Does education raise productivity, or just reflect it?

It is clear that education has an important effect on wages paid in the labour market
However it not clear whether this is due to the role that education plays in raising the
productivity of workers (the human capital explanation) or whether education simply reflects the ability of the worker (through a signalling role). In this paper we describe and implement, using a variety of UK datasets, a number of tests from the existing literature for discriminating
between the two explanations. We find little support for signalling ideas in these tests. However, we have severe reservations about these results because our doubts about the power of these tests and the appropriateness of the data. We propose an alternative test, based on the response of some individuals to a change in education incentives offered to other individuals caused by the changes in the minimum school leaving age in the seventies.
Using this idea we find that data in the UK appears to strongly support the human capital explanation.