29 Nov 2010

How Does Scott Sumner Really Feel About “Income”?

Economics, Federal Reserve, Financial Economics 10 Comments

I was reading Scott Sumner’s blog and was moved to send him this email (with links added):

I was going to do a really smarmy post once I realized this, but I want to give you a chance to defend yourself. Here’s my problem:

(A) You have been arguing that income is a meaningless concept, and that people don’t use it to plan their consumption.

(B) You have been arguing that we are in a recession because national income did not grow quickly enough, and so people’s plans are all screwed up. In fact you want Bernanke to stop talking about inflation and start talking about income.

I’m sure you can see at least the apparent problem.

Bob

With permission, I reprint Scott’s reply:

Same word, completely different concept. When you say “income” you include stuff like capital gains, which is not a part of GDP. I am interested in production, not income, as the term is applied to individuals.

BTW, NGDP isn’t really the optimal monetary policy target, but on the other hand the optimal target (nominal wages) is not a good Fed target for all sorts of pragmatic reasons. So I use NGDP as a good compromise. OK, now fire away!

Scott

I don’t really have much else to say. If I understand him, Scott is saying that firms and households make long-term contracts based on their estimates of what national output will be, but don’t care how much their accountants say their own incomes will be.

If I have indeed correctly understood Scott, I guess all I have to say is, I don’t think that’s a useful foundation for a model on which to advise Bernanke.

10 Responses to “How Does Scott Sumner Really Feel About “Income”?”

  1. Silas Barta says:

    You’ve mentioned another Sumner idea there that I find loopy and have criticized him for (can’t find my past comment at the moment):

    firms and households make long-term contracts based on their estimates of what national output will be

    This will come as news to many businesses. I do not know any firm that bases decisions on NGDP. This has a loose *correlation* with what firms care about, but it is “screened off by”[1] metrics they do care about and can actually measure. For example, car makers don’t care about what national expenditures will be; they care about what expenditures on *cars* will be. Yes, normally they will go together, but it is the *difference* that a car maker must be keenly aware of. They can’t just think about general purchases in deciding whether how much and what to produce.

    (What if a private investigator made business decisions based on NGDP? What about Consulting By RPM?)

    Generally speaking, I find Sumner to commit, in epic proportions, the error of noticing a correlation, and believing that the correlation can be exploited causally to get favorable outcomes. Basically, ignoring Goodhart’s Law, the same fallacy that leads people to believe that, since college education is correlated with individual success, if we make more people go to college, those people will have more success.

    And like the example I gave to you in email, Sumner thinks that, because stock price correlates with and enterprises value, then you can completely ignore other metrics when deciding how much an enterprise is worth. But the enterprise’s stock price only arises out of valuations by people using *other metrics*! Without those metrics, there can’t be a stock price in the first place. *Someone* has to be using a method that goes beyond asking, “gee, what does the market think this asset is worth”?

    That’s (part of) why I think he is in error to think that the historical correlations with NGDP are evidence that NGDP targeting will be more effective than any of the other targets the Fed has tried.

    [1] i.e. rendetered independent of what firms care about when you condition on …

  2. Scott Sumner says:

    Silas, You are missing the point. If each individual cares about the income in his sector, then the aggregate of all individuals and firms care about the aggregate of income in all sectors. When I talk about expectations, I am not talking about the expectations of car companies, but the average level of expectations for the entire economy.

    I am well aware of Goodhart’s law, but I don’t see how it applies to a policy that targets the goal veriable. Goodharts law claims that if you start targeting some intermediate variable, then the relationship between the intermediate variable and the goal variable will break down. But NGDP expectations are my goal variable.

    Stock prices aren’t determined by corporate income, they are determined by corporate expected cash flow.

    Bob, See my response to Silas. Also note that capital gains are unpredictable (efficient markets), so the aggregate forecast of everyone’s individual income (not knowing cap gains) will be close to the aggregate forecast of national income.

    I’ll grant you this; although I still think actual income is a pretty meaningless idea (at the individual level) I agree that expected income has some role in the kind of contracts people sign. But for wage contracts I’d say wage income is better than total income, and for debt contracts I’d say future net cash flow is better than income.

    Obviously there is SOME correlation between income as convetnionally measured and varuous contracts, but there are much better variables than individual incomes, which are a very different concept from the “income” that goes into NGDP.

    • bobmurphy says:

      Scott wrote:

      Also note that capital gains are unpredictable (efficient markets), so the aggregate forecast of everyone’s individual income (not knowing cap gains) will be close to the aggregate forecast of national income.

      This is off the cuff, and I reserve the right to say “oops,” but I think you are way off here, Scott.

      Efficient markets doesn’t mean that the mathematical expectation of capital gains is zero. E.g. back in 1995, if I held a ten-year (no coupon) bond priced at $1000, the expected capital gain over the course of a year would not be 0%.

      If I understand your viewpoint, you are saying a regular person shouldn’t care–at least when deciding whether he can go to Disneyland this month–that his bond portfolio is appreciating or not. (Now it’s true, you have also said that you yourself care more about capital gains than your wages; but that just proves to me you don’t really believe what you are saying about accounting income.)

      So if I’m right, then in an economy with large capital gains, the aggregate of everybody’s accounting income need not come close to GDP, right? If they do in practice, it’s not because of efficient markets, but because it just so happens that GDP is a lot bigger than the capital gains.

      Let me give an exaggerated example: Suppose aliens promise that starting in 2012, they will do something to the Earth’s soil so that it produces 100x as much food per acre. They will also give Exxon–and only Exxon–the technology to produce sweet crude oil from water and salt.

      If everybody believed the aliens, then the value of farmland and Exxon stock would skyrocket, right away. Yet actual GDP couldn’t physically do so, and in fact might even go down, depending on the specifics. (Consumption would go way up, but investment might go down.)

      Yet when it came to people’s plans, they should all certainly take into account the aliens’ promises. It would be incredibly suboptimal for people to base their decisions just on GDP this year.

      (I know I know, you are going to say they should look at the expected future path of GDP. But I’m trying to show you that capital gains take into account that information too.)

      • Silas Barta says:

        That also shows the problem with the GDP metric: it’s not clear that consumption *as measured for GDP* would even go up, because people could just spend their consumption (and investment) dollars on something else rather than increasing them on net.

        Or for an even clearer example, imagine the aliens gave everyone Star Trek replicators. Then, there might be no exchanges involving money, forcing GDP to near zero, despite consumption sustainably skyrocketing!

        (PS: Thanks for the italics cleanup.)

        • bobmurphy says:

          Well “consumption” (C) would plummet, according to Krugman and Sumner. I realize you know this, just clarifying.

          The government would have to run huge deficits in order to boost AD and prevent massive unemployment.

  3. Silas Barta says:

    @Scott_Sumner:

    If each individual cares about the income in his sector, then the aggregate of all individuals and firms care about the aggregate of income in all sectors.

    Only if you want to jump headfirst into a fallacy of composition. You said that businesses make plans based on NGDP, not car-NGDP, shoe-NGDP, etc., and that this is why we need to prop up NGDP — so it will lead carmarkers and shoemakers to form “good” plans. But once you recognize that no firm is basing its plans on NGDP itself, the argument falls apart.

    From the fact that “people will spend more in nominal terms next year”, it does not follow that that any producer will predict the spending will be in *their* sector. And if they did, for some reason, use that metric, but people had become much poorer and were shifting their preferences, (i.e. how it is right now) they would end up with a massive economic miscalculation that compounds the problem (because they would be tooling up for purchases that won’t happen).

    Do I need to dig up the thread where you claimed that carmakers base decisions directly on NGDP expections, or are you ready to disavow that?

    But NGDP expectations are my goal variable.

    Then you’re just shifting the level at which you commit the fallacy. It’s like if I were running a communist country and wanted to show the capitalists that we can make shoes just as well as they can, so I order factories to meet a quota of X shoes. Then someone points out that I haven’t proven anything because the shoes suck and fall apart quickly. Your response is like saying, “so what? Number of shoes produced *is* my goal variable!”

    Well, Scott_Sumner, if NGDP is your goal variable, then it’s a goal no one cares about as an end (i.e. terminal value rather than instrumental). No one cares that NGDP is high if everyone’s miserable. NGDP is only useful to the extent that it actually reflects economic health. But ones you recognize that there can be divergence between NGDP and genuine economic health, it is no defense to say it’s your goal — it’s *not* the goal you should be caring about.

    Stock prices aren’t determined by corporate income, they are determined by corporate expected cash flow.

    Great! Then I made the right call by never arguing what you just rejected, didn’t I?

    I wasn’t defending the position (that Bob_Murphy wasn’t even taking either) that corporate income suffices to determine stock price. My point was that the “buck” cannot “stop” at stock price when asking what an enterprise is worth. Rather, someone, somewhere, must be using metics other than the stock price, which you had deemed the be-all end-all.

    And nor is it the case that there’s one holy grail number that is “the” determinant of the stock price, as you seem to suggest. Numerous factors enter into it. For example, it’s not *just* the expected cash flow that matters, but also future costs, the value of assets, current cash flow, intellectual capital accumulation, etc etc. But there’s the kicker: several of those do rely on the concept of “income” that cannot be ignored by looking at a stock price. If everyone ignored such income calculations in favor of the stock price, there would be no stock price! — or at least, certainly not an informative one.

    So yes, you were in error to suggest the stock price as an alternate metric for company valuation, since that price must come from some other valuation that does *not* reference the stock price, just as Bob_Murphy said.

    (Note: In my earlier agreement with much of your “income is meaningless” post, I don’t think I ever agreed with the specific point that income has *zero* uses, just that the lumping together for tax purposes is stupid, as you claimed. In that respect, I agree with Bob_Murphy’s criticisms here.)

  4. Silas Barta says:

    Oops, the two paragraphs after “But NGDP expectations are my goal variable” aren’t supposed to be in italics — those are my reply, as you probably figured out. A little clean-up help, Bob? 🙂

  5. Scott Sumner says:

    Bob and Silas, Imagine a little old lady in a house in Phoenix, Her social security income is $30,000, and the value of her house falls by $100,000. She doesn’t sell her house. So her income is negative $70,000. My point is that the income figure of negative $70,000 is pretty meaningless. I can’t imagine any decision that would be based on that income figure. Will that determine her consumption this year?

    I’m not saying that her capital loss has zero effect on her consumption, I can imagine her tightening her belt a bit because she is poorer. But adding social security income and capital losses is adding apples and oranges. You guys seem to think her negative $70,000 income means something, but I can’t imagine what it means. What use is it?

    In the alien example, the capital gains would lead to more consumption, but a much smaller increase in consumption that what would result from an increase in salaries.

    I agree that expected capital gains aren’t precisely zero, but the big variations are mostly unexpected.

    Silas, I said the car companies care about NGDP, not that it is all they care about. I’m sure when estimating future car sales the future expected levels of national income play a big role, but other factors such as oil prices also play a role in their forecasts. So I don’t see any contradiction.

    In my NGDP targeting discussion I am saying that firms collective, in aggregate, make decisions based on expected NGDP. It’s no differnet from saying investors care about expectated inflation, even though inflation rates vary from one city to another.

    • bobmurphy says:

      Scott, I think I finally put my finger on what is going on here (i.e. why we keep passing as ships in the night). I mean, besides the fact that I am right and you are wrong. 🙂

      I am not trying to come up with some “necessary and sufficient factors to determine consumption at time T.” Someone needs to know the present value of his capital, in order not just to plan this period’s consumption, but also a path of future consumption.

      So when I say, “Income is how much consumption you could engage in this period, without reducing your capital going into next period,” that’s not to say it’s a *goal*, but rather a benchmark. If I say, “Celsius has absolute zero at -273 degrees, and temperature helps us cook,” it would be weird for you to object, “But Bob, who in the world uses a recipe that calls for -273 degrees?!”

      With your example, the little old lady presumably had a lifetime consumption plan. She figured she was going to live for x more years, consuming y per year, and then leave her estate valued at z to her heirs.

      Oops, now this year her house goes down by $100k, while her SS payment is only $30k. So she reevaluates. She realizes that if she consumes the same amount this year as in her original plan, she is going to start next period with $100k less in wealth than her previous plan called for; eventually something will have to give, down the road–she can’t simply ignore the fact that her house went down. She reoptimizes and probably consumes a decent amount anyway this year, though it will be less than in the original plan. And it will also be less in subsequent years, relative to the original plan.

      In a world of certainty, “income” might not be all that useful, but as I tried to say in my original article, I think it’s extremely useful to help people update their plans in light of surprises.

      It’s true, this period’s income doesn’t have a one-to-one correspondence with this period’s consumption, but (a) neither does any other concept you can offer, and (b) who ever said we needed such a thing?

  6. Silas Barta says:

    @Scott_Sumner: I can’t imagine any decision that would be based on that income figure. Will that determine her consumption this year?

    Many things, but she’ll certainly have to behave differently knowing that, on top of her 30k income, she has lost some of the option value of selling her house. Not taking into account this change in wealth could lead to disastrous decisions.

    Just because other factors can interact with your wage income doesn’t mean it’s a meaningless exercise to look at a household income statement.

    (btw, I don’t know if this matters, but the -70k figure is a red herring, since the government doesn’t consider unrealized capital gains/losses in their definition of income, so there’s nothing for you to complain about there in the tax code.)

    In my NGDP targeting discussion I am saying that firms collective, in aggregate, make decisions based on expected NGDP. It’s no differnet from saying investors care about expectated inflation, even though inflation rates vary from one city to another.

    Then your claims about NGDP targeting don’t follow. If firms care only about a specific kind of NGDP, then they won’t base any action on learning that the Fed is keeping NGDP in toto at 5% growth. Because the benefits of that policy assume that businesses react to this *aggregate* metric, and that assumption is false, the results you expect will not actually happen.

    (Incidentally, sometimes NGDP falls because people don’t want the crap that is offered for sale. At what cost should they be “re-educated”?)