EMPLOYMENT ANALYSIS: CONCEPTS, INSTRUMENTS Ekkehard ERNST Ekkehard - - PowerPoint PPT Presentation

SUMMARY O

T i i

OF TRAIN

Training summary

• Employment analysis: Concepts, instruments,

NING

p y y p applications

• Labour market information systems

l j i h d d li i

• Employment projections: Methods and an application

I f anything is unclear, please ask !

EMPLOYMENT ANALYSIS: CONCEPTS, INSTRUMENTS

Ekkehard ERNST

Port of Spain, Nov 2nd 2011

Ekkehard ERNST

OVERVIEW

O i Overview

• Concepts for labour market analysis

p y

• Measuring the labour market
• Labour market analysis: Approaches
• Labour market analysis: Models
• Labour market analysis: Policies

TRAINING OB

T i i bj ti

BJECTIVE

Training objectives

• Get an overview of commonly used employment

ES

y p y indicators

• Understanding broad trends

d i i l i i

• I ntroduction in analysis instruments…
• …and application to current problems

LABOUR MARKET ANALYSIS: CONCEPTS

LABO

Labour market analysis: Some basics

OUR MA

• Observed employment is the result of several,

decentralised decisions

RKET CO

decentralised decisions

Participating in the labour market Finding gainful employment

ONCEPTS

Deciding how many hours to work

• In technical language we write:

( ) ( ) LF

UR HOURS ETF × − × = 1

• Notation:

Notation:

ETF: Total employment (full‐time equivalent) HOURS: Hours worked per person UR: Unemployment rate LF: Labour force

LABO

Labour force

OUR MA

• The labour force is constituted by all those who are

contributing to productive employment

RKET CO

co t but g to p oduct e e p oy e t

Covers both employed and job seekers Does not cover people deciding to stay or become inactive...

• h

i bl t t k l t ( th

ONCEPTS

...or who are incapable to take up employment (e.g. those with health problems) Typically covers people above age 15.

• Inactivity can be a choice:

Taxes are too high for second earners (women) to seek for employment employment Social assistance is too generous Opportunity costs are too high (in comparison to the wage that can be earned)

LABO

Employment

OUR MA

• Employment counts from the first hour...

Employment does not indicate the number of hours worked

RKET CO

...nor the type of work carried out. It is only a numeric head‐count indicator of all those who contribute to a country’s productive capacity

ONCEPTS

contribute to a country s productive capacity

• Without employment covers different statuses:

Job seekers, i.e. Who would like to work but can’t find employment Inactive, i.e. who do not or cannot work (physically or mentally weak people) mentally weak people) Those who would like to work but have given up to search, i.e. discouraged workers

LABO

Types of employment

OUR MA

• Several forms of employment...

Dependent employment (wage earners)

RKET CO

Dependent employment (wage earners) Self‐employment (independent workers) Own account workers (e.g. entrepreneurs) I f l l t ( ith t l b t t)

ONCEPTS

Informal employment (e.g. without proper labour contract) Temporary employment

• not all of which work full time:
• ... not all of which work full time:

Full‐time employees (regularly work more than 30 hours per week) Part‐time employees (regularly work less than 30 hours per week, sometimes very few hours: even 1 hour counts !)... ...which sometimes is involuntary

LABO

Working hours

OUR MA

• Different aspects of working time

Normal working hours

RKET CO

Normal working hours Over‐time working hours Regular working hours

ONCEPTS

• For labour market analysis only regular working

hours are relevant

Productive capacity increases with every hour, whether it is overtime or not Regulation of overtime constitutes, however, important incentives for employers wrt expansion of workforce The marginal productivity may decline as average working hours increase working hours increase... ...but the average productivity (per worker) increases in any case

LABO

Unemployment

OUR MA

• Unemployment concerns all those without a job who:

Have not been working over the last week/month, not even

RKET CO

for an hour... ...are looking for a job...

• and ready to take up an occupation immediately (i e no

ONCEPTS

...and ready to take up an occupation immediately (i.e. no health problems, child care issues, etc.)

• Some job seekers are unemployed for longer spells:

Typically unemployment spells above 6 months are considered to be long‐term People tend to loose skills (both technical and non‐technical, People tend to loose skills (both technical and non technical, “soft” skills) LTU are more difficult to mobilise and activate to return to l t employment

LABO

Labour market flows I

OUR MA

• Most labour market surveys only cover “stocks”

Current situation of interviewed person in the labour

RKET CO

Current situation of interviewed person in the labour market No regard to dynamic aspects: “What have you be doing 1 th/ t / ?”

ONCEPTS

1 month/quarter/year ago?”

• Labour market theory makes use of flows

The extent to which employment is created depends on The extent to which employment is created depends on how difficult it is for an employer to find new workers It also depends on his/her expectations regarding future developments Finally, it also depends on wage earners expectations and salary requirements y q

LABO

Labour market flows II

OUR MA

• Labour market analysis needs more information

Effects of policies depend on the speed of flows more

RKET CO

Effects of policies depend on the speed of flows more than on the impact of stocks Additional sources of information can be used but are t it h l f l l d i d l b f

ONCEPTS

not quite as helpful as properly designed labour force surveys

• Some proxy indicators
• Some proxy indicators

Vacancy information, help‐wanted‐index, online ads Unemployment duration and probabilities of finding new employment

LABOUR MARKET ANALYSIS: MEASUREMENT

MEA

Employment trends across countries…

ASURING

80

Employment‐to‐population ratios (2007 vs. 2010)

THE LAB

ISL IDN KAZ NZL NOR PER RUS SWE THA

70

BOUR M

AUS AUT BEL BRA CAN CHL CHN COL CYP CZE DNK EST FIN FRA DEU IDN IRL ISR JPN KOR LUX MUS NLD PHL POL PRT ROM RUS SVK TWN UKR GBR USA VEN

60

2010

ARKET

BEL BUL HRV EST GRC HUN ITA LVA LTU MKD MLT MDA POL SLV ZFA ESP TUR

40 50

MAR

30 30 40 50 60 70 80

2007

MEA

Employment to population ratios

ASURING THE LAB

WPOP ET EPR =

BOUR M

WPOP

l l

ARKET

• ET: Total employment
• WPOP: Working‐age population, i.e. all people 15 years

and above and above EPR: Employment‐to‐population ratio

MEA

Employment index ‐ Calculations

ASURING

• Take a particular date as base year, e.g. 2005

Take a particular date as base year, e.g. 2005

THE LAB

Take a particular date as base year, e.g. 2005 Calculate the relative level of following years with respect to that base year

BOUR M

BaseYear t BaseYear t

ET ET Index ET

+

× = − 100

ARKET

• Some words of caution

When grouping countries, add the absolute employment

BaseYear

levels first before constructing the index Try to find a base year with a particular meaning (e.g. peak of the cycle) p y )

MEA

…and at the regional level: Global shifts in employment

ASURING

Employment developments (index, 2005=100)

THE LABOUR MARKET

MEA

Country level reactions of employment during the crisis

ASURING

• G20 countries have lost up to 5 percentage points of their employment rate
• In these countries, the unemployment rate more than doubled !

THE LAB

ent rates in pp.)

BOUR M

4

• 2

in employme q3 to 2009q3,

ARKET

• 6
• 4

Changes (2008q Spain nited States Canada South Africa Turkey ed Kingdom Italy Australia n Federation France Germany Japan Brazil blic of Korea Mexico Argentina China Indonesia U S Unit Russian Repub

MEA

Temporary employment took the largest hit

ASURING

Temporary employment in the EU (%‐change year‐on‐year)

THE LABOUR MARKET

MEA

Employment adjustment: Hours‐Job count mix differs across countries

ASURING THE LABOUR MARKET

MEA

Sectoral developments…

ASURING THE LABOUR MARKET

MEA

…have accelerated during the crisis

ASURING

Finl nland and Sw eden eden Sw itzerland and Japan apan

Sectoral restructuring and house price developments

THE LAB

Net Nether herlands ands Ita Italy Ko Korea Au Austria Be Belgium Chi Chile Finl nland and

H

• u

sin g d e p re ssio n

BOUR M

Portugal ugal Uni United S ed Stat ates es Au Australia Franc ance No Norway Canada anada Net Nether herlands ands

ARKET

Hungar Hungary Spai pain Greec eece New New Z Zeal ealand and Poland and Czec ech Re h Republ public Portugal ugal D k D k Uni United K ed Kingdom ngdom Irel eland and Slov

• vak

ak Re Republ public Sloveni enia Estoni

• nia

Hungar Hungary

H

• u

sin g b u b b le

Q1 Q1-2001 Q1-2002 2002 Q2-2003 2003 Q3-2004 2004 Q4-2005 2005 Q1-2007 2007 Q2-2008 2008 Q3-2009 2009 Q4-201 2010 Tu Turkey Denm enmar ark

MEA

Sectoral adjustment

ASURING

• How to calculate the intensity of employment reallocation

across sectors?

Si l l l i di

THE LAB

Simple to calculate indicator Only one number Most commonly used: Lilien indicator

BOUR M

( )

2 / 1 1

log log ⎥ ⎦ ⎤ ⎢ ⎣ ⎡∑ Δ − Δ =

J j t d jt d jt

E E E E Lilien

ARKET

1

⎦ ⎣ =

j t

E

• where j: sector, J: number of sectors, t: year/quarter/month, d:

i i d hi h l dj i id d (1 5 time period over which sectoral adjustment is considered (1 year, 5 years, 1 quarter, etc.)

• The indicator will change depending on:

the sectoral detail (number of sectors J) the period over which change is considered (i.e. d)

MEA

Unemployment developments,

ASURING

200

Evolution of unemployment (2005‐2010, 2005q4 = 100)

THE LAB

150

BOUR M

100

ARKET

50 Western Europe Eastern Europe and CIS Southern Europe North America and Oceania Asia Latin America

MEA

…long‐term unemployment, …

ASURING

180

Percentage increase in numbers of long‐term unemployed, Q1 2009–Q1 2010

THE LAB

90 120 150

BOUR M

30 60

ARKET

‐30

huania nmark stonia reland Cyprus Latvia States Spain

• rway

Finland weden ngdom

• vakia
• rtugal

ulgaria epublic

• venia

Greece Turkey France ungary Italy Japan rlands elgium Poland Malta Brazil mbourg Africa Austria

• mania

rmany ia, FYR Croatia Lith De E I C United N F Sw United Kin Slo Po Bu Czech Re Slo G T F Hu Nethe Be P Luxem South A Ro Ge Macedoni C

MEA

…and inactivity increased during the crisis...

ASURING

Inactive population (in % of working‐age‐population)

THE LABOUR MARKET

MEA

…against the background of large scale under‐employment

ASURING

Time‐related under‐employment (in % of labour force), latest year available

THE LAB

20.0 25.0

BOUR M

10.0 15.0

ARKET

0 0 5.0 0.0

MEA

Youth unemployment has accelerated

ASURING THE LABOUR MARKET

MEA

Youth unemployment by region

ASURING THE LABOUR MARKET

MEA

Unemployment flows I

ASURING

Unemployment in‐ and outflows in US, Canada, Japan and UK 35% 2.5%

THE LAB

30% Outflow

BOUR M

25% 2.0%

ARKET

20% Inflow 15% 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009* 1.5% Inflow

MEA

Unemployment flows II

ASURING

9% 0 8% Unemployment in‐ and outflows in France, Germany and Italy

THE LAB

9% 0.8%

BOUR M

7% 0.6% Outflow

ARKET

5% Inflow 4% 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009* 0.4%

EMPLOYMENT ANALYSIS

EMP

Questions for employment analysis

PLOYMEN

• How to stimulate employment

Is employment growing in line with output/productivity? Are certain categories benefiting more/less from output

NT ANALY

Are certain categories benefiting more/less from output growth? Which sectors contribute to employment most?

YSIS

• How to improve income generation

Can wages increase without creating unemployment? Do policies need to adjust to stimulate/restrict wage growth? p j / g g Is the economy gaining/loosing competitiveness?

• How to enhance labour market chances

Can workers switch easily between jobs/firms/sectors? Do job seekers find quickly re‐employment? How do policies need to adjust to stimulate labour market reactivity

EMP

How to analyse employment developments

PLOYMEN

• Three levels of analysis

Over the medium run: Link between employment and h/ d i i

NT ANALY

growth/productivity Over the short run: Trade‐off between higher wages and lower unemployment

YSIS

Dynamic analysis: Understanding labour market flows

• The (appropriate) level of analysis depends on the

availability of data:

Key Indicators of the Labour Market: Contains information on employment and growth for all 182 ILO information on employment and growth for all 182 ILO member countries KILM also has information for wages but with less coverage So far only limited information available for flows

EMP

Medium‐run analysis: The Okun’s curve I

PLOYMEN

• Traditional approach to labour market analysis:

Statistical relationship between output growth and

NT ANALY

Statistical relationship between output growth and unemployment Alternative: Elasticity between growth and employment All t i htf d l l ti f l t

YSIS: OK

Allows straightforward calculation of employment developments once GDP estimates have been carried out

KUN’S CURVE

EMP

Medium‐run analysis: The Okun’s curve II

PLOYMEN

• Underlying approach to the GET (Global employment

Trends)

NT ANALY

e ds)

• Comes in two varieties

Identify different elasticities depending on whether we are in

YSIS: OK

a recession or a boom OR: Use historical elasticities

• Bottom‐up approach: Use sectoral elasticities and

KUN’S CU

Bottom up approach: Use sectoral elasticities and aggregate

• Based on annual data

URVE

EMP

Introduction to employment elasticities I

PLOYMEN

• Quantitative measure for measuring the “employment‐

intensity” of growth

NT ANALY

te s ty o g o t

%‐change in employment given a 1‐percentage point change in economic growth E l t d l dd d ( t t) d d i t

YSIS: OK

Employment and value‐added (output) are needed input variables Examine how growth in output and employment evolve

KUN’S CU

together over time Can examine for population subsets – e.g. women, men, youth

URVE

youth Can be applied at sectoral level as well, e.g. change in aggregate output in relation to employment by sector

EMP

Introduction to employment elasticities II

PLOYMEN

• When value‐added and employment data correspond

to precisely the same group:

NT ANALY

to p ec se y t e sa e g oup:

%‐change in labour productivity given a 1‐percentage point change in economic growth S t d i d t l l t d i

YSIS: OK

Sector and industry‐level trends in an economy

• Analysing structural changes in employment:

KUN’S CU

Movement from agriculture to higher value added sectors Labour absorbing versus labour shedding industries

URVE

EMP

Calculating employment elasticities

PLOYMEN

• Two main methods:

“Arc”‐elasticity (spreadsheet calculation):

NT ANALY

( ) ( )

= − ⎛ ⎝ ⎜ ⎞ ⎠ ⎟

εi

i i i

E E E Y Y Y

1

/ /

YSIS: OK

“Point”‐elasticity (using econometric regressions):

( )

− ⎝ ⎠

i i i

Y Y Y

1

/

KUN’S CU

Y E ln ln β α + =

1:

URVE

2:

β β = ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ ∂ ∂ → ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ ∂ = ∂ E Y Y E Y Y E E ⎠ ⎝ ∂ ⎠ ⎝ E Y Y E

EMP

Which method should I use?

PLOYMEN

• The “arc” method is preferred when there are very few

data (e.g. year‐over year comparison):

NT ANALY

data (e.g. yea o e yea co pa so ):

Computationally simple (can be worked out by hand or in a spreadsheet) L d t l til lt

YSIS: OK

Leads to volatile results

• The “point” method is preferred when there are

KUN’S CU

several observations:

Provides more stable results Gi l i hi b h i bl h

URVE

Gives average relationship between the variables over the period in question, instead of relationship between start‐ and end‐points

EMP

Relationship between elasticities, productivity and employment

PLOYMEN

• When output and employment correspond to the same

group, there is a special relationship:

NT ANALY

g oup, t e e s a spec a e at o s p:

Y = E

P

i i i

×

YSIS: OK

Δ

Δ Δ

Y = E P

i i i

+

KUN’S CU

where ∆ represents the growth rate of a particular variable

• We then have:

URVE

We then have:

ε = 1 −

ΔP

ε =

ΔE

ε = 1

ΔY

ε = ΔY

EMP

Elasticities, productivity and employment: numeric example

PLOYMENT ANALY

GDP Growth Difference Arc Employme nt Elasticity + Arc

YSIS: OK

GDP Growth Employment growth Productivity growth (Productivity+ Employment) in GDP growth Employment Elasticity Productivity Elasticity Productivity Elasticity 1985 8.7 0.8 7.8 8.7

• 0.1

0.09 0.90 0.99 1986 6 3 0 2 6 5 6 4 0 0 0 03 1 03 1 00

KUN’S CU

1986 6.3

• 0.2

6.5 6.4 0.0

• 0.03

1.03 1.00 1987 5.8 5.7 0.1 5.8 0.0 0.99 0.01 1.00 1988 6.4 0.5 5.9 6.4 0.0 0.08 0.92 1.00

URVE

1989 4.8 2.7 2.1 4.7

• 0.1

0.55 0.44 0.99 1990 4.6 4.1 0.4 4.6 0.0 0.90 0.10 1.00

EMP

Interpreting employment elasticities

PLOYMEN

G

NT ANALY

GDP growth Employment elasticity Positive GDP growth Negative GDP growth

YSIS: OK

elasticity ε < 0 (-) employment growth (+) productivity growth (+) employment growth (-) productivity growth

KUN’S CU

0 ≤ ε ≤ 1 (+) employment growth (+) productivity growth (-) employment growth (-) productivity growth (+) employment growth ( ) employment growth

URVE

ε > 1 (+) employment growth (-) productivity growth (-) employment growth (+) productivity growth

EMP

Employment may be more elastic even with lower GDP growth

PLOYMEN

Employment elasticities versus GDP growth in Asia (200‐2004)

9 10

NT ANALY

7 8 9

%) China Vietnam

YSIS: OK

5 6

ual GDP growth (% India Bangladesh Thailand K

KUN’S CU

3 4

Average annu Sri Lanka Singapore Malaysia Philippines Pakistan Korea Indonesia

URVE

1 2 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Employment elasticity

EMP

Sectoral elasticities in Pakistan, 1992‐2004

PLOYMEN

Agriculture Value Added Industry Value Added Services Value Added

NT ANALY

Agriculture Industry Services Added Growth (average annual %) Added Growth (average annual %) Added Growth (average annual %)

YSIS: OK

Pakistan 0.19 1.31 1.18 2.8 4.3 4.3

• With sufficient sectoral information elasticities can be

KUN’S CU

• With sufficient sectoral information, elasticities can be

split:

Information on employment elasticity by sector

URVE

But: No information about cross elasticities: “How much employment in sector 1 is modified with a 1%‐ change in value added in sector 2” change in value added in sector 2

EMP

Main characteristics of Okun‘s elasticities

PLOYMEN

• Backward‐looking
• Dynamic

NT ANALY

y a c

• Volatile
• No “ideal” figure

YSIS: OK

• Misleading when used in isolation
• Complementary when used with GDP growth,

KUN’S CU

unemployment rates and other LMI

URVE

EMP

Problems with the Okun’s curve

PLOYMEN

• Problems:

Non‐stable relationship: Intensity of job creation varies over

NT ANALY

the business cycle Non‐stable relationship II: Intensity of job creation changes with sectoral adjustment

YSIS: OK

with sectoral adjustment Problems of statistical identification: Growth‐employment elasticity differs depending on the time horizon at which to

• KUN’S CU

look What is the right horizon?

URVE

EMP

Applying the Okun‘s curve to current problems: G20

PLOYMEN

• Estimating employment elasticities using the “Point”‐

elasticity:

NT ANALY

elasticity:

Problem: Often time series are very short, especially for employment

YSIS: OK

Missing data

• Solution:

KUN’S CU

Using panel data Group data from different countries Calculate panel‐wide elasticity on a larger sample

URVE

EMP

Applying the Okun‘s curve to current problems: G20

PLOYMEN

• But:

How to take into account differences across countries?

NT ANALY

How to take into account differences across countries? Are the country elasticities the same ?

• Solution:

YSIS: OK

Use fixed effects for level differences across countries Differentiate coefficients across countries Group data to improve efficiency of estimation

KUN’S CU

Group data to improve efficiency of estimation

GDP ET β

URVE

it it i i it

GDP ET ε β α + + =

EMP

Applying the Okun‘s curve to current problems: G20

PLOYMEN

Country Country‐specific coefficient Argentina 0.31

NT ANALY

Australia 0.57 Brazil 0.23 Canada 0.57 China 0.02 F 0 35

YSIS: OK

France 0.35 Germany 0.29 India 0.04 Indonesia 0.12 Italy 0.31

KUN’S CU

y Japan 0.28 Korea 0.34 Mexico 0.13 Russia 0.32

URVE

Saudi Arabia 0.24 South Africa 0.77 Spain 1.18 Turkey 0.32 United Kingdom 0 47 United Kingdom 0.47 USA 0.59

EMP

Applying the Okun‘s curve to current problems: G20

PLOYMENT ANALYSIS: OKUN’S CURVE

EMP

Applying the Okun‘s curve to current problems: Fiscal multiplier I

PLOYMEN

• Okun’s elasticities can also be used for policy analysis:

How much employment can be generated from a 1%‐

NT ANALY

How much employment can be generated from a 1% increase in public spending How should public spending and taxation evolve over the b i l ? P /A /C t li l?

YSIS: OK

business cycle? Pro‐/A‐/Counter‐cyclical?

• Extended Okun’s curve estimation:

Add an estimate of the impact of public spending/deficit on

KUN’S CU

Add an estimate of the impact of public spending/deficit on GDP to the employment elasticity

1

URVE

2 1 it it i i it

di P bli S GDP GDP ET δ ε β α + + + + =

2 it it i i it

ding PublicSpen GDP ε δ γ + + =

EMP

Applying the Okun‘s curve to current problems: Fiscal multiplier II

PLOYMEN

1.4 1.6 1.8 2.0 ultipliers

Emerging economies

NT ANALY

0.4 0.6 0.8 1.0 1.2 Employment mu

YSIS: OK

1.6 1.8 2.0 rs

Advanced economies

0.0 0.2

Argentina SouthAfrica Mexico Botswana Kenya China Short-term multiplier Long-term multiplier

KUN’S CU

0 6 0.8 1.0 1.2 1.4 ployment multiplier

URVE

0.0 0.2 0.4 0.6

erland Japan states nited gdom rance stralia many Italy

Emp

Switze J United s Un King Fr Aus Germ Short-term multiplier Long-term multiplier

EMP

Applying the Okun‘s curve to current problems: Fiscal multiplier III

PLOYMEN

• ver the degree of pro-cycality of government spending (1991-2008)

Net employment creation in Sub‐Saharan African countries

NT ANALY

0.1

YSIS: OK

0.1 0. yment growth % p.a.)

KUN’S CU

• 0.2

Net employ (in %

URVE

• 0.3

Low procycality Medium procycality High procycality

THE

Short‐term analysis: The Phillips curve

PHILLIPS

• Driving question:

How do wages and prices react to unemployment changes?

S CURVE

How do wages and prices react to unemployment changes? Can unemployment be lowered without driving up prices? How does employment react to a macroeconomic shock?

• Statistical observation:

Higher unemployment rates are correlated with lower rates

• f price inflation
• f price inflation

As the unemployment rate goes down, inflation starts to accelerate Question: Can this relationship be exploited by policy makers?

THE

Origins of the Phillips curve

PHILLIPS

4 00 5.00

S CURVE

2000Q4 2004Q4 2005Q3 3.00 4.00 n rate 2003Q2 2006Q4 2.00 Inflatio 2002Q1 0.00 1.00 2 00 1 00 0 00 1 00 2 00 ‐2.00 ‐1.00 0.00 1.00 2.00 Unemployment gap

• Negative slope when plotting unemployment against

inflation rate over a full business cycle (here: USA)

THE

Traditional Phillips curve

PHILLIPS

• Traditional Phillips curve:

First observed by William Phillips for the United Kingdom

S CURVE

First observed by William Phillips for the United Kingdom Statistical relationship between inflation rate and unemployment W l fi d f th t i Was also confirmed for other countries But: Did not remain constant for longer time periods

• In the 1970s the relationship broke down
• In the 1970s the relationship broke down

Unemployment remain high despite accelerating inflation Stagflation The entire Phillips curve seemed to have shifted upwards

Short‐ and long‐run Phillips curve

• New labour market concept: the structural

unemployment rate (the long‐run Phillips curve) u e p oy e t ate (t e o g u ps cu e)

Unemployment rate at which inflation is neither accelerating nor decelerating (NAIRU) At th t t th i i t f ll d ith t At that rate the economy is running at full speed without

• verheating nor deflating

Bringing unemployment rate further down is unsustainable

• ver the long‐run
• Question: What affects the structural unemployment

rate? rate?

Depends on the indicator: HP‐filter vs. structural model Unionized wage bargaining, employment protection Lack of product market competition

THE

Modern formulation of the Phillips curve

PHILLIPS

• Inflation expectations play an important role:

Price changes only in reaction to anticipated unemployment

S CURVE

Price changes only in reaction to anticipated unemployment gaps Can be combined with backward looking elements: Inflation i t d t l k f i f ti l l h i h bit persistence due to lack of information or slowly changing habits

• Inflation as a weighted average of past and (expected)

f t i fl ti future inflation:

G U l E β

t t t t

ntGap Unemployme E + + =

+ − 1 1

π β απ π

THE

Phillips curve vs. wage curve

PHILLIPS

• The Phillips curve is a short‐cut for a more elaborate

model of the labour market

S CURVE

• de o t e abou

a et

Prices are influenced by wages and capacity constraints at the firm level W i fl d b i d th l t Wages are influenced by prices and the unemployment gap Wage‐price spiral determined simultaneously by demand conditions on both labour and product markets

( )

NAIRU U E w

t wu t B w t F w t

− + Δ + Δ = Δ

− +

β π β π β

π π 1 1

( )

Y Y w w E

t px t pwB t pwF t

− + Δ + Δ = Δ

− +

β β β π

1 1

THE

Phillips curve example

PHILLIPS

Employment reaction to a temporary real wage shock

• Phillips curve models allow

full specification of the economic dynamics

S CURVE

economic dynamics

Employment reaction to a temporary technology shock

• They can take into account

y differences in structural characteristics of the labour k t ( LM fl ibilit ) market (e.g. LM flexibility)

MAT

Modern labour market analysis: A primer

TCHING A

• Starting point: The Beveridge curve

At every point in time there is co‐existence of open job vacancies and unemployed job seekers

AND UNE

p y j Depending on the position in the cycle an economy moves up and down the curve Sometimes the entire curve moves, due to structural and policy changes

EMPLOY

, p y g

MENT FLOWS

MAT

Understanding labour market flows I

TCHING A

Decomposing unemployment dynamics

t t t t t

OUT IN E L U − = Δ − Δ = Δ

AND UNE

t t t t t

Labour force growth as a function of history and incentives

T L L β β β Δ Δ Δ

EMPLOY

Lt t L t L t L L t

Tax u L L ε β β β α + + Δ + Δ + = Δ

− − 3 1 2 1 1

MENT FL

• Labour force growth determined by

Historical trends (persistence) Di d k ff (i β lik l b i )

LOWS

Discouraged worker effect (i.e. βL2 likely to be negative) Tax incentives (and other non‐tax measures such as child care provisions, etc.)

MAT

Understanding labour market flows II

TCHING A

Decomposing employment creation

t t t

tion JobDestruc Hiring ET − = Δ

AND UNE

t t t

g

Hiring as a function of the matching rate

EMPLOY

between vacancies and job seekers

( )

t t t

U V m Hiring , =

MENT FL

• Change in employment determined by

Hiring intensity

LOWS

Rate of job destruction The facility with which new vacancies V are matched with job seekers U job seekers U

MAT

Understanding labour market flows III

TCHING A

• Hiring depends on incentives to open vacancies (i.e. job

creation)

AND UNE

Demand factors: Investment, private consumption, external demand Persistence effects: Past employment rates Relative prices: Wages, user cost of capital

EMPLOY

p g , p Financial markets: Real share prices Demand pressure on the labour market (labour market tightness)

MENT FL

AD w ET n JobCreatio β β β α + + +

Job creation as a function of demand and supply factors

LOWS

JC t t t t t t JC t

r V U AD w ET n JobCreatio ε β β β β β β α + + + + + + + =

− − 1 6 5 4 3 2 1 1

MAT

Understanding labour market flows III

TCHING A

• Job destruction is a function of

Relative prices: Wages, real interest rates, tax wedge Schumpeter effect: TFP import competition

AND UNE

Job destruction determined by technological and competitive forces

REER r TFP tion JobDestruc β β β α + + + =

Schumpeter effect: TFP, import competition

EMPLOY

JDt t JD t JD t JD t JD t JD t JD JD t

AD w IMP REER r TFP tion JobDestruc ε β β β β β β α + + + + + + + =

6 5 4 3 2 1

MENT FL

• Wages

Negotiation (Nash bargaining) between firms and workers Distribution of producer rent (matching rent)

LOWS

Distribution of producer rent (matching rent)

Wages depend on reservation wages and bargaining power

( ) ( )

t t t t t

U V UB w , 1 Π + − = γ γ

MAT

Data and methodology

TCHING A

• Data:

Macro data from OECD

AND UNE

Unemployment flow estimates by Elsby et al. (2008)

‐Estimated flows based on LFS information on unemployment duration

EMPLOY

duration ‐ Match job creation/destruction rates under certain assumptions

• Methodology:

MENT FL

Start with single‐equation identification The estimate system of equations Full macro‐model on the basis of GMM

LOWS

Full macro‐model on the basis of GMM

MAT

Determinants of unemployment outflows II

TCHING A

• Decomposition of unemployment outflows shows that:

Demand components play an important role (>40%)

• d

f f l l ff ( )

AND UNE

Indication for some financial accelerator effect (>30%) Relative prices (wages) more moderate role (<20%)

EMPLOYMENT FLOWS

MAT

Determinants of unemployment inflows II

TCHING A

• Decomposition of unemployment inflows shows:

No Schumpeterian effect from import penetration (strong d d ff )

AND UNE

demand effect) Job churning due to changes in interest rates and TFP growth

EMPLOYMENT FLOWS

MAT

A simple macro framework I

TCHING A

• Unemployment flows influence each other:

Higher job destruction rates increases unemployment pool... h k h f f f ll

AND UNE

This makes it cheaper for firms to fill vacancies... This increases hiring and job creation rates... ...which makes it more difficult for other firms to find new

EMPLOY

labour... which lowers job creation rates, etc....

• An aggregate supply curve to understand interest rates:

MENT FL

• An aggregate supply curve to understand interest rates:

The basic labour flow model assumes fixed interest rates and productivity

LOWS

To analyse macroeconomic employment dynamics we need an aggregate supply curve

MAT

A simple macro framework II

TCHING A

• First step: Only fiscal policy reaction curve

Mutual dependence of unemployment flows on each other

• l

f h f h l b k

AND UNE

A policy reaction function to the state of the labour market Long‐term interest rate purely determined by changes in government debt

EMPLOY

No considerations to short‐term variations in private savings (assuming historical trend) Considering different fiscal and labour market policies

MENT FL

t i t ij t t t t t

Policy LM Macro Outflows Inflows

, , 1

ε α + + + + + =

−

Considering different fiscal and labour market policies individually

LOWS

t p t pj t t t t

• t
• j

t t t t t j

Inflows Outflows Policy Policy LM Macro Inflows Outflows

, , 1 , ,

ε α ε α + + + = + + + + + =

− t r t rj t t t t t p t pj t t t

Savings Debt Policy RIRL f f y

, , , ,

ε α + + + + =

MAT

Assessing policy effectiveness: Job destruction

TCHING A

9.5

10 Labour market spending: Contributions to job destruction (short- vs. long-term)

AND UNE

6.1 0.8

5 1

n %)

EMPLOY

• 3.9

0.8

• 1.6
• 5

ntributions (i

MENT FL

• 5.7
• 5.7
• 6.7

12 9

• 6.6

5

• 10

Con

LOWS

• 12.9
• 15

Training expenditures Public employment services Hiring incentives Unemployment benefits Direct job creation

Short-term effect tfl Long-term effect tfl

• n outflows
• n outflows

MAT

Assessing policy effectiveness: Job creation

TCHING A

39.2

40 Labour market spending: Contributions to job creation (short- vs. long-term)

AND UNE

25.7

30 4

n %)

EMPLOY

15.7 15.6

20

ntributions (in

MENT FL

5.3 4.0 3.5 3.5 7.5 2.8

10

Con

LOWS

Unemployment benefits Hiring incentives Training expenditures Public employment services Direct job creation

Short-term effect

• n outflows

Long-term effect

• n outflows
• n outflows
• n outflows

MAT

Policy effectiveness depends on macro environment: Public debt

TCHING A

Low Intermediate High

Public debt level

Government consumption

Low Intermediate High

Public debt level

Non-wage government consumption

Employment multipliers at different levels of public debt

AND UNE

4 6 8

nt estimate

Low Intermediate High 5 10 15

nt estimate

Low Intermediate High

EMPLOY

• 2

2

Coefficien

• 5

5

Coefficien

MENT FL

5 10

stimate

Wage government consumption

150 200

stimate

Spending on public employment services

LOWS

• 5

Coefficient es

• 50

50 100

Coefficient es

Median of coefficient 5% confidence interval Low Intermediate High

Public debt level

Low Intermediate High

Public debt level

MAT

Policy effectiveness depends on environment: Structural unemployment

Hiring incentives

TCHING A

100 150

stimate

Hiring incentives

AND UNE

50

Coefficient es

EMPLOY

• 50

Low Intermediate High

Structural unemployment rate

Training expenditures

MENT FL

40 60

estimate

g p

LOWS

20 20

Coefficient e

• 2

Low Intermediate High

Structural unemployment rate

MAT

Policy effectiveness depends on environment: Financial crisis times

TCHING A

15 Low Intermediate High

Financial stress tercile

Government consumption

50 Low High

Financial stress tercile

Public employment

AND UNE

5 10

efficient estimate

30 40

efficient estimate

EMPLOY

• 5

Coe

20

Coe

Direct job creation Unemployment benefits

MENT FL

300 400

mate

Low Intermediate High

Financial stress tercile

100

mate

Low Intermediate High

Financial stress tercile

LOWS

100 200

Coefficient esti

50

Coefficient esti

• 100

• 50

MAT

A simple macro framework III

TCHING A

• Second step: Endogenous short‐term interest rates

Taylor rule for interest rates

• fl

d

AND UNE

New Keynesian inflation determination Aggregate demand determined by state of the labour market

EMPLOYMENT FLOWS

MAT

Using second‐step macro model for policy simulation

TCHING A

• Estimation and simulation

Using GMM method to estimate the full model using panel d

AND UNE

data Simulating the resulting model for the “average G20” country Shock the model with the 2009 unemployment shock, i.e. the

EMPLOY

baseline scenario should yield the average decline in employment growth Create three counter‐factuals: One austerity scenario and two

MENT FL

Create three counter factuals: One austerity scenario and two public deficit scenarios (spending vs. tax reduction) Here: Only two alternative scenarios depicted

LOWS

MAT

The second‐step estimated macro model

TCHING AND UNEMPLOYMENT FLOWS

MAT

Employment recovery: The baseline scenario

TCHING AND UNEMPLOYMENT FLOWS

• Baseline scenario

Baseline scenario

Recovery in employment by 2017 to pre‐crisis trend growth rates

MAT

Employment recovery: Additional stimulus

TCHING AND UNEMPLOYMENT FLOWS

• Additional stimulus pushes employment up

MAT

Employment recovery: Austerity measures

TCHING AND UNEMPLOYMENT FLOWS

• Austerity worsens employment outlook

EXERCISES

Exercises Exercises

EXER

Exercise 1: Labour market indicators

RCISES

• Calculate labour market information from basic data for

Trinidad

Employment‐to‐population ratio Unemployment rate Sectoral employment rates p y Sectoral adjustment (annual frequency) Average hours worked

EXER

Exercise 2: Employment elasticities

RCISES

• Calculating employment elasticities for Trinidad

Calculate annual GDP and employment growth rates

• l

l l l l ( h d) Calculate yearly employment elasticities (Arc‐method) Calculate two 5‐year elasticities Calculate the Point‐elasticity Examine results

• On the basis of these employment elasticities…

how big is the jobs gap between current and pre crisis …how big is the jobs gap between current and pre‐crisis employment developments? …how big would the jobs gap be next year with a GDP growth rate that is only half as big as the current trend?

EXER

Exercise 3: Phillips curve for Trinidad

RCISES

• Construct a structural unemployment rate for ToT:

Use an historical average

• l

l f l d Calculate an HP filtered version

• Extract price inflation information

Estimate a basic Phillips curve p How does the Phillips curve change with different estimates for the structural unemployment rate How much more unemplyoment is being generated by How much more unemplyoment is being generated by bringing the inflation rate down by 1 percentage point?

The End The End