The short-term effects of the war on Ukraine’s economic activity

Editors’ note: This column is part of the Vox debate on the economic consequences of war.

The recent full-scale Russian aggression in Ukraine and the subsequent introduction of martial law have imposed severe restrictions on data collection and processing by Ukraine’s state statistics agencies. The lack of timely information makes it difficult to assess the magnitude and pace of change in the Ukrainian economy and therefore has an impact on policy-making. Timely assessment of human and capital losses1 Valuable input from ongoing discussions on optimal post-war restructuring strategies, including estimates of GDP contraction (Garikano et al. 2022, Baker et al. 2022).

Figure 1 Region GDP vs Twitter by region (2019)

To understand the short-term effects of the war on economic activity, we combine high-frequency predictions of gross regional products in the Ukrainian region. Our predictions are based on variables and specifications identified in the literature to accurately track GDP during catastrophic events such as natural disasters or armed conflict: night light intensity, online search behavior, and social media activity.

Our preliminary analysis shows that at the beginning of the war economic activity fell by about 55% from its pre-war level but recovered by about 85% in April (Konstantinescu et al. 2022).

Proxy of economic activity

We build on development based on recent advances in economics, which combine obsolete big data with changes in economic activity. Although incomplete, they are widely used to supplement, improve, and even replace conventional systems of economic activity (for example, during natural disasters) when they are in short supply.

In large countries with diverse economic bases, regional considerations of relevant local shocks add important details to understanding overall developments (Fingleton and Szumilo 2019). It is also noteworthy that neither the position nor the timing of military activity on either side of the conflict is likely to be outward for economic activity. Since 30% of Ukraine’s GDP comes from Kiev and Kiev Oblast, but less than 2% from Kherson Oblast, regional activity is crucial for accurately assessing the macro effects of war. A conventional war is inherently a geographical crisis, with losses of varying intensity over time and spreading unevenly in space. Therefore, to track its effects (even on a macro scale) requires geographically isolated analysis at high frequencies. These dimensions, with limited data available, determine the potential space for potential nookcasting and forecasting solutions.

In the context of a macro analysis, high frequency indicators are more difficult to interpret than volatile and established statistical data. Nevertheless, they act as reliable proxies in emergencies, as indicated during the Covid-19 epidemic when the use of noisy but high-frequency data became a major advantage (Woloszko 2020, Chetty et al. 2020).

Figure 2 National GDP of Ukraine: Annual 2012-2021 and Monthly January-April 2022

Night light

Nightlights have been used successfully to estimate economic activity in both national (Chen and Nordhaus 2011, Henderson et al. 2011, 2012) and sub-national scale (Sutton et al. 2007, Doll et al. 2006). If night lighting is a normal good, then the change in its cost reflects the change in income. Since the intensity of light can be regularly measured from space, it is a popular tool through which advanced methods are available. The main concern when using this method is that remotely felt nightlight data may be affected by the weather so they cannot be used reliably at very high frequencies. During wartime, night lights may be less reliable for some additional reason. First, the lights may be turned off to avoid the target.2 Second, fires from enemy bombings do not reflect economic activity while the smoke they emit may obscure other activities. Third, the armed forces are strategically using light to confuse their opponents. Although well-established, we consider this method to be the least reliable for our work in this context and have the potential to devalue economic activity during the war.

Figure 3 Actual GDP varies between 2021 and March 2022

Google Trends

Ettredge et al. (2005) kickstarted, and Choi and Varian (2009) further established, the relevance of Google search engine data in the near-term forecasting of a wide range of casting and economic variables. Recent applications to gauge GDP nationally are found in Goetz and Knetsch (2019). For any search term (or a group of words like ‘subject’ or ‘category’), Google reports the popularity of requests received for this search over other searches, and a regional breakdown is also available. One problem with this approach is that it works on shares rather than search queries. This means that during war it can be biased by changing the total number of searches.


Transient social media interactions serve as a valuable GDP proxy in both cross-country applications as well as fine geographical granularity as indicated in Indaco (2020) and Ortega-Bastida et al. (2021).

We use a number of Twitter posts that have a picture and geographical location matching the Ukrainian Oblast administrative boundaries. Oblast-level shares of tweets (out of total tweets) are positively correlated with oblast-level GDP shares (outside national GDP), as shown in Figure 1. Twitter users use such posts to communicate explicit costs so changes in this activity are highly correlated with changes in revenue and consumer trends. The advantage of this approach is that, unlike Google trends, it responds to changes in the local population.

Since pre-war Ukrainian regional GDP is measured annually, we are limited by the data to estimate the suit of our model in the annual frequency. The model specification is guided by the literature quoted above. All the values ​​we reported have been adjusted for inflation and marked in 2004 hryvnia.

Economic activity in 2022

Figure 2 shows the data collected at the national level. It shows measured GDP and sample forecasts for 2012-2020, as well as non-sample forecasts for 2021 and 2022 (2022 data is plotted monthly showing an annual value for each month). The best in-sample performing model employs Google Trends and Twitter and tracks measured GDP before 2021. It accurately predicts 3.5% of GDP growth in 2021 (close to the standard measured in national data) by the Bank of Ukraine). Although the 2022 monthly data probably contains a lot of noise, they suggest an intuitive decision: economic activity experienced a dramatic push in March (-45% compared to 2021) but recovered in April (-15% compared to 2021). Adding nightlights and static effects of the region to our chosen specification makes little difference, but the use of nightlights alone suggests that economic activity decreased by about -40% between 2021 and April 2022, which is our low-threshold forecast. Overall, there are two sources of change: (1) changes in the territories controlled by Ukraine and (2) changes in activity in the controlled areas. Figure 3 shows the changes in economic activity that contributed to the national total of different regions between 2021 and March 2022. In March, the occupied territories made no contribution (in red), the regions that were not directly affected (yellow) contributed slightly less than in 2021 but some central and western regions (green) experienced higher levels of activity. The anecdotal evidence suggests that extensive internal migration and temporary strong migration may be responsible for any positive effects.

Figure 4 Daily GDP Model: Google Trends

Our final image uses only Google Trend data but increases the frequency of daily monitoring (Figure 4 shows the annual daily GDP level). Although the data is volatile, this practice is useful for showing two important points. First, it suggests that in the early days of the war, the level of economic activity dropped sharply by about 40%. Second, economic activity increased at the end of March. At the beginning of April, it was about 80% of the average level recorded in 2021. Further analysis shows that a large portion of the growth at the end of March came from Kiev City and Kyiv Oblast. Since these two areas are important for the country’s GDP, winning the Kiev war has had very significant economic consequences. Our indicators further suggest that activity in Russian-occupied territories has increased since the invasion before 2014 and was higher in 2022. We emphasize that these results reflect the situation at the end of April 2022 and may vary depending on how the war begins.

Author’s Note: The opinions expressed in this column do not necessarily reflect the views of the authors and the National Bank of Ukraine.


Becker, T, B Eichengreen, Y Gorodnichenko, S Guriev, S Johnson, T Mylovanov, K Rogoff and B Weder di Mauro (2022), “A Blueprint for the Reconstruction of Ukraine”,, 7 April.

Chen, X and W. Nordhaus (2011), “Using Enlightened Data as Proxies for Economic Statistics”, PNAS 108 (21): 8589-8594.

Chetty, R., JN Friedman, N. Hendren, M. Stepner, and The Opportunity Insights Team (2020), “How have Covid-19 and sustainability policies affected spending and employment? A New Real-Time Economic Tracker Based on Private Sector Data ”, NBER Working Paper.

Choi, H and H Varian (2009), “Predicting the Present with Google Trends”, Google Inc.

Constantinescu, M, K Kappner and N Szumilo (2022), “Estimating the Short-Term Impact of the War on Ukraine’s Economic Activity”, unpublished manuscript.

Doll, CNH, JP Muller and JG Morley (2006), “Mapping Regional Economic Activity from Night Light Satellite Images”, Environmental economy 57: 75-92.

Ettredge, M, J Gerdes and G Karuga (2005), “Using web-based search data to predict practical economic statistics”, Communication of ACM 48 (11).

Fingleton, B and N Szumilo (2019), “Simulating the Impact of Transport Infrastructure Investment on Wages: A Dynamic Spatial Panel Model Approach”, Regional sciences and urban economics 75: 148-164.

Garicano, L, D Rohner and B Weder di Mauro (2022), “The Economic Consequences of the War”, Debate.

Goetz, TB and TA Knetsch (2019), “Google data on the bridge equation model for German GDP”, International Journal of Forecasting 25: 45-66.

Henderson, V, A Storeygard and DN Weil (2011), “A Brilliant Idea for Measuring Economic Growth”, American Economic Review 101 (3): 194-99.

Henderson, V, A Storeygard and DN Weil (2012), “Measuring Economic Growth from Outer Space”, American Economic Review 102 (2): 994-1028.

Indaco, A (2020), “From Twitter to GDP: Estimating Economic Activity from Social Media”, Regional sciences and urban economics 85: 103591.

McLaren, N and R Shanbhogue (2011), “Using Internet Search Data as an Economic Indicator”, Bank of England Quarterly Bulletin 2011 Q2: 134-140.

Ortega-Bastida, J, AJ Gallego, JR Rico-Juan and P Albarran (2021), “A Multimodal Method for Regional GDP Forecasting Using Social Media Activity and Historical Information”, Applied Soft Computing 111.

Sutton, PC, CD Elvis and T. Ghosh (2007), “Estimates of Gross Domestic Product on a Sub-National Scale Using Nocturnal Satellite Images”, International Journal of Ecological Economics and Statistics 8 (S07): 5-21.

Woloszko, N (2020), “Tracking GDP Using Google Trends and Machine Learning: A New OECD Model”,, 19 December.




Leave a Reply

Your email address will not be published.