Maximum Cap on Entry

I assume the introduction of a centralized policy which places a maximum number of entrants per subgroups at a given area. The maximum number is set to 2 per subgroup, thus a maximum entry of six coffee shops per year per tube station. I have introduced this shock (policy) in 2007 in order to compare the counterfactual results with the actual behavior of the independent coffee shops (actual data). Notice that the maximum entry of independents is 4 during this period (2007-2009), while for the C-Ss it is 2 and for the other C-Ss is 3. As a result this counterfactual analysis will mainly influence the independents entry behavior.

In order to simulate the response of the independent incumbents, after this shock, I have used the coefficients estimated from the panel probit and their error terms15_{. In a sense}

I have assumed that the unobservable terms, which are included in the error term, remain unaffected by the introduction of this shock. I have simulated in each period the probability of exit, summing over the coefficients which were multiplied by the unchanged variables and multiplied by the new Entry variables and then added the true error term. To be more specific, the equation that generates the prediction for a given year is (ignoring the time and

14_{See the following: http://camden.gov.uk/ccm/content/business/business-regulations/licensing-and-}

permits/licences

/entertainment-related-licences/tables-and-chairs-licence.en

Figure 4.2: Average Number of Independents per tube station

space subscripts for easiness):

ˆ

y= ˆb0+ ˆb1Xtrue+ ˆb2XN ew+ ˆε

Where ˆy are the predicted values (new values), the subscript “true” is for the true variables, “New” for the variables that are affected by the counterfactual analysis and ˆεis for the error term which was generated by the panel probit with the true data. Notice that for each year after 2007 I have recalculated the number of incumbents per subgroups (since the shock had an impact on the number of incumbents). I have then generated the average exit rate and multiplied that with the number of incumbents that had not exited in the previous period, according to the counterfactual result, but had exited according to the true data. This approach allows me to calculate the number of exits for a part of the sample where I lack estimated errors16_{. The next figure (4.2) plots the results of the exercise with respect to}

the number of independent coffee shops.

16_{I implicitly assume that the error term for this category of coffee shops is not different from all the others.}

Furthermore, by multiplying with the estimated exit rate I have assumed that these coffee shops (new entrants that become incumbents) will behave in the same manner with all the other incumbents in the market.

On the horizontal axis I have plotted the time period of the data and on the vertical the average number of independents per tube station. The red dashed line corresponds to the counterfactual number of independents per tube station (OwnindeppureNEW; this is the number of independents on average) and the black discontinuous line refers to the actual number of independents (Ownindeppure). Interestingly, the number of independents increases immediately after the shock (introduction of regulation) but in the following year it decreases to a point that reaches a lower level than the pre-shock period. Thus, the intervention fails to safeguard or even increase the number of the independents in the market since after 3 years the number of independents decreases by approximately 14% ((11-9.5)/11). Notice that this holds even though the number of exits has decreased17_{. It seems that the decrease on the}

number of exits is not enough to compensate for the decrease on the number of entries. The last result, supports the displacing and replacing ideas presented in the theoretical chapter of this topic. This is mainly because the least profitable and most inefficient incumbents are forced to exit from the market at an earlier stage. Consequently, the counterfactual results suggest that a policy of a maximum cap on the number of entrants will definitely generate a loss of independents.

Furthermore, notice that the entry of C-Ss and other C-Ss is largely unaffected by this policy and that the exit rates of these groups are lower than the exit rates of the independents. However, the entry decision of the C-Ss (small and large) is not endogenized. As a result I am expecting that the effects on independents from the presented policy will be smaller than the actual (real) effects from implementing this policy. To be more specific, by restricting the entry of independents I am expecting an increase on the entry of the C-S (both small and large) while in this scenario I assume that their entry remains unchanged. This will increase the likelihood of exit (more exits) and as a result decrease the number of independents even further. For example, assume a market with 4 independent entrants, 1 C-Ss and 0 other

C-Ss. By placing a cap I am imposing 2 independent entrants, 1 C-S and 0 (zero) other C-Ss. However, it should be expected that the C-Ss will respond to this policy by increasing their entry, for example 2 C-Ss instead of 1. As a result, I underestimate the effects on the number of independents in the market of Central London. However, the analysis is useful since it provides a maximum bound (independents are going to be necessarily smaller in number) on the number of independent coffee shops.

To conclude, this policy can be adopted by a planning authority in order to promote growth in other retail sectors, within a metropolitan area, that have been underdeveloped (for example efforts to establish a shopping centre). However, in the case of coffee shops this policy will largely benefit the C-Ss and the decrease of the coffee shops will be disproportionately against the independents, thus a loss on product differentiation. The driving force of this result is that the independents have a higher entry and exit rate. This means that by restricting entry the independents’ exit rate remains relatively high and the net result is a decrease on the number of independents. In other words, a selection process is constrained, a process which guarantees that the less profitable independent incumbents will be displaced or replaced by a more profitable independent and as a result the entry will cancel out exit. The next section presents a counterfactual exercise where the cap is more targeted.