Assessing the Level of Poverty and Utilization of Government Social Programs Among Tobacco Farmers in Indonesia

Abstract

Introduction

Studies examining profit suggest that former tobacco farmers do as well or better than current tobacco farmers. Research has yet to examine the relationship among current and former tobacco farmers, poverty, and receipt of government social assistance. This type of research is critical to understanding the direct and indirect subsidization of tobacco growing. This study analyzed tobacco farmers’ poverty levels and receipt of government social assistance programs.

Aims and Methods

We designed and conducted an original four-wave economic survey of current and former tobacco farming households in Indonesia between 2016 and 2022. We then used descriptive analysis and probit regression for panel data to estimate the relationship between tobacco farming and poverty status.

Results

Tobacco farmers’ per capita income and poverty rates vary across years. The poverty rate was significantly higher in the year with a higher-than-normal rainfall as it negatively affected farming outcomes. During this year, the poverty rate among current tobacco farmers was also higher than that of former tobacco farmers. Regression estimates from the panel data confirm the association between tobacco farming and the likelihood of being poor. We also found a high share of current tobacco farmers who receive government social assistance programs, such as cash transfer programs and a universal healthcare program.

Conclusions

Our findings show high poverty rates—particularly during bad farming years—and high rates of government social assistance among tobacco farmers. The high rates of government assistance among tobacco farmers living in poverty show that the government is indirectly subsidizing the tobacco industry.

Introduction

Tobacco farming is promoted as a lucrative crop by the tobacco industry, particularly in low- and middle-income countries.^1–4 Tobacco is presented as an income earner that contributes as a revenue source for the government and an important livelihood option for farmers.^2,5,6 Tobacco-growing countries often adopt this narrative in their approach to tobacco production, either directly or indirectly supporting tobacco supply. This entrenched narrative is then used to oppose tobacco control efforts, with many of the major tobacco-growing countries being the strongest opponents of domestic and global tobacco control.^7–9 Although, on the surface, tobacco appears to be a lucrative commodity, evidence consistently suggests that many facets of tobacco farming remain problematic.^10,11

Indonesia remains the only country yet to ratify the Framework Convention on Tobacco Control in the Asia Pacific region.^12,13 Indonesia is 1 of the top 10 tobacco-producing countries in the world, though it uses only 1.15% of total commercial agricultural land for tobacco farming.^14,15 Tobacco farming is mainly concentrated in the Central Java, East Java, and West Nusa Tenggara regions and comprises only 0.03% of the gross domestic product.¹⁴ Indonesia has been a net importer of tobacco leaf in the past two decades to fulfill domestic cigarette and kretek production.¹⁴ The political economy of tobacco production has hindered advancement of tobacco control measures, due in large part to the power of tobacco interests in the country. For example, cigarette prices in Indonesia are among the most affordable in the world and tobacco taxes remain well below international standards.^16,17 Public health groups have advocated strongly for an excise tax increase and reform of the complex-tiered tax structure.¹⁸ The Government of Indonesia raised taxes by merely 10% in 2023 and 2024, which is unlikely to outpace inflation and economic growth and will not effectively reduce consumption in Indonesia.¹⁹

Tobacco consumption remains high as of 2019 (71.3% of men aged 15 and above and 3.9% of women in the same age category) in Indonesia.²⁰ High levels of tobacco consumption have resulted in an estimated economic loss of US$45.9 billion with almost 2 million tobacco-related illnesses and over 230 000 tobacco-related deaths in 2015.^10,21 Despite the negative health, social and economic consequences of consumption, many government agencies remain supportive of the tobacco industry, framing health-based regulatory efforts as an attack on sovereignty and national pride, emphasizing that tobacco is an important source of government revenue.^10,22 For example, pro-tobacco industry arguments in the national media state that regulations targeting kretek, a clove-flavored cigarette produced in Indonesia, can lead to a loss of national character and that external foreign powers are trying to kill the tobacco industry via regulations.²²

While the policies on tobacco are divided between pro-industry growth and tobacco control measures, the former remains dominant. The Ministry of Industry released a “Tobacco Industry Roadmap” around the same time as the Ministry of Health released the “Tobacco Control Roadmap.”^23,24 The tobacco industry roadmap outlined several government supports provided to tobacco companies through the current multi-tiered tax structure, and policy targets that aimed to increase tobacco production by 12%, an annual cigarette production of 260 billion cigarettes, by 2020.^23,24 Within this roadmap there is recognition of the health risks of tobacco consumption and it presents familiar messaging promoting now widely discredited harm reduction approaches like low-tar cigarettes and social responsibility programs, which are known to be used as a form of marketing by the tobacco industry.²³ The government’s approach to the tobacco industry is premised on the assumption that tobacco remains a lucrative economic commodity.²³

Despite the prominence of this narrative of tobacco as a lucrative economic commodity in Indonesia and elsewhere, there is an emerging body of literature across multiple tobacco-growing countries that challenges this narrative.^6,25–28 This body of research has focused on the economic conditions of tobacco farming households but has not specifically examined these economic conditions concerning poverty and government social assistance. The emphasis on poverty levels is important because governments often support households that fall within the category of poverty. However, providing government support to tobacco farmers facing poverty is often overlooked and contradicts the industry narrative of tobacco being a lucrative commodity for farmers. Concerning this nearly ubiquitous industry narrative, there is a need to examine how tobacco farming is associated with farmers’ income and poverty levels and how governments directly or indirectly support production.

In this study, we analyzed data from four waves of a longitudinal survey conducted among tobacco and non-tobacco farmers between 2016 and 2022 to understand the association of tobacco farming with poverty levels and the extent of government social assistance for farmers in Indonesia. We analyzed income and poverty levels among farmers and the government support tobacco farmers receive. Within this analysis, we compared the levels of profit, poverty, and government support received among current and former tobacco farmers in each of the four farming seasons and assessed the data trends across seasons to provide a picture of smallholder tobacco farming as an income earner. Using panel data regression, we then examined the relationship between tobacco farming and poverty status. The results of this analysis provide insights into the economics of tobacco growing in Indonesia and the mechanisms of support provided by the government to subsidize production through social assistance programs.

Materials and Methods

We conducted a four-wave Tobacco Farmer Survey (TFS) in Indonesia between 2016 and 2022. In wave 1, we obtained a nationally representative sample of tobacco and former tobacco farmers from Indonesia’s top three tobacco-producing regions: East Java, Central Java, and Nusa Tenggara Barat. Tobacco production in these regions accounted for 90% of total tobacco production in Indonesia. We followed the same households of current and former tobacco farmers across waves to generate panel data. However, due to resource constraints, waves 2, 3, and 4 TFS only included current and former tobacco farmers from East and Central Java. The share of tobacco farmers in these two regions accounted for about 80 percent of tobacco farmers nationwide. We obtained four-wave panel data of 646 tobacco and non-tobacco farmers. The waves 1 to 3 TFS research reports provided a detailed discussion regarding the sampling procedure.^29,30

The survey instruments for the TFS were initially based on the World Bank’s living standards measurement study and recent studies on the political economy of tobacco farming.^29,30 The TFS instrument comprises more than 25 topics, including household characteristics, planting and land use, agricultural inputs, tobacco and crop sales, and non-farm businesses. We used a computer-assisted personal interviewing method to collect data from current and former tobacco farmers.

The main dependent variable we analyzed in this study is poverty status. To identify poverty status, we first calculated and analyzed the household income per capita of current and former tobacco farmers. Following established studies in the literature, we defined household income as the sum of tobacco farming profit, non-tobacco farming profit, household enterprise profit, wage income, and other income, minus farming costs, rent, and household labor costs.^28,31–34 Household labor costs are incorporated in the calculation of household income because they are an estimate of opportunity costs borne by households’ labor, which assumes that if farmers were not laboring in the tobacco fields, they could at very least sell their labor at the lowest prevailing wage. We argue that this is a reasonable assumption because these districts typically have varied and robust local economies beyond tobacco growing. We used district-specific agricultural minimum hourly wages and the reported household labor hours for farming to estimate household labor costs. Lastly, we divided household income by the number of total household members to obtain household income per capita. We converted the household income per capita to US$ using wave-specific (purchasing power parity) PPP exchange rates from the OECD database.^35,36

We calculated the poverty rate using the headcount ratio measured by per capita household income and a poverty line of $1.90 in 2011 PPP a day per person. The wave-specific poverty line was calculated using the 2011 PPP and adjusted for inflation between 2011 and the wave-specific year. A household is considered poor in a particular wave if the per capita household income is strictly below the poverty line. Before the analysis, we dropped outliers belonging to farming households with income lower than the 5th and higher than the 95th percentiles in each wave. The final analytical dataset consisted of unbalanced panel data of 623 current and former tobacco farmers. Analysis using balanced panel data of 389 current and former tobacco farmers yielded similar results and can be accessed in the online supplementary analysis. The online supplementary analysis can be accessed through the following link: https://bit.ly/tobacco-poverty-supplementary.

We summarized current and former tobacco farmers’ characteristics and farming outcomes in Table 1. Wave-specific summary statistics are presented in Table S1 of the online supplementary analysis. The typical current and former tobacco farmers were poor, but the average is higher among current farmers than former tobacco farmers. Current tobacco farmers generated lower income per capita than former tobacco farmers, which can be explained by lower crop sales, income from farming enterprises, and other income.

Table 1.

Summary Statistics, in Median

	Former tobacco farmer	Current tobacco farmer	Total
1 if household is poor, median	1	1	1
	(0.495)	(0.468)	(0.477)
1 if household is poor, mean	0.571	0.676	0.649
	(0.495)	(0.468)	(0.477)
Income per capita, US$ PPP	588.6	328.9	399.6
	(997.3)	(1610.5)	(1486.9)
Share of land for tobacco	—	33.33	25
	—	(18.20)	(21.41)
Cultivated land, hectare	0.804	0.750	0.750
	(150.7)	(132.4)	(137.4)
Share of land for non-tobacco	0	0.667	0.750
	(0.0394)	(0.182)	(0.215)
Quantity of tobacco produced, kg	—	307.8	307.8
	—	(1330.2)	(1330.2)
Total input tobacco, PPP	—	245.7	139.5
	—	(945.4)	(850.9)
Total input crop, PPP	584.4	341.5	397.2
	(865.4)	(747.6)	(785.5)
Income from tobacco, PPP	—	331.2	10.77
	—	(3535.1)	(3077.5)
Crop sales, PPP	363.9	256.5	275.7
	(2912.5)	(2725.7)	(2777.5)
Income from farming enterprises, PPP	22.03	0	0
	(825.6)	(1120.8)	(1052.4)
Income from non-farming enterprises, PPP	0	0	0
	(106.9)	(198.1)	(179.0)
Other income, PPP	94.18	64.02	69.25
	(1418.1)	(898.6)	(1064.0)
HH profit, PPP	2473.1	1448.2	1719.3
	(4300.5)	(8226.2)	(7451.3)
HH profit per area, PPP	2782.7	1648.4	1938.7
	(30 519.3)	(29 581.2)	(29 959.1)
Resource per area, PPP	3757.3	1998.2	2371.7
	(170 546.0)	(306 049 867.5)	(263 491 688.6)
HH labor in hours, tobacco	—	368	192
	—	(636.9)	(602.7)
HH labor in hours, crops	140	0	15
	(354.3)	(340.4)	(348.7)

In the analysis section, we further tested differences among current and former tobacco farmers. We used bar graphs and confidence intervals to compare household income per capita and poverty rates between former and current tobacco farmers. Using the Mann–Whitney U test, we tested differences in means of household income per capita between the two groups of farmers. Using the proportion test, we tested differences in poverty rates between the two groups. Analyses using median and parametric tests yielded consistent results.

We used probit regressions for panel data to estimate the relationship between tobacco farming and poverty status. The model specification for the probit regression is:

$${\displaystyle \begin{array}{l}{\displaystyle Pr(\mathrm{p}\mathrm{o}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{t}\mathrm{y}\mathrm{\_}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{u}{\mathrm{s}}_{it})=\mathrm{\Phi }({\beta }_0+{\beta }_1\mathrm{t}\mathrm{o}\mathrm{b}\mathrm{a}\mathrm{c}\mathrm{c}{\mathrm{o}}_{it}+\gamma X_{it}+a_i+u_{it})}\end{array}}$$

(equation).

where i indicates household, t indicates time, poverty_status indicates households’ poverty status, and tobacco indicates either tobacco farming status or the share of a household’s land used for tobacco farming. The function Ф is the cumulative distribution function of the standard normal distribution.

The vector X includes household characteristics such as a log of total cultivation area, log of labor hours, log of assets, log of agriculture wage, log of non-agriculture wage, an indicator of whether farmers enter a contract with a leaf-buying entity, household size, head of household age, head of household’s years of schooling, and wave-specific fixed effects (FE). We also included dummy variables to indicate households with missing labor hours, assets, agricultural wages, nonagricultural wages, and years of schooling. The term a indicates farmer-specific unobserved characteristics, while u indicates idiosyncratic shocks.

We estimated the model in Equation using the probit FE model to take advantage of the rich panel dataset and to estimate average marginal effects. We used the FE model to accommodate the time-invariant household-specific unobserved effects.³⁷ Estimating the tobacco farming effect using FE probit analysis may result in a biased estimate due to incidental parameter problems.^37,38 Therefore, we used the jackknife bias corrections by Fernández-Val and Weidner (2016) for panel data and a user-written Stata package probitfe for the estimation.³⁹

Estimation using random effects requires a stronger assumption that the household-specific unobserved effects are uncorrelated with tobacco farming, an implausible assumption in our context. In any case, our results show that probit random effects estimations yield qualitative results similar to our fixed effect probit analysis.

We used the Stata (V.17.1) statistical package for the analysis.

Results

Overall, there was a change in income per capita across the four waves. Figure 1 presents household income per capita differences between current and former tobacco farmers. Income per capita was significantly lower in waves 1 and 4 than in waves 2 and 3, particularly for current tobacco farmers. This suggests that a typical tobacco farming household barely generated income in waves 1 and 4, while a typical tobacco farming household generated positive income in waves 2 and 3.

Figure 1.

Per capita household income of current and former tobacco farmers, in US$ PPP

One of the variables that explained the difference in income per capita across waves is weather conditions. Previous studies found that 2016 (wave 1) was considered a “bad” year for farming, as the higher-than-normal rainfall negatively affected income per capita. On the other hand, 2017 and 2019 were considered “good” years as rainfall was close to the long-run average. The waves 2 and 3 TFS reports showed that tobacco farmers generally stated 2017 and 2019 as favorable for tobacco growing in terms of weather.^11,14,30 The abnormal weather in 2016 negatively affected current and former tobacco farmers, but current tobacco farmers disproportionately fared worse. Cultivated land for tobacco, tobacco leaf production, and prices were significantly lower in wave 1 than in waves 2 and 3. Moreover, tobacco farmers bore significantly higher hired labor costs in wave 1.^11,14,30

While also adversely affected by the weather in the “bad” year (2016), former tobacco farmers consistently generated positive income across waves. Their income was also significantly higher in the relatively “good” years (2017 and 2019). We found that a typical former tobacco farmer between waves 2 and 4 generated higher income than a typical former tobacco farmer in wave 1 (Mann–Whitney U test, p-value of .0000, .0060, and .0003, respectively). A typical former tobacco farmer in wave 3 also generated higher income than a typical former tobacco farmer in wave 2 (Mann–Whitney U test, p-value of .0017).

Moreover, a typical former tobacco farmer’s income was significantly higher than a typical current tobacco farmer in wave 1 (Mann–Whitney U test, p-value of .000), in wave 3 (Mann–Whitney U test, p-value of .026), and in wave 4 (Mann–Whitney U test, p-value of 0.003). There is no significant difference in per capita income between the two farmers’ groups in wave 2 (Mann–Whitney U test, p-value of .4472). Using the same data, a previous study also found that first-wave tobacco farmers who switched to non-tobacco farming in wave 2 generated higher income per hectare.¹¹ Explanations for the difference in income between current and former tobacco farmers include former tobacco farmers having a more diverse income portfolio and bearing lower costs for agricultural inputs, hired labor, and household labor.¹¹

Figure 2 shows that poverty rates among former and current tobacco farmers were higher than the national poverty rate as measured by the poverty headcount ratio at $1.90 a day using the 2011 PPP. More than half of current (68%) and former tobacco farmers (57%) were considered poor, while the national poverty rates in 2016, 2017, 2019, and 2021 were 10.9%, 10.6%, 9.41%, and 10.4%, respectively. The figure also shows the dynamics of poverty among current and former tobacco farmers across waves, corresponding to the income dynamics discussed previously. The share of tobacco farming households that were considered poor was significantly higher in wave 1 compared to wave 2 (proportion test, p-value of .0000), wave 3 (proportion test, p-value of .0000), and even wave 4 (proportion test, p-value of .0000). Poor tobacco farming outcomes driven by the unfavorable weather was one of the main explanations for the high poverty rate in wave 1.

Figure 2.

Poverty rate among current and former tobacco farmers.

There were noticeable differences in poverty rates between former and current tobacco farmers, particularly in waves 1 and 3. Poverty rates among former tobacco farmers were significantly lower than poverty rates among current tobacco farmers in wave 1 (proportion test, p-value of .0000), wave 3 (proportion test, p-value of .0627), and wave 4 (proportion test, p-value of .0158). While wave 3 was a relatively successful tobacco farming year, former tobacco farmers earned significantly higher non-tobacco crop income in the dry season, nonagricultural wages, and other income compared to current tobacco farmers.³⁰ Poverty rates among former and current tobacco farmers were similar in wave 2.¹¹ Both current and former tobacco farmers enjoyed a successful farming year relative to other seasons.

Poverty rates among former tobacco farmers did not change significantly between waves 1 and 2. However, the poverty rate among these farmers was significantly lower in wave 3. One of the explanations was higher income driven by higher non-tobacco crop income in the dry season, nonagricultural wage income, and other income in wave 3 among former tobacco farmers.⁴⁰ Poverty rates among former tobacco farmers did not change significantly between waves 3 and 4 (proportion test, p-value of .1349).

We now turn to the results of the probit FE regressions using panel data to further understand the relationship between tobacco farming and poverty status. The estimated coefficients from analyses of the average marginal effects are presented in Table 2. We found that tobacco farming is associated with a higher likelihood of being poor. An estimate from the probit FE regression suggests that being a current tobacco farmer is significantly associated with a higher likelihood of being poor by about 11.66 percentage points. The estimates from the regressions are remarkably consistent with findings from the graphical and descriptive statistics that poverty rates were generally higher among current tobacco farmers than former tobacco farmers.

Table 2.

Relationship Between Tobacco Farming and Poverty, Average Marginal Effects From Probit Random Effects

	A: Probit panel using FE, wave 1–4	B: Probit panel using FE, wave 1–4	C: Probit panel using RE, wave 1–4	D: Probit panel using RE, wave 1–4
1 if current tobacco farmer	0.1166***		0.0818***
	(0.023)		(0.0227)
Share of land for tobacco		0.0099***		0.000995*
		(0.0034)		(0.000525)
Observations	1342	1342	2126	2126
Controls	Y	Y	Y	Y
Time F.E.	Y	Y	Y	Y
S.E.	Jackknife	Jackknife	Cluster robust	Cluster robust

The signs *, **, *** indicate significance at the 1%, 5%, and 10% levels, respectively. Estimated average marginal effects of control covariates are not shown for brevity but are available in Table S2 of the supplementary analysis. Households with the same outcomes over time are dropped from the estimation using the probit fixed effects. The estimation of standard errors for the random effects model is clustered at the household level. We did not collect assets during the wave 4 tobacco farmers survey.

The survey also collected information on the total land for cultivating tobacco and non-tobacco crops. Given this information, we calculated the share of land for tobacco farming for each household. Previous studies have shown that tobacco farming is associated with higher farming costs and lower income, resulting in higher poverty levels. Therefore, we hypothesized that a higher share of land dedicated to tobacco farming is associated with a higher likelihood of being poor. The results in Table 2 support our hypothesis. Generally, a higher share of land dedicated to tobacco farming is associated with a higher likelihood of being poor. An estimate from a probit FE regression using panel data suggests that an additional percentage point of land dedicated to tobacco farming is associated with a higher likelihood of tobacco farming by 0.99 percentage points.

We conducted several sensitivity analyses to check for robustness. First, we show wave-specific estimates in Table S3. Except in wave 2, the results are consistent in that a higher share of land for tobacco farming is associated with a higher likelihood of poverty. The main explanation is that the tobacco farming outcomes were better in wave 2. Estimated coefficients from probit random effects are quite similar. Table S4 also shows that regression results using balanced panel data yield consistent results.

Given the robust findings on the relationship between tobacco farming and poverty, we analyzed the receipt of government social programs among current and former tobacco farmers. As shown in Table S5, we found a high share of farmers (24.3%) were beneficiaries of various government social protection programs. This general finding is further supporting evidence of the high poverty rate among farmers. Households who were beneficiaries of government social programs typically, but not always, received a Social Welfare Card (Kartu Perlindungan Sosial, KPS) or a Family Welfare Card (Kartu Keluarga Sejahtera, KKS). Figure 3 shows analyses of the receipt of KPS and KKS among farmers. The share of former and current tobacco farmers who received KPS and KKS was higher in later waves, although the cross-wave differences are not statistically significant. The higher share of KPS and KKS receipt in later waves can be explained by the government’s broader expansion of social protection programs, as the receipt of KPS and KKS is the basis for households to receive various social protection programs.⁴¹

Figure 3.

Receipts of various forms of social assistance programs among current and former tobacco farmers.

The share of current tobacco farmers who received KPS and KKS appears to be higher than former tobacco farmers, particularly in waves 1, 3, and 4. However, the differences are significant only in wave 4 (proportion test, p-value .0113). This aligns with the previous findings that the share of poor households among current tobacco farmers was higher in wave 4 than in wave 3.

The government of Indonesia implements various social assistance programs to support poor and vulnerable households, as shown in Figure 3. These programs include cash transfers—both conditional (Program Keluarga Harapan) and unconditional (Bantuan Langsung Tunai)—, for-poor rice subsidy (Raskin), subsidized or paid health insurance (BPJS-PBI or Kartu Indonesia Sehat), and an education cash transfer (Bantuan Siswa Miskin or now Program Indonesia Pintar programs).

The shares of former and current tobacco farmers who received these programs were quite high. In wave 3, almost 15 percent of former and current tobacco farmers received benefits from the cash transfer program. The shares increased significantly in wave 4, partially owing to the expansion of the cash transfer program during the Covid-19 pandemic. In waves 1 and 2, more than 60 percent of former and current tobacco farmers received benefits from the Raskin. The share of former and current tobacco farmers who received benefits from the Raskin was quite low in 2019 and 2021 because the national government transitioned the program to a non-cash food assistance program (Bantuan Pangan Non-Tunai, BPNT). Many former and current tobacco farmers also received the universal healthcare coverage program (KIS or BPJS-PBI). We found that current tobacco farmers tend to receive higher levels of support across most social assistance programs.

Discussion

Our analysis suggests that many smallholder farmers often struggle with low and uneven income, but it is generally worse for the farmers growing tobacco. We found that over 50 percent of current and former tobacco farmers were below the poverty line in 2016, 2017, 2019, and 2021 while the national poverty average remained lower than or around 10 percent.^42,43 Our panel data analysis also confirmed that tobacco farming and a higher share of land dedicated to tobacco farming are associated with a higher likelihood of being poor. Current tobacco farmers generally received more social assistance from government programs than former tobacco farmers. Tobacco farmers face higher poverty levels and greater income fluctuation because of tobacco farming. These findings align with other research findings from Indonesia and other tobacco-farming countries such as Kenya and Malawi, demonstrating that tobacco farming is often less lucrative than decision-makers and industry narratives suggest.^6,28,29,44

While poverty and government support are important indicators of the precarity of tobacco farming, the crop does remain entrenched in Indonesia. As in other tobacco-growing countries, tobacco farmers receive loans, inputs, and other supports unavailable for many other crops, mainly in contracts with tobacco leaf-buying companies.²⁶ Such support appears to build loyalty towards the tobacco industry.²⁶ However, access to financing facilities for tobacco farmers, as indicated in wave 2 survey data, also impacts perceived tobacco profitability.¹¹ While easy access to financing provides the ability to enter each growing season with the necessary inputs, poverty levels of farmers remain high, highlighting the exploitative nature of many of these contracts, and the need for long-term solutions.

Despite the notion that tobacco as a commodity is lucrative, tobacco farmers depend highly on many of the government’s key social assistance programs. As results indicate in 2016 and 2017 (waves 1 and 2 of the survey), over 60% of tobacco farmers indicated they were recipients of government social assistance. The dual dependency of tobacco farmers, first on industry loans and second on government social assistance programs, creates a dependency cycle that perpetuates poverty and drains government resources. Importantly, this analysis points to how the government indirectly subsidizes industry activity, contributing to the massive profit margins derived from an exploitative supply chain and tobacco companies leveraging the false narrative of prosperity to influence government decisions.³⁵ If the narrative of tobacco being a lucrative economic commodity is true, we should have seen poverty rates lower than the national average among tobacco farmers with little to no reliance on government support, or poverty rates should have been at least higher than non-tobacco farmers. However, as we have illustrated in our results, we do not see either. In this research, we have highlighted that tobacco farming has caused disproportionate poverty among current and former tobacco farmers compared to the national poverty levels in Indonesia. As highlighted above, high poverty levels also lead to high dependency on social assistance. Additionally, previous research has indicated that shifting to alternative crops will positively impact smallholder farmer household income.²⁹ Given the unreliability of tobacco as a crop, its poor performance as a consistent and high revenue earner, and evidence suggesting that there are better alternatives, it will be sound economic and agricultural policy for the government to support alternative crops and/or other off-farm livelihoods. Supporting alternative crops may improve farmers’ income and reduce the likelihood of poverty and reliance on the government’s social protection programs. Additionally, given Indonesia’s high level of domestic tobacco consumption, reducing tobacco farming might also contribute to better public health outcomes.^10,45

The relationship between tobacco consumption and poverty has been an important line of inquiry for tobacco control researchers and advocates, and the linkages are well established.^46–48 Given the narrative that tobacco is a prosperous economic commodity, the linkages between poverty and tobacco farming are less well-understood and remain understated. This study provides an important insight into the poverty levels among tobacco farmers in a relatively lucrative tobacco-growing country and how the government supports these farmers through social assistance programs, thereby serving as an indirect subsidy for tobacco growing. This line of inquiry is critical to inform discussions of tobacco-growing alternatives and the government’s potential to reallocate resources indirectly subsidizing tobacco growing to support alternatives.

Conclusion

Poverty and associated government assistance programs indicate how unlucrative tobacco is as an agricultural commodity for farmers. The research findings of this study point clearly to high rates and higher likelihood of poverty among tobacco farmers and correspondingly high rates of utilization of government assistance. This analysis provides novel insights into how tobacco is linked to the broader political economy of Indonesia and suggests that this line of inquiry can provide a deeper understanding of the relationship between government and tobacco as an economic commodity. Our findings run counter to the narrative of tobacco being a lucrative economic commodity for farmers and indicate that the government of Indonesia is indirectly subsidizing the tobacco industry through the provision of social assistance to tobacco farmers who are living in poverty.

Funding

This research was supported by the Office of the Director, National Institutes of Health (OD) and the National Cancer Institute (NCI) under Award Number R01TW010898; NCI through a CRDF Global grant; the World Bank; and the American Cancer Society. Its contents are solely the authors’ responsibility and do not necessarily represent the official views of these funders.

Declaration of Interests

The authors have no competing interests to declare relevant to this article’s content.

Author Contributions

Gumilang Sahadewo (Conceptualization [Equal], Formal analysis [Equal], Investigation [Equal], Methodology [Equal], Writing—original draft [Equal], Writing—review & editing [Equal]), Raphael Lencucha (Conceptualization [Equal], Formal analysis [Equal], Investigation [Equal], Methodology [Equal], Writing—original draft [Equal], Writing—review & editing [Equal]), Shashika Bandara (Conceptualization [Equal], Formal analysis [Equal], Investigation [Equal], Methodology [Equal], Writing—original draft [Equal], Writing—review & editing [Equal]), Jeffrey Drope (Conceptualization [Equal], Funding acquisition [Lead], Investigation [Equal], Methodology [Equal], Writing—original draft [Equal], Writing—review & editing [Equal]), and Firman Kartaadipoetra (Conceptualization [Equal], Data curation [Lead], Methodology [Equal], Writing—review & editing [Equal]).

Ethics Approval

IRB of the Morehouse School of Medicine, the IRB of record for the American Cancer Society, and the IRB of SurveyMeter.

Data Availability

The data underlying this article will be shared on reasonable request to the corresponding author.