Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Episode #111: Len Kiefer
177 Plays7 years ago
Len Kiefer is the Deputy Chief Economist at the Federal Home Loan Mortgage Corporation, better known, as Freddie Mac. Freddie Mac is the government sponsored enterprise that helps keep mortgage money flowing to home buyers and home owners, across America....
The post Episode #111: Len Kiefer appeared first on PolicyViz.
Recommended
Transcript
Introduction and Welcome
00:00:11
Speaker
Welcome back to the Policy Viz Podcast. I'm your host, John Schwabisch. Happy New Year, everyone. I hope the new year is kicking off well for you. I am excited because on today's episode, I'm actually live. Well, I guess all podcasts are live to some extent, but I'm actually in person here in Northern Virginia with Len Kiefer, who is the deputy chief economist at Freddie Mac. Len,
Len Kiefer's Role and Journey to Data Visualization
00:00:31
Speaker
welcome to the show. Hey, thanks for having me. Thanks for bringing me in to Freddie. It's always exciting to be here. You guys actually have a very nice sound studio here. Yeah. That's impressive.
00:00:39
Speaker
You have a pretty interesting, I'd say, Twitter feed at the very least, because that's where I've sort of gotten to know you. Lots of different data visualizations, doing a lot of different stuff, mostly with housing and the economy. And I want to get to that in a little bit. But maybe we can start by having you talk a little bit about your background and how you got here to Freddie and the sort of work that you're doing here. Yeah, sure. So I'm an economist working in our economic and housing research group here at Freddie Mac. And so what we do broadly is help people understand what's going on in the economy.
00:01:05
Speaker
housing and mortgage markets. So we do a lot of different research and publications that we'll put out. Part of the Twitter that you mentioned is part of an effort to engage people, to get them sort of enticed to sort of read more about some of the trends that we're looking at. I came here, I wasn't a housing economist initially. My background, I'm a PhD economist, was a macro economist. And when I came out of school, I was a professor at Texas Tech University. My wife was also an economist.
00:01:30
Speaker
She took a job in DC. I made the wise decision to follow her and ended up here in DC and ended up at Freddie Mac. And that was actually pretty much my introduction to data visualization. As you may imagine, a lot of economists sort of figure things out, but they're not always the greatest visuals, particularly a decade or so ago. And so when I was working at Freddie, folks in the group I started with were really interested
00:01:54
Speaker
in presenting information effectively and really got me introduced to the idea of data visualization and really taking it very seriously.
Data Visualization Culture at Freddie Mac
00:02:01
Speaker
Interesting. And that sort of helped me, I think, get started to think about this and it's ongoing, a lot of learning, thinking about new ways to do it and every day I'm sort of figuring more things out. Right.
00:02:09
Speaker
And over that time period, how have the tools changed? So I know you're a big R guy from what I've seen. So have the tools really changed what's been used here and what you've been using? Yeah. When I first started, I mentioned the group I was with actually was big into S+, which was a precursor to R in a lot of ways. And so we learned a lot. It was actually eye-opening for me, as I had done a lot of things in MATLAB or in Excel coming out of school.
00:02:36
Speaker
And so the S plus sort of the structure things we had a kind of a set of visualizations that we used to help come play to look at some complex sort of data analytics and it was really effective. And so that sort of got me introduced and then I shift groups got away from it.
00:02:50
Speaker
Did a lot in, you know, Excel, not really got into R until a few years later when I started seeing some cool things that people were doing. I saw some, you know, actually some statistical things that I wanted to do. So I actually started using R and then I discovered, hey, this is actually a lot, helps me make a lot of cool visualizations and I started just sort of going down that path.
00:03:10
Speaker
And the community is so great. I got engaged with some folks there. And so I just started trying to build on that. And so for you, R is the entire data process. Like, you're doing your data analysis and your statistics works and your visualizations. Because I know a lot of people who just do their data vis in R, but maybe they're doing their statistics in state or SAS. Maybe not S-plus anymore. But you're using it for the whole thing? Well, it varies, really. Certain things I'll do in R, certain a lot of stuff that I share on social media,
00:03:39
Speaker
is done in R. Actually at work, though, a lot of the things we'll do will be in other tools. We have a wide variety of tools people use, like even within my small team.
00:03:48
Speaker
Some people use SAS, some people use Stata, some people will use R. It really varies. And so it depends on the project, who's working on what. If you're sort of sharing, you might want to use a different tool. So R is becoming increasingly popular. But a lot of times, that's not the tool of choice. So to work together, you might need to use a different tool. So I use a variety of things. But I'm increasingly using R in my own work.
00:04:12
Speaker
in my own research. And so I found it pretty helpful. Right. Now you mentioned that when you got here, data visualization was already something that was seemingly important. But what is the workplace culture like when it comes to visualizing data, communicating data? Lots of people I talk to, they might be the only person in the organization that's interested in this. Freddie's huge. I mean, thousands of people, right? So what's the culture like when it comes to communicating data?
00:04:40
Speaker
You know, it's very interesting, John, because this organization, as you mentioned, is really large. Freddie Mac has thousands of employees. At least hundreds of them, maybe even thousands, are working on some form of analytics. There is just a ton of different data visualization, communication going on a lot internally, but a lot externally. There's a lot of complex analysis going on. So there's a ton of people working.
00:05:01
Speaker
And I think the organization has really realized that if we're going to sort of move into the modern sort of age, understanding the data is going to be important. So data visualization, I think, is an important tool. And I'm seeing that across the organization. And it's increasing over time. We brought on a lot of folks that are coming from out of school. I think there's been more focus on data visualization in school. And so you're seeing a lot of people with a lot of great ideas.
00:05:22
Speaker
And so I think it's a really interesting workplace. A lot of people are really stepping up their game across the organization. And so I think that's been really encouraged. And there's been a lot of people really focused on that. And when you think about it, we've got this huge amount of data. I mean, that varies, depends on what you mean by big data. But in the mortgage space, you've got millions of records over decades. To analyze that and really understand what's going on, these tools, I think, are very valuable. And pull out insights from this massive data is a big project.
00:05:52
Speaker
Yeah. And are you seeing that buy-in all the way up and down the organization? When I was at CBO, for example, the director that I was working for at the time was really into it. And I think that helped spread the gospel, as it were, to, yeah, this is something that's important, something we're going to prioritize. So do you see that? Do you find that all the way up and down the chain? I don't know how flat or hierarchical it is, but are you finding that sort of buy-in throughout the organization?
00:06:18
Speaker
I think so, increasingly. Some senior officers, I think, recognize that, the folks that are handling the high levels of modeling across the company. They've been very engaged in this, and they recognize this as an important avenue. Another different way is there's different levels of engagement, but I have seen that in support. Certainly, a lot of it isn't explicit like, we're going to make great data visualization. It's effective visualization. I'm going to be able to communicate clearer
00:06:44
Speaker
I think that's had engagement at the highest levels of the organization, and that translates data visualization as a part of that.
Tools and Techniques in Data Visualization
00:06:50
Speaker
I've seen a pretty strong support of that, and I've seen a lot of investments around the company, people having different training, different seminars,
00:06:57
Speaker
To help sort of recognize that is in one important way that we can sort of help be better at what we do Can you talk a little bit about the different audiences that people are talking to so there's clearly the Inside you're working on a millions of observations. You're trying to figure it out There's social media which I want to talk about a little bit because I'm kind of keenly interested in how you view like effective Twitter visualizations, for example
00:07:22
Speaker
I assume there's academic conferences and there's a board meeting and there's other policymakers. So, you know, how do people think about these different audiences? It's interesting, John. So, in my position, my role in the company, I have the opportunity to speak to a wide variety of audiences. So, I'm sort of outward facing the chief economist role and my support. We are out talking to people constantly. It's either, you know, our business partners around the country or even around the world. We're talking, you know, bond investors.
00:07:47
Speaker
There's just a lot of different types of people and they're more sophisticated to less sophisticated folks and so having sort of that effective communication takes a different set of skills than if you're talking to the expert, you know, folks that are looking at things around sort of internal models or understanding.
00:08:03
Speaker
what's going on in particular sort of areas that Freddie Mac is sort of working on. And so that's sort of why, I see a lot of different specs, there's a very specialized, very focused group that's looking at maybe one particular model at a particular area of the mortgage market. There the audience is very sophisticated, they know sort of the language, the jargon, whereas if you go out to a general audience, which we're increasingly trying to engage with to help them understand sort of what's going on in the mortgage market there, you have to keep
00:08:28
Speaker
things at a much higher level. And so the skills that sort of work in one area don't necessarily work in others, but I think there's all across the board certain principles, things that you've written on.
00:08:39
Speaker
Others on the show have talked about effective communication is absolutely crucial. And do you work with the analysts who maybe, you know, they might have a project where they need to communicate to sort of a non-specialist audience and maybe they're used to communicating to the academics or the researchers. You know, how do you work with folks who maybe not had a lot of experience communicating with that wider non-specialist audience?
00:09:04
Speaker
Yeah, that's actually a big challenge for my role because, like I said, our group is talking externally. A lot of times the new folk that we bring on, like I've brought on two new economists to my staff, excellent training, but they haven't really thought about things as an external... They've written their dissertation, right? Yeah, they're used to talking to their advisor or other experts. Talking to a general audience is tough for them to start out, but we've had some support in terms of training. We've brought in folks
00:09:28
Speaker
and also just talking to people about the importance of it. And I think also living it is important in a sense of, you know, you could make a class, but it's giving examples. You know, we'll do presentations. I'll work with my team to split things together and talk about sort of why we want to do things in a certain way. I think that is probably the best way to get an understanding and an empathy and understanding what the audience is doing.
00:09:49
Speaker
is actually thinking when they engage with their stuff. Yeah, that's great. So I know you have a love of R, and I know you have a love of getting stuff out on social media using R. Can you talk a little bit about what you like most about R, and also then how you think about putting things on social media? Because what I notice about your Twitter feed is the prime example is that the visualizations work great on social. And I think a lot of people aren't really thinking about visualizations for social.
00:10:19
Speaker
They have a project and then maybe they take a screenshot of it and they put it on social. But it doesn't always fit. It doesn't always work great. And yet yours seem to really fit that medium particularly well. So let's start with the with the R part. It looks like a lot of stuff you're doing is in R. What's your love of R? I think the great thing about R is the community of folks that are engaged in it. So there's a wide variety of people. It's open source. You've got a lot of new people coming in with great ideas. And the senior people are actually pretty
00:10:48
Speaker
uh... well-engaged you know i was starting out i was sharing some stuff i was talking to folks you know i could see a lot of followers wouldn't realize sort of their sort of level of engagement with the community right and what they actually in general will be very supportive of people in sharing their things talking with them uh... and so that's i think a real positive yeah so i've been able to set sort of a positive feedback you get engaged with people and share things they'll retweet it or share it uh... and you can talk with people in this is a pretty good discussion there and what's exciting is this all these people with these great ideas
00:11:16
Speaker
that are coming up with new ways to visualize things, new packages and things that make it relatively easy.
00:11:21
Speaker
to sort of plug in and do things, whereas instead of having to figure it all out on your own, you can see a cool visualization. Actually, in a lot of cases, be plugged in and make your own within hours or minutes even in some cases. And that's a really positive thing. You get a nice feedback, and then you can engage. And in general, people are pretty open about that and supporting people. And that's really positive. And on the other hand, you have pretty powerful tools on the back end, the folks that are actually supporting and developing this.
00:11:48
Speaker
actually made the tools from analytic perspective, data manipulation, really easy. And that sort of helps. Because a lot of the trouble is I've got this data and this awful structure, it's trapped in a PDF on Excel spreadsheet, how did I get it out? I've made a lot of tools that actually make that a lot less painful than it used to be, you know, a few years back. But the data you're often getting is millions of records, presumably over time.
00:12:11
Speaker
So presumably a lot of it is not the PDF problem. Are there particular challenges you have with getting, I don't want to use the word big data, but it's large data, getting that into R and visualizing it effectively?
00:12:26
Speaker
Yeah, I mean, that's where, again, the tools are getting better. There are other packages that make it a lot easier than it used to be when I was starting out. The hardware has also gotten better. So the computers can handle fairly large data, largish data. I can remember a decade ago trying to look at, for example, the American Community Survey, microdata, one year's worth of data. It's about three million records. Hard to put on a machine. Now you can run it almost instantly. And so that becomes a lot easier. And so for that kind of size data,
00:12:55
Speaker
It works pretty well. And a lot of the things I share and talk about is ways to aggregate it. Because that's what I do a lot of time thinking about is I have these millions or hundreds of thousands of records. How do I tell a story? How do I get out of that the insight that I'm interested in?
Insights and Strategic Thinking in Data Analysis
00:13:09
Speaker
And so how to roll that up and aggregate. And that's where ours has been pretty effective.
00:13:12
Speaker
in terms of increasing the speed at which I can write. I don't know if you have an answer to this question, but what you just talked about is something that I think rings true for a lot of people is how do they get down from this big data set into something that they can communicate or evaluate, I think, quickly and easily.
00:13:32
Speaker
What is your thought process when you have... I mean, we just take the ACS as a simple one, right? You have three million records. You want to talk something about the mortgage market. I mean, I don't know if this is like a step-by-step thing, but what is your thought process? Someone said to you, how do I take three million records and get to this graph or this story or this point that I want to get out? How do I do that?
00:13:54
Speaker
That's a tough question, but I mean, that's one of my advantage. I think I wouldn't have had a great answer if I hadn't worked at Freddie because I mentioned when I started out, I mean, that's what people were doing. They were taking, you know, maybe not millions, hundreds of thousands, large numbers of records from something you would put in a spreadsheet, right? Too big for that. And they were trying to figure out ways to help understand sort of how models were working. And so people have come up with strategies for how to do that. They had done a lot of thinking.
00:14:18
Speaker
It's not just a danger, and I fall into this sometimes, it's just to start running the computer and start running the analysis and start creating the visualizations, which is fun. Sometimes you've got to take a step back and actually think about, well, what am I trying to say? And that's where I think some of the experts that are here, they really know, people I'd worked with and still work with, they really know these data, they understand.
00:14:39
Speaker
And I think actually if you think about folks that have been in the industry for a while that had to suffer, you know, punch cards, very slow computers, they had to be very strategic about the type of analysis they did because it was so expensive. And so that I think being able to work with folks like that and then also new people who have
00:14:55
Speaker
No idea, because everything is super fast. I think that creates a nice mix, and that's where having a broad ecosystem of people working on different problems really helps. That's the great thing. That's why I'm lucky to be in this role. There's a lot of other places like that, but specifically here. They've got that insight and said, okay, I kind of know, either from a statistical perspective, how we should look at this. How can I translate that into the right kinds of aggregations?
00:15:19
Speaker
Yeah, that's really interesting having that generational match. I want to talk quickly about social media. So when you are creating a graph or set of graphs for a tweet, are you thinking about it specifically for the tweet? I feel like your Twitter feed does a really nice job. You seem to be doing a nice job of thinking about that. But I guess the real question is, how are you thinking about the different output types for the visuals?
00:15:47
Speaker
I like Twitter and social media because it's a chance to engage and they have now with the support for graphics or short animations, you can tell a little story. And so I'm often doing these things that I'll share. A lot of times are part of something. Maybe they're part of a presentation that I've used, but it can be broken off. So I do have in mind, I'm thinking, okay,
00:16:06
Speaker
For example, my team works on something called the Primary Mortgage Market Survey. It's a weekly survey of mortgage rates in the United States or what a typical prime borrower would expect to face if they were going out to try and get a mortgage today. And so we've been doing that, Freddie Mac, since 1971. We have 40-year history every week.
00:16:25
Speaker
And so part of it is like, well, what's the story? Thinking about what's been happening week to week and what's the longer term perspective on that? And so since I'm doing that every week for several years now, and I'm thinking, well, how can I engage people? Is there a story? Sometimes there's not and things are relatively stable or how things moved. I think there's always a bunch of stories there that you can tell. And so how can I tell it? Well, if you have an interesting visual, I think I can get some attention there and it can help people to think about
00:16:52
Speaker
For example, if you look at the week-to-week trends, maybe rates have moved up a few basis points. But if you look at a longer-term perspective, they are super low. And so showing that, I think, helps sort of reinforce that point for different folks. And so I think of that as a way to maybe engage folks. Oftentimes, we'll do some research piece. We recently wrote about house price bubbles, whether or not we're in one, what's going on with that. We could compose some different visuals to try and capture all the different data that we looked at for that.
00:17:21
Speaker
How can we sort of tell that story? How do we get people interested in that story?
00:17:25
Speaker
to show different sort of trends sort of in the housing market. And you've got hundreds of metro areas, 50 states plus DC. So how can you show that? Well, think in a ways to aggregate that and put that into a graph. It won't tell the whole story, but it might get people engaged. And so that's the lead to, hey, this thing's kind of interesting. We found this particular angle and here's the link to the larger, the longer report. That's often the case. Sometimes it's just a question. Some of it's like I'm chewing on something, I'm thinking about it. Let's put it out there.
00:17:53
Speaker
Just see a lot of smart people out there. Maybe they have a thought or maybe someone else will find this interesting. I've done that. I've engaged. I've seen different folks out there that are in my space sort of sharing their research piece or something they've seen.
00:18:05
Speaker
sort of sparks an idea in my own mind. Maybe I can go and do that. So it's part of being engaged in the community. So that's interesting because I would guess that a lot of the work that you all do is driven by what's obviously going on in the economy, what's going on in policy. That's what's determining and requests, I would guess, from stakeholders of we need this report on this thing. But if you ever had a case where you found something interesting just by sort of happenstance, you sort of dug in this way.
00:18:31
Speaker
And then the reaction from social media led you or your team to say, let's dive into this a little bit more. There's clearly a demand for it. A little bit. I mean, I think there's often if it's a real hit, it'll be on both sides. Right. So there'll be things that people are wondering about and then people will also be asking us about. And so there is that connection there. I don't think I've ever heard a case yet.
00:18:50
Speaker
But they keep trying, where I've gotten a comment, and it's like, oh, wow, I've never heard of that. But that may happen. And there are certain micro aspects of it, little angles on it. People will point things out. But that's what part being engaged
Research on House Prices and Market Trends
00:19:05
Speaker
is. You can not only show people what you've been doing, but find out what others are doing. And in some cases, you get some interaction. That's the best. Yeah, that's great. So you've put out a lot of things, obviously, both you and Freddie, obviously.
00:19:20
Speaker
Do you have any recent projects, new projects that you're particularly excited about, things that you didn't think you were going to find, things that maybe you put out on Twitter that were sort of any of these cases where it's like, whoa, we put this out. I wasn't quite seeing that this was going to happen, this reaction. I was going to get these comments back.
00:19:37
Speaker
Well I mean I think one of the things that we've been focused on because it's so important to Freddie Mac is sort of what's going on with house prices. So a lot of my visualizations have to do with trends in house prices. We Freddie Mac have a house price index show a lot of different ways to visualize it to think about you know what's been going on with home values around the country. And we wrote a piece on you know a house price bubble.
00:19:57
Speaker
Is the U.S. headed to another house price ball? If you look at recent trends in prices, well above income. We talked about how to think about that, ways of looking at it, and actually part of that outgrew out of some visualizations that I'd done a few years ago, just looking at metro area trends, because you've got 300 metro areas. It's hard to consume it all at once, and not likely one statistic is going to be that. So having a variety of visuals helped. We sort of came to a conclusion at that point.
00:20:22
Speaker
But it's going to be ongoing because it's so important to understand in sort of what's going on in the housing market. We'll continue to track that and be something that we'll be looking at in the years, months to come over the next year.
Conclusion and Listener Engagement
00:20:34
Speaker
Right. So on the research side, are we in a housing bubble? No, that was what our current, at least currently, that's what it was, but we'll see going forward. Well, thanks so much for coming on the show. I'm really enjoying the Twitter feed and I'll put links to
00:20:48
Speaker
all this and the Freddie research that's going on. And this has been great, really fascinating. Thanks so much for coming on the show. Yeah, thanks for having me. And thanks, everyone, for tuning into this week's episode. If you have comments or questions, please do let me know. I'll put links to all the work that's going on at Freddie and links, of course, to Len's Twitter feed. So if you have comments or questions, please let me know. So until next time, this has been the PolicyViz podcast. Thanks so much for listening.