Journalof Econometrics 2023年第1期

Volume 232,Issue 1，January 2023

——更多动态，请持续关注gzh：理想主义的百年孤独

1.Time seriesanalysis of COVID-19 infection curve: A change-point perspective

COVID-19感染曲线的时间序列分析:变化点视角

Feiyu Jiang, Zifeng Zhao, Xiaofeng Shao

In this paper, we model the trajectory of the cumulativeconfirmed cases and deaths of COVID-19 (in log scale) via a piecewise lineartrend model. The model naturally captures the phase transitions of the epidemicgrowth rate via change-points and further enjoys great interpretability due toits semiparametric nature. On the methodological front, we advance the nascentself-normalization (SN) technique (Shao, 2010) to testing and estimationof a single change-point in the linear trend of a nonstationary time series. Wefurther combine the SN-based change-point test with the NOT algorithm(Baranowski et al., 2019) to achieve multiple change-point estimation. Using theproposed method, we analyze the trajectory of the cumulative COVID-19 cases anddeaths for 30 major countries and discover interesting patterns withpotentially relevant implications for effectiveness of the pandemic responsesby different countries. Furthermore, based on the change-point detectionalgorithm and a flexible extrapolation function, we design a simple two-stageforecasting scheme for COVID-19 and demonstrate its promising performance inpredicting cumulative deaths in the U.S.

在本文中，我们通过分段线性趋势模型对COVID-19累计确诊病例和死亡病例(对数尺度)的轨迹进行建模。该模型通过变点自然地捕捉了疫情增长率的阶段性变化，并由于其半参数性质而具有较大的可解释性。在方法方面，我们提出了新生的自我规范化(SN)技术(Shao, 2010)，以测试和估计非平稳时间序列的线性趋势中的一个单一变化点。我们进一步将基于sn的变点检验与NOT算法(Baranowski et al.， 2019)相结合，实现多重变点估计。利用所提出的方法，我们分析了30个主要国家累计COVID-19病例和死亡的轨迹，并发现了有趣的模式，这些模式可能对不同国家的疫情应对有效性产生相关影响。此外，基于变点检测算法和灵活的外推函数，我们设计了一个简单的COVID-19两阶段预测方案，并在预测美国的累计死亡方面展示了其良好的性能

2.Nowcasting theoutput gap

预测产出缺口

Tino Berger, James Morley, Benjamin Wong

We propose a way to directly nowcast the output gap using theBeveridge–Nelson decomposition based on a mixed-frequency Bayesian VAR. Themixed-frequency approach produces similar but more timely estimates of the U.S.output gap compared to those based on a quarterly model, the CBO measure ofpotential, or the HP filter. We find that within-quarter nowcasts for theoutput gap are more reliable than for output growth, with monthly indicatorsfor a credit risk spread, consumer sentiment, and the unemployment rateproviding particularly useful new information about the final estimate of theoutput gap. An out-of-sample analysis of the COVID-19 crisis anticipates theexceptionally large negative output gap of −8.3% in 2020Q2 beforethe release of real GDP data for the quarter, with both conditional andscenario nowcasts tracking a dramatic decline in the output gap given the Aprildata.

我们提出了一种方法，使用基于混合频率贝叶斯VAR的贝弗里奇-尼尔森分解来直接预测产出缺口。与基于季度模型、CBO潜力测度或HP滤波器的方法相比，混合频率方法产生了类似但更及时的美国产出缺口估计。我们发现，与产出增长相比，产出缺口的季度内即期预测更可靠，信用风险利差、消费者情绪和失业率的月度指标为产出缺口的最终估计提供了特别有用的新信息。对COVID-19危机的样本外分析预计，在公布该季度的实际GDP数据之前，2020年第二季度的负产出缺口将达到- 8.3%，考虑到4月份的数据，条件和情景nowcast都将跟踪产出缺口的大幅下降。

3.Time varyingMarkov process with partially observed aggregate data: An application tocoronavirus

具有部分观测汇总数据的时变马尔可夫过程:对冠状病毒的应用

C. Gourieroux, J. Jasiak

A major difficulty in the analysis of Covid-19 transmission isthat many infected individuals are asymptomatic. For this reason, the totalcounts of infected individuals and of recovered immunized individuals areunknown, especially during the early phase of the epidemic. In this paper, weconsider a parametric time varying Markov process of Coronavirus transmissionand show how to estimate the model parameters and approximate the unobservedcounts from daily data on infected and detected individuals and the total dailydeath counts. This model-based approach is illustrated in an application toFrench data, performed on April 6, 2020.

分析Covid-19传播的一个主要困难是许多感染者没有症状。因此，感染者总数和已恢复免疫接种者总数尚不清楚，特别是在疫情早期。在本文中，我们考虑了冠状病毒传播的参数时变马尔可夫过程，并展示了如何估计模型参数，以及如何从被感染和被检测个体的每日数据和总每日死亡计数中近似未观察到的计数。这种基于模型的方法在2020年4月6日执行的法国数据应用程序中得到了说明。

4.Nowcasting in apandemic using non-parametric mixed frequency VARs

Florian Huber, Gary Koop, Luca Onorante, Michael Pfarrhofer,Josef Schreiner

This paper develops Bayesian econometric methods for posteriorinference in non-parametric mixed frequency VARs using additive regressiontrees. We argue that regression tree models are ideally suited formacroeconomic nowcasting in the face of extreme observations, for instancethose produced by the COVID-19 pandemic of 2020. This is due to theirflexibility and ability to model outliers. In an application involving fourmajor euro area countries, we find substantial improvements in nowcastingperformance relative to a linear mixed frequency VAR.

本文利用加性回归树发展了非参数混合频率变量的后验推理的贝叶斯计量经济学方法。我们认为，在面临极端观测的情况下，回归树模型非常适合宏观经济临近预测，例如2020年COVID-19大流行产生的模型。这是由于它们对异常值建模的灵活性和能力。在一个涉及欧元区四个主要国家的应用中，我们发现，与线性混合频率VAR相比，近预报性能有了实质性的改善。

5.How to go viral:A COVID-19 model with endogenously time-varying parameters

如何传播病毒:具有内生性时变参数的COVID-19模型

Paul Ho, Thomas A. Lubik, Christian Matthes

We estimate a panel model with endogenously time-varyingparameters for COVID-19 cases and deaths in U.S. states. The functional formfor infections incorporates important features of epidemiological models but isflexibly parameterized to capture different trajectories of the pandemic. Dailydeaths are modeled as a spike-and-slab regression on lagged cases. Our Bayesianestimation reveals that social distancing and testing have significant effectson the parameters. For example, a 10 percentage point increase in the positivetest rate is associated with a 2 percentage point increase in the death rateamong reported cases. The model forecasts perform well, even relative to modelsfrom epidemiology and statistics.

我们估计了一个具有内生性时变参数的美国各州COVID-19病例和死亡的面板模型。感染的函数形式包含了流行病学模型的重要特征，但可以灵活地参数化，以捕捉大流行的不同轨迹。每日死亡被建模为滞后情况下的尖峰-平板回归。我们的贝叶斯估计显示，社会距离和测试对参数有显著影响。例如，阳性检测率每增加10个百分点，报告病例的死亡率就增加2个百分点。该模型的预测效果很好，甚至与流行病学和统计学的模型相比也是如此。

6.Nonparametriccomparison of epidemic time trends: The case of COVID-19

流行时间趋势的非参数比较:COVID-19的案例

Marina Khismatullina, Michael Vogt

The COVID-19 pandemic is one of the most pressing issues atpresent. A question which is particularly important for governments and policymakers is the following: Does the virus spread in the same way in differentcountries? Or are there significant differences in the development of theepidemic? In this paper, we devise new inference methods that allow to detectdifferences in the development of the COVID-19 epidemic across countries in astatistically rigorous way. In our empirical study, we use the methods tocompare the outbreak patterns of the epidemic in a number of Europeancountries.

COVID-19大流行是目前最紧迫的问题之一。对各国政府和政策制定者来说，一个特别重要的问题是:病毒在不同国家的传播方式是否相同?还是疫情的发展存在显著差异?在本文中，我们设计了新的推理方法，允许以统计严谨的方式检测COVID-19疫情在各国发展的差异。在我们的实证研究中，我们使用这些方法来比较该流行病在一些欧洲国家的爆发模式。

7.Who should getvaccinated? Individualized allocation of vaccines over SIR network

谁应该接种疫苗?通过SIR网络进行疫苗的个性化分配

Toru Kitagawa, Guanyi Wang

How to allocate vaccines over heterogeneous individuals is oneof the important policy decisions in pandemic times. This paper develops aprocedure to estimate an individualized vaccine allocation policy under limitedsupply, exploiting social network data containing individual demographiccharacteristics and health status. We model the spillover effects ofvaccination based on a Heterogeneous-Interacted-SIR network model and estimatean individualized vaccine allocation policy by maximizing an estimated socialwelfare (public health) criterion incorporating these spillovers. While thisoptimization problem is generally an NP-hard integeroptimization problem, we show that the SIR structure leads to a submodularobjective function, and provide a computationally attractive greedy algorithmfor approximating a solution that has a theoretical performance guarantee.Moreover, we characterize a finite sample welfare regret bound and examine howits uniform convergence rate depends on the complexity and riskiness of thesocial network. In the simulation, we illustrate the importance of consideringspillovers by comparing our method with targeting without network information.

如何在不同个体之间分配疫苗是大流行时期的重要政策决策之一。本文利用包含个体人口特征和健康状况的社会网络数据，建立了有限供应条件下的个体化疫苗分配策略。我们基于异质交互sir网络模型对疫苗接种的溢出效应进行建模，并通过最大化包含这些溢出效应的估计社会福利(公共卫生)标准来估计个性化疫苗分配政策。虽然这个优化问题通常是一个NP-hard整数优化问题，但我们表明，SIR结构导致了一个子模块目标函数，并提供了一个计算上有吸引力的贪婪算法来逼近一个具有理论性能保证的解决方案。此外，我们刻画了一个有限样本的福利后悔约束，并考察了其一致收敛速度如何依赖于社会网络的复杂性和风险。在仿真中，我们通过比较我们的方法与没有网络信息的目标来说明考虑溢出的重要性。

8.Sparsespatio-temporal autoregressions by profiling and bagging

稀疏时空自回归分析和套袋

Yingying Ma, Shaojun Guo, Hansheng Wang

We consider a new class of spatio-temporal models with sparseautoregressive coefficient matrices and exogenous variable. To estimate themodel, we first profile the exogenous variable out of the response. This leadsto a profiled model structure. Next, to overcome endogeneity issue, we proposea class of generalized methods of moment (GMM) estimators to estimate theautoregressive coefficient matrices. A novel bagging-based estimator is furtherdeveloped to conquer the over-determined issue which also occurs in Changet al. (2015) and Dou et al. (2016). An adaptive forward–backwardgreedy algorithm is proposed to learn the sparse structure of theautoregressive coefficient matrices. A new BIC-type selection criteria isfurther developed to conduct variable selection for GMM estimators. Asymptoticproperties are further studied. The proposed methodology is illustrated withextensive simulation studies. A social network dataset is analyzed forillustration purpose.

考虑一类具有稀疏自回归系数矩阵和外生变量的时空模型。为了对模型进行估计，我们首先从反应中提取外生变量。这就产生了一个概要的模型结构。其次，为了克服内生性问题，我们提出了一类广义矩估计方法来估计自回归系数矩阵。为了克服Chang et al.(2015)和Dou et al.(2016)中也出现过的过度确定问题，进一步开发了一种新的基于bagging的估计器。提出了一种自适应的前向后贪心算法来学习自回归系数矩阵的稀疏结构。本文进一步提出了一种新的bic型选择准则，用于GMM估计量的变量选择。进一步研究了其渐近性质。所提出的方法是说明了广泛的模拟研究。为了说明目的，分析了一个社会网络数据集。