Upload
lemien
View
216
Download
1
Embed Size (px)
Citation preview
Welfare costs of business fluctuationsGali, Gertler, López-Salido - ReStat 2007
What are the costs of business fluctuations?
Lucas(1987, 2003) Representative agent framework, no
unemployment consider costs of fluctuations in consumption,
without taking source of fluctuation into account
show that costs of fluctuations are very small
Here:
Analyze costs associated with fluctuations in consumption and employment (i.e. leisure)
Recessions:Loss of output and consumption ( low output and consumption and high leisure, households would want to work more)
1
Theory-based measure of the variations in aggregate economic efficiency:
The gap between the marginal product of labor (mpn) and households consumption/leisure tradeoff (mrs)
Use representative agent – no unemployment, only variation in hours.
Neglect costs of efficient fluctuations (neglect that consumption variability is costly even if it is caused by efficient fluctuations driven by technology shocks)
Neglect costs of distortions of relative prices and wages (due to price and wage stickiness)
2
Inefficiency gap gapt = mrst - mpnt
Employment is too low relative to the efficient level if the marginal product of labour (mpn) is greater than the marginal disutility of working, measured in consumption units (mrs)
3
FirmsAssume standard production function function (constant elasticity of output with respect to hours)
Y = ANα
MPN=dYdN
=Aα Nα−1=α YN
Lower case denotes logs:mpnt = log(α) + yt - nt
mpnt = yt - nt (neglecting constant)
Relate to markups in goods and labor markets
Wage-taking firms and no labor adjustment costsPrice setting:
In logs:
pt = μpt + (wt – mpnt)
μpt = pt – (wt – mpnt)
= mpnt - (wt - pt)= (yt – nt)- (wt – pt )= - st
(st is log of wage share or log of real unit labor cost: S = WN/PY)
4
Households
MRS is the marginal rate of substitution between consumption and leisure, i.e. the marginal disutility of work, measured in terms of consumption
Logsmrst = σct +φnt -
( reflect changes in preferences, low frequency)
Boom: N and C are high => UNt high and UCt low
=> MRSt high
Wage markup μwt = (wt – pt) – mrst
(difference between wage and marginal disutility of work, expressed in terms of consumption)
5
Can estimate the inefficiency gap empirically by
gapt = mrst - mpnt = σct + φnt - - (yt – nt)
Can also decompose the gap in two markups by adding and subtracting the real wage
gapt = [mrst- (wt - pt)] - [mpnt – (wt – pt) ]gapt = - μw
t - μpt
= - { μwt + μp
t}
In steady state
gap = - (μp + μw ) < 0
Empirical decomposition in price markup and wage markup is only possible if the observed wage reflects the true cost of hiring an additional unit of labor (the wage is allocational)
This is not possible if there is wage smoothing so that the firm provides “insurance” to risk averse workers, i.e. that wage is smoothed over the cycle so as to reduce income fluctuations, while employment is nevertheless set at the efficient level.
6
Measurement
gapt = mrst - mpnt = σct +φnt - - (yt – nt)
Baseline case: 1/ φ is the Frisch wage elasticity of labor supply Micro data: 1/ φ in interval 0.05-0.5Macro data : 1/ φ is unity or higher
Here : assume φ = 1
σ is the coefficient of relative risk aversion (the inverse, 1/σ, is the intertemporal elasticity of substitution)Direct estimates of 1/σ in interval 0.1 – 0.3 Balanced growth: σ = unity
Here: assume σ =1
Choose “low” values of φ and σ => Bias against finding large costs of fluctuations
Allow for low-frequency changes in preferences
by fitting a third-order polynomial in time
7
Relations hold up to additive constant => can identify variation over time, but not levels
8
Gap_t = c_t + 2n_t - y_t
Gap comoves with the business cycle (shaded areas represent NBER-dated recessions)
9
Changes in wage markup dominate variation in efficiency gap
Wage markup c_t + n_t dominates
While price markup y_t - n_t fluctuates less
10
The gap is interpreted as a measure of inefficient changes in markups.
However, we must explore an alternative interpretation, namely that labor supply preferences change, in which case the short run volatility in our gap measure does not indicate ineffiency
Yet no support for alternative interpretation:
One can reject test of no-Granger causality of detrended GDP, nominal interest rate and yield spread on gap measure (i.e. gap depends on these variables, which it should not if the gap just indicate labor supply preferences)
Furthermore, monetary policy shock also affects gap
11
Welfare costs
Assume that
= constant
12
NB: typo in (19): (1-Φ) should be Φ
NB: typo in expression for ψ: (1-Φ) should be Φ
13
14
15
16
Discussion
Focus in this paper:
Recessions involve lost output, and fluctuations involve inefficient tradeoff consumption – leisure
Other points:
representative agent, no unemployment recession give increase utility of leisure neglect costs associated with effects on
income distribution neglect other costs of unemployment
o self confidence, o loss of human capitalo social costs, etc
costs associated with wage and price inflation when wages and prices are rigid
costs of efficient fluctuations Calculations based on mean (gap) = 0
o if downturns are more persistent, then mean (gap) >> 0, implying that the welfare loss would be much greater
o neglect hysteresis (may be caused by loss of human capital)
17