In: Economics
total expenditures per year ($1,000s, averaged over groups of breweries of same capacity) |
capacity (1,000 bb1s) per year |
195.22 | 1 |
810.27 | 2 |
591.78 | 3 |
1,342.35 | 4 |
1,915.61 | 5 |
2,003.14 | 6 |
1,727.58 | 7 |
2,863.54 | 8 |
884.61 | 9 |
3,508.42 | 10 |
1,216.58 | 11 |
4,170.70 | 12 |
3,831.40 | 13 |
3,963.27 | 14 |
5,409.94 | 15 |
4,871.02 | 16 |
5,532.12 | 17 |
4,654.85 | 18 |
4,008.82 | 19 |
4,571.65 | 20 |
4,459.94 | 21 |
5,948.31 | 22 |
4,774.38 | 23 |
2,159.74 | 24 |
3,690.03 | 25 |
5,756.83 | 26 |
3,980.78 | 27 |
4,218.47 | 28 |
4,816.53 | 29 |
8,391.56 | 30 |
8,542.18 | 31 |
4,627.00 | 32 |
9,564.62 | 33 |
6,231.67 | 34 |
2,902.75 | 35 |
5,392.48 | 36 |
2,991.46 | 37 |
1,144.31 | 38 |
1,660.64 | 39 |
7,650.07 | 40 |
2,752.20 | 41 |
13,112.65 | 42 |
13,652.02 | 43 |
5,102.94 | 44 |
10,290.01 | 45 |
12,307.84 | 46 |
6,235.05 | 47 |
7,582.25 | 48 |
6,344.05 | 49 |
5,909.06 | 50 |
13,162.89 | 51 |
8,955.16 | 52 |
15,875.48 | 53 |
7,617.45 | 54 |
4,553.70 | 55 |
3,568.83 | 56 |
20,227.45 | 57 |
14,570.19 | 58 |
5,826.64 | 59 |
6,750.42 | 60 |
8,385.15 | 61 |
9,710.43 | 62 |
13,921.87 | 63 |
25,251.10 | 64 |
16,335.71 | 65 |
11,041.33 | 66 |
24,352.56 | 67 |
10,122.02 | 68 |
10,518.31 | 69 |
23,886.06 | 70 |
19,895.86 | 71 |
17,598.34 | 72 |
28,384.11 | 73 |
30,223.77 | 74 |
22,292.95 | 75 |
19,171.24 | 76 |
25,677.27 | 77 |
21,597.64 | 78 |
38,039.96 | 79 |
38,107.86 | 80 |
27,894.94 | 81 |
31,374.81 | 82 |
35,944.08 | 83 |
24,587.37 | 84 |
25,887.29 | 85 |
28,862.45 | 86 |
39,225.46 | 87 |
41,124.94 | 88 |
33,463.49 | 89 |
33,827.72 | 90 |
31,207.26 | 91 |
42,523.71 | 92 |
51,865.68 | 93 |
59,537.79 | 94 |
41,635.65 | 95 |
58,594.86 | 96 |
44,490.04 | 97 |
54,895.81 | 98 |
68,593.77 | 99 |
72,877.54 | 100 |
Let us first plot the data as follows. We may observe that the output and cost have a non-linear relation. Hence, we may use higher order polynomial to estimate the cost function. Consider the following specification of the cost function to be estimated:
, where denotes capacity (1,000 bb1s) per year.
The regression results using excel can be presented below. The estimated resgression equation is:
As we may find, the linear, quadratic and cubic terms are statistically significant as their individual p-values are less than all conventional level of significance. Further, the terms are explaining more than 90% of variations in cost since the R-squared = 0.91. The overall regression is significant as evident by the p-value corresponding to the F-statistics. Hence, the above estimated cost function fits the data well.