This study uses a systematic literature review (SLR) design that follows the guidelines of PRISMA 2020. SLRs were chosen because they allow for the systematic and transparent identification, evaluation, and synthesis of evidence ( Moher et al., 2009). This design is suitable for answering complex research questions and involves different types of socio-economic outcomes from various study designs ( Petticrew & Roberts, 2008). The PRISMA flowchart used in this study is presented in Figure 1.

The eligibility criteria in this systematic review are determined based on the PEO (Population, Exposure, Outcome) framework. The PEO framework was chosen because the research questions did not involve a comparison group, suitable for observational and qualitative studies in the environmental and social fields. Full details of the inclusion and exclusion criteria are presented in Table 1.

Population: A study that includes the human population in Indonesia, including households, farmers, forestry workers, indigenous communities, women, the poor, and people living in areas with deforestation or forest-related programs.

Exposure: Deforestation, forest loss, or land use change as the main variables. These exposures can be in the form of oil palm expansion, food estates, REDD+, illegal logging, small-scale mining, forest fires, or post-disaster relocation.

Outcome: At least one socio-economic outcome, which includes social policy, social protection, welfare, poverty, or livelihood. Studies that only report environmental outcomes (such as carbon stocks, biodiversity) or macroeconomic outcomes (such as national GDP) without any correlation to socio-economic aspects are excluded.

In addition to the three PEO components above, the eligibility criteria also include the following aspects. The databases used are only Scopus, ScienceDirect, and Springer. The year of publication is limited from 2016 to 2026. The accepted language is English. Document types included were only original empirical studies articles, while books, book chapters, proceedings, review articles, editorials, commentaries, systematic reviews, and meta-analyses were excluded. Full text access must be open access (available free of charge immediately after publication). The research location must be in Indonesia.

Literature searches were conducted on three electronic databases: Scopus, ScienceDirect, and Springer. These three databases were chosen because they have a wide range of environmental, social, and policy sciences, and provide adequate filter features. The last search date was March 31, 2026. Details of the three databases along with the date of access are presented in Table 2.

The search strategy was developed based on the PEO (Population, Exposure, Outcome) framework. For the Scopus database, the search strategies used are as follows: (TITLE-ABS-KEY (deforestation OR “forest loss” OR “land use change”) AND TITLE-ABS-KEY (“social policy” OR “social protection” OR welfare OR poverty OR livelihood) AND TITLE-ABS-KEY (Indonesia)) AND PUBYEAR >2015 AND PUBYEAR <2027 AND (LIMIT-TO (DOCTYPE, “ar”)) AND (LIMIT-TO (LANGUAGE, “English”)) AND (LIMIT-TO (OA, “all”)). The three blocks are combined using the Boolean AND operator. The filters applied include the year of publication (2016–2026), language (English), document type (research article), and open access (open access).

Study selection was conducted in three main stages by two independent reviewers (Author 1 and Author 2). The first stage is title and abstract screening based on the eligibility criteria in Table 1. The second stage is the removal of duplicates using Mendeley software. The third stage is full-text screening to decide on the final inclusion. Any discrepancies between the two reviewers are resolved through discussion or involve a third reviewer if no consensus is reached. This selection process does not use automation tools. Complete study selection flow, including the number of studies identified (n = 144), after title and abstract screening (n = 126), after duplicate removal (n = 124), after full-text screening (n = 18), and number of included studies (n = 18).

Data from each study that met the criteria were extracted using a pre-designed structured form. The form includes information about the identity of the study (author, year, title, journal), population characteristics (location, sample size, demographic group), exposure (type of deforestation, duration, intensity), outcome (reported socio-economic domain, specific indicators, measurement methods), key findings (direction of relationship, effect size if any, p-value or confidence interval), as well as funding sources. Extraction was carried out by two reviewers independently, then the results were compared to ensure consistency. If there is unclear or missing data, the study authors are contacted by email a maximum of twice.

Due to the high heterogeneity in the study design, exposure type, and outcomes measured, no statistical meta-analysis was performed. Instead, the thematic synthesis method developed by Thomas & Harden (2008) is used. The thematic synthesis process consists of three stages. The first stage is line-by-line coding of the results of each study. The second stage is the formation of a descriptive theme by grouping similar codes into initial themes that are still close to the original text. The third stage is the development of an analytical theme that goes beyond the original study content to generate new interpretations, answer research questions, and identify patterns and contradictions.

To explore heterogeneity, a subgroup analysis was conducted based on three characteristics: geographical area (Sumatra, Kalimantan, Java-Bali, Sulawesi, Nusa Tenggara, Maluku-Papua), type of exposure (oil palm, food estate, REDD+, illegal logging, mining, fires, general land cover change), and study design (cross-sectional, longitudinal, qualitative, mixed methods). The results of the analysis of these subgroups are integrated into a thematic synthesis to uncover variations in relationships between contexts. Sensitivity analysis was not performed because there was no meta-analysis; Instead, findings from studies with high risk of bias were discussed separately and compared with findings from higher-quality studies.

A total of 18 articles met the eligibility criteria and were included in this systematic review. In summary, most of the studies are located in Sumatra and Kalimantan ( Chrisendo et al., 2020; Duker et al., 2019; Edwards, 2019; Hasudungan et al., 2024; Herwirawan et al., 2017; Howson, 2018; Pramono et al., 2021; Sulistyorini et al., 2025), some in Java ( Fattah et al., 2025; Garcia-Fry et al., 2022; Rahman et al., 2017; Widianingsih et al., 2024), and the rest cover national areas or inter-island comparisons ( Edwards, 2019; Fattah et al., 2025; Pramono et al., 2021; Sulistyaningrum & Nasution, 2026; Syahza et al., 2020). Only a few studies include Nusa Tenggara and Papua ( Herwirawan et al., 2017; Pramono et al., 2021). The dominance of studies in Sumatra and Kalimantan reflects the intensity of deforestation and expansion of oil palm plantations on the two islands, but also indicates the limited representation of eastern Indonesia. The complete characteristics of each study are presented in Table 3.

In terms of research design, quantitative studies dominate with cross-sectional and longitudinal approaches ( Chrisendo et al., 2020; Edwards, 2019; Fattah et al., 2025; Herwirawan et al., 2017; Sulistyaningrum & Nasution, 2026). Qualitative studies with case study, ethnography, and photovoice approaches are also quite numerous ( Duker et al., 2019; Howson, 2018; Spiegel, 2020; Sulistyorini et al., 2025; Widianingsih et al., 2024; Wisnaeni & Najib, 2025; Yeny et al., 2022). Some studies used mixed methods, such as Hasudungan et al. (2024), Garcia-Fry et al. (2022), and Fawzi et al. (2020). This diversity of designs enriches the understanding of the relationship between deforestation and socio-economic outcomes from various perspectives.

The most studied type of deforestation exposure is oil palm expansion ( Chrisendo et al., 2020; Edwards, 2019; Hasudungan et al., 2024; Howson, 2018; Sulistyorini et al., 2025). National food estate programs and REDD+ have also received significant attention ( Duker et al., 2019; Howson, 2018; Wisnaeni & Najib, 2025; Yeny et al., 2022). Several studies examined deforestation due to illegal logging, small-scale mining, forest fires, and land cover change in general ( Fawzi et al., 2020; Herwirawan et al., 2017; Pramono et al., 2021; Spiegel, 2020; Sulistyaningrum & Nasution, 2026). The most commonly reported socio-economic outcomes were poverty and livelihoods, followed by welfare (income and food security), while social policies and social protection were less often measured directly.

Some studies have found that oil palm expansion is associated with a reduction in poverty rates at the district level. Edwards (2019) reports that a 10 percentage point increase in oil palm planted area from 2000 to 2015 is associated with a 6.1 percentage point faster poverty reduction and a growth in per capita household expenditure of about 9 percent faster than in districts with smaller oil palm areas. These findings are reinforced by Fattah et al. (2025) which shows that village funds consistently reduced poverty and unemployment in almost all regions of Indonesia from 2015 to 2023, even though land-use changes have a negative impact on the growth of the agricultural sector.

However, the aggregate findings hide a sharp inequality at the household level. Hasudungan et al. (2024) reports that although some households have experienced increased income and market access from palm oil expansion in Sumatra and Kalimantan, the poor face higher food insecurity, decreased nutritional diversification, and increased dependence on food purchases. Chrisendo et al. (2020) reinforces these findings by showing that in Jambi Province, oil palm adoption by smallholders is positively correlated with diet quality through increased income, but this effect is highly uneven. Households that are able to adapt tend to be more prosperous, while vulnerable groups (landless farmers, women, and poor households) actually experience a decrease in welfare. Herwirawan et al. (2017) adds that in the semi-arid border region of East Nusa Tenggara, deforestation is not significantly correlated with poverty or village population, indicating that the relationship between the two variables is highly dependent on the local ecological and socio-economic context.

Fattah et al. (2025) reports that Java-Bali shows the highest policy effectiveness in reducing poverty, while Papua-Maluku-Nusa Tenggara faces the greatest institutional challenges. These differences reflect the quality of governance, access to infrastructure, and development history that differ between regions. In areas with weak institutions, the positive impacts of deforestation (if any) are more difficult to feel by the poor.

Deforestation has a profound impact on people’s livelihoods. Sulistyaningrum & Nasution (2026) reports that forest degradation in Indonesia caused GDP losses of around IDR 822.8 billion, income losses of IDR 260.6 billion, and the loss of about 4,735 jobs in the forestry sector in 2022. The most dramatic impact was seen in indigenous communities. Sulistyorini et al. (2025) documents how deforestation due to the expansion of oil palm plantations in the Dayak Tomun customary forest area, Central Kalimantan, from 2004 to 2005 has transformed forest landscapes into plantations, depriving communities of access to forest resources that are the basis of traditional livelihoods. Spiegel (2020) using photovoice methods on six women in Central Kalimantan shows that small-scale gold mining and deforestation have reduced agricultural space, disrupted forest resources, and forced changes in often unfavorable and unstable livelihoods.

On the other hand, communities are responding to deforestation by diversifying their livelihoods. Garcia-Fry et al. (2022) reports that after the relocation due to the 2010 Merapi eruption in Yogyakarta, around 63 percent of farmers diversified their livelihoods, and around 94 home business entrepreneurships (HBE) were formed. This diversification reduces transportation expenditure to monthly income from around 21–26 percent to 11–18.4 percent, as well as lowers CO2 emissions. Rahman et al. (2017) shows that agroforestry based on durian-cassava and teak-maize yams has a higher NPV value and benefit-cost ratio than the slash and slash-burn system (swidden). Agroforestry also strengthens land use rights, increases social prestige, strengthens social cohesion, and creates job opportunities.

Intervention programs designed to divert livelihoods away from destructive activities have also shown mixed results. Fawzi et al. (2020) reports that the chainsaw buyback and entrepreneurship program in Gunung Palung National Park succeeded in diverting 50 poaching pairs to 90 new businesses with income equivalent to the regional minimum wage. However, the business failure rate reached 33 percent and the rate of back to logging was only 6 percent. The problem is that the number of active loggers in the national park has not decreased significantly as new loggers fill the gap. Howson (2018) and Duker et al. (2019) also criticize alternative livelihood programs in REDD+ projects that do not take into account social and gender dynamics, thus exacerbating the exclusion of marginalized groups.

Welfare in the context of deforestation includes both monetary and non-monetary dimensions. Evidence from 18 studies shows a systematic trade-off between increased monetary income and a decrease in non-monetary aspects of welfare, especially in vulnerable groups. Hasudungan et al. (2024) reports that in Sumatra and Kalimantan, oil palm expansion increases income and cash flows for some households, but poor groups experience food insecurity, decreased nutritional diversification, increased dependence on food purchases, and risk of malnutrition. Chrisendo et al. (2020) found that in Jambi, oil palm adoption was positively correlated with diet quality through increased income, while women’s off-farm work was also positively correlated with nutrition. However, this positive effect is only enjoyed by households that have adequate assets and adaptive capacity.

Sulistyaningrum & Nasution (2026) shows that informal workers in the forestry sector have lower access to banking services and food security than formal workers, but access to health services does not differ statistically significantly. These findings indicate that social protection in Indonesia still does not adequately reach informal workers in the forestry sector. Pramono et al. (2021) reports that the surge in Chinese foreign investment through the Belt and Road Initiative (BRI) project in Indonesia has led to land clearing, habitat fragmentation, water and air pollution, carbon emissions, and risks to indigenous peoples and local livelihoods.

On the other hand, a local wisdom-based approach shows the potential to improve well-being without sacrificing ecological functions. Syahza et al. (2020) reports that the use of peatlands in Riau Province for agriculture, plantations, forestry, and other economic ventures makes a major contribution to the welfare of coastal and rural communities, as long as it is done carefully by maintaining hydrological, biodiversity, and carbon storage functions. Rahman et al. (2017) also confirms that agroforestry is not only more economically profitable but also reduces pressure on the surrounding forests.

Top-down social and social protection policies have shown generally disappointing results. The national food estate program in Central Kalimantan’s peatlands is the most obvious example. Yeny et al. (2022) reports that food estate development has a medium to high level of risk, with community activities and changes in farming methods as the highest sources of risk. Wisnaeni & Najib (2025) reinforces this finding with a profound ecological-legal critique. The food estate program is considered to have failed to achieve food production targets, caused deforestation and peat degradation, increased carbon emissions, displaced the livelihoods of the Dayak indigenous people, and violated the principle of constitutional decentralization. The absence of the Free Prior and Informed Consent (FPIC) process is one of the root problems.

The REDD+ project has also received similar criticism. Howson (2018) reports that the implementation of the Lamandau River REDD+ project has exacerbated the exclusion of marginalized groups, especially landless farmers, women, and rubber tappers without access to land. The REDD+ market mechanism creates “slippery violence” through the closure of protected areas, the entry permit system, and the uneven distribution of incentives. Duker et al. (2019) compares two REDD+ projects in Indonesia and Ethiopia, concluding that the neglect of smallholder agriculture interests makes it difficult to achieve emission reduction and welfare improvement targets.

However, not all social policies fail. Fattah et al. (2025) shows that village funds allocated since 2015 have consistently reduced poverty and unemployment in almost all regions of Indonesia. Although village funds are not specifically designed to respond to deforestation, their success shows that village-based approaches that provide autonomy and flexibility to local communities are more effective than centralized mega projects. Widianingsih et al. (2024) also highlights that changes in land use for the benefit of villages in Tarumajaya Village, Bandung, through proposed land use by large companies, can increase village resilience, reduce agrarian inequality, and promote social justice.

A synthesis of 18 studies reveals some common patterns. First, the relationship between deforestation and socio-economic outcomes in Indonesia is heterogeneous and highly contextual, with no universal relationship. This is reflected in the findings of Edwards (2019) which show that palm oil expansion reduces aggregate poverty at the district level, while Hasudungan et al. (2024) and Chrisendo et al. (2020) report that poor groups actually experience increased food vulnerability and decreased nutritional diversification. Second, there is a systematic trade-off between aggregate economic benefits (increased income, reduction in aggregate poverty) and losses to vulnerable groups (decreased food security, loss of access to resources, increased vulnerability), as documented in Sulistyaningrum & Nasution (2026) on GDP losses and income from forest degradation.

Third, the most negatively affected groups are those who are already structurally vulnerable: landless farmers, women, indigenous peoples, and informal workers. Sulistyorini et al. (2025) and Howson (2018) show how Dayak indigenous communities and marginalized groups in REDD+ projects are losing access to the forest resources on which they base their livelihoods, while Spiegel (2020) through the photovoice method reveals the impact of deforestation on women in Central Kalimantan. Fourth, social and social protection policies that are top-down and ignore the local context are likely to fail, as evidenced by Yeny et al. (2022) and Wisnaeni & Najib (2025) on food estates and Howson (2018) and Duker et al. (2019) on REDD+. In contrast, locally-based approaches such as village funds ( Fattah et al., 2025), agroforestry ( Rahman et al., 2017), peat local wisdom ( Syahza et al., 2020), and participatory programs ( Fawzi et al., 2020; Widianingsih et al., 2024) show potential for success.

Fifth, there is significant regional variation, with Java-Bali showing the highest policy effectiveness, while Papua-Maluku-Nusa Tenggara faces the greatest institutional challenges, as revealed in Fattah et al. (2025). Herwirawan et al. (2017) also shows that in East Nusa Tenggara, deforestation is not even significantly correlated with poverty, indicating the importance of the local context. Sixth, gender roles are very important but often overlooked; women are more negatively affected by deforestation and gender-insensitive programs, as documented in Chrisendo et al. (2020), Howson (2018), and Spiegel (2020). A summary of the relationship directions for each outcome is presented in Table 4.

There were some limitations in the evidence synthesized from the 18 studies. First, geographic bias is very clearly visible. Most of the studies are located in Sumatra and Kalimantan, while eastern regions of Indonesia such as Papua, Maluku, and Nusa Tenggara are highly under-represented. This is acknowledged in Herwirawan et al. (2017) which examines the border region of North Central Timor, as well as Fattah et al. (2025) which shows that Papua-Maluku-Nusa Tenggara faces the greatest institutional challenges yet few studies have addressed the region in depth. As a result, the generalization of the findings throughout Indonesia has been limited, as the ecological and socio-economic characteristics of the eastern region differ significantly from the western region of Indonesia.

Second, the dominance of cross-sectional design over longitudinal studies is an important methodological limitation. The majority of included studies use cross-sectional designs, such as Sulistyaningrum & Nasution (2026), Edwards (2019), and Herwirawan et al. (2017), which are only able to capture relationships at a single point in time without being able to draw strong causal conclusions. Only a few studies use longitudinal design, such as Garcia-Fry et al. (2022), Chrisendo et al. (2020), and Fattah et al. (2025). These limitations make it difficult to determine the direction of the cause-and-effect relationship between deforestation and socio-economic outcomes. In addition, non-standard outcome measurements between studies make it difficult to quantitatively compare effect measures, such as the differences in poverty and welfare indicators used between Edwards (2019), Chrisendo et al. (2020), and Fattah et al. (2025).

Third, the limitations of the review process also need to be noted. Only 18 studies met the strict inclusion criteria (open access, English, 2016–2026), so there may be relevant publications missed, especially from grey or Indonesian-language literature. In addition, no studies simultaneously measured all five socio-economic outcomes within a single analytical framework, so the interactions between outcomes (e.g. between poverty, livelihoods, and social policies) could not be directly analyzed. The potential for publication bias is also a concern, where studies with positive or significant results are more likely to be published than studies with negative or non-significant results.

Based on the above conclusions, the following recommendations were formulated for three stakeholder groups. For governments to be able to integrate socio-economic impact analysis, including food security, livelihoods, and gender, in every land conversion licensing process as well as strategic projects such as food estate and REDD+. Require the implementation of Free, Prior, and Informed Consent (FPIC) for indigenous peoples affected by deforestation. Implement a temporary moratorium on food estate expansion in peatlands and primary forests until a participatory impact study is available. Strengthen village-based approaches through the allocation of village funds for conservation-based alternative livelihood programs.

In addition, researchers can standardize multi-location longitudinal studies with standard outcome indicators to enable future meta-analysis. Expand the scope of the research area to the east of Indonesia, especially Papua, Maluku, and Nusa Tenggara, which are still under-represented. Use participatory approaches such as photovoice and focus group discussions to capture the perspectives of women and indigenous peoples that are often overlooked.

Finally, for development practitioners to prioritize agroforestry and community-based forest management over large-scale monoculture projects. Include adaptive social protection components, such as temporary income guarantees and conditional food access, in any alternative livelihood program. Actively involve women in the planning, implementation, and evaluation of programs as they are most affected but often excluded.