Newsroom

Patrick L. Anderson Wins National Award For Comparing Human Experts With Machine Learning

FOR IMMEDIATE RELEASE

East Lansing, Michigan—September 12, 2018— The National Association for Business Economics (NABE) has awarded Patrick L. Anderson, Principal and CEO of Anderson Economic Group, its annual Edward A. Mennis Award, which is given annually for outstanding writing in the field of business economics. Mr. Anderson will also be presenting the report at the Annual NABE conference in Boston on September 30th, 2018. Mr. Anderson’s paper, “Business Strategy and Firm Location Decisions: Testing Traditional and Modern Methods” will be published in the Business Economics journal.

On September 7th of last year, Amazon released its RFP to build a second corporate headquarters (HQ2). Shortly after that, the Anderson HQ2 Index was released. The Index was strikingly accurate in predicting which cities would be selected to move onto the next round in the site selection process for Amazon’s next headquarters. Following the release of the Index, Mr. Anderson decided to apply different methods to the prediction, and to compare the success of those methods.

“These findings are important for both traditional and high-tech firms. They demonstrate that true human thinking beats machine learnings when important decisions loom ahead,” said Patrick Anderson. “This award confirms that carefully assessing the underlying strengths of cities, including costs, talent, and the high-tech workforce, is the right way for businesses to make important location decision.”

In his report, Anderson examined this debated question of where Amazon will open its second headquarters. The paper applies three approaches to the Amazon decision: an income model; a value model; and machine learning models. The paper found that expert judgement from a business economist outperformed the machine learning models. Additionally, the model of a value-maximizing firm outperformed the income model using the same data, and captured valuable intuition that a traditional model would have missed. The paper questions whether machine learning or human input is more valuable when making an important business decision.

Metro Areas predicted by the AEG Methodology, and by Amazon’s shortlist include:
 
•    Atlanta
•    Boston
•    Chicago
•    Dallas
•    Denver
•    Indianapolis
•    Los Angeles
•    Miami
•    Montgomery County
•    Nashville
•    New York City
•    Newark
•    Northern Virginia
•    Philadelphia
•    Raleigh
•    Washington D.C.