of-sample fit. Thus, the algorithms can rank variables according to their relative ability to predict the target variable. We can thus whittle down the correlates of growth identified by Sala-i-Martin to the ones that robustly contribute to prediction. Thus, we are able to identify those variables that best predict growth and recessions 5 years out, without any of the inherent contradictions outlined above.
In our analysis, a country in a particular year is the observational unit. We structure the data so that the target (growth or recession) is 5 years out. For example, the first period contains covariates for 1971–1975, while the target is growth, or an incidence of recession, in the 1976–1980 period. Looking at growth in 5-year periods is standard in the literature. However, choosing the dependent variable or target 5-years out is, to our knowledge, new in the literature. This data structure is therefore our first innovation toward developing a truly predictive model. Our targets are economic growth and recessions.
We also report the marginal effect of these variables on economic growth and recessions through partial dependence plots or PDPs. The PDPs provideinsights on the pathways of economic growth. They tell us how changing a variable affects the target over the range of that change. Thus, we are able to say (with some sense of the confidence that comes from estimates of predictive accuracy) whether, over a certain range, a particular variable has a greater or lesser effect on growth, whether it affects growth negatively or positively, as well as identify other ranges where the variable does not affect growth. Thus, if we find that Investment is an important predictor of growth, the PDP shows us how an increase in investment affects growth over the range of that increase. In fact, we find that the covariates of growth affect growth in consistently non-linear ways. A parametric point estimate cannot capture this non-linearity. The information in PDPs is particularly useful to policy makers, when, for instance, it comes to understanding how countries with different levels of investment may respond differently to changes in a policy lever. It also has implications for the process of developing theoretical models of growth in that these models need to take into account these non-linearities.
The growth literature’s focus on growth accounting and regressions, and therefore on the correlates of growth, ends up generating long lists of possible correlates of growth. Such lists hamper standard econometric techniques since they are plagued by a number of problems—parameter heterogeneity, model uncertainty, the existence of outliers, endogeneity, measurement error, and error correlation (Temple 1999), to name a few. In the following chapters, we suggest that Machine Learning can help circumvent some of these problems. Thus, Machine Learning methodologies that create parsimonious lists of the covariates of growth that are validated by out-of-sample fit can be particularly useful in the growth literature. They can complement current econometric methodologies, and, at the same time, they can offer fresh insights into economic growth.



++++++++++++++++论坛里2025-2026年新书如下+++++++++++++++++
【气候金融研究资料】Climate Finance in the Net-Zero Transition  https://bbs.pinggu.org/thread-16690755-1-1.html
【经济分析共轭对偶研究资料】Conjugate Duality in Economic Analysis  https://bbs.pinggu.org/thread-16689103-1-1.html
【旅游经济研究】The Economics of Tourism Destinations Theory and Practice https://bbs.pinggu.org/thread-16689101-1-1.html
【金融大模型研究资料】2026 Large Language Models in Finance   https://bbs.pinggu.org/thread-16686488-1-1.html
【金融会计AI资料】Python for Accounting and Finance A Mind-Mapping Approach https://bbs.pinggu.org/thread-16686486-1-1.html
【行为经济学研究】Behavioral Economics https://bbs.pinggu.org/thread-16682572-1-1.html
【行为经济研究分析资料】Social and Economic Behavior of AI Agents Theory https://bbs.pinggu.org/thread-16682564-1-1.html
【人工智能会计资料】AI_in_Accounting https://bbs.pinggu.org/thread-16675417-1-1.html
【金融智能资料】The Social and Economic Behavior of AI Agents Theory https://bbs.pinggu.org/thread-16675404-1-1.html
【金融商务研究资料】Decoding AI Unleashing the Future of Business and Finance https://bbs.pinggu.org/thread-16675388-1-1.html
【金融计量时间序列分析】Econometrics, Finance, and Time Series Analysis https://bbs.pinggu.org/thread-16672146-1-1.html
【贝叶斯 计量经济学】Bayesian Econometrics and Their Applications https://bbs.pinggu.org/thread-16672144-1-1.html
【金融估值与计量经济学】Financial Valuation and Econometrics https://bbs.pinggu.org/thread-16672143-1-1.html
【机器学习经济】Business Data Science: Machine Learning and Economics to Optimi https://bbs.pinggu.org/thread-16672019-1-1.html
【经管研究方法】Modern Machine Intelligence Approach for Financial and Economic https://bbs.pinggu.org/thread-16671989-1-1.html
【行为金融学资料】Behavioral Finance Theory and Application https://bbs.pinggu.org/thread-16670911-1-1.html
【金融分析提示词】AI Prompts Financial Analysis 100+ Practical Prompts Samples  https://bbs.pinggu.org/thread-16665733-1-1.html
【金融资料】Lectures on the Theory and Application of Modern Finance with R CGPT https://bbs.pinggu.org/thread-16665722-1-1.html
【金融经济数据处理】Introduction to Python for Quantitative Finance Productivity https://bbs.pinggu.org/thread-16665695-1-1.html
【金融编程参考资料】Quantitative Finance with Case Studies in Python https://bbs.pinggu.org/thread-16641019-1-1.html
【因果推断与机器学习】Causal Inference and Machine Learning  https://bbs.pinggu.org/thread-16619975-1-1.html
【因果机器学习与AI】Applied Causal Inference Powered by ML and AI  https://bbs.pinggu.org/thread-16584734-1-1.html
【英文经济学资料】Economic Analysis Through Mathematics Tools and Techniques https://bbs.pinggu.org/thread-16578330-1-1.html
【英文人工智能资料】Integrating Artificial Intelligence (ChatGPT) into Marketing https://bbs.pinggu.org/thread-16578273-1-1.html
【英文经济资料】Dynamic Modeling and Econometrics in Economics and Finance https://bbs.pinggu.org/thread-16578258-1-1.html
【英文经济资料】Economics of the Energy Crisis Environment, Policy and Security  https://bbs.pinggu.org/thread-16578252-1-1.html
【因果机器学习与AI】Applied Causal Inference Powered by ML and AI https://bbs.pinggu.org/thread-16584734-1-1.html
【英文经济学资料】Economic Analysis Through Mathematics Tools and Techniques https://bbs.pinggu.org/thread-16578330-1-1.html
【英文人工智能资料】Integrating Artificial Intelligence (ChatGPT) into Marketing  https://bbs.pinggu.org/thread-16578273-1-1.html
【英文经济资料】Dynamic Modeling and Econometrics in Economics and Finance https://bbs.pinggu.org/thread-16578258-1-1.html
【英文经济资料】Economics of the Energy Crisis Environment, Policy and Security https://bbs.pinggu.org/thread-16578252-1-1.html
【金融科技大模型资料】Finance and Large Language Models  https://bbs.pinggu.org/thread-16554562-1-1.html
【大模型金融科技资料合集】Large Language Models Ops for Finance https://bbs.pinggu.org/thread-16554546-1-1.html
【气候金融研究资料】Financing Climate Action India in a Global Context https://bbs.pinggu.org/thread-16554489-1-1.html
【智能金融科技资料】AI For the Finance Professionals https://bbs.pinggu.org/thread-16554467-1-1.html
【英文金融科技资料】 Deep Learning in Banking Integrating AI for Next-Generation https://bbs.pinggu.org/thread-16534089-1-1.html
【英文金融研究资料】Empirical Finance（实证金融） https://bbs.pinggu.org/thread-16535447-1-1.html
【最新公司金融资料】Corporate Finance（公司理财） https://bbs.pinggu.org/thread-16536256-1-1.html
【英文经济学资料】Ethical Economics And Sustainable Development The Role of Mora https://bbs.pinggu.org/thread-16532745-1-1.html
【英文行为金融学资料】Behavioral Finance and Asset Prices: Influence of Emotions https://bbs.pinggu.org/thread-16533070-1-1.html
【最新英文资料】2026 Data Science in Finance and Accounting https://bbs.pinggu.org/thread-16530558-1-1.html
【最新英文资料】2026 Foundations of Artificial Intelligence  https://bbs.pinggu.org/thread-16530566-1-1.html
【最新英文资料】2026 Sustainable Digital Finance  https://bbs.pinggu.org/thread-16530574-1-1.html
【最新英文资料】2026 Signature Methods in Finance https://bbs.pinggu.org/thread-16530594-1-1.html
【最新英文资料】Generative Artificial Intelligence A Law and Economics Approach https://bbs.pinggu.org/thread-16530608-1-1.html
【最新英文资料】2026 Mastering Financial Markets with Python New Horizons  https://bbs.pinggu.org/thread-16530671-1-1.html
【最新英文资料】2026 Artificial Intelligence in the Digital Era Economic, Legisl https://bbs.pinggu.org/thread-16530675-1-1.html
【最新英文量化金融资料】Quantitative Finance with Case Studies in Python https://bbs.pinggu.org/thread-16531139-1-1.html
【最新英文量化金融资料】Quantitative Finance An Introduction to Investments  https://bbs.pinggu.org/thread-16531269-1-1.html
【最新劳动经济学资料】Economics, Philosophy and the Neglect of Labour https://bbs.pinggu.org/thread-16532435-1-1.html
【最新经济学研究资料】The Economics of Immigration https://bbs.pinggu.org/thread-16532536-1-1.html
【最新英文经济研究资料】新全球经济秩序The New Global Economic Order  https://bbs.pinggu.org/thread-16532720-1-1.html
【英文AI金融资料】Ultimate FINGPT for Financial Analysis: Build, Train, and so https://bbs.pinggu.org/thread-16528627-1-1.html
【英文经济学资料】Theories and Models in Economics An Empirical Approach to Meth https://bbs.pinggu.org/thread-16528619-1-1.html
【英文金融科技算法资料】Generative AI in FinTech Revolutionizing Finance https://bbs.pinggu.org/thread-16520982-1-1.html
【英文经管资料】Research and Management Quantitative Methods in Business and Eco https://bbs.pinggu.org/thread-16520930-1-1.html
【英文管理学研究资料】AI for Qualitative Research A Hands-On Guide for Manage https://bbs.pinggu.org/thread-16520910-1-1.html
【英文金融科技资料】Advanced Digital Technologies in Financial and Business M https://bbs.pinggu.org/thread-16518009-1-1.html
【英文数字经济类资料】Data-Driven Modelling and Predictive Analytics in B and  https://bbs.pinggu.org/thread-16517990-1-1.html
【英文商务智能资料】AI, Machine Learning and IoT for Smart Business Management https://bbs.pinggu.org/thread-16517975-1-1.html
【英文计量经济学资料】Machine Learning for Econometrics and Related Topics https://bbs.pinggu.org/thread-16517965-1-1.html
【英文计量经济学资料】Econometrics with Machine Learning https://bbs.pinggu.org/thread-16517909-1-1.html
【英文环境经济学资料】Economics of the Environment: Theories, Policies, and Prac https://bbs.pinggu.org/thread-16514727-1-1.html
【英文金融科技资料】Mastering Financial Markets with Python https://bbs.pinggu.org/thread-16497417-1-1.html
【英文金融科技资料】AI in Financial Decision Making  https://bbs.pinggu.org/thread-16496535-1-1.html
【英文经济学资料】Probability Theory for Quantitative Scientists https://bbs.pinggu.org/thread-16496522-1-1.html
【英文计量经济学资料】Machine Learning for Econometrics https://bbs.pinggu.org/thread-16496057-1-1.html
【英文计量经济学资料】Modern Series Methods in Econometrics and Statistics https://bbs.pinggu.org/thread-16496039-1-1.html
【英文计量经济学资料】Time Series Econometrics https://bbs.pinggu.org/thread-16496034-1-1.html
【英文金融科技】PY金融数据分析Data Analytics for Finance Using Python https://bbs.pinggu.org/thread-16458389-1-1.html
【英文资料】会计与审计研究工具及方法Accounting and Auditing Research Tools and S https://bbs.pinggu.org/thread-16458370-1-1.html
【英文资料】行为金融学Behavioral Finance Limited Rationality in Financial Market  https://bbs.pinggu.org/thread-16458354-1-1.html
【英文经济学资料】Applied Behavioral Economics Theory, Method, and Practice https://bbs.pinggu.org/thread-16429463-1-1.html
【英文经济管理资料】Taxation in the Digital Era 数字时代的税务 https://bbs.pinggu.org/thread-16429464-1-1.html
【英文经济学资料】Agricultural Economics and Policy农业经济与政策 https://bbs.pinggu.org/thread-16429466-1-1.html
【英文金融技术资料】Deep Learning: Advanced Techniqes For FinanceDeep Learning https://bbs.pinggu.org/thread-16429471-1-1.html
【英文金融投资资料】BUSINESS VALUATION IN THE LAW https://bbs.pinggu.org/thread-16430700-1-1.html
【英文金融资料】Computational Methods in Finance金融计算方法 https://bbs.pinggu.org/thread-16434977-1-1.html
【英文金融资料】Computation and Simulation for Finance金融计算与仿真  https://bbs.pinggu.org/thread-16435036-1-1.html
【英文金融科技资料】Computational Intelligence for Autonomous Finance Challenges https://bbs.pinggu.org/thread-16435100-1-1.html
【英文金融数学资料】Mathematics Computational Finance计算金融数学 https://bbs.pinggu.org/thread-16435130-1-1.html
【英文金融资料】Artificial Intelligence and Finance Competition, Crimes and Finance https://bbs.pinggu.org/thread-16435177-1-1.html
【英文资料】AI in Accounting, Auditing and Finance AI会计审计金融应用 https://bbs.pinggu.org/thread-16435305-1-1.html
【英文资料】The Unaffordable Price of Static Decision-making Models Challenges https://bbs.pinggu.org/thread-16398858-1-1.html
【英文资料】不动产金融建模基础 Foundations of Real Estate Financial Modelling https://bbs.pinggu.org/thread-16397953-1-1.html
【英文资料】深度学习经济研究Deep Learning Models for Economic Research https://bbs.pinggu.org/thread-16397938-1-1.html
【社会学研究方法英文资料】Social Research Methods and Applications Qualitative M https://bbs.pinggu.org/thread-16217907-1-1.html
【研究方法英文资料】The Art and Science of Quantitative Research https://bbs.pinggu.org/thread-16217918-1-1.html
【管理学英文资料】量化风险管理PY Quantitative Risk Management Using Python  https://bbs.pinggu.org/thread-16217936-1-1.html
【金融智能】Mastering Financial Markets with Python New Horizons in Technical An  https://bbs.pinggu.org/thread-16697530-1-1.html
【行为金融学小册子】A Behavioral Finance Approach to International Monetary https://bbs.pinggu.org/thread-16697508-1-1.html
【网络分析经济金融应用】Network Analysis for Economic, Business and Financial https://bbs.pinggu.org/thread-16697500-1-1.html
【金融智能】Financial Statistics with Python Methods, Regression, Risk Analysis https://bbs.pinggu.org/thread-16697492-1-1.html
【因果推断与机器学习】Causal Inference and Machine Learning In Economics, Social  https://bbs.pinggu.org/thread-16697303-1-1.html
【计量经济学非线性建模】Non-Linearity in Econometric Modeling A Practical Appro https://bbs.pinggu.org/thread-16694207-1-1.html
【计量经济学实证习题集】Exercise Empirical Economic Research and Econometrics  https://bbs.pinggu.org/thread-16694198-1-1.html
【人工智能金融科技研究】Artificial Intelligence in Finance, Accounting and Busin https://bbs.pinggu.org/thread-16693943-1-1.html
【经济学资料】数字经济学习资料Digital Economics（中英文合集） https://bbs.pinggu.org/thread-16692726-1-1.html
【社会学研究方法资料】Social Network Analysis in Second Language Research https://bbs.pinggu.org/thread-16692044-1-1.html
【供应链创新研究资料】Innovations in Computational Logistics and Supply Chain https://bbs.pinggu.org/thread-16691659-1-1.html
【供应链管理研究资料】Principles of Supply Chain Management A Balanced Approach  https://bbs.pinggu.org/thread-16691655-1-1.html
【供应链管理资料】Logistics and Distribution Management Understanding the SC  https://bbs.pinggu.org/thread-16691651-1-1.html
【供应链管理研究资料】Supply Chain Analytics Principles and Applications https://bbs.pinggu.org/thread-16691649-1-1.html
【定量研究及基础统计学资料】Scientific Research and Methodology--Quantitative RS https://bbs.pinggu.org/thread-16711002-1-1.html
【量化行为研究资料包】Quantitative Behavior Analysis  https://bbs.pinggu.org/thread-16710992-1-1.html
【社会学研究方法资料】Social Research Methods Qualitative and Quantitative Appro https://bbs.pinggu.org/thread-16710756-1-1.html
【人工智能商业应用】AI in Business Management Bridging the Gap https://bbs.pinggu.org/thread-16710747-1-1.html
【经济学资料】The Economics of Political Philosophy A Treatise on Socio-Liberal https://bbs.pinggu.org/thread-16709748-1-1.html
【经济金融数学】Economic Analysis Through Mathematics Tools and Techniques DM  https://bbs.pinggu.org/thread-16709664-1-1.html
【会计计量与经济估值价值资料】Accounting Measurement and Economic Value https://bbs.pinggu.org/thread-16709651-1-1.html
【原材料经济学资料】Introduction to Atomic Economics https://bbs.pinggu.org/thread-16709635-1-1.html
【智能经济学】AI in Socio-Economic Systems Novel Human-Centric Solutions https://bbs.pinggu.org/thread-16709616-1-1.html
【因果推断资料】Modern Randomization-Based Design and Analysis for Causal Infere  https://bbs.pinggu.org/thread-16708746-1-1.html
【因果研究方法】Every Coin Has Two Sides Causal Inference in Pharmaceutical Stat https://bbs.pinggu.org/thread-16708740-1-1.html
【金融算法交易机器学习资料】Algorithmic Trading via AI Machine Learning with R https://bbs.pinggu.org/thread-16705689-1-1.html
【金融交易与资产管理】Generative AI for Trading for Asset Management https://bbs.pinggu.org/thread-16705566-1-1.html
     【量化金融数据分析资料】Quantitative Python Unlocking Datas in Finance https://bbs.pinggu.org/thread-16715466-1-1.html
     【金融数据分析资料】Data Analytics in Finance https://bbs.pinggu.org/thread-16715462-1-1.html
【智能机器学习金融服务监管】AI and ML Governance in Financial Services https://bbs.pinggu.org/thread-16705072-1-1.html
【金融智能工具应用】AI-Powered Accounting withExcel and Power BI https://bbs.pinggu.org/thread-16705064-1-1.html
【社会经济行为AI】The Social and Economic Behavior of AI Agents Theory and Exper https://bbs.pinggu.org/thread-16701768-1-1.html
【金融经济AIGC资料】Generative AI for Business and Economics Theory and App  https://bbs.pinggu.org/thread-16701717-1-1.html

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏0 回帖