Satellite data have been used to observe Earth phenomena in a wide range of scientific disciplines, but recent advances in access to satellite data and in the technology required to process them are expanding their use in the social sciences, especially for disadvantaged countries that lack in the capacity to produce statistics or the types of statistics available. Satellites observe the world on the same basis, at regular intervals, and with detailed geographic coverage.

North Korea is one of the most underdeveloped countries in the world in terms of statistic data. It publishes very limited economic statistics, and while the outside world may provide estimates of GDP and other statistics, the reliability of these estimates is often questioned due to the lack of raw data. Therefore, recent attempts have been made to analyze the North Korean economy using satellite data.

This study makes a new contribution to the literature on North Korea’s economy using satellite data in two ways. First is the object of the study. Existing studies utilizing satellite data usually analyze at the regional and national levels. Considering the importance of enterprises and the abundance of satellite data and spatial information on enterprises, this study uses satellite data on 179 major enterprises in North Korea to create indicators to measure corporate activities, which are then aggregated at the sub-industry level to create industry indicators. In North Korea, microdata at the firm level is almost non-existent except for the database on frequency of news reports. Industry level statistics are also limited to Bank of Korea’s estimates on GDP by high-level industry classification, namely light and heavy chemical industries for manufacturing, and output of a few items, making it difficult to analyze detailed industries. This study utilizes satellite data on major North Korean companies to generate firm-level indicators of entrepreneurial activity, which are then aggregated at the industry subcategory level to generate industry-level indicators for 41 industries.

Second, this study utilizes new satellite data. Previously, most of the satellite data used to observe North Korea’s economy were based on nighttime light levels. This is because nighttime illumination is widely used and reliable as an indicator of income level in the field of economics due to the close correlation between electricity consumption and economic activity. However, the usefulness and reliability of nighttime light has been questioned in North Korea’s unique economic environment. This is because nighttime light observations are made around 1 a.m., when economic activity in North Korea is unlikely to be active, and electricity consumption may not necessarily reflect economic demand when considering the authorities’ monopoly on electricity distribution.

In this study, we aimed to observe the North Korean economy by utilizing land surface temperature captured during the daytime and SAR satellite imagery which can be observed regardless of weather conditions, in addition to the nighttime light data. The surface temperature data was used to index the temperature of the production sites of enterprises, and the SAR satellite images were used to measure the changes in goods loaded in outdoor areas within the enterprise area. Nighttime light data was used to measure the amount of light generated by enterprises at night. As all three are generated by different methodologies utilizing different satellite data, the indicators are expected to reflect different aspects of North Korean enterprises and industries by observing different aspects of firm activities, such as disparities between night and day, inside factories and outdoor areas.

The use of satellite data to generate economic statistics is a recent development and there is no established methodology. It is also the first time that such an attempt has been made for North Korean firms and industries. Therefore, this study evaluates the explanatory power and limitations of the satellite-based economic indicators generated in this study at the firm and industry level. At the firm level, we evaluate the alignment of satellite-based firm indicators with production and investment activities and the macroeconomic environment as reported by North Korean media for nine major North Korean factories. At the industry level, we generated industry indicators by weighting and averaging the enterprise indicators across sub-categorized industry units, and compared them with economic statistics such as trade and output estimates.

The organization and main content of each chapter is as follows. First, Chapter 2 reviews previous studies on economic analysis using satellite data, especially attempts to measure economic activity, and describes the research methodology and data used in the literature. Four steps were taken to derive satellite-based economic indicators: first, target firms were selected, spatial information of firms was acquired and overlaid with satellite data, and economic indicators were generated by processing satellite data at the firm level. As described above, three types of indicators were generated: the “temperature gap” metric using temperature is the average temperature gap between the production and non-production areas of a company’s premises. The "illuminance gap" metric, which uses nighttime light levels, subtracts the national average nighttime light level from the average nighttime light level of the enterprise to control for the effects of power supply at the national level. Finally, using SAR satellite imagery, we created a "load change" metric, which is the percentage of the outdoor area of a company’s property that shows a change in visible goods between two consecutive satellite images.

We then examined and briefly characterized the changes in the distribution of the three indicators (temperature gap, illuminance gap, and load change) by time, industry, and company size. In the case of temperature gap, the level of temperature gap progressively decreased throughout normal years (2013-2016), the UN sanctions period (2017-2019), COVID-19 period (2020-2022), and Post-COVID period(2023) in the lower-end companies, while in the upper-end companies, there was no significant change in the UN sanctions period and a decrease in temperature gap was observed in the COVID-19 period. For illuminance gap, the level of illuminance gap consistently improved across the four periods, especially for larger companies. In terms of load change, the lower-ranked firms showed similar characteristics to the temperature gap as the level of load change decreased over time, while the higher-ranked firms showed mixed changes during the UN sanctions and COVID-19 periods, but a consistent increase in the area of load change during the post-COVID period. By industry, the temperature gap was higher in the heavy chemical industry, including primary metals and transportation machinery, while the illuminance gap was higher in the transportation machinery, electrical and electronic, and chemical industries, and load change was higher in the transportation machinery, machinery, and electrical and electronic industries. Within industries, the variation between firms was higher in light industries such as textile and apparel and food and beverage services. Finally, by firm size, temperature gap was strongly positively correlated with firm size, while illuminance gap and load change were not.

Chapter 3 examines the explanatory power of the three satelliteindicators for nine major enterprises: Kimchaek Iron and Steel Complex, Cheonlima Steel Complex, Heungnam Fertilizer Corporation, Namheung Youth Chemical Complex, Taean Heavy Machinery Complex, Sinuiju Textile and Chemical Complex, Hamhung Disabled Soldiers’ Essential Plastic Goods Factory, Kim Jong Suk Pyongyang Textile Factory, and Pyongyang Goksan Factory. The extent to which the satellite indicators explained the production and investment activities of major enterprises varied across enterprises and time periods, and was not complete. Nevertheless, certain conclusions can be drawn: temperature gap is related to the production of enterprises, especially in the heavy chemical industry. The illuminance gap tended to increase during periods of large-scale capital investment and construction, suggesting that it is generally related to investment activities. In the case of Kim Jong Suk Pyongyang Textile Factory and Pyongyang Goksan Factory, it was also consistent with production activities. In the case of load change, unlike other indicators, it was difficult to find a consistent relationship with business activities due to the short time period (2017-2023), but in the case of the Taean Heavy Machinery Complex and the Hamhung Disabled Soldiers’ Essential Plastic Goods Factory, we found a relationship with business activities. On the other hand, we also detected a possibility of measurement error due to the influence of temperature and illumination in the surrounding area, and upward and downward rigidity of the temperature gap variable, among other factors.

Chapter 4 examines the explanatory power of satellite indicators compared to economic statistics at the industry level. We examine the correlation between economic statistics and satellite-based industry indicators for a total of 16 industries for which we have output estimates at the industry subdivision level, or for which we can construct trade statistics on intermediate goods imports, final goods imports, or final goods exports related to the industry. Only the temperature gap and illuminance gap variables generated industry-level indicators, while load changes were not used in this chapter due to the small number of firms for which data were available.

The analysis shows that temperature gap is correlated with economic statistics at least moderately for nine heavy chemical industries and four light industries, with higher correlation coefficients for heavy chemical industries. In the case of illuminance gap, economic statistics and satellite-based industrial indicators were correlated for two heavy chemical industries and one light industry, showing a lower correlation with industrial production than temperature gap. To assess the overall explanatory power of satellite-based industrial indicators, we conducted a fixed-effects model analysis of the correlation between trade statistics and temperature and illuminance. The results show that each of the temperature and illuminance gaps are significantly and positively correlated with the economic statistics, but only the temperature gap is significant when considered together, suggesting that the temperature gap is more relevant to production. Also, when separating heavy and light industries, the positive relationship is significant only for heavy industries.

The above results show that the satellite-based enterprise and industry indicators derived in this study reflect economic activities at the enterprise and industry level to some extent. However, there are limitations to the satellite-based economic indicators developed in this study that prevent them from accurately measuring North Korea’senterprise and industrial production. We found a number of limitations in the satellite data itself, such as glare in the nighttime data, and in the methodology for calibrating the satellite data and derivingindicators, such as making considerations for firm size and seasonality. Therefore, it is recommended that the indicators developed in this study be used as a supplementary data source. Nevertheless, given the scarcity of North Korean economic statistics, this study is significant in that it presents a new methodology for analyzing the North Korean economy using satellite data and confirms certain explanatory power. Further development of the methodology can be expected to expand our understanding of the North Korean economy.