经鸿之粟 | 三季度高质量论文汇总

This paper discusses a network game in a local-average setup, where players' payoffs depend on the social norms they confront. I focus on an optimal targeting intervention problem, where a planner aims to maximize social welfare by altering individual characteristics ex ante. First, I decompose orthogonally the optimal intervention into the eigenspaces of the social welfare matrix. In what follows, I present the limit forms of the optimal intervention as the budget approaches innity or zero. I then identify the errors arising from approximating the optimal intervention with its various limit forms. Finally, this paper explores the intervention problem under incomplete information, where the planner switches from mean to variance intervention based on the comparison of their unit costs and the aggregate budget size.

In China, state owners make minority equity investments in private firms. We study the allocative implications of such government investments using a two-sector DSGE model with financial friction and idiosyncratic productivities. In our model, private owners are more productive than state owners but face tighter financial constraints. Equity investment by a state owner alleviates the private owner’s financial constraint but dampens its own productivity, consistent with Chinese firm-level data. Under this setup, only private owners with sufficiently high productivity accept such investment, while only state owners with sufficiently low productivity make such investment. As a result, expansion of such government investment improves capital allocation within each sector and across sectors. Our analysis shows that financial liberalizations, including liberalizing interest-rate controls and reducing the loan-to-value gap between sectors, stimulate private owners’ demand for such government investment but discourage state owners from making it, thus generating an ambiguous effect on aggregate productivity.

This paper considers a semiparametric single index model in which the dependent variable is subject to a nonparametric transformation. The model has the form G0(Y)=g0(XTθ0)+e, where X is a random vector of regressors, Y is the dependent variable and e is the random noise, the monotonic function G0 , the smooth function g0 and the index vector θ0 are all unknown. This model is quite general in the sense that it nests many popular regression models as special cases. We first propose identification strategies for the three unknown quantities, based on which estimators are then constructed. The kernel density weighted average derivative estimator of δ (proportional to θ0 ) has a V -statistic representation and its asymptotical normality is established under the small bandwidth asymptotics. The kernel estimator of the transformation function G0 is a functional of the conditional distribution estimator of Y given XTθ0 and is shown to be √ n-consistent and asymptotically normal. The sieve estimator of g0 is shown to enjoy the standard nonparametric asymptotic properties. A specification test for the single index structure and extension to allow for endogeneous regressors are also developed. In addition, data-driven choices of the smoothing parameters are discussed. Simulation results illustrate the nice finite sample performance of the proposed estimators and specification test. An empirical application to studying the impact of family income on child achievement demonstrates the practical merits of the proposed model.

Chernozhukov, Fernández-Val, and Luo (2023, CFL (2023) hereafter) considered a distribution regression model subject to sample selection with a binary selection mechanism. In this paper, we show how to identify and estimate a semi-parametric distribution regression model subject to a censored selection rule. With censored selection, we do not need to impose the usual outcome exclusion restriction or exclusion of the level of the latent selection variable from the selection sorting function for model identification, unlike CFL (2023). We propose new semiparametric estimators and corresponding inference procedures for model parameters and related functional parameters. We apply our method to investigate wage inequality in the UK for the period 1978–2000 using the Family Expenditure Survey (FES) data. Our findings reveal that (i) the selection sorting exclusion and outcome exclusion restrictions imposed by CFL (2023) are rejected; (ii) there is negative selection into work at most quantile levels for females, but not for males; (iii) in contrast to CFL (2023), our selection sorting effect pattern does not offer clear evidence on assortative matching or glass ceiling in the UK labor market; (iv) the latent gender wage gaps after correcting for selection bias are about 25%–50% of CFL (2023)’s estimates, and are also significantly smaller than the observed wage gaps; (v) similar to CFL (2023), there exists some strong evidence on gender discrimination in the UK labor market.

We examine the causal impact of ideological education on students’ political participation and career choices by exploiting China’s staggered rollout of a high school curriculum reform that emphasized political indoctrination. Using nationally representative survey data on college students that the authors collected, we find that exposure to the new curriculum increases the likelihood  of  joining  the  Chinese  Communist  Party  by  14%  and  raises  the  probability  of securing state-sector jobs after graduation by 15%. These results highlight the powerful role of ideological education in shaping students’ political alignments and career trajectories.

This paper reports a field experiment on Shanghai cabdrivers’ labor supply, analyzing the data using an expectations-based reference-dependent model that allows daily income- and hours-targeting. Our main innovation is to elicit the cabdrivers’ income and hours expectations, twice a day. We find that expectations indeed affect labor supply in a way predicted by a reference-dependent model, and hours expectations have a stronger influence than income expectations. Both expectations are found to be correlated with their most recent historical average values. While income expectations do adjust within the day, hours expectations are sticky. The findings suggest that the targeting effect based on hours expectations plays a more important role than traditionally thought.

This article proposes a structural change test for copula models based on the kernel smoothing method. The proposed approach enables adaptable estimation of the dynamic marginal distributions, either parametri-cally or semi-parametrically. The test statistic is formulated via the weighted quadratic distance between the local smoothing copula and the empirical copula function, using pseudo-observations of marginal distributions. The test statistic is pivotal with an asymptotic standard Normal distribution, irrespective of the marginal distributions, parameters, and estimations, and is consistent against a wide range of smoothly transitioning structural changes as well as abrupt structural breaks for copula models. Monte Carlo simulations show that the test performs well in finite samples and outperforms existing tests in the case of periodic changes.

Clustered data analysis is an important topic in data science. A well established approach is to assume all clusters share the same unknown parameters of interest, and the difference between different clusters is formulated and accounted for by cluster effects. Whilst this approach works very well in many issues, such as exploring the global impact of an explanatory variable on the response variable, it does not provide much insight about individual attributes of each cluster. Assuming different clusters have completely different parameters would result in too many unknown parameters, which would lead to large variances of the final estimators. Following the idea of homogeneity pursuit proposed in Ke et al. (2015), various modelling approaches are proposed in recent literature to group the unknown parameters and explore the individual attributes in clustered data analysis. However, most of them are either difficult to implement or require each cluster to have reasonably big cluster size. In this paper, we propose a new approach, which is easy to implement and does not require any cluster to have big size, and establish its asymptotic properties without assuming the size of any cluster tends to infinity. We also conduct intensive simulation studies to show the approach works very well when sample size is finite. Finally, we apply the approach to a well known financial dataset to show its superiority in exploring individual attributes in clustered data analysis.

We revisit the concept of reclaim-proof allocations proposed by Ekici (2013) for house allocation problems with existing tenants. As a concept of core, the definition of reclaim-proof allocations assumes that when a coalition blocks an allocation, an agent in the coalition is allowed to bring her allocated object into the coalition, even when it is privately owned by an outsider. We propose a new notion of core called the effectual core by restoring the feasibility of coalitional blocking in Ekici’s definition. Our main result shows that the effectual core, while by definition weaker than reclaim-proofness, is actually equivalent to it. Together with Ekici’s results, it is then immediate that an allocation is in the effectual core if and only if it is produced by the You request my house-I get your turn (YRMH-IGYT) mechanism (Abdulkadiroğlu and Sönmez, 1999) and if and only if it is a competitive allocation.

大数据时代的到来为CPI的预测带来了前所未有机遇和挑战，充分利用高维数据信息， 发展可解释的机器学习预测模型，对于理论发展和现实实践均具有重要意义。为此，本研究构建了包含9个类别239个变量的中国月度宏观经济数据库，并对比了包含传统时间序列模型、正则化回归、因子模型和集成算法等在内的13个模型在大型数据集下对CPI的预测能力。进一步地，基于控制变量的思想构建了机器学习衍生算法，对相关的结果进行解释和机制分析。结果表明，随机森林和XGBoost具有良好的预测效果，尤其是在中长期预测中表现出了较大优势。通过进一步的分析发现它们的优势在于非线性的模型设定和非稀疏的变量处理，前者使得模型中的变量关系更加符合实际，而后者能够充分地利用大数据信息。同时，这两个模型也筛选出了自回归项、价格、就业等在CPI预测中更加合理且重要的变量类别。

洗钱不仅破坏金融体系稳定，还助长犯罪、腐败和恐怖主义活动，对国家经济安全以及国际声誉造成负面影响。随着数字经济的发展，义务机构(包括金融机构和特定非金融机构)和监管机构面临的反洗钱风险发生了变化。本文在Stackelberg博弈框架下讨论数字经济发展中的重要支撑——数据要素对洗钱者、义务机构和监管机构行为的影响，并基于2013～2023年60个国家和地区的面板数据，实证检验了数据要素发展对反洗钱风险的影响。研究发现，反洗钱风险相关主体的策略选择与数据要素相对边际贡献影响有关，数据要素发展降低了反洗钱风险，且主要通过数字支付覆盖率的提升和监管机构数字化能力的提高实现。数字支付覆盖率的提升和监管数字化转型，提高了金融交易透明度，提升了监管机构的监测和分析能力，使义务机构和监管机构能够更有效地监测和追踪资金流动。异质性检验发现，与高收入国家相比，数据要素发展更显著地降低了非高收入国家的反洗钱风险。本文研究对于理解数据要素发展如何重塑反洗钱风险、加大对数据要素开发利用的投入、进一步推动监管机构数字化转型、提升反洗钱工作的有效性具有理论和政策价值。

晚婚、晚育和家庭少子化已成为我国人口发展面临的瓶颈，厘清婚育行为背后的深层次原因对推动我国人口高质量发展实现中国式现代化具有重要意义。文章基于2010-2020 年中国家庭追踪调查面板数据，展示了我国劳动力人力资本错配程度在不同行业和职业上的分布，并实证检验了人力资本错配对婚姻和生育的惩罚效应及其影响机制。研究表明，人力资本错配显著降低个体的结婚率和生育率，拖后个体的结婚和生育时间，并减少生育子女的数量，婚姻推迟是造成生育水平下降的核心原因。这一影响在男女间不存在显著差异，女性并没有因为工作上的人力资本错配选择回归家庭。个体收入下降与工作不确定性增加是人力资本错配对婚育行为造成负面影响的主要影响机制。

作为制造与服务融合发展的新型产业形态，服务型制造推动了制造业从以产品为中心向以人为中心转型，并已经成为我国制造业转型升级的重要方向。加快发展服务型制造，是我国在深刻洞察产业资本发展规律的基础上，有效提升制造业企业核心竞争力的实践抓手，更是推进新型工业化、加快制造强国建设的战略选择。