As one studies through school and college, we might not appreciate in those settings the reach that statistics will have in important decision making across many different fields and interests. Over a career though, we need to make many important decisions that influence the course of events for an enterprise.
What is the value (not price) of any insurance coverage you might need to cover risk? If you ran a retail shipping company last year and anticipated greater holiday congestion due to 'frozenomics' weather, what decisions can you make before or after the holidays to mitigate the operations crush? What is the likelihood that a movie or book project that you are working on will be profitable? What strategy decisions can be made to make any project more accessible or profitable, or both? How does a new electric carmaker design batteries and value their respective warrant liabilities? How should energy drilling decisions change based upon civil unrest in eastern Europe?
These are the sorts of questions we need to consider each day, and they are inherently statistical problems in disguise. Our ability to better understand and be exposed to quantitative models, as a result, would help us address these questions.
Some recent broad examples include the US Food and Drug Administration walking back the idea that trans fats are safe ingredients in food consumption, but this could have been resolved by looking at better copula analysis, linking nutrition to heart disease, stroke and other risk factors. Or economists bidding up the price of technology stocks, as baby boomers feel social technology innovation is such a highly productive tool. Yet simple actuarial analysis, and time series analysis on long-term productivity changes would show that these tools are not allowing an organically slow growth and aging population to achieve the sorts of 4% plus growth that was enjoyed in the 1970s and earlier.
In my own work, I have often had to team up with leading non‐quantitative people in order to solve team problems. The better versed I am at statistics, the stronger my ability to influence better outcomes. And we just noted a sample of the many different types of problems, which require quantitative fluency to solve.
I got my start out of college, working in the management consulting industry. I travelled the US and saw businesses in action. I also saw the gap in quantitative understanding among leading professionals, across many of my client industries. Even as I had engineering, and business degrees, both arguably with some quantitative component, I had returned to graduate school to study statistics. An advanced theoretical degree at the top statistics programs offers a rich background to perform interesting applied work later in one’s career.
After graduate school, I worked on Wall Street for a decade, before running two important areas for the government. The first was the analytics group for the $700 billion TARP program. Then the policy, research and analysis department for the head of the Pension Benefit Guaranty Corporation. Today I advise heads of organisations, and lecture on statistics and analytics, at Georgetown and at Rutgers universities. Without a statistics and communications background, these opportunities would not be possible.
What I see now, over the past two decades of analytics work, is that the developed world is forging new pathways into the data and information revolution. More data for sure (often from clicks on anyone’s portable tablet), but how do we ensure that the statistical modelling of this new digital information is robust? After all, we have seen a number of recent cases where bad analysis has led to large secular issues. Note the incorrect modelling of the housing growth and the default probability of subprime loans.
This is why statistics is a top demand field. Since if there are not enough trained statisticians to perform the rigorous analytics needed, then less knowledgeable people will. And the more the latter happens, the more incorrect probability and statistics applications continue to become cemented with crude rules of thumbs that go against the progress we should expect. It takes a skilled quantitative mind as we enter this next century, to step beyond simple calculations and ask the broader cognitive questions about pattern recognition. And even knowing why statistical models change, and why data may be corrupt.
This is where there is a fine line with the new commercial drive for 'Big Data' initiatives. One must be careful to not expend enormous resources, yet again, on merely finding banal connections in limitless data, regardless of significance or underlying reality. One can spend enormous resources mining Twitter data to determine who would be great to solicit a complex financial product to, meanwhile being completely oblivious to the massive differences in macroeconomic constructs that could freeze up credit conditions globally. My recent article, 'More connected; not more productive', explores this disconnect between what’s occurring now and how one would should see things with a long lens on.
As we push forward into this new century, we will often leverage important probability and statistics ideas that were around many hundreds of years ago, at the origins of statistics. Many of them were zealously created to focus on themes of chance from the then gambling halls of Monte Carlo, and continue to be bedrock in the development of modern algorithms and valuation models on the floors of Wall Street or Canary Wharf. Or we have leading central banks process econometric data, to understand time variant relationships among data, but they still need to appreciate the student t-distribution developed in a much earlier time. And at the Guinness brewery.
So as I conclude, I urge you to seek out a statistics curriculum and continue to practice it as you develop in your career. You’ll find it quite enjoyable and be able to share your insights with the world. The new industrial revolution we are on may seem muted, but it offers the best growth opportunities for those desiring to process data and make smart decisions for their societies. Statistics will help you manage risks in your own personal life. Allowing one to think about the limited sample size and inherently uncertain outcomes that probability models can help quantify.
We live in a world that has been in recent years defined by increasingly tight margin pressures on budgets. Along with business and government executives requiring a workforce well adept in (over)analysing large data sets and using modern statistical models to efficiently deliver conclusions. The practice and communication of this quantitative topic will allow you to leave a strong impression on how society is progressing over the next hundred years. Make sure that you are an ambassador, not just loyal to one specific academic or professional community. Society needs people such as yourself to step up and educate them on the types of quantitative ideas that are important.
The views expressed in the Opinion section of StatsLife are solely those of the original authors and other contributors. These views and opinions do not necessarily represent those of The Royal Statistical Society.