Technology and Development

Even when technologies are appropriate and potentially profitable, adoption depends on the decisions of owners, managers, workers, and consumers. These decision-makers may misperceive returns, face risk, distrust providers, resist organisational change, or fail to learn from others. Since analogous issues have been extensively studied in agriculture (Suri et al. 2024), we focus here on manufacturing and services.

A first set of frictions concerns experimentation. Firms may not know the returns to a new product, input, or practice before trying it. If experimentation is costly or risky, adoption can remain low even when average returns are high. Recent evidence shows that small firms may underestimate returns, that temporary subsidies can induce experimentation, and that insurance or return options can raise adoption by limiting downside risk (Duflo et al. 2011, Bai et al. 2025b, Killeen 2025). Related work shows that simplified rules of thumb and targeted information can improve business practices and adoption decisions (Drexler et al. 2014, Hanna et al. 2014, Kremer et al. 2019).

A second set concerns trust and observability. Firms and consumers may distrust providers, forget to act on profitable opportunities, or avoid technologies that make behaviour more visible. These frictions are especially relevant for digital technologies. Reminders and trust-building can increase take-up of fintech products, while greater observability can reduce adoption when it strengthens monitoring by principals or owners (Houeix 2024, Gertler et al. 2025b).

Inside firms, adoption also depends on incentives. Workers may resist technologies that raise firm productivity but reduce their own earnings or rents. Managers may fail to sustain new practices when organisational routines are fragile or when key personnel leave. Evidence from manufacturing and management experiments shows that changing compensation contracts, improving management practices, and retaining managerial capacity can be central to whether adoption occurs and persists (Bloom et al. 2013, Atkin et al. 2017, Bloom et al. 2020, Adhvaryu et al. 2022, Bloom et al. 2025). Labour-market institutions can further shape whether upgrading survives adverse shocks (Farrokhi et al. 2024).

Learning across firms can either accelerate or limit diffusion. Networks among managers and entrepreneurs can spread information about practices, products, and technologies (Fafchamps and Quinn 2018, Cai and Szeidl 2018, Asiedu et al. 2023). Production networks can also transmit adoption, as in the diffusion of improved machinery across linked firms (Chaurey et al. 2025). Recent evidence highlights interactions among coworkers and across hierarchies as additional margins shaping diffusion inside organisations (Cai et al. 2024, Meki 2025). But learning may be weaker among direct competitors, where firms have less incentive to share useful information (Hardy and McCasland 2021, Cefalà et al. 2025). The broader point is that adoption is not only a production-function choice. It is mediated by information, incentives, trust, and the organisation of firms and networks.

Policy implications. These frictions imply a different role for policy than in the infrastructure or human-capital sections. The relevant instruments are often narrower: temporary subsidies for experimentation, insurance against downside risk, credit or matching grants, information provision, management support, contract redesign, or trust-building.

The case for these policies is strongest when the friction is local and identifiable. Firms may underexperiment when the upside is hard to observe and the downside is privately costly. Workers may block adoption when compensation contracts make them bear the losses. Users may avoid digital tools when adoption exposes them to monitoring, taxation, or tighter control. In these cases, broad subsidies may be less useful than policies that change the specific constraint.

Finance remains important, especially because it interacts with risk and learning. Credit constraints can prevent firms from trying technologies with high long-run returns, and this logic also appears in macro-development work on state-dependent industrial policy under financial frictions (Itskhoki and Moll 2019). But the general lesson is not that adoption should always be subsidised. It is that adoption failures often come from the decision environment of the adopter, not only from the average profitability of the technology.

For full reference list see the end of the conclusion chapter.