Asia & Pacific Expert Group on Disaster-related Statistics

Issue Paper 1, Draft ver. 1.0

Direct Economic Loss Statistics

Direct economic loss is the monetary value of total or partial destruction of physical assets resulting from a disaster. Itshould indicate, at least via proxy, the amount of economic losses directly experienced from a disaster.The scope of measurement is restricted to direct impacts to economic assets as defined according to the System of National Accounts(SNA, 2008)[1] as these are the items that can be valued in monetary terms consistently with current standards for economic statistics.

In this brief note, we review two basic methods for calculating an aggregate indicator for direct economic loss according to this definition. We analyze the two proposed indicators and associated measurement or analytical demands according to two criteria: (i) data limitations and feasibility and (ii) relevance and fitness for purpose.

In a technical paper (UNISDR, 2015) for measuring direct economic loss[2], UNISDR developed amethodology foran aggregated indicator for direct economic loss for the purpose of global monitoring if implementation of the Sendai Framework on Disaster Risk Reduction. The methodology consists of three steps. The 1st step is to collect data on physical damageper disaster or hazardous event. The 2nd second step is to apply a unit cost (i.e. average cost) value as a multiplier for the identified physical damage in order to convert observed impacts to assets into monetary terms. In the final step, the value as calculated through this unit value method can be converted to US dollars or other currencies for international comparisons. In order to make the values comparable over time, it would be necessary to regularly update the unit cost multipliers, as is done, for example, for price indices, to adjust for inflation (or deflation) over time.

Mathematically, this direct economic loss indicatoris calculated using the following formula:

Direct economic loss=

(a)Number of physical assets affected*(b)Size of the physical assets*(c)Unit Cost (construction cost per unit)

Thus, the required variables for calculating this indicator are (a), (b) and (c).

An alternative to the unit cost measurement method, proposed by the Asia andPacific Expert Group in the Draft DRSF is to use the sum of observed actual replacement costs. Wherever available, the directly observed data on the costsof reconstruction or replacement of assets after a disaster can be summed together to produce the overall economic loss by type of asset and in aggregate for the affected areas. For cases where transactions could not be observed, estimation could be used so that important losses will not be missed, as long as the estimations are the best available approximation of a market price-based valuation of actual reparations costs incurred.

Data limitations & feasibility of measurement

One important limitation of the proposed indicator from UNISDR is that it does not account for differences in the extent of the damages to the assets -e.g. a house that received minor damages versus a house that is totally destroyed. In theory, the extent of damages to the assets could be introduced through an additional variable (d), which could be derived through the use of some type of assessment or observation by recovery experts (e.g. categorization by partially damaged, damaged, or destroyed). Statistics from some national disaster management agencies already include some information or some categorizations of extents of damages to assets, but it is not yet clear whether this information could be harmonized to produce comparable statistics for an additional economic loss variable (d).

If the data are available, extent of damages could be the percentage of the damage in proportionto the physical size of the asset (e.g. proportion in square meters). If the extent of damage is measured as a percentage of the total size of the asset it could work as a weighting variable (d) for the above equation. If the variable is measured in absolute size of damages or destruction (e.g. square meters), then the variable is effectively a more accurate measurement of thevariable (b). Hence, (b) size of the physical assets can be replaced by this measurement of size of actual extent of damages.

In any case, some guidance from the Expert Group on the physical measurement of impacts to assets would be useful to help better harmonize and improve qualities of direct impact statistics across countries more generally. Even for cases where reliable statistics on economic loss are consistently produced and disseminated in monetary terms, it is a good practice to also maintain statistics on impacts to assets in physical terms, as much as possible.

Some general recommendations on physical units for impacts to assets can already be made quite logically. For example,impacts to roads should be measured in kilometers and impacts to agricultural land should include statistics calculated in area (hectares), preferably according to types of crops. But the units of measurement for other types of assets are less obvious and can vary significantly across countries or across collections of statistics for individual disasters (examples include number of units or buildings by type of building, numbers of dwellings, number of households or people affected, number of floors or other measures of size of buildings, and so on). The expert group is collecting additional information from member States on current practices related to units of measurement of direct material impacts in order to develop some general recommendations on this issue.

The UNISDR has not yet specifieddetails for amethodology for deriving unit costs. According to the proposed equation, unit costs are a multiplier applied to the measurements of physical size of impacts, based on average expected values, for estimating overall impacts in monetary terms. The compiler faces several immediate questions for selection of an appropriate average cost, or unit cost multiplier. First, the compiler would need to decide whether it is the unit (or average) value of the building or infrastructure at the time that it was built, adjusting for depreciation (or appreciations, if relevant) or the expected unit cost at the time of the disaster (or when the reconstruction happens). Also, what data source should be used? In some cases there will be public or commercial records for property values for certain kinds of infrastructure but the information may not be part of the government’s official statistics collections and so some kind of validation mechanism will be needed to ensure that available estimations used for calculating average unit values of assets are aligned in scope and with methodological standards for asset valuation.

As already mentioned above, an important weakness of an average cost method for valuing economic loss is how to incorporate not only the number and size of assets affected, but the extent of the damages. This problem is avoided if actual reparations costs incurred (or estimations of those costs) to rebuild assets after a disaster are collected and summed because in this way the extent of the damages, at least from an economic perspective, are already incorporated in the market-determined costs.

On the other hand, using a replacement costs method exhibits the problem that the assets (e.g. buildings) may be reconstructed differently. It's important for analysts of the statistics to consider that the contextafter a disaster is never quite the same as before. The concept of "building back better", for example, has become a commonly cited objectivefor communities to improve their long-term resilience after a disaster. If the “building back better” portion of replacement costscan be separately identified, than the investments should be included in measurement of disaster risk reduction expenditure and not as part of the economic loss. However, if such data are not available, the actual reconstruction costs may still need to be utilized as the best available estimation of economic loss, perhaps allowing for some necessary adjustments, where appropriate.

Relevance & fitness for purpose

For either method described in this note, some further guidance should be developed onhow to account for relatively extreme values, e.g. extremely valuable properties, very poor communities, or for rural communities. Extremely high or low values of properties will affect the relevance of the aggregateresults and some relevant measures of economic loss could be useful to develop economic loss indicators that are more closely related to experiences of the affected population at the local level. For example, for very poor communities, the average or actual reconstructions costs could be very small or almost insignificant compared to GDP or other economic indicators on a national scale, although these impacts could have been devastating or at least highly significant for that particular community. In rural communities, there may not always be monetary appraisals or transactions for reconstructions after a disaster, particularly if reconstruction becomes a community effort, or utilizing salvaged materials.

One possible solution for this issue is to distinguish dwellings by different types structure related to the value, so that analysts can study impacts according to the different categories of affected asset or population groups. For example, the 2015 Survey of Impacts of Climate Change on Human Life (ICCHL) Programme[3] ofBangladesh presented statistics on damaged or destroyed dwellings of different values that are related to different socio-economic or regional factors (see example below).

Source: 2015 Survey of Impacts of Climate Change on Human Life (ICCHL) Programme, Bangladesh Bureau of Statistics

This categorization of affected dwellings by type provides a context to add meaning to the values calculated for the losses. Categorizations will vary between countries but the underlying principle of this type of presentation is to produce a more reliable account of total economic losses and improve the possibility for disaggregated analyses (e.g. the extent of the impacts as experienced by socio-economic groupings or by regions.)

The relevance or fitness of different approaches to measuring economic loss varies depending on the purpose of the analyses. For example, for producing aggregate indicators used for monitoring high-level targets (i.e. national and international monitoring of the Sendai Framework Targets and SDGs), proxy measurements may be sufficient to indicate overall trends over a long time period. However, as these indicators will be used in the global indicator monitoring systems, there is also a need to ensure international comparability in methods. Thus, in reviewing relevance in this context, it's also important to consider strengths and weaknesses in the approaches with respect to comparing the measurements across economies. The replacement cost method would provide internationally comparable indicators and the method would be most aligned with the existing standards for economic statistics used to produce the well-known aggregated indicators such as GDP. A unit cost approach also could be implemented for international comparability to the extent that there is a reference method for estimating average costs for each of the major types of assets (dwellings, hospitals, roads, etc.) that could be consistently utilized for calculating the statistics over time.

The statistics on direct impacts in physical terms, i.e. the number and size of assets affected, are useful in their own right for analysis of disaster impacts and should be retained in loss and damages databases as much as possible. Thus, even for cases where replacements costs could be measured directly, collection of statistics on impacts to infrastructure in physical terms (i.e. variable (b) in the average unit cost method equation above) should be collected and harmonized, where possible, acrosscountries and across disasters.

For disaggregated analyses, e.g. analyses by geographic areas or for specific types of assets, generally the replacement cost method will be better fit because the statistics are compiled based on individual observations. Calculating value based on an estimation of average cost, by its nature, is not as well suited for disaggregated analyses. On the other hand, depending on the purpose of the study, statistics in physical terms (km, number of buildings etc.) may in some cases actually be the most relevant statistics for making comparisons of the scope and factors for intensity of direct impacts.

Special Case of Agriculture

For consistency and to allow for aggregation across the economy, thesame valuation principles and methodological standards should be applied for measurement of economic losses across the different types of assets across the economy and across disaster occurrences. However, agriculture has been highlighted in the DRSF and on the basis of results from the DRSF Pilot Study as one important exceptional case for special attention.

If economic loss indicators are restricted to measurement of fixed assets, than certain components of agricultural establishments such as improvements to land, machinery and equipment, livestock (“cultivated biological resources” in the national accounts terminology) and other repeat use cultivated assets (e.g. trees cultivated for fruits and nuts) are included as fixed assets and, in principle, damages or destruction to these assets should be valued for calculating the direct economic losses in alignment with the principles described above.

However, the growing crops that may have been destroyed by a natural hazard are technically classified differently; either as part of land improvement or, more likely, as work-in-progress output, and not part of fixed assets.[4]

Although single-use growing crops are not technically accounted as assets,these crops are the main source of income for farmers and case studies[5] show that they are particularly vulnerable to destructionfrom disasters. Therefore, it is proposed that growing crops should be counted and valued as part of direct economic loss measurements for the farmer.

A special approach to valuation is needed for the special case of growing crops, which based on the expected value (e.g. using the closest relevant market value for mature crops of the same type) to estimate the losses to the owners of the damaged agricultural land and improvements made for the purpose of producing the foregone output. This measurement approach is recommended regardless of the stage of production for the growing crops, as long as some significant actions had taken place prior to the disaster and was subsequently damaged or destroyed.

The recommendation is essentially aligned with the unit cost approach as suggested in UNISDR (2015) as information on the area and volume estimates for affected agricultural land will need to be converted to monetary values using an estimated average expected market value for the affected crops.

The reason for making this special distinction for growing crops is because there is also a scope within the broader framework of disaster-related statistics to estimate the impactsfrom a disaster on future output or employment (e.g. through models of the economy such as input-output tables). These impacts will be experienced over a longer period, in the months and possibly years after a disaster and are called indirect impacts in the current terminology that has emerged following the adoption of the Sendai Framework (previously, indirect impacts were often simply called “losses” in post disaster assessment studies). Thus, it is important to specifygrowing crops as a special and unique exception identified by the Expert Group for measuring direct economic losses. Besides this exceptional case, all other relevant measurements related to impacts on products not yet realized(including work-in-progress or inventories)should be treated as indirect impacts. The basic principle is described by diagram 1 below, which shows the potential scope of overall economic impacts from a disaster – both direct and indirect:

Diagram 1: Total potential economic impacts from a disaster


[1]

[2]Concept note on Methodology to Estimate Direct Economic Losses from Hazardous Events to Measure the Achievement of Target C of the Sendai Framework for Disaster Risk Reduction: A Technical Review

[3]Bangladesh Disaster-related Statistics 2015: Climate Change and Natural Disaster Perspectives, Bangladesh Bureau of Statistics (BBS), Ministry of Planning

[4]The SNA defined land improvement in paragraph 10.79 as “actions that lead tomajor improvements in the quantity, quality or productivity of land”, but generally crops other than those cultivated for repeat use over multiple accounting periods (such as trees cultivated for fruits and nuts) would seem to be more appropriately classified as works-in-progress rather than as part of fixed assets (depending somewhat on its state of processing at the time of the disaster. Work-in-progress is defined in SNA paragraph 10.134 (“consists of output produced by anenterprise that is not yet sufficiently processed to be in a state in which it is normally supplied to other institutional units”) and further described for the agriculture case in 10.140

[5] See, e.g. Expert Group Pilot Study Report (Phase I),