so I'm going to start in today's lecture by just first giving you a motivation for why we should be interested in studying public economics and then I'll talk a little bit about the logistics of the class and give you an outline of the course so what is public economics at a broad level public economics basically focuses on answering two types of questions first how do government policies affect the economy so for instance if we have the Romney tax plan instead of the Obama tax plan at a very broad level what are the impacts going to be on the economy in terms of the number of people who are working the distribution of income and so forth second is a normative question how should we optimally design policies in order to maximize welfare so if we have a choice of different social welfare programs or education finance policies or tax policies which of these policies should we actually pick in practice in order to maximize economic growth then address distributional concerns I think there are three motivations for studying these questions the first is just the practical relevance the sheer practical relevance as you can see in the current election of knowing what the answers to these questions are the second is academic interest and the third is methodological so I'm going to spend some time on each of these three motivations and while talking through those three aspects of the course hopefully you'll get a flavor for some of the things we'll do during this class so motivation one practical relevance you know I think the simplest reason to be interested in public economics and the way I got interested in the field is I think many of us come to economics with an interest in improving the world in some way nning in increasing economic welfare social welfare and that I think directly leads to an interest in public economics why is that almost every economic intervention that we think of typically occurs through government policy that is it involves public economics through generally two channels so first we think of various types of price intervention so taxes welfare programs social insurance the provision of public goods all of these are basically the government changing prices in the economy in some way right so the government might make put it put a tax on carbon for instance to make use of polluting materials more costly in order to help the environment so any issue that you're interested in like environmental economics ultimately ends up having some public economics aspect the second domain of intervention is regulation so things like minimum wages FDA regulations zoning laws labor laws all of these things also involve the government involved the type of analysis that we'll talk about in this class another direct reason that we're interested in studying the role the government is that it just plays a big role in the economy directly as an employer even in the US which has one of the smallest government sectors one-sixth of Americans are employed directly by the government now the stakes in answering these questions are extremely large because of the broad scope of the policies that were analyzing right so for instance tax reforms are immediately going to affect millions of people and so these are extremely important questions from that point of view and there's a very contentious debate on the appropriate role of government in society so you see this in all you know in almost all public political debates including current campaigns so for it to take one concrete example Mitt Romney argues that replacing Medicare with his proposal of a decentralized private insurance system will improve health outcomes and reduce costs obama and contrast argues that the romney proposal will worsen health outcomes and raise costs now only one of these guys can be right right by definition and in principle we should be able to figure out the answer to that question now you you both sides cite economists who support their view so you know part of the problem is that we don't really know what the answer is but you can see that injecting science into these debates has great practical value and i think we're starting to get to a place in public economics where we have what looked more like scientific answers to questions like this now the second motivation for all of you who are might take this course as students is the academic interest of this field so public economics is typically the endpoint for many other subfields that you might study so for instance macro development labor and corporate finance questions often at some ultimate deep-level are motivated by a public economics question so for instance there's a large literature macroeconomics on the costs of business cycles and intertemporal models of household behavior like life cycle models or models with credit constraints buffer stocks and so forth all of those theories at the end of the day matter because we were interested in things like studying unemployment insurance or studying the effects of fiscal stimulus and so it's very natural to combine public economics with one of these other fields as part of your graduate coursework another example is studies and labor economics of the employment effects of the minimum wage similar kind of rationale right you're ultimately interested in the policy question so I think beyond the specific issue of comment combining two fields I think the other thing that's useful is that understanding public finance can help ensure that you work on relevant topic so I think especially in applied micro nowadays there's a lot of focus on finding clever identification methods to answer some empirical question and I think the concern is that when you approach a problem in that way you are not necessarily working on something that is important or that other people are going to find interesting say boring when you're on the job market I think that often ends up not being an issue in public economics because it's very easy to say you know you should be interested in understanding the impacts of health insurance or tax policy very few people would argue with that okay the third motivation which I actually think is perhaps the most important one is that this is a very exciting time in public economics and applied micro more generally in terms of the kinds of work that's being done in the field and I would say public economics is at the frontier of a methodological transformation and applied micro more broadly and so the way to think about this I think is that people are increasingly taking up taking a very data-driven approach to answering the important policy questions that we've been talking about and so rather than you know in the past when there was a lot of purely theoretical work or more recently purely empirical work there's a focus on combining theory and data in a very particular way that I think is powerful in its ability to answer a lot of important questions and so what you'll see is that the course combines a broad set of skills apply theory applied econometrics and simulation methods and that's a useful skill set for many applied fields in economics so we're going to talk about a bunch of different topics in the class which I will come back to later in the lecture but all of those are going to reflect a broader set of methodological themes and so I want to highlight those to start the class so that you see how the various topics we talk about connect back to these basic themes there are any questions please feel free to interrupt that at any time okay so the first important theme is as I just mentioned connecting theory to data so modern public economics tightly integrates theory with empirical evidence to derive quantitative predictions about policy so take specific questions like what is the optimal income tax rate say on top income earners or what is the optimal unemployment benefit level we now I think have the tools where we're actually able to come up with numbers based on some theory and empirical evidence for what we think these the answers to these questions are under certain well-defined Social Welfare functions and you what I find quite interesting is that you see the translation not just from academic journals but to the political debate and you see editorials for instance in the Wall Street Journal or New York Times citing the results of papers that are exactly answering these questions using the types of methods that we're going to talk about so there's a lot of practical impact using this approach now that differs strongly from the traditional approach even you know 10-15 years ago the type of work that was being done in this field which was primarily theoretical work or numerical simulations so typically the problem in public economics theory as and I think theory more more broadly is that unless you make strong assumptions most generals were general versions of theoretical models of the type we'll talk about in this class make pretty weak predictions so you know to take an example in the tax literature you get the prediction that the optimal tax rate should be between zero and a hundred percent and while that you know might sound trivial that's actually as you'll see not totally trivial to establish mathematically but that's pretty much all you can get from a purely theoretical model right and you can see that if you went to advise presidential campaign with that type of it by – you wouldn't really get very far so the other technique was numerical simulation so let me make a lot of assumptions about the parameters of these models and then make a quantitative prediction about what the optimal policy would be so there you can get much further but the problem is you're making incredibly strong assumptions about what utility functions look like what behavioral responses to policies look like and so forth and it's very hard to know whether your results are robust to alternative versions of those assumptions and in practice they typically are not so the theme of this course we're going to focus on modern work typically you know within the past 10 to 15 years is that it derives grow bust formulas from theoretical models that can be implemented using well identified empirical estimates so we start from the types of theoretical models that people were analyzing in the past but then distill those models to a small set of parameters that matter for a policy question like optimal tax policy and then develop really credible research designs to estimate those parameters so you end up using the tools of applied microeconomics labor account liber economics and so forth to estimate parameters that are highly relevant to these models and answer these these policy questions so you'll see that in each of the different topics that we talk about now that naturally leads to the second theme which is the use of quasi-experimental empirical methods so what you'll see is that throughout the course you'll basically get an illustration of various quasi experimental research designs for instance events studies regression discontinuity synthetic control all of these are the types of methods that we'll talk about in the class and while we're talking about those different types of methods I think you'll pick up a number of practical lessons and applied econometrics so for instance you know by the end of the course hopefully you'll be able to answer all these questions what is identification by functional form you often hear that term discussed in seminars and people usually think that that's not a good idea so why is that when is propensity score reweighed useful approach when does it lead to credible estimation what drives weakens from problems and how can they be fixed or another question that often discussed in seminars are you estimating a local average treatment effect of late or an average treatment effect when should you be concerned about estimating a late as opposed to an 8 what matters in in different cases now all of these issues at some level you learn in a theoretical econometrics class but at least in my experience when you actually come to applications there are a number of practical issues that arise which you can really only learn by doing this type of work and seeing how it's done in practice one thing that I'll emphasize throughout the class is the use of nonparametric graphical techniques rather than parametric regression models and so I want to give you an example of what I mean by that just to start a very simple stylized example so I'm using here a paper by a famous statistician named Ann scam from 1973 where he illustrated the value I think of nonparametric analysis in a very simple way so what is nonparametric analysis mean at the simplest level it just means plotting the data or looking at the data directly rather than imposing the assumption that the relationship between two variables is a straight line or is quadratic or something like that so let's take four different data sets okay which hypothetically are show you the relationship between years of schooling and earnings so people who get more schooling earn more here's data set one where you see that you know there's a straight line relationship between years of schooling and earnings if you look at these data points and you estimate a coefficient of 1/2 with a standard error point 1/2 now here's a second data set where you also if you just ran an OLS regression and did not look at the data you would conclude that the regression coefficients 1/2 and you get a standard error of 0.1 – here's a third data set that also produces exactly the same regression coefficient and standard error and here's a fourth data set that also produces the exact same regression coefficient standard error now you can see if you actually look at the data those four relationships are very different right and you wouldn't make the same statistical inference from those four graphs even though if you just ran the OLS regression which is the typical thing that people use to do you would conclude exactly the same thing in all four cases so this is obviously a cooked up illustration to make this point but my sense is that this captures something much broader that's very important in the field that you'll see that the most compelling evidence now is nonparametric evidence that doesn't rely on any particular functional form assumptions about linearity or quadratic relationship and so forth and essentially what you see with the modern research designs like regression discontinuity or synthetic controls difference and difference is that they're harnessing this type of nonparametric approach combined with specific quasi experimental designs to deliver these credible estimates okay now that leads me to what I see is the third theme of this literature which is what's good popularly called big data or the use of administrative data in the context of economics so what you'll find is that compelling implementation of quasi experimental methods typically requires a tremendous amount of data so just to go back to this graph I what you'll typically see as a rule of thumb is if you get a T statistic of 2 when you run a regression p-value of less than 0.05 which is typically the bar people are shooting for you will find that if you draw the graph associated with that it'll look terrible in the sense that you won't actually find it to be very convincing in order to get a convincing graph that is convincing nonparametric identification you need a tremendous amount of data because you're looking at each of these points visually in some sense or another way to think about it is if you implement a regression discontinuity design you're essentially using data very close to a threshold cutoff and throwing away most of the other data in your sample so you need to start with an enormous sample in order to be well identified ok so the I think the reason the field is really developing very rapidly now as we increasingly have access to such datasets so for example we have scanner data on consumer purchases with tax and Social Security records which I will talk about a little bit more in a second we have school district databases which kids and have give you their test scores and teacher assignments and so forth which we'll talk about toward the end of the class you have information on peer and social networks from websites and other sources and these datasets have really allowed us to do a number of new things that are very exciting so just to show you their importance in the field I want to show you what tabulation that we put together documenting the use of survey data and administrative data in leading economics journals so take four of the leading economics journals AEE RG p QJ in econometrics this chart here plots the fraction of articles that use survey data sets pre-existing survey data sets so what I mean by that are data sets like the current population survey the survey of income and program participation the panel study of income dynamics if you took a class like this save twenty years ago or even ten years ago all of the empirical evidence you talked about would be from studies that use these type of data sets and in fact my own thesis which I wrote here at Harvard used somewhat used data from the survey of income and program participation so what you can see is a striking trend where today a a very small percentage of articles are using survey based data sets among a microt empirical studies right so for instance in the QJ that number looks something like five percent in one of the latest issues and the way I think about this is this is basically like two papers the QJ you got 1600 submissions in this year so if you're starting with you know survey data set you're basically aiming and you're aiming for these types of girls which is what you should be you're looking at like a 2 and 1600 odds which is obviously not a great situation to be starting from now wait what is substituting for the use of survey data they're basically two things one is that people design their own data sets like small experiments where I've you know let's say randomize something in a school I then collect my own follow-up data on those students and then I write a paper using that so there's some of that but then the other major source of data is what you see here the use of administered data in these same journals and you can see this very sharp upper trend and these are data sets like what I was describing things like school district databases tax records and so forth the fundamental distinction being that I don't need to come and survey you in order to get the information I automatically get the information because it's recorded in some way when you scan an item at the grocery store or when you pay your taxes or when social security updates its records and so forth so what is the benefit of this type of administrative data concretely first it provides higher quality information right so there's virtually no missing data or attrition so take the case of tax data you know basically either you're paying taxes in some way or you're dead more or less right so in the US so you're you're rarely going to miss people in contrast if you take typical survey data sets like the CPS there's a 31 percent non-response rate for income and that is you might have seen your first year econometrics classes creates all kinds of econometric problems to deal with where you have selection bias because you only have information on 70% of the people and so forth so what you find with these admin data sets is a lot of the basic econometrics issues that we worry about like attrition and selection end up being non issues and you can focus on really developing a credible research design with this high quality information the second feature of these data are that they permit a longitudal long tracking over long periods so one of the things that we'll talk a little bit about in this class but we're just starting to get evidence on are the long term impacts of policies typically because we haven't had the data to analyze the effects of policies safe 20 or 25 years later but with administrative data sets because you're linking people by these identifiers like social security number you can track people over even say 20 plus years like in these studies of the long-term impacts of early childhood education which we've recently done here and look at things for instance like how the quality of the teachers you had in elementary school does a paper will talk about toward the very end of the class how the quality of the teachers you had when you were saying kindergarten or in third grade affect your earnings when you're 28 years old and you can see that the quality of teaching an elementary school does actually have substantial effects on earnings and that ends up having various policy implications for how you want to hire teachers incentivize teachers and so forth the other benefit of using these large data sets is their just their sheer size right so the tax data for instance are 2,000 times the size of the CPS and so one benefit of that is that you get smaller standard errors but that's not the most important one it's that you can develop new research designs right so to illustrate this illustrate some of the type of work that we'll do in this class let's look at the Earned Income Tax Credit so this is from another paper that we've recently written with John Friedman and Emmanuel Saez so one of the programs will talk about a fair bit in this class is called the Earned Income Tax Credit which is kind of the biggest anti-poverty program in the u.s.
Is one way to think about it it provides cash transfers for low-income individuals that increase with the size of your earnings so if you're earning zero dollars you get a zero EITC and then as you earn more you get about 40 cents on the dollar from the government okay depends the exact number depends upon the parameters of the structure of your family and then the refund is maximized at some point then there's a plateau region and then it gets cut back okay so I just want to point out that this income level here maximizes the size of your tax refund why is that because as you go to the right you're paying various other taxes to the government like payroll taxes right so if you could pick one point to be in the US tax system like if you could just make up your income you'd want to make up an income of nine thousand you know three hundred dollars or whatever that cutoff is all right so here now is the distribution of income in the US just plotted using the population tax data so this is a histogram it shows you the percentage of people in thousand dollar income bins just you know from 0 to 40 K and take first the one child case which is the schedule that I was showing you before you can see that there's a big fairly big spike in the distribution exactly at that point that maximizes the size of the tax refund right so a bunch of people have figured out that this is the point that I want to be at in the tax system or at least the point that I want to tell the IRS that I'm at in the tax system in order to get the biggest refund for people with two kids the kink the refund maximizing point is shifted over a little bit and so you can see that for those people the spike again lines up exactly with the point at which they maximize the size of the refund right so the basic point here is that a bunch of people are responding to tax incentives and we'll talk about this in more detail later but we think a lot of this is just non-compliance that's intuitive because it's very hard to imagine that you actually managed to earn exactly nine thousand three hundred dollars right yeah that's something that you can write down on the tax form if you're say self-employed but let's say you're working at Walmart it's hard to go to your Walmart you know manager and say I want to work exactly this many hours so that I maximize my EITC refund so now what's interesting about this and this is what I want to illustrate here with these data is that increasingly you see the use of research designs that would not have been feasible with the types of data sets we had before so we can look at this distribution by ZIP code in the US and that's what this map shows you so this is a heat map which plots the fraction of people who are reporting exactly that refund maximizing level of income all right so think of this as like the fracture people who figured out the system and figured out how to maximize the size of their tax refund so the dark areas are places with higher levels of response to the Earned Income Tax Credit and the light areas have lower levels of response so this map is drawn for 1996 and you can see that there just a few dark spots 1996 is when the EITC really became big in the u.s.
It was expanded under the Clinton administration so think of this as kind of year zero for the program and so you can see that they're just a few spots like the southern tip of Texas New York the southern tip of floor where there's substantial response right when the program is introduced now let's look at what happens over time says 1999 2002 2005 2008 you can see like it looks like this knowledge is kind of disseminating from southern Texas right as the there's spread of response to the EITC over time and so we'll talk in more detail about this paper later on but this basically gives you a research design to contrast places where like North Dakota where people don't seem to be responding very much to the EITC with other places in order to figure out the impacts of the program on the local economy right which is obviously a design that you could not have implemented with previously available data resources okay now a next important theme in the field is that work in public economics is heavily influenced by insights from many other fields but in particular psychology in economics behavioral economics and I think that's because of growing evidence that people fail to optimize in various dimensions and that creates a direct role for government intervention right it raises new policy questions and it suggests new policy instruments so to take one example there's growing evidence that people don't really pay attention to their retirement savings accounts the work of David Laibson Bridgette major and others and that creates an important set of questions in terms of what the government should be doing should the government mandate that people save a certain amount should it subsidize certain types of retirement accounts should it perhaps have defaults that automatically enroll people in certain types of plans there are costs and benefits to all of these different policy options you wouldn't really have thought of them prior to thinking about the behavioral economics literature but I would say now they're front and center in this field and just to give you an example in the context of what we were just talking about so those maps with the EITC higher and lower response what are the what what are the characteristics of the places where people are responding much more to the EITC turns out that the single strongest predictor which by itself has an r-squared of 0.6 which is quite high is the density of AIT C filers in your area so you could kind of see that if we go back to this map the places where people are not responding are the places that don't have a lot of density of eid c filers or for that matter I mean just North Dakota but density of people in general not just the ITC fathers so you can see that the really dense places are the places where people are responding and what's the intuition for that we think it's something like agglomeration or peer effects if you're out in the middle of nowhere and you're eligible for this program you've never heard about it and nobody's going to tell you about it if you live in an apartment building where lots of other people are claiming the EITC you yourself are likely to hear about it and figure it out right and so that's addressed that information and peer networks might be a key determinant of response to policies and so one of the things that might come out of this is rather than expanding expenditure on the EITC by increasing that 40 cent match to a 50 cent match maybe what we should be doing is getting the information out there to the places where people are not responding which could potentially be much cheaper right so we're currently spending 50 billion dollars a year on the EITC maybe it would be much more effective in increasing labor supply of low income individuals if it was structured differently okay so let me now start to get into the material a little bit and just start with a broad overview of what the government does so government expenditures account for about one third of GDP in the US and that is actually relatively low relative to most other developed countries so in many European countries more than 50% of GDP is government expenditure a key feature of the US government is decentralization we have government at multiple levels one-third of the spending is done at the state or local level and two-thirds is done at the federal level so one natural question right away is why are we doing some of the spending like on public schools at the local level and why some of the spending done like on social insurance being done at the federal now here's a time series of what federal government revenue and expenditure look like now I apologize but this is actually the labels are switched so you should correct that in your notes so revenue did not spike during World War two the expenditure did and what you can see is that you know we're roughly 20% of GDP now that numbers crept up a little bit Romney is arguing that we should have a cap of 20 percent of GDP to bring expenditure back down you can see that we were running a surplus around 2000 but then after that we've had deficits and so forth and at a broad level revenues and expenditures look roughly aligned which you would hope they would be but you know increasingly we know that that is not the case which is exactly the problem of the growing debt and question is how we address that now here's what government spending looks like by country you can see that there are vast differences so as I was saying the u.s.
Is considerably blowing the this is adding state and federal now right so that's where we're at 30% u.s. considerably lower than these other countries and so you know one of the big picture questions is what's the optimal level of government size at some level is it that's going to depend upon the details of what the government is doing it but do we want kind of a Sweden Swedish type of social safety net where you get several months of maternity leave when you have a child paid for by the government and so forth or do we want something more like the US or do we want to go more in this direction you can see that there are big questions here and substantial variation now where are we collecting our money in in the US so federal tax revenues have changed considerably over time so in 1960 the vast majority of the revenue the government got was coming from income taxes and corporate taxes and to a much lesser extent payroll taxes and excise taxes so what our payroll taxes they are levied on wage earnings directly rather than on total household income so for instance the Social Security tax that's automatically deducted when you get paid where FICA is that's what's called the payroll tax that income tax that you pay at the end once you add in you take it account of all your deductions and capital income and so forth that's what we call the income tax so the main chip that you can see from 1960 to today is that there was a substantial expansion of payroll taxes right and so to someone on why that is why are we collecting a lot more money from the payroll tax today instead of income taxes what what changed but yep Medicare social insurance programs more broadly Medicare Social Security we're spending a lot more on the social insurance programs that are financed by payroll taxes right and so that's probably the single biggest shift in the nature of revenue collection in the u.s.
And at the state and local level there are also some changes somewhat smaller we used to collect a lot of money from property taxes typically used to finance schools that's become much less important but what has become more important are income taxes and federal grants so that is money being collected by the federal government and then being passed on to the state in the form of a block grant that the state can spend on various things okay now what what is the government spending the money on exactly as we just talked about the biggest the single biggest shift and this motivates the nature of research today is that we spend a lot more money on social insurance programs and transfer today than we did in 1960 so taken particularly the case of help 2.9 percent in 1960 23 percent to date and projected to continue to grow very rapidly right if you just linearly extrapolate it's going to be a hundred percent not only a government spending but of GDP so obviously something will change before then hopefully but that's the major shift and so if you look at what the government is actually spending money on the vast majority of it are transfers from one person to another so if I'm unemployed and you have a job you're effectively giving me money for the government if when you retire somebody's giving you money paying through paying for it through payroll taxes etc right so all of this stuff basically looks like that and then we spend a little bit of money on national defense on education on things like roads and infrastructure and so the consequence of that is in the earlier literature economists didn't really have very much to say about optimal spending right because economists don't have much to say about the best way to build a highway for example but if you are when the expenditures on things like Social Security or health insurance there is actually a lot of economics and that there are incentive effects there are welfare effects and so forth and so there's a huge growing literature on these programs more recently now the way in which the u.s.
Collects its money is actually quite different from most other countries and the single biggest difference is that we don't use much consumption taxation in fact at the federal level we use virtually no consumption taxation all the taxation is of income rather than expenditures right whereas in these other countries you tend to have very large value-added taxes or taxes at the point of sale so I take in particular developing countries like Mexico you can see that consumption taxes are huge there what why do we some and guess why that is yeah yeah it's much harder to enforce an income tax than it is a consumption tax typically and so develop developing countries tend to rely on consumption taxes for that reason now even developed countries rely on consumption taxes quite a bit like the value-added tax because many economists tend to think that value-added taxes are actually more efficient than income taxation and we'll touch upon this briefly later on in the class the US for whatever political reason has not gone in that direction okay so now to broadly organize the class the question we always ask to start analysis of any of these topics is when is government intervention necessary that is we take a market first government second approach so why is that because you know from your introductory economics classes that the private market outcome is typically efficient under a pretty broad set of conditions so we usually think that private markets are going to do a good job in terms of delivering economic outcomes that we like and so the government should only be fixing up what goes wrong with the private market and that's very much the way you'll see we'll approach each one of these questions why is it that we actually need the government to do something and conditional on the government doing something what you know what should the government do so the course can be abstractly split into two parts the first is how can government improve efficiency when the private market is inefficient the second is what can the government do if the private market outcome is undesirable because of redistribution concern so the private market outcome might be efficient but we might not like its distributional properties and so we still might want to do something as a social planner so to illustrate let's take the simplest possible exchange economy where let's say you have ten apples that you're dividing between Amy and Bob the efficient frontier is to have all of the ten apples allocated to one or the other so that you're on the blue line right and we think that the private market is typically going to deliver an outcome that's going to be on that frontier and so then from an efficiency point of view we're not worried about it it's not like we could have made Amy better off without hurting Bob right we're on the Pareto efficient frontier now there going to be some cases where the private market actually delivers an outcome like this where sumit apples some of the resources are essentially being wasted and we can make everyone better up and so we want the government to intervene and so the first set of topics in the class or the first half of public economics is basically about how we improve efficiency what are the circumstances where we can improve efficiency now there are situations where we the market might deliver an efficient outcome and so we're not concerned about improving efficiency but we're unhappy with the distributional properties so if amy gets all the apples and Bob doesn't have any we might not like that and so then we might want to implement tax or transfer policies to shift the outcome to something that is more desirable from a social welfare point of view so what are the conditions under which the private market provides a Pareto efficient outcome this should be familiar to you from prior economics classes we know from the first welfare theorem that there are three conditions needed for the private market to provide an efficient outcome the first is that there are no externalities the second is that we have perfect information and the third is that we have perfect competition and the way I think to think about this class is that it's basically a class about the failure of the first of all program like what are the cases where the first welfare theorem fails and what can we do about it so let's talk about each of these failures in turn so start with externalities so the issue with externalities is that markets may be incomplete due to a lack of prices essentially so I think the best way to think about externalities is not so much that there's some property of certain types of transactions but that they're really about missing markets so for instance pollution would not be thought of as an externality if there were a market for pollution if there was a price I had to pay every time I emitted pollution as a firm right there would be no externality there and so we think in practice that there are not markets for many things people care about and so achieving the efficient solution well and we'll talk about all this in more detail requires some sort of organization to coordinate individuals like if all of us are having externalities on each other we want somebody to come in and coordinated one to reach a better equilibrium and what is that organization typically it's the government right the government serves that function of dealing with large-scale externalities at some level by putting in restrictions on pollution or imposing taxes and so forth so this is why the government so it makes sense that we need the government to intervene when we have externalities that's why the government funds public goods but there are important questions of what public goods the government should be providing and what are the best tools to correct externality so one of the questions we'll talk about for instance is I could regulate the level of pollution or I could impose a tax on pollution these are two different policy instruments and in a very well-known paper our colleague Marty Weitzman here talked about the trade-off between using price methods versus quantity methods and we'll talk about that paper later in the class so the second set of failures also has to do with incomplete markets and that's asymmetric information so when some people have more information than others we know that markets will generally fail so the classic example is adverse selection often in the context of health insurance so take an example where everybody knows their health status healthy people are not going to want to buy insurance if it's priced for the average person because I'm going to think you know I don't want to pay a thousand dollars let's say for this health insurance plan because my probability of getting sick is very low so I'm going to drop out but that's going to make the average person in the population sicker health insurance company's gonna have to raise premiums in order to not go broke and that cycle keeps continuing until the market breaks down so this is a situation where mandated coverage could potentially make everyone better off and government intervention could be warranted another example which is highly relevant actually in the current economy is capital market imperfections or credit constraints where essentially there's similar type of asymmetric information problem that I would be willing to lend you money if I knew that you were a reliable borrower but I have no way of knowing if you're reliable or if you're like one of those other guys is not they pay me back so I end up putting much tighter restrictions on lending than I otherwise would and this can potentially have very important long-term consequences tying back to that earlier theme of looking at long term impacts what are the impacts of providing access to credit for education if we expand the Pell Grant program in the US which enables a lot of people to go to college does that have high returns does the government sort of get its money back in terms of tax dollars does it increase GDP in the long run these are all questions on which we have some evidence but we could use much more evidence going forward I think they're very important question a third example which is not really about asymmetric information per se but it's sort of incomplete markets is markets for intergenerational goods right so in the in the standard first welfare theorem type of analysis neoclassical model we assume that all the participants are at the table that is everybody who's affected by an economic decision has their interests represented in some sense in the economy but future generations are not their interests are not being directly internalized right when we make choices about what to invest in or how much to pollute and so forth and so you might think that the government is acting as a representative not only of the current generations but of future generations as well and so may basically make people more patient than they otherwise would be taking into account these future generation okay so all of those are issues we'll talk about especially in the context of social insurance the third failure is imperfect competition so when markets are not competitive there's again a role for government regulation so we know that when we have monopolies they're not going to go to the point where price equals marginal cost and especially when we have natural monopolies like electricity or telephone markets there are you you naturally end up wanting to have to regulate these suppliers in order to get them to behave in the way that you'd like from a social point of view that topic for historical reasons I think is left two courses on industrial organization and is not typically covered in public economics now intellectually my view is that that doesn't necessarily make sense the methodological approaches that we're going to talk about here in the public economics course I think could well be applied to issues related to regulation there's a very interesting set of work to be done I think in that direct and there are some people starting to do this type of work so for instance Glenn while at Chicago is applying some of the tools that we'll talk about here to IO type of problems what you'll notice is also good place to discuss the difference between the types of methods that have become popular in IO versus labor and public economics you'll notice if some of you are taking all of these classes that there's a very different style of empirical work that you'll be exposed to and I highly encourage you to learn about all the different approaches io tends to take more of a structural approach where we're trying to character fully identify the model and then characterize what the equilibrium looks like and what the impacts of policy changes are on equilibria public economics has gone in the direction of being less reliant on estimating all the parameters of the theoretical model but that did not naturally put some restrictions on the types of questions we're able to answer but within the domain of questions we answer I think we're we're able to obtain answers that are much less dependent on strong modeling assumptions the final set of motives which is what we touched upon a little bit earlier are not market failures but individual failures so this is not something that we really would have talked about in the traditional public economics class but increasingly it's clear that this is very important if people don't optimize then government intervention may be desirable so for instance we might want to force people to save for retirement for instance by a social security program independent of whether there are any market failures and I actually think this is extremely important because intuitively and from looking at the work in this area while I think we can formulate reasons for government intervention in markets like Social Security so you know there might be some asymmetric information in annuity markets that makes the government provide these social security annuities in practice I think what it really boils down to is that if we didn't make people prepare for retirement some way we'd end up with a bunch of people who showed up with no assets when they were retiring because they hadn't planned properly you have very high elderly poverty rate and you want to do something about that so you end up having social security so I think while it sometimes theoretically elegant to come up with market failure type motivations for some of the programs we're going to discuss in reality what seems to resonate with politicians and the public is that it's really about addressing individual failures in many circumstances and I think further developing this literature as a direction that's going to be important so a key conceptual challenge this literature is much much earlier stage right and so there's still a bunch of conceptual issues to be worked at one of the most important ones is how you avoid the paternalism critique how is the government supposed to know what's better for you than you yourself so who is the social planner to come in and say you should be saving more or you should be wearing a seatbelt or you should not be smoking et cetera right and that's really important in terms of doing theoretical work in this area because if you don't answer that question you basically have no discipline in terms of determining optimal policy so if you're respecting individual's preferences which is the traditional approach then you have a well-defined objective function which you can talk about how you maximize when you're subject to various market failures and that's what a lot of public economics is about but when I say you know individuals are maximizing utility function you and I is the social planner want to maximize utility function B and B is just something I've made up and I have no discipline on what B is then obviously I have no discipline on what policy recommendations come out of that and so one of the challenges is how do you do behavioral economics in a well-defined systematic way when you don't trust individuals own utility function okay now the last set of topics is redistribution right so even when the private market outcome is efficient it may not have good distributional properties and I think we're increasingly seeing see this all the time now in the in the popular debate that efficient markets seem to deliver very large rewards to a very small set of people right so this is a discussion about top income inequality in the US and how a tiny share of people have an enormous amount of wealth and the income in the economy both in the US and many other countries and so people tend not to like that outcome and so you know what can we do about that while the government can redistribute income through the tax and transfer system like the Earned Income Tax Credit by taxing high incomes and so forth now that might naturally lead you to think well why don't we just have the government do everything right if the government is can deal with this problem of distribution and potentially implement the types of outcomes that the market implements why do we limit government intervention in any way and so you know the way to think about that is we one solution to all these issues would be for the government to oversee all production in allocation in the economy right like a socialist approach so this goes exactly against that market first government second approach that I said we've taken this class and I want to talk about why we don't approach it from this perspective I think there are some deep problems with this solution that relate basically to informational issues the first is that the government is not really in a position to aggregate people's preferences and technology to choose the optimal level of production and the optimal allocation of resources so to take an example you know suppose I was to determine how everyone in this room would be allocated some fixed set of resources if I want to know what everybody's utility function looks like so that I can give the people who really like candy bars more candy bars and the people who like more apples more options right now how am I going to elicit those preferences if I go around and ask everyone there's no way you can have a incentive compatible mechanism designed there because you have no reason to tell me the truth right you should say you know I love candy bars and I love apples so give me everything right and so there's no real way for the government to elicit information properly that's one deep problem the second deep problem is that even if the government manages to implement some allocation you have to take into account that the government can't fully control really what people do and so if I tell you I'm going to perfectly redistribute resources across individuals in some way like let's say I equally split the number of apples across people no matter what then you essentially have no reason to put in any effort in production right your incentives are completely distorted because the rewards you get there no relation to the allocation you're going to be assigned by the government and so what that generates in practice is a situation where the government can't actually do everything you have to rely on the private market to do most of the stuff and then you fix it up but the consequence especially the second point is that there are sharp trade-offs between the costs and benefits of government intervention so that picture before where we had the red dot moving along the frontier to implement to improve distribution that's actually never feasible right what we actually end up doing is as we distribute resources from a me to Bob we end up moving into the interior of the set because it's infeasible to give resources to Bob and take them away from Amy without distorting Bob's incentives to work for instance okay and so you end up then having to answer this question which I see is kind of the part of the political debate at some level how far in am I willing to go in order to achieve more equity right that's the basic idea of the equity efficiency great often it arises in numerous contexts that we will talk about so another way to organize the class a related way is that they're going to be three types of questions one can think about answering the first is a set of issues related to what people call positive analysis what are the observed effects of government programs and interventions in practice so to take an example suppose I change tax rates by 10% what actually happens in practice how much less to people work what happens to say GDP if we're able to measure such aggregate effects and so forth all of that is positive analysis the second set of issues which typically builds on the first is normative analysis what should we do if we can set the optimal policy so as we've been talking about what's the design of these various programs usually what you'll see is that you need to take the lessons learned from positive analysis in order to say something meaningful normatively a third set of issues which we will not talk about in this class but would be addressed in a political economy class are what's typically called public choice developing theories to explain why the government behaves the way it does so one way to think about that is it's a positive analysis of the government itself so why does the US have more income taxation while European countries have more consumption taxation there must be some reason for that and we can develop theories of why the institutions have developed such that these countries are doing different things right a related issue is to identify optimal policy given political economy concerns so in the type of normative analysis we'll talk about in this class it's going to be as if politics didn't doesn't exist it's like what would we do optimally if we could just set the policy but in reality you're always going to face political constraints and so a more nuanced way to think about it is what's the optimal policy subject to the fact that I actually have to get this thing approved by the voters some model and that's the standard criticism of pure normative analysis and that's more of a political economy approach which i think makes a lot of sense but it's not going to be what we cover in this class so here's an outline of what the course will look like so in the next couple of lectures we will talk about the very basic issues of tax incidence and the efficiency costs of taxes so these are examples of positive analysis so when I have changes in tax rates how is the distributional burden of that tax shared across agents in the economy how does it affect the level of output etc then we'll take those tools and use them to address a set of normative issues in optimal tax analysis so what's the optimal design of consumption taxes what's the optimal way to treat savings what's the optimal income tax system what's the best way to design the transfer system we'll talk about a bunch of theory and evidence and in all of these sections let's spend some time focusing on one particular empirical literature that's very large which is the effects of income taxes on labor supply because that's viewed as a key thing the income tax is central in most developed economies and the key dimension it distorts as the amount people work so there's a big impaired empirical literature that also connects to labor on that topic we'll then do a couple of lectures on corporate taxes you'll see that the corporate tax auditor is actually pretty old because people haven't been working on it for a while and we actually didn't even teach it the past couple of years but I've incorporated those lectures this year because my sense is that there is a lot of interesting stuff to be done in that area and so I want people to be exposed to it will then do a section on social insurance so these last two lectures you can think of as being more focused on expenditures the first four lectures that first four sets of topics are on a collection of tax revenue so social insurance and then public goods and externalities so how do we correct situations where we have externalities where we want to provide public goods and there I'm adding a lecture this year on education policy which you know one of the issues is should we really think of education as a public good now in practice you know that's not entirely clear but anyway we have the government intervening substantially in education markets and what's the best way for the government to intervene that is a really hot area and so I think it's useful to talk a little bit about those issues from a public finance perspective