In this video we are going to show you some incredibly sophisticated corporate finance and corporate strategy analyses that is only possible if the corporate finance MBA is also equipped to understand coding.

Why are we posting this video?

In a recent post we discussed how MBA-styled financial modelling is already outdated for high-end corporate strategy work. Many readers wrote in saying it could not be true since the schools have not changed their curriculum or are just starting to merge finance, strategy and analytics: here and here.

First off, cutting edge thinking rarely originates in business schools alone. And business schools tend to be very slow to adapt when/if they eventually do. In these examples from the New York Times all you see are the schools putting MBA’s alongside technologists and hoping for something to happen by osmosis. Hope is not a survival strategy.

MBA programs are being incredibly shortsighted since they are merely looking to adopt the business models from the tech companies, like lean processes, modularization etc.

They should go much further and teach their students some core coding skills because as this video shows, some work can only be done by a very intelligent MBA in corporate finance who can also code.

Big Data is allowing corporate strategists to better advise and interact with clients. It more than just flashy apps, but a fundamental re-wiring of the way we do strategy. Learning some basic code will create an entirely more productive, creative and effective group of management consultants.

Remember, we have been here before: when MBA schools and consulting firms refused to adopt spreadsheet programing as a core skill. Learning to code is just another way to analyze problems and a better way than relying on creaky spreadsheets. As Theodor Levitt showed in people should avoid being entranced by a product, in this case spreadsheets, and focus on what the client wants, in this case better and more effective analyses with Big Data.

Since spreadsheets where the traditional means to the end, of analyzing clients, we should adapt to some form of deeper coding since that is a better means to the very same end.

Finally, this is not to say PhD’s in the STEM fields will displace management consultants anytime soon or ever – they will not. The same way MBAs needed to learn to work with spreadsheets, they now need to complement their skills with some very basic coding skills. This is not the end of the MBA by any measure. It is just a process of adapting slowly. What we will see is an MBA graduate who adds technical skills to her/his repertoire in the very same way financial modelling skills became essential in the 1990’s.

In the same way electrical engineers did not displace management when electricity became widespread, coders will not replace management when coding becomes widespread.

Subscribers to our executive programs will have detailed access to the tools and step-by-step methods to build these sophisticated analyses.

Enjoy this.

Michael

Training by ex-McK, BCG et al. Partners

Receive free exclusive episodes to advanced strategy and case interview training programs, plus a chapter from Bill Matassoni's Memoir. This is the ONLY way to sample FC Insider material.

Where else can you learn from ex-partners?

Sign up to receive exclusive FC Insider episodes. Start now:



Privacy Policy

Comments

11 responses to Big Data Changes Strategy & Finance

  1. Hi Jikku,

    Good questions. These require long responses and they are covered in extreme detail in the upcoming video updates to the Corporate Strategy & Transformation Study AND the Big Data Study so I will not repeat them here.

    Note, that scenario planning is NOT about creating a flexible strategic response. That is a myth and misunderstanding of scenario planning. An option is an option and I can think of no strategy option pursued by a company that is flexible. They all get rammed aside when things change. If scenario planning truly worked, as practitioners incorrectly assume it should, we would see a lot more companies who could respond well to changes.

    A simple sanity check will prove this out. How can a company develop a strategy that can work in all environments well? A strategy that works in all environments does not work in any one environment well and that is a bad strategy. A company is either perfectly organized to compete in a, for example, low price environment or a high price environment. It cannot magically switch overnight as things change.

    We provide the shortlist of options to consider. That is our job!

    We will teach the correct way to build flexible strategies as part of the Corporate Strategy & Transformation Study. It is a new addition to that study.

    Michael

  2. Dear Michael,

    1) How does using ‘big data to analyze and select from the 5000+ strategic options using the efficient frontier curve’ compare with ‘scenario planning’ where the company prepares a flexible strategic response for few key high impact risks/unknowns that may eventually play out as you execute strategy? For instance are both tools used for a different purpose where efficient frontier is used to choose a strategy while scenario planning is to prepare for unknowns as you implement the strategy ?

    2) Do executives struggle to decide on an option from the efficient frontier curve as there are 5000+ strategic options to choose from ? I understand intuitively that choosing an option on the frontier curve enables me to get the highest return for a certain level of risk. However the executives will not know what are the various variables and their relationships that the model used to create the frontier graph and I wonder whether they are confident in making a business judgement to choose an option just from the frontier curve ? In the past, they might have been provided with 4 strategic options but wasn’t it easy for them to analyze which factors are different amongst the options and more confidently make a business decision?

  3. Hi Florian,

    Yes, but they do a very basic version of it. It gets more complicated as one tries to mimic reality. I have not seen anyone solve the problems we are trying to solve.

    And yes, modelling the optimal position of a company is very different from modelling a portfolio of equity. The former is harder to accomplish – as you will see in the healthcare study.

    Michael

  4. Hi Michael,

    I have a bit a newbie question: do you think investment funds already do these analysis at company level, when they want to buy or sell stocks?
    For example, if an investment fund compared what a company plans to do to achieve result x vs what could be the optimal routes to result x, for an equivalent risk profile.
    Most investors obviously use efficient frontier models, although I assume they look at the volatility of each stock, not at what the company behind the stock could achieve.

    Cheers,
    Florian

  5. Hi Everyone,

    This video has sparked a lot of interest and we have received a lot of emails. We will post more about this over the year, and especially when we begin the next study where we will be deploying this technique at a client. Especially for the technical details which take a long time to type up in the comments section!

    In terms of the actual coding skills you need, I think the language is not important. Focus more on the skill you need. You need to be able to build financial models and databases on a cloud server and link that to an application based interface. We use Tableau in general, but anything will do.

    The main skill to learn is build data-based models which can sit in a cloud. The other skill is the type of co-variance analyses required in the model. A LOT of MBA’s have written in saying they know how to do this.

    I would say you know how to do this in a very basic way and what we have done here is far more complicated so take some time to think about it more. We are essentially working out the relationship between about 50 variables over a 10 year period while collecting data at hourly or daily intervals and working out the degree of volatility. So the other skill is to learn how to do this from both a theoretical viewpoint and from the basis of actually coding it.

    There are a lot of practical hurdles to doing this. For one, picking a base variable is pretty tough.

    Michael

  6. Hello Nandini,

    How are you?

    It is interesting you are mentioning healthcare. We are busy negotiating to redo a study done in the 1980’s by BCG but to adapt it using some of the principles of Big Data we describe above.

    I do believe that will be incredibly challenging and exciting since there is so much to do in healthcare. Healthcare is ripe for some creative Big Data thinking.

    And not just by building pretty apps, but fundamentally rewiring the way we analyze key issues. If the issue turns out to be improving the efficacy of treatment then we will definitely take it on since that is the key issue – knowing which treatments are useful and which are not, and by default which costs should not be incurred. Conventional analyses like NPV and real options does not work so well here so it is ripe for disruption.

    If we go ahead, we will obviously post everything on this website.

    You are right though, Big Data is much more clearly applicable to operations and operating functions. It is tougher to apply to corporate strategy since only 2 of the 3 levers apply.

    Michael

  7. Hi Michael, this is a great intro to using Data Science/Big Data in a management function. There are several studies out there (most notably this McKinsey one: http://www.mckinsey.com/insights/business_technology/big_data_the_next_frontier_for_innovation) that call out the skills gap in data analytics in management functions. But it is hard to visualize concrete applications of Big Data in functions such as corp strategy while it is easy to imagine examples in retail sales, O&G exploration, Healthcare, etc. I am really interested in seeing more such examples pertaining to typical corporate management functions. Thanks!

  8. Hi Yuanming,

    That is a great question.

    The reason we leave out the error bars or confidence intervals is because clients assume we will only use data where the statistical confidence is high.

    We do not want them to have to think about which options (dots on the exhibit) are statistically significant. That is why we are paid. When presenting any analyses, you want to keep it as simple as possible. Therefore, only the statistically acceptable data is used.

    In the training videos we go into a lot of detail on the calculations and treatment of variability in the models – and it is best not to type out the explanations here since they are depicted better on graphics.

    You are right though, all points on the x and y axes are the means of the distributions.

    It is quite exciting since we could not do this as easily just a few years ago. Imagine what we will be able to do in 5 to 10 years?

    Michael

  9. This is a really exciting demonstration! Michael, I have one question and wish you could help clarify it. Thanks!
    My question is “How to find out the potential error bar around the return and the risk?” I am interpreting these two metrics of each dot as mean values of a distribution, therefore the error bar could further give us a sense of the confidence in these data points.

  10. Hi Aylwin,

    Thanks for this. I still believe you need human interpretation of the data since business is not a pure science. A great degree of judgement is required which models are not even close to simulating.

    Michael

Leave a reply

You must be logged in to post a comment.