Introduction

Financial analysis needs factual data and an explanation of the variability of these data. Data state facts but need more knowledge regarding how these facts materialised. Furthermore, understanding causality is crucial in studying decision-making processes.

The Financial Document Causality Detection Task aims to develop an ability to explain, from external sources, why a transformation occurs in the financial landscape as a preamble to generating accurate and meaningful financial narrative summaries. Its goal is to evaluate which events or chain of events can cause a financial object to be modified or an event to occur, regarding a given context.

Participants will need to provide the cause or effect within given segments. There are 2 subtasks: one in English and one in Spanish. The English dataset has been sourced from various 2017 UK financial annual reports from the corpus made available by UCREL in Lancaster University. The Spanish dataset has been extracted from a corpus of Spanish financial annual reports from 2014 to 2018. These datasets are comparable in both languages to allow for testing of multilingual models.

Traditionally, the extraction of cause-effect relationships has been extractive, as seen in the previous editions of FinCausal. In 2025, it is presented as a generative AI task, where questions about causes or effects are posed, and the system responses are evaluated using exact matching and similarity metrics.

Causality

A causal relationship involves stating a cause and its effect, meaning that, according to the text, two events are related, and one triggers the other. In summary, there are two types of causes:

1. Justification of a statement.
   For example: “This is my final report since I have been succeeded as President of the Commission as of January 24, 2019.”
2. The reason explaining a result.
   For example: “In Spain, revenue grew by 10.8% to 224.9 million euros, due to an increase in cement volume accompanied by a more moderate price increase.”

We are interested in both types of causes because they are useful for studying and understanding decision-making in a company. Additionally, in the second type, causes that lead to quantifiable effects are highly relevant for financial analysis.

Both types of causes can be agents or facts. Effects can be quantified or not, but they must not be expectations, hypotheses, or future projections.

Dataset

The dataset will consist of three parts: context, question, and answer:

• Context: The original paragraph from the annual reports.
• Question: It is formulated to find the other part of the relationship, either the cause or the effect. It will always be abstractive, meaning it should reflect the content of the cause or effect being asked about, but not exactly match the provided context. For example:
  • Why did X (effect) happen?
  • What is the consequence (effect) of X (cause)?
• Answer: The answer will be the cause or effect previously questioned, extracted verbatim from the text, making it extractive. If a complex relationship appears (such as a causal chain of three or more elements or a complex relationship that is not a causal chain), a maximum of two questions will be asked.

Here are some examples:

English Subtask

Context	Question	Answer
In October 2016, we announced an implementation agreement to sell ACR to two Shenzhen government sponsored investment companies, subject to regulatory and other approvals. This approval process remains ongoing and, as a result, we did not value ACR on an imminent sales basis as at 31 March 2017 .	Why was ACR not valued on an imminent sales basis as of March 31, 2017?	This approval process remains ongoing
The Board has resolved that, in view of the size of the Board, it is most appropriate for matters of remuneration to be dealt with by the Board as a whole.	What was the implication of the Board’s size?	it is most appropriate for matters of remuneration to be dealt with by the Board as a whole

Table 1: Sample for English

Spanish Subtask

Context	Question	Answer
Es cierto que se ha registrado una desaceleración de la actividad en la segunda parte de 2015 respecto a lo observado en el primer semestre debido a la ralentización de la demanda mundial, el agotamiento de algunos factores cíclicos y el aumento de la incertidumbre, en parte ligada al ciclo electoral español, pero las previsiones indican que se alcanzará un incremento del PIB de alrededor de un 2,7% para 2016.	¿A qué se atribuye que se haya registrado una desaceleración de la actividad en la segunda parte de 2015 respecto al primer semestre?	a la ralentización de la demanda mundial, el agotamiento de algunos factores cíclicos y el aumento de la incertidumbre, en parte ligada al ciclo electoral español
En este ejercicio, se han consolidado las ventas a empresas de grandes flotas de vehículos, por lo que se ha creado un departamento especializado en esta operativa y se ha iniciado el proyecto de internacionalización de renting con acuerdos con otros operadores para dar servicio a clientes que se expandan fuera del territorio.	¿Qué consecuencias ha tenido la consolidación de ventas a empresas de grandes flotas de vehículos en este ejercicio?	se ha creado un departamento especializado en esta operativa y se ha iniciado el proyecto de internacionalización de renting con acuerdos con otros operadores para dar servicio a clientes que se expandan fuera del territorio

Table 2: Sample for Spanish

For both subtasks, participants can use any method they see fit (regex, corpus linguistics, entity relationship models, deep learning methods) to provide the cause or the effect questioned.