Manzana Insurance

Operations Management ____________________________________________________________ _________________________ Case Study Manzana Insurance [pic] Presented to: Prof B. Mahadevan Submitted By Group 12 (Section B) Peeyush Razdan (0811115) Shalekh Banka (0811124) Shalem Anand Tirkey (0811125) Shreshth Sharma (0811128) Sumeet Mittal (0811131) Manzana Insurance As per the case facts, Manzana Insurance’s Fruitvale branch is the least performing branch and the senior VP seeks a report on the same. Their competitor Golden Gate (backed by its corporate parent generated a price war to gain market) is performing much better in terms of most metrics of Insurance business. 1) Problems faced by Manzana Insurance (Fruitvale branch) a) High Turn Around Time (TAT) Agents in the insurance sector are mediators who act as an interface between the client and the insurer. Hence, the quality of service, measured by average TAT, to the agents (who are common to the competitors) is of primary importance. TAT for Fruitvale has deteriorated to 6 days (1991) from 5 days (1990), while Golden Gate’s offer of 1 day TAT is luring agents away from Manzana. The number of late renewals is also increasing and quite high compared to Golden Gate. b) Geographic/Territorial Allocation to Underwriting Teams Leads to an Uneven Task Distribution We observe that the geographic allocation of agents to the underwriting teams is not optimal. This method has an inherent problem; there can be a surge in requests from a geography which might overload that particular team while the other teams might be idle due to lack of requests from their geography. Hence, their pooling should be rather than using the geographic allocation. c) Primary focus on RUNs rather than RERUNs and Increasing late Renewals Currently, Fruitvale is focusing on RUNs for getting new customers and compromising the service towards the RERUN requests for the existing customers. Loss of focus on RERUNs has led to increase in its TAT causing agents to move towards Golden Gate. The number of late Renewals has increased from 20% to 44% over the last year which has led to a significant increase in the Renewal loss rate from 33% to 47%. Renewal is a low premium – high volume business. Exhibit 5(in case) shows that a new policy and policy renewal give similar revenues of $6724 and $6205 respectively. The commission given to agents is 25% and 7% respectively. So in fact renewal of old insurance brings more revenue due to reduced commission percentage. Also, the time taken for renewal of policies is less, thus more renewals can be done in a given time. d) FCFS Scheduling may not be always optimum Most of the departments follow FCFS scheduling which may not be optimum. Ex. a policy with an effort of 4 hrs would delay all the other smaller policies behind it, effectively delaying 10 simple requests for serving one difficult one. Moreover, the scheduling policy followed across various departments is not same. While some strictly follow FCFS others do it on type of policy ex: RUN preferred over RERUNs. Also, The RAP is given preference over RERUNs by the rating team. The RAPs in fact take more time for processing than RERUNs. Also, only 15% of all quotes result in new policies. Thus, the Manzana Fruitvale branch seems to be servicing RAPs at the cost of RERUNs. e) Huge backlog of policies The existing backlog of policies is quite high, due to which any new policy received is not attended instantly (due to FCFS) further adding on to existing backlog, essentially creating a cascading effect. 2) Process Flow and Capacity Analysis We have used process flow, capacity analysis, Lead Time Analysis and Average daily Workload Analysis to analyze the current situation at Manzana’s Fruitvale Branch. ) Process Flow Exhibit 1 shows the process flow diagram with the existing capacity across the four major team Distribution clerks, Underwriter teams, Raters, Policy writers. b) Capacity Analysis Usage of Mean time for capacity analysis is ideally favored over 95% Standard Completion Time (SCT) parameter as 95% SCT is a highly conservative. It assumes every request to be of longer duration (95th percentile with respect to request duration) and does not take into acco unt the time saved while executing smaller requests. Mean on the other hand takes into account that fact that time consumed on longer jobs is compensated by the time saved on smaller jobs. Capacity analysis based on Mean Time (Exhibit 2) shows the bottleneck for RUNs is the Distribution step, for RAPs is underwriting, for RAINs is again Distribution and for RERUNs is policy writing. The same capacity analysis when done using 95% SCT (Exhibit 3) shows Underwriting step to be bottleneck for all the 4 types of policy requests. We observe that there is high standard deviation for most of the steps (especially Underwriting Step) and is comparable to mean value itself. Ex: For RERUNs Underwriting step mean is 18. 7min while the standard deviation is 19. 8min. Hence we are using 95% SCT for determining bottleneck step. And thus we consider Underwriting Stage as the bottleneck for the whole system. c) Lead (Service) Time Analysis Using Little’s Formula we have done Lead Time Analysis (Exhibit 4) which shows that on an average Lead Time is approximately 2 days (2. 10). As we have seen, throughput on the other hand is approximately 6 days which is much higher than the average Lead Time. This suggests that the longer throughput time is because of allocation problems described later. d) Average Daily Workload Analysis The argument of allocation problem is further strengthened by our Average Daily Workload Analysis (Exhibit 5). This analysis has been done taking into account the relative proportions of various types of requests on a daily basis. On the basis on mean processing time for all the steps it comes out that the average daily workload for each step is less than 7. 5 hrs (the stipulated working time). Inferences Above analysis shows that problem at Fruitvale branch is not due to capacity constraint but it is actually due to allocation issue. Predominantly, allocation problems are in underwriting step. Currently allocation is done on the basis of Geographical/territorial lines which lead to uneven workload over time as well as inefficient allocation leading to longer TPUT time. Exhibit 5 shows that RUNs on an average take 50% (Exhibit 6) more processing time than RERUNs hence territory1 which is heavily loaded with RUNs has a higher RERUN loss even though overall it is least loaded. This necessitates a better allocation scheme rather than FIFO. Shortest Job First Scheduling may be used to minimize TPUT but it may delay a high priority request (ex a RERUN close to expiry) and also requires a priori estimation of various time factors. Currently, the system follows FCFS though not strictly, since some departments prioritize based on type of request ex: Underwriting favors RUNs over RERUNs. We would suggest an improved priority scheduling over FIFO which would be determined based on the type of request, agent’s importance, client importance, expiry time etc. et across uniformly across the steps. 3) Recommendations a) Distribution of reports to teams based on priority rather than territorial As found in previous section the current workforce should suffice the existing requirements if there are no backlogs and even if there are, it should not take more than 2 days. The problem was identified in the allocation of the policy request based on territory. We recomm end a collective (instead of territorial) request processing system with more intelligence added to the distribution system, which were also identified as bottlenecks for RUN and RAIN. This distribution system would first prioritize the requests and then allocate them to the underwriting team on an optimal basis ensuring even distribution of total requests, individual requests RUN, RERUN etc. This would remove the existing anomaly of having an overloaded team and an idle team concurrently. They should also keep monitoring the progress of the process and remove blocking issues that may result in a further delay of other policies. To expedite this process, it can also be automated by the help of computers. These changes should substantially reduce the TAT and the delayed RERUNs. The priority should be decided based on the following factors instead of current system purely on the type of request 1) Type of request – RERUN, RUN, RAP, RAIN 2) Already waited time – request that have waited for a long time should be upgraded 3) Estimate of total time required 4) Agent’s priority 5) Client’s priority 6) Expiry Time (Incase of RERUNs) 7) Other factors determining the business value of the request Hence overall, this would reduce the intensity of bottlenecks at DC and UT. b) Increase focus on RERUNs RERUNs, which generate maximum revenue, should be given a higher priority based on their proximity to the expiry date. This ensures customer retention and gives sufficient time to agents for renewal. c) Reduce Backlogs Reduction in current backlogs to minimum will help Fruitvale achieve a day’s TAT as explained. This can be done by working overtime, getting temporary teams possibly from other branches, and increase number of working days temporarily. d) Use Mean instead of SCT We observe that SCT is not a correct approach to evaluate the process flow since the 95th percentile customer’s time considered as a benchmark to set up the system results in very conservative estimates. Initially, other statistical methods like mean + n times standard deviation etc. can be used to achieve desired level of service. Following that, a record including TAT for each request should be maintained so that it can be used for future reference and for reducing the noise in mean that we recommend to be used in future. Exhibits Exhibit 1: Process Flow Diagram Exhibit 2: Capacity Analysis based on Mean Processing time |Capacity |RUNs |RAPs |RAINs |RERUNs | |Distribution |26. 8 |36. 00 |41. 38 |64. 29 | |Underwriting |30. 96 |35. 53 |59. 73 |72. 19 | |Rating |47. 68 |55. 64 |54. 96 |47. 68 | |Policy Writing |31. 69 |NA |41. 67 |44. 91 | *darkened cells show the bottleneck Exhibit 3: Capacity Analysis based of 95% Standard Completion Time Capacity |RUNs |RAPs |RAINs |RERUNs | |Distribution |14. 05 |16. 70 |26. 43 |41. 67 | |Underwriting |12. 59 |15. 43 |27. 33 |21. 50 | |Rating |32. 06 |40. 59 |40. 27 |39. 05 | |Policy Writing |25. 20 |NA |31. 65 |33. 58 | *darkened cells show the bottleneck Exhibit 4: Lead (Service) Time Analysis using Little’s Formula |Requests in progress |82 | |Requests served per day (in steady state) |39 | |Lead /Service Time (Requests in progress/Requests served per day) |2. 1 | Exhibit 5: Average Daily Workload Analysis |Demand Analysis |Requests in 120 days (1991) |Demand per Day | |RUNs |624 |5. | |RAPs |1524 |12. 7 | |RAINs |451 |3. 758333 | |RERUNs |2081 |17. 34167 | |Total |4680 |39 | |Mean Processing Time |RUNS |RAPS |RAINS |RERUNS |Workers/Teams | |Â   |(5. 0) |(12. 70) |(3. 75) |(17. 34) |Total Time |hrs per day | |Distribution |356. 20 |635. 00 |163. 13 |485. 52 |1,639. 85 |6. 83 | |Underwriting |226. 72 |482. 60 |84. 75 |324. 26 |1,118. 33 |6. 21 | |Rating |392. 60 |821. 69 |245. 63 |1,309. 17 |2,769. 09 |5. 77 | |Policy Writing |369. 20 |N. A. |202. 50 |868. 73 |1,440. 43 |4. 0 | Exhibit 6: Uneven distribution due to territorial allocation |Policies |Total |% diff with avg |RUNs |% diff with avg |RERUNs |% diff with avg |Rene wal lost | |/Territory | |1315 | |208 | |693 | | |1 |1151 |14. 24% less |274 |31. 73% more |636 |8. 2% less |403 | |2 |1393 |5. 93% more |179 |13. 94% less |840 |21. % more |227 | |3 |1402 |6. 66% more |171 |17. 7% less |605 |14. 54% less |296 | | | | | | | | | | | | | | | | | | | | | | | | | | | | Processing Time (using mean values) | |Distribution |Underwriting |Rating |Writing |Total |Baselined (w. r. t. minimum) |Baselined (w. r. t. RERUN) | |RUN |68. 5 |43. 6 |75. 5 |71 |258. 6 |169. 35 |150. 09 | |RAP |50 |38 |64. 7 |NA |152. 7 |100. 00 |88. 62 | |RAIN |43. 5 |22. 6 |65. 5 |54 |185. 6 |121. 55 |107. 72 | |RERUN |28 |18. 7 |75. 5 |50. 1 |172. 3 |112. 84 |100. 00 | | ———————– Distribution clerks Number : 4 Capacity: Runs 26. 27 (58. 63), Raps 36 (72. 28), Rains 41. 37(195. 65), Reruns 64. 28(290. 32) Underwriter teams Number: 3 Capacity: Runs 30. 96(12. 59), Raps 35. 52(15. 42), Rains 59. 73(27. 32), Reruns 72. 19(21. 49) New Requests, Renewal requests Request for underwriting Policy writers Number: 4 Capacity: Runs 31. 69(25. 19), Raps NA, Rains 41. 67(31. 64), Reruns 44. 91(33. 58) Raters Number: 4 Capacity: Runs 47. 68(32. 05), Raps 55. 64(40. 58), Rains 54. 96(40. 26), Reruns 47. 68(39. 04) Policy issuing request Rating request Manzana Insurance Operations Management ____________________________________________________________ _________________________ Case Study Manzana Insurance [pic] Presented to: Prof B. Mahadevan Submitted By Group 12 (Section B) Peeyush Razdan (0811115) Shalekh Banka (0811124) Shalem Anand Tirkey (0811125) Shreshth Sharma (0811128) Sumeet Mittal (0811131) Manzana Insurance As per the case facts, Manzana Insurance’s Fruitvale branch is the least performing branch and the senior VP seeks a report on the same. Their competitor Golden Gate (backed by its corporate parent generated a price war to gain market) is performing much better in terms of most metrics of Insurance business. 1) Problems faced by Manzana Insurance (Fruitvale branch) a) High Turn Around Time (TAT) Agents in the insurance sector are mediators who act as an interface between the client and the insurer. Hence, the quality of service, measured by average TAT, to the agents (who are common to the competitors) is of primary importance. TAT for Fruitvale has deteriorated to 6 days (1991) from 5 days (1990), while Golden Gate’s offer of 1 day TAT is luring agents away from Manzana. The number of late renewals is also increasing and quite high compared to Golden Gate. b) Geographic/Territorial Allocation to Underwriting Teams Leads to an Uneven Task Distribution We observe that the geographic allocation of agents to the underwriting teams is not optimal. This method has an inherent problem; there can be a surge in requests from a geography which might overload that particular team while the other teams might be idle due to lack of requests from their geography. Hence, their pooling should be rather than using the geographic allocation. c) Primary focus on RUNs rather than RERUNs and Increasing late Renewals Currently, Fruitvale is focusing on RUNs for getting new customers and compromising the service towards the RERUN requests for the existing customers. Loss of focus on RERUNs has led to increase in its TAT causing agents to move towards Golden Gate. The number of late Renewals has increased from 20% to 44% over the last year which has led to a significant increase in the Renewal loss rate from 33% to 47%. Renewal is a low premium – high volume business. Exhibit 5(in case) shows that a new policy and policy renewal give similar revenues of $6724 and $6205 respectively. The commission given to agents is 25% and 7% respectively. So in fact renewal of old insurance brings more revenue due to reduced commission percentage. Also, the time taken for renewal of policies is less, thus more renewals can be done in a given time. d) FCFS Scheduling may not be always optimum Most of the departments follow FCFS scheduling which may not be optimum. Ex. a policy with an effort of 4 hrs would delay all the other smaller policies behind it, effectively delaying 10 simple requests for serving one difficult one. Moreover, the scheduling policy followed across various departments is not same. While some strictly follow FCFS others do it on type of policy ex: RUN preferred over RERUNs. Also, The RAP is given preference over RERUNs by the rating team. The RAPs in fact take more time for processing than RERUNs. Also, only 15% of all quotes result in new policies. Thus, the Manzana Fruitvale branch seems to be servicing RAPs at the cost of RERUNs. e) Huge backlog of policies The existing backlog of policies is quite high, due to which any new policy received is not attended instantly (due to FCFS) further adding on to existing backlog, essentially creating a cascading effect. 2) Process Flow and Capacity Analysis We have used process flow, capacity analysis, Lead Time Analysis and Average daily Workload Analysis to analyze the current situation at Manzana’s Fruitvale Branch. ) Process Flow Exhibit 1 shows the process flow diagram with the existing capacity across the four major team Distribution clerks, Underwriter teams, Raters, Policy writers. b) Capacity Analysis Usage of Mean time for capacity analysis is ideally favored over 95% Standard Completion Time (SCT) parameter as 95% SCT is a highly conservative. It assumes every request to be of longer duration (95th percentile with respect to request duration) and does not take into acco unt the time saved while executing smaller requests. Mean on the other hand takes into account that fact that time consumed on longer jobs is compensated by the time saved on smaller jobs. Capacity analysis based on Mean Time (Exhibit 2) shows the bottleneck for RUNs is the Distribution step, for RAPs is underwriting, for RAINs is again Distribution and for RERUNs is policy writing. The same capacity analysis when done using 95% SCT (Exhibit 3) shows Underwriting step to be bottleneck for all the 4 types of policy requests. We observe that there is high standard deviation for most of the steps (especially Underwriting Step) and is comparable to mean value itself. Ex: For RERUNs Underwriting step mean is 18. 7min while the standard deviation is 19. 8min. Hence we are using 95% SCT for determining bottleneck step. And thus we consider Underwriting Stage as the bottleneck for the whole system. c) Lead (Service) Time Analysis Using Little’s Formula we have done Lead Time Analysis (Exhibit 4) which shows that on an average Lead Time is approximately 2 days (2. 10). As we have seen, throughput on the other hand is approximately 6 days which is much higher than the average Lead Time. This suggests that the longer throughput time is because of allocation problems described later. d) Average Daily Workload Analysis The argument of allocation problem is further strengthened by our Average Daily Workload Analysis (Exhibit 5). This analysis has been done taking into account the relative proportions of various types of requests on a daily basis. On the basis on mean processing time for all the steps it comes out that the average daily workload for each step is less than 7. 5 hrs (the stipulated working time). Inferences Above analysis shows that problem at Fruitvale branch is not due to capacity constraint but it is actually due to allocation issue. Predominantly, allocation problems are in underwriting step. Currently allocation is done on the basis of Geographical/territorial lines which lead to uneven workload over time as well as inefficient allocation leading to longer TPUT time. Exhibit 5 shows that RUNs on an average take 50% (Exhibit 6) more processing time than RERUNs hence territory1 which is heavily loaded with RUNs has a higher RERUN loss even though overall it is least loaded. This necessitates a better allocation scheme rather than FIFO. Shortest Job First Scheduling may be used to minimize TPUT but it may delay a high priority request (ex a RERUN close to expiry) and also requires a priori estimation of various time factors. Currently, the system follows FCFS though not strictly, since some departments prioritize based on type of request ex: Underwriting favors RUNs over RERUNs. We would suggest an improved priority scheduling over FIFO which would be determined based on the type of request, agent’s importance, client importance, expiry time etc. et across uniformly across the steps. 3) Recommendations a) Distribution of reports to teams based on priority rather than territorial As found in previous section the current workforce should suffice the existing requirements if there are no backlogs and even if there are, it should not take more than 2 days. The problem was identified in the allocation of the policy request based on territory. We recomm end a collective (instead of territorial) request processing system with more intelligence added to the distribution system, which were also identified as bottlenecks for RUN and RAIN. This distribution system would first prioritize the requests and then allocate them to the underwriting team on an optimal basis ensuring even distribution of total requests, individual requests RUN, RERUN etc. This would remove the existing anomaly of having an overloaded team and an idle team concurrently. They should also keep monitoring the progress of the process and remove blocking issues that may result in a further delay of other policies. To expedite this process, it can also be automated by the help of computers. These changes should substantially reduce the TAT and the delayed RERUNs. The priority should be decided based on the following factors instead of current system purely on the type of request 1) Type of request – RERUN, RUN, RAP, RAIN 2) Already waited time – request that have waited for a long time should be upgraded 3) Estimate of total time required 4) Agent’s priority 5) Client’s priority 6) Expiry Time (Incase of RERUNs) 7) Other factors determining the business value of the request Hence overall, this would reduce the intensity of bottlenecks at DC and UT. b) Increase focus on RERUNs RERUNs, which generate maximum revenue, should be given a higher priority based on their proximity to the expiry date. This ensures customer retention and gives sufficient time to agents for renewal. c) Reduce Backlogs Reduction in current backlogs to minimum will help Fruitvale achieve a day’s TAT as explained. This can be done by working overtime, getting temporary teams possibly from other branches, and increase number of working days temporarily. d) Use Mean instead of SCT We observe that SCT is not a correct approach to evaluate the process flow since the 95th percentile customer’s time considered as a benchmark to set up the system results in very conservative estimates. Initially, other statistical methods like mean + n times standard deviation etc. can be used to achieve desired level of service. Following that, a record including TAT for each request should be maintained so that it can be used for future reference and for reducing the noise in mean that we recommend to be used in future. Exhibits Exhibit 1: Process Flow Diagram Exhibit 2: Capacity Analysis based on Mean Processing time |Capacity |RUNs |RAPs |RAINs |RERUNs | |Distribution |26. 8 |36. 00 |41. 38 |64. 29 | |Underwriting |30. 96 |35. 53 |59. 73 |72. 19 | |Rating |47. 68 |55. 64 |54. 96 |47. 68 | |Policy Writing |31. 69 |NA |41. 67 |44. 91 | *darkened cells show the bottleneck Exhibit 3: Capacity Analysis based of 95% Standard Completion Time Capacity |RUNs |RAPs |RAINs |RERUNs | |Distribution |14. 05 |16. 70 |26. 43 |41. 67 | |Underwriting |12. 59 |15. 43 |27. 33 |21. 50 | |Rating |32. 06 |40. 59 |40. 27 |39. 05 | |Policy Writing |25. 20 |NA |31. 65 |33. 58 | *darkened cells show the bottleneck Exhibit 4: Lead (Service) Time Analysis using Little’s Formula |Requests in progress |82 | |Requests served per day (in steady state) |39 | |Lead /Service Time (Requests in progress/Requests served per day) |2. 1 | Exhibit 5: Average Daily Workload Analysis |Demand Analysis |Requests in 120 days (1991) |Demand per Day | |RUNs |624 |5. | |RAPs |1524 |12. 7 | |RAINs |451 |3. 758333 | |RERUNs |2081 |17. 34167 | |Total |4680 |39 | |Mean Processing Time |RUNS |RAPS |RAINS |RERUNS |Workers/Teams | |Â   |(5. 0) |(12. 70) |(3. 75) |(17. 34) |Total Time |hrs per day | |Distribution |356. 20 |635. 00 |163. 13 |485. 52 |1,639. 85 |6. 83 | |Underwriting |226. 72 |482. 60 |84. 75 |324. 26 |1,118. 33 |6. 21 | |Rating |392. 60 |821. 69 |245. 63 |1,309. 17 |2,769. 09 |5. 77 | |Policy Writing |369. 20 |N. A. |202. 50 |868. 73 |1,440. 43 |4. 0 | Exhibit 6: Uneven distribution due to territorial allocation |Policies |Total |% diff with avg |RUNs |% diff with avg |RERUNs |% diff with avg |Rene wal lost | |/Territory | |1315 | |208 | |693 | | |1 |1151 |14. 24% less |274 |31. 73% more |636 |8. 2% less |403 | |2 |1393 |5. 93% more |179 |13. 94% less |840 |21. % more |227 | |3 |1402 |6. 66% more |171 |17. 7% less |605 |14. 54% less |296 | | | | | | | | | | | | | | | | | | | | | | | | | | | | Processing Time (using mean values) | |Distribution |Underwriting |Rating |Writing |Total |Baselined (w. r. t. minimum) |Baselined (w. r. t. RERUN) | |RUN |68. 5 |43. 6 |75. 5 |71 |258. 6 |169. 35 |150. 09 | |RAP |50 |38 |64. 7 |NA |152. 7 |100. 00 |88. 62 | |RAIN |43. 5 |22. 6 |65. 5 |54 |185. 6 |121. 55 |107. 72 | |RERUN |28 |18. 7 |75. 5 |50. 1 |172. 3 |112. 84 |100. 00 | | ———————– Distribution clerks Number : 4 Capacity: Runs 26. 27 (58. 63), Raps 36 (72. 28), Rains 41. 37(195. 65), Reruns 64. 28(290. 32) Underwriter teams Number: 3 Capacity: Runs 30. 96(12. 59), Raps 35. 52(15. 42), Rains 59. 73(27. 32), Reruns 72. 19(21. 49) New Requests, Renewal requests Request for underwriting Policy writers Number: 4 Capacity: Runs 31. 69(25. 19), Raps NA, Rains 41. 67(31. 64), Reruns 44. 91(33. 58) Raters Number: 4 Capacity: Runs 47. 68(32. 05), Raps 55. 64(40. 58), Rains 54. 96(40. 26), Reruns 47. 68(39. 04) Policy issuing request Rating request