GDP is important, but it’s not that important

Given that I’m writing a dissertation on the history of national income accounting, I hate to say this, but… GDP just isn’t as important as some people want to make it out to be. Some of the worst offenders in this genre of claim seem to be, unsurprisingly, GDP’s biggest critics. Let’s take an example from an op-ed in this week’s New York Times*, Our Mismeasured Economy by Lew Daly at Demos. The editorial follows in a nearly 100-year old tradition of criticizing how national income statistics handle hard to measure, non-market production, in this case government output. Daly argues, sensibly enough, that the way we handle government is ad hoc and arbitrarily rules out the possibility that government could actually add value (we explicitly assume that the value of government output is equal to what we pay for it, no more, no less).

That’s all well and good, but what bothers me is the over-the-top way in which Daly motivates his critique. Here’s the opening line:

Today’s polarized debates about the role of government often boil down to a single issue: the size of government compared with the size of the overall economy, as measured in gross domestic product.

Really? Are we following the same debates? Because although I’ve certainly seen reference to the size of government (and in fact, we see examples of these claims as far back as the early 1930s), they do not seem to me to be a dominant mode of debate at the present juncture. To be fair to Daly, I have not done a systematic content analysis of contemporary ‘debates about the role of government’, but I would be shocked if even a small percentage of these debates (5%?) explicitly or implicitly referenced the size of government as measured in the national accounts. And far fewer “boil down to a single issue” in those terms. Think, for example, of the recent Hobby Lobby case, and other debates around the Affordable Care Act. This debate is about ‘the role of the government’, but it’s not about the ‘size’ of the government but about its intrusiveness: can the government mandate that private companies provide certain kinds of care to their workers? Think also of the NSA wiretapping scandals. Again, the proper role of the government is at the center of the debate, but not the government’s size as a percentage of GDP.

Daly’s op-ed makes a number of sensible points about what we miss if we focus our debate on the productivity of government on the national accounts (though I’m not sure I agree that the fix is to change how we measure GDP as opposed to, say, coming up with alternative measurements of government productivity and restricting our analysis of GDP to where such a number makes the most sense – GDP is built on a bedrock of “market epistemology”**, and I doubt it will ever move far from that principle). But there’s no need to tee those claims up with an overblown one about the centrality of GDP to contemporary political debates about the role of the government.*** GDP is important because of its diffuse implications for how we think about the world, as well as some more narrow technical uses (such as the World Bank’s categorization of “least developed” countries, see, e.g, Jerven’s work) – but it’s not quite so woven into technical systems as, say, inflation statistics, which directly determine wage increases and social security benefits. And so I get that it’s a bit tougher to talk about why getting GDP ‘right’ is so important. But maybe that means we should be having a different debate, about what the right ways to measure and think about the productivity of government are, rather than a narrow technical one about GDP. Somehow I doubt that the Tea Party is going to stop complaining about the Affordable Care Act if the government’s share of GDP goes down a point.

* H/T to Beth Berman for sending this piece along.
** I define market epistemology as the belief that markets provide the best or only definitive information about economic value. Market epistemology shapes debates about the production boundary and in turn the boundary of the economy – that is, it shapes what we decide to count and how we decide to count it, especially for difficult cases like unpaid housework, government output, and owner-occupied housing. See, e.g., chapter 4 of the dissertation I should be writing instead of this blog post!
*** At this point, I’d also like to fully embrace the irony of using a single editorial as a case to motivate a more general argument about the perils of motivating a general argument about discourse from a handful of cases. I can dig up more if you’d really like, I read this stuff for a living.



  1. I always use GDP portions to make these points:

    (1) Finance doesn’t matter; everyone’s fears are overblown. It’s a tiny percentage of GDP.
    (2) Government exploded by an order of magnitude between about 1890 and 1940 (I think). What did we get for that tradeoff?
    (3) The majority of GDP is paid to workers.
    (4) The majority of GDP is created by small and medium sized businesses, not multinational corporations.

    Those are common talking points among libertarian academics, I understand, and you probably know what we’re inferring regarding each one so I won’t hammer.

    But I’m wondering if you’ve looked at the history of how those points have gotten used as levers in debates over how we conceptualize the economy *once we’ve agreed where the boundary of it is,* a boundary that it sounds like you’re doing an apt job of highlighting.

    • A few thoughts.
      (1) I actually think this is a case where GDP is misleading. Greta Krippner coves this well in her 2005 article, and the updated version in her book. Finance is relatively small as a percentage of the labor force. It’s a bit bigger as a percentage of GDP, but still small. But as a percent of profits, it’s enormously outsized. And for various reasons she discusses, that’s incredibly important. So, financialization is defined as a mode of accumulation, a shift in how profits are made, towards finance, not a shift in the proportion of GDP.
      (2) There’s a messy and interesting history to the measurement of government in the national accounts that bears on this question (and is going to be part of my dissertation, chapter 4, I think). The current way we count government – and the way we’ve done so since about 1947 – treats all government expenditure as final product (not counting straight up transfer payments, IIRC). So, if gov’t spend $1 trillion, gov’t counts as $1 trillion towards GDP. Effectively, government is treated as a final consumer. And that implies the kind of logic you are using – that government ‘eats up’ a portion of the total production pie. But, as economists have noted since before GDP was called GDP, much of government product is actually intermediate production. Roads are used by consumers, but also businesses to move around goods. Courts are used to resolve disputes between businesses, not just individuals. And so on. So, it’s actually quite illogical to count all government expenditure as being final product (like cars); some should count as intermediate production (like the steel that goes into the cars) and thus gov’t as measured in the NIPA is probably ‘bigger’ than it should be. Kuznets himself made such adjustments in the 1930s and 1940s, but lost a debate with his successors at Commerce about institutionalizing the practice. I can tell you more about this, or send you the chapter in a few months…
      (3) Unobjectionably true, but trends are probably more meaningful here. Labor’s share has been declining steadily (contra the first ever stylized fact from Kaldor which observed a constant labor share).
      (4) Probably true (I haven’t looked lately), though between outsourcing and the difficulties of accounting for the productivity of multinationals, it’s hard to know what to make of this. Given Jerry Davis’s death of the corporation, “large” vs. “small” companies may not mean what they used to mean. For a fun insight into the problems of accounting for multinationals, check out Puerto Rico’s GDP (the weirdness being that all of the value of viagra accrued to Puerto Rico because it was manufactured there!).

      So, I’m not sure if that answers your question, but I guess what I’d say is, exactly how those boundaries are drawn is consequential for how those debates play out.

      • Financialization is not a thing in the way that it is mostly received, as confirmation of the centralizing tendency of wealth accumulation. As people substitute toward liquid assets, returns to them diminish. They substitute back toward other things. Finance has expanded some; it’s not a big deal. Profits are an accounting fiction. What matters is the real consumption opportunities people enjoy, and those continue to explode across the income distribution.

        Interesting point on (2). Actually Paul Pieper at UIC made a similar point in his dissertation at Northwestern, but I forgot the details. The conclusion was somewhat opposite, if I remember right. I think I’ll email him about it and send the point along to you.

        Labors share has not been declining. Labor’s income increases have come in the form of nonpecuniary and in-kind benefits that don’t counted as wages. There is actually a sociological point somewhere in there about how the epistemic community decides what is an isn’t a compensatory wage. This is an important point because of how this alleged divergence is being thrown around by the inequality people. It is, of course, to economists, a straightforward distinction to economists — whatever workers take home (gleaned shop nails or not) that contributes to their stream of utility is part of their wage rate. There is a giant divergence between the way bureaucrats and public politicos are interpreting these figures and the way economists do.

        More evidence that Callon et al. have it all wrong when they claim that the economy is cultural construction dictated from Harvard econ. (the gender people are better on this — social constructions emerge in interaction (through male and femaleness in the household etc), rather than coming down from on high, some group of experts)

        The size of businesses follows the same kind of zipf/power law/etc that the size of any other income, city population, network agglomeration distribution does. The idea that large businesses lead the economy in terms of wresting control with profits (krippner above on finance), or by setting social norms that the rest of everyone else follows along, is wrong. There would be precisely zero innovation in such an economy, and was for over 9,000 of settled agriculture before egalitarianism and tolerance took hold via liberal ethics.

        • Re: Labor share – it’s still declining even if you look at total compensation and not just wage compensation (though not as fast). See, e.g., this post by Jared Bernstein.

          • You mean the top line in that Excel graph that says “compensation share?” I see a constant trend with stochastic variation over the long term. Looks like the trend got a bump when Unions got stronger in the 60s/70s, and that has been slowly whittled away as they’ve declined. That’s good for unskilled black teenagers.

          • If you want to talk about misleading statistics and unskilled black teenagers, check out Becky Pettit’s Invisible Men (excellent review by Kieran here).

  2. Yessir, we ought to end the war on drugs tomorrow. It really breaks my heart that people are mad at Facebook and worried about exploitation while one of the most obvious instances of the violent oppression of minorities continues under the very same premise that runs up the backbone of the corporation scare: protecting people.

  3. justaguy

     /  August 10, 2014

    As someone who does research on China, I would point out that Chinese state economic development policy is an area in which GDP is incredibly influential. Chinese state development policies measure “economic growth” in terms of GDP, and incentivize officials to maximize GDP growth by using it as a central criteria for their job evaluations. During the reform era, the Chinese state has given local governments more administrative leeway in terms of how they enforce regulations, and made them more reliant on local tax revenues for their administrative budgets. The naive assumption is that the sum total of lots of local administrative units maximizing GDP growth will be an overall healthy economy. In practice, this leads local officials to overlook environmental regulations, land ownership, labor laws, etc. in order to maximize local GDP and local tax revenues.

    So, various trends that the Chinese state recognizes as obstacles to long term economic growth – such as pollution and economic inequality – are created by policies enacted to maximize GDP.

    • Thanks! That’s a great example, and one that I think also showcases how GDP is *not* used in most of the world. So, yes, I think there’s evidence that in China, GDP is actually more of a statistical target, and not just an indicator. We can imagine a spectrum of target to indicator, with, say, core CPI used for inflation targeting on one end, and the Case-Shiller index on the other. In China, GDP is closer to a target, here in the US and most of the rest of the world, it’s not (though it’s still important, as I tried to say!). I’ve read a few things on regional GDP targeting (and associated allegations of statistical manipulation) in China but I’d love any further recommendations you have for good (English-language) sources on the topic!

%d bloggers like this: