The popularity of digital wallets has increased in many markets, but cash is still the main form of payment in most countries. Nicolas Crouzet, Apoorv Gupta and Filippo Mezzanotti investigate the adoption of electronic payments by Indian retailers after 2016, when the country replaced its two largest denominations of currency with new bills. They point to coordination failures as a likely obstacle to the diffusion of fintech payment systems.
The continuous pace of improvement of financial technologies over the past decade has raised expectations about their potential to improve financial inclusion, particularly in developing countries, where fostering access to financial services remains a key goal for policymakers. One canonical example is digital wallets. These services, generally cheap (or free) and easy to adopt, allow consumers to deposit money on a digital platform which is then used to make peer-to-peer payments or shop at retailers. Retailers can also adopt these services to receive payments. Examples in developed economies include Zelle, Quickpay, or Revolut. In the developing world, PayTM in India and AliPay in China have led the way.
While their use has increased substantially in many markets over the last decade, these products are still far from dominant, and cash still represents the main payment form in most countries. The relatively slow diffusion of these technologies (like many other productive technologies) has puzzled both economists and policymakers. Why is their adoption so slow?
In our study, we provide evidence of the importance of coordination failures in technology adoption decisions. Coordination failures arise when decisions to adopt new technology are complements across users — that is, when the private value of adoption for each single user depends positively on the number of other users that use the technology. In these situations, expectations of low adoption can become self-fulfilling. Financial technologies are network goods, making decisions to adopt them act as complements across users, and that creates scope for coordination failures. While the concept of coordination failure is well understood in economics, we know less about how important this mechanism can be at slowing down adoption, particularly in the financial technology area. Also unclear is the potential role that policy can play in addressing this issue.
To better identify the role of coordination failures, we study the adoption of an electronic wallet technology by retailers after the 2016 Indian demonetisation. On 8 November 2016, the Indian government announced that it would void the two largest denominations of currency in circulation and replace them with new bills. At the time of the announcement, the voided bills accounted for 86 per cent of the total cash in circulation. The public was not given advance warning and the bills were voided effective immediately. A two-month deadline was announced for exchanging the old bills for new currency. To do so, old bills had to be deposited in a bank. However, withdrawal limits, combined with frictions in the creation and distribution of the new bills, meant that immediate cash withdrawal was constrained. As a result, bank deposits spiked but cash in circulation fell. Cash transactions became harder to conduct, but funds remained available for use in electronic payments. Importantly, though the shock was very large, it was also temporary, as cash availability had normalised shortly after January 2017.
We use this unique episode in the history of monetary economics to understand how agents adopt financial technology. The technology we consider was the largest provider of non-debit card electronic payments in India at the time of the events. The provider offers a digital wallet consisting of a mobile app that allows customers to pay at stores using funds deposited in their bank accounts. Payment is then transferred to retailers’ bank accounts via the app. The pecuniary costs associated with the adoption of this technology for retailers are small. in fact, there are no usage fees, and all that is required to join the platform is to have a bank account and a mobile phone, both of which were common in India by 2016. In our study, we focus specifically on the adoption decision of retail businesses.
We find that the demonetisation caused a persistent increase in the size of the platform, that is, the total number of merchants using it. Most interestingly, it caused a persistent increase in the adoption rate, that is, the number of new merchants joining the platform. Taken together, our results suggest that coordination problems could be an important obstacle to the diffusion of fintech payment systems and that temporary interventions can be sufficient to overcome these coordination problems. However, we highlight an important caveat with the use of temporary intervention: they can also exacerbate initial differences in adoption across markets.
Sudden spike in electronic payments
The demonetisation led to a large aggregate increase in the use of electronic payments. Within a week, the number of transactions nationwide for our company jumped 150 per cent, then roughly doubled each week over the next three weeks. Additionally, the usage didn’t drop as soon as the cash problem was fixed a few months later. While the growth slowed substantially, people kept using the app long after the cash crunch was normalised.
The aggregate evidence suggests that the temporary contraction in cash led to a persistent increase in the adoption of fintech payments. However, this finding alone does not necessarily establish that complementarities played a role in the process. To further investigate this aspect, we study a dynamic technology adoption model in which firms face a choice between two payment technologies (cash and the electronic wallet), and where the electronic wallet features positive adoption externalities — the profits from operating under this technology increase with its rate of use by other firms. The model confirms that network effects — when more users signed up, others were more likely to sign up as well — played a key role. We find that in a hypothetical scenario where network effects were absent in the technology, the adoption rate would have been about 45 per cent lower after the demonetisation.
The model predicts that with complementarities, a large but temporary shock, on top of durably increasing the overall number of users, also increases its growth rate in a persistent way. In other words, the number of new firms joining the platform every period remains high even after the shock has dissipated. The reason is that, with complementarities, the initial adoption triggered by the shock, by temporarily expanding the platform, increases the relative future value of adoption for other firms. This “snowball” effect can generate persistence in the increase in adoption rates.
We then test whether this prediction can be confirmed in our data. To do this, we provide an empirical design to estimate the impact of the cash contraction on adoption across districts. Specifically, we exploit variation across districts in the importance of “chest banks” to identify variation in exposure to the shock. In the Indian system, currency chests are branches of commercial banks that are entrusted by the central bank with cash-management tasks in the district. These branches receive new currency from the central bank and oversee its distribution locally. We find that districts where chest banks account for a larger share of the local banking market experienced a smaller cash crunch during the months of November and December 2016. We then show that the districts that were more exposed to the cash crunch also experienced a larger and more persistent increase in total adoption following the demonetisation. Crucially, higher exposure also predicts a larger increase in the number of new firms joining the platform, even after restrictions on cash withdrawals are lifted.
We present one caveat, though. Our model suggests that when the shock is large but temporary, adoption responses exhibit state-dependence. With this term, we refer to a process in which adoption is not uniform but instead crucially depends on the pre-shock adoption rate. The intuition for this result is simple: all other things equal, higher pre-shock adoption rates increase the strength of adoption externalities, making it easier to reach the tipping point beyond which the platform has sufficient critical mass to continue growing even after the initial shock dissipates. Our empirical evidence seems to support this implication. For instance, we show that districts located closer to ”payment hubs” — cities where the penetration of the technology was already high before the demonetisation — displayed a statistically and economically stronger response to the shock, both in the short- and long-run. Thus, the demonetisation also exacerbated initial differences in adoption across regions.
Our study provides important evidence on the importance of coordination frictions, and the role policy can play in removing those frictions, in the adoption of technologies that feature network externalities — a feature becoming increasingly common in the new economy. We highlight how large but temporary interventions can have permanent effects on adoption because they effectively act as devices that help firms overcome coordination frictions. However, in those environments, because of state-dependence, an intervention that is too brief can also exacerbate inequality in adoption rates. Policymakers may therefore face a trade-off between the duration of the intervention, and how much it will exacerbate the initial difference in adoption rates across markets or regions.
- This blog post is based on Shocks and Technology Adoption: Evidence from Electronic Payment Systems, in the Journal of Political Economy.
- The post represents the views of its authors, not the position of LSE Business Review or the London School of Economics.
- Featured image provided by Shutterstock
- When you leave a comment, you’re agreeing to our Comment Policy.