Although the cost of computation has continued to decline rapidly since the 2000s, especially when new innovations are taken into account, rather than computational power, the binding constraint on growth is more likely to lie in organisational constraints and the need for co-invention, write Diane Coyle and Lucy Hampton.
In the search for explanations for slower productivity growth since the mid-2000s in many countries, one possibility is a slower pace of progress in digital technologies. In this paper we show that the cost of computation has continued to decline rapidly, taking into account innovation in chip types and cloud computing. This is a continuation of its long-run trend; the decline has slowed since the mid-2010s, but not earlier. To the extent that the productivity slowdown is linked to technology use, the explanation is more likely to lie in elements of the input bundle other than computational power, such as human and organisational capital.