Nearly three quarters (72%) of the Fire Department"s $210 million budget, and just over 1,500 (79%) of its 1,900 budgeted FTEs for FY 2001-02 are devoted to fire suppression and emergency medical services. Thus, planning for the employee resources to perform these services is a crucial activity that ensures adequate resources for maintaining high-quality service in the community.

The Department"s original model for calculating required staffing was only recently brought into use. This section of the report examines the Department"s staffing model, its underlying assumptions, and procedures for application of the model. More specifically, this section of our report does the following:

· Reviews various components of the Fire Department"s staffing model, the basis for overall staffing requirements, and specific factors that help determine the necessary amount of relief for fixed-post positions;

· Compares the Department"s model with a "best-practices" model;

· Evaluates, then recommends appropriate relief factors for different fixed-post positions; and

· Discusses externalities that might positively or negatively influence results of the model.

It is important to note that much of the data and accompanying analyses in the report are based on the Department"s initial model. The model used for budget purposes in FY 2001-02 incorporates subsequent revisions made by the Department after concerns were shared by the Budget Analyst during the annual budget review.

However, it is also important to note that the model has multiple applications. It can be used to determine the Department"s staffing requirements, and assist with the determination of the Department"s annual budget requirement. It can also be used to gauge historical leave trends by employees., . While historical trending should also take into account adjustments for any changes of which the Department is aware -such as those resulting from MOU negotiations, labor-market conditions and other factors -historical trends provide the most accurate indicators of future workforce behavior. Finally, and perhaps one of its most powerful roles, is the model"s use as a tool for managing absences that drive the Department"s overtime cost. This is the feature with the greatest potential for cost savings in the Fire Department.

Overview of the Fire Department"s Staffing Model

The Fire Department"s original staffing model represents the first effort by the Department to establish a rational process for estimating staffing needs. It is a sophisticated financial tool that encompasses a broad range of service, classification and pay categories. Most importantly, the model is a welcome addition to the assortment of techniques available to help control Departmental expenditures -especially overtime. We found this effort to be highly creditable and sought to validate and, if necessary, strengthen the model for its continued use.

One component of the model is shaped by sections of the Department"s labor agreements that deal with minimum daily staffing requirements. The positions included in this grouping are fixed-post positions, meaning that the position must be filled 24 hours per day, seven days per week, and 365 days per year. The Chief of the Department determines the level of service the Department will provide and the organizational structure through which that service is provided. Staffing for the desired level of service flows from the Chief"s determination.

At this writing, a total of 354 uniformed personnel from different operating divisions are required on each shift to staff the units assigned to the Department"s 42 stations and fireboat (details of this deployment are provided in Table 2.2.1, below). Included in the group of 354 uniformed personnel are those assigned to Emergency Medical Services functions. However, the Central Emergency Communications Center (CECC) personnel, also referred to as "Radio," are not included in either direct fire suppression activities or in the model despite the fact that the Radio Unit contains fixed-post positions and currently utilizes uniformed personnel. (A separate model was developed subsequently for "Radio" by the Department.) Finally, the model factors-in a variety of approved leaves to help arrive at an overall Relief Factor.

Table 2.2.1
San Francisco Fire Department
Minimum Daily Staffing Requirement

The initial application of this model yielded the following results for the FY 2001-02 budget process:

· An FTE estimate of 3.51 was established as the underlying requirement to fill one position continuously for 24 hours daily, 365 days per year;

· This 3.51 FTE per fixed-post position was derived mathematically, whether using work hours or work days, as follows:

· Relief factors of 8 percent for vacation, 7 percent for sick pay, 6 percent for disability pay, and 4 percent for comp time -totaling 25 percent-- were generated by the model based on a combination of then-current experience in the first three of these leave categories and Departmental practice in the remaining one;

· These factors were calculated by dividing the variable average daily absences in selected approved-leave categories by the fixed Minimum Daily Staffing Quota of 352 for fire suppression activities (the number originally used, but is now revised to include two Arson positions);

· The combined relief factor of 1.25 applied to the base of 1,235.52 FTEs (3.51 x 352 Daily Staffing Quota) meant that 1,544 FTEs are required to fully staff fire suppression activities, while at the same time accommodating a certain level of expenditure for approved leaves and overtime. While this is somewhat more than the 1,490 uniformed positions authorized in the current-year Suppression budget, the approved budget contains a reduction in uniformed attrition savings equivalent to an increase of 41.58 FTEs, which brings the total FTEs much closer to the number generated by the model as the requirement (12 positions less than calculated using the Department"s assumptions).

· Including overtime and premium pay, an approved budget in excess of $151 million.

Our approach to validating this original model is described in the following section of this report.

Comparison of Model with "Best Practices"

The Fire Department"s original staffing model is more accurately described as a staffing utilization model. It defines uniformed personnel of different classifications and ranks in terms of the numbers required for fire suppression activity on a continuous, year `round, day-to-day basis. Positions in this kind of setting are generally referred to as fixed-post, meaning that relief is required when there is an absence, to ensure uninterrupted coverage. Such positions are most commonly found in public safety areas of law enforcement, emergency medical services, fire protection, and related communications systems.

Working with the Controller"s Office, the Budget Analyst performed a management audit of the City and County of San Francisco"s 911 Emergency System (1991), and addressed some of the issues associated with its fixed-post staffing. A major technical document used regularly in the Budget Analyst"s office is the "Staffing Analysis Workbook for Jails,"¹ published by the National Institute of Corrections. While this document focuses on custody operations, the model is directly transferable to any operation which requires continuous coverage. We found the Fire Department"s model to be generally consistent with our own direct experience analyzing fixed-post staffing, as well as with the basic construction and intent of the material in the Workbook. While some specific aspects of the Fire Department"s model differ, the methods used by the Department are acceptable and generally accomplish the same purpose. These are described, below.

· It was unnecessary for the Department to develop a staff coverage plan for fire suppression activities because the minimum daily staffing requirement is explicitly set forth by SFFD management and reinforced through current labor agreements;

· Similarly, as work shifts are published annually in the "21-Tour Watch Chart," the Fire Department utilizes net annual work hours, a convention that facilitates the calculation of full-time equivalents (FTEs);

· The model encompasses all of the different classifications and ranks involved in fire suppression activities;

· The Department utilizes actual experience from an existing internal attendance and reporting system as the basis for determining its various relief requirements; and

· Vacation -both annual and intermittent, sick leave, compensatory time and disability leave are taken into account for relief-factor calculations.

On the other hand, there are some aspects of the Department"s model that are inconsistent with the standard practices in other jurisdictions.

· The model did not incorporate all leave categories, or accurately account for leave used; and, did not incorporate the impact of limited duty work restrictions on overall staffing needs; and,

· The model did not include impacts on staff availability that result from the assignment of uniformed personnel to the Consolidated Emergency Communications Center (CECC).

The Department tracks all other approved leave types -e.g., jury duty, personal leave and military leave. While use of these miscellaneous leave categories does not occur in a substantial quantity, it is sufficient to make a statistical difference and it warrants inclusion in the model. Therefore, we disagree with the Department"s exclusion of these leave types from the model.

Emergency Medical Services (EMS) positions are fixed-post and included in the model, as noted above. This involves aggregation of the EMS personnel leave experience with that of firefighters, when there is evidence to suggest that it might be different (see a subsequent section of this report for details).

At this writing, EMS personnel work the same shifts as firefighters. However, different proposals are now under consideration for the EMS work shifts to change. Irrespective of the outcome regarding work shifts, the model should apply to this group because it involves fixed-post positions. If the work shifts do change, the model can accommodate any workweek configuration. Given the nature and volume of calls for emergency medical services and the corresponding level of activity for EMS personnel, the leave experience for this group should be tracked separately.

As noted previously, positions in Radio also are of the fixed-post variety. However, they are not included in the model because they are indirectly rather than directly involved in field suppression activities. In addition, the work shifts for this group are different from those of firefighters. While uniformed personnel are currently assigned to the Radio Group, conversion to civilian positions has always been planned. However, irrespective of the eventual staffing classifications that might be used, the staffing model should currently include the positions assigned to the Radio group because the assignment impacts the number of staff available for other duties. Similar to the EMS group, leave experience for Radio should be tracked separately from the suppression group. After sufficient longitudinal data are collected for Radio, a determination can be made regarding appropriate factors for relief.

In summary, we compared the Department"s staffing model to a combination of the literature and our experience, and determined that the Department"s model is readily usable with some recommended modifications. The following sections of this report address issues pertaining to the range of relief factor calculations, alternate calculations, and different levels of leave experience by different groups.

Calculating Relief Factors

The Department maintains, through its Assignments Office, an automated system for tracking staff attendance, leaves of all types, and rank and location assignments. Various reports are routinely prepared from these data and the system is capable of generating many types of custom reports as well. While we made full use of both report types for this audit, the principal report relied upon for validation of the staffing model was the Daily Average Absences Report (DAAR). Exhibit 2.2.1 provides a sample of the monthly summary.

This report shows actual activity in each leave category as represented by specific headings in the report and, based on source data from the Assignments Office, sets forth the numbers of uniformed Department personnel with suppression-related assignments, the numbers scheduled and at work each day, and the numbers of employees absent for a variety of approved leaves. This report shows the number of employees who were hired on their scheduled day off to work overtime (WDO). Data in the Daily Average Absence Report, shown in Exhibit 2.2.1, provide the foundation and other critical elements for most of the calculations in the model. We obtained these same data from the system covering the past seven years. Our trend analysis focused on the entire seven years as well as the period covering the last four years in which the impact of the paramedics" merger with the Fire Department was most apparent.

Table 2.2.2, below, provides an overview of the principal DAAR components with actual amounts and percentages entered for the FY 2000-01 daily averages. A narrative description of the relief factor equation and calculation follows.

Table 2.2.2
San Francisco Fire Department Staffing Components Used for Calculating Relief Factors

Definitions:

A - Includes all eligible suppression-activity employees who are on payroll and expected (assigned) to report for duty on a given day.

B - Contains the various categories of approved leave that are shown in detail in Exhibit 2.2.1.

C - Represents the difference between A and B.

D - The MSR results from the Chief of Department"s determination of service level and the MOU deployment specifications.

The MSR represents the number of personnel the Department is committed to make available for duty at 8 A.M. and 8 P.M. each day. When the number of employees on duty before WDOs is less than the MSR, employees on their scheduled day off can be hired to make up the difference. Such employees, who indicate their availability in advance, are selected on the bases of rank and seniority to fill specified vacancies. They are referred to as WDOs, or (employees) Working (on their) Day Off. WDOs are paid overtime at the rate of time and one-half for these assignments.

As stated previously, the DAAR Report tracks employee absences for those employees who are scheduled for work. By dividing the number of scheduled employees into the number of employees who are absent from work, the percentage of employees who are absent can be determined. Simply stated, this calculation provides an estimate of the average amount of leave taken by each employee on the payroll. In a fixed post environment (such as the Fire Department), this calculation can also be viewed as a "relief factor" (i.e., the number of additional employees who must be employed to perform the duties of staff when they are absent). For example, a relief factor of 25% means that for every four persons that are scheduled for duty, a fifth person must be employed to cover for vacation, sick and other leave, if the post is to be staffed on a 24-hour, 365 day basis.

When calculating the relief factor, the Department divided the average number of employees on leave each day by the minimum staffing requirement. This mathematical approach is valid. However, the Department"s estimates of employee leave patterns were higher than those historically experienced by the Department, overstating the Department"s relief staffing need by approximately 4.0%.

The Department used certain leave category figures that were for short time segments (only eight months of data were used to calculate vacation usage), and they produced relative high percentages in several categories (vacation, IDV/TC). Using the minimum staffing requirement number of 352 as the divisor, the Department computed a total relief factor of 25 percent.

By way of comparison, the Budget Analyst used the same absence categories for two comparisons, one based on the four-year average and another based on the most recent fiscal year. This resulted in relief factors of 21 percent for each of the analyses, which was four percent less than the factor computed by the Department. These computations are shown in Table 2.2.3, below.

However, the Department did not include a relief computation for light duty personnel, which if added would have increased the Department"s total relief factor by four percent, to approximately 29 percent. This overage is significant when considering that the base number of personnel required to provide daily minimum staffing of 354, exceeds 1,242 employees.

A 4.0% variance in the relief factor, when applied to this base, equates to approximately 50 employees. At an average annual cost of $85,000 per employee, the assumptions used by the Department overstated the Department"s staffing need by over $4.2 million. Since the Department continues to budget for light duty employee relief with overtime, this difference in assumptions would have resulted in an excess budget amount for FY 2001-02 of over $4.2 million.

The Budget Analyst included light duty as an absence category to demonstrate the impact on the average daily WDO amount by accounting for all factors affecting overtime usage. Table 2.2.3 shows how selected leave category assumptions can affect the outcome of the model.

Table 2.2.3
Comparisons of SFFD Relief Factor
Computations Using Alternative Assumptions

As shown in the table, actual leave usage patterns for vacation, sick, disability and other categories resulted in a 21 percent relief factor, depending on whether four-year or one year data totals are used in the computation. These relief factor computation results compare with the 25 percent relief factor calculated by the Department using the same leave categories for a shorter period (8-months). When absence from scheduled assignment due to light duty is also factored in, the computation of the relief factor based on four-year and one-year data totals increases to 24% and 25%, respectively, while the Department"s would have increased to approximately 29%. As illustrated previously, this results in a substantial increase in the need for uniformed relief. Further, it can result in an inflated estimate of overtime needs.

Another comparison that we conducted focused on WDOs, the principal overtime driver. This comparison was intended to help determine if there was a leave usage trend or pattern based on seasonal influences. We compared WDOs in the first half of the fiscal year with those in the second half over the past seven years (see Exhibit 2.2.2). Within that period, four of the years showed the second half of the year having a higher incidence of average daily WDO usage than in the first half of the year. But the amounts and percentages of difference were small -except in FY 1999-2000. On the other hand, for three of the seven years analyzed, the first half showed a higher incidence of average daily WDO usage than in the second half, and all of the differences were substantial. However, nothing in the WDO data suggests a need for special treatment of information in the Average Daily Absence reports as a general approach to calculations for the Department"s staffing model.

It is important that the methodology used by the Department be mathematically precise, and conform with the best practices for calculating relief factors. The benefits of such precision and conformity are described below:

· A longer, organizationally relevant data timeline smoothes out trends and protects the Department from anomalous spikes in either direction,

· The model results are more defensible when clear, simple and comprehensive,

· The model can be applied with confidence, others can be easily trained in its use, and

· Future applications and analyses are based on consistently defined, and updated data.

In Section 1.7 of this report, the Budget Analyst recommends specific targets for three categories of leave and a targeted range for WDOs. These targets are suggested for use in the model to illustrate how it serves as a management tool to attack absenteeism and help control overtime spending. Exhibit 2.2.3 provides an analysis of historical relief factors and the targets recommended by the Budget Analyst.

Other Leave Experience

Because leave experience for different groups may differ, it is important to segregate the analysis for major worker groupings. For example, the Department"s overall amounts of sick and disability leave usage clearly jumped following the 1997 paramedic merger, but it was unclear whether the increase resulted from disproportionately high use by paramedics that joined the Department, or if there was simply a general usage increase by all field suppression personnel. While leave information for EMS personnel was combined with firefighters" data in system records prior to January 2001, the data was not as robust or reliable as system data and access going forward from January 2001. Therefore, it was difficult to distinguish the specific source of increases in this case.

Leave histories from the DAAR include paramedics, which provides a reasonable surrogate for calculating relief factors for other EMS personnel. However, there are two compelling reasons for tracking the use of leave for EMS workers separately in the future: (1) the paramedic function is fundamentally different from that of firefighting, and paramedic leave usage patterns as a group may actually be different from those of firefighters; and, (2) in the event that management changes paramedic work shifts in the future, the calculations for FTEs and relief factors can be based wholly on paramedic experience rather than an aggregate experience that may or may not be directly comparable. Note that our relief factor calculations only included H3 Paramedic/Firefighters and H33 Rescue Captains.

Positions in the Central Emergency Communications Center (CECC) are fixed-post positions. Persons in these positions work 12-hour shifts and a different workweek cycle than do firefighters⁴. As fixed-post positions, there is a need to calculate their required FTEs and relief factors separately. Leave experience for Radio should be tracked separately from both EMS and the suppression group.

Data limitations notwithstanding, we examined data covering the last half of the fiscal year to obtain some abbreviated sense of the Radio Group"s leave experience. From January through June 2001, it appears that the Radio group experienced a lower incidence of total absences than the Suppression unit (17% vs. 21%), with the most noticeable differences occurring in the areas of disability (1% vs. 5%) and limited duty (0 vs. 2%).

Uncertainty over demographic comparability of Radio and Suppression obscures any conclusions that might be drawn, so with total absences at the 17% level and the base FTE requirement of 4 per fixed-post position, it seems that the Radio group should have between 103 and 112 budgeted positions. This range covers the 105 required staff positions recommended by the Budget Analyst in an earlier audit.⁵ Further, we understand that the current staffing level in Radio of 54 persons is considered by the Fire Department to be adequate. However, we were unable to fully reconcile filled and budgeted positions, the number of Radio personnel on extended leave and the amount of overtime used by this unit. Again, we recommend that this area be tracked separately and examined more closely by the Department.

Beyond the longitudinal deficiency of these data, the overriding issues are that

· Even with limited data, it is clear that leave patterns vary for different groups of workers. Therefore, for purposes of calculating FTEs and relief factors, tracking and reporting leave activity should be done by individual group rather than aggregated for the whole workforce;

· The impact of the combined EMS and suppression leave usage on the calculated relief factor is minimal at this time because of the relative size of the two groups.

Conclusions

The San Francisco Fire Department uses an automated staffing model to determine its minimum daily staffing requirement for the 354 fixed-post positions that must be filled on a daily basis. The model incorporates an overall relief factor that reflects employees" use of vacation, sick, disability, compensatory time off, military and other approved leave.

Our evaluation of the original model revealed the following : (a) it does not incorporate significant variables, such as the average number of employees scheduled for duty, and limited-duty assignments into its calculations for relief; and, (b) some of the variables used in the model overstated the Department"s historical use patterns.

By (a) including limited-duty assignments and all other leaves, (b) either using more accurate estimates of the number of employees scheduled for duty, or consistently applying alternative calculations, (c) estimating more realistic, time-sensitive leave averages, and (d) separately tracking leave usage for EMS and Dispatch personnel, the staffing model can become a more accurate instrument.

By implementing these recommendations, the Department would be provided with a more reliable basis for staff planning and projecting overtime cost driven by employee leave. Estimated budget savings from changes to the original model would total approximately $4,224,500 per year.

Recommendations

The San Francisco Fire Department should:

2.2.1 Continue use of the different automated staffing models to calculate the overall relief factor and FTEs required for all fixed-post positions in Suppression, Emergency Medical Services and the Radio Group;

2.2.2 Utilize all leave categories in calculating the relief factor.

2.2.3 Utilize historical data from the Daily Average Absence Report, based on averages covering the previous four year period.

2.2.4 Separately track and report on attendance and leave usage for personnel in Emergency Medical Services and in the Radio Group.

2.2.5 Prepare written procedures that describe construction and application of the models sufficient for other Finance staff to appropriately and correctly make use of them.

2.2.6 Work with staff in the Assignments Office and Human Resources Division to ensure accuracy of leave data captured in the Department"s automated system.

Costs and Benefits

There would be no cost to implement these recommendations.

The Department would have a more reliable basis for staff planning budgeting, and would be able to more accurately project overtime costs. Budget savings from implementing the recommendations would be at least $4.2 million annually based upon changes from the Department"s original model.

Exhibit 2.2.1
San Francisco Fire Department
Daily Average Absences Report-Summary by Month

Exhibit 2.2.2
Comparisons of Average Daily WDOs in
First and Second Halves of Fiscal Year

Exhibit 2.2.3

Exhibit 2.2.3 provides an analysis of the major categories available in the Daily Average Absence Report with the focus on those used by the Department, the numbers and percentages for both seven- and four-year averages, and the recommendations of the Budget Analyst for use in the Department"s staffing model.

Analysis of Relief Factors and
Budget Analyst"s Recommendation

¹ The first edition was published in 1988 and the second edition issued in March 2001.

² Includes use of vacation (annual and intermittent daily), sick, (compensatory) time coming, disability and limited duty.

³ Includes jury duty, military, disciplinary, bereavement, personal, workers compensation and educational.

⁴ Employees in the Radio Group have a 42-hour work week based on the following: employees work 12-hour shifts, from 0700 hours to 1900 hours and from 1900 hours to 0700 hours, three days on and three days off, and the cycle is repeated every six weeks.

⁵ Management Audit of the CCSF 911 Emergency System, 1991 (with the Controller"s Office)