of regional unemployment. Section 7 evaluates the responses of unemployment to the
actual shocks that occurred during the sample period. Finally, Section 8 concludes.
2 The Chain Reaction Theory (CRT)
An important dimension of the unemployment problem is that employment, wage setting,
and labour force participation decisions are characterised by significant lags, and these lags
interact with one another. The main salient feature of the chain reaction theory (CRT) is
the use of dynamic multi-equation systems to model the structure of the labour market,
and analyse the evolution through time of the unemployment rate. The predictions of
the CRT lie in stark contrast to the unemployment rate predictions of the structuralist
theory8 which estimates single-equation dynamic models.9
In the context of autoregressive multi-equation models, movements in unemployment
can be viewed as "chain reactions" of responses to labour market shocks. The network
of interacting lagged adjustment processes is the propagation mechanism for these chain
reactions and are well documented in the literature.10 For example, firms’ current employ-
ment decisions commonly depend on their past employment on account of costs of hiring,
training, and firing; current wage decisions depend on past wages due to staggered wage
setting; labour force participation decisions depend on the past labour force on account
of costs of entering and exiting from the labour force.11 By identifying the various lagged
adjustment processes, the CRT can explore their interactions and quantify the potential
complementarities or substitutabilities among them.
To illustrate the workings of the CRT consider the following simple system of labour
demand, wage setting, and labour supply equations:12
nt = α1nt-1 - γwt + εtn, (1)
wt = α2wt-1 - δut + εtw , (2)
lt = εlt, (3)
where nt is employment, wt is real wage, and lt is fabour force, the autoregressive para-
meters are ∣α1,α2∣ < 1, γ and δ are positive constants, and the error terms εn, εw, and
εl are strict white noise processes independent of one another. All variables are in logs.
8 Phelps (1994) gives a complete description of the structuralist theory.
9See Karanassou, Sala, and Snower (2006) for a detailed comparison of the chain reaction and struc-
turalist theories.
10 See, for example, Layard and Bean (1989), Lindbeck and Snower (1987), Nickell (1978), and Taylor
(1980).
11Of course, the employment, wage, and labour force adjustment processes may arise for reasons other
than the ones given above.
12 For ease of exposition, and without loss of generality, this illustration ignores constants and explana-
tory variables. In Section 4 we estimate an extended version of this labour market model by including
constants, several explanatory variables and the second lags of the dependent variables.