Unraveling Financial Behaviors (2): Understanding the Dynamics behind Remittance Transaction Cancellations by Migrant Users

SentBe users can cancel their remittance request with a simple one-click process, free of charge, before the actual transfer amount is transferred from their bank accounts to SentBe. It was initially conceived for users’ convenience. Initially designed for user convenience—allowing adjustments to the sending amount or corrections to the recipient's bank account information—it has proven to be a distinctive feature not found in traditional banking services. Unlike conventional banks, which lack such flexibility, SentBe empowers users with this option.

However, a surprising revelation unfolded as a specific group of users began exploiting this cancellation feature strategically. Intrigued by the potential use of this feature as a "free look-back" option to optimize remittance timing, researchers from the study mentioned in the preceding blog post delved into the data at the user level to understand the patterns of this feature's usage.

Our transaction data affirmed the validity of this idea. Two critical fields in our dataset, namely Application Time (t1) and Application Time (t2), shed light on whether a user employed the cancellation feature. The former marks the time when a user initiates a transaction order, while the latter indicates the time when the user completes the order by either depositing the amount to SentBe’s account or canceling the order. This distinction allows us to classify transactions into Cancellation or Remittance. An example of this differentiation is illustrated in Table 1, depicting the actions of an Indonesian user.

Table 1. Example of Remittance and Cancellation by an Indonesian User

Source: Agarwal, S., Cho, S., Choi, H.-S., & Klapper, L. (2021). "Optimizing the Use of Fintech for International Remittances by Migrant Workers." KAIST College of Business Working Paper Series. Link

Moreover, researchers identified users who consecutively canceled multiple orders within a short timeframe. Table 2 illustrates an instance of sequential cancellations by a Filipino user. To delve into specifics, the user's initial order request was placed at 8:42 am on January 17th, 2017. However, this transaction was canceled at 2:36 pm on the same day. This action didn't conclude the user's interaction. Almost immediately, a new order was initiated at 2:37 pm, only to be annulled once again in less than an hour, precisely at 3:14 pm. Subsequently, the user repeated this cycle, canceling the order nine more times before ultimately executing a remittance transaction two days later. This detailed account provides a comprehensive view of the user's engagement with the cancellation feature, revealing a strategic and iterative approach to optimize remittance timing.

Table2.  Example of Sequential Cancellation by an Filipino User

Source: Ibid.

In both cases, users canceled transactions when the spot exchange rate at time t2 was advantageous compared to time t1. Let's delve into the specifics using Table 1, where spot rates are quoted in Indonesian rupiah to 1,000 Korean won. At time t1, the user initiated an order for 2,150,000 Korean won, equivalent to transferring 26,264,400 Indonesian rupiah. Fast forward to time t2, and 26,174,100 Indonesian rupiah would have been sent with the same order of 2,150,000 Korean won. As the spot rate at time t1, which was guaranteed by the order, was more favorable than at time t2, a remittance transaction based on the order at time t1 was processed.

Now, consider the scenario from Table 1 where the same user placed another order the next day, this time for 2,005,000 Korean won, only to cancel it a day later. Examining the shift in spot rates, the value of 1,000 Korean won increased from 12,178 Indonesian rupiah at time t1 to 12,199 Indonesian rupiah at time t2. Calculating with the order amount in Korean won, the user could have sent 24,458,995 Indonesian rupiah at time t2, whereas 24,416,890 Indonesian rupiah would have been sent at time t1. Consequently, the user opted to cancel the original order at t2, showcasing a keen awareness of favorable exchange rate dynamics and an adept use of the cancellation feature to maximize remittance value.

In Table 2, where the user consistently canceled orders 11 times, the spot rates exhibited a consistent upward trend from the first t1 to the 11th t1. At the outset, the order amount at time t1 would have been 42,212 Philippine pesos, and by the 11th order at time t2, it had risen to 42,717 Philippine pesos. Through these strategic cancellations, the user capitalized on the fluctuating spot rates. By canceling the 11th order and placing a new one, the user could send an additional 505 Philippine pesos. However, this meticulous approach continued until the 12th order at t2, where the user eventually executed the transfer. At this juncture, with SPOTt2 dropping to a level where 1,000 Korean won equaled 42.499 Philippine pesos, the user could only send 287 Philippine pesos more compared to the condition of the initial order. This exemplifies the user's adept navigation of spot rate dynamics to optimize the remittance amount.

Figure 1. Example of Sequential Cancellation

Source: Ibid.

Figure 1 illustrates the sequential cancellation pattern of the user detailed in Table 2. The x-axis represents time and date, while the y-axis denotes the spot rate of the Philippine peso per 1,000 Korean won. The black solid line traces the fluctuation in spot rates over time. The dotted red line marks the instances of order placement and cancellation. On this graph, the left end of each red line signifies an order, while the right end indicates the cancellation time. If an order remains uncanceled, the line appears in blue. Consequently, the 11-step staircase-like dotted red line represents the 11 instances of cancellation made by the user from the morning of January 17th to the evening of January 18th. Throughout this period, spot rates exhibit detailed fluctuations, but the overall trend is upward. The user's ultimate order, placed around 6 pm on January 18th, involved depositing the order amount to SentBe nearly a day later, allowing for processing with a more favorable spot rate established the day before.

Through these strategic actions, the user ultimately gained an amount proportionate to the difference between the spot rates at the initiation and conclusion of the cancellation sequence. Interestingly, a clear pattern emerged as the user showed a reduced inclination to use the cancellation feature during the day, possibly due to its alignment with working hours. On January 18th, despite the spot rates being notably higher than at other times, no actions were taken during this period. Although missing a more optimal spot rate, the user maximized the amount sent with the final order decision. It seems fortuitous for the user that the overall trend in spot rates decreased subsequently.

Table 3. Summary Statistics of Optimality Scores

Source: Ibid.

Of the total sampled users, 45% appeared to have used this cancellation feature at least once, and 20% had employed sequential cancellations similar to the user depicted in Table 2. The study also analyzed how much these groups were more optimal in selecting a beneficial timing in remittances using the Optimality Score (OS). Assessing the optimality of the exchange rates applied to actual transactions, the OS significantly increased in the group of users who had used the cancellation feature. Table 3 presents the OS of the cancellation group as 0.062, while it is 0.015 for those who have never used the cancellation option. In cases where users have used the cancellation function sequentially multiple times, the OS is shown at a much higher level, 0.162.

Figure 2. Average Optimality by the Usage of Cancellation

Source: Ibid.

Figure 3. Average Optimality by the Usage of Sequential Cancellation

Source: Ibid.

Depicting variations in the Optimality Score (OS) based on the application of both standard cancellation and sequential cancellation, Figure 2 illustrates that the cost-saving effects stemming from the cancellation features are notably pronounced. Specifically, a more substantial increase in OS is observed, primarily driven by transactions involving the sequential cancellation functionality.

This outcome aids in identifying that SentBe users, predominantly migrant workers from South and Southeast Asian countries, have strategically utilized the cancellation option as an arbitraging tool on SentBe’s digital remittance service platform, resulting in more optimal decision-making.

In addition, it is noteworthy that the cancellation feature was not initially designed to serve this purpose but emerged organically based on user experiences. SentBe, as a non-bank financial service provider, implemented this feature to accommodate the necessary period for users to transfer their requested amounts in the local currency to SentBe’s legally registered account. The study's results underscore the importance of staying attuned to user behaviors, as these insights might reveal latent needs and contribute to further service improvements.

The unexpected discovery prompted SentBe to reconsider the depth of its users', possibly many overlooked, financial behaviors and recognize potential areas for service enhancement. One notable implication arising from this study is that migrant workers might have a heightened need for sensitivity to exchange rates. This is particularly significant because the price levels in their home countries are lower than in Korea, making even a slight change in the sending amount have a more substantial impact on the receiving amount for their remittance recipients.

Regardless of the income or price level of their home countries, human beings are universally inclined to seek economic gains with specific efforts whenever possible. This observation aligns with a universal characteristic where individuals are keen on monitoring stock price fluctuations to make timely decisions, even during working hours, through stock exchange apps. The study suggests that such financial behaviors, often overlooked, can play a crucial role in shaping the preferences and decision-making processes of financially excluded individuals when utilizing digitalized financial services.