Estimating the Impact of Ecommerce On Productivity Growth

By / 7.31.2017

In an earlier post, I estimated that the expansion of ecommerce since 2007 is saving American households 64 million hours per week in shopping time.  What impact does this have on measured productivity? This is not an easy question. Unpaid shopping hours are part of  “household production,” which is generally excluded from official calculations of GDP. That omission is a problem, because it undervalues the unpaid time that people contribute to their households, ranging from cooking and childcare to commuting.

The best  way to tackle the impact of ecommerce on productivity is to build up a consistent set of national accounts integrating  both the increased importance of data for economic growth (See Moving Beyond the Balance Sheet Economy), as well as shopping as an economic activity requiring both market hours (the retail worker) and nonmarket hours (driving to the store, choosing items from the shelves and so forth).  For our purposes here, though, we are going to do a back of the envelope calculation to estimate the impact of ecommerce on productivity growth.

According to the BLS, the number of hours worked by employed workers in the nonfarm business sector rose by 2.1% from 2007 to 2016, or an increase of 79 million hours of work per week.* Suppose that we adjust for the reduction in household shopping hours, on a 1-to-1 basis.**  Then total weekly hours only rise by 15 million from 2007 tp 2016, or only a 0.4% gain in hours. The annual percentage growth in hours goes down from 0.2% to virtually nothing.

This calculation suggests that factoring in ecommerce could raise measured productivity growth in the nonfarm business sector by 0.2 percentage points from 2007 to 2016. This is a significant difference, but obviously not enough by itself to reverse the slowdown in productivity growth. However, the relative gains in productivity should continue as the market share of ecommerce increases.

This type of productivity analysis, which integrates the impact of the data-driven economy on both market and household production, can be extended to other innovations as well, such as autonomous vehicles and artificial intelligence.


*The BLS reports annual hours in the nonfarm business sector, which I divided by 52 to get weekly hours.

**We could make a case for the adjustment factor to either be higher or lower than 1-to-1. We could value the household hours at the minimum wage, in which case they are less valuable than retail and fulfillment center worker hours. Or we could say that being stuck in traffic driving to the mall or waiting on checkout lines is really annoying, which would make those saved hours more valuable.