Study of the Determinants of Demand for Propane

Download Report

Transcript Study of the Determinants of Demand for Propane

Study of the Determinants of
Demand for Propane
Prepared and Presented By:
Joe Looney
J.D. Laing
Ernest Sonyi
Background
• Propane is used for:
– Heating homes
– Heating water
– Cooking
– Drying clothes
– Fueling gas fireplaces
– As an alternative fuel for vehicles
Background
• Propane is used in the Petrochemical
industry to make:
– Plastics
– Alcohols
– Fibers
– Cosmetics
Background
• Propane is used in agriculture for:
– Crop drying
– Weed control
– Fuel for farm equipment
– Fuel for irrigation pumps
Propane Use by Sector
Objectives
• To determine the variables and
interactions of these variables upon the
demand for propane in the United States
• Attempt to use econometric data to show a
direct relationship between propane prices
and propane demand
Assumptions
• The weekly supply data for propane reflects a
replenishment of used stores of the product in
the market
• Meets all five assumptions for regression
–
–
–
–
–
The model makes sense
There is a significant statistical relationship between variables
There is an acceptable percent variation between variables
There is no problem with autocorrelation
There is no problem with multicollinearity
Hypotheses
• H1: Demand for Propane is explained by
poultry production
• H2: Demand for Propane is explained
temperature
• H3: Demand for Propane is explained by
prices
• H4: Demand for Propane is seasonal
Variables
• Dependent Variable
– Weekly quantity of Propane supplied to the market
• Independent Variables
– Weekly poultry slaughter counts
• Used as a measure of Agricultural impact on demand
– Spot Prices for Propane in Texas, the Midwest and
Northwest Europe
– Weekly temperature average for the United States
– Weekly temperature averages for the Northwest,
Northeast, Southwest and Southeast regions of the
United States
Variable Identification
• Endogenous
– U.S. demand for
propane
• Exogenous
– Temperature averages
– Poultry slaughter rates
– Propane prices
Methodology
• Used WinORS to analyze data
• Weekly data was collected covering the time
span between 2004 and 2007
• Stepwise regression was used to identify the
statistically significant variables
• Ordinary Least Squares was used to test for:
–
–
–
–
Normality
Homoscedasticity
Autocorrelation
Multicollinearity
Linear Demand Model
Qx = 2226.016 -2.352Tx -7.154Ux-5.108Nx127.42D1-262.222D2-122.885D3
Qx = Weekly quantity of propane supplied
Tx = Weekly spot price of propane in Texas
Ux = Weekly temperature average for the US
Nx = Weekly temperature average for the Northeast US
D1 = Spring Dummy Variable
D2 = Summer Dummy Variable
D3 = Fall Dummy Variable
*all other variables were determined statistically insignificant by the stepwise model
Preditictive Ability Graph
Constant Variance Graph
Statistical Significance and
Coefficient of Determination
• The P-Value is 0.00001 therefore it
satisfies the 99% confidence interval
Root MSE
SSQ(Res)
Dep. Mean
Coef. Of Var. (CV)
R-Squared
Adj R-Squared
154.774
3808838.392
1234.163
12.541%
74.089%
73.111%
Homoscedasticity and
Normality
• P-value for White’s is > .05 therefore the data is homoscedastic
• Correlation for Normality is below the approx. Critical Value
therefore the data is not normal (flaw of this model)
White's Test for Homoscedasticity
P-value for White’s
Correlation for Normality
Approx. Critical Value
6.81
0.86988
0.9945
0.999
Autocorrelation
• Durbin value should be > 2, in this case the value is
close enough to not reject the data
Rho
Durbin
Durbin H
D Lower Limit
D Upper Limit
Ho: Rho = 0
Rho: Pos & Neg
Rho: Positive
Rho: Negative
0.019
1.935
n/c
1.651
1.817
Do Not Reject
Do Not Reject
Do Not Reject
Multicollinearity (VIF)
• All values for VIF for all independent variables should be
less than 10 to ensure no multicollinearity
Independent Variable
VIF
Mont Belvieu, TX Propane Spot 1.042
Price FOB (Cents per Gallon)
Weekly Temp Avg - National
15.241
Weekly Temp Avg - NE
14.241
Summer Dummy
6.407
Spring Dummy
3.714
Fall Dummy
2.393
Elasticity
•
The elasticity of all the independent variables is inverse and inelastic
Independent Variable
Elasticity
Mont Belvieu, TX Propane Spot -0.18828
Price FOB (Cents per Gallon)
Weekly Temp Avg - National
-0.35136
Weekly Temp Avg - NE
-0.22362
Summer Dummy
-0.03033
Spring Dummy
-0.07443
Fall Dummy
-0.02297
Conclusions
• We reject the hypotheses that the demand for propane is explained
by poultry production or temperature
• We accept only the hypotheses that demand for propane is
seasonal and is explained by prices specifically the spot price of
propane in Mont Belvieu, TX
• The only flaw in the model is that the data may not be normal
because the correlation for normality is just below the approx. critical
value
• The spot price of propane in Mont Belvieu, TX is inversely inelastic
to the demand for propane in the U.S. with an elasticity of -0.18828,
therefore a 10% increase in the spot price would result in only a
decrease of 1.8% in demand nationally