Manzana Insurance
The business problem: Profits are down. Why? Operating Profit Variances Revenues
2Q '89 $ Gross Premiums 8,218 Comm. + Exp. 933 Net Rev. 7,285
% 100.0% 11.4% 88.6%
2Q '91 $ 8,901 1,126 7,775
% 100.0% 12.7% 87.3%
Variances $ % 683 193 1.3% 490 -1.3%
Losses Gross Profit
4,273 3,012
52.0% 36.7%
6,453 1,322
72.5% 14.9%
2,180 (1,690)
20.5% -21.8%
Operating Expenses Operating Profit
1,244 1,768
15.1% 21.5%
1,443 (121)
16.2% -1.4%
199 (1,889)
1.1% -22.9%
18.5% 1.4% 80.1%
2,172 133 6598
24.4% 1.5% 74.1%
Gross Premium Breakdown New Policies Endorsements Renewals
1,523 113 6582
* Plus Program accounts for $97 of operating expense increase.
•Essentially flat revenues (but increasing % from new policies) • Increased losses • Increased commissions & expenses • Increased operating expenses
649 20 16
5.9% 0.1% -6.0%
Some hypotheses & observations •
Increased losses may be due to shift in mix to newer policies – – –
•
Adverse selection: new policies may be initiated by clients “anticipating a loss” To get PLUS bonus, the underwriters may be too lenient in their assessment of risk. Due to poor service, Manzana may be getting only the new policies that other insurers don’t want !
Old policies have a higher contribution net of commissions and expenses.
Policy Margins ('91) Gross Premium Commision Gross Margin
•
% 100% 25% 75%
Old $ 6,205 434 5,771
% 100% 7% 93%
And old policies take less time to process!
RUN RERUN
•
New $ 6,724 1,681 5,043
Dist. 68.5 28.0
UW 43.6 18.7
Rating Writing 75.5 71.0 75.5 50.1
Conclusion: Old policies are the most profitable and least costly to serve, so the high renewal loss rate is driving Manzana’s profits down!
Why the poor turnaround time? Look at utilization: Basic input data
Arrival Rates ('91 6 months)
RUNS no. request % #days rate (#/day)
350 7.5% 120 2.92
rate(#/min)
0.00648
RAPS convert. RAPS* RAINS RERUNS to RUNS 1798 451 2081 274 38.4% 9.6% 44.5% 120 120 120 120 14.98 3.76 17.34 2.28 0.03330
0.00835
0.03854
0.00507
Mean Processing Times (min.) Distribution Clerks Underwriting Raters Policy Writers
RUNS RAPS RAINS RERUNS 68.5 50.0 43.5 28.0 43.6 38.0 22.6 18.7 75.5 64.7 65.5 75.5 71.0 na 54.0 50.1
Utilizations Distribution (4 clerks) Underwriting Teams Territory 1 no. request rate (#/min) Utilization Territory 2 no. request rate (#/min) Utilization Territory 3 no. request rate (#/min) Utilization Rating (8 clerks)
Policy Writing* (5 writers)
RUNS 11.1%
RAPS 41.6%
RAINS RERUNS 9.1% 27.0%
RUNS
RAPS
RAINS
TOTAL 88.8%
RERUNS TOTAL
162 0.0030 13.1%
761 0.0141 53.6%
196 0.0036 8.2%
636 0.0118 22.0%
96.9%
100 0.0019 8.1%
513 0.0095 36.1%
125 0.0023 5.2%
840 0.0156 29.1%
78.5%
88 0.0016 7.1%
524 0.0097 36.9%
130 0.0024 5.4%
605 0.0112 21.0%
70.4%
6.1%
26.9%
6.8%
36.4%
TOTAL 76.3%
RAPS conv. RUNS to RUNS 9.2% 7.2%
RAINS RERUNS TOTAL 64.0% 9.0% 38.6%
* The 274 RAPS converted to RUNS used for Policy Writing utilization. NOTES: 1) Based on Exhibit 7, '91 number of requests 2) Assumes 4 wks/month, 5 days/week, 7.5 hrs/day operation 3) If you assumed the same 28.4 min. weighted avg. processing time in each territory, then the utilizations for territory 1,2 and 3 are (resp.) 92.3%, 83.0% and 70.8%. The above method is more accurate since it accounts for variations in mix across territories.
Observations •
High utilization in distribution (89%) and underwriting (70%-97%)
•
Unbalanced utilization in underwriting makes things worse –
•
Territory 1 has 97% utilization, which is dangerously high
Division of territories on geographical lines eliminates the economy-of-scale benefits of pooling
Economy of scale advantages 50 Separate territories
40
Wq
30
20
Waiting time is lower in pooled system at any given utilization.
s=1 s=2 s=3
Pooled territories
10
0 0.80
0.85
0.90 Utilization ( ρ )
0.95
1.00
Priority and release rules •
RUNS/RAPS/RAINS given priority in underwriting – –
•
RERUNS delayed even more than they would be under FCFS Makes a bad situation worse at underwriting for RERUNS
RERUNS released only 1 day in advance –
– –
Ostensibly to get best information to reevaluate risks, but how much more information is gained in a few days on a policy that has been in force for a year or more? No chance to be on time This leadtime is completely controllable, unlike the RUN/RAP/RAIN leadtime.
Problems with the current quotation policy •
Double counting Example: 5 jobs
3 jobs
1
2
Question: If each job takes one minute to complete at each station, how much time does it take to clear the system? • •
Standard completion time (SCT) too conservative (95th percentile of processing time) The 95th percentile of the sum of 5 random times is MUCH less than the sum of the 95th percentile of each time (statistical averaging)
Manzana is quoting itself out of business!
A reasonable course of action •
Pool underwriting teams to take advantage of economies of scale
•
Keep RERUNS low priority, but release them a week or so in advance so they have a chance of being on time (exact leadtime requires analysis) – –
•
RUNS/RAPS and RAINS are time-sensitive work and cannot be delayed RERUNS are not time sensitive if they are released far enough in advance (“background work”)
Develop a realistic TAT quotation policy.
Some simulation results: FIFO, no pooling RUNS
RERUNS
Simulation results: FIFO with pooling RUNS
RERUNS
Simulation results: Priority for RUNS with pooling RUNS 1-Day turnaround time is actually feasible!
RERUNS
Releasing RERUNS 4 days prior to due date is more than enough to guarantee 100% ontime!
Manzana Insurance: Key Lessons •
Diagnose profitability by customer segment .. – – – –
contribution margin (commissions) acquisition costs (PLUS program) indirect costs (e.g. insured losses) workload (e.g. processing time)
Often, repeat customers are the most profitable! (Zero Defections) •
Most managers do not understand the causes and effects of queuing, and this can lead to very bad decisions – – – –
•
staffing for maximum utilization ignorance of pooling economies (geographic organization) wrong priorities (time-sensitivity vs. profitability as basis for priorities, e.g. RUNS vs. RERUNS) release policies (RERUNS) & due date setting (TAT calculation)
A simple analysis of utilization + knowledge of queues can go a long way toward diagnosing and solving leadtime performance problems.
