Varieties of cultural crowdfunding: The relationship between cultural production types and platform choice

Purpose Due to the unique nature of the Cultural and Creative Industries (CCIs), the impact of crowdfunding on them is more significant than on other industries. This study investigates the association between crowdfunding campaigns in four different categories of cultural production and each campaign promoter’s decision regarding platform choice. Design/methodology/approach We classified cultural productions according to the Cultural Enterprise Framework. We collected data from 1,465 successful, reward-based, culture crowdfunding campaigns from five Nordic countries (Denmark, Finland, Iceland, Norway, and Sweden). We used binary logistic regression for estimation purposes.


Introduction
i.e. record companies, film distributors, videogame publishers (Ebbers and Wijnberg, 2012). Therefore, the notion of distribution channel better reflects the multiple potential benefits derived from the process beyond fundraising, such as increased social standing and cultural capital. It achieves this by incorporating and emphasising the value derived from the potential benefits and externalities of successful crowdfunding beyond funding a project (Mollick and Kuppuswamy, 2014). With supplementary we suggest that, for many artists, crowdfunding is not, per se, a replacement for other sources of financing. Nor is it a rolling, ongoing financing model, partly because there are social constraints built into asking the same crowd for support multiple times (Davidson and Poor, 2015, pp. 301-302).
The existing literature on reward-based crowdfunding explores facets and aspects of CCF, predominantly the model's potential for artists and creators (Mollick, 2014;Bannerman, 2013) and the patrons' motivations for supporting campaigns (Boeuf et al., 2014;Jian and Shin, 2015). There is less attention given to the role of the crowdfunding platform as a 'relational mediator' (Ordanini et al., 2011) between cultural producers and crowds. To the best of the authors' knowledge, there are no existing studies on the relationship between a cultural production's type and the promoter's choice of platform to host the campaign. For many promoters in non-English-speaking countries, the choice of which platform to use is a dichotomous option between a local or an international one. A local platform predominantly caters to a country-specific market, with its interface, presentation and contributions set to local language and currency. International platforms, with Kickstarter and Indiegogo as principal exponents, have English as their default language and the USA as the country of provenance. A differential factor is that the default language of an international platform expands the global reach of a campaign by increasing the number of potential connections between the promoters and the patrons beyond the scope a local platform may provide.
Platform usage by CCF promoters within the EU (excluding Norway and Iceland) points to a near equal overall division of usage between local (53%) and international platforms (47%) (De Voldere and Zeqo, 2017). Variations first become visible when comparing platform usage at the sectoral level -as examples, performing arts campaigns used 223 different platforms (out of a potential 380), while videogame projects, in comparison, used 61 platforms (ibid, p. 77). A higher number of different platforms in use in the former sector indicate a local bias, while in the latter context, less variation and diversity indicate a preference for an international platform. Although De Voldere and Zeqo (2017) provide information on local and international platform usage overall and by sectors, the determinants of these variations are not fully established (p.78 -86).
Therefore, we investigate the factors that may determine the choice of the type of platform are, using a sample of platforms and successful CCF campaigns in the Nordic countries (Denmark, Finland, Iceland, Norway and Sweden). Our finding is that productions that are, culturally speaking, 'rooted', or anchored, because of language barriers, sectoral category and other cultural traits, are more likely to use a local platform -their cultural affinity predicts the likelihood of choosing a local platform. In the case of complex cultural productions that are culturally adaptable because their language and cultural traits are no obstacle to wider diffusion, then the promoters are more likely to use an international platform.
In the next section, we present a review of the literature and formulate hypotheses. Section three presents the research methodology, including the data collection procedure and analysis. Section four presents the findings, and section five a discussion of the results. Finally, section six concludes with a summary, limitations, and future research directions.

Historical funding channels for the CCIs
The cultural industries are those sectors of the economy engaged in the production and distribution of cultural commodities (goods and services), usually differentiated from other economic activity by their output's higher cultural (symbolic, aesthetic and experiential) rather than utilitarian value, irrespective of commercial value (Throsby, 2001, p. 112;Power, 2002, p. 105;European Commission, 2010, p. 5).
The cultural industries are also a subset of the creative industries (Jones et al., 2015, p. 5;UNCTAD, 2010, p. 5). The latter is a broader sectoral class of activities reliant on individual creativity and skills for producing goods and services, and whose economic value resides in some form of intellectual property (Throsby, 2008). Within the European context, these two closely allied groups of industries make up the Cultural and Creative Industries (Pratt, 2012).
CCF is, in relation to funding ventures and projects in the CCIs, a digital development of a historical model of arts funding, the subscription-based patronage model of funding (Williams, 1981;Swords, 2017). The logic of subscription-based patronage is to collect individual payments, from people subscribing, to underwrite the production costs of artistic works, such as publishing a book or producing a concert. The contributor (patron) receives a copy of the book or a concert ticket in exchange for their contribution, while the artist or producer reduces financial risk. Notable examples of subscription-based patronage are the publishing of the books of Mark Twain (Hill, 1963), the musical works of Mozart (Link, 2003), and the drawings of naturalist and painter Jean Jacques Audubon (Cummings, 2015).
The main difference between CCF and subscription-based patronage is the limits in terms of reach and uptake of the latter. The basic subscription-based patronage model depended on agents, such as salespeople, to reach the potential subscribers on a one-to-one basis (Hill, 1963, p. 26). Later, the adoption of a variety of successive media and communication platforms to disseminate calls on a oneto-many basis exponentially increased the potential reach. Some prominent examples of campaigns are Joseph Pulitzer using his newspapers to fund the construction costs of the pedestal for the Statue of Liberty (Bannerman, 2013) and film director John Cassavettes employing a radio programme to finance his directorial film debut Shadows (Matthews et al., 2014). With the emergence of the internet, and before the advent of specialised platforms, the band Marillion used their website to finance a tour of the USA in 1997 (Leyshon et al., 2016). Thus, patronage by subscription serves as a model for CCF, and CCF provides a historical funding model with traction by merging a classic model of preselling with the potential boundless diffusion provided by the internet.

The unique aspects of CCIs in the crowdfunding context
Artists or creatives as entrepreneurs in the CCIs differ from entrepreneurs in other sectors relative to their ambitions, financing challenges, and preferences. Many artists acting as 'cultural entrepreneurs' do not perceive moneymaking as a primary goal of their activities. They often appreciate other values, such as professional recognition as an artist or creator above economic incentives (Swedberg, 2006). Therefore, they tacitly accept working under resource-constrained conditions, relying on heuristic schemes, such as 'bricolage', or 'making do' with whatever resources are available for the production of works (Stinchfield et al., 2013). Accessing finance is generally considered challenging for entrepreneurs because of the difficulties of evaluating a project's or venture's economic potential or performance, often explained as the effect of uncertainty, caused by asymmetric information between the provider and the receiver of funding (Andreoli, 2018, p. 151). However, in the CCIs, uncertainty and subsequent rejection is primarily a result of symmetrical ignorance (Caves, 2001, p. 3). The effect of this type of mutual ignorance, e.g. the producer knows the qualities of the output and the process, yet the audiences and consumers may show their appreciation individually or collectively in unpredictable ways, goes beyond missing, lack of, or unequal distribution of information. The fact is that neither party knows the other's intentions, or more succinctly, nobody knows (Caves, 2001). Some of the reasons explaining this situation are differentiated output qualities, such as novel, innovative product characteristics, and demand fluctuation because of social contagion, like the perceived notions of trends and fashion that change suddenly (Caves, 2001, p. 6;Kretschmer et al., 1999, pp. 64-66;Throsby, 1994, p. 4). The result is a market with an oversupply of available potential 'candidates' for commercialisation.
From a financing perspective, this leads to a selection problem for incumbent firms (Kretschmer et al., 1999), while the artists' and creators' willingness to work regardless of financial gain under uncertain conditions in densely populated markets bias their financing preferences towards 'free money' over financial risk-taking. Adversity towards risk manifests itself in a predilection for public funding or private investments in the form of non-returnable advances (Fleming, 2007;Wilson and Stokes, 2005) over models with personal obligations, such as debt financing (Borin et al., 2018). The combined effects of these aspects are reasons why many consider the CCIs as sectors with small-scale and project-based producer organisations finding it challenging to demonstrate the growth potential necessary for attracting the interest and money of traditional investors (Fleming, 2007).
CCF can remedy some of these issues because it alleviates uncertainties and addresses aesthetic and professional contradictions by a reversal of the market selection process (Andreoli, 2018, p. 49). In the CCI, market selection typically implies that an artist or creative must search, identify and make several customised pitches and presentations to potential companies or public institutions who may (or not) accept to provide funding. With CCF, the publication of the project takes place on a public platform, and the patrons decide whether they want to provide support. Individually, the patrons may not provide as much financing as a traditional intermediary, but the sum of all the small contributions, often presales, enables the promoter to demonstrate market validation and potential. A successful CCF campaign becomes one of several potential ways to gain visibility by demonstrating a demand for a product or the existence of a potential audience. Success through CCF effectively mitigates issues of uncertain demand and symmetrical ignorance. Interpreted this way, CCF becomes a conduit for breaking into the 'mainstream' system of cultural production (Galuszka and Brzozowska, 2017).
CCF creates opportunities, increases diversity, provides better access to funding, and leads to a higher level of consumer and audience engagement through participatory processes (Bannerman, 2013, Rykkja et al., 2019. However, in real terms, success is volatile, hinging on a range of factors related to the type of cultural production, leading to situations where most projects either succeed by a small margin or fail by a large one (Mollick, 2014). Culture as a contextual variable and the project's match with the communities of patrons are crucial factors for determining the success of specific types of campaigns (Josefy et al., 2017, p. 176). For example, even if a project is thriving locally, there are no guarantees it will flourish elsewhere. Platforms may help promoters overcome geographical distance (Agrawal et al., 2015), but that is not always the case, as others find that contributors are people living close to the promoter or are part of his/her inner social circle of friends (Mendes- Da-Silva et al., 2016).
In relation to motivations behind contributing, specific projects have aspects of gifting and exchange of commodities built into them (Galuszka and Brzozowska, 2016), making participation and the ability to influence processes as necessary as the gifting of the product itself. There are also cases where being a member of a community has higher potential value for the contributor than the material rewards offered (Boeuf et al., 2014). However, social ties might produce negative connotations, as patrons may feel that contributing becomes a pro-social obligation (Bernard and Gazel, 2017). Conversely, other campaigns feature projects with a clear commercial orientation, where pre-purchasing a specific product is the primary motivation and most frequently used type of reward (Steigenberger, 2017;Thürridl and Kamleitner, 2016). In these cases, CCF favours projects and promoters with a pre-existing network of supporters (Davidson and Poor, 2015;D'Amato, 2016).

Hypotheses development
Defining typologies of cultural productions, particularly if the purpose is differentiating territorial embeddedness, e.g. the local, national or international scale, is a complicated matter to which statistical agencies supporting or mapping the economic impact of the cultural sector have dedicated considerable efforts and time (Eurostat, 2018;KEA, 2006;Throsby, 2008). Such efforts have often used criteria helpful in determining the origin of finished and standardised products (like when determining the nationality of a film production or a literary work). Hence, using such criteria is more challenging in categorising heterogeneous and often unfinished production activities, such as different types of cultural productions looking for financing through CCF. Our approach in this study is to adopt some of the abovementioned criteria to the field of crowdfunding by proposing a specific framework for classification. The framework itself is one of the contributions of this study.
Two crucial factors concerning the choice of a platform are scope and scale. The reach, or scope, of a given production, may limit its potential geographic reach. As an example, if a cultural production relies on language as an essential nonmaterial component and the language in question is not a world language, then the ability to appreciate the output is limited linguistically and geographically to those able to understand the language. The effect is a limit to the potential scope of the cultural production. It becomes, culturally speaking, 'anchored' to a geographical zone where the language concerned is spoken and understood. The same applies to productions with contextual elements -if a production requires personal attendance, or has an orientation towards a specific, local community, then content and delivery may also limit interest geographically, and consequently the production's reach relative to the interest of potential patrons.
The second factor is that financial needs and production requirements dictate the scale of production, implying that cultural productions vary in complexity, ranging from developing complex, unique, and resource-demanding prototypes to the reproduction, distribution and marketing of existing commodities.
The level of complexity may also indicate degrees of cultural adaptability, e.g. complexity in relation to financial requirements may translate into less reliance on linguistic components in the delivery. The more expensive or unique the cultural production is, the more adaptable it becomes to attracting the interest of a non-national patron.

[FIGURE 1 ABOUT HERE]
Page 8 of 27 In the lower-left quadrant of Figure 1, we have the category of cultural affinity. Cultural productions with cultural affinity are productions relying on a common linguistic or cultural understanding of the project for appreciation, and they have an orientation towards reproduction. The essential premise is that these cultural projects have a built-in language or other cultural component acting as a barrier for export and international reach. The literature on international trade in cultural goods provides some evidence for this proposition. Variables related to a common language and cultural proximity predict the potential for the bilateral flow of cultural goods (Disdier et al., 2010), with a common language increasing bilateral trade by as much as 56% (DEPS, 2007, p. 19). Language also acts as a stronger barrier for uptake than cultural distance in import markets (Fu, 2013, p. 810). One reason may be that many 8 Our essential argument is that cultural productions differ in respect to their scope and scale and these differences dictate their potential geographic reach (scope) and financial requirements for production (scale). If we want to assess whether scope, scale, or a combination of both are determinants guiding platform choice, the different types of cultural productions need categorisation, so that their influence and effect on determining platform may be measured and accounted for.
The literature on arts marketing provides a model, the Cultural Enterprise Framework (Colbert, 2007, pp. 8-10). The model has four quadrants and two continuums. The horizontal continuum is the enterprise's market orientation with 'market' (for-profit) and 'product' (not-for-profit, artistic values) at the extremes. The vertical continuum has activity, in the form of the production of a prototype or reproduction (of an existing work) at the extremes. Along with factors such as the size of the operation and legal status, the model enables differentiation between prototype production and distribution of reproduced content, by commercial, for-profit enterprises in the CCIs and organisations and institutions in the not-for-profit art sector (ibid., p. 9).
In this study, we adapt the Cultural Enterprise Framework (see Figure 1) by defining the horizontal continuum as the cultural scope of the product (local anchoring to international adaptability) to classify cultural productions according to how restricted the specific project, or sector, would be relative to market reach. The vertical continuum indicates the type of production according to its level of complexity. The extremes indicate at one end the production of an expensive, complex prototype and, at the other, the reproduction and distribution of already existing works. There are four quadrants in the model, each corresponding to a distinct category of cultural productions, with the lower two predominantly featuring productions oriented towards reproduction, and the upper two towards production.
consumers have to invest more time and mental effort to familiarise themselves with a cultural expression with a different origin than the local one, a phenomenon referred to as cultural learning (Park, 2015). As an example, nearly three-quarters of book sales in the Nordic countries are works published in a local language i , even if the inhabitants of the Nordic countries are among the most proficient users of English as a foreign language in the world (EF, 2019). We also assume that a lower average CCF campaign goal, as is the case with music and sound recordings (Davidson and Poor, 2016) may indicate that music artists use CCF as a channel primarily to enter and position themselves in the local market.
Statistics on Scandinavian music exports reveal that besides Sweden (export share of turnover of 19%), the share for the other two countries (Denmark and Norway) is 5-6% (Portnoff, 2017, p. 23). Thus, we hypothesise that: H1: Cultural productions with a high degree of cultural affinity are more likely to use a local platform.
Cultural productions with contextual content are productions with an output that has geographic limitations to its broader dissemination in addition to linguistic or cultural barriers, e.g. projects that do not travel because of an experiential component that requires personal attendance such as visits to festivals, performances and venues. Most of these productions also have an individual uniqueness to them, pointing towards prototype-like orientation and higher costs of delivery. We know from the literature that geography, local infrastructure and cultural offering all condition crowdfunding (Le Béchet et al., 2018), and that contextual elements predict community interest (Josefy et al., 2017). Many cultural productions and producers are said to have scalable, international growth opportunities (Fleming, 2016, p. 116), but in reality, cultural capital, or name and brand recognition, may affect wider online interest and hence what is being funded (Sorensen, 2012). In itself, this would not limit interest were it not that the only difference between these types of productions and those with cultural affinity is their production orientation and hence delivery costs. Linguistic or culturally non-adaptable content may still feature prominently, providing geographic restrictions that limit the scope. Thus, we hypothesise that: H2: Cultural productions with a contextual element are more likely to use a local platform.
We define complex cultural productions as those where the aim and outcome are of producing complex, resource-and design-intensive commodities. Provided that the content of the project overcomes limitations imposed by scope (cultural reach), the assumption is that these campaigns are more likely to use an international platform. All sectors within the CCIs have much higher costs of production (fixed and sunk) for prototypes with marginal costs of delivery (Towse, 2014, p. 8;Caves, 2001), and thus benefit greatly from economies of scales. These fixed and sunk costs are significantly higher for cultural productions in certain sectors compared with others. In practical CCF terms, it means that campaigns from specific sectors must raise a higher amount of funding, and may achieve their goals by successfully reaching out to platform users beyond the members of the promoter's inner social circle of friends. For instance, while the overall amount raised by reward-based campaigns was €6,211 in 2016, videogame campaigns managed to raise €43,897 and design campaigns €11,000 (De Voldere and Zeqo, 2017, p. 88 and 90). Campaigns from these sectors used an international platform in respectively 59% (design) and 87% (videogames) of cases (ibid, p. 79, 86). Consequently, we assume that the probability of successfully financing a cultural production requiring substantial funding to cover fixed production costs is higher on an international platform. Therefore, we hypothesise that: H3: Cultural productions of a complex nature are more likely to use an international platform.
The composite motives category includes cultural productions of commodities with content reliant on composite, visual semiotic language coupled with an orientation towards presentation and distribution.
If the campaigns have a production-based orientation, the financial requirements for prototypes are medium to low. Promoters of these productions might additionally have composite motives to use CCF.
What binds these productions and promoters together is a process of 'market creation', also referred to as pioneering entrepreneurship (Khaire, 2017, p. 16). Thus, the primary objective for the promoter(s) is to create value or demand for new artistic expressions and products that in themselves may have a composite nature, in particular arts and design goods. For other types of productions, such as tabletop games, international platforms provide an established market of patrons, as exemplified by the volume (over a third -37.4 percent) of the game campaigns on Kickstarter that are board games (Roedenbeck and Lieb, 2018). Considering their primarily visual aesthetics which make these productions culturally adaptable for international markets and the existence of an established user base of patrons with interest in some of these types of productions on international platforms, we hypothesise that: H4: Cultural productions with composite motives are more likely to use an international platform.

Methodology Data
We collected data from successfully funded crowdfunding campaigns for cultural productions, promoted by a national or resident of one of the five Nordic countries, on either a local or an international platform, with a start date after 1 January 2010 or an end-date before 31 December 2016.
There are several reasons for choosing the Nordic countries as a setting for the study. Primarily, several of these countries were early adopters of CCF. In 2013, Iceland registered the highest number of projects per million inhabitants on Kickstarter (176), with Denmark and Sweden positioned among the top twelve European countries (Barbi and Bigelli, 2017). Secondly, government funding for culture is stable and suffered no drastic cutbacks after the financial crisis in 2008 (Nordicstatistics, 2018). Thirdly, the Nordic countries present a different cultural and geographical setting compared with other CCF studies which have frequently used data from a single global platform (Kickstarter, e.g. Mollick, 2014;Josefy et al., 2017) and language (English, like Cox and Nguyen, 2018). Finally, the Nordic region is a large and growing crowdfunding market in the European context. In 2017 they were the third-largest market, with 449 million euros of funding raised across the different models and an average annual growth rate of 67 percent between 2012 and 2017 (Ziegler et al. 2019, pp. 74-75). Nonetheless, there are distinct differences between countries in relation to uptake, regulatory frameworks, public intervention, and prevalence of crowdfunding business models.
The names of local platforms come from an iterative online and offline search strategy guided by a snowball sampling approach (Creswell, 2013). By reading reports (such as Myndigheten för Kulturanalys, 2013), checking the websites of industry associations (Nordic Crowdfunding Alliance), and dialoguing with platform operators, we drew up an initial list of 52 platforms. Of these, 32 did not have any projects of interest for this study (i.e. none of the campaigns was actively engaged in CCF), five had relevant campaigns with all starting and ending after 31 December 2016, and five had nonsearchable campaign archives (they folded, or stopped operating). Ten local platforms: Boomerang We devised a project evaluation matrix with eight pre-defined cultural fields (Literature, Music, Performing Arts, Visual Arts, Film, TV and Radio, Heritage, Videogames, and Design) supplemented by the operational definition of industrial sectors from the concentric circles model of the CCIs (Throsby, 2001;Throsby, 2008, p. 150). This tool serves to identify, categorise and classify cultural productions according to their sectoral affiliation. The identification of campaigns relied on manual searches using the platform's proprietary search engines, with the filtering of results by project categories and geographic location where possible. Returned results went through a qualitative assessment for inclusion using our evaluation tool. If the campaign passed the test, for each campaign, we manually coded the name of the platform, crowdfunding model, sectoral classes, financial information, year, and demographic information about the promoter (country, city, contact details). We adopted two distinct approaches to the manual coding of data. Primarily, in some cases, we classified and categorised the campaigns differently from the classification systems used by individual platforms.
The rationale was to ensure unity of categories by classifying all campaigns according to one typology when collecting data across multiple platforms. Secondly, our definition of success was the platform's criterion. Thus, for a reward-based platform operating on an 'all-or-nothing' basis, only the campaigns raising finance above the defined goal were recorded. Analogously, if the platform operated on a 'keepit-all' or 'minimum amount' basis, successful campaigns were those raising an amount above the stated minimum, or indicating that the promoters had opted for a model allowing them to 'keep all' financing, regardless of the amount raised.
We initially identified 1,665 campaigns. Upon review, we discarded 179 from the analysis (were not successful on closer review, lacking cultural dimension, equity crowdfunding, and cases where the promoter is not from, or based in, a Nordic country). The final dataset comprises 1,465 successful campaigns meeting all criteria for further analysis, spread over 48 sectors.

Operationalisation of variables
Before testing the formulated hypotheses, we needed to operationalise the dependent, independent and relevant control variables. Our dependent variable is platform choice, and the independent variable is cultural production types, which is divided into four categories (see Figure 1). As we can associate a campaign's financial goal, the promoter's country of residence or origin, and the platform's financing model with platform choice, we controlled for these three variables.
To measure platform choice by means of distinguishing campaign promoter choice between international or local platforms, we dummy-coded a binary variable with 1 for international (US-based) platforms and 0 for local, where local is a national platform from any of the Nordic countries. The four different categories of cultural production types, i.e. cultural affinity, contextual content, complex production and composite motives were also dummy-coded as binary variables. For example, among Iterative searches collecting, evaluating and classifying campaigns took place over two time-periods (May-September 2016 and September-October 2017) and were conducted by a research assistant with previous academic and working knowledge of cultural and creative projects. The authors subsequently reviewed and checked the compiled data independently. Checks included control of categorisation accuracy, consistency with inclusion criteria, data completeness (missing values), and checks for significant omissions of projects by repeating some of the same searches on the same platforms. For a few platforms, the owners corroborated geographic information and data for the exact year of the end of the campaign.
Post-collection, and before running analyses, and using the adapted Cultural Enterprise Framework, we classified all campaigns into one of the four quadrants of the 4C model of cultural productions. The assignment of each campaign observation into one of the types was a qualitative exercise guided by the arguments and assumptions underlying our hypotheses.

Results
As the dependent variable is a binary variable, and independent variables include both binary and continuous variables, logistic regression estimation is the most appropriate method for hypothesis testing (Glonek and McCullagh, 1995). We estimate five models, of which the first one is only a control variable model. Models 2-5 include one type of cultural production as an independent variable, in addition to the control variables. The models are expressed in equation (1) to (5) as follows: Page 13 of 27 13 the 1,465 campaigns, we dummy-coded 1 for the 818 cultural affinity campaigns and 0 for the others.
Likewise, we operationalised the other three cultural production types. We used the financial target amount set by the campaign promoter to represent the financial goal. To reduce skewness in the data, we took the natural log of the financial goal. Note that we converted all values to euros using appropriate exchange rates (yearly average) before the log transformation, where applicable. For each of the five countries, we used a dummy-coded variable to control for a country effect. For example, 140 campaigns were coded 1, indicating Norway as the country of origin or residence of the promoter of the campaign, and 0 otherwise. Lastly, we created a dummy variable for the platform-financing model. CCF campaigns using platforms with 'All or Nothing models' are coded with 1, and 'flexible models' are coded with 0.

[TABLE 3 ABOUT HERE]
The interpretation of regression estimates in logistics regression is different from linear regression. In Table 3, we report the exponential betas, which represent the odds ratios. An odds ratio is the probability of one event over another when both events are mutually exclusive. The odds-ratio of platform choice in this context can be expressed as follows: Based on equation (6), equal probability of two mutually exclusive events yields an odds ratio value of 1. Hence, an odds ratio higher than 1 indicates that the probability of choosing an international platform is higher than that of choosing a local platform, and lower than 1 indicates the opposite (Hair et al., 2009). Thus, based on Table 3, the probability of using a local platform is 3 times (exp of 0.33~0.25/0.75) higher for cultural affinity productions and about 1.29 times (exp of 0.78~0.44/0.56) higher for contextual content productions (although the latter is not statistically significant). Similarly, the probability of using an international platform is 4.74 and 2.23 times higher for campaigns with complex production and composite motives, respectively.
To summarise, we find support for H1, that is, campaigns with a high degree of cultural affinity, are more likely to choose a local platform (exp =0.331, p<0.001). H2, which predicts that campaigns with a contextual element are more likely to choose a local platform, is not supported (exp =0.782, p=0.192). H3 is supported, indicating that a campaign with complex production is more likely to choose an international platform over a local one (exp =4.736, p<0.001). Furthermore, H4 is also supported, as campaigns that contain a composite motive are more likely to choose an international platform (exp =2.232, p<0.001). Overall, the three control variables have a significant effect on platform choice (p<0.001).
As a post-hoc test, we investigated whether the results of the hypothesis testing also hold at the national level. We estimated the same regression models at an individual country level. Table 4 presents a summary indicating some degree of country-to-country divergence. H1 and H3 are supported in four of the countries, but not Sweden. Identical to the original result, H2 is rejected for all countries. H4 is supported in the context of Sweden and Finland but not for Norway, Denmark and Iceland.
14 Table 3 presents the estimation results of the binary logistic regression. All the Variance Inflation Factor (VIF) values of the five estimated models are below 3, under the recommended cut-off value of 10 (Hair et al., 2009), which confirms the non-existence of any multicollinearity problem. The correct average percentage of predicted platform choice is 77.5%. As for explanatory power, the pseudo-R-square values indicate that the model explains between 30% and 45% of the promoters' platform choices.

Discussion
The support for H1 is in line with the argument that cultural productions with a linguistic element, such as the publishing of a book, or production of a theatrical play, require some level of cultural affinity for appreciation and consumption, and are subsequently more likely to choose a local platform. In addition, the support for H3 is in line with the argument that complex cultural productions that require a high level of funding to cover fixed costs of production are more likely to choose an international platform.
Moreover, in line with H4, projects where the promoter might have composite motives, meaning that there are no potential linguistic or cultural barriers for choosing a local over an international platform, are more likely to choose an international platform.
In contrast to previous studies, we find that the choice of platform is conditional on the content type and orientation of the project. A general recommendation to use Kickstarter due to the higher level of success that cultural productions experience on the platform (Cox and Nguyen, 2018) may not apply in all contexts. Our study provides evidence that content language is a linguistic barrier relative to potential reach and scope for CCF campaigns from non-English speaking countries with small domestic markets for cultural commodities, with the practical consequence that promoters are more likely to choose a local platform. In cases of cultural productions facing other cultural barriers because of fragmented and competitive market conditions, as is the case with recorded music, the campaigns are more vulnerable to geography and distance than was demonstrated by a previous study (Agrawal et al., 2015). Our results are more in line with findings related to the effects of distance in other non-English-speaking countries, like the case of Brazil (Mendes- Da-Silva et al., 2016). An explanation might be that those musicians resorting to CCF are predominantly in an early career phase, and use their campaigns to target family, friends and people in their core social network. Musicians have, on average, less demand for capital to produce an album than a complex production, such as a film or a videogame (Davidson and Poor, 2016;Barbi and Bigelli, 2017). Therefore, they are less incentivised to compete for space, attention and contributions on an international platform.
Surprisingly, we find that contextual elements, content that is site-specific and requires personal attendance, are not a determinant of platform choice. These cultural productions are neither more nor less likely to choose a local or international platform. It might be the case that culture and its corresponding local community is an important contextual variable for success (Josefy et al., 2017), but not as a factor determining the choice of platform. Nor do the findings provide any link between local infrastructure and campaigns, as was evidenced in a study set in France (Le Béchec et al., 2018).
At the opposite end of both continuums, we have the category of complex productions. Relative to reach, their symbolic value and aesthetic form are visual. Hence they are less impeded by linguistic barriers, and thus they compete for attention on an equal footing with productions from English-speaking countries. We provide evidence in favour of the hypothesis that promoters of these productions are more likely to choose an international platform. The financial requirements, and added reach, might explain the preference. Successfully funded campaigns on Kickstarter in the categories of fashion, film, and games, have average goals that are 53%, 101%, and 371% higher, respectively, than the average goal of music projects -the typical campaign in the cultural affinity category (Barbi and Bigelli, 2017).
Complex productions greatly benefit from the network effects of a large international platform and the extra user base; this is reflected in the promoters' choice of platform.
One of the values generated through CCF is online presence and attention. Online visibility is something that major firms in many CCI sectors (record companies, book publishers, or film producers), monitor to identify new artists, creators, and products to market and exploit (Waldfogel, 2018, pp. 184-185).
Projects with a composite motive use CCF to gain a marketing push. Thus, with the added reach of an international platform, all things being equal, cultural productions with composite motives are as likely to thrive in an international environment as much as in a local one.
In relation to the post-hoc analysis, in the case of Sweden, public intervention and facilitation may explain why we find no support for the hypotheses except H4. Sweden functions as the Nordic hub for many CCI's, such as fashion and the music industries. Besides, public authorities acknowledged the potential of CCF and financed testing of match funding models. The platforms Fundedbyme and Crowdculture both received funding from Sweden's innovation agency Vinnova in 2010 iii . Besides, there were more articles on crowdfunding published in Swedish media in 2010 than in the international press (Myndigheten for Kulturanalys, 2013, p.11). These factors provide some explanation to general uptake and a stronger local platform structure for CCF in Sweden than in other Nordic countries.
One theoretical contribution of our study is the 4C typology of cultural productions, which may be of value as a framework for differentiating cultural productions according to their potential for geographical spread and diffusion and their requirements concerning funding. The study adds to the literature on geography and crowdfunding by implying that not all distance can be automatically circumvented via crowdfunding (Agrawal et al. 2015) while also explaining some of the motives and reasons as to why certain types of cultural productions remain predominantly local (Mendes-da-Silva et al. 2015). Linguistic barriers, market access, attention and industry structure are additional factors that play a role in determining the choice of platform.
In terms of managerial implications, we evidence that if the cultural affinity (local anchoring) is strong, and the financial requirements are limited, due to a reproduction orientation of the production, a local platform may be sufficient in achieving campaign goals without incurring costs of international campaigning. Conversely, if the international orientation is stronger, and the financial requirements higher because of a production orientation, an international platform may prove more effective thanks to wider potential exposure. For the platforms, the findings are of interest as they evidence that there may be strategic benefits of specialising in promoting certain types of cultural productions, as delineated by the 4C model.bb

Conclusion
This study investigates whether the choice of the platform used for cultural crowdfunding campaigns differs by the degree of scope (local anchoring vs. international adaptability) and scale (production vs. reproduction). We used a sample of 1,465 successfully funded CCF campaigns for cultural productions promoted on reward-based crowdfunding platforms over six years, from 2010 to 2016. The promoters of these campaigns are artists and creatives from the five Nordic countries. We model platform choice as a binary option between using a local or an international platform. Cultural productions are classified into four categories -cultural affinity, contextual content, complex productions and composite motive, adapting a model for classifying cultural organisations taken from literature on arts marketing. Each category exhibits a different set of barriers or potential, guiding our assumption as to which type of platform the promoter is more likely to choose. We find that the choice and type of platform used depends on the orientation of a given project and that, in particular, linguistic barriers, the complexity of the project, and marketing potential, all condition the choice of platform, with different levels of support and prominence between countries.
Meanwhile, as is the case with most studies, this study has certain limitations. First, we included only successfully funded projects. Including both successful and unsuccessful campaigns might have given a different result. Secondly, the categorisation of cultural productions is a subjective assessment, and there could be other ways to conduct this exercise. Thirdly, our study samples projects from five small Nordic countries, but repeating the same comparative study based on cultural productions in other parts of Europe or in countries from a different continent, might give different results. Finally, there are several variables, such as brand name strength, familiarity, and equity, that may also influence decisions towards international platforms, since the name of platforms in larger markets, such as "Kickstarter" in the English-speaking context, or "Verkami" in the Spanish-speaking one, have often become synonymous with the activity of crowdfunding. This may explain the lack of significant effect concerning contextual content, as promoters potentially are torn between the local relevance of the production and the brand appeal of international platforms (i.e. festival or event participants in an experience with international attraction).
Furthermore, in addition to studies examining the generalizability of our findings in new national contexts and time periods, future research may also benefit from studies utilising qualitative research design. Interviewing artists would potentially reveal some additional variables and factors predicting choice of platform. There are also numerous variables related to the promoter not included as control variables in this study, such as gender (Gafni et.al, 2019), and personality type (Davidson and Poor, 2015), which may be used in future research. In any case, predicting choice depends on a preceding classification of cultural productions, and we believe that our 4C model can be useful for similar studies in the future.

Contextual Content
Cultural Affinity

Composite Motives
Complex Production International adaptability Production Reproduction Local anchoring