Worldwide, tropical cyclones (TCs) have dominated weather-related disaster damages during the past two decades (Hudson et al. 2019). The TCs develop in tropical and subtropical regions where they are referred to as hurricanes, typhoons or willy-willies (Van der Sommen, Pearson & Boggs 2018). On a yearly basis, these storms create billions of dollars in damages across many places (Chari, Ngcamu & Novukela 2020; Ishizawa, Miranda & Strobl 2019). In Southern Africa, TCs are the second-most prevalent and impactful disaster, after drought (Centre for Research on the Epidemiology of Disasters [CRED] 2019). Between 1998 and 2017, the TC became the costliest disaster, with reported losses amounting to US$1.3 billion in Southern Africa (Wallemacq, House & McClean 2018). This amounted to 46% of recorded economic losses of all hydrometeorological and geologic disasters during that period. The economic losses induced by TC include damage to or destruction of infrastructure, family assets and ecosystems (Hoque et al. 2017). They also interrupt individual or family consumption behaviour, government expenditures and investment decisions (Haque & Jahan 2016). The damages are a combined action of high destructive winds, torrential rains, landslides, storm surges and flooding (Hoque et al. 2016).

The impact of TC is a function of several factors, including the physical characteristics of the storm itself, economic characteristics of the affected place, its topography and the preparedness level of the affected community, amongst others (Bueti, Ginis & Rothstein 2014; Duvat, Volto & Salmon 2017; Kabir, Salehin & Kibria 2015). As evidenced in Seychelles, high magnitude TC (Category V storms according to the Saffir-Simpson scale) can generate strong surface winds and heavy rains that can cause huge economic damages (Duvat et al. 2017). Economically developed regions tend to suffer greater losses than the less-developed ones (Haque & Jahan 2016; Ishizawa et al. 2019). Furthermore, economies with many interindustry linkages can suffer severe disruptions which can spread to neighbouring countries (Van der Sommen et al. 2018).

In March 2019, Cyclone Idai, a category IV storm, landed in Zimbabwe through Mozambique (Mavhura 2020). It was the worst storm to hit Zimbabwe in the last five decades. The cyclone brought very strong winds of about 195 km/h (120 mph), torrential rains and floods, which caused extensive damage to public utilities, residential areas and agriculture (Munsaka et al. 2021). The water, sanitation and hygiene (WASH) infrastructure in individual homes, health and educational institutions and public buildings were extensively damaged. Amongst the damaged infrastructures, the most critical were water pumping stations and reticulation networks (World Bank 2020). The cyclone also induced landslides and mudslides, which cut off the roads and buried more than 70% of houses in the townships of Chimanimani and Chipinge districts (Chanza et al. 2020). For close to a month, some communities in Chimanimani district were isolated because of blocked roads, a situation that forced the government of Zimbabwe to airlift relief materials using the army. Other families went for more than a year living in temporary shelters without electricity, running water and other basic services (Dube, Wedawatta & Ginige 2021). To date, a couple of studies on Cyclone Idai have been carried out, focusing on what went wrong and right during the disaster in Zimbabwe (Chanza et al. 2020; Dube et al. 2021; Mavhura 2020; Munsaka et al. 2021); however, the direct economic losses of the storm have not been ascertained in order to inform disaster risk reduction (DRR) policy and practice in Zimbabwe.

Under the direction of the United Nations Office for Disaster Risk Reduction (UNDRR), an open-ended intergovernmental expert working group (OIEWG) developed methodological guidelines for determining direct disaster economic loss for Target C of the Sendai Framework for Disaster Risk Reduction (SFDRR) (UNDRR 2017). Although the OIEWG methodology was principally designed to assist countries in consistently monitoring their progress in achieving Target C of the SFDRR (UNDRR 2017), it can be adapted at a microscale to estimate the disaster economic losses and therefore inform DRR policy and practice. Furthermore, the methodology has not been widely tested in African countries. Therefore, this study explicitly tested the applicability of the OIEWG methodology by assessing the direct economic losses induced by Tropical Cyclone Idai (TCI) in Zimbabwe. Thus, the key questions driving the study were: in what ways did TCI directly affect the economy of Zimbabwe? How can the OIEWG methodology for Target C be applied to determine the direct economic losses induced by TCI in Zimbabwe? The next section will discuss some perspectives and challenges in determining economic losses from disasters.

Economic losses from disasters are either direct or indirect. Direct losses involve the ‘physical damages and destruction of physical assets’ while indirect losses are ‘interruptions of supply chain as a result of disasters’ (Okuyama 2007:116). Thus, direct economic losses are derived from tangible assets such as roads, bridges and crops, while the indirect losses are intangible and include loss of income or revenue to business interruption or missing assets (UNDRR 2017).

In many cases, the determination of economic losses in the aftermath of a disaster is performed using post disaster needs assessment (PDNA) methodologies. Depending on the scale of the disaster, two PDNAs may be conducted, a rapid needs assessment and a more comprehensive one. The rapid needs assessment is usually completed within 72 days of the disaster taking place, while a more comprehensive PDNA takes several weeks or months. The rapid assessment may not be accurate because it is usually hurriedly performed in order to provide statistics for relief, humanitarian appeals and solidarity aid (Longenecker et al. 2020; Yuan & Liu 2018). A more comprehensive PDNA provides detailed and more reliable economic loss data; however, many of the comprehensive PDNA are only calculated after large-scale disasters. Yet a significant number of small-scale and recurring events that negatively impact communities rarely receive such comprehensive assessments and documentation (Osuteye, Johnson & Brown 2017). Some under-estimations and over-estimations of the actual loss are likely to occur during PDNA (Natho & Thieken 2018). In addition, defining the timeframe at which reconstruction costs of the damaged or destroyed physical assets is attributed to the disaster(s) is a challenge.

There are many country-specific and sector-specific methodologies of estimating economic losses induced by disasters. Firstly, Ishizawa et al. (2019) used monthly night light images to calculate the economic impact of TC in the Dominican Republic. Although this method is effectively used where there is limited public data, it has one major limitation: it is mainly restricted to yearly data, partly because such night light data are usually available on yearly basis. This tends to mask most of the economic damages within the year under study (or at other higher temporal frequencies). Secondly, Haque and Jahan (2016) used an input–output economic modelling framework to estimate losses from Cyclone Sidr in Bangladesh. Although the model managed to comprehensively trace the interactions amongst various sectors of the Bangladesh economy using relatively less data, it had two major weaknesses. Firstly, the model maintained a linear and static nature and secondly, it failed to respond to changing prices of goods and services and the economies of scale experienced during production. Other technical models that are used to determine disaster economic losses include the computable general equilibrium (CGE) and the social accounting matrix (SAM) (Mainar-Causapé, Ferrari & McDonald 2018; Xie et al. 2014b). Although the CGE techniques are nonlinear and capable of factoring in changes in prices of goods and service, they are well-suited for long-run equilibrium analyses, a situation which often leads to underestimation of the disaster losses. Unlike the CGE technique, the SAM model is too rigid and requires a large amount of data in order to come up with upper bounds for the estimates (Okuyama 2007). Furthermore, both the SAM and CGE techniques hardly distinguish between direct and indirect losses.

In addition, some countries use replacement costs, that is, the cost of repairing or replacing the damaged buildings or assets with materials of almost the same kind and quality, to estimate the disaster loss (Natho & Thieken 2018). Although the replacement method might be easy to calculate, it may provide values that are more than the actual physical loss because of price increases soon after a disaster. Other countries, especially in Europe, use insurance compensation mechanisms to determine direct economic losses of a disaster in the housing sector (Surminski & Thieken 2017). However, asset insurance is rarely available or is not affordable in many developing countries, including Zimbabwe. In Europe, countries such as Germany use cereal yield and price per ton data to calculate losses in the agriculture sector (Natho & Thieken 2018). As for the roads sector, Germany uses the length of the affected paved roads as a variable to calculate direct economic losses. In addition to these methods, some advanced countries use remote sensing techniques to estimate the direct disaster-economic losses (Fan et al. 2017).

In view of the given perspectives and challenges, determining the actual economic loss is a complicated matter. While direct damage and destruction to infrastructure and assets are relatively easy to monetise (i.e. conversion of physical value into economic value), intangible costs such as loss of biodiversity, stress and inconveniences caused are difficult to measure in monetary terms (Hudson et al. 2019). Furthermore, some disasters create non-linear cascading effects over time that are hard to capture during the PDNA. For example, the Chinese Ice Storm of 2008 resulted in significant losses in the demand and supply sides of industry and commerce, which were difficult to measure when the disaster occurred (Xie et al. 2014a). In other cases, disasters bring in economic benefits through donations, remittances, relief funds and humanitarian assistance that may outweigh the costs incurred by the affected regions. For example, following the 2004 Indian Ocean Tsunami disaster, the international humanitarian response brought in untied aid of about US$13.5 bn, which exceeded the estimated losses and damages of about US$10 bn (Telford & Cosgrave 2007). Likewise, flooding in the Muzarabani district of Zimbabwe creates residual moisture, which brings many economic benefits to riparian communities who depend on off-rainy season farming (Mavhura 2017). Therefore, in such situations where disasters bring in certain benefits to the affected communities, there may not be any incentive to undertake economic loss estimates induced by the disasters. Furthermore, many developing countries do not have the capacities to consistently quantify the economic loss (Natho & Thieken 2018).

The results of this study are first presented as per disaggregated sector of the direct economic loss attributed to TCI. The disaggregation enabled understanding the needs of the specific sectors of the economy which were hard hit and therefore needed prioritisation during the recovery and rehabilitation stages. The disaggregation also enabled designing the policy implications needed for DRR. After the disaggregation, the overall economic loss (compound indicator) is discussed next.

This study explicitly adapted the OIEWG methodology for Target C of the SFDRR in order to assess the direct economic loss induced by TCI in Zimbabwe. The methodology used a sectoral approach, which included the agriculture sector, services (education, health and WASH), housing, critical infrastructure (energy, road transport and telecommunication) and cultural heritage. The findings revealed that TCI inflicted huge losses in most sectors of the economy. Nearly two-thirds of the losses were channelled through the housing and critical infrastructure, while the agriculture sector lost more than a quarter of the total estimated direct economic losses. As with many economic impact analyses of disasters (Merz et al. 2010; Okuyama 2007), the accuracy of the current estimates can be debated depending on the methodology used and availability of data. For example, the value of the assets in the damaged housing units and other buildings could not be established because of the unavailability of such data across the affected sectors. However, the sectoral analysis approach showed that Zimbabwe needs to prioritise housing, critical infrastructure and agriculture during the recovery and reconstruction phases from TCI. As Zimbabwe’s economy is strongly based on farming, which accounts for approximately 12% of the total GDP (Frischen et al. 2020), the restoration of agricultural livelihoods would contribute to the attainment of the sustainable development goals (SDGs) in Zimbabwe, especially the first three SDGs related to eliminating poverty and hunger and promoting good health and well-being. The sectoral analysis approach also underlined the need to strengthen future mitigation efforts in the housing, agriculture and critical infrastructure sectors, amongst others.

The huge economic losses from TCI were a function of several factors, including the physical characteristics of the cyclone itself, its temporal scale, economic activities of the two most affected districts of Chimanimani and Chipinge and their topographies. As a high-magnitude storm (Category IV storm according to the Saffir-Simpson scale), TCI generated strong surface winds and heavy rains that caused mudslides and rock falls, which damaged housing units, crops, road infrastructure and service sectors, amongst others (Mavhura 2020). The cyclone made its landfall just prior to the harvesting of summer crops, thereby destroying many crops such as maize and bananas. During that time, most soils were already saturated with water, conducive conditions for mudslides and rockfalls which destroyed the built environment (Chanza et al. 2020). River discharge was also high, to the extent that floods swept away roads, bridges and fields. Most of the damages occurred in densely populated townships located in mountainous places with enhanced economic activities, as well as in farms where crops were about to be harvested. In view of the high likelihood of future climate and weather-related disasters in Southern Africa (Chikoore, Vermeulen, & Jury 2015), Zimbabwe faces the risk of economic losses if it fails to adopt and effectively implement adequate DRR strategies. Strengthening DRR strategies in the built environment, including improving the building standards and retrofitting buildings using the ‘build back better’ (3Bs) principle, has been found to be critical in Southern Africa Development Community countries including Zimbabwe (Dube et al. 2021; Owusu-Sekyere, Lunga & Karuaihe 2021). Like Australia, Nepal and Sri Lanka, Zimbabwe can use the 3Bs principle as an aid in determining the post-TCI recovery, reconstruction and rehabilitation best practices (Fernandez & Ahmed 2019). Also critical is the need to strengthen cyclone forecasting and warning systems as a DRR strategy. Zimbabwe can learn from countries such as Hong Kong that have made advances in cyclone forecasting and warning systems and constructing cyclone-resilient infrastructure in order to minimise direct damages (Wong & Choy 2018).

The study used the replacement costs in order to estimate the direct economic losses. This involved calculating the cost of repairing or replacing the damaged or destroyed goods with materials similar to the original ones used before the cyclonic disaster (Natho & Thieken 2018). Such replacement costs were calculated using the OIEWG methodology developed for monitoring progress in achieving the SFDRR Target C (UNDRR 2017). In this study, the methodology worked well because it distinguished between direct and indirect losses induced by TCI. The methodology also worked well with a sectoral analysis approach where uniform indicators of damage and destruction of tangible assets were used. However, the OIEWG methodology was linear in nature and failed to factor in changes in prices of goods and services soon after the disaster. Therefore, the actual replacement estimates might have been higher than what has been provided in this study. Future studies will have to use other methods that factor in price changes from the local markets. The OIEWG methodology also provided room for double counting. For example, losses in the WASH sector were estimated independently from the housing and other services sectors, yet the WASH infrastructure is closely attached to such sectors.

As a result of limited data availability (quantity and quality), this study only calculated the direct economic losses induced by TCI. In some cases, most of the value of the assets in damaged or destroyed houses and building was not included because of insufficient data. Yet the true costs of disasters also include the value of assets in the damaged sectors and the indirect losses which arise from interruptions of the supply chain induced by the disaster (Okuyama 2007). Although calculating the value of the damaged assets places an additional burden on data collection, and indirect losses are difficult to measure, the two are critical in informing DRR policy and practice (Merz et al. 2010). This is because some disasters create nonlinear losses in the demand and supply sides of industry and commerce (Xie et al. 2014a). Hence, data collection needs to be rigorous and future studies that focus on sectoral indirect losses induced by disasters are needed in order to better inform DRR policy and practice.

This study adapted the OIEWG methodology for Target C to determine the direct economic losses induced by TCI in Zimbabwe. The methodology used a sectoral analysis approach. The results revealed that TCI inflicted huge losses in most sectors of the economy, notably the housing, agriculture and critical infrastructure including road transport, energy and telecommunication. It can be concluded that the OIEWG methodology works well in estimating direct economic losses attributed to disasters. The sectoral extent of damages induced by TCI indicated that Zimbabwe needs to prioritise the housing, critical infrastructure and agriculture during the recovery and reconstruction phases. The same approach also underlined the need to strengthen future mitigation efforts in the housing, agriculture and critical infrastructure sectors, amongst others. There is a need to adopt and effectively implement adequate DRR strategies in the built environment. Strengthening DRR strategies is critical, including improving the building standards using the 3Bs principle. Also critical is the need to strengthen cyclone forecasting and warning systems and construct cyclone-resilient infrastructure in order to minimise direct damages.