TEC: A young conference for an emerging field
Amazon VP and chief economist for digital streaming and advertising Phil Leslie on economists’ role in industry.
TEC is a young conference because only recently has it become common for economists to take operational roles in industry.
“Economists have only been really working in industry for about the last 15 years,” says Amazon vice president and chief economist for digital streaming and advertising Phil Leslie, who will be a panelist in a TEC conference session titled “Outlook for Econ Careers in Tech”. “Companies like Yahoo, eBay, and Microsoft created research labs and would have a handful of economists along with other scientists. They focused on writing research papers for academic publications using these cool new sources of data.
“Then, about 10 years ago, Amazon decided they also wanted to start hiring economists. But they were going to take a different approach. The economists were going to be embedded throughout the businesses, and they were going to be focused on business problems.”
At the time, Leslie was an associate professor at the Stanford Graduate School of Business. “I was a skeptic,” he says. “I thought, ‘Wow, you're going to have a hard time recruiting academics to do that, because all the other companies are putting them in research roles.’ However, it turns out that that first set of academics — and here Pat Bajari deserves a ton of credit — were so successful at driving real value for Amazon that, quickly, word spread throughout the company that these economists were actually really helpful and valuable. As a result, demand for economists grew substantially.
“Now you can go to top economics departments, and the PhD students will tell you they're considering academic jobs and jobs in industry. Industry jobs weren't anything anybody even mentioned five or ten years ago. Amazon pushed the boundary on creating an entire new career path, and NABE [the National Association for Business Economics] has positioned themselves as the industry association that is really focused on what it means to do economics in industry.”
Cause or correlation?
Leslie believes that one of the things that economists bring to industry is a focus on causality. “When you say ‘causal modeling’ to any economist, they’ll say, ‘Yeah, that's what economists do,” Leslie says. “‘We do causal modeling.’”
“Something that's always been an important part of any economics training is understanding the distinction between correlation and causation,” he continues. “For example, the role that advertising plays in driving sales. You may see products that have a lot of advertising, and they also get a lot of sales. It doesn't necessarily mean that the advertising caused those sales. Sometimes, companies will by default put more advertising into their most successful products.
“So how would you go about figuring out what the causal effect of advertising is on sales? That's where you get things like experimental variation or holdouts or approaches around instrumental variables or other types of methodologies that economics have developed for getting at causal effects.”
Leslie points to an example from his own work at Amazon. An Amazon Prime membership comes with a host of benefits — fast, free shipping, free two-hour grocery delivery, free video, and free music, among others.
But, Leslie says, “we don't necessarily know which of those benefits are most important to our customers. Should we be providing more free shipping? Faster free shipping? Should we be providing more video content? Should we be improving the music benefit or reading benefit or gaming benefit? So we built a structural econometric demand model of the bundled services for Prime members that allows us to understand how different benefits create value for our customers.”
Theory meets data
One of the other distinguishing aspects of economics, Leslie says, is its emphasis on theoretical models. “It’s an important way that economists are different from other types of data science experts at Amazon,” he says.
Still, he adds, economics isn’t nearly as theoretical as it used to be. “Back in the 1970s and the 1980s, economics was mostly about theoretical and mathematical modeling,” he says. “There was not much data analysis. Today, that has flipped around. Econometrics and empirical research are the primary focus.
“A common criticism of the theory models was that they are built around assumptions that may or may not hold in the data. I think people started to realize that for a theory to be truly useful, we have to be able to test it and validate the underlying assumptions with data. Today, economists have made great strides in combining theory and data. It’s another reason that industry has begun to take a greater interest in economics.”
More information about Amazon’s presence at TEC 2020 is available here.