Steve Landsburg goes "meta" on me

Steve Landsburg has what I think is the best response so far to my "Debt is too a burden on our children unless you believe in Ricardian Equivalence" post.

[Update: I strongly recommend Steve's latest post, which puts everything together.]

Steve says "I want to explain what [Nick] means, and why it’s wrong." But he doesn't really say I'm wrong. (I think that's just a rhetorical flourish!)

He does something more interesting. He takes a public choice perspective. He is asking who my audience is, and who I'm trying to convince. He goes "meta" on me.

Here is my version of Steve's argument. (The following are my words, not Steve's).

'What's wrong with you macroeconomists is that you think in terms of aggregates, not individuals.

Some individuals are Ricardian. They care about their grandkids, foresee the future tax burden on their grandkids created by the debt, and increase their bequests to their grandkids to fully offset that burden. So those individuals won't care whether current government spending is financed by taxing or borrowing. They don't care if the taxes fall on themselves or on their grandkids.

Other individuals don't care about their grandkids. They want to shift the burden of current government spending off themselves and onto their grandkids. So those individuals want current government spending to be financed by borrowing rather than taxing. They want the taxes to fall on their grandkids and not themselves.

So no individual will see any problem with borrowing to finance current spending. They either don't care, or else prefer it. Nick doesn't have an audience for his post.'

Again, those were words I have put into Steve's mouth. Read the original to decide if my version is fair and accurate.

I think it's a really neat argument. I didn't see it coming.

It's not just a "neat" or "clever" argument. It raises a rather fundamental question about any economic writing that is supposed to be policy-relevant. "Who's your audience?"

Here's my response.

Suppose I have a friend who cares about his grandkids, and who reads the New York Times, and believes everything he reads in the New York Times. And suppose the New York Times tells him that the debt is not a burden on his grandkids. What should I do? I should write an open letter to the New York Times saying that the debt is too a burden on our grandkids, unless we as individuals take offsetting action and increase our bequests to our grandkids in response. And hope that the people who write for the New York Times will be convinced by my argument and change their minds. Which is what I did.

(Again, just to repeat what I have said elsewhere, the fact that deficits may have costs on future generations does not mean we should not run deficits. Because deficits may have benefits too, both on us and on future generations. Nearly everything we do has costs as well as benefits, and we should consider both.)

In other news, Bob Murphy has done a 180 handbrake turn on the question of the debt burden. Well done Bob! To paraphrase (OK, totally distort) Keynes: 'When I hear a convincing counter-argument I change my mind; what do you do sir?'

I really wish all of us, myself included, could be always as open and honest (and funny) about changing our minds as Bob has been.

But what interests me most about this is how Bob describes how wrenching a paradigm shift it was. That's exactly how I felt when I made the same 180 turn in the 1980's. It ain't easy, when you have been so used to seeing the world one way, and convinced that is the right way to see it, to suddenly switch to seeing it the other way. It's jarring.

And, Bob is no Keynesian. Bob came to his original perspective from reading Ludwig von Mises, not Abba Lerner. This question of the burden of the debt cuts across the dimension of big government vs small government. Not everything is ideology.

Comments

[J.V. Dubois]

Nick: First thing is, that I am in the same position as Bob is now. I believed in "we owe it to ourselveses" in a way that was not correct. Your simpe overlapping-generations-apple-economy model cured me of it and there is no comming back now. This is just another example of influence of WCI on me and my thinking. And it is kind of a response for Krugmans's recent blog (which I really like) where he asks:

"First, there is no such thing in modern America as a pundit respected by both sides. Second, there are people writing about economic issues who are a lot less confrontational than I am; how often do you hear about them? This is not a game, and it is also not a dinner party; you have to be clear and forceful to get heard at all."

My answer would be, "read WCI". I have written this recently on and I must repeat it, that it is stunning to see that there are people like Nick that are not only willing to discuss their ideas with broad audience, but they are actually gettind something from it and returning later to sharpen their arguments or present a fresh way of seeing things. I am recently (since I discovered WCI) also having this "Kung-fu master" feeling that Bob has. It is very much like my experience with genius professor of Math in my University, who was able to solve complex mathematical problems in his mind and who was always willing to engage students in a discussion proving them wrong or right in matter of seconds and in the process being able to refine his own thought on issues he was working on independently.

Ok, enough of praise. I will muddle the things a little more as I too am finding this cognitive dissonance too hard to deal with and my mind is too egoistic to give up easily. Assume that we are dealing with your model except that there is growth between generations and there are capital gains. My idea is nothing new, it was actually part of the old Mankiw whitepaper. The point is, that if there are irresponsible parrents that want to use government spending to consume more, this should affect the return on capital for those responsible parents that think about future generations and invest. I realize that this is probably just restatment of Samuelson "interest on debt under growth is sustainable", but I wonder if market cannot correct irresponsibility so that this Samuelson position holds at all times.

[JKH]

“So no individual will see any problem with borrowing to finance current spending. They either don't care, or else prefer it. Nick doesn't have an audience for his post.'”

I see it the opposite

Your audience is everybody

With borrowing, everybody has the option of creating either effect for their kids

With taxation, they lose the option of passing it on as soon as the tax/transfer occurs

With borrowing, they retain that option until the last revision to their will

Options are better; options have value; and these are very long dated options

The rational person should care about that value

[Greg Ransom]

And even this "microeconomic" perspective assumes that the knowledge problem doesn't exist -- i.e. the most important of all microeconomic problems doesn't exit (esp. as the knowledge problem involves coordination of alternative production paths across time -- where government often leads to systematic knowledge coordination _errors_ across time by microeconomic evaluators/choosers.)

Steve/Nick writes,

"'What's wrong with you macroeconomists is that you think in terms of aggregates, not individuals.

Some individuals are Ricardian. They care about their grandkids, foresee the future tax burden on their grandkids created by the debt, and increase their bequests to their grandkids to fully offset that burden. So those individuals won't care whether current government spending is financed by taxing or borrowing. They don't care if the taxes fall on themselves or on their grandkids.

Other individuals don't care about their grandkids. They want to shift the burden of current government spending off themselves and onto their grandkids. So those individuals want current government spending to be financed by borrowing rather than taxing. They want the taxes to fall on their grandkids and not themselves.

So no individual will see any problem with borrowing to finance current spending. They either don't care, or else prefer it. Nick doesn't have an audience for his post.'"

[Nick Rowe]

J.V.: First, Thanks!

Normally, the more future-oriented people are (including how much they care for their kids), the lower the rate of interest will be. But the other thing that affects the rate of interest is how early in their lifespans people earn (or get) their income. In models like Samuelson's, people live for 2 periods, and if they get most of their endowment in the first period they want to save for their old age but can't. So the rate of interest is very low (it can easily be negative). And a PAYGO government pension plan and/or national debt sustainable Ponzi can solve a problem that the private sector can't.

JKH: Oooh! Neat too. And I didn't see that one coming either!

[JKH]

Landsburg:

"Unless! If we — the current generation — foresee all this, and care about our grandchildren, we’ll choose to (in effect) undo what the government has done by saving our tax cuts and giving them as gifts to our grandchildren (presumably as part of their inheritance). This restores every generation’s consumption to the original status quo."

That’s incorrect.

The tax cuts are saved automatically as a matter of both macro and micro financial accounting.

The macro is a forced result of the current deficit (a la MMT net financial assets).

The micro saving is a forced result of the macro saving. It’s merely a matter of the distribution of that saving. Neither macro saving or its distribution (somewhere) can disappear so long as the debt hasn’t been paid down with taxes.

So, the bonds are a permanent record of saving (until they are paid off), no matter who owns them.

The only question is whether the bonds themselves are gifted rather than sold to the next generation.

Somebody in the current generation has to end up with the bonds.

Those are the ones with the decision.

It’s got nothing to do with a saving decision beyond the point at which the deficit was actually incurred.

The next generation has to inherit them, not buy them, in order not to be down financially, if taxed to repay them.

That’s got nothing to do with a “saving decision” by the current generation.

It only matters who ends up with the bonds that are either sold or bequeathed. The current generation has already done its saving, before that decision occurs.

[JKH]

P.S.

Or, the current generation could sell the bonds to the next generation, and then bequeath them that money. But that's just a wash, equivalent to beqeathing the bonds in kind. Again, that's got nothing to do with "saving a tax cut". The money came from selling the bonds, not "saving a tax cut". That saving occured as soon as the bonds were issued, macro and micro.

[Ram]

Why can't we just say the following:

PuS (public savings) = T (tax revenue) - G (government expenditure)
PrS (private savings) = Y (national income) - T - C (consumption)
Y = C + I (investment) + G + NX (net exports)
S (national savings) = PuS + PrS = T - G + Y - T - C = Y - C - G = I + NX

In period one, a bigger budget deficit lowers PuS, which (unless this induces offsetting rises in PrS) lowers S, which lowers either I, NX, or both, which raises C, G, or both. In period two, a bigger budget surplus raises PuS, which (unless this induces offsetting declines in PrS) raises S, which raises either I, NX, or both, which lowers C, G, or both. Thus, if our generation incurs government debt, while the next generation pays it off, then we enjoy higher C/G, and they suffer lower C/G.

In short, more government debt today burdens future generations, so long as Ricardian equivalence does not obtain. Am I missing something, or is this just a less elegant way of telling your story?

[Nick Rowe]

Ram: because in my first post in this series, I=S=NX=G=0 by assumption in all periods. Y was constant. And I still got a burden on future generations.

Y=C+I+G+NX (or, I-S+G-T+NX=0) is a useful identity for some purposes. But it isn't very useful for intergenerational accounting. Because it provides a cross-generational snapshot at a point in time, and not a picture that allows us to compare across generations over time. Andy Harless is better at explaining this.

[Nick Rowe]

JKH: I think you are correct. I think I could restate what Steve said to make it correct (in terms of *desired* saving, or in terms of *some* individuals saving, etc.).

[Ram]

Ah, OK, guess I missed the importance of I=S=NX=G=0 the first time through.

[JKH]

Nick,

Once more, I think Landsburg has missed the most important point of your post. That is the question of whether or not future generations inherit or buy the bonds that were originally bought by the current generation.

If the next generation buys the bonds, they swap them for money, which the current generation can use to consume products that might otherwise have been consumed by the next generation. That means the current generation has consumed twice – once using the transfer, and once using the proceeds from the bond sale. That double consumption effect is what is necessary to transfer a burden to some future generation, provided that some future generation is taxed to pay down the bonds. Your model lays that out pretty neatly.

This has nothing to do either macro or micro with human behaviour in terms of people deliberately “saving the tax cut”. The tax cut was saved automatically at the macro and micro level as soon as the bonds were issued. From there, it’s actually a question of the velocity of money and multiplier effects and various financial transactions as to how the saving that’s connected with the tax cut gets transferred around the system and ultimately distributed at the micro level. But whatever the distribution, the entire amount is only saved once as an aggregate quantum at the macro level, and that occurs as soon as the bonds are issued.

Here is my example to illustrate this point:

Every last taxpayer that gets the tax cut can go to Vegas and blow the whole thing, and that has nothing to do whatsoever with the final resolution of whether or not there is a burden for the future generations. That money that gets blown on the Vegas trip ends up redistributing the initial net saving and related money associated with the tax cut to airlines, their employees, Vegas hotels, and blackjack dealers. All of those agents may end up saving the money and buying the bonds with it. And they may choose to either sell or bequeath the bonds to their descendents.

That choice of those blackjack dealers et al has absolutely nothing to do with “saving the tax cut”. The tax cut is saved automatically at the macro level as soon as the bonds are issued. From there, it’s a matter of whether those blackjack dealers et al in their retirement choose to bequeath the bonds to their children, or sell them to their children for proceeds which the retired dealers then blow on a trip to Hawaii. It’s that final Hawaii trip in effect that transfers the contingent burden to their children – contingent on a future government decision to tax them in order to pay down the bonds.

So “saving the tax cut” is absolutely irrelevant to this problem.

BTW, I attempted to follow that whole discussion on Ricardian equivalence (prodded by Krugman on Cochrane) that sort of paralleled your generational theme with Krugman, and it’s not clear to me that any of your profession really understands “what is meant” by Ricardian equivalence. What I mean by that is that it seems to me that some of most esteemed names amongst your colleagues have been involved in a debate wherein everybody has a different interpretation of what is meant by Ricardian equivalence. And if the best names in the business all disagree with each other, what possible absolute truth can it represent in terms of a reliable and consistent meaning?

[marcel]

@JV Dupois, who wrote:

[Krugman wrote:] "First, there is no such thing in modern America as a pundit respected by both sides. Second, there are people writing about economic issues who are a lot less confrontational than I am; how often do you hear about them? This is not a game, and it is also not a dinner party; you have to be clear and forceful to get heard at all."

My answer would be, "read WCI".

I think that Krugman, with typical United Statesian provincialism, used America to mean The United States. It is the opposite of synechdoche, the device of indicating something by referring to one of its parts (sails for ships, etc.) In this case, a part is indicated by referring to a large collective or aggregate to which it belongs.

So, I refute your answer!

[marcel]

(By "large" in the 2nd paragraph of my comment above, I of course meant "larger")

[vimothy]

Isn't it more that it's a bit harder to say what RET means for G than what RET means?

IIRC, Nick has a post on that round here somewhere...

[Rob]

If one assumes a static population of potential taxpayers and a form of savings that grows at the same rate as the interest on the bonds the government sold then would not an individual who cared about his grandkids be better off paying the tax if he he was expecting to have more than one grandchild (and he wasn't relying on the other 3 grandparents to make a similar legacy)?

In fact the only person who actually be better off not paying the tax would be someone with no descendants.

[Nick Rowe]

JKH: "Nick, Once more, I think Landsburg has missed the most important point of your post. That is the question of whether or not future generations inherit or buy the bonds that were originally bought by the current generation."

I don't think he missed that bit.

Take the simple example of a closed economy with no investment. Then when the government makes a transfer payment, the accounting determines that they *must* save it all (actual private saving is identically equal to actual government dissaving).

But if they are non-Ricardian they will *try* to dissave it, and the result will be a higher interest rate and/or higher output (depending on if the model is classical or keynesian) until they stop trying to dissave it.

And if they are Ricardian they will *desire* to save it all at the current rate of interest and level of output, so the transfer will have zero effect on interest rates or output.

JKH and vimothy: Suppose there's an increase in G financed by borrowing, what happens? I like to split this into 2 questions:

1. an increase in G financed by an increase in T;

2. plus, a cut in T financed by borrowing.

Ricardian equivalence tells us the second has zero effect. So we now concentrate on the first question. And the answer to that depends on what that G is spent on. For example, if G is a perfect substitute for C, it has zero effect. The government taxes me $1 and spends it on apples and gives me the apples. I just buy $1 less apples myself.

A lot of the confusion comes from people having different implicit assumptions about whether G is a substitute for C. John Cochrane implicitly assumes perfect substitutibility. Paul Krugman implicitly assumes zero substitutibility. They each think the other is an idiot.

I stayed out of it this time around. I couldn't be bothered. Last time I intervened was when the Chicago guy from the World Bank explicitly said that it all depended on how productive an investment G was. And he was right. And PK and a lot of people missed his point. Because if G is highly productive, an increase in G financed by borrowing would *reduce* future tax liabilities. So you get a reverse Ricardian effect, and a very big multiplier.

[Nick Rowe]

Rob: I think you are onto something there.

[Steven Landsburg]

Steve says "I want to explain what [Nick] means, and why it’s wrong." But he doesn't really say I'm wrong. (I think that's just a rhetorical flourish!)

Agreed.

Again, those were words I have put into Steve's mouth. Read the original to decide if my version is fair and accurate.

Your version is fair and accurate.

But as far as your audience goes: You're assuming an audience that doesn't think clearly about how big a burden govt policies impose on their grandchildren. This is an audience that has been bombarded with warnings from bloggers, journalists and politicians about how big that burden is. If the audience is misestimating the size of the burden, I do not see any reason to assume they've underestimated as opposed to overestimated. So when you address this audience and say "Increase your estimate!", you are, as far as I can tell, equally likely to be pushing them in the right or the wrong direction.

[Steven Landsburg]

JKH: That money that gets blown on the Vegas trip ends up redistributing the initial net saving and related money associated with the tax cut to airlines, their employees, Vegas hotels, and blackjack dealers.

You seem to be assuming that hotel and airline services are provided at zero cost (and therefore at an enormous mark-up). In a world where everything is costless, there is surely no such thing as a debt burden.

[JKH]

Nick 8:09

Thanks. That's helpful on Ricardian.

It seems to me then that much of the discussion is conflating multiplier dynamics with Ricardian dynamics.

Shouldn't a "standardized" explanation of RE incorporate multiplier scenarios as well where G is inevitably involved, to make it generally clearer? I.e. as you have I guess? i.e. shouldn't any RE discussion be fully consistent with clearly stated multiplier assumptions where G is even mentioned?

Conversely, unless potential or assumed multiplier scenarios are covered in the discussion, it shouldn't even reference G expenditure in the context of RE?

[JKH]

Steven L.,

"You seem to be assuming that hotel and airline services are provided at zero cost (and therefore at an enormous mark-up). In a world where everything is costless, there is surely no such thing as a debt burden."

I think I was implicitly assuming, to make things very simple, that all that Vegas related expenditure added to GDP, generating additional equivalent income (ultimately) to the providers of inputs – income that ended up being saved. So I think that assumes a multiplier of 1 on the tax cut (not sure), and shifts the location of the associated saving from the transfer recipient to the ultimate income earners behind the provision of the Vegas related services. That includes the costs of the related hotel and airline services, including salaries, other upstream inputs and the income behind those, cost of capital, etc.

[JKH]

meant transfer, not tax cut above

[JKH]

Nick,

"But if they are non-Ricardian they will *try* to dissave it, and the result will be a higher interest rate and/or higher output (depending on if the model is classical or keynesian) until they stop trying to dissave it."

I think that corresponds to my Vegas example, where the transfer recipient hits the town. But somebody ends up saving as per my reply to Steven L. above; its forced by the original transfer. Saving as an already realized flow and now stock becomes the hot potato. And there is still the issue of whether the current generation sells or bequeaths the bonds to the next, as per the blackjack dealer's retirement decision.

[Greg Ransom]

Steve, the key thing is people have no firm or consistent idea of what significance this all may have for their own affairs, or when this significance might come into play in a way that makes a difference for their own plans.

And the inability to plan is _one of the burdens_ all of this debt accumulation has on prospect for the successful coordination of production and consumption plans across time of particular persons and entities.

Steve writes,

"But as far as your audience goes: You're assuming an audience that doesn't think clearly about how big a burden govt policies impose on their grandchildren. This is an audience that has been bombarded with warnings from bloggers, journalists and politicians about how big that burden is. If the audience is misestimating the size of the burden, I do not see any reason to assume they've underestimated as opposed to overestimated. So when you address this audience and say "Increase your estimate!", you are, as far as I can tell, equally likely to be pushing them in the right or the wrong direction."

[Ralph Musgrave]

Nick’s entire thesis is based on the assumption that oldies get youngsters to buy bonds off them. Nick and supporters seem to have no interest in the real world – in the empirical evidence. That is, do oldies actually sell bonds to youngsters?

Well they do in that the above process is a normal part of the pension system. That is, oldies sell bonds (and other assets) to fund their retirement. But the $64k question is: “do oldies BOOST their consumption in retirement by the above means given an INCREASE in government debt?”. And my answer to that is “I doubt it”. Reason is that even if there were no government debt at all, those planning their pensions could choose any standard of living they want in retirement (within the constraints posed by their income when young) by using other assets, or using pay as you go or “unfunded” pension schemes.

Ergo I don’t see a strong reason for thinking that the creation of government debt, or an expansion therein, places a burden on future generations (given Nick’s other quite reasonable initial assumptions – closed economy, constant full employment, etc)

[JKH]

Ralph,

“But the $64k question is: “do oldies BOOST their consumption in retirement by the above means given an INCREASE in government debt?”.”

Consider again the dynamic of Nick’s model.

The first cohort gets to consume twice – once when they get the original transfer funded by the bonds; once when they sell the bonds to the next generation. They’ve double dipped. It’s their double-dipping into consumption, with the help of the sale of outstanding bonds, that causes a chain reaction whereby that creates a burden for the ultimate holders of the bonds who are taxed to pay them down.

So it’s their choice to sell the bonds rather than bequeath them to the next generation that results in the double-dipping of consumption. And that double-dipping of consumption is inherent to the potential for a future burden in the event of a future tax to pay down the bonds. But it’s the fact that they got the original transfer that put them in the position to make a choice with that particular consequence.

If the bonds weren’t issued, and the same cohort faced a similar decision to consume or not out of other savings, it obviously doesn’t have the same implication for Nick’s model because Nick’s model isn’t even operative, given there was no original bond issue. No bond issue; no tax problem; no burden question.

“Reason is that even if there were no government debt at all, those planning their pensions could choose any standard of living they want in retirement (within the constraints posed by their income when young) by using other assets, or using pay as you go or “unfunded” pension schemes.”

That’s all true except for the fact that there’s no contingent tax liability in respect of paying down the bonds – so the specific tax threat that is described in the problem doesn’t exist, and it’s that tax that creates the final net burden. The fact that the problem is moot when the pre-existing conditions that are essential to the problem haven’t been created doesn’t mean there’s no difference in the final outcome.

Aside from that, people should decide whether they’re interested in playing the game of assessing the logic of Nick’s model or trashing it because its specifications aren’t a perfect match for the complexity of the world and don’t provide globally comprehensive 401K balance sheet transitions, etc. But it’s really not cricket to do both, unless the difference between those two things is specified.

[Ken]

As the relative layperson in the group, this all seems a bit silly.

1. In all periods, the aggregate amount of apples produced and consumed is unaffected by the debt. This is what Krugman is pointing out and he is correct. (I just read some of Murphy's posts and it surprises me that he ignores this simple fact in his own examples and scenarios.)

2. The debt can change the distribution of apples consumed.

3. Depending on expectations, some people might end up consuming fewer apples over time then they had expected. One good example is if they own government debt that they think can be honored on a real basis without increasing taxes. This is where Rowe/Murphy are correct.

What am I missing?

[Ken]

To be clear, I do understand that the debts created by one generation could affect the distribution in future generations. Whether that makes Krugman "wrong", though, I think becomes a matter of interpretation of his intended point. (To me, Krugman has a consistently cavalier attitude towards government redistribution of wealth.)

[JKH]

"In all periods, the aggregate amount of apples produced and consumed is unaffected by the debt... What am I missing?"

It's affected in the first period if the transfer payment is used to consume apples and in addition proceeds from selling the bonds are used to consume apples. That's double dipping on consumption.

And its affected in the final period when the tax causes a decline in apple consumption in that period. That's the zero consumption that is the offset to the front end double dipping.

It's not the outstanding debt that's the issue - its the use of the transfer payment funded by debt creation to consume apples, and the non-use of the tax payment by the taxpayer to consume apples.

[Nick Rowe]

Steve: Thanks!

"But as far as your audience goes: You're assuming an audience that doesn't think clearly about how big a burden govt policies impose on their grandchildren. This is an audience that has been bombarded with warnings from bloggers, journalists and politicians about how big that burden is. If the audience is misestimating the size of the burden, I do not see any reason to assume they've underestimated as opposed to overestimated. So when you address this audience and say "Increase your estimate!", you are, as far as I can tell, equally likely to be pushing them in the right or the wrong direction."

Maybe. But it's also an audience that has been told (if my interpretation is correct, and the audience may have adopted the same interpretation), that the burden of the debt is zero (except for tax disincentives, debt to foreigners, etc.).

[Ken]

Thanks JKH.

I think I'm correct that in all time periods the aggregate consumption of apples is unchanged regardless of debt policy. I haven't seen an example given by Nick or Bob Murphy that shows a change in aggregate apple consumption in a given period. In fact, it is impossible based on the assumptions of their simple models.

I think I was missing, though, Nick's point that due to overlapping generations, the older generation could take consumption from overlapping younger generations in exchange for bonds that cannot be repaid on a real basis without increased taxation (assuming no GDP growth). In other words, the distribution effects could cross generations due to overlapping generations. Again, though, the aggregate consumption in a given time period is unaffected. If someone thinks that last point is wrong, I'd love to be corrected.

At least that's my latest understanding.

[JKH]

Ken,

Hmm - maybe - I don’t know to be honest.

But let the time period of the overlap approach zero.

And consider the dividing line between generation B and C, where C gets taxed.

Two choices:

- B sells bonds to C, and B consumes with the money received from C for the bonds, while C forgoes consumption; C uses the bonds to pay the taxes, i.e. the bond maturity proceeds pay the taxes

- B wills bonds to C, and C consumes with the money he didn’t pay B for the bonds; C uses the bonds to pay the taxes same as the first case

In the first case, B consumes within epsilon preceding B's death

In the second case, C consumes within epsilon following B's death

Technically - feasible depending on the non-zero period for consumption choice, overlapping two different periods

Realistically crazy, I know, but I think that was my intuition for the model

Either way, I’m not sure the interpretation of that aspect is critical for the model

[wh10]

JKH if you have time, you're needed on aisle 3 http://worthwhile.typepad.com/worthwhile_canadian_initi/2012/01/the-30-years-debt-burden-non-war.html?cid=6a00d83451688169e20168e5132e84970c#comment-6a00d83451688169e20168e5132e84970c

[Adam]

Sorry to keep repeating myself, but do you honestly think anyone see a deficit, concludes there will be higher future taxes, and increases the bequest to their kids?

I don't see anyone (or at least no significant group of people) reaching that conclusion and I see even fewer taking that action.

Do economists really live in world where this type of action actually happens?

[Steve Roth]

Nick, since I'm still having trouble deciding what I believe here, I'd like to see this problem thought through in MMT world, where the government never borrows. It just prints dollars when it wants to buy something. Let's assume either no taxes, or taxes exactly sufficient to keep inflation steady. Might be clarifying?

[Nick Rowe]

Steve: If the deficit is money-financed, and the money pays no interest (it's like currency) then it's different. The big difference concerns the "No Ponzi Condition". (Read the #4 option in my previous post). This is one of the important reasons why people like me want to use monetary policy, rather than fiscal policy, to increase Aggregate Demand in a recession.

But suppose, just for the sake of argument, that we start at full employment. Then the government decides to cut taxes, or give everyone an annual transfer payment, and it finances the resulting deficit by printing zero interest money. That is the "helicopter money" thought-experiment (except the helicopter keeps on flying every year. If the helicopter doubles the money supply every year, then the price level would double every year, and the value of money people held would be halving every year, exactly offset by the new money they got given by the government. So the transfer is paid for by an inflation tax.

[Nick Rowe]

Adam: "Do economists really live in world where this type of action actually happens?"

let me give you a small-scale example. We had our street paved a couple of years back. The local government gave us a choice:

1. Each of us could pay the cost in one lump sum. (Balanced budget, with an increase in taxes to pay for the increased government spending).

2. Each of us could have the cost amortised on our taxes over 10 years at 5% interest (IIRC). (Deficit-financed increase in government spending, because the local government would have to borrow the money to pay for paving the road).

That choice made almost no difference to me. I almost flipped a coin before deciding what to do. I was behaving exactly like Ricardian equivalence says I would.

If I had less savings, and couldn't borrow (except at much higher than 5% interest), and if I were living hand-to-mouth, then it would have made a big difference to me. I would have strongly preferred option 2, because option 1 would have meant I would have to cut back my current consumption to pay the lump sum tax bill.

Does that example scale up? Well, that's the big question. Econometricians have tried to test it. Not very conclusively. One neat paper I saw ages ago, by Mankiw, estimated that about 50% of the population seemed to act like 1, and the other 50% like 2. That's my guess. Though I'm sure it depends on a lot of things.

[vimothy]

JKH and vimothy: Suppose there's an increase in G financed by borrowing, what happens? I like to split this into 2 questions...

Nick, thanks for that.

Like the Kocherlakota thing, seems like much fuss over nothing and mostly people talking past each other.

[Rob]

Nick, Is that really Ricardian Equivalence or is that just choosing the option which works out the cheapest based on your expectations of a number of future variables ? If they had chosen a third option of paying for it by selling 100-year bonds would you really have bought the bonds or taken out other forms of savings to cover your descendants tax liabilities ?

[Nick Rowe]

Rob: If I had been totally indifferent between paying for the road in 1, 10, 100, whatever years, then it would be exactly Ricardian Equivalence. And it made very little difference to me.

"If they had chosen a third option of paying for it by selling 100-year bonds would you really have bought the bonds or taken out other forms of savings to cover your descendants tax liabilities ?"

Basically, roughly, yes. I'm thinking past my lifetime.

BTW, nobody believes Ricardian Equivalence holds exactly.

[Rob]

Having now read the Wikipedia article on "Ricardian Equivalence" twice I consider myself something of an expert on the subject. It seems that Ricardo didn't really believe in it himself, and it is mainly used as a tool for anti-Keynesians to claim that deficit spending by the government is doomed to fail since people will just save the amount of the additional spending.

I'm pretty much a paid-up member of the Anti-Keynsian League , but can see this theory has some serious issues. Clearly there will be a tendency for people to save more in the expectations of future higher taxes as a result of government borrowing. If I know taxes will double next year I will probably save more now. But because of time preference probably not to the full extent of the expected tax increase. If the tax increase was expected to be 1000 years from now time preference would almost certainly eliminate any effect on my current spending no matter how much I care about my descendants.

Ricardian Equivalence claims that 100% of government borrowing will be factored into current spending. Common sense says that time preference, ignorance, and selfishness will all mean that it will be less than 100%. In addition things like chances of default, inflation, Bob Murphy becoming president and abolishing all taxes , all make us less likely to save the full amount. My guess is that for most expected tax increases greater than 10 years away the effect of Ricardian Equivalence would be 0%.

I have not seen one good defense of this theory. Can you point me to one ?

[Nick Rowe]

Rob: time preference doesn't invalidate REP. The higher is the rate of time preference, the higher is the equilibrium interest rate, and the bigger the increase in future taxes. It's a wash. Ignorance is debatable. Some people might underestimate, others might overestimate. (You might argue that ignorance leads people to underestimate *changes*.) Not leaving bequests will invalidate REP. So will immigration, etc.

Again, nobody believes it 100%. Nobody believes any theory 100%. But then nobody believes the exact opposite either - that people ignore the future. The truth very probably lies somewhere between those two extremes. The most sensible thing to say is that there is *some* truth in REP.

The original paper is by Robert Barro "Are bonds net wealth".

[vimothy]

Rob, I've got one: the Ricardian Equivalence theorem tells you exactly what conditions need to be satisfied for Ricardian Equivalence to hold.

[Nick Rowe]

What vimothy is saying (I think) is that Ricardian Equivalence is like the Modigliani-Miller Theorem of corporate finance (actually they are almost brother and sister). The MMT says that the corporate debt-equity ratio is irrelevant. That is a very surprising result. We "know" that it *must* be false. But it's also an incredibly useful theorem. Because by telling us under what conditions the MMT is true, it tells us where to look if we want to explain what determines the debt/equity ratio. RET tells us where to look if we want to explain why tax vs bond finance matters. That makes it a very useful theory.

[Nick Rowe]

Yep. Arguing *whether* RE is true is pointless. Arguing *why* RE might be false is useful.

[vimothy]

Arguing *whether* RE is true is pointless. Arguing *why* RE might be false is useful.

Nick, that's well put.

[Rob]

Ok. I can see that Time Preference is not relevant to the discussion. Thanks for clarifying that.

If all we're talking about is a tendency for govt borrowing to lead to an increase in savings levels at any given interest rate then that is a useful theory.

However , anyone who claims that RET proves that govt borrowing and govt taxing are equivalent in their effects is making a claim that is not supported by this more limited interpretation of the theory.

Steve L. is not claiming that: He's just claiming that those who want to save to offset taxes will do so. My original point about this being irrational under certain circumstances (where paying the present tax will be cheaper than paying the future tax for many descendants) still seems valid. Doesn't that itself further weaken RET ?

[Greg Ransom]

"Arguing *whether* RE is true is pointless. Arguing *why* RE might be false is useful."

This is common in the human sciences. Figuring out why Plato and Russell, for example, were wrong about concepts / language / math turned out the be very hard -- and ultimately very fruitful. Ditto figuring out why Mill & Carnap & Popper & Nagel & Hempel were wrong about the nature of science and scientific explanation. Etc.

In economics note well:

Those who claimed that we could do without _interest_, _money_ and _private property_ led to some of the most important advances in all of economics, e.g. to the work of Wieser and other on interest, and the work of Mises and others on the the role of money prices and property rights in social coordination.

Similarly those who insisted that "just prices" could be determined led to the beginnings of modern economic science, e.g. the work of the Spanish Schoolmen.

But note well, in these cases arguing *whether* _xyz_ is true or possible led directly to singly fruitful understandings of *why* _xzy_ must be false or impossible.

Similar things have happened in macroeconomics ...

[Nick Rowe]

Rob: "Doesn't that itself further weaken RET ?"

At the individual level I think it does. At the aggregate level it might all (roughly) cancel out. Those with more kids will save more, and those with fewer kids will save less.

Greg: nice comment.

[Rob]

Plus they will pressure for the tax option and object to the public debt , not be indifferent as he claims "In this story, nobody who is alive today actually objects to the public debt. So Nick cannot use this story as a reason for you, me or anyone to become a deficit hawk."

[JKH]

Nick,

Why isn’t Ricardian equivalence a tautology at the macro level?

Suppose the government cuts taxes and pays for that by issuing bonds.

At the point of the tax cut and bond issuance, private sector (actually non-government) income and saving increases. Before any associated spending begins, the balance sheet of the government has increased by a net liability and the balance sheet of non government has increased by a net asset (the bonds). Before any associated spending begins, the balance sheet of non government has increased by the bonds and by an increase to its net wealth (equity) as that tax cut constitutes an increase to its disposable income.

At the macro level, that non government net wealth or equity or saving can never disappear, unless and until the government taxes it back.

The proof of that is that the worst the rest of the non government can do in theory is that the rest of its balance sheet goes to zero net worth or zero equity. It can’t go negative. The balance sheet equity of individual agents can go negative, but the full macro balance sheet can’t. The reason for that is that the lower bound for real asset value is zero, and all financial assets and liabilities otherwise net out to zero.

So if the rest of the non government balance sheet equity can’t go below zero, and if the bonds that created original additional net asset value for non government remain outstanding, then the saving or net worth or equity value associated with those bonds and that tax cut can’t disappear – not until the bonds are paid back with taxes.

And that means that the saving to pay back the tax cut always exists at the macro level, so long as the bonds are outstanding. Under any circumstances, at the macro level, the bonds that were issued represent the saving that can be used to pay back the tax.

And that’s because the cash that the government gives the holders of the maturing bonds comes from the repayment of the original tax cut, and at the same time the tax repayment comes from the cash proceeds of the bond maturity.

Both these things happen simultaneously at the macro level. The transaction is basically an asset liability swap at the macro level – non government swaps a maturing bond in exchange for extinguishing its own tax liability, and government swaps its own bond liability in exchange for the repayment of a tax cut.

Ricardian Equivalence is a tautology at the macro level.

QED?

Therefore, Ricardian Equivalence is entirely a micro distribution issue?

[Max]

"Nick, since I'm still having trouble deciding what I believe here, I'd like to see this problem thought through in MMT world, where the government never borrows. It just prints dollars when it wants to buy something. Let's assume either no taxes, or taxes exactly sufficient to keep inflation steady. Might be clarifying?"

I'm not Nick and I could be off base, but here's what I think. Given:
- fixed interest (below GDP growth)
- fixed government spending
- fixed inflation
- variable taxes

...the deficit and taxes would follow a "random" path determined by private demand. There would be no discretionary control over the deficit.

If it is desired to smooth taxes then the MMT government could invest in recessions rather than cut taxes.

[Nick Rowe]

JKH: "Nick, Why isn’t Ricardian equivalence a tautology at the macro level?"

I'm not sure I follow your argument. But take my original story.

Suppose Ricardian Equivalence were true. We would then observe the old A's giving (not selling) their bonds to the young B's (their kids) and not eating any of the B's apples.

Suppose Ricardian Equivalence were false. We would then observe the old A's selling their bonds to the young B's, and eating 110 of the B's apples.

RE cannot be a tautology, if its truth or falsity give different observational outcomes.

[JKH]

Nick,

That's the micro argument, and I fully understand it, which I think you've acknowledged. It has to do with inter-generational distribution of the bonds at the individual level - and the fact that you're comparing two different scenarios for individual A's reflects this. Your model illustrates this micro behavior factor perfectly.

My macro argument is entirely different. What it says is that the saving that is required to pay the taxes always exists at the macro level, and that is a tautological fact. The proof is that the mere action of paying taxes is a reduction in cumulative macro non government saving - the same saving that was created by the original tax cut and reflected in the value of the bonds issued. So any Ricardian issue of whether or not somebody saved to pay taxes is always micro - as per your example of different behavior of the A's.

Put another way, whoever is stuck with paying the taxes is doing it out of the same cumulative saving at the macro level.

[Nick Rowe]

JKH: Ah. I get you now.

At the macro level (and this would be clear if I specified my model more fully) we would also observe a higher equilibrium interest rate if RE were false and a lower rate if it were true. Because under RE people would *want* to save the transfer, and if RE were false the rate of interest would need to rise to persuade them to want to save the transfer.

[vimothy]

BTW, I think JKH's macro-level accounting explains very well the motivation behind RET and the question, "Are government bonds net wealth?"

[Rob]

1. Govt bonds needs to be sold: Interest rates will need to rise to clear the market.
2. Assume no RET - the interest rate would be set to the market clearing rate to allow the bond to sell
3. Add in RET - Surely this would mean additional demand for savings on top of this and move the interest rate down.

The greater the tendency of the future tax payers to offset by more present savings the cheaper the government can borrow the money at.

And if interest rates are zero and no-one is borrowing then the higher the RET effect the less effective will be govt stimulus funded by borrowing (which will lead to offsetting savings)

[Steve Roth]

Nick: "If the helicopter doubles the money supply every year, then the price level would double every year"

Really??

To paraphrase Nick: I thought this had been resolved decades ago.

Nobody believes this quantity-price proportionality theory, do they? There are so many things wrong with it, both theoretically and empirically. Long-term there's correlation, but the direction of causation is seriously questionable. Causation may hold shorter-term, but proportionality to quantity decidedly doesn't. The empirics (and Friedman's theory) say it's about flows of new money -- *changes* in stock levels, and the rate of change in those flow changes, not the stock levels themselves. And even there the effects aren't proportional.

And no, it doesn't seem that using it as a simplified explanatory/exploratory conceptual tool is useful. It's sort of like saying:

Strong winds blowing in from the Pacific and pushing against the mountains of western Washington cause air pressure to increase, pushing clouds out of the area.

It's not just simplified, it's wrong. So using it to think about the weather takes you down false paths that lead nowhere useful.

??

[Steve Roth]

JKH: "the saving that is required to pay the taxes always exists at the macro level, and that is a tautological fact"

Absolutely right, and this is exactly where I get hung up -- or at least diverge. Because in aggregate, if you ignore financial assets as the NIPAs do, there are no "savings." Every dollar spent -- whether on income or investment -- is a dollar of income.

The only reason the word exists in the NIPAs is because "saving" is *defined* there as Income - Consumption Spending (hence, tautologically, Savings = Investment Spending, which is a very odd thing; spending is saving??), instead of Income - Spending (which is an accurate definition of "Savings" on the individual level). The NIPAs are set up this way to depict a barter economy using transparent money and no wealth, debt, credit, or financial assets. It allows us to calculate real "production," but ignores a whole lot of our financialized economy.

Income - Spending ... a.k.a. Savings ... equals zero. A tautology.

This is exactly what the MMTers point out: privately held net financial assets are unaffected by how much individual people and businesses save or spend in a given year. Absent the government and external sectors, the change is, tautologically, zero.

Just to say that I think a huge amount of economic discussion gets derailed by this failure to understand the difference between individual savings and aggregate (nonexistent) "saving."

IOW: individual saving is, on the aggregate level, merely *not spending.* It only reduces the flow of money; it doesn't, can't, affect the stock -- because every dollar leaving the stock instantly appears in the stock again, in somebody else's account. Spending = Income.

Better to discuss the flows between sectors (with financial assets included) like the MMTers do. (Though it's unfortunate that they refer to these flows between sectors over a given period as sectoral "balances." Implies a stock when they're actually talking about flows. Unnecessarily confusing, and semantically inconsistent with their stock-flow consistent methodology.)

[JKH]

Steve Roth,

Right, if I follow your general train of thought there. There are a lot of algebraic relationships between slicing and dicing of accounts as reflected in national income accounting versus MMT’s translation of national income accounting. “NFA” as a stock as defined by MMT is pretty much the same thing as the government debt, or the cumulative government deficit as a cumulative flow. And that cumulative flow constitutes marginal saving by non government, as a flow. So the fact that the debt even exists is proof that the cumulative marginal non government saving it represents always exists as a stock of non government savings, somewhere in the economy. It doesn’t disappear, and it doesn’t have to be “resaved” at the macro level. And it’s in that sense that I say Ricardian equivalence is a tautology at the macro level – because the cumulative saving that is required to pay the taxes that pay down government debt already exists, and is reflected in the outstanding stock of government debt itself. If you look away from the various micro complexities of the detailed distribution of who owns the debt and who pays the taxes, etc., you’re left with the macro simplicity of that tautology. At least that’s my view of it.

[vimothy]

Romer's nice explanation of RE is germane:

“The result of the irrelevance of the government’s financing decisions is the famous Ricardian equivalence between debt and taxes. The logic of the result is simple. To see it clearly, think of the government giving some amount D of bonds to each household at some date t1 and planning to retire the debt at a later date t2; this requires that each household be taxed amount D*exp[R(t2)-R(t1)] at t2. Such a policy has two effects on the representative household. First, the household has acquired an asset—the bond—that has present value as of t1 of D. Second, it has acquired a liability—the future tax obligation—that also has present value as of t1 of D. Thus the bond does not represent “net wealth” to the household, and it therefore does not affect the household’s consumption behaviour. In effect, the household simply saves the bond and the interest the bond is accumulating until t2, at which point it uses the bond and interest to pay the taxes the government is levying to retire the bond.”

Advanced Macroeconomics, David Romer

[JKH]

Vimothy,

Romer makes sense to me as a micro scenario where an individual household has acquired the bond. A variation on this is the case that has been discussed lately, which is where the individual household has received a transfer (or a tax cut). So this scenario is combined with the household’s decision not to spend the transfer/tax cut, saving it instead, buying the bond with it, and waiting to be taxed, at which time it effectively exchanges the maturing bond for the tax liability.

That’s fine, but what it misses is the point (I’ve made) about an RE tautology at the macro level. The non government saving that was created by the original deficit spending (transfer or tax cut) exists in stock form so long as the debt exists. So the economy as a whole never has to save anymore than it already has in order to extinguish the final tax liability. The difference between macro and micro is only the distribution of who has the bonds and/or the saving at the micro level, the stuff that adds up to what was effectively predetermined at the macro level when the original deficit was created. In the base case here, the individual household makes a conscious decision to ring fence its share of an expected tax liability by immediately saving it in the form of the bonds.

Romer’s interpretation of net wealth or lack thereof has an analogue at the macro level, particularly in MMT terms. That would be the statement that the RE adjusted NFA position is actually zero – because non government NFA would be offset by non government GFL (gross financial liability), where the GFL is the tax liability.

In that context, one way of approaching the RE debate from the MMT perspective is to say that this is all rubbish – people don’t save to pay an expected future tax liability. But the reason MMT argues this is not so much because it would be necessarily incorrect to acknowledge the offset I described under the assumptions stated, but because the expectation of a future tax liability AT ALL is rubbish. And this view relates closely to the alleged inherent desire for NFA on an ongoing basis, the empirical case for the interest rate being less than the growth rate in the long run, and the fact that the underlying issue is not insolvency but inflation risk.

So there are two problems that I see with it – the first, that RE is a tautology at the macro level, requiring no additional intentional saving at the macro level. The macro numbers have to balance via double entry bookkeeping, whatever the dust up may look like at the micro level. The second is that RE is irrelevant to the degree that taxes won’t be required anyway, which means a fortiori that micro saving intentions are additionally irrelevant (even though macro will always balance anyway).

[marcel]

Ken wrote:

I think I'm correct that in all time periods the aggregate consumption of apples is unchanged regardless of debt policy. I haven't seen an example given by Nick or Bob Murphy that shows a change in aggregate apple consumption in a given period. In fact, it is impossible based on the assumptions of their simple models. ... Again, though, the aggregate consumption in a given time period is unaffected. If someone thinks that last point is wrong, I'd love to be corrected.

I don't believe this is right, at least, it's not what (I thought) Krugman was thinking about. I cannot find it now, so perhaps I imagined it, but my understanding is that Krugman was thinking of a situation with unemployed resources, apples not produced (and therefore not consumed) because farmers did not expect demand to be sufficiently high (to sell them at a profit). Farmers could produce and sell more, profitably, if that did not lead to a reduction in prices.[1] If the debt is distributed to individuals who will use it to purchase (and presumably consume) more apples at current prices -- say individuals who are currently unemployed or otherwise liquidity constrained -- farmers will produce and sell more, etc., etc. Consumption rises in the current period. BdL has a (mildly windy) example here (see the discussion of Alice, Beverly, et al).

If there are not unemployed resources, then aggregate consumption in any period is left unchanged. If there are, due to insufficient demand, then debt, either government as in this discussion or private as in BdL's example, is a way reduce their level and increase consumption now. Between greater depreciation due to higher activity and greater investment due to expectations of higher profits, consumption in future periods is also likely to be different as a result of today's debt.

[1]This leads to a side discussion of the market structure that the farmers find themselves in, since in the ideal version of the competitive market, each individual farmer believes that (s)he has no affect on prices, so can increase output effectively without limit taking the current price as given... No need to address this here.

[vimothy]

JKH,

Very interesting comment, thanks.

I think your (initial) observation that RET rests on a tautology and your observations re RET vs MMT are especially astute. For Barro, the household cannot spend the extra wealth, because it must save to pay its extra taxes. For MMT, the household cannot spend the extra wealth either, but in the opposite sense: since the bond will always be there on the macro level, the household does not need to try to save anything; that saving is already assured. That's a nice insight.

And I agree that, assuming the MMT case with regards to the ability of the government to finance its operations without taxes, this result becomes kind of academic.

Remember that Romer is referring to the representative household above. Here, that household operates according to the same perfect logic as your macro-level accounting. The representative household is a modelling convention that represents or averages those effects you identify at the micro level.

What's more, the model itself is just a set of rules, essentially (with the representative household as a dynamic optimizer). Because of the macro-level tautology you identify, it must be the case that the representative household responds in that way at the micro level (given relevant particular conditions). If the bonds are irrelevant on aggregate, they must be irrelevant to the representative household too.

MMT operates according to a different set of assumptions, of course, and so it interprets this is result in quite a different way. But it's interesting to see how and where the two approaches overlap--perhaps a theme worth developing.

[vimothy]

Perhaps my comment above isn't very clear.

The rep household is a consequence of modelling the entire sector as if it was an individual. In other words, what if all households felt the aggregate effect equally--in which case, there would only be one household, the sectoral average or "representative" agent. So whatever holds for the whole sector must also hold for the representative household.

Since bond and tax obligation exactly cancel out on aggregate, whatever maximised lifetime representative household consumption prior to the govt borrowing maximises lifetime consumption after it. Government bonds are not net wealth.

[JKH]

Vimothy,

Very clear, thanks, and I agree.

How’s this, to play it back again, in a slightly modified way:

IF you assume that the economy, macro and micro, expects a tax liability in order to pay down the bonds, and IF you assume that there is a micro representative agent distribution of bonds proportionate to the micro representative agent distribution of the tax liability, THEN there is an economic offset to the so-called MMT net financial asset agent at BOTH the macro and micro level in the form of the tax liability AND the bonds are associated with zero net wealth at both levels AND Ricardian equivalence holds at both levels.

However, REGARDLESS of the micro distribution of either bonds or tax liabilities, IF the economy at the macro and micro level expects a tax liability in order to pay down the bonds, then saving to pay the taxes MUST exist in the form of the bonds (as net financial assets), REGARDLESS of any micro attempt to save to pay the taxes exists or not, AND Ricardian equivalence must hold at least at the macro level even though agents may not attempt to save for that purpose at the micro level. Moreover, similar to the first case, bonds are not net wealth at the macro level, but they may well be net wealth at the micro level, depending on distribution.

[JKH]

P.S.

The national income accounts reconciliation of MMT NFA lies in the fact that NFA represents saving while the government’s debt position represents NFL dissaving (NFA is after all derived from a sector manipulation of national income accounting). The sum of the two is flat, which in a way is consistent with a net zero wealth position and consistent with Ricardian equivalence at the macro level. However, one must believe that the bond position is indeed an effective liability that must be repaid with taxes in order to validate this consolidation.

However, MMT does not believe that bonds must be repaid or that they equate to an inevitable tax liability. That’s key to the rationale for why MMT tears apart the consolidation in the first place, and puts the NFA interpretation into prominence.

So it seems that there are different dimensions to Ricardian equivalence at macro and micro levels. But more importantly perhaps, the question of whether or not the bonds are treated as a true liability and whether or not there is a true tax liability as a result hangs over the whole thing. In fact, the very question of whether or not people save to pay the tax, macro or micro, is dominated by the question of whether or not the assumption of a tax liability is valid in the first place. That’s what MMT objects to in the idea of RE, I think (and so do a lot of others).

Returning to the difference between RE macro and RE micro:

The paradox of thrift says roughly that if everybody tries to save too much, the economy will suffer as a result. The paradox of thrift is a real economy effect, because S declines.

In a way, this notion of a macro level Ricardian equivalence tautology is a sort of monetary paradox of plenty. It says that no matter how much people at the micro level might not save in order to pay a tax in respect of an assumed future debt repayment, the saving will always be there at the macro level in order to pay down that debt. It’s a monetary economy effect, because both NFA and the debt are monetary forms of cumulative saving and dissaving, and because NFA cumulative saving doesn’t disappear just because of the failure of micro agents to save for taxes.

[vimothy]

JKH @ 09:18 PM,

Nicely struck, sir! A succinctly stated comparative analysis of Modern Monetary Theoretic-Ricardian Equivalence effects under homogeneous- and heterogeneous-agent scenarios--

Very cool.

It got me thinking about your MMT net financial asset agent. Does MMT have a theory or model (possibly implicit) that explains the behaviour of this neoclassical representative agent analogue? When the government increases NFA, why does act the way it acts (whatever that might be)?

[JKH]

Vimothy @ 10.52 p.m.

"Does MMT have a theory or model (possibly implicit) that explains the behaviour of this neoclassical representative agent analogue?"

I don't know; not that I'm aware of. You'd have to ask the head MMTers about that. My guess is they'd say the representative agent approach is unnecessary. Their motivation in general tends to be geared toward understanding deficit economics at the macro level, in order to position their rationale for promoting more deficit spending flexibility than mainstream. So I think they’d avoid the representative agent approach for the most part, and for that reason, wouldn’t pay much attention to the idea of Ricardian equivalence or any debate around it.

BTW, I don’t see understanding MMT as being necessary to understanding the financial accounting for the macro around RE, although it doesn’t hurt, particularly in the way it analyzes the NFA piece. But I haven’t seen what I’ve developed here specifically in MMT.

[Bob Murphy]

Hey Nick, thanks for the kind words. Even now, the results of the paradigm shift are still ramifying (is that the verb I want?) through my psyche.

Let me encourage you in your disagreement with Steve Landsburg: I believe Steve that he has been thinking clearly about this from Step 1, but I can't believe he doesn't see why you and I believe that Krugman, Dean Baker, Yglesias, et al. have been convincing people that what we do today in terms of deficits can't possibly affect our great-grandkids, except because of deadweight losses from income taxes etc.

[Bob Murphy]

Yikes! That was an awful sentence. To clarify:

* Landsburg has been right all along, and I didn't understand him at first.

* Krugman et al. have been totally misleading on this stuff. Maybe they actually are aware of all these subtle issues, and it's just a coincidence that they are writing things that would have bolstered me in my earlier state of ignorance, but I doubt it.

[JKH]

Nick,

Here’s your apple model, interpreting Cohorts A, B, and C as “cohort traders” with long, short, or flat positions in apples, as they evolve in their apple trading books:

A opening position: flat apples (receives apples by transfer; pays apples to buy apple bonds)

A closing position: long apples (receives apples by selling apple bonds)

B opening position: short apples (pays apples to buy apple bonds)

B closing position: flat apples (receives apples by selling apple bonds)

C opening position: short apples (pays apples to buy apple bonds)

C closing position: still short apples (pays apples to satisfy apple tax liability; receives apples from maturing apple bonds)

All apple positions are marginal to pre-existing apple inventory. So a trader’s short position in apples represents borrowing from apple inventory.

All of this works exactly the same way when you substitute money for apples.

Traders are enormously reliant on accounting.

So were you, Nick, in your original post.

:)

[vimothy]

JKH,

In response to your PS,

It might be worth distinguishing between some of the various reasons that RE fails in the MMT analysis.

Firstly, in the institutional-MMT case, there is no macro or micro level tax liability corresponding to the NFA. Government bonds are net wealth.

However, one possible complication is the extent to which government bonds affect aggregate real consumption possibilities. At full employment (assuming this is the norm under institutional-MMT), their first order effect must be purely redistributive—perhaps with second order effects on the composition of output. So in that sense, it seems to me, government bonds are not net wealth.

Next, in the current institutional set-up, if the growth rate of the debt is less than the growth rate of the economy, then the debt never has to be paid back—in which case it makes no sense to save the bonds. In fact, rational actors with access to the model should know this, so we shouldn't expect them to put to one side bonds for a tax liability that doesn’t exist even at the macro level. Government bonds are net wealth.

However, as a matter of historical record, under current arrangements, governments do sometimes run surpluses. And even if they do not, higher current deficits might still result in lower future deficits at the margin (recall that the debt growth can be decomposed into primary deficit and debt service effects), which implies nominal micro and real micro and macro effects even if the nominal macro question is settled as a matter of accounting tautology.

In purely nominal terms, the savings that are required to pay the taxes may exist as an accounting reality at the macro level, but there exists uncertainty at both the micro and macro levels as to the real debt burden that household sector faces as a result of receiving the bonds, and hence its real cost in terms forgone consumption, and the timing of that forgone consumption, if it must be forgone. In addition, there is a question about the nominal micro distribution of bonds versus tax liability. Therefore (with an eye to maximum generalizability), there is ambiguity as to whether rational households should treat government bonds as net wealth. It seems to me that this conclusion is the most appropriate given current realities, even given my ostensibly MMT-orientated analysis here.

[vimothy]

Another question I find interesting is how MMT envisions the behaviour of the aggregates after the switch to institutional-MMT. For instance, do we at some point reach saturation / satiation point, after which no further additions to NFA are necessary; perhaps MMT would predict balanced primary deficits, with steady growth to NFA from the debt service? Does it derive a stability rule, perhaps, for NFA under a long term full employment scenario? dNFA=0? dNFA=dY?

[vimothy]

JKH @ 12:51 AM,

Thanks.

I expect you're probably right about MMTers' view of rep agents. But I'd also say that it's a strange view, given the epistemological similarities and methodological implications of considering an aggregate sector as an aggregate sector (MMT view), as opposed to considering an aggregate sector as its representative agent or sectoral average (neoclassical baseline models). In fact, I'd say the two approaches are, in this sense, exactly equivalent.

But leave aside the language of the rep agent. Why according to MMT does an aggregate like the household behave in the way it does? Or, one could ask the same question but at the level of individual and heterogeneous households.

Since it identifies a continuous lack of net nominal government expenditure at the macroeconomic level as the most prominent cause of misery in the economy, this is necessarily MMT's most important point of focus. But perhaps this comes at the expense of other considerations, at times. In particular, MMT is relatively uninterested in the behaviour of individual firms and households taken qua individual firms and households. Why do they act? How does MMT distinguish its analysis from the neoclassical line? It seems to me that MMT lacks a theory of the real economy, particularly its supply side. Doubtless this reflects my own lack of engagement with the appropriate literature, but such is my superficially informed impression, in any case.

[JKH]

Vimothy,

You make some very interesting points there (your third and second last, before seeing your last).

First, I think you know that I am not an MMTer. As a confirmed outsider, I’m an observer who appreciates what I think are some specific real strengths within the realm of MMT analysis of monetary systems. I think they’ve made a significant contribution. But my interest in speculating on their thought processes in total is limited by the fact that I’m deliberately not a member of the club.

Second, I haven’t seen the phrase “institutional-MMT case” before to my recall, but I think you mean a case in which the government is following MMT policy more or less.

Third, just before final points, let me recast something. I’d describe the following thought process as the ‘Ricardian equivalence interior monologue” of a representative agent facing such a consideration:

- Do I expect that the government will want to pay down debt at some point?
- If so, I must expect a government surplus at that stage
- If so, that means some sort of incremental tax burden, compared to the counterfactual of no surplus
- And if all that turns out to be true, do I want to save in a compartmentalized way in order to reserve for my share of that expected incremental tax burden?

Quite apart from those considerations, what I said before is true. If the government decides to pay down debt, the saving required to do that is always available at the macro. The question from there, depending on how the agent answers those questions above, is how the individual sets out his stall. To the degree that individuals expect taxes, and to the degree that they decide they want to save for those taxes, they will presumably scramble for their share of the tautologically determined macro saving that already exists to do so. And that will have consequences for other things. And there I think you need to drill down into micro modelling to figure out how all that will work. In my mind, the outcome will depend on such things as money supply, money velocity, multipliers, etc. etc. Too complicated for me to get into or attempt here. To the degree that individuals will attempt to save their share of macro saving that already exists, I can even see the paradox of thrift working at the margin of those who believe in Ricardian equivalence, because money velocity and multiplier effects would slow marginally down as a result.

Regarding the accounting question overall, it is a very important principle of accounting that the numbers best represent what it is you’re trying to communicate. There is a big difference between a certain tax liability and a tax liability that is possible but not certain. There are ways of dealing with the difference using probability analysis and/or option pricing techniques. But nobody knows the future with certainty. So the accounting for Ricardian equivalence concerns (which I think becomes accounting for the certainty or uncertainty of future taxes) needs to reflect your view of Ricardian equivalence importance. The implicit implication of the MMT NFA representation I think is that RE is bunk. Whereas the implicit implication of the consolidated national income representation of saving (where NFA is offset by NFL on consolidation) is that RE is fully operative. That’s the way it seems to me, anyway. And if you come to point where there are actual surpluses, then RE would seem to be in quasi-live mode, although you didn’t necessarily know that before the surpluses actually started up.

Your final question relating to NFA saturation and the consequential deficit spending trajectory is VERY interesting. But you really need to ask an MMTer like Scott Fullwiler for an answer to that. I could speculate on the answer, but I’d really like to draw the line there.

[vimothy]

JKH,

Thanks for the detailed reply. That's very helpful.

I do understand that you're not an MMTer, BTW. But I'm interested in your perspective as an informed outsider--certainly, it's clear who has the comparative advantage here. There also seems to be some overlap between the accounting-consistent framework you use to approach these problems, and the framework used by MMTers. Would that be a fair statement? I could imagine that MMTers might draw similar conclusions, given similar assumptions. But perhaps that overstates the case.

Before attempting to respond more fully, I'd like to ask a related question that might help to pin down some of the assumptions that drive the difference between neoclassical and MMT analyses. To the extent that I understand it, I agree that the accounting tautology you've a identified ensures that the macro level savings required to pay the tax obligation are always present. But is there an accounting convention or technique that takes into account changes in their real value? If so, does this affect your analysis of assured macro savings?

Forgetting about the micro case and attendant distributional issues here for the sake of simplicity, I don't know if it follows that the real macro savings are assured, given the fact of assured nominal savings alone.

[vimothy]

In other words, at the macro level, debt is swapped in when the deficit spending occurs, and swapped out when the tax is paid to retire the debt, so that the net nominal macro effect is a wash. But there might be fluctuations in its real value happening beneath that, so that the net real macro effect is non-zero.

[JKH]

Vimothy,

I’d say what I share with MMT mostly is the importance of using accounting logic and accounting frameworks as a subset of economic analysis. That’s a pretty broad shared interest. On the accounting, again I think you have to design the accounting system to reflect what it is you’re trying to measure. The general hierarchy for conventional accounting is along the lines of book value, market value, fair value, economic value, etc. These are all variations on the measurement of asset value. When you get into options pricing, you’re starting to build in specific assumptions about probability weighted measures into your accounting system. Although all asset values even at original book value reflect implicit probability and risk assessments of value and the price paid for assets at the time of the transaction.

At the macro level, government debt normally appears in economic analyses at book value. It can be market to market for specific purposes. But the tax liability that might be presumed to pay down debt is very much a probabilistic judgement. It can only exist as an expected tax liability at the best of times, until the point where the government actually announces the tax change, at which point agents can start to record tax liabilities on the basis of certainty. So accounting systems would regularly reflect the latter, but not necessarily the former. It would have to be a pretty exotic accounting system that reflected expected Ricardian tax liabilities. But maybe that sort of accounting is implicit in a lot of economic models. I don’t know. I do know that all of this debate about the debt burden is irrelevant unless you expect something by way of debt repayment, which means deliberate surpluses. That’s the precondition before trying to assess whether somebody is even attempting to save for this reason at the micro level. In summary, I’d say this area of economic analysis is pretty distant from any formal type of accounting, which is why it’s interesting to see that the explosion on the blogs recently includes some models that are in effect rudimentary accounting systems for the simulation of this debt burden problem.

[vimothy]

JKH,

Thanks.

Perhaps I should expand my point a little more.

Between any two arbitrary time periods there is a non-zero, and non-trivially non-zero, chance that the real value of the outstanding government debt might change; that is, since the debt is denominated in nominal terms, it's real value changes with changes in the price level. This implies (I think) that the macro level savings necessary in the event that the debt is later retired cannot be guaranteed in real terms.

In fact, thinking more about it, if that issue cannot be resolved from within the MMT framework, could this be a least upper bound for how far it can take you with regards to analysing the macro debt dynamic? Admittedly, there are a lot of implications to think through; maybe I overstate the case here too. This again raises the question of what drives the behaviour of non-government economic agents in the MMT model, which need to be brought in here. Hmm.

It also seems to me to be the case that, not worrying any more about the real-nominal distinction, at the micro level, households could face lower marginal future deficits as the result of current deficits, which might be an important influence on the actual sectoral level response. Say that the government is following a growth rule for the debt, where the primary deficit is the target variable that policy makers set to offset the debt service so that proportionate debt growth is stable at whatever the desired level (a long run goal of zero change to proportionate debt, say). Then, at the margin, higher primary deficits in the present means lower primary deficits in the future. I think this is an overlooked point of interest.

In anticipation of a possible response, it's obviously the case that NFA are at the same level either way (by definition of the debt growth rule); however, it's equally obvious that decomposition of NFA into primary deficit and debt service flows is distributional, but that this distribution might have implications for aggregate behaviour.

[Steve Roth]

Vimothy, JKH:

Also an MMT watcher, something of a convert for reasons similar to JKH's. Thanks for your good discussion.

Vimothy: "considering an aggregate sector as an aggregate sector (MMT view), as opposed to considering an aggregate sector as its representative agent or sectoral average (neoclassical baseline models). In fact, I'd say the two approaches are, in this sense, exactly equivalent."

I don't think this is right. JKH's MMT RA seems to be just a fractally identical pie slice of the aggregate, as opposed to an agent interacting with other agents -- with the resulting emergent macro properties.

I think you're right that MMTers don't think in RA terms because their descriptive macro model doesn't require it. They don't generally seek micro foundations for that model. If you look at the macro simulations built by Steve Keen (which work in very MMT/Godleyesque mode, though I think he'd resist being called an MMTer), he algorithmically models the behavior of different sectors in their interactions with other sectors -- banking, government, producers, purchasers -- but not down to the level of individual human agents.

This is definitely a weakness one could proclaim for MMT. The intersectoral algorithms might not properly model the emergent properties resulting from micro-level decisions. (Though I have to add that neoclassical microfoundation assertions, given their ridiculous starting assumptions -- necessary to make their models even work -- and their failures of prediction, are simply laughable.)

One can imagine true agent-based micro simulation modeling, algorithmically encoding the empirical insights on human economic behavior/decision-making from Kahneman, Tversky, et. al. into the simulated agents' behaviors -- plus some other necessary theorizing perhaps derived from neoclassical insights -- eventually working up from the bottom to meet or enrich those sector-based algorithms, rather like the Atlantic and Pacific rail lines, or the French and English chunnel sections meeting in the middle. But the work in that area is almost nonexistent so far. Some Santa Fe Institute folks have been making noises about such a simulation for a while, but I don't know of anything to show for it yet.

On government bonds: when issued or retired, they don't change the aggregate quantity of private NFAs. They just replace cash with bonds, then the reverse. Interest payments by the government do increase private NFAs, just like other government spending does.

"one possible complication is the extent to which government bonds affect aggregate real consumption possibilities"

This is the key issue, it seems to me: to what extent do the two scenarios (higher or lower deficit spending) affect the quantity of real-economy purchases/sales, expenditure/income -- GDP?

If higher deficits now result in a smaller economy in the future (relative to the counterfactual), a burden has been imposed -- Generation C, D, or E doesn't get to eat as many apples. Is that the case?

I still don't feel like I know the answer to that question. And I don't feel like the apple-economy model will answer it for me. I *think* the problem is: apples can be consumed; money can't. Modeling apples as money, or vice versa, might be resulting in an insoluble conundrum.

[vimothy]

Apologies for the duplicate post--I certainly wouldn't mind if somebody deleted either one of them.

[Nick Rowe]

spam filter is playing up again. I just rescued 3 posts, but 1 was a duplicate, so i unpublished it. Let me know if anything is still missing or wrong.

[Steve Roth]

JKH: "I’d say what I share with MMT mostly is the importance of using accounting logic and accounting frameworks as a subset of economic analysis."

Yes. I call them the accounting-based school of economics. Maybe its because I'm a businessperson that it has such appeal to me.

"you have to design the accounting system to reflect what it is you’re trying to measure. "

Yes again. The NIPAs, for instance, are designed to estimate the production -- measured by purchases and sales -- of real goods and services. As a result of how that's achieved, they're not useful in representing the whole financialized economy. "Income" in the NIPAs, for instance, doesn't count capital gains on financial assets. And I've said already how the NIPA definition of Savings is weird and unwieldy. Which all means you have to agglomerate the NIPAs and the Fed Flow of Funds Accounts to see what's going on. It seems like Wynne Godley's chart of accounts, which does include financial assets and is "stock-flow consistent," is a more promising accounting model.

"the tax liability that might be presumed to pay down debt is very much a probabilistic judgement."

At best. Risk vs. ("radical") uncertainty. Which brings us back to Keynes...

[vimothy]

Steve,

Thanks for your comment.

My point regarding rep agents is that they are in reality purely aggregate creatures, but simply treated as individuals within the context of a particular model for reasons of expedience.

To put it bluntly, I might say that a rep agent model is the result of not taking microfoundations seriously either.

Think of this. You note that Godley modelled aggregate sectors simply as aggregate sectors. But, taken on its own and considered in relation what I'm saying, this is question begging; namely, in that it leaves the question of why sectors act they way they act completely unanswered. The sector behaves in a particular fashion. How? Why? And what do you draw on to explain this if not its component parts? If it has no component parts, what exactly does it aggregate?

[vimothy]

Nick, You got it exactly right, thanks.

[vimothy]

Steve,

Apologies, I obviously wasn't paying attention when I first read your comment. I can see that you understand where I'm coming from.

[JKH]

Vimothy,

I must be missing something. In a monetary economy, the real value of the debt is identically equal to the real value of the saving that was originally and always associated with it, no? They are nominally equal, so they must be equal in real terms.

I seem to recall a discussion ages ago when I actually took an accounting course that the accounting profession had generally rejected the idea of real value accounting as being unviable and perhaps not terribly useful in any regular, systematic, conventional framework. But real value accounting can always be created at the option of the user, I suppose.

Steve,

I may be an outlier but I actually find the NIPA definition of saving to be very useful.

“Which all means you have to agglomerate the NIPAs and the Fed Flow of Funds Accounts to see what's going on.”

That’s an essential point. The flow of funds intersects orthogonally with NIPA.

The micro correspondents are income statement, balance sheet, and sources and uses of funds.

NIPA macro corresponds to income statement micro, and Fed flow of funds is actually an amalgam of balance sheet and sources and uses of funds perspectives.

I can’t imagine how else to do this stuff. I find the existing set of statements at both macro and micro levels to be totally coherent, although it can be challenging to integrate.

[Steve Roth]

JKH: "I may be an outlier but I actually find the NIPA definition of saving to be very useful."

No I'm certainly the outlier in finding it problematic. But you may be an outlier in actually understanding it -- with its necessary relationship to the FOFs.

"I can’t imagine how else to do this stuff."

That's an important statement. I see two possibilities:

1) I'm an internet crank who can't grasp the obvious, or

2) We're so used to thinking in NIPA terms that any other way of thinking is almost impossible. (Have you read Godley's "stock-flow consistent" paper. The word "savings" never appears. It's not needed.)

I'd give 60%+ odds to #1.

Two examples of why I think the NIPA "savings" usage is conceptually problematic/confusing (along with its associated accounting identities):

o It gives people the impression that if individuals save more (a.k.a. spend less) in a period, there will be more loanable funds available at the end. You hear this everywhere, from bozo to Bernanke. When in fact spending (a.k.a. not saving) doesn't reduce the stock of money available for lending; it just transfers it from one account to another (in passing, triggering the transfer of real goods). More velocity/turnover, stocks unchanged.

o If I withdraw $10,000 from my business bank account and buy ten computers for my employees, that's called "Saving" (Income - Consumption Spending), calculated in the NIPAs as a residual.

Spending is Saving?

The idea is that fixed assets are "savings." Okay, arguably so. But nobody uses the word that way. Wrong? No. Confusing? Contributory to further confusions? Heck yes.

But back to the original question: in MMT World, does deficit spending today mean a smaller economy (less apples produced/sold/bought/consumed) in Generation C, D, or E? And is that world a valid model for answering this question? I'm still flummoxed.

[JKH]

Steve,

I’m afraid I haven’t read Godley’s stuff yet, but I know that to date I have a sufficiently high regard for the views of those who regard him highly, that I would as well.

Regarding your first example, I think the antidote is to understand that in any given period covering flows, NIPA expenditures and output must equal NIPA income (roughly speaking). Which in a closed economy with a balanced budget, means I = S. Which means realized saving necessarily infers realized investment. And investment creates saving by logical deduction in that sense. From a NIPA perspective, using similar logic to loans create deposits, saving is the logical requirement that matches investment.

The best post I think I’ve seen so far on the internet regarding this sort of relationship was done by Andy Harless, notwithstanding that MMT is saturated with this sort of thinking. (I’m never sure that Harless realizes how close he is to MMT thinking in some of the stuff he writes. He strikes me as a hybrid.)

http://blog.andyharless.com/2009/11/investment-makes-saving-possible.html

Regarding your second example, I think it’s a question of what’s classified as investment. And investment means real investment. There is no concept of monetary investment in NIPA, or in the generally accepted definition of investment in economics, as far as I know.

The actual granular classification scheme for investment in NIPA is another issue, but that’s detail. If you research the detail, you can interpret it. I’ve never worried about whether or not computers should be classified as an investment, but I see it being logical at the level of businesses. Don’t know about household computer purchases.

“Spending is Saving? The idea is that fixed assets are "savings." Okay, arguably so. But nobody uses the word that way. Wrong? No. Confusing? Contributory to further confusions? Heck yes.”

I think this is problematic.

Spending CAN be investment.

Spending can be on consumption goods or investment goods.

And the QUANTITY of investment in a closed, balanced budget economy EQUALS the QUANTITY of saving.

BUT, the equivalence of quantities does NOT mean the equivalence of concepts or distinct economic flows.

In this latter sense, investment is NOT the same thing as saving in a monetary economy.

Investment is a component of real economic output. The economy produces consumption goods and investment goods. I’m not sure that too many people understand that the investment component in Keynesian type analysis actually refers to the PRODUCTION of investment goods by a given business enterprise, and not the actual PURCHASE of investment goods by a business enterprise.

Saving is a component of monetary income (in a monetary economy).

Monetary saving matches real investment in a closed, balanced budget economy.

(Nick’s the expert on how to deal with this stuff in a purely real barter economy.)

“The idea is that fixed assets are "savings."

Again, that’s problematic.

Fixed real assets are classified as the stock of cumulative investment goods produced (allowing for depreciation).

In a closed balanced budget economy, savings are the stock of accumulated monetary income that has not been spent on consumer goods, net of any monetary saving that has been written off in a monetary sense due to depreciation or other reductions in cumulative investment. Adjustments for budget and current account imbalances are required from there.

I think the “residual” thought process problem that you refer to is a function of the multiplicity of the Keynesian output and income categories:

For example, in a closed balanced budget economy, G and T would net out, and the remaining categories would be C, I, and S. And provided you knew the classification of C and I, it would be trivial to identify I and S directly on the expenditure side, rather than go the residual route of identifying S as that which is not C on the income side.

But the requirement for residual logic comes in because of the inevitable imbalances in both the government budget and the current account. It gets tricky to identify S without going into residual mode in that real world. But all the pieces of the puzzle fit, as they’re defined.

“But back to the original question: in MMT World, does deficit spending today mean a smaller economy (less apples produced/sold/bought/consumed) in Generation C, D, or E? And is that world a valid model for answering this question? I'm still flummoxed.”

That is definitely a question for the MMTers.

BTW, one reason I’m reluctant to take a stab at this sort of question (apart from the fact that I’m not an MMTer) is that it hits one of several big sore points with me about MMT – which is that from my perspective, MMT studiously avoids the whole issue of financial planning and risk management as it is generally pursued in matters of financial management. I have a really big problem with that sort of attitude in dealing with the uncertainty of the future, in the broadest sense of financial management.

Anyway, hope the rest makes some sense. This stuff is not easy, but my impression is that economics in general is not easy, and there are no short cuts to understanding the best way to think about things, whatever that may be.

[Steve Roth]

JKH, thanks. I think I get everything you're saying here. It's the real goods vs. money I'm still struggling with.

Yes, investment/consumption is somewhat arbitrary, time-wise; lunch is an investment in the afternoon's work, and we consume drill presses over many years by using them. (I just assumed that the NIPAs count business computer purchases as investment.)

Thanks *very* much for the Harless link. Excellent and very helpful. Nice to see that others more versed than I also consider this puzzling, and worth considering -- that it's not totally obvious. He's saying/thinking a lot of the things I'm saying, but generally better.

These two lines especially help:

"Savings are defined as unconsumed income."

"now you’re not talking about saving income; you’re talking about saving resources. Resources have to be “saved,” in the sense of “not used up by consumption,”"

This goes back to my apples versus money issue. Apples are produced using inputs, and are consumed. Money isn't. There's a flow of apples. There's a circuit (circuits?) of money. There's an inequivalence there that makes it hard for me to grasp. "Supply" seems to have very different meanings and implications in the two contexts. (Hence my discomfort with the "demand for money" concept.)

When you buy a hamburger, that's not consumption. It's consumption *spending* -- a transfer of money from your account to someone else's, *triggering* the creation/transfer of real goods.

When you eat a hamburger, that's consumption. Ditto when you use a drill press.

So in what seems to be a very real sense, the unconsumed income -- savings -- from this year is stored in that drill press, not in bank savings/financial assets -- because that investment purchase/money transfer triggered the creation of a drill press that we didn't have before; we don't have any more financial assets. That's why it's hard for me to think of it as "savings," which in our financialized economy generally refers to money.

I'm going to write this up at more length over at my place, stop burdening Nick's comments thread. But don't hesitate to respond more here if inclined. Any aid in helping me to look less stupid is much appreciated. ;-)

[vimothy]

Steve,

Interesting thoughts in your last comment at 2012 at 07:50 PM. This is a good observation:

"So in what seems to be a very real sense, the unconsumed income -- savings -- from this year is stored in that drill press, not in bank savings/financial assets -- because that investment purchase/money transfer triggered the creation of a drill press that we didn't have before; we don't have any more financial assets."

If you like, have a look at my response to JKH below.

I'd also be interested to find out more about your opinion of the different modelling strategies on offer. When you wrote that neoclassical assumptions are ridiculous, did you have something specific in mind? What's ridiculous about it, and what would you replace it with?

JKH,

Perhaps an example would make things a bit clearer.

Say that the govt sells a bond to some proper subset of the household sector—let’s call them “bondholders”—in period t1, for 100 units of nominal currency, and let its real value also equal 100.

This bond pays out 110 in nominal terms at some future period tk, at which point the government levies taxes equal to 110 that fall equally on the whole household sector.

Say that the price level, which we’ve normalised to 100 at t1, falls unexpectedly over time so that at tk it is 90.

Now, the real value returned to the bondholders (which is obviously is equal to the real value of the taxes levied on the household sector to retire the bond in tk) is greater ex post than ex ante. The bondholders received an unexpected windfall from taxpayers. Their real return should have been 110, but instead it was 122.

In this deflationary scenario, one could imagine the real present value of the debt growing larger than the present value of the economy's productive capacity—meaning that it could become quantitatively impossible for the government to discharge the debt through taxes; a kind of “real terms insolvency”. (For a quick e.g., fix aggregate per period production at 1000 in real terms throughout and let it equal 0.1 per period in nominal terms after the deflation. The government would need to save total production for 1100 periods between t1 and tk to retire the debt).

BTW, to clarify, I don’t know if this has any implications for the possible use of money creation to resolve the problem.

[JKH]

Vimothy,

We know the nominal present value of cash flows is the future expected nominal value discounted by the nominal interest rate.

We know expected future nominal cash flows (expectation is subjective assumption), and we know the nominal interest rate by observation and/or interpolation/extrapolation.

Then, the assumed inflation rate and the corresponding derived real interest rate are what they are.

With a zero bound and deflation, you get high real rates of interest. That’s a monetary policy problem, with potentially disastrous real economic consequences. I don’t see it as a problem in nominal present value calculation or any derived real value result.

Nominal taxes are required to pay down nominal debt.

So I’m not seeing where the problem is.

Again, high real rates combined with the zero bound are a monetary policy problem. If not fixed, the real output of the economy could spiral down, in part due to the real value of debt. That becomes a nominal contraction with massive debt write-offs.

Every nominal write-off is magnified in real value terms. Nominal values are actual and observable or assumable. Real values are unobservable and derived.

If you want to market assets and liabilities to “real value”, you may get some weird results for the real value of equity, I guess. But at the end of the day, both nominal and real equity are just accounting plug items.

The worst that can happen for nominal valuation is a nominal discount rate of zero, in which case present value equals future value. But in such a real rate environment, you have to adjust all expected future cash flows for the probability of default in an economic contraction. That’s why equities wouldn’t go to infinite value.

Transactions occur at nominal value, whatever may be the real value perceptions.

As far as government finance is concerned, nominal value debt is either retired or it isn’t, with nominal value taxes. That’s undeniably true. Deflation can produce a real economy performance trap, depending on government policy. Austerity with deflation isn’t working so well, as we see in Europe.

That's all off the top of my head. Forgive the pun, but I’m not seeing the value added in any comprehensive attempt at real value accounting, which may be why you don’t see it standardized anywhere. And my suspicion is that any potential math conundrum you may foresee because of attempted real value accounting must be offset by necessary rational expectation of real economic contraction and corresponding expected and actual losses in both nominal and real terms (e.g. discounting equity cash flows at a zero nominal rate). And remember that nominal losses (actual and expected) are magnified in real terms under deflation, so that can wipe out a lot of inconvenient math on the real side.

So, you may be right, but I’m not seeing this as an accounting problem.

[Nick Rowe]

JKH: "So, you may be right, but I’m not seeing this as an accounting problem."

It isn't an accounting problem. It's a behavioural problem. The maintained assumption of economists is that real desired saving (and investment) depend on real interest rates.

[Steve Roth]

Nick on my previous re: money stock and price level, just to be sure, I graphed this.

http://research.stlouisfed.org/fredgraph.png?g=4gd

Since July 1981:

5-6x growth for M1 and M2 (and M3 was on track to far exceed that), with lots of variance.

2.5x for CPI and core CPI (with little variance).

[123 (TMDB)]

What would Paul Krugman say?

Since the real Krugman is unavailable, I have asked my inner Krugman. The result was a meta critique. "It is wrong to apply the overlapping generations model to government debt, as it should be applied to the total debt (public and private). Only by applying the overlapping generations model we will get fair comparison between the policies of the Bush and Obama administrations. Before the crisis, various methods were available to individuals who wanted to increase their spending by reducing the spending of other cohorts (second mortgages, zero downpayment mortgages and all other ways to collect "tax" revenue from your unborn children). Right now the availability of these Bush-era methods is reduced, and all we have is the faster growth of government debt instead. As the growth of the total debt is slower now than before the crisis, the current problem is the insufficient burden on the future generations (assuming the revealed pre-crisis preferences of NYT readers were right). This problem was not important enough to include in the NYT piece. Of course, the main problem is that the debt cycle has stopped (see "Debt, Deleveraging, and the Liquidity Trap"). And by the way, the phrase "some of our children will pay money to other children" was also intended to remind readers that they can choose to which group their children will belong."

I also wanted to ask my inner Krugman whether an inheritance paradox exists (i.e. when you try to increase the bequest in a liquidity trap you reduce the total lifetime spending of your descendants because of various first order and second order effects), but instead of an answer, I received a "stack overflow at line 47923" error from my Krugman emulator.

[Adam]

Nick - Thank's for the response. I actually don't have any doubt that people respond to taxes that they see and understand (ala the window tax post your colleague put up). My problem isn't with that step. If people know there will be higher future taxes, I have no doubt that enough people respond to that have a measurable effect.

My problem is getting from a deficit today to higher future taxes. It seems to be an article of faith for some to whose political biases it appeals, but I think there is enough uncertainty in that connection (in the form of other tax, fiscal and monetary decisions and expected growth) that I'm skeptical of much real world reaction to it. And, of course, I'm skeptical that many people can or would chose to do anything about their children's future taxes (i.e., most people don't have much to bequest).

So we're starting from a debatable premise and then projecting rational reactions to that as though it's true, which to me seem like two pretty big potential sources of error married to each other.

[Michael Carroll]

Ok, lets try this in the context of the apple model.

I want to introduce fiat money and claim that debt with zero percent interest IS fiat money and that the govt can use this to manipulate the nominal interest rate.

Look at this net worth statement for each generation (please excuse rounded numbers):

A 1000 apples + 110 apple debt asset (includes 10% interest)
B 1000 apples + 119 apple debt asset (includes 8% interest) + 2 apple dollars
C 1000 apples + 126 apple debt asset (includes 6% interest) + 4.5 apple dollars (note the money unlike the apples can be inherited)
D 1000 apples + 131 apple debt asset (includes 4% interest) + 7 apple dollars
E 1000 apples + 133.5 apple debt asset (includes 2% interest) + 8 apple dollars
F 1000 apples + 133.5 apple debt asset (includes 0% interest) + 8 apple dollars

Which the is the same as.

F 1000 apples + 141.5 apple dollars

By issuing fiat money, the government converts a promise to rob EACH next generation by 10 percent MORE into a promise to rob SOME future generation by EXACTLY 141. 5 apples. That is a promise that the government can put off perpetually without burdening any future generation.

I do have to question whether I have violated any key assumption of a one good model when I introduce fiat money? Did we do the same when we introduced debt?

(apologies Nick, I tried to post a version of this to an earlier thread and that may be in moderation)

[vimothy]

JKH,

I'm thinking that the problem might be this—what, if anything, ties the nominal identities to the real identities? The nominal value of the asset equals the nominal value of the liability, and the real value of the asset equals the real value of the liability (ex post identity), but there's nothing there that enables us to go easily from nominal to real or vice versa. We need something like behavioural equations from economic theory to do that.

In the scenario I outlined above, in the infinite-deflationary limit, real debt becomes infinite (that is, its relative value in terms of consumption units becomes infinite). But no corresponding real savings exist that could possibly discharge the debt, since aggregate production would also need to be infinite, and it does not change. There is the asset itself, of course. Its ex ante real value exactly offsets the ex ante real burden of the tax liability.

But the difference between bondholders and households is a non-empty set. We cannot simply write off the value of the bondholder’s asset against the taxpayer’s liability. That’s equivalent to an admission that the real savings don’t exist (no consumption is transferred). If the government simply taxes the bondholders to pay the bondholders, that also looks like effective default (with the same implications for consumption).

Ex post, after the debt is written off, the real value of the tax liability and the real value of the bond are both identically equal to zero. Real savings did not increase with the real value of the debt.

What this shows I think is that the savings are not guaranteed in real terms, despite the guarantee that there exists an asset of equal nominal value to the tax liability at the macro level.

[JKH]

V,

“We cannot simply write off the value of the bondholder’s asset against the taxpayer’s liability. That’s equivalent to an admission that the real savings don’t exist (no consumption is transferred). If the government simply taxes the bondholders to pay the bondholders, that also looks like effective default (with the same implications for consumption).”

This is rough and could be shorter, but I can’t be bothered to edit it:

Make it simple by assuming a closed economy.

C + I + G = C + S + T
I + G = S + T
G – T = S – I

The left hand side, when G exceeds T, is a budget deficit expressed as a positive number

The right hand side then must be a positive number to match.

Which means S exceeds I

(G – T) is the budget deficit

(S – I) is equal to the non government sector’s net financial asset position as a flow increment generated by that budget deficit

So NFA is the mirror image of the deficit, and equal to the bonds issued to fund the deficit

Notice that S = I + (S – I)

Which means total saving is the amount required to fund investment, plus NFA

But NFA is the mirror image and equal to government bonds

So NFA represents saving by the non government sector

All of the above is standard MMT decomposition

Over time, you simply add up the deficits, bonds, and NFA as the cumulative result of flows and you get the representation of debt and stock NFA

The stock NFA represents savings, where savings is the stock that corresponds to cumulative saving flow

So you have:

Government Debt = NFA as saving
Government NFL = non government NFA

And the point is that the nominal savings that corresponds to nominal debt never disappears UNTIL the non government sector is taxed

Taxes destroy NFL and the NFA that matches it 1:1

And ALL of this holds true, REGARDLESS of what happens to the real value of real physical assets in the real economy, or what happens to the real value of nominal financial assets that are captured in the (MMT) “horizontal” section of the financial economy.

Try it. Try taking the nominal value or real value or both of everything else in the economy down to zero, and see what happens.

Nothing can affect the equation Gov NFL = non Gov NFA and nothing can affect the values in that equation other than government deficits or surpluses.

And the NFA pays for the taxes. Non government just cashes in their bonds in exchange for extinguishing a tax liability. And so on as I’ve explained elsewhere.

So the nominal value of the bonds ALWAYS equals the nominal value of NFA, and because of that, the real value of each as a financial item is always equal, no matter what the price level. There’s absolutely no real economy even required to make this true, other than the identification of a price level so nominal financial values can be deflated to real values.

BTW, taxing to pay down bonds is equivalent to defaulting on bonds without the tax. Exactly the same effect.

And taxation is equivalent in flow terms to negative saving, so as NFA is taxed down, it unwinds the stock of NFA saving.