I’ve been mulling over matters statistical – which if turnout at this week’s ONS International Economic Statistics conference was anything to go by are becoming increasingly popular – and recently have been focused on the ‘production boundary’. This divides the activities included in GDP from those that are not. The theory is that GDP just includes monetary transactions at their exchange value. The practice is different: non-marketed government activities are included, and so is ‘imputed rent’, the hypothetical amount owner-occupiers pay themselves to live in their homes. The argument is that as people switch between renting anfd owning, and as countries differ in the proportion of owner-occupiers, it would distort comparisons over time or across countries to leave it out.

The consequence is that in the main figures we use to assess the health of the economy, businesses and governments are considered productive and households are not. Feminists have long disagreed, as women have mainly done household work. So indeed did Simon Kuznets. Today I came across (courtesy of this post) this quotation from his 1947 article in the Journal of Economic History, ‘Measurement of Economic Growth’:

[A]t least three major institutions are to be distinguished: the family, the business enterprise, and the state. Unless a measure of total output is to reflect the growth of a given institution alone, it obviously should include all economic production within the family, the business enterprise, and the state. Yet most measures of national income note only market-bound output, including almost all state production but omitting large portions of productive activity which, not being market-bound and forming an integral part of family life, are not considered properly economic. There is a definite choice here between totals more comprehensive but less homogeneous and those less comprehensive but more homogeneous.

However unimportant this difficulty may appear for short-term studies, in the long periods implied in measuring economic growth the problem is of too large a magnitude to be dismissed easily. Such long periods are characterized by important shifts in the weights of these different institutions, and reducing the scope of measurement will necessarily produce a substantial bias. Of the quantitatively impressive growth of total output in this country, as measured in the ordinary estimates of national income, a large part is to be associated with the extension of the business at the expense of the family sector. Consequently, one important prerequisite for a more efficient measurement of economic growth lies in the inclusion of such sectors of production that easily escape the statistical eye. As specific examples we may cite the capital formation involved in the work of American farmers in bringing virgin land into cultivation, or the work within the old- fashioned large family, so much of which has been taken over in recent decades by business firms.

It is pretty clear to me that – in addition to the steady shift toward households purchasing ever-more services previously produced at home (from child care to ready meals and restaurants) – digital activities are blurring this boundary. We’re undertaking online more work that used to be marketed, for example acting as our own travel agents and bank tellers – and for that matter supermarket cashiers as more stores introduce the terrible machines that will increase their measured productivity by substituting household labour for paid labour. We’re producing digital public goods such as open source software, which create their own difficulties for the national accounts. We’re using household assets like cars and spare rooms to earn money in the “sharing” economy.

Anyway, this is what I said at the conference this week – there’s a paper in preparation, which I’ve been working on in my role as an ONS Fellow. Not surprisingly, this was controversial with some of the national accountants in the audience!