Phil Bak is the CEO of ACSI Funds & Exponential ETFs. He is widely regarded as an expert in the management, development and trading of ETFs. Prior to joining ACSI, Bak was a managing director at the New York Stock Exchange, where he initiated market structure enhancements and worked with asset managers, regulators and liquidity providers to ensure a fair and orderly secondary market for ETFs. ETF.com caught up with Bak to talk about his company’s new ETFs based on customer satisfaction and brand value, and how those two ideas can be investable.
ETF.com: Last November, your firm launched its first ETF, the American Customer Satisfaction Core Alpha ETF (ACSI). How does it select its underlying stocks?
Phil Bak: The American Customer Satisfaction Index is our sister company. It’s a private company, and it creates the data.
It has a proprietary economic model it uses to translate the data, and then we have an exclusive license on the data to create the investable index, what we call the ACSII, which is the American Customer Satisfaction Investable Index.
We use the index data to create a more traditional portfolio, manage things like sector constraints and some basic risk levels, and use that investable index for the ETF.
ETF.com: The last presidential election showed us just how off polls can be. Who are you surveying, and why do you feel this is reliable data?
Bak: One of the things we saw around the election was that polling is not always 100% accurate. But what was even more glaring is that the difference between polls can be pretty dramatic: different methodologies are used, as well as how those polls manage their sample and their demographics, and how questions are phrased.
Polls are not the same, and an interview or a survey is not the same as another interview or survey from somebody else. It’s crucial that people understand the quality of the data they’re getting is rigorous and has a process in place that makes sense.
The ACSI data was created at the University of Michigan by researchers. We have several Ph.D.s that work on it today. There is a proprietary economic model involved, and we normalize for over seven demographics.
What really makes this data special is we’re not polling people for their opinions on goods and services and companies that they have not directly interacted with. We’re drilling down to find the exact specific customers of companies. We survey people who have actually bought a certain car.