Manpower Recruitment Agency in Pakistan , Manpower Recruitment Agencies in Pakistan
Recruitment in Recovery
-
Upload
jorden-kirkland -
Category
Documents
-
view
26 -
download
2
description
Transcript of Recruitment in Recovery
04/19/23 1
Recruitment in Recovery
Mark SandersUtrecht School of Economics, Netherlands
and
Riccardo WeltersUniversity of Newcastle, Australia
04/19/23 2
Motivation• Outflow from unemployment fails
to increase in proportion to the hiring rate. Why?– Self Selection/Sorting– Signaling– Recruiting Strategies– Search Behavior
04/19/23 3
In the Literature• Burgess (1993): Unemployed job-
seekers benefit less than proportional from hiring rate increases due to increasing competition from employed job seekers.
• Russo (2000): Firms switch to more expensive advertising in tight labor markets to maintain a target arrival rate of applicants per vacancy.
04/19/23 4
Our Main Argument• Firm and job-seeker search
behavior interact.• This interaction helps explain why
the outflow rates move less than proportional to hiring rates.
• And has important ALM policy implications.
04/19/23 5
Facts for the Netherlands• Unemployed rely more on LEO (72% vs.
13%) than employed.• Employed rely more on adds (54% vs.
27%) than unemployed.• In tightening markets:
– Ads become more (36-49%) and LEO less (11-8%) effective in matching.
– Ads are more frequently used by firms, LEO less.
04/19/23 6
Strategy• Build a search model that:
– Predicts the search channel switch – Predicts the recruitment channel
switch– Allows for the interaction to produce
counter cyclical outflow/hiring rates for unemployed.
– Test these predictions in a dataset for the Netherlands
04/19/23 7
The Model
Employed
On-the-JobSearchers
Choose search effortin Channel 2
UnemployedSearchers
Choose search effort in both Channels
FirmsOpen vacancies and choose recruitment channelHires
Firing
Open
Closed
Search Channel 1
Search Channel 2
Recruitment Channel 2Recruitment Channel 1
04/19/23 8
Testable Hypotheses• Hypothesis I: In tight labor markets OJS increases,
increasing the probability of filling a vacancy in channel 2.
• Hypothesis II: In tight labor markets firms therefore switch to channel 2.
In tight labor markets unemployed job searchers increase total search effort.
The allocation of search effort between channels depends on firm recruitment channel switch (into channel 2) and the on-the-job search response (out of channel 2).
The effect of tightness on outflow is ambiguous.
• Hypothesis III: The least competitive unemployed searchers will switch to channel 1
first/more.
04/19/23 9
The Data• OSA Supply Panel:
– 4.000 persons 1994-2000 pooled– On Job Search yes/no– Search channel information only for
unemployed
• OSA Demand Panel:– Only one year used (2001) 800 firms
04/19/23 10
The Results• In a logit on OJS(1,0) we find the vacancy rate has a positive
and significant impact controlling for education, sector, contract type and experience. This supports HI.
• In an ordered logit on the importance of open recruitment channels (1-5) we find the vacancy rate has a positive and significant impact, controlling for size class, private-public and educational level of workforce.
• Similarly for the importance of the public channel (1-5) the effect is insignificant (not negative!).
• But in a logit on preference for the public channel (1,0) the vacancy rate again has a negative significant impact. Supporting HII.
• In a logit on choosing open search channels for unemployed job searchers we find the aggregate vacancy rate has a small positive impact, controlling for education (+), search duration (0), self-confidence (-) and interaction between education and confidence (mixed). No strong support for HIII.
04/19/23 11
The Results• We can accept HI and II.• But must we reject HIII?
– Institutional changes– Other channels are not considered– “Most intensely used channel” may
not be the relevant dependent variable
04/19/23 12
Hard ConclusionsThe model works so our logic is sound.The data supports the key assumptions.
But……to prove our point:
We need to probe the data furtherControl for institutional changeImprove our tightness (per channel) measureRun an ordered logit on all possible channelsBring in search intensityOther suggestions?
04/19/23 13
Tentative ConclusionsIff we can prove our point:
Unemployed job searchers require assistance in tightening labor markets to compete in the open channelSo that ALM-policy effort should be pro-cyclical.
04/19/23 14
The Model
),)1((
),(
22222
1111
LvsLusuLsmLm
LvsuLsmLmfeu
fu
Matching in closed (1) and open (2) channel:
),/)1(/(/)),/1/1(,/(
),/(/)/,(
22222222222222
111111111
feufeu
fuuf
svusvusmvmsθvsθsφ
svusmvmθssφ
Flow probability of filling a vacancy through (1) and (2):
04/19/23 15
The ModelJob finding flow probability for unemployed JS:
Job finding flow probability for employed JS:
21221
2121
222
22
11
11
)1()1(
ususus
vvss
usus
vs
us
vseuu
iu
iu
eu
iu
u
iu
u
222
22
)1(
usus
vseu
ie
e
04/19/23 16
The Model
Employed(1-u)L
On-the-JobSearchers
UnemployedSearchers
uL
Firing Rate=λ(1-u)
Hiring Rate=φ2e+(φ1
u+φ2u-φ2
e)u
Channel 1φ1
uuLChannel 2
φ2e(1-u)L +φ2
uuL
04/19/23 17
The Model
))(,/( 2,12,12,12,12,12,12,12,1 VJsvusφscrV fuf
First Order Condition on search effort:
Firms choose search effort per channel:
2,1
2,1
2,1
2,1 (.)'
(.) fsφ
φr
c
Jr
sf1,2 is negative in channel specific flow cost and
the marginal effect on the probability of filling the vacancy through that channel and positive in interest rate and job value as well as, obviously, the probability of filling the vacancy through that channel.
04/19/23 18
The Model
02,1 V
Together with the FOC his implies:
Firms open vacancies in both channels:
2,1
2,1
'φ
cJ
As the marginal probability is decreasing in the vacancy rate:Higher job value increases number of vacancies in both channels.Higher costs will reduce vacancies in that channel.
04/19/23 19
The Model
JψλwprJ e )(
Implies:
Value of a filled vacancy (job):
eψλr
wpJ
Allows for expressing optimal search effort per channel in aggregate variables only. Effort in a channel depends positively on the effort of job searchers in that channel. Tightening marketsWill cause firms to shift towards the open channel.
04/19/23 20
The Model
)( UWψzrU u
Implies:
Unemployed Job Searchers:
0(.)
)(.))('(.)'(2
21
u
su
su
ψr
zrWψψ iu
iu
Which implies that unemployed job searchers set effortsuch that marginal probabilities equalize. This implies they switch to channel 1 when employed job searchers search effort increases in channel 2.
04/19/23 21
The Model
wμψUWλswrW eie )()1( 2
Setting effort to maximize yields:
Employed Job Searchers:
μψ i
ese 1
(.)' 2
Which implies that employed job searchers set effortin response to a wage mark-up and reduce effort when unemployment or the search effort of unemployed in channel 2 increases.
04/19/23 22
The Model
ββ
wwJVwJwWUwWw 1)))(()(()))(()((maxarg*
Which yields:
Solve for equilibrium wage:
ee μψs
zβw
21
)1(*
And closes the model.
04/19/23 23
The ResultsTable 1: Job search decision employees, pooled 1992-2000, clustered1
OJS[1,0]=βθ+γX+εModel 1
Logit
Vacancy Rate 1.053 (0.03)
Education: Primary education 0.663 (0.10)
Lower vocational 0.662 (0.07)
Intermediate vocational 0.816 (0.08)
Higher vocational 0.950 (0.11)
University reference
Contract type: Fixed term (future) 5.238 (0.52)
Fixed term (no future) 1.78 (0.13)
Permanent reference
Experience 0.954 (0.00)
Sector Dummies Jointly significant
04/19/23 24
The ResultsTable 2: Recruitment intensity in open channels, 2001
Open[1…5]= βθ+γX+ε
Model 1 Model 2 Model 3
Tightness Proxy Vacancies/Employment Difficult Vacancies/ Employment
Difficult Vacancies/ Vacancies
Β 5.27 (3.03) 8.92 (2.64) 1.74 (2.94)
Size 0 - 9 0.24 (6.25) 0.24 (6.21) 0.32 (3.29)
Size 10 - 24 0.41 (5.30) 0.41 (5.29) 0.52 (2.65)
Size 25 - 49 0.65 (2.44) 0.64 (2.49) 0.75 (1.26)
Size 50 - 100 0.78 (1.30) 0.80 (1.18) 0.93 (0.30)
Size > 100 reference reference reference
Private sector 0.59 (2.40) 0.58 (2.45) 0.56 (2.03)
Public sector reference reference reference
75-100% < MAVO 0.50 (3.62) 0.51 (3.50) 0.54 (2.33)
50-74% < MAVO 0.64 (2.35) 0.62 (2.51) 0.84 (0.69)
25-49% < MAVO 1.32 (1.79) 1.36 (1.92) 1.46 (1.97)
< 25% < MAVO reference reference reference
Sector Dummies Jointly significant Jointly significant Jointly significant
04/19/23 25
The ResultsTable 3: Recruitment intensity in public channels, 2001
Public[1…5]=
βθ+γX+ε
Model 1 Model 2 Model 3
Tightness Proxy Vacancies/Employment Difficult Vacancies/ Employment
Difficult Vacancies/ Vacancies
Β 2.68 (1.61) 1.92 (0.76) 1.28 (1.26)
Size 0-9 0.37 (4.05) 0.38 (4.06) 0.47 (2.07)
Size 10-24 0.59 (2.88) 0.60 (2.78) 0.47 (2.83)
Size 25-49 0.91 (0.51) 0.90 (0.58) 1.07 (0.30)
Size 50-100 0.71 (1.69) 0.72 (1.60) 0.81 (0.81)
Size > 100 reference reference reference
Private sector 0.79 (0.99) 0.77 (1.12) 1.00 (0.00)
Public sector reference reference reference
75-100% < MAVO 1.08 (0.38) 1.07 (0.34) 1.23 (0.51)
50-74% < MAVO 1.10 (0.51) 1.10 (0.50) 1.34 (1.17)
25-49% < MAVO 1.02 (0.09) 0.99 (0.03) 0.94 (0.31)
< 25% < MAVO reference reference reference
Sector Dummies Jointly significant Jointly significant Jointly significant
04/19/23 26
The ResultsTable 4: Recruitment preference for public channel, 2001
PublicPref[1,0]=
βθ+γX+ε
Model 1 Model 2 Model 3
Tightness Proxy Vacancies/Employment Difficult Vacancies/ Employment
Difficult Vacancies/ Vacancies
Β 0.18 (2.08) 0.03 (2.92) 0.80 (1.00)
Size 0-9 3.34 (5.35) 3.36 (5.41) 2.06 (2.15)
Size 10-24 2.29 (4.57) 2.40 (4.81) 1.76 (2.15)
Size 25-49 1.87 (3.23) 1.92 (3.32) 1.73 (2.18)
Size 50-100 1.26 (1.04) 1.26 (1.01) 1.13 (0.45)
Size > 100 reference reference reference
Private sector 1.04 (0.29) 1.06 (0.37) 1.08 (0.41)
Public sector reference reference reference
75-100% < MAVO 1.64 (2.52) 1.63 (2.48) 1.07 (0.24)
50-74% < MAVO 1.59 (2.37) 1.61 (2.42) 1.27 (0.94)
25-49% < MAVO 1.05 (0.27) 1.02 (0.11) 0.83 (0.75)
< 25% < MAVO reference reference reference
Sector Dummies Jointly significant Jointly significant Jointly significant
04/19/23 27
Table 5: Channel choice unemployed job searcher, pooled 1994-2000, clustered1
Open[0,1]=
βθ+γX+ε
Model 1 Model 2 Model 3
Vacancy rate 1.072 (3.46) 1.044 (1.96) 1.067 (3.50)
Higher education 1.174 (0.53) 9.118 (2.08) 2.516 (3.05)
Lower education reference reference reference
Search duration (mths) 0.995 (0.52) 0.997 (0.29) 1.002 (0.60)
Search duration (>1 y.) 1.000 (0.49) 1.000 (0.46) -
Search duration (< 1 y.)
reference reference -
Confidence (very high=1..very low=5)
1.165 (1.78) - -
C=1: very high conf. - reference 9.24*10-7 (17.31)
C=2: high conf. - 3.365 (1.08)
referenceC=3: neutral - 3.860 (1.98)
C=4: low conf. - 2.443 (1.05)
C=5: very low conf. - 5.883 (2.17) 3.967 (1.75)
Dummies C - Jointly significant Jointly significant
C=1xE=1 - reference 2.96*10-8 (-)
C=2xE=1 - 0.073 (1.66)
referenceC=3xE=1 - 0.090 (1.85)
C=4xE=1 - 0.179 (1.25)
C=5xE=1 - 0.039 (2.46) 0.146 (2.30)
Interactions CxE - Jointly significant Jointly significant