Chosen by the permanently employed ie equation 6 and

chosen by the permanently employed (i.e., equation (6)) and the permanently
unemployed (i.e., p * = 0). More formally, outsiders maximise
( )
[ ]
) ( ) 1 ( ) ( c U w U e c U
o
− − + τ . The first order
condition is
( ) ( ) [ ]
0 ) ( ) 1 ( *) ( ) 1 ( ' = − − + − − c U w U e p p w U e
op
o
o
τ τ , where p * is given by equation (6). Since
0 >
op
e , then * p p
o
< , i.e., outsiders would choose a lower p than insiders do.
8
Using a lobbying model, Gaston and Nelson (2004) show that unions and employers pressure policy-makers
to ease tax burdens and cap increases in unemployment benefits when wage bargaining is decentralised. When
the risk of unemployment is lower and collective bargaining is more centralised, the demand for publicly-
provided unemployment insurance is higher because workers prefer higher employment risk-higher wage
contracts.
9
However, this effect is weighted by ) ( α θ − . In fact, if active labour market programmes do not appreciably
increase steady state employment, then the tax effect is negligible. Government-funded training programmes in
the United States are commonly found to have negative rates of return (e.g., LaLonde, 1995). Specifically,
participants are no less poor or do not have significantly lower probabilities of unemployment after completion
of the programmes. It seems somewhat paradoxical then, that generally speaking, government-sponsored
training schemes are politically popular. On the other hand, Fougère et al. (2000) find that the training
component of French active labour market programmes has small productivity and employment effects, but that
the programmes do lower wage costs for employers.

9
generosity of unemployment benefits and spending on active labour market programmes.
There are also wage formation effects that are likely to be more pronounced in more
unionised economies. Finally, the labour market programmes themselves are interrelated.
The connection between the policies is not only determined by the government’s budget
constraint, but also through the way in which the labour market policies influence steady state
unemployment rates and hence, worker welfare. In the next section, we seek to uncover the
empirical relationship between the types of labour market policy as well as to investigate the
primary determinants of both active and passive labour market policies.
III. The Empirical Findings
In this section, we use cross-sectional, time-series data for OECD countries to
investigate various measures of openness, political variables and other controls relevant to
analysing the determinants of modern labour market policies.
10
Before proceeding to the
econometric analysis, we provide a brief overview of the times series behaviour of active and
passive labour market policies in OECD economies. The trends are contrasted with the time
series behaviour of measures of economic openness. We also provide a selective survey of
related empirical studies to further motivate our empirical specification and facilitate
comparison.