Columns that reveal the world
- Getting up close and personal with the researchers -

I think that most students come into contact with accounting for the first time when they enter university. Accounting is written as "calculate and match," so the image of "calculation" is strong, but if you ask whether accounting = calculation, that is not quite the case. For accounting, calculation is not the goal but a means. Accounting is "accounting" in English, and "account for ~" means "to explain to ~," so accounting means explaining the results of your business activities. To do this, you record your economic activities using a method called bookkeeping. In other words, calculation is only a means, and when I start my university classes, I often tell my students that it is important to learn how to use the numbers obtained by calculation, or how they are used in society. This is because I want them to know from the beginning that there is a much bigger world beyond just calculation.

When I first started studying accounting in university, I still had a strong image of it being just calculations, but my perception changed when I read in a book written by a professor who would later become my mentor in graduate school that "accounting is the language of business." I was shocked to learn that accounting is a language for communication, and that accounting is the study of communication. After encountering this phrase, accounting suddenly became much more interesting, and I also came to understand its importance as an academic field and its necessity in society.

When I first started studying accounting, the task of fitting the numbers together was fun, like a puzzle, but as I entered a seminar in my third year and began to study accounting more professionally, I realized that accounting is not difficult or rigid, but rather flexible and supple. I gradually came to understand that accounting is not just about recording transaction facts, but also contains the judgments and opinions of managers. For example, the numbers are filled with the thoughts of company managers, such as how much a company is spending on research and development, and how much resources are allocated to each business division. Recently, there are more accounting figures that include estimates, but they contain more judgments from company managers that take future predictions into account. Accounting is an academic field that contains both rigidity and flexibility, and plays an important role in society as a communication tool. Learning these things made me want to study accounting more professionally.

In my classes, I have previously taught under the theme that "accounting is one of the three sacred treasures of business, alongside English and IT," but since last year, the main theme has been "Thinking about accounting as data science in the VUCA era."

VUCA is an acronym for Volatility, Uncertainty, Complexity, and Ambiguity. It was originally a military term, but was taken up at the World Economic Forum (Davos Conference) in 2016 and has attracted attention as a keyword to describe socio-economic trends. Volatility means that things are not constant and change in various ways. Uncertainty means that changes in things cannot be predicted. Complexity means that there is not one cause for changes in things, but that multiple factors are intertwined. Especially in the current COVID-19 pandemic, the causal relationships of things are thought to have become more complex. The spread of COVID-19 was called a pandemic in medical terms, but at the same time, in a socio-economic context, it can be said that one of the reasons for this is that it has brought about an infodemic (a state in which a large amount of unreliable information is spread mainly on the Internet). Ambiguity means that the causal relationships and future of things become unclear due to the three factors mentioned above.

From a macro perspective of the socio-economy, when we consider climate change such as global warming and abnormal weather, former US President Trump's America First policy, the UK's withdrawal from the EU, and the current spread of COVID-19, we can see that we are living in a VUCA era. From a micro perspective of corporate management, this is even more evident, as the rapid development of IT and AI, as well as the impact of COVID-19, are causing major changes in the industrial structure itself. For example, while megatech companies such as GAFA (the initials of major IT companies "Google", "Amazon", "Facebook", and "Apple") in the US and BATH (the initials of major Chinese IT companies "Baidu", "Alibaba", "Tencent", and "Huawei") in China, have grown significantly, there are also industries that have suffered great damage, such as transportation, food service, and entertainment, and are being forced to respond to DX (digital transformation) and seek survival strategies in the COVID-19 pandemic.

Even over a longer span of time, the drastic changes in the competitive environment for companies are astonishing. Looking at the market capitalization rankings* of listed companies, in 1989, the year Japan changed from Showa to Heisei, seven of the top 10 companies in the world were Japanese. Thirty years later, in 2019, the era changed from Heisei to Reiwa, and of the top 10 companies in the world that year, seven were American and two were Chinese. Even Toyota Motor Corporation, the highest ranking, was ranked 43rd by a Japanese company.

So, what kind of skills are needed in this VUCA era? There are many different skills that are needed, but I would like to mention "data science" in particular. Here, I would like to think of data science as "a method to scientifically analyze data." For example, the airline industry is said to have seen its performance worsen due to the impact of the new coronavirus, but if you look at Japan Airlines' (JAL) accounting figures for this year, you can see that although the company's business performance is naturally in the red, if you look closely at the company's financial condition, you can see that they are trying to increase their capital to endure this period of turmoil, and that they are in a financial state that can withstand even if the deficit continues for some time to come. Before blindly fueling anxiety, you should first look carefully at the numbers and data in front of you and interpret them correctly. This is the basis of data science.

In the COVID-19 pandemic, accounting control, such as cash flow and inventory turnover, is even more necessary. I believe that the ability to objectively grasp the output data and interpret it in detail will become increasingly important. "Accounting is the most powerful business data." That is my message.

*[STARTUP DB, "The last market capitalization ranking of the Heisei era. What are the 30 years that created the gap between Japan and the world?", 2019, https://media.startup-db.com/research/marketcap-global]

As I said at the beginning, "accounting is a communication tool," so far, the accounting field has focused on numerical data such as sales and profits. Companies have set the management goal of increasing sales and profits, and investors have made investment decisions based on whether or not they have achieved that goal. However, in recent years, data other than numerical data has also become important. There are three main reasons for this. First, when evaluating a company's value, there has been a growing tendency to place emphasis not only on sales and profits as actual results, but also on intangible assets such as the brand power that the company possesses. Second, companies are now expected to contribute to the environment and society, not just profits, such as ESG (Environment, Social, Governance). And in the VUCA era, more detailed explanations of current performance and future plans are now required. These are the three reasons.

In light of this demand for communication that goes beyond traditional accounting figures, in recent years I have been researching corporate financial reports using a cutting-edge method known as text mining. Simply put, text mining is the extraction of "meaningful information" buried in large amounts of text data using a technology called natural language processing. "Mining" generally refers to the digging up of rare minerals from large amounts of earth and sand in mines, but text mining can be said to replace this with text data.

Using text mining techniques, it is easy to specify a word on Twitter and find a specific tweet. For example, if you specify the word "accounting," you can instantly retrieve 100 tweets that currently contain this word. This is the first stage of text mining, and the next step is natural language processing. Here, a technique called "morphological analysis" is used to divide the retrieved text into words, and frequently used words are extracted from this, making it possible to visualize the results. The size of the text changes depending on the number of tweets, so it is immediately clear what types of text are being tweeted the most.

Words in tweets containing "accounting"

Using these text mining tools, my co-researchers and I analyzed coronavirus disclosures in financial results reports for the fiscal year ending March 2020. In response to the serious impact of the global spread of COVID-19 on management activities and the dramatic increase in uncertainty about the future business environment, the Financial Services Agency and stock exchanges have requested that companies provide comprehensive disclosures about how COVID-19 will affect corporate performance, cash flow, and management strategies. In response, we thought we would use text mining to analyze what explanations companies provided in their financial reports.

We conducted an analysis using the financial results reports of 2,345 companies listed on the stock exchange for the fiscal year ending March 2020, and created a "co-occurrence network" from the text of the financial results reports. This involves analyzing word-to-word combinations to determine which words are frequently combined and how closely linked those words are. If we expand the connections between those words further, it becomes a network. By analyzing this, we can identify the words that are at the center of discussion and clarify the relationships with the words associated with them.

《Co-occurrence network using financial results summary for the fiscal year ending March 2020》

It is natural that corporate sales will decrease due to the new coronavirus, but it is clear that companies are devoting more explanation to whether the company is properly managing its finances to continue to survive and what its financial condition is. Investors can also see that "I see, this is what you should pay attention to when reading a company's financial statements." It is believed that visualizing the text will help facilitate communication between management and investors. Some people may think that since financial statements are written in Japanese, it is okay to just read the text without using programming, but the total number of words in the financial statements of 2,345 companies is 5.9 million, and the total number of characters is 11 million, so it is difficult for humans to read all of them. Using a programmed computer, it can be read instantly and visualized in an easy-to-understand way.

Text mining also reveals the tone that companies use to explain things. This is called "sentence analysis." For example, during the COVID-19 pandemic, negative words and words with high uncertainty, such as "decrease," "uncertainty," "difficulty," and "slowdown," stand out, while a small number of positive words, such as "strengthening," "improvement," and "stability," are also used by certain companies. This is my hypothesis, but I believe that such analysis can also reveal that companies that have been properly implementing BCPs (Business Continuity Plans) based on disaster prevention measures even before the spread of COVID-19 may be properly analyzing highly uncertain situations and responding positively to them.

Some may wonder whether such research can really be called accounting. However, accounting is a field of communication, and in recent years, the importance of non-numerical data has been increasing. Programming and other skills are also required to read language, which is unstructured data. In order to analyze data released by companies from a more multifaceted perspective, text mining techniques are likely to become increasingly effective in accounting in the future.

What I want to convey through my research is the joy of academic study and research. I believe that university is a place where you don't just learn about known problems and their solutions, but also discover unknown problems and develop the ability to confront them. What I value most in this regard is "fun." If you don't find it fun, you won't be able to continue.

It is said that in the world of accounting, a major paradigm shift occurred in the 1960s, from normative research to empirical research. Roughly speaking, normative research is an approach that builds a theory that explains how accounting systems and standards "should be." Empirical research, also known as empirical research, is an approach that builds a theory (hypothesis) that explains how accounting systems and standards "are" and verifies it using real data. Since then, empirical research has become mainstream, and more than 50 years have passed since then. I believe that research using IT and AI may bring about a new paradigm shift in accounting, and I am currently working on this research.

So what will happen in the future? I predict that accounting will undergo at least another change. It will be a research field that takes into account human "characteristics" and "emotions." There is no doubt that IT and AI will be introduced into the accounting world more than ever in the future, but if the subjects who speak the language of accounting are human, the subjects who listen to it are also human. Human information processing is always influenced by emotions. Accounting is sometimes misunderstood as an impersonal discipline that only deals with numbers, but there are also unique research projects such as the aforementioned sentiment (tone) analysis of managers, and for example, analyzing the photos and signatures of managers in annual reports to analyze the relationship with the company's performance. In the future, we may be asked even more about how to read a person's personality and emotions from huge amounts of business data.

Accounting is the language of business, and is evolving into an academic field of business communication that includes not only numbers but also natural language. Furthermore, it is set to be further driven by IT and AI. I look forward to seeing this field become an even more exciting field in the future. (Published in June 2021)