This is the seventh in our series of posts by Ph.D. students on the job market this year
The tripling of area planted with tropical oil crops since the 1990s represents the largest transformation of global food and agricultural systems since the Green Revolution. The area planted for oil crops since the 1970s has expanded by over 150 million hectares, three times that of all cereal crops in the same period (Byerlee, Falcon, and Naylor, 2016). Tropical oil crops feature in most agricultural and food policy debates: genetically modified organisms, food versus biofuels, small farmers versus agribusiness, mono- versus inter-cropping, land grabs, and the environmental footprint of food consumption. The most prominent debates concern clearing forests across the tropics to plant oil crops, particularly oil palm, and the haze that regularly blankets Southeast Asia. Palm oil is the world’s most consumed vegetable oil—ubiquitous in everyday products from food and drink to soap and cosmetics—and one of the world’s most socially contested industries.
The most important questions surrounding the tropical oil crops boom involve managing complex economic—environmental trade-offs, yet there is surprisingly little systematic evidence on how the global palm oil boom has affected welfare in rural communities across the developing world where it is grown. My job market paper asks whether the world’s largest modern plantation-based agricultural expansion—that of palm oil in Indonesia—has been pro-poor. It offers two contributions. The role of agriculture in economic development and poverty reduction has a rich history in development economics (see, e.g., Gollin (2010) and Dercon and Gollin (2014) for reviews), but relatively little attention has been paid to plantation and cash crops with starkly different characteristics to other forms of agriculture (Hayami, 2010). My paper sheds new light on the poverty elasticity of plantation-based cash crop production.
Second, I provide new evidence for a salient policy debate on palm oil across the developing world, where the welfare impacts of tropical oil crops are typically neglected in debates that, often for good reason, tend to focus on environmental issues.
Indonesia’s palm oil boom
I estimate the impacts of the massive expansion in Indonesian palm oil production on local poverty over the 2000s. The area of Indonesia planted with oil palm increased from under four million hectares at the turn of the century to around 12 million hectares today. Smallholder farmers account for most this growth, roughly tripling their area planted to over five million hectares (Directorate General of Estate Crops, 2015). Combining administrative data on local oil palm acreage at the district (kabupaten) level with district poverty estimates calculated from the National Socioeconomic Survey (SUSENAS), I relate decadal changes in the share of each district planted with oil palm to changes in district poverty over the same period. Using the changing share of district area planted allows me to compare the poverty elasticity of oil palm land against alternative uses for rural land, such as forestry, rice, and rubber. I opt for long differences because impacts are unlikely to be fully realized immediately. Companies must establish the necessary infrastructure, hire laborers, prepare land, and plant seedlings. Smallholders similarly need time to switch livelihood, prepare land, plant seeds, then wait for their first harvest around two and a half years later. The main empirical concern is oil palm being endogenously planted due to different underlying conditions, with expanders on different growth and poverty reduction paths. I attempt to overcome this with an intuitive instrumental variable approach.
Exploiting agricultural expansion where it makes the most sense to grow
My identifying variation comes from detailed geo-spatial data on agro-climatic suitability for oil palm, as estimated by the Food and Agriculture Organization’s Global Agro-Ecological Zones (GAEZ) model. I instrument the change in the share of each district’s area used for oil palm with its average agro-climatically attainable palm oil yield, taking the mean value of pixels in each district polygon. Oil palms only grow under certain agro-climatic conditions (humid low-land tropics), and potential yields and profits in each district affect the likelihood that district will have oil palms planted. The main concern with this instrument is that a key input to the oil palm GAEZ productivity model could affect productivity of similar tropical crops and thus welfare through agricultural productivity in other sectors. This is a common challenge with external instruments, particularly those relating to climate and weather (Bazzi and Clemens, 2013; Sarsons, 2015). Using a crop-specific instrument reduces this threat, but I go a step further. GAEZ attainable yield data are available for most of Indonesia’s major crops, so I control for other key crops’ potential yields to further restrict the identifying variation to that relating only to oil palm and not shared suitability characteristics with other crops. Even with this precaution, I must rely on the assumption that agro-climatically suitable oil palm expansion districts are not on different poverty trends for a reason other than oil palm. This exclusion restriction seems reasonable, as placebo estimates on pre-boom poverty trends show that highly suitable expanders were not on different poverty trajectories, and there is no statistically significant relationship between potential yield and poverty in regions that did not increase their planted area.
- Districts that increased their area planted with oil palm achieved more poverty reduction than otherwise similar rural districts with less or no increase in planted area. The magnitude of the estimated poverty reduction from increasing the district share of oil palm land by ten percentage points from my preferred instrumental variable estimator is around 40% of the poverty rate. The figure below shows my main result in the raw data, comparing the average poverty rate of rural districts with palm oil expansion against those without and the national district average.
2. Palm oil expansion districts experienced more rapid increases in household expenditures for people in agriculture and the bottom quintile, and greater provision of public goods most related to agricultural manufacturing, namely roads and electricity. Palm oil expansion also tends to coincide with a sustained boost to primary, industry, and total district outputs, but no discernible impact on services (data are taken from SUSENAS and the World Bank’s DAPOER).
3. A simple policy simulation based on my most conservative estimate suggest that at least 1.3 million rural Indonesians were lifted from poverty exclusively due to growth in the palm oil sector over the 2000s. The palm oil boom thus accounts for a sizeable share of the roughly ten million people lifted from poverty in this decade when national economic growth and poverty reduction slowed, and anti-poverty programs expanded rapidly.
Indonesia’s recent smallholder-led palm oil expansion provides an important case study of how geographically dispersed pro-poor growth can reach remote rural regions. While environmental concerns are likely to remain an important issue in global policy debates about the future of palm oil, income effects may be the first-order issue in communities where it comes from.
Ryan Edwards is a Postdoctoral Research Fellow at Stanford University. His job market paper is available from his website.