Doing stuff for free

As I occasionally muse on Twitter, Diane Coyle’s stuff is fascinating and wonderful. If you’ve read her GDP book or any of her papers (and you ought to) I’ve nothing really to add.

But I have made a cut-out-and-keep picture of my understanding of some of the issues raised in trying to capture changes to GDP and productivity (with apologies to Diane). This is for me rather than you. It’s a bit messy, but here it is. And I like it enough to share, because prosthletise as I do, not everyone is familiar with the basic outline of these ideas.

If you’ve read this far, I expect you have one of 3 thoughts: 1) I already know this; 2) you’ve taken something simple and made it look complex, kthxbye; 3) can you explain? The rest of this blog helps people with thought number 3.

The big black line is the Production Boundary: a threshold between stuff that happens when you Do Stuff For Money (ie, captured in GDP), and when you Do Stuff For Free (ie, everything else).

Breaking the diagram into smaller pieces, let’s start with the up arrows:

When you find a new and better way to Do Stuff For Money (DSFM), this tends to be good for GDP growth and is good for the growth of GDP/ hour worked (generally thought of as productivity growth). On the diagram, this is #1 arrow on the left and captures all sorts of things from the invention of the harvester-reaper, to the roll-out of basic management training. Many if not most of the things we *all* agree about as boosting productivity do so through this arrow.

When you find a new and better way to Do Stuff For Free (DSFF), this doesn’t *directly* touch GDP growth or productivity growth. On the diagram, this is the #2 arrow on the right and captures things like the first order effects of the invention of Wikipedia or the computer language Python, learning how to cook better food, writing a blog etc. This saves us time and/ or is an improvement on what came before. It is good stuff that increases what economists would call the consumer surplus, but won’t deliver a first order boost to GDP.

There are a lot of other lines on this chart. All the other lines cross the black Production Boundary line.

When a red line crosses the production boundary from right to left it boosts GDP. When a red line goes from left to right it reduces GDP.

Famously, marrying your housekeeper reduces GDP as they pass from DSFM to DSFF (#3), even if the stuff they do doesn’t change (although in fairness it might). Divorcing your spouse and paying them to do stuff they would otherwise do for free (#4) strikes me as less conventional, but would deliver a GDP boost. What either means for aggregate GDP/ hour I guess depends upon how much you pay them per hour (as a monetary measure of their value) versus how the output per hour of the rest of the population is valued.

And now we come to some interesting bits.

Inventing and manufacturing something like a washing machine that saves us time and improves our non-market lives boosts GDP and consumer surplus (so arrow #5 is upward sloping). That is, if people generally do their own washing rather than pay someone else to so do. And so the activity of washing clothes shifts from the DSFF to the DSFM side of the Production Boundary.

And there have been some innovations like the self-scan checkout that move tasks from the DSFM side of the Production Boundary to the DSFF side (#6). Here, firms boost output/ profit per employee by substituting a combination of capital and customer labour for their own low skill labour. Unlike the washing machine, this is a slight hit to consumer surplus, getting shoppers to do the work that firms had previously paid staff to do (although maybe reduced queue-times make up for the hit; also, maybe some people enjoy scanning). So the end result is a slight decrease in consumer surplus and higher firm profit.

Online travel agents – like self-scan check outs – boost measured output per employee by shifting some of the work to the customer. But I reckon that the proposition is a cheaper outcome for the customer (#7). So GDP takes a first order hit as offline travel agencies shrink.

And lastly there is the case that Tim Harford wrote up about Microsoft Office. Tim’s mischievously makes the case that having office workers all writing their own PowerPoint presentations, doing their own expenses, etc, rather than what they are presumably experts in doing is an insult to the principle of division of labour, and a negative drag on productivity. I’ve popped this one in at #8 sloping down from DSFF to DSFM rather than just being an arrow of questionable direction sitting on the left hand of the production boundary. The logic is built on Tim’s blog: that having to do stuff in which you are not expert but which is not very taxing (ie, make bad PowerPoint presentations) may be a welcome relief from your actual work. It might even be thought of as a pretty rubbish, but fully-paid, work break. I don’t think I agree with Tim, but I 1) can’t think of another good example off the top of my head, 2) am keen to get symmetry in the diagram, 3) love Tim’s stuff.

Who cares?

Well, I think everyone actually cares about each of GDP, GDP/ hour and general welfare. But I also think that different people are incentivised to really care a lot about one angle more than another.

Government treasury departments care a lot about raising and spending money. For this, GDP is the most immediate concern. It is hard to tax stuff that is neither money flow or money stock, and GDP is money flow. Wikipedia may be awesome, but it is hard to get a cut of the welfare they deliver in order to pay nurses’ and teachers’ salaries. Things that boost GDP in some taxable way are good for governments that seek to resource spending decisions with tax.

Financial analysts and investors care about revenue, profitability, and spare capacity. For this, GDP is important, but output per employee is – I reckon – even more so. Changes in output per hour inform at the macro level the amount of spare capacity there is in an economy, and a boost in output/ hour can boost on a sustainable basis the ultimate level of GDP in a non-inflationary manner (keeping central banks away from the brake pedal). Output/ hour is also important for understanding how corporate profit margins will change. For example, if firms automate, they can produce more revenue per employee; if the cost of automation is not high this will boost profitability, helping boost firm values. And firms and investors will care a lot if some innovation (that they don’t own) shifts stuff across the production boundary. It could maybe wipe out an industry’s viability.

Citizens care about welfare improvement. We shouldn’t obsess over GDP growth if it is associated with reduced consumer surplus. That said, the correlation between measured GDP per capita and different measures of welfare is strong (maybe related to governments’ ability to intervene to raise outcomes being contingent on taxing GDP, don’t know). Have a follow of the brilliant @MaxCRoser for a plethora of great charts, many of which show that places with plenty of GDP tend to get better health outcomes. Like this chart of maternal mortality and GDP/ capita, reproduced here from

There is sooo much more to write on this, but frankly, I am poorly qualified to write it and Diane Coyle’s stuff reads much better. Also, check out the Bean Review.


2 thoughts on “Doing stuff for free

  1. I have read too D Coyle’s book and I think that like your discussion here it is quite interesting, but also largely irrelevant and in some ways wholly misguided, because you are mistaking GDP for GDI, which is something completely different, and that “mistaking” is usually highly politically motivated.

    GDP as the name says it was and should be defined is a list of physical quantities, from tons of steel to hours of medical consultation and km/tons of freight transported. That’s an unwieldly list, but it does represent the productive capacity of a political economy (gross of depreciation). It is very important that GDP should not be multiplied by a price vector to create a GDP index, because it makes production not comparable across time, because of the incommensurable price vectors involved. Instead it is very useful to compare physical quantities produced across time, because they do reveal much about the actual production process.

    GDI instead is the sums of all incomes received by domestic participants, and only somewhat related to GDP, in part because income does not come only from production, in part because where income and production are related, they are related by price vectors that are incommensurable across time.

    The “misunderstanding” of GDP, that is the confusion with GDI, is rooted in the misdescribing of GDP used for military planning purposes in WW2 as the indicator that military planners used to understand the ability of a country to *finance* (total) war: that’s quite absurd because:

    * Military planners look at GDP, physical quantities, to understand the ability of a country to *sustain* (total) war in the physical sense, the cost be damned. What matters i how many planes/tanks/ration boxes can be produced, not tax levels etc.

    * The ability of a country to *finance* (total) war is not related to GDI, its income, or taxes on it, because military planners will commandeer whatever they can, the cost be damned; but to the ability of the country to spend gold, because they cannot commandeer imports, and no exporter will accept the currency of a country that might lose a war.

    In 1941 the UK had to surrender (to Roosevelt instead of Hitler, fortunately) because while the productive capacity of the domestic economy was largely unchanged, the english gold reserves had run out and finacing imports was no longer possible. That had little to nothing to do with GDI (or GDP, as the UK’s gold production is negligible). Historian C Ponting writes:

    “In August 1940, with Luftwaffe bombs dropping on the nearby streets, the Churchill government met to discuss the fact that Britain might soon lose the war. This had nothing to do with the planes circling overhead or the lack of any serious means of taking the war to Germany; but with the amount of gold left in the vaults of the Bank of England. By commandeering the gold they were supposedly protecting on behalf of European governments in exile, and by carefully regulating government purchases, the British government might, just, remain solvent until November (a proposal to force the British people to hand in their gold jewelry was turned down because the economic benefit would be trivial in comparison to the negative effect on morale)”

    GDI instead matters as to financing *peacetime* government debt, because it relates indeed to the fiscal capacity of a political economy over years of peace.


  2. “the correlation between measured GDP per capita and different measures of welfare is strong”

    That “measured GDP” sound “optimistic” because a big chunk of GDP is not measured at all but is “estimated”, or is “hedonically” improved.
    If that instead is translated to “gross domestic *income* per capita [except where there is high inequality] and different measures of welfare” that does not sound incredibly surprising: higher income drives better living standards…


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s