**2. Literature Review**

Although the goal of this paper is intertwining success factors and local policies with economic results of cycle logistics projects in Europe, the literature review proposed in the above section embraces a wider perspective. In fact, it considers those factors and policies also affecting the adoption of bikes for the mobility of people as well as active travel behaviors at large. This was a deliberate methodological choice geared to adopt a more comprehensive perspective of analysis in order to capture all possible elements affecting the cycle logistics field. In fact, this approach avoids the siloization of cycle logistics-related factors and policies: they cannot be and are not considered separately from the "whole picture", which includes the use of bikes for personal purposes and active travel behavior at large. In fact, many factors and policies concerning cycle logistics are included in more comprehensive political programs, affecting the wider area of bike mobility and active travel. Finally, the reason behind this wider slant of analysis is justified by considering that cycle logistics solutions for freight transport always require some supporting policies from public authorities as well as the fact that they implicitly concern the adoption of bikes by individual users. Therefore, all of those factors and policies affecting the overall use of bikes for individual purposes prove to be relevant to cycle logistics solutions.

The search strategy implemented in Scopus, the Elsevier database, included the following terms: cargo bike, cargo-bike, cargobike, cyclelogistics, cycle-logistics, cycle logistics, cycle mobility, bike + mobility, active travel, bike + economy, bike + policy, cargo cycle, cargo-cycle, cargocycle. The resulting documents were then selected by analyzing their contents with a qualitative review, and all the co-authors independently reviewed each selected paper. Finally, they shared their independent evaluations with each other and identified the documents to be used for the literature review. Moreover, additional searches were con-

ducted on the Internet by checking for additional relevant sources—e.g., from specialized websites—to integrate with the initial documents.

#### *2.1. Experiences from the UK and Ireland*

Local authorities in the UK have targeted large logistics companies in order to incentivize them to adopt cargo bikes into their supply chain. Moreover, communication campaigns of cargo cycle programs at the local level have emphasized both perception issues and lack of awareness and regulation as factors impacting negatively on the implementation of cycle logistics initiatives [7]. Ref. [21] demonstrates that promoting policies in the UK should address three pillars: incentive system, risk perception and availability and maintenance of infrastructures. Some of these key factors are common among cycle logistics and cycling at large, despite significant definitory differences and obstacles that characterize each strain. Finally, extant studies found some relationships between the impacts of both cycle logistics and cycling at large, but the definitory elements are not shared among them: it is a question of convergence of two different strains (cycle logistics and cycling at large) toward some similar results [18,21–30].

An additional factor lies in the impacts: an inverse relationship between the adoption of bike delivery solutions and the number of obese citizens, including bike messengers, has been found, thus proving the broad extent of social impacts partly attributable to cycle logistics [22–26]. In fact, cycle logistics projects in the UK have several impacts, since they contribute to reduce the pressure on the National Health Service; delivering goods by bike is positively linked to health benefits and is proven to help tackle urban mobility issues, which are directly responsible for 70% of substances threatening public health [21,27]. In Scotland, different research studies have identified a complex set of shared impacts—e.g., demographic, economic and infrastructural—related to both cycle logistics and cycling at large [18,28–30]. In Ireland, cycle logistics have been included in two ad hoc governmen<sup>t</sup> programs in order to increase both individual and socio-economic benefits [9]. Tackling safety issues in Irish urban transportation networks is key in order to build on the reputation of cycling as safe in cities and, especially, to ensure the business viability of bike delivery projects [18]. In this context, risk perception, infrastructures and attitudinal aspects have been identified as key factors [18,31,32]. However, the current strong political will in many towns in the UK is not the only enabling factor supporting cycle logistics and cycling at large. In fact, the traditional good public transport systems and the urban infrastructures adverse to car use in city centers act as complementary factors in the British scenario since they help in decreasing traffic levels and, at the same time, enable the timely delivery of goods and, hence, the increase in customer satisfaction for cycle logistics businesses [33]. On the contrary, public transport proved to be poorly designed in Ireland since it provoked a 29.7% drop for non-motorized commuting and a 37.5% increase in car use from 1986 to 2006, thus revealing a negative context factor and discouraging the start-up of bike delivery businesses [9,34]. Many authors have investigated socio-economic and transportation- and household-specific factors in major Irish cities: supporting policies, infrastructures, car ownership and socio-demographic status at large have been identified as some of the more relevant issues impacting non-motorized transport, thus including cycle logistics [9,35–39].

#### *2.2. Experiences from Greece and Italy*

An analysis of factors impacting the adoption of bikes as a standard transportation means in Greece suggested that women's eco-cyclist inclination tends to make them ask for bike delivery services more often than men [40]. Nonetheless, the existing literature about the Greek scenario sheds light on how the gender factor is mitigated by other variables— i.e., demographic, economic and environmental—and gender may play a varying role depending on the relative significance of such factors interacting with it [40]. Demographic and cultural ones—e.g., marital and education status—often have a low impact, while age may be associated with environmental concern [41]. Finally, low income has been shown

to be the most relevant economic aspect affecting the demand of bike delivery services, thus giving policy makers and managers a relevant insight into unexploited targets for new mobility solutions in Greece [40].

In Italy, factors such as the network and the topological shapes of many cities made it necessary to test several pilots which emphasize the environmental and social benefits of such initiatives [42]. These Italian projects build on previous research, which has already demonstrated that 51% of all trips for goods transportation in European cities can be realized by bike, and 19% to 48% of the overall mileage currently performed by combustion engine vehicles can be done by cargo bike [7,43]. In this context, the main factors determining the success or failure of pilots were the size and weight of goods to be delivered compared to the load capacity of cargo bikes; the relevance of time windows for the delivery; the impacts on brand image and corporate social responsibility; cost levels; availability of a supporting network and reliability of enabling technologies [42]. Finally, the cycle logistics scenario in Italy proves to be primarily affected by the social (e.g., visibility and green image, low energy consumption, service quality and coverage), physical (e.g., load capacity vs. goods size, weight and number, technological reliability and battery duration) and political (e.g., better quality of life for citizens, re-use of public facilities and incentives) factors of the socio-ecologic model proposed by [16].

#### *2.3. Experiences from Scandinavia*

An analysis of Scandinavian countries, especially Denmark, showed that accessibility and availability of both safe infrastructures and parking facilities—together with high urban density—enable the start-up and development of cycle logistics projects as well as the adoption of bikes at large [20]. Other authors recognize the key role of supporting policies [16,44–46].

#### *2.4. Experiences from Central Europe*

In the Netherlands, the use of bikes for the mobility of both goods and people already accounted for 30% of overall local trips in 1997, thus showing a strong cultural integration of bikes into Dutch society [47]. As for cycle logistics projects, commercial deliveries in Dutch cities are generally planned for short trips within the 3.5-km threshold, while in Germany, they are feasible within 2 km [47]. Differences between the average trip thresholds in the Dutch and German scenarios are due to cultural factors as well as urban density and infrastructural factors [47]. In Germany, recent research efforts on inner-city courier shipments have identified the specific vehicle choice of "messengers"—i.e., freelance couriers—as one of the main drivers for the adoption of bike delivery solutions [48]. In turn, vehicle choice is affected by several variables at the individual—e.g., demographic, attitudes and values—technological—e.g., accessibility and availability of enabling technologies—and economic—e.g., price and availability of information—levels [48]. As for technological aspects, technical innovations adopted by cycle logistics initiatives in Germany and France have been combined with new concepts and configurations of urban mobility systems. They have been successful, especially when associated with urban micro-consolidation centers as well as technologies for reduced driver fatigue and improved range and payload [48–51]. However, recent studies have shown that effective commercial transport solutions in city centers always come out of a multitude of factors harmoniously combined with each other, such as organizational structure of supplier firms, demand patterns, technical prerequisites and cultural inclination to accept a modal shift from customers, firms and messengers [48].

#### *2.5. Experiences out of Europe: Australia and the United States of America*

Many research works report that Australian bike-based businesses are worse positioned compared with their competitors in North America, China and Europe [6,52–54]. The findings sugges<sup>t</sup> that adverse reactions to safety helmets being compulsory together with trip distance may affect messengers' vehicle choice and undermine the success of

Australian cycling and cycle logistics programs (see also [39,55,56]). Other studies conducted in Australian and American cities sugges<sup>t</sup> that individual factors, including messengers' vehicle choice, are key in order to nurture bike delivery or bike sharing schemes as well as cycling at large [41,56–65]. As for mandatory helmet usage, it is not perceived as an advantageous factor because of the need for either purchasing or always carrying such safety accessories; studies show that it is detrimental in both Australia and the USA, with other factors being equal [17,55,66–68]. Moreover, urban density and availability of infrastructures are also recognized as relevant factors in Australia and the USA [69–71].

However, commercial deliveries in the United States are not effective because of the noticeable differences with the network and topological shapes in European inner cities [72–74]. Although there has been a sound debate about primary causes of the flop of Australian programs, the field research is still poorly grounded and calls for further empirical research efforts.

#### *2.6. Experiences from Asian Countries*

As regards the Malaysian scenario, customer needs and political factors play a key role—e.g., need for door-to-door transport, spread and availability of dedicated infrastructures and environmental aspects [4,75]. Other studies confirm that socio-economic impacts were found to be significant [4,76].

As Western societies have shown a strong commitment to cycle logistics and cycling at large (see Sections 2.1–2.5), likewise, Asian ones have proven to be strategically engaged in them as well because of their relevance to national agendas [77–79]. Policy implications also call for a dramatic change in population distribution in some Asian cities, such as in Malaysia, where a foreseeable "donut cake" distribution due to the downstreaming phenomenon in urban centers will generate a downsizing of economic activities and residential density in downtown areas, impacting, in turn, the operations and the development of bike delivery businesses [4]. Downstreaming in cities will be amplified by other local policies such as priority being afforded to motorized vehicles and lack of non-motorized promotion, resulting in a clearly disadvantageous scenario for cycle logistics projects [4]. These results call for taking the opportunity to design transportation networks suitable for cycle logistics and integrating them with existing road networks in Malaysia. Moreover, policy makers should push towards the adoption of priority policies reversing the current ones, as it happens in Europe—i.e., ensuring priority to cyclists and bike messengers, excluding them from turning or one-way direction constraints [4].

However, Malaysia and European and Northern American countries are not the only ones to cope with political, safety and socio-demographic factors, since they have proven to be relevant also to the Japanese national agenda [80,81]. In particular, the lack of effective safety regulation in Japan called for the implementation of active safety policies and countermeasures which also sugges<sup>t</sup> an increase in the viability of commercial deliveries in cities [80,82–85]. This way, it has been possible to adopt more effective policies in order to tackle both bike messengers' and goods' safety in Japan as well as it is being done in other Western societies such as Ireland and in developing countries [18,20,28,35–38,47,86–89].

#### *2.7. Experiences from Developing Countries*

In the last century, policy makers in developing countries have focused on motorized transportation, thus promoting and designating urban development as a hindering context for cycle logistics projects. This approach to planning and implementation of activities has dominated the transportation arena, even though non-motorized travel is even more significant in developing countries than in Western societies. The reasons behind this have been recognized to be the poorly grounded literature and the poor dissemination of research results in this field, also at the power elite level [89]. Moreover, policy makers hold their responsibility since they did not relevantly take into account the positive impacts of non-motorized transportation in terms of environmental, energy and socio-economic benefits. Therefore, they neglected the need of promoting the start-up of

bike delivery services—e.g., by incentivizing them or realizing supporting infrastructures. Their carelessness towards the multifaceted negative effects of traditional combustion engine vehicles—e.g., increase in congestion, energy consumption, pollution, costs, pressure on health system, safety and security—is even more alarming [90,91]. Policy makers and people at large have perceived, for many decades, that the dilemma between motorized versus non-motorized transportation could be associated with rural versus urban and developed versus non-developed, while the most advanced studies prove that cycle logistics may provide a significant contribution to cope with many issues affecting urban contexts worldwide [89].

A further challenge appears on the horizon: most cities in some developing countries, e.g., India, are not able to satisfy the need for investments and measures concerning infrastructures, safety, land use and incentive systems geared to serve their growing cycling population, including bike messengers, and the overall viability of their delivery services in inner cities [37,86,88,89,92–101].

In conclusion, where policy makers overlook the need for an integrated set of interventions geared to promote cycle logistics and cycling at large, this may result in limited success or even in failure of supporting policies. Sometimes, this political issue comes from the desire to maintain good relationships with relevant shares of voters, and other times from evidence of weak political capacity [102–106]. Finally, the paradigmatic shift towards bike delivery services is linked to a set of factors and interventions to tackle them. In the following, all relevant factors and policies are analyzed together with their relative significance across geographical areas and countries. Furthermore, corresponding measures are identified that prove effective in modifying the ways in which goods are delivered in cities. In this context, the comprehensive subject of this paper—i.e., Europe-wide analysis—is particularly suitable for defining policy implications in the broader field of sustainable urban transport. The focus on cities and urban areas at large is further justified by recent research results showing that even European or nationwide transport policies depend on their success at the local level [6,9].

#### **3. Data Collection and Methodological Approach**

We must point out that the profit and profitability variables help when assessing the potential of an individual project to achieve business objectives or to produce an economic result based on both an effective and efficient use of resources, within the context in which such a project is implemented. In fact, in general, profit in itself is not sufficient to prove the economic appeal of an investment in a project and whether it is worth pursuing, except when dealing with a business company. On the contrary, the concept of profitability applies, more generally, to all kinds of economic organizations, including non-profit. Profitability has been calculated as a dimensionless value according to the standard definition—i.e., sum of present value of cash flows over 5 years divided by initial investment. On the contrary, profit is not dimensionless—the currency is euros—and it has been calculated by subtracting the normalized costs in the Cyclelogistics project (e.g., for bike purchasing, maintenance, insurance and messenger pay) from the overall income [107,108].

Such hypotheses depending on profit and profitability as economic, financial results are tested by utilizing normalized data that are calculated at the European level [108]. Therefore, such data do not reflect any region-specific features, as they are generated by definition and by construction with a normalization process throughout the Europe-wide scenario, which was the context of the analysis considered by the source study [108]. To the best of the authors' knowledge, the data included in this research document are the most comprehensive and reliable, so far, and are endowed with an official element, being the result of activities supported also by the European Commission.

Both profit and profitability are calculated with reference to the specific organization implementing the cycle logistics service. Therefore, profit and profitability refer to the organization's business/project level, whereas those values of profit and profitability are registered as economic, financial results.

The overall methodological approach and the corresponding steps are summarized below (Figure 1) and discussed in detail in the rest of this section.

**Figure 1.** Methodological approach and stages.

Following this approach, in the first stage, critical factors and policies mentioned in Section 2 have been clustered by area and country. The results are shown in Appendix A (Table A1) and Appendix B (Table A2).

In the second stage, both critical factors and policies relevant to those projects run in Europe have been selected and grouped by category (Table 1a,b).



In the third stage, beyond factors and policies identified with the above analysis, a complementary study has been conducted in order to take into account also projectspecific factors characterizing 50 cycle logistics solutions implemented in Europe (Table 2a). The main type of bike is considered as the key feature since other project-specific factors— such as size range of delivery, ease of driving and parking, price per delivery, cost per bike, etc.—result to be affected by it. In particular, the bike models considered are traditional, trailer and cargo bikes as well as tricycles. Traditional bikes are those standard models used also by urban citizens for their own private purposes. If used for commercial deliveries, they only allow transporting small-sized goods and in a limited quantity, since they do not have large cargo accessories, neither in front nor behind. They are generally cheaper

than other bike models in terms of purchase, maintenance and insurance costs and ensure a higher speed and ease of driving and parking in cities. Cargo bikes have a large box, generally in front, which allows transporting even big-sized goods as well as a large quantity of small- to medium-sized items. They are much more expensive than traditional bikes and more difficult to park and drive. Trailer bikes lie in the middle between traditional and cargo ones and are endowed with a small cargo behind. As for tricycles, they are the largest and more capacious bikes as well as the ones with the highest costs. Moreover, they have the lowest average speed as well as the lowest ease of driving and parking in cities.

**Table 2.** (**a**) Project-specific factors (sources: [107,109] and projects' websites). (**b**) Average economic arguments from [108]. (**c**) Average data on the European cities scenario from [108].



**Table 2.** *Cont.*

In the fourth stage, economic results have been calculated and normalized with reference to the data referring to the average European city as provided by the Cyclelogistics project [108], as reported in Table 2b,c. In detail, profit-and-loss data have been calculated and normalized coherently with the consolidated average economic arguments on page 19 of the aforementioned data from the Cyclelogistics project [108], whilst the average data on the European cities scenario have been extracted from page 9 of the same reference [108]. In this context, normalization means using the same average economic arguments in the aforementioned report in order to calculate profit and profitability of cycle logistics projects. This way, the methodological approach adopted allowed us to compare project data homogeneously. Profit and profitability measures for each project have been calculated over 5 years depending on available data from the Cyclelogistics project [107,108] and by rounding yearly profits down with a 0.1 correction coefficient. This way, we have reduced the overall profit (and also the profitability inferred from it). This helps us in further challenging our research hypotheses by assuming a more pessimistic scenario in terms of profit and profitability levels as well as further mitigating the risk of their overestimation. In addition, profits have been prudently calculated by considering only the share of bikebased trips explicitly devoted to goods delivery in European cities, thus assuming a scenario with an even more pessimistic underestimation.

At first, these assumptions could sound like an attempt to simplify calculations in order to be able to perform a rough estimate and comparison among projects. Anyhow, as a matter of fact, that is currently the only way to perform such an analysis because of the lack of sufficient and homogeneous data. Such a shortage of data concerns both the non-profit projects funded by the European Commission and those projects run by private start-ups. Profit and profitability have been calculated in order to obtain an insight into the potential of each individual project to achieve business objectives or to produce an economic result based on both an effective and efficient use of resources. In fact, in general, profit in itself is not sufficient to prove the economic appeal of an investment in a project and whether it is worth pursuing, except when dealing with a business company. On the contrary, the concept of profitability applies, more generally, to all kinds of economic organizations, including non-profit. Profitability has been calculated as a dimensionless value according to the standard definition—i.e., sum of present value of cash flows over 5 years divided by initial investment. On the contrary, profit is not dimensionless—the currency is euros—and it has been calculated by subtracting the normalized costs in the Cyclelogistics project (e.g., for bike purchasing, maintenance, insurance and messenger pay) from the overall income [108]. Economic results and their descriptive statistics calculated in Microsoft Excel® are reported below (Tables 3–5).


**Table 3.** Area-/country-specific policy and factor categories, average profit and profitability levels.

## **Table 3.** *Cont.*


**Table 3.** *Cont.*


◦ In this paper, "Central Europe" includes only continental European countries and excludes the Italian peninsula, Greece, the Balkans and the Scandinavian peninsula.

**Table 4.** Profit and profitability statistics by area-/country-specific policy and factor category.


◦ In this paper, "Central Europe" includes only continental European countries and excludes the Italian peninsula, Greece, the Balkans and the Scandinavian peninsula.


**Table 5.** Project-specific factors, average profit and profitability levels.

In the fifth and last stage, possible associations of factors and policies with economic results have been analyzed with IBM® SPSS® Statistics 24. Some analyses have been made in order to make inferences about data and to understand whether the observed pattern is real or due to chance. Before using IBM® SPSS® Statistics 24, the dataset was cleaned up by deleting overlapping data concerning cross-country-specific factors and policies.

Therefore, in the fourth section, mean and standard deviation were calculated again in IBM® SPSS® Statistics 24, together with other advanced statistics.

As an additional note, despite the many variables that could be identified in the literature review, the availability of data was limited to some of them. Moreover, data for some variables were only partly available. Finally, the only variables with full data available for a quantitative analysis were related to profit, profitability, geographical area and type of bike.

Finally, we state our four research hypotheses, based on the above explication of the variables:

**Hypothesis 1 (H1).** *The profit distribution varies across categories of geographic area, that is, across the different geographic areas, not across each country pertaining to a specific geographic area.*

**Hypothesis 2 (H2).** *The profitability distribution varies across categories of geographic area, that is, across the different geographic areas, not across each country pertaining to a specific geographic area.*

**Hypothesis 3 (H3).** *The profit distribution varies across categories of bike model.*

**Hypothesis 4 (H4).** *The profitability distribution varies across categories of bike model.*

#### **4. Results and Discussion**

The preliminary step of the statistical analysis was conducted in order to verify whether data distributions of profit and profitability are normal or not. Checking the normality of distributions is relevant since this methodological step impacts the choice of the statistical tests to adopt (e.g., parametric vs non-parametric tests) in order to ensure the reliability of results.

In the following, Table 6a,b show normality tests on area-specific profit and profitability. Table 7a,b show normality tests on project-specific profit and profitability.

**Table 6.** (**a**) Preliminary analysis of normal distribution hypothesis of area-specific profit and profitability: skewness and kurtosis. (**b**) Preliminary analysis of normal distribution hypothesis of area-specific profit and profitability: Kolmogorov– Smirnov and Shapiro–Wilk tests.



**Table6.***Cont.*

a Lilliefors Significance Correction.

**Table 7.** (**a**) Preliminary analysis of normal distribution hypothesis of project-specific profit and profitability: skewness and kurtosis. (**b**) Preliminary analysis of normal distribution hypothesis of project-specific profit and profitability: Kolmogorov– Smirnov and Shapiro–Wilk tests.



**Table 7.** *Cont.*

a Lilliefors Significance Correction.

The results show that all data distributions are not normal since the prevailing tests of normality—i.e., Kolmogorov–Smirnov and Shapiro–Wilk—lead to reject the normal distribution hypothesis.

Therefore, parametric tests—i.e., one-way ANOVA—were not conducted, whilst nonparametric ones were conducted: the Kruskal–Wallis H test was applied [110] since it is more appropriate than the Mann–Whitney U one. In fact, in our analysis, all independent categorical variables—i.e., both area- and project-specific factors—have more than two levels. In Table 8a,b, the results of the Kruskal–Wallis H tests are reported.

The results of the first statistical tests conducted–from the Kolmogorov–Smirnov to the Shapiro–Wilk one—are aimed at proving the reliability and suitability of the second step of the statistical analysis—i.e., Kruskal–Wallis H test [110]—which challenges the research hypotheses with corresponding null hypotheses. The final results prove that the only null hypothesis rejected is the one related to the bike model, thus confirming that the hypothesized dependence is true and significant. On the other hand, the across-the-region hypothesis is not supported, thus showing that there are no specific regional features affecting profit and profitability results more than others.

#### *4.1. Discussion on the Statistical Tests of H1 and H2*

In the following, statistical results concerning our research hypotheses on categories of geographic area are discussed. In particular, such hypotheses—which are negatives of the null ones reported in Table 8a—state that:

**Hypothesis 1 (H1).** *The profit distribution varies across categories of geographic area, that is, across the different geographic areas, not across each country pertaining to a specific geographic area.*

**Hypothesis 2 (H2).** *The profitability distribution varies across categories of geographic area, that is, across the different geographic areas, not across each country pertaining to a specific geographic area.*

**Table 8.** (**a**) Kruskal–Wallis H test of area-specific factors and policies. (**b**) Kruskal–Wallis H test of project-specific factors.


Asymptotic significances are displayed. The significance level is 0.05.

As a first remark, such hypotheses are rejected. In fact, both profit and profitability distributions are the same across categories of geographic area (Table 8a). In this case, multiple comparisons were not performed because the overall test does not show significant differences across such categories.

Considering profit and profitability performances shown in Tables 3 and 4, the statistical analysis of area-specific variables (Table 8a) shows an overall significance of success factors and policies in the European context. It also proves that there are no single factors or policies having a relevantly higher impact than others on the likelihood of success of cycle logistics projects. Although differences between distributions depending on the area are not significant, we can still analyze data in Table 4 in order to obtain a deeper understanding of the determinants of such a phenomenon. Data concerning profit and profitability by area highlight somewhat high values in terms of mean and standard deviation, except for Scandinavia. Mean values associated with projects in "Central Europe" prove to be higher than the overall average, whilst those in the "UK and Ireland" and "Greece, Italy and the Mediterranean islands" are just below it. "Scandinavia" has the lowest mean values. One of the main reasons behind that may be found by considering that cycling is an activity deeply rooted in Scandinavian cultures, especially in Denmark, which is the application context of the two projects considered for this area. Therefore, Danish people are used to transporting both small- and big-sized goods by themselves, thus not calling for bike delivery services. For instance, in 2008, IKEA invested in bikes—and trailers, if needed—at selected stores in Denmark (and also in Sweden) so that customers can ride home for free with their new purchases [111,112]. Although Danish projects show the lowest profit level, their profitability level goes high, much more than in any other area in Europe. A possible interpretation is that the high degree of accessibility and availability of suitable infrastructures in those countries helps with lowering the cost of some items—e.g., vehicle maintenance, insurance and, hence, purchase of new bikes. On the other hand, it helps with increasing service levels—e.g., deliveries are more likely to be made on time and customers are more satisfied. Moreover, supporting economic measures prove to nurture

bike delivery organizations as well. Finally, such delivery services tend to be more effective, efficient and, hence, with a higher profitability in those countries than in others.

#### *4.2. Discussion on the Statistical Tests of H3 and H4*

In the following, statistical results concerning our research hypotheses on categories of main type of bike are discussed. In particular, such hypotheses—which are negatives of the null ones reported in Table 8b—state that:

#### **Hypothesis 3 (H3).** *The profit distribution varies across categories of bike model.*

#### **Hypothesis 4 (H4).** *The profitability distribution varies across categories of bike model.*

As a first remark, both profit and profitability distributions vary across categories of main type of bike (Table 8b). In this case, multiple comparisons were performed because the overall test does show significant differences across such categories.

In detail, the statistical analysis proves how profit and profitability distributions change according to the main type of bike utilized in cycle logistics projects and, thus, recognizes that some bike models have a relevantly higher impact than others on the likelihood of success of bike delivery businesses. Multiple comparisons concerning profit and profitability distributions across categories of the project-specific variable have been performed in order to understand which types of bike have the most significant impact on profit and profitability. The multiple comparisons have been analyzed by means of adjustment using the Bonferroni correction for multiple tests and they are reported in Figure 2 (concerning profit) and Figure 3 (concerning profitability).

In Figure 2, the pairwise comparison for profit shows the highest value for the node representing projects using contemporarily cargo bikes and tricycles. Note that this result is coherent with the data reported in Table 5, where the mean values associated with the two bike models used together are the highest ones for profit and also for profitability. Hence, that appears to be the best modal configuration. Furthermore, tricycles and cargo bikes taken separately prove to have a relatively significant impact on the profit variable and to generate relatively high values for both profit and profitability. By contrast, traditional or trailer bikes seem to not to provide such a relevant impact on profit, even though traditional bikes have a profitability mean value higher than the overall average in Table 5. On the contrary, if we analyze the paired use of different modal configurations, trailer bikes combined with big-sized bike models—i.e., either cargo bikes or tricycles or both of them—appear to be the best matched choice. In particular, trailer bikes combined with cargo bikes and/or tricycles seem to broadly cover the whole demand span. In fact, such configurations of paired bike models can satisfy altogether the delivery needs of different goods— i.e., small- to big-sized—and cope flexibly with different topological features and network shapes at the same time. As further proof, the trailer–tricycle–cargo bike configuration including all those types of bikes is the one having the highest impact on profit distribution among the three significant configurations in Figure 2. In fact, it covers a more extended demand span than either the trailer–tricycle or trailer–cargo bike configurations.

In Figure 3, the pairwise comparison for profitability shows the two highest values for those nodes representing projects exploiting cargo bikes and tricycles together as well as those adopting traditional bikes. Again, note that this result is coherent with the data reported in Table 5, where the mean values associated with the two bike models used together are the highest ones for profitability and also for profit. Analogously, traditional bikes have one of the highest values in terms of average profitability in Table 5. Hence, those appear to be the best modal configurations. The node representing tricycles, taken separately, has a relatively high value as well, but the possible matches with other modal solutions do not show statistical significance. Similarly, in Table 5, tricycles alone generate the highest mean value in terms of profit, as well as a profitability level above the overall average. On the contrary, the paired use of different modal configurations shows how trailer bikes are the

most flexible bike model since they can be fruitfully combined with either traditional bikes or cargo bikes and tricycles. The significant impact on profitability shown by trailer bikes together with cargo bikes and tricycles could be explained by the broader coverage of the whole demand span, since those matched bike models satisfy the delivery needs of heterogeneously sized goods, similarly to data concerning the profit variable (Figure 2) already discussed. On the other hand, trailer and traditional bikes together also significantly affect profitability. In this case, it is worth mentioning that those bike models imply quite a low cost (Table 2a) in terms of purchase, maintenance and insurance. Moreover, they require quite a short time for goods delivery due to the high ease of driving and parking, low load capacity and high average speed in cities. Those features characterizing both trailer and traditional bikes prove how they allow efficient use of resources, thus justifying the significant performance in terms of profitability in Figure 3.


**Figure 2.** Multiple comparisons of differences concerning profit reported in Table 8b. Each row tests the null hypothesis that the Sample 1 and Sample 2 distributions are the same. Asymptotic significances (2-sided tests) are displayed. The significance level is 0.05. Significance values have been adjusted by the Bonferroni correction for multiple tests.

**^ĂŵƉůĞϭͲ^ĂŵƉůĞϮ dĞƐƚ ^ƚĂƚŝƐƚŝĐ ^ƚĚ͘ƌƌŽƌ ^ƚĚ͘dĞƐƚ ^ƚĂƚŝƐƚŝĐ ^ŝŐ͘ Ěũ͘^ŝŐ͘ dƌĂŝůĞƌďŝŬĞͲĂƌŐŽďŝŬĞ** ϰ͘ϱϬϬ ϰ͘ϬϮϴ ϭ͘ϭϭϳ Ϭ͘Ϯϲϰ ϭ͘ϬϬϬ **dƌĂŝůĞƌďŝŬĞͲdƌŝĐLJĐůĞ** Ͳϴ͘ϱϬϬ ϱ͘ϮϬϬ Ͳϭ͘ϲϯϱ Ϭ͘ϭϬϮ ϭ͘ϬϬϬ **dƌĂŝůĞƌďŝŬĞͲdƌĂĚŝƚŝŽŶĂů** ϭϯ͘ϬϬϬ ϯ͘ϵϯϭ ϯ͘ϯϬϳ Ϭ͘ϬϬϭ Ϭ͘ϬϬϵ **dƌĂŝůĞƌďŝŬĞͲĂƌŐŽďŝŬĞ͖dƌŝĐLJĐůĞ** ϭϳ͘ϱϬϬ ϱ͘ϮϬϬ ϯ͘ϯϲϱ Ϭ͘ϬϬϭ Ϭ͘ϬϬϴ **ĂƌŐŽďŝŬĞͲdƌŝĐLJĐůĞ** Ͳϰ͘ϬϬϬ ϰ͘ϲϱϭ ͲϬ͘ϴϲϬ Ϭ͘ϯϵϬ ϭ͘ϬϬϬ **ĂƌŐŽďŝŬĞͲdƌĂĚŝƚŝŽŶĂů** Ͳϴ͘ϱϬϬ ϯ͘ϭϲϵ ͲϮ͘ϲϴϮ Ϭ͘ϬϬϳ Ϭ͘Ϭϳϯ **ĂƌŐŽďŝŬĞͲĂƌŐŽďŝŬĞ͖dƌŝĐLJĐůĞ** Ͳϭϯ͘ϬϬϬ ϰ͘ϲϱϭ ͲϮ͘ϳϵϱ Ϭ͘ϬϬϱ Ϭ͘ϬϱϮ **dƌŝĐLJĐůĞͲdƌĂĚŝƚŝŽŶĂů** Ͳϰ͘ϱϬϬ ϰ͘ϱϲϳ Ϭ͘Ϭϵϴϱ Ϭ͘ϯϮϰ ϭ͘ϬϬϬ **dƌŝĐLJĐůĞͲĂƌŐŝďŝŬĞ͖dƌŝĐLJĐůĞ** ϵ͘ϬϬϬ ϱ͘ϲϵϲ ϭ͘ϱϴϬ Ϭ͘ϭϭϰ ϭ͘ϬϬϬ **dƌĂĚŝƚŝŽŶĂůͲĂƌŐŽďŝŬĞ͖dƌŝĐŝLJůĞ** ϰ͘ϱϬϬ ϰ͘ϱϲϳ Ϭ͘ϵϴϱ Ϭ͘ϯϮϰ ϭ͘ϬϬϬ

**Figure 3.** Multiple comparisons of differences concerning profitability reported in Table 8b. Each row tests the null hypothesis that the Sample 1 and Sample 2 distributions are the same. Asymptotic significances (2-sided tests) are displayed. The significance level is 0.05. Significance values have been adjusted by the Bonferroni correction for multiple tests.
