Wednesday, August 13, 2014

Chris House on stimulus spending

Back in March, Chris House wrote a blog post explaining why he thinks that tax rebates make better stimulus than government spending. He concludes that tax rebates are the best form of stimulus, and that government spending projects should only be undertaken if the projects would pass a cost-benefit test in the absence of any stimulus effect. House:
If a project doesn’t meet the basic cost / benefit test, then it shouldn’t be funded, regardless of the need for stimulus...If the social value of a government project exceeds its social cost then we should continue to fund the project whether we are in a recession or not. If the social value falls short of the social cost then, even if the economy is in “dire need” of stimulus we should not fund it. If we really need stimulus but there are no socially viable projects in the queue then the government should use tax cuts... 
If the direct social benefit of a bridge is $100, then all the government needs to consider is whether the cost of building the bridge is greater or less than $100. If you then tell me that, because we are in a recession, there are additional stimulus benefits from the project (e.g., the workers who build the bridge take their new wage income and buy goods and services from other businesses further stimulating demand, increasing employment, and so on.), the government should exclude these additional benefits from its calculation.
I don't understand this assertion at all. It makes no sense to me.

Let's consider a simple numerical example. Suppose that the economy is in a recession. And suppose that, because of the Zero Lower Bound or whatever, the pure fiscal "multiplier" is substantial. Specifically, suppose that $100 of tax rebates will increase GDP by $110. In this case, stimulus spending is a "free lunch."

Now suppose that instead of doing tax rebates, the government can build a bridge. The social benefit of the bridge is $90, and the bridge would cost $100. In the absence of stimulus effects, therefore, the bridge would not pass a cost-benefit analysis. For simplicity's sake, suppose that spending money on the bridge would create exactly the same stimulus effect as doing a tax rebate - spend $100 on the bridge, and GDP goes up by $110 from the stimulus effect.

In this case, the net social benefit of spending $100 building the bridge is $90 + $110 - $100 = $100.
And the net social benefit of spending $100 on a tax rebate is $110 - $100 = $10.

Bridge wins!

In fact, it turns out that the bridge wins even with a pure multiplier of less than 1! As long as the multiplier is greater than 0.2, in fact, it's worth it for the government to build the bridge. This will mean that the apparent multiplier of bridge-building will far exceed the "pure" multiplier. In this case, the apparent multiplier from bridge-building will be 2.

This fits the results of Auerbach and Gorodnichenko (2012). It also fits with the simple Keynesian theories you'd read in an Econ 102 textbook. It also fits the results of Bachmann and Sims (2011). And if true, it means Chris House is completely wrong.

Now let's relax the assumption that the pure stimulus effect is equal for the two cases. Suppose that government spending creates waste, for example. Or suppose that due to the particular nature of the mechanism that makes stimulus effective in the first place, the people who build the bridge will tend to stick most of their fee in a bank instead, rather than spending it as people would do if they got a tax rebate. Concretely, suppose that due to government waste or reduced stimulus effect, the pure stimulus benefit of building the bridge is only $30 instead of $110 (a pure multiplier of only 0.3 for government spending vs. 1.1 for tax rebates).

In that case, the net social benefit of building the bridge is $90 + $30 - $100 = $20. Bridge still wins! House is still wrong!

The difference between bridge-building and tax rebates can be stated in simple terms. If you give people a tax rebate, they may stick the whole thing in the bank, completely negating the stimulus. If you pay unemployed people to build a bridge, they may stick 100% of their earnings in the bank - but now you have a new bridge.

So basically, I don't see how House is doing his cost-benefit analysis. His conclusion is diametrically opposed to Econ 102 textbook Keynesianism, which is fine, but I feel like there should be some justification. Almost everyone - even John Taylor! - thinks that infrastructure spending makes better stimulus than tax rebates, not worse. Chris House asserts that the exact opposite is true, and that's fishy.


  1. Rob Rawlings4:07 PM

    Shouldn't you take into account the private benefit of the $100 tax rebate as well as the effects on GDP ?

    If that private benefit exceeds $90 then bridge loses!

    1. That's included in the $110.

    2. Rob Rawlings4:31 PM

      How ?

      You have:
      Bridge is $90 + $110 - $100 = $100 (you include the $90 value of the bridge)
      Tax rebate is $110 - $100 = $10.(you don't include the value of whatever the receivers of the rebate spent it on).

      If they bought $100 worth of liquorice and they value this more than the bridge, then the tax rebates wins over the bridge.

    3. Nope, because in that first example I assumed the bridge builders will also buy the same $100 worth of licorice with the salary they earn from building the bridge.

      To understand more clearly, see my second-to-last paragraph. Suppose that instead of buying anything with their tax rebates, people stick all of it in the bank. And suppose that instead of doing anything with their bridge-building salary, the bridge-builders stick all of it in the bank. The only difference between these outcomes is that in one of them, you have a nice new bridge (and the bridge-builders are tired out).

    4. The difference is that the government did something. People like Rawlings don't like that. They want employment to be dependent on the whims of the private sector, investors, managers and job creators. Government can only use tax payers' money to induce those people or bail them out when their house of cards falls down.

    5. Rob Rawlings5:08 PM


      Scenario 1: The govt prints $100 and uses it to pay a builder to build a bridge.

      Scenario 2: The govt prints $100 , gives it 100 people who club together and pay the same builder to build a tunnel.

      In both scenarios the builder spends the money and kicks of the multiplier process.

      If the people value the tunnel they chose over the bridge the government chooses then the bridge is a bad option

      Obviously once you start looking at different propensities to save v spend from the direct investment v private rebate option then the story will change in favor of pubkic investment


      1. I thought you were assuming "other things equal" at least in the first part of your post anyway.

      2. If the govt is targeting a certain spending level then it can just keep increasing the tax rebate until it hits the target and then its a straight utility comparison between public spending and private spending again (the fact that additional saving has been created to achieve this is a costless accounting fact).

    6. unanimous6:54 PM

      But when the private sector is saving or reducing debt too much, scenario 2 won't happen, because the private sector is not in tunnel building mode. The private sector by definition is in a mode that needs to spend more.

    7. Rob Rawlings7:05 PM


      The idea is that by reducing taxation the private sector will increase spending. They will also increase savings so the tax breaks will need to be bigger than the desired spending increase (if these savings are not channeled into additional investments by the banking system)

    8. Anonymous7:26 PM

      Scenario 1 is a false comparison to Scenario 2. The true match would be that 'The govt prints $100, gives it 100 people who club together and pay the same builder to build a tunnel, but first, they have to build a bridge.'

    9. Rob Rawlings7:43 PM

      why? In both cases $100 is spent building a physical asset.

      The real difference is that you would probably have to give them more than $100 to get them to spend the $100 as they would want to save some part of it. But I think that difference is more apparent than real.

    10. In case (2), the government's financial position is stronger; it has $100 of debt but owns a bridge (if it wanted to, it could privatize the bridge and pay off at least some of the debt), whereas in case (1) it has $100 of debt and owns nothing. But surely we don't want to say this difference in financial position represents a "net social benefit"? It has to be more complicated than that...

    11. Rob Rawlings8:08 PM

      Yes, that's a good point. It would seem that is is better for the government to own a bridge rather than just giving money away for nothing in return.

      But if its using fiscal policy to stabilize AD then I don't really see what difference it makes as all its doing is adjusting the size of the money supply to hit its target.

      Leaving aside whether public or private spending is optimal , I think a bigger issue is that infrastructure projects like bridge building may take so long to plan and implement that they are not a optimal short-term way of targeting AD. Tax adjustments are simply more viable.

    12. James6:11 AM


      How about:

      "Scenario 1: The govt prints $100 and uses it to pay a builder to build a bridge.

      - The builder then spends the $100, giving $1 to 100 different people. They then club together and pay another builder $100 to build a tunnel.

      Scenario 2: The govt prints $100 , gives it 100 people who club together and pay the same builder to build a tunnel."

      We could make up lots of different examples.

      However, Noah's point is simple:

      1. both the govt spending and the tax rebate increase GDP by $110.

      This means $110 worth of goods and services are sold as a result of either the govt spending or tax rebate. Perhaps the $110 is clubbed together and spent on building a tunnel, who knows.

      2. However, the tax spending also adds a new bridge.

      So in both cases you have $110 increase in GDP, but in the govt spending case you also have a new bridge, with a value of $90.

    13. James6:13 AM

      sorry, that should be: "2. However, the govt spending also adds a new bridge."

    14. Rob: Just to jump in here, note that the govt. doesn't print the money that is used to build the bridge; it borrows it.

    15. Rob Rawlings9:33 AM

      I don't think you specify anything abut the source of the $100 in your post but fair enough.

      Taken a step back I suppose what you post is really saying is that when the government borrows money it is better off buying something (like a bridge) with the money than just giving it away. Given a certain multiplier the increase in GDP will be the same but it will also have a bridge.

      That seems valid to a certain point.

      But if the govt is using policy to target an optimal level of output then this stops making sense. The govt in that case needs to be using deficit spending to hit that target. When it hits the target the normal rules will apply as to which public projects are worth borrowing for and which not. If is has used the deficit to build bridges that don't meet these criteria then this will be an de-optimization compared to deficit funded tax cuts.

    16. Rob Rawlings7:05 PM


      The idea is that by reducing taxation the private sector will increase spending. They will also increase savings so the tax breaks will need to be bigger than the desired spending increase (if these savings are not channeled into additional investments by the banking system) "

      Please read Krugman on this, particularly deleveraging.

    17. Anonymous11:29 AM


      I'm not sure why you think the 'normal rules' will backdate to decisions made under different circumstances. If the government needs to be doing deficit spending, then why not build something out of it? Tax rebates are conceptually the same idea government spending wise as the infamous 'hole digging' spending, i.e., giving people money for no change in production.

      With that comparison in mind it might make more sense why it doesn't matter that the 'normal rules' will apply at some point in the future. Sure the bridge wouldn't be a good idea to build when the normal rules apply but then again randomly cutting taxes by deficit spending isn't a good idea when the normal rules apply either!

    18. Rob Rawlings11:52 AM

      As I said it make sense if viewed as a one-off action.

      But I see macroeconomics theory as providing a guide to optimizing output through time. If policy has that end in mind then govt-funded bridge-building seems a bad idea (by the time you've planned and built the bridge macroeconomics conditions have changed). If tax cuts are considered unfair in some way though and you really need to see a return for your dollars then just use the borrowed money to fund a subsidy on final sales.

    19. Rob Rawlings11:58 AM

      And of course if you drop Noah's assumption that the money has to be borrowed then the distribution of new dollars to optimize AD and the borrowing of dollars to fund long-term projects (or control the interest rate) become totally separate things driven by different criteria.

    20. Anonymous12:21 PM

      Ah, my mistake then. Yes of course you're right that if the spending won't be done today then it makes sense to consider the longer term dynamics for when it will be spent. I think the objection to the original House post was that he seemed to be setting up the static scenario and saying it was a worse idea even then.

      You probably weren't responding to me with the fairness comment, but if you were I wasn't actually objecting on 'something has to be produced' grounds, so we agree there.

      As for the last, that sounds a lot like you're commenting that ideally we should just increase the supply of money rather than do deficit spending. Don't know if it was what you meant, but it sounds like you're saying fiscal policy is second-best to monetary policy, which I think most people reading this blog would agree with. Which is probably true, but rather counter to the original problem's assumption that the fiscal multiplier is large (whether that's ever a realistic assumption depends on your brand of monetarism).

    21. Rob Rawlings12:59 PM

      Yes, generally I would favor monetary policy to address AD stabilization.

      But at the ZLB I think "helicopter drop" style monetary expansions probably work better.

    22. Noah: Maybe you should move your second to last paragraph to be earlier in the article. If you readers are getting confused early on in the article, maybe the writer of the article is to blame.

    23. Rob Rawlings9:41 AM


      When Noah says "The difference between bridge-building and tax rebates can be stated in simple terms. If you give people a tax rebate, they may stick the whole thing in the bank, completely negating the stimulus" - this ignores the fact that the tax rebate would be saved and not spent is a sign that one of the causes of the recession is balance sheet issues. By allowing people to increase their savings the stimulus IS working. If they would have liked to have increased their saving and they end up with a bridge that they don't value as much they are worse off. (while the builder of the bridge is only marginally better off as he had to give up labor for the $100).

  2. I think the key to his analysis is just before the bit you cite:

    "This type of stimulus spending will increase employment and GDP but it won’t really enhance social welfare. True, we might get the beneficial effects of the stimulus but we could achieve that by simply giving the workers the money without requiring that they dig the holes. If we simply give out the money, GDP increases by less but social well-being goes up by more since the work effort and time wasn’t required."

    I think he's implying that if you distribute cash directly, the inputs that would've gone to making the inefficient bridge will instead be utilized more efficiently by the holders of the helicopter dollars.

    1. Right, see my second example, in which I also make that assumption.

  3. John Aziz4:17 PM

    It's about first order and second order effects. First order effects of tax cuts raise GDP zero. Second order effects of tax may raise GDP a lot, but we don't know for sure. First order effects of $x spending raises GDP by $x. Second order effects may raise or lower GDP, but again we don't know for sure. We can argue for days about the best ways of estimating second order effects, but if we're desperate (and the zero bound is, I think, by definition a rather desperate situation) it is the first order effects we should concentrate on.

    1. But we do know for sure what the first order effects are. Right....

  4. Tax rebates usually increase non-permanent consumption, booze, dope, and fun at night. There is nothing permanent; a one time hurrah! when one receives a check from IRS.

    Noah should have analyzed what WPA did for the infrastructure and self-esteem of people

    I had an opportunity to go to White Sands in 1980s and was able to see the work done under WPA for the park; including benches with southwestern design and motif, a lasting accomplishment. The young park ranger was very proud telling the stories about all that was done under WPA. No tax rebate will ever do that.

    1. Tax rebates are better than nothing. They can be used to pay down debt and "deleverage."

      Noah: "Almost everyone - even John Taylor! - thinks that infrastructure spending makes better stimulus than tax rebates,"

      Even Martin Feldstein!

  5. Chris House asserts that the exact opposite is true, and that's fishy.

    It's not fishy, it's faith-based.

    Everything private -- say, consumption spending -- is better than anything public -- say, public works. One doesn't have to prove this, it just is.

    (You can't refute a theology -- and that's what you're up against here.)

  6. Why have you chosen House as your blog-nemesis? I only hear about him when you talk about him.

  7. What I dislike is talking about cost/benefit like they are independent of the macro economy and interest rates. The cost may be up front but the benefit is payable over years and when rates are low that benefit mushrooms. He wants to put a figure on it separate from reality.

  8. Your right. House's math is nonsense. But wait, there's more.

    First off, tax cuts do not spur GDP growth.

    Second, something like 47% of the population don't pay any federal income tax. [Gee, do you suppose there; a connection there] These are the people who have the least and will surely spend that next stimulative dollar, if they could only find it. They won't find it from a tax cut.

    The obtuseness of right wing economics is an awesome spectacle.


    1. Calling it right wing economics is misleading. What we have is agreement that government should build bridges. People who disagree are right-wing. But that's the easy way out for Neoclassical Economics. It's these little concessions that hide the fact that they are wrong on the big picture.

      It is simply false that we know how a dollar spent will rebound through the economy. Your post at Angry Bear is an anecdote. Only a tiny percentage of economists even consider the amassing of massive amounts of data to solve the question to even count as economics. It's absurd.

      Pretending like House's nonsense is the only wrong stuff is how an obviously wrong paradigm refuses to die.

  9. Maybe House thinks that if you give homo economicus a $100 tax cut he's definitely going to give it to an unemployed person to produce something worth at least $100. So if the government pays the same person $100 to build a bridge that's only worth $90, that has to be worse. Why let a real government project allocate resources when imaginary individuals do it so much more efficiently?

  10. Seemouse7:05 PM

    I dunno I'm still trying to understand this - doesn't the stimulus spending on the bridge have to be repaid in the future through tax increases, whereas the tax rebate doesn't?

    1. Anonymous9:21 AM

      The tax rebate is from the government borrowing just as much as the stimulus spending it. The only way the government can reduce taxes by $100 without borrowing is to reduce spending by $100.

  11. Argosy Jones7:43 PM

    Do we even have an objective way of comparing 'social benefit' and 'social cost'?

    1. O'Shaughnessy7:50 PM

      Oh yeah we got that - it's whatever the government says it is.

  12. Anonymous7:56 PM

    You mentioned bridges! I have a book called High Steel about the building of the Golden Gate and Bay Bridges. Obviously huge wastes of resources that should never have been built. San Francisco would be a prosperous city if they had not been built.

  13. I don't think your criticism works if your assumption is that both the bridge and the tax rebate increase GDP by $110 in total. In case 1, a bridge is built plus $10 worth of other stuff from the net knock on effects (presumably also with social benefit). In case 2, the tax rebate produces $110 worth of additional GDP by assumption; stuff that presumably also has social benefit. Maybe someone even built a $110 bridge! To the extent that the workers and the tax rebatees both buy licorice, the net licorice production only increases by $10 in the bridge world while the rebate world licorice production may have gone up by $100 or $110. Whatever was built, you can't double count the bridge and the spending in case 1 but only count the spending benefit in case 2 while ignoring the social value of the production it was spent on. The only way your example works is if you mean the GDP actually goes up by $210 in the Bridge example (bridge plus stimulus), but that's just an argument that the bridge is a better stimulus and doesn't really push back on Chris House's argument.

    1. Nope. See the argument in the second-to-last paragraph (zero multiplier case).

    2. Rob Rawlings12:37 AM

      A zero multiplier world would be a world where no-one spent any of their income and the state controlled the whole economy.

      Assuming a non-zero multiplier then you really need to compare the utility of $100 of additional public spending against $100 of additional private spending - not $100 of public spending against $100 of tax rebate - as these are 2 quite different things.

    3. Noah, a zero multiplier would mean that when the government built the bridge GDP did not increase; i.e. another bridge that would have gotten built did not. You have to subtract the lost bridge.

    4. dlr: No, I mean a zero multiplier on tax rebates. The apparent multiplier on govt. spending will be higher, and will include the bridge value. Google "balanced budget multiplier" and read the Wikipedia article to see a simple version of this idea.

  14. Why worry about paying back the $100 borrowed to build the bridge in the future?

    Why build bridges when you can just dig holes to increase GDP?

    1. James6:22 AM

      a bridge is generally more useful than a hole, I guess.

    2. You can fill a hole in again and get more GDP. Much better than a bridge.

  15. In a discussion over the weekend I put it to an economist why he thought equations were so important for making policy. Lincoln didn’t consult economists when he guaranteed the bonds and gave away the land that built the railroads. Congress didn’t consult models when it gave piles of money to every state for the purpose of creating universities that improved agriculture and other productive technologies. And yet these were both fantastic macro-economic decisions.

    The opportunity cost of economics is policy makers who look at the world and say, “We need stuff and we have the money to buy it. But I have no idea what to do. Let’s ask a PhD from the University of Chicago.”

    Only a rank moron, or somebody under the influence of neoclassical economics, knows what he needs, knows that he has the money to buy it, and yet doesn’t know if he should use the money he has to buy the things he needs.

  16. I honestly don't understand how this happens again and again. You go to the trouble to look things up on the internet to back your claims, but the things you look up aren't data, aren't evidence, aren't anything, really.

    How does the solution to a (big) Sudoku Puzzle tell us how the world works? It can't. No matter how internally coherent it is (and that's what makes Sudoku work, the coherence), Sudoku puzzles aren't facts about the world.

    1. I hate it when the Internet fails to back up my claims. Stupid Internet. Almost as stupid as the University of Chicago per your earlier post.

  17. Assume that your hypothetical $100 tax rebate produces zero social benefits--no externalities--and it'll still wind up making the recipient of the tax rebate $100 happier. Shouldn't that be included in the arithmetic, too?

    1. Anonymous9:18 AM

      No, the tax rebate either has to come from current government spending or from future taxes. Either way some other person (or the same person in the future) is out $100. That's where the -$100 comes from in the social benefit calculations.

      If there's some reason to expect that the recipient of the tax rebate will be happier than the other person is harmed then there's a net gain, but by converse if the recipient is less happy than the other person is harmed there's a net loss. Usual convention is to assume that on average the transfer itself conveys no benefit.

    2. Got it. So, leaving aside the multiplier effect (which we can assume is the same whoever spends the money), the social benefit must be greater than the personal benefit lost because the money was taken out of private hands and spend by an unfeeling government. That's logic even a less than-arithmetist like me can understand.

  18. Isn't this just the flip side of the "Broken Window" story?

    In the "Broken Window" story, the idea is that the shop keeper buys window repair rather than clothing: same amount of transaction, but the net results is that the economy has produced less clothing. In this story, the net result is an additional bridge.

    The "Broken Window" story, IMHO, suffered from a model of unending cyclic economic transactions with poorly justified entry points, ending points, no multipliers, no savings alternative, and no distributional effects. Noah pins down most of these, though I haven't noticed him pointing out that tax rebates go to the rich who may bank them while bridge building money gets paid out largely as wages to workers who generally spend them.

  19. Anonymous11:25 AM

    (1) The government gives $100 to a builder to build a bridge and he buys a BMW from Germany.

    (2) The governemnt issues a $100 tax break and consumers use it to buy things from China

    Straw man arguments and false dichotomies are being used as distracting issues form the real problem and economists are the instrument of this propaganda..

  20. Seamus12:37 PM

    Isn't the point that we should be allocating resources to those areas which will prove sustainable - if the bridge basically doesn't make sense then put the money where consumer choice will lead to a more efficient, rational allocation of resources. All that steel and concrete wasted and all those licorice shops not being built. We send the wrong signals into the marketplace. Make more steel, make more concrete, other people needing that stuff have to pay more, work in construction, keep driving to work, and so on and on.
    Anyway, if it was me I'd keep both tools in my macro toolbox, some marginally not cost-effective infrastructure projects, and payroll tax cuts, among others.


  21. "If the social value of a government project exceeds its social cost then we should continue to fund the project whether we are in a recession or not. If the social value falls short of the social cost then, even if the economy is in “dire need” of stimulus we should not fund it. If we really need stimulus but there are no socially viable projects in the queue then the government should use tax cuts..." – Chris House

    Noah, I think I see perhaps the point Chris is trying to make, but you have to make very clear the crucial first part – "If the social value of a government project exceeds its social cost then we should continue to fund the project whether we are in a recession or not." We're horrendously, comic-tragically short of that. To optimize social welfare, for efficiency in optimizing total societal utils, especially over the long run, we should be spending ridiculously more on basic scientific and medical research, smart infrastructure, Heckman-style early development; prenatal care and nutrition, high quality and educational daycare and pre-school; smart education and training, teen pregnancy programs, and so on. We're just horribly short on this due to, Big gummit always bad; there's no such thing as externalities (including the pink elephant of economics, positional externalities), asymmetric and poor information and expertise, zero marginal cost goods,…

    To put us into the if part of this statement would be a gigantic leap forward for us, and especially our children and grandchildren. So, the problem Chris "worries" about is a very high class one.

    However, suppose we live in this dream world where it's hard to find any more public investments that stand on their even when not in recession. In that artificial idealized case (and this includes government spending and taxation which considers plummeting marginal utility of dollars, and positional and other negative externalities), then perhaps Chris has a point.

    The drawback of the tax cut, stimulus-wise, is that much of it can be saved. In fact, if it's a tax cut for the rich, the only kind Republicans really like, in truth, then the vast majority is saved. Their marginal propensity to consume is extremely low.

    I think what Chris House could say is that if say 95% of the tax cut is just being saved, then in this idealized model, we could just make the tax cut for over $2,000 and we would beat the bridge at $100. You could then say, ok, 20 bridges, but if the other 19 are pretty much useless, then over the long run they provide nothing more, but over the long run, or very long run, by contrast, the $1,900 in savings will be spent on things which arguably provide a lot more utility. I think in a long run, simple, very unlike our reality, model, Chris might be able to show at least Pareto-efficiency with the tax cuts.

    1. I think in this kind of highly idealized model, you perhaps would just juice up the tax cuts until the stimulus benefit was zero, the recession was ended. The cost to society of the tax cuts might, in effect, be zero, at least the part saved; it's just sifting money around, the part that goes into saving, no real cost, so you just don't worry about it and keep juicing the tax cut up until enough money is spent to end the recession.

    2. But the "if" part of Chris's statement is just so tragically far from the truth, and that really has to be stressed.

      Another issue: This high marginal propensity to save makes deficit-neutral stimulus really something that could be good. Tax a dollar and spend it and you get an extra dollar in spending immediately. While the tax especially if on the rich might result in only 5 cents less in spending. Economists in the past used to talk about this a lot more, like Samuelson.

    3. I mean, suppose there were a tax cut of $100, and it was all saved, 100%. So, 0 stimulus benefit, and 0 government spending benefit. House would not say $0 benefit and $100 cost, so -$100. He would perhaps say the real cost of the saved money is 0 to society. It was just pieces of paper with dead presidents pushed from one pocket to another. You have to not count that in your cost, at least your real cost to society as a whole.

    4. Yeah, it looks like I'm right on Chris House's thinking. From his response post today:

      "The second error is more important. In his calculation, Noah is counting the out-of-pocket revenue outlay for the tax cut as a cost. This isn’t correct. The tax cut is a transfer. There are no direct social costs associated with the tax cut."

  22. I think there is an error in the computation. You are not including the cost of building the bridge. You are including the $100 as a cost , but this is just a transfer.

    This becomes clear when you assume that the multiplier is zero. Suppose you tax and take $100 from Person A and give it to Person B. If Person B has a higher propensity to consume, or to take risk, etc, perhaps the multiplier is more than 1. But assume that Person A and Person B are exactly the same.

    In the case of a tax and transfer, the net social benefit is zero: Person A: -100 and Person B: +100

    Now suppose you tax Person A and pay Person B to build the bridge. And the value of the bridge is $45 for each person.

    Now you have, Person A: +45 - 100, Person B: +45, +100. This gives you the +90 in your calculation. But this is not correct. The $100 going to Person B is not the cost of building the bridge, it's a transfer. You have to pay for materials and Person B's time and effort. If you are giving $100 to Person B to go build the bridge, then his benefit is not +45 +100, but +45 +100 - cost of materials - cost of time and effort.

    The argument would work if the $90 in the computation was a net benefit, but then that would be assuming the conclusion (e.g. assuming that social benefits exceed social costs which was House's point).

    1. Nope, no error in my computation. The real cost of the bridge in this simple example is leisure. ;-)

  23. This comment has been removed by the author.

    1. I think the math in your example is incomplete. The consumer who gets the tax break also buys something with the $100, and let's say this is actually worth $100 to the consumer. The net social benefit is then $100 + $110 - $100 = $110. Then the comparison with the bridge has the tax cut winning.

    2. Not sure where that $110 in your equation comes from.

  24. Anonymous6:06 PM

    Noah, this is a very interesting idea you are presenting, but I don't you've taken it far enough. What if, after giving the rebate or commissioning the bridge, the government suddenly increases the income tax? Then all of the spending can be recovered, and in one case you have a brand new bridge, but in the rebate case you get nothing. What's more, the government can spend the recovered money again for more public works, and then recover it again, and again and again.

    You've discovered a way of revitalizing the infrastructure of the US for free!

  25. House seems to have overlooked the fact that the presence of underemployed resources changes the cost benefit analysis. When the interest rate hits the ZLB many long lived investment become attractive that were not when interest rates were zero. Likewise, if labor of the kind going into a project is unemployed, then the economic cost of using it is less than the wages paid. Fiscal "stimulus" in a recession is needed because government do not generally employ cost benefit analysis for projects and so fail to ramp up investment during recessions.

  26. Anonymous8:48 PM

    I didn't bother reading the comment section (I imagine this point was already brought up).

    Let's say that people were going to come together and invest money in human capital, or new houses, or whatever. Let's say it was worth $100. Then you take the money away. Then you spend it on a bridge. Is it really $110 still? Or just $10? This seems like a terrible broken window fallacy. The others bring up the point that with a tax cut they reinvest it as well.

  27. Rob Rawlings9:04 PM

    Thinking about this a bit more.

    Assume a zero multiplier. Here are the 2 choices

    1. Build the bridge

    To keep things simple assume you can build the bridge with labor alone (and the raw materials are free). Say the cheapest you can hire the builder for for is $100 . You calculate the value to the community of the bridge and it comes in at $90. The builder accepts the job at the lowest price at which his fee is greater than the value he gives his leisure. He is very marginally better off - lets say by $1

    - The builder is better off by $1
    - the community is $90 better off with the bridge
    - the govt is $100 in debt , so $100 worse off

    Net gain: -9

    2. Tax Rebate

    Assume you just give the $100 to the community as tax rebate

    - The community is $100 better off
    - The govt is $100 worse off
    - The builder is unaffected

    Net gain: 0

    You're better off giving the money away than building a bridge whose costs exceeds its value.

    Is there something wrong with my logic ?

    1. Bill Ellis11:25 PM

      Your assumption that the builder just gets just a dollar is fixs the outcome.

      The hundred spent goes to everyone who has added value to the raw materials on the way to them becoming a bridge. (all that stuff does not come free. )

      The way to figure what part the 100 spent on the bridge went to "Builder" and and all the other people who added value to materials and services required to build the bridge.... is to subtract the price of the raw materials before they were were touched by people.

      That would have to add up to way more than "one dollar to the builder".

    2. You asked: there something wrong... Yes, it is illogical, totally, not a shred of consistency anywhere. However, you will make a dem doing repub things, which is to line the pockets of rich and hasten the demise of poor.

    3. Josh L12:26 AM

      Rob: I think you're losing part of the benefit. You're not taking into account that the original 100$ has to go "somewhere". The builder should not be better off by 1$- he should be better off by 100$ (otherwise why are you spending 100$ on the bridge- if you are, and the builder only gets 1$ benefit, you're below the .2 multiplier)

      Assuming materials are 0 for now ( for simplicity),

      gov't spends 100$ on bridge:

      The builder is now 100$ richer (this part is equal to the tax cut).

      But now you have a 90$ bridge for "free".

      So you have the physical 100$+the 90$ from the convenience/utility of the bridge= 190$-100$(cost)=90$.

      Look where you say in 2: "the community is 100$ better off". You need to do that same transfer for part 1, the money can't just disappear.

      Another way to word it that may help:

      Let govt=A, builder = B

      For the bridge, when you transfer 100$ from A to B, you end up with 0$ (transfer is neutral)+90$ of a bridge.

      For the tax cut, you transfer 100$ from A to B, you end up with 0$.

      The "potential" labor of the builder is wasted. Some use (almost no matter how inefficient) is beter than no use.


      It's like the classical "pay a man to dig a hole" , except you pay him to actually build anything with any utility > 0$ (because why not? it's free value). It's like getting a free bridge when you do the net-zero (assuming no multiplier) transfer.

    4. Rob Rawlings9:43 AM


      On: "The builder is now 100$ richer (this part is equal to the tax cut).". He has $100 extra dollars, but he's also put in a lot of labor that he wouldn't have done (by my assumption) for only $99 - so I think its reasonable to say he's only $1 better off.

      Put simply: The community is better off with $100 in cash that they can spend how they like (even if they chose to save ever dollar) than a bridge that is worth only $90 to them. The reason is that the costs (in the case of the bridge in my version - the builders labor) are higher than the benefits (the value the community assigns to the bridge).

      If this is happening in a recession then the fact that community is saving so much probably indicates that money (and other savings vehicles) are overvalued. The solution is to increase the supply of money not to build bridges whose costs don't justify the benefits.

  28. "
    The difference between bridge-building and tax rebates can be stated in simple terms. If you give people a tax rebate, they may stick the whole thing in the bank, completely negating the stimulus. If you pay unemployed people to build a bridge, they may stick 100% of their earnings in the bank - but now you have a new bridge.

    No, the bridge builders have to pay federal taxes on the earnings. The recipients of the tax rebate don't have to pay federal taxes on the rebate.

    1. Also...

      The bridge builders are tired. They spent time they could have been doing something else acquiring the money they put in the bank. The tax rebate recipients spent time doing what they wanted to do, and they put money in the bank. Maybe the tax rebate recipient went out and gathered pinecones that they will use to make christmas wreaths to sell or give away.

      Next, we have to look at who is receiving the helicopter drop. The construction guys may be well off and put the money in the bank, whereas the general tax rebate recipient might be more likely to spend the money on. What the various people buy is also important. When we build the bridge, will there be a bottleneck on acquiring steel and cement and bridge engineers? Or will the increased steel and cement and bridge engineers lead to higher production and lower unit costs for these items?

      The tax rebate recipients are spending money on what people want to buy. They are increasing production and lowering unit production costs on things that people want to buy. (Unless there's some sort of bottleneck in that production path.)

    2. And, I've never seen a clear discussion as to why a helicopter drop should be done with borrowed money instead of freshly printed cash (well ... actually an update to a database...)

    3. And, finally... Noah, you should not have shifted the discussion from tax cuts to tax rebates. House said "tax cut". You said he said "tax rebate". And from then on, the conversation rapidly deteriorates into a confusing morass. Tax cuts are stupid. Helicopter drops are cool.

  29. There is a reported infrastructure deficit in the US of a level not known by myself. It would seem that the idea of the alternatives of building bridges or reducing taxes to generate growth, is missing the main point of infrastructure deficit. But the building bridges solution would reduce the infrastructure deficit, even although it is not the reason proposed in the post for the spend. As long as it is not a bridge on the road to nowhere. In some cases there is an argument to increase taxes and build bridges or any form of useful infrastructure.

    There is a post on the inet somewhere about cost benefit analysis of infrastructure spend with a comparison between the UK and France. The former does a lot of the analysis and so has a poor road system in comparison to the latter who just build fantastic roads everywhere and benefit from them. I think that the post it is by Professor Lewis in the UK.

  30. Anonymous9:13 AM

    Even worse than saving the money, wealthy special interests use their tax breaks to invest overseas and offshore domestic jobs. Tax cuts can lead to domestic job loss.

    When a government like the US is captured by wealthy special interests, the tax cuts are distributed to the wealthy who already spend all they want and are denied to the unemployed (who get nothing from a tax cuts) who have unmet demand.

    The goal of the wealthy Malefactors and their apologists is to transfer money upward and limit its distribution downward. Public infrastructure that can be used by anyone distributes money more equally and in a way that is not as favorable to the wealthy. That is why they create arguments against stimulus no matter how fallacious. The arguments are not made in good faith. The arguments are made in support of a goal: reducing public goods and services.
    -jonny bakho

  31. Government checks are government purchases (money for nothing). The cost of producing nothing is zero, and the benefit of consuming nothing is zero. Thus, the net benefit of government checks is zero.

    If government purchases stimulate the economy, then government checks stimulate the economy, too. Government checks, therefore, strictly dominate government spending with negative net benefits.

    During recessions, costs of production tend to fall due to excess supplies of labor and capital. Thus, recessions often make otherwise wasteful purchases cost-effective. This rationale for spending, however, has nothing to do with stimulating the economy. Even if the monetary authority fully offset the stimulative effect of the spending, it would still be cost-effective.

    Regardless of the need for stimulus, only government purchases that pass the cost/benefit test have merit.

  32. Warren Nichols2:56 PM

    Although various commenters have objected to your computation, you still do not seem to realize that it is quite wrong.

    You assume that, given the state of the economy, if the government issues a check for $100 to be used either to build a bridge or as a tax rebate, then GDP will increase by $110. Calculating the gross benefit to society in the case in which the additional $110 of goods and services includes the bridge as $90 + $110 is simply not correct. The bridge is part of the extra goods and services, and is being counted twice.

    Chris House’s point, in your context, would seem to be that if the extra bridge with $90 social value were to be replaced by an extra bike that someone purchased with their $100 rebate check then society as a whole would be better off. Social value is the sum of values to individuals, so providing a bike to a person who values that bike at $100 is better than providing society with a bridge for which the sum of the values to individuals is $90. But this analysis does not go far enough. If the buyer of the extra bike is almost indifferent between buying the bike or saving the $100, and the fully employed bike-makers are almost indifferent between making that extra bike or quitting at the usual time, then the value to society of the extra GDP represented by the bike would be almost nil. Everyone would be almost as happy if the extra bike had never been built. In contrast, the value of the gain from the job by the unemployed bridge-workers, who are far from indifferent between building the bridge or going fishing, would be quite substantial. More broadly, House does not seem to take into account the additional gains to society that come when stimulus spending is directed towards under-utilized resources.

    ZLB issues do not come into your comparison. Whatever gains are to be had by government borrowing from the job bridge or to issue a tax rebate. Of course, increase in GDP per dollar of government expenditure may well be considerably more when the government spends the money directly, in which case the societal cost of the financing that produces a given increase in GDP must be considered.

    I think that the most relevant of Keynes’s insights were not about the ZLB per se, but rather about what to do when an economy may be stuck at a high-unemployment equilibrium. Even doing something very wasteful, such as building many airplanes and tanks and then having someone blow them up for you, might move the economy to a much more favorable low-unemployment equilibrium. There was a natural experiment that, at least for several decades, gave considerable credence to this notion. You know far better than I do how much support it has today; Chris House seemed to dismiss the notion as being “silly”. For someone in support, the most direct way to get the process started would seem to be to do the stuff that most directly puts unemployed people back to work. But this takes time, and tax rebates can start the process more quickly – people might even increase spending in anticipation of the rebate. Thus the proper discussion may not be which method of stimulus to use, but rather what the best combination of the two may be.

    1. Warren Nichols3:06 PM

      (There was a copy-and-paste error in the second-to-last paragraph. Corrected version below.)

      ZLB issues do not come into your comparison. Whatever gains are to be had by government borrowing, whether from financing at a low interest rate or even from printing money, will be the same whether the $100 check is used to pay for the bridge or to issue a tax rebate. Of course, increase in GDP per dollar of government expenditure may well be considerably more when the government spends the money directly, in which case the societal cost of the financing that produces a given increase in GDP must be considered.

  33. It was mentioned by Rob Rawlings that your assumptions are wrong. I assume that the $90 value out of $100 bridge is there as a comparison to what ammount of value a $100 could in case it was used in a markety way.

    So let's do your example for private sector again. Government borrows $100 and hands out some tax rebates (for instance lowers payroll taxes to low earners). Let's assume that these low earner will spend the money in the same way, maybe insulating their houses? So you have the same spending multiplier except that the value bought for that $100 is actually $100 of insulation instead of $90 of bridge.

    Tax rebate wins.

    1. Rob Rawlings11:49 AM

      Noah's logic assumes the $100 will have the same flow effect on GDP once it gets into the hands of the population

      He is claiming that if the govt gets the $100 into their hands via paying them to build a bridge rather than just giving it to them via a tax rebate they will be better off because they will have the $100 plus a bridge rather than just the $100.

      I think this is wrong. He is forgetting the cost of the bridge. If the community gives up $100 of labor (or other resources) for a bridge that (by his assumption) is only worth $90 to them then this is bad business no matter how deep the recession is. They would be better off with the $100 and no bridge.

    2. Warren Nichols3:49 PM


      You are correct that Noah is forgetting the cost of the bridge. But the situation is more complicated than just that.

      The $100 rebate is a red herring. It is not making people $100 better off, because the government intends to tax it back in the future.

      Because the bridge-workers are unemployed, they value the time it would take to build the bridge at less than usual. If they value it at $70, that is the cost to society of building the bridge (ignoring material costs), so the net benefit to society is $90 - $70 = $20.

      My thinking is further explained in a comment below.

    3. Rob: Is this really what Noah is thinking? Then it is clearly a mistake, he is forgetting the cost of the bridge.

      Warren: If indeed workers are willing to work for less so that for instance you can build $90 bridge with $70 costs then the project then fulfills Chris House's criteria - it is profitable by $20 even without taking into account any multiplier effect. Then it totally makes sense to do that project and I believe Chris would have no problem with that.

    4. Warren Nichols7:52 AM


      I agree.

      Another way of looking at it, using Economics 101 terms, is that even though the bridge results in a "consumer surplus" of $(-10), there is a "producer surplus" of $30, and society is $20 better off.

      Noah says in a comment that "the real cost of the bridge in this simple example is leisure." I don't think that he would criticize Chris House's cost/benefit analysis and then knowingly assign leisure a value of zero; I think that he knows that he made a mistake but is unwilling to fess up.

    5. Rob Rawlings9:19 AM


      Can you clarify your position on this ?

    6. Anonymous12:52 PM

      The government spends $100 on the bridge. That is the cost of the bridge. What are you ranting and raving about?

  34. Anonymous12:27 PM

    What Warren Nichols said...

    But repeating it in my words...

    Suppose the second order stimulus effects (in terms of GDP and Welfare) are the same regardless of whether the government writes a $100 check for a bridge or writes a $100 check as a tax rebate.

    So we only need to compare the first order effects.

    In terms of GDP, the bridge wins. The $100 spent on bridge building is part of GDP (it has a first order impact as well as second order impacts) while the $100 tax cut has no first order impact on GDP (and the same second order impacts as the bridge case).

    When we look at first order impacts in terms of welfare, it gets complicated. You assume that the bridge increases welfare in terms of first order effects by $90. But you assume that the tax cut does not increase welfare at all in terms of first order effects.

    In other words, when you are looking at the bridge, you are considering both first and second order welfare effects. The $90 first order effect and the second round stimulus effects.

    However, when you are looking at the tax cut, you are considering only second order welfare effects. 0 in first order effects and only second order stimulus effects.

    What you have done makes sense for GDP, but not for welfare.

    To clarify this further, many intermediate macro textbooks show that the balanced budget multiplier is 1 (assuming interest rates and prices are fixed) in terms of GDP. Suppose the government taxes people $100 more (this decreases economic activity by second order effects) and then spends that $100 on bridge building (this creates a bridge, and then increases economic activity by the same second order effects as the decrease caused by the $100 tax increase). In all, taxing people $100 to build a bridge that costs $100 (assuming fixed interest rates and prices - i.e. a horizontal AS curve) increases GDP by $100.

    But what about welfare? This is trickier. On one hand, they have a new bridge (and enjoy positive second order effects from increased spending). On the other, they have paid $100 more in taxes (and suffer negative second order effects from decreased spending).

    1. Anonymous2:12 PM

      The first order effects of the bridge are a waste of money - they cost $100 (in labor and other resources) and are worth only $90.

    2. Warren Nichols4:08 PM

      I had one of Noah's main assumptions wrong. Please see my second comment below, and my reply to Rob above.

  35. Warren Nichols2:48 PM


    I need to take back part of what I said in my previous comment. I now realize that your assumption “spend $100 on the bridge, and GDP goes up by $100 from the stimulus effect” actually means that you are assuming that if $100 is spent on the bridge, then GDP will go up by $210. The additional goods and services will consist of the $100 bridge and $110 of other stuff.

    This is a rather bold assumption. After making it, and according to the way you compute social value, your disagreement with Chris House seems to come down to: House does not realize that bridges are free.

    But, at least as I understand it, your computation of social value is rather suspect. You say that the net social benefit of spending $100 on a tax rebate that results in a $110 dollar increase in GDP is $110 - $100 = $10. It appears that you are treating $110 as the benefit to society of the rebate, and $100 as the cost to society. In this situation, it is not that simple.

    In this situation, the government is sending $100 to someone, but it is intending to collect $100 from someone as well, either on April 15 of the next year or (plus interest) further in the future. To a first approximation, the cost to society of the tax rebate, after all is said and done, is zero. Even if the tax rebates result in no change to GDP (say because they are all saved), the net cost to society of providing citizens with those dollars today and taxing them back in the future is minimal.

    Members of society have a choice between activities which are recorded as GDP and those which are not. If society as a whole chooses a GDP of $x, rather than a GDP of $(x+110), the presumption is that, all else being equal, the social value of a GDP of $(x+110) is less than the social value of a GDP of $x. This is perhaps clearer at the individual level. If your spouse made you work longer than you wanted to, you would produce more, have more money, but consider yourself to be worse off, as the money did not fully compensate you for how much you valued your time.

    The dollar value of the rebate check can be ignored in a first calculation of social value. The dollar value of the increased GDP must be netted against how much the people producing the corresponding goods and services valued their time; “time is money”. With the increase in GDP considered to be the increased value to society, you seem to considering the cost to society of producing the corresponding goods and services to be zero. This is not correct; in particular, bridges are not free.

    1. Warren Nichols4:24 PM

      Typo in line 2: should be " $110 from the stimulus effect."

      Last paragraph: Second sentence should be "The gross social value of the increased GDP ..."

  36. All this worship of government infrastructure spending is sad. Personally, I'm not completely anti-government infrastructure spending, but I don't worship it. Take the intercontinental rail-road. Railroad executives had been given 23 million acres of land and $64 million in taxpayer money. The land has been estimated to be worth $400M in 1880 dollars, and much of that was sold by the railroads. But this doesn't begin to touch on the waste and corruption that quickly blossomed. In the end, travel between the coasts was shortened from six months to six days or so.

    You might think that that sounds like a massive social benefit, one that should make the railroad companies owning the connected lines very profitable. Less than three years after the joining ceremony at Promontory Point, the UP declared bankruptcy due to vast inefficiency and waste. The CP ultimately went under as well. The shoddy workmanship on the line led to continuous repairs to bridges, viaducts, tunnels, rails, ties, and beds. The routes had been chosen to maximize incoming money during the construction, NOT with an eye to sustainable profit during operation of the line. The truth is, the average person didn't need or could not afford products from the other coast, even when travel time was cut to a week, nor did he actually need to travel back and forth. Even if cost of travel had been reduced, it was still prohibitively expensive for most uses. It would cost skilled laborers working 60-hour weeks almost a month's salary for a one-way ticket to ride 'emmigrant' class, which was a bench seat. It would take over two months salary to ride in a sleeper, not counting the week of lost wages to get there. Today, of the nearly 2000 miles of railroad that started in Iowa and ended in San Francisco, only a few hundreds of miles of the original road-bed is still in use.

    Alternatively, James Hill built a transcontinental railroad without public money and just a few land grants, and was one of the few transcontinental railroads that never went bankrupt. "What we want is the best possible line, shortest distance, lowest grades, and least curvature we can build." In January 1893, the Great Northern Railway ran more than 1700 miles from St. Paul to Seattle. He took care to route it on the most direct and least-steep route, while making sure to serve an economic purpose along the way, encouraging and building up an economy along the way if necessary. He encouraged settlement and farming along the route with the railroad's profits, so that each section paid for itself, rather than simply being a connection between east and west. The strategy of making a profitable regional railroad that also happened to span the country, eventually, versus just making an intercontinental railroad, was the primary secret of his success.

    The real history of the Transcontinental railroad, the companies and corruption involved, the costs, and the outcome, stand in STARK contrast to the glowing picture Obama and government stimulus junkies like to paint for the low-information voters.