Data-Driven Science, Economics, Public Policy, and Business Decision Making

The availability of modern machine learning and text analytics tools enable researchers and business practitioners to raise their predictive analytical capabilities to a new level. The models shown below demonstrate how Wordstat and Minitab products work together to enhance the scientific method.

Theory development and hypothesis testing become much more effective when unstructured data (text) can be mined from journals and social media and then used as economic predictor variables. Model testing using machine learning becomes much more efficient when literally hundreds of predictor variables can be tested for significance at once in a single equation.

The examples shown below give insights as to how today’s powerful computational capabilities are enhancing the philosophical paradigms of the scientific pioneers like Charles Peirce and Karl Popper.

Case Study 1—Taking Economic and Business Forecasting to the Next Level

Predicting the DOW Jones Closing Prices

Traditionally, economists have relied upon financial indices such as the Fed Funds Rate or the Volatility Index  (VIX) for predicting important economic time series such as the DOW closing price. Wall street analysts have long known that sentiment drives the DOW, but have had difficulty obtaining the data in a cost-effective way.

In this example we extracted a representative sample of daily news titles between 8/29/2022 and 3/12/2023 and performed topic modeling using Wordstat. This produced daily time series on 293 candidate topics. We then ran the daily frequency counts of those topics as predictor variables for the Daily DOW closing price using Minitab’s Treenet regression modeling tool. This analysis reduced the problem to a small number of candidate predictors. Then after removing serial correlation by inserting the DOW price at lag 1 into the model, we were left with two significant financial predictors (the Fed Fund Rate lag 1 and the Volatility Index Lag 1). We are also left with one significant sentiment topic (“Died Suddenly” from the COVID Vaccine). Refer to the upper left panel in Figure 1 below. The model exhibits multi-collinearity between economic predictors; so correlation coefficients are provided to show the proper signs.

Analysis of the significant predictor variables is instructive in how predictive modeling can change paradigms and move science forward. Economists have long debated the competing theories of the net effects of changes in the federal funds rate on stock markets. Keynesians predict a positive relationship between changes in the federal funds rate and stock prices. Whereas, real activity theorists predict a negative relationship. Our analysis shows a positive coefficient and hence supports the Keynesian perspective for the period of analysis. Refer to the paper: Kunaey Garg –  The Effect of Changes in the Federal Funds Rate

The emergence of the text variable “Died Suddenly from the COVID Vaccine” as a significant positive predictor of the DOW was surprising, yet upon reflection makes sense. It is an instructive illustration of how predictive modeling can quantify hidden truths that can lead eventually to changing paradigms and move science and mindsets forward. Our research into this variable reveals that the concept of “Died Suddenly” largely began in early 2021 by the early warnings from Robert Kennedy Jr and groups of physicians such as Sherri Tenpenny, Lee Merritt, Vladimir Zelenko and others. Example: Brian Shilhavy Editor, Health Impact News  Also in early 2021 Dr. Jane Ruby courageously began using her daily internet show and social media platform as a vehicle for whistleblowers, physicians and medical researchers to warn the public of the potentially fatal effects of the COVID vaccines. Example: Demaryius Thomas, ex-Broncos star wide receiver, dead at 33.  Eventually, Wall Street researchers such as Edward Dowd began quantifying and broadcasting the number of excess deaths and disabilities to the U.S. Workforce resulting from the COVID vaccine. Example: Tyler Durden – Vaccination Analysis. This work was followed by the release of the documentary “Died Suddenly” which has been viewed by millions alerting them to the dangers.

Our interpretation of the positive correlation between the DOW and the number of daily mentions in news titles for this topic suggests that the market sees substantial economic value from the revelations to the public of the dangers of the COVID vaccine.

Figure 2 illustrates the prediction model in Minitab that calculates the DOW based upon the above four variables.