Tuesday, November 08, 2011

Let them eat NGDP


This weekend, at an INET conference, Mike Konczal of Rortybomb made the excellent point that the econ blogosphere is probably adding more new policy ideas to the public discourse right now than academia itself. He identified NGDP targeting as the most important and dramatic example.

And boy was he right! NGDP frenzy has reached truly astonishing levels. I have my doubts that the Fed will actually adopt an NGDP target, but there is no denying that the blogosphere and much of the popular press has fallen for this idea like girls at my college fell for Thom Yorke back in 1999 (no I'm not bitter, why do you ask?). These days Scott Sumner says "NGDP" like Herman Cain says "9-9-9." Now, I am all for looser monetary policy, whether it comes in the form of an NGDP target or not. But NGDP has reached the point where, like Radiohead, its cult status is actually starting to annoy me. 

Take, for example, this post from Matt Yglesias (an extremely sharp dude) on how NGDP is the "actual thing":
[I]t seems natural to many people to take [real GDP and inflation] as “given” and understanding NGDP (= GDP + inflation) as somehow constructed and exotic. But actually NGDP is, relatively speaking, the simple quantity here. It measures total spending in the economy. You count everything, add it all up, and you’ve got your NGDP...Social construction enters the picture when we try to move from this nominal quantity to the allegedly “real” one... 
[W]e have all these different measures of inflation...[but it's]not directly measurable the way NGDP [is]...Nor is it particularly clear what, in principle, “real” GDP is measuring. Think up all the classic critiques of GDP as a measure of human welfare...When you shift from GDP to “real” GDP, though, you’re not measuring the total quantity of spending anymore.
I just don't think this is the right way of seeing it.

Matt's main point is that because inflation is mismeasured, real GDP ("RGDP") is also mismeasured. Sure, that's true (though of course NGDP itself is not measured with anything remotely approaching certainty). But that doesn't mean NGDP is "better" than RGDP. An accurately measured number is not necessarily more useful than a less accurately measured number, if the latter is something we care about and the former is something we really don't.

In terms of economic well-being, we do not care about NGDP. Or at least, we shouldn't. To see this, take the example of Zimbabwe. In 2007, Zimbabwe's NGDP grew at a rate of over 60,000%! No, that is not a typo. The "total quantity of spending" in Zimbabwe at the end of 2007 was more than 60,000 times what it was at the beginning of 2007. For real.

Anyone still think NGDP is the "actual thing"?

The problem with NGDP is that you can't eat it. It really is just numbers on a page. Normally, it is a reasonable approximation to a measurement of stuff we can eat. But when someone starts playing around with the numbers on the page, as in Zimbabwe's hyperinflation in 2007, NGDP stops being any kind of measure of stuff we can eat. 

In other words, NGDP has a potentially large source of mismeasurement in it, and that mismeasurement is inflation itself. When we try to take NGDP and recover some kind of measurement of how much eat-able stuff our economy produces (i.e., RGDP), we have to try to account for the mismeasurement represented by inflation. And of course we end up mismeasuring inflation, as Matt points out, but all that means is that we are mismeasuring our mismeasurement of RGDP. Inflation is still the first-order mismeasurement.

Actually, the world abounds with cases like this. It's easy to measure how many hours a worker sits in front of a desk, but hard to measure how much useful work (s)he actually accomplishes in that time. But it's the latter we care about, not the former.

Now, I am not saying that the U.S. is in danger of hyperinflation. It's not. Nor am I saying that RGDP is a perfect or even a great measure of human happiness. It's not. My point is that NGDP is not supposed to be a measure of how well the economy is doing. People who support NGDP targeting, like Scott Sumner, support it because they believe that NGDP is a good indicator of how much money we should print. Not because they think that NGDP is intrinsically good. 

NGDP may be less indirectly measured, but RGDP is still the "actual thing."

Like I said, I support printing money and buying stuff. And I think OK Computer was a rockin' album (and Kid A and The Bends were good too). But thinking of everything in terms of NGDP does not make all of macroeconomics suddenly simple and comprehensible and predictable. And Radiohead didn't save rock. Sorry, fans, but there it is.


Update: Scott Sumner agrees about Radiohead. Unfortunately, he says I'm "criticizing NGDP" and "missing several key points." Huh?? If you read the above post, you'll notice me saying:
My point is that NGDP is not supposed to be a measure of how well the economy is doing. People who support NGDP targeting, like Scott Sumner, support it because they believe that NGDP is a good indicator of how much money we should print. 
In Sumner's "rebuttal," he writes:
NGDP is...the proper indicator of nominal shocks...[it's]a nominal shock indicator, not an indicator of living standards.  I’d be the first to admit that RGDP, with all its faults, is a better measure of living standards than NGDP...[but NGDP is more] useful in developing optimal monetary policy.
Not to be excessively whiny, but is that not exactly what I just said???

So exactly what "key points" am I missing? And how on Earth am I "criticizing NGDP"? My gosh, people. NGDP is an economic indicator, not a fair damsel in need of gallant knights to defend her maiden virtue. Settle down!

19 comments:

  1. But RGDP is not the "actual thing" either. (In fact, that was the point of the person who said they couldn't eat iPhones.) Hamburgers and gasoline and diamond rings and movies and physical therapy and depositions and lectures and root canals are the actual things. RGDP is an attempt to combine a bunch of actual things into a single abstract thing that exists only in the mind of economists. It's true that you can come up with a more elegant theory if you ignore the fact that there is more than one good, but that doesn't mean that the theory accords more closely with "actual fact" than a less elegant theory which recognizes that there are multiple goods and chooses to measure them in terms of a single actual good used as numeraire. There is no unique "actual thing," but at least money is an actual thing, while RGDP units are not.

    ReplyDelete
  2. Yes. So?

    The "actual thing" we care about is a social choice function.

    ReplyDelete
  3. In the post you claim that "RGDP is still the 'actual thing.'" I've disagreed. Are you changing your position by saying "yes" to my disagreement?

    Sure, if you're looking for a measure of social welfare, and your only choices are NGDP and RGDP, then obviously RGDP is a better choice. RGDP is a social construct that happens to be a vaguely credible measure of social welfare. It's not an actual thing. NGDP is an actual thing: the total monetary value of all the goods and services produced[1]. It doesn't happen to be very useful for the specific purpose of measuring welfare, but for positive economics, we don't need to measure welfare. (And if you ask me why I care about RGDP, it's not primarily because I think it's a good measure of welfare.)


    1. OK, there's still a lot of social construction involved in defining what goods and services are produced, but it's an order of magnitude less, much closer to being an actual thing, if you measure them in terms of a single good such as money.

    ReplyDelete
  4. Are you changing your position by saying "yes" to my disagreement? No, only that we are using the term "the real thing" in different ways.

    NGDP isn't actually a "real thing" in the sense that you mean, since much of it is imputed, and does not involve actual dollars changing hands.

    RGDP equals NGDP in the base year, so in that year, RGDP and NGDP are equally as real and measured no better or worse than each other.

    "Measures of social welfare" are value judgments...normative statements.

    ReplyDelete
  5. "...we are using the term 'the real thing' in different ways."

    But you haven't (so far) argued that I'm using the term incorrectly, so I don't see what your original point was. RGDP is RGDP, which is in one sense an actual thing and in another sense not. NGDP is NGDP, which is in one sense an actual thing and in another sense not. I don't think Matt Yglesias will disagree that they both have valid uses.

    "NGDP isn't actually a "real thing" in the sense that you mean, since much of it is imputed, and does not involve actual dollars changing hands."

    That's (part of) what the footnote to my last comment was about.

    "RGDP equals NGDP in the base year, so in that year, RGDP and NGDP are equally as real and measured no better or worse than each other."

    But for RGDP, we only care about intertemporal comparisons, so the base year value by itself is irrelevant. It is meaningless to say that RGDP=X in a particular year unless you are comparing it to another year.

    "'Measures of social welfare' are value judgments...normative statements."

    And one of the main purposes of RGDP is to help make (or formalize) such value judgments. I think this tends to support Matt's argument that it is a social construct.

    Re-reading your post, I think you're misinterpreting Matt's point. I don't think he means to suggest that NGDP is somehow generically "better" than RGDP, only that it's a more clearly defined concept. I can explain fairly precisely to a non-economist what NGDP means (conceptually, though admittedly there's a lot of fudging in the actual computation). I can't explain precisely, even to another economist, what RGDP means, at least not in a few minutes. To define precisely the concept of RGDP would require several hours and a fair amount of math, and even then the definition wouldn't be satisfying (at least not to me). Of course it's easy to fudge a definition by pretending that relative prices don't change, but in real life relative price changes are very important.

    ReplyDelete
  6. Harless:
    "Re-reading your post, I think you're misinterpreting Matt's point. I don't think he means to suggest that NGDP is somehow generically "better" than RGDP, only that it's a more clearly defined concept."

    I agree with this. We wouldn't be discussing NGDP if the Senate wasn't blocking Obama's jobs bill or if the Fed was fulfilling its dual mandate.

    NGDP measures the velocity of money. If a dollar turned over six times in the course of the year, it would produce $6 nominal GNP. But if the same dollar, for unanticipated reasons turned over only five times, there would be only $5 in resulting economic activity - a gross shortfall when it was multiplied by the entire money supply.

    At Bernanke's recent press conference, a New York Times journalist asked him if they were considering NGDP targeting (go blogosphere!) and he answered "no" because their focus on inflation and "price stability" is working.

    But the subtext of the question was that people don't think the Fed's current system is working - Bernanke admits the forecasts are "unsatisfactory"(!)- and looking at the output gap of what the economy could produce (trend NGDP) and is producing would highlight this.

    I support NGDP level targeting but would be happy if Bernanke decided to target higher inflation instead.

    ReplyDelete
  7. "Hamburgers and gasoline and diamond rings and movies and physical therapy and depositions and lectures and root canals are the actual things. RGDP is an attempt to combine a bunch of actual things into a single abstract thing that exists only in the mind of economists."
    Like god, a unicorn and NGDP. All are metaphysical concepts. Read the logical positivists like Moritz Schlick, Neurath or Carnap. By the way, they influenced the Austrian school of economists, who had great difficulties to accept the aggregate concepts of national income accounting.

    ReplyDelete
  8. RGDP nor NGDP are measures of social welfare. They are more or less accurated indicators of the state of economy.
    The state the economy is not the state of social welfare. The economy can be in cero enemployment, and people not happy at all because they ´d like to work less, Or get better medicare assistance.
    Economy statistics, like GDP, are essential to measure change in the time. Because we know that teher are relationships that works: growt and employment are correlated; productivity through time translates in less hours worked for more product... Leisure growes through times. mor leisure means more time to creative work... and so on. To measure all that is important.

    ReplyDelete
  9. I think Matt's point can be quite succinctly summarised by saying that there is a 'fact of the matter' as to what NGDP is, because there is a 'fact of the matter' as to how much money is exchanged for final goods and services (whatever those goods are). The disambiguation of NGDP into Real GDP + inflation is the output of an essentially contestable model of what 'really' constitutes that set of good and services - i.e. what makes one good the same as another, what makes one different and how to value those differences (as well as substitutiion effects). NGDP targeting fans (of whom I am one) don't want to the Fed to target 5% nominal gdp growth(/level growth) because nominal gdp growth is good, but because we believe it is the monetary rule most conducive to maintaining full employment and higher long-run (real!) GDP growth. And we think this (in part) because while NGDP is subject to potential mismeasurement it doesn't have the potential for additional model error like inflation does (see Scott Sumner on housing costs in CPI). Does that make sense?

    ReplyDelete
  10. I believe Matt Y. made it perfectly clear in numerous earlier posts that the chief reason he favors a move to targeting spending stability rather than targeting price stability is that the former will allow for a higher rate of inflation. And he thinks a higher rate of inflation is a good thing in the current context.

    Then Krugman endorsed the "communications strategy" of supporting NGDP level targeting as a way of pushing for a higher expected rate of inflation, but without having to use the dread word "inflation."

    http://krugman.blogs.nytimes.com/2011/10/30/a-volcker-moment-indeed-slightly-wonkish/

    Then Scott Sumner decided he wanted to "ban" discussions of inflation from his comments section.

    http://www.themoneyillusion.com/?p=11607

    Now Matt has discovered that inflation is such a conceptually fraught and confusing concept that we should avoid it, in favor of the more easily measurable concept of NGDP.

    This is all politics; not economics. The inflationists don't like using the word "inflation", because its a loser politically, so they are attempting to obfuscate their inflationism under a technocratic alphabet soup to bamboozle the rubes.

    ReplyDelete
  11. "This is all politics; not economics. The inflationists don't like using the word "inflation", because its a loser politically, so they are attempting to obfuscate their inflationism under a technocratic alphabet soup to bamboozle the rubes."

    "Bamboozle" huh? The only people I see agreeing with you about "inflationism" is the Wall Street Journal editorial board, Paul Ryan and various other sundry conservatives. It's either them or ultra-hardcore lefties like Henwood who argue devaluing is the "lazy way out."

    What's your solution? Write letters to Republican Senators to pass Obama's job bill and lower unemployment? Might as well ask them for a pony too.

    ReplyDelete
  12. Might as well ask them for a pony too.

    NGDP targeting is a pony too. Rather it's two ponies: one that might take some of the better-off toward Rainbow Land, and another that will cart the most vulnerable off a cliff.

    If the NGDP targeters could tell a single, clear, causally compelling story about what they are trying to accomplish, and how they are trying to accomplish it, they might be able to convince skeptics like me. But they won't, because they can't. The story changes every day.


    Some days it's about expectations; some days it's not about expectations.

    Some day's it's to increase inflation; the next day inflation is not a meaningful concept.

    Some days it's super-massive asset purchases; some days they are told we don't need more asset purchases.

    Every time some MMer pipes up to say "tomahto", some other MMer comes back the next day to say "tomayto."

    ReplyDelete
  13. "some days it's about expectations; some days it's not about expectations.

    Some day's it's to increase inflation; the next day inflation is not a meaningful concept.

    Some days it's super-massive asset purchases; some days they are told we don't need more asset purchases."

    If you're confused about what proponents are saying I'd suggest rereading Romer's New York Times editorial.

    It was somewhat arrogant of you to proclaim that it was "sad" to see Romer "fall under the sway" of this new "fad."

    Why not argue the issue on the merits and the economics?

    If you demonstrate confusion about what respected economists like Romer are arguing, you'll be less likely to convince people on the fence that your position is correct.

    The best strategy is to state your debate opponent's position at its strongest.

    ReplyDelete
  14. Anonymous5:13 PM

    In Rainbows is the 'actual thing'.

    ReplyDelete
  15. This comment has been removed by the author.

    ReplyDelete
  16. Peter, Romer focused on increased confidence and expectations of future growth, and a temporary climb in both inflation expectations and actual inflation.

    Now Sumner and Yglesias say we shouldn't talk about inflation anymore. So I guess there is nothing about which either I or anyone else should change our "inflation expectations."

    So now you tell me what the "proponents" are arguing. They are all over the map. Which proponents? Sumner? Woolsey? Christenson? Yglesias? Romer? Krugman? Put it all together and distill a clear, straight, causal story from them that goes from the Fed adopting the new target to an improved economy. I submit all you get are vague clouds of suggestions and ambiguity. It's voodoo.

    Of course, if I am allowed to have inflation expectations, then if I become convinced by the Fed's powerful "commitment" to bringing NGDP back to trend, it is open to me to believe to expect almost the entire amount of the increase to be due to a rise in inflation. I don't see how that is supposed to increase my overall economic confidence level. Given the fact that I am not due for another performance review or raise until the end of next year, then if I expect inflation between now and then, I'm going to have to save more, despite the inflation, to accommodate my schedule of fixed expenses to the expected rise in prices.

    ReplyDelete
  17. "I submit all you get are vague clouds of suggestions and ambiguity. It's voodoo."

    I'd submit again that you're confused.

    "Of course, if I am allowed to have inflation expectations, then if I become convinced by the Fed's powerful "commitment" to bringing NGDP back to trend, it is open to me to believe to expect almost the entire amount of the increase to be due to a rise in inflation."

    Why would you do that?

    "I don't see how that is supposed to increase my overall economic confidence level."

    Doesn't matter what you think. It's not all about you. It matters what most people think.

    "Given the fact that I am not due for another performance review or raise until the end of next year, then if I expect inflation between now and then, I'm going to have to save more, despite the inflation, to accommodate my schedule of fixed expenses to the expected rise in prices."

    You're right, 3 percent inflation would be nightmare. 4 percent ... wheelbarrows of cash ... fascism.

    Or if you were unemployed, a growing economy would create job openings which needed to be filled. A tighter labor market would allow people to negotiate wage increases in line with inflation.

    The Fed should do whatever it can to meet the NGDP level target. If it tries many things and fails it should explain why it's failing and call more loudly for fiscal action. I doubt it would fail though. They can always print more money.

    ReplyDelete
  18. Desolation Jones8:49 PM

    Hey, where did you find the Zimbabwe NGDP? I was looking for it a while ago, but I couldn't find. Now if I google it today, this post is the first result to come up.

    ReplyDelete
  19. To get Zimbabwe's 2007 NGDP growth, first find their real GDP growth (somewhere in the neighborhood of -10%) and then adjust for 66,000% inflation. Voila.

    ReplyDelete