Information is power: ICT and agricultural productivity

One reason for low agricultural productivity in developing countries is that farmers lack sufficient information or guidance in the use of modern agricultural methods. This article examines the impact of an Information and Communication Technology initiative in Bangladesh, involving call centres that enable farmers to consult experts regarding agricultural practices. It finds that, post intervention, villages with access to phone service experienced a 50% reduction in agricultural inefficiency at the plot level.

Regardless of differences in geographical factors like land or climate, one critical reason why farming is less productive in poor countries is that farmers do not use modern technologies and still follow old traditional farming methods (Foster and Rosenzweig 2010). Evidence suggests that farmers often lack sufficient information or guidance in the usage of modern agricultural methods (Magruder 2018). Access to expert advice and training can help farmers learn about better farming methods (Anderson and Feder 2004), but these in-person services are costly and remain out of reach for many farmers (Fabregas et al. 2019). Does remote access to advice on modern agricultural practices improve agricultural production efficiency? In a recent study (Chakraborty, Negi, and Rao 2025), we explore how Information and Communication Technology (ICT) interventions – specifically the ones using mobile phone-based advisory to deliver and diffuse information – help address inefficiencies in rice production in Bangladesh.

Agriculture remains the backbone of Bangladesh's economy, employing over 40% of the population and contributing significantly to rural livelihoods (Asian Development Bank, 2023). Yet, the sector continues to struggle with low productivity, particularly in rice farming, despite favourable geographical conditions. The average yield of rice in Bangladesh is around 4.9 tonnes per hectare, which is significantly lower than other major rice-producing countries (Figure 1). This productivity gap is often attributed to poor access to information about modern agricultural practices, limited use of high-yield inputs, and suboptimal resource allocation (Alam and Kijima 2024).

Figure 1. Rice yield of the top 20 rice producers

Note: Figure plots the rice yield (bars) and total rice production (dashed line) for the top-20 rice producing countries.

Source: Based on data from FAOSTAT.

In rural Bangladesh, most farmers rely on traditional farming practices and informal sources of knowledge, such as neighbouring farmers or local traders, which may not always provide accurate or updated information. Additionally, in-person agricultural extension services aimed at disseminating farming knowledge have limited reach and are costly to operate, leaving vast swathes of farmers without timely access to expert guidance.

The promise of ICT-based extension services

To address these challenges, the Bangladesh government launched Krishi (Agricultural) Call Centres in 2014. These centres aimed to provide affordable, timely, and need-specific advice through mobile phones, enabling farmers to consult agricultural experts regarding crop management, pest control, fertiliser use, and other agricultural practices. This intervention sought to bridge the information gap and improve productivity, especially among smallholder farmers.

Our study leverages a nationally representative panel dataset from the Bangladesh Integrated Household Survey (BIHS), conducted in 2011-2012, 2015 and 2018-2019, to evaluate the impact of the Krishi Call Centres Intervention (KCCI). We leverage the variation in phone service availability across regions and over time, and the introduction of the call centre service in 2014 to measure its impact on agricultural inefficiency at the level of individual plots. Figure 2 shows the growth of phone service availability in the sampled villages over time.

Figure 2. Evolution of phone coverage in the villages surveyed over the years

Source: Bangladesh Integrated Household Survey (BIHS).

Significant reductions in agricultural inefficiency

Our main outcome variable is a measure of inefficiency in rice production. To measure this, we combine plot-level input usage and location data from the BIHS dataset and obtain the potential yield data from the Global Agro-Ecological Zones (GAEZ) dataset. The potential yield represents the maximum attainable yield for each possible combination of cultivation inputs, namely type of water supply and level of complementary inputs, given the geographical endowments of each GAEZ plot. First, we match each household's location with a GAEZ plot and calculate the potential yield corresponding to the input use. We then calculate the plot-level inefficiency measure as the percentage difference between the actual and potential yield, which varies with respect to both geographical conditions and input usage.

A key finding of our study is that the intervention led to a 50% reduction in plot-level inefficiency in villages with availability of a phone service. Thus, by enabling farmers to seek expert advice, the call centre service helped them make better use of inputs, adopt modern practices, and reduce inefficiencies. Furthermore, we find that the impact of the intervention on inefficiency is driven primarily by plots that use rainfed farming and not by the ones that use tractor(s). Figure 3 documents these findings.

Figure 3. Reduction in plot-level inefficiency post-intervention

Notes: (i) The outcome variables for both regressions are plot-level inefficiency in rice production. Regression 1 focuses on the overall impact of the intervention, whereas Regression 2 focuses on the heterogeneous impact of the intervention by input usage at the plot level. (ii) Coefficient (1) documents the point estimate of the overall impact from Regression 1 with a 90% confidence interval. Similarly, coefficients (2) and (3) present the point estimates of triple-difference coefficients from Regression 2 with 90% confidence intervals.

Technical notes: (i) Difference-in-differences (DiD) is used to compare the evolution of outcomes over time in two groups, where one was impacted by an event or policy – in this case, access to the call centre service – while the other was not. A 'triple difference' estimate captures the differences in average outcomes of plots using rainfed farming and tractor(s) from the average outcomes of plots using irrigation and not using tractors. (ii) A 90% confidence interval indicates that, if the experiment was repeated over and over with new samples, 90% of the time the calculated confidence interval would contain the true effect.

Mechanisms behind improved efficiency

We further explore the mechanisms driving the observed reductions in inefficiency. One prominent mechanism was the change in the quantity of input usage post-intervention. Farmers using rainfed water increased their application of fertilisers and pesticides. Additionally, there was evidence of increased labour hours dedicated to farming. This suggests that, before the intervention, rainfed plots had lower and possibly sub-optimal input usage, but it increased post intervention.

Figure 4. Changes in amount of input use among rainfed farmers post-intervention

Notes: (i) For each outcome variable, the figures show the point estimates of ‘triple-difference’ coefficients from the same regression with 90% confidence intervals. (ii) Panel (a) captures the impact of the intervention for plots that reported using rainfed farming. Similarly, panel (b) captures the impact of the intervention for plots that reported using tractors.

Conversely, farmers who already had access to high-level inputs, such as tractors, showed little to no change in input usage. This reinforces the idea that the intervention was particularly beneficial for farmers who practiced traditional farming methods and lacked prior access to expert agricultural advice. Moreover, we find evidence of increased interactions with extension services post-intervention. After the intervention, rainfed farmers in villages with mobile phone access reported a higher likelihood of in-person consultation with agricultural agents or visits from extension officers. This is an indication that call centres were successful in demonstrating the utility of in-person consultation from extension agents.

Role of geographic networks and spillover effects

Another important insight from the study is the role of geography-based social networks in amplifying the impact of intervention. We find that farmers geographically closer to households with phone access experienced positive spillover effects. This suggests that agricultural knowledge gets diffused within communities, benefiting even those without direct access to the service.

We also find that households located in geographically remote areas with fewer social ties, as determined by the number of households within a 5-kilometre radius, benefited the most from the intervention. This is noteworthy as traditional agricultural extension services often fail to reach such remote farmers. By reducing their reliance on traditional, often inaccessible information sources, the call centres bridged a critical information gap.

Conclusion and implications

Findings from this study demonstrate the potential of ICT in improving agricultural productivity. By providing farmers with timely and relevant information, the intervention increased the take-up of modern inputs by farmers and reduced production inefficiencies. As developing countries continue to grapple with low agricultural productivity, scaling up and optimising ICT-based advisory services are critical in achieving higher agricultural productivity.

Our study offers several important policy insights for Bangladesh in particular, and other developing countries in general, which face similar agricultural challenges. Expanding mobile-based advisory services can help deliver need-specific and timely critical information, particularly for smallholder and remote farmers. Improving mobile connectivity in rural areas can help expand the scope and effectiveness of such services by elevating agricultural productivity. The study finds that farmers using traditional, low-cost input benefited the most, suggesting that tailoring advisory services to these farmers could yield the highest productivity gains. Additionally, leveraging existing social networks to disseminate agricultural knowledge can significantly enhance the impact of these advisories. The study also highlights that geographically remote farmers, who often lack access to traditional extension services, gained the most from the call centre support. Thus, targeted outreach programmes for such farmers could help reduce inefficiencies and boost productivity. Sustaining farmers' engagement with ICT-based services through periodic training, follow-up calls, and personalised feedback can be essential in ensuring sustained adoption and use of modern inputs and practices.