« Is a need for fiscal stimulus a symptom of a poorly-designed social safety net? | Main | Shouldn't Taylor Rules include Fiscal Policy? The Fiscal exit strategy. »


Feed You can follow this conversation by subscribing to the comment feed for this post.

Nick, A few comments:

1. As a practical matter it is impossible to index the tax system against inflation. It's just too complicated for the average person to handle, unless you abandon taxes on capital. Come to think of it, we should not tax capital, so maybe that's a good idea.

2. Pushing people into higher tax brackets with inflation might slow the economy, as you suggest. But what if the trend rate of inflation is positive? Then tax rates might rise every year, whether you are in a recession or not. They would merely rise slightly faster in booms.

3. I still think countercyclical fiscal policy makes no sense if the monetary authority is doing its job. Consider the following two monetary policies:

A. An instrument setting expected to produce on-target AD.
B. An instrument setting expected to fail to produce on-target AD.

Only in case B would fiscal stimulus be of any use at all. By why would a central bank ever choose case B?

I guess we'd have to ask Mr. Bernanke the answer to that question.

I believe Singapore has lower taxes on the employer's share of labor income during recessions. That reduces the problem of sticky wages. It also might help deal with unemployment that is not caused by inadequate nominal spending. In my view this policy could be helpful in small countries that are at the mercy of big blundering neighbors--like Singapore and Canada. For those countries a nominal shock in the big neighbor becomes a real shock to their own economy. So despite my quibbles I think you end up with some good ideas.

Scott Sumner would say you need a macro futures market, and I think this is the best one could do.

Scott: Yes, if monetary policy were perfect, we wouldn't need to worry about automatic fiscal stabilisers. But I'm a bit of a "belt and braces" man myself. Or a believer in Murphy's Law. In any case, even if expected AD were perfectly stabilised by monetary policy, I don't think that means that actual AD would necessarily be perfectly stabilised, so there could still be a role for automatic fiscal stabilisers.

You could always adjust the tax brackets and benefits to *target* inflation, if that were a constant. E.g. increase tax brackets and benefits by 2% per year regardless of actual inflation.

I don't believe that the long-run macro problem is that significant. The PBO just published a fiscal sustainability report. It demonstrates that with the expected structural deficit and their demographic and economic assumptions, Canada needs to improve the budget balance by about 1% of GDP to be sustainable through their planning horizon of ~70 years. I think that tweaking tax rates slightly to ensure debt:GDP is long-run sustainable will have minimal effects on tax policy optimized for micro and short-term macro concerns. In other words, design a system that handles the first two well, and slight nudges to this policy can take care of the long-run macro.

So as an policy to balance 1 & 2, one could look to a negative income tax, so the marginal tax rate is always >0. It could be a flat tax, so that for a given level of revenue, the marginal rate is minimized. In order for it to satisfy 2, it would need to be non-negligible, so something on the order or 30 - 50% marginal. As Scott says, we could lean more towards the micro side we can also rely on monetary policy to take care of the short-term macro. Having 30 - 50% marginal rates might mean the automatic stabilizers require less drastic swings in monetary policy, which might not be a bad thing.

Great post, Nick!

Thanks Andrew: Within the current Canadian context, I tend to agree with you on the first point. Small nudges would be enough to keep us with a long run sustainable fiscal policy, and there's no great urgency. But at other times, or in other places, it might be more of an issue. Like the US today.

What about indexation (to inflation)? On micro grounds, all tax and benefit rules should be 100% indexed to inflation. A purely nominal shock should not affect optimal real taxes or benefits.

This is impossible of course. Not to want, but impossible in practice. Even when money creation is expected in quantity; it still isn't helicopter dropped. That makes all the difference.

"On micro grounds, all tax and benefit rules should be 100% indexed to inflation."

It is impossible both in theory and in practice, as you have debt contracts denominated in nominal interest rates, which forces inflation to affect the economy.

One way you can see this clearly is to imagine a CB that tried to maintain a "real" overnight target rate:

inflation goes up --> cut the nominal rate --> inflation goes up more (all things equal).

Inflation goes down --> raise the nominal rate --> inflation falls further (all things equal)

The CB would not be able to defend a "real" overnight target rate because the equilibrium is unstable. I.e. you cannot both have a well-defined real overnight target and a well-defined price level.

Suppose, on the other hand, that the private sector signs debt-service contracts in real terms, but the CB operates with a nominal overnight target.

In that case, inflation increases and the CB hikes rates, but the non-CB rates are cut (as they are inflation indexed). This forces banks to refuse to roll-over debt at the old "real" rate and to charge a higher "real" rate to new borrowers. Therefore "real" interest rates increase purely due to nominal effects. And this will cause the "real" economy to respond to these nominal effects.

And of course if both the CB and the credit markets suffer from "money-illusion" and demand nominal rates in their contracts, then inflation will have large effects on patterns of investment and savings in the economy.

So it is impossible to have an economy with a well-defined price level -- i.e. with money -- that is not influenced by nominal effects. Not even in theory, and this is independent of menu costs or other rigidities.

I think monetary policy is a very poor tool for addressing demand failures.

Vis-a-vis the business sector, firms are funded with long term liabilities, and you do not want to artificially lower these costs in response to a temporary recession, as it will distort the cost of funds of the firm during the recovery. That is a pro-cyclical policy, in that the during the recovery, the lower cost of funds leads to artificially high returns, and then the cost of funds rise to match, leading to another bust. So you do not want to adjust a long term cost in order to provide a short term investment boost.

The best way to boost the likelihood of a business to invest in the face of a short term demand failure is to increase demand -- e.g. revenue. A stable revenue stream for the business sector is what optimizes investment decisions, and let the cost of funds float. Even without government intervention in the financial markets, rates will fall prior to a recession and they will rise when it ends. I.e. rates are already adjusting, whereas demand is not. So directly address demand, rather than trying to manipulate rates in such a way that more demand arises from borrowing.

Moreover, business cost of funds do not respond nearly as well to overnight rate cuts as does consumer borrowing -- particularly real estate borrowing. Monetary policy is asymmetric in its effectiveness, influencing bank lending rates much more than corporate bond yields or equities. And banks do not lend to productive businesses -- banks lend almost solely on real estate.

So if you disaggregate "investment" into productive and unproductive investment, then changes in the overnight rate are primarily reflected in an increase in unproductive household investment -- real estate investment, rather than in capital formation or business investment.

Therefore a policy to lower rates in response to a demand shortfall biases the economy to be more dependent on household real estate debt rather than business investment. I.e. you may emerge from the recession with robust "demand", but the composition of that demand has changed, and you are distorting the economy in a long term way by burdening households with more debt and shifting overall investment towards unproductive uses. This forces rates to keep falling, and has many other negative consequences. There are only so many bullets in that gun, and it has not served us well, even though we escaped many demand-led recessions this way.

The best way to address a demand failure is via fiscal policy, as this allows a short term increase in revenues without a long term shift in business cost of funds or a long term shift in household balance sheets. You want the intervention to be both brief and to not interfere with the process of capital allocation, biasing one form of investment over another. That alone rules out zero rate shifts, due to the enormous asymmetry they have on mortgage rates vs. baa rates. I prefer high and constant capital gains taxes (which are positively correlated with capital formation), together with high marginal rates, unemployment benefits or work programs during the bust. The exact opposite of the policies that got us into this mess.

In this way, when the cost of capital is lower than the return on capital -- i.e. unexpectedly high returns that occur during booms -- then excess capital gains and dividends will be taxed and the government will receive an increase in revenue, draining demand during those booms. When the cost of capital is higher than the returns on capital (so businesses are liquidating and unemployment is rising) the government will drain less revenue during those busts. This allows for "automatic" counter-cyclical deficit spending without biasing the overall cost of capital or investment composition in the economy.


Your analysis shows why thinking about monetary policy in terms of a overnight interest rate is a disaster.

Care to elaborate? The overnight rate is the transmission mechanism for monetary policy as it is currently implemented.

RSJ: "The CB would not be able to defend a "real" overnight target rate because the equilibrium is unstable. I.e. you cannot both have a well-defined real overnight target and a well-defined price level."

YES! This is exactly what Bill and I (and Scott Sumner, and others) have been talking about, when we say that the long run price level is indeterminate under an interest rate target. This is known as the "Wicksell Problem". And this is the underlying reason why we want to stop thinking of monetary policy in terms of setting an interest rate.

There needs to be some sort of "nominal anchor" in the system to make the price level determinate. If the CB targets some nominal variable, like the money supply, or nominal price of gold, or NGDP, or something, then the price level will be determinate. But the nominal interest rate does not count as a "nominal variable in this context; as you rightly observe, if the CB tries to hold the nominal rate constant, but everyone else cares about real rates, the resulting real disequilibrium will cause either accelerating inflation of accelerating deflation. (We can think of this as the CB needing to have "money illusion" to make the price level determinate).

Nick, it sounds like you are looking for a monetary price anchor as a replacement for the no longer applicable convertible fixed rate regime using gold as the standard. Or perhaps I'm not understanding you correctly.

BTW, MMT'ers suggest setting a price anchor in terms of the floor price of labor through a permanent job guarantee.

fire and forget forget fiscal policy?

I don't think its doable, for political and practical reasons.

1) Political:
The two sides have very different ideas about fiscal policy so as political power shifts the policy /will/ shift.

2) Practical reason:
Fiscal policy isn't completely macro scale. A lot of the point of fiscal policy is removing or creating arbitrage opportunities with government to control human behavior. This ties in with political problems, because I don't think that government will ever want to give up this power.

Ahh, I see what you mean, but monetary policy does not have an interest rate target, the target is (currently) a linear combination of output gap and unemployment. The rate is called a "target" rate between FOMC meetings, but everyone knows that the rate itself is the jumping variable. Does it matter if you are using discreet time (e.g. CB policy meetings), so that your jumping variable is constant in "real" time, or must you use a continuous time jumping variable? It doesn't seem to be an important distinction, or perhaps I am missing something?

Nick, I am not sure what you mean by "if monetary policy was perfect." My argument merely depends on it being unbiased. If monetary policy is unbiased then over- and under-shooting the AD target should be equally likely in a recession. In that case there is no argument for fiscal stimulus in a recession, indeed fiscal contraction would be equally likely to be helpful.

So if by "perfect" you mean unbiased, then yes, I am calling for perfect monetary policy. But if by perfect you mean "is expected to hit the target precisely" then perfection has no role in my argument. Is being unbiased too much to ask for from the Fed? Suppose you found that every time it rained weather forecasters were likely to overestimate the chance of rain over the following few days. Would you merely say "well, nobody's perfect?"

I think your suggestion on adjusting tax brackets for trend inflation does address the concern I had. And again, I think your proposal makes some sense for smaller countries that aren't really able to control the AD facing their export industries. I only oppose fiscal stabilization policies in countries like the US, which can control AD.

Lord, I do think a macro futures market would be nice, but that assumption is not necessary to make my argument work.

tjxfh: "Nick, it sounds like you are looking for a monetary price anchor as a replacement for the no longer applicable convertible fixed rate regime using gold as the standard. Or perhaps I'm not understanding you correctly."

I think you are understanding me correctly. I am looking for some sort of price level anchor. The gold standard is a bad one. Price level path targeting would be better, but that still leaves open the precise way to implement it (e.g. a short run interest target, as at present, or something else, which I would prefer, because interest rate control only seems to work as long as people retain their "faith" that the BoC can hit its ultimate target, and that faith was challenged recently).

"BTW, MMT'ers suggest setting a price anchor in terms of the floor price of labor through a permanent job guarantee."

Yes. In principle a policy like that could serve as a nominal anchor, if it were the *only* target of monetary policy. (Just replace "gold" with "labour"). I'm annoyed at myself right now, because about 6 months ago I managed to get my head around MMT monetary policy in this regard. And I vaguely remember there was some problem with it. I think because MMTers were also trying to hit some other (incompatible?) target at the same time? But my memory has failed me. Damn!

Doc: yes, you are probably right, unfortunately. But we could still imagine each political party putting forward its own fire-and-forget policy.

RSJ: I too don't think the distinction between continuous and discrete time should matter for this question.

But if the central bank tries to target *any* real variable, like the output gap and/or unemployment, you get exactly the same price level indeterminacy problem.

Scott: "Nick, I am not sure what you mean by "if monetary policy was perfect." My argument merely depends on it being unbiased. If monetary policy is unbiased then over- and under-shooting the AD target should be equally likely in a recession. In that case there is no argument for fiscal stimulus in a recession, indeed fiscal contraction would be equally likely to be helpful."

You are losing me here. If automatic fiscal stabilisers can work with a very short lag, shorter than the forecast horizon at which monetary policy sets expected future AD at target, then automatic fiscal stabilisers ought to be able to help keep actual AD on target. For example, if unemployment increases, the newly unemployed know they will start getting employment insurance cheques, and will start spending them, even before Statistics Canada knows that the unemployment rate has increased.

Long-term planning of public infrastructure rehabilitation & construction projects would help too. There should be some mechanism to "fast-track" already planned-for projects when there is a construction slump, and conversely to "slow-track" them when there's a construction boom.

The long-term growth, development and rehabilitation of infrastructure needs to be planned a decade in advance, which would give a 1 to 2 year leeway in "fast-tracking" or "slow-tracking" projects.

The problem we have now is that we slashed infrastructure investment far too much in the 80's and 90's to deal with deficits. Since 2000 shortages of skilled labour has made it very difficult to ramp up investment.

I think that makes a lot of sense. Thanks again, Alex.

Alex: That reminds me about the discussion back in the Fall of 2008/Winter 2009 when Stephen was saying that infrastructure might not be the best stimulus given that construction sector employment was holding up (and seems to have held up pretty well). I made the suggestion that we should have a buffer of infrastructure projects with all the engineering work done ready to be put to tender and started (maybe a three to six month lead time from pulling the trigger to work commencing). Perhaps one could use the overnight rate as the signal, that is, once the central bank begins cutting rates.

Andrew F: "Perhaps one could use the overnight rate as the signal, that is, once the central bank begins cutting rates."

And even on standard Cost/Benefit micro grounds, that would make sense. The NPV of an investment project will be higher the lower are interest rates. So there are infrastructure investment projects that would not meet the positive NPV criterion at high interest rates, that would have positive NPV at a low enough interest rate. (Though, it's usually a longer-term interest rate that is relevant here).

Or indexing unemployment compensation and setting its duration proportional to unemployment futures. Or automating statistics gathering to provide real time economic information. Most are still geared towards quarterlies though which limits how dynamic this can be. Direct response programs like unemployment and welfare would be more dynamic than tax policy unless withholding can be adjusted in real time. I don't see any reason fiscal and monetary policy can't complement one another catching each others deficiencies.

I would also like to say shifting the duration of treasury debt over the cycle, the longer the cycle the longer the duration and reversing in downturns to reduce long rates as well, but I am not sure how much of a shift is possible in short order.

Nick, I don't see how fiscal policy can be defended on the shorter lag argument. In the case of the US, it has been 18 months since monetary policy went off course. If the Fed had been targeting 12 month forward NGDP in September 2008, there would have been no need for Congress to have passed a fiscal stimulus in early 2009. Remember, one of the most powerful influences on CURRENT AD is future expected monetary policy, and future expected NGDP. If expectations of NGDP 12 months forward had stayed on target, then 3 to 6 month forward NGDP expectations would not have fallen enough to justify fiscal stimulus.

I don't generally agree with people like Krugman on fiscal policy, but I'd also note that he claims that almost all the punch comes from spending, not taxes and transfers. But spending on government projects is precisely where the long lags occur. Perhaps tax rebates can help a bit, but I doubt they have a significant impact.

Perhaps fiscal stimulus can help a bit when NGDP remains depressed for many years, but of course that is virtually the definition of inept monetary policy.

The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad