It's easier to have a sensible fiscal rule with an NGDP level-path target

It's a simple point, perhaps an obvious point, that only needs a short blog post, but I think it's worth making.

The Brits seem to be arguing about whether they should have some sort of rule (or charter) for fiscal policy, and if so, what it should look like. I am a bit skeptical, because once you get into the details (like Simon Wren-Lewis is doing here) it all starts to sound very complicated and likely to cause lots of arguments about what should be counted as what.

But set my skepticism aside.

And even if you are skeptical about the feasibility of a formal fiscal rule, it's a useful thought-experiment to help us be conceptually clear about what we want fiscal policy to look like. And one thing we want fiscal policy to look like is sustainable in the long run. And that is true regardless of whether we think fiscal policy is or is not needed to help monetary policy stabilise aggregate demand.

A sensible fiscal rule would not let the debt/GDP ratio wander off over time towards plus infinity or minus infinity. And since the debt (with the exception of some indexed bonds) is mostly nominal debt, we are talking about a ratio of debt to nominal GDP.

Suppose the Bank of Canada had a (say) 5% Nominal GDP level-path target. The "level-path" bit is important, because it means that if the Bank of Canada misses its target one year, it will make up for that miss in subsequent years to bring the level of NGDP back onto the same path as if it had not missed.

That would make it a lot easier to write down a sensible fiscal rule. For example, if the fiscal rule aimed at a long run target of 40% debt/NGDP ratio (just to keep the arithmetic simple), the rule would say the government is allowed to run a deficit of 2% of GDP on average. Because the debt and NGDP would then both be growing at the same 5% on average, so their ratio would stay the same. It's sustainable.

If you don't have an NGDP level-path target, nobody know what NGDP will be be over the coming decades. So nobody knows what average deficit would be sustainable. The fiscal rule would have to be changed whenever forecasts for long run NGDP growth are changed. And let's face it, economists aren't very good at forecasting the level of NGDP over the coming decades. But that won't stop us arguing about it, if it matters for fiscal policy.

(The Bank of Canada's 2% inflation target helps a bit, but not as much as an NGDP level-path target would help. Because: we can't forecast real GDP very well; the CPI inflation rate (which the Bank targets) is not the same as the GDP deflator inflation rate; the Bank of Canada does not fix its past misses with an inflation target, unlike a price level path target.)

Sure, an NGDP level-path target only fixes one problem with getting fiscal policy right. But every little bit helps.

Comments

[Ralph Musgrave]

What’s wrong with this fiscal rule. 1, abolish government borrowing, as advocated by Warren Mosler and Milton Friedman. 2, When stimulus is needed, have the state print money and spend it and/or cut taxes.

A further merit of that rule is that it disposes of the distinction between fiscal and monetary policy, or if you like, merges them. Given that economists have absorbed millions of person hours and millions of dollars discussing the relative merits of fiscal and monetary policy over the decades, there would be a huge saving in that those economists would have to find something more useful to do.

[Paul]

Why would a level goth path rule for gross public debt be inferior to one related to a target share of (N)GDP? Setting a debt growth path below that of NGDP would reduce the share of debt as a proportion of the economy, but it wouldn't be trending to negative infinity.

[Nick Rowe]

Paul: What do you want to reduce the debt/NGDP ratio to? Suppose it's 20%. OK, replace my 40% with 20%, so my 2% becomes 1%. But yes, you will need to run smaller deficits than 5%xdebt/NGDP to bring the debt/GDP ratio down to 20%, if it's initially above 20%. But once it's there, my 1% figure will tell you what's consistent on average with keeping it there.

[Nick Rowe]

Ralph: sorry, but Typepad keeps putting you into spam.

Because adding one extra constraint (Thou may not sell bonds) normally means you can't do as well as you could without adding that extra constraint. Sometimes you want to spend a lot, without raising taxes enough to pay for it, and you don't want to increase AD. Like a temporary emergency. So you issue bonds, and let the central bank tighten monetary policy to keep AD on track.

But we have been through this many times, and never get anywhere.

[Ralph Musgrave]

Nick,

You say “Sometimes you want to spend a lot, without raising taxes enough to pay for it, and you don't want to increase AD”. Yes: Milton Friedman argued that the amount of extra spending in war time can be so large that borrowing could be better than raising taxes. But otherwise, i.e. in peace time, he saw no good reason for government borrowing, and I pretty much agree. I.e. even the annual spend on the largest infrastructure projects is peanuts compared to total public spending, so I see little excuse for borrowing.

The only possible weakness in Friedman’s argument is the well known Golden Rule, namely the idea that governments should borrow to invest. The arguments for and against the Golden Rule are complicated. Kirsten Kellerman had a paper in the European Journal of Political Economy (“On a popular misinterpretation of the golden rule…”) criticising the idea, i.e. she supported Friedman. And there’s a big question mark over the golden rule which Kellerman may or may not have mentioned, which is that education is a form of investment. But no one, for some strange reason, ever suggests that the education budget should be funded via borrowing.

But let’s assume that golden rule quandry is settled and that we agree that the optimum amount of public investment to fund via borrowing is X% of GDP and that we’re talking about peace time not war time. There is then the question as to whether government AS CURRENCY ISSUER and controller of aggregate demand should borrow. My answer to that is “never” or “practically never”. I.e. I suggest (to repeat) that if extra AD is needed, the state should simply “print and spend” and/or cut taxes. But if anyone thinks there are good reasons to borrow in the latter circumstances, I’m all ears.

[Majromax]

@Ralph:

> I.e. even the annual spend on the largest infrastructure projects is peanuts compared to total public spending, so I see little excuse for borrowing.

Tax rates are sticky. If the government decides $1bn of public spending is desirable but no AD stimulus is necessary, it is administratively simpler to raise that via a bond issue than via setting tax rates.

Furthermore, on a more practical level borrowing ensures that money is diverted from the most marginal private-sector investment. Assessing a tax is only equivalent if markets fully clear and all agents are not liquidity-constrained.

> But no one, for some strange reason, ever suggests that the education budget should be funded via borrowing.

Tell that to those with student debt.

More practically, education operating expenses are a flow, not a stock. Although the economic benefit of one teacher-year is paid off over time, we wish to have a consistent and steady flow of teacher services. That means that "funding education via borrowing" is nearly the same as "funding education via taxes," as taxing to pay down earlier-accrued education (akin to a mortgage) replaces taxing to pay for current services.

> There is then the question as to whether government AS CURRENCY ISSUER and controller of aggregate demand should borrow. My answer to that is “never” or “practically never”. I.e. I suggest (to repeat) that if extra AD is needed, the state should simply “print and spend” and/or cut taxes.

This is isomorphic to the combination of a balanced budget, a slowly-changing stock of debt, and a central bank operating through open market operations on that stock of debt. In the long run, a slightly less-than-balanced-budget might be necessary to ensure proper liquidity in the debt market.

On a practical level, "print and spend" is difficult because governments work better when the decision on whether stimulus (or contraction) is necessary is well-separated from the decision of what services the government should provide.

[Nick Rowe]

Majro and Ralph: even if taxes aren't sticky, and can be changed instantly, there's the tax-smoothing argument for not planning to change taxes. The marginal deadweight cost of an extra dollar of tax revenue is an increasing function of the tax rate. So you minimise the total deadweight cost of a given Present Value of tax revenue by holding tax rates as constant as you can over time. (It's sorta like diminishing Marginal Utility given you an incentive to buy insurance, and to smooth your consumption over time).

[Frank Restly]

Nick,

A better rule would be interest expense as a % of tax revenue. It is possible for the debt level to be a low percentage of nominal GDP and yet interest payments consume most / all of federal expenditures.

Example:
Federal Debt = $5 Trillion
Interest Expense on Federal Debt = $250 Billion
Annual Nominal GDP: $20 Trillion (Debt is 20% of nominal GDP)
Annual Tax Revenue: $250 Billion (Interest expense is 100% of available tax revenue)

This is not a sensible way to run things. Any decline in nominal GDP and tax revenue means the government starts missing payments.

Better way is to take Ralph's advice and government doesn't issue debt at all. That means either government prints and spends money into existence (bye, bye central bank) or government sells risk bearing securities (equity).

[Britmouse]

Nice post Nick. I think you're underselling a bit the benefits of NGDPLT for sensible fiscal policy; regardless of the fiscal rule, the main benefit for NGDPLT particular over IT is that tax revenue is highly predictable even in the face of significant productivity shocks. Basically every Budget for the last eight years we have had a "surprise" that output/hour is not growing at 2%, so even with CPI usually at or above target, NGDP & hence tax revenue is way below expectations.

[Nick Rowe]

Frank: there is some point in having an (interest on debt)/NGDP ratio target instead of a debt/NGDP target. Though interest rates can change unpredictably. But in both cases being able to predict future NGDP would help the government figure out if the deficit was consistent with that target over the longer term, because it would know the denominator.

Britmouse: Thanks. Hmmm. Yes, I think that's a valid point, if tax revenue is a fairly stable function of NGDP, which seems plausible. But we would also need to consider transfer payments, to make a stronger case.

[Frank Restly]

Nick,

Would it surprise you to learn that interest rates and total government debt can go up, but annual interest expense can go down? Hint, governments can borrow for terms longer than a year and don't have to pay coupons.

[Frank Restly]

Simple example:

Scenario #1:
Government borrows $100 at 5% for one year and another $40.45 at 6% for two years compounded. Interest and principle are repaid when the bond comes due.
Total Debt = $100 + $40.45 = $140.45 at start of year #1
Total repayment at end of year #1 = $105, Interest = $5
Total repayment at end of year #2 = $45.45, Interest = $5

Scenario #2:
Now suppose government instead borrows $80 at 6% for one year, $32.18 at 7.2% for two years compounded, $17.53 at 8.4% for three years compounded, and $10.84 at 9.6% for four years compounded.

Total Debt = $80 + $32.18 + $17.53 + $10.84 = $140.55 at start of year #1
Total repayment at end of year #1 = $84.80, Interest = $4.80
Total repayment at end of year #2 = $36.97, Interest = $4.80
Total repayment at end of year #3 = $22.33, Interest = $4.80
Total repayment at end of year #4 = $15.63, Interest = $4.80

And so despite taking on more debt at higher interest rates in scenario #2, the government pays less annual interest than in scenario #1.

[Nick Rowe]

Frank: No it would not surprise me. And yes I do understand why, and did not need your "hint". Nor your long-winded examples. Stop commenting now.

[Ralph Musgrave]

Nick,

Re taxes being allegedly sticky, they can’t be all that sticky given the frequency of changes in tax rates, direct and indirect in the UK. For example VAT was adjusted twice (down and then up again) in the recent crisis. And at least once a year (and more often twice a year) at budget time (and/or Autumn budget time), the UK finance minister changes various taxes: fuel taxes, income tax and so on.

Re funding education via borrowing, student loans are just a recent development, and given the problems involved they could easily be abolished. Moreover, education for those under university age is still funded via tax.

Also, the reason for student loans was not that it was suddenly thought desirable to fund education via loans per se: the reason was that while BASIC education (learning to read and write etc) is arguably a basic human right, and hence that that ought to be provided for free, when it comes to higher skill levels, there’s a good argument for having the relevant investment pay for itself in strictly commercial terms.

Re the need for a “less than balanced budget might be necessary”, I’m baffled. In the UK the finance minister George Osborne has advocated a balanced budget and has been widely ridiculed. Far as I know there is NOT ONE SINGLE ECONOMIST who supports him. I.e. there is no question but that almost permanent budget deficits are needed.

And given a need for a budget deficit, having the state print and spend, and/or cut taxes is the obviously way to do it. I.e. I don’t see the clash between “print and spend” and the need for budget deficits.

Re your final point, namely that the decision as to HOW MUCH stimulus is needed should be separate from the question as to the exact nature of extra public spending (or tax cuts) I quite agree. The former is a technical question for economists and the latter is a purely political question. But that separation is easily achieved under a “no borrowing / print and spend” regime. All you do is have some committee of economists, perhaps at the central bank, decide how much to print and spend (much like the existing Bank of England Monetary Policy Committee) while the exact way to spend that money is left entirely to politicians.

An arrangement of that sort was advocated in this submission to the UK’s Vickers commission on banking:

http://b.3cdn.net/nefoundation/3a4f0c195967cb202b_p2m6beqpy.pdf

[Britmouse]

Yup, at least in the UK tax receipts have been stable at around 33-35% of NGDP - and increasingly stable since the 1990s really - no doubt in part because changes to the tax system have become less dramatic over this period. Even in 2008/9 the cyclical fall in receipts/NGDP was less than 2pp peak to trough.