Sunday, October 14, 2012

Nick Rowe Agonistes

I’m astonished, once again, that a topic as fundamental and important as the impact of national deficits on future “generations” (quotes explained below) is still a debated issue in economics.  But apparently it is.  And once again it has created a flurry of blog posts, from Dean Baker, a frustrated response from Nick Rowe who wonders why his overlapping generations model has not settled this issue, and responses to Rowe from Brad DeLong, Dean Baker, Mark Thoma, and Paul Krugman.  I’ll also point to Robert Murphy here (worth reading for the comedy value alone, although I think he may have been too hard on David Brooks, who seems like a perfectly nice gentleman) and here, and Robert Waldman at Angry Bear here,   And of course as long as you’re in the neighborhood you can look at my own posts on phantoms in the dark, starting here…but if you’re in a link-following mood, start with Baker and Rowe.  They are where the latest flurry started.


I’m not sure that my additions to this will help resolve anything if the sheer mass of intellect represented at the end of those links has not.  But heck---it can’t hurt to make my comments, can it?  Particularly when the issue is this important.  It relates to many of the topics that are prominent in politics right now, and probably will be for the next decade or two at least.  National debt, the aging population, income distribution and redistribution, are all part of this discussion.  Professor Rowe’s overlapping generation model in particular is almost a direct picture of the impact of the creation of social security, and of the loss of social security if that should ever happen. 

The reason for the quotes around the word “generations” above is that this word has caused some confusion in the conversation.  Most of the papers and blog posts that discuss this topic have not actually been concerned with generations, with groups of people born at roughly the same time who travel through the same bit of history together: they use that word only to indicate that they are considering accounting periods---years, generally, or decades---that are far apart from each other in time.  They are asking whether debts incurred in one year will have a negative economic impact in some other year far in the future, a year populated by members of future generations, perhaps, but considering each period on its own merits without trying to separate the multiple cohorts that share earth at any time or follow each cohort through its time here.  But others do really mean cohorts; and sometimes, particularly when politicians are using it, the meaning intended by using that word slips from one interpretation to the other within the same argument, or even within the same sentence.

In the list of links above Krugman, Baker, Thoma and DeLong seem to me to intend the first meaning; they seem to be talking about the economic potential available to people alive at a particular future period.  But Nick Rowe and Robert Murphy, in particular, are looking instead at the second meaning.  They specifically write about, and give examples of, overlapping cohort-type generations, looking at the total consumption available to each generation through their lives, the total that they have consumed after passing through all the individual time periods in their bit of history.  I want to comment primarily on the examples discussed by Rowe and Murphy, but I’ll introduce that by saying that I don’t think the other writers actually respond to Rowe’s model.  Krugman recognizes the terminology issue but talks right past it when he says:

“First, however, let me suggest that the phrasing in terms of “future generations” can easily become a trap. It’s quite possible that debt can raise the consumption of one generation and reduce the consumption of the next generation during the period when members of both generations are still alive.”   

The emphasis is his, and that matters---because Rowe’s overlapping generations model, in his own example of it, and mine that I presented here and reproduce below, and very explicitly in Murphy, shows that consumption can be raised for the lifetime of the present generation and reduced for the lifetime of one or more generations in the far future, long after the present generation has left the field.  In other words, this model is explicitly not talking about “the period when members of both generations are still alive.”

Here’s the example I used in this post to explain it:

Each of the diagonal arrows represents a single generation, with consumption represented by apples available during two separate periods of their lives.  The first diagonal-line generation gets what Murphy at one point calls the “free market” allocation, which is a large number of apples when they are young, and then a small number when they are old, with a total of 4 apples over their lifetime.  Generation 2, counting from the left, has the good fortune to spend its young years under a free-market allocation, but then benefits from a redistribution---Social Security, perhaps---during its old years, with the result that as a generation they get to consume 5 apples over their lifetime.  Then they die, and new generations take their place.  The next generation spends both young and old periods under the redistribution scheme that helps the elderly, and they get 4 apples over their lifetime, and they die.  But then the redistribution scheme is repealed, and the fourth generation from the left has to accept redistribution away from itself during its young period, but does not benefit from redistribution toward itself during its old period.  Over their lives they only get to consume 3 apples.   The generation that benefited from all of this was absolutely not alive in the same time period as the generation that lost out.  And there’s nothing that requires the redistribution scheme to last only two generations; it could last fifty generations, and when it is removed the final generation that pays in when it is young and receives no benefit when it is older will pay the cost, perhaps hundreds of years later. 

I’m grateful to Professor Rowe for explaining this model to me in a comment after I first mentioned that I was puzzled about it.  I think the insight in this model is important.  We instituted Social Security in 1935, and generations have benefited from it ever since.  Now there is talk about changing it, and there are many who would like to end it or expect it to end on its own by going broke.  We can see that if it does end the last generation under SS runs some risk of having its lifetime consumption reduced.

But I may need another comment from Professor Rowe to clarify the model’s conclusions.  I don’t see this model as having anything directly to do with debt or deficits.  Nothing in the picture above mentioned how the redistribution was financed.  If the redistribution during the middle two periods is financed entirely through taxes, the same benefit is experienced by generation 2, and the same loss by generation 4.  And, of course, it’s possible to create debt for other reasons than redistribution of incomes: if a debt leaves everyone’s incomes the same through time, how does the overlapping generations model indict it?  If the redistribution it creates is random with respect to age in its impact, so that the while income may be redistributed between people, the result does not imply a redistribution between cohorts, how does an overlapping generations model indict it?  This model seems to be directed very narrowly on debt created to finance redistribution between generations.   And finally, the cost to a future generation only occurs when the redistribution is reversed.  If it is maintained forever, through taxes or through new debt, or even through the creation of new money, there is no future generation that endures a cost. 

Nick Rowe responses to this last criticism this way:

“Some people might argue that the interest rate is below the growth rate, and will stay that way, which means the economy is dynamically inefficient, ponzi schemes are sustainable, and what the economy really needs is a government-run ponzi scheme which really would not be a burden on future generations, because the debt plus interest would be rolled over forever, with no tax increase on any future generation. That's OK too, if you can convince us your assumption is correct.”

But that’s the way it has worked for a very long time.  National debt is rarely really paid off completely with budget surpluses, but it is often reduces as a share of GDP through GDP growth, which will happen only if growth is greater than whatever new deficits must be financed, including interest.  It is possible that the economy is dynamically inefficient with regard to federal debt due to a general political disapproval of debt in the public sector.  It seems arguable, at least, that it is the assertion that the future will be different from the past, the assertion that the debt will harm us more in the future than it does right now, that requires evidence.  (I don’t think that models and evidence to show that are impossible: total federal debt in the United States is now more than 100% of GDP, and there are big expenses looming in the middle distance.  There is a wide discussion right now, as there should be, about the impact and limits of debt overhang, with some people arguing that a national debt that is too large actually reduces growth.  But that’s a completely different argument than the overlapping generations model, and it’s an argument that’s still just beginning.)

And finally, even if we convince ourselves that this overlapping generations model directly addresses debt, rather than redistribution however financed, or otherwise convince ourselves that debt not only can but must impose a burden on some future cohort through their lifetime, that doesn’t really impact the argument that is being made by Krugman and the others.  They are time-slicers, arguing that debt may redistribute income within a future year but that it doesn’t burden the national population as a whole, young and old together, during that year; they aren’t really looking at cohorts, so their “zombie idea” can’t be killed by a model that may illuminate the redistributional impact of debt on cohorts.  They can agree with the overlapping generations model, and still maintain that current debt doesn’t impose a burden on future time slices or accounting periods. 

The overlapping generations model does not deny that, in fact.  During each vertical time slice in the picture above the economy creates 4 apples altogether, time period after time period.  I think that part of the point of the model is that some cohorts may be burdened over their lifetimes even if no time period suffers any economic damage from debt-financed redistribution schemes.

No comments:

Post a Comment