INSTITUTE AND FACULTY OF ACTUARIES

EXAMINATION

4 October 2019 (am)

Subject CP2 – Modelling Practice

Core Practices
Paper 2

SUMMARY

Institute and Faculty of Actuaries

Objective
The CEO of ABC parcel delivery company has requested some support to understand the potential
profitability for a parcel delivery company, under a number of scenarios.
There are currently four drivers who have an allocation of parcels, with a minimum number per day and a
maximum amount they can deliver. Payment for deliveries is different per driver. An additional driver is
available to deliver any surplus parcels not allocated to any of the original four drivers.
ABC have a total of 550 parcels to be delivered per day and using a random scenario generator need to
assess how often they are likely to achieve their target profitability.
The scenarios considered are:
- Scenario 1:
o Allocating parcels to drivers in order of the cheapest driver first
o Allocating parcels to drivers in order of the maximum potential number of parcels they
will deliver
- Scenario 2:
o Repeat above assuming drivers I and II have car issues and have to reduce the number of
deliveries by half.
- Scenario 3:
o Repeating the initial calculations assuming that the number of parcels to be delivered
increases from 550 to 650.
- Scenario 4:
o Changing the charging structure, so all drivers paid the same amount – determining this
amount in order to achieve the target profit 100% of the time.

Data
The following data was provided by CEO of ABC parcel delivery company.
- Number of parcels to be delivered per day: 550

Driver Price paid per parcel Minimum number of Maximum number of

($) parcels parcels
I 2.25 40 50
II 1.25 100 500
III 3.5 75 200
IV 3 100 700
V (only if any surplus 3 N/A N/A
parcels)

- Amount ABC receives per parcel: $5

- Target profit per day: $1,450

The statistics department within the Actuarial firm provided 400 random numbers, 4 sets of 100 numbers
for each driver.
A check on the data has been completed that the random numbers are between 0 and 1 – this is shown in
the chart below:

Check on data between 0 and 1

1.2

0.8

0.6

0.4

0.2

0
0 20 40 60 80 100 120

Assumptions

- Data provided by ABC is correct and accurate for modelling purposes

- Number of parcels are rounded down to nearest integer
- 100 simulations is sufficient to provide a good indication of possible options to draw
conclusions on profitability of the Company.
- Delivery of parcels is within an area that the drivers can make all the deliveries allocated to them
- No other expenses incurred by ABC Parcel Delivery Company e.g. no overheads to store parcels
or get them to the delivery drivers
- The total number of 550 parcels is guaranteed every day and no seasonal fluctuations
- No new entrants to the parcel delivery market which would change the availability of any of the
drivers

Methodology
Scenario 1:
Number of potential parcels delivered per driver
The number of parcels per driver and simulation calculated as the minimum of those they would be
willing to deliver (from the data) and the random number multiplied by the maximum possible number of
parcels. The number of parcels rounded down to the nearest integer.
For example, for driver II, multiply the random number from each simulation by 500, and apply a floor of
100.
Allocation to cheapest driver first
Based on the data table the driver allocation is in the following order: II, I, IV, III.
The parcels are allocated based on the numbers calculated above, ensuring that the driver gets their
minimum, if fewer parcels left than the minimum then 0 parcels are allocated.
Overall, there are 550 possible parcels for delivery. So, for each simulation, allocate the number of
parcels to driver II first. As this will always be less than 550, then next look to driver I's allocation of
parcels for that simulation. At each allocation point, will always check if total delivery of 550 achieved.
Any remaining left over allocated to the next driver as per the order above, with the same conditions on
checking whether the 550 total parcels achieved.
If any parcels remain after completing the above allocation to all four drivers then the remainder allocated
to driver V.

Allocation to the drivers who deliver the most first

Based on the data table the driver allocation is in the following order: IV, II, III, I
The parcels are allocated based on the numbers calculated above, ensuring that the drivers get their
minimum. If fewer parcels are left than the minimum then 0 parcels are allocated.
If any parcels remain after completing the above allocation then the remainder allocated to driver V.

The allocation process works similar to the first example, but as Driver IV could deliver 550 parcels
under 1 simulation, then there will always be a check point needed for all allocations to the other drivers
if any parcels are allocated to them.

Calculation of profit per simulation

The income received by ABC is calculated as the total number of parcels to be delivered multiplied by
amount charged per parcel, i.e. 550 x 5 = $2,750
The outgoes are calculated per driver as the drivers charge per parcel multiplied by the number of parcels
delivered for that simulation. The outgoings then vary per simulation.
The profit is calculated as income less outgo.
Summary information
For each method of calculation the following summaries are calculated – the minimum, maximum, and
mean amount of profit from the simulations. Also calculate how many of the simulations where the target
profit amount is reached.

Scenario 2
Similar calculations to scenario 1 completed, except by halving the number of parcels for Drivers I and II.
Achieved by reducing the number from the 'potential number of parcels' section by 50%.
All other calculations remain the same.

Scenario 3
Similar calculations completed as per scenario 1, except that the total number of parcels allocated has
increased from 550 to 650.

Scenario 4
Using the calculation from scenario 1 but where the driver order is based on most deliveries, the price
paid to each driver is changed. This is linked to a new parameter and goalseek is used to set this price so
that the amount paid per driver is the same for all 5 drivers. The goalseek is set so that the price paid per
driver is set so that the target profit is reached 100% of the time.

Results
Scenario 1
The table below summarises the results from scenario 1:

Allocation method Minimum profit Maximum profit Mean profit($) % simulations

($) ($) where target
reached
Cheapest driver 1,222 1,995 1,544 56%
first
Most parcels first 1,100 1,853 1,315 22%

The above results show that allocating to the cheapest driver first generates better profits – this makes
sense as driver V delivers a large number of parcels, but also has a very minimum allocation, so is likely
in most simulations to get most of the deliveries. Given the low charge of driver II at $1.25 per parcel,
then if the maximum allocation of parcels (500) given to driver II then ABC will earn $1,875 ((5-
1.25)*500) profit from this driver alone.

The results for the allocation to most parcels delivered first reaching the target in fewer simulations is
sensible, as the driver who delivers the most (driver IV) is also more than twice as expensive as driver B.
Therefore allocating more parcels to driver IV means much lower profits.
Chart 1: Simulation results for allocation to cheapest driver first

As can be seen from the graph there are a number of simulations when the target profit is reached.

Graph to show scenario results for allocation to driver who delivers the
most first
2000

1500

1000
Profit $

500

0
1
6
11
16
21
26
31
36
41
46
51
56
61
66
71
76
81
86
91
96

Scenario Number

Scenario profit target profit

Chart 2: Allocation of parcels to most delivered first

The above chart shows that there are fewer simulations where the target profit is reached.
Scenario 2
If drivers I and II experience issues with their cars, then the following table shows a summary of the
results:

Allocation method Minimum profit Maximum profit Mean profit($) % simulations

($) ($) where target
reached
Cheapest driver 1,118 1,547 1,304 18%
first
Most parcels first 1,100 1,521 1,214 3%

Overall, these results show how reliant ABC are on driver II in particular being able to make their
deliveries in order to reach the target profit.
Assuming driver II only delivers 250 parcels, then the reduction in profit for ABC would be (3-1.25)*250
= $438, assuming that payment to the other driver is on average $3 per parcel compared to $1.25 for
driver II. This also assumes that driver II can deliver their full number of parcels but based on the random
numbers this is the extreme. So a reduction in profits less than $438 is to be expected.
Also, we can see from the results that the minimum profit is almost unaffected as a result of the car
issues.

Scenario 3
Increasing the number of parcels available for delivery from 550 to 650 to assess if the increased income
will help ABC reach their target profit.
The table below provides a summary of these results:

Allocation method Minimum profit Maximum profit Mean profit($) % simulations

($) ($) where target
reached
Cheapest driver 1,419 2,195 1,738 93%
first
Most parcels first 1,300 2,100 1,577 72%

Based on the average we can see that if ABC agree to deliver more parcels per day then they are more
likely to achieve their profit target under both allocation options.
Allocating to the cheapest driver first gives a very likely probability of meeting the target profit per day.

Scenario 4
By making the pay more equivalent across the drivers, then ABC could achieve their target profit by
paying the drivers $2.36 per parcel. Overall this makes sense as the available funds to pay the drivers for
outgoing is (2750-1450)=1300 so dividing this by 550 parcels is $2.36.
Chart 3: Comparison of the amounts charge by driver

Comparison of amounts charged by driver

4
3.5
Amount paid per parcel $

3
2.5
2
1.5
1
0.5
0
I II III IV V (only used if
parcels not
delivered)

Driver

Original amounts paid Target payments

As shown in the above graph:

 Driver I amount paid per parcel are only slightly reduced

 Driver II sees the largest increase in amount paid per parcel – this is reasonable given driver II
delivers the most under this parcel allocation method
 Driver III sees the largest reduction in amount paid
 Drivers IV and V see a similar reduction in amount paid per parcel

Conclusions
To reduce potential that ABC does not meet their daily target then it would be better to pay the same flat
rate to all drivers.
The final results of the Company profit will depend on multiple factors including availability of drivers,
number of parcels to be delivered. So there would be no guarantee that the target profit could be reached
every day.

Next steps
- Validate the data provided
o In particular, the number of parcels per driver and costing
o Compare to other parcel delivery companies
o Compare the costing of parcel deliveries to other comparable companies
- Consider pricing of parcel deliveries (and payments to drivers) based on size and weight
- Consider pricing based on distance of deliveries
o For drivers
o For charging of customers
- Increase number of drivers to enable increase in parcel deliveries
- Increase the simulations to get a more stable distribution of results of profits for ABC parcel
delivery company
- Other expenses – include costings of holding areas for parcels, and potential refunds to customers
if parcels not delivered in tack.
- Combine an increase in parcel deliveries with the equal costings
- If equal pricing for drivers is not mandatory, consider drivers who are cheaper to maximize profit
o But consider potential reputation issues if drivers are not as good as the more expensive
drivers
- Investigate the maximum number of parcels that ABC would need to deliver in order to reach
their target profit 100% of the time.
- Peer review of the model
- Determine if expand to deliver parcels for just individual consumers or larger Companies – e.g.
could ABC set up a contract with a large retail company to deliver products for them
- Consider impact on expenses if expansion bigger than the overhead costs can handle e.g.
managing the number of parcels
- Consider impact of tax on the profitability
- Consider other shock scenarios e.g. Drastic increase in cost of fuel impacting on drivers potential
to make an income; Drivers unable to deliver due to extreme weather

END OF SUMMARY