Community-based feedback to improve land pooling for planned urbanization: A case study of Thimphu, Bhutan

Land Use Policy xxx (xxxx) xxxx

Contents lists available at ScienceDirect

Land Use Policy journal homepage: www.elsevier.com/locate/landusepol

Community-based feedback to improve land pooling for planned urbanization: A case study of Thimphu, Bhutan☆,☆☆ Katie Farrin*, Jade R. Laranjo, Nina Ashley Dela Cruz Asian Development Bank, Philippines

A R T I C LE I N FO

A B S T R A C T

Keywords: Urbanization Land pooling Land acquisition Hedonic pricing

Land pooling can potentially be an effective strategy for public infrastructure projects, especially in small but growing cities where rural-to-urban transition gives rise to the prospect of well-planned greenfield development. Seen as a more inclusive and less costly alternative to land acquisition, land pooling has been used for urbanization projects in Japan, Germany, Korea, United States, Australia, India and Thailand, among others. However, little empirical evidence exists on the experience of communities who participate in land pooling—particularly in developing Asia. This paper reviews the literature to compare outcomes of land acquisition and land pooling; presents a simple model to illustrate the gains from land pooling; and, to evaluate land pooling in practice, uses an original primary data set from 1200 households in four Local Area Plans (LAPs) in southern Thimphu, Bhutan, to assess land price growth as well as project beneficiary satisfaction with the consultation, planning, and construction phases of a land pooling project. Results indicate that land prices in the LAPs increased, due in part to additional public services but in large part to a rise in urbanization over time. Among original landowners, results from a community feedback survey show that not all households have access to the services that were to be provided when land pooling was implemented. Thus, land pooling participants benefited from the increased land prices but not all were satisfied with the quality of works, particularly in terms of a lack of maintenance and recreational areas. Increased property prices also have implications for the availability and affordability of rental housing, as no pooled land was reserved for low-income housing in the LAPs. The case study has implications for government and multilateral development bank policy for land pooling and similar mechanisms for planned urbanization.

1. Introduction Land ownership may be worth more than the market price of the land, and land’s intrinsic value may be even higher among low-income people. Especially in rural areas, land serves many purposes for the poor: as a productive livelihood and hedge against food security; a foundation for the formation of social networks; collateral for borrowing; insurance against unemployment; an access point for public services; a facilitator for political and civic engagement; and a basis for shelter (Cotula et al., 2006). The rise of urbanization has put mounting pressure on agricultural lands, and displacement of rural people in the name of city expansion and urban development often means that original owners of or settlers on the land cannot share in the gains of

urbanization; in fact, they most likely suffer the largest welfare losses—particularly if they are resettled outside of the newly urban area. An alternative to land acquisition, land pooling is a land consolidation and redistribution technique used to redefine ownership of land and create a new configuration of plots that is more appropriate for urban use. Each landowner in a defined area makes a proportionate contribution of land for urban development; these contributions produce enough land for the provision of public roads, infrastructure, social facilities, and open space, after which the remaining pooled land is reshaped to fit an urban plan and returned to the original landowners. Land pooling may be an effective strategy for public finance of infrastructure projects, particularly in small but growing cities, where ruralto-urban transition gives rise to the potential for well-planned

☆

Katie Farrin is an Economist with the Millennium Challenge Corporation, Washington, DC, USA; Jade R. Laranjo is an Associate Economics Analyst and Nina Ashley Dela Cruz is a consultant with the Economic Research and Regional Cooperation Department, Asian Development Bank, Manila, Philippines. This work was funded by the Asian Development Bank. ☆☆ This paper represents the personal views of the authors and does not necessarily reflect positions of the Asian Development Bank or the Millennium Challenge Corporation. ⁎ Corresponding author. E-mail address: [email protected] (K. Farrin). https://doi.org/10.1016/j.landusepol.2019.104257 Received 29 April 2019; Received in revised form 22 August 2019; Accepted 29 September 2019 0264-8377/ © 2019 Published by Elsevier Ltd.

Please cite this article as: Katie Farrin, Jade R. Laranjo and Nina Ashley Dela Cruz, Land Use Policy, https://doi.org/10.1016/j.landusepol.2019.104257

Land Use Policy xxx (xxxx) xxxx

K. Farrin, et al.

greenfield development.1 Not only is land pooling potentially a more inclusive means of ruralto-urban transition, where original landowners maintain a stake in urban development, it also can reduce the public finance burden of urban development by reducing or eliminating the cost of land acquisition. Land pooling and similar mechanisms such as land trusts and town planning schemes are likely to be increasingly used in urbanization projects where land acquisition would otherwise create budget constraints and time delays for infrastructure building and planned urban expansion. Without timely implementation of urban planning, unplanned urbanization can lead to increased environmental deterioration, settlement on marginal lands and degradation of basic services such as drinking water, sanitation and waste disposal and treatment. Urban sprawl can also occur, increasing energy inefficiency and the cost of community infrastructure and public services (Bhatta, 2010). Land pooling can mitigate the negative impacts of unplanned urbanization and avoid issues common to land acquisition in developing countries: insufficient (i) laws and regulations; (ii) procedure for land acquisition and compensation; and (iii) land valuation approaches for compensation (Ghimire et al., 2017). In Bhutan, planned infrastructure spending by the public and private sectors is more than 6 percent of the country’s GDP; the majority (more than 5 percent of GDP) of infrastructure investment comes from the government budget (ADB, 2017a). Fueling the need for large infrastructure expenditures is the speed of urbanization in Bhutan since the 1960s, when fewer than 5 percent of the population lived in urban areas; as of 2018, urban dwellers make up more than 40 percent of Bhutan’s population (Fig. 1). Thimphu, Bhutan’s capital and largest city, saw rapid population growth from rural-to-urban migration in the late 1990s and early 2000s, prompting the government to institute a comprehensive plan for urbanization (the Thimphu Structure Plan, or TSP, which was approved in 2003); part of the TSP required preparation of Local Area Plans (LAPs) to detail the greenfield development of infrastructure networks and other amenities in what would be small, planned cities (Norbu, 2018). Urban infrastructure—including roads, water and sewerage, and power and telecommunications systems—was constructed in the newly formed LAPs of Thimphu, where residents ceded some of their land in exchange for the provision of public amenities. The land in the LAPs had been largely used for agriculture but was subject to zoning changes when the LAP areas were incorporated into the Thimphu municipality; the land use change allowed for organized urban expansion and increased densities in the LAPs. In four LAPs in southern Thimphu—Babesa, Changbangdu, Lungtenphu and Simtokha (“southern LAPs”)—where urban infrastructure was funded by the Asian Development Bank (ADB, 2017b), land pooling contributions were close to (but not exceeding) 30 percent. LAPs were also planned with designated space for community amenities like recreational areas, police stations, healthcare facilities and schools. With the land pooling and infrastructure works project completed, landowners in the land pooling area generally have (i) higher total incomes than renters, with a large proportion of income earned through property rental; (ii) non-agricultural jobs, having shifted out of agriculture to other employment sectors when their land became incorporated into Thimphu’s municipal area; and (iii) greater access to

Fig. 1. Urbanization Dynamics in Bhutan, 1960-2018. Source: World Bank World Development Indicators, accessed 30 July 2019.

public services. Land values in the southern LAPs have increased significantly, which implies the municipality can potentially increase landbased tax revenue in the newly urbanized areas.2 However, welfare outcomes are more varied when households describe their experience with timeliness, quality, and completion of works they expected under the LAP. Further, the pace of urbanization has resulted in strain on the built-up public infrastructure, with some public services poorly maintained or unavailable for some households. Results are even more skewed when we consider the livelihoods of renters, who face high costs of living and rental poverty as a result of the property price boom. The study contributes to the knowledge base on land pooling for urbanization by: 1) Documenting outcomes of southern LAP land pooling projects in Bhutan; 2) Conducting selected analytical exercises to quantify the expected benefits that can be generated through land pooling; 3) Discussing the potential for expanding the use of land pooling as a public finance mechanism for infrastructure projects—not only in Bhutan, but more generally in urbanization projects in developing Asia; and 4) Offering policy implications and considering changes that may be necessary to successfully implement land pooling-based infrastructure financing projects. The rest of the paper is organized as follows. Section 2 reviews the literature on land acquisition and land pooling; Section 3 presents a simple theoretical model of land acquisition versus land pooling; Section 4 details the project area and land pooling timeline in the southern LAPs of Thimphu; Section 5 summarizes the data and data collection process for a household survey of original and new landowners, as well as renters in the area; Section 6 estimates a hedonic pricing model to show how public works have affected housing prices in the urbanizing area; Section 7 outlines results from the qualitative portion of the household survey, where respondents rate the various stages of land pooling implementation; and Section 8 concludes with policy recommendations for land pooling. 2. Comparing land pooling and land acquisition: literature and experience 2.1. Compulsory land acquisition for urbanization

1

Greenfield development refers to development on previously undeveloped land, as opposed to brownfield development, which refers to re-development or infill in previously urbanized or industrialized areas. In Thimphu, while the redrawing of plot lines through the land pooling process ensured no destruction of structures, very few structures (which largely consisted of household homes and barns) existed in what had been a rural, agricultural area. Landowners were compensated for the loss of crops and orchard trees due to the land pooling process.

The objective of land acquisition is land assembly for infrastructure 2 While issues of tax reform and increased tax efficiency are beyond the scope of this paper, the need for improvements in tax-take to finance operating, maintenance, and capital expenditure costs of urbanization-related infrastructure is a lesson learned from this case study.

2

Land Use Policy xxx (xxxx) xxxx

K. Farrin, et al.

homelessness, marginalization, health risk, loss of access to public services, and social disarticulation (Patel et al., 2015). City upgrading projects are particularly harmful to the displaced populations when resettlement sites are distant from the original settlements, far from employment opportunities, and with little access to public transportation. (Several case studies provide detail on development-induced displacement effects on informal settlers, where resettlement locations put the displaced population at a disadvantage. See, for example, Abebe and Hesselberg, 2013 for Addis Ababa, Ethiopia; Nizuke et al., 2019 for Kigali, Rwanda; Patel et al., 2015 for Ahmedabad, India; and Shaw and Saharan, 2018 for Kolkata, India.)

development that supports urbanization, industrialization or agricultural productivity growth; its mechanism is the transfer of land from local communities to the government. The design and implementation of compulsory land acquisition policy (also called eminent domain) varies across countries, but shared experience suggests the persistence of common challenges: mass acquisition of land to expand urban center boundaries; relocation of affected populations to poorly developed areas; persistent land-related conflicts; inadequate or late compensation for acquired land; low credibility of or corruption within the institutions responsible for land acquisition; and displacement of indigenous communities from their traditional employment and livelihood (Lozano-Gracia et al., 2013; Kusiluka et al., 2011; Sarkar, 2007). Moreover, while the impact of lost livelihood varies widely by country, resettlement projects largely affect rural communities in countries where income inequality is already pervasive. The most heavily impacted areas (measured in lost agricultural income for affected populations) are sub-Saharan Africa and Southeast Asia (Davis et al., 2002). Long-term effects—such as persistent food insecurity—have also been observed where people are displaced from ancestral lands (Valente, 2009). Recommendations for safeguarding the welfare of affected indigenous communities include providing adequate consultation prior to land acquisition exercises; fair and prompt compensation; and strengthened programs that increase awareness of land laws and policies, as well as access to formal institutions responsible for the provision of justice (Kusiluka et al., 2011). Adverse effects of compulsory land acquisition policy on local communities have been exacerbated by the speed of urbanization, and many governments have enacted welfare-improving changes to laws that regulate the implementation of land acquisition and resettlement programs in response to pushback from the public. For example, Malaysia’s Land Acquisition Amendment Act (2016) addresses land acquisition issues in response to a scarcity of suitable land for underground mass rapid transit (MRT) and high-speed rail (HSR) infrastructure development. In addition to the provisioning for the acquisition of underground land, landowner compensation limits for appeals (i.e., objections to the compensation amount) increased by 66 percent for individual landowners and doubled for government and corporate landowners (Conventus Law, 2018). In India, the Right to Fair Compensation and Transparency in Land Acquisition, Rehabilitation and Resettlement (RFCTLARR) Act (2013) requires states to pay almost four times the market value for rural land and twice the market value for urban land that is subject to compulsory acquisition. Land acquisition under the RFCTLARR has led to delays of flagship urban development projects because rules on compensation, rehabilitation, and resettlement for affected populations are not consistent with several related laws, such as the National Highways Act of 1956, the Metro Railways (Construction of Works) Act of 1978, the Railways Act of 1989 and the Electricity Act of 2003 (KPMG, 2013). In China, the People’s Republic of China Land Administration Law of 1986 (PRCLAL) mandates compensation of up to 16 times the land’s average annual production value to agricultural households affected by compulsory land acquisition (Chan, 2000). Although revisions to land acquisition law are prolandowner, they also heavy the burden for public financing of urban infrastructure projects. While land acquisition for planned urbanization has its downfalls, the alternative of fast-paced, unplanned urbanization raises a number of challenges, including (i) inadequate supply of serviced land for housing development; (ii) rapidly increasing land prices; (iii) a lack of affordable housing for the poor and new migrants; (iv) squatting on public and private land; (v) underutilization of land to its full potential where tenure security is weak; and (vi) threats to food security and existing farmland due to informal settlement (Nguyen et al., 2018). Informal settlements that arise in the absence of sound urban expansion policy provide a unique challenge for urban planning, and development-induced displacement of informal or slum populations has been shown to have negative impacts including landlessness, joblessness,

2.2. Land pooling as an alternative to land acquisition: experience in Asia Land pooling, which is also known as land consolidation, land readjustment, land replotting, and land redistribution, is a land management technique that consolidates, reshapes, and reallocates land for the purpose of planned urbanization. Potential advantages of land pooling include land provision for public works at low or no cost to the government, shared costs and benefits of urbanization projects among landowners, timely distribution of urban-fringe landholdings for orderly urban expansion, mitigation of excessive land speculation, and delivery of a fair supply of land for new housing development (Archer, 1986). Land pooling has been widely used in Australia, Canada, Germany, India, Japan, South Korea, Taiwan, Thailand, and the United States, to name a few. The experience in these countries shows land pooling is politically, financially, and administratively feasible across a range of environments. Its transferability to developing Asia, however, is still being tested. The Japanese land readjustment model—adapted from earlier experience in Germany—is often used as a blueprint for similar land pooling schemes in Southeast Asia. Land readjustment was the country’s means of rehabilitation and urban reconstruction after natural disasters (the Great Kanto Earthquake in 1923) and conflict (World War II), and the practice has been institutionalized through the Land Readjustment Law (1954). Land readjustment projects can be initiated and executed by both public and private individuals and entities, with the 1954 law giving landowners formal rights to participate in the project implementation body and local public corporations a wider scope to implement land readjustment projects (Sorensen, 2000). By 2000 almost a third of Japan’s urban area (361,132 ha over 10,878 projects) had been redesigned through land readjustment; the use of land readjustment for urban development at this scale—in terms of total area and percent of urban area in a country—is unprecedented (Home, 2007). South Korea’s land pooling experience began under the Japanese occupation with the Urban Planning Act of 1934, which extended the city limit of colonial Seoul by incorporating suburban areas for the purpose of planned urban expansion. After the occupation and the Korean War, follow-up legislation for land pooling came in the form of the Land Readjustment Act of 1966 and the Urban Development Act of 2000 (Lozano-Gracia et al., 2013). The Korean model differs from the Japanese model in that little central government support was offered to finance land readjustment. Self-financing of land pooling projects was thus the norm in South Korea, with landowners contributing 30 percent of their land for public facility construction and an additional 20 percent for sales to cover project start-up costs (ADB, 2010); in Japan, the land contribution was significantly smaller: 20 percent for public infrastructure space and 10 percent for reserve land to be sold (Schnidman, 1988).3 Rapid urbanization in South Korea was enabled by 3 An infrastructure-finance-forward feature of land pooling schemes in many Southeast Asian cities is that pooled land can be reserved for sale—going beyond infrastructure development uses generally associated with the goals of land pooling—while ownership rights of pooled land parcels are returned to the

3

Land Use Policy xxx (xxxx) xxxx

K. Farrin, et al.

1976 (GTPUDA) built upon the updated Bombay Town Planning Act of 1954. Land pooling and readjustment picked up particularly in Gujarat, and more so after further revisions to the GTPUDA in 1999 (Yoshino et al., 2018). A salient feature of land pooling in Gujarat is the rollover of land pooling revenues for investment into newer land pooling schemes; because significant up-front capital expenditure needs for infrastructure and services are a recognized shortcoming of land pooling projects, the revolving funds mechanism used in the Gujarat Town Planning Scheme (TPS) hurdles this obstacle by capturing land value gain from land sales in older, more established TPS projects and using it to cover start-up costs of new TPS projects (Mathur, 2014). The city of Ahmedabad in Gujarat is one example of the success of the use of the TPS. Since the 1999 amendments to GTPUDA, 50 TPS covering over 5,000 ha have been completed and an additional 47 (adding almost 5,000 ha of coverage) are under preparation; the total population served by the post-1999 TPS in Ahmedabad is 1.5 million (Mathur, 2013). The unique attribute of the TPS that makes it a hybrid land pooling model is the transfer of payments between the government and landowners. TPS landowners do not simply donate land for development, but rather are compensated at lower, current-use rates for the portion of the land ceded by the household. However, landowners also then must pay betterment fees back to the government. Betterment fees are determined at the start of the TPS, in a process where infrastructure and administrative costs of the TPS are determined; compensation to each landowner is calculated at the semi-final value; a final (post-project) value for each plot is systematically assessed; 50 percent of the increase in land value is taken as a betterment fee; and the net value owed to the landowner is given as final compensation (Ballaney, 2008). Other innovations in land pooling are taking place in India. In one of the largest land pooling initiatives in India, Amaravati, the new capital of Andhra Pradesh, is being formed through the pooling of 33,000 acres of plots from land-owning farmers, with over 59,000 readjusted and serviced plots returned (APCRDA, 2017). Landless agricultural laborers working in the area are eligible to receive a “livelihood restoration benefits” package that includes: (i) payment of pensions of INR 2500 per month per family for a period of 10 years, annually adjusted for inflation; (ii) employment under the Mahatma Gandhi National Rural Employment Guarantee Act (MGNREGA) for up to 365 days per year; (iii) access to interest-free loans for enterprise development; (vi) access to skill upgrading and training opportunities; (v) provision of subsidized food; and (vi) free access to medical and educational benefits in designated facilities (APCRDA, 2018). Finally, recent legislation has paved the way for even more land pooling in India’s capital. The Delhi Land Pooling policy (DLP) was approved by Delhi Development Authority in September 2018, allowing the city to tackle its urbanization challenge: to develop the estimated 20,000–24,000 ha of land needed to accommodate the increasing population by 2021, the DLP liberalizes land policy to enable developers to directly acquire land from farmers or landowners through partnerships in place of forceful acquisition by the state. The DLP is expected to lead to the development of over 4 billion square feet of development over 10 years, including 1.45 million new residential units (KPMG, 2013). Mongolia’s Urban Redevelopment Law of 2015 serves as the legal basis for urban redevelopment through land readjustment. At least 80 percent of land rights holders must agree to a project plan before establishing a residents’ association and formulating a project plan to be approved by the district or city governor; rights holders can participate in these projects through a rights conversion process (land-to-land or land-to-apartment) without surrendering their land rights. A tripartite contract among the district governor, rights holders and the project implementation team sets rules for urbanization projects, including the option to sell reserved land or building floors to retrieve part of project costs. Land pooling in Mongolia has been part of the effort to control urban sprawl in the capital, Ulaanbaatar, where ger areas—traditional, unplanned, and un-serviced tent settlements that have been rapidly expanding around the capital—will be reorganized and serviced

its land readjustment policies; 84 percent of all its cities have used land pooling, and Seoul has redeveloped more than half its land area through 17 land readjustment projects. Another difference between the Korean and Japanese models of land pooling is a 1980 provision of Korean law that mandates a portion of land pooling proceeds be set aside to finance 30 percent of low-income housing costs (Schnidman, 1988). Thailand also adapted land readjustment from Japan, after a 1987–1989 Japan International Cooperation Agency (JICA) “Applied Technology for City Planning” program provided academic and technical advice to spur the launch of land readjustment in the country. A Land Readjustment Fund was established in 1993 to jumpstart land readjustment projects. The Land Readjustment Act of 2004 governs the implementation of land readjustment, outlining how the public and private sector can optimize urban planning and ensure active community participation in readjustment projects. As of 2010, land readjustment projects were ongoing in 13 Thai cities (Puasakul, 2010). Indonesia was the first Southeast Asian country to implement land pooling; a project in Bali began in 1981, covering almost 80 ha of land in Denpasar (Yoshida, 2003). Known as land consolidation in the local terminology, several land pooling projects were implemented before a formal legal mechanism was established through the Regulation of the Head of the National Land Agency No. 4 of 1991 (De Souza et al., 2018). Technical assistance (also from JICA) helped scale up the use of land pooling across the country, with 274 projects implemented in 27 provinces by the early 2000s (Yoshida, 2003). In Indonesia, however, land consolidation has been hampered by an insufficiently developed land titling and registration system (World Bank, 2007) and the absence of a central government funding mechanism to cover project costs in land pooling areas (Sitorus, 2005). The 1991 regulation explicitly requires agreement by 85 percent of rights holders, whose land covers at least 85 percent of the project area, for approval of a land consolidation project (De Souza et al., 2018); this is a significantly higher quorum compared to go-ahead approval rates for land pooling projects in other countries. In Taiwan, budget and personnel shortages faced by local municipalities spurred the Act of Promotion of Private-Owners Initiated Land Readjustment (1979); incentives such as tax deductions and low-interest loans were offered to encourage private owners to form a collective unit to undertake land readjustment by themselves (De Souza et al., 2018). An interesting feature of land pooling in Taiwan is a practice known as zone expropriation, which is founded on the country’s land expropriation law rather than the land pooling and readjustment law. Zone expropriation is, de facto, a modified and compulsory form of land pooling; project land is expropriated, serviced and subdivided into building plots with some of these plots (up to 40 percent of the project land) transferred back to the landowners as compensation for their expropriated land. Landowners can choose to be compensated in cash, but nearly all landowners choose to receive “equivalent-value land.” Local governments implementing the zone expropriation projects can also sell about ten percent of the project land as cost recovery plots (World Bank, 2007). In India, a rich history of a hybrid land pooling method for infrastructure development began in the early 1900s, with the Bombay Town Planning Act of 1915 serving as the foundation for the first legal framework for land pooling and readjustment in the country. Land pooling applications in India continued to gain traction after the partition of Bombay resulted in two states refining the land pooling law of the colonial period: the Maharashtra Regional and Town Planning Act of 1966 and the Gujarat Town Planning and Urban Development Act of

(footnote continued) original landowners in some proportion. In the case of land pooling in the southern LAPs of Thimphu, no reserve land contribution was required of land pooling participants. 4

Land Use Policy xxx (xxxx) xxxx

K. Farrin, et al.

land near its original location. There are several important parameters when assessing land acquisition outcomes of households under the four scenarios described above. The initial endowment of land for a representative farmer is L . Let pR , pN , and pA be land prices for rural, near-acquisition, and acquisition areas, respectively, with associated growth rates ΔpR , ΔpN , and ΔpA . Intuitively, land price growth rates will be lower in rural and urban periphery areas as opposed to the land acquisition area, where urbanization is in full swing and where public infrastructure will be built that will push up the value of nearby land ( ΔpR < ΔpN < ΔpA ). Land rental costs are a fixed proportion, θ , of land prices for all areas (this assumes rental price growth moves with land price growth). The displaced farmers who have become landless are assumed to be compensated a fixed proportion of the value of land, πR or πN , in the area in which they have resettled. Considering non-land income opportunities, let wR , wN , and wA be wage rates in the three areas. Rural wages for displaced households may or may not be higher than near-urban and urban wages despite higher land prices in urban areas, as (i) labor moving into the city might depress the urban wage; and (ii) skills of displaced households might be better suited to agriculture, resulting in a lower wage for urban jobs for which the displaced households are relatively less experienced. Finally, households who own land can rent all or part of the land (designate this choice proportion as γ ) and can sell all or part of the land (designate this choice proportion by λ ) in any given year (sales are assumed to occur at the end of the year, so rental income can accrue for all of the rented land before the sale). Landless households face a rental cost that is covered by wages and choice proportion, μ , of compensation money. As an alternative scenario, land pooling could be used for public infrastructure provision in an urbanizing area. Under land pooling, land owning households give up an agreed-upon portion of their land, α , without monetary compensation by the government but in exchange for access to public services, so that they keep the remaining (1 − α ) portion of their original land. Prices for this land—because it will be near the new public services in the urban area—will follow the same growth as acquired land under the acquisition scenario ( ΔpA ). After displacement through land acquisition or urbanization and no resettlement through land pooling, we can characterize farmers by their land or compensation assets and income opportunities in their resulting area of residence (Table 1): Rather than being displaced, farmers in the acquisition area would prefer land pooling if wages and land income opportunities (sales and rental value from the original land holding) in the acquisition area result in a greater stream of benefits over time relative to in near-acquisition and rural areas. Land replacement is always preferred to compensation within the same area. Using the simplest case as an illustration, we compare scenarios (i) and (v) where land prices are equal at the start of a 25-year land pooling project. If wages are also assumed to be equal and growing at an equal rate in nearby and project areas, the difference in benefits across scenarios depends on the share of land contributed to land pooling (α ) and the difference in land price growth rates ( ΔpA − ΔpN ). In the absence of land rental or sales, the asset value growth over time

through a ger-to-apartment scheme (JICA, 2017). Urbanization in Nepal shares similarities with that of Bhutan. As one of the least urbanized but most rapidly urbanizing countries, Nepal faces issues of unplanned growth, haphazard land subdivision and construction of buildings in areas with substandard or nonexistent public infrastructure or service provision. To better guide urban development, the government enacted the Town Development Act of 1988 (which supersedes the Town Planning Projects Implementation Act of 1973). Under the 1988 Act, Town Planning Implementation Committees (TPICs) implement land pooling projects. The first land pooling pilot launched in Gongabu, Kathmandu Metropolitan City, in 1988, after which more land pooling projects (12 by the year 2000) were implemented, mostly in the Kathmandu Valley (De Souza et al., 2018). Like the TPS in India, the land pooling pilot in Gongabu featured the collection of reserve land from original landowners; this reserve land was converted to serviced plots as part of the land pooling, and was sold to pay back the loan from the revolving fund Kathmandu Valley Town Development Committee that was used for infrastructure investment. 3. A simple model comparing land acquisition and land pooling While land pooling and readjustment is not a new concept—even in developing Asia—governments have traditionally used eminent domain as a tool to acquire lands for the provision of public infrastructure. Not only does land acquisition for urbanization come at a significant cost to governments (large, upfront land acquisition costs can make up a large portion of project costs, not to mention additional costs and time delays that can occur when landowners mount legal challenges against land acquisition), it can potentially have negative impacts on the livelihoods of displaced households. Consider a rural, land-owning household in an area subject to land acquisition for urbanization. Assume the household falls under one of the following possibilities: (i) Cash or in-kind land replacement is enough to resettle the household on a plot of land of equivalent size and quality near the original land holding; (ii) Cash or in-kind land replacement is enough to resettle the household on a plot of land of equivalent size and quality, but far from the original land holding in an area that is not urbanizing; (iii) Cash or in-kind land replacement is insufficient for the replacement of land of the size and quality near the original land holding, so the household becomes landless and rents in the same area in which it used to own land; (iv) Cash or in-kind land replacement is insufficient for the replacement of land of the size and quality even outside the land acquisition and urbanizing area, so the household becomes landless and rents farther away in an area that is not urbanizing. Because the exercise of eminent domain is a relatively non-consultative process, households take their land acquisition compensation as given. Further assume, whether payments are made in cash or in kind under scenarios (i) and (ii), household will always purchase or keep replacement land in the area near or far, respectively, from the acquired land it previously owned. Under scenarios (iii) and (iv), any land given in kind as a replacement is sold at market value and the household becomes landless in the area near or far from the acquired land; the household will use the payment or land proceeds, in combination with wage income, to finance its rental payments. Finally, assume a household that cannot afford to buy land in a far, non-urbanizing area (scenario (iv)) cannot afford to rent in the closer, urbanizing area. In practice, land acquisition where compensation in land or cash is significantly delayed could effectively put households in a lower-opportunity scenario—even if promised compensation, timely delivered, would have been enough for the household to maintain ownership of

Table 1 Assets and returns, land acquisition and pooling scenarios. Source: Authors.

5

Scenario

Starting Value of Assets

Income opportunities

(i) Nearby land replacement (i) Rural land replacement (i) Compensation for nearby rental (i) Compensation for rural rental (i) Land Pooling

pN L pR L πN pN L

wN + γθpN L + λpN L wR + γθpR L + λpR L wN + (μπN − θ) pN L

πR pR L (1 − α ) pA L

wR + (μπR − θ) pR L wA + γθpA L + λpA L

Land Use Policy xxx (xxxx) xxxx

K. Farrin, et al.

Fig. 2. Land Pooling and Land Acquisition Asset Values. Source: Authors’ simulation.

illustrates the difference in welfare outcomes under land pooling versus what can be considered a best-case land acquisition case. We consider four combinations of land pooling contribution and price growth rate difference, where a low land pooling contribution of α = 0.285 and a high land pooling contribution of α = 0.6 represent differing land pooling regimes (the low contribution is more representative of the case in Thimphu, whereas the higher contribution is representative of some Indian land pooling projects). Land price growth is assumed to be 3 percent annually in near-urban areas, while annual urban-area price growth is assumed to be 5 percent under the low price growth difference case and 8 percent under the high price growth difference case. Under all cases but the low price growth difference, high land pooling contribution combination, land pooling assets exceed land acquisition assets at some point in the 25-year project life (Fig. 2). Because a larger land pooling contribution results in a lower starting land asset value, high land pooling contributions result in a longer “break even” span, from the start of the project until land pooling asset values equal land acquisition asset values. For example, under high price growth difference assumptions, land pooling landowners with high land pooling contributions must wait 20 years to be better off than land acquisition owners whose land has been replaced in a nearby area, but those with low land pooling contributions are better off relative to land acquisition owners after approximately 8 years.

Fig. 3. GDP Growth in Bhutan, 1994-2018. Source: World Bank World Development Indicators, accessed 30 July 2019.

a zoning change that brought agricultural areas under the municipality, the loss of agricultural lands in rural-zoned areas had already begun in the absence of a proper urban plan; according to the National Environment Commission, 583 acres of paddy lands across six districts surrounding Thimphu were converted to other uses from 1999 to 2002 (ADB, 2004). In the pre-expansion municipality of Thimphu, informal settlement had also become more prevalent due to a housing shortage that would have required 600 new dwellings be built per year to keep up with population growth but not to absorb the existing housing deficit; in 2002, more than 10 percent of the population of Thimphu were thus living in squatter settlements without proper basic services (ADB, 2002). A comprehensive master plan—the 2002–2027 Thimphu Structure Plan (TSP)—was approved by the Cabinet in 2003 after a series of public consultations with various stakeholders including land and property owners. The implementation of the TSP required the preparation of detailed Local Area Plans (LAPs) for the planning of infrastructure networks and other amenities, facilitated through a land pooling process. LAPs cover small areas (about one square kilometer) and are expected to accommodate access (direct or within walking distance) to public services to a population of around 12,000 (Norbu, 2018). Initially, land for urban development and public services was to be acquired by the government, but it was later decided that land pooling

4. The study area Bhutan is a small, landlocked country in the Eastern Himalayas, bordered by Asia’s largest two countries, China and India. It has a population of about 727,145, growing at 1.3 percent annually (ADB, 2018). Despite increased urbanization, more than half of Bhutan’s population remains in rural areas and relies heavily on agriculture. Per capita gross domestic product (GDP) has more than tripled since the mid-1990s and GDP growth rates are positive but variable from 1994 to 2018, with a slowing of growth since 2018 (Fig. 3). Thimphu, Bhutan’s capital city, covers about 19 percent of the total population of the country and is in a valley in the central western part of the country, surrounded by mountain ranges and rivers. In response to growing urbanization, Thimphu’s urban area was expanded from 8 to 26 square kilometers in 1999. While the expansion came in the form of 6

Land Use Policy xxx (xxxx) xxxx

K. Farrin, et al.

Fig. 4. Before and After Land Pooling, Southern LAPs of Thimphu. Source: Norbu (2018).

regulatory constraints to implementing an alternative to land acquisition for urbanization. The absence of legal grounds for land pooling eventually led to the passing of the Local Government Act (2009, 69) (Land Pooling Rules and Regulations), which defines land pooling and outlines provisions on land pooling and guided land development to carry out planned development in line with the goal of ensuring the timely and sustainable provision of urban services while minimizing displacement and acquisition. Land registration, prohibited land transactions, and land use conversions are defined in the Act. Per the legislation, two thirds of the landowners are required to support the land pooling approach for development plans to proceed (World Bank and Ministry of Works and Human Settlement Bhutan, 2014). Prior to the Local Government Act of 2009, the Land Act of Bhutan 2007 superseded the Land Act of Bhutan 1979 as the governing law behind land acquisition. The 2007 Act served to improve the management, regulation, and administration of ownership and use of land for socio-economic development and environmental well-being. It established an autonomous National Land Commission, reduced the number of land categories from 20 to 7; and made provisions for compensation in substitute land or cash payment or both by the government upon acquisition. With respect to compensation for land acquisition, the 2007 Act gives landowners in rural areas the option to choose what kind of

would instead facilitate the process. In consultations, landowners indicated they preferred pooling to acquisition due to a variety of issues with land acquisition, including inadequate compensation offered by the government, the possibility of land acquired from farmers being allotted as serviced plots to businesses, information asymmetry on market prices and opportunities that was perceived to favor more educated and rich community members, a rushed planning process and lack of transparency, and distorted land and housing markets due to speculation among few large landowners (Norbu, 2018). An Asian Development Bank (ADB) loan ($24 million) funded the construction of public works in the Southern LAPs (Figs. 4 and 5). The project resulted in the construction of (i) 33.7 km of secondary and tertiary roads with associated drainage to connect the four local areas to major service roads and the Babesa expressway; (ii) a 31.9 km network of secondary and tertiary water distribution pipes; (iii) a 28.5 km sewerage collection network; and (iv) 25.7 km ducting for power and telecommunication networks with 265 access shafts for cable and associated structures (ADB, 2017b). While the layout of the LAPs was planned to include other community service areas (e.g., parks, police stations, schools, hospitals), these were not funded under the ADB infrastructure project. The land pooling experience in Thimphu shed light on legal and

Fig. 5. The Study Area. Source: ADB project documents, n.d. Central Thimphu is the area directly to the northeast of Changbangdu (area 6 on the map). Serbithang (area 10 on the map), while part of another ADB-funded project, was not part of the land pooling project in the southern LAPs. 7

Land Use Policy xxx (xxxx) xxxx

K. Farrin, et al.

compensation they would like and calls for the revision of the compensation rate every three years by the Property Assessment and Valuation Agency (Royal Government of Bhutan, 2007).

Table 2 Initial Estimated Population of Landowners in each LAP. Source: Authors’ calculation.

5. The data 5.1. Sample selection In the empirical analysis, primary data are collected through a household survey covering the southern LAPs. A sampling frame of households who owned land prior to the formation of the LAPs (“original landowners”) was initially constructed using land and mortgage records for Chang Gewog4 provided by Thimphu Thromde.5 A field team in Thimphu was tasked with locating mobile numbers for these households (mobile numbers were not included in land and mortgage records obtained), as well as for households who currently own land in the LAP but did not own land in the LAP to the infrastructure project (“new landowners”). Sampling was to be stratified at the LAP level, with original landowners from each LAP selected through purposive sampling to obtain a near-census of these households, followed by purposive sampling of new landowners to also obtain a near-census. The remainder of the sample would then consist of renters who do not own any land in the LAP. Both original and new landowners identify their rental properties in the interview process; renters are sampled through selecting a household at random in an apartment building (apartment building floors are randomized, and on selected floors enumerators alternate between knocking on doors on the left- or righthand side, moving to the next door on the selected floor if the initial respondent refuses the interview or is unavailable). Due to a lack of complete digital records, in combination with the available data from the 2017 Bhutan Census of Population and Housing being disaggregated only to the municipality level, a full listing of the population of owners or renters in the LAPs was unattainable. A subset of paper-based land records was collected, and this was used to roughly estimate the population of owners in each LAP (Table 2). The population of renters remains unknown but is estimated to be much higher than the population of owners given the nature of the LAPs (apartment buildings are primarily filled with renters, and fewer landowners—if any—occupy units in the buildings they own). Based on previous studies conducted in the northern land pooling areas of Thimphu, the response rate was expected to be approximately 70–80 percent. Available (but incomplete) mortgage records allowed a rough matching to the land records to tentatively identify the number of original landowners and new landowners. From these combined data, an overall target of 900 owners and 300 renters, with a priority on old landowners, was set as the initial target sample (Table 3). Citizen representatives for each LAP are to maintain contact information for landowners; however, these representatives were not willing to assist the survey team by providing these numbers due to incomplete contact information records, time constraints, and political sensitivities surrounding the upcoming national election in Bhutan. Without a reliable list of owners and contact details from which to draw a sample, convenience sampling was the chosen alternative. Field teams spent one day manually gathering contact details of building owners by asking LAP residents and populating lists. Using the contact information from the manual listing, enumerators scheduled appointments to interview owners in the following days. This process was repeated until manual listing activities in all four LAPs had covered all landowners present in the area. Despite considerable efforts in the manual listing, it

LAP

Total unique landowners

Proportion of unique landowners

Babesa Changbangdu Lungtenphu Simtokha Total

353 235 411 252 1251

28.2% 18.8% 32.9% 20.1% 100%

Table 3 Initial Target Sample. Source: Authors’ calculation. Category

Original Sample

Reserve Sample

Targets complete (70% response)

Original Landowners New Landowners Renters Total

586 314 300 1200

0 351 0 351

410 490 300 1200

was not possible to conduct interviews with all landowners. Constraints to obtaining a near-census of landowners include: (i) Mobile numbers of landowners did not register calls (this happened occasionally when tenants provided mobile numbers for landlords, but mobile numbers for the landlords had been changed or were otherwise out of service); (ii) Landowners had recently sold off their land or buildings and the new tenants or neighbors could not definitively identify landowners (online rental payments are common in the southern LAPs, and many tenants making electronic transfers to their landlords’ accounts had never even seen their landlords); (iii) Absentee landowners owning land or buildings in the LAPs refused to come to Thimphu for interviews, or were out of the country and could not be contacted to schedule an interview; (iv) Landowners agreed to interviews but did not conform to the appointments made, even after several attempts to reschedule; (v) Owners of vacant plots could not be identified in several LAPs (even neighbors had no knowledge of who owned adjacent plots); (vi) Owners outrightly refused to be interviewed or suggested their caretakers be interviewed instead (caretakers are not expected to know the level of detail required in the survey). Overall, an estimated 20–25 percent of landowners refused to be interviewed or could not be contacted. While we cannot determine the characteristics of the non-participating land owners, observational evidence indicates that those who were unwilling or unable to be interviewed are likely to be higher income households with higher-thanaverage land ownership; among those unavailable to be interviewed, it is likely that their rental income is higher than average, as many of the non-contactable households had moved abroad and were managing apartment buildings or hotels at arms-length. Initially, it was expected that there would be enough owners to reach two thirds of the sample (n = 900) so a sampling interval of three was deployed for renters, i.e. one renter selected for every third apartment building. However, as fieldwork progressed it was clear that the total number of owners that could be surveyed would be closer to 600, so the target and sample interval of renters was adjusted to reflect this (Table 4). Random selection of and adjustment of building-level sampling intervals for renters results in a sample where renters are proportional to owners for survey responses in each LAP.

4

Gewogs are administrative units in Bhutan representing a bloc of villages; gewogs are one level of administration below districts. Chang Gewog is comprised of areas in southern Thimphu, including the southern LAPs. 5 Thimphu Thromde is the local administrative government, or municipality, of Thimphu. 8

Land Use Policy xxx (xxxx) xxxx

K. Farrin, et al.

Table 4 Final achieved sample by LAP and respondent type. Source: Authors’ calculation.

Table 6 Average Monthly Wages. Source: Authors’ calculation.

LAP

Renters

Original landowners

New landowners

Total

%

Babesa Chang Bangdu Lungtenphu Simtokha Total

175 116 169 148 608

111 46 102 107 366

69 42 73 44 228

355 204 344 299 1,202

29.5 17.0 28.6 24.9 100

Average Monthly Wages by Tenancy and LAP Monthly Wage (Nu.) Tenancy Type Renter Original Landowner New Landowner LAP Babesa Changbangdu Lungtenphu Simtokha

5.2. Survey instrument The survey instrument covers demographics, expenditures, and income-generating activities of households and includes a series of modules on land transactions and experience with the land pooling process and outcomes. Original landowners answer additional survey modules designed to generate objective and subjective welfare comparisons of pre- and post-land pooling time periods. Borrowing from methods used in participatory impact assessment, households interviewed are instructed to create their own timeline of land ownership or rental events; knowing when and where households bought, sold, and/ or rented property informs the research of differential effects of the land pooling project. Depending on the length of ownership or tenancy, survey respondents are also asked to evaluate the land pooling process in three distinct periods: (1) consultation to formation of a LAP; (2) formation of a LAP to the start of urban infrastructure works; and (3) construction to completion of urban infrastructure. All respondents are asked about current conditions, including completeness and maintenance of existing infrastructure and amenities.

Average Rental Income and Payment (Nu.) Monthly Rental Income (Residential) Tenancy Type Original Landowner New Landowner Renter

5.4. Land prices We base our land price analysis on the transaction roster of the household survey, in which households are asked to itemize land transactions made since 2005 (some households had earlier records and provided these as well when available). The transaction roster includes questions on year of sale or purchase, price, land area, structures on the transacted plot, services available on the transacted plot, and reason for buying or selling. In total, 365 sales and purchases were recorded. However, transactions are dropped if they display inconsistencies with the current plot roster, where more detailed information is recorded for plots still owned by the household. The final sample for land transactions is reduced to 260 once inconsistent transaction observations are dropped. While it would be ideal to have more detailed plot characteristics in the transaction roster of the survey, most respondents could not recall information at this level of detail. Typically, respondents would produce a legal title at the interview, which they

Table 5 Distribution of Plot Service, Southern LAPs. Source: Authors’ calculation. Distribution of plot service by LAP and Overall, Original Landowners (%)

Paved road Water Sewerage Drainage

Overall

Babesa n = 111

Changbangdu n = 46

Lungtenphu n = 102

Simtokha n = 107

n = 366

100.0

91.3

93.1

100.0

97.0

96.4 85.6 73.9

93.5 69.6 82.6

98.0 89.2 76.5

86.9 93.5 66.4

93.7 86.9 73.5

101,315 92,328 Monthly Rental Payment 10,959

total household income for landowners. Renters spend 42 percent of their wage income on housing expenses at the median (Table 7). More interestingly, a breakdown of rental expenditure by income quintile reveals that annual rental expenditure does not vary greatly by household income, with median annual expenditures across quintiles ranging from 100,000 to 120,000 Nu. (approximately $1447-1736). Low variation of rental expenditures across income levels indicates that households in the lower income quintiles suffer from housing stress, where the median household in the lowest income quintile spends over 90% of its income on rent; only in the fourth and fifth income quintiles do we see evidence of access to affordable housing (i.e., households spending less than a third of their income on rent) (Fig. 6). This is consistent with one criticism of land pooling schemes where land is not set aside from low-income housing: land readjustment and infrastructure provision may result in artificially inflated land and property prices in the project area, which eventually reduces housing supply for low-income populations (Schnidman, 1988). Disparity in rental expenditure share across income quintiles in the project area may be exacerbated by the existence of housing subsidies for civil servants, who tend to earn more than the average citizen.

Focusing on the sample of original landowners, who were asked to recall the situation before land pooling and compare it to the present, current access to services within the LAPs is not universal. While most original landowners had access to paved roads and water after land pooling, fewer had access to sewerage and drainage in LAP areas (Table 5). Lack of access means that the household had never had access to a service directly on an owned plot, or that the plot previously had the service (before or after land pooling) but the plot no longer could access the service due to some level of disrepair. Examining the full sample, wages in the area are relatively high but vary by land ownership status and across LAPs. On average, renters report lower monthly wage incomes than original and new landowners, while there is no significant difference between wages across original and new landowners. Reported monthly wages are the highest in Simtokha, followed by Babesa, with lower monthly wage incomes in Changbangdu and Lungtenphu (Table 6). While landowners earn significantly more in wage income relative to renters, rental income for owned property is an even larger chunk of

LAP

41,241 26,157 26,037 45,905

Table 7 Average Rental Income and Payment, Owners and Renters. Source: Authors’ calculation.

5.3. Summary statistics

Service Type

26,217 48,458 49,081

9

Land Use Policy xxx (xxxx) xxxx

K. Farrin, et al.

Fig. 6. Median Annual Rental Expenditure by Income Quintile. Source: Authors

buildings would be expected to add more value to a plot, as generated income potential from these types of buildings is higher than for singlefamily homes. Total Services captures access to public services as a result of urbanization and the land pooling process. Because some households had access to some public services before land pooling, one service on its own may not have a significant effect on price—particularly when the quality of that service is not known prior to land pooling. However, infrastructure provision in the land pooling process was done as an entire package, with all four services coming online during the construction period. As the land pooling construction phase progresses, we expect to see a higher count of total services and a corresponding positive relationship between number of services and land price. Additionally, because all four services were offered at relatively the same time, including a categorical variable for each service on its own in the model is likely to introduce multicollinearity; the more highly correlated the individual service variables, the more difficult it is to attribute differences in price to a change in a single variable under the assumption of ceteris paribus. Adding a set of indicators for LAP captures location-specific characteristics that can determine price. Changbangdu, the omitted LAP, is the closest to central Thimphu; thus, we expect coefficients on Lungtemphu, Simtokha, and Babesa variables to be negative to capture the price premium of being closer to the city center. While more granular spatial data measuring the location of the plot from central Thimphu would provide more information on the relationship between land prices and travel time to the town center, efforts made to collect GPS data during survey implementation were unsuccessful; further, data on travel time to central Thimphu were contemporaneous and thus did not match the reference period for the transaction for which price data were collected.

would use to supplement their recall information when responding to questions about the plot at the time of its purchase or sale. Without more detailed information about the plot at the transaction time, the hedonic pricing model presented in the following section is restricted to a simpler set of variables that affect the value of land in the southern LAPs. The simplification of the hedonic pricing model and the small sample size for plot transactions are caveats of the analysis; identifying the causal impact of public works on land prices over time would require more detailed land title records or a panel data study, where households are surveyed over multiple periods. 6. A hedonic pricing model for the southern LAPs of Thimphu 6.1. Model specification Based on a theory of consumer demand, hedonic pricing models assume that items can be valued by their individual characteristics (Rosen, 1974). Empirically, this means an item’s price can be regressed on the characteristics to determine the way in which each characteristic uniquely contributes to the overall composite unit price. Thus, hedonic pricing models are often used in real estate analysis. Using the land transaction data from the household survey, we estimate the following loglinear model: K

lnpnt = β0t +

∑ βkt znkt

+ εnt

k=1

Where lnpnt is the log form of the price per square meter for each t are plot-specific characteristics that are transaction n in year t. z nk hypothesized to affect land prices through quality differentiation. Determinants of price in the hedonic model are as follows:

• With Structure: dummy variable equal to 1 if the land has a building on it at the time of transaction, and 0 otherwise; • Total Services: variable ranging between 0–4, equal to the number of •

6.2. Results Results of the hedonic pricing model are presented in Table 8. Estimated coefficients on all LAP location variables are negative, although Lungtenphu is not significant (Lungtenphu is the closest LAP to Changbangdu, followed by Simtokha and Babesa). As expected, price per square meter declines the farther away a plot is from the city center. The presence of a structure on the plot at the time of sale is positive, large in magnitude, and highly significant. The Total Services variable also takes on the expected sign and is significant, with each additional service available at time of transaction increasing price per square meter by 13%.

major services to which the plot has access to road(s), water, drainage and sewerage at the time of transaction; LAP: dummy variables equal to 1 if the transacted land is in the LAP, and 0 otherwise (Changbangdu is the omitted LAP).

Because building a structure comes at an additional cost, we expect the coefficient on With Structure to be positive to indicate the additional value of land when it includes a building—most often in Thimphu this is a family home, multi-family residential apartment building, or mixeduse building. Unfortunately, transaction data does not differentiate across structure type, as this information is not included in land titles, and, during survey piloting, land-owning households were not often able to recall what kind of structure was on the property at the time of purchase if they had since rebuilt on the plot. Intuitively, multi-family

7. Landowners’ assessment of the land pooling process Because land pooling is designed to be an inclusive process that benefits original landowners, it is important to understand these 10

Land Use Policy xxx (xxxx) xxxx

K. Farrin, et al.

8. Conclusions and policy implications

Table 8 Determinants of Land Prices, Southern LAPs. Source: Authors’ regression analysis.

While the benefits of land pooling can create a win-win situation for governments (improved public finance for urbanization) and citizens (increased participation in the urbanization process and shared gains of land value uplift), land pooling projects are relatively new to developing Asia and there is room to improve by learning from early adopters. While land pooling has the potential to be a more inclusive urban development plan for original owners or inhabitants of a project area, how the land pooling and land readjustment process is conducted affects both objective and subjective outcomes for land pooling participants. Social safeguards and resettlement policy should adapt with the evolving nature of urban development projects, allowing for land pooling but also taking the landless into account through provisioning of low-income housing or rent control in the land pooling area. The experience in Thimphu suggests that land pooling was effective in providing initial structure to the urban expansion process, but that a lack of more innovative financing mechanisms and the absence of a plan for low-income housing provision constrained the potential of land pooling in the southern LAPs. For Bhutan and other countries in developing Asia, land pooling projects—particularly in greenfield areas where cities are small but growing and readjustment of landowners’ plots is still relatively less complicated—would benefit from the following set of recommendations:

Dependent Variable: log price per sq. m. (Nu.), detrended −0.818*** (0.315) −0.0904 (0.317) −0.683** (0.318) 1.112*** (0.195) 0.130** (0.0614) −0.144 (0.309) 242 0.234

Babesa Lungtenphu Simtokha With Structure Total Services Constant Observations R-squared Robust standard errors in parentheses

*** p < 0.01, ** p < 0.05, * p < 0.1.

owners’ perceptions about their participation in and the final result of the land pooling process. A qualitative line of questioning in the survey allowed for the collection of household-level data that include (i) preference rankings for community amenities that were to be provided to residents under the LAP; (ii) quality perceptions for individual amenities provided to LAP residents; and (iii) Likert-scale questions that elicit satisfaction of original landowners with the consultation, planning and construction phases of the LAP. Households’ welfare will improve the most where services they most value are available. However, across the sample, only 48% of households had access to the service they ranked as most important. Taking it one step further, households not only want a public service to be available, but they want available public services to be high quality. In this respect, only 33% said the most important service was accessible and high quality. This shortcoming of public service provision indicates a need for increased community engagement in project planning stage, financing for maintenance and expansion of services as LAPs grow. Responses to questions about satisfaction with land pooling project phases indicate that original landowners are the most dissatisfied when they did not approve of the original land pooling process (Table 9).

8.1. Collect a comprehensive set of baseline data of affected households before the implementation of land pooling projects This includes data on informal settlers and on landless households who are employed in the project area (e.g., as agricultural laborers). To date, studies on land pooling have been largely qualitative in nature due to shortcomings in household-level data availability for the affected populations. Monitoring and evaluation plans for future land pooling projects should include a solid research component and household survey design so that project impacts can be more rigorously estimated.

8.2. Allow for visionary master planning Land pooling—or any other form of planned urban development—should be viewed through a long-term lens. Consultation and planning for land readjustment and infrastructure building should include discussions and decisions on how to finance and implement maintenance, upgrading and expansion of urban infrastructure as cities urbanize. In Thimphu, inadequate infrastructure maintenance, in combination with high population growth that tested the carrying capacity of the new urban infrastructure, was cited as one of the biggest shortfalls of the land pooling project.

Table 9 Satisfaction with Land Pooling Phases and Approval of Initial LAP Plan. Source: Authors’ calculation. Satisfaction with Land Pooling Phase

Consultation Very dissatisfied Dissatisfied Neutral Satisfied Very satisfied Don’t know Planning Very dissatisfied Dissatisfied Neutral Satisfied Very satisfied Don’t know Building Very dissatisfied Dissatisfied Neutral Satisfied Very satisfied Don’t know

Approved of original LAP plan (%) No

Yes

3.7 14.81 44.44 33.33 3.7 0

0 2.84 14.18 69.5 8.51 4.96

3.7 25.93 51.85 14.81 3.7 0

0 2.84 19.15 68.09 9.22 0.71

3.7 11.11 48.15 33.33 0 3.7

0 9.22 24.82 56.74 8.51 0.71

8.3. Plan to include landless households who will become renters in a land pooling area and allow for opportunities for low-income households to become property owners Price speculation may be an unwanted effect of land use change where newly urbanized areas experience rapid development. Because land from land pooling projects is turned over to original landowners—who are likely to be middle-to-high-income—the land value uplift from land pooling is often captured by the already better-off. Examples like the cross-subsidization of low-income housing through reserve land sales from South Korean land pooling projects and the pension system for the landless implemented in the Amaravati land pooling system are points of reference for the design of more inclusive land pooling regimes.

11

Land Use Policy xxx (xxxx) xxxx

K. Farrin, et al.

8.4. Make legal and regulatory changes to facilitate an urban infrastructure financing plan

From Experience and Challenges Ahead. IIED, FAO ISBN: 1-84369-632-0. Davis, K.F., D’Odorico, P., Rulli, M.C., 2002. Land grabbing: a preliminary quantification of economic impacts on rural livelihoods. Popul. Environ. 36 (2), 180–192. https:// doi.org/10.1007/s11111-014-0215-2. De Souza, Francisco, Felipe, Ochi, Takeo, Hosono, Akio, 2018. Land Readjustment: Solving Urban Problems Through Innovative Approach, 1st edition. Japan International Cooperation Agency Research Institute, Tokyo. Ghimire, S., Tuladhar, A., Sharma, S.R., 2017. Governance in land acquisition and compensation for infrastructure development. Am. J. Civ. Eng. 5 (3), 169–178. Home, R., 2007. Land readjustment as a method of development land assembly: a comparative overview. Town Plan. Rev. 78 (4), 459–483. Japan International Cooperation Agency, 2017. Project for Improvement of Capacity for Implementation Skills of Ulaanbaatar Master Plan Sub-Project 2: Outline of Urban Redevelopment Law and Mechanism of Urban Development Projects. Ulaanbaatar. . KPMG, 2013. DDA Land Pooling Policy. Accessed February 26, 2019. http://naredco.in/ notification/pdfs/DDA%20land%20pooling%20policy.pdf. Kusiluka, M.M., Kongela, S., Kusiluka, M.A., Karimuribo, E.D., Kusiluka, L.J.M., 2011. The negative impact of land acquisition on indigenous communities’ livelihood and environment in Tanzania. Habitat Int. 35 (1), 66–73. https://doi.org/10.1016/j. habitatint.2010.03.001. Lozano-Gracia, Nancy, Young, Cheryl, Lall, Somik V., Vishwanath, Tara, 2013. Leveraging Land to Enable Urban Transormation. World Bank Policy Research Working Paper No. 6312. Mathur, S., 2013. Self-financing urbanization: insights from the use of town planning schemes in Ahmadabad, India. Cities 31, 308–316. Mathur, Shishir., 2014. Innovation in Public Transport Finance: Property Value Capture. Routledge, London. Nguyen, H.M.V., Pham, A.T., Nguyen, L.D.L., Diep, V.T., 2018. Decisions on socialization by urban land pooling approach to renovation of landscapes of Ho Chi Minh City embankments. MATEC Web of Conferences 193. https://doi.org/10.1051/ matecconf/201819301018. Nizuke, A., Sliuzas, R., Flacke, J., Maarseveen, M., 2019. Livelihood impacts of displacement and resettlement on informal households—a case study from Kigali, Rwanda. Habitat Int. 86, 38–47. Norbu, G., 2018. Land pooling in Thimphu, Bhutan. Global Experiences in Land Readjustment. United Nations Human Settlements Programme (UN-Habitat), Nairobi. Patel, S., Sliuzas, R., Mathur, N., 2015. The risk of impoverishment in urban developmentinduced displacement and resettlement in Ahmedabad. Environ. Urban. 27 (1), 231–256. Puasakul, U., 2010. Land readjustment in Thailand. Asia-Pacific Housing J. 4 (13), 35–39. Rosen, S., 1974. Hedonic prices and implicit markets: product differentiation in pure competition. J. Polit. Econ. 82 (1), 34–55. Royal Government of Bhutan, 2007. The Land Act of Bhutan 2007. http://oag.gov.bt/wpcontent/uploads/2010/05/Land-Act-of-Bhutan-2007_English.pdf. Sarkar, A., 2007. Development and displacement: land acquisition in West Bengal. Econ. Polit. 42 (16), 1435–1442. Shaw, A., Saharan, T., 2018. Urban development-induced displacement and quality of life in Kolkata. Environ. Urban. https://doi.org/10.1177/0956247818816891. Schnidman, Frank, 1988. Land Readjustment. Urban Land (February). Sitorus, S., 2005. The Implementation of Land Readjustment in Indonesia. Working Paper. Yogyakarta: National Land College. Sorensen, A., 2000. Conflict, consensus or consent: implications of japanese land readjustment practice for developing countries. Habitat Int. 24, 51–73. Valente, C., 2009. The food (In)Security impact of land redistribution in South Africa: microeconometric evidence from national data. World Dev. 37 (9), 1540–1553. https://doi.org/10.1016/j.worlddev.2009.01.005. World Bank, 2007. The Evolving Role of World Bank Urban and Shelter Projects: Addressing Land Market and Economy-Wide Constraints. Final Report No. 70429. Washington, DC. . World Bank and Ministry of Works and Human Settlement Bhutan, 2014. Resettlement Action Plan for Lower Taba Local Area Plan. Accessed February 27, 2019. http:// documents.worldbank.org/curated/en/763861468201547323/pdf/ RP16310SAR0RI000Box385182B00PUBLIC0.pdf. Yoshida, T., 2003. Comparative analysis on land consolidation projects between Indonesia and Japan. J. Asian Archit. Build. Eng. 2 (2), 111–116. Yoshino, N., Paul, S., Sarma, V., Lakhia, S., 2018. Land Acquisition and Infrastructure Development Through Land Trust Laws: a Policy Framework for Asia. ADBI Working Paper Series No. 854.

To complement master planning and updated property valuation practices, legal and regulatory changes—particularly to the taxation system—should be part of an urban infrastructure financing plan. This includes building or strengthening a regulatory framework to allow for instruments like tax increment financing or betterment fees. 8.5. Digitize land records and institute updated processes of valuation for land and property prices Transparent and available data on land ownership, land prices, and housing units per parcel paves the way for a streamlined land pooling process. While property rights in Bhutan are well established, the country’s land registration system has only recently begun to be digitized; details on parcel ownership, use, structure type(s), and prices over time—which would provide insights on the impact of public infrastructure provision on land prices and assist in determining potential changes to the tax system—are nearly impossible to track without a well-maintained online database. Data on infill and parcel occupancy rates over time would assist in ensuring that urban planning meets carrying capacity needs of a growing city. Improving the digital land registration system would also serve as a strong backbone to analysis that could be used for the transparent and systematic setting of betterment fees from land value uplift. References Abebe, G., Hesselberg, J., 2013. Implications of urban development-induced resettlement on poor households in Addis Ababa. Ghana J. Geogr. 5, 32–50. Andhra Pradesh Capital Region Development Authority, 2017. Amaravati Status Report. Presentation Delivered at Amaravati Deep Dive Workshop. 14-15 December. . Andhra Pradesh Capital Region Development Authority, 2018. Amaravati Sustainable Capital City Development Project (ASCCDP): Final Revised Draft Resettlement Action Plan for 10 Sub-Arterial Roads. Amaravati. . Archer, R.W., 1986. The use of the Land-Pooling/Readjustment technique to improve land development in Bangkok. Habitat Int. 10 (4), 155–165. Asian Development Bank, 2002. Technical Assistance to the Kingdom of Bhutan for Housing Sector Reform. Manila. . Asian Development Bank, 2004. Bhutan: Country Environmental Analysis. Manila. . Asian Development Bank, 2010. Corridor Densification by Floor Space Index-Linked Land Use Control and Infrastructure Financing Mechanism. Urban Innovations and Best Practices, Manila. Asian Development Bank, 2017a. Meeting Asia’S Infrastructure Needs. Manila. . Asian Development Bank, 2017b. Urban Infrastructure Development Project: Project Completion Report. Manila. . Asian Development Bank, 2018. Key Indicators for Asia and the Pacific 2018. Asian Development Bankhttps://doi.org/10.22617/FLS189512-3. Ballaney, Shirley, 2008. The Town Planning Mechanism in Gujarat, India. The International Bank for Reconstruction and Development/The World Bank, Washington, DC. Bhatta, B., 2010. Causes and consequences of Urban growth and sprawl. Analysis of Urban Growth and Sprawl From Remote Sensing Data. Advances in Geographic Information Science. Springer, Berlin and Heidelberg, Germany. Chan, N., 2000. Land acquisition compensation in China—problems and answers. Int. Real Estate Rev. 6 (1), 136–152. Conventus Law, 2018. Malaysia - Key Changes of the Land Acquisition (Amendment) Act 2016. Accessed February 26, 2019. . http://www.conventuslaw.com/report/ malaysia-key-changes-of-the-land-acquisition/. Cotula, L., Toulmin, C., Quan, J., 2006. Better Land Access for the Rural Poor: Lessons

12