Employee Screening: Employee Screening: Theory and EvidenceTheory and Evidence

Fali Huang Peter Fali Huang Peter Cappelli Cappelli

SESS, SMU SESS, SMU Wharton, Wharton, UPennUPenn

Dec. 18, 2006 Dec. 18, 2006

Introduction

To elicit effort from employees when effort is difficult to observe, a firm may

screen job candidates carefully to select workers with intrinsic motivation, or

monitor employees intensively to prevent shirking.

We analyze the relationship between screening and monitoring in the context of a principal-agent model,

and test the theoretical results using a national sample of U.S. establishments.

Contributions to Related Literature

Personnel economics

High-performance work practices are associated with less monitoring, but screening is often neglected in the literature. Holmstrom and Milgrom 1991; Ichniowski, Shaw, and Prennushi 1997. Cappelli and Newmark (2001).

Social PreferencesWorkers with higher intrinsic motivation make extra effort and receive higher wages. But this has not been tested on any national sample of firms.

Gift exchange (Akerlof 1982); organization identity (Akerlof and Kranton 2005).

Model: Production and Monitoring A continuum of principals and agents with equal mass. Production function:

making effort shirking

Prob. of producing h p q

Prob. of producing 0 1-p 1-q

Cost of effort c . Making effort is socially optimal: hp-c > hq. Monitoring:

monitoring intensity:

Shirking is caught by principal with prob.

Total monitoring cost:

[0,1]im ( ),ip m

i mm k

Heterogeneous Agents and Screening

Agents differ in conscientious levels:

A cooperative type agent has conscientiousness > 0: he feels guilty of the amount if he shirks.

A selfish type agent has zero conscientiousness. The proportion of cooperative types is ρ (0,1).∈ Screening:

Prob (detecting a selfish type agent) = r (0,1); ∈Prob (detecting a cooperative type agent) = 1.

Screening principal hires the 1st agent perceived to be coop.

Prob (a hired agent is cooperative type):s

1 r1 . #

Wage and Utility

The wage for an agent with is

Utility function:

,i iw b d i

Baseline Wage Incentive Pay

(1 ( )) ( ) ( ) , if shirks;( , , , )

( ) ( ) , if not;i i i

i i ii

p m u d v bU b d m

u d v b c

Make effort if ( ) ( ) .i ip m u d c

Timeline

Incentivepackage

announced.

( , , , )i ib d m s

Agentsapplyfor jobsat theirpreferredprincipals.

Principalsscreen jobcandidates,hire the firstone perceivedcooperative,rejects others.

Agentsconsume thebase wage,choose tomake effortor shirk.

Principalsmonitor,not payincentivewage ifshirking detected.

Screen Effort Choice MonitorMatchingPackage Choice

Equilibrium: Principals max profit, Agents max utility, Market clear.

The Equilibrium

Proposition. (1) In the equilibrium, a proportion of principals screen job

candidates with intensity , while others do not screen. The optimal solution is uniquely determined, where

The optimal wages are

(2) Both types of agents prefer to work for screening principals. A cooperative type agent makes effort while a selfish one shirks if hired by a screening principal, while neither shirks under a non-screening principal.

s

s

( , )ss m

0s

sm

.s nw w

NES97 data The data come from establishment‑level surveys of

employment practices conducted by the U.S. Bureau of the Census in 1997 for the National Center on the Educational Quality of the Workforce.

It is a nationally representative sample of private establishments with more than 20 employees.

It appears to be the best data available In terms of representativeness, response rate, and breadth of questions about work practices and organizational characteristics.

Variable Selections The screening selectivity is measured by two variables:

Candidates#: The number of candidates interviewed for the job opening of a typical production employee;

about 90% of the firms interview between 2 and 10 candidates, while the mean is 7 (SD 8).

Attitude Screening: The importance of a candidate’s attitude (mean 4.4 out of 5) in hiring decision, or the average of the importance of attitude and communication skills (mean 4.1).

The monitoring intensity is measured negatively by

Employee-Supervisor Ratio: mean 19 (SD 21)

Teamwork: Percent of production employees in teamwork (16%)

Control VariablesSchooling for production employees

Working hours per week for the production employees

Computer usage (supervisors): percent using computers in their jobs

Computer usage (production employees)

Union ratio: percent of non-managerial, non-supervisory employees covered by a collective-bargaining agreement

Months to reach job proficiency for a typical new hire

Ratio of women: percent among permanent employees

Minority ratio: percent among permanent employees

Job rotation: percent of non managerial/supervisory employees currently involved

Average employee benefits

Five size dummies and 21 industry dummies

Teamwork .013**(.006)

.013**(.006)

.118***(.040)

.108***(.040)

.0019***(.0005)

.003***(.001)

.011***(.003)

.008***(.003)

Screening Selectivity

Candidates# Attitude Screening

OLS (1) OLS (2)2SLS

(1)2SLS

(2)OLS Probit

2SLS(1)

2SLS(2)

Employee-supervisor ratio

.020**(.009)

.019**(.009)

.298***(.112)

.284**(.113)

.0014*(.0007)

.0042**(.0018)

.022***(.0087)

.019**(.0079)

Screening on Work Exper.

_.075(.296)

_.238(.36)

_.256***(.051)

_.173***(.026)

Screening on Acad. Perform.

_1.08***(.279)

_.77**(.35)

_.020(.047)

_.267***(.026)

Observation 2049 2037 2031 2020 2156 1902 2134 2123

R-square .044 .054 _ _ .18 _ _ _

F-stat in 1st stage regression

_ _ 73.34 9.72 _ _ 10.21 75.65

Table 2: The Trade-Off between Screening and Monitoring

Table 3: Wages and Screening Selectivity

OLS (1) (2) (3) (4) (1C) (2C) (3C) (4C)

Candidates# .0020** (.0008)

.0019** (.0008)

.0019** (.0008)

.0018** (.0008)

.021*** (.0073)

.019*** (.0073)

.011 (.008)

.015 (.010)

Attitude Screening

.030*** (.009)

.029*** (.010)

.35** (.14)

.71*** (.25)

Candidates#-Residual

-.02***(.0073)

-.018**(.0074)

-.010(.0084)

-.014(.010)

AttitudeScreen-Residual

-.32**(.14)

-.68***(.25)Screening on

Work Exper..015(.01)

.010(.01)

.0008(.01)

-.32(.20)

Screening on Acad. Perfor.

.009(.01)

-.00(.01)

.02*(.01)

-.025(.16)

WorkExperiences-residual

.32(.20)

AcademicPerformance-residual

.036(.16)

Observation 1882 1876 1877 1873 1241 1238 1238 1237

R-square .59 .59 .59 .59 .54 .54 .54 .55

Concluding Remarks

We analyze the relationship between screening selectivity and monitoring intensity in the context of a principal-agent model and test the theoretical results using a national sample of U.S. establishments.

We found a substitution relationship between screening selectivity in work ethic and monitoring intensity, and a complementary relationship between screening selectivity and high performance work practices (especially teamwork).

High screening selectivity leads to high wages.