VoxEU Column

How information matters for adopting a new technology in Bangladesh

In many developing countries, agricultural productivity has remained very low due to sluggish diffusion and adoption of new efficient cultivation methods. This column describes the results of a field experiment in Bangladesh in which randomly picked farmers received training in a new technology. An increase in the number of farmers in a village who received training increased the adoption rate of the technology among ‘untreated’ farmers. The findings suggest that frequent information and training are the easiest ways to help disseminate a new technology and encourage its adoption.

In many developing countries, especially in South Asia and sub-Saharan Africa, agricultural productivity has remained very low due to sluggish adoption and diffusion of new efficient cultivation methods, which are critical for food security and economic growth. Most frictions impeding the adoption of new agricultural technologies are rooted in imperfect information – that is, they stem from farmers' uncertainty, costs of learning and limited knowledge of the new technologies. Although this has strong empirical support (Moser and Barrett 2006, Barrett et al. 2010, Conley and Udry 2010, Jack 2013, Barrett et al. 2018), we still know little about the role of information transmission and risk attitudes on technology adoption decisions.

In a recent paper (Islam et al. 2018), we study this question by investigating the determinants of the adoption of the System of Rice Intensification (SRI) technology by farmers in rural Bangladesh. In particular, we focus on the importance of peers and risk attitude as the main factors affecting the adoption of the SRI technology. 


We use the case of Bangladesh because improving agricultural productivity has been critical in facilitating poverty alleviation and food security. Rice is Bangladesh’s largest crop and the main staple food for the 180 million people of the country. Furthermore, rice cultivation accounts for 48% of total rural employment (Sayeed and Yunus 2018). It also provides two thirds of the caloric needs of the nation, along with half the protein consumed. Its contribution to agricultural GDP is about 70%, while its share of national income is one sixth. However, crop yields of rice in Bangladesh remain low because of limited adoption of new innovations by farmers. We focus on SRI, which is a climate-smart, agro-ecological methodology aimed at increasing the yield of rice by changing the management of plants, soil, water, and nutrients. Specifically, SRI involves early, careful transplanting of single seedlings with wider spacing, in fields that are not continuously flooded and have optimum water management, with actively aerated soil containing a higher proportion of organic matter. Proponents of SRI claim its use increases yield, saves water, reduces production costs, and increases income, and that its benefits have been observed in more than 40 countries.

Experimental design

In collaboration with BRAC, an international development organisation based in Bangladesh, our randomised controlled trial (RCT) was conducted over two years – 2014/15 (year 1) and 2015/16 (year 2) – in 120 villages across five districts in rural Bangladesh. A set of 60 villages were randomly allocated to one-year training (referred to as T1 villages) and treated farmers only received one-time training in year 1. For the other 60 villages (referred to as T2 villages), treated farmers received the same training twice (i.e. in both the first and second year). Within each village some farmers were treated, and some were not. This number randomly varies between villages. Therefore, the percentage of treated farmers in each village varies exogenously and is referred to as the exposure rate for each untreated individual i in village v. Figure 1 depicts the distribution of the exposure rate across farmers in the 60 T1 villages (dashed curve) and the 60 T2 villages (solid curve).

Figure 1 The distribution of exposure to the training treatment across farmers


We estimate the impact of the exposure rate on a dummy variable equal to 1 if the untreated farmer adopts the SRI technology, and 0 otherwise. We control for the exogenous characteristics of the untreated farmers, which include age, income, land size, household size, occupation and education. We also include a year fixed effect.


If we pool together all the 120 villages, our results indicate that there is a positive and significant impact of the exposure rate on the adoption rate of the SRI technology. In particular, an increase of 10% in treated farmers in a village increases the average adoption rate for an untreated farmer residing in the same village by 2.2%. If we split the 120 villages into T1 and T2 villages, we find that the effect of the exposure rate on the adoption rate of the SRI technology is no longer significant for the T1 villages, but is significant and has a stronger impact for the T2 villages. Indeed, an increase of 10% in T2 treated farmers in a village now increases the rate of adopting SRI technology for an untreated farmer residing in the same village by 4.2%.

Figure 2 visualises these results. It displays this distribution for the 120 villages (blue curve), the 60 T1 villages (red curve), and the 60 T2 villages (green curve). If we consider this distribution for the 120 villages, we see that in villages where the percentage of treated farmers is 40%, the (predicted) adoption rate for untreated farmers is 5%; when the exposure rate is equal to 80%, the (predicted) adoption rate is close to 22%. For T1 villages, these numbers are 6% and 10%, respectively, while for T2 villages we obtain 3% and 36%, respectively. The diffusion rates are thus highly non-linear with higher fraction of farmers adopting as additional percentage of farmers are receiving training.

Figure 2 How adoption of technology increases with exposure to trained farmers

We provide a simple theoretical model that helps interpret these results. Our model suggests that when untreated farmers receive more accurate and precise information about a new technology from their treated peers, they are more likely to adopt the new technology. In our model, the lower the variance of the noise of the quality of the technology, the more accurate the information transmitted to the untreated farmer, and the more likely the latter is to adopt the SRI technology. This implies that the T2treated farmers provide the untreated farmers with a more accurate and precise information on SRI technology than the T1treated farmers due to their more advanced training. 

The role of risk aversion in technology adoption

It is well known that risk aversion plays an important role in new technology adoption, especially in poor socioeconomic context such the one in which we conduct our field experiment. We elicit farmers’ risk aversion using a lab-in-the field experiment in which they complete a simple gamble-choice task. We then estimate a similar econometric model but introduce the risk aversion of the farmers and the cross effect of risk aversion and exposure rate. 

We find that risk-averse untreated farmers are less likely to adopt than risk-loving untreated farmers. We also find that when the fraction of treated farmers increases, more untreated farmers adopt the SRI technology; however, the more risk averse they are, the lower the impact of this is on the adoption rate of untreated farmers. Our results suggest that untreated farmers who are more risk averse are less influenced by their treated peers. A trained farmer's impact on his untrained peers increases if he himself adopts SRI technology. 

More generally, our results indicate that the crucial determinants of adoption and diffusion of new technology are the accuracy and reliability of information transmission about the quality of technology circulated among farmers and their degree of risk aversion. In terms of policy implications, we believe that when a new technology is as complex as the SRI, most farmers will be reluctant to adopt it. Our findings suggest that frequent information and training are the easiest ways to help disseminate a new technology and encourage its adoption.


Barrett, C B, M R Carter and C P Timmer (2010), “A Century-Long Perspective on Agricultural Development,” American Journal of Agricultural Economics 92: 447-468.

Barrett, C, A Islam, A Malek, D Pakrashi and U Ruthbah (2018), “The Effects of Exposure Intensity on Technology Adoption and Gains: Experimental Evidence from Bangladesh on the System of Rice Intensification,” unpublished manuscript, Monash University.

Conley, T G and C R Udry (2010), “Learning About a New Technology: Pineapple in Ghana,” American Economic Review 100: 35-69.

Islam, A, P Ushchev, Y Zenou and X Zhang (2018), “The Value of Information in Technology Adoption: Theory and Evidence from Bangladesh”, CEPR Discussion Paper 13419.

Jack, B K (2013), “Constraints on the Adoption of Agricultural Technologies in Developing Countries. Literature Review,” Agricultural Technology Adoption Initiative, J-PAL and CEGA.

Moser, C M and C B Barrett (2006), “The Complex Dynamics of Smallholder Technology Adoption: The Case of SRI in Madagascar,” Agricultural Economics 35: 373-388.

Sayeed, K A and M M Yunus (2018), “Rice Prices and Growth, and Poverty Reduction in Bangladesh,” background paper to Commodity Markets, Economic Growth and Development, the UNCTAD-FAO Commodities and Development Report 2017.

945 Reads