Operations Strategy (Class 4) Capacity Strategy and Operational Hedging

Operations Strategy/Risk Mitigation & Operational Hedging
5/20/15
Operations Strategy (class 4)

Capacity strategy and Operational Hedging
Capacity strategy
Analyze the value of flexibility as operational hedge
Explain the concept of risk management
q
Case:
Seagate Technology
Common Categories of Flexibility

Flexibility measures the ability to adapt or change. (Upton)
Mix Flexibility
Volume
Flexibility
New Product
Flexibility
The ability to produce multiple products. This may refer to

simultaneous production, e.g., different models
produced at the same time on the same assembly line, or to
the ability to easily switch production from one product to
another.
The ability to increase or decrease the production rate of a
particular product (without a significant increase in cost or
decrease in performance)
The ability to introduce new products into the operating

system (sourcing, production and distribution)
Operations Strategy/Risk Management & Operational Hedging
Allon
Slide 2
Allon
5/20/15
Obstacles to achieving flexibility

Flexibility costs more but its benefits are difficult to measure, value
and convey
Competition and customer perception may preclude using flexibility
It is often unclear which features of a process must be changed to
enhance its flexibility:
Flexibility is driven more by people and management than by
process structure (policies/procedures/culture)
Slide 3
Allon
Assembly and Test Operations

Contribution margins
Cheetah
Assembly
Cheetah: $400
dedicated resource
Test
Barracuda
Assembly
dedicated resource
Barracuda: $300
shared resource
Proposed capacity plan

Cheetah Assembly: 300K units @ $30K per 1K units
Barracuda Assembly: 300K units @ $20K per 1K units
Test: 600K Units @ $80K per 1k units
CapEx = $40M (fixed)+9M (C Assy) + 6M (B Assy) + $48M (Test) = $103M
Allon
Slide 4
Allon
5/20/15
Valuation of network capacity under uncertainty

Demand D
Invest in
capacity K
Market has
demand D
Sales
Mismatch units
and cost
Operating
profit
Profit deviation
from mean
Pessimistic 25%: (150, 350)
(150, 300)
(0, -50) = - $15M
$150M
-$41.25M
Most likely 50%:
(300, 300)
(300, 300)
(0, 0 ) = $0
$210M
$18.75M
Optimistic 25%: (450, 250)
(300, 250)
(300, 300, 600)

CapEx = $103M
(-150,0) = -$60M
-$18.75M
$195M
$191.25M
$3.75M
$24.59M =
sqrt(expected
squared
deviation)
E(NPV) = $191.25M - $103M = $88.25M

E(profit variability) = st.dev of operating profits = $24.59M
Slide 5
Allon
Benchmark: Full Information

What if capacity is built once demand is known?
Capacity plan is perfect match to demand
Operating profit = $300 x 300k + $400 x 300k = $210 M
NPV = $210M - $103M = $107M
ROI = 107/103 = 104%
What is the impact of uncertainty?
Allon
Slide 6
Allon
5/20/15
Determining the Best Capacity

Build
Capacity
Observe
Actual
Demand
Produce and
Sell
Based on forecast
Outcome A: Demand = Capacity
No Demand/Capacity Mismatch.
Sales = Demand
Based on actual demand

Outcome B: Demand < Capacity
Demand/Capacity Mismatch.
Sales = Demand
BUT
You wish you had invested in less
capacity!
Outcome C: Demand > Capacity

Demand/Capacity Mismatch.
Sales = Capacity
BUT
You wish you had invested in more
capacity!
Reminds me
of overage
cost
Reminds me
of underage
cost
Canonical Newsvendor Model!
Slide 7
Allon
Simplification: Reduce variety or complexity:

Single Product or Dedicated Test
Imagine Seagate only sold the Cheetah drive
Capacity cost =
30K per 1K units
Capacity cost =
80K per 1K units
Cheetah
Assembly
Contribution
margin
$400
Test
Would set Assembly Capacity = Test Capacity

Two-resource problem equivalent to a one-resource problem
Capacity cost = 30K
+80K=110K per 1K units
Contribution
margin
$400
Resource
Determine optimal capacity, K*

Set Assembly and Test Capacities equal to this K*
Allon
Slide 8
Allon
5/20/15
Determining the Best Capacity for a single resource:

Marginal Analysis for 1D Cheetah Problem
Demand Distribution
Capacity cost =
$110 per unit
Demand
Contribution
margin
$400
Resource
Probability
Cumulative
Probability
150
0.25
0.25
300
0.50
0.75
450
0.25
1.0
Should we invest one more unit above 150?

Marginal value = .5*$400 + .25*$400 = .75*$400 = Pr(shortage)*$400
Marginal cost = $110
Add to investment MV > MC
= as long as service level < Critical Fractile = cu = 290 = 290 = 0.725
cu + co 290 + 110 400
= 1D newsvendor model
K* = 300 so set capacities of Cheetah Assembly and Test both at 300
With single product, optimal safety capacity = 0!
Same for single Baracuda product.
Slide 9
Allon
The real problem is more complicated

Necessity and Value of Network Analysis
Contribution
margins
Cheetah: $400
Cheetah
Assembly
dedicated
resource
Test
Barracuda
Assembly
dedicated
resource
shared
resource
Barracuda:
$300
We need an holistic approach

because Test is a shared resource
Network analysis is needed to capture:
(shifting) bottlenecks in a multi-resource network
correlated demands
operational flexibility (substitution, rev. max) and hedging
Allon
Slide 10
Allon
5/20/15
Capacity for a Multi-Resource Network
If we have multiple resources, we have multiple constraints and capacity is multidimensional!

A given capacity portfolio (vector), constrains later production decisions:
Capacity = KC
Cheetah
Assembly
Capacity = KT
Cheetah Production xC < KC
dedicated resource
Barracuda Production xB < KB
Test
Capacity = KB
Cheetah + Barracuda Production

xB + xC < KT
shared resource
Barracuda
Assembly
dedicated resource
Dont forget: we also dont want to produce more than demand in any given scenario
If Test is the bottleneck, how should we allocate its capacity between
Cheetah and Barracuda?
Slide 11
Allon
Proposed Capacity Plan

Barracuda
in 1000 units
800
700
Test
Capacity 600
Note: boNleneck
depends on the
demand scenario
(shiRing boNleneck)
500
400
PessimisUc
(0.25)
Expected
(0.5)
Barracuda Assy. 300

Capacity
200
100
0
OpUmisUc
(0.25)
Feasible
ProducUon
100
200
300
400
Cheetah Assy.
Capacity
Allon
500
600
700
Cheetah
800 in 1000 units
Test
Capacity
Slide 12
Allon
5/20/15
Newsvendor Networks:
Marginal Analysis for Barracuda Assembly
Barracuda
in 1000 units
Increase Capacity from 300 to 301

Sell 1 more Barracuda in PessimisUc Scenario
Sell 0 more in other scenarios
MV = 0.25*300 = 75
MC = 20
Marginal prot = 75-20=55 posiUve
800
700
Test
Capacity 600

Sell 1 more Barracuda in PessimisUc Scenario
Sell 0 more in other scenarios
MV = 0.25*300 = 75
MC = 20
Marginal prot = 75-20=55 posiUve
500
400
PessimisUc
(0.25)
Expected
(0.5)
Barracuda Assy. 300

Capacity
200
100
0
OpUmisUc
(0.25)

Sell 0 more Barracuda in all scenarios
MV = 0
MC = 20
Marginal prot = -20 negaUve
Feasible
ProducUon
100
200
300
400
500
Cheetah Assy.
Capacity
600
700
Cheetah
800 in 1000 units
Test
Capacity
Slide 13
Allon
Newsvendor network analysis:
Marginal analysis of value of capacity in a network
Adding one unit of Barracuda final assembly capacity increases expected operating
profit by 25%*$300 = $75 > CapEx of $20, as long as Kb < 350. Marginal profit is
negative beyond 350.
Similarly, adding one unit of Cheetah final assembly capacity has expected marginal
value of 25%*$400 - $30 = $70, as long as Kc < 350. But if we increase Kc beyond
350:
test is constraining us and the marginal value of an additional cheetah capacity unit is that we
can make one more Cheetah drive but one less Barra drive (due to test constraint). The
expected marginal value is 25%*($400-$300) = $25 < CapEx of $30, which is suboptimal.
Increasing both cheetah and test capacity is also suboptimal: marginal value = 25%*$400 =
$100 < $30+$80
Sizing of a capacity portfolio starting guideline:
P(resource i is bottleneck)
Marginal cost of capacity i
Marginal return of capacity i
Maximize value using marginal analysis of overage versus underage!
Allon
Slide 14
Allon
5/20/15
Operational Capacity Hedge:

K = (350, 350, 600)
Barracuda [1000 units]

700
What is utilization of the three

capacities?
600
500
Pessimistic
25%
400
Expected
50%
Optimistic
25%
Original capacity
proposal
100
0
0
100
200
300
400
In what sense is this capacity plan an

operational hedge?
1. Upstream excess capacity illustrates
one of the operational hedging
strategies: capacity reserves mitigate
demand risk
2. Less excess capacity is needed in
testing, which illustrates another OH
strategy: pooling mitigates demand
risk
3. To exercise the switching real option
embedded in test, we need upstream
capacity imbalance
300
200
Safety capacity in each scenario

More upstream capacity than
downstream > patently unbalanced
capacity portfolio
500
600
700
Cheetah [1000 units]
Slide 15
Allon
Seagate Technology:
Excel formulation and Mean-Variances
Production Margin per thousand (in unit of $1000)
Cheetah
$
400
Barracuda
$
300
CapEx per thousand (in unit of $1000)
Cheetah
$
Barracuda
$
Testing
$
Fixed Cost
$
30
20
80
40,000
Demand
Probability
(In unit of thousand)
Cheetah
Barracuda
Scenario A
0.25
Scenario B
0.5
150
350
300
300
450
250
Contingent Production
Cheetah
Barracuda
Sum(Cheetah+Barra)
Op. Profit
(in unit of thousand)

150
350
500
$
165,000
300
300
600
210,000
350
250
600
215,000
Probability
Expected Op. Profit
Std. Dev. of op. profit
Investment (CapEx)
Net Incom
ROI
0.25
$
$
$
$
Allon
0.5
0.25
Expected Value
$
88,250
$
93,500
$
94,500
ROI
86%
80%
90%
Capacity (in thousands)

350
350
600
200,000
20,310
105,500
94,500
90%
Summary comparison 3 capacity portfolios

Coordinated:
(300, 300, 600)
Full Insurance:
(450, 350, 700)
E(NPV)-Optimal
(350, 350, 600)
Operations Strategy/Risk
Management
& Operational
Hedging
With
deterministic
demand
Scenario C
0.25
(300, 300, 600)
Slide
16
107,000.00
$
$
$
104% $
Std dev
24,590
31,820
20,310
-
Mismatch cost
$
18,750
$
13,500
$
12,500
$
Allon
5/20/15
Mean-Variance Analysis of Operational Hedging:

Example via Capacity Portfolio Imbalance in Seagate
10
NPVOptimal K*
Initial
Sales-plan
driven K
Full
insurance K
For whom is K* optimal?
Reduce risk:
What is the optimal riskreducing capacity

portfolio?
When is operational
hedging most effective?
Expected
Value ($)
-4
-6
Standard deviation
of value ($)
Slide 17
Allon
Optimal Capacity Portfolio Investment:

Implementation Challenges
The optimally installed capacity will never be all used simultaneously
The current sales-plan driven capacity planning practice will never lead to the
optimal capacity vector
Which organizational changes are needed so the optimal capacity vector is

chosen?
Incentives: mktg-sales v. finance-ops
Demand & sales planning & forecasting
Integration and coordination (ERP)
Allon
Slide 18
Allon
5/20/15
Demand, Production, and Capacity Planning:

Pitfalls of Sales-Plan driven investments
Still happens:
- Tech
- Medical devices
- Automotive
Resource Planning
&
Capital Budgeting
Prod &Inv
Planning
Informal
Feedback
Sales
Plan
Demand
forecast
Master
Production
Scheduling
MPS
CRP
CRP = How much
capacity is required
to produce MPS?
MRP
Mat
INV
CRP &MRP
Records
Shop-floor &
Vendor Systems
Slide 19
Allon
Learning Points from Seagate Case:

Operational hedging through capacity portfolios
Capacity portfolio investment is simultaneous investment in multiple resource

types. The basic question is to determine the capacity for each resource type.
Dont do sales-plan driven capacity planning

i.e., dont build capacity for one single demand scenario
Explicitly capture forecast uncertainty

Agree on a demand forecast, i.e., set of scenarios, capturing correlation
Build capacity based on the forecast.
Use newsvendor logic to balance underage and overage consequences.
Use a portfolio approach

Best capacity for resource X depends on capacity choice for resource Y.
Network model captures moving bottlenecks and operational flexibility
Allon
Slide 20
Allon
10
5/20/15
Learning Points from Seagate Case:

Operational hedging through capacity portfolios
Tailor operational hedges to specific risks:

1. Adding different amounts of reserves (insurance or safety capacity) to
different resources.
2. Flexible resources pool demand risk and need less safety capacity. This also
enables them to exercise their real switching option to maximize profits.
The above mitigate operational risk: they reduce the expected mismatch cost
and thus add value.
They also mitigate financial risk: profit variability is reduced.
Slide 21
Allon
GM Chevy
2 assembly plants:
Arlington, TX: 75,000 units/yr
Fairfax, KS: 200,000 units/yr
2 stamping plants:
Pontiac, MI
Fairfax, KS [contiguous, same location as assembly]
A die-set (for stamping hood, roof, etc) costs about $30 to $50M
Should we have one die-set in each plant, or 1 set in Fairfax (and ship
to Arlington)?
Allon
Slide 22
Allon
11
5/20/15
Approach
Value proposed capacity plan
Consider other simple capacity investments
Give bounds on value
Build intuition
Towards an value-maximizing investment
Strengths and weaknesses
Allon
Slide 23
Allon
12

Operations Strategy (Class 4) Capacity Strategy and Operational Hedging

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Operations Strategy (Class 4) Capacity Strategy and Operational Hedging

Hochgeladen von

Copyright:

Verfügbare Formate

Operations Strategy/Risk Mitigation & Operational Hedging