Unraveling Financial Behaviors (1): Analyzing the Sensitivity of Migrant Worker Users to Foreign Exchange Rates

SentBe has undertaken a meticulous effort to illuminate the financial behaviors of our migrant users, a demographic often overlooked within the financial landscape. Despite a significant lack of comprehensive studies on the distinctive behaviors of migrant workers, SentBe has diligently amassed granular data, focusing on their remittance activities through our platform. In pursuit of this objective, SentBe has engaged in collaborations with esteemed researchers from globally recognized institutions, such as the Korea Advanced Institute of Science and Technology (KAIST), National University of Singapore, and the World Bank, by facilitating the sharing of transaction data.

This study delves deep into the financial behaviors of financially excluded individuals—our migrant users—employing transaction data to unveil unique characteristics. A noteworthy revelation surfaces concerning the optimal nature of their remittance transactions in relation to exchange rates. This exploration not only enriches our comprehension of the financial landscape but also thrusts into the spotlight the financial dynamics of a segment that has historically been underrepresented.

Are migrant worker users sensitive to foreign exchange rates when sending money home? The unequivocal answer is affirmative. After thorough analysis, it has been established that workers within our sample exhibit a discerning sensitivity to foreign exchange rates when remitting funds home. To be precise, users sent 0.02% more in amount relative to exchange rates of 5 days before and after their actual transaction date.

Before delving into the data analysis, a notable observation is that transactions are concentrated during the week of the 10th day each month (Figure 1), coinciding with their monthly salary payment day. Our marketing team has strategically aligned promotions around this period. An assumption underlying this is that migrant users tend to send money back home promptly after receiving their monthly salary, irrespective of currency rate fluctuations. This aligns with insights from user focus group interviews, where it was revealed that these users often constitute the primary income source for their families in their home countries. There exists a robust trust-based mutual commitment, with families anticipating regular remittances.

Figure 1. Transaction Counts Across Days of the Month (2016 to 2024)

Source: SentBe

However, this nuanced sensitivity doesn't imply an outright indifference to foreign exchange rates. The study introduces the Optimality Score (OS) to assess the optimality of the exchange rates applied to users' actual transactions. The OS utilizes daily exchange rates from five days before and after ([-5, +5]) the date when remittance transactions were requested and processed. This methodology allows for the calculation of counterfactual amounts, revealing what the cost would have been if users had conducted transactions on other days within the 10-day period. This analysis provides insights into whether transactions were executed at the optimal exchange rate within the specified 10-day horizon.

Figure 2 serves as an illustration of the OS graph, plotting the daily exchange rate in the window of [-5, +5] days relative to the rate on August 28th, 2018, which is the date when a Vietnamese user practically sent money to Vietnam via SentBe. The amount of the original remittance in the receiving currency, Vietnamese dong, is normalized, with the amount set to 1. The black line depicts the relative amounts reflecting the exchange rates of the [−5, +5] days concerning the actual transferred amount. The average relative amount stands at 0.9917, translating to an OS of 0.0083 — the difference between the remitted amount and the average relative amount. Consequently, we can deduce that this user, who sent 10,000,000 Vietnamese dong, approximately equal to 407 US dollars, would have sent 9,917,000 Vietnamese dong, about 403 US dollars if the user had sent the money 5 days earlier or later.

Figure 2. Example of the Optimality Score

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

The OS can result in a negative value. Another Vietnamese user's average relative amount is presented as 1.0068, with the difference between this average value and the real transaction amount ending up as -0.0068. Assuming the user sent 10,000,000 Vietnamese dong on January 22nd, 2019, 10,068,000 Vietnamese dong, about 410 US dollars, would have been sent during the 10-day timeframe.

Figure 3. Example of the Optimality Score

Source: Ibid.

The overall findings demonstrate that individuals in our sample consistently opt for the highest (most advantageous) exchange rate within the [−5, +5] day window surrounding the actual transaction, with an average optimality score of 0.020 (Table 1).

Table 1. Summary Statistics of Optimality Scores

Source: Ibid.

This trend depicts distinct patterns between users who send money around salary payment dates (the 10th to 18th date of each month) and users who send money on other dates. Figure 4 illustrates that users sending money on non-salary dates send more money at more favorable exchange rates than users sending money on salary dates.

Figure 4. Optimality Scores Trend by Remittance Timing

Source: Preliminary draft "Optimizing the Use of Fintech for International Remittances by Migrant Workers (2021)."

The patterns of the OS appear to vary by users' nationality, in addition to whether transaction timing belongs to salary or non-salary dates. Figures 5 and 6 showcase notable contrasting behaviors among users from Bangladesh, Cambodia, and Malaysia. These nationalities' users seem to save more by capturing a more beneficial level of exchange rate during non-salary dates. The charts for Thailand, Vietnam, and Indonesia indicate that the OS during non-salary dates is more likely to be positive, while the OS during salary dates is generally negative. Interestingly, users from the Philippines and India appear to send money home with the least loss, regardless of whether their remittance is done during salary or non-salary dates.

Figure 5. Optimality Scores Trends by Nationality of Users Sending Money during Salary Dates

Source: Ibid.

Figure 6. Optimality Scores Trends by Nationality of Users Sending Money during Non-Salary Dates

Source: Ibid.

Previously, SentBe's marketing activities primarily involved distributing promotion coupons on occasions like users' national or religious holidays. The findings of this study have led SentBe to adopt a new perspective in basic marketing, delving deeper into the overlooked financial behaviors. Unlike assuming users would send money right after receiving their monthly salary, SentBe has been sending SMS to users, informing them about exchange rates being better than the previous week. This approach is expected to prompt users to reconsider remittance timing and make informed decisions. Similar to how we subscribe to news updates when trading stocks online, there are potentially more strategies to be applied in digital remittance services for migrant workers.