"Does monetary policy have (bad) distributional consequences?"

Some questions are bad questions. This is one of them. We can get a clearer and more useful answer if we change the question. We can avoid wasting a lot of time arguing at cross purposes.

Here's a better question: "If we used fiscal policy instead of monetary policy to remove a shortage of aggregate demand, would that switch from one policy to another have distributional consequences?" The answer is almost certainly yes, though what those distributional consequences are will depend on the exact nature of the fiscal policy.

Here's an even better question: "If we assume that monetary policy always responds to changes in fiscal policy to keep aggregate demand at the same level, will changes in fiscal policy have distributional consequences?" Now the answer is certainly yes, though what those distributional consequences are will depend on the exact nature of the fiscal policy.

[Update: now I think about it, I think my third question is actually the same as my second question; just a clearer way of saying it.]

It all comes down to the "assignment question": what policy instrument is assigned to which target? Fiscal policy is multi-faceted, because there are lots of different types of government expenditures and lots of different types of taxes. Monetary policy isn't. You need lots of numbers to describe fiscal policy, and only one number to describe monetary policy. Fiscal policy can in principle be used to change the distribution of income/wealth in almost any way you want; monetary policy can't. It makes sense to assign income distribution to fiscal policy and aggregate demand to monetary policy. If you don't like the distribution of income, then change fiscal policy, and let monetary policy adjust in response to keep aggregate demand at the right level.

This post is in response to many people I have heard asking the title question and arguing against using monetary policy to increase aggregate demand because of the distributional consequences. But most recently it is in response to Ashwin's post in Macroeconomic Resilience.

"A quantitative easing program focused on purchasing private sector assets is essentially a fiscal program in monetary disguise and is not even remotely neutral in its impact on income distribution and economic activity. Even if the central bank buys a broad index of bonds or equities, such a program is by definition a transfer of wealth towards asset-holders and regressive in nature (financial assets are largely held by the rich). The very act of making private sector assets “safe” is a transfer of wealth from the taxpayer to some of the richest people in our society. The explicit nature of the central banks’ stabilisation commitment means that the rent extracted from the commitment increases over time as more and more economic actors align their portfolios to hold the stabilised and protected assets."

There's a lot I would disagree with in that paragraph.

First, it is the central bank's threat to buy an indefinitely large amount of an indefinitely wide variety of assets for an indefinitely long time, conditional on NGDP (or whatever) being below target, that does the job of increasing expected and hence actual aggregate demand. Not the actual purchases.

Second, the large current size of central banks' balance sheets, and the high prices of the assets held by central banks, is a consequence of tight monetary policy, not loose. A credible threat to buy more assets conditional on NGDP being below (a sensible) target would mean central banks would have to shrink their balance sheets by selling assets, and would cause the prices of those assets to fall. Low nominal interest rates are a consequence of low NGDP expectations.

Third, monetary policy works not by changing central banks' (current and expected future) stocks of assets but by changing central banks' (current and expected future) stocks of liabilities. Anybody can buy, sell, and hold those assets; only central banks (and counterfeiters) can issue high powered money. Central banks are small players in the market for all other assets, but are very big players indeed in the market for high powered money.

But mostly, I think Ashwin is asking the wrong question. I can't stop people asking the question "would loosening monetary policy to increase aggregate demand cause the distribution of income/wealth to worsen?". My guess is that some people would indeed be made worse off, like people who hold a high proportion of their wealth in safe nominal bonds (though I'm not sure if that counts as a worsening of inequality or not). But I can't imagine that anyone would use that as a reason for not increasing aggregate demand. After all, those who would probably benefit the most would probably be those with the highest risk of unemployment, who are likely to be among the most poor. I hope nobody would argue "we just have to live with the recession, because if we cured it some rich people would benefit".

It all comes back to the question of the assignment of targets to instruments. Like most economists for the last 40 years, I would assign aggregate demand to monetary policy and the distribution of income to fiscal policy. You use monetary policy to target aggregate demand, because that's about the only job it's good at, and you don't want to ignore an instrument when you don't have enough instruments to leave one idle. And you use fiscal policy to target income distribution, because it's good at that job and monetary policy is bad at that job. Then ask: "if monetary policy responds to fiscal policy (and everything else) and keeps aggregate demand where it should be, will fiscal policy changes have distributional consequences?" The answer is: "of course they will". It's just standard micro public finance.

Comments

[K]

Nick,

The issue to me is any kind of discretionary targeting. Look at QE2. The market rallied very strongly in the months preceding the announcement (nothing to do with anticipated asset purchases or monetary expansion - it was just anticipation of a more long range low rate policy). How did the market anticipate the announcement? Who knows. Nobody told me about it. But every once in a while there were news stories coming out that the FOMC was increasingly ready to act. If I were CEO of a large US bank and could get an FOMC member on the line whenever I wanted, I'd appreciate that. "Don't tell me how you're going to vote at the next meeting, of course. That would be wrong. I'm just calling to see how you feel about the economy, and to let you know what I'm seeing. Just from a *purely* personal perspective!" All perfectly legal, BTW. Ka-ching!

That's the problem with discretionary policy of an unaccountable
regulator: trillions of dollars of rents to be collected. This a problem with *any* imperfectly defined target. Even an NGDP, or capital market target is vague unless the Fed says announce once and forever *exactly* how any deviation will be returned to target. There's always discretion. Which brings us back to the issue of "targeting" vs "backing." While similar in principle, backing with capital assets (stocks and bonds) is not discretionary, and is therefore neutral in practice as well as in principle.

[Nick Rowe]

K: I think I agree with your rules vs discretion point.

I'm not sure I follow your backing vs targeting point. Maybe you are saying that the *composition* of assets in the central bank's balance sheets will have distributional consequences (because it's effectively fiscal policy), and should not be left to discretion. Fair point. Tricky when we get to LOLR operations, but I wouldn't deny that LOLR operations have distributional consequences.

[K]

No, I wasn't clear, Nick. I'm in favour of backing because it *doesn't* have distributional consequences. By "backing" though, I don't mean the Fed buys stocks with its senior liabilities and returns the profits/losses to the shareholders (government). That doesn't tie the unit of account to the value of the assets, because there's a junior claim (equity) on the assets of the Fed. Purchases therefore have significant consequences, as does purchase policy. That opens us up to all the problems of discretionary policy.

What I'm suggesting is that we create a huge open-ended fund that buys a broad cross-section of capital assets. Anyone can redeem their fund units for underlying assets, or vice versa. Now the value of the units is just the value of the assets divided by the number of units. It never changes as units are created or destroyed, and it can't ever go broke. The Fed doesn't care about the number of fund units which is not an element of policy. It's endogenously determined. The only decision is the purely nominal decision on the split rate of the units which, ideally, is set once and forever, but not a disaster if it has to be changed.

[dlr]

K -- do you anticipate these fund units become the medium of exchange? In any case, how can you be confident that changes in time preference, real expected returns on marginal investments or risk aversion (for RBC or sunspot reasons) won't cause welfare reducing fluctuations in prices and AD problems?

Mike Sproul or maybe even John Cochrane (or a more zealous FTPL proponent) might argue we already have a price level determined by the real backing of capital assets, except that the capital assets are reflected in the government budget constraint which derives its value from the interaction of taxing power and private capital -- regardless whether it its directly convertible or explicitly held by the government.

[Mike Sproul]

dlr:

Yep; that's what I would say all right. I think Cochrane would generally agree, but unfortunately he never seems to blog about it.

I would add that when the Fed issues 100 paper dollars in exchange for $100 worth of bonds, stocks, land, furniture, or whatever, nobody's net worth is affected, so there are no distributional effects. If I really wanted to rant and rave, I'd go on to say that aggregate demand is a nonsensical concept, but one thing at a time.

[Alan T.]

Nick,

I find it striking that you take deliberate redistribution for granted in your post. Here in the U.S., I talk to many people who say that redistribution of income or wealth is theft. How should I reply? In my experience, economists often refuse to answer that question, on the grounds that the ethics of redistribution, like any other question about ethics, is not an economic question. But I think it is important for public policy to be able to justify redistribution.

[Farid Elwailly]

At the zero bound the Fed would still be able to target NGDP by threatening to use (or actually using) unconventional means. Why isn't a signal to the Federal government that an increase in fiscal spending is indicated and will be allowed to boost AD a valid "unconventional" tool. I think this exclusion of fiscal spending as an AD boost at the zero bound is artificial and unwarranted. It smells of ideology to me based on the notion some have that fiscal spending will somehow cause more harm in all cases.

[K]

dlr: "do you anticipate these fund units become the medium of exchange?"

I support equal access of all agents in the economy to any official medium of exchange. An electronic transaction account, no paper claims. To the extent that others (banks, credit card issuers, whatever) want to provide alternative exchange media, linked to the official unit of account, for anonymous transaction purposes, or whatever, I don't care.

"In any case, how can you be confident that changes ... won't cause welfare reducing fluctuations in prices and AD problems?"

Though nominal rigidities are important, personally, I don't think they are anywhere near as important in driving fluctuations in AD as are (perfectly rational) changes in people's subjective expectations in the face of a totally unknowable future. It is the availability of money as a (false) consumption guarantee that enables the dynamic that transitions the economy into the paradox of thrift sunspot. In a world of Bayesian subjective probability measures posterior measures become very fat tailed and the system becomes extremely fragile against large shifts in consensus expectations.

Since nominal rigidities are mostly *downward rigidities*, I don't see nominal rigidities becoming a serious problem if you run a relative high split rate (interest rate), e.g. Nick's suggested 7%. I'd even go for 5%. The nominal interest rate would then be 5%. NGDP growth might also be at 5%, if, as I suspect, people do not turn out to be risk averse vs the general market portfolio (beta=1) once the numeraire becomes that
portfolio. But if people are still market risk averse, then NGDP growth would run under 5%. I don't see how NGDP growth could run unusually low for very long when the market portfolio is guaranteeing 5% nominal annual returns. Think about what would happen to fear if we could never observe the broad market going up and down. The only thing that would be moving is relative asset prices and some exotic derivatives known as "real return bonds," that very few people ever trade. Would we have real return bond bubbles, instead of market crashes? Possibly, but I doubt it.

If you take away risk premium fluctuations (a contradiction in terms - risk premia *can't* fluctuate in intertemporal equilibrium or we'd have to have an aversion to risk premium risk), there is not much risk in the market. This is at the root of the equity premium puzzle. In a zero risk-premium, 5% nominal return world, the only risk is how future nominal output is going to be split between capital, labour and government. Doesn't sound very "risky."

"Mike Sproul or maybe even John Cochrane (or a more zealous FTPL proponent) might argue we already have a price level determined by the real backing of capital assets"

I would say that too. Almost. It has the real backing of taxation which is backed by the value of capital and wages. But it's not *determined* by the value of capital because the CB targets inflation. I.e. the tax "asset" of the government is *sufficient* to cover the CPI guarantee, but the value of money is not equal to value of the capital assets that ultimately back it. The ability to tax may ultimately *not* be sufficient to make good on any given guarantee (inflation) which is one more reason why capital asset value *targeting* is dangerous. We need actual backing.

[Mike Sproul]

K:
"the tax "asset" of the government is *sufficient* to cover the CPI guarantee, but the value of money is not equal to value of the capital assets that ultimately back it."

Exactly. The value of money looks like a written put, with the issuer's assets on the horizontal axis and the value of money on the vertical axis. We could also say that money is priced like corporate bonds, in the sense that the issuer's net worth can usually fluctuate without any effect on the value of the bond. But if the issuer's net worth becomes negative, then the value of the bond will move in step with the issuer's net worth.

[Steve Roth]

"it is the central bank's threat to buy an indefinitely large amount of an indefinitely wide variety of assets for an indefinitely long time, conditional on NGDP (or whatever) being below target, that does the job of increasing expected and hence actual aggregate demand. Not the actual purchases."

It is the *threat* that I will shoot you that causes you to turn over your wallet. Not the actual shooting. (The shooting can have the desired effect, but it's unnecessary if the threat is credible.)

But you and I both know that I'm going to be stealing your wallet every day for the rest of our lives. If I don't shoot you today when you refuse to hand it over, the threat won't be credible tomorrow.

So yeah: the threat generally does the job. But the shooting can also do the job, and periodic actual shooting is necessary for the threat to do the job.

[Scott Sumner]

Excellent. I was going to write a similar post in response, but yours is better than what I would have written. There's a simple counter-argument to Ashwin. In Australia the base is 4% of GDP, and in Japan it's 22% of GDP. By his logic Japan is "using monetary stimulus" and Australia isn't. Obviously the opposite is true.

[Ralph Musgrave]

Nick,

Strikes me your main criticism of Ashwin’s argument (para starting “But mostly…) consists of putting words into Ashwin’s mouth. Ashwin (according to you) argues that because a monetary boost has undesirable distributional effects, therefor a boost or “increase in aggregate demand” is undesirable.

Nope. He simply argues (as did I on my own blog on 11th March this year) that the undesirable distributional effect of stimulus done by monetary policy alone is an argument for not doing stimulus this way. (I actually argued that the distinction between monetary and fiscal policy be abolished, i.e. that every extra dollar of fiscal spending should be new money – which is very much what MMT advocates.)

Re your point about the effectiveness of the “central bank's threat to buy an indefinitely large amount . . . assets . . ”, this “threat” is currently very empty, isn’t it? That is, central banks are quite clearly operating on the basis of “try a little QE and see what happens”. They certainly are not threatening to buy mega quantities of assets relative to the quantities already bought.

[ac]

This paper (http://emlab.berkeley.edu/~ygorodni/CGKS_inequality.pdf) assesses empirically whether expansionary monetary policy in the US increases or decreases inequality. The answer is that inequality FALLS with expansionary policy.

[David Pearson]

Nick,
I don't think Ashwin is asking the question, "is QE a good policy assuming it works?"

[Nick Rowe]

Alan T. : "I find it striking that you take deliberate redistribution for granted in your post. Here in the U.S., I talk to many people who say that redistribution of income or wealth is theft. How should I reply?"

Dunno. You might say that fiscal distribution does in fact take place, whether they agree with it or not, and they need to recognise that fact.

I think I would say: Look, we talk about what is the *best* monetary policy, and fiscal policy, so we are already making some sort of ethical judgments. I would then sketch out two different ethical approaches to redistribution: a utilitarian one "the poor get greater benefits from an extra $ than the rich"; and a sort of contractual/public good one "Look, most of us rich want to give to the poor, but none of us is rich enough to handle this all ourselves, so we join a club and all chip in to help the poor, and we don't want any free riders in our club, so you have to pay your dues".

[Nick Rowe]

Farid: both the opposition to and the support of fiscal policy can smell of ideology to me. Both opponents and proponents of fiscal policy are very tempted to argue for a lower or higher *level* of G, when fiscal policy for AD control is about making G (in Old Keynesian models) or the growth rate of G (in New Keynesian models) *vary systematically inversely with the business cycle*. Good fiscal policy cuts both ways.

But that still leaves open the question: is it a good thing, on *microeconomic* grounds, to have G vary countercyclically? Ideally, we should be building lots of schools in years when there are lots of kids, not when there's a recession on. Let monetary policy handle the recession, let fiscal policy handle getting the number of schools to match the number of kids.

[Nick Rowe]

Steve: "So yeah: the threat generally does the job. But the shooting can also do the job, and periodic actual shooting is necessary for the threat to do the job."

The shooting analogy isn't perfect. Because if the conditional threat to buy assets is credible, the central bank would actually sell assets. I threaten to go to the North of the herd of cattle, to drive them South. But if my threat is credible, and they head South, I actually follow the cattle south.

Scott: thanks!

Ralph:"...this “threat” is currently very empty, isn’t it? That is, central banks are quite clearly operating on the basis of “try a little QE and see what happens”. They certainly are not threatening to buy mega quantities of assets relative to the quantities already bought."

They aren't really threatening at all. They ought to be making a *conditional* threat: "if NGDP is less than x we will buy unlimited amounts, and if NGDP is above x we will sell unlimited amounts". Instead it seems to me to be precisely what you say; "let's try a little QE and see what happens"!

[Reverend Moon]

@Scott Sumner
What does the base have to do with anything?

And can you please explain what YOU mean when you say things like tight or loose monetary policy?

[Nick Rowe]

Rev Moon: since Scott is busy, and is perhaps unlikely to check back in here, and I can guess how he would reply anyway, he would reply:

Tightness or looseness of monetary policy is measured by how slowly or quickly expected future Nominal GDP is growing.

The monetary base is the medium of account, and what central banks ultimately control.

[Mark A. Sadowski]

Another question might be if we use monetary policy to stimulate aggregate demand will this have a positive effect on equality?

After experiencing over 30 years of declining rates of inflation and simultaneously rising levels of inequality the answer to this question would seem to be intuitively obvious. And in fact the empirical evidence strongly suggests that the answer to this question is in fact yes.

There appears to be four channels by which this comes about:
1) Factor Shares of Income
2) Savings Distribution
3) Earnings Heterogeneity
4) Financial Income

1) Factor Shares of Income
Disinflation during the eighties and the nineties was been accompanied by a significant rise in the profit share of national income in most OECD countries or, equivalently, by a reduction in the labor share. This suggests that changes in the rate of inflation are non-neutral with respect to the distribution of factor income. The consequences of inflation upon inequality thus may largely be the indirect result of the effects of inflation upon factor shares. The mechanism by which this comes about is fairly simple. Accelerating inflation is correlated to falling unemployment rates, falling unemployment rates lead to greater labor bargaining power, and greater labor bargaining power is correlated with lower markups. Furthermore, higher inflation rates create greater price dispersion leading to greater competition among producers to limit markups. This hypothesis has been tested with a panel of 15 OECD countries over the period from 1960 to 2000 and a robust positive relationship between inflation and the labor share was obtained:

http://pareto.uab.es/wp/2000/46000.pdf

2) Savings Distribution
An unexpected increase in interest rates or decrease in inflation will benefit savers and hurt borrowers thereby generating an increase in consumption inequality to the extent that savers are generally wealthier than borrowers. This hypothesis was tested with respect to the United States over the period from 1952 to 2004 and it was found that within the household sector, the main losers from accelerating inflation were rich, old households, whereas the main winners were young, middle-class households with mortgage debt:

http://www.stanford.edu/~schneidr/inflation1_7.pdf

3) Earnings Heterogeneity
Labor earnings are the primary source of income for most households and these earnings may respond differently for high-income and low-income households to monetary policy shocks. This could occur, for example, if unemployment disproportionately falls upon low income groups. This hypothesis was tested using US data from 1948 through 2002 and found that disinflationary monetary policy typically lowered the employment to population ratios of minorities and less-skilled by increasing their unemployment rates and not decreasing their labor force participation rates. Furthermore African American unemployment rates, particularly teen African American unemployment rates are more sensitive to disinflationary monetary policy than whites and out-of school teenagers and out-of- school men and women with no than a high school degree (ages 16 to 24) bore the brunt of disinflationary monetary policies through higher unemployment rates:

http://www.npc.umich.edu/publications/working_papers/paper10/03-10%20rev.pdf

4) Financial Income
It appears to be an empirical fact that aggregate financial income tends to rise sharply while business income declines after contractionary monetary policy shocks. While the decline in business income is much larger, it is offset for high income households by the increase in financial income. Further, the top 1% of the income distribution receive approximately 30% of their income from financial income, a much larger share than any other segment of the population. This suggests that total income for the top 1% likely rises even more than for most households after contractionary shocks. The financial income channel of income inequality has been examined by the following paper:

http://emlab.berkeley.edu/~ygorodni/CGKS_inequality.pdf

This paper also finds additional evidence in favor of the earnings heterogeneity hypothesis by showing that contractionary monetary policy shocks not only lower labor incomes at the bottom end of the distribution, but raise labor incomes at the upper end of the distribution.

The paper concludes with the following rather relevant and provocative observation:

"Finally, the sensitivity of inequality measures to monetary policy actions points to even larger costs of the zero-bound on interest rates than is commonly identified in representative agent models. Nominal interest rates hitting the zero-bound in times when the central bank’s systematic response to economic conditions calls for negative rates is conceptually similar to the economy being subject to a prolonged period of contractionary monetary policy shocks. Given that such shocks appear to increase income and consumption inequality, our results suggest that standard representative agent models may significantly understate the welfare costs of zero-bound episodes."

And finally, for those who are convinced that discretionary fiscal stimulus must have a positive impact on equality I leave the following paper:

Distributional Impact of the American Recovery and Reinvestment Act: A Microsimulation Approach
by Ajit Zacharias, Thomas Masterson, and Kijong Kim

Abstract:
"Over the last two decades, those at the bottom of the income scale have seen their incomes stagnate, while those at the top have seen theirs skyrocket. Without intervention, the recession that began in December 2007 was likely to exacerbate this trend. Will the American Recovery and Reinvestment Act of 2009 (ARRA) be able to keep the situation from getting worse for those at the bottom of the income scale? Will ARRA reverse the upward trend in inequality that we have seen in the recent past? We employ a microsimulation of ARRA to address these questions. We find that, despite a large amount of job creation, ARRA is likely to have little impact on overall income inequality, or on the income gaps between relatively advantaged and disadvantaged groups."

http://www.levyinstitute.org/pubs/wp_568.pdf

[Steve Roth]

"Steve: "So yeah: the threat generally does the job. But the shooting can also do the job, and periodic actual shooting is necessary for the threat to do the job."

The shooting analogy isn't perfect. Because if the conditional threat to buy assets is credible, the central bank would actually sell assets. I threaten to go to the North of the herd of cattle, to drive them South. But if my threat is credible, and they head South, I actually follow the cattle south."

But if it's a repeated game, if you never head north the threat isn't credible.

Sure, if the threat is credible based on past experience, simply setting expectations works.

But it's not turtles all the way down.

[Bonnie]

I think the issue of inequality itself is subject to interpretation and generally has a perceptional problem that the only way to foster equality is by taxation, government expenditure and through interventions in markets as the people who benefit from transfer payments supposedly have a better quality of life. Even if one agrees that there is an inequality problem, of which I am skeptical because people move in and out of degrees of wealth or lack thereof over time and buying the concept at face value means accepting an assumption of a static population of want, does it necessarily mean that solving it in such a way is the only way? It seems archaic and draconian to me.

The statistics on efficiency of government transfer programs don’t stack up to what one might expect from the levels of spending, revenue and borrowing. Implications from the statistics, which are casually brushed under the rug, is that there are way too many undeserving hands dipping in the pot as the money goes from taxpayer or bond buyer, through the bureaucracy, and finally makes it to recipients with little left over to serve the initial intent. That is a byproduct of handling this issue within the political system and the structure of transfer programs, and there is scarcely any difference in effect between fiscal stimulus and monetary stimulus when considering only whether rich people are getting richer as a result. The only difference is which set of rich people are becoming richer, asset holders or politically connected rent seekers.

We are apparently in a situation where we have a choice to make between options that are satisfying to almost no one for one reason or another, but as I see it, doing nothing leaves the people who are currently most in need worse off and adds to their ranks by the day. We have lived with the drawbacks of both methods of providing adequate levels of liquidity in the system for decades and I don’t understand why now seems like a really good time to be politically paralyzed with everyone wanting it their way or the highway. Everyone loses in that environment, and I think it would be much better if we could come to some sort of understanding about how to stop the bleeding and move forward from here.

[Ritwik]

Nick

I don't think you're avoiding arguing at cross purposes here. You've re-assumed everything that Ashwin doesn't want you to assume. The entire monetary/ fiscal AD/redistribution split assumes that monetary policy is distributionally neutral. Ashwin makes the claim that it it not, perniciously not in the case of QE type asset purchases. If your argument is that fiscal policy should be set up in such a manner to neutralise any distributional effects caused by monetary policy, then you're asking it to hit a constantly moving target.

Ashwin further makes the claim that in the world that exists today, a quasi-fiscal helicopter drop is the best central bank policy, as it is truly distributionally neutral. He also says in the post that a central bank targeting a given market price has a simpler job establishing its credibility (an hence achieving its aim)than one that is trying to purchase assets of certain durations and risk profiles to stimulate the economy. These are specific, fine-grained claims about 'which' monetary policy and you've done a bit of disservice to the discourse by bringing it back to the paleo fiscal/monetary distinction and re-hashing the point about a credible central bank not having to much. You wanted to avoid cross-talk, but you're talking straight past his points!

[Ashwin]

Nick - on your first and second points, the first two-thirds of my post explains that in a world where significant negative interest rates on reserves are not feasible and government bonds function as safe repo-able collateral in the banking system, the threat of buying government bonds is an ineffective threat. Without the ability to enforce negative interest rates on reserves, the very concept of high-powered money is meaningless.
On the third point, the BoE by now owns around a third of the existing stock of UK govt bonds, the SNB has amassed around 300 bn CHF of foreign currency reserves in its bid to maintain the CHF peg so central banks are most certainly not small players in any of these markets.

If the CB is targeting a market variable such as the CHF peg, then the argument that the bazooka never needs to be used may be valid. But this is not the case with NGDP targeting. Even with market variables, the SNB has accumulated 300 bn in forex reserves in the process of defending its peg so the argument is disproven in the real world.

On the neutrality question, I'm suggesting that we start with a more neutral solution and you're suggesting that the distortionary regressive policy is fine because we can combine it with another tax to remove the distortion. IMO expecting the fiscal authority to search out all those who benefit from Fed kremlinology and tax back their gains is rather optimistic.

[Nick Rowe]

Hi Ashwin!

Perhaps I should have been more clear. My post was not a general response to all that is in your post. I was responding only to one small part of your post, where you made a point about distribution, that others too have made.

Suppose AD is currently too low, and we use monetary policy to increase AD to where it should be (you and I will disagree on whether that's possible, but that's not what this post is about). Many people (most notably the unemployed, but also some asset holders) will gain as a result. Some people (those who hold safe nominal assets like government bonds) may lose as a result.

"IMO expecting the fiscal authority to search out all those who benefit from Fed kremlinology and tax back their gains is rather optimistic."

That is not how I would want fiscal policy to respond, even if it were feasible. For example, I would not want fiscal policy to increase taxes on those who got jobs and use the proceeds to compensate government bondholders. Instead, I would prefer fiscal policy to do its traditional job, basically put higher taxes on the rich and pay benefits to the poor. The job of fiscal transfers is not to try to restore the distribution of income to what it would have been under bad monetary policy where there is mass unemployment, for example.

[Ashwin]

Nick - apart from the point you made on distribution, you have three very specific disagreements all of which I have rebutted either in my post or in the previous reply. The paragraph which you quote built on the very specific institutional arguments that I had made in the earlier sections of the post.

On the point on distribution, I fail to see how money-financed helicopter drops are not a superior solution to asset purchases - same impact, less distortion.

[Nick Rowe]

Ashwin: a helicopter drop equals (just by the math): an open market operation purchase of bonds + a lump sum transfer payment financed by the sale of bonds. All I've done is broken it down into two parts, monetary and fiscal.

Now let's ask my question: suppose monetary policy kept AD to where we wanted it to be. Would you also want to do a bond-financed lump sum transfer payment to all people currently alive, in order to improve the distribution of income? If you answer "yes" to that question, then go ahead and do it. I think I would answer "no" to that question, at least for countries with a relatively high debt/GDP ratio, because I would worry about the burden of future taxes. You might disagree. But regardless of how we would answer that second question, I think my way of asking the question is clearer. It lets us separate out the effects of increasing AD and the effects of fiscal policy on the distribution of income.

Let me try this thought experiment, to try to get to the root of our differences. Suppose you were in charge of fiscal policy. And suppose the economy were humming along nicely, with AD where it should be, and you had fiscal policy and the distribution of income where you wanted it to be (given the limitations of fiscal policy). Then all of a sudden the central bank screwed up, tightened monetary policy, caused AD to fall, and caused a recession. What would you do?

Would you:

A. Tell the central bank to stop screwing up, and loosen monetary policy to get AD back to where it was, and leave fiscal policy at the optimal setting you had it at before?

B. Ignore the central bank, and change fiscal policy to increase AD and move fiscal policy away from the optimal setting for distribution?

I would choose A. If you choose B you are throwing away one policy instrument, which means you face a worse trade-off with your remaining policy instrument.

Let me try a third tack. When you think about the distribution of income, what matters to you? Is it:

A. the actual distribution of income (e.g. the Gini or something like that)?

B. The distribution of income relative to what the distribution of income would be if AD is too low and the economy is in a recession (i.e. who gains and who loses as a result of getting the economy out of a recession by increasing AD)?

I would answer A. In my opinion, the distribution of income in a recession has no moral claim to be any sort of status quo against which to consider gains and losses. And when you talk about who gains and who loses from using monetary or fiscal policy to escape a recession you are giving the distribution of income in a recession some sort of moral status it does not warrant. The distribution of income in a recession is in no sense some sort of "natural order of things" which needs to be preserved.

What matters is for distribution is how rich and poor people are when we are out of the recession. That's what fiscal policy should address. It does not matter how rich and poor people are relative to how rich and poor they were in a recession.

(I'm still not sure I am explaining myself clearly enough).

[Nick Rowe]

Sometimes, just sometimes, like now, I wish I were better at math. Then I could talk about envelope theorems, when you have two targets (AD and income distribution), and two instruments (monetary and fiscal).

[Min]

The threat is greater than the execution, as a rule. However, this sounds like a hollow threat.

"First, it is the central bank's threat to buy an indefinitely large amount of an indefinitely wide variety of assets for an indefinitely long time, conditional on NGDP (or whatever) being below target, that does the job of increasing expected and hence actual aggregate demand. Not the actual purchases."

If you don't buy stuff, then I am going to buy stuff. Even though my buying stuff doesn't really do much. What kind of a threat is that?

[Nick Rowe]

Min: an indefinitely large and indefinitely permanent increase in the stock of base money is a bit different from you and me buying stuff.

[Ashwin]

Nick - let me break down my arguments. My first point is that when the CB pays interest on reserves, there is nothing such as high-powered money and there is no hot potato effect from asset purchases of any sort. The monetarist argument is to enforce a penalty rate on reserves to get rid of excess reserves (As Scott Sumner suggests) - my argument is that small negative rates make little difference to safe-haven flows and large negative rates will lead to people hoarding bank notes.

In such an environment, asset purchases by the CB only work to the extent that they have fiscal effects - so the comparison is not between monetary and fiscal policy but between two types of fiscal policy. The expectations channel works not via the hot potato but via the various channels through which asset price inflation leads to CPI inflation or NGDP growth.

In essence, I'm asserting that the CB buying private sector assets is fiscal/industrial policy. Instead I'm suggesting that we adopt that tried and trusted method of all banana republics in creating inflation which is monetisation of fiscal deficits - except that we do it in a systematic and non-corrupt manner that benefits every citizen equally.

[Nick Rowe]

Mark: Bottom line: both theoretically and empirically, a tightening of monetary policy would worsen inequality and a loosening of monetary policy would improve equality.

Sometimes I wonder (I think others have wondered) if there isn't a conspiracy between those in safe jobs and those with safe nominal assets to keep monetary policy tight and the economy in recession. Us boomers are to blame?

[Nick Rowe]

Ashwin: currently, the Fed's and the Bank of England's balance sheets are abnormally large. Any policy that succeeded in getting AD (NGDP) higher and growing again at a reasonable rate would mean those balance sheets would have to shrink.

Suppose we decided to use money-financed transfer payments (helicopter money) instead of money-financed purchases of assets as the mechanism by which we would get AD higher and growing again at a reasonable rate. If that policy were successful, and AD did indeed increase and start growing again at a reasonable rate, the central bank's balance sheet would need to contract, the stock of base money would need to fall, and we wouldn't be using helicopters. We would be using vacuum cleaner operations to suck money back out of people's pockets and burning it. A credible threat to fly the helicopters if NGDP was too low, if it succeeded in raising NGDP to target, would mean we would have to put those helicopters into reverse, and make them into vacuum cleaners. What your policy would really mean is not lump sum money-financed transfer payments, but lump sum money-destroying taxes. That is not going to be good for the distribution of income. The fiscal aspects of that policy are really bad.

The alternative policy, of shrinking the balance sheet by selling off assets, is a much better alternative.

[K]

Nick,

"Any policy that succeeded in getting AD (NGDP) higher and growing again at a reasonable rate would mean those balance sheets would have to shrink."

Not so.  Assuming no IOER, the quantity of excess reserves and the short rate are complementary variables. I.e. the quantity of excess reserves can be non-zero if *and only if* the interbank rate is zero; the interbank rate can be non-zero if *and only if* the quantity of excess reserves is zero. So the size of the balance sheet can be anything *whatsoever* (and is economically irrelevant) until the CB decides to actually hike the short rate. Then excess reserves have to be zero. So the *only* economically relevant parameter is the time at which the CB says it's going to leave the ZLB (and of course the path of rates thereafter). There is no point in the future in which the actual quantity of excess reserves makes any difference whatsoever.

[Nick Rowe]

K: if NGDP were higher and growing at some reasonable rate the overnight rate (interbank rate) would have to be above zero.

[K]

"if NGDP were higher and growing at some reasonable rate the overnight rate (interbank rate) would have to be above zero."

OK...? I'm not sure how that relates to what I'm saying. Just to be clear: at the ZLB, the quantity of excess reserves, current or future, has *zero* impact on NGDP. Zero! And after we exit the ZLB, the quantity of excess reserves is... zero!

So what I'm saying is that the path of the size of the Fed's balance sheet is a ridiculously elaborate ritual dance the only consequence of which is the time at which the path hits zero, an utterly bizarre manner of communicating the timing of the first rate hike.

*Then* there's the asset side of the purchases which will or will not have stimulatory effects depending on the beta of those assets. If buying positive beta stocks can make the market go up, then it's absurd to believe that buying negative beta bonds can do the same. And if bonds are positive beta, then I think that introduces some doubt whether the economy is desirous of more inflation.

[Nick Rowe]

K: "So what I'm saying is that the path of the size of the Fed's balance sheet is a ridiculously elaborate ritual dance the only consequence of which is the time at which the path hits zero, an utterly bizarre manner of communicating the timing of the first rate hike."

And I would say that communicating the timing of the first rate hike is a bizzarre and very disfunctional manner of communicating the central bank's NGDP target path.

[K]

Nick,

I have no issue with your NGDP target. I am asking how the bank *achieves* the target. What I'm saying is that there is nothing in the path of the quantity of excess reserves other than the time at which it hits zero that has *any* impact on that target. The path of rates *does* impact it. What is your mechanism by which the quantity of reserves affects NGDP while the short rate is at zero? (That's *all* you got given that quantity is zero after we leave the ZLB).

[Ashwin]

Nick - the vacuum cleaner can simply be a broad-based consumption tax.

And in a system with IOER, the Fed can even raise rates without hoovering up all the excess reserves. In fact, that is one of the benefits that the Fed itself keeps referring to - IOER gives it the ability to engineer a graceful exit. And this is not extraordinary - the ECB, BoE and others have been running similar corridor systems for a long time. For example, a large part of the ECB's LTRO operations came straight back to it via its deposit window (deposits with the ECB pay interest).

[Nick Rowe]

K: You want to know what concrete steps the bank should take?

[K]

Nick: "The Fed puts on its best James Dean (oops, Marlon Brando, thanks Andy) voice and replies: "What have you got?""

So asset purchases? I already *know* that asset purchases can have real effects. I discussed that above. What I'm saying is that the quantity of money is irrelevant. You're not *at all* reading what I'm writing! QE is exactly as useful as the treasury issuing t-bills and buying whatever assets the Fed's been buying. I'm not saying that doing that, or just threatening to do it, can't work. I'm just saying the quantity of reserves matters to *nobody*.

Break it down, like you say:

1) Fed exchanges reserves for T-bills (useless)
2) Fed exchanges T-bills for other assets (possibly useful depending on beta)

Quantity of excess reserves is *irrelevant*.

[Nick Rowe]

Ashwin: OK. There are three (and more) ways to vacuum up the excess supply of money that results from the success of our credible threat that gets NGDP back up and growing again:

1. Broad-based consumption tax vacuum cleaner.

2. The CB pays a high enough interest rate on reserves to persuade banks to hold large enough reserves to mop up the excess money.

3. The CB sells bonds.

Now 2 and 3 are roughly equivalent fiscally. In 2 the CB pays out more interest, and in 3 the CB earns less interest. So in both cases the CB earns less profit and hands over less profit to the fiscal authority. In fact, 2 is really equivalent to the CB issuing bonds and selling them to banks. (The only difference is that the bonds might have a slightly longer duration).

Now let's compare 1 to 3. Here's a thought-experiment. Suppose the central bank were doing 3. Would you want also to do 4 at the same time, where 4 is a broad-based consumption tax used to retire bonds?

Notice that doing 1 is exactly the same as doing 3 plus 4 together. 3 increases the stock of bonds in public hands, and 4 reduces that stock of bonds again, leaving only the increased broad-based consumption tax.

If you think that 4 is a good idea distributionally, because (say) you want to make the current generation pay to reduce the debt burden on future generations, then you could make the case for doing 1. But you could equally well have the fiscal authorities do 4, and the monetary authorities do 3, because the results would be the same as doing 1.

[Nick Rowe]

Ashwin: or, let me try it this way:

I can imagine a world in which all government deficits are financed by printing money, and all government surpluses are used to reduce the stock of money. There would be no government bonds (or the stock of bonds would not change). And (at least in principle, ignoring policy lags etc.) it would be perfectly possible to stabilise AD in that world. But you have just taken away one degree of freedom for government policy. Maybe you want to have temporarily high G, and/or temporarily low T, for various micro reasons, but for macro reasons you don't want to increase AD. When you have two policies, fiscal and monetary, that can move independently (at least temporarily) you can do that. When you only have one policy, fiscal/monetary, you can't do that.

[michellehersew]

That's a good post.

[Min]

Moi: "If you don't buy stuff, then I am going to buy stuff. Even though my buying stuff doesn't really do much. What kind of a threat is that?"

Nick Rowe: "Min: an indefinitely large and indefinitely permanent increase in the stock of base money is a bit different from you and me buying stuff."

So the threat is this?

"If you don't buy stuff, then I am going to create money and buy stuff with it."