Economic aspects of evacuation and resettlement after a radioactive fallout in Sweden

Abstract

High energy prices recently have moved nuclear power back into the limelight. The biggest risk of nuclear industry has been large-scale accidents that give rise to ground deposition of long-lived fission products such as ¹³⁷Cs, notably Chernobyl in 1986 (Ukraine) and Fukushima in 2011 (Japan). In Japan, extensive land remediation of residential areas was carried out at an estimated direct cost between 16 and 41 billion Euros. We have studied a hypothetical radioactive fallout scenario in Sweden and then applied a cost–benefit analysis on remediation of urban land and resettlement of evacuees. Direct costs for remediation of amounts to ⁓100 million Euro/km² (2020 price levels). For an average city in Sweden the costs related to evacuation and decontamination greatly exceed the potential monetary benefits from averting radiation induced cancers. Thus, based solely on financial factors, it is concluded that an exhaustive evacuation and resettlement is not monetary cost-effective.

Introduction

Release of radioactive isotopes following accident, sabotage, terrorism, or acts of war could affect large areas of land and incur serious risks of radiological health consequences for many people. Dealing with such events may be both urgent and very complex and just like many other incidents such as pandemics or major natural disasters, requires knowledge, planning and preparation. Since the issues are complex and policy response must be fast, it is important to in advance develop contingency plans that are relatively detailed or explicit.

In this note we focus on one particular aspect of radioactive contamination—the time profile of the hazard created, and the net benefits of carrying out countermeasures. The initial radiation from a fresh nuclear fallout may be very high due to the contribution from many short-lived gamma rays emitting fission products (such as ¹³¹I, ¹³²Te, ¹³²I, ¹⁴⁰La) [1]. Even though this initial exposure of external radiation can be partly averted by protective measures as sheltering or temporary evacuation, it is the remaining ground deposition of long-lived radioactive elements that will pose the threat over long-term [2]. This is not immediate damage as with e.g. an earthquake, but the deposition of radioactive elements poses a threat of late effects in terms of radiation induced cancer that can occur many decades after on-set of exposure. The momentaneous threat itself will decrease over time due to the combined forces of radioactive decay and ecological turnover as the radioactive particles tend to migrate into deeper soil layers, leading to shielding of the external dose and a decrease in root uptake that altogether will reduce the threat they pose. In addition, we have human time preferences that imply we discount problems occurring far in the future [3].

Decontamination of land that has been affected by fallout is associated with large monetary costs. The Japanese government has prioritized a large-scale restoration of affected areas after the accident in Fukushima. Up to and including 2019, between €16 and €41 billion was set aside for land remediation [4] and several areas are now habitable [5]. Furthermore, Japanese Ministry of Health reports that ⁓1150 km² of urban and semi-urban land were included in this decontamination project [6] rendering a cost estimate of 62 MEuro per km². In a Swedish multidisciplinary project carried out in the beginning of the 2020s [7] research was done on remediation measures after nuclear accidents in the scale of those in Chernobyl in 1986 and Fukushima in 2011. The experience from these accidents in terms of economic costs, notably by [8–10] were used as starting points for modeling costs and effects in the remediation of a Swedish fictitious RN-accident.

In this part of the study, we have collected data about prices of different types of lands from Statistics Sweden and real estate prices from Swedish Real Estate Statistics for 2022 [11–13] and calculated how much production loss for a certain period would be. Based on those data, cumulative and total benefits and costs are calculated, according to the formula presented here, developed from [8–10]. To provide a more concrete example, the cost–benefit calculations have been focused on the epicenter of the hypothetical fallout with the entire Lund city of 26.4 km² needing evacuation and remediation. Furthermore, we have made some assumptions based on experiences from Fukushima regarding the return frequency among the evacuees and to which extent the returned population contributes to resuming the economic activities in the decontaminated area.

Material and methods

Cost–benefit function

In Fig. 1 an expression is given of the flow of monetary benefit in an area, B(t) (Euro km⁻²), both direct and delayed indirect costs from an area that is subject to remediation measures due to a radioactive fallout leading to an initial effective dose rate, d(t) at time t. B is a time-integral of the monetary flow rate, b (Euro y⁻¹ km⁻²). One aim of this expression is to find parameter settings for which B is maximized, often with respect to the timing, t (y), of the events represented in the expression. The remaining parameters included in this expression are described in Table 1.

Figure 1.

Schematic illustration of the economic significance of the various components in the net benefit function B(t) per unit evacuated area.

Table 1.

Variables included in the cost–benefit function, B(t), given in Fig. 1.

Variable (units)	Description
b(t) (Euro y−1 km−2)	Net monetary flow for a specific km2 in a geographical area.
d(t) (Sv y−1)	External effective dose rate at time t in the affected area.
	The monetary value of a statistical healthy year (in prices at the year zero level) (Euro HEALTHy−1). Swedish GDP per capita is ⁓50*103 € y−1, and may represent a coarse estimate of the monetary value of any healthy year lost due to cancer disease.
r (inh km−2)	Population density in the affected area.
P (y Sv−1)	Loss of productive life years due to radiation induced cancer.  This expression is fetched from Munro (2013), where P is the probability coefficient for fatal cancer per unit effective dose (case Sv−1) as defined in IAEA 1994 [8]. P is the probability of fatal cancer induced by radiation and is set to 0.05 cases per Sv−1. Factors β and δ account for non-fatal cancers and hereditary affects, which according to Munro is set at 0.01 and 0.013. The factor nrad represents the average number of life years lost due fatal cancer and is set to 13 y per cancer case in IAEA [8]. Hence this product becomes P = 0.05*(1 + 0.01 + 0.013) *13 = 0.66 cancer free y per Sv-1, which is thus the estimated loss of cancer free year per Sv−1. In Munro [9] this factor is simplified to P = 1.
fs	Shielding factor of residential buildings in the area; default is set to 0.4.
r(t)	A time-dependent function describing the decay of the external effective dose rate. An expression for r(t) derived from Jönsson et al. [14] has been used here.
tr (y)	Time from start of evacuation until resettlement. For simplicity it has been assumed that evacuation starts at t = 0.
Sevac (Euro inh−1 y−1)	Monetary loss rate of capital of real estate during evacuation. In this study which has an area representative of Swedish cities and towns in terms of population density, we have used a value of 5 million SEK (≈500 000 000 euros) when the value of different properties within the area are taken into consideration and is therefore proportional to the population density ρ. Typical household size per single residence is set at 2.4 members. Sevac(t) is in turn given by Sevac·(r/2.4)·(1-Exp(−(ln2/Ω)·t) where it is assumed that the buildings’ values are depreciated over a life-time given by the parameter Ω, (see below) in absence of service and renovation and that even after return, the costs of restoration and renovation will equal to that of the annual depreciation rate. In this model we assume that the value depreciation of a property stops when evacuated owners return. Hence, Sevac is multiplied by the return rate, Rret(t) (See below).
Ω	Depreciation rate (y−1) of real estate and infrastructural capital value. In this example we have set Ω = 2.31 y. This value will represent a situation where we assumed that the full capital value of a residential house has been completely depreciated after 15 y.
aDRD (y inh−1)	Cost of productive life years lost due to Disaster Related Deaths (DRDs). Yanovskyi et al. [10] estimate this number to be, on average, 11 years per DRD case. The likelihood of a DRD case in a population combined with the estimated life years lost due to evacuation in order of 10 years is set to 0.11 years per evacuated average person.
Scleanup (Euro km−2)	Direct costs of decontamination per unit area in the affected region. For a typical Swedish semi-urban residential area, a sum of 100 MEuro in 2020 year price level has been assumed [15].
d (%)	The discount rate. In this study a value of 3.5% has been used in accordance with other economic assessments in Sweden, with the Transport Agency’s Analysis Method for Socioeconomic Calculation Values for Transportation Sector (ASEK), as the main reference point. In the latest version ASEK 8.1, it is recommended to use 3.5% as discount rate for future costs and benefits[16].
(t)	Time-dependent decontamination efficiency as elaborated in e.g. Rääf et al. [17].
br (I(t), Rret(t),  ) (Euro inh−1)	Net monetary flow rate per returning inhabitant by restoring infrastructure and production in the affected region per unit evacuated area (Euro y−1 km−1). In this study we assume it is a function of average income from employment, I(t), fraction of evacuees returning to Lund, Rret, and fraction of returnees being in employment upon return, s. It will hence be described by the following formulae: where we assumed an average annual income in the city of Lund at I0 = 50 kEuro y−1 per employed returned inhabitant at time of the accident (t = 0). We have also assumed an annual income increase of 3.5%, that is, I=I0·1.035^t in the expression in Fig. 1. Furthermore, the first term in this expression represents the net loss rate from the fraction of the evacuated population who would be in employment if the accident had not taken place, assumed to be s = 50% of the city of Lund. The second term represents the annual net gain from the fraction of returnees who become actively employed (see below).
Rret(t)	Fraction of evacuated population returning to the remediated region. This fraction can be as low as 50% in Sweden according to Rasmussen et al. [18]. In this study it is assumed that 15% of the evacuees will return within 1 y, of which s = 50% will be in employment (irrespective of age). The fraction of returned will then increase by 10% each year until reaching a maximum of 50% around t = 13 y.

Study area and design

The city of Lund in Southern Sweden was selected as our study area. A fictitious nuclear power plant fallout, with an initial effective dose rate of 1 Sv y⁻¹, and with a subsequent time-pattern of dose rate mimicking the one from the Chernobyl fallout in Sweden [14] has been assumed to occur over the city of Lund. The value of the initial dose-rate (1 Sv y⁻¹) was selected to mimic the reference scenario described in [10] from which the estimates of the term a_DRD in Fig. 1 has been taken. Out of our three geographical scenarios, Lund represented the middle one in terms of cost of a combination of inhabited area, agricultural land surrounding the affected area, importance of infrastructure and demographics. For simplicity, we have opted the whole Lund city as affected area in based on that the affected area in our model consists of only of inhabited area, thus buildings and roads. The city consists of 26.4 km² with a population of 95,000, which means that the average population density is approximately ρ = 3600 inhabitants per km². The surrounding agricultural lands, which are classified among the highest valued land areas in Sweden, are not taken in the model but we recommend that they should be considered when a decision is finally taken about type and degree of remediation.

In this study, a societal perspective with costs and benefits related to evacuation is used, remediation and return of evacuees. While costs are centered at the affected area itself, benefits are bound to the inhabitants and can occur both at the area with returnees and at relocation areas for those who do not return after remediation. For return of evacuees to economic activities, an age interval of 20–67 is used, because in Sweden by the age of 20, most have finished secondary school and either start working or studying at university level, the reference year of retirement in 2023 is 67 [19]. That age interval makes up 65% of the population of Lund. It is assumed that the evacuated persons will get employment in relocation areas after some time, with 50% of those in the working-age interval of 20–67 being employed in year three after the accident. The employment rate will gradually increase by ⁓10% per year reaching the national average level of ⁓80% [20] at year 8 after the accident. Hence, assuming a similar age-structure among the returnees as the initial population ⁓0.8*0.65 = 50% (=σ in Table 1) of the total population will then be in active employment.

In the chosen scenario of Lund city as affected area, direct and indirect costs are calculated based on some underlying assumptions. One assumption is that the direct loss of capital in terms of buildings affected by the incident has a depreciation life of 15 years, meaning that the values of the affected buildings are depreciated wholly within this period. As most of the loss happens within the first years after the accident, it is assumed that the depreciation process follows a logarithmic growth path. It is further assumed that when people return from relocated areas to their homes in the affected area, the costs of restoration and renovation of their homes equal that of annual depreciation rate, so the depreciation of buildings continue as unaffected. Another assumption is that the affected persons will get employed at an increasing degree at relocated areas so that the value they create will offset some of the costs of evacuation and decontamination after the accident.

Results

Based on the assumptions underpinning the study, results of the calculations show a large concentration of costs in the first few years after the fallout. Disaster-related deaths, as elaborated in [9], occur primarily in direct relation to relocation of evacuees and loss of capital due to abandonment of the affected area stretch to a decade, given that the buildings are not serviced and kept in shape. Nevertheless, benefits of avoiding cancer cases will be felt long after the accident as it takes time for radiation to lead to cancer and give symptoms. After some years, assumed 3 years in this study, a share of the working-age population, here taken as 20–67 years, will find jobs at either the relocation area or at the affected area in case of returning. Here it is hypothesized that in year 3, some 50% of those in age-interval 20–67 will be employed and that will increase by 10% each year so that in year 8, some 80% are employed. As that is the mean employment rate in Sweden [20], it is set as an asymptotic cap value, meaning that from year 9 onwards, it is assumed that 80% of 20–67 are employed and that retirement of some will be filled with children turning into adults.

At 2020 years values, the monetary values of offsetting cancer risk will be miniscule, compared to the costs of evacuation, disaster-related deaths and remediation of the affected area. In addition to an estimated clean-up cost for decontamination of 26.4 km² urban area of 26.4*100 MEuro (=S_cleanup in Table 1)=2.65 GEuro, the net costs in total accumulate during the first 5 years to about €41 billion, meaning around €440,000 per inhabitant, if associated to only those in Lund. Nevertheless, in real life, many more are affected. Since Lund is close to two strategically important highways connecting Sweden with continental Europe, contamination in Lund would divert the traffic so many more than Lund inhabitants would be directly affected. Even indirectly, inability to use valuable agricultural land around Lund would affect those businesses associated with farming industry and local tourism. This can be compared to Japan, where the total financial loss due to the Daiichi nuclear power plant accident was estimated at $210 billion [21]. Nevertheless, the loss was calculated for a much larger area than Lund city’s 26.4 km² and as a result of that accident, ⁓165 000 people were evacuated [22], i.e. almost twice the population of Lund. With the assumption of the same pattern of return as in the case of Fukushima in Japan, most of those in working age will stay in their relocation areas, while most of the elderly might return. Nevertheless, experiences from Fukushima only covers ⁓10 years, so it is difficult to theorize further than that. Orita et al. [23] had nevertheless found out through a study of quality of life among relocated from Fukushima that those who had returned were happier and valued their quality higher, which might influence those who have yet to return.

As can be seen from the plot in Fig. 2 there is an initial remarkable loss of capital directly after the fallout accident. However, that is attenuated by the discount effects as well as the evacuees being employed either elsewhere or in the affected area after return. The initial financial loss is, nevertheless, so vast that even after the returnees are being employed from year 3 and onwards, it takes for the affected persons and their offspring ⁓50 years to cancel out the net loss rate (Euro y⁻¹).

Figure 2.

Four components of the net benefit rate, b(t) (euro y⁻¹), by evacuation of 26.4 km² of the Lund city area as a function of time after a fictitious nuclear power plant fallout, giving an initial external dose rate of 1 Sv y⁻¹ and with a radionuclide composition equaling that of the Chernobyl fallout in Sweden [14].

In comparison to the financial costs of evacuation and remediation, the potential future gains from averted cases of radiation-related cancer cases, and subsequent loss of life-years, are small. While there is an accumulated latent risk of cancer in the long run, many of the persons also risk catching other diseases during the latency period before debut of cancer symptom, and a large part of the general morbidity in the cohort of evacuees might not be associated with their radiation exposure. Evacuation itself is associated with high risk for disaster-related death (DRD), which in case of Daiichi nuclear power plant accident affected ˃2200 people, i.e. ˃1 out of 100 evacuated [24]. The potential number of people being diagnosed with radiation-induced cancer in the future might not exceed the DRD cases, considering the expected number of cancers (excluding thyroid cancers) attributed to Chernobyl nuclear disaster among the ca 115 000 evacuees in Belarus, Russia and Ukraine (not including liquidators) is assumed to be ˂300 (This number being derived by conservatively taking the factor 0.07 (older cancer detriment used by ICRP(1991) [25]) times 3585 manSv (Collective Dose according to Table 6 in [26]). Based solely on monetarised values of cancer-free life-years for Lund’s 95 000 inhabitants, i.e. almost 0.4 billion euros within the first 50 years after the accident, total evacuation and decontamination is not cost-effective.

Discussion

The cost–benefit function defined in Fig. 1 can essentially be simplified as follows: where D_averted represents the cumulative averted dose during evacuation, D_residual the corresponding cumulative dose after the return, A_DRD the cumulative cost of loss of disaster related deaths, and T_r the return time of the evacuees. From the results of our parameter estimates specific to the region of Lund in Sweden it can be seen that the cumulative monetary return is essentially governed by the last three terms, the initial but decreasing cost of clean-up vs the net benefit in restoring the semi-abandoned infrastructure of the evacuated city, and the return rate, R_ret, of the evacuees. Nevertheless, a more exhaustive sensitivity analysis could in the future be conducted to further investigate the relative importance of these parameters.

As the scale of the radiation accident in this study is hypothetical, it is difficult to gage the effects of such accident outside the considered area in Lund. In reality, a potential accident might affect areas beyond the city borders to varying extent, depending on some external factors such. The very valuable agricultural lands surrounding Lund city would make cost-effectiveness calculations a matter for the national farming industry since substantial long-term loss of agricultural production can be anticipated due not only to disposal of food produce initially exceeding maximum permitted contamination levels stated by EU [27], but also to a larger scale the loss of consumer confidence in the food products even long after the products fall below those exemption levels. Hence, even if the results herein do not support that there is a monetary benefit of an exhaustive evacuation and decontamination after a radioactive fallout, taking also into consideration other factors, such as the long-term effects for the agriculture in the affected region and the trust of members of the public in the authorities, could lead to different conclusions.

Studies about returnees in Fukushima, Japan, mostly point to a disproportionate share of returned being elderly men [28, 29]. A reason behind that is that many were involved in agriculture and could not stay away from their land and properties for long. Nevertheless, even among city dwellers, higher share of those who had returned were elderly, indicating that a higher proportion of persons in working age stay where they are relocated after evacuation. In the case of the city of Lund, which has a relatively young population thanks to the university with thousands of students as well as firms hiring newly graduated students, it can be assumed that a total evacuation and decontamination will probably be difficult to justify due to low cost-effectiveness of the decontamination projects. In this case, other factors such as the value of the surrounding lands, centrality of the city, cultural values associated with certain monuments and buildings, as well as the highly important maintenance of public trust.

According to Organization of Economic Cooperation and Development (OECD), governments’ responsiveness in times of crisis is a major component of trust in government [30]. According to a Meta analysis by Conal Smith (2020), GDP growth is positively related to public trust so that a 10 percent increase in public trust is associated with a 0.5 percentage point increase in real GDP growth [31]. In the case of Sweden, where, according to the OECD, ⁓40% trust the national government to do their tasks well [32], an increase of trust to 50% could have meant a real GDP growth of 3.3% instead of 2.8% [33] for 2022, and in real monetary terms it would mean ⁓30 billion SEK (≈3 billion euros) more in economic growth. Thus, if governments act in response to a major accident, such as nuclear accident, more than direct economic costs and gains should be considered to have robust decisions that are acceptable for the involved stakeholders.

Conclusions

Calculations in this study show that direct and indirect costs related to evacuation and resettlement after a radioactive fallout are much higher than potential monetary benefits. However, in other cases, with essential infrastructure for the country, other factors than sheer financial ones are important to take into consideration, rendering resettlement beneficial even if the costs exceed the financial benefits.

It is important to note that the study is based on a hypothetical case and that much of the economic values are based on assumptions related to primarily the nuclear accident of Daiichi nuclear power plant in Fukushima, Japan. With some cross-checks about size of the area and the people affected, as well as the total financial loss, the results in this paper are reasonable. Nevertheless, for more precise calculations, further research and more robust data are needed.

Conflict of interest

None declared.

Funding

None declared.