Unit a: Unemployment

EMAC /Lecture

March
 Outline

Definitions

Stylised facts
Okun's Law

E ficiency wages
Causes
Solow condition
Shirking model
Macroeconomic e ficiency wage model
 Objectives/outcomes for U

* Analyse the relationship between output growth and unemployment.

* Use the shirking model of e ficiency wages to obtain and describe the non-shirking condition.
* Use e ficiency wage and monopoly union models to analyse the causes of changes in wage and
unemployment rates.
* Use macroeconomic labour market models to describe hysteresis.
* Use macroeconomic data to explore the relationship between unemployment and economic growth.
 Literature

* Sorensen & Whitta-Jacobsen (SW): Ch. -

* Marinkov & Geldenhuys, : Okun's coe ficient for SA
/ Definitions
 Definitions

* Working-age population (WAP) = all people between and years old

* Labour force (L) = working-age people who are willing and able to work
* Includes employed (E) and unemployed (U)
* LFPR = U+EWAP
* Not economically active: working-age people who are not willing and/or able to work
* Unemployed = all people who are able and willing to work
* Who don't have a job, and have taken active steps to find one (strict)
* Who don't have a job, but have not taken active steps to find one (broad; includes discouraged workers)
* Unemployment rate = u = UL
E
* Employment rate e = 1 − u = L
* Absorption rate (a) = employed people as fraction of working-age population
 Types of unemployment

* Voluntary vs. involuntary

* Cyclical vs. structural
* Cyclical: Due to temporary negative supply/demand shocks (recessions)
* Arises due to nominal wage rigidity in most neo-classical models (but real wages may be lexible; are not lexible enough)
* Use stabilisation policy (expansionary fiscal and monetary policy)
* Structural: Due to real wage rigidity
* Types of structural unemployment:
* Frictional unemployment (takes time to start new job)
* Search unemployment (takes time to find best available job; have to weigh cost of search and benefits of search )
Frictional and search = relatively small component of overall unemployment (see SW, . )
* E ficiency wage considerations (e fort depends on wages, high wages required for high e fort)
* Trade unions (push up wages to maximise average income of members)
* Requires use of structural policy (labour market/product market reforms, education)
 Unemployment due to rigidities (F . )

* Rigidity (stickiness) prevents wage from reaching w* to clear market ...

* SR nominal rigidity leads to cyclical (SR) unemployment...
* LR real rigidity leads to structural (LR) unemployment
/ Stylised facts
 Some facts about unemployment (SW, . )

* In most (advanced) economies, no long-run trend in unemployment

* There is a lot of variability in unemployment in the short-run
* Measured using variance, standard deviation, coe ficient of variation
* There is a lot of persistence in unemployment
* Measured using autocorrelation coe ficient
* = correlation between current and lagged unemployment
* Long-term unemployment ( months plus in many countries; year plus in SA) tends to vary positively with
overall unemployment
* Large di ferences in unemployment rates across geographical areas, over time
* In SA: lowest unemployment rates in WC (and Northern Cape).
* Free State: slightly higher than national rate (tends to be one of the provinces with highest o ficial rate...)
* Large and persistent di ferences in unemployment rates across for people with di ferent levels of educational
attainment
* Unemployment tends to be lowest for those with post-secondary education ...
 Okun's Law

* Another important fact is that unemployment and economic growth are strongly correlated
* Specifically, the change in the unemployment rate (∆u) is strongly negatively correlated with the GDP
growth rate (or the change in the GDP growth rate)
 Okun's Law: SA application – Marinkov& Geldenhuys,

* Authors find evidence of Okun's Law for SA (across wide variety of specifications
* Also find evidence of (weak) correlation between employment and output growth; unemployment and
output growth ... no evidence of ”jobless” growth
* But: unemployment increased in SA for about consecutive years (between and ), even though
there was positive economic growth during some of these years. Problem = this growth was too low ...
 Decomposing ∆u

* Relationship between u, E and Y:

* ∆u = gLF − gE
* Let gLF = gLFPR + gWAP , while gE = gY − gAPL
* ∴ ∆u = gLFPR + gWAP + gAPL − gY = gReq − gY
* If gReq > gY ⇒ ∆u > 0
* If gReq < gY ⇒ ∆u < 0
* If gReq = gY ⇒ ∆u = 0
 SA: ∆u, gReq and gY

* Sustained increase in unemployment during s and s, because output growth (GDP) was less than
required growth
 Estimating Okun (MG )

* In its simplest form, Okun's law postulates a negative relationship between changes in u and GDP growth
(e.g., see F . )
* These days, Okun's law is understood to explain the relationship between unemployment and growth in the
SR (over the business cycle)
* In modern specifications, Okun's Law postulates a negative relationship between the cyclical components of
unemployment and output
* uc = β0 + β1 yc + error, where uc = (u − ū) and yc = (y − ȳ)
* To obtain yc and uc , need time series filters that can decompose series into trend (structural/natural) and cyclical
(short-run) components (e.g. HP, BP, linear-detrending, first-di ferencing, etc.)
* Test: H1 : β1 < 0
 Estimating Okun: extensions

* More sophisticated specifications will include lags of uc and yc to account for persistence and the fact that
higher output doesn't immediately translate into lower unemployment.
* E.g. uc = β0 + β1 uct−1 + β2 yct−1 + ...
* Asymmetries can be incorporated to allow di ferent responses during recessions and upswings
* e.g.: uc = β0 + β1 (ycdown ) + β2 (ycup ) + ...
* Asymmetries due to: restrictions on firing and firm investment in specific training of employees (stronger in
upswing) or if firms are more likely to fire during downswings than hire during upswings (stronger in downswing)
 Estimating Okun: results

* Main result: reject H (irrespective of type of specification): there is therefore a negative relationship
between cyclical unemployment and cyclical output in SA
* Little evidence of asymmetries
* This means there is scope for expansionary fiscal and monetary policy to lower unemployment in SA
* But: most of SA unemployed are LR/structurally unemployed – expansionary policies can only go so far.
Structural policy (education, some labour market reform) may also be required
/ E ficiency wages
 E ficiency wages (SW, . )

* E ficiency wage = idea that productivity of workers depends positively on wages they receive
* Specifically, that workers will work hard (high productivity) if wage paid by firm is greater than income they
can earn if not working for firm (unemployment benefits; wages at other firms)
da d a 2
* a = a(w), where dw = a′ > 0 and dw2 = a
′′
< 0 (increasing, concave)
* Under e ficiency wages, profit maximising firms may end up paying workers a wage that is higher than the
market-clearing wage, leading to unemployment
 Causes of e ficiency wages (SW, . )

* Reduce labour-turnover
* Worker recruitment
* Worker-disciplining device
* Reciprocity
 Solow condition

* Suppose that Y = aL and a = a(w)

* If price = constant and normalized to one, then R = Y = aL = a(w)L, while C = wL
* This implies that Π = a(w)L − wL
* To maximise profit: Πw = ΠL = 0
* Πw = ∂Π∂w
= a′ (w)L − L = 0 ⇒ a′ (w)L = L ⇒ a′ (w) = 1
* ΠL = ∂L = a(w) − w = 0 ⇒ a(w) = w ⇒ a(w)
∂Π
w
=1
* Since 1 = 1 ⇒ a′ (w) = a(w)
w
* a′ (w) = marginal; a(w)/w = average ... to max profit, average and marginal productivity must be equal .. this
also means that ...
a′ (w)
* a(w)/w = 1 ... the real wage elasticity of productivity = to max profits ...
* Solow-condition
Solow condition: e ficiency curve (SW, F . )
Solow condition: e ficiency curve
Solow condition: e ficiency curve
Solow condition: e ficiency curve
 Couple of notes about this diagram

* EC = e ficiency curve
* v = outside option
* w∗ = wX = profit maximising wage
* a/w = average productivity
* To max profits, choose w so that marginal productivity = average productivity (a′ = a/w)
* This only occurs at w∗ = wX
* At w, (average) productivity is greater, but marginal is less; cost e fect of higher wages exceeds productivity e fect
* Note: slope of tangent = marginal; slope of ray = average
 Shirking model (SW, . )

* Firms will pay workers w if they exert required e fort â

* But firms will fire workers who put in less e fort (i.e. if a < â)
* Worker is shirking if a < â
* Putting in e fort is costly (disutility of work): c(a) = cost of e fort, where c′ > 0 and c′′ > 0 (increasing and convex)
* Firms can only monitor workers imperfectly, and catch shirkers with probability 0 < q < 1
* Since e fort is costly, and all shirkers who are caught are fired, let shirking workers put in no e fort (a = 0)
* Since e fort is costly, those who don't shirk put in exactly the required e fort (a = â)
* For non-shirking workers, their utility is given by Uns = w − c(â)
* For shirking workers, utility is given by Us = (1 − q)w + qv, where v = outside option
* Their utility = weighted average of wage (if not caught) and outside option (if caught)
 No-shirking condition

* To not shirk, utility from not shirking must be at least as great as utility from shirking
* w − c(â) ≥ (1 − q)w + qv
c(â)
* Solving for wage: w∗ ≥ v + q ... so-called non-shirking condition (NSC)
c(â)
* Firms will pay minimum required wage for required e fort, so that NSC is w∗ = v + q
* Note that:
∂w∗
* ∂v
=1>0
∂w∗
* ∂q
= −q−2 < 0
∂w∗ 1 dc da
* ∂c
= q da
× dw
>0
∗
* This means that w increases if
* the outside option increases
* probability of catching shirkers decreases
* cost of e fort is increases (or if required e fort is greater)
NSC
Higher v
Higher v
Lower v
Lower v
Higher q
Higher q
Lower q
Lower q
 Partial equilibrium model (SW, . )

* If productivity does not depend on wage, then firm will choose wage B (max profit by min wage)
* But if productivity depends on wage, then firm may choose point o f Ls (e.g. A), to max profit
* Now: wage increase leads to higher productivity and revenue, but also to higher cost. Must find point at which
revenue e fect o fsets cost e fect to maximise profit
* Now, to max profit, firm must minimise not w, but w/a = unit labour cost (ULC): if wages increase by %,
productivity must also increase by % to keep w/a constant ...
* Unemployment possible ... but if w that maximises profit lies at B, then firm will choose it ...
 A macroeconomic e ficiency wage model (SW, . )

* Monopolistic competition: each firm's product is a little di ferent from those of other firms. Each firm is a
”local monopolist” and a price setter, but firms compete because their products can substitute for each other
( )−σ
* Demand function: D(Pi ) = PPi Y
n
* Y = GDP; P = price index; Pi = firm's price; σ = price elasticity of demand
* Note: σ is indicator of product market competition: large sigma = lots of competition
* Production function: Yi = ai Li
* ai = (wi − v)η , 0 < η < 1 (eta = wage elasticity of e fort)
 Price and wage setting

* Goal of firm = max profit

( )
wi
* Profit: Πi = (pi )−σ nY pi − (wi −v)η , where wi , pi are real/relative wages and prices
* The firm's problem is now to choose wi and pi to max profits
( )
wi wi
* The price that achieves this is pi = σ−1
σ
× (wi −v) η ≡ m (wi −v)η
v
* While the wage that maximises the firm's profit is wi = 1−η
* Implication:
* Price that maximises profit = mark-up over ULC (mark-up = m = σ−1
σ
) ... higher sigma, lower m
1
* Wage that maximises profit = mark-up over outside option (mark-up = 1−η ... higher eta = higher mark-up
 Wage curve (WS)

* Let outside option be v = ub + (1 − u)w ... income that can be earned if not employed by firm
* b = unemployment benefits (UIF in SA)
* In SA, grants also part of outside option
* Firms are symmetric: use same labour, production function, etc... Also, face same outside option, so they set
the same wage ..
v ub+(1−u)w
* Therefore, w = 1−η = 1−η = 1−η/u b
1

* So w is just mark-up over unemployment benefit

* WS is negatively sloped in w-u space (F . ): with higher u, v is lower, which means wage for given e fort level
is lower, which means less wage pressure
* If w is negatively sloped in w-u space, then it is positively sloped in w-e space (e = 1 − u): w = 1−η/(1−e)
1
b
* Higher benefits (b) shi t WS up
WS curve: w-u (F . ) and w-e space
 The price curve (PS)

wi
* Firms set prices to max profits using pi = m (wi −v) η

* All set same wage; so all set same relative price ... so all set same price ...
wi
Pi /P = 1 ⇒ pi = 1 ⇒ 1 = m (wi −v) η ∴ mw = (w − v)
η

* Since v = ub + (1 − u)w: mw = [u(w − b)]η = (1 − e)η (w − b)η

* This relation implies a negative relationship between e and w (positive between u and w) – the price curve
(PS): with higher e, v is higher; this means lower productivity at current wage, which means ULC (= w/a)
increases, which raises prices, which leads to lower real wage
The price curve (F . ), plus PS in w-e space

PS shi ts down if b and/or m increases

 Macroeconomic equilibrium

* Macroeconomic equilibrium given by intersection of WS-PS

* WS-PS model:
* w = 1−e−η
1−e
b (WS)
* mw = (1 − e)η (w − b)η
* Let b = cw (c = replacement ratio)
* In equilibrium: e∗ = 1−c−η
1−c ⇒ u∗ = η
1−c ... u increases in eta and c
* GDP = Y = aeL ... so increase in e will lead to increase in GDP
Macroeconomic equilibrium (F . )
 Implication of macro equilibrim: in the e ficiency wage WS-PS model, ū > 0...

* In the e ficiency wage WS-PS model, there is necessarily unemployment in equilibrium ...
* Why?
* Recall that v = ub + (1 − u)w = (1 − e)b + ew and a = (w − v)η
* Positive e fort only if w > v.
* w > v only if u > 0 (e < 1)
* To ensure positive e fort, firms will increase wages if they are close to outside option ... but this means that there will be
unemployment in equilibrium.
* In the partial equilibrium model (see Solow condition discussion), unemployment is possible, but not guaranteed
in equilibrium ...
* Recall that, if productivity depends on wage, a firm may choose a wage that lies above the labour supply curve Ls to
maximise its profit
a′ (w)
* The wage that the firm chooses is determined by the Solow condition: a(w) = 1 ⇒ a(w) = a′ (w)
* Unemployment is possible, but if the w that maximises profit lies on the labour supply curve, then the firm will choose this
wage, and there is no unemployment ...
 Increase in b (F . )

* With increase in b, WS shi ts up: higher b leads to higher v, and firms must increase w to maintain e fort at all
levels of e ...
* With increase in b, PS shi ts down: higher b leads to higher v, which leads to lower a at prevailing wages. This
leads to higher ULC and higher prices, and lower real wages at all e ...
* Higher benefits lead to lower employment, higher unemployment, and lower GDP (e fect on wages
ambiguous)
Increase in b
Increase in b
Increase in b
 Decrease in b

* With decrease in b, WS shi ts down: lower b leads to lower v, and firms can lower w to maintain e fort at all
levels of e ...
* With decrease in b, PS shi ts up: lower b leads to lower v, which leads to higher a at prevailing wages. This
leads to lower ULC and lower prices, and higher real wages at all e ...
* lower benefits lead to higher employment, lower unemployment, and higher GDP (e fect on wages
ambiguous)
Lower b
Lower b
Lower b
Lower b
 Increase in sigma

* Increase in sigma leads to lower mark-ups and lower prices, which leads to higher real wages at every e ... PS
shi ts up
* With higher e and w, u is lower and GDP is higher
higher sigma
higher sigma
 lower sigma

* Decrease in sigma leads to higher mark-ups and higher prices, which leads to lower real wages at every e ...
PS shi ts down
* With lower e and w, u is higher and GDP is lower
lower sigma
lower sigma
lower sigma
 Using e ficiency wage WS-PS to explain di ference in ū or ē between countries

* In this model, di ferences in ē (and therefore ū) between countries may be due to:
* Lower unemployment benefits, ⇓ b ... or grants in SA case ... (does not mean grants are 'bad’, just that there are
trade-o fs)
* The country with lower unemployment benefits has lower WS and higher PS than the country with higher unemployment
benefits...
* Lower mark-ups, ⇓ mp ⇒⇑ σ = less market power... potentially due to more stringent competition policy ...
* The country with more product market competition (less product market power for firms/producers) has lower PS than the
country with less product market competition (more market power for firms/producers)
* So, in the e feciency wage WS-PS model, country with lower (LR) unemployment (ū) has lower
unemployment benefits and/or more product market competition ...
 In closing

* Work through this lecture's homework questions

* Next time: Unit Unemployment (contd.) – Trade union models (SW, ch. ).
* Also: hysteresis (C&S , ch. . )
* Assignment ...
* Covers U -U (L -L )
* Consider homework questions marked [for assignment purposes] for inspiration
* Will be available from Monday, March. Submit by Friday, March.
* This is an individual assignment. All answers that you submit must be your own.