We examine whether economic growth at the district-level in Vietnam is accompanied by environmental degradation and climate change. We do this using high resolution night-time satellite imagery of luminosity as a proxy for economic activity, along with satellite imagery on temperature and air pollution. We find that in Vietnamese districts economic activity is accompanied by increases in temperature – regardless of whether a district is urban or rural. We also find that economic activity is accompanied by an increase in Nitrogen Dioxide (NO2) density and particulate matter (PM 2.5). These results suggest that economic growth in Vietnam may be driven by environmentally unsustainable practices stemming from industrialisation, poor waste management and fossil fuel consumption.
We conduct the analysis using several datasets. The first primary dataset is nighttime satellite image from NASA’s Black Marble catalogue. The data is available monthly from January 2012 and has a resolution of 15 arc seconds (approximately 500m × 500m at the equator). In our previous analysis, we found a strong and positive linear relationship between Vietnam’s nighttime luminosity (from satellite imagery) and district GDP. Since there is little to no GDP data at the district level, we use luminosity as a strong proxy for economic activity.
The second primary dataset is Copernicus’s ERA5 which contains the ambient temperature measured hourly from 1959 to the present. We perform aggregations to obtain the monthly average and maximum temperatures of Vietnam an the district level. The temperature variables are measured in degrees Celsius.
The third dataset contains PM 2.5 air pollution data from 1990-2019 that is granular to the second administrative level (districts) from the Global Health Data Exchange (GDHx). In order to obtain district-level values of PM 2.5, we take the average value of all pixels that fall within a district. This data is only available annually (rather than monthly).
Finally, we use district-level data on the average Nitrogen Dioxide (NO2) amount collected from January 2019 – Jul 2020, which was provided through the Big Data Observatory’s collaboration with Space4Good.
The relationship between Economic Development and Temperatures
The following images map the relationship between luminosity, average temperature and maximum temperature for Vietnamese districts for each month from January 2014 – December 2021.
Figure 1: Monthly log luminosity (left), average temperature (center), and maximum temperature (right) for districts in Vietnam from January 2014 to December 2021
We see that luminosity is concentrated in urban centers along the coastal regions of Vietnam. This pattern is mirrored in the average and maximum temperature data. However, provinces in the Southeast, Central Highlands, and Mekong Delta regions also experience high average and maximum temperatures – potentially due to their proximity to the equator. Vietnam’s Northeastern and Midland regions, where there is limited luminosity, typically experience high temperatures during summer months.
When assessing the time series of luminosity for Vietnam, we see that luminosity is at the lowest during the early summer and at its peak during the winter. This seasonality could be driven by days being shorter during winter months. As a result, this drives the consumption of artificial light in human-populated areas, especially in dense urban centers. Another key observation is that luminosity – like Vietnam’s official GDP – shows a gradual increase over time.
Figure 2: Monthly line graph of the total monthly luminosity Vietnam (January 2014-December 2021)
Figure 3: Monthly line graph of (1) monthly average temperature, and (2) monthly maximum temperature for Vietnam (January 2014-December 2021).
As seen from Figure 3, Vietnam’s monthly average and maximum temperature between 2014 and 2021 possesses strong seasonality. Yet, both metrics, unexpectedly, do not seem to be increasing over time.
To assess the relationship between luminosity and average temperature, we conduct an ordinary least-squares regression using year-month fixed effects (as a method to control for seasonality), and provincial-level fixed effects. The results (presented in Table 1 of the Appendix), indicate that higher amounts of luminosity are accompanied by higher temperatures.
According to previous literature, one reason for this could be due to urban areas having higher temperatures than rural areas. The literature suggests this is driven by urban areas having roads and buildings which absorb heat more than in vegetation-dense rural areas. To test this hypothesis, we examine the relationship between growth of luminosity and growth of temperatures. This is one method of addressing urban-rural differences as it measures whether any district (urban or rural) which experience growth in luminosity experience similar growth in temperatures. The results (presented in Table 2 of the Appendix) indicate that areas (whether urban or rural) which experience an increase in luminosity, similarly experience a growth in temperatures.
In order to further examine whether urban areas experience higher temperatures than rural areas, we plot the differences in Figure 4 below. We see that for average temperature measures there is no noticeable difference between urban and rural areas.
Figure 4: Vietnam’s monthly average temperature split by urban vs rural areas (2014-2019).
The connection to Air Pollution
Another indication of climate change and environmental degradation is air pollution. It is particularly relevant to Vietnam, whereby the economy is heavily reliant on industrialization for growth. Therefore, it is important to examine whether this relationship between luminosity and temperature could be driven by air pollution (an externality of industrial practices). Using Nitrogen Dioxide and PM 2.5, we test the assumption that an increase in economic activity leads to an increase in air pollution, which may in turn lead to increases in temperatures. .
Nitrogen Dioxide (NO2) is one of six widespread air pollutants that has national air quality standards to limit them in the outdoor air. Below, we analyse the monthly relationship between NO2 density with luminosity (economic activity) from January 2019 – July 2020 (the entire time series produced by Space4Good).
From figure 5 below we can see that luminosity and NO2 density are mostly concentrated in urban areas – in Hanoi, Hai Phong and along the east coast. We conduct a further regression analysis to assess the relationship between luminosity and NO2 density. The results (contained in Table 3 of the appendix) demonstrate a strong correlation – meaning that luminosity is accompanied by NO2 for districts in Vietnam.
Figure 5: Monthly log luminosity and average NO2 mass per area for provinces in Vietnam from January 2019 to July 2020
In addition to NO2, we examine the relationship between economic activity and particulate matter (i.e. PM 2.5). Again we find clustering of PM2.5 towards Hanoi, Hai Phong and Ho Chi Minh City. This seems to align with the clustering of higher temperatures in Vietnam.
Figure 6: (From left to right) Annual log luminosity, average temperature, and average air pollution for provinces in Vietnam from 2014 to 2019
When graphing average PM2.5 for urban and rural areas we find that urban areas experience higher measures than rural areas (see Figure 7):
Figure 7: Line graph of Vietnam’s annual average air pollution by urban and rural area (2014-2019).
Through further regression analyses (see Appendix Table 4), we find a strong relationship between luminosity and average PM2.5. This means that regions which have higher levels of economic activity experience higher measures of particulate matter. Moreover, there is a strong relationship between temperature and particulate matter, suggesting that regions which are warmer have higher measures of PM2.5.
In this study, we establish a strong and positive association between luminosity, a proxy for economic growth, and both temperature and air pollution. This relationship is even more pronounced when annual, provincial, and population density effects are accounted for. This indicates that districts which experience higher levels of luminosity (and luminosity growth), experience higher measures of both temperature and air pollution measures. We also find that, contrary to previous studies, temperatures in Vietnam are not significantly higher in urban areas than in rural areas.
Dell, Melissa, Benjamin F. Jones, and Benjamin A. Olken. Climate change and economic growth: Evidence from the last half century. No. w14132. National Bureau of Economic Research, 2008. https://www.nber.org/system/files/working_papers/w14132/w14132.pdf
Huong, Tran Thi. “Green Economic Development Policy in Vietnam in the Context of Global Climate Change.” Sociology 9.2 (2019): 45-54. https://www.davidpublisher.com/Public/uploads/Contribute/5d3f9aad9c383.pdf
Mentaschi, L., Duveiller Bogdan, G.H.E., Zulian, G., Corban, C., Pesaresi, M., Maes, J., Stocchino, A. and Feyen, L., Global long-term mapping of surface temperature shows intensified intra-city urban heat island extremes, GLOBAL ENVIRONMENTAL CHANGE-HUMAN AND POLICY DIMENSIONS, ISSN 0959-3780, 72, 2022, p. 102441, JRC123644. https://publications.jrc.ec.europa.eu/repository/handle/JRC123644
World Bank. 2022. Climate Change Risks for Sub-National Governments and Public Infrastructure Assets & Investments, https://docs.google.com/document/d/1tigdUwGqkWU4TiJHgosMo5-37ARft7a6YxkhEZUMYIo/
World Bank. 2022. Developing a new methodology to measure nightlights, https://docs.google.com/document/d/1riOzxF_XB-NfpD3QxP8WS1_WfEnd3JGYg-OsO7kI4cQ/
Table A1: Column 1 measures the relationship between log luminosity and the average temperature. Column 2 measures the relationship between log luminosity and the maximum temperature. Both models use year fixed effects (year-months are omitted for better readability).
Table A2: Column 1 measures the relationship between monthly luminosity and average temperature growths. Column 2 measures the relationship between monthly luminosity and max temperature growths. Both models use fixed effects (omitted for readability).
Table A3: The relationships between the monthly log luminosity and average NO2 mass per area . All models use fixed effects (omitted for better readability).
Table A4: The relationships between the annual (1) average temperature, (2) maximum temperature, and (3) log luminosity and average PM 2.5 . All models use fixed effects (omitted for better readability).
Figure A1: Vietnam’s monthly max temperature by urban/rural areas (2014-2019).
Figure A2: Vietnam’s monthly luminosity growth by urban/rural areas (2014-2019).
Figure A3: Vietnam’s monthly average temperature growth by urban/rural areas(2014-2019).
Figure A4: Vietnam’s monthly max temperature by urban/rural areas (2014-2019).