Table 4.2: Descriptive statistics of hospitality and cultural firms within coffee shop municipalities after the implementation of the 2012 drug policy
Note: ∗p <0.1,∗ ∗p <0.05,∗ ∗ ∗p <0.01.Firm characteristic variables are reported in thousands of EUROS.
Table 4.2 presents the descriptive statistics (mean, mean differences, and standard deviation) of hospitality- and cultural firms in all-, treated-, and control coffee shop municipalities after the implementation of the local sales restriction in the post-treatment year of 2012. Though the discrepancy of the operating result between treated and control coffee shop municipalities has reduced, the issue is that the financial performance in other areas has decreased in 2012 as the net profit, the operating result, and the profit-and-loss account is reported as being statistically significant.
Prior qualitative studies have shown that the profits of firms located within control- coffee shop municipalities should increase in 2012 as a response due drug tourists being unable to purchase marijuana in Southern coffee shop municipalities. Additionally, qualitative studies on this theme have shown that the number of crimes associated with drugs increased after the initial introduction of the local sales restriction to only municipal residents. The descriptive statistics presented pro-vides a possible indication that the drug policy might have not had a significant negative impact on the profitability of firms within treated coffee shop municipalities.
However, due to data aggregation and that firms within specific industries may not be present in the same Southern coffee shop municipalities, the propensity score specification had to be changed in order to create a control group that could achieve ”strong ignorability” by dropping the variable median household income and the percentage change in the number of suspects in coffee shop municipalities.
Table 4.3: Propensity score balancing test - hospitality industry
Note: ∗p <0.1,∗ ∗p <0.05,∗ ∗ ∗p <0.01.U = Unmatched sample, M = Matched sample. Soft drug crime rate is the dependent variable. Nearest-Neighbor matching algorithm with 6 neighbors with a caliper size of 0.01.
As outlined by Rosenbaum and Rubin (1983). The balancing results presented in Table 4.3 indi-cates that most of the covariates are close or within the bounds of ”strong ignorability” indicating that the group of Northern coffee shop municipalities are similar to Southern coffee shop munici-palities after the matching procedure.1516
Table 4.5 represents the balancing test for firms in the cultural industry. As with Table 4.3, when examining the balance of the mean values, the bias difference, and whether the chosen char-acteristics are significantly different between Northern- and Southern coffee shop municipalities, the results indicate that the matching procedure was successful in constructing a control group that also fall within the threshold of ”strong ignorability”.17
15Except the number of residents that receive unemployment welfare at a bias difference of 8.1.
16In addition, I checked matching results for pair, kernel, and nearest-neighbor matching with 4 and 8 neighbors for matching consistency. Other algorithms give similar matching results indicating that the matching specification is robust.
17However, while the number of university students is almost at 10% (weak ignorability), literature indicates that importance of the variable to the matching procedure should be taken into account when determining if the balancing test has achieved acceptable results.
Table 4.4: Propensity score balancing test - cultural industry
Note: ∗p <0.1,∗ ∗p <0.05,∗ ∗ ∗p <0.01.. U = Unmatched sample, M = Matched sample. Soft drug crime rate is the dependent variable. Nearest-Neighbor matching algorithm with 6 neighbors with a caliper size of 0.01.
The matching result for the hospitality industry left a sample with 17 treatment- and 40 control observations leading to a total of 399 observations over 7 years from 2009 to 2015. The matching result for the cultural industry left a sample with 7 treatment- and 19 control observations leading to a total of 182 observations over 7 years from 2009 to 2015.
Table 4.5: Rosenbaum Bound Test on propensity score specification sensitivity - hospitality indus-try
Note: ∗p <0.1,∗ ∗p <0.05,∗ ∗ ∗p <0.01.Soft drug crime rate is the dependent variable.
I used the Rosenbaum bounds test to examine whether the propensity score specification was sen-sitive to the possibility of unobservable influences altering the probability of treatment selection.18 The Rosenbaum bounds test checks whether the propensity score specification is sensitive to the influence of an unobserved factor, Γ, that is introduced within the propensity score specification.
18The user package rbounds by Gangl (2004) was used for the Rosenbaum bounds test.
The notion behind the test is that while the magnitude of Γ increases, the assumption that the probability of treatment assignment being random due to observable characteristics being similar between treated- and control observations is invalidated which would imply that it could result in a counter-factual group that does not accurately mirror the treatment group.
The magnitude of Γ is reported in the first column of Table 4.5 with the corresponding upper-(SIG+) and lower bound (SIG-) significance which report the significance of the value of Γ. Low value of Γ would indicate that the selection process of the PSM model is significantly influenced by an unobserved variable indicating that if two observations with the same set of observed char-acteristics have different percentages of receiving treatment, this would undermine the assumption that the probability of assigning treatment is random. Duvendack et al. (2011) states that a value of 2.0 would indicate strong insensitivity to unobserved influences.
Table 4.6 presents the Rosenbaum bounds test for the propensity score specification for the hospitality firms. The value of Γ becomes non-significant in the lower-bound significance level when Γ is equal to 1.6. While the recommendation for the social sciences is a value of 2.0, a value of 1.6 would argue that the propensity score specification could be influenced by unobservable influence and that the propensity score specification might need to be re-evaluated.
Table 4.6: Rosenbaum Bound Test on propensity score specification sensitivity - culture industry
Note: ∗p <0.1,∗ ∗p <0.05,∗ ∗ ∗p <0.01.. Soft drug crime rate is the dependent variable.
Table 4.7 presents the Rosenbaum bounds test for the propensity score specification for the cultural firms. The Γ value becomes insignificant at 2.1 when examining the Lower(SIG-) column which would indicate a strong insensitivity to unobservable influences. However, it is important to note that the Rosenbaum bounds test merely test for the sensitivity of how strong an unobservable factor has to be before it can influence treatment assignment. This does not exclude that there is already bias within the specification itself.
Table 4.7: Effect of the Soft Drug Sales Restriction in coffee shop Municipalities on the Hospitality Firm Profitability (PSM-DID)
Note: ∗p <0.1,∗ ∗p <0.05,∗ ∗ ∗p <0.01.. Clustered standard errors are grouped at municipal level and shown in parenthesis. Treatment effect of the policy is captured by Equation 4.1 Columns 1 to 5 are measured in thousands of EUROS.
Table 4.7 presents the treatment effect of the policy on the financial performance of firms within the hospitality industry located within Southern coffee shop municipalities.
Column 1 shows the net profit of hospitality firms within coffee shop municipalities as the out-come variable and is not reported significant indicating that the policy did not significantly impact the Southern coffee shop municipalities when compared to Northern coffee shop municipalities.
Column 2 presents the cost of sales of hospitality firms within coffee shop municipalities and is not reported as significant.
Column 3 presents the wages, salaries, and premiums paid to employees working within the hospitality industry and is not statistically different from Northern coffee shop municipalities.
Column 4 presents the operating profit of hospitality firms within coffee shop municipalities as the outcome variable and is also reported as insignificant. Column 5 presents the net result of hospitality firms within coffee shop municipalities as the outcome variable and is also reported as insignificant.
From the results presented in Table 4.7, the foremost conclusion is that the policy implemented in 2012 did not have a significant impact on the profitability of hospitality firms. One possible explanation is that the drug tourists that were deterred through implementation of the policy were not considered to be a significant factor for the profitability of firms within the hospitality industry.
Another is that the reduction in profitability is only noticed within specific geographical locations within the municipality such as those closely located near a coffee shop.
In order to check whether the difference-in-differences model with propensity score matching satisfies the parallel trend between treatment- and control group as outlined in Section 4.3, I use Autor (2003) as seen in Equation 4.2 by introducing yearly interactions terms to see the whether there are any significant differences between the treatment and control group in 2010 and 2011 before the policy was implemented. If the coefficient reported by the interaction term for the year 2010 or 2011 is significant, this would indicate that the difference-in-differences model and propensity score matching did not satisfy the assumption and that there is significance factor that was not captured by matching on observable characteristics and thus results from the model would be considered biased.
Table 4.8: Hospitality Firm Profitability Before and After the Implementation of the Soft Drug Sales Restriction in coffee shop Municipalities (PSM-DID)
Note: ∗p <0.1,∗ ∗p <0.05,∗ ∗ ∗p <0.01.. Clustered standard errors are grouped at municipal level and shown in parenthesis. Treatment effect of the policy is captured by Equation 4.2. Columns 1 to 5 are measured in thousands of EUROS.
Table 4.8 shows the impact on the hospitality industry in Southern coffee shop municipalities and the possible fade-out of the treatment effect. Column 1 presents the net profit of the hospitality industry as the outcome variable and reports no significance in the pre-treatment years of 2010 or 2011. Other year interactions terms are also not reported as significant indicating that the policy had no significant effect on firm profitability. Though not significant, the change in the magnitude of the coefficient reported in 2011, 2012, and 2013 might indicate that the revenue of hospitality firms was slightly reduced when compared to the control group.
Column 2 reports the cost of sales of the hospitality industry as the outcome variable and the pre-treatment period years of 2010 and 2011 are reported as insignificant. While the year interac-tion terms of the coefficient reports no significance in the post-treatment years, when examining the difference in the magnitude of the coefficient in Column 2 between 2011 and 2012 shows a decrease in the coefficient’s magnitude following a similar pattern as Column 1 even though there is no reported significance. This seems to suggest that the policy may have had a small effect as the policy deterred drug tourists from visiting the Southern coffee shop municipalities. Prior research on this issue seems to suggest that drug tourists and the domestic clientele of coffee shops within Southern coffee shop municipalities traveled to the nearest coffee shop municipality that had not implemented the sales restriction of marijuana. However, without more precise details on the movement and purchasing behavior on drug tourists and domestic drug users, this is merely an assertion.
Column 3 presents the wages, salaries, and social premiums paid to employees working in hos-pitality firms. The change between 2011 and 2012 to a positive coefficient indicates that payment to employees for services rendered within firms located in Southern coffee shop municipalities are
larger than those in Northern coffee shop municipalities. However, there is no explanation without more in depth data to explain the large change in the coefficient’s magnitude to explain.
Column 4 presents the operating result of hospitality firms as the outcome variable and reports no significance in the pre-treatment period. As with Column 1, which reported the net profit, the magnitude reduction between 2011 and 2012 indicates that the profitability was reduced. There-fore, it is a possibility that firms within the Southern coffee shop municipalities have additional costs as the operating profit represents the earnings before interest and taxes. However, the coef-ficient in Column 4 reports no strong statistical difference between Northern- and Southern coffee shop municipalities.
Column 5 reports the net result of firms within the hospitality industry as the outcome variable and reports no significance in the pre-treatment period. As the net profit presented in Column 1, the implementation of the policy might have resulted in a strong change in the magnitude of the coefficient when comparing the magnitude between 2011 and 2012. Column 6 presents the net result which is the profit-and-loss statement of firms as the outcome variable as reports no significance in the pre-treatment period. As the net profit presented in Column 1 and operating profit in Column 5, the implementation of the policy might have resulted in a strong change in the magnitude of the coefficient when comparing the magnitude between 2011 and 2012. However, the magnitude changes in the following year indicating that even the relatively insignificant impact did not alter the financial performance of hospitality firms.
Table 4.8 shows that the estimation results are robust due to no reported significance in the pre-treatment period for outcome variables. While no significance was shown in the post-treatment period indicating that the policy did not have a statistical effect on the profitability of the hospi-tality industry, there was a strong reduction in the coefficients’ magnitude in 2012. Therefore, the policy might have reduced profitability when examining Columns 1, 4, and 5. However, the exact relationship between the different outcome variables indicates that other influences may have an significant impact on the profitability of firms.
The fixed effects in combination with the matching procedure should control for macro shocks to the economy but the aggregation from firm level to municipality level might obfuscate the results.
The other industry of interest is the cultural industry including institutions such as museums, art exhibitions, galleries, and other cultural related firms.
Table 4.9: Effect of the Soft Drug Sales Restriction in coffee shop Municipalities on the Cultural Firm Profitability (PSM-DID)
Note: ∗p <0.1,∗ ∗p <0.05,∗ ∗ ∗p <0.01.. Clustered standard errors are grouped at municipal level and shown in parenthesis. Columns 1 to 5 are measured in thousands of EUROS.
Table 4.9 presents the policy impact on the profitability of firms within the cultural industry using propensity score matching with difference-in-differences. Column 1 presents the net profit of firms within the cultural industry as the outcome variable and is reported as statistically insignificant.
Column 2 presents the costs of sales of firms in the cultural industry as the outcome variable
and is reported as insignificant. Column 3 presents the wages, salaries, and social premiums paid to employees working in cultural firms as the outcome variable and is also reported as insignificant.
Column 4 presents the operating profit of firms within the cultural industry as the outcome variable and is reported as insignificant. Column 5 reports the net result of firms within the cultural industry and is reported as insignificant.
From the results presented in Table 4.9, the policy had no effect on the profitability of firms within the cultural industry located within Southern coffee shop municipalities. Unlike the hos-pitality firms presented in Table 4.7 and 4.8, the difference between the cultural firms in either Northern- or Southern coffee shop municipalities is smaller when comparing the coefficient’s mag-nitude. However, due to no significance being reported, it corroborates the possible aspect that the policy did not have a significant effect on the profitability of firms that are traditionally tied to tourism. Therefore, the drastic reduction in drug tourism seems to have had no impact on the local economy of Southern coffee shop municipalities.
To ensure that the results presented in Table 4.9 are credible, I use the yearly interaction terms to check if there is a statistical difference between the Northern- and Southern coffee shop municipalities within the pre-treatment years of 2010 and 2011.
Table 4.10: Cultural Firm Profitability Before and After the Implementation of the Soft Drug Sales Restriction in coffee shop Municipalities (PSM-DID)
Note: ∗p <0.1,∗ ∗p <0.05,∗ ∗ ∗p <0.01.. Clustered standard errors are grouped at municipal level and shown in parenthesis. Columns 1 to 5 are measured in thousands of EUROS.
Table 4.10 presents the profitability of the cultural industry in Southern coffee shop municipalities.
Column 1 reports the net profit and is insignificant indicating that the policy did not have a negative effect during its implementation or in the following years. Column 2 presents the cost of sales as the outcome variable and reports no significance.
Column 3 presents the wages, salaries and premiums as the outcome variable and reports no significance indicating that the policy had no effect on the number of personnel working for firms.
Column 4 presents the operating profit of firms as the outcome variable and it is reported as insignificant. Column 5 presents the net result of firms as the outcome variables and is reported as insignificant.
The results from Tables 4.9 and 4.10 indicate that the policy did not have the expected negative impact as theorized by public officials and firms. Therefore, based on these results the policy can already be considered a success. In Chapter 3, the results indicated that while the policy initially increased the total number of crimes between 2010 and 2011, overall the soft drug crime rate was reduced. Possible factor that may explain the reason for insignificance might lie in the assumption that a major factor in their decision to visit the Netherlands is to experience soft drugs.19 Table 4.11: Effect of the Soft Drug Sales Restriction in coffee shop Municipalities on the Hospitality Firm Assets (PSM-DID)
Note: ∗p <0.1,∗ ∗p <0.05,∗ ∗ ∗p <0.01.. Clustered standard errors are grouped at municipal level and shown in parenthesis. Columns 1 to 5 are measured in thousands of EUROS.
Table 4.11 presents the treatment effect of the policy on the assets owned by firms within the hos-pitality industry located within coffee shop municipalities. Column 1 presents the intangible assets of firms within the hospitality industry as the outcome variable and is reported as insignificant indicating that the policy had no significant impact to cause a difference between the Northern-and Southern coffee shop municipalities. Column 2 presents the tangible assets of firms within the hospitality industry as the outcome variable and is reported as insignificant.
Column 3 presents the value of the inventory owned by firms within the hospitality industry as the outcome variable and reports no significance. Column 4 presents the firm’s liquidity as the outcome variable and reports no significant difference between the Northern- and Southern coffee shop municipalities indicating that the implementation of the policy had no effect.
19Due to the relative small size of the Netherlands, the fact that the Southern provinces are considered rural and most tourists spend their time in the Northern municipalities might be the reason why the policy did not significantly impact tourism within the Southern coffee shop municipalities. If tourists desire to experience soft drugs, the short distance between Northern- and Southern provinces should not impede tourists.
Table 4.12: Hospitality Firm Assets Before and After the Implementation of the Soft Drug Sales Restriction in coffee shop Municipalities (PSM-DID)
Note: ∗p <0.1,∗ ∗p <0.05,∗ ∗ ∗p <0.01.. Clustered standard errors are grouped at municipal level and shown in parenthesis. Columns 1 to 5 are measured in thousands of EUROS.
Table 4.12 presents the yearly treatment effect of the policy on the assets owned by firms within the hospitality industry. Columns 1 and 2 presents the intangible and tangible assets owned by the hospitality which are both reported as insignificant indicating the policy did not result in firms increasing the value of their assets. Column 3 reports the value of the inventory owned by hospitality firms and is reported as insignificant. Column 4 presents the firm’s liquidity as the outcome variable. Firm’s liquidity represents assets that can easily be exchanged to cash such as cash in the bank, investments, and so forth. While also not reported as significant, the magnitude of the coefficient is reduced when the policy is implemented in 2011 indicating that firms may have used their liquidity to overcome possible reductions in the net profit during the same year as seen in Table 4.7.