Like other central banks around the world, the Fed is grappling with the disruption that financial technology — or fintech — poses on the banking system. Patrick Harker, a former Wharton dean who is president of the Federal Reserve Bank of Philadelphia, shared his views on the potential of fintech, the missing factor in cryptocurrencies — and why going to a four-year college might not be the path for everyone. He was interviewed by Wharton finance professor Richard J. Herring in a fireside chat at the “Fintech and the New Financial Landscape” conference, which the Philadelphia Fed recently hosted. (Listen to the podcast above. You can also watch a video of interview using the player below.)
An edited transcript of the conversation follows.
Richard J. Herring: How are the innovations around fintech changing and challenging the way the Fed has traditionally done business? The Fed has all of the problems that the incumbent financial institutions have. It has legacy systems all over the place, plus I think it has even more trouble changing them because it has the obligation to stay connected with thousands of financial institutions. It has a very serious responsibility to maintain the security of the system and to ward against cyber security threats.
It also has a huge coordination problem. Yet the pressures that current financial institutions are facing to do things more quickly, more accurately, are surely appearing inside the Fed as well. The Fed in some sense is a granddaddy of all people who sit on top of big data, which presumably could improve the way you facilitate transactions, enhance bank supervision, improve economic forecasting to some extent, and I am sure there are other applications. How is the Fed thinking about these challenges?
Patrick Harker: I will start with the standard Fed disclaimer: I am not the Fed. I can only say what I think and not for anybody else in the Federal Reserve System.
People think of us obviously as a monetary policy engine for the country and policymakers, but we are also a bank. People often forget that the Federal Reserve System’s 12 reserve banks make up the bank and [provides] banking products. We have the wholesale product office in New York, the retail product office headquartered in Atlanta, Cash [Product Office] headquartered in San Francisco.
Who are our customers? They are banks and the U.S. government. We are the fiscal agent for the U.S. government. Out of St. Louis and Kansas City [Mo.] [federal reserve banks], there are lots of activities related to Treasury services that we provide. This kind of activity gets sprinkled all across the system. For example, we just built a very large and complex post-processing payment system for the U.S. Treasury. We built it here and we handed it over to Kansas City to run and put in operation. As a bank, we have lots of legacy systems. Fedwire, Fedwire Securities, ACH — you can go down the list. We have the same problem large banks have of how to modernize those systems.
… By the way, everybody thought for years that cash was going away. We still all have cash operations. It has not gone away; below our feet [at the Philadelphia Fed building] is a lot of cash. That’s not going away anytime soon, at least it doesn’t seem like it. But in addition to the legacy systems we have, we have two other constraints imposed on us.
“The real challenge for the Fed … is how do we provide the basic rails of the financial system so that the money can move efficiently, and do that in a way that everybody can have access to it?”
One is we can’t cherry-pick. We can’t pick our customers and say, ‘we like you better than you. You are profitable, [but] you are not profitable.’ Ubiquity is clearly important in the Federal Reserve System. The second constraint that is on us with the monetary policy act is that we shouldn’t be in businesses where the private sector can provide that [product or service] efficiently and effectively.
That raises a set of questions that we are, frankly, still debating. Several years ago, before I got here in 2015, the system launched this conversation about faster payments and more secure payments. That process I think many of you probably … have been involved in some way, shape or form. But that conversation continues.
It comes down to a fundamental discussion that the system, ultimately the board of governors, has to make a decision on: What businesses are we in? And if we are not going to be in it, can the private sector provide it? And if the private sector provides it, are they going to make sure that the small community bank in Altoona, Pennsylvania, can plug in?
The real challenge for the Fed, as we think through the emerging landscape of technology and changes in technology, is how do we provide the basic rails of the financial system so that the money can move efficiently, and do that in a way that everybody can have access to it? Everybody. That is really the challenge.
This conversation is going on. We have not come to any conclusion. There was a Federal Register notice recently about asking questions around real-time gross settlement systems or real-time systems that other central banks in the world have put in place. Whether we do that or not is still an open debate; there is no conclusion on that.
Herring: You mentioned the other central banks in the world, and it is true you have to worry not only about what is going on here, and especially in the Federal Reserve District given where we sit, but the fact that the dollar is the main reserve currency for the world, and you have to somehow interact with lots of other central banks, some of which are experimenting with issuing their own digital currencies.
I am sure this is something that your R&D function has to worry about, as well as cryptocurrencies that I assume you are not really thinking about adopting but have to worry about nonetheless. What is the thinking about digital currencies?
“It comes down to a very fundamental question: Do you trust that currency?”
Harker: We’ve had digital currencies since the advent of the computer. There is only $1.5-ish trillion of currency outstanding, hard currency, in the world today. The vast majority of money moving around the system is in digital form. … Digital has been in existence since the first mainframes; it is going to continue to be in existence. That’s how we move money across the economy.
The question about cryptocurrencies — again, it’s still an active discussion about the pros and cons, the advantages and disadvantages of different types of technology. In the case of currency, we can talk about and we can argue about different technology platforms, blockchain, invariance of blockchain, and so forth. To me, though, it comes down to a very fundamental question: Do you trust that currency?
The vast majority of Americans and people in the world are walking around with dollars in their pockets, not even thinking about whether they trust [the currency] or if [one dollar] is actually worth a buck. They just walk around and assume it’s a buck. Why? Because generations of life experiences [have shown that] this is worth a buck. When they walk into a store, it’s worth a dollar.
That is not necessarily the case for some of the emerging cryptocurrencies. There, they do seem to behave a little more commodity-like with variable prices than, say, the buck. So, trust is really at the center. We only believe that piece of paper is worth something because there is a central bank that backs it up, and the American people believe that we will actually back it up.
If you don’t have that, it is hard to build a broad-based use of some other currency. Now, I don’t dismiss the fact that people could build that trust over time. There is no question that people could build that trust over time. But that is ultimately what it comes down to — if people don’t believe it is worth what you say it is worth, it is not worth what you say it is worth.
Herring: It goes back to the fundamental principle that at root all finance is a ‘confidence game,’ in the best sense of the word, of course. But the attempts by some central banks to introduce digital currency go a little further than you would suggest. … Sweden, I think, has gone a long way toward displacing real currency with digital currency.
Harker: Here is my litmus test. When we get rid of the penny, we can talk about digital currency. We can’t even agree on getting rid of the penny, which costs more than a penny to make. I don’t think we as a society are … there yet. I think the American public still likes having cash.
“This will continue to be a partnership between traditional banking institutions and startups.”
Herring: How are you coping with a system that is getting permeated by fintech, not just the freestanding fintech firms, but partnerships between fintech firms and existing banks, and fintech divisions within banks? They are relying on all kinds of technology that is very difficult for an outsider to understand, yet you are still obliged to think about fair lending practices and a whole range of laws that pertain to basically transparency in data and decision-making.
Harker: First, we need to build our expertise [internally and through] conferences, which is something that we need to do not just in Philadelphia but across the system. But it is beyond just assembling all of [the conference speakers and attendees] so we can learn from you — and that is really what this is about, us learning from you, and having this dialogue — we also need to start investing in different types of talent within a system.
For Philadelphia, we are investing in a small but, potentially over time, growing group of machine-learning folks. We hired our first individual this summer, and we are in the market to look for a second person. That [team] will be a resource for many across the bank to start to think about machine learning. One area is just simply understanding what it is, and most importantly, what it isn’t. Second, in our supervisory role we are going to increasingly be asked to opine in some way, shape or form on the relationships the banking institutions have with other vendors or their own internal teams on what these algorithms are.
We are also going to have to figure out some very deep issues around consumer compliance — fair lending for example. This was brought up by [Fed] Governor Lael Brainard [in an earlier speech]. You throw a bunch of data in, and it discriminates against women in hiring. Who knew? The institutions can rightly say, we never meant to do that.
[The algorithm] just trained on that data in a way that we didn’t understand. And that is an honest answer. So, one of the things we are doing in addition to hiring some internal resources is also partnering with our colleagues at Penn computer science. Typically, Feds always partner with local economic departments and business schools. Here we wanted to reach out to the computer science department.
“There is clearly a role for regulation in the system. We need to make sure that the system remains safe and sound.”
… We hosted a conference about a year and a half ago — half [of the attendees were] macroeconomists and half machine-learning people. We were trying to understand what that intersection is — and there are some deep issues there. For example, one of the issues [Penn computer and information science professor Michael] Kearns is working on is, how do you build fairness into a machine-learning algorithm? … It doesn’t know the concept of fairness — and it is actually a deep technical issue of how to do that.
Again, we are not going to solve that problem. Hopefully, Kearns and his colleagues will. But we need to understand it, because on the flip side we are asking banks, ‘Are you complying with these laws?’ If they say, ‘Well, I don’t know, the algorithm is just telling me so.’ That is not a good answer … but we need to understand it together.
Last but not least, we in our supervisory function have lots of data that we analyze for patterns, both at micro-prudential levels and at macro-prudential levels. For example, in Philadelphia, we are the repository of all consumer loans in the system. We have something [at the Fed] called the Consumer Finance Institute and a big part of that grew out of the payment card center here many years ago, out of the credit card industry.
It is a large group within supervision, regulation and credit focused on trying to look for patterns within an institution and across auto lending, home equity — you name it. That is a pattern recognition problem. What are we seeing? What are those patterns? We have great statistical minds here and econometric minds doing that work, but can we enhance it some way with some of the advances in machine learning?
I don’t know the answer to that, but part of what the Fed does and should do is try to work with the academic community and the industry to enhance the frontier of knowledge in these areas. That is what we are working on.
Herring: Another aspect of your responsibility is to think about how to integrate these innovations and fintech firms into a system that will be more efficient, still fair and transparent, and one that has a measure of competitive equality. I guess that raises a question of whether there should be a special fintech charter, whether we should have a sandbox to experiment with this sort of thing. In many ways we seem to be behind the rest of the world in thinking about how to deal with this set of new players.
“Part of what the Fed does and should do is try to work with the academic community and the industry to enhance the frontier of knowledge in these areas.”
Harker: I am very sympathetic to the idea. As companies … scale and get to scale, the idea that you would have 50 [sets of state rules for fintech] as opposed to one [federal regulation] probably makes less and less sense. There is clearly an advantage to having a national solution at a certain level. I wouldn’t say [it is necessarily fitting] from day one for an institution as they are starting up, but at some level having that national charter or whatever, that is out there, … I am sympathetic to that. We can quibble over the details of that, and there is a lot of quibbling we could do, but I think the concept at least, it makes sense.
But the other thing that is very clear is that this will continue to be a partnership between traditional banking institutions and startups. Not for everyone, but for a lot of institutions, [it makes sense] to be that third-party vendor, that white label institution behind the scenes … particularly for small community banks. One of the great advantages of this country that we often don’t see as an advantage is that we do have a great diversity in the types of banks we have.
For example, coming out of the recession one of the first De Novo banks, probably the first De Novo bank in the country is here in this district, the Bank of Bird-in-Hand [in Bird-in-Hand,] Pennsylvania. They know how to serve the Amish and Mennonite community who have very unique needs when it comes to their banking practices. For example, the Amish don’t have insurance on the barn because if the barn burns down, the community rebuilds the barn. There are little things like that — but they are not so little.
You need to understand that. The third-party vendors, the core processors, are going to be critical in this effort. It is not just the bank itself, but they will have to work on integrating the advances in fintech. That is the Jack Henrys [fintech company that manages the technology needs of many banks] of the world and so forth, the ones that are going to be critical players. We need to understand more what they are doing and how to regulate them as well because they are playing such a large role.
“What are the jobs … that have a 95% or better chance of being automated in the foreseeable future? That number is about one in five within the Third District.”
Herring: You do have people providing services that are parallel to existing bank services, but under a very different set of compliance costs and regulatory prohibitions. At what point do you need to rethink what it means to be a bank?
Harker: On the fintech side, on the small business and consumer side, you are right. There is other pressure on banks, too, with shadow banking and being a funding source for commercial activity. You are seeing, on both sides, the banks are being squeezed. That said, I still think that there is clearly a role for regulation in the system. We need to make sure that the system remains safe and sound. It is what we were chartered to do after many, many chaotic moments in American history.
Herring: Let me now turn to the longer-term issue, which is that many of these innovations are certainly sold under the principle that they are really relieving workers of routine, dull, mind-numbing chores, and freeing up workers to think about more important, more engaging problems.
The issue, of course, is that the people who are being freed up from these routine chores are often not well-equipped to take on the more challenging, more interesting problems. And that could, to some people’s way of thinking, create longer-term unemployment issues, which will have to become a concern of the Fed over the long haul.
Harker: The bottom line is the machines are coming. I don’t agree with my old student [Tesla founder] Elon Musk that the human race is [under] threat [from AI]. But there is clearly change underway, and the pace of change is accelerating. … We have always had new technology, we’ve always had Luddites, we’ve always had the need to retrain, to get the new skills. This has been true since the Iron Age and the Stone Age, but now the pace is just accelerating.
In Philadelphia, we just released a report where we asked the following question within this district as an example of what could be happening nationally: What are the jobs by geography, within the Third District [that the Fed oversees], and by city, … that have a 95% or better chance of being automated in the foreseeable future? That number is about one in five within the Third District [covering eastern Pennsylvania, southern New Jersey and Delaware]. It’s a big number … and it will vary a lot.
In Reading, Pennsylvania, 22% of the workforce is in manufacturing. They never lost their small to mid-sized manufacturing base, and they are actually good at it. The answer for Reading is going to be different than the answer for Philadelphia or the answer for Wilmington, Delaware. The message in that report is look at the data, understand what your challenges are, and then we need to rethink. … We need to stop thinking of workforce training and development programs as a social good and a government program, and start thinking of them as an investment.
“If I look at the state of the economy today, the biggest challenge we face right now is the lack of workers. I am hearing this everywhere.”
One thing we are hearing loud and clear is there is a skills mismatch in the workforce and people can’t find workers. The growth is being limited by their ability to find the labor they need. We have done a lot of work in this area — I don’t see this [technology] as a threat, where Skynet (AI dictator from the ‘Terminator’ movies) is going to be active, we are all going to go away — but rather this is a chance. It is not going to happen overnight. Companies are going to make decisions on their capital expenditures in these technologies. Some of them are already doing it because they can’t find the workers, so they decided, ‘Well, the cost to me, the bar is now much lower because I can’t find the workers anyway.’
We need to rethink the model of workforce development. We have released a system-wide book … of really interesting papers on a conference we held in Austin, [Texas]. We took it one step further and said, ‘What would such a thing look like?’ We have announced here in Philadelphia [a very different model], in partnership with Philadelphia Works, which is the local government training program, a very large technology employer in the city, and a nonprofit called Social Finance. Social Finance will seed with some other money the startup of this skills-based training program that are developing skills specifically for that employer.
One of the problems in these workforce training programs and why they fail is they are based on the ‘train and pray’ model. We think we know what they want, we’ll train you for something we think we know and then we’ll just pray you get a job — as opposed to employers being engaged from the get-go and developing the skill base. So, the seed money will come, but then as the employers get the workers they need, they will reimburse. In this case, Philadelphia Works for that benefit of getting that worker with that skill set they need, and then this will start to create what we hope is a virtuous cycle. In this case, digital skills are necessary for this employer, but [what’s needed] will vary a lot by city, by the company you work with.
The point is, let’s stop thinking of this as some social good — and it is — but start thinking about it as a real, hardcore business investment that we need to make, a public and private partnership, to move the needle. If I look at the state of the economy today, the biggest challenge we face right now is the lack of workers. I am hearing this everywhere — [the need for] workers with the right skill set.
“There’s a guy in Lehigh Valley that is paying starting diesel mechanics more than $100,000 a year and can’t find [workers].”
You see this in the JOLTS data — the Job Openings and Labor Turnover Survey data. For several months, openings have been higher by a lot than the people who are unemployed. And there is this lack of dynamism — whether people don’t have the skills or they’re in the wrong place geographically and unwilling to move. There are lots of potential reasons, but it is one of the limiting growth factors in the growth of the U.S. economy. We need to bring more people off of the sidelines and into the workforce.
Herring: It is not only a productivity measure but also a full employment measure. It is often said that we are much better at providing credentials and skills, and it at least goes some direction toward addressing the gap. The Germans seem to have gone even further. Do we have a full understanding of why the German system of apprenticeships doesn’t seem to take hold in the U.S.?
Harker: This is going to be an odd thing for a lifelong academic, former university president to say: We are sending too many kids right away to a four-year college. Given the demographic shift and the college-age population in decline, I actually think it is going to get worse.
There are many good, what we call opportunity occupations [around] — jobs that pay above minimum wage where there is a future, where you don’t need a four-year college degree. [Of course,] you almost always need something beyond high school, such as an associate’s degree, and at some point you may go back and get that four-year degree as you move up and you become a supervisor.
But the idea that for everybody in America the path is high school and then college — it doesn’t fit everyone. What we are finding is that there are these severe shortages, [for example, in jobs like] machinists. Ask the local people in your neighborhood or in your communities. [Another example are jobs for] diesel mechanics. There’s a guy in Lehigh Valley [in Pennsylvania] that is paying starting diesel mechanics more than $100,000 a year and can’t find [workers].
These are good-paying careers. And it could be a career that goes somewhere else over time. Germans have embraced that [idea of apprenticeships] culturally. It is much more of a cultural issue than [in the U.S.] We send the message that everybody goes to college. When we dug deeper on why this is, some of it comes down to the incentives that we set up for guidance counselors in high school.
What is the measure of success of a guidance counselor? How many kids go to college. That’s it. It’s not how many kids have gotten a great career as an electrician. It’s just not what they measure. And so some of it comes down to even just generally how we think about what are good, solid career paths for people in our society.