Florists

Universiteit Maastricht
Faculty of Economics and Business Administration
Maastricht, 25 November 2009
Name: Leijs, D.
ID number: I555622
Study: International Business Economics
Course code: Industrial Organization EBC2005
Tutor Name: Kuypers, T.
Individual report

Table of contents:

1. Description of the industry page 2 2. The regression page 2, 3 3. Economic interpretation page 3, 4 4. A new variable page 4 5. Conclusion page 5

6. Appendix A (original output) page 6 7. Appendix B (output with extra variable) page 7

1. Description of the industry

The US florists industry includes about 22.000 retail flower shops with total revenue of $7 billion. The industry is highly fragmented, which means that no firm dominates the market and there is a lot of competition. This is also confirmed by the fact that the 50 largest firms only count for 6% of the market. Entry barriers are low, one can entre easily without dealing with high sunk costs or doing large investments. Opening a shop will cost you only $30000 or $40000 to get started.
Most important products are cut flower arrangements (count for 55% of industry revenue), potted plants (count for 15% of industry revenue) and loose cut flowers. Often, stores also sell vases, artificial flowers and other gift items. Best selling flowers are roses, carnations and lilies.

The demand for flowers depends mainly on discretionary consumer spending. That is why shops have to marketing themselves. Individual shops also have to compete with supermarkets and other mass merchandisers who can sell flowers at lower prices because of economies of scale. Because consumers are not willing to travel large distances for their flowers, most of the firms in the florists industry serve local markets.

The industry of florists is labour-intensive. Employees have to put together flowers arrangements, take orders, care for flowers, and serve walk-in customers. Sales per employee are about $60.000.

2. The regression

Following the second regression equation of Carree, we find a summary output. This output is added in appendix A

To get to the number of establishments (N), we first find our industry-specific constant, which is -6,0981 in the florists industry. Second, the logarithm of the population of all US states is taken. Because this is our highest coefficient in the summary output, population is our most important variable. Which of course makes sense: an increase in people (and so in potential customers) in a certain state implies more local markets available for entry. Third, we see that income is also an important determinant. When real income per capita will go up with 1%, the number of florists is almost going up with 0,5%. Because real income is also a real number, again the logarithm is taken to come to a percentage. Looking at our density variable, we see this is a negative one. This negative relation is straightforward: when states have a higher level of concentration of the population, there will be fewer establishments necessary because the average distance between consumer and establishment is lower. The same is true for urbanization. So far our model is as expected. Together with real income per capita, poverty measures the purchasing power of the population. Because poverty already is a percentage of our population, this number is used in our model and no logarithm is taken. Also the percentage of people over 65 years is added to our model, this one measures the distribution of the population. Expected was a positive or negative relation, depending upon the industry. In the case of florists it is a relatively small, positive number. Older people tend to be related to the number of establishments in the florist industry.

Looking at our regression statistics, we see that two out of 6 variables are not significant at a 10% significance level: density and poverty have respectively p-values of 0,4964 and 0,6236. The relation of density, as well as poverty, to the number of establishments (N) is not meaningful. All other variables are significant and have very low p-values. This can be regarded as very strong evidence that the regression relationship is significant. Our R2 of 0,9877 confirms that the power of our regression is large, as 0,9877 almost equals 1.

Our F value of 560,5064 is larger than F0,1. It is even larger than F0,01 and therefore we have extremely strong evidence that our regression model is significant, and so the relation between our intercept and the other variables.

3. Economic interpretation

Variables that were found to be significant are population, real income, urbanization and people over 65. These variables make sense: population is significant in every industry because of the scale effect. Real income is significant because flowers are in some way luxuries. Flowers are not necessarily to survive, but can add something extra to life. Urbanization is highly significant: when average distance between consumer and establishment is lower, fewer establishments are needed. For our variable over 65 there is strong evidence that the relationship is significant. With a p-value of 0,0017, over 65 is one of the top three most strongest regression relationships, together with population and urbanization.

4. A new variable
Appendix B includes the summary output with a new variable: female. Adding this variable from 1993 to our original model makes our regression more accurate. Because our variable ‘female’ deals with a p-value of 0,0865, it is significant at our 10% level. Adding a ‘male’ does in this case not make sense, since males have a p-value of 0,1024. By out regression it is proven that there is a positive relation (coefficient of 0,0454) between the number of establishments and the number of females living in a certain state. Therefore, we add this variable to our model.
But there is more. As can be seen in our output, when our new variable is added, R2 goes up from 0,9877 in our original model to 0,9885 in our new model. The power of the regression is thus increased and our variables explain 98,85% of the total variation in the number of establishments. Our standard error is going down, which is also a positive fact. This number decreases from 0,1096 to 0,1070 and there are less errors in the regression.
Again, F shows that there is extremely strong evidence that our regression model is significant; our value of F is even larger than F0,01.
In our summary output we have a t value of 1,7563. If we wish to carry out a test at the 0,1 level of significance, we have to use the critical value t0,1/2 = t0,05 = 1,683 which is based on
49 – (7 + 1) = 41 degrees of freedom. Because in every case (leaving density and poverty behind) |t| > 1,683 (in this case, a minus does not indicate a negative number) and the p-values related to the intercept are less than 0,1 we have evidence that our variables are significantly related to our intercept.

5. Conclusion

From our summary output due to the second equation of Carree, it is proven that the industry of florists is in high competition. This can be a consequence of low entry barriers. In the output R2 is high, which will not be the case in a monopoly or other dominant markets. There are two insignificant variables: density and poverty. They tend to have no real influence on the number of establishments of florists. Population, real income, urbanization and people over 65 do change the number of florists in a certain state. After adding the number of females, the power of our regression is increased and we have 5 out of 7 variables that are significantly related to the number of establishments which makes this a powerful model.

Appendix A

Appendix B