Government memo, 2022-04

EFTA00141334¶

FEDERAL BUREAU of PRISONS¶

Risk Analysis of Overtime, Augmentation, and Incentive Payment¶

April 2022¶

EFTA00141335¶

Table of Contents¶

Overview¶

Executive Summary¶

Background & Current Processes¶

Summary of Approach¶

Overtime & Augmentation Analysis Correlation Chart¶

Section 1: Overtime Usage, Connections¶

Current Usage of Overtime(Overall Snapshot, Reasons for Overtime) Connection:Augmentation Not Associated with Overtime

Connection:Incidents Not Associated with Overtime

Connection:Leave(Sick&AWOL)Associated with Overtime

Connection:Vacancies Not Associated with Overtime

BOP's Overtime Tool

Section 2: Augmentation Usage, Connections¶

Overview: Overtime and Augmentation

Current Usage of Augmentation(Overall Snapshot)

Connection: Incidents Not Associated with Augmentation

Connection: Leave(Sick and AWOL)Not Associated with Augmentation

Connection: Vacancies Not Associated with Augmentation

Section 3: BOP Incentive Analysis¶

Summary of BOP Incentives Overview of Insights¶

Recruitment & Relocation Incentive Usage¶

Retention Incentive Usage – Findings and Insights¶

Connection: Retention Incentives Not Associated with Staff Separations¶

Connection: Unclear if Current Retention Incentives are Cost Effective¶

Connection: Retention Incentives Vary Across Institutions, Vary Within Institutions¶

Case Study: North Central Region (NCR) Psychology Services¶

Appendix¶

Summary of Analysis & Findings¶

Recruitment Incentive Usage Summary¶

Relocation Incentive Usage Findings¶

OBJECTIVE¶

To review and analyze the risks associated with BOP’s increased usage of overtime and augmentation and analyze the effectiveness of recruitment, relocation and retention incentives.¶

EFTA00141336¶

Executive Summary¶

This document highlights the bureau of Prisons’ usage of overtime, augmentation and incentives (from 2017-2021) to uncover potential drivers of, and effectiveness of, usage.¶

Key Takeaways¶

Trends of both Overtime and Augmentation have both been increasing over time, disputing the hypothesis that they are used to substitute each other¶

85% of incentive payments are focused on retention incentives¶

Retention incentives do not appear to reduce staff separations¶

3¶

EFTA00141337¶

Background & Current Processes¶

In response to staffing issues, the Government Accountability Office (GAO) published a report in Feb. 2021 outlining shortfalls across the bureau. This analysis seeks to satisfy the recommendation of the bureau to conduct a risk assessment of its overtime and augmentation use on its staff, inmates, and institution security, and assess the outcomes of the incentives it utilizes.¶

CURRENT PROCESS¶

Overtime¶

• Overtime reports can be pulled from the Roster Scheduling Software that is viewable at the Central, regional, and local level

• The Roster Scheduling Program makes improvements on some current processes associated with assigning overtime (e.g., current training certifications are captured in program, so it is clear who is qualified for a shift, and Overtime Authorization forms are automatically saved in the system)

• Correctional Programs Division (CPD) developed a separate Overtime Tool housed in SAS that tracks overtime spending by institution, region, and overall agency

• The Overtime Tool is accessible at the regional level. Regional leaders have the authority to share this information at the local level (usually institution Wardens and Captains). The report used for the Overtime tool is updated by local administrators on a monthly basis

Augmentation

·The responsibility of tracking augmentation is at the local level; how accurately the augmentation time code is utilized depends on each institution ·Some institutions utilize an augmentation log to track the reason and the frequency of staff augmentation to provide transparency to institutional staff¶ ·Posts that are augmented for part of a shift may not get coded as being augmented, potentially resulting in an under-reporting of the practice

¶

Incentives¶

• Wardens request incentives from their respective regional office who approve incentives at their own discretion

• Funding for incentives comes from an institution’s operational budget

• There does not appear to be a standard policy bureau-wide that articulates reasoning behind retention incentive variances (for the same position at the same institution) or ties current retention incentives to quantifiable metrics defined in terms of an agency’s goals (per OPM’s guidance)

• Incentives are sent to Staffing and Employee Relations (SERS) to be reviewed annually as required by OPM

OBJECTIVES OF ANALYSIS¶

Uncover risks associated with increased use of overtime hours per staff¶

Uncover risks associated with increased use of augmentation hours per staff¶

Analyze the usage of recruitment, relocation, and retention incentives¶

4¶

EFTA00141338¶

Summary of Approach¶

Our approach centered on testing several hypotheses to best uncover potential risks associated with BOP’s increased use of overtime and augmentation and usage of incentives.¶

	Section 1: Overtime Risk Analysis	Section 2: Augmentation Risk Analysis	Section 3: Incentive Usage Analysis
Data Inputs	·Overtime and Augmentation hours for all BOP staff by institution and region(FY17-FY21) ·Inmate related incidents by institution and region(FY17-FY21) ·Sick and AWOL leave by institution and region(FY19-FY21) ·Vacancies across all institutions(FY21)	·Per-person hiring, relocation and retention incentive spending for FY17-FY21 across all 122 BOP institutions ·BOP exit survey data 2016-2020 ·Cost to hire*
Hypotheses	·Overtime and Augmentation usage are used to substitute each other(i.e., as one increases,the other decreases) ·The institutions with the highest overtime and augmentation usage are the most short-staffed ·Institutions with high levels of overtime and augmentation experience more safety incidents ·Institutions with increased usage of overtime and augmentation utilize more sick leave and absent without leave(AWOL)	·Retention incentives reduce staff separations ·Retention incentives are connected to hard-to-fill locations and positions
Analyses Conducted	1. Review BOP's current overtime and augmentation per staff yearly trend overall and regionally 2. Review BOP's current overtime hours by custody and non-custody staff 3. Test relationship of overtime and augmentation between inmate incidents,sick&AWOL usage,and vacancies with regression analysis(The strength of relationships is scored usingR².In statistics,R²is the proportion of the variation in the dependent variable that is predictable from the independent variable.For the purposes of these analyses,anR²value of.50or greater is considered a strong relationship) 4. Review BOP's current Overtime Tool to address gaps and opportunities for improvement	1. Review summation of incentives given to staff 2. Breakdown recruitment,relocation,and retention incentives by percent of staff and the top positions receiving them 3. Analyze retention incentive variance of top BOP positions receiving incentives 4. Calculate average retention incentive spending per staff 5. Test relationship between separation rate and retention incentive spending with regression analysis
Limitations	·Unable to retrieve workplace injuries date to test relationship with overtime and augmentation ·Vacancy data was limited to2021and could not assess prior years ·Unable to retrieve number of programs delayed/cancelled to test relationship with augmentation	·Unable to retrieve cost associated with hiring new staff at BOP to compare to cost of staffing incentives	*Source:Automatic Data Processing(ADP)

5¶

EFTA00141339¶

Overtime & Augmentation Analyses Correlation Chart¶

This chart encompasses the three types of analyses conducted against Overtime and Augmentation data. Analyses that proved a correlation are accompanied with a check mark, while analyses that did not show a correlation are accompanied with an X.¶

Analyses Conducted Against Overtime and Augmentation Data	Correlation with Overtime?	Correlation with Augmentation?
Inmate Incidents	X	X
Sick Leave and Absent Without Leave(AWOL) Usage	√	X
Vacancies	X	X

When Overtime and Augmentation were compared against one another, it did not appear that overtime and augmentation are used in substitution.¶

6¶

EFTA00141340¶

Overtime Risk Analysis¶

7¶

EFTA00141341¶

Current Usage of Overtime – Overall Snapshot¶

A review was conducted using overtime usage data from 2017-2021 to assess the usage trend across all institutions. Below are the initial observations of usage overall and by region.¶

Key Findings¶

• Overtime usage (hours) has increased significantly from 93 hours per staff in 2017 to 182 hours per staff in 2020*

• While the overtime usage has gone down minimally from 2020, the 2021 average is approximately 95% higher from its 2017 average

• The region with the highest overtime usage (hours) per staff in 2020 is South Central, but the Western region had the highest percent increase from 2017 to 2020

• In 2021, the top 5 institutions with the highest Overtime per staff were and Big Spring FCI, Brooklyn MDC, Coleman Complex, Forrest Complex, and Yazoo City Complex and the highest vacancies were Beaumont, Butner Complex, Florence Complex, Thomson USP, and Yazoo City Complex

*The increase of overtime from 2019-2020 could have been due to COVID-19¶

8¶

EFTA00141342¶

Current Usage of Overtime – Reasons for Overtime¶

A review was conducted using overtime usage data from 2017-2021 to assess the usage trend across all institutions. Below are the initial observations of overall usage broken out by reasons for overtime (custody, outside hospital, other).¶

Key Findings¶

• Custody overtime hours of 963 thousand in 2017 was relatively low compared to outside hospital hours of 1,822 thousand*

• However, over the years the gap between the two has been closing

• In 2019, custody overtime hours surpassed outside hospital by 22%

• In 2021, custody overtime hours surpassed outside hospital by 37%

• Other reasons for overtime have remained relatively low compared to custody and outside hospital, however it would be beneficial to understand the full reasons for overtime

*This data is solely based on reported overtime hours in total and not on the full reasoning behind use of overtime¶

9¶

EFTA00141343¶

Connection: Augmentation Not Associated with Overtime¶

An analysis of overtime was conducted using overtime hours and augmentation hours for individual institutions for the time period of 2019. The purpose was to test the hypothesis of: Overtime and Augmentation usage are used to substitute each other (i.e., as one increases, the other decreases).¶

Real world data significance requires $ R^{2} > . 50 $ for consideration. Each blue dot represents an institution.¶

Key Findings¶

• Assessing augmentation and overtime by fiscal year 2019, the $R^{2}$ value is .20 – Meaning it is not significant enough to conclude that there is a strong relationship

• Although the relationship is not strong, it is leaning towards a positive one. Since institutions have anecdotally relayed that overtime and augmentation are used to substitute each other, it would be anticipated that they would have a negative relationship (as one increased, the other decreases)

• When assessing the 5-year usage trend analysis for overtime and augmentation, they both follow the same trend—both have increased over time

Connection

It does not appear that overtime and augmentation are used in substitution.

10¶

EFTA00141344¶

Connection: Incidents Not Associated with Overtime¶

Risk analysis of overtime was conducted using overtime hours and inmate incidents data $ _{1} $ for individual institutions for the time period of 2019. The purpose was to test the hypothesis of: Institutions with high levels of overtime experience more incidents.¶

Real world data significance requires $ R^{2} > . 50 $ for consideration. Each blue dot represents an institution.¶

Key Findings¶

• Assessing at the incidents and overtime by fiscal year, the $ R^{2} $ value is .29 – Meaning it is not significant enough to conclude that there is a strong relationship

• While an increase in overtime use can lead to incidents, it cannot be concluded that it is a direct affect of overtime with the data provided

• Additionally, it is acknowledged that incidents represent an extreme outcome of increased institution safety risks—meaning that there might still be increased safety risks with increased overtime usage, even if those do not connect to an increase in incidents

Connection¶

There does not seem to be a relationship between inmate incident and staff overtime.¶

1 Inmate incident data is a consolidated report of allegations, verified assaults, and minor incidents for inmate-on-inmate and inmate-on-staff incidents. Time period of 2019 was used to factor out COVID-19. The strength of relationships is scored using R2. In statistics, R2 is the proportion of the variation in the dependent variable that is predictable from the independent variable. For the purposes of these analyses, an R2 value of .50 or greater is considered a strong relationship.¶

11¶

EFTA00141345¶

Connection: Leave (Sick & AWOL) Associated with Overtime¶

Risk analysis of overtime was conducted using overtime hours of all institutions and leave $ _{1} $ usage during the same time period of 2019. The purpose was to test the hypothesis of: Institutions with increased usage of overtime utilize more sick leave and absent without leave (AWOL).¶

Real world data significance requires $ R^{2} > . 50 $ for consideration. Each blue dot represents an institution.¶

Key Findings¶

• Assessing the leave and overtime by institutions, the $ R^{2} $ value is .52 – Meaning it is significant enough to conclude that there is a strong relationship

• There is a positive relationship between overtime and leave usage meaning when overtime is high, a similar pattern can be seen with leave usage (Although it cannot be said for certain it is a one-for-one relationship of overtime and leave usage as other factors such as staff tenure need to be considered)

• An increase of overtime could mean that more staff call out, which could be a result of burn out or decreased morale

Connection¶

Instances of institutions with increased usage of overtime could be an indicator of increased sick and AWOL leave usage.¶

1 This analysis includes sick and AWOL leave hours. Time period of 2019 was used to factor out COVID-19.¶

12¶

EFTA00141346¶

Connection: Vacancies Not Associated with Overtime¶

Risk analysis of overtime was conducted using overtime hours of all institutions and vacancies during the same time period of 2021. The purpose was to test the hypothesis of: The institutions with the highest overtime usage are the most short-staffed.¶

Real world data significance requires $ R^{2} > . 50 $ for consideration. Each blue dot represents an institution.¶

Key Findings¶

• Assessing the vacancies and overtime by institutions, the $ R^{2} $ value is .18* – Meaning it is not significant enough to conclude that there is a strong relationship

• There does not seem to be a relationship between overtime hours and vacancies

• Since overtime is used to supplement the shortage of staff, there might be other driving factor(s) leading to increase overtime use (i.e., staff callouts)

*The weak relationship between Overtime and Vacancies could be due to a position not being filled as a result of a staff member on extended leave or staffing guidelines not being the most up to date¶

Connection¶

Vacancies are not directly associated with overtime hours.¶

*The regression analysis was based on availability of vacancy data¶

13¶

EFTA00141347¶

BOP’s Overtime Tool¶

BOP created a tool in SAS to retroactively track overtime spending budget of its facilities going back as far as 2009. Below is an overview of its current capabilities and the opportunities for improvement.¶

Current Capabilities and Features¶

High-level coding of reasons for overtime (Custody, Outside Medical, Other)¶

Ability to filter by agency, regionally, and facilities to drill down further¶

Track overtime spending regionally and/or by facilities to assess spending against their allocated budget¶

Access to Regional Directors for awareness on overtime spending¶

Data pulled from local administrators that pull from financial system (UFMS) on a monthly basis¶

14¶

EFTA00141348¶

Augmentation Risk Analysis¶

16¶

EFTA00141349¶

Current Usage of Augmentation – Overall Snapshot¶

A review was conducted using augmentation usage data $ _{1} $ from 2017-2021 to assess the usage trend across all institutions. Below are the initial observations of usage overall and by region.¶

Key Findings¶

• Following a similar pattern of overtime, augmentation usage (hours) has also increased significantly from 7 hours per staff 2017 to 13 hours per staff 2021

• While the augmentation usage has gone down minimally from 2020, the 2021 average is approximately 86% higher from its 2017 average

• The region with the highest augmentation in 2021 is North Central, but the Western region had the highest percent increase from 2017 to 2021

• In 2021, the top 5 institutions with the highest Augmentation per staff were and Berlin FCI, Sheridan FCI, Thomson USP, Waseca FCI, and Williamsburg FCI and the highest vacancies were Beaumont Complex, Butner Complex, Florence Complex, Thomson USP, and Yazoo City Complex

$$ _{1} \text{Augmentation data might be underrepresented as its dependent on the augmentation code input } $$¶

17¶

EFTA00141350¶

Connection: Incidents Not Associated with Augmentation¶

Risk analysis of augmentation was conducted using augmentation hours and incidents $ _{1} $ data for individual institutions for the time period of 2019. The purpose was to test the hypothesis of: Institutions with high levels of augmentation experience more incidents.¶

Real world data significance requires $ R^{2} > . 50 $ for consideration. Each blue dot represents an institution.¶

Key Findings¶

• Assessing at the incidents and augmentation by fiscal year, the $ R^{2} $ value is .13 – Meaning it is not significant enough to conclude that there is a strong relationship

• There does not seem to be a relationship between augmentation hours and number of incidents that are inmate driven

• Although, there is no relationship found, there could be a risk to safety of institutions as incidents are only one factor considered.

Connection

There does not seem to be a relationship between inmate incident and staff augmentation.

$$\text{1} \text{Inmate incident data is a consolidated report on allegations, assaults, and minor incidents for inmate-in-inmate and inmate-on-staff. Time period of 2019 was used to factor out COVID-19.}$$¶

18¶

EFTA00141351¶

Connection: Leave (Sick and AWOL) Not Associated with Augmentation¶

Risk analysis of overtime was conducted using augmentation hours of all institutions and leave $ _{1} $ usage during the same time period of 2019. The purpose was to test the hypothesis of: Institutions with increased usage of augmentation utilize more sick leave and absent without leave (AWOL).¶

Real world data significance requires $ R^{2} > . 50 $ for consideration. Each blue dot represents an institution.¶

Key Findings¶

• Assessing at leave and augmentation by institutions, the $ R^{2} $ value is .09 – Meaning it is not significant enough to conclude that there is a strong relationship

• There does not seem to be a relationship between augmentation hours and leave usage

• Since augmentation might be underreported (it is unplanned and does not impact operational budget the way that overtime does), the leave usage relationship could be understated as it would be anticipated that augmentation would follow a similar trend of overtime and leave

There does not seem to be a relationship between sick and AWOL leave usage and staff augmentation.¶

$$ _{1} \text{Leave in this analysis includes sick and AWOL leave hours. Time period of 2019 was used to factor out COVID-19.} $$¶

19¶

EFTA00141352¶

Connection: Vacancies Not Associated with Augmentation¶

Risk analysis of overtime was conducted using augmentation hours of all institutions and vacancies during the same time period of 2021. The purpose was to test the hypothesis of: The institutions with the highest augmentation usage are the most short-staffed.¶

Real world data significance requires $ R^{2} > . 50 $ for consideration. Each blue dot represents an institution.¶

Key Findings¶

• Assessing the vacancies and augmentation by institutions, the $ R^{2} $ value is .13 – Meaning it is not significant enough to conclude that there is a strong relationship

• There does not seem to be a relationship between augmentation hours and vacancies

• Since augmentation is used to supplement the shortage of staff, there might be other driving factor(s) leading to increase augmentation use (i.e., staff callouts)

Connection¶

Vacancies are not directly associated with augmentation hours.¶

The regression analysis was based on availability of vacancy data¶

20¶