2.2 Screening

Signaling:

the worker moves first, sending a credible signal to the firm about his type. The firm pays the worker the wage corresponding to his type. Example: an applicant mentions a university degree in the cover letter. University graduates have, on average, higher ability than non-graduates. Hence, the firm will offer a higher wage to graduates than non-graduates.

Screening:

the firm moves first, putting job applicants to a test that helps identify their type. Example: interviews and probation periods.

—>Screening and signaling can happen simultaneously and be done by both applicants and firm.

• Screening for credentials, often done automatically.

• Screening is often multi-stage to minimize costs per applicant

  • Formal assessment through tests, exercise, etc.

  • Face-to-face interviews.

  • Probationary periods.

• Information:

  no screening tool can give perfect information about the agent type, statistical errors are possible.

• Efficiency:

  screening activities do not make anyone any more productive, so in a sense are “wasteful”. Sometimes signaling is wasteful, too.

• Procedural issues:

  rat races, legal restrictions in case of probationary periods, subjectivity of interview impressions.

—>Therefore, more screening is optimally required for high-value jobs to maximize the value of the worker-job match.

• Were they better matches than the average of the initial 1200 applicants?

• But, better or tougher selection implies higher recruitment and screening costs.

• Utility analysis is an HR technique used to calculate the costs and benefits of screening.

• We study the Naylor-Shine utility model: Here the quality of selection is the average standard score on a measure of job performance for the selected group.

• Other utility models are explained in Cascio and Boudreau (2008) Investing in People, (see chapter 8).

  • Taylor-Russel Model: The proportion of individuals in the selected group who are considered successful.

  • Brogden-Cronbach-Gleser Model: The dollar payoff to the organization resulting from the use of a particular selection procedure

• Suppose there is no screening, you just hire very first random people who applies.

• How much more productive will this worker be compared to the average?

• The expected productivity of hired workers without selection is just population average. So, the answer is 0.

• Now, suppose your recruitment procedure selects with a selection rate α, that is, the top share α of the applicants (based on a recruitment test) are hired.

  —>Intuitively, if the test is positively related to productivity, they will be more productive than population average. But how much more?

Exercise: Which assessment measure has high/low validity?

• Assessment centers .36

• Biographical data measures .35

• Cognitive ability tests

• Conscientiousness tests .22

• Employment interviews (structured) .46

• Employment interviews (unstructured) .58

• Graphology .02

• Integrity tests .46

• Reference checks .26

• Work sample tests .33

• Can be done directly if data on individual worker performance are available.

• But such data are often unavailable. So the use of reasonable “guesstimates" is ok.

• Option 1: compute the actual average score of the selected applicants.

  But: not applicable when we are interested in assessing the expected benefits of screening procedures before actually using them.

• Option 2: Calculate 𝐸(𝑥|𝑥 > 𝑥𝑎 ) under a certain assumed distribution of x.

• Most frequently assumed distribution is normal. If 𝑥~𝑁(0,1) (by normalisation), 𝑥I𝑥 > 𝑥𝑎 follows truncated normal distribution.

Suppose a random variable z follows a normal distribution. Then we took only the values 𝑧 ≥ 0. The distribution of 𝑧I𝑧 ≥ 0 is called truncated normal and is illustrated right.

• The values of 𝜆𝛼are tabulated

• The interpretation of 𝜆𝛼 is the height of the normal curve at the cutoff point 𝑥𝛼.

• By the normal distribution assumption, the applicants selected with the selection rate 𝛼 are 𝑟*𝜆𝛼/𝛼 standard deviations more productive than population average.

• How much is this in monetary terms?

• Gains from selection = 𝑆𝐷 ⋅ 𝑟*𝜆𝛼/𝛼 , where

  SD = the standard deviation of lifetime productivity distribution (some workers are more productive over their working life than others).

  —>These gains must be set against the costs of selection = 𝑐/𝛼 , where c is per-capita recruitment costs.

• The net utility from selection is 𝑢 = 𝑆𝐷 ⋅ 𝑟 𝜆𝛼/𝛼 − 𝑐 𝛼

  = tradeoff between the gains and costs of selection, and the factors affection this tradeoff. It´s highly non-linear in 𝛼, implying there exists an optimal selection rate than maximizes the benefits of selection net of costs.

• Tougher selection (= lower 𝛼 or higher 𝑐) is more profitable when the test score is more reliable (=higher 𝑟) and ability makes a large different to productivity (=larger SD).

• It can assess the impact of various HR practices affecting recruitment, selection and C&B in terms of the value they deliver through higher productivity of the selected applicants.

• Screening means using an observable indicator that is correlated with performance, to make inferences about a worker's future performance in the firm.

• Naylor-Shine utility model is a useful way of thinking about the roles of toughness of selection (α), screening tool validity (r), productivity gains (SD) and costs (c) of selection, and the likelihood of turnover (p) in shaping the net gains from selection and thereby determining the selection practices firms use.

• Key insights: Tougher selection (i.e. lower α or higher c) is more profitable when the test score is more reliable (i.e. higher r). Ability makes a large difference to productivity (i.e. larger SD), and the likelihood of turnover is small (i.e. lower p).

• Multi-stage selection may be more profitable than single-stage

Signaling:

the worker moves first, sending a credible signal to the firm about his type. The firm pays the worker the wage corresponding to his type. Example: an applicant mentions a university degree in the cover letter. University graduates have, on average, higher ability than non-graduates. Hence, the firm will offer a higher wage to graduates than non-graduates.

Screening:

the firm moves first, putting job applicants to a test that helps identify their type. Example: interviews and probation periods.

—>Screening and signaling can happen simultaneously and be done by both applicants and firm.