78 Technology For Understanding Policy
We are joined by Max Ghenis (https://twitter.com/MaxGhenis), co-founder of Policy Engine https://twitter.com/ThePolicyEngine) for a conversation about ways technology can help folks understand the impact of public policy proposals.<br><br>### Resources and Shoutouts:<br>- I, Daniel Blake (film)(https://www.imdb.com/title/tt5168192/){<br>- What we owe the future (book)(https://www.goodreads.com/book/show/59802037-what-we-owe-the-future)
Top Keywords
- policy 0.014
- income 0.007
- open source 0.006
- policies 0.006
- state 0.006
- source 0.005
- poverty 0.005
- much 0.005
- government 0.004
- open 0.004
- system 0.004
- policy engine 0.004
Transcript
Hello. I'm Ryan Cook, and this is Civic Tech Chat, a show that looks at the way technology, politics, and policy impacts the world around us. The tools we use, the way services are delivered, and how we talk about and set policy all shape our society. We'll gather around and have a chat about these things together and more. Before we get started, I do wanna let you all know that we've started a Discord for the podcast. There will be a link with an invite down in the episode description. Do feel free to go check that out. It's a small community right now, but hoping to grow it. It's a great way to reach out to me and let me know things that you might want us to cover or to just hang out and talk about civic tech. Anyway, let's go ahead and start the show. Max, thank you so much for joining us here on Civic Tech Chat. Could you introduce yourself and tell us a bit about what you do?
Yeah. Thank you for having me, Ryan. My name is Max Gennis, and I'm the cofounder and CEO of PolicyEngine. We are a nonprofit that builds free open source software to compute the impact of public policy. So specifically, we have a web app that lets anyone design a custom reform to the tax and benefit system, also check their own taxes and benefits, and compute what those kind of reforms would mean for society as a whole and, individual household.
Max, what would you say is your personal why? That thing that drives you to get out of bed each morning and do what you do.
Yeah. I in some sense, it's pedantry maybe. Thinking about there are a lot of really important questions out there that people don't always have answers to. And I hope that our software can make those kind of conversations more data oriented. And from a more serious perspective, I think poverty alleviation has been a major driving force for me. It's sort of why I, left my cushy tech job and went back to school for economics to study the especially global poverty, but also we're working in The US and The UK where there's a lot of poverty. And there's a lot of policy lack of transparency. People don't know how public policy affects them in really material ways in terms of understanding how their net income at the end of the month is going to be affected not just by how much they work and how much they earn, but also government policies that represent in developed countries a third of GDP. So these are really important questions that we hope we're, shedding some light on.
Are there any videos, podcasts, books, or other pieces of media that you'd recommend out there for folks that are listening?
One piece I've really enjoyed recently is called I, Daniel Blake. It's a movie set in The UK about someone who has a medical catastrophe and loses their job and has to deal with the government benefit system and just has a real difficult time and showcases how both policy design and transparency can actually affect people's real lives. It's actually a drama to fiction, so it's also just really entertaining. But I think it has a lot of truth in to it that's motivated me. One other book I've been reading kind of on the other end of the spectrum is called what we owe the future by William McCaskill, who is the creator really of the effective altruism movement, which aims to encourage people to do the most good they can with resources they have available. And this book is in particular about thinking about the long term future of humanity, And this is, maybe an ambitious vision or view of what we're doing, but I I hope that we can play some part in creating a society that is more empirical, has better epistemics, and just is more caring about individual households and connecting those views of individuals back to the interest of society as a whole.
So, Max, you mentioned that, you were, part of the founding team there for PolicyEngine. Can you tell us a bit about what that is and why you wanted to start it?
Pulse yeah. PolicyEngine is a a nonprofit that we started last year, and it actually grew out of another nonprofit I created called the UBI Center, which is a think tank that researches universal basic income policies. And my cofounder and I both worked there, and we really designed the underpinnings of policy engine for ourselves. It was an open source project, but our mission was to help ourselves understand this very complex system of the existing tax benefit laws and understand how various reforms to it that might have something to do with universal basic income could actually work, how they would affect households, what kind of tax rates they might imply, what kind of work incentives, poverty rate inequality, all that, and then we built a front end for it. So that front end also helped us navigate our own questions as policy researchers, but we pretty quickly realized that this was relevant beyond UBI policies. So and relevant hopefully to other think tanks as well. So that's why we spun it off into a new nonprofit in The UK last year in October and earlier this year, a few months ago in The US as well.
Looking over some of the resources you provided to kinda help me get a grasp for, what policy engine is about, and I think related to some of your comments about seeing it as being more than about those, like, particular policy areas is, there's a comment there that mentions that you'd like to see policy engine become a software backbone for economic policy globally. Can you talk a little bit about what that what, like, what would that look like?
Yeah. When people econometricians or statisticians, when they want to do a new analysis using some regression technique like ordinary least squares or some other kind of approach clustering, they very rarely start from scratch. They pretty much use existing software packages. Maybe that's Stata or some package in r or Python. And that, I think, first of all, saves them a lot of effort, and it also makes sure that their results are more consistent and accurate than if people were doing it from scratch every time. This really has not translated yet to the area of policy analysis. So many different think tanks when they're doing their analyses that score policy proposals or government departments doing the same or academic researchers. They will build their own usually bespoke model just to capture the part of the policy that they are most interested in. And first of all, that means it's usually off limits to people who are not in those organizations. So it's rarely open source, and, also, it's usually incomplete. It's pretty opaque in terms of how accurate it is. Some of them are very well resourced and build very high quality software, but we think there's an opportunity to help unify all those kinds of rules engines. They're really translating a large part of this is translating tax code benefit rules into code and then applying that to answer relevant questions for public policy making. So we think we can make policy analysis more efficient, more transparent, and more accessible to the public and not just the the top think tanks by having an institution that's doing that across lots of different organizations. And so far, we've done that in The UK and The US. US is very complicated as I think we might talk about soon. And so I think that sets a precedent for doing it in other countries as well.
How can tools like the one we're talking about help engage folks in this policy making process that might not otherwise be engaged?
I think I'm, to some degree, an instance of this. So I spent a lot of my career as a data scientist in tech, and I was curious as many people are about public policy design. But I didn't really feel like I had the tools to answer some of the questions I was most interested in, again, because a lot of these models are within certain institutions that are closed off. But I I found a organization called the policy simulation library that was building an open source tax model, and I was able to, first of all, answer some of the questions that I had myself by reforming the tax system using their tools. And, ultimately, I contributed to that project, and that's kind of the predecessor to what we're doing today. Certainly, I I think my background's a bit more technical than the average American, but a lot of people are curious if you just go on Twitter and see the kind of policy debates that come out, how is some policy that just came out going to affect me? The Inflation Reduction Act just came out, and we're actually in the midst of modeling those tax credits and rebate programs. They're quite complicated, and I think there have been some parts of the programs that have not been widely reported, such as tax credits are not refundable. So if you don't owe tax liability, you're probably not going to benefit from it. So having a tool where you don't have to do any coding, you can just go in, enter your household information, and see the personalized impact of that policy, I think, can help people move into understanding it better. And then over time, I would also hope that we can get them to think about the impact of those policies, not just on them, but also on society by presenting it in the same user interface.
I think you're touching on this a bit already, but, I imagine that there's difficulties potentially in trying to take these kind of, like, quantitative measures and translating them into real world impacts like you're talking about. Things that would be understandable. For example, if I were to look at and go, oh, what would do this to either my, like, my livelihood or the way I'm kinda interacting day to day? How difficult is it to to try to make that that leap?
I think the societal metrics are harder than the personalized ones. We have a couple screens on our product that if you design a policy reform, it starts by telling you the impact on society. So because we are really geared toward policy analysts. But we try to start with relatively simple metrics. How much is the policy going to cost? What's the impact on the poverty rate, which has been widely reported? For example, the child tax credit reforms and the American Rescue Plan. So people seem to care about those kinds of simple metrics, and then also the share of people who come out ahead and behind from policy. If you scroll down a bit, you can see the Gini index and other inequality indexes that are more for people who really understand how those measures work. But I I think the key to what we're doing is also sharing those personalized impacts. So when you see that policy reform, you can see how is this going to affect my tax liability, how's it going to affect the amount of benefits I receive, and some of those together, that's a net impact on your net income. So some policies might increase your taxes, but also increase your benefits. It's kind of going to be ambiguous what the net impact is going to be. But you can put all those things in dollars and cents and tell people, you know, at the end of the month, this is what it's going to look like for your your household, checkbook.
Something in there that can maybe at times also be difficult to make connections to is you mentioned, like, here's the, change in taxes you're paying versus services you're getting. I sometimes wonder if it's difficult sometimes to connect those two things together where, oh, like, yes, I've increased the amount of taxes I'm paying, but there is a real increase in services, that I benefit from that. It's potentially like a net gain, but maybe it's hard to see in, like, the pure numbers of, of legislation that's come out. Is is that something that can be, like, difficult to communicate around with folks as they're trying to understand the impacts of these sorts of policies?
Yeah. Definitely. There are some policies that are easier to put in the same number system as others. So if you just have a refundable tax credit or something like that, that's basically just cash. There's other cash assistance programs too. On the other end of the spectrum, you probably pay property taxes that fund the park or your school, and that's not part of our product. I think that's pretty hard for people to even wrap their head around in terms of how much they value that service. And then there's a lot in between. So everything from Medicaid to food stamps or broadband subsidies. These and there are actual economic studies showing people tend to value those, aside from health care, but certainly food stamps. People value at roughly 90¢ on the dollar, because they are constrained. They it's sort of conditional on spending a certain amount on food. Broadband subsidies also. And there's also the application process. So this resource might be available to you, but how much is it worth it to or what's the cost to you of actually going through the process of applying and keeping up with the paperwork and all that? So I think these are really interesting empirical questions. We've tried to take as neutral stance as possible on it. So just saying here's the cost to the government of providing this value or the benefit you receive and just letting people, make their own judgments about how much it's worth. And but over time, I I think we would love to explore those kind of subjective questions as well.
As we talked beforehand, you mentioned that there have been some challenges as you've gone from starting in the in The UK to now thinking about the policy in The US space. Can you talk a bit about the those difficulties as you've made that that transition?
The US is a beast in terms of public policy. The UK system and The US system on the face have kind of similar structures. So you have a progressive system. The nominal tax rates increase with your income. They both have means tested benefit programs. Many of them are catered especially toward families with children. And you have unemployment benefits, benefits for people with disabilities, etcetera. But having implemented both sets of policies, even at the national level, The US is just way more complicated. We have many more parameters. So even our tax system, we have seven brackets instead of three in The UK. We have a pretty wide range of parameters that say how a program phases out. For example, the earned income tax credit has many more parameters than the equivalent program in The UK. And I think the biggest source of the complexity is the state level variation. So not only do federal programs vary by state, for example, the food stamp program, the utility allowance, the amount that you can write off from your income if you pay utilities, that varies depending on the state as do a number of other parameters of that program. But, also, every state has its own policies, its own state income tax, its own benefit programs. And something also that makes more complicated but is kind of a similar between The US and The UK is all these programs are much more complicated for low income households, especially low income households with children. So but, yeah, The US is just you can kind of count up the lines of code, and we're only three states through, all 50 states. But I think we're going to find that the repo is a lot slower to download for The US than The UK over time.
So I I imagine there being these well, I guess, 50 states with very different, policy environments could potentially be a bit of a a problem. Like, well, how do you how do you scale your effort? How have you all been kind of approaching that challenge if if you don't mind me asking?
We've gone where just in terms of prioritization by state, we've started with the places that have relevant policy conversations happening. So there was a program in Massachusetts that we want to analyze. We have a relationship with a organization in Maryland that we want to enable, and then Washington has a new program. They also have very relatively simple tax system. So those were our three. We're working on California, New York, Pennsylvania, some of the big states next. But, in terms of scaling, it will require maintenance. I think we're you know, once the policies are done, we're gonna have to make sure that we're checking every year. Are these changing or not? Ultimately, we would love to move to a system where it has enough visibility with other researchers or even people in government that we could guide them through at least filing GitHub issues to let us know when some policy is changing, so it's not just us doing this, and ultimately, potentially have them be real partners and co owners of this collective, software product that hopefully could be useful for lots of different, parties.
An issue that I imagine is also somewhat related to the ease of of interacting with with, various, state governments at this level is this idea of, like, level of professionalization of, state legislatures. So this is an issue and a topic they may show itself through thinking about things like what are the staffing support levels that each legislator might have, how much time do they get to spend in session, what tools do they have available to them to do things like research and actually, like, draft, what will ultimately become bills. And I'm sure there's other things that I'm not remembering in the moment as I I try to ask this question. But, what opportunities do you see for technology to help out in legislatures that might have needs in, you know, one or more of those areas?
I should say, first of all, I think state legislative analysts, especially, and lots of people in state government are really doing heroic work. If you compare what resources they have, as you were saying, to the resources available to the federal government, and those folks also have, they really do amazing work. It's just a lot going on and not a lot of time, and legislators are part time in many cases. So you're sort of working, in a very concentrated part of the year, and it's just, very difficult. So and as you said, a lot of these states vary in terms of their staffing levels. So I'm based in California. Our legislative analyst office has over 50 people, and they don't just analyze the policies that are coming out of the legislature. They'll do primers on various topics throughout the year. They really do great work. At the same time, I think they, open source and other kinds of newer technologies are not usually embraced immediately by folks in government for a variety of institutional reasons, and, it's hard to hire really top notch software people and things like that. So I think we can help them and learn from them in terms of their expertise and helping them perhaps get more into the open source space. And if there are some policies that they don't need the special private dataset, a lot of the time, these state analysts do have data that means that their estimates are going to be more accurate than the public could do, at least for now. But there's a lot of questions that people have that they want more exploratory power, including legislators themselves. And we we hope that for the especially the resourced less resourced kind of teams, we can provide that kind of insight in a more rapid fashion.
I I think what I'm kinda hearing from you there is, like, the idea is to try to, I guess, like, provide the tools to, whether it's the legislator themselves, remembers that staff to kinda chase down curiosity. Maybe they have a question about, like, oh, if I do this, like, what's the the impact? Is that kinda what I'm hearing from you?
Yeah. Spontaneity. And, like, you have a shower thought on let's do this tax credit. Like, let's enable them to actually explore that idea in the moment instead of having to file a ticket to some organization to do the most full robust analysis they can. Sometimes there's folks on the team who have their they know their way around the spreadsheet and some microdata or some tool so they can do some of that work. But I think I've personally learned a lot just using policy engine and other tools like it that provide this immediate feedback. It just lets you explore a much more rich set of ideas and refine your own understanding of how the laws work.
And as you've worked with states or done exploration, through laws and policies there, are there any things that you've discovered that have surprised you along the way so
far? There have been some bugs in some laws. So, just as a couple examples, Missouri has a deduction on its state income tax where you can deduct part of your federal income tax liability. And the federal government also has what's called the state and local tax deduction. So you basically have a circular reference there, and it's not totally clear how the government intends for people to address that. Another example is Washington recently introduced, what's called the working families tax credit, and they seem to have made an arithmetic error in providing one of the percentages in the law. So rather than fixing it, they pass a new law that enabled the department of revenue to change that number depending on each year's values. So I think those kinds of things also just reveal that there are people who are making these decisions who just don't have the tools that they need to, to answer basic questions like, are how are someone actually going to fill out this tax return? How is they how are they actually going to complete this program and receive it? And can we see some graphs and some experimentation around how that all could change depending on how we refine the law?
In the conversation we had to prepare for this episode, we talked a bit about a thing called the rules as code concept. Can you talk a bit about that for folks that are listening that might be less familiar with it?
Yeah. Rules as code is sort of the hashtag. There's also computational law. Rules as code as a movement is this idea really of addressing the kinds of issues I just spoke about where if you can put all the rules that affect people that the government has in some computational coded platform, then you can address a lot of the potential bugs that might come up, have a more responsive government. You can make that as a service that lots of people can use to explore different parts of the law. I think there's also a lot of pretty grand ambitions for this movement that people explore. For example, you could imagine really using version control system, the systems that lots of folks use with Git, to say every proposal that aims to reform the law, that's really a diff. Right? That's kind of a pull request or something like that. So can we unify those technologies between legislation and get inversion control to have a much better understanding of what people are actually proposing. And then if you hook that into a platform somewhat like ours where other platforms are estimating the impacts of those laws, you can see directly this proposal is going to change this. That's what the impact is going to be on the budget, poverty, maybe carbon emissions, air pollution, kind of you can go down the list, but you need as a framework to understand all that and to make that all happen. You need to be able to implement these, words in the laws as something like Python code.
I guess then I could see it. You know, the a subcommittee could have a, a linter for the subcommittee style rules for for, for for bills put forward.
Exactly. Yes. Two spaces only. Yeah.
Cool. If there are, folks out there listening today that say they're listening to this conversation and they're like, man, this, like, policy data, this technology space, kinda where all these things are meeting, that might be something that I've I'm interested in myself in trying to, like, get involved in. What sort of advice would you give that person that's kinda now wondering, well, crap. Like, what do I what do I do to get started?
I would find an open source project if you can. That's certainly been very valuable to me. I was able to find a project around tax credits, which was an area I cared a lot about because it is a major influencer of the poverty rate and, outcomes like that. So find the outcome that you care about in the policy area and see maybe there's already an open source project that's exploring it. I'll just say there's a couple other projects that have really influenced me. Open states is another. So they scrape legislation from state websites, and they then make that service available to anyone to explore, what's going on in every state. There's a tool from the New York Fed. They have a dynamic general equilibrium model on GitHub. Both OpenSASE is also on GitHub. There have been open source models for pandemic spread over time. But I haven't seen, for example, an open source climate model to say this is, for example, the Inflation Reduction Act. What is that going to mean for carbon emissions? There were a few different analyses that were reported in the media, and they had pretty different estimates of both under baseline policy and the IRA, what emissions would look like. So I think having some open source option to explore those kinds of really world changing bills can be really powerful. But there's so many great commune communities on GitHub and elsewhere to just get your feet wet. Try using the software yourself if it doesn't meet your needs entirely. Just poke through the code.
Max, thank you so much for joining us here on Civic Tech Chat. I know, folks will, listening to this will find some nuggets to take into their day and, learn from.
Thank you so much, Ryan.
You can follow us on Twitter using the handle at civic tech chat. Visit us on the web at civictech.chat, or subscribe to us for content updates wherever it is you download your podcasts.
