Impact of Forecasting on the Bullwhip.pdf
-
Upload
vitorellivitorelli -
Category
Documents
-
view
213 -
download
0
Transcript of Impact of Forecasting on the Bullwhip.pdf
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
1/34
19.04.2007 1
Erkan BAYRAKTAR
Baheehir University, stanbul
Kazm SARI
Beykent University, stanbul
Impact of Forecasting on the Bullwhip
Effect
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
2/34
19.04.2007 2
Impact of Demand Forecasting onthe Bullwhip Effect Introduction
Previous Studies
Our Study:
Winters Method (Triple Exponential Smoothing)for demands with linear trend and seasonality
Simulation Model
Design of Experiment Analysis of Simulation Ouput
Conclusion
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
3/34
19.04.2007 3
The Bullwhip Effect
The bullwhip effect represents the phenomenon
where orders to supplier tend to have larger
variance than sales to the buyer (i.e., demand
distortion) and this distortion propagates upstream
in an amplified form.
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
4/34
19.04.2007 4
Retailers Orders to Wholesaler
Increasing Variability of Orders up the Supply Chain (Lee et al. 1997b)
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
5/34
19.04.2007 5
DecreaseProfitability
DecreaseCustomer service
rate
IncreaseLead Time
IncreaseTransportation cost
IncreaseInventory cost
IncreaseProduction cost
High Bullwhip EffectPerformance
Measurement
(Chopra et al.2001, p.363)
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
6/34
19.04.2007 6
Metters (1997) studied the impact of the bullwhip
effect on profitability by establishing an empiricallower bound on the cost excess of the bullwhip
effect. Results indicate that the importance of the
bullwhip effect to a firm greatly depending on thespecific business environments and eliminating the
bullwhip effect can increase product profitability by
10-30%.
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
7/34
19.04.2007 7
Causes of the Bullwhip Effect(Lee et al. 1997a, 1997b)
Demand forecasting Order batching
Price fluctuation
Rationing and shortage gaming
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
8/34
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
9/34
19.04.2007 9
2. Previous Studies: Simple
Exponential Smoothing
Chen et al. (2000b):
Correlated demand AR(1) pattern
Order-up-to inventory control policy
The larger the smoothing parameter (), the larger the
bullwhip effect,
Longer lead times (L) lead to larger increase in bullwhip
effect,
A retailer facing a longer lead time, L, must use a smallersmoothing parameter (), in order to reduce the bullwhip
effect
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
10/34
19.04.2007 10
3.Previous Studies: Douple
Exponential Smoothing
Chen et al. (2000b):
Demand with a linear trend
Order-up-to inventory policy
Smoothing parameters , , and lead time have asignificant impact on the bullwhip effect,
Larger the smoothing parameters, the larger thebullwhip effect,
The increase in variability (bullwhip effect) doesnot depend on the magnitude of the linear trend.
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
11/34
19.04.2007 11
Winters Method (Triple
Exponential Smoothing) for
Demands with Linear Trend &Seasonality
A Simulation Study
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
12/34
19.04.2007 12
Design Of Experiment
The purpose of the design of experiment is to
analyze the impact of :
Smoothing parameters (, , ),
Lead time (L),
Strength of the seasonality,
on the bullwhip ratio.
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
13/34
19.04.2007 13
Independent Variables
Smoothing parameters Alpha (0.01, 0.25, 0.50),
Beta (0.01, 0.25, 0.50),
Gamma (0.01, 0.25, 0.50),
Lead time Low ( 1 week ),
Medium (3 weeks),
High (5 weeks),
Strength of seasonality Low seasonality Medium seasonality
High seasonality
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
14/34
19.04.2007 14
Demand Types
Demand generatorsnormal()noise
season
t)]52
2(sin[season
t)slopebase(Demandt +
+
+=
30
15
5
SEASON
10021000Low
Seasonality
10021000Medium
Seasonality
10021000High
Seasonality
NOISESLOPEBASEDEMAND TYPE
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
15/34
19.04.2007 15
1000
1200
1400
1600
1800
2000
2200
2400
0 10 20 30 40 50
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
2000
0 10 20 30 40 50
1300
1400
1500
1600
1700
1800
1900
2000
2100
2200
0 10 20 30 40 50
High Seasonality Medium Seasonality
Low Seasonality
Demand types with
different strength of
seasonality that is used insimulation analysis
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
16/34
19.04.2007 16
Simulation Study: Supply Chain Structure
Consist of two members:
one retailer and one manufacturer,
In each period, t, the retailer observes his inventoryposition and places an order, qt , to the manufacturer
After the order is placed, the retailer observes andfills customer demand for that period, denoted by Dt
Any unfilled demand is backlogged
There is a fixed lead time between the time an orderis placed by the retailer and when it is received at theretailer
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
17/34
19.04.2007 17
Simulation Study: Forecasting Technique
It is assumed that retailer uses the winters (triple
exponential smoothing) method to forecast demandwhich is formulated as
n= 1,2...s
Ft+n : forecast at period t+n,
Lt : level component of demand at period t,Tt : trend component of demand at period t ,
Set+n-s : seasonality index for the same period in previous year,
(Abraham and Ledolter, 1983, p.170)
sntettent SnTLF ++ += )(
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
18/34
19.04.2007 18
Simulation Study: Forecasting Technique
s-1t
1
11t
e
11
t
s-1t
1t1t
S)1(S
)1()(
)()1(S
+
+
+
+
++
+
++
+=
+=
++=
t
t
tttt
te
L
D
TLLT
TLDL
(Abraham and Ledolter, 1983, p.170)
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
19/34
19.04.2007 19
Simulation Study: Ordering Policy
We assume that the retailer follows a simple order-up-to inventory policy in which order-up-to point isestimated from the observed demand as
: Forecasted demand over L periods
L : lead time + review period ( 1 )
: standard deviation of the demand over L periods
z : is a constant chosen to meet a desired service level,
L
t
L
tt zDS
+=
L
tD
L
t
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
20/34
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
21/34
19.04.2007 21
Simulation Output Analysis
The length of the simulation run is 520 weeks.
First 156 weeks are used to estimate the initial
parameters for the forecasting model,
To reduce the impact of random variations,ten replicates were conducted for each
combination of the independent variables.
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
22/34
19.04.2007 22
Simulation Output Analysis
Simulation Output Analysis indicates that
bullwhip ratio significantly influenced by
the
Smoothing Parameters (, , ), Lead Time,
Strength of the seasonality.
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
23/34
19.04.2007 23
Multiple Range Test
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
24/34
19.04.2007 24
Interaction Effect
Between SmoothingParameters & Lead
Time
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
25/34
19.04.2007 25
Interaction Effect
Between SmoothingParameters &
Seasonality
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
26/34
19.04.2007 26
Further Analysis of Gamma Parameter
Bullwhip Ratio for Various Levels of Gama
In order to better understand the influence of the
gamma parameter on bullwhip ratio we further
make an analysis for various levels of gammaunder different conditions.
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
27/34
19.04.2007 27
Further Analysis of Gamma Parameter :Bullwhip
Ratio for Various Levels of GammaEstimated Marginal Means of Bullwhip Effect
Gama
,91,81,71,61,51,41,31,21,11,01Es
tima
t
ed
Marg
ina
l
Means
17
16
15
14
13
12
11
Although, generally increasing values of gamma() increases thebullwhip ratio, in some points at the medium level of gamma()
the bullwhip ratio is smallest, not at small values of gamma ()
parameter
Duncana,b
120 2,3621
120 2,3736120 2,3866
120 2,4154
120 2,4177
120 2,4702
120 2,4831
120 2,5469120 2,6271
120 2,7147
,110 ,872 ,372 1,000 1,000 1,000
Gama,31,21
,41
,11
,51
,61,01
,71,81
,91
Sig.
N 1 2 3 4 5 6
Subset
Natural logarithm(Ln) transformation is made to
the dependent variable
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
28/34
19.04.2007 28
Conclusion
While seasonality gets stronger, the bullwhip effect goes
down. Intuitively, it can be said that the bullwhip effect is compensated by
the variability generated by the seasonality.
The impact of gamma ( ) parameter on the bullwhip ratio is
minor when compared with the other smoothing parameters Increase in lead time (L) leads to an increase in the bullwhip
ratio, but when alpha () and beta () parameters are chosen small, this
increase is very small when compared with higher values of alpha (),and beta ()
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
29/34
19.04.2007 29
Conclusion...
Choosing alpha (), and beta () parameter small is
important in order to reduce the bullwhip ratio,especially when the lead time (L) is long or/andstrength of seasonality is low.
The impact of the gamma( ) parameter on thebullwhip ratio is different form the alpha () andbeta () parameters. While smaller alpha () and beta() lead to smaller bullwhip ratio, for the gamma( )parameter, small and high values lead to largerbullwhip ratio, but medium values of gamma( )lead to smaller bullwhip ratio.
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
30/34
19.04.2007 30
Future Area of Research
Assess the impact of bullwhip effect on the
performance measures of the supply chain
(e.g., total cost of the members, total chain
cost, service level of chain members, andservice level of the chain).
Investigate other techniques of time series
analysis for seasonality.
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
31/34
19.04.2007 31
Thank you...
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
32/34
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
33/34
19.04.2007 33
-
8/11/2019 Impact of Forecasting on the Bullwhip.pdf
34/34
19.04.2007 34