64. DesiotNG AND MANAGING THE SUP®LY CHAIN
3.3 RISK POOLING
Consider the following case,
Risk Pooling
ACME, a company that produces and distributes elec
tronic equipment in the Northeast of the United States,
faces a distribution problem. The current distribu
tion system partitions the Northeast into two markets,
each of which has a single warehouse, One ware-
house is located in Paramus, New Jersey, and the sec~
‘ond is located in Newton, Massachusetts. Customers,
typically retailers, receive items directly from the
‘warehouses; in the current distribution system, each
customer is assigned to a single market and receives,
deliveries from the corresponding warehouse.
The warehouses receive items from a manufactur
ing facility in Chicago. Lead time for delivery to each
of the warehouses is about one week and the manu-
facturing facility has sufficient production capacity to
satisfy any warehouse order. The current distribution
strategy provides @ 97 percent service level; that is,
the inventory policy employed by each warehouse
is designed so that the probability of a stockout is,
3 percent. OF course, uniilled orders are lost to the
competition and thus cannot be satisfied by future
deliveries
‘Since the original distribution system was designed
‘over seven years ago, the company’s newly appointed
CEO has decided to review the current logistics and
distribution system. ACME handles about 1,500 dif-
ferent products in its supply chain and serves about
10,000 accounts in the Northeast.
ACME is considering the following alternative
strategy: Replace the two warehouses with a single
‘warehouse located between Paramus and Newton that
will serve all customer orders. We will refer to this,
proposed system as the centralized distribution sys-
tem. The CEO insists that the same service level,
97 percent, be maintained regardless of the logistics,
strategy employed.
Obviously, the current distribution system with
‘wo warehouses has an important advantage aver the
single warehouse system because each warehouse is
close to a particular subset of customers, decreasing
delivery time. However, the proposed change also has
‘an important advantage: it allows ACME to achieve
cither the same service level of 97 pereent with much
lower inventory or a higher service level with the
‘same amount of total inventory.
Intuitively this is explained as follows. With ran-
dom demand, it is very likely that a higher-than-
average demand at one retailer will be offset by a
Jower-than-average demand at another. As the number
of retailers served by a warehouse goes up, tis like-
lihood also goes up. Indeed, this is precisely the third
principle ofall forecasts described at the beginning of
Section 3.2.2: aggregate forecasts are more accurate,
How much can ACME reduce inventory if the
company decides to switch to the centralized system
‘but maintain the same 97 perceat service level?
‘To answer that question, we need to perform amore
Figorous analysis of the inventory policy that ACME
should use in both the current system and the cen-
tralized system, We will explain this analysis for two
specific products, Product A and Product B, although
the analysis must be conducted for all products,
For both products, an order from the factory costs
‘$60 per order and holding inventory costs are $0.27
‘per unit per week. In the current distribution system,
the cost of transporting a product from a warehouse to
‘customer is, on average, $1.05 per product. Its esti-
‘mated that in the centralized distribution system, the
transportation cost from the central warehouse will
be, on average, $1.10 per product. For this analysis,
wwe assume that delivery lead time is not significantly