Analysis of Gas Prices in Howard County, Maryland


Appendix A—Economic Modeling Assumptions and Explanations



Download 0.59 Mb.
Page9/21
Date29.07.2017
Size0.59 Mb.
#24498
1   ...   5   6   7   8   9   10   11   12   ...   21

Appendix A—Economic Modeling Assumptions and Explanations

A.1 Assumptions


RESI assumed the following when analyzing the changes in retail prices of gas stations.

  1. Infrastructure changes did not occur during the period. The period reviewed in the annual analysis is too short to determine if there were major infrastructure improvements in an area, but RESI can determine that the stations’ locations have been in existence for quite some time. RESI assumes that to change infrastructure would take longer than the current study period, and therefore did not add a variable for this into the analysis.

  2. Gas stations have perfect information about competitors’ prices. RESI assumes, since prices are posted for the public to view and information is available about gas prices in a region through website such as AAA and GasBuddy, that retailers in an area have perfect information about what their competitors’ prices are on a given day or time.

  3. Rack prices are reported before the next period, and retailers will have a chance to change their prices accordingly. In the 2012 cross-sectional model, RESI assumes that the change is very instantaneous but that the declining prices may hinder changes in retail prices as retailers may worry about sudden shocks and price reversals to the higher historical values.

  4. Gas stations’ competitors and locations are known. RESI assumes that any new placement of gas stations will do so with zoning restrictions and adhere to an optimizing Hotelling location model.36 Hotelling proposed that, given two firms located at separate ends of a line, one firm’s potential for sales to consumers would depend on the distance between that firm and its competitor, as well as which way consumers traveled. For example, take the line below with two gas stations.



β

α

B

A

Hotelling stated that if the firms are unable to change position and the distances are recorded as α and β, then those distances times the cost to transport the good (or in this case search for the good) must not exceed the price offered by either A or B.37 If the good is homogenous, consumers will tend to go to the closest firm to avoid extra costs, but if the distance is equal, then they will choose whatever firm is cheapest.38

A.2 Results


Figure : Annual Analysis Regression on Margin for 2002–2012

Variable

Impact Multiplier

t-statistic

Statically Significant at 5% (Y/N)

Intercept

-3.011

-4.575

Yes

Household Income Change

0.144

0.330

No

Percent Unbranded

-0.177

-2.264

Yes

Stations per 1,000 Residents

0.466

3.381

Yes

Population Change

-10.023

-3.066

Yes

Dummy (Anne Arundel)

-0.393

-6.420

Yes

Dummy (Baltimore)

-0.372

-8.546

Yes

Dummy (Baltimore City)

-0.303

-5.278

Yes

Dummy (Carroll)

-0.385

-6.199

Yes

Dummy (Frederick)

-0.116

-1.421

No

Dummy (Harford)

-0.676

-9.727

Yes

Dummy (Montgomery)

0.406

5.586

Yes

Rack Change

0.545

8.500

Yes

Tax Current Period

0.584

1.649

No

Tax Lagged-One Period

-2.472

-4.388

Yes

Sources: RESI, OPIS, U.S. Census
Figure : Annual Analysis Regression on Margin for 2002–2012 Statistical Parameters

Statistic

Estimation

R-square

0.871

F-statistic

31.196

Durbin-Watson

2.201

Sources: RESI, OPIS, U.S. Census


Figure : Cross-Sectional Analysis Regression on Margin for 2012

Variable

Impact Multiplier

t-statistic

Statically Significant at 5% (Y/N)

Intercept

3.386

6.814

Yes

Population in thousands

0.000

4.711

Yes

Household income per 1,000 residents

0.089

1.461

No

Difference in competitor prices

0.434

10.650

Yes

Percentage of Unbranded

0.212

3.577

Yes

Percent Permitted

-0.439

-8.598

Yes

Percent CUP

-0.097

-7.128

Yes

Distance to Major Road

-0.005

-5.040

Yes

Distance to Next Station

0.015

5.017

Yes

Rack Prices

-1.103

-6.580

Yes

Dummy—located within a conditional use permit zone

0.015

2.214

Yes

Dummy—located within a nonconforming zone

0.012

1.667

No

Dummy—located within Columbia, MD

0.107

9.358

Yes

Sources: RESI, OPIS, U.S. Census
Figure : Cross-Sectional Regression on Margin for 2012 Statistical Parameters

Statistic

Estimation

R-square

0.637

F-statistic

137.747

Durbin-Watson

1.814

Sources: RESI, OPIS, U.S. Census



Download 0.59 Mb.

Share with your friends:
1   ...   5   6   7   8   9   10   11   12   ...   21




The database is protected by copyright ©ininet.org 2024
send message

    Main page