Blog

Mapping Smallholder Farmers in Indonesia to Inform Agriculture Policies

Pulse Lab Jakarta
Aug 8, 2019

As the world’s population continues to increase, governments have been stepping up efforts to ensure that citizens have access to safe and adequate food supply. The Sustainable Development Goals are also undergirded in the ambition to end hunger, which is crucial to sustain humanity. Favoured with fertile soils, abundant natural resources and a conducive climate, Indonesia has for years been a key player in the agriculture sector, helping to meet food consumption demands domestically and in the region. The agriculture sector is also a key economic driver and provides employment to millions of smallholder farmers. Collaborating with the Directorate of Cooperatives and Small and Medium-sized Enterprises (SMEs) in the Ministry of National Development Planning (Bappenas), Pulse Lab Jakarta developed an interactive visualisation and analysis dashboard that uses the latest agriculture census data to identify smallholder farmers, with respect to the commodities they produce and the size of their landholdings to generate insights that can inform policies relevant to SMEs in the agriculture sector.

The Livelihoods of Many Smallholder Farmers

In Indonesia, roughly a third of the land is used for agricultural purposes and the sector contributes significantly to the country’s Gross Domestic Product. Consisting largely of smallholder farmers (defined as having landholding less than two hectares) and large plantations, millions of Indonesians depend on this sector to earn their everyday livelihoods. Smallholder farmers, who typically come from agriculture-dependent households, are also part of the segment of SMEs in Indonesia. Information nonetheless regarding SMEs (in the case of smallholder family farmers) that contribute to the agriculture sector remains limited.

Having a better understanding of how many of these smallholder farmers exist; the kinds of commodities they produce; and the land size used can enable the Government to provide more fitting support. Ranging from building practical infrastructure, delivering agriculture and business training programmes, de-linking productivity with climate anomalies/shocks, and providing agriculture production subsidies and credit, such support and policies, all have the potential to improve incomes that can lead to more sustainable livelihoods for smallholder farmers.

The Making of the Dashboard

The agriculture sector in Indonesia is vast, yet not all smallholder farmers are recognised in official statistics. Furthermore, there’s a dearth of information on productions by smallholder farmers, especially at a subdistrict level. The 2013 Agriculture Census, the latest of six agricultural censuses ever conducted in the country, is the most comprehensive data set currently available. The census includes six agriculture categories, namely food crops, horticulture, estate crops, livestock, fishery and forestry.

Within this data set though, there is no indication of which farmers are considered SMEs. Through discussions with Bappenas’ Directorate of Cooperatives and SMEs, we thus decided to experiment with land size as a proxy to determine smallholder farming SMEs. For instance, farmers with landholdings less than one hectare were categorised as SMEs for the food crops category. While for the horticulture category, farmers with landholdings smaller than half a hectare were classified as SMEs.

Once the data was processed and the classifications for smallholder farming SMEs for each agriculture category was established, our next move was to prototype a visualisation and analysis dashboard that could cover all 34 provinces in Indonesia, including data on the national and sub-district levels.

Interactive, Intuitive and Informative

The agriculture census in Indonesia is carried out every ten years, where door-to-door interviews and community surveys are conducted using paper questionnaires. The dashboard was intended to not only present the data collected in a way that was digitally interactive (allowing users to easily switch between provinces and survey commodities of interest), but also do so in a visually appealing way. After rounds of iterations alongside our Bappenas colleagues to ensure optimum user experience, we came up with a prototype that allows data overlaying and multiple presentation views.

Users can select one of the six agriculture categories to view how a particular type of agriculture is distributed across provinces. Visualised above, after selecting one of the categories (for example “Tanaman Pangan” which means food crops), users can also make further selections from a list of commodities (for example “jagung” which means corn). From there, they’re given two options to choose between: the land size used for the commodity (“Luas Lahan”) and the number of households involved in activities related to this commodity (“Rumah Tangga”).

The dashboard also visualises the data at the national level, as well as at the provincial and sub-district levels.

National Level: the colours indicate the scale of the number of households involved with growing corn (yellow showing fewer households and dark blue showing a higher number of households)

Provincial and Subdistrict Levels: Looking at Nusa Tenggara Timur province as well as its sub-districts, the colours indicate the scale of the number of households involved with growing corn (with yellow showing fewer households and light green showing a higher number of households). Users can also select a particular sub-district to view information regarding land size used and number of households involved.

The dashboard offers different data presentation views, allowing users to observe the commodity proportion as well as the number of households associated with it at both the national and provincial levels. From the screenshots below, we see that corn “jagung” accounts for nearly 17 per cent of all agriculture commodities nationally, while again looking specifically at Nusa Tenggara Timur province, corn is the largest commodity in terms of land use.

Across all 34 provinces, we are also able to get a sense of each province’s rank based on the number of households involved in corn production (the top three being Nusa Tenggara Timur, Jawa Timur and Jawa Tengah based on the recent census data).

The dashboard also provides useful visualisations on the amount of land that is used for a particular category of agriculture, which helps with categorising which farmers fall into the SMEs segment. Again, using land size as a proxy to determine SMEs, the areas with the light blue shade indicate smallholder farmers who may be considered as SMEs since the land use is one hectare or less for food crops.

Having an understanding about the type of agriculture commodity, the size of landholding, characteristics of the farm (and the related household) used for production can provide the Government with useful information that can inform resilience efforts when smallholder farmers experience economic shocks. For instance, if smallholder farmers experience sudden, unexpected fluctuations in the price of corn, the Government can better assess what measures to take to render quick support based on the number of households dependent on corn production, with even greater efficiency if the data is up to date. Similarly, should smallholder farmers in a specific region become affected by drought or other natural disasters, the dashboard can help predict which commodities are likely to be impacted and at what scale.

Scaling Up Beyond Prototyping

Using the data from the 2013 agriculture census as the single data source, the dashboard was developed as a rapid prototype. We acknowledge that determining smallholder farming SMEs is multidimensional and to be more accurate, we cannot rely on the size of landholding alone. Therefore earlier this month, the Directorate for Cooperatives and SMEs showcased the features of the dashboard to other ministries and government institutions to discuss how the dashboard could be further developed by integrating additional data sets from other ministries to provide a more comprehensive understanding about how the agriculture sector intersects with other areas of national development, as well as to come with a multidimensional proxy for smallholder farming SMEs.

Going forward, Bappenas as the custodian of the dashboard, is positioned to explore additional data sets from other ministries (especially with the Satu Data initiative now being implemented) to also ensure the livelihoods of smallholder farmers are safeguarded through long-term resilience planning. Potential data sets that may be integrated include land ownership and land usage data from the National Land Agency, as well as secondary data on the number of smallholder farmers from the Ministry of Agriculture, and agriculture-related statistics from Statistics Indonesia (BPS). Information regarding the size of land used for agriculture purposes can help the Government differentiate between smallholder farmers and large plantations, which in turn can improve the Government’s efficiency in evaluating and issuing land certificates for agriculture-related SMEs.

Our team is currently discussing how data regarding irrigation from the Ministry of Public Work and Housing Indonesia may also be combined with weather information to provide insights on suitable periods for planting and harvesting. The Ministry of Agriculture currently generates forecasts and crop suitabilities for upcoming seasons, however such information often loses context, by layering such data into this dashboard, it will be possible to better understand current crop production trends and their drivers. Another use case will be to develop alternative credit scoring schemes for farmers, for instance by overlaying nearest water bodies (as a proxy for access to water), nearest markets/roads (proxy for sale), onto the mapped plots, such data can facilitate credit provision to farmers who are in need of capital.

Mapped farm-level data in combination with additional remotely sensed biophysical data layers of climate and soil, can help identify groups (“homologs”) of farmers, with similar household characteristics and biophysical conditions, but with differing yields, meaning some with high yields (referred to as positive deviants), and some with low yields, but belonging to the same group (or homolog). This type of insight can help build farmer to farmer learning alliances, connecting and transferring knowledge between farmers with high yields, with those with lower yields, within the same homolog, thereby using data-derived insights for innovative and tailored agriculture extension programs. We are also interested in harnessing alternative sources of data such as from satellite image and social media to see how these data sets can be fused with traditional agriculture data to inform broader policies related to national economic growth, food security and sustainable living.