Wednesday, March 4, 2015

Mankiw on dynamic scoring

Greg Mankiw has a nice op-ed on dynamic scoring

The issue: When the congressional budget office "scores" legislation, figuring out how much it will raise or lower tax revenue and spending, it has been using "static" scoring. For example, it assumes that a tax cut has no effect on GDP, even if the whole point of the tax cut is to raise GDP.

This is obviously inaccurate. But, as Greg points out, there is a lot of uncertainty in dynamic scoring.

How much will a tax cut raise GDP, and thus potentially not cost as much in tax revenue? (Tax revenue = tax rate x income, so if income rises a given reduction in tax rate costs less in tax revenue.)

By what mechanism? Keynesians will analyze the issue through a multiplier. The tax rate cut puts money in people's pockets, they spend the money, that raises income, and so forth. Other economists focus on the incentives of a tax cut rather than the income transfer. A tax rate cut can induce people to work, save, invest, go to school, etc.  They will come to different answers, especially for policies that emphasize transfers (often with bad incentives) or that emphasize incentives.

How much will policy change growth rates? Long run growth really swamps everything. And the connection between policy and growth is especially hard to nail down.

Greg doesn't really come down on how to solve this issue. I have two suggestions:

1) Embrace uncertainty. It's a fact, we don't know the elasticities, multipliers, and mechanisms that well. So stop pretending. Don't produce only a single number, accurate to three decimals. Instead, present a range of scenarios spanning the range of reasonable uncertainty about responses. The CBO presents a range of fiscal scenarios already.

2) Transparency. Calculations should be utterly transparent and reproducible. If you don't like the labor supply elasticity assumption, you should be able to change the number and produce a new forecast. Scoring should capture "if you think x, then the answer will by y."

Good policy will not result from the illusion of certainty.

Greg also opined on the second round effects, how policy might change economic outcomes which might change future policy. Here I'll side with the old fashioned approach -- let's not go there! The science of forecasting future congressional reactions to events is, let us say, a bit less certain (even) than that of assessing private-sector behavioral responses.

Update:

Greg responds:
Dynamic scoring requires the solution of a general equilibrium model. To solve a dynamic GE model, you need to specify how the government is going to satisfy its present-value budget constraint. You might be tempted to ask the model what happens if the government cuts taxes and never does anything else. But you won't get very far. The model will tell you that the government has to do something else eventually, and it won't tell you what will happen if the government tries to do something impossible.
Greg is right. Though this hasn't bothered CBO scoring yet. Year after year the CBO releases budget forecasts in which debt to GDP ratios climb inexorably; the CBO proclaims this "unsustainable," and life goes on.

Let's try to compromise. A rule that "dynamic scoring models must satisfy a long run restriction in which debt/GDP is no greater than 100%" might work. But one does not have to do huge changes to many models to accomplish that fact. It would be good to have a common benchmark assumption about long run policy so different short run policies can be compared. For example, score all policies in the first 20 years with a common assumption about how debt / GDP at the end of 20 years is resolved.

Where I would rather not go is more detailed political modeling of future congressional actions, especially ones with large distortions.

And for many policies this will not be a huge issue. For example, if we get rid of energy tax boondoggles, one can calculate many interesting behavioral responses, but it is a drop in the bucket of the big social security/medicare/pensions/slow growth debt nexus.

23 comments:

  1. What is wrong with a static calculation? If 100,000 cartons of cigarettes are sold, and taxes raise 1 cent, then estimate $1,000 of revenue. As long as the tax adjustments are marginal, it should be fairly close to accurate - and I think also more transparent to lay people like me. If the revenue only goes up by $500, then even hillbillies like me can understand that you're at the far end of how much you can honestly tax it for revenue, and we could probably understand and follow through on laws that require tax increases to have 85% efficiency or better (not an economist - as I'm using it here, that means my listed example would have 50% efficiency)

    I understand all bets are off with large changes. Here in Kansas, they're considering a massive cigarette tax spike. The local discussion I've heard mostly hinges upon increased smuggling, pointlessness of giving our governor any more money to waste, etc, but nothing about proposed revenue.

    I guess what I'm saying, is the scoring for voter benefit, or expert benefit? I think the second-order effects are discussed separately for the average voter, and the average politician can't be expected to be any more intelligent than that.

    Sam Coulson

    ReplyDelete
    Replies
    1. because the CBO is almost never grading 1 cent tax increases on cigs. Making it legible for the layperson is part and parcel of the problem. The ACA was 20,000 pages- pumping out 1 number in terms of GDP growth makes it "understandable" but hides the fact that it isn't at all understood. It is nothing more than political ammo (Obama-care will decrease the deficit) and it actually dangerous because the legislation gets written to make the CBO's scoring (which they know in advance) more favorable, even though its wildly inaccurate in most cases.

      Delete
  2. Economic growth models (v.g. Barro, Lucas) have been doing this for ages, but in qualitative terms. They explain, instead of predicting, what will happen upon a increase/decrease in tax/government spending. Of course this isn't as fancy as providing very inacurate estimates that are bound to be wrong, but it is faithful to the true nature of economics, explanation, not prediction. Of course some measure of estimate will be needed for CBOs to work their budgets, but a plain old sensitivity analysis could well suffice, in fact embracing the uncertainty that you've asked for. Just show a couple of projections based on different assumptions and let people judge.

    ReplyDelete
  3. 1) There is so much noise in estimating these parameters, you may as well go with the null hyothesis.
    2) The CBO is bad and forecasting future year budgets anyway. Maybe, I shouldn't say bad. Forecasting is hard to do, especially about the future. We shouldn't be trying to budget with to high of a degree of precision. The father out the CBO tries to forecast the harder it is to do.

    The CBO should stop making 10 year forecasts. They should forecast this years budget and next years budget and that is it! In which case the dynamics of the scoring are far less relevant.

    ReplyDelete
  4. John,

    "For example, score all policies in the first 20 years with a common assumption about how debt / GDP at the end of 20 years is resolved."

    If debt / GDP is the benchmark against which all government policy (tax, spending, and otherwise), then focus on the debt and let GDP take care of itself.

    "How much will a tax cut raise GDP, and thus potentially not cost as much in tax revenue?"

    Instead ask the question, if government sells equity, how much is debt reduced.

    "Embrace uncertainty." - No, embrace the certainty that when government relies on equity financing, it's debt will certainly fall.

    ReplyDelete
  5. Well...all fine and good. The problem is when a political party gets its hands on dynamic scoring and then justifies any action on the basis of the putative positive results that are pending.

    Although I confess I would like to see dynamic scoring applied to a holiday on FICA taxes.

    Then offset losses to the Social Security and Medicare and trust funds through central bank-financed purchases of Treasuries.

    Since QE is nothing more than a swap of reserves for Treasuries there should be no problems with this -- the problem will be how to handle such a robust economy.

    ReplyDelete
    Replies
    1. "The problem is when a political party gets its hands on dynamic scoring and then justifies any action on the basis of the putative positive results that are pending."

      I think this is the biggest concern with dynamic scoring...on the other hand static scoring is pretty much guaranteed to be wrong, so which is really worse?

      Delete
  6. OT: The Fed said it bought $4 trillion in bonds in QE. It has a ba;acne sheet of $4.5 trillion.

    The banks have $2.5 trillion in excess reserves.

    Um, what happened to the other $1.5 to $2 trillion?

    Was that...stimulus? Money spent on other assets or goods and services?

    ReplyDelete
    Replies
    1. A portion of the $2 trillion will be required bank reserves. And the majority of the rest should be cash in circulation (both in the US and abroad).

      Delete
  7. "For example, score all policies in the first 20 years with a common assumption about how debt / GDP at the end of 20 years is resolved."

    Although this is essential if dynamic scoring is to be of any use, it is easier said than done. Recent published dynamic scoring exercises in the UK have assumed that any revenue losses from tax cuts can be balanced (in terms of sticking to the intertemporal budget constraint) by costlessly raising lump-sum poll taxes. Not surprisingly, this unrealistic assumption means that almost any tax cut will be scored as having significant dynamic benefits, because it involves replacing a distortionary tax by a nondistortionary tax. Furthermore, because the CGE model used does not allow for externalities, this means that it also appears attractive to remove Pigouvian taxes on pollution and other 'bads'.

    It's true that these exercises can still be useful in comparing the effects of alternative tax cuts, where the cuts are scaled to have the same effect on the Government's budget balance. But typically lobbyists and agencies do not work like this - instead they produce a report which sets out a spurious figure for the absolute benefits of an individual policy which is attractive to them for political or other reasons.

    ReplyDelete
  8. "big social security/medicare/pensions/slow growth debt nexus"

    interesting....maybe an idea for a separate post to elaborate?

    ReplyDelete
  9. There is the Republican's nightmare. The CBO could use historical data for their dynamic scoring. Given the feeble record tax cuts have in stimulating the economy, it may turn out that a tax increase would appear more desirable. One report like this with the historical data to back it up will get this stupid idea out of circulation.

    ReplyDelete
    Replies
    1. If you're referring to temporary "here's some cash now go spend it" tax cuts like the stimulus rebate checks under Bush or Obama's temporary payroll tax break, then I completely agree. If we're talking about longer term reductions in rates, like the cuts under Kennedy/Johnson and Reagan, that's another story.

      Delete
  10. Dear Professor Cochrane,
    Will you post your thoughts on the Lee-Rubio Tax proposal, a potential application of dynamic scoring?

    ReplyDelete
  11. "For example, it assumes that a tax cut has no effect on GDP, even if the whole point of the tax cut is to raise GDP."

    The number one reason for tax cuts in the last thirty years is to pander to favored constituencies. The number two reason for tax cuts is to hamstring government for ideological reasons ("starve the beast")

    Given that the real cost of capital is at historic lows I think it is hard for anyone to argue with a straight face that: (1) the United States needs more investable funds; and (2) that cutting taxes is just what we need to encourage the accumulation of those additional investable funds.

    On the other hand, a "tax cut" designed to reduce the effective marginal tax rate on poor people (say to a maximum of 50%) as an over riding principle on the withdrawal of benefits under social programs with rising income might actually operate to increase the GDP.

    ReplyDelete
    Replies
    1. Absalon,

      "Given that the real cost of capital is at historic lows I think it is hard for anyone to argue with a straight face that: (1) the United States needs more investable funds; and (2) that cutting taxes is just what we need to encourage the accumulation of those additional investable funds."

      Several flaws with your argument:
      1. Loans are negotiated on a nominal / not real basis. Saying that the real cost of capital is at historic lows is misleading. The central bank lends at a nominal cost, the real cost is calculated using backward looking data.

      2. For a functioning economy, both lender and borrower should be content with the terms of a loan. Your argument might make sense if the central bank was the only game in town. To say that the real cost of capital borrowed from a private bank is at historic lows also means that the real return on borrowed capital is at historic lows.

      Tax policy should be structured so that both interests (borrowers and lenders) are served.

      Delete
    2. Frank - there is so much wrong with your comment I hardly know where to start. (1) I said real and I meant real. Nominal rates are also low. (2) Capital does not just flow through bank loans. I was actually thinking mostly of the stock market where P/E over 25 are fairly common for large stable companies, companies are engaged in massive share buybacks and the Shiller CAPE is very high. Corporate bond rates are very low compared to the last forty years across the quality spectrum. (3) There are two trillion dollars in excess bank reserves. Companies do not want to raise capital and are aggressively returning capital through share buybacks. If we reduce taxes on capital income to encourage more accumulation of capital, what good will it do if there is no place that wants to employ that extra capital. (4) The evidence from the way capital markets are operating now suggests strongly to me that cutting taxes on capital income will not lead to more saving / investment / job growth / tax revenue.

      Delete
    3. Okay,

      We can do stocks instead.

      http://www.nyxdata.com/nysedata/asp/factbook/viewer_edition.asp?mode=tables&key=333&category=3

      Total shares outstanding (Jan, 2008) - 421 billion
      Total shares outstanding (Jan, 2015) - 471 billion

      Yes there was a falloff in February of this year. But the point is, those "massive" buybacks you are referring to are the exception rather than the rule.

      http://www.wsj.com/mdc/public/page/2_3021-peyield.html

      S&P 500 PE Ratio sits at around 18.09. That's an implied nominal cost of capital of equity of about 5.5% + a dividend yield of 1.97% gives a total nominal cost of equity capital of about 7.5%. Subtract about 1-2% for inflation and the "real" cost of equity capital is about 6.5% to 5.5%.

      And I thought you said the real cost of equity capital is low?

      "Companies do not want to raise capital and are aggressively returning capital through share buybacks."

      Then why have the total number of shares outstanding on the NYSE increased consistently since 2008?

      "If we reduce taxes on capital income to encourage more accumulation of capital, what good will it do if there is no place that wants to employ that extra capital."

      First, the evidence is contrary to your statement, those extra shares sold indicate that there is some place that wants to employ that extra capital? Second, it is my opinion that capital income should be treated no differently than ordinary income because it is too easy to game the system. An example would be a person claiming to be independent contractor (not a laborer) and so all of his / her income is corporate earnings, not wages.

      And finally, I stick by what I said:
      1. Measuring the "real" cost of debt capital using interest rates on new loans and prior inflation data is misleading. And, if you look at the spread between corporate BAA and the CPI inflation rate, you will see that it is well within historic norms (currently at about 4.6% inflation adjusted).

      I am using corporate BAA data from here:
      http://research.stlouisfed.org/fred2/series/BAA

      2. In a functioning economy, both borrower and lender should be content with the terms of a loan. This is controversial?

      3. And finally, tax policy should be structured so that both interests (borrowers and lenders) or if you prefer (shareholders and share issuers) are served. Again, this is controversial?


      Delete
  12. The Wall Street journal says the P/E on the S&P 500 was 20.5 as of last Friday.

    I don't care how many shares are outstanding if the numbers are being increased by, for example, share splits. What matters is how much money are companies raising by new share issues vs. how much money are companies spending on share buybacks.

    Dividends get paid from the "E" part of the P/E" so it is conceptually wrong to add the earnings yield to the dividend yield. An earnings yield of 5% is cheap by historical standards.

    I don't have a philosophical problem with taxing corporations and then giving the shareholders some credit for the fact that dividends represent income that has already been taxed once.

    I don't care how the borrower and lender "feel" about their deal.

    My search says that BAA is yielding nominal 4.6%. A large existing, profitable company that wants to expand is paying corporate income tax around 35%. If it borrows money to expand it can deduct the cost of borrowing from the existing profit and save about 35% of the interest as taxes not paid. Let's assume that inflation is 1.5%. Then that nominal 4.6% becomes about 1.5% net real when the income tax and inflation effects are factored in. That is cheap capital. If we leave out the tax effect, we have a cost of capital of about 3% which is still cheap.

    ReplyDelete
    Replies
    1. Absalon,

      "What matters is how much money are companies raising by new share issues vs. how much money are companies spending on share buybacks."

      And you know how much net share issuance is from what source? The only thing you have provided is a blanket statement about "massive share buybacks" with no reference source. How net shares have been sold in say - the last 4 years, 8 years, pick a time frame?

      "Dividends get paid from the "E" part of the P/E" so it is conceptually wrong to add the earnings yield to the dividend yield."

      Incorrect. Money is fungible. You have no idea as an outsider where the dividend payments came from. They could have been borrowed, they could have been obtained from new share sales, they could have been obtained from liquidation of physical assets, or a million other sources. You can look at net earnings and conclude that earnings exceeded dividends paid out, but you can't determine the actual funding source for each dollar of dividends paid. See the dividend payouts that Lehman was making just before it went bankrupt. Were those dividends all from earnings?

      "An earnings yield of 5% is cheap by historical standards."
      Again, it is my belief that dividends are part of the cost of equity capital since their funding source can be anything. Disagree if you like.

      "My search says that BAA is yielding nominal 4.6%."
      Link?

      Now the interesting part:
      "If it borrows money to expand it can deduct the cost of borrowing from the existing profit and save about 35% of the interest as taxes not paid."

      A company can deduct the nominal cost of borrowing, not the real cost. Would tax policy be improved if the real cost of borrowing could be offset?

      And if the real cost of private borrowing can be offset with tax policy, is there ever a need for higher inflation or a higher inflation target by the central bank?

      Delete
  13. Frank

    http://research.stlouisfed.org/fred2/series/BAA

    ReplyDelete
  14. Some comments/suggestions

    1. Model transparency is great for economists who know (or believe they know) the strengths and weaknesses of specific models. For the rest of us, how about some model validation data together with the model forecast. If the model predicts high inflation as a result of deficits (e.g., scoring a tax cut), show whether the model correctly "predicts" our present low inflation.

    2. I love the Bayesian suggestion to present a range of possible outcomes rather than the median or mean outcome! But with reservations: Imagine how politicians would use ranges: "This could cost up to [upper bound]". Also, ranges depend on tail modeling -- Gaussian vs. t for example -- so they are even more uncertain than the mean.

    3. There's the Churchill line about Democracy - the worst form of government except for all the others. That's basically the defense of static scoring. You might have to accept it because dynamic scoring doesn't serve the scoring purpose, which is to give Congress and the public numbers that have not been cooked.

    ReplyDelete

Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.