What drives economic growth? Why are some regions wealthier than others? How does climate change affect the economy?
These fundamental questions continue to concern economists, politicians, and scientists. In answering them, new questions often arise, such as: what is the correct scale for observing and measuring changes in economic productivity? When studying the economic impacts of climate change, what role does spatial granularity play in determining our findings?
Granularity as a concept refers to the extent to which a large entity is composed of different pieces. In the context of spatial analysis, this usually relates to grid-cell size, or the various levels of administrative boundaries in a given area, which could be national, regional or state, county, municipal, or even neighborhood level. Drivers of economic growth at the national level may differ from the drivers of economic growth at the neighborhood scale. Weather and climatic conditions are also experienced locally and can even differ over just a few kilometers. As such, granularity is an important consideration when designing an economic study or analysis, especially when looking at the impacts of climate change.
Despite this, many studies that explore the economic impacts of climate change continue to rely on data applied to entire countries. For example, they might examine changes in temperature and precipitation and relate this to changes in economic output, averaged over a nation. On the one hand, such analyses are common because national indicators like Gross Domestic Product are generally the most accessible forms of economic data, and are available for almost every country including with sectoral detail. On the other hand, such analyses feel immediately problematic when we think of extremely large countries like Russia, China, or the United States where there are many different economic and climate zones. The climate effects felt in a dry region like Phoenix, Arizona will differ vastly from those felt in a much cooler region, for example, Portland, Oregon. Mitigation and adaptation measures can also vary a lot within a given country. A national assessment using average temperatures or precipitation would surely obscure substantial within-country heterogeneity.
This diversity in local conditions holds up for many other aspects of society beyond just climate, from political beliefs and ethnicity to land use. These differences are reflected in the multitude of sizes and sectoral makeup of various regional economies within a country. There is beauty to this diversity, and we should make use of it to better understand our economies and how they respond to shocks - climate or other.
In our most recent publication, we echo these sentiments. We advocate the need for more granular economic data, which we believe is a powerful tool for unveiling economic trends that operate at scales finer than national.
As our main contribution, we provide DOSE, a global, sub-national database of reported economic output from 1960 onwards for more than 1,600 regions across the globe. For most regions, we also provide a sectoral breakdown of output, divided into three categories: agriculture, manufacturing (industry), and services. The key strength of our database beyond its spatial and temporal coverage is the fact it only consists of “official” (reported) values that we collected from various national statistical offices and supranational bodies like EuroStat or OECD. There are naturally some gaps in the data, but we have not estimated anything ourselves and refrained from using interpolation to fill them, unlike other similar data products. We demonstrate a simple use of our data in the figure shown above. In the upper panel, one can clearly identify within and between-country differences in output for a specific year, while the lower 3 panels tell us more about the sectoral breakdown of these regional economies.
Each row of our database represents a reported annual observation for a given region in a specific year with a whole range of “ready-to-use” processed data: real and nominal per capita income (gross regional product) both in aggregated form and broken down into the three aforementioned sectors, in local currency and US dollars. We also provide regional average temperature and precipitation data for each annual observation. Finally, we include a unique identifier for each region that facilitates the matching of DOSE with spatial boundary data. From here, users can link DOSE to other geospatially referenced datasets.
In providing economic data at a more granular, sub-national level, we hope to enable future research that unveils economic intricacies both between and within countries. Using methods like fixed effects regression, researchers can use our data to isolate the effects of a particular phenomenon on the economy whilst accounting for differences in time-invariant conditions across space. We hope the insights discovered with our product will aid in the development of more targeted economic policies and campaigns, particularly in the context of climate change. We aim to regularly expand DOSE both in time and over space.