Effects of sea level rise on economy of the United States

ABSTRACT

We report the first ex post study of the economic impact of sea level rise. We apply two econometric approaches to estimate the past effects of sea level rise on the economy of the USA, viz. Barro type growth regressions adjusted for spatial patterns and a matching estimator. Unit of analysis is 3063 counties of the USA. We fit growth regressions for 13 time periods and we estimated numerous varieties and robustness tests for both growth regressions and matching estimator. Although there is some evidence that sea level rise has a positive effect on economic growth, in most specifications the estimated effects are insignificant. We therefore conclude that there is no stable, significant effect of sea level rise on economic growth. This finding contradicts previous ex ante studies.

KEYWORDS:

• Barro type growth regression
• climate change
• economic growth
• sea level rise
• spatial autoregressive model
• USA counties

Acknowledgments

We would like to express gratitude to an anonymous referee who provided us with a useful feedback and constructive inputs and comments. At the Department of Economics at Sussex, we would like to thank to many of the members of faculty and PhD students for their valuable comments and suggestions. Among many, we expressly thank Alexander Moradi for providing us with detailed feedback, useful advises and constructive discussion.

Disclosure statement

No potential conflict of interest was reported by the authors.

Research materials

Association of Religion Data Archive (htp://wtww.thearda.com/Archive/ChCounty.asp).

Center for Operational Oceanographic Products and Services (http://tidesandcurrents.noaa.gov/about.html).

Permanent Service for Mean Sea Level, Holgate et al. 2013 (http://www.psmsl.org/data/obtaining/complete.php).

United States Census Bureau (http://www.census.gov/support/USACdataDownloads.html).

United States Census Department of Agriculture (https://www.ers.usda.gov/data-products/rural-urban-continuum-codes).

Notes

1. This method is not the same as the typical 3SLS used for estimation of simultaneous equations models, which is described for example in Greene (2002). Therefore, the residuals do not need to be corrected as in case of typical 2SLS or 3SLS (expect of adjustment for heteroscedasticity, which we discuss in Section 5.1 and adjustment for spatial patterns which we discuss below).

2. We also tried other matching algorithms besides the propensity score matching. These include Mahalanobis distance and its generalization, where the optimal weights of each covariate are found by a generic search algorithm (Diamond and Sekhon 2014). However, we obtained the best matchings (in terms of balance) applying the propensity score method, therefore we do not present results of the other matchings.

3. Assuming unequal variance in the two groups.

4. We use the same methodology as Higgins, Levy, and Young (2006) and Rupasingha and Chilton (2009). We attempted to replicate the results of Rupasingha and Chilton (2009), but we did not obtain precisely the same estimates as we do not have their dataset available. However, our estimates are not qualitatively different from those of Rupasingha and Chilton (2009) and as in their paper, some of our estimates turned out to be insignificant or having different sign than expected. These include for example per capita highway and education expenditures (viz Section 5.6).

5. We found it more sensible to choose the cut-offs 0 mm/year and 5.5 mm/year than using quantiles because the distribution of the sample sea level rise is very specific. For most counties, sea level rise is equal to zero or to a very small positive value, for few cases it is extremely high and for even fewer cases it is negative and close to zero.

Additional information

Funding

This research was financially supported by CEC DG Research and Innovation through the RISES-AM project [grant number 603396].