Wage Setting and Unemployment

Steinar Holden, August 2005

Wage setting and unemployment

Characteristics of wage setting in many OECD countries

· widespread collective bargaining

· large variation in union density

· extension laws lead to high bargaining coverage

· widespread employment protection legislation (EPL), although with large variation across countries

· Wages are set in bargaining, or

· unilaterally by the employer, possibly due to weak bargaining power of employee or union

o efficiency wage or monopsony

· Not appropriate to view the wage setting as the result of “the invisible hand”, ensuring equilibrium between supply and demand

· Not clear that it is appropriate to view wages as set by worker/households, as in some recent literature

Theoretical framework

Imperfectively competitive firms, facing a downward sloping demand curve,

Production function firm i (employment N only input)

Yi = F(Ni), F’ > 0, F’’ < 0

Demand (Qi is the producer price, Q the aggregate output price level, E is the elasticity of demand, Y aggregate demand)

Profits in firm i (Wi is nominal wage, τ is payroll tax rate)

πi = QiF(Ni) – Wi(1+τ)Ni

Right-to-manage assumption:

Firms set prices (and employment) to maximize profits, for given wages

Prices set as markup

over marginal cost

product market comp. up => E up => markup down

Leads to profit maximising labour demand

and indirect profit function

From firms’ price setting, and aggregating over firms

· One-to-one correspondence between employment and real wage, with no direct effect of aggregate demand, Y, for given real wage.

More elastic product demand, E up, (could be interpreted as greater product market competition) leads to higher employment for given real wage.
Union utility

P is consumer prices

Wages set in firm-specific bargain,

Nash bargaining solution

where

· π0 and u0(U, Z) are the disagreement points, u01 < 0, u02> 0,

· U is unemployment and

· Z is an indicator for wage pressure variables, like unemployment benefits, “union strength”, industrial conflict legislation, etc.

· disagreement points depend on what motivates bargainers to reach agreement, costly delay or risk of breakdown (Binmore et al, Rand J. 1986)
The first order condition of the Nash bargain

The solution can be written on the form

producer real wage

= W(wedge, U, wage pressure, prod. market comp)

+ - + -

(where the effect of aggregate output Y is subsumed under U, invoking Okun’s law)

Empirical evidence:

· aggregate times series

· evidence based on micro data

The Wage Curve (Blanchflower and Oswald)

· Evidence for many industrialised countries suggest an “empirical law”, that the elasticity of real wages wrt unemployment is about -.1

· If unemployment doubles, real wages fall by 10 percent

· Empirical support for dynamic fluctuations around a long-run wage curve, rather than a Phillips curve

· Wage curve does not require unions and bargaining

· Theoretical derivation under efficiency wages

· Evidence suggest stronger effect of unemployment in non-union setting

· Equilibrium unemployment Un given by wage and price setting behaviour

· Increased wage pressure, Z up,

=>wage curve shifts up => W/Q up and Un up

· More elastic product demand, E up

=> price curve shifts up => W/Q up and Un down

Wages and prices set in nominal terms, based on expectations

· U < Un possible if wage and price growth is above the rates expected in the price and wage setting

· Increasing wage and price growth for U < Un

· But: wage and price growth depend on more than labour market pressure

· If W(1+τ)/Q above the wage curve, wage growth may increase even if U > Un

· No direct effect of aggregate demand Y on Un.

· Price level adapts to make aggregate demand equal to aggregate supply, given by Un, the production function and labour supply

· However, in open economy:

Y up => Q up => P/Q down

=> wage curve down => Un down

In the long run, the price curve becomes flatter, in the limit horizontal, fixing the producer real wage, due to requirement of

o expected rate of return to capital

o balanced foreign trade

· Individually, workers and unions may still gain from increased bargaining power

· Collectively (nationally) higher wage pressure gives only rise to higher unemployment, with no long run impact on real wages

o negative external effects in wage setting

§ on consumer and input prices,

§ on unemployment

§ public budget, and

§ from envy/status

· Yields strong incentives to incomes policy and coordinated wage restraint

European unemployment – the aggregate evidence

Large variation over time and across countries

Econometric evidence based on panel data

· Interaction of stable institutions and shocks/baseline variables which shift over time

o Layard et al (1991) (increased wage pressure from 1970, the replacement ratio, real import price, etc)

o Ball (NBER, 1996) (monetary policy, institutions and hysteresis)

o Blanchard and Wolfers (EJ, 2000) (TFP growth, the real interest rate, labour demand shifts, etc)

· Changing institutions

o Belot and van Ours (JJIE, 2001) (interaction between institutions)

o Nickell, Nunziata and Ochel (EJ, 2005)

Studies agree that institutions are important, but vary on additional variables and interactions
Nickell, Nunziata and Ochel (EJ, 2005)

20 OECD countries, 1961 - 1995

· Higher unemployment in country-years with

o strict EPL (not significant)

o high benefit replacement ratio

o long benefit duration

o interaction of benefit duration with repl. ratio

o increased union density

o low coordination in wage setting

o low interaction of coordination and union density

o high total employment tax rate

o low interaction of coordination and empl. tax rate

o high owner occupied housing (not significant)

o labour demand shock

o TFP shock

o money supply shock (not significant)

o real interest rate (not significant)

· Based on dynamic simulations, Nickell et al find that changing institutions can explain 55% of the 6.8 percentage point increase in European unemployment from 1960s to 1990-95.

o changing benefits can explain 39%,

o increased labour taxes 26%,

o shift in union variables 19%, and

o movements in EPL 16%

· Nickell et al argue that model with institutions and shocks make no real contribution to understanding the change in unemployment as compared to the model with changing institutions.

Critique of the institutional explanation

Baker, Glyn, Howell and Schmidt (2004),

Freeman (2005)

Evidence based on institutional variables fragile due to

· uncertain classification of institutional variables

· “economic Darwinism”, where measures are constructed ex-post of researchers who were not unaware of unemployment developments

· coefficient estimates unstable, often insignificant or with “wrong” sign

· no apparent bilateral relationships (consistent with evidence in OECD, 2004)

o bilateral relationships in one decade often disappear or are reversed in the next

Specifically on Nickell et al (2005)

· regressions include country-specific time trends, and for some countries, the estimated time trends implies that unemployment would have been negative unless institutions had changed

· coefficient estimates too large to be plausible (e.g. one unit increase in coordination reduces unemployment by about 6 – 7 percentage points, depending on interaction variables)

· annual data gives the impression of high precision that is not warranted, given the slow and uncertain changes in institutions

· EPL not significant

Calibrated models of stylised European and US economies (Ljungqvist and Sargent, JEP 1998)

· general equilibrium search model with stochastic layoffs, accumulation and depreciation of skills

· workers choose search intensity, job acceptance, and quitting behaviour

· Welfare states with high unemployment benefits work well under low economic turbulence, where the availability of “good jobs” counteract generous unemployment benefits

· Under high economic turbulence, high benefits reduce job acceptance for laid-off workers, leading to long-term unemployment and skill deterioration

· Explains low European unemployment in the 1950s and 60s, high unemployment as of the 1970s, and long duration of unemployment in Europe

· But not clear that turbulence has increased (e.g. no increase in standard deviation of annual industry employment changes), nor that the model can explain huge variation between European countries