Local labour market impacts of climate-related disasters: a demand-and-supply analysis

ABSTRACT

Using US county-level data from 1991 to 2015, the labour market impacts from various climate-related disasters are examined. It is found that different disasters have statistically and economically different impacts on local employment and wage. The standard economic demand–supply analysis can help to understand the heterogeneous impacts of disasters on these labour market outcomes.

KEYWORDS:

• labour market
• climate-related disasters
• principal component analysis (PCA)
• spatial Durbin error model (SDEM)

JEL:

• Q54
• R11

ACKNOWLEDGEMENTS

The authors thank Professor Paul Elhorst, guest editors of this special issue, and two anonymous reviewers for valuable comments and suggestions that greatly improved the manuscript. The authors are thankful to the participants at Southern Economic Association Annual Meetings, Southern Regional Science Association Meetings and NE1749 for helpful suggestions.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

Notes

1 Census data are available for 1990, 2000 and 2010. Social capital data are available for 1990, 1997, 2005, 2009 and 2014. We use the most recent census or social capital data in the panel regression equation (1 and 2).

2 It includes the number of establishments in religious, political, social, business, professional, labour, sports and recreational organizations.

3 The recent literature suggests that including lag terms of the endogenous variable may not fully address the problem. Reed (2015) suggests that using more lagged terms of the endogenous variables as instruments can address the issue of potential bias caused by endogeneity. We also used two lags of the endogenous variables as instruments and ran a two-stage least square (2SLS), and the results remain similar. Unfortunately, we could not implement the same approach with the SDEM model on panel data.

4 The main source of this data set is the ‘Storm Data and Unusual Weather Phenomena’ from the National Climatic Data Center (NCDC). Owing to the NCDC’s change in reporting procedures, every event listed in its storm data set that had exact damage values assigned was entered into the database from 1995 onwards. However, only selected events with property or crop damage greater than US$50,000 were recorded from 1990 to 1995 (HVRI, 2013).

5 Natural disasters are often multi-county in scope. The correlation between D_it (disaster in county i) and VD_it (disaster in neighbouring counties) is high for some natural disasters such as drought, heat, hurricane and winter weather. For the rest of the disasters, the correlation coefficients vary between 0.40 and 0.65. These values also depend on the spatial weight we used to establish the spatial relationship between counties. Considering these high correlations, the estimated spillover effect may also capture the effect of disaster severity, as severer disasters are more likely to affect multiple neighbouring counties. Given the data’s limitations, it is impossible to distinguish further the true spatial spillover effect from this effect of disaster severity.

Additional information

Funding

This work was funded by the National Institute of Food and Agriculture (NIFA) [grant award number 2017-68006-26233].