FinTech for Social Good: A Research Agenda from NLP Perspective

Chung-Chi Chen, Hiroya Takamura, Hsin-Hsi ChenarXiv 2022 · DOI · arXiv

Introduction

In recent years, the Natural Language Processing (NLP) community has focused heavily on financial technology (FinTech), primarily through the lens of short-term market prediction and sentiment analysis. However, as financial markets become increasingly accessible to the general public, a new challenge emerges. This paper proposes a strategic shift in the research agenda: moving from purely profit-driven models toward FinTech for Social Good. By leveraging NLP to bridge the gap between financial inclusion and financial literacy, researchers can help protect vulnerable investors and promote sustainable, long-term impact investing.

The Problem: Inclusion Without Literacy

The "democratization of finance"—driven by mobile banking and low-barrier trading applications—has successfully achieved massive financial inclusion. However, this accessibility has significantly outpaced financial literacy. In Taiwan, for instance, while 80% of new brokerage accounts are opened by individuals in their twenties, nearly 40% of that demographic is classified as financially illiterate.

The "gamification" of trading apps often encourages high-frequency trading without providing the necessary tools for risk assessment. Traditional FinTech tools exacerbate this by focusing on short-term monetary returns, often ignoring the long-term social and environmental impacts (ESG) of investment decisions. There is a critical need for NLP tools that do more than predict price movements; we need systems that educate users and analyze the broader consequences of financial activity.

A Two-Pillar Framework for NLP in FinTech

The authors propose a framework that applies NLP techniques to two specific pillars of social good: Active Investor Education and Long-term Impact Investment.

1. Active Investor Education (Active Feedback)

Rather than providing passive manuals, NLP can facilitate "active" education by acting as a real-time sounding board for investors.

  • Scoring Investment Rationales: Instead of predicting if a stock will rise, models should evaluate the rationality of an investor's logic. By scoring user-provided rationales against established financial theories, NLP can identify logical fallacies or over-reliance on hype.
  • Risk Reminder Generation: Moving beyond binary "Buy/Sell" signals, models should generate natural language "risk reminders." These reminders outline specific upside and downside scenarios—such as how a specific supply chain constraint might invalidate a current market rating—helping investors think more critically about their positions.

2. Impact Investment (Long-term Analysis)

To promote sustainable development, NLP must move from time-series forecasting to complex narrative analysis.

  • ESG Identification: NLP can be used to extract and synthesize non-monetary risks and opportunities related to Environmental, Social, and Governance factors that are often buried in unstructured text.
  • Scenario Planning: The authors suggest moving toward "plausible scenario generation." For example, if a climate anomaly occurs, a model should be able to generate a narrative describing the specific impacts on an industry’s sustainable operations, allowing investors to plan for long-term volatility rather than short-term noise.

Key Insights and Conceptual Shifts

This research agenda introduces several paradigm shifts for the NLP community:

  • Rationality vs. Profitability: A profitable trade is not always a rational one. An investor might profit due to sheer luck despite a flawed premise. NLP should focus on correcting the underlying reasoning to foster sustainable investment habits and prevent future losses.
  • The Evaluation Paradox: Traditional NLP metrics like precision and recall are insufficient for risk reminders. If a model warns of a risk that does not eventually materialize, it is not necessarily "wrong." In a social good context, the model succeeded by highlighting a potential danger that the investor needed to consider.
  • Human-Centered AI: The objective is not to replace human decision-making with better trading bots, but to create "augmented intelligence." This technology should empower individuals, protecting them from the pitfalls of impulsive, game-like trading environments.

Limitations and Future Directions

Transitioning to this new agenda requires overcoming significant technical hurdles:

  • Dynamic Knowledge Integration: Financial markets move faster than standard model training cycles. A major challenge is injecting "yesterday's news" into static models like FinBERT without requiring a full retrain.
  • Standardizing Evaluation: There is currently no automated standard for evaluating the quality of generated financial scenarios or risk reminders. While human evaluation is the gold standard, it is difficult to scale.
  • Utilization vs. Replacement: There is a risk that convenient AI interfaces might further erode critical thinking. Future work must focus on designing interfaces that encourage users to learn and engage rather than simply following an AI-generated prompt.

By refocusing NLP research on these social objectives, the community can ensure that the democratization of finance leads to genuine empowerment rather than increased financial vulnerability.

Summary updated

Abstract

Making our research results positively impact on society and environment is one of the goals our community has been pursuing recently. Although financial technology (FinTech) is one of the popular application fields, we notice that there is no discussion on how NLP can help in FinTech for the social good. When mentioning FinTech for social good, people are talking about financial inclusion and green finance. However, the role of NLP in these directions only gets limited discussions. To fill this gap, this paper shares our idea of how we can use NLP in FinTech for social good. We hope readers can rethink the relationship between finance and NLP based on our sharing, and further join us in improving the financial literacy of individual investors and improving the supports for impact investment.
BibTeX
@misc{chen2022fintech,
  author = {Chung-Chi Chen and Hiroya Takamura and Hsin-Hsi Chen},
  title = {{FinTech for Social Good: A Research Agenda from NLP Perspective}},
  year = {2022},
  eprint = {2211.06431},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2211.06431},
}
Figure