Friday, 15 June 2012

Public Sector Employment: Bad for Local Manufacturing, Good for Local Services

For a while now, I have been looking at what happened to local authority employment in the early to mid-2000's to figure out what, if any, impact public sector employment has on private sector employment. We've just published our first set of results from this work and the results are pretty interesting.

When we look at a relatively short period of time (2003-07) growing public sector employment has little impact on overall private sector employment. BUT this doesn't mean public sector employment has no effects. Our estimates suggest that each 100 extra public sector jobs in a local authority 'creates' 50 additional jobs in (non-tradable) services, but 'destroys' 40 jobs in manufacturing. In short, increasing public sector employment is bad for local manufacturing, good for local services.

This difference makes sense. Local restaurants and shops benefit from the spending of public sector workers, local cleaning firms benefit from demand from organisations that employ those workers. What about manufacturing? It's easy to see why it doesn't benefit - neither public sector workers nor organisations buy much from local manufacturing firms. But how to explain the negative effect? There are at least two possible channels. First, increasing local public sector employment pushes up wages and house prices. This is bad news if you are a local manufacturing firm trying to compete with China. Second, higher public sector pay attracts away good administrators, accountants and entrepreneurs who would otherwise work in the private sector. Again, this is bad news for local manufacturing. We aren't able to say which of those channels are at work, but anecdotal evidence suggests that both probably play some role.

This isn't the end of the story, however. Using different data we look at changes over a longer time period (1999-2007). This data isn't quite as good, so the results need to be interpreted more carefully, but the results are interesting. Over the longer time period, those negative effects on manufacturing are even stronger (they lose about 80 jobs for every additional 100 public sector workers). Again, this makes sense - the channels through which manufacturing gets hurt take time to work through. What about local services? Over the longer period, this loss of local manufacturing employment offsets the increase in public sector employment so there's no longer a beneficial effect on local services. In short, over longer time periods public sector employment crowds out private sector employment.

There are, of course, some caveats. In particular, we have to use econometric techniques to allow for the fact that manufacturing has been declining and that public sector employment may be moving to offset this. The details are in the paper, but the effects I talk about above shouldn't be driven by that general decline. Also, the effect of public sector cuts may not exactly mirror that of the public sector increases that occurred during the period of our study. That said, there's no strong argument to think that the overall effect should be different on the way down as opposed to on the way up (even if the timing might differ).

In terms of politics, there is something in this for everyone. Labour can point to the short run effects of public sector job cuts (they will be bad for total employment in affected areas). The coalition can point to crowding out of manufacturing and the longer run crowding out of total private sector employment. As ever, our research can't tell us which of these is more important - this depends on what policy makers want to achieve. But it does give us a feeling for what the short and long run changes might look like and suggests that for areas experiencing large public sector job cuts, things are likely to get worse before they get better.

1 comment:

Anonymous said...

The logic that 1) higher rates of local employment drive-up wages, that 2) organisations compete for labour, and 3) a small proportion of manufactured goods are sold locally dictates that employment growth in all non-manufacturing sectors will ‘destroy’ manufacturing jobs. But surely the strength of any such causal relationship depends more on the nature of the sector ‘destroying’ the manufacturing jobs than whether the jobs are in the public or private sector.

Given manufacturing would suffer catastrophically at a national scale if the domestic consumer market was reduced only to people who worked for organisations that did not compete with manufacturers for labour, your paper reads like a backhanded argument for not locating moveable public sector jobs in areas of proportionately high manufacturing employment.

Yet given the falling demand for labour experienced in most parts of the UK since 2007, the degree to which organisations ‘crowd one another out’ should have reduced substantially in areas where relatively high proportions of the workforce are engaged in manufacturing; areas which typically have experienced substantially rises in unemployment. Meanwhile, the reliance of non-tradable services on the rump of a shrinking workforce in these places continues to grow. The net effect should be that in manufacturing towns more private sector jobs are now created for each local public sector job than your paper estimates.

In London, where a tight labour market still exists and public and para-public sector employment has grown during the downturn, the converse is likely occurring (though the effect will mostly be felt by tradable service producers rather than tradable goods producers). This adds to the inefficiencies caused by high and spiralling London housing costs diverting personal spending of workers away from more productive parts of the economy. In the case of the public sector, a strong efficiency argument thus exists for dispersing discretionary jobs away from the capital, especially given that, even if the public sector were privatised, there is no evidence to suggest the crowding-out effects on any other sectors, including manufacturing, would be reduced.