How financially viable is smallholder forestry? A case study with a high-value tropical forestry species

This paper evaluates the financial performance and risk of high-value tropical forestry under the challenges faced by smallholders, using Vanuatu sandalwood as a case study. We developed a financial model to predict returns from smallholder-based sandalwood plantations. The model was used to investigate the following issues typically faced by smallholders: (i) what is the financial impact of smallholders harvesting trees at young ages because of fear of theft or cyclone damage? (ii) how does the opportunity cost of labour impact on the financial returns of sandalwood? and (iii) what are the current opportunities for smallholders to finance the establishment of sandalwood plantations and how might these be improved? Data were collected from expert interviews and relevant literature. We found that smallholder sandalwood can be a profitable investment, but tree security issues and environmental risks are leading to early harvest and an associated 64% reduction in potential returns. To improve the profitability and attractiveness of sandalwood plantations to smallholders, the following key issues must be addressed: (i) risks associated with tree theft need to be dramatically reduced; (ii) earlier returns from planting systems must be planned for smallholders with more immediate needs; and (iii) greater access to financial services is needed, including loans with competitive borrowing rates, particularly to increase the scale of planting for individual smallholders that might be constrained by labour. Theft mitigation can include fencing, patrolling, microchip tagging, all of which add significant maintenance cost. Formalised social and governance structures within and between villages, may be more cost-effective in improving tree security. Incorporating agricultural crops into systems helps produce earlier financial returns, and more rapid payback. Accessible loans can allow for upscaling of smallholder systems that are limited to household labour for plantation management, harvesting, and processing the logs.


Introduction
Demand for timber and other forest products is expected to grow rapidly over the coming decades due to the combined effects of increasing world population, urbanisation and prosperity (Midgley et al. 2017b). At the same time the area of natural forest available to meet this demand is in decline due to the combined effects of deforestation, forest degradation and protection in forest reserves (FAO 2020;Warman 2014). As a response to this reduction in supply and increased market demand for forest products, the area of forest plantations has been expanding (Evans 2010;Payn et al. 2015). While industrial plantations comprise the majority of this expansion (Carle and Holmgren 2008;McEwan et al. 2020), smallholder commercial forestry is also increasing in areas where land available for larger plantations has become constrained (Maraseni et al. 2017b;Midgley et al. 2017a). Smallholder forestry has become prevalent in places like Vietnam (Maraseni et al. 2017b), Java (Race and Wettenhall 2016) and northeast and central Thailand (Boulay et al. 2012;Schirmer et al. 2015). Globally, the area of planted forests for productive purposes are owned by public institutions (~ 50%), smallholder farmers (~ 32%) and corporate organisations (~ 18%) (Portin and Lehtonin 2012).
Smallholder forestry has the potential to address poverty through the diversification of income sources, as well as reduce reliance on scarce natural resources (Boyd et al. 2007;de Sousa et al. 2016;Grieg-Gran et al. 2005;Maryudi et al. 2015). There is increasing evidence demonstrating smallholder forestry can be a commercially viable livelihood activity in many parts of Asia (Boulay et al. 2013;Cuong et al. 2020;Harrison et al. 2005;Maraseni et al. 2017a;Phimmavong et al. 2019;Sabastian et al. 2014). The viability of cultivating high-value trees will depend on conditions such as secure land and tree tenure, reliable production systems and risk mitigation, sufficient commercial value, strong demand for their products and accessible markets (Belcher et al. 2005;Byron 2001). In Pacific Island Countries commercial smallholder forestry has potential (Raymond and Wooff 2006), provided production systems are compatible with existing livelihood strategies (Walters and Lyons 2016) and issues with logistics and market access can be addressed (Blumfield et al. 2018).
Sandalwood (Santalum spp.) has considerable potential for cultivation in the Asia-Pacific region. The most widely marketed species is Santalum album L., which has been utilised since antiquity (Aftel 2004;Gunn 2016). Other related species in the Pacific, including the focus species in this study, Vanuatu sandalwood (Santalum austrocaledonicum Vieill.), have been exploited over the past 200 years (Merlin and Fleetham 2012;Shineberg 1967). All sandalwood species provide high-value, lowvolume, non-perishable products (Page et al. 2012b). Their fragrant heartwood oil has traditionally been used for religious and customary purposes and is now used for cosmetics, aromatherapy, soaps, perfumery and medicines (Goswami and Tah 2018;Page et al. 2010a;Rhind 2013;Sultana and Rahman 2018). Sandalwood is also used for production of incense sticks and carving logs, the latter being the most valuable sandalwood product (Kumar et al. 2012;Page et al. 2012a;Page et al. 2010b). Despite the development of large-scale sandalwood plantations in Asia-Pacific countries, especially Australia, China and India , demand still exceeds supply, resulting in positive prospects for expansion of smallholder organically-grown sandalwood (Tate 2012;Thomson 2020).
There are various constraints limiting smallholders taking advantage of the market opportunities for sandalwood production. High initial investment, long maturity periods and high risks make forestry less appealing to smallholders with immediate needs (Hoch et al. 2009;Pagiola et al. 2007;Rahman et al. 2015). Also, the opportunity cost of labour is dynamic due to seasonal fluctuations in job opportunities, and the availability of family labour is limited (Martyn 2015). Adding to that, credit is often unavailable for smallholders who are unable to meet institutional requirements, including the provision of collateral (Arano et al. 2004;Boscolo and Whiteman 2008;Larson and Ribot 2007;Zhou et al. 2016). Financing the establishment of plantations is difficult, as small loans are usually provided only for short terms and require a minimum level of literacy and high-degree of trust between the stakeholders involved (Boscolo and Whiteman 2008;Larson and Ribot 2007). Even when credit can be obtained, high interest rates and short-term repayment periods can limit their application in small-scale forestry investments.
In Vanuatu, sandalwood has high market liquidity and the risk of theft is very high. Estimates indicate that heartwood development may start around age 7 and increase at a rate of about 2.5 kg per tree per year thereafter (Page et al. 2012b). The harvest of older logs provides higher percentage of heartwood in the log. However, older logs from natural forests are no longer available due to overexploitation, and farmers opt for an early harvest (around age 8-12) of their planted trees for fear of theft, loss of trees due to cyclone damage and to meet more immediate needs for cash. Also, the value of labour is not stable and reflects the opportunity costs at a given time. Unlike the production of fruits and nuts, for example, sandalwood in mixed systems has a flexible labour demand, which is an advantage. However, the first processing step, de-sapping the logs to obtain heartwood (i.e., removal of the outer sapwood), is labour intensive and might impact on other household activities.
Using Vanuatu sandalwood as a case study, the aim of this research is to evaluate the financial performance of small-scale tree growing of a high-value tropical forestry product under the challenges faced by smallholders in terms of risks, labour demand and access to credit. We develop a financial model to investigate the financial returns to smallholders from growing sandalwood. We then use the model to explore the following questions: (i) what is the financial impact of smallholders harvesting trees at a young age because of fear of theft or cyclone damage? (ii) How does the opportunity cost of labour impact on the financial returns of sandalwood? and (iii) what are the current opportunities for smallholders to finance the establishment of sandalwood plantations and how might these be improved? The paper starts with a brief description of the case study. Then, we outline the methods used to develop the financial models and the scenarios investigated. This is followed by the main results and discussions. Finally, the conclusions and policy implications are presented.

Description of the location and species studied
Vanuatu is located in the Southwest Pacific, and comprises 83 small islands with a population of over 270,000 as of 2017 (World Bank 2018). Its economy is based on agriculture, fishing, tourism and offshore financial services (Commonwealth Secretariat 2018). The climate varies from tropical in the north to subtropical in the south (Vanuatu National Statistics Office 2002). The mean monthly temperatures of the centrally located capital Port Vila range between 27°C and 22°C. On the northern islands, annual rainfall averages about 4000 mm (Vanuatu National Statistics Office 2002). The country is highly prone to cyclones. Vanuatu had 71 tropical cyclones between 1981 and 2011, 41% of which were category 3 or higher (Australian Bureau of Meteorology and CSIRO 2014).
Santalum austrocaledonicum is one of 16 species of sandalwood and is native to New Caledonia and Vanuatu. It is an hemi-parasitic shrub to small tree, typically 5-12 m tall, with crown width of 4-8 m and maximum DBH between 40 and 50 cm (Thomson 2006). Vanuatu sandalwood is adapted to dry areas of the country (Page et al. 2012b) with mean annual rainfalls of 800 to 1,500 mm (Gillieson et al. 2008) and an annual dry season of two to five months (Page et al. 2012b). It prefers welldrained neutral to slightly acidic or alkaline soils, but it grows in a broad range of edaphic conditions including shallow and infertile soils (Gillieson et al. 2008;Thomson 2006). This means that sandalwood can be grown in areas marginal to agriculture (Page et al. 2012a), where the land opportunity cost is low.
Vanuatu sandalwood produces a highly valuable aromatic oil within its heartwood that is in high demand in international incense stick and fragrance markets. The most valuable product is a clear grain log used for decorative carving, followed by heartwood for oil extracted through distillation and then powdered heartwood used for incense products (Page et al. 2012c). Primary processing involves removing the sapwood from around the heartwood, typically undertaken manually by harvesters using a large machete. A second stage cutting of any remaining sapwood is undertaken by traders following purchase from resource owners. The chips from this second cutting retain some heartwood and can also be sold for the manufacturing of incense and other products (Page et al. 2012c).
In Vanuatu the extraction and export of sandalwood products has long been an important source of income for resource owners (Page et al. 2012a;Page et al. 2010a). This industry has been operating since the mid-1800s (Shineberg 1967) based on harvesting of naturally recruited trees. Over the last 40 years, the industry has been characterised by highly variable and declining annual yields and short-duration harvest moratoriums (Berry 2005;Bule and Daruhi 1990;Gillieson et al. 2008;Lui and Smith 2007;Tate 2012;Vira and Smith 1995). In response to declining wild resources the Vanuatu Department of Forests launched the Sandalwood Policy (2002), which promoted sandalwood planting (Page et al. 2012b), and proposed a moratorium on exports of unprocessed wood (Gillieson et al. 2008). Under this policy, licensees could export sandalwood in the form of oil, powder, spent biomass, carving wood or carvings (Page et al. 2012a). These efforts have resulted in increased rates of sandalwood planting on some islands since 2000 (Thomson 2006) and by 2016 18,000 households representing 32.6% of those across the country had planted sandalwood (VNSO 2017). When compared with other cash crops this proportion of households was lower than for coconut (41.9%) but equivalent to kava (32.4%), and greater than cocoa (16.3%) and coffee (3.5%).
The Sandalwood Policy (2002) has not, however, resulted in widespread domestic processing and value adding of Vanuatu sandalwood. Unprocessed logs, in the form of clean heartwood, represent the bulk of exported sandalwood products (Pacific Horticultural and Agricultural Market Access 2017). Nevertheless, the sandalwood industry supports both rural and urban employment and income generation (Page et al. 2010a). The size of trees harvested, and quantity and quality of heartwood is decreasing each year and a shift from natural to planted sandalwood in Vanuatu will be needed for the sustainability of the industry (Tate 2012).

Research methods
We developed a financial model to explore the key factors affecting the financial viability of sandalwood plantations established by smallholders. Data for the financial model were sourced from a comprehensive review of the literature and complemented and validated with information obtained from experts in Vanuatu and Australia. The experts were national and international researchers, business owners, sandalwood growers, government officials and farm managers. The financial model was developed in Excel and used a previous model developed by Harrison and Harrison (2016b) as a starting point. The model incorporated the production system proposed by Page et al. (2012c) with 444 sandalwood trees, 556 intermediate host plants and 111 long-term host plants per hectare. A mortality rate of 20% after planting was assumed (Thomson 2006). Silvicultural practices in the model included weeding up to year seven (Harrison and Harrison 2016a), and pruning up to year four (own assumption). Labour needs were based on Harrison and Harrison (2016a) and are presented in Table 1.
In the base-case scenario, harvesting of 50% of the trees takes place at age 15 and final harvest occurs at age 20 (Page et al. 2012a;Page et al. 2010a). Heartwood production was calculated with a simple regression model using age seven as the beginning of production of heartwood and a yield of 18 kg at age 18. For each 18 kg of heartwood, same amount of sapwood and 2.5 kg of second cut chips can be obtained (Page et al. 2012c). This ratio was used to calculate the amount of each product in our model. From the heartwood produced, it was estimated that only 5% would serve as carving timber, and the remaining would be sold as heartwood. We conservatively assumed that smallholders would harvest and sell 0.25 kg of seed per tree per year from year nine onwards. Harrison and Harrison (2016a) used a figure of 1 kg per tree per year. Based on our experience in Vanuatu, we believe that a lower estimate better reflects the opportunities for the commercialisation of seeds, although more seed is likely to be available per tree. Cost for seed collection, processing and selling are presented in Table 2 along with other financial parameters used in the model.
Financial indicators -Internal Rate of Return (IRR), Net Present Value (NPV) and Land Expectation Value (LEV) -were calculated for each scenario. The discount rate was defined as a rounded figure of the sum of a risk-free discount rate plus a risk premium. The risk-free discount rate chosen was equivalent to long-term government bonds, of 7.58, minus inflation rate of 2.8% in accordance with the value for Vanuatu in 2019 (World Bank 2020). The risk premium for lending for Vanuatu was unavailable, so the value for Fiji of 3.5% was used (World Bank 2020). The World Bank interest rates are lower than commercial rates, and possibly only available to farmers in development projects. The resulting discount rate (r) of 8.28% is consistent with the rates used in the past by Harrison and Harrison (2016b) and Thomson et al. (2011), and a rounded figure of 8% was used in this study. Calculations were done in USD using a conversion rate of USD 1 = VUV 114. The following was carried out to address each specific research question: i) Apart from the harvesting system of the base-case scenario, the currently practiced harvesting system was also assessed. Bourne (2019) estimated that in recent years, 90% of the sandalwood being sold in Vanuatu are between 13 and 15 years old, although a recent survey of growers suggest that trees of 8-12 years are also being cut. The relative value of such young trees is VUV 800-1,500 per kg (Bourne 2019) and much lower than prices paid for more mature wood. Younger trees also do not contain larger diameter heartwood logs for craftwood. A twelve-year cycle and a price of VUV 1,150 per kg were selected to represent the current practice. At this stage the heartwood yield is low and so too the presence of higher-graded products. Postponing harvesting to around year 17 to 20 could bring higher returns as there is a much higher content of heartwood and potential for small carving logs. However, these extra years of production also mean increased risk of illegal harvesting and exposure to cyclones. The Land Expectation Value (LEV) for the base-case scenario and for the current practice were calculated for comparison.  ii) Sandalwood tending is often part of the management of the farming system as a whole and estimating labour inputs for sandalwood is not straightforward. Farming labour often relies on household members and management practices frequently take place when the opportunity cost of labour is low. Nevertheless, in some circumstances, smallholders have alternative sources of income as daily wages for hired labour or work in urban areas. Informally people who work for others in rural areas are paid between VUV 500 and VUV 1,000 per day depending on the individual agreement. A sensitivity analysis was carried out using no cost of labour and an hourly wage of VUV 100, which represents a low wage for rural activities. iii) Smallholders might lack the capital to invest in a farming system focused on sandalwood production. At the same time, the lending alternatives might limit the involvement of smallholders in sandalwood production. A sensitivity analysis was carried out to assess the suitability of financial services to smallholders, using the discount rates of 10, 17 and 28% based on the loan opportunities for smallholders in the country as discussed below.
Lending is often unavailable for smallholders who have no capacity to provide collateral. The Vanuatu Agricultural Development Bank had, by June 2011, about 800 loans with an average of VUV 450,000 (USD 3,947) each (McCaffrey 2011). It makes both secured and unsecured loans, lending to smallholders who cannot access loans from commercial banks. Even when access to credit is possible, they are often unsuitable for long-term ventures such as tree growing. In Vanuatu, loan terms vary between one week for small amounts and seven years for larger amounts at commercial banks (McCaffrey 2011). For instance, Credit Corps provides loans to small-and medium-sized business with a 30% deposit, using vehicles as collateral and a declining interest rate of 16%, but for 2 or 3 years only, and VANWODS offer four different loans, all of which are for up to 20 weeks (McCaffrey 2011). Loans that may suit sandalwood tree growing for smallholders considering timeframe and deposit and collateral requirements are presented in Table 3.

Results and discussion
Our analysis indicated that sandalwood tree growing can be a viable livelihood activity for smallholders in Vanuatu. Table 4 presents the financial indicators for the different scenarios. The NPVs found in this study are comparable to or higher than found previously. Page et al. (2010a) had a NPV of USD 21,786/ha for agroforestry with sandalwood planted at 3 × 4 m, interplanted with garden species, with harvesting at ages 15 and 20, and a discount rate of 10%. Harrison and Harrison (2016b) estimated a NPV of USD 23,631/ha with a risk adjusted discount rate of 8% for a sandalwood agroforestry system also containing cacao, sweet potato and vegetables. Sweet potato was added to the system for weed control and cash crop for the early years of the project. For a smallholder sandalwood (S. album) system in India Divakara et al. (2018) reported a NPV of USD 34,500/ha (IRS 2,584,914) (using a discount rate of 15%, based on prevailing interest rates of 9%, it is unclear how the authors arrived at the 15% value) and LEV of USD 37,100/ha (IRS 2,778,782) with a rotation period of 15 years and non-commercial host. Comparing to the results of this study and those mentioned above, a NPV of USD 106,000/ha for a sandalwood plantation with 100 trees per hectare and a rotation of 20 years, as calculated by Thomson et al. (2011) using a discount rate of 8% (it is unclear how the discount rate was formulated), is optimistic. Sandalwood competes with other trees such as whitewood (Endosperumum medullosum) and mahogany (Swietenia macrophylla) in Vanuatu as an option for smallholder forestry. Whitewood in Vanuatu was found to have a net present value (NPV) of between USD 6,586 and 8,257/ha (depending on silvicultural regime) at a discount rate of 5% (Grant et al. 2012). At a 10% discount rate the NPV approached zero for all production regimes. The more favourable financial performance of san-dalwood found in this study (NPV of USD 14,573 to 52,362/ha and IRR of 23 and 28%) is reflected in the greater level of household adoption of sandalwood (32.6%) than whitewood (17.8%) (VNSO 2017).

The financial impact of smallholders harvesting trees at a young age
Despite the potential financial attractiveness of sandalwood investments, smallholders are achieving only a third of the potential returns under the current practices of early harvesting. The early harvest of sandalwood at age 12 led to a decrease in LEV by USD 42,493/ha, or 64%, in comparison to the base-case scenario. The earlyharvest scenario also has a lower IRR of 23% (  cost for which NPV is zero for the early harvest scenario is also less than a third of the base-case scenario. This is because there are several fixed components in the cost regardless of the volume harvested, such as the costs for tree planting and early maintenance. Over 20 years of project in the base-case scenario, it is expected 628 person-day of work per hectare, or an average 31 days of work per year per hectare. In the scenario with harvesting at age 12, 549 person-day of work per hectare or an average of 46 days of work per year per hectare are needed. At increased rates of tree protection from 5 min per tree per year to 15 min per tree per year, there is a 10% drop in the return to labour in the base-case scenario. Improved tree security is essential for smallholders to delay harvesting to improve returns. The theft of sandalwood trees is a genuine issue in most sandalwood producing countries. Penalties for sandalwood theft in Western Australia are of AUD 200,000 (Biodiversity Conservation Act of Western Australia 2016) and in Tonga are double the value of the sandalwood stolen (Sandalwood Regulations of Tonga 2006). While both legal mechanisms relate to native sandalwood, the Vanuatu Government could consider fines for the theft of planted sandalwood for risk management. However, the theft of trees and agricultural crops from private property can be difficult to prove and prosecute (Isaac et al. 2017;Mortimer et al. 2005). In India sandalwood theft is prevalent and growers need to invest in boundary walls and patrolling after trees reach 10 years (Mishra et al. 2018). Viswanath et al. (2010 reported that 50% of the maintenance costs of plantations was attributed to security. Research has been undertaken where sandalwood trees are tagged with microchips, containing accelerometers to detect vibrations associated with cutting, with a capacity for remote signalling (Patil and C. S. 2019). While electronic tree theft control systems might be technically feasible (Hamza et al. 2013) they are likely to be cost prohibitive for smallholders in Vanuatu. Sandalwood theft may need to be mitigated through risk management mechanisms from formalised social and governance structures within and between village communities.
Mitigating the damaging effect of cyclones is also an important consideration for sandalwood producers in Vanuatu. Tungon and Tabi (2015) conducted a rapid sur- vey of a range of planted tree species following severe tropical cyclone Pam 1 , and reported that sandalwood had moderate tolerance to the extreme winds with minor damage on the western side of the island, but complete defoliation and stem breakage and wind thrown trees along the eastern and southern coasts. Following the same cyclone crossing the island of Erromango, Mowatt (2019) reported up to 20% planted sandalwood losses in exposed conditions, but only minor structural damage to trees planted in sheltered contours. Mowatt (2019) recommended two methods for mitigating the damage caused by cyclones, firstly the strategic retention of existing vegetation when planting sandalwood to provide wind protection and canopy reduction pruning prior to the storm to reduce damage. This approach recognises that plantings comprised of mixed trees species (between six and eight) are less prone to wind damage than monocultures (Herbohn et al. 2017). Interestingly retaining some vegetation during site preparation was found to reduce the costs of this preparation and subsequent weed control, and anecdotally promoted early growth in the sandalwood (Mowatt 2019). Depending on the timing of the cyclone within the rotation, many of the windblown and damaged sandalwood trees can be salvaged. Both the early-harvest (12 years) and base-case scenario are equally exposed to the effects of cyclones for the first 12 years. Cyclone exposure of the base-case scenario from years 13 to 20 can potentially result in up to 20% losses in a given event, with a proportion of these potentially being salvage. We carried out complementary calculations to include additional pruning to the base case scenario to reduce the impacts of cyclones. Additional pruning was considered between years 13 and 20 (assuming 5 min per tree per year). This increases the labour demands of the project by 25 days and represents an additional discounted cost of USD 297. This value looks relatively low in relation to the potential mitigation benefits. Nevertheless, its upfront cost might be high for resource-poor farmers, and the probability of impacts by cyclones and the potential intensity of damage need to be assessed in more detail.

The impacts of the opportunity cost of labour on the financial returns of sandalwood
Labour demand is mostly concentrated around harvesting time and initial planting years (Fig. 1). Return at the end of the project is higher with lower discount rates and opportunity cost of labour. Reducing the hourly cost of labour from USD 1.93 to USD 0.88 does not cause a major change in the outputs of the model. When removing the cost of labour from the annual cashflows, the IRR of the project increases from 28 to 32% and NPV increases by about 6%.
If the conditions related to the sale price of sandalwood products, cost of capital and labour demands are met, sandalwood can be a very competitive labour opportunity. The financial performance of cocoa production in Vanuatu was calculated as between USD 9.60 (VUV1,100) and USD 21.90 (VUV 2,500) per family day of labour (McGregor et al. 2009) equating to a present value of USD 13 to 30 based on 2.8% inflation. This is similar to the USD 26.55 per person day of labour for a sandalwood plot harvested at 12 years of age found in this study. We have been unable to locate studies of financial performance for other key cash crops in Vanuatu to permit comparison with sandalwood.
The opportunity cost of labour did not have a large impact on the NPV, indicating that returns over the cycle of a project are high enough to compensate for the foregone income from other activities. However, returns are not distributed evenly along the project life and smallholders are often unable to cope with long periods of negative project balance. For an earlier payback, other products could be incorporated to the system. Inclusion of fruit and vegetable species in agroforestry systems focused on a long-term product can help to service short term loans or meet immediate cash needs (Appiah 2001;Harrison et al. 2016). The selection of companion crops is important. In Laos, Phimmavong et al. (2019) demonstrated higher NPV and LEV in short rotation Eucalyptus grown with rice in the first year compared with Eucalyptus monoculture, but no such difference in the latter compared with Eucalyptus grown with cassava. In India sandalwood agroforestry systems intercropped with pigeon pea (Cajanus cajans) for the first six years (Divakara et al. 2018), Indian gooesberrry (Phyllanthus emblica) or horse gram (Macrotyloma uniflorum) (Viswanath et al. 2010) had a slightly lower NPV, benefit-cost ratio and LEV compared with sandalwood without intercropping. However, both these studies suggested that intercropping was preferred by farmers due to the higher equivalent annual income compared to planting sandalwood alone. The incorporation of other products in the system can act as an efficient risk management mechanism. The provision of income at the earlier stages of sandalwood production could potentially lead to longer harvesting cycles and higher volumes of heartwood accumulated, but further research is required. The current opportunities for smallholders to finance the establishment of sandalwood plantations and how these might be improved Under all scenarios, annual cashflows remain negative for at least nine years, at which time seeds are harvested for sale. This highlights the importance of the revenue from seed harvest in the middle years of the project. Revenue from all products (i.e. craftwood, heartwood, sapwood, second cut chips and seeds) is USD 73,493/ha and USD 105,111/ha at ages 15 and 20 years respectively.
The base-case project has substantial upfront costs of USD 3,265/ha, of which about 21% are labour costs. Adding the initial investment to the discounted costs of maintenance and protection activities up to year eight, USD 4,692 are needed before any cash inflows are received. Discount rate plays a significant role in the profitability of the activity. Payback happens at year 11 for the base-case scenario and the one with a 10% discount rate and at year 14 for r = 17% (Fig. 2). For the scenario with r = 28%, the inflows by year 20 are just enough to make up for the establishment and maintenance costs, as the mean IRR for the sandalwood plantation investment was 28%.
In Vanuatu smallholder sandalwood plantings are primarily established using their own equity, particularly labour. Therefore, any increase in the scale of sandalwood planting is limited by labour availability and thus would require external inputs such a finance. The IRR calculated in this study of 28% is at the top of the range of the previous studies on sandalwood plantations of 16-28% (Page et al. 2010a;Thomson et al. 2011). However, smallholders only source of credit is often through informal money lenders. In such circumstances, interest rates tend to be high. In Ethiopia for example, informal money lending had interest rates of 50% and over (Duguma 2013) and in Pakistan between 31 and 36% (Hussain and Thapa 2012). This emphasises the need to enhance smallholder access to formal credit. Even though 28% is the discount rate for which the NPV equals zero, the NBV Small Loan (Table 3) has appropriate magnitude for sandalwood investment and its interest rate is the closest to reality for most smallholders in developing countries.
The amount needed to invest is compatible with all three borrowing opportunities. However, the cost of capital (interest charged plus opportunity cost of income foregone) for smallholders might be too high. Only the business loan with an interest rate of 10%, providing loans from USD 438 to USD 43,860 with a variable loan term, might be suitable for a profitable investment in sandalwood tree growing. Despite the profitability of the investment under such circumstances, the requirement for a 50% deposit of the loan can constrain smallholder access to the financial service. Poorer   Fig. 2 Project balance and annual net cashflow per hectare for smallholder sandalwood plantations with discount rates (r) of 8, 10, 17 and 28% smallholders have low levels of asset liquidity (Daniels 2001) and the provision of deposits or collaterals for loans can be a barrier for them. Alternatives have been created to enable smallholders to access credit. For instance the warehouse receipt system available in the USA, Latin America, Asia and Africa, (Chapoto and Aboagye 2017;Coulter 2009). In this system farmers can deposit farm production (grains) at a warehouse and use it as a collateral for a loan. While land titles can be used as collateral for loans across many market economies (Besley and Ghatak 2010;Deininger and Feder 2001), most land in Vanuatu is communally owned and under informal customary title making it challenging to mobilise as collateral (Curtin and Lea 2006). Besides the need for collateral or deposit, the term of the loans is often insufficient for forestry. The seven-year term of the ANZ Personal Finance loan is too short considering the project balance for r = 17% is negative until year 14.
Smallholder sandalwood production using owner equity is likely to continue to be the main mode of sandalwood expansion in Vanuatu. Van Der Meer Simo et al. (2020) demonstrated that independent smallholder agroforestry production of the aromatic bark tree Persea kurzii over seven years combined with rice and banana production had greater NPV than a range of contract models growing eucalypts on a similar rotation period. While Persea kurzii serves as a good example of independent smallholder-led agroforestry systems, their profitability is dependent upon maintenance of high value markets. In Thailand the NPV of short rotation (seven years) Eucalyptus for both contract and independent smallholder growers was positive, with no significant difference between them (Boulay et al. 2013). Therefore, independent smallholder forestry systems that use the owner's equity can be competitive with those using external finance or company support, although scale might be limited by labour for smallholder grown Vanuatu sandalwood.

Policy implications
We have demonstrated that smallholder-based sandalwood forestry systems can profitable and competitive with other labour opportunities. The Vanuatu Forest Policy promoted the establishment of 20,000 ha of planted forests by 2020 by way of largescale plantations, community forestry plantations and woodlots (Vanuatu 2013). Community and smallholder contribution to this target is constrained by different types of risks and limitations typical to the smallholder context in remote areas. In order to improve the profitability of the systems we make the following policy recommendations.
To scale up smallholder investment in sandalwood planting, their access to credit must be enabled and financial services need to be suitable for the nature of smallholder forestry. Interest rates must be low enough to maintain the financial profitability, magnitude of loans must match initial investment, loan term must take into account the long-term feature of forestry and the requirements for loans must be consistent with what smallholders can meet. There have been efforts to increase access to rural credits at reasonable interest rates in Vanuatu for decades, through the Comprehensive Reform Program, followed by the country's Priorities and Action Agenda and several national institutions created to provide inclusive financial services (McCaf-frey 2011). However, the reach of financial services can still be improved, since only 13% of people had access to bank accounts in 2007 (ADB 2007), and in 2011 McCaffrey (2011) estimated that 19% of the population used financial services. Loans targeted at smallholder forestry or incentives will be required to expand the industry.
The attractiveness of smallholder sandalwood growing can be increased with the incorporation of other components into the system for earlier financial returns and faster payback. Techniques to reduce damage by cyclone need to be promoted. Most importantly, smallholders can benefit greatly from increased tree security. Enhancing governance and social structures within and across villages might be the most financially viable means of increasing tree security.

Conclusions
Because smallholders in other parts of the world, particularly in remote locations, face challenges similar to those present in Vanuatu, our findings are also relevant beyond the Vanuatu context. Our case study demonstrates that by addressing limitations typical to smallholders, and particularly in remote locations like Vanuatu, small-scale forestry can become an attractive activity to complement local livelihoods. There is the opportunity to tackle several regional and global goals with smallholder forestry. Small-scale tree growing can help address several of the United Nations Sustainable Development Goals, including numbers 1, 2, 6, 7, 8, 13, 15 and 16 (De Jong et al. 2018;Katila et al. 2017;Mbow et al. 2014;Waldron et al. 2017). Addressing issues related to smallholder access to financial services, risks to tree plantations, and timing of returns from forestry can lead to higher levels of tree planting by smallholder farmers and better outcomes of these initiatives.
It is crucial to acknowledge the limitations of smallholder forestry and provide smallholder forestry investors with realistic information regarding expected outcomes and risk. Despite the intention to lift people out of poverty through forestry, encouraging people to commit to an unprofitable activity can hinder the emergence of better livelihood alternatives (Arnold and Pérez 2001). To make the financial models developed in this study publicly available for broader use and better decision making regarding investments in sandalwood, a user interface has been developed and can be accessed on: https://sites.google.com/site/agroforestrytreeproducts/. The website contains scientific information and the financial models for sandalwood and other relevant native species from Vanuatu.