How to Incorporate the Costs of Weather- and Climate-Related Natural Disasters into Measurements of Economic Growth

Weather- and climate-related costs have risen significantly in the United States over the last four decades. Many of these costs are associated with natural disasters, including devastating wildfires and floods. These costs can be due to loss of assets, as when a home burns in a wildfire, or loss of income, as in crop losses from heat and drought. The largest U.S. natural disasters are tracked by the National Oceanic and Atmospheric Administration (NOAA), which monitors disaster events costing $1 billion or more in real terms.¹ These billion-dollar-plus disasters increased from an average of three disasters per year between 1980 and 1989 to 17 per year between 2014 and 2023.

To take the house example, faster losses of housing means we need to build more housing just to maintain the existing stock. That is, net investment in housing is lowered by the faster losses. Technically, at the national level of data, these greater losses add to depreciation, which is the rate at which we expect investments to disappear over time. But currently, the higher rate of disasters is not included in the data, even though natural disasters reduce the value of capital assets and in turn lower societal well-being. Leonard Nakamura, emeritus economist with the Federal Reserve Bank of Philadelphia, and Brian Sliker, who worked as an economist with the Bureau of Economic Analysis (BEA), address these deficiencies in the national accounting of natural disasters in their paper, "Climate Shocks in the Anthropocene Era: Should Net Domestic Product Reflect Climate Disasters?"²

Disasters are a problem because losses from them vary a lot. Natural disaster asset losses, the authors explain, are input into one of two categories in the system of national accounts: "consumption of fixed capital" (the current technical term for depreciation) or "other changes in the volume of assets." The choice of where to place the depreciation figures can have big implications for interpreting changes in macroeconomic data from one quarter to the next. Hurricane Katrina, for example, led to large quarterly swings in national income in 2005, which clouded interpretation of the data when analyzing the health of the economy. Back then, natural disaster asset losses were placed in "consumption of fixed capital," but since 2009, these losses have been placed in "other changes in the volume of assets,"³ which does not affect net income. However, removing them completely means that the acceleration of asset losses doesn't show up in depreciation. The authors argue that the rising trend in these costs should show up, telling us that in general, assets are less permanent than they once were as our climate and weather change.

To calculate this new trend, they used NOAA's Billion-Dollar Weather and Climate Disasters (BWCD) dataset from 1980 to 2023.⁴ Much of the information in the dataset comes from the Federal Emergency Management Agency and private insurance companies, although many other sources are also used.⁵ The disaster data are divided into seven types: drought, flood, freeze, wildfire, and three varieties of storms (tropical cyclone, winter, and severe). The costs captured in the BWCD dataset include damage to buildings, vehicles, public infrastructure (roads, bridges, and levees), electrical infrastructure, offshore energy platforms, and agricultural assets, as well as wildfire suppression costs.

Between 2014 and 2023, the average annual cost of large disasters in the United States, based on the BWCD dataset, was $109 billion — equivalent to 0.45 percent of gross domestic product (GDP). Because these figures exclude natural disasters that cost less than $1 billion, the true impact of weather- and climate-related disasters is even larger.

To account for the costs of billion-dollar-plus natural disasters in national accounting, the authors estimated the depreciation and measured its impact on net domestic product and related aggregate statistics for the period from 1980 to 2023. (Net domestic product is defined as GDP minus depreciation of capital goods.) The authors used an econometric technique called Poisson pseudo-maximum-likelihood regression to facilitate their analysis.

The monetary costs of billion-dollar-plus weather- and climate-related disasters, they show, rose substantially between 1980 and 2023, with the trend in costs rising 4.9 percent annually in real terms. This translates to an approximate doubling of the costs of these natural disasters every 14 years. The costs of these disasters, they note, vary greatly from year to year, but their overall upward trend over the last 44 years is pronounced. "Much of this trend in costs is likely due to climate change," they suggest, but without a detailed study, it is not possible to attribute the increase in the costs of large natural disasters specifically to climate change. Instead, the authors focused on the costs of these large disasters and how measures of capital depreciation can be improved to enhance national accounting.

Adding the weather- and climate-related cost trend into depreciation calculations leads to significantly faster depreciation of real assets: They found it reduced the level of net domestic product by 0.4 percent and had a slightly negative impact on net domestic product's growth rate. However, when they also included the residual shocks component (in other words, the short-term variations in disaster costs) in their calculations, it led to abrupt changes in net domestic product. They argue the residual should go into "other changes in the volume of assets."

Using the authors' new approach, the trend in costs of natural disasters could be reflected in net national product without introducing significant short-term swings in the data, which can occur because of the unpredictability of natural disasters. Their results may lead to improved measures of aggregate economic activity and thus societal well-being.⁶