Skip to main content
Factors that affect businesses have now become complex. You can't just rely on experience or intuition to bring success to your business. Accurately predicting the future is becoming all the more essential.
Business Dynamics is a technology that helps you look at your business scientifically and anticipate the future accurately. What should you do to guide your business to success? We listened to three researchers at Hitachi's Yokohama Research Laboratory who are passionately devoted to the research of Business Dynamics. This technology helps companies to develop their business strategies.
NAGAOKAIt's the application of System Dynamics (SD) to business. SD was developed in the 1950s by Jay Wright Forrester, a professor at the Massachusetts Institute of Technology. It looks at the internal structure of a system's behavior quantitatively. To understand a system, SD first creates a model of its complex cause-and-effect relationships. It then expresses the model with mathematical formulas to assess it quantitatively.
TANAKAThere's a rising trend in recent years where companies seek to expand globally and grow more of their service businesses. But when they try to make specific plans, they run into the reality of "There are too many uncertain elements—we don't know how to design the business" and "We don't know where possibilities of failure lurk." What are the keys to business success? If we use Business Dynamics, we can visualize the answers to these questions.
NASUThere are existing technologies for visualizing a business, such as modeling its elements in a tree form. With these methods, however, you couldn't model trade-off relationships—for example, a relationship where prioritizing some elements leads to sacrificing others. The advantage of Business Dynamics is that you can include such a relationship in the model.
NAGAOKAWhat led me to begin this research in the first place was my involvement in another research effort at the time, and I wanted to turn it into a successful business.
Previously, I had been a member of the new business planning and the market research. So I'm very familiar with planning new businesses. As part of this work, I unfortunately have seen businesses fail. But I've also been able to keenly experience the joy of spreading our research into the world as new businesses.
These experiences led me to think, "When planning a business, I want to take risk into account and increase the business's likelihood of success. To do this, I wonder if I can research its measurability." So I began this research.
Figure 1: Example of uncertain elements involved in business
TANAKASystem Dynamics is made up of two stages, qualitative analysis and quantitative analysis.
First, for qualitative analysis, we create a causal loop diagram (CLD) to get an overall picture of the system we're analyzing. In a CLD, arrows—called "links"—show the cause-and-effect relationships between elements—called "nodes." Nodes are key elements that explain the overall image of a system. If you analyze a business's characteristics, like its sustainability and profitability, nodes are elements that affect the business.
I'll explain briefly what can be expressed by a CLD. Let's take for example an organization that has just been launched. When profit rises, the organization invests to build up its strength, for example by adding people and expanding its systems. As a result, the number of employees grows. As employees increase, that by itself has effect, so sales also rise. As sales rise, profits go up too. So, as you can see, there is a positive loop flowing through the elements "profit," "organizational strength," "number of employees," and "sales." But if you simply add employees without a plan, the cost price increases. When the cost price climbs, a negative loop occurs—profit drops. By arranging elements in a CLD, we can graphically represent a variety of phenomena that can occur in a system in terms of its elements' cause-and-effect relationships.
Next, based on the CLD we created, we carry out quantitative analysis to see how the actual numbers stack up. We represent a system's phenomena in a different model called the System Dynamics (SD) model, and depict the phenomena as equations. We then apply numbers to equations we derived to quantitatively evaluate the system.
Figure 2: System Dynamics, the basis for Business Dynamics
NAGAOKAEven if you understand the theory of System Dynamics, you can't simply try to actually apply it and model a business. Human subjectivity is a part of modeling. Because of this, completely different models can be produced by different people. So we've developed technologies to help make modeling more systematic. In other words, these methods complement modelers' skills and experience so that they can achieve results above a certain level.
NAGAOKAThere are three major technologies. The first is Business Structure Templates. This toolset allows you to understand a rough structure of your business. The second is Component Models. This toolset lets you reuse models that are considered to be effective based on past results. The third technology is Normative Models. This toolset lets you analyze your completed model by comparing it with patterns of structures that are prone to failure and patterns that anticipate growth. These patterns are extracted from analyzing past results. Through this pattern matching, you can evaluate your business.
Figure 3: Three modeling-support technologies
NAGAOKAAt first, we took an approach where previously created models were broken down into components so new models could be created by combining the components.
However, as we sought to model businesses using the component models we created, it became clear that we first had to define a "larger framework." In other words, while it was fine to push forward with creating a model using component models, there weren't standards for identifying whether the elements included in the model were sufficient. To determine whether elements are missing, we first needed a framework that lets us understand what kinds of elements we should investigate.
So we developed Business Structure Templates.
NASUWe assumed that indicators well-known in the business world should serve as their foundation. We then elaborated on the templates by applying a variety of case studies. Right now, the templates have several forms according to the field of the business being modeled.
NAGAOKAFigure 4 is an example of a template. It includes BSC (Balanced Score Card), 3C (Customer/Competitor/Company), and PEST (Political, Economical, Social, Technological) analyses.
The most important indicator when evaluating a business is usually earnings. So, BSC is incorporated into the template to examine relationships between earnings and business operations and earnings and customers. For 3C, factors affecting business are changes in customer behaviors and how competitors respond to your company's actions. From the standpoint of PEST, external factors that you can't control, such as regulations, affect your business as you globalize and shift towards the service industry. So these factors are incorporated into the template.
Figure 4: Example of a Business Structure Template, a modeling-support technology
TANAKAA business planner can break down an idea in his mind if he uses this Business Structure Template. He first extracts keywords from his mind, places them up in the framework we've prepared, and create a CLD.
Through this process, you can see areas where discussions are concentrated and areas where they are not. For example, when we lay out the keywords on the template and organize their cause-and-effect relationships, if there are few keywords lined up under "Competitors"—from 3C's standpoint of "Customers, Company, and Competitors"—then you know more discussion is needed in that area.
So the Business Structure Templates allow you to become concretely aware of your business.
NAGAOKAIt is that creating new models of businesses in areas we have not dealt with before.
We can make a CLD according to a template. However, the value indicators that we should apply differ completely depending on the industry. Because of this, we need a look of time to understand the causative relationships in an area of business. We need to interview experts, read books, and conduct reviews.
Now, Component Models, a modeling-support technology, should help us reduce the time we need for study and boost our efficiency when creating models. However, we don't have enough Component Models right now. It's a lot of work, but right now we are modeling a variety of businesses and increasing our stock of Component Models.
TANAKAMany people in new industries and fields have approached us, and we found there are still quite many fields we don't know much about. This has made me realize once again how broad Hitachi is.
NAGAOKAThat's right. There are times when I think, "Hitachi has such a business," and "We've just gotten new business patterns."
NASUFor quantitative analysis, I've been working on expressing business fluctuations accurately with mathematical equations. In other words, I've been working on SD models.
For example, I worked on modeling project management. I tried to model the cause-and-effect relationships between scored items like "progress" and "communication" and to mathematically express these relationships. These items are used by a system project manager to evaluate a project's status. Our model used actual scores as input and sought to predict changes in future conditions. In creating this model, it was difficult to determine how best to mathematically express the causal relationships.
For example, it's well-known that if a project doesn't have enough communication, workers won't sufficiently check things like a product's specification. So they have to redo their work, and the project becomes delayed. When expressing this phenomenon as a formula, we have to come up with a formula that captures the question "When the amount of communication changes, how does this affect progress?"
One further complication is that the scores themselves are given by people. So it is difficult to capture this perspective with a mathematical formula, even if the scoring can be measured. In this case, I made hypotheses about the characteristics of the project management field and about scoring trends common among project managers by analyzing past scoring data. I then expressed these hypotheses mathematically. To verify my hypotheses, I ran the past scores through my mathematical formulas. As you can see, it's very difficult to express the know-how in a person's head as math equations. It's very difficult to create explicit formulas of tacit business knowledge.
NASUThat's right. In the first place, there aren't enough data available. So unlike the case of project management, you can't derive formulas based on past data.
We need an approach where we mathematically express hypotheses we developed based on our deepened understanding of systems in new businesses and the know-how of experts in those fields. Above all, verifying of new businesses takes time. We can test whether the model express a past phenomena or not, by using past data. We can't, however, find out whether the results of a simulation of the future are really correct unless we look into the results of the business as it progresses year by year.
Still, even though examining the hypotheses about a business takes time, the advantages of modeling and simulating it are tremendous. We can confirm imaginable business risks with simulations. Based on the results, we can determine strategies and actions we should take at that point in time.
NASUWhen I try to perceive a business structurally and treat it as a social system in a computer to predict the future, I can feel the great gap between real-life business and model systems. Our research is to fill this gap. I feel this is a rewarding field as we try to ultimately build a bridge connecting theory and reality.
TANAKABefore I joined this research group, I supported business planners in organizing their ideas. In this work, I used causal loop diagrams. So I'm interested in how we can use CLDs to organize ideas.
I'm drawn by two major questions. The first is, "What kinds of circumstances make it easy for people to understanding something?" Because a variety of people are involved in business, they have their own causal loop diagrams sketched out in their minds. There are times when they have to combine these diagrams to share their awareness together. At that time, there are people who think causal loop diagrams are chaotic and difficult to understand. So that people can share their awareness smoothly, what we need is a notation system for CLD that is easy to understand by everyone. Coming up with such a system is one of my interests.
The second question is, "How do people actually behave?" People often act irrationally and make irrational decisions that do not match theory and equations. I want to include this irrationality into models. I hope to create models that incorporate fields like behavioral economics and psychology.
NAGAOKANo matter how much we appeal to business workers about the importance of the technologies we're researching, there are still many people who think, "The results are the same no matter if we use them or not."
Now, a phrase I like is "Theory without practice is empty; practice without theory is blind." I have seen that some projects based on experience and intuition had turned out to be disappointing and wasteful. I hope we can spread a culture to the everyday workplace where workers will review the overall shape of a business at appropriate timing—for example, when the business environment changes—and make flexible corrections.
(Publication: November 18, 2013)