IMF：超越年平均值：季节性气温对美国各县就业增长的影响（英）.pdf

IMF Working Papers describe research in progress by the authors and are published to elicit comments and to encourage debate. The views expressed in IMF Working Papers are those of the authors and do not necessarily represent the views of the IMF, its Executive Board, or IMF management. 2023 JUN Beyond the Annual Averages Impact of Seasonal Temperature on Employment Growth in US Counties Ha Nguyen WP/23/142 I thank Rabah Arezki, Bas Bakker, Adolfo Barajas, Rudolfs Bems, Andrew Berg, Mai Dao, Mercedes Garcia-Escribano, Hui He, Toan Phan, Koralai Kirabaeva, Vladimir Klyuev, Anton Korinek, Ruy Lama, Emanuele Massetti, Rodolfo Maino, Joe Procopio, Nooman Rebei, Nikola Spatafora and Maryam Vaziri for very helpful comments and feedback, and Ruchun Li for editorial help. I am grateful to Berkay Akyapi and Emanuele Massetti for introducing to me climate data via Google Earth Engine. 2023 International Monetary Fund WP/23/142 IMF Working Paper Institute of Capacity Development Beyond the Annual Averages Impact of Seasonal Temperature on Employment Growth in US Counties Prepared by Ha Nguyen* Authorized for distribution by Mercedes Garcia-Escribano June 2023 IMF Working Papers describe research in progress by the authors and are published to elicit comments and to encourage debate. The views expressed in IMF Working Papers are those of the authors and do not necessarily represent the views of the IMF, its Executive Board, or IMF management. ABSTRACT Using quarterly temperature and employment data between 1990 and 2021, this paper uncovers nuanced evidence on the impact of seasonal temperature within US counties higher winter temperature increases private sector employment growth while higher summer temperature decreases it. The impacts of higher temperature in mild seasons, fall and spring, are statistically insignificant. Moreover, the negative impact of higher summer temperature persists while the positive impact of higher temperature in the winter is more short-lived. The negative effects of a hotter summer are pervasive and persistent in many sectors most significantly in “Construction” and “Leisure and Hospitality” but also in “Trade, Transport, and Utilities” and “Financial Activities.” In contrast, the positive effects of a warmer winter are less pervasive. The employment effect of a hotter summer has been more severe in recent decades. RECOMMENDED CITATION Nguyen, H. 2023. Beyond the Annual Averages Impact of Seasonal Temperature on Employment Growth in US Counties. IMF Working Papers, 2023/142 JEL Classification Numbers C33, C55, E24, O44, Q54 Keywords Climate change; temperature; employment; US counties Author’s E-Mail Address Hnguyen7imf.org WORKING PAPERS Beyond the Annual Averages Impact of Seasonal Temperature on Employment Growth in US Counties Prepared by Ha Nguyen1 1 I thank Rabah Arezki, Bas Bakker, Adolfo Barajas, Rudolfs Bems, Andrew Berg, Mai Dao, Mercedes Garcia-Escribano, Hui He, Toan Phan, Koralai Kirabaeva, Vladimir Klyuev, Anton Korinek, Ruy Lama, Emanuele Massetti, Rodolfo Maino, Joe Procopio, Nooman Rebei, Nikola Spatafora and Maryam Vaziri for very helpful comments and feedback, and Ruchun Li for editorial help. I am grateful to Berkay Akyapi and Emanuele Massetti for introducing to me climate data via Google Earth Engine. IMF WORKING PAPERS Beyond the Annual Averages Impact of Seasonal Temperature on Employment Growth in US Counties 2 Contents I. Introduction 4 II. A Theoretical Framework 6 III. Data and Empirical Specification 9 Data 9 Empirical Specifications . 10 IV. Main Findings . 11 Annual Regressions . 11 Main Findings . 12 V. Summer and Winter Impacts Across State’s Climate . 15 VI. On the Mechanisms of the Summer Temperature Effects . 18 VII. On the Mechanisms of the Winter Temperature Effects . 21 VIII. Effects of Temperature by Decade . 23 IX. Robustness Checks . 26 Not Using County Employment Weights 26 Dropping Extreme Employment Growth . 27 Dropping Recession Quarters 27 Controlling for Natural Disasters 28 Controlling for Precipitation 29 X. Conclusions 30 References . 30 Figures Figure 1 Dynamic Impact on YoY Employment Growth to a One Degree Fahrenheit Higher Temperature 14 Figure 2 Average Summer Impact by State . 16 Figure 3 Average Winter Impact by State 17 Figure 4 The Effect on Employment Growth of Higher Summer Temperature by Sector 19 Figure 5 The Effect of Higher Winter Temperature by Sector 21 Figure 6 Average Employment Shares in the Summer and Winter in a County 23 Figure 7 Average Annual Increase in Summer Temperature by State over 1990 and 2021 . 24 Figure 8 Dynamic Impact of YoY Employment Growth to a One Degree Fahrenheit Higher Temperature Regressions are Unweighted 26 IMF WORKING PAPERS Beyond the Annual Averages Impact of Seasonal Temperature on Employment Growth in US Counties 3 Figure 9 Dynamic Impact on YoY Employment Growth to a One Degree Fahrenheit Higher Temperature Top and Bottom 1 percentile of Employment Growth Data are Dropped 27 Figure 10 Dynamic Impact on YoY Employment Growth to a One Degree Fahrenheit Higher Temperature Employment Growth Data for Recessionary Quarters are Dropped . 28 Figure 11 Dynamic Impact on YoY Employment Growth to a One Degree Fahrenheit Higher Temperature Controlling for Natural Disasters . 29 Figure 12 Dynamic Impact on YoY Employment Growth to a One Degree Fahrenheit Higher Temperature Controlling for Precipitation . 29 Tables Table 1 Summary Statistics . 10 Table 2 Impact of Annual Average Temperature on YoY Growth of Annual Average Employment 11 Table 3 Impact of Temperature on YoY Private Employment Growth . 12 Table 4 Relationship between Employment Effect and a State’s Climate . 18 Table 5 Impact of Temperature by Decade 25 IMF WORKING PAPERS Beyond the Annual Averages Impact of Seasonal Temperature on Employment Growth in US Counties I. Introduction Climate change is the biggest challenge for humankind. Temperature is rising. The global average temperature is already about 1.2 degree Celsius higher than the pre-industrial level. Droughts, wildfires, and massive storms are starting to occur more frequently with devastating effects. Understanding the impact of rising temperature, the most basic manifestation of climate change, on economic activity is fundamental to adaptation and mitigation efforts. The economic literature has generally found that higher temperature hurts economic activity. Early literature examines the relationship between average temperature and aggregate economic variables e.g., Sachs and Warner, 1997; Gallup, Sachs, and Mellinger, 1999. It finds that hotter countries tend to be poorer. However, this relationship might be driven by omitted variables such as country institutions. Recent literature uses fluctuations in temperature within a country or a region to control for slow-moving characteristics see for example, Dell et al., 2012; Cashin et al., 2017; Colacito et al., 2019; Letta and Tol, 2019; Acevedo et al., 2020; Kahn et al., 2021.1 It finds that higher temperature reduces the economic growth of poor countries Dell et al., 2012; Acevedo et al., 2020 and the US Colacito et al., 2019. The negative effects run through reduced total factor productivity growth Letta and Tol, 2019, and reduced investment and labor productivity Acevedo et al., 2020; Kalkuhl and Wenz, 2020. Burke et al. 2015 document the non-linear effect of temperature economic growth rises with average annual temperature until around 13 degrees Celsius and drops after that. This paper examines the dynamic effects of temperature on the private sector’s employment growth at a local level, namely US county, and high frequency, namely quarterly. Going to the county and quarterly levels allows for more precise temperature measurement. Therefore, it could estimate the effects of temperature more precisely and uncover the subtle effect of seasonal temperature. This paper focuses on job growth as the main economic outcome. Jobs are featured prominently in the USs discussions of climate change mitigations. Many worry that climate change mitigation efforts will hurt jobs AFP, 2022. This paper finds that higher temperature, on average, hurts jobs in the US. Using data between 1990 and 2021, this paper discovers opposing effects of higher temperature in the winter and summer. On average, within a county, higher summer temperature reduces private sector employment growth, while higher winter temperature increases it. The impacts of higher temperature in mild seasons, fall and spring, are statistically insignificant. The findings showcase the heterogenous and nuanced effects of temperature shocks. This paper finds interesting dynamic effects of seasonal temperature. Higher summer temperature hurts economic activity in the current and following quarters. A temporary one-degree Fahrenheit F higher summer temperature decreases year-over-year YoY employment growth of that summer by 0.063 percent. It also decreases YoY employment growth of the following fall and winter by 0.08 and 0.075 percent, respectively. In contrast, the positive impact of higher temperature in the winter is more short-lived. A temporary one-degree Fahrenheit warmer winter boosts YoY employment growth in that winter by 0.05 percent but has statistically insignificant effects on employment growth in the following spring and summer. In sum, the negative impacts of 1 Also see recent surveys by Dell et al. 2014 and Auffhammer 2018 IMF WORKING PAPERS Beyond the Annual Averages Impact of Seasonal Temperature on Employment Growth in US Counties higher temperature in the summer are larger and more persistent than the positive impacts of higher temperature in the winter. Therefore, the average employment effect of higher temperature across seasons is negative. The economic literature typically examines the impact of annual average temperature on annual economic outcomes e.g., see Deschnes and Greenstone, 2007; Dell et al., 2012; Burke et al., 2015; Acevedo et al., 2020; Kalkuhl and Wenz, 2020; Akyapi et al., 2022. However, since temperature can vary greatly within a year, from freezing winters to scorching summers, this paper argues that seasonal temperature is a better approximation of weather than annual temperature.2 More importantly, the economic structures of different seasons could be very different. For example, construction, travel, and tourism are expected to rise in summer and fall in winter. Therefore, examining the effects of seasonal temperature on seasonal economic activity could offer new insights to complement the existing analyses using annual average temperature and annual-average economic outcomes. In addition, working with the country-average temperature is also not ideal since even within a country, temperature can vary greatly. A country, or even a US state, may have several climate zones. A case of localized temperature, such as at the county level, can be made here. Nevertheless, granular analyses come with their challenges and issues. First, and the most obvious issue is the lack of high-frequency economic data at the local level. One reason why employment growth is chosen as the main variable of interest is that the US has reliable quarterly data at the county level more on that in section III. The second, and more conceptual issue is labor mobility at the local level. At the country level, labor mobility is relatively restricted. At least in the short-run, workers have to stay in a country and try to find work with a temperature shock. But an analysis at the local level, such as US county, implies labor mobility is much less restricted. People could move in and out of a county to work in another county in response to a temperature shock. Therefore, the effects of temperature on employment with labor mobility can be larger than without. Deryugina and Hsiang 2017 and Colacito et al. 2019 examine the impacts of seasonal temperature at US county and state levels, respectively. However, they still use annual economic outcomes, which could mask interesting dynamic effects of seasonal temperature. This analysis complements their analyses by not focusing on annual economic outcomes but on the high-frequency impact of temperature on quarterly employment growth in US counties. By adopting this local and high-frequency empirical framework together, it unveils novel and interesting dynamic effects of seasonal temperature. It could also shed light on the mechanisms by documenting the effects in each industry and how they propagate over the next quarters. In other words, by observing temperature’s impacts on different sectors at a high frequency, instead of being diluted by the annual averages, the paper can provide additional insights into the mechanisms. This paper finds that the negative effects of a hotter summer are pervasive and persistent in many sectors most significantly in “Construction” and “Leisure and Hospitality” but also in “Trade, Transport and Utilities” and “Financial Activities.” Employment growth in these sectors may get directly hit by rising temperature. It is also possible that some of the lower employment growth is indirectly affected due to input-output linkages between different sectors or the aggregate demand effect. For example, job growth in “Financial Activities” could be dampened due to a slower financial service demand from “Construction.” In contrast, the positive effects of a 2 For example, highest daily temperature in Washington D.C. United States in 2021 ranges from the mid-30s Fahrenheit in the winter to the mid-90s Fahrenheit in the summer. The average annual temperature for Washington D.C. is about 70-degree Fahrenheit. If we use this annual average of 70-degree Fahrenheit in our analyses, we might be mistaken that Washington D.C.’s weather is more moderate while in fact, it has a cold winter and a hot summer. IMF WORKING PAPERS Beyond the Annual Averages Impact of Seasonal Temperature on Employment Growth in US Counties warmer winter are less pervasive, only in “Construction,” “Leisure and hospitality,” and “Natural Resources and Mining.” It is also more short-lived. The richness of county-level data allows for the examination of the effect by US state – which is another important contribution. This paper discovers a relationship between the negative effects of a hotter summer with a state’s summer climate hotter states have more severe negative impacts of higher summer temperature. Some cooler states e.g., Alaska and Massachusetts even benefit from the higher summer temperature. On the other hand, the relationship between the impact of higher winter temperature and a state’s winter climate is not as clear. An important point of discussion is how would these findings on short-term responses help us predict the long- term responses to hotter climates It has been argued that the short-run responses to temperature fluctuati