Mathematical Optimization Université Paris-Dauphine

The Lockbox Problem

Consider a national firm that receives payment via checks from all over the country.

• the Postal Service guarantees a service time that depends on the traveled distance

• for each travel day the firm loses the money related of the interest lost

• the firms can open offices (called lockboxes) in different cities to handle the checks and
reduce the travel times. Conflicting tradeoff between:

– the cost of opening a lockbox

– additional interests gained since the checks are cleared faster

Consider a firm in the United State which receives payments from 4 regions:

• The average daily total payment from each region is as follows: 300,000 $ from the
West 120,000$ from the Midwest, 360,000$ from the East and 180,000$ from the South.

• The firm consider the option of opening lockboxes in L.A., Cincinnati, Boston and
Houston. Operating a lockbox costs 90,000$ per year.
The average days from mailing to clearing is given in the table:

From – To L.A. Cincinnati Boston Houston

West 2 4 6 6
Midwest 4 2 5 5
East 6 5 2 5
South 7 5 6 3

We can calculate the losses due to lost interest for each possible assignment. For
example, if the West sends to Boston, then on average there will be 1,800,000 (= 6
· 300, 000) in process on any given day. Assuming an investment rate of 10%, this
corresponds to a yearly loss of 180,000. We can calculate the losses as follows:

From To L.A. Cincinnati Boston Houston

West 60 120 180 180
Midwest 48 24 60 60
East 216 180 72 180
South 126 90 108 54

• Write a MIP to minimize the total yearly costs

 (Explanation: consider the first year and the case the West sends to Boston, for 6 days no
money can be invested, money instead which could have been invested and would have given
an interest at the end of the year.)

Decision variables:
(
1 if lockbox j is opened (j = 1, . . . , n)
yj =
0 otherwise
(
1 if region i (i = 1, . . . , m) sends to lockbox j (j = 1, . . . , n)
xij =
0 otherwise

• n = number of lockboxes (j = 1, . . . , n)

• m = number of regions (i = 1, . . . , m)

• αij = annual cost of assigning region i to the lockbox j

• βj = annual cost of lockbox j

n X
X m n
X
min αij xij + βj yj (1)
j=1 i=1 j=1
n
X
xij = 1 i = 1, . . . , m (2)
j=1
m
X
xj ≤ myj j = 1, . . . , n (3)
i=1
xij ∈ {0, 1} i = 1, . . . , m, j = 1, . . . , n (4)
yj ∈ {0, 1} j = 1, . . . , n (5)

• Each region must be assigned to one lockbox

• region can only be assigned to an open lockbox

 min 60x11 + 120x12 + 180x13 + 180x14 (6)
+48x21 + 24x22 + 60x23 + 60x24 (7)
+216x31 + 180x32 + 72x33 + 180x34 (8)
+126x41 + 90x42 + 108x43 + 54x44 (9)
+90y1 + 90y2 + 90y3 + 90y4 (10)
x11 + x12 + x13 + x14 = 1 (11)
x21 + x22 + x23 + x24 = 1 (12)
x31 + x32 + x33 + x34 = 1 (13)
x41 + x42 + x43 + x44 = 1 (14)
x11 + x21 + x31 + x41 − 4y1 ≤ 0 (15)
x12 + x22 + x32 + x42 − 4y2 ≤ 0 (16)
x13 + x23 + x33 + x43 − 4y3 ≤ 0 (17)
x14 + x24 + x34 + x44 − 4y4 ≤ 0 (18)
xij ∈ {0, 1} i = 1, . . . , 4, j = 1, . . . , 4 (19)
yj ∈ {0, 1} j = 1, . . . , 4 (20)
(21)

• Additional valid inequalities can be added to restrict the search space and accordingly
improve the performances of the solvers.

• For example the following constraints can be added:

xij ≤ yj i = 1, . . . , m, j = 1, . . . , n (22)

• Different formulations of the same problem can have different linear programming re-
laxation.

• A good formulation is a formulation with a tight linear programming relaxation, i.e.,

small gaps between the values of the integer solution and the LP relaxation solution.