Transitional shelter strategies in context: cash, private insurance, rental subsidies, private, public and temporary housing after the Great Eastern Japan Earthquake and Tsunami (GEJET)


This paper reports on the transitional shelter response in Japan after the Great Eastern Japan Earthquake and Tsunami (GEJET). It is typical that  international lessons  regarding post disaster shelter focus on the design of temporary housing and comparing the cost, size, coverage and speed of delivery of temporary housing units.  However, we argue that such debates need to  consider the  institutional context that links pre- and post-disaster housing processes. This paper presents the international debate on transitional shelter with reference to data from this and previous disasters and regulation and literature about Japan; it thus describes and then explains the GEJET shelter response in the context of housing policies and conditions before the disaster . We therefore provide the information on both pre- and post-disaster housing processes and anlayse the framework and context  in one place for the GEJET, in contrast to previous works that tend to have a less holistic approach. The paper concludes with recommendations for strategic decisions and analysis aimed at international institutions involved in disaster response and recovery.

1. Introduction

In 2011, Lord Paddy Ashdown chaired the Humanitarian Emergency and Response Review which looked at the way the UK government responds to emergencies. The review found that:

 “[p]roviding adequate shelter is one of the most intractable problems in international humanitarian response……arguments between experts over design, quality and cost can slow the process, and weak coordination in the sector often leads to a wide variance in what is provided.” (HERR, 2011)

This paper asks what can be learnt from the government of Japan’s transitional shelter strategy after the Great Eastern Japan Earthquake and Tsunami (GEJET). Comparing only the design and cost of temporary housing units may make the Japanese response appear irrelevant to other contexts. For example, in Haiti pre-assembled units cost 1,500-3,000 USD while pre-fabricated units in Japan cost 35,000 USD. Instead, this paper looks beyond these indicators to why the government's strategy began and then unfolded as it did in order to derive more interesting lessons and pertinent comparisons for disaster response.

The GEJET destroyed or damaged 620,802 homes and 561 km2 hectares of land along the Tohoku[*] coastline (BRI and NILIM, 2011; EEFIT, 2011). The event led to 19,000 fatalities and economic losses, in 2012, of USD 210bn (Munich Re 2012a, Swiss Re, 2012).

This paper covers the response in Iwate, Miyagi and Fukushima Prefectures where 102, 345 homes were completely destroyed. The government of Japan’s transitional shelter strategy originally planned support for 116,000 affected households by supplying 52,000 temporary housing units (46%) and subsidising households to rent what was expected to be up to 63,000 publicly owned housing units (54%). However, as the response unfolded, 136,000 households received support: 52,000 moved into temporary housing (38%), only 18,000 into public rented housing (13%) and an overwhelming 66,000 households opted to use the subsidy to rent private accommodation (48%).


The paper is based on a review of the publicly available data and literature (limited to resources available in English) as well as key informant interviews during an EEFIT mission in May 2013.

The paper uses the term transitional shelter to describe the overall policy approach in Japan. The temporary, collective shelters where people initially sought refuge are called evacuation centres[†] and pre-fabricated housing units are called temporary housing or temporary houses.

It is organised in terms of international policy; pre-disaster housing and temporary housing processes in Japan; the strategic response to the GEJET and how it compares to other responses with similar coverage; a discussion of the legal, functional and technical capacities that shaped the response; and conclusions about which lessons apply only in Japan and which may be applicable beyond Japan. This analysis is in support for Japan’s reconstruction strategy which made international learning a specific objective.


2. International Policy on Shelter and Housing after Disasters

This analysis puts the Japanese case in the context of international policy on shelter and housing after disasters. Firstly, in the recent revision of the World Bank’s Post-Disaster Needs Assessment (PDNA), there is a renewed emphasis on baseline assessment, not just of needs, but of capacities including functional (e.g. successful creation and management of policies and legislations) and technical capacities (e.g. knowledge about service delivery in, say, the housing sector) (GFDRR, 2014). Section XXX looks at these capacities in Japan.

Secondly, there is a strong distinction, in the World Bank’s new Disaster Recovery Framework (DRF), between private and public assets and the role of government in the recovery of both (GFDRR et al., 2014). Section XXX looks at the ways that Japan anticipates this.

Lastly, there are recurrent calls in the international humanitarian sector to think of shelter after disasters as a “process not a product” (Davis, 1978; IFRC, 2013a) (XXXX); to consider – as per the government of Japan’s strategy – transitional shelter and settlements (IRP, 2012; UN, 2008); and to, thus, set standards with reference to what already exists and what might be appropriate and acceptable (Sphere project, 2011). Seen in a broader framework of disaster management, whether conceived as cyclic processes of preparedness, disaster response and recovery (Alexander, 2002; Baird et al., 1975; UNDP and UNDRO, 1991) or as parallel processes of response, recovery and risk reduction (IFRC, 2013b), these frameworks encourage questions about:


·         who decides when, where, what and how to build housing;

·         why people live, or build, their homes in certain locations or in certain ways;

·         who pays; who owns and who rents; and

·         how these factors might affect different households and places after a disaster.

Section XXX discusses how these factors affected the transitional shelter response in Japan and what this means for transferring lessons.

2.1 Transitional Shelter and Settlements: confusions and criticisms

It is worth introducing the transitional shelter approach because this terminology – even though its meaning is confused and its application contested – has been adopted by the knowledge-sharing project sponsored by the Government of Japan and the World Bank that was initiated after the GEJET, Learning from Megadisasters, (The World Bank Institute, 2012).

This approach was originally conceived to promote a “wider understanding of the settlement options selected by the entire affected population” and allow these options to be supported (Shelter Centre, 2010). It was intended to capture the idea that people who lose their homes, flee a disaster or seek refuge in nearby evacuation centres, have to find alternative accommodation until they can return to rebuild or repair or until they decide to resettle elsewhere (OXFAM, 2005; Shelter Centre et al., 2010; UN, 2008).

Three sources of confusion over transitional shelter are addressed here:

·         pre-packaged or pre-fabricated shelter kits are often called "transitional shelters" when, in fact, the shelters themselves are just one component of the post-disaster transitional shelter response;

·         this leads to the assumption that the thing in transition is the shelter (that the shelter kit will be incrementally upgraded and improved) rather than the people affected by the disaster who might prefer to move rather than transition a kit into a permanent home.

·         Any subsequent focus on facilitation of movement or relocation then is assumed to be made possible by making a shelter kit lightweight and mobile rather than taking a broader view of what would enable people to dwell elsewhere, perhaps by directing support to alternatives like rental accommodation.

The main criticism of transitional shelter has been levelled at the use of pre-fabricated shelters or kits. This comes from the confusion over terminology but is also because, in some contexts, pre-fabricated kits may be regarded as intrinsically limited:

·         in scale because they are expensive and thus limited in number;

·         in scope because large, uniform floor areas and single storeys are only suitable in some locations and for some households (Crawford et al. 2014);

·         in the possibilities they offer for building capacity (Lyons et al., 2011; Schilderman and Lyons, 2011) and addressing the reality of building practices (Bendimerad, 2004; Langenbach, 2009; Langenbach et al., 2006)

·         in their impact on local economic recovery (Burnell and Sanderson, 2011; Clermont et al., 2011; Crawford, 2011; UN-HABITAT, 2012).

Whether these limitations are applicable to the deployment of pre-fabricated housing in Japan is of particular relevance to this analysis of the specific, historical conditions which made this option viable after this disaster.

3. Pre-disaster housing processes in Japan

Given the controversy over pre-fabricated housing in other contexts, it is useful to put the scale, speed, quality and costs of temporary housing after this disaster into the context of the ‘normal’ functional and technical capacities of Japanese housing and pre-fabricated housing sectors.

3.1 Housing, land and property: supply, demand and controlling speculation

Figure 3.1 shows annual house completions in Japan. For reference, Japan's rates of completion at the turn of the century are roughly 6 times the rate of the UK but for only twice the population and one that is shrinking. Among the drivers for these high rates are the preference for new build rather than “second-hand” houses (80% of completions in Japan compared to 5% in the UK) and the short design life of housing (20- 30 years) (Koo and Sasaki, 2008). On the supply side, light touch planning control makes it easy to get permission to build (Mori, 1998) and there has also been a historical tendency to use housing policies to manage the national economy which has entailed facilitating low-cost housing finance to boost domestic demand and control inflation (Barlow et al., 2003). Further, specialist pre-fabricated housing manufacturers supplied about 20% of the market for detached, family houses in Japan and about 14% of all housing completions or 160,000 housing units per year (W. Johnson, 2007).


In addition, a third of all completions (75% of houses) is owner-driven: commissioned by individual, owner-occupiers. This is critical because it means that “the majority of housebuilding is nonspeculative”, which as we shall see has consequences for innovation in the housing supply chain, and “does not involve land development, as the plots are already in the ownership of the housebuilder's customer” (Barlow and Ozaki, 2005) but it also means that land prices make up a significant proportion of housing costs for individuals (Zetter, 1986).

3.2 Housing policies: targeted welfare provision

As for more general public policy norms in Japan, outlined in Section XXX, and similar to the model that has historically operated in the UK and USA – housing is based on a) general self-help and b) help for “households unable to help themselves” via targeted provision of welfare. This means publicly owned housing provided at subsidised rents to people whose eligibility is based on income and welfare criteria (elderly, disabled, single-parent families) (Horita, 2006).



Figure 3.1 Annual rates of house building in Japan (Horita, 2006; Kitamura, 2011) with UK rates for comparison (Department for Communities and Local Government, UK, 2013)

3.3 Scrap and build: cultural and economic attitudes to transience and permanence

The short lifespan of houses has been attributed to Japanese perceptions that buildings are “transient rather than permanent” (W. Johnson, 2007) but, amplifying this possible cultural preference, are three other conditions: land prices, from the second world war until the crash in 1990, rose faster than building prices so the amount spent on a home is dominated by the cost of land leaving less money to invest in longer lasting buildings; during the same period quantity rather than quality drove supply and older housing had become obsolete in terms of more recent floor area standards; and certain building types (wood-framed dwellings of one or two stories that were exempted from structural calculations) are thought to lack seismic resistance (Koo and Sasaki, 2008).


The disconnection of  the value of buildings from the value of land means that house value depreciates over time more like a consumer than an investment good (Barlow and Ozaki, 2005). this has in the past been driven by family ownership of land and (at least until the 1990s) employment in a single location for life combined with a “strong cultural attachment to the land in Japan" and parcelisation and redistribution to small-scale tenant farmers after the second world war (Zetter, 1986). There may also be cultural and religious preferences for modernity and renewal that (W. Johnson, 2007) seem to have contributed to what has been termed a ‘scrap and build’ approach to housing (Barlow et al., 2003).

3.4 Innovation: mass customization not mass production

These structures of land and housing markets have nurtured pre-fabricated housing suppliers. These companies XXXX emerged from Japan’s manufacturing sector in the 1950s, a process summarised in Table 5.6





As mentioned, the majority of housebuilding has happened on family plots so housing developers do not traditionally generate revenues from land banks and have had to innovate in "product and process"  such that "[u]nderpinning the mass customised approach is the way the large suppliers have been able to use standardisation (the complete and consistent interchangeability of parts) and preassembly of components and complete subassemblies (such as timber and steel-frame systems and external cladding), to move from a focus on economies of scale in production towards economies of scope." (Barlow and Ozaki, 2005).

3.5 Builders: big companies with massive capacity

More than 90% of house builders are small, local (traditional timber beam and post) contractors supplying fewer than 10 houses annually (REF). Of the remaining 10% of companies, 344 have the capacity to deal with orders of more than 100 units. Among these big companies are names including Daiwa House Industry Co., Sekisui House Ltd. (1928) and Misawa Homes Co. (Kitamura, 2011). Ten years ago, Sekisui and Misawa were supplying 60,000 and 30,000 pre-fabricated units per year (Barlow et al., 2003).


As the preceding subsections argue, the size and capacity of these firms and the choice and possibilities they have created for pre-fabricated housing are due to a number of interacting factors that are particular to Japan and undoubtedly contribute to the rapid delivery times that are presented later in Figure 5.3.


What these headings show is that who decides, who builds, and why and how housing is designed, built and financed are related not just to individual families but to a larger historical, political and industrial context that has been previously documented.



4. Previous disaster processes: laws and  lessons from Kobe

The DRF notes therefore that legal clarity is needed over the degree of responsibility taken by governments or other institutions for repairing or replacing private housing assets; restoring critical public infrastructure; subsidising or facilitating housing recovery; and relying on private insurance (GFDRR et al., 2014).


Indeed, one of the opening quote’s “intractable problems” in responding to large scale loss of housing after disasters is that, although housing is a private good, it often takes precedence over other sectors after disasters because of its importance for recovery and “direct impact on affected populations” (GFDRR et al., 2014).

4.1 Legal clarity: definitions, regulations and expectations

In Japan, the regulatory framework anticipates these decisions in several fundamental ways (Edgington, 2010).

Firstly,  the national government has responsibility for recovery (fukkyū), which traditionally means clearing debris (including from private land) and restoring important public infrastructure rather than private property, economic or individual recovery but it also means that the government is accountable for making reliable and public estimates for how long this might take and then for delivering against these targets, as exemplified by public statements on the amount and methods of estimating debris issued by the government (Office of the Prime Minister of Japan, unknown)REF.

Secondly, norms are based on principles of self-reliance (Horita, 2006) such that taxpayer money cannot be used to rebuild private property or to subsidise the development of private property. The development of private property has been interpreted in the past to include the building of temporary housing on private land.

Thirdly, and in line with these principles, the provision of temporary housing, made available free of charge, is preferred over distributing cash for rebuilding as the former can be considered an "in kind" welfare benefit for those who have lost homes (interpretation of the Disaster Relief Law). Priority is given to the vulnerable and the definition of temporary housing is that it is expected to last for two years (interpretation of Building Standards Law) (IRP, 2012).

Fourthly, as reported by the General Insurance Association of Japan, residential insurance penetration was only 23.7% of the population prior to the event in 2010. Although this has crept up after this and previous events (Miyagi Prefecture is now reporting the highest penetration in Japan at 43.5%) and it is supported by a simple claims adjustment system (IMF, 2012; World Bank, 2012; EEFIT, 2013), insurance may accelerate recovery by cushioning some households but it does not appear to be intended or to function as an instrument for accelerating reconstruction across the affected population.

4.2 Political debate and institutional learning: temporary housing after previous disasters

The GoJ’s Basic Guideline to Recovery after the GEJE and tsunami prioritised a regional approach to planning, setting a broad, high-level framework that allowed for different local options and where the housing objective was placed alongside the need to ensure employment (livelihoods), take particular account of the needs of the elderly; and draw on, and control and subsidise the cost of renting, public housing (Government of Japan, 2011).

These key elements can be traced back at least to the Great Hanshin Earthquake that struck the city of Kobe in 1995. As for the GEJET, the regulatory framework allowed for provision of temporary housing units intended to last for 2 years, and 29,278 (plus 3,168 outside Kobe) were provided. Three years after the earthquake in April 1998, 45% of temporary housing was still occupied (about 14,000 households) and national and international criticism had been levelled at the living conditions in these units for the slow pace of public housing provision. Elderly and low income people were found to be living in isolated, cold, noisy, small (20-26m2) pre-fabricated housing units that had no communal facilities and were installed on the outskirts of the city and this appeared to result in higher incidents of 'kodokushi' or “solitary and initially unnoticed deaths” (Habitat International Coallition, 1996; Kako and Ikeda, 2009). This prompted a policy change designed to open up housing options not only by increasing the target number of new public housing units to be reconstructed but also by speeding up repair and rehabilitation, subsidizing private rental housing (government leases it and then lets units at subsidised rents) and through a further subsidy to reduce the rent of existing public housing.


The transitional shelter strategy has to be seen in the context of ‘normal’ housing policy and the Kobe response: alongside the vast number of new build completions in ‘normal’ conditions, post-disaster capacity for reconstruction is also vast with figures of 4,000-6,000/month reported for Kobe (Hirayama, 2000). Further, according to a detailed sociospatial analysis of the Kobe response, the ‘normal’ problems associated with the two tier housing policy outlined above carried over into the post-disaster housing logic (Horita, 2006). Among these problems were a geographic concentration of disadvantage and the negative impacts that followed from this on community support or communal activities like basic upkeep, high rents for low quality private rental housing and limited market interventions to stimulate the supply of low-cost private housing options.


Elderly and low-income groups also became concentrated in temporary housing because alongside the general vulnerabilities associated with being elderly or on a low-income, there were housing-specific vulnerabilities, correlated with age and income,that led to and then amplified and protracted the vulnerability of those who found themselves in temporary housing.


Firstly, proportionately more low-income (often the renters) and elderly people (often the owners) had been living in downtown old, wooden, multi-family rental housing and wooden terraced housing whose numbers had been disproportionately damaged by the earthquake (Edgington, 2010; Hirayama, 2000). This lower quality housing suffered more damage but, precisely because it was of lower quality, it had also been the bedrock of the lower cost private rental sector (Hirayama, 2000). As a consequence, this population group was disproportionately represented in evacuation centres.

Next, eligibility for temporary housing was based on still being stuck in an evacuation centre so they were transferred in the same proportions to temporary housing and, although eligibility for permanent public housing was prioritised for people still in temporary housing the allocation was slow. Finally, proportionately more of this vulnerable population were unable to return home because rates of reconstructing in the older parts of the city from which the vulnerable population had come were slower due to lack of space and complicated land re-planning processes (Habitat International Coallition, 1996; Sorensen, 2000for densely packed areas of the city with small plots and many owners (Edgington, 2010; Hirayama, 2000) required government intervention in the housing market to prevent rent inflation (Habitat International Coallition, 1996). Elderly people, who were also the landlords supplying the rental inventory and depending on rental income for their livelihoods, were not able to access the low interest loans made available after the disaster because they had lost their source of income and collateral, their housing asset, and faced age-discrimination from lenders.

Taken together and looking back at Table 5.1, the learning from the Kobe recovery is clear in the policy response to the GEJET: supply of public housing was a key priority for the tsunami-affected areas, overlooked after the Kobe earthquake and, later, understood to be home to an elderly, low-income population.




5. Post-disaster shelter and housing processes after the GEJET

An excellent summary of the approach to transitional shelter in Japan has been compiled by the Government of Japan and World Bank (IRP, 2012). This section borrows from this summary and adds further disaggregation by prefecture, uses international comparisons and draws on data about the pre-existing housing stock.

5.1 Strategic options and individual choice 

Transitional shelter options were based on three large scale programmes:

·         Support for people to move into private rental housing: information on available units was provided by the Centre for Information on Public Houses for the Affected, set up by MLIT on 22nd March 2011; a rental subsidy was paid directly to the disaster-affected tenant household for up to 2 years. This was extended by another 12 months in April 2012 (Asia and Japan Watch, 2013)

·         Temporary housing units: procurement was financed by the national government, production and construction was sub-contracted to manufacturers by the prefectural governments (BRI and NILIM, 2011) and site selection and planning permission was by municipal governments[‡]. Although prefectures had pre-agreements with housing manufacturers in place prior to the earthquake (IRP, 2012), the scale of the crisis meant that an international invitation to tender was launched on behalf of Iwate, Miyagi and Fukushima prefectures on 15th April 2011 (Federation of Housing and Community Centers, 2011). These units were intended and specified for a two year period of occupation.

·         Making government-owned or public housing available as rental accommodation.


The number of people that chose each option is summarised in Table 5.1.

Table 5.1 Transitional Shelter Options and Choices (compiled from various sources)


Number of houses allocated or chosen

Number of houses supplied

Number of houses allocated or chosen

Number of houses supplied

Temporary housing (mostly prefabricated)





Government-owned accommodations





Public housing





Private rental housing











Source: (BRI and NILIM, 2011)

Source: (IRP, 2012)

5.2 Cash, Rental Subsidies and Market Interventions

Alongside temporary housing, the government quickly announced a package of financial measures so that: “victims who have lost immediate family members will receive ¥350,000 (US$4,200) per dead or missing member; households whose homes were destroyed (approximately 110,000), will receive ¥350,000 (US$4,200) each; households whose homes were severely damaged, (approximately 127,000) will get ¥180,000 (US$2,160) each... Other forms of support available to the victims include donations, tax exemptions, student tuition waivers, debris removal and various kinds of loans.” (EEFIT, 2011) and replacement of household white goods for 280,000 affected households (Japanese Red Cross Society, 2011).

Figure 5.3 compares the shelter costs per unit area in recent disasters in other countries and Figure 5.4 shows a comparison of the different household support packages provided in Japan. Two points worth noting are that: a) in line with regulation, none of the compensation packages are sufficient or intended for the reconstruction or repair of private housing and b) even with the extension of rental subsidies by a further 12 months, the cost of this transitional shelter option is far less than the pre-fabricated temporary housing units.

Market interventions were also in place. As well as small-scale initiatives to promote handicraft production for people in temporary settlements (IRP, 2012), plans for commercial space similar to those for temporary housing were proposed including: designating temporary sites and spaces for recovering commerce [KII Ref] and providing limited-term rental support for industrial machinery (Iwate Prefecture, 2011).

The unit costs of temporary housing were also controlled through prior arrangements with suppliers, open tenders (advertised internationally but required to be in consortium with a Japanese partner) and prior standardised, detailed specifications (Federation of Housing and Community Centers, 2011).

In terms of support to facilitate reconstruction, the average cost for rebuilding a house is reported to lie between JPY15m-30m with repair costs above 3m but reconstruction of private assets was not facilitated by the transitional strategy, not permitted in some locations and not expected for households from areas that would be re-planned. Land compensation reported in one municipality was a payment of 70% of the price of unusable land (Oska, 2013). Government reconstruction grants were made available from October 2011 that could cover up to 10% of costs for mortgaged properties and 5% for those owned outright.

Residential insurance penetration, as noted, was 23.7% nationally and 33.6% and 14.6% in Miyagi and Fukushima respectively. The simple claims adjustment system allowed 60% of claims to be paid within 2 months, 80% within 10 weeks and 90% within 5 months (IMF, 2012; World Bank, 2012; EEFIT, 2013). This represented 741,000 claims with a value of JPY 1,200bn (47% of total insured residential losses and approximately 20,000USD per claim).

GoJ met the costs was through a combination of the national government issuing bonds, enforcing extra taxes and public sector pay cuts and through private means (including insurance) to cover residential and some commercial property.[s1] 


Comparison of Different Unit Costs and Losses

Range reported or estimated in USD

Cost of temporary housing unit (5,000,000 JYP)



Cost of rental subsidy (1,440,000 JYP for 24 months)



Average loss per home (total losses divided by units damaged)



Reported reconstruction cost per home (Oska, 2013) 15-30,000,000 JPY



GoJ reported grants per household 1.5m-3m JPY



GoJ estaimted average grant per household (total grants divided by total damaged units)



Average value of insurance payout per claim








Range suggested in USD (based on 80.98 JPY to 1 USD)

Total losses (USD)



Private buildings (residential, commercial and industrial)




Total residential losses (assuming 23,7% penetration, means 31.2bn is roughly 23.7% of total)



Public infrastructure



Total insured losses



Total insured residential losses (78% of insured losses)



private nonlife insurance (56%)



cooperative mutual insurance companies (44%)



Total destroyed or damaged homes



Estimated residential reconstruction cost based on total units and reported average cost per unit (Oska, 2013)



Cost to GoJ of disaster relief (incl temporary housing) 2011-2013 budget)





Cost of temporary housing unit (5,000,000 JYP)



Cost of rental subsidy (1,440,000 JYP for 24 months)



Cost of other relief



Cost to GoJ of waste disposal (incl temporary housing) 2011-2013 budget)



Cost to GoJ of total reconstruction grants



Total insured residential losses



Claims paid out within 5 months



Approximate shortfall to be made up by private means






5.3 Settlements and Locations

Space availability (location, quantity and quality of space) and land availability (ownership and lease options) are factors in designating sites for temporary housing.

5.3.1 Space

In terms of space, at least three issues have been observed in the literature and field visits:

·         A reluctance on the part of the affected population to occupy designated temporary sites because of fear of tsunamis and perceptions that sites were too close to the sea (e.g. a park designated in the Sendai port area) or because of exposure to other hazards (e.g. Matsushima town where the Roke typhoon on 22nd September 2011 inundated buildings, nearly reaching the water marks left after the tsunami and whose “heavy rainfall also caused landlsides in the tsunami-affected areas near temporary shelters on high ground or near mountains”) (Suppasri et al., 2014)

·         conflicting uses or demands on sites where space was also needed for debris sorting and storage or apparently available spaces that would have eaten into school grounds or sports facilities (e.g. the Ooya Green Sports Park in Motoyoshi) (Johnson, 2011) 

·         temporary sites with poor access or transport links where measures included waiving parking charges (MHLW, 2011) and including a space allocation “transitional car parking”[§]

5.3.2 Land

With construction immediately prohibited in the disaster-affected areas – legally and practically because of debris – sites for temporary housing were limited and sometimes at some distance from people’s communities of origin. This also caused delays in delivery of temporary units (IRP, 2012). Compounding these issues was difficulty identifying and getting permission from private land owners[**].


Areas where building was to be prohibited had been designated within months by April 2011 and were enforced by national law (ADRC and IRP, 2011b). Strict interpretation of national law would have meant not building temporary housing on private land but this was relaxed in some cases to allow fixed-term leases of privately owned land[††] (but no declarations of eminent domain as in other disasters).

5.3.3 Settlement

After the Kobe earthquake which hit a dense, established city, the few temporary sites that could be set up held a large number of households and individual households found themselves in an impersonal and isolating situation (Edgington, 2010; Habitat International Coallition, 1996). Learning from this, the GoJ adopted a principle of networked relocation, wherein people were encouraged to organise into groups of 5 households before being allocated temporary housing (IRP, 2012).

The data suggest that the average number of units per site was 44 in Iwate, 56 in Miyagi and 89 in Fukushima (ADRC and IRP, 2011a). For comparison, after the Kobe earthquake 16 sites hosted 4,400 units each with two further sites hosting 1,000 units each (Edgington, 2010).




5.4 Design, construction, specification and life-cycle

Among the criteria applied in the invitation to tender were a minimum production capacity for suppliers of 100 units (excluding all but the largest suppliers) and lead time to completion of 2 months; a 2 year aftercare agreement; structural performance criteria[‡‡]; and fire insurance premiums. The price per unit was not fixed but the tender invitation gave an indicative price based on previous emergencies of 5,000,000 JPY(shown for comparison in Figs XXX and XXX) (Federation of Housing and Community Centers, 2011).

The criteria also required as an essential condition of contract that housing specifications be “determined so that materials can be easily recycled after demolition”. It also offered suppliers a lease option: where units would be returned to the supplier after two years. These both presuppose that demolition and recycling are the likely end of life option for these units. Evidence from other post-disaster settings suggest that the fate of temporary units other than demolition include: sale, squatting or renting, all unlikely given the regulated context of housing and temporary housing in Japan; refurbishment for future disasters; reuse as community buildings; integration into a new core house (C. Johnson, 2007).

Although the authors could not find any systematic analysis of reuse and recycling of units after disasters in Japan, in 1999 the Japanese Ministry of Foreign Affairs reported that 500 of the temporary housing units used after the Kobe earthquake and no longer occupied because people had moved into longer-term alternatives, would be sent to Kosovo to house returning refugees (MoFA, 16/071999).

The current condition of the temporary housing units installed after the GEJET is not known. However, now that their occupation has exceeded their specified life-span and aftercare agreements, tracking their ongoing maintenance costs, disposal costs and reuse value is an important indicator in overall cost-benefit evaluation of the recovery process.

5.5 Coverage and Speed

The first rental housing units were identified as ready or available within 11 days (MHLW, 2011) while the first temporary housing units were nearing completion after 4 to 8 weeks. Within four months of the disaster, 75% of the 450,000 people who had sought refuge in evacuation centres had been able to move to alternative accommodation (IRP, 2012).

Although the pace of delivery was criticised (Washington Times, 2011), Figure 5.1 shows the progress of delivery by shelter option following the earthquake and tsunami in Japan and Figure 5.2 shows the progress following the earthquake in Haiti in 2010.

By May 2011, the Ministry of Environment had set a deadline of March 2014 for the debris clean up: this was to be a 3 year operation (UNEP, 2012). The life-time specified for temporary housing and the duration of rental subsidies set out in the initial plan (and in law) was two years but the extent of post-tsunami land use changes and the sheer scale of debris meant that rental subsidies had to be extended (Asia and Japan Watch, 2013) and temporary housing had to be upgraded (IRP, 2012).


Figure 5.1 Number of transitional shelter packages delivered after GEJE (ADRC and IRP, 2011a; BRI and NILIM, 2011; IRP, 2012; MHLW, 2011)


Figure 5.2 Number of transitional shelter packages delivered after the Haiti earthquake (EPYPSA, 2011; UCLBP and IASC CCCM Cluster, 2013; UN-HABITAT, 2012)



Figure 5.3 International comparison of shelter projects (Ashmore et al., 2012, 2011, 2010; IRP, 2012).


Figure 5.4 Household support packages in Japan (Ashmore et al., 2012; Brasor and Tsubuku, 2011; Federation of Housing and Community Centers, 2011; IRP, 2012; Japanese Red Cross Society, 2011; Japan Statistics Bureau, 2008)




6. Discussion: pre-conditions and surprises

Several aspects distinguish this transitional shelter response from other cases at a similar scale. Firstly, prior debates and norms were in place: citizens were registered and public and private responsibilities were demarcated in regulation. This meant that certain options were excluded including support for immediate reconstruction and repair and cash hand-outs. There had been criticism and analysis of previous disaster responses and apparently learning about the consequences for vulnerable groups of protracted periods in temporary housing on large, remote sites had been institutionalised.

Secondly, a recent census had been carried out and housing surveys were in the public domain which allowed rental stock to be identified, promoted as an option and then subsidised through direct cash transfers to households. These data and the sharing of this information allowed this to happen within days.

Thirdly, detailed socio-spatial and disaggregated analysis of previous events showed how high-level policies have played out for different people and the granularity of population data allows the allocation of housing to be evaluated relative to pre-disaster housing conditions by location and by family type and status[§§] (Edgington, 2010; Hirayama, 2000; Horita, 2006; Maly and Shiozaki, 2012).

Just as for housing policies generally, it is at the municipal level that the trade-offs play out between finding safe sites for resettlement and isolating people socially or economically; between defining eligibility criteria to target the most vulnerable and stigmatising certain groups; and between the households, regardless of whether they qualify for immediate relief, that are able to access or capitalise on other sources of support or finance and those that cannot.

6.1 Flexible, high level strategy: accommodating local differences and longstanding stigma

Within and between prefectures, there were significant prior variations that affected the post-disaster response. In terms of the local population, economy and access to markets and services, remote, proudly self-reliant fishing villages shared the coastline with marinas and tourist resorts, ports, industrial food processing facilities, a steel plant and two nuclear facilities but people were already choosing to migrate away from these areas.

Time series data for net migration from 1993 to 2012 shows that the three prefectures had been experiencing outward migration until 2008, particularly Fukushima, although in 2008 the rate of outward migration had fallen slightly, most steeply in Miyagi. Some key informant interviews suggested that there were declining opportunities for young people to find work locally[***]. Supporting evidence for this claim is beyond the scope of this report but it appears to track long term national trends of an aging population, a declining agricultural sector, an urbanising population and a general increase in youth unemployment (OECD, 2008; World Bank and Commission on Growth and Development, 2008; Zetter, 1986).

The pattern of housing tenure has in Iwate and Fukushima has thus remained less “urbanised” i.e. there are fewer indications that a shortage of supply of land and housing has pushed up prices and created a rental market. This is not the case in Miyagi where the proportion of households privately renting is approaching the rate of owner-occupation and average rents are higher than in Iwate and Fukushima.

The housing and land survey gives average monthly rents in Tokyo, Iwate, Miyagi and Fukushima as 77,000, 40,000, 47,000 and 40,000 Yen respectively (Japan Statistics Bureau, 2008). The proportion of households in public sector rental housing is small suggesting that this was not a prevalent housing option prior to the disaster, unlike urbanising Tokyo where this is higher as a proportion.

These pre-conditions – unpopular public housing and under-occupied private housing – meant that the take up (Table 5.1) of the public housing option was far lower than anticipated (based on previous disasters) and private rental apartments ultimately proving to be more “popular due to lower prices, higher comfort, and greater versatility” (IRP, 2012).

6.2 Rental subsidies and pre-fabrication: controlling prices, avoiding handouts and a history of manufacturing and innovation spurred by constraints on land speculation

For those unable to find or unwilling to relocate to rental accommodation, the alternative was temporary housing generally on a designated temporary site. There was high-level recognition of the trade-offs between safe sites for temporary settlement and the need to make sure communities are connected to each other and their sources of livelihood (promotion of networked relocations) and between promoting self-reliance and the reality that elderly and low-income groups have fewer options and resources (prioritising vulnerable people for temporary housing).

In addition, the procurement of temporary housing units was based on pre-arranged contracts between prefectures and housing suppliers and an industrial supply chain already configured to react quickly and deliver thousands of units. This supply chain emerged from a number of specific historic, cultural and economic characteristics of the housing market in Japan.

There was also a pre-defined and harmonised national specification which, among other criteria, set standards for floor space. Moreover, these standards had been set in a context where national space standards for ‘normal’ housing existed already and the relationships between floor space, tenure and geographic location were open to analysis.

These pre-conditions did not entirely forestall problems on the ground. Firstly, several factors had an impact on the amount of space available for transitional settlement sites and for what might be considered an appropriate and adequate standard for covered living space[†††]. These included the obvious localised impacts that the disaster had had on land and housing and whether the coastal areas were flat (more space but more damaged land) or steep (less suitable space) or had been previously inhabited or uninhabited (Ashmore et al., 2012; IRP, 2012). The average floor space of normal housing was two to four times the standardised 30m2[‡‡‡] temporary units and differed by prefecture. Table X shows the average floor areas of dwellings in Tokyo (64m2), Iwate  (126m2), Miyagi  (100m2) and Fukushima  (117m2). Again Miyagi is the exception: it has more apartment dwellers (pressure on land encourages vertical construction and multiple occupancy), more RC frame buildings and fewer households owning land.

Secondly, and related to the pre-disaster living conditions the popularity of pre-fabricated housing in Japan - particularly among higher income groups (W. Johnson, 2007) - did not necessarily carry over into acceptability or full occupation of pre-fabricated temporary housing. This standard, low cost specification is a very different product from high end, customised housing units in terms of floor area, appearance and finishes (as shown by comparing Figure 5.6 and Figure 5.6). Thus, temporary housing was not always the option of choice when there were alternative transitional shelter options available, notably subsidized private rental housing.


Thirdly, the number of units procured for this response outstripped the pre-positioned supplies and the delivery and construction capacity. Problems included bottlenecks in labour and material supplies and in poorly insulated, poorly built units arriving on site (IRP, 2012).

Finally, the Japan Times reported criticism that Iwate, Miyagi and Fukushima only procured 18, 1 and 36 per cent respectively of their temporary housing locally (Brasor and Tsubuku, 2011). Indeed, when prefectures began to procure temporary housing beyond their pre-arranged contracts, an international tender was launched (Kitamura, 2011) but it is notable that the tender invitation, released to make up a shortfall of about 7,000 units, set as a requirement an apparently low threshold of production capacity – 100 units (in fact this threshold excludes all but the largest Japanese players) – but only included as a “consideration” the “utilization of local materials as well as the creation of job utilizing local construction workers”. Evaluating the impact of systemic measures to stimulate the local, rather than prefectural or national, economies via the construction of temporary housing is beyond the scope of this paper.

Despite its success in terms of speed, coverage, quality and cost control, this transitional shelter strategy was plagued by challenges common to other transitional shelter strategies including:

·         Local preferences that were different from the numbers planned and anticipated by the national government;

·         Difficulties of meeting (and being seen to meet) equitably the needs of the most vulnerable with limited resources;

·         Delays in finding space and securing land for temporary sites and housing units;

·         Downward pressure on quality and upward pressure on costs of temporary housing; and

·         Dissatisfaction with the quality of housing units and a longer than anticipated stay away from home.




Table 5.2 Pre-Disaster Options and Choice


Net Migration











Table 5.3 Pre-disaster Living Space and Land


Type of housing

Type of construction

Land ownership












Table 5.4 Pre-disaster location, livelihood and care



Urbanisation, livelihood, economy



Figure 5.5 Predominant transitional shelter option in each affected prefecture (IRP, 2012)








Table 5.5 Summary of characteristics by prefecture compiled by the author from (ADRC and IRP, 2011a; IRP, 2012; Suppasri et al., 2013)



Housing totally collapsed

Land area affected by the tsunami

Number of temporary housing sites under construction by June

Number of temporary housing sites still in planning  by June

Temporary housing planned

Temporary housing as a percentage of totally collapse housing

Preferred option


Ria-Coastal Areas







Mostly temporary housing (prefabricated)


Ria  coast (Kesennuma to upper Ishinomaki); Coastal plains (Lower Ishinomaki to Yamamoto)







Mostly private rental apartments


Coastal Plains







Mostly private rental apartments



Table 5.7 Average floor areas by tenure type, Japan (Japan Statistics Bureau, 2008). Areas are in square metres (m2) and represented as squares for comparison but not to indicate the footprint of actual floor areas.

Owned houses

Rented (owned by local government, Urban Renaissance Agency or public corporations)

Rented (owned privately - wooden)

Rented (owned privately – non-wooden)

Average floor areas in Japan by tenure















Table 5.8 Average floor areas by prefecture (Japan Statistics Bureau, 2008)





Average floor areas in affected provinces











7. Conclusions: how is this relevant elsewhere?

The government of Japan set a clear policy framework that had been informed by experience particularly after the Kobe earthquake; by evidence thanks to national and publicly available data with the open research that this then allows; and, by a social contract that defined public expectations of government and determined prior to the event what people would be entitled to, how temporary units should perform and how much they would cost.

The transitional shelter response to the large scale destruction of housing following the GEJET had a number of features that were specific to Japan not least the preference for pre-fabricated units over the distribution of cash of an equivalent value because:

·         prior norms discouraged hand-outs and defined temporary housing as a service not an asset. Although direct cash transfers are fungible, flexible and easy to transfer in Japan, this option was limited to small packages of compensation or the in-kind welfare benefit of a temporary housing service with the expectation that all but the most vulnerable will meet the costs of rebuilding their homes by private means.

·         transitional shelter included public and private rental subsidies and cash of equivalent value to pre-fabricated housing would have dwarfed the rental subsidy creating either a more expensive response or one that could be perceived as inequitable.


·         cash could not solve the housing problem: there were just not enough available housing units in the right locations to meet needs. the massive loss of land and housing stock meant that private and public rental housing was not available immediately and could not accommodate all those affected and options for relocation and migration were initially limited because such a vast area had been affected

·         the speed, cost and quality of temporary housing was achieved through prior agreements with pre-fabricated housing suppliers or tenders based on a standard specification..Delivering against these agreements was possible because of the internationally unique capacity of these suppliers.

·         reconstruction of private assets was not facilitated by this strategy, not permitted in some locations and not expected for households from areas that would be re-planned.  Insurance, with its low coverage and small average claims, was not an instrument for accelerating reconstruction, though it cushioned recovery for some. Reconstruction grants were made available from October 2011 that could cover up to 10% of costs for mortgaged properties and 5% for those owned outright.

This paper argues that this was possible because of a number of specific, historical conditions and the particular geography, people and places that were affected by this event.

7.1 Learning for other regions and cities in Japan

The transitional shelter policy in Japan played out differently in urbanising areas with, for example, higher take up of the private rental option in areas with pre-existing “urban” features such as lower rates of owner-occupation of land and buildings, higher rates of privately rented apartments.

This has implications for the future: lessons learned from an urban earthquake (Kobe) did not apply wholesale to a coastal tsunami (GEJET) and lessons from the GEJET may require interpretation before their application to disasters in other parts of Japan.

Japan has long been planning for its next massive urban earthquake. Sophisticated simulations  of the post-disaster shelter options that households might choose after an earthquake in Tokyo have been developed as a way to inform decision-making (Sato, 2011). These simulations suggest that in Tokyo, the need for shelter would outstrip Japan’s annual capacity to supply pre-fabricated temporary housing, not to mention the space and land constraints that will make it difficult to install.

There is a need, then, to identify spaces and build an inventory of vacant or unoccupied public and private housing stock to inform the post-disaster shelter response. This level of preparedness – one that depends on data, city-wide and community planning – is an essential complement to ongoing efforts to prepare as individuals and communities with training, stocking up on household supplies, putting aside "grab-bags" with essential documents and being aware of the needs of vulnerable neighbours.

7.2 Learning for other national and municipal contexts

Prefabricated housing has played a critical role in this and other Japanese post-disaster responses and, following similar learning, the pre-positioning of transitional shelter materials and standard designs is now being pursued at a strategic level in other countries, for example, Peru, frequently affected by disasters (USAID, 2012).

Two features distinguish this approach in Japan and made it a viable option. Firstly, the strategic assumption and public expectation that this temporary housing would arrive within weeks. Secondly, that the specification of temporary housing – although its size and cost were markedly higher than in other contexts – had been pre-agreed in light of both national space standards (Fukushige and Ishikawa, 2013) and evidence on the ‘normal’ living conditions of different socio-economic groups (Housing Stats 2008). In other settings, this standard-setting process often happens in real-time with limited information against which to benchmark and, as a consequence, disagreement over standards (HERR, 2011).

Critically, seen in the context of housing policy, attitudes to welfare and public housing, temporary housing in the Japanese context is perhaps better described as provision of a temporary housing service for a limited period of time rather than provision of a housing asset that will become a permanent home, with all the in-situ investment and upgrading (or reinstallation in a new location) that this might entail. This and the relatively small role played by grants (as a percentage of overall reconstruction costs) and insurance (as a percentage of the population) also challenge thinking that justifiesthe high costs of transitional shelter kit assets  by equating it with the value of housing losses (usually far higher than a kit).

Where these expectations are not proscribed or clear, questions about whether high value household assets can realistically be distrusted equitably must surely be asked. Further, where there is no prior local capacity to deliver pre-fabricated housing in this way, assumptions about speed, cost, quality and land availability should be documented in any account of strategic decision-making and then tested and evaluated against the structure and capacity of the firms and builders that were supplying housing before the disaster. In particular, this option should be considered against the costs and timeframes of possible alternatives that affected people might choose to pursue in the absence of clear policies or funded options and with low expectations of receiving support.

Japan is not alone in the world in having both a recent census and housing surveys in the public domain nor institutionalised and regulated prior norms for welfare or disaster response. International humanitarian response could benefit from an audit of these data in priority countries and cities, not to identify the barriers to their own intervention (XXX) but the systems and support already in place to facilitate recovery.

Decisions about assistance packages were dependent on fundamental and distinct assumptions about how long people might need and expect assistance to last, which, in turn, was based on the government’s estimates for how long it would take to: a) allocate alternative housing and deliver temporary units; b) clear debris; c) reinstate public infrastructures; d) plan or re-plan areas for reconstruction; and e) rebuild private assets. Although the government was well placed from previous experiences, it did not necessarily get these numbers right but because these statements were in the public domain, they were held to account and had to explain why they were realistic and reliable.


Perhaps, then, key points for comparison are not what was achieved after the GEJET but what was assumed to be possible in the first instance and the reliability of public assumptions.


[*] Refers to region of six prefectures (ken): Akita, Aomori, Fukushima, Iwate, Miyagi and Yamagata.

[†] Evacuation structures and evacuation centres are not covered here. Chapter 10 and Section 11.1.5 of the 2011 EEFIT Report deal with this issue as do other recent articles (EEFIT, 2011; Suppasri et al., 2014, 2013).

[‡] Key informant interview, Arup 8th June, Tokyo

[§] Key informant interview, 6th June, community mobiliser Miura San, Oye District

[**] Key informant comments, 20th November 2013

[††] Key informant interview, 6th June, community mobiliser Miura San, Oye District

[‡‡] 1m snow loads and 30-34m/s wind speeds

[§§] Data exist at this level of granularity in Japan but were not accessed by the EEFIT team, either because they are not in the public domain, subject to data protection or not translated.

[***] Key informant interview, 6th June, community mobiliser Miura San, Oye District

[†††] The international humanitarian standard for covered living space recommends that “people have sufficient covered space to provide dignified accommodation…” where “essential household activities can be satisfactorily undertaken, and livelihood support activities can be pursued as required.” but emphasises that this must be appropriate and adequate for the specific context.

[‡‡‡] the floor areas of rental apartments obviously varied depending on what was typically available. Temporary housing units were largely single storey, serviced pre-fabricated units requiring utility connections and basic services on site.


 [s1]Joey, do you know if Oska,s figures were for Ishinomaki only? Or an average for the whole response?


This section is my attempt to put together the costs, losses and insured losses in the residential sector against the costs and funding of the transitional strategy and full reconstruction.


I have tried to pull together the figures from your original chapter and they work out at similar orders of magnitude by different ways of estimating.


Do you have a figure for total residential losses that does not include commercial and industrial? I have back calculated a rough number but if you have a definitive one, all the better.