This government, alongside most of its predecessors, is concerned about social mobility. A society in which one's prospects are largely or wholly determined by chance of birth is not one with which many will feel comfortable. But any strategy to increase social mobility must be long term, multi-faceted and cautious in its claims.

As the coalition government prepares to launch its own strategy for tackling social mobility, recent work at IFS exploring the literature on social mobility has highlighted some important conclusions that the government would be wise to bear in mind.

First, countries with higher income inequality tend to have lower social mobility (at least when using income-based definitions of mobility). In an unequal society there is further to travel to get from the bottom to the middle or the top. The UK has relatively high income inequality and low social mobility. It is therefore likely to be very hard to increase social mobility without tackling inequality.

Particularly in a context of high levels of inequality such as that in the UK it is important to be clear what one is trying to achieve through increased social mobility. It is obvious that pursuing relative social mobility implies downward mobility for individuals from rich/middle income families. In a world in which the consequences of downward social mobility are significant, there will be many who find this mobility very uncomfortable.

It also matters whether the government is more concerned about improving the mobility of the most disadvantaged or those somewhat further up the social spectrum. Policies aimed at improving the mobility of the most disadvantaged or the least skilled can be very costly. In part this is because the UK labour market appears to be "hollowing out", by which we mean there are increasing numbers of high skill and low skill jobs, and fewer in the middle. So it may be harder and more costly to help those at the very bottom than it will be to help those somewhat above the bottom. Any comprehensive social mobility strategy is likely to want to deal with both of these groups and may need to treat them quite differently.

One set of interventions which we know are important are those aimed at very young children, as the recent Field Review and Allen Review have highlighted. But it is equally important to understand that they will never be enough by themselves. The evidence is clear that early investments are most productive if they are followed up with later investments. Important findings in this area are that:

• Continuing to increase the supply of graduates and highly skilled workers has the potential to reduce wage inequality (or at least slow down increases) and therefore help (in relative terms) those at the bottom.

• Cognitive skills are highly valued in the labour market, and basic skills such as literacy and numeracy have higher economic returns in the UK than in many other countries. But effective interventions in adulthood that improve cognitive skills are not easily found.

• There is emerging evidence that later inventions targeted at improving non cognitive skills (such as time management, teamwork, leadership skills, self-awareness and even self-control) may be more effective. Certainly there is clear evidence that such non cognitive skills are highly valued in the labour market.

• Finally, interventions that change students' decisions at key points (e.g. the decision about whether to stay in full-time education beyond age 16), rather than their skills directly, could still have a positive impact on education outcomes and hence social mobility. These will be most productive where they also increase subsequent educational attainment.