Bridging the Pay Gap with Analytics Margrt Vilborg Bjarnadttir - - PowerPoint PPT Presentation

Margrét Vilborg Bjarnadóttir Robert H. Smith School of Business | University of Maryland With David Anderson (CUNY), Cristian L. Dezső (Smith) & David Gaddis Ross (UnFL)

Bridging the Pay Gap with Analytics

People Analytics

Target the right applicants Workforce requirements Employee satisfaction and productivity Identify high- value employees at risk

People Analytics

+ sentiment analysis + voice analysis Understanding the

• rganization

People Analytics

People Analytics

• Increased focus on measuring, reporting and

understanding the workforce

– diversity, gender pay equity, skills gaps, labor utilization, retention rates, real-time feedback,

• rganizational network analysis

Gender Pay Equity

• State/country level legislation

Gender Pay Equity

• State/country level legislation

Why Now?

• State/country level legislation
• Business initiatives

$

AGE

$

AGE

Ln(W) = βX

wages equals A combination of terms that can and should explain wages including ...

βfemale

– the adjusted gender pay gap

And then What?

Reducing the Gap

• The goal is to close (or reduce) the gap while

accounting for operational constraints

• min

i i

s t

• δ
• ln

F iF i i i

c w T i

• δ
• . .

s t

• δ
• 0,

i

i

• δ

Minimize the cost of raises Reach the target goal Do not hand out negative raises

Reducing the Gap

• The goal is to close (or reduce) the gap while

accounting for operational constraints

• min

i i

s t

• δ
• ln

F iF i i i

c w T i

• δ
• . .

s t

• δ
• 0,

i

i

• δ

Cost Savings

The naive approach Optimization

Insights

• Who gets a raise?

– Employees with Low Wages – Women Who “Resemble” Men – Men Who “Typify” Men

Insights

Effect on the pay gap per $ men women A woman who’s raise will negatively affect the paygap A man who’s raise will positively affect the paygap

Insights

Effect on the pay gap per $ men women Underpaid Overpaid Little or no relationship between the residual and pay-gap influence

Ensuring Fairness

• Balancing efficiency with fairness

– Fairness first

• Prioritizing fairness over efficiency

– Balanced approach

• Weighting fairness and efficiency

The naive approach Optimization Fairness first Balanced approach

Cost Savings

Applying Analytics to the Pay Gap

• We formulated the demographic pay gap problem as

a optimization problem

– Provide a provable optimal heuristic to solve it – Insights into which employees influence the gap

• Cost savings

– The optimization approach saves up to 50% of the cost compared to the naïve method – Fairness models typically save between 10 and 20% Empowering the HR analyst

Real Time Data Driven Decision Making

Measurement

Know, understand and quantify the problem

Intercept close

SALARY MODEL RESULTS

Explanatory Variable Value p-value Significance

• 0,061

A gender gap

• f 5.92%

Real Time Data Driven Decision Making

Develop a plan

RESULTS

Measurement

Real Time Data Driven Decision Making

Develop a plan Measurement Stay Vigilant

Understand in real time the effects of hiring and promotional decisions

• How will raises affect my pay-gap?
• How is the pay-gap progressing?
• During a salary review cycle, how do we optimally

balance performance with pay-gap considerations?

Summary

• People Analytics have taken off

– Need for real-time data-driven decision support – We need to translate the math into user friendly tools

• Using smart analytics we can address demographic

pay gaps fairly and efficiently

• The paper: The unintended consequences of On a Firm's Optimal Response to

Pressure for Gender Pay Equity, available at https://ssrn.com/abstract=2798938

• Patent filing: U.S. Provisional Patent Application Number 62/248515
• Business application: PayAnalytics LLC

margret@rhsmith.umd.edu