Following your recent debate, it would be generous to claim that improvements in English National Health Service (NHS) productivity were driven by the rollout of choice and competition. The improvements, in any case, are not dramatic. Year-on-year increases in NHS expenditure meant that input growth increased rapidly during the early 2000. Output growth lagged slightly behind because of restrictions on doctors’ working hours and lead times before new facilities come on stream. Since 2004–2005, inputs and outputs have been growing at a similar rate; we are now getting out of the NHS what we put in.

What explains this improved situation? It is partly due to a slowdown in recruitment and use of agency staff. It is also because the NHS is treating more people and offering better quality care, evidenced by lower waiting times and hospital mortality rates. It is not obvious that choice and competition have driven these improvements. Hospitals were forced to look at their staffing by concerns about deficits; waiting times have fallen primarily because of the ever more demanding target regime, complemented by payment-by-results.

Indeed, choice and competition may have contributed negatively to productivity. Choice has raised expenditure through the implementation of Choose and Book, but does not contribute directly to output (although it might indirectly have led to improved capacity utilisation), and the qualitative benefit of being offered a choice is not easily quantifiable. Independent sector treatment centres may have provided a competitive edge in some localities but were providing less than 1.5% of elective activity in 2007–2008, despite being paid for somewhat more than this amount.

The search for evidence is likely to be frustrating. Both ‘choice’ and ‘competition’ are used to describe a diverse range of policies that tend to be implemented or strengthened along with other policy changes, thereby making it difficult to disentangle their specific contributions. This reality is no bad thing; it is better to have more policy levers than objectives. Rather than relying on one lever at the expense of others, the challenge is to design a package of policies that complement one another. In all likelihood, this package will comprise examples from each of the Trust, Command and Control; Voice; and Choice and Competition Models.