Entrepreneur: Start & Grow Your Business

An efficiency-based approach on human resource management: a case study of Tainan County fire branches in Taiwan.


by Lan, Chun-Hsiung^Chuang, Liang-Lun^Chang, Chi-Chung
Public Personnel Management • Summer, 2007 •

Introduction

Disaster prevention defines three major missions: fire prevention, disaster rescue and emergently medical service relating to the lives of people. (1) People expect that the fire department can adopt proper countermeasures for disaster prevention and they also hope that fire branches can dispatch rescue teams efficiently to execute their missions. Currently, the performances of fire branches commonly adopt single or few rules to measure the performances of fire prevention, hazardous materials management, disaster rescue and duty supervision, but emergently medical services, educational training, fire investigation and general administration are not taken into consideration. Current performances tend to cause arguments because of lacking objectivity and justice. Therefore, the most important thing to do is to look for a proper measure with objectivity and justice to evaluate the performance efficiencies of fire branches and then adjust reasonable resources for those inefficient fire branches.

In 1978, Harry pointed out that efficiency, effectiveness and productivity are three major parts of performance. (2) In 1988, Fortuin (3) placed the organizational goal in two categories: efficiency and effectiveness. The efficiency is defined as the ratio between input and output, (4) and the effectiveness is defined as the achieving level of the expected production output by a production system. (5) In fact, efficiency and effectiveness represent different levels of performance, and there is no guarantee that both of them can be achieved simultaneously. However, an efficient organization must handle both of them well, and use the most efficient way to pursue maximum effectiveness. (6)

There are many measures of performance evaluation: the Ratio Approach, the Regression Analysis, the Multiple Criteria Analysis, the Analytic Hierarchy Process, the Balanced Scorecard, the Delphi Hierarchy Process, the Total Factor Productivity (TFP), and the Data Envelopment Analysis (DEA). (7) Among these methods, the DEA is the most suitable way to measure the performance efficiency of nonprofit organizations because of its multi-indication character. The performance efficiencies of fire branches have to be reasonably measured by multiple inputs and outputs, and the function relationship between inputs and outputs are unknown in advance. (8) In this research, DEA is selected as the measuring method of performance efficiencies for fire branches because of its characteristic of multi-indication, and thus the relative efficiency of each fire branch can be determined by comparing the quantitative data of inputs and outputs. (9)

The Data Envelopment Analysis (DEA) was proposed from Charnes, Cooper and Rhodes in 1978. Originally, the DEA is applied to measure the performance efficiency of the public or nonprofit organization, but later is applied to many beneficial organizations. The model of DEA is shown by the ratio of output/input and has the same meaning of the so-called TFP. (10) The DEA is based on the concepts of Pareto Optimality and Frontier to calculate the relative efficiencies of the whole decision making units (DMUs) in order to determine their performances, especially for the similar decision making units. (11) In fact, the DEA uses the separated programming via the fractional programming and then transfers the process to linear programming in order to find out the values of the relative efficiencies for the whole decision making units (DMUs) and to determine the inefficient DMUs. (12) This study is trying to measure the relative efficiency for each fire branch under the double duties of disaster prevention and the security of peoples' life and property. Besides, not only can the DEA strengthen the justice on the judgment of performance efficiency for each fire branch and provide an excellent referenced guideline for the resource allocation of each fire branch, but it also can offer the new thinking to measure the performance efficiencies of fire branches.

This study aims to assess the performance efficiencies of fire branches by using 35 fire branches of Tainan County Fire Bureau as an example. Currently, the Tainan County Fire Bureau consists of office duty and field duty sectors, including six sections, one fire center, three fire corps and 35 fire branches. The employees in the fire branch are regarded as long-term workers (13) because each fire branch has to operate all day, and thus the employees in the fire branch should be trained to handle multiple tasks, including water supply, etc. According to relevant laws and ordinances, the manpower of the fire branch is arranged relative to the population and the size of the area. (14) The current allocation of fire protection resources merely considers the location and its associated response time. (15)

The resource allocation for each fire branch has no rules to be followed, and the current allocation depends on the resource distributor; the differing characteristics of the city and country, the governmental budget subsidiaries and the scale of fire branch are not considered. Therefore, the current method often causes a biased assessment of performances. This study, however, considers the aspects of control area, loadings on fire duties and government budget in order to establish a reasonable method to assess the performances of fire branches.

The DEA is conducted in the first-stage of this study. The second stage, according to the future estimated trend of output to select a proper strategy, is (ORA)--where output trend is steady or decreasing--which is also called or Multi-Stage Resource Allocation Approach (MSRAA). In MSRAA, output tendency is increasing. The solutions of ORA are recommended from the contribution index of each input item. The MSRAA is a quantitative approach presented in this study to allocate resources and will be then described in detail in this study. Through the two approaches of ORA and MSRAA, the decision maker can adopt different strategies based on the assessment of the future output trend. Furthermore, these two strategies will function as a referenced guideline to resolve the long-term existing difficulties in a way that reasonably eliminates or allocates resources while making decisions.

The Determination of Input and Output Items

This study focuses on the investigation of the performance efficiency for each fire branch; the production function of DMU is not assumed, so that the DEA is chosen as the assessing measure of performance efficiency in this study. In fact, the DEA includes two different models: Charnes, Cooper and Rhodes, and Banker, Charnes and Cooper. Both of them have two options--input orientation and output orientation. Because a fire branch tries to minimize its input usage of resources to maintain current performance, this study adopts the input-oriented model of Charnes, Cooper and Rhodes to conduct the efficiency analysis for each fire branch. In 1989, Golany and Roll thought that the selection of input and output items was very important while executing DEA. (16) Generally speaking, the determination of the input and output items for the DEA model should be paid more attention. So, the common way to determine the input and output items is to interview with organization officers and then to analyze the organization and management objectives, literature reviews, and experiences. (17)

Therefore, four input items (the number of on-duty personnel, the on-duty cost, the total vehicle displacement and the vehicle maintenance fee) and five output items (the number of fire cases, the number of rescue cases, the number of public service cases, the number of listed fire protected spaces and the number of fire hydrants) were selected as variables in this article for assessing efficiency.

The definition of each variable is given in table 1, and the input and output values for each DMU are listed in table 2. Table 3 describes the correlation coefficiencies between input items and output items of DMUs. From table 3, there exist positive correlations between each input item and each output item. This means that the relationship between each variable complies with the characteristic of "isotonicity," which is the basic assumption of Data Envelopment Analysis. Backward elimination (18) can then be applied in order to delete the input and output items with zero weight in sequence until the weight of each left item is nonzero (i.e. if the weights of input/output items are zero, those items are eliminated). After executing the backward elimination, the previous selected items cannot be deducted from this study. The weights of input and output items for each DMU are listed in table 4.

Empirical Analysis

Frontier software was applied to investigate 35 fire branches of Tainan Fire Bureau in Taiwan by using the input and output data from 2003 to perform the efficiency analysis and potential improvement analysis. The efficiency analysis is described below.

The production efficiency derived from the Charnes, Cooper and Rhodes model of DEA includes the technical efficiency and the scale efficiency. The production efficiency, the technical efficiency, the scale efficiency and the return to scale of each fire branch in Tainan County are listed in table 5. For example, the production efficiency of Baihe branch is 0.6656, its technical efficiency is 0.6775 and the scale efficiency is 0.982. It reveals that the production inefficiency of Baihe branch is mainly due to its technical factor because its technical efficiency (0.6775) is smaller than the scale efficiency (0.982). The analyzing results of DEA for those 35 fire branches in Tainan County are described as follows:

Firstly, the production efficiencies from 14 branches among 35

branches are equal to one. Secondly, regarding to the technical

efficiency, there are 21 fire branches whose technical efficiencies

are equal to one. Thirdly, the scale efficiencies of 14 fire

branches among 35 branches are equal to one. Fourthly, for

analyzing the return to scale, there are three fire branches which

have been categorized into the decreasing return to scale (DRS).

Those three DRS branches mean that they can try to decrease their

scale for efficiency improvement. Fourteen fire branches are in the

category of constant return to scale (CRS); this indicates that

these 14 branches have already reached the optimal production

scale. The 18 fire branches left are in the category of increasing

return to scale (IRS) meaning that those 18 IRS branches can try to

amplify their scales for efficiency improvement. The detailed

information of DRS, CRS and IRS for those 35 fire branches is

listed in table 5.

Resource Strategies

Based on future output trends, this section presents two strategies, the MSRAA and the ORA.

When the future output trend is steady or deceasing, it can be inferred from the potential improved targets and the improved ranges of input/output items that the relatively inefficient units do not require further inputs; on the contrary, they should properly trim their resources. The target values and improved ranges of input/output items for each fire branch are listed in table 6. The ORA strategy is to trim the values of input items on the basis of the DEA report because the output items of each fire branch cannot be changed by us. Taking the Shanhua branch as an example, its present input values (the number of on-duty personnel, the on-duty cost, the total vehicle displacement, the vehicle maintenance fee) are 10, 181.817, 26.002 and 224.685 respectively. The contribution indexes (shown in table 7) of these four input items are 0%, 96.3%, 0% and 3.7% in order, and the target values of input items (shown in table 6) are 8.71, 177.38, 23.23, and 219.2 respectively. Based on the contribution indexes, the ORA strategy recommends that Shanhua branch has to improve "the on-duty cost" to its target value of 177.38 first because its contribution index of 96.3 percent is the largest, then improve "the vehicle maintenance fee" to the target value of 219.2 because such item's contribution index 3.7 percent is just below the contribution index of the on-duty cost. This way, the relative efficiency of the Shanhua branch is improved. The detailed information of contribution indexes for each DMU is listed in table 7.

On the other hand, if the future output trend is increasing, the MSRAA strategy is performed. The aim of MSRAA is to balance the workload of relative efficient branches with other inefficient branches when the future estimated output trend is increasing. Thus, the workloads of those efficient branches can be deducted if resources can be added into them.

The main ideas of MSRAA are as follows:

1. A unit of allocated resources is defined and relative efficient DMUs are grouped. Those efficient DMUs are regarded as candidate DMUs to supplement resources because their work loading is higher than those inefficient branches while the future trend is increasing.

2. The allocated capacity (the number of candidate DMUs) is checked. (If allocable resources are greater than or equal to the allocated capacity, a unit allocated resource is added into each candidate DMU and then one can proceed to step five; otherwise go to the next step.)

3. Compute the total improved performance efficiencies of the whole DMUs by adding a unit allocated resource to a candidate DMU individually, and order these candidate DMUs in accordance with their total improved performance efficiencies. To be more specific, the most improved one is assigned the number "one," followed by the less improved ones--"two," "three," and the like. If a "tie" occurs, a DMU with less resources has the priority.

4. Based on the order of the previous step, resources are assigned to those candidate DMUs one by one according to their order (from the least to the greatest) until the entire allocable resources are exhausted.

5. Calculate the relative efficiencies of the whole DMUs where resources are added and regroup these newer efficient DMUs as the candidate DMUs (the allocated capacity). If remaining resources exist, these remaining resources are considered as the allocable resources for the next step. Now return to the step 2 to check the allocated capacity or go to the final step.

6. Generate the suggested resource allocations in the final step, and the MSRAA is complete.

An Exemplified Case of Resource Strategy

Bryan (19) mentioned that while the population keeps increasing, the losing costs of fire cases including the control cost of fire and the system cost of fire would increase. Therefore, fire branches of Tainan County, where there is a positive population growth, still need to supplement resources. Taking the supplementation of firefighters as an example, there are 20 firefighters to be assigned to fire branches, and a firefighter is determined as an allocated unit. In the first stage, there are 14 efficient fire branches evaluated from the DEA report, and these 14 efficient fire branches (i.e. the allocated capacity is 14) are considered as the initial set for the first stage. Assign 14 firefighters into fire branches in the initial set (each fire branch has assigned a firefighter) because the allocable resources (20 persons) is greater than the allocated capacity (14), and then calculate the relative efficiencies of fire branches after adding 14 firefighters in the initial set for the first stage. The aforementioned calculations are listed in table 8. At this time, the remaining number of firefighters is six (i.e. the allocable firefighters for the next stage is six). From the fluctuation of relative efficiencies, it is found that the Syuejia branch, the Rende branch, and the Gueiren branch would turn into inefficiencies after adding a firefighter for each. Such phenomenon means that the original on-duty firefighters of these three branches before adding firefighters have achieved the optimal scale. Therefore, there are 11 branches (14 - 3 = 11) whose relative efficiencies are still 100 percent after adding a firefighter. Order the above-mentioned 11 branches as the initial set of the second stage, and then the resource allocation (RA) set of the second stage is determined. The allocable resources of the second stage are six firefighters (20 - 14 = 6). Table 9 shows the detailed process to assign the remaining six firefighters consecutively to the most prior six branches in the RA set at the second stage, and then calculate the performance efficiencies of the whole fire branches (35 fire branches). After conducting the MSRAA, these 20 firefighters are reasonably allocated into fire branches that are really in need of resources. After adding those 20 firefighters by using MSRAA, the number of efficient fire branches has been increased from 14 to 18 branches (shown in table 10). It indicates that the workloads of fire branches in Tainan county have been slightly balanced. The entire relative efficiencies of each stage for conducting the MSRAA are listed in table 10.

Conclusions

Nowadays, performance evaluation is an important topic in the field of management sciences. This topic is highly valued by administrative organizations and/or businesses because a great performance is always the guarantee of management. (20) Performance evaluations also are important measures for organizations that want to see how they've executed or organized certain goals, and how they've stimulated the morale and the efficiency of the workplace. The aspects of effectiveness, efficiency and productivity would impact the operation of organizations if resources were not reasonably allocated. Fortunately, the proposed strategies in this study will help to solve such difficult and complicated problems.

By the investigation of the changes of future output trends in this study, decision makers can quickly determine their resource strategy, ORA or MSRAA. If the future estimated output trend is decreasing or steady, the inputs of those relatively inefficient branches require a proper reduction, but how to accurately and reasonably adjust those input items and the priority of them have been fully discussed in this article. The proposed ORA strategy enables decision makers to determine the reduction of resources for each relatively inefficient branch based on the contribution indexes of input items. And, since the future estimated output trend is increasing, resources of input items are encouraged to supplement, and accurately and reasonably allocate resources into right branches has to be considered. Indeed, the MSRAA is capable of reasonably allocating resources. These two resource strategies can help decision makers quickly obtain a referenced guideline for resources allocation on the basis of the relative efficiency for each DMU.

As a matter of fact, a fire branch has the responsibility of public safety, and therefore its scale of fire resources has to be seriously considered. If the input resources of relatively inefficient branches are greatly reduced, the public safety of locality will be affected. Consequently, the decision maker has to consider the appropriate scale of fire resources for each fire branch while performing the ORA strategy, Based on the increasing population of Tainan County, the future fire protection duties will accordingly become heavier. Therefore. the proposed MSRAA strategy in this study will play an important role in the allocation of future fire protection resources. Although this study takes the allocation of fire manpower as an example to demonstrate the MSRAA strategy, the input resources such as the number of fire vehicles (total vehicle displacement), the maintenance fees, and the on-duty cost all can be allocated through the proposed MSRAA strategy. In sum, this study provides an efficiency-based quantitative approach to adjust or allocate fire protection resources, and further establish an executive prototype of the new era to pursue higher efficiency.

Chun-Hsiung Lan

3F, No.221

Nanya W. Rd. Sec.2

Panchiao 220

Taiwan R.O.C.

chlan@mail.nhu.edu.tw

Liang-Lun Chuang

Tainan County Fire Bureau

Tainan County, 730

Taiwan R.O.C.

pb35kimo@yahoo.com.tw

Chi-Chung Chang

Nanhua University

Dalin, Chiayi 622

Taiwan R.O.C.

mgenera1168@yahoo.com.tw

Notes

(1) National Fire Agency Ministry of Interior R.O.C: http://www.nfa.gov.tw.

(2) Harry, P.H. (1978). The Status of Productivity Measurement in the Public Sector. Public Administration Review, 38(1): 28.

(3) Fortuin, L. (1988). Performance Indicators-Why, Where and How, European Journal of Operational Research, 34:1-9.

(4) Farrell, M.J. (1957). The Measurement of Productivity Efficiency, Journal of the Royal Statistical Society, Series A, 120(3): 253-281.

(5) Szilagyi, jr. A.D. (1984). Management and Performance, 2nd ed., New Jersey: Scott, Foresman and Company.

(6) Robbins, S.P. (1994). International Management, 4th ed., New York: Prentice-Hall.

(7) Clarke, R. L. (1992). Evaluating USAF Vehicle Maintenance Productivity Over Time: An Application of Data Envelopment Analysis. Decision Science, 23(2): 376-384.

(8) Banker, R.D. and Morey, R.C. (1986). Efficiency Analysis for Exogenously Fixed Inputs and Outputs, Operations Research, 34(4): 513-521.

(9) Insurance Service Office. (1980). Fire Suppression Rating Schedule, edition 6-80, P30.

(10) Gleason, J.M. and Dariod, T.B. (1982). "Toward Valid Measures of Public Sector Productivity: Performance Measures in Urban Transi," Management Science, 28(4): 237-243.

(11) Banker, R.D., Charnes, A. and Cooper, W.W. (1984). Some Models For Estimating Technical and Scale Inefficiencies in Data Envelopment Analysis, Management Science, 30(9): 1078-1092.

(12) Charnes, A., Cooper, W.W., and Rhodes, E. (1978). Measuring the efficiency of decision making units, European Journal of Operational Research, 2: 429-444.

(13) Traut, OA., Larsen, R., and Feimer, S. (2000). Hanging on or fading out? : Job satisfaction and the long-term worker, Public Personnel Management, 29(3): 343-351.

(14) Schaenman, P.S. (1974). Measuring Fire Protection Productivity in Local Government, Boston: National Fire Protection Association.

(15) Coleman, R.J., Granito, J.A., and Hickey, H.E. (1979), Managing Fire Services International City/County Management Association, 42-43.

(16) Golan, B. and Roll, Y (1989). An Application Procedure for DEA, OMEGA, 17(3): 237-250.

(17) Kao, C. (2000), Data Envelopment Analysis in Resource Allocation: An Application to Forest Management. International Journal of Systems Science, 31(9): 1059-1066.

(18) Hwang, S.N. and Chang, T.Y. (2003). Using Data Envelopment Analysis to Measure Hotel Managerial Efficiency Change in Taiwan, Tourism Management, 24(4): 357-369.

(19) Bryan, J.L. (1979). Managing Fire Service, International City Management Association ICMA, p. 363.

(20) Thompson, J.R. and LeHew, C. (2000). Skill-based pay as an organizational innovation, Review of Public Personnel Administration, 20(1): 20-40.

Chun-Hsiung Lan is a professor and a chairman of the Graduate Institute of Management Sciences at Nanhua University in Taiwan. He earned his Ph.D. at Tamkang University's Department of Management Sciences in Taiwan. His research interests are in the fields of efficiency management, resource strategy, operations research, calculus of variations and soft computing. He has published many international journal papers regarding the aforementioned fields of study.

Liang-Lun Chuang is the director of the Tainan County Fire Bureau's Emergency and Rescue Command Center, located in Taiwan, and a Ph.D. student at the Graduate Institute of Management Sciences at Nanhua University, Taiwan. He received his bachelor's degree from the Department of Fire Sciences at Central Police University, Taiwan in 1990, and completed his master's degree at Nanhua University (Taiwan) in 2004. His research interests lie in the fields of efficiency management and forest fire rescue.

Chi-Chung Chang is a manager of Quick Service Restaurant in Taiwan and a Ph.D. student at the Graduate Institute of Management Sciences at Nanhua University, Taiwan. He received his bachelor's degree from the National Taiwan University of Sciences and Technology, Taiwan, in 1984, and completed his master's degree at Nanhua University (Taiwan) in 2004. His research interests lie in the fields of efficiency management and operations research. Table 1: The Definitions of Input and Output Items

input/ No. output Name of Item Definitions 01 input number of on-duty The monthly average on-duty persons

personnel of the fire branch during the

period of assessment (person) 02 input on-duty cost The business expenses and vehicle

fuel expenses of the fire branch

during the period of assessment

(thousand dollars) 03 input total vehicle The displacement of fire vehicles

displacement of the fire branch during the

period of assessment (cc /1000) 04 input vehicle The maintenance fee of fire

maintenance fee vehicles of the fire branch during

the period of assessment (thousand

dollars) 01 output number of fire The number of fire cases occurred

cases within the control area of the fire

branch during the period of

assessment (case) 02 output number of The number of emergency rescue

emergency rescue cases of the fire branch during

cases the period of assessment (case) 03 output number of The number of public services,

public-service such as bee capture, snake capture,

cases water supply, etc, made by the fire

branch during the period of

assessment (case) 04 output number of listed The total number of protected

fire protected spaces listed in the fire branch

spaces during the period of assessment

(house) 05 output number of fire The number of fire hydrants listed

hydrants in the fire branch during the

period of assessment (hydrant) Table 2: The Values of Input and Output Items for Each DMU DMUs Input 1 Input 2 Input 3 Input 4 Sinying 18 336.165 53.819 449.005 Fire Branch Liouying 6 129.276 18.952 170.932 Fire Branch Yanshuei 7 145.558 18.952 170.932 Fire Branch Baihe 10 223.796 37.953 232.998 Fire Branch Houbi 7 132.648 18.952 179.245 Fire Branch Dongshan 7 138.794 18.952 179.245 Fire Branch Dongyuan 5 99.457 10.726 108.867 Fire Branch Syuejia 7 160.664 37.953 224.469 Fire Branch Jiangjyun 5 89.125 10.726 100.540 Fire Branch Beimen 6 117.883 18.952 169.570 Fire Branch Jiali Fire 8 186.725 29.729 366.802 Fire Branch Sigang 6 99.530 10.726 125.492 Fire Branch Cigu 6 123.401 18.952 199.746 Fire Branch Madou 16 315.491 46.179 316.562 Fire Branch Siaying 6 133.056 18.952 170.932 Fire Branch Lioujia 7 164.292 18.952 179.245 Fire Branch Guantian 7 177.484 18.952 179.392 Fire Branch Shanhua 10 181.817 26.002 224.685 Fire Branch Danei 6 102.197 18.952 179.245 Fire Branch Anding 5 114.020 10.726 83.928 Fire Branch Yujing 7 156.231 18.952 170.932 Fire Branch Nansi 6 130.523 10.726 100.554 Fire Branch Nanhua 6 120.707 10.726 108.867 Fire Branch Sinhua 8 157.979 18.952 170.932 Fire Branch Shansun 10 210.668 18.952 145.994 Fire Branch Gueiren 9 231.614 27.178 270.124 Fire Branch Wunsian 7 164.243 18.364 179.245 Fire Branch Rende 7 184.667 46.179 286.750 Fire Branch Guanmiao 9 218.04 18.952 241.310 Fire Branch Yongkan 9 174.008 16.452 140.756 Fire Branch Sinshih 8 182.156 18.364 162.620 Fire Branch Nanke 13 209.819 23.319 107.506 Fire Branch Dawan 8 195.469 37.953 286.750 Fire Branch Yanhang 10 157.374 18.952 249.623 Fire Branch Fusing 17 289.29 46.179 283.312 Fire Branch DMUs Output 1 Output 2 Output 3 Output 4 Output 5 Sinying 212 1658 175 533 664 Fire Branch Liouying 91 596 87 123 207 Fire Branch Yanshuei 119 762 115 187 237 Fire Branch Baihe 125 810 76 175 232 Fire Branch Houbi 77 545 49 120 134 Fire Branch Dongshan 59 432 45 69 105 Fire Branch Dongyuan 19 224 52 16 35 Fire Branch Syuejia 79 607 83 231 265 Fire Branch Jiangjyun 51 360 26 71 205 Fire Branch Beimen 59 241 69 40 113 Fire Branch Jiali Fire 128 1069 98 210 383 Fire Branch Sigang 81 564 31 109 175 Fire Branch Cigu 102 457 49 80 162 Fire Branch Madou 144 1220 191 242 189 Fire Branch Siaying 53 654 82 84 117 Fire Branch Lioujia 91 575 82 91 122 Fire Branch Guantian 110 668 85 235 187 Fire Branch Shanhua 92 836 120 160 317 Fire Branch Danei 30 253 48 24 63 Fire Branch Anding 99 600 34 156 135 Fire Branch Yujing 38 451 93 95 91 Fire Branch Nansi 16 321 42 56 65 Fire Branch Nanhua 18 212 75 34 54 Fire Branch Sinhua 178 1049 140 280 205 Fire Branch Shansun 40 391 72 55 97 Fire Branch Gueiren 203 1221 105 278 309 Fire Branch Wunsian 141 859 76 216 117 Fire Branch Rende 163 819 128 247 209 Fire Branch Guanmiao 236 798 80 187 238 Fire Branch Yongkan 168 1124 86 160 326 Fire Branch Sinshih 124 917 96 227 139 Fire Branch Nanke 40 342 88 53 327 Fire Branch Dawan 89 1012 67 405 213 Fire Branch Yanhang 119 894 54 484 219 Fire Branch Fusing 157 1263 82 240 182 Fire Branch Table 3: Correlation Coefficiencies Between Input and Output Variables

Number of On-Duty Total Vehicle Vehicle Number

On-Duty Cost Displacement Maintenance of Fire

Personnel Fee Cases Number of 1 0.923 0.733 0.641 0.484 On-Duty Personnel On-Duty Cost 0.923 1 0.812 0.746 0.627 Total Vehicle 0.733 0.812 1 0.850 0.499 Displacement Vehicle 0.641 0.746 0.850 1 0.632 Maintenance Fee

Number of Number of Number of Number of

Emergency Public- Listed Fire Fire

Rescue Service Protected Hydrants

Cases Cases Spaces Number of 0.685 0.634 0.532 0.599 On-Duty Personnel On-Duty Cost 0.774 0.730 0.600 0.607 Total Vehicle 0.682 0.661 0.612 0.578 Displacement Vehicle 0.774 0.631 0.732 0.676 Maintenance Fee Table 4: The Weights of Input and Output Items for Each Fire Branch Fire Branch Input 1 Input 2 Input 3 Input 4 Sinying 0.646659 1.00E-06 0.214788 0.138553 Fire Branch Liouying 2021684 1.00E-06 1.00E-06 0.685738 Fire Branch Yanshuei 1.973753 1.00E-06 1.00E-06 0.610545 Fire Branch Baihe 1.423995 1.00E-06 1.00E-06 0.402548 Fire Branch Houbi 1.00E-06 2.234261 1.00E-06 1.00E-06 Fire Branch Dongshan 1.324998 1.174018 1.00E-06 1.00E-06 Fire Branch Dongyuan 1.00E-06 1.00E-06 5.017617 1.00E-06 Fire Branch Syuejia 1.776252 1.00E-06 1.00E-06 0.618562 Fire Branch Jiangjyun 1.712203 1.00E-06 1.00E-06 2.341881 Fire Branch Beimen 1.00E-06 2.851680 1.00E-06 1.00E-06 Fire Branch Jiali 2.249996 1.00E-06 1.00E-06 1.00E-06 Fire Branch Sigang 1.00E-06 3.125864 0.373861 1.00E-06 Fire Branch Cigu 1.109726 1.716473 1.00E-06 1.00E-06 Fire Branch Madou 1.00E-06 1.00E-06 1.00E-06 1.418376 Fire Branch Siaying 2.999997 1.00E-06 1.00E-06 1.00E-06 Fire Branch Lioujia 2.463639 1.00E-06 0.119034 1.00E-06 Fire Branch Guantian 2.032505 1.00E-06 0.151452 0.391078 Fire Branch Shanhua 1.00E-06 1.781181 1.00E-06 0.073213 Fire Branch Danei 1.00E-06 3.289379 1.00E-06 1.00E-06 Fire Branch Anding 1.00E-06 1.00E-06 1.00E-06 5.349879 Fire Branch Yujing 2.497762 1.00E-06 0.081352 1.00E-06 Fire Branch Nansi 1.00E-06 1.00E-06 5.017616 1.00E-06 Fire Branch Nanhua 1.00E-06 1.00E-06 5.017616 1.00E-06 Fire Branch Sinhua 1.00E-06 1.00E-06 2.839749 1.00E-06 Fire Branch Shansun 1.00E-06 1.00E-06 1.00E-06 3.075498 Fire Branch Gueiren 1.999996 1.00E-06 1.00E-06 1.00E-06 Fire Branch Wunsian 2.217190 1.00E-06