Starting from Laursen and Salter's (2006) contribution, a number of studies have tried to address the relationship between openness and performance. However, various authors have so far used the same label – performance – to measure different dimensions such as: firm's performance (Zubielqui, Jones, & Lester, 2016; Rass et al., 2013), Research & Development (R&D) performance (Suh & Kim, 2012), firm's success (Nitzsche et al., 2016), firm growth (Love et al., 2011), innovativeness/innovation performance (Chen, Chen, & Vanhaverbeke, 2011; Egbetokun, 2015; Spithoven, Frantzen, & Clarysse, 2010) and product/NPD performance (Gesing, Antons, Piening, Rese, & Salge, 2015; Praest Knudsen & Bøtker Mortensen, 2011).

We list the number of extant quantitative studies which address the relationship between openness and performance – with particular attention to the variables used in the analysis – in Table 1.

We argue that extant contributions present two main limitations: sample specific restrictions and the definition of the output “performance” variable.

As regards the first issue, all the contributions presented in Table 1 use country or sector-specific samples. While this choice favors the recognition of specific patterns of openness, it does not allow confrontation of results across various contexts, thus failing to address the existence of cross-national or cross-sectoral patterns over time. In addition, much literature focuses on various “innovation-related” measures as the performance output, including scale measures for innovation performance (Cruz-González, López-Sáez, Navas-López, & Delgado-Verde, 2015; Pullen, De Weerd-Nederhof, Groen, & Fisscher, 2012), continuous measures of innovation outputs, such as the number of new or significantly improved products (Hwang, Kim, & Kim, 2009; Mention, 2011), or percentage measures, such as the share of sales from innovative products/services (Chen, Vanhaverbeke, & Du, 2016; Gesing et al., 2015). While these measures provide objective indicators of the output of innovation efforts put through by companies, they do not directly address the effect on firm overall performance: using ratios of innovation over firm revenues does not guarantee that the firm will experience an increase in the latter, since the increasing incidence of innovation may come at expenditure of overall sales results.

To avoid distortion of results, we rely on objective and comparable performance measures of firm performance, namely firm sales, shares value and employment levels.

OI literature is yet to find consensus on the openness – performance relationship. In the next paragraphs, we review significant extant contributions in the field.

Chen et al. (2011) show that both scope and depth of openness have a positive effect on the company's innovation performance. Similarly, Pullen et al. (2012), study the effects of firm network characteristics on innovation performance and find that goal complementarity is positively related to innovation performance. Rass et al. (2013) posit a positive relationship between the implementation of OI instruments and performance, and a positive relationship between the former and firm social capital. Results show that firm social capital acts as a positive moderator, positively influencing firm performance in the implementation of OI practices. Gesing et al. (2015) address the effects of governance modes and collaboration types on firm NPD performance, measured as the share of revenues from products that are new to the firm and the share of revenues from products that are new to the market. They find that formal collaborations with both market-based and science-based partners show a stronger positive and significant effect on revenue shares from new-to-the-firm and new-to-the-market products compared to informal collaborations. Egbetokun (2015) finds that the breadth of firms’ network portfolio – the number of external knowledge sources – has a positive effect on both the firms’ overall innovation and product innovation performance. In addition, the variety of informal collaboration has a positive impact on firms’ innovativeness, while the variety of formal arrangements has a minor impact, as it only partially increases the overall innovation performance. Nitzsche et al. (2016) show that absorptive capacity, open culture, connectivity, strategic and structural flexibility all have a positive and significant effect on firm's innovation success.

Suh and Kim (2012) study the effects of four different type of collaboration activities – in-house R&D, technology acquisition or licensing, R&D collaboration and networking – on firm R&D performance, measured in terms of product/service innovation, process innovation and patents produced by firms, find that non-collaboration (in-house R&D) is an efficient strategy, while networking is the most inefficient strategy when considering all innovation outputs. Again, Laursen and Salter (2006) find that both search breadth and depth have an inverted U-shaped relationship with innovation performance. Similarly, Garcia Martinez, Zouaghi, and Sanchez Garcia et al. (2017) find an inverted U-shape correlation between the impact of alliance portfolio diversity on firm incremental and radical innovation performance. Love et al. (2011) find that firms’ internal organization, the presence of external public support and increasing competition, and the creation of ties with commercial laboratories have a positive effect on the firm innovative performance – innovative sales – while increasing openness is positively but non-significantly related with innovation.

In short, extant OI literature has so far focused on defining the level of openness of the company – measured through the extent of relationships with external counterpart, failing to consider the value of the acquired assets, and their effects of performance in comparison with the development of the same products on performance. In addition, while the use of “innovation” related indicators as the output of the innovation process provides a useful indicator on the success of innovation-related activities, it does not tell us how such output influences firm's overall performance, i.e. revenue growth. Finally, we see heterogeneous views in defining the relationship between openness and performance, since the multiple works show that openness has beneficial effects on performance, while other studies argue that the relationship is negative, or rather non-linear. Hence, we seek to contribute to extant literature by proposing an empirical analysis of the effects of openness – focusing on the nature of acquired intangible assets versus internally developed assets – on performance, measured in terms of overall sales, firm shares value, and the level of employment. To do so, we use an unbalanced panel data of 329 European companies listed in the five major markets – United Kingdom, Germany, France, Italy and Spain, over the period 2008–2013. Details about the research design are presented in the following section.

The empirical analysis relies on a panel regression analysis. The data collection process consisted of two main steps. First, a preliminary analysis was conducted on all companies listed on stock market exchanges (London, Frankfurt, Paris, Milan and Madrid) of the five largest European countries. All these countries have adopted IFRS (International Financial Reporting Standards) provisions, hence the data is assumed to be comparable.

The data collection process started with the collection of all 2178 companies listed on the abovementioned Stock Exchange markets for fiscal year 2008. From this first group, we excluded all the companies that did not provide financial information using IAS/IFRS principles. Then we continued excluding firms that did not meet our selection criteria: identifiability, control and the existence of future economic benefits. Moreover, all the companies involved in financial services were excluded from the sample.1 The same collection procedures have been accepted in previous studies (Denicolai, Ramirez, & Tidd, 2014; Denicolai, Cotta Ramusino, & Sotti, 2015; Denicolai, Ramirez, & Tidd, 2016). From the initial sample, we were able to select 328 companies following all the eligibility criteria. We extended the original sample by including the following years, up to 2013. The final sample consists of 328 companies over 2008–2013, thus 1968 firm-year observations. From this value, we subtracted a number of companies which have been removed from listing or acquired in the reference period. The final sample is an unbalanced panel data consisting of 1896 firm-year observations. Country and industry distribution of the sample are presented in Tables 2 and 3. We complemented the data with the extraction of the same dataset using Bureau van Dijk's ORBIS Software.

Finally, we included control variables, namely industry effects and country effects through, using dummy variables. In addition, we controlled for innovation propensity of firms, expressed as the logarithmic value of amount of R&D expenditure reported at the end of each year included in the sample, in Euros. A summary of the variables used is presented in Table 4.

The identification strategy is based on panel data regression where we are able to control for firm-level fixed effects and firm-specific linear time trend and on a GMM-system estimator (Blundell & Bond, 1998).

Our baseline empirical specification is the following:

In our baseline specification (1) we are not able to control for firm level time-invariant heterogeneity that could affect performance and OI practices at the same time. In order to include these factors, we rely on a second specification where we add firm-specific Fixed Effects (FEs) and year dummies:

We also estimate a more demanding specification including both firm specific intercepts and firm specific time trends, allowing them to follow different trends in the outcome variables, which is:

This may address the concern that firms who saw an improvement in performance associated to an increase in OI activities may have been already on an increasing trend before the investments in OI practices.

Finally, we include a quantile regression analysis in order to check for differences in the effect of OI at different percentiles of the outcome variables.

Table 5 shows the summary statistics of the variables used in the econometric analysis. In particular, firms in the sample spend, on average, 36 million Euros per year in development of intangible assets and 52 million Euros per year in acquisition of intangible assets over the 2008–2013 period. As regards dependent variables, firms in the sample have an average turnover of 3.9 Billion Euros on average, employ 16.800 full-time personnel per year and have an average stock price of 17.77 Euros over the selected period.

The following tables (Tables 6–8) show the results of the econometric models used in the study. All tables consist of two panels. Panel A reports the results using the expenditure in internally generated intangible assets and panel B those using expenditure in externally acquired intangible assets. Column (1) presents the estimates of the specification with triple interactions (Eq. (1)), column (2) shows the estimated values from the equation with firm-specific FEs, in column (3) firm-specific time-trends are added to the specification in column (2). Finally, in columns (4)–(6) we present estimates for the quantile regressions at the 50th, 80th and 20th percentiles.

Table 6 investigates the effect of expenditure on intangible assets on firms’ turnover.

Results in Column (1) show that both development and acquisition of intangible assets is significantly correlated with turnover (0.095, p-value<0.001 and 0.127, p-value<0.001, respectively). If we assume firms are homogeneous, it seems that intangible assets generate positive results on turnover, irrespective of their source. Specifically, the acquisition of external intangible assets generates a vis-à-vis greater effect on the dependent variable. The argument is aligned with the finding of positive effects of openness – measured as the extent of acquisition of external intangible assets (i.e. Chen et al., 2011; Rass et al., 2013).

Adding controls for firm fixed effects and year dummies in column (2), results still show a positive and significant correlation for both asset development and acquisition with turnover (0.069, p-value<0.001 and 0.047, p-value<0.001, respectively). By controlling for firm heterogeneity, results show that both development and acquisition of intangibles have a positive effect on turnover, and that the effect of the former exceeds the latter: a 1% increase in expenditure for development generates a 0.069% increase in turnover, while a 1% increase in expenditure for acquisition generates a 0.043% increase. While the argument is still aligned with the positive effect on turnover generated by inbound openness, results in this specification show that the degree of incidence is vis-à-vis lower than the one generated by development.

To address this issue, we control for the presence of firm-specific time trends in column (3). Again, results return a positive and significant correlation with turnover for both development (0.025) and acquisition (0.026), both significant at the 10% level. By isolating the effects of firm-specific time trend, we see that external assets have a marginally greater effect on turnover than internally developed assets: a 1% increase in the former leads to a 0.025% increase in turnover, while a 1% increase in the latter leads to a 0.026% increase in turnover. Results show that positive effects are consistent throughout the specifications, and their influence on the dependent variable, once we add controls, is almost equal.

For further specification, we added quantile regressions to verify whether there are differences in magnitude when compared to firm size. Results in column (4)–(6) show that the effects of development are positive and consistent for larger firms – in the 50th and 80th percentile of the sample – while for relatively smaller firms they are not. In contrast, the positive relationship between acquisition of intangibles and turnover is consistent throughout size classes, with a marginally greater effect for firms in the 50th percentile (0.024, p-value 0.001) than in the 20th and 80th percentile (0.017, p-value 0.001 and 0.0014, p-value 0.01, respectively).

In this light, we argue that asset acquisition (i.e. increasing inbound openness) generates positive results on firm performances, irrespective of their size, while development generates positive results (an increase in sales) only when companies are sufficiently large to exploit the outcomes of internal R&D. This, we believe, confirms that internal and external intangible assets are complements, rather than substitutes (Denicolai et al., 2016).

In Table 7 we repeat the same analysis with a different dependent variable, stock prices. Results in column (1) show a non-significant relationship between development of intangible assets and financial performance (shares value), and a positive relationship between acquisition and the dependent variable (0.075, p-value<0.1). By isolating the effects of sector, country and year specification, the expenditure on development of internal intangible assets does not appear to be correlated with financial performance, while expenditure on acquisition seems to be positively correlated with an increase in stock prices.

In column (2), results for both variables are both positive and consistent (0.067 and 0.045, respectively). Both results are consistent at the 5% level.

When controlling for firm-specific time trends in column (3), results show contrasting results to column (1): development is positively associated with firm stock price (0.066), and the relationship is consistent at the 5% level. In contrast, the effect of acquisition is negative (−0.007), although close to zero and non-significant.

Quantiles in column (4)–(6) show that development is not-significantly related with share values across all groups, while asset acquisition has a positive and significant effect for relatively larger firms in the 50th and 80th percentile (0.031 and 0.037, respectively), and such effects are significant at the 1% level, while for relatively smaller firms – in the 20th percentile – the relationship is positive but non-significant.

Here, we argue that results are consistent with the resource-based view of the firm (Barney, 1991) since firms’ proprietary technologies may represent the source of competitive advantage on the market, hence they reflect a positive effect on investors’ expectations which in turn increases the stock value. In contrast, acquisition of external technology does not influence financial performance, since purchased technology is seldom unique, and therefore easily replicable by competitors.

Again, in Table 8 we repeat the same analysis, using the level of employment as the dependent variable.

Results in column (1) show the presence of a positive and significant relationship between development of intangible assets and the level of employment (0.044, p-value<0.05). A 1% increase in firm's expenditure for development of intangibles creates a 0.044% increase in the number of employees. In contrast, acquisition of intangible assets is not significantly correlated with employment.

In column (2), both explanatory variables are positively and significantly related with the dependent variable at the 1% level, and the effect of development (0.052) is higher than the effect of acquisition (0.027): while both actions increase firm employment, the impact of the former is vis-à-vis greater than the latter.

Results in column (3) show that – however – only development appears to be positively and significantly related with the level of employment (0.035, p-value<0.05). Acquisition, in contrast, has a positive effect on the dependent variable, although it is close to zero (0.007) and non-significant. Quantile distribution shows that the effects of development for relatively smaller groups of firms – 20th and 50th percentile – are positive and significant at the 1% level (0.027 and 0.031, respectively), while for firms in the 80th percentile the effect on employment is non-significant. Acquisition of intangible assets, in contrast, generates positive and significant results on employment across all size classes.

We argue that – in terms of employment – development of resources generates higher returns: when investing creating new technology, companies tend to bring in additional employees to conduct R&D. Instead, when they leverage on the acquisition of technology, they tend not to increase the number of employees, but rather rely on the firm's absorptive capacity (Cohen & Levinthal, 1990; Denicolai et al., 2016) to integrate the acquired assets.

Summarizing the results, we see that when controlling for sector-, country-, and year-specific effects, development and acquisition of intangible assets generate different results: both affect the economic side of performance of a firm, measured in terms of turnover. Their effects on the financial and human capital side of performance, however, is different: development of intangible assets generate a positive effect on the level of employment, while acquisition of external intangible assets has a marginally positive effect on financial performance – measured as the value of firm stocks.

Controlling for firm-specific fixed effects, we find a positive and significant effect of both internally developed and external intangible assets. Results are consistent across all models: both variables generate positive results in terms of economic, financial and human capital performance. For all three dimensions, the effect of development is marginally greater than acquisition.

In column (3), we find that both explanatory variables are positively related with firms’ economic performance, and that the effects of both variables on turnover is almost equal: a 1% increase in development of intangibles generate a 0.025% increase in turnover, while the same increase in acquisition generates a 0.026% increase. In addition, we find that development of intangible assets is correlated with the other two performance dimensions, while acquisition is not.

Quantile distributions show that acquisition has a positive and consistent correlation among all class sizes with the economic and human capital dimension of performance, while in terms of financial performance the effect is significant only for relatively larger-sized firms (in the 50th and 80th percentile of the sample). Differently, development of assets generates a positive result on economic performance only for relatively larger firms, and a positive effect on the level of employment only for relatively smaller firms, while we find no significant correlation with financial performance.

We argue that the article provides many contributions to existing literature: first, it clarifies the dimensions considered to evaluate the level of openness, the level of performance and the relationship between the two dimensions. Then, it builds on extant literature to overcome existing limitations, proposing the measurement of inbound openness through the use of objective indicators, the amount invested in the acquisition of externally developed intangible assets and the amount invested by the firm in the development of intangible assets. Such variables provide more reliable indicators of firm propensity to openness than scale measures adopted so far by OI literature (Egbetokun, 2015; Ferreras-Méndez, Newell, Fernández-Mesa, & Alegre, 2015; Laursen & Salter, 2006). In addition, the paper proposes the study of the effects of development versus acquisition on three different dimensions for performance: (1) economic; (2) financial; and (3) human capital.

We argue that in this form the paper suffers from three main limitations: first, since it studies the effects of openness versus development using a sample of European companies, it misses out on the behavior of firms from different regions. In addition, since the sample only refers to listed company, we cannot control for smaller-sized firms. Finally, we did not control for the types of intangible assets – whether internal or external – the firms in the sample invest into, therefore missing on sorting out differences among various categories (patents, trademarks, concessions, etc.).

We believe the article is of interest for both scholars and practitioners. For researchers, the paper provides the basis for overcoming current limitations of OI studies in evaluating the relationship between openness and performance. The inclusion of objective measures of firm propensity for openness – i.e. the level of investment in external intangible assets – provides a more reliable indicator than scale measures derived from secondary data. In addition, the definition of multiple dimensions of performance helps in highlighting the presence of different effects of development versus acquisition across the dimensions.

For managers, we argue that the choice in investing on development of intangibles versus acquisition leads to similar positive results in terms of economic performance, although it seems that the latter is marginally more effective than the former. In contrast, such investments do not directly affect other dimensions such as financial and human capital performance, although we see a marginally significant lagged effect of development on employment.

For scholars, the work leaves room for the proposition of similar quantitative studies in different regions, provided that financial data are exhaustive and adopt the same accounting standards across countries. In addition, the inclusion of small- or medium-sized firms could highlight the presence of differences of behavior across size-classes, as posited by previous works (Vrande, de Jong, Vanhaverbeke, & de Rochemont, 2009). Finally, the inclusion of indicators for the amount invested by each company in different intangible asset classes could provide more suitable implications for professionals in the innovation field.