We can look at the effects from only two recessions, 2001 and the Great Recession, because the CPS does include certain key variables prior to 1994. It also lacks a variable on bonus payments. In contrast to the NAWS, the CPS data does not record whether an immigrant is undocumented. Therefore, we focus on immigrants in general and form interaction terms between immigrant status and the recession dummies. Otherwise, we use as similar a set of demographic variables as possible. Table 4 presents the regression results for the ln wage and weekly hours in the three sectors. In none of these three sectors did either recession affect the wages of non-immigrants or of immigrants. Presumably, wages are sticky in these sectors, partially due to union and other contracts and minimum wage laws. The unemployment rate had a statistically significant effect only in the construction sector, and that positive effect is small, as in the agricultural sector. The 2001 recession did not affect the weekly hours in these sectors. The Great Recession reduced weekly hours by 1.3 hours in the hotel sector but not in the other sectors. Thus, for most employed workers in these three sectors, weekly hours remained constant during recessions. This result contrasts with that in the agricultural sector where weekly hours rose substantially during recessions.But they most often use traditional technologies, despite the existence of more modern alternatives. This puzzle has sparked extensive research to understand which barriers constrain technology adoption. Researchers have focused most particularly on credit and insurance markets failures and on information frictions . Focusing on the latter, decades of research show that knowledge indeed plays an important role for adoption . Recognizing this, governments have invested heavily in dispatching agricultural extension workers throughout rural areas to transmit information from scientists to selected farmers — with the hope that information will then flow to others via social networks. But these public-sector efforts are widely recognized as having had limited impact . In advanced economies, by contrast,berry pots private businesses have become the main source of information about technological innovations for potential clients. Farmers rely on commercial input dealers not only to purchase inputs but also to be informed about available technological options.
These dealers are in turn informed by the companies whose products they sell and by public agricultural agencies and research institutions . In developing countries, could there similarly be a greater role for private business-motivated entrepreneurs in the public agricultural extension system? The optimal boundary between government and private entities in the provision of public services is an extensively studied topic in economics . Empirical research has found that outsourcing public service delivery to private firms can improve provision in many sectors such as potable water, health, food distribution, and education . Unlike these sectors, developing country agriculture has yet to see a large role for private business in delivering public services. Agriculture thus offers an opportunity to study whether private input suppliers can internalize benefits from the spread of information and become agents in the delivery of public services. Our experiment provides information on new technologies to local input suppliers , rather than to selected farmers as in the business-as-usual approach. Specifically, we give agrodealers access to a new seed variety for their own learning. Taking this business-oriented approach to agricultural extension could influence farm-level adoption in different ways. On the one hand, input suppliers — unlike contact farmers — have economic incentives to spread information. They stand to reap profits when suggesting the use of new inputs to farmers. Repeated interactions with farmers over time may discipline them into providing high-quality recommendations benefiting their clients.1 On the other hand, agrodealers might purposefully under-provide quality . Or they may recommend products that maximize their own profits instead of customer welfare, as has been observed in numerous other sectors . These concerns about securing through agrodealers adoption of the right technologies by the right farmers are akin to concerns about the quality of public services provided by the private sector . Therefore, our analysis seeks to answer both whether informed agrodealers increase adoption and whether they do so for the farmers that stand to benefit the most from the innovation. We study these questions with an experiment where information services are provided to local seed dealers. Our sample consists of 72 blocks, spread across 10 districts in the Indian state of Odisha.
We consider the dissemination of Swarna-Sub1, a new and profitable flood-tolerant rice variety that reduces the damage caused by flash flooding.We partnered with the government extension service to support their conventional activities in 36 control blocks. This included providing seed minikits to the contact farmers on whom they rely to use the new technologies and spread information about them in their social networks. It also involved carrying out large-scale “cluster” demonstrations where many farmers grow the new variety on contiguous pieces of land, and organizing farmer field days to share results from demonstrations. These are all activities the government extension service would do with adequate resources, but we supported them as part of the experiment to make sure that they were carried out and that our control group reflects business-as-usual activities at their best. We provided the exact same quantity of seeds and the same information to input dealers in the 36 treatment blocks. We did not support any conventional extension activities as was done in the control blocks. These dealers are highly local, small-scale businesses, selling seeds and sometimes other inputs such as agro-chemicals. They were free to choose how to use the demonstration seeds. The key distinction between this treatment and the standard mode of agricultural extension is that information constraints are being relaxed on the supply side, rather than on the demand side for new technology. The treatment tries to leverage the natural incentives created by the marketplace for private businesses to transmit information to their clients. Turning to the results one year later, we find that the dealer-based approach to the diffusion of information increases adoption of Swarna-Sub1 — the variety being introduced — by over 56 percent, i.e. from 6.3 to 9.8 percent of farmers. We further find that the average farmer in the treatment blocks is cultivating 69 percent more land with the variety, and the cultivated area of adopters increases on the intensive margin by about 9 percent. Consistent with these results on farm-level adoption, we find that the treatment triggered a supply-side response.
The magnitude of this response is similar to the increase in adoption. By the 2018 season, two years after the new seeds had been introduced, dealers in treatment blocks were about 59 percent more likely to have Swarna-Sub1 in stock. There is also some evidence that informing agrodealers causes a change in local input production. Treatment blocks were producing 40-50 percent more Swarna-Sub1 seeds during the three years after the intervention. An immediate next question is whether farmers induced to adopt in the dealer treatment are those with relatively higher expected returns. Dealers may invest little in promoting a particular input to the right farmers if something else provides them higher profit margins. Dealers in our setting principally sell seeds that are produced by the state-run seed corporation — that fixes both wholesale and retail prices equally for all seed varieties. This aspect of our setting eliminates a differential price motivation for dealers promoting the new technology. We consider a different possibility where dealers might encourage adoption by farmers with the greatest potential benefits if they want to maintain a good reputation as providers of high-quality advice. To consider this, we test for treatment-effect heterogeneity according to past flood exposure — an important determinant of returns.We find that the treatment effects of dealer-based extension are largest for households that are expected to have the highest returns. We would not expect this type of heterogeneity if dealers were sharing information indiscriminately or without considering the possible benefits for farmers. This offers a first piece of evidence that dealers are concerned with their reputation when sharing information. We then explore whether dealers proactively promote the new variety and what might explain their behavior. For this, we sent “secret shoppers” to around 300 dealers to inquire about new rice varieties in the third season of the study — two years after the seeds were introduced. We find direct evidence that the treatment changes what dealers say to farmers coming into their shops. Dealers in treatment blocks are about 25 percent more likely to mention Swarna-Sub1 when listing the new varieties to consider. When specifically asked for a recommendation,hydroponic grow system we find that dealers are less likely to recommend older types of seeds and in some cases more likely to recommend trying SwarnaSub1. These findings suggest that relaxing information constraints on dealers leads them to share this information with farmers. As a final piece of evidence, we ran an experiment to test whether business incentives and reputational concerns motivate dealers to spread information. In partnership with a local NGO, we revisited dealers during the fourth season to further solicit recommendations. Each dealer was randomized into one of two treatments. In the first treatment, someone visited the dealer and asked which farmers, locations, and varieties would be best for a demonstration where farmers would cultivate a new variety and then the NGO would organize a meeting with other villagers to explain its attributes. Importantly, the name of the dealer giving the recommendation would be advertised during the meeting. We refer to this treatment as the reputation treatment because the dealer’s identity would be publicly displayed as part of the demonstration. In the second treatment, the program was presented as one where the NGO would not name the dealer and would collect the harvest after the demonstration and redistribute it as seeds to other farmers. This treatment reduces the salience of reputation for the dealer and lowers their business motivation by reducing the demand that would be created by the demonstration.
We find that priming dealers to consider their reputation and business motivation changes the advice they give. In particular, it causes them to suggest different locations, types of farmers, and seed varieties. Starting with location, when presented with a candidate list of villages for the program, dealers in the reputation treatment are more likely to suggest a village not on that list. Most of this is explained by treatment dealers being more likely to suggest their own villages. Dealers in the reputation treatment also spend more time thinking of which farmers to recommend. They are more likely to suggest neighbors or other people in their own village. Finally, the reputation treatment causes dealers to recommend less common seed varieties. Taken together, these findings suggest that business interests and reputational concerns play a role in the information sharing process for dealers.These results help interpret the findings in our main experiment. Concerns over market share and reputation could be factors causing dealers to promote an improved seed — even if it does not provide them with higher profit margins. In sum, we present the first evidence on how leveraging private agents on the supply side of input markets can help disseminate agricultural technology more effectively than the conventional methods used by the public sector. An important channel through which this happens is that dealers, when they gain information, spread it actively to farmers. Importantly, the gains from leveraging agrodealers are largest in areas where the technology delivers the highest benefits to farmers. Reputational concerns may be one of the factors that motivates dealers to match the technology to farmers with the highest expected returns. These results contribute to the literature on public-private partnerships by showing how using private-sector input suppliers to spread knowledge can improve the delivery of public services.Moreover, it does so at a lower cost compared to expanding the current practices used in the public sector. It is cheaper to visit a small handful of dealers compared to carrying out large-scale demonstrations and training. Our findings contribute to the literature on agricultural technology adoption in the developing world.Several studies consider different methods to improve the provision of extension services. For instance, cell-phone-based monitoring of extension workers can help solve agency problems and increase the number of farmers reached . Alternatively, researchers have considered efforts to better select, incentivize, and train the farmers that the extension service partners with . Finally, field days to share information from demonstrations is a commonly used technique . This literature has focused on ways of reaching more farmers via the traditional channels — either through direct outreach or by learning from the contact farmers that were trained by extension workers.