Download PDF - Ivie
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
supermarket placed in the same area where a consumer lives is T ,meanwhile<br />
the driving time to commute to a different area is aT ,wherea>1.<br />
Let p(c l |θ i ,c k ) denote the probability to buy in constituency c l of a consumer<br />
of type θ i who lives in constituency c k . From the assumptions, it is<br />
easy to check that:<br />
1 k = l<br />
p(c l |θ 1 ,c k ) =<br />
0 k = l<br />
a<br />
k = l<br />
p(c l |θ 2 ,c k ) =<br />
a+1<br />
1<br />
k = l<br />
a+1<br />
Then, the expected number of consumers who will buy in region c k are<br />
given by<br />
π(c k )=n 1k +<br />
a<br />
a +1 n 2k + 1<br />
a +1 n 2l<br />
k =1, 2,l =1, 2,k = l.<br />
Supermarkets’ problem to choose specific locations to open their new<br />
premises, may be modelled then as zero-sum game. Total sales of a supermarket<br />
depend then on its own location and on the location of its competitor.<br />
Let us denote by π i (c l ,c k ) the total sale of company i =1, 2 when supermarket<br />
1 is placed in region c l and supermarket 2 is placed in c k . We have<br />
then,<br />
π 1 (c l ,c k ) =<br />
π 2 (c l ,c k ) =<br />
1 n 2 1l + 1(n 2 21 + n 22 ) k = l<br />
n 1l +<br />
a n a+1 2l + 1 n a+1 2k k = l<br />
1 n 2 1l + 1(n 2 21 + n 22 ) k = l<br />
n 1k +<br />
a n a+1 2k + 1 n a+1 2l k = l<br />
The payoff can be written as a matrix as follows, where supermarket 1<br />
chooses rows and supermarket 2 chooses columns:<br />
c 1 c 2<br />
1<br />
c 1 n 2 11 + 1(n 2 21 + n 22 ), n 11 + a n a+1 21 + 1 n a+1 22,<br />
1<br />
n 2 11 + 1(n 2 21 + n 22 ) n 12 + a n a+1 22 + 1 n a+1 21<br />
c 2 n 12 + a n a+1 22 + 1 n 1<br />
a+1 21, n 2 12 + 1(n 2 21 + n 22 ),<br />
n 11 +<br />
a n a+1 21 + 1 n 1<br />
a+1 22 n 2 12 + 1(n 2 21 + n 22 )<br />
9