# U.S. Retail Coffee Market (A)

Topics: Prediction interval, Statistics, Consumer price index Pages: 5 (850 words) Published: February 23, 2014

1.(A) MARKET SIZE OF TOTAL RETAIL COFFEE MARKET IN 1983 AND 1988

MARKET SIZE IN 1983 – \$ 5441.4 ± 248.3 MILLION
MARKET SIZE IN 1988 - \$ 9169.2 ± 248.3 MILLION
Total Retail Market size for coffee for the producers will mean the total sales in dollars that will be achieved in the desired years. To forecast these sales, we need to determine first what sales are dependent upon.

As a first step, we examine a plot of sales in \$ over previous years (EXHIBIT A). We see that 1977 and 1978 show unusually high sales. This can imply that sales do not necessarily depend on time. This is confirmed by a regression of sales in \$ with time. Even though the R-squared value is not low 56.9%, the actual sales for 1978 does not lie in the 95% confidence interval predicted our regression (=(-7016.76 + 129.2*78 + 2(619))) (EXHIBIT B). From reading the case study, price seems to be a big factor determining sales. The Brazil frost disaster that destroyed supply of coffee and as a result raised prices can explain the surge in sales \$ in 1977 and 1978. It would be reasonable to resume that there is a relationship between prices and sales in \$.

Running a regression of sales in \$ with price (EXHIBIT C), we find a VERY strong positive relationship with a very high value of R-square 98%. This means that 98% of the variation in the sales in \$ data can be explained by prices. The result is statistically significant as absolute value of the t-statistics is above 2. Thus we establish the following relationship: Sales = 405.7 + 1539.3(price) ± 2(124.15)

To predict sales in \$ for the years 1983 and 1988, we need to estimate the price for these years. According to the case,
“One major producer predicted continuing price declines in fiscal 1979 to an average of about \$2.55, followed by some increase to about \$2.70 in fiscal 1980 and 1981. Prices might then be expected to inflate at the same rate as the US Consumer Price Index, barring supply disruptions.” Assuming that this is accurate and that no supply disruptions occur in the future, we can get the price for 1983 and 1988 by inflating the 1981 price of \$2.70 by CPI percentage change of each year. To calculate the CPI percentage change, we run a regression of CPI % change with time (EXHIBIT D). We get the following statistically significant positive relationship with a high 68.5% r-square value:

CPI % change = -28.7 + 0.47(year)
When we plug in the years 78-88 in the formula, we get the corresponding CPI% change for those years (EXHIBIT E (i)). We have assumed that assume that after 1981 the percentage change in price for coffee has been equal to the percentage change in CPI (EXHIBIT E (ii)): CPI % changet = Price % changet

Using these price % change, we can calculate the price(t) for a given year with the help of the price from the previous year price(t-1) using the following formula (EXHIBIT E (iii): Pricet = pricet-1(1+ (%change in pricet/100))

Once we have Pricet we use our regression model from EXHIBIT C to get Salest (EXHIBIT D(iv)) and since we know there might be some error in the forecasted sales, we use 95% confidence intervals around our final answers.

(B) SPLIT OF MARKET BY MAJOR SEGMENTS IN 1983 AND 1988
The major segments in the Retail Coffee Market are:

Caffeinated v Decaffeinated (EXHIBIT F)
o Running a regression (EXHIBIT G) of Caffeinated segment share with time we get: Caff Sharet = 135.27 + (-0.62(t))
Plugging in the desired vales we get:
Caffeinated
1983

83.81%

16.19%

1988

Decaffeinated

80.71%

19.29%

Ground v Soluble (segmented into freeze-dried v spray-dried) (EXHIBIT H) o Running a regression (EXHIBIT I) of Ground segment share with time we get: Ground Sharet = 110.35 + (-0.64(t))
o Running a regression (EXHIBIT J) of freeze-dried segment share with time we get: FD Sharet = -69.78 + (1.07(t))
Plugging in the desired vales we get:
Ground

Soluble
Spray-dried

Freeze-dried

1983

57.23%

23.74%

19.03%

1988...

Please join StudyMode to read the full document