• 00:20    |    
    So the first topic I´m planning to discuss this morning is about fiscal stimulus packages, I guess is the current preferred target, but more broadly, I´mn gonna try to be looking into how government spending and tax rates affect the macro economies.
  • 00:41    |    
    So, on the government spending side, this is partly looking for what are usually called Keynesian multipliers, this has become very popular blade, aftern being dead in the sort of policy, academic-literature for twenty years or more, so it´s kind of a surprising development.
  • 00:59    |    
    But, in the face of this ongoing financial macroeconomic crisis, a lot of governments have decided that Keynes, basically, had things correct all the timen and that multipliers are very important.
  • 01:12    |    
    And that at least something that´s helpful in the current environment is large government fiscal spending packages, usually focusing more on the spendingn side rather than on the tax side, but I´m gonna be trying to look at both things.
  • 01:27    |    
    So this is an ongoing research program, this is relying on U.S. macroeconomic data, but very long-term data, which turns out to be quite important, in termsn of isolating the effects related, particularly to government expenditure, on things like real gross domestic product.
  • 01:46    |    
    So the basic idea, in terms of the recent policy implications, don´t really go much beyond the general theory of John Maynard Keynes from 1936, which is thatn in some depressed situations, the government having an increase in its purchases of goods and services can have a positive effect on output, which is not only perhaps positive but also mightn involve a multiplier.
  • 02:12    |    
    So that the GDP responds more than one-to-one, in that kind of model, to an increase in government purchases. The Obama administration of the United Statesn is assuming that the multiplier is about one and a-half, a number which is completely pulled out of the air actually, it doesn´t really come from any empirical analysis that I know about.
  • 02:35    |    
    But it was sort of the number you needed, in order to get the stimulus package that they had proposed, to produce the results they wanted, that´s as far as In can tell about where this number came from.
  • 02:46    |    
    And quite often recently, you see the administration saying things like even though the economy is still depressed, because we´ve had this program it´s say,n so and so many number of jobs, and that all comes from this kind of multiplier analysis and then some connection between the real gross domestic product and employment.
  • 03:07    |    
    So they´re really somehow taking this very seriously, and not only that, the U.S. government is advising other governments to have similar kinds of packagesn and, of course, some governments have implemented large scale fiscal stimulus packages of this type, so in that sense is quite current.
  • 03:28    |    
    So, as I noted in a recent Wall Street Journal column; if the multiplier was even lower, so they´re assuming a number like about one and a half, if then multiplier was a more modest 1.0, even that would be quite amazing, if true.
  • 03:44    |    
    A multiplier of 1.0 means that if the government expands its purchases, lets say to build a bridge or something, by one unit, that the GDP rises by one unit,n which essentially means that for society, the bridge is free,
  • 04:02    |    
    and then, of course, if the bridge is useful that´s a really good deal because you get this useful public infrastructure project without having to give upn other resources so, in particular you don't have to give up anything from private consumption, or from private investment, because the GDP itself, expands by enough to accommodate the extran government outlay, for example, for the bridge.
  • 04:26    |    
    As I´ve mentioned, the U.S. government is assuming a multiplier bigger than one, about one and a-half and that means that the whole procedure is even moren kind of magical or attractive, because not only do you get the bridge for free when the government expands its purchases by one unit, you also get another half of a unit of private uses ofn output increasing, either consumption or investment.
  • 04:53    |    
    So, the bridge is free and you also get extra consumption or investment, that´s what it means if a multiplier is bigger than one. And that´s why it wasn always the famous result from Keynes´s General Theory, that if the government does even useless things, like filling up holes, that in some kind of situations this can be desirable, becausen even if the bridge is no good you get the extra private consumption and investment.
  • 05:17    |    
    So, this is all very puzzling and the question is what is it about the privateeconomy and how it´s functioning that could allow this to be the right answern with respect to government policy interventions?
  • 05:32    |    
    You know, as an aside in the early 1980s, when the Reagan administration came into power there were some pretty extreme arguments on the policy side but moren on the supply side, along the lines that if you cut taxes you get such an expansion of reported income that tax revenues would actually increase.
  • 05:51    |    
    So, that was another kind of free-launch result that you were in an environment where lower tax rates would produce such a response in the private economy,n you would actually get more revenues.
  • 06:03    |    
    So there were certainly no trade-off, then it was obviously good to cut tax rates if you actually got more revenues. So that seemed to be an extreme supplyn side vision, which was to some extent popular in the early 1980s, and now I think we´ve kind of replaced that with a fairly extreme demand side view which corresponds to this multipliers beingn bigger than one.
  • 06:26    |    
    So, it´s not too complicated to see why you might get this kind of result, the multiplier bigger than one. I mean, in some ways the private economy isn malfunctioning, there´s something wrong with the price system.
  • 06:37    |    
    The reason you get the multiplier bigger than one, is that there is an underlying assumption that you have idle economic resources, in the form of labor andn capital that are unemployed.
  • 06:49    |    
    And then, somehow, when the government comes in and expands certain projects, you use the idle resources and that´s what allows you to produce freely, notn only the goods and services that go to the government for bridges or whatever, but also the additional resources that go to private consumption and investiments.
  • 07:09    |    
    So, somehow, it´s utilization of the underutilized factors of production, that underlies the Keynesian multiplier.
  • 07:19    |    
    So, this is argued to be attractive because these resources really were idle, they really were somehow not being used in a productive way, and then then question is: why does not the private economy function in such a way that when there are these idle resources available, that they become utilized?
  • 07:34    |    
    What is it about the private market economy that´s malfunctioning?So that the private business and the rest of the private economy didn´t figure out thatn there were these idle resources that could be employed and then you could have a better result involving more consumption and investment.
  • 07:52    |    
    So, the Keynesian story was never too clear, if you talk about the General Theory; Keynesian economists often said that the problem was that the wages andn prices were sticky and they were too high, and that´s why the private economy wasn´t doing the right thing.
  • 08:10    |    
    But if that were really the story, then expansionary monetary policy would be enough to get you out of that problem. If it was just that wages and pricesn weren´t falling enough, to where they should by essentially inflating the economy with more monetary policy, you would get around that problem.
  • 08:27    |    
    So I think, in fact, something more serious must be involved then the sticky wages and prices; but what it is, in terms of the underlying macroeconomicn theory, I think has not been well specified or well worked out.
  • 08:39    |    
    Now if you do a standard cost-benefit analysis for government activity, so again if you go back to the bridge, and you try to decide is it worthwhile to putn in the resources to build a bridge, the usual cost-benefit analysis assumes that if a bridge costs one unit then that really is the social cost of building the bridge.
  • 08:59    |    
    And then the present value of the benefits has to be enough to more than upsetthat in order for the project to be attractive.
  • 09:08    |    
    That´s the standard cost-benefit approach to whether government should be doing some activity or not, of course there are many operational complications inn implementing that, but that´s the spirit of the approach.
  • 09:18    |    
    So, if you put that in to the macroeconomic context, the implicit assumption is that the multiplier is zero, so that if the government comes in and builds an bridge, that´s gonna drive out private uses of GDP, private consumption and investment, one to one.
  • 09:36    |    
    And, in order for that to be worthwhile, it has to be that the ultimate benefits you get from the bridge, in terms of what´s gonna happen in the future, maken that worthwhile.
  • 09:45    |    
    But it´s not a free-launch kind of argument about having more government activity, and I´m not saying that you wouldn´t wanna have infrastructure investment,n but it has to be investment that is justified on these cost-benefit grounds that´s the standard approach.
  • 09:59    |    
    Okay now, the key question really, to me, which is what this project is about, is an empirical one: what do multipliers really look like?
  • 10:08    |    
    Giving how longstanding a question this is, it´s surprising to me, there´s really not much in the way of what I would regard as serious empirical evidence onn this issue.
  • 10:21    |    
    It´s surprisingly scanty, in terms of what you can look at, to assess what is the reasonable number to assume for this Keynesian multiplier; is it greatern than one? Is it positive? Does it depend on how much idle resources are present, like how big the unemployment rate is? Is the multiplier larger in a depressed situation such as today?
  • 10:43    |    
    And the usual assumption, which is maybe be reasonable is that the multiplier is bigger when there´s a lot of unused resources, when the unemployment isn high, where there´s a full employment economy, people often think maybe the multiplier is much smaller, in perhaps zero.
  • 10:57    |    
    But the empirical evidence on these matters is amazingly lacking, and that´s what this research project is really about, it´s using the long term U.S. data,n in this case, to try to assess these questions, particularly about the extent of the government spending multiplier.
  • 11:15    |    
    So as I mentioned in the U.S., the government is always thinking that the multiplier is somehow one and a half, and all their calculations are predicated onn that, I don´t believe that number came from any serious empirical evidence.
  • 11:27    |    
    So our objective in this work that Charles Redlick and I are carrying out, is to fill what I view is this major, and increasingly important these days, gapn in the empirical evidence, about what these kinds of multipliers look like, what is the real basis for things like the stimulus packages.
  • 11:50    |    
    So, as I indicated, this would be for a long-term U.S. macroeconomic data, but when you see how I get the results, I think you can think about how this mightn apply in the context of other countries.
  • 12:02    |    
    There are some issues that arise that make actually the U.S. long-term data, a particularly good place to isolate the relevant empirical effects.
  • 12:14    |    
    But then one could also think about where else could this analysis be applied; I just got a note from somebody at the World Bank saying that they´ve beenn assigned this topic.
  • 12:24    |    
    The World Bank basically wants to advice countries, I suppose the IMF also, telling them that stimulus packages are a good idea, so he was assigned the taskn of coming up with empirical evidence that would support this.
  • 12:38    |    
    Now he sort of said to me that they already know what they want to advice, what the answer is, that somehow these stimulus packages are a good idea so he isn supposed to, as he understood his directive, to find supporting empirical evidence for that.
  • 12:51    |    
    But this is a serious guy, and I know that if he´s gonna do this project, he´s gonna do it in the real way and he´s gonna try to come up with what he regardsn as evidence but, like Charles Redlick and myself, he is surprised that there isn´t a lot of existing evidence that looks convincing to rely on, to come up with this important policyn implications.
  • 13:13    |    
    The research is gonna look first at effects, so the depending variable, the thing that I´m gonna be trying to explain is real per capita gross domesticn product.
  • 13:22    |    
    And I´m interested on how that responds to certain government interventions, particularly on the fiscal side here. So the things that we´re especially goingn to look at are, first off, real per capita government purchases, and that´ll be the thing that looks like Keynesian multipliers, you know,
  • 13:39    |    
    usually it´s a multiplier, but it´s bigger than one, and we are gonna look at what the numbers end up being. It´s gonna be important in our research thatn we´re gonna distinguish defense purchases, which of course are all from the central or the federal government, I guess everywhere, but in the U.S. data in particular.
  • 13:58    |    
    And then we are gonna have other government purchases which are both, federal and state, and local, so it´s gonna be some total of non-defense governmentn purchases.
  • 14:09    |    
    Now the programs that are proposed are not really about the military, like the Obama administration is certainly not proposing a big expansion in then military expenditures, and as far as I know they are not planning any wars, which would be related to maybe having a lot more military spending.
  • 14:26    |    
    They´re proposing a big increase in non-defense government purchases and they´ve already put some of that into effect, a lot of it looks like what they calln infrastructure investment.
  • 14:36    |    
    However, it´s going to turn out to be the case in the data, this will be the long-term U.S. macro data, it´s gonna be a lot easier to isolate the effect onn GDP from the military purchases, and I will talk about why that´s true, and it´s gonna be much harder to infer from the data what the multiplier is from the non-defense part, which is in factn what they are concentrating on.
  • 15:00    |    
    So, maybe we can get back to that after I look at some results.
  • 15:03    |    
    But anyway, we´re gonna distinguish the defense part of government purchases from non-defense. Now, at the same time, I´m also gonna be looking at the taxn side, and how changes in taxes affect production; the real per capita gross domestic product.
  • 15:23    |    
    Taxes are gonna appear in two ways in this analysis; first, in terms of incentives or distortions, what´s gonna matter is some kind of marginal tax rate fromn doing something, like a marginal tax rate from the income tax.
  • 15:38    |    
    That´s what´s gonna provide incentives for people that do things like work, or produce, invest , and so on. We´ve constructed some new measures, that I´lln discuss, that try to gage that for the overall U.S. economy; some kind of marginal tax rate time series, which I´ll discuss shortly.
  • 15:59    |    
    You can also look instead at tax revenues, so if you think about fiscal deficits, then that´s gonna be expenditure on the one side and tax revenues on then other side, and of course you can have variations in tax revenues that are different from variations in tax rates that apply at the margin.
  • 16:17    |    
    Often they go together, often if you have a tax cut, you´re cutting rates and you´re also cutting revenue, but that´s not always the case. So, for example,n the biggest tax rate cut in the post World War II period in the United States is the second Reagan tax cut, which were initiated in 1986 and applied particularly in 1987, and 1988, there weren major cuts in marginal tax rates from the individual income tax.
  • 16:46    |    
    But the program overall was designed to be revenue neutral. It was at least designed, with some assumptions; to generate the total of revenue, basically byn closing a lot of loopholes, deductions and so on, at the same time that the rates were decreased.
  • 17:04    |    
    So that´s an example of a change where the tax rates went down a lot, but at least perspective revenues didn´t change. Now usually, like if you take the 2003n tax-cut-program from the second George Bush, that was the cut in tax rate and the perspective cut in tax revenues.
  • 17:27    |    
    So then in that case, which is more typical, they went together. But that´s not always the case. This is just a simple form of this equation for determiningn real per capita gross domestic product, which I´m gonna be using, so there´s nothing very sophisticated in terms of economic theory in here, I´m afraid.
  • 17:50    |    
    This is set up particularly to isolate a kind of multiplier coefficient which is what is so relevant thinking about these stimulus packages such as the U.S.,n one from earlier this year.
  • 18:02    |    
    So little wise real per capita gross domestic product, so the verb on the left hand side is basically the growth rate of real per capita gross domesticn product; from yRt minus 1yRt.
  • 18:16    |    
    Everything I´m gonna be doing here is with annual data, which is gonna be necessary for the long-term analysis but there are other reasons why probablyn something higher frequency than annual is not gonna be too useful here.
  • 18:30    |    
    But this is all gonna be annual, so the period is gonna be one year. So, the variable here is gonna be the variable that picks up the potential multiplier,n which is gonna be this coefficient beta 1.
  • 18:47    |    
    So Gt is real per capita government purchases. In practice, I´ll be thinking about defense purchases, all which is from the federal government, or I cann think about non-defense purchases from the total government sector.
  • 19:02    |    
    But Gt is gonna be real per capita government purchases, so this is the change in the level of real per capita government purchases from one year to then next, and that´s expressed as the ratio to the previous years´ real per capita gross domestic product.
  • 19:20    |    
    Another way to look at this, you can think about this variable as having Gt minus one dividing this, in which case this would be the growth rate of real pern capita government purchases, and then you´d get another Gt minus 1 on top, and that would be divided by the y, so that would be the ratio of government purchases to GDP.
  • 19:42    |    
    So this variable can also be expressed as the growth rate of the government purchases multiplied by how big the government purchases are as a share of then domestic product.
  • 19:52    |    
    But this is set up, if you think about this change in government purchases and this change in GDP, so beta 1 is telling you how much extra GDP you get from an unit change in this government purchases.
  • 20:09    |    
    And that´s why beta 1 is gonna be the multiplier coefficient. So we care about whether beta 1 is positive, is GDP really rise with a bigger government.
  • 20:18    |    
    Work I´ve done on long-run economic growth, suggest than in a long-run, bigger government sectors is probably negative for GDP reducing the growth rate. Thatn would mean here that beta 1 would actually be negative, but anyway, you wanna know is it positive, and then the multiplier case is where is not only positive but bigger than one.
  • 20:40    |    
    And again, the Obama administration in the U.S. is assuming that beta 1 is about 1 and a half. So I wanna know whether the data are consistent with that justn the long-term U.S. data, the same we could be doing for lots of other countries, perhaps.
  • 20:54    |    
    Now tau tis gonna be some measure of the tax rate, and I´ll describe empirically what we were doing with that, ´cause I think a major contribution from thisn research after exactly the time series for the U.S. that we have, which is gonna correspond to this.
  • 21:12    |    
    So if you think about this as a marginal tax rate, so a higher tax rate would be a bigger distortion, that would tend to discourage economic activity, laborn supply, investment, production etc.
  • 21:24    |    
    So the expectation would be that if the tax rate rises from Y(t-1) that GDP would go down, and maybe the timing is not purely contemporaneous, maybe there´sn some lag.
  • 21:37    |    
    So maybe, it´s not the Y tvalue that´s associated with the Yt value here, there might be some lag response to, for example an increase in tax rates.
  • 21:47    |    
    And that´ll turn out to be true actually, in our estimates that when a tax rate goes up, the GDP doesn´t seem to fall immediately as within the year, but itn seems to be more of a delayed response, one to two years, but we can look at the results later.
  • 22:05    |    
    So, there could be other variables and that might include things like tax revenues rather than tax rates. So the Keynesian view doesn´t really focus onn incentive effects from the tax system. It thinks basically about income effects, so if there is more taxes, more tax revenue is going to the government, disposable income is lower, and peoplen consume less and all that stuff.
  • 22:29    |    
    So it´s all about income effects; where this is about substitution effects related to the distorting influences from the tax system. But we could also haven the income effects associated with changes in real tax revenues, rather than tax rates, and as I mentioned, they don't always move together, or at least not perfectly.
  • 22:47    |    
    We might have some other variables, so I´ll discuss that; but these are the kinds of fiscal influences that I´m gonna try to isolate, and as I mentioned isn somewhat unusual in terms of research, they seem to have very current application to this policy interventions, not only in the United States, but those being advocated, for example, by then IMF, and I guess the World Bank is trying to get into that business, I don't know who else.
  • 23:11    |    
    So, it´s already noted, beta 1 is the Keynesian multiplier, we would´ve thought after all these years he would have had a better idea of what all this is,n and another thing I wanna look at, which has a lot of currency these days, is; is it true that beta 1 is higher when the economy has more slack, which gauged by the unemployment rate?
  • 23:33    |    
    So, a lot of people assume that that´s true but I know of no empirical evidence on that but that´s one of the things I wanna to look at. And as I mentioned In expect beta 2 less than zero, that´s sort of distorting the effects relating to the tax system, but it might be picked up.
  • 23:50    |    
    Now let me talk about what the key empirical problem is in implementing this, and here I think is where the application to other countries might comen in.
  • 24:00    |    
    So suppose instead of g, I´m trying to figure out what´s the effect of the government buying up more goods and services, on the total of gross domesticn product.
  • 24:16    |    
    So supposed that instead of gthere, I had valuated from Wal-Mart, I had how much output was produced, how much valuated by Wal-Mart in period t, and I putn that on the right hand side, and I have some theory that when Wal-Mart feels like producing more, it contributes to the aggregated economy.
  • 24:35    |    
    I think you would say that that was probably silly, it´s probably the case that what Wal-Mart is doing is responsive, mostly responding to what is happeningn in the rest of the economy.
  • 24:46    |    
    So most of the response you would guess would be reverse causation, if I picked out a single company and put it here instead of g, you might get pretty goodn explanatory power and I haven´t done the Wal-Mart thing.
  • 25:00    |    
    I have a feeling that if you put the change in Wal-Mart valuated on right side, you get pretty good explanatory power for GDP, and certainly if you putn enough corporation it would fit very well, but what does it mean?
  • 25:11    |    
    Most of this is when GDP goes up, then whatever it was that influenced GDP to go up, is also going to influence Wal-Mart to produce more goods and services,n ´cause that wouldn´t be very meaningful to stick that in the right hand side, and the question is why are government purchases different.
  • 25:29    |    
    This is about how much the government is doing in terms of something about using goods and services to build bridges or to fight wars or whatever, so why isn it different from Wal-Mart?
  • 25:40    |    
    I think it is the essential problem here. So formally, the problem is that the variable g, might be endogenous with respect of real GDP, and then would worryn about reverse causation from GDP to government purchases, rather than try to isolate, which is an exogenous increase in government activity and how that affects the total of goods and services,n which is a GDP.
  • 26:06    |    
    This is why the distinction in the long-term U.S. data, between defense purchases and non-defense purchases is important, because if you think about defensen purchases, particularly those driven by war and peace, they´re all very likely to be exogenous with respect of what´s going on with the real gross domestic product, at least it´s much moren convincing.
  • 26:29    |    
    Especially if you look at the variations associated with World War II, which are by far the biggest variations in government purchases; it´s fairlyn convincing that most of that did not have anything to do with responding to the GDP, most of that was for some exogenous event from having to do with the war.
  • 26:46    |    
    So it´s not that I particularly wanna look at wartime spending and see how it affects GDP, but in terms of this identification issue, and getting somen exogenous variation of the thing I wanna study, that´s why this scenario might be particularly compelling.
  • 27:03    |    
    So, I´m gonna be in that sense focusing on variations in defense spending, especially those associated with wars and figure out what the multiplier beta 1isn in that circumstance.
  • 27:14    |    
    Now, when I wrote this column in the Wall Street Journal I observed that many macroeconomists in the U.S. think that World War II finally got the Unitedn States out of the last phase of the Great Depression.
  • 27:27    |    
    So, of course the war starts in 1939, but the U.S. activity and expenditure, doesn´t really start until 1941 in a big way, but it´s still true that then economy has a lot of slack, the unemployment rate in 1940 is over 9%, still not nearly as high as it was in 1933, but still, quite high.
  • 27:49    |    
    So, many macroeconomists believe that the big expansion of government defense purchases starting in 1941, was responsible for getting us out of the last partn of the Great Depression, and reducing the unemployment rate from 9% to 5, to 3, to almost nothing in the peak of the war.
  • 28:08    |    
    No,w in order for wars to be a good arena to isolate the multiplier, you need something else; so, particularly for people who were here yesterday, I talked an lot about how the history of world macroeconomic disasters is focused just proportionally on war time.
  • 28:27    |    
    Most countries, for instance, during World War II suffered greatly because of destruction of capital, and people and so on, and that was a big negative onn gross domestic product for those places.
  • 28:39    |    
    So, that would not be a good candidate for looking for a Keynesian multiplier, because there is another effect on the supply side there, if you aren destroying a lot of factories and workers and all that stuff, so that would not be a good place to look, to try to isolate this Keynesian demand multiplier.
  • 28:59    |    
    But the United States is sort of perfect, because basically you had this vast expansion of purchases of goods and services, and had no destruction ofn physical plan domestically, and even the number of people who got killed is, actually, not that big for the U.S. relative to the population.
  • 29:17    |    
    So in order for this to be a good experiment you need this big increase in government purchases associated with wars, but you don´t want a lot of then destruction that accompanies many of the wars in history, such as those that I highlighted in the discussion of rare events and disasters that I talked about yesterday. So, this works for then U.S., but would not be real good for Germany.
  • 29:42    |    
    I would not be doing this analysis for Germany, and particularly in terms of what happened in World War II. You´d have to hold constant to destruction, andn there´d be no way really to do that, so... but for the U.S. it´s ok.
  • 29:56    |    
    Now, the problem is, I might get good estimates about the multiplier for defense expenditures; but there are good reasons to think that non-defense purchasesn will have different effects.
  • 30:08    |    
    Actually, I think the usual argument would be that the effects would be smaller for non-defense purchases, we can discuss that, in which case the multipliern estimate for defense should be an upper bound.
  • 30:20    |    
    I´m gonna look at non-defense purchase also, but this identification issue is gonna turn out to be quite important, this Wal-Mart problem.
  • 30:28    |    
    So, I wanna discuss the data that I´ m looking at, which are particularly data that pertain to tax rates, U.S. long-term data, and information related ton government purchases. Those are two critical kinds of time series that I´m looking at, trying to relate to gross domestic product.
  • 30:48    |    
    So let me talk about the data. So, what I put up here first, is the one that´s related to somekind of measure of marginal tax rates; again, this is gonna ben for the U.S. The basic concept here is gonna be some kind of income weighted marginal income-tax rate that each individual or household faces.
  • 31:13    |    
    So conceptually, if you talk about a particular year, there would be some marginal tax rate imposed on any particular household. And the parts of the taxn system that I´m gonna look at here are the individual income tax, which is either federal or state, and local in the U.S.; but mostly federal, and then the social security payroll tax is alson quite big.
  • 31:36    |    
    So those are the taxes that I´m gonna be looking at. So, any individual in this environment in the particular year has a particular number for what then marginal tax rate is, which is how much you would pay an additional tax if you earned another dollar of income in year t.
  • 31:50    |    
    That´s the marginal income tax rate for a family or a person, gonna somehow aggregate this up across all the households. I´ll be doing that by weighting eachn individual marginal rate by some measure of income; so, it´s gonna be an income weighted average marginal tax rate. That´s the series that we´ve constructed, which I think is sort of aggregaten summary of distortion from the government with respect to income taxation.
  • 32:17    |    
    So, some parts of taxation are not in this, but it does have a large part, particularly of the federal government, it has most of the federal revenues,n actually.
  • 32:25    |    
    Now, I originally carried out an analysis like this, back in the 1980s, and at that time we constructed a series back from when the income tax started inn 1916, I guess, up to the mid 1980s.
  • 32:42    |    
    So, what we´ve done here is modified the procedure and updated from the mid 1980s to 2006. And, it´s actually easier to do now, because the National Bureaun of Economic Research, in Cambridge, MA, not too far from Harvard, has a whole program for simulating the U.S. tax system where you can figure out things like average marginal tax rates.
  • 33:07    |    
    So, we were able then, to update from 1986 to 2006, using NAT, and also include state and local income taxes, which we hadn´t been able to do before, and wen also included the social security payroll tax, which is pretty large.
  • 33:24    |    
    Here is the answer, this starts in 1910, but basically the marginal tax rate is zero in 1910, it´s before the federal tax system.
  • 33:32    |    
    You probably don´t know or care about this history, but the U.S. tried several times earlier in history, to implement the income tax during the civil war,n and I think it was in the 1880s or 1890s, but the Supreme Court said it that was unconstitutional, that you couldn´t have this form of direct taxation because the Constitution didn´t providen for it.
  • 33:54    |    
    This was at a time when there was more serious nature in terms of rules in the U.S., and if it didn´t say so in the Constitution, it was likely to ben described as unconstitutional.
  • 34:06    |    
    So, that got rid of the tax that´d been in effect, I think it was in the 1890s. Then when Woodrow Wilson was president, he wanted to expand the government,n and it was clear that in order to do that, he needed a broad-based tax system, so he put in the income tax and he got a constitutional amendment passed, in order to do that.
  • 34:26    |    
    So that´s when the Federal Income tax started, I think 1916 was the first year it was actually put into effect. There are some state and local income taxes,n that were earlier; but basically, it´s zero in 1910. And the social security payroll tax starts in 1937, so that´s later.
  • 34:44    |    
    So, conceptually this is adding up, the federal individual income tax, state and local income taxes, social security payroll tax, and then doing this averagen marginal construction, based on individual incomes you used as weights.
  • 35:00    |    
    Now, one part of this calculation you might complain about; the social security payroll tax, is being viewed as a tax, but to some extent it is true that then promised pension benefits you get, depend on how much you contributed to it during your work history.
  • 35:16    |    
    And to the extent that you´re getting offsetting private pension benefits because you work and pay more social security tax; that´s not really a tax, son there could be some subtraction there.
  • 35:28    |    
    Of course we didn´t know how to measure that, and we haven´t done anything with that so this views it just as another form of tax, but collected as a payrolln tax. It includes both, the part paid by workers directly; and the part paid by employers, so it´s basically fifty-fifty. But we don't think that that´s a meaningful division in terms ofn distortions.
  • 35:51    |    
    This series picks up the major events which people have described qualitatively. So for example, these are the Reagan tax cuts, the first one is the 1981 ton 83; and the second one, is implemented in 86 and then it comes from 86 to 88.
  • 36:09    |    
    And as I mentioned that´s, in some ways the most, at least in the post World War II period, that´s the biggest marginal rate cut. So the number on then vertical axis is the affective average marginal tax rate, so 0.3 means 30%.
  • 36:23    |    
    So, I should say in terms of the level, the average marginal tax rate today is about 36% implied by this construction. There are some issues that arisen because the marginal rate cuts that were put into effect in 81 were phased in; there was a planned cut from 81 to 84, initially
  • 36:42    |    
    And if you think about it, it may not be such a good idea, to phase in a tax rate cut. So you´re saying that I´m gonna to cut tax rates by 5 percentagen points, but I´m only doing that percentage point right now.
  • 36:57    |    
    So, that´s telling people that the rate today is higher than it´s gonna be next year or the year after, that might not have such good implications for then economic activity.
  • 37:06    |    
    And indeed, some people have claimed that this phasing was a negative; there was a recession in the U.S. in 81, 82, and some people claim that the phasingn may have had something to do with that, I´m not sure.
  • 37:20    |    
    But anyway, so this is the 86 when, these are the two, there are two Bush-tax cutting programs, recently 2001, 2003; but the 2003 one had a much biggern marginal rate effect, so it´s a bigger reduction on the marginal tax rate here.
  • 37:39    |    
    The Kennedy-Johnson tax cuts in the mid 1960 are often mentioned also; of course these are Democrats, unlike the two previous cases that were from then Republican party.
  • 37:51    |    
    So this is a pretty substantial marginal rate cut associated, originally proposed by President Kennedy, and then implemented by Johnson after Kennedy died.n That´s this one here.
  • 38:03    |    
    There´s this one under Nixon, the main change there was that the top rate on earned income, used to be 70% just before this, and they reduced the top taxn rate on wages and some other so-called earned incomes to 50% in that legislation, so that´s another cut.
  • 38:24    |    
    If you look at where are the big increases in tax rates, well the biggest is during World War II, this is tremendous increase, ´cause before World War II,n the income tax didn´t apply for most of the population at all. So that´s in this period, and then there´s this tremendous expansion of the scope of the income tax system.
  • 38:46    |    
    As part of that they introduced income tax withholding, I don´t know if you know this story, but Milton Friedman was involved with doing that, he was veryn embarrassed by this fact.
  • 38:59    |    
    Of course, Milton was very good in a technical sense, in designing this program, which I think was run by Ken Gailbraith, I believe that that´s true, he wasn in the same office, but then Milton has written to the effect, that you can´t possibly raise a lot of revenue from an income tax unless you have withholding.
  • 39:18    |    
    It´s just too hard to get people to pay after the fact, and that was just the system in the U.S. until 1943, when they introduced tax withholding, whichn mechanically is extremely effective at raising money, raising revenue.
  • 39:32    |    
    But then of course, Milton went out to say that therefore you could have a much bigger government, ´cause it was easier to finance, and that´s what happened,n and he said well that was a big mistake, I shouldn´t have been involved in that enterprise.
  • 39:44    |    
    For the big tax increases first of the World War II, there´s a very large increase in the average marginal tax rate from 1971 to 1981. That´s not because ofn the legislated changes raising tax rates, it´s because the system was not indexed, and just rising prices and rising real incomes put people into higher rate brackets.
  • 40:10    |    
    And now, it´s a very big effect because the marginal rate had been about 28% in 1971, and it rose to a peak of 42% without legislated changes, being veryn important over that period. Since then the system is much more indexed.
  • 40:28    |    
    There were some celebrated tax increases in the early 1990s; it was thought that the first George Bush lost the election in 92, because he went against hisn pledge "Read my lips: no new taxes," by raising taxes; but actually is pretty moderate, is not big.
  • 40:46    |    
    And then Clinton, in 1993, had a tax rate increase, which he had promised during the campaign, and he fulfilled that promise. But overall the tax raten increases in the 90s were moderate, they weren´t very large.
  • 41:01    |    
    There´s also a big increases in taxes associated with the Korean War, which is over here, after decreasing after World War II. And finally, it´s often beenn remarked, in terms of the policy of Franklin Roosevelt, in such that they actually raised tax rates during the Depression, which is supposed to be sort of the height of crazy governmentn policy.
  • 41:23    |    
    So, if you look here the Depression, the worst of the Depression is really 1932, 33, and tax rates were actually rising in that period in some kind ofn attempt to balance the budget. So first there was Hoover and then, there was Roosevelt, came in at 1933.
  • 41:42    |    
    So, this is one piece of the data, so I´m gonna to try to see what difference does it make for the macro economy if you do things to this tax rate, like then Reagan tax cuts, or Bush, so that´s the central question.
  • 41:57    |    
    The other major variables that I´m looking at, for this fiscal study is government purchases, distinguishing defense from non-defense, so let me show youn what those data look like.
  • 42:13    |    
    Okay, so the blue graph is the change in defense purchases, real per capita, expressed as a ratio to the previous years, real per capita gross domesticn product; that´s the relevant variable that will be there in the multiplier analysis.
  • 42:28    |    
    So, you look at how much did you increase defense purchases over the year and express that as a ratio to the level of GDP basically, with some adjustment forn population which isn´t too important.
  • 42:44    |    
    Well, you can see why wars are critical here, in terms of doing an experiment that might allow you to isolate some effect, it´s almost like a perfectn randomized experiment, to try to see what´s the effect i of a terrific increase or decrease in these purchases on the economy.
  • 42:59    |    
    So, the dominant thing obviously is World War II, which is this. So this starts in 1941; interestingly, you would´ve thought that the U.S. would´ve had moren of a buildup in 1939 and 1940, but it´s not much in the data.
  • 43:13    |    
    There´s actually more of a buildup at the beginning of World War I, which starts in 1914, but the U.S. doesn´t get in until 1917. There´s more of a responsen before 1970, than there is here before 1941. Although, it doesn´t wait until December of 1941, it starts earlier, it shows up very heavily during the year.
  • 43:35    |    
    So, the first observation here would be 1941, that´s big, and that´s particularly interesting for this exercise because as I mentioned the unemployment raten is really high, so this is the effect of the government coming in and spending a lot on stuff when the economy has a lot of slack.
  • 43:54    |    
    That´s important for some of this analysis. And there isn´t too much data related to that issue. So this accumulates through 1943, 44 is still high, and thenn 45 is a reduction, it´s not merely so much.
  • 44:10    |    
    And this big negative here is the demobilization after the war, where there´s a tremendous decrease in government purchases for the military.
  • 44:18    |    
    So, particularly in 1946 there´s a tremendous reduction, which is like minus 26% of the GDP, of the previous years´ GDP.
  • 44:26    |    
    So you can see the peak numbers are something over 20% of GDP, is the increment in defense vending in the one year relative to GDP, so there´s a tremendousn impulse.
  • 44:37    |    
    So the main other things that show up here, World War I is also big, that´s this; and again, there´s a working down a demobilization after the war. Then Korean War is big, not as big as the World Wars, but the Korean War is very noticeable, something like 5 or 6%.
  • 44:58    |    
    So, that´s something you can hope to use to isolate the multiplier, and there´s some demobilization afterwards. And you can see if you look at the whole restn of the series, the blue, there´s not much.
  • 45:12    |    
    And maybe, surprisingly that includes the Vietnam War, which is actually pretty trivial. There´s a little blip up in defense purchases for the Vietnam War,n but it´s quite minor, and it actually becomes negative, while the war is still going on, that´s this guns and butter business in the U.S. where we´re trying to have the fight in Vietnam War,n but not compromising anything else.
  • 45:34    |    
    And part of that was that they really didn´t have much of an increase in defense expenditure. And you can see the rest of it, is basically nothing. So youn can try to isolate what´s the effect of defense expenditures on GDP, it´s gotta be the wars that need the focus. There´s a little bit of other stuff but not much.
  • 45:56    |    
    So the red line is the non-defense purchases relevant, so it´s a change in non-defense purchases relevant to GDP? So let me tell you what´s the biggestn observation there is this here, this is the New Deal period.
  • 46:13    |    
    So a lot of Keynesians these days, are talking about this, this is 1934 through 1936, under Franklin Roosevelt, where these tremendous "expansion" of variousn kinds of infrastructure activities, the building bridges or putting up parks or whatever. So you can see it here, it´s about 3% of GDP, at the max, so very small relative after war timen experiences but not negligible.
  • 46:45    |    
    So, the usual claim is made by people today in supporting the stimulus packages; this was a great idea we just didn´t do enough, we needed to do thisn bigger.
  • 46:56    |    
    Like Japan in the last decades, which had tremendous infrastructure investment and ran up a public debt; they didn´t do enough, and that´s why it didn´tn work. I don't know when you ever declare failure, if you don't do the government purchases then things go badly, well that´s because you didn´t do it.
  • 47:17    |    
    Then,if you do it and it still goes badly, it´s because you didn´t do enough. That´s the load that the Obama administration is in these days. I mean, non evidence is ever gonna say: This fiscal package was a mistake.I don't know what data would give you that answer according to them.
  • 47:32    |    
    So the only other things that´re pretty big in the data, are associated with the war; so for example, during World War II, there´s a reduction in other typesn of government expenditure, so that´s a negative, and then it comes back up after the war.
  • 47:50    |    
    So more or less like other activities, the government is absorbing all kind of resources for the war, other uses of goods and services by the government thatn don't involve the war, tend to be crowded out, that´s why there´s this negative, and then afterwards you have this buildup in demand for non-defense government activities; so, there´s somen increase here and then World War I has a similar kind of pattern.
  • 48:11    |    
    If you look at the rest of it, you can see there´s not much there. It´s going to be pretty tough to isolate the effect of these variations in non-defensen purchases on GDP.
  • 48:22    |    
    And there´s two reasons why this is a problem, here´s this discussion over here. I really wanna know what is the effect of an extension of this non-defensen government activity, which is what is particularly being designed in the current fiscal stimulus package.
  • 48:37    |    
    There´re two reasons why it´s gonna be very hard to take this out of the data; one, is the variations are relatively small, particularly compared withn defense; and especially, if you don't have the New Deal period, those are the biggest, but even those are not that big.
  • 48:53    |    
    The second problem is the Wal-Mart problem; the variations that occur are probably, to a large extent, reactions to the economy, the GDP rather than then reverse.
  • 49:04    |    
    So, for example, if you look today in the U.S., you look at California, which is in a very difficult financial situation, they´re suffering from the badn economy and partly what they´re doing, maybe not enough, is cutting back on government expenditures.
  • 49:19    |    
    So that´s the response to the economy, they´re cutting back on outlays for education and other things. So you wouldn´t wanna say that the reason that then California economy is doing badly, is because they cut back on government expenditure, mostly, it´s a response to the economy, it´s endogenous.
  • 49:34    |    
    So that´s the second reason; that´s hard because I think most of the movements that do occur in non-defense purchases, are probably endogenous; and moreover,n they´re not anyhow that big. So there´re two problems in isolating the non-defense part.
  • 49:49    |    
    So let me talk about some results that we have, so these are estimates of this equations that I had earlier somewhere. This simple equation, everything is inn this form, the dependent variable is basically the growth rate of real per capita GDP, from t-1 to t.
  • 50:08    |    
    Everything is annual data, but one thing I forgot to say in terms of annual data, the tax rates really are annual. For an individual the tax rate you payn doesn´t depend on when, during the year, you earn the money; the tax rate really is an annual variable.
  • 50:25    |    
    So, it doesn´t make too much sense to have like a quarterly series on marginal tax rates, there´s something wrong with that, so that´s one reason why annualn date is probably more relevant, but there are other things that come up.
  • 50:38    |    
    Anyway, so the dependent variable is the growth rate of real per capita GDP. Then I got these various variables on the right hand side, and the governmentn purchases variables entered in the form that´s specified there.
  • 50:51    |    
    It´s kind of important which time period you are looking at, if you remember what the data look like for government purchases; the war time stop it´sn absolutely critical for the defense spending. So it´s gonna matter a lot, do you have the wars as part of your time series, and if you start in 1954, pretty much you don´t have any wars,n because Vietnam isn´t big enough.
  • 51:11    |    
    So, starting in 1954 really means just after the Korean War and pretty much you don´t have much variation in the military purchases.
  • 51:19    |    
    So in terms of these starting dates, this is the sample after the Korean War, which is not much military action after that, in terms of ofn quantitatively.
  • 51:32    |    
    I mean, the U.S. is always fighting a war somewhere, but it doesn´t show up in these numbers if things have any consequences. So in 1950, it means, I´ven concluded the Korean War; 1946, means you have the aftermath of World War II; 1939 means you have all of World War II in the data;
  • 51:49    |    
    1930 means you´re including the Great Depression observations, and finally; 1917, means it concluded World War I, at least the part where the U.S. isn present; 1917 to 18, and also, there´s a pretty big economic contraction, with the drop in 1921, but that´s also included when you start in 1970.
  • 52:11    |    
    But those are the different sample to be considered, the data always go up in 2006. Now 2006 is the last data at the moment to which we have the averagen marginal tax rate series, but I believe any day we will be able to update that to 2007, but there is some lag in information being provided.
  • 52:30    |    
    This first variable is this government purchases, in the form that I mentioned before, is the change in purchases divided by GDP, the coefficient on that isn supposed to be a multiplier type of number.
  • 52:44    |    
    This is just for the defense purchases, so particularly over the longer samples that are driven mostly by war and peace, and World War II is particularlyn dominant when World War II is part of the sample.
  • 52:55    |    
    I have some variables which is the lag of unemployment rate, which is intended to incorporate some kind of normal business cycle adjustment.
  • 53:05    |    
    So, the idea is that if the unemployment rate is high, it tends to be followed by an economic recovery, which shows up as higher the normal growth of realn per capita GDP, that´s what I expect. So if that´s the holding fix another influence is to try to isolate the fiscal effects, so that's what this variable is.
  • 53:23    |    
    This third variable is this question: Does the multiplier depend on how much economic slack there is in the economy?And it's hard to isolate that in then shorter time series, particularly in the post World War II data, you don't have enough variation in the government purchases variables.
  • 53:46    |    
    But over the longer term I have some information, so the idea here, is this is gonna be the multiplier coefficient, and this is gonna ask is that coefficientn affected by the level of the lag unemployment rate.
  • 53:59    |    
    So the usual view is that when there's more slack, like in 1940, when the unemployment rate was still high, the response to a given dollar purchases inn regard to GDP is supposed to be greater, and if that's true, the coefficient in this variable should be positive.
  • 54:15    |    
    Now, this is set up, so that this variable is measured as each of its two pieces are deviations from their own mediates, so the unemployment rate here is,n actually, entered relative to the median over the long term, which I think is 5.7% or something.
  • 54:31    |    
    The point of that is that this variable, the coefficient on it, is gonna represent the multiplier when the unemployment rate is at its median, which is 5.7%n in the U.S. case.
  • 54:43    |    
    So that would be sort of the typical multiplier when you have a typical level of slack in the economy, when the unemployment rate is at its median, and thisn variable will tell you if the unemployment rate is above or below the median, what happens to the multiplier, that's what that will pick up.
  • 55:00    |    
    And then, the contention in the current context, is because we have a lot of slack, the multiplier is bigger than average, to the extent that I can assesn that it would be from this road.
  • 55:12    |    
    This variable is the change in that average marginal income tax rate that I showed you. Tauis the symbol for the average marginal tax rate; it turns out thatn with the biggest effect is with the one year lag.
  • 55:27    |    
    So, this is the change upon one year earlier and then the growth of GDP in the following year. That turned out to be what matters most in the data. I don'tn have any theory of why it should be particularly a one year lag.
  • 55:40    |    
    This next variable is something that comes up, particularly about the observations related to World War II. If you remember the tax rate data during Worldn War II, there' s a tremendous increase in the marginal tax rate.
  • 55:53    |    
    The results suggest that in that environment the response to the higher tax rate is not nearly as big as it usually is. Now one reason for that is that then government is commanding more of the use of resources during World War II.
  • 56:10    |    
    The most extreme example is they drafted millions of people and told them to go work as members of the military, and then, the fact that they are facing an higher tax rate, doesn't matter for their work, or at least for the fact that they were drafted.
  • 56:24    |    
    But, more generally, the government is commanding use of a lot of resources and not using the market system, at the extent they're doing that, I expect then effect of the tax rate to be weaker. And that's what I'm looking for in that variable.
  • 56:37    |    
    Finally, I'm trying to hold constant some influences related to the state of the credit markets, and monetary policy could come in here, and this isn particularly important, if we're trying to incorporate the Great Depression data into the sample.
  • 56:53    |    
    So the measure I've used about the state of the credit market, is this interest rate spread between some form of corporate bonds, which are DAA rated bondsn versus U.S. treasury bonds.
  • 57:07    |    
    So the idea is, that when things are going bad, like for the last year, various spreads get much higher, there's a bigger distinction between the risky andn the low risk stuff, that is between the corporate and the U.S. treasury, and that's thought to be a negative influence on economic activity.
  • 57:27    |    
    So, I really, I don't, for the purpose of this project, I don't care so much about the variable while I'm trying to hold constance of other things in ordern to isolate the effects of the fiscal variables.
  • 57:40    |    
    So, this is one possibility for doing that, which turns out to be particularly important during the Great Depression. So, those are all the variables, so letn me give you the nature of the results.
  • 57:53    |    
    The way these equations are estimated is all mostly squares, the only difference is that the interest rate spread is a variable you would think as beingn endogenous.
  • 58:06    |    
    That is to some extent, when the economy is doing worse, it's probably gonna show up on credit markets as higher spreads. So to hold that constant, I'm usingn as an instrument for this, the lag value of the spread.
  • 58:18    |    
    So, aside from that instrument, which is just the lag value of this one variable, all the other variables are entered into the system, and it would ben ordinarily these squares, except for this one minor income.
  • 58:32    |    
    Now the most important part of that is the assumption that the contemporaneous change in defense purchases is exogenous with respect to GDP.
  • 58:41    |    
    And I think that's most convincing for the big war time variations, so that'll be an issue. But I'm treating this as exogenous and that's the particularn question, if I had non-defense there, if I had Wal-Mart, I think you should complain about that and I think defense is special in that regard or that's the whole idea, anyway.
  • 59:04    |    
    So, let me show you what these coefficients, these are the coefficients on defense purchases, so these are estimates of the multiplier, associated withn defense outlays, much of which are driven by war and peace. But not a lot of domestic destruction from war, that's important.
  • 59:21    |    
    If we start in 1954, the number does not have much variation in that variable, mostly Vietnam, and then a little bit of other stuff. The estimate of an coefficient that's positive, but is actually statistically insignificantly different from zero.
  • 59:39    |    
    So this will be a multiplier on the half, but completely unreliable if that were the only data. If you start in 1950 and put in the Korean war, which isn pretty big, up and down, you get a point estimate which is not so different, but now statistically significant.
  • 59:57    |    
    That's because the Korean War has a lot of information, about these "exogenous" variations in these types of government outlays.
  • 01:00:06    |    
    Similarly, to include 1946 or 1939 in particular, that puts in the whole World War II experience; so the World War II experience is not just onen observation, you go through the whole year from 41 to 46.
  • 01:00:22    |    
    You got all kinds of different observations, in 41 you got this big increase when the unemployment rate is still high, then you got even larger increases,n and they slow down and there's a big negative number in 1946, and that's all important data telling you about the effect of government purchases.
  • 01:00:37    |    
    So now, you see the point, estimate is not too different, it's around 0.6, but the standard error is way down, that's because you get a lot ofn information,
  • 01:00:48    |    
    from the variation in defense purchases from World War II. The estimate becomes much more precise, so if you are one to assume that the same numbern applies, including the war experience, you can really pin down the multiplier associated with defense purchases.
  • 01:01:04    |    
    So that's this 0.6 number here. I´ll talk about this other stuff in a minute, but you can see the rest of these numbers are also similar. But with then whole sample, which includes World War I, you still have around 0.6.
  • 01:01:19    |    
    So that's a multiplier significantly positive; it's bigger than zero, and it's significantly less than one. So, it's certainly way less than 1.5, which isn the Obama team number.
  • 01:01:31    |    
    The unemployment rate lag works the way you would expect, a higher unemployment rate means more potential to recovery; you get higher growth.
  • 01:01:43    |    
    I don't care about that per se, I'm just trying to hold that concept to isolate the other things. Let me forget about this variable for the moment, youn can't really isolate that just from the Post World War II data, but I´ll come back to that.
  • 01:01:55    |    
    This is the change in the marginal tax rate, so this is also a great interest, the usual supply side view, from distortion stand point, is when the taxn rates go up, the economic activity will go down, and that effect shows up here as being significant for a one year lag; that's this coefficient here.
  • 01:02:14    |    
    So -0.6 means that if the marginal tax rate goes up by one percentage point, 0.01, the growth rate of GDP goes down by 0.6 of the percentage point pern year.
  • 01:02:28    |    
    This is not exactly like a multiplier, but it's something that connects the tax rate, which is like 0.4 or 0.3 or a recent number, a change in that ton this response of GDP, in a proportionate sense.
  • 01:02:46    |    
    So this is significantly less than zero. Now the irony of this is that Obama's Chair of the Council, Christina Romer, has a study out there for the postn World War II period, using a different tax measure, but finding that higher taxes are negatively related to GDP, I´ll talk about her findings later, but that's about taxes not about spending,n and she was sort of co-opted by the administration.
  • 01:03:19    |    
    The administration wants to focus on spending, not tax cuts. So she had to start saying that we´ll see how spending is more important than taxes.
  • 01:03:29    |    
    But her empirical results, which haven't even been published yet, they're very current, are actually about taxes and are actually not inconsistent withn this kind of a result, but I can say more about that a little later.
  • 01:03:40    |    
    This is this interest spread thing. It's supposed to be bad credit market conditions and it does have a negative effect, but I think mostly this isn important try to incorporate the Great Depression in the sample, where this spread is enormous.
  • 01:03:54    |    
    Okay, so this is starting in 1950, I mentioned, we have a much more precise estimate of the government purchases multiplier, the other results are not son different.
  • 01:04:07    |    
    So, starting 1950, 1954, you get very similar tax rate effects. If we start in 1946, the tax rate effect becomes much less clear. I´ll try to discuss somen more results in a little more time, we don't have this fully resolved, but I´ll tell you what I know about this.
  • 01:04:27    |    
    But this effect becomes much less clear, so in the post war data, which is also what Christina Romer was looking at, you got these pretty clear negativen responses to higher tax rates in terms of economic activity.
  • 01:04:40    |    
    But in this longer sample, it's less clear cut, so this includes World War II, and as I mentioned before, the multiplier as a similar point estimate muchn more precise, but the tax rate thing is much less clear than it was in this example.
  • 01:05:02    |    
    Now if I start in 1939, so I have this magnificent, "World War II evidence", it's possible to look for more fine results, to look at interaction ofn various things, you got enough variation to be more sophisticated, you can't possibly isolate it in the post World War II data.
  • 01:05:21    |    
    So I´ve included two more variables here, which you just can't estimate, precisely at all if you start later, but if you start in 1939, which means, I'mn including all the World War II variations, which are dramatic, then you can look at two more things; one is this interaction here, which tells you if the multiplier is higher when there's moren slack.
  • 01:05:43    |    
    And indeed, it's true here, this is significantly positive. So, let me tell you what this coefficient means. So, this is in terms of the unemploymentn rate.
  • 01:05:45    |    
    So, if you think about raising the unemployment rate above its median, 5.7% or 0.057, if you raise it by 0.01 above, it means that you multiply that by 5,n so five times 0.01 is 0.05, the multiplier is higher by 0.05 if the unemployment rate is one percentage point above its median.
  • 01:06:20    |    
    So the multiplier at the median is estimated to be here 0.65, but if the unemployment rate is higher by one percentage point, it would be 0.70. To get then multiplier up to one, in this set of results, you need to raise it by 0.35, above 0.65, which means that the unemployment rate has to be 0.07 above its median.
  • 01:06:45    |    
    So the median is about 6, so you need the unemployment rate to be about to 13% in the U.S., to get the estimated multiplier up to one, for thesen results.
  • 01:06:58    |    
    It shouldn't take these numbers completely seriously, because the identification of this interaction term, is pretty tenuous and it can't be done at alln with these post World War II data.
  • 01:07:11    |    
    And it's really building up particularly, the fact that World War II starts when the unemployment rate is still high. So, specially in 1940, then unemployment rate is 9%, in 1941, the unemployment rate is still 5.5% or something. So, the real information about what difference does it make, is coming from that, and that´s why it's not alln that reliable, but it's enough here to be statistically significant.
  • 01:07:37    |    
    To get the number up to 1.5, you need the unemployment rate to be something like 20%. I don't know it would really hold over that whole range, but that'sn what you´d get.
  • 01:07:56    |    
    So, I don't know whether Paul Krugman would be happy with this result, it does have the sign that he wanted. ´Cause when he rants in his column, he talksn about how any idiot knows that when there's slack the multiplier is really big,so it does have the right sign for him, but he probably wouldn´t like the magnitude we talk about.
  • 01:08:14    |    
    This is another interaction term which you really can't isolate without having this longer data that has World War II in it, so this is looking at then question;
  • 01:08:27    |    
    Is the response of the economy for the tax rate, is it weaker when the government is commanding a lot of stuff, such as enduringWorld War II, and I´ven measured that by the share of defense outlaid in the GDP?
  • 01:08:40    |    
    This is really an imperfect proxy, but the idea is that defense purchases rise enormously during the war and the government, along with that, isn commanding a lot of the resources, is not just relying on the market.
  • 01:08:53    |    
    Now, I´m taking how big the defense purchases are, is some proxy for how much the government is directly controlling the economy, ´cause, I didn't known how to measure that directly.
  • 01:09:01    |    
    So, the 0.75, what that really means is that the effect of the tax rate, which is negative, is attenuated when the government is commanding more of then resources, that's what the sign means.
  • 01:09:15    |    
    That the tax response is weaker and that's actually what the World War II experience looks like; it is this tremendous increase in the marginal taxn rate.
  • 01:09:24    |    
    Going back to 1930, it means you include the Great Depression, and it's putting a lot of pressure on this model, because this model doesn't really haven the capability of explaining the Great Depression exactly. There's something about that that's not being explained by fiscal variables for sure.
  • 01:09:43    |    
    The one variable that's doing most of the work for the depression is this interest rate spread, which is astronomical, and credit markets are basicallyn not functioning in the early 1930s; so, not only is this coefficient negative, but to fit the Depression it´s trying to put a lot of weight on this.
  • 01:10:05    |    
    That's why the magnitude of this coefficient became so big; we're looking at some different functional forms trying to see if there's some way to get thisn in a way that's stable over the different time periods.
  • 01:10:16    |    
    Because obviously, this is not the same coefficient as the one that applies here, but I think is that you're getting out into a range where the creditn spread is so large, having a particularly negative effect on economic activity, and this is just a linear form. And I think the linear form isn't right.
  • 01:10:35    |    
    But we're looking at that. Anyway, that's what goes on during the depression; it doesn't affect too much the things we really care about, which is then fiscal estimates.
  • 01:10:48    |    
    So, the multiplier is basically the same as it always is. The tax stuff looks like it's negative, but this is not statistically significant by itself,n this interaction term is also not significant all by itself, so there's some tax stuff but it's not... it's hard to isolate it, because there's so much else gonna during the Great Depression,n this isn't definitive.
  • 01:11:13    |    
    And then this is what happens when you incorporate World War I. So, this is basically a similar set of results, the credit spread is not as big inn magnitude as it is here, and I think we can probably figure out a better specification, and it makes it look more stable over the different time period, but I'm not sure about that.
  • 00:00    |    
    Initial credits
  • 00:20    |    
  • 00:42    |    
    Keynesian multipliers
  • 01:28    |    
    Keynes's theory
    • Basic idea
    • US multiplier effect
    • Multiplier variable
      • Multiplier bigger than one
      • Reagan administration
      • Idle resources
      • Private market economy
      • Wages and prices
      • Cost-benefit analysis for government activity
      • Macroeconomic context of multipliers
  • 10:0.14999999999997726    |    
    Empirical multipliers
    • Empirical evidence
    • Long-term US data
      • Government purchases
      • Real per capita government purchases
      • Effect of military purchases on GDP
    • Taxes
      • Tax revenues
      • Tax cuts
  • 17:34.5    |    
    GDP equation
    • Tax rates
    • Marginal tax rate
    • Tax revenues
  • 23:50    |    
    Key empirical problem
    • Defense purchases
    • War time effects
    • Non-defense purchases
  • 30:51    |    
    Marginal tax rate
    • US average marginal income-tax rate
    • Social security tax
    • Average marginal construction
    • Social security payroll tax
    • Reagan tax cuts
    • US tax cuts
      • Marginal tax rates 1971-1981
      • Korean war
    • Changes in US defense and non-defense purchases
    • Increment in defense spending
    • War-time government spending
      • World War II
      • Endogenous variables
  • 49:49    |    
    Equations for GDP growth
    • Equation variables
    • Multiplier coefficient
    • Change in marginal income-tax rate
    • Nature of the results
    • Interest rate spread
    • Contemporaneous change in defense purchases
    • Coefficients on defense purchases
    • Unemployment rate
    • Change in marginal tax rate
    • Tax rate effects
    • Multiplier effect in the unemployment rate
    • Interest rate spread
  • 01:10:37    |    
    Final words
  • 01:11:33    |    
    Final credits

Public Spending and the Money Multiplier

New Media  | 17 de julio de 2009  | Vistas: 124

Robert J. Barro looks at John Maynard Keynes’s theory on public spending and the money multiplier effect. Barro uses long-term US GDP rates to analyze the impact of government spending on the country's economy. According to the Keynesian theory, governments spend public money in order to multiply it, thereby creating even more benefits for society. However, Barro refers to the lack of empirical evidence supporting the current US administration’s assumption that the multiplier effect is 1.5. Barro also classifies government spending into defense and non-defense purchases and compares the two. At the same time, Barro’s research also looks at how changes in taxes affect production and the real per capita GDP. According to the long-term data Barro is working with, big events, like wars, can force a state to increase the country's rate of resource absorption.