Mostly Economicsも取り上げているが、表題のNBER論文が上がっている。原題は「Business in a Time of Spanish Influenza」で、著者はクレムソン大のHoward Bodenhorn。

Mandated shutdowns of nonessential businesses during the COVID-19 crisis brought into sharp relief the tradeoff between public health and a healthy economy. This paper documents the short-run effects of shutdowns during the Spanish flu pandemic of 1918, which provides a useful counterpoint to choices made in 2020. The 1918 closures were shorter and less sweeping, in part because the US was at war and the Wilson administration was unwilling to let public safety jeopardize the war’s prosecution. The result was widespread sickness, which pushed some businesses to shutdown voluntarily; others operated shorthanded. Using hand-coded, high-frequency data (mostly weekly) this study reports three principal results. First, retail sales declined during the three waves of the pandemic; manufacturing activity slowed, but by less than retail. Second, worker absenteeism due to either sickness or fear of contracting the flu reduced output in several key sectors and industries that were not ordered closed by as much as 10 to 20% in weeks of high excess mortality. Output declines were the result of labor-supply rather than demand shocks. And, third, mandated closures are not associated with increases in the number or aggregate dollar value of business failures, but the number and aggregate dollar value of business failures increased modestly in weeks of high excess mortality. The results highlight that the tradeoff between mandated closures and economic activity is not the only relevant tradeoff facing public health authorities. Economic activity also declines, sometimes sharply, during periods of unusually high influenza-related illness and excess mortality even absent mandated business closures.


MRブログでアレックス・タバロックが「重要な新たな論文」として紹介しているが、表題のNBER論文が上がっている。原題は「Group Testing in a Pandemic: The Role of Frequent Testing, Correlated Risk, and Machine Learning」で、著者はNed Augenblick、Jonathan T. Kolstad、Ziad Obermeyer、Ao Wang(いずれもUCバークレー)。

Group testing increases efficiency by pooling patient specimens and clearing the entire group with one negative test. Optimal grouping strategy is well studied in one-off testing scenarios with reasonably well-known prevalence rates and no correlations in risk. We discuss how the strategy changes in a pandemic environment with repeated testing, rapid local infection spread, and highly uncertain risk. First, repeated testing mechanically lowers prevalence at the time of the next test. This increases testing efficiency, such that increasing frequency by x times only increases expected tests by around √x rather than x. However, this calculation omits a further benefit of frequent testing: infected people are quickly removed from the population, which lowers prevalence and generates further efficiency. Accounting for this decline in intra-group spread, we show that increasing frequency can paradoxically reduce the total testing cost. Second, we show that group size and efficiency increases with intra-group risk correlation, which is expected in natural test groupings based on proximity. Third, because optimal groupings depend on uncertain risk and correlation, we show how better estimates from machine learning can drive large efficiency gains. We conclude that frequent group testing, aided by machine learning, is a promising and inexpensive surveillance strategy.


というNBER論文が上がっている。原題は「The Determinants of Fiscal and Monetary Policies During the Covid-19 Crisis」で、著者はEfraim Benmelech(ノースウエスタン大)、Nitzan Tzur-Ilan(同)。

As countries around the world grapple with Covid-19, their economies are grinding to a halt. For the first time since the Great Depression both advanced economies and developing economies are in recession. Governments and central banks have responded to the pandemic and the economic crisis using both fiscal and monetary tools on a scale that the world has not witnessed before. This paper analyzes the determinants of fiscal and monetary policies during the Covid-19 crisis. We find that high-income countries announced larger fiscal policies than lower-income countries. We also find that a country’s credit rating is the most important determinant of its fiscal spending during the pandemic. High-income countries entered the crisis with historically low interest rates and as a result were more likely to use nonconventional monetary policy tools. These findings raise the concern that countries with poor credit histories – those with lower credit ratings and, in particular, lower-income countries – will not be able to deploy fiscal policy tools effectively during economic crises.


「Credit Booms, Financial Crises and Macroprudential Policy」というNBER論文を清滝信宏氏とガートラーらが書いているungated版(4月時点のWP)日本経済国際共同研究センターの昨年8/22のマクロ経済学ワークショップでのスライドペンシルベニア大の昨年5/3のコンファレンスでのスライドSemantic Scholarの一昨年12月のスライド)。以前、同じ著者(Mark Gertler(NYU)、Nobuhiro Kiyotaki(プリンストン大)、Andrea Prestipino(FRB))のNBER論文を紹介したことがあったが、そちらの論文のモデルを基に行われた研究のようである。

We develop a model of banking crises which is consistent with two important features of the data: First, banking crises are usually preceded by credit booms. Second, credit booms often do not result in a crisis. That is, there are "good" booms as well as "bad" booms in the language of Gorton and Ordonez (2019). We then consider how the optimal macroprudential policy weighs the benefits of preventing a crisis against the costs of stopping a good boom. We show that countercyclical capital buffers are a critical feature of a successful macroprudential policy.

*1:cf. これ


「Testing, Voluntary Social Distancing and the Spread of an Infection」というNBER論文をアセモグルらが上げている。著者はDaron Acemoglu(MIT)、Ali Makhdoumi(デューク大)、Azarakhsh Malekian(トロント大)、Asuman Ozdaglar(MIT)。

This paper studied the effects of testing on social activity and voluntary social distancing in the context of an epidemic. Social activity levels determine the (endogenous) social network over which contacts take place and an infection spreads. Testing enables authorities to identify and isolate infected individuals who spread the virus, and has been identified by the recent literature on COVID-19 and policymakers as a key tool for combating epidemics. Our analysis, however, shows that the impact of testing on the spread of an epidemic may be more complex because, knowing that tests will lead to the isolation of infected individuals, agents can increase their social activity levels and refrain from voluntary social distancing. As a result, our analysis established that the effects of testing on the spread of the infection can be non-monotonic—greater testing can lead to higher infection probabilities.
Our analysis also characterized the optimal testing policies. The same forces that lead to nonmonotonic comparative statics also imply that a benevolent social planner may prefer to leave her testing capacity partially or fully unused—because increasing testing can make the spread of the virus more likely. This implies that testing should often be combined with mandatory social distancing measures—which ensure that the adverse behavioral effects of testing can be countered by preventing excessively high social activity levels.
Our paper is part of a growing literature on the interaction between economic incentives and epidemiological dynamics. Two high-level contributions of our approach are to conceptualize the problem of endogenous behavior as one of social network formation and to use the percolation model rather than the SIR dynamic model. Both of these contributions can be useful beyond the confines of our specific question, but the robustness of our conclusions to relaxing both assumptions and adopting different modeling strategies need to be investigated. Other interesting areas for research include the analysis of optimal testing and tracing when tests lead to type I and type II errors and policy is constrained by privacy considerations and non-obedience (both in acquiescing to testing and following mandatory social distancing guidelines). Another interesting avenue is to enrich the setup to incorporate more heterogeneity and richer economic, social and epidemiological interactions so as to enable quantitative policy analysis.


というNBER論文をカーメン・ラインハートらが上げているungated版VoxEU記事)。原題は「Coping with Disasters: Two Centuries of International Official Lending」で、著者はSebastian Horn(ミュンヘン大)、Carmen Reinhart(ハーバード大)、Christoph Trebesch(キール世界経済研究所)。

Official (government-to-government) lending is much larger than commonly known, often surpassing total private cross-border capital flows, especially during disasters such as wars, financial crises and natural catastrophes. We assemble the first comprehensive long-run dataset of official international lending, covering 230,000 loans, grants and guarantees extended by governments, central banks, and multilateral institutions in the period 1790-2015. Historically, wars have been the main catalyst of government-to-government transfers. The scale of official credits granted in and around WW1 and WW2 was particularly large, easily surpassing the scale of total international bailout lending after the 2008 crash. During peacetime, development finance and financial crises are the main drivers of official cross-border finance, with official flows often stepping in when private flows retrench. In line with the predictions of recent theoretical contributions, we find that official lending increases with the degree of economic integration. In crises and disasters, governments help those countries to which they have greater trade and banking exposure, hoping to reduce the collateral damage to their own economies. Since the 2000s, official finance has made a sharp comeback, largely due to the rise of China as an international creditor and the return of central bank cross-border lending in times of stress, this time in the form of swap lines.


というNBER論文が上がっているungated(SSRN)版)。原題は「Biases in Long-Horizon Predictive Regressions」で、著者はJacob Boudoukh(IDC ヘルツェリア大)、Ronen Israel(AQR Capital)、Matthew P. Richardson(NYU)。

Analogous to Stambaugh (1999), this paper derives the small sample bias of estimators in J-horizon predictive regressions, providing a plug-in adjustment for these estimators. A number of surprising results emerge, including (i) a higher bias for overlapping than nonoverlapping regressions despite the greater number of observations, and (ii) particularly higher bias for an alternative long-horizon predictive regression commonly advocated for in the literature. For large J, the bias is linear in (J/T) with a slope that depends on the predictive variable’s persistence. The bias adjustment substantially reduces the existing magnitude of long-horizon estimates of predictability.


 Rt:t+1 = α1 + β1Xt + ut+1
 Xt+1 = ω + ρXt + vt+1                      (1)
 E[β^1 - β1] = (σuvv2)E[ρ^ - ρ] ≒ - (σuvv2){(1+3ρ)/T}

 Rt:t+J = αJ + βJXt + εt:t+J                      (2)
研究者が、J期毎のサンプルという重複の無いサンプル長T/Jのデータ(nolと表記される)を用いて(2)式を推計すれば、標準的な通常の最小二乗法(OLS)が適用される。しかし、Jが大きいとサンプルサイズが小さくなるのが一般的であるため、研究者は重複する全データ(olと表記される)を用いて(2)式を推計する。データが多いと推計値の漸近的効率性は向上するが、自己相関のある誤差によってOLSの標準誤差の特定の誤りも生じる。そのため、研究者は、これまで開発された様々な分散不均一性・自己相関(heteroscedasticity and autocorrelation=HAC)調整済み推計値のどれかを使って標準誤差を調整する。中でもNewey and West(1987)がファイナンス研究では良く使われる。


 E[β^Jol - βJ] = (1/T)[ J(1+ρ) + 2ρ{(1-ρJ)/(1-ρ)} ](σuvv2)

小サンプルのHAC標準誤差の推計に付き纏う問題のため、研究者たちは、(2)式の重複回帰の代替策として、非重複回帰や、1期のリターンの予測変数のラグ和Σj=1...JXt-jへの回帰(Jegadeesh (1991)/Hodrick (1992))や、(1)の構造が含意する長期の係数を用いることがある。本稿では、それらの長期回帰の派生形についても小サンプルバイアスを導出した。幾つかの興味深い結果が得られたが、中でも驚くべきは、すべてのρとJについて、小サンプルバイアスの方が重複/非重複回帰の話よりも深刻であった、ということである。また、人気のあるJegadeesh (1991)/Hodrick (1992)の代替策にも深刻なバイアスがあった。

*1:cf. ここ