Explaining Telecoms and Electricity Internationalization in the European Union: A Political Economy Perspective

One consequence of the liberalization of certain services in the European Union was that a number of formerly inward-looking incumbents in telecommunications and electricity rapidly transformed themselves into some of the world’s leading Multinationals. However, the precise relationship between liberalization and incumbent internationalization is contested. This article tests three persuasive arguments derived from the political economy literature on this relationship. The first claims that those incumbents most exposed to domestic liberalization would internationalise most. The second asserts the opposite: incumbents operating where liberalization was restricted could exploit monopolistic rents to finance their aggressive internationalisation. The third argument claims that a diversity of paths will be adopted by countries and incumbents vis-à-vis liberalization and internationalization. Using correlation and cluster analysis of the sample of all major EU telecoms and electricity incumbent Multinationals evidence is found in favour of the third hypothesis. Internationalization as a response to liberalization took diverse forms in terms of timing and extent and this is best explained using a country, sector and firm logic.


Introduction
From the 1980s, the European Commission (EC) started to embark seriously on forging market integration in the network industries, particularly in telecommunications and electricity, even though it had long enjoyed significant legal competence in the field. 1 This new, liberalized policy environment which gradually extended over these two sectors substantially changed the business options available to incumbents. In particular, liberalization 'enabled' these previously inward-looking domestic incumbents to contemplate, and pursue, expansion abroad. As a consequence of the new policy environment, dozens of incumbents -previously perceived by some politicians as inefficient 'lame ducks' fit only for privatization during the new economic policy emerging from the 1980s (Crafts 1991) -rapidly transformed into highly respected, world class Multinational Corporations. Their emergence perhaps provided evidence at last of a new dawn of European 'international champions', this time not in the traditional industrial sectors (Hayward 1995), but in the network industries since, though business reached many corners of the globe, the overwhelming bulk of investment was in other EU countries (Clifton, Comín and Díaz-Fuentes 2007). The policies which underlined their emergence could be understood as a response to a concern that European business, including network industries, had to adapt to new technological and competitive challenges from the United States, Japan and beyond. Market integration in the network industries, it was anticipated, would result in a smaller number of more competitive firms better able to confront global challenges. From the 1990s, a significant number of these incumbents internationalized and now figure not only as some of Europe's but also the world's largest Multinationals. EDF, Telefónica, E.ON, Deutsche Telekom, France Télécom and RWE ranked in the world's top 25 non-financial Multinationals in 2006 (UNCTAD 2008). Fifteen years earlier, none of these firms figured in the top 100 ranking. Given this development, it would appear that European policy-makers met with some success. Now, liberalization was a 'prerequisite' of incumbent internationalization, because it reduced or removed previous restrictions on investment and ownership across borders. Internationalization would not have been prioritised or even permitted when nationalized incumbents were domestic monopoly public service suppliers. However, the precise relationship between incumbents' internationalization and liberalization is highly contested. Considerable tension has been generated around the perception that some incumbents embark on aggressive internationalization strategies in other countries which are relatively more exposed to liberalization -even daring to take over 'their' national 'jewel in the crown' -whilst the 'aggressors'' home governments delay or restrict liberalization in that sector. While this perception could generate disquiet in any industry, it is particularly alive in energy and communications, which have long been considered of national strategic, economic and social importance. Indeed, far from their strategic role becoming obsolete in the twenty-first century, new modes of terrorism have used network industries to organise (communications) and deliver (transportation and postal services) terror. For market integration to be successful, it is essential that a level-playing field is created and that it is perceived that all players stick to the rules of the game. To this end, common liberalization deadlines are set, and the EC uses various disciplinary instruments to 'punish' non-compliers. The problem of 'asymmetric behaviour' has been partially addressed in the drafting of the new electricity directive (EC 2009) through the so-called ´Gazprom clause' which stipulates prospective acquisitions by vertically integrated firms can be blocked if the target incumbent has unbundled. In practice, liberalization as a process is rarely implemented identically in different settings: the way that policy is understood, and the speed and depth of its implementation, invariably differ. Political economists ascribe the different ways in which a policy such as liberalization is implemented on the ground to the various and multiple pressures States receive from businesses, trade unions, NGOs, as well as the extent to which the State can respond (Smith 2001;Thatcher 2001;Henisz and Zelner 2006). Purposeful delay -or the perception of purposeful delay -could bring market integration to a stand-still ('why should we open up, with all the political headaches it involves, if they aren't?'). 2 Thus, the question of States' and firms' responses to liberalization cuts to the heart of the political economy of the integration process.
There are several persuasive arguments in the political economy literature on the relationship between internationalization and liberalization. Three main approaches will be tested here. The first argument underlines the logic of EC policy in this field: sectoral liberalization leads to the erosion of the incumbents' market share, exposing managers to the 'cold winds' of international competition. Fearful of being left behind in the 'race' to internationalization -investment opportunities are limited in these sectors -managers are pressurised to exploit firm economies of scale and know-how in new or more lucrative markets abroad. So, faster, deeper liberalization at home is associated with greater incumbent internationalization. The second argument is less optimistic: incumbent managers, faced by the challenges presented by liberalization, will lobby government to restrict or delay liberalization at home whilst, simultaneously, exploit opportunities opened up by relatively earlier liberalizing countries abroad. High-risk business abroad is supported by 'softer touch' liberalization, so greater incumbent internationalization is associated with slower and partial liberalization. The third argument is more influenced by comparative political economy literature. Liberalization is met by rational behaviour of States and firms but, because institutions matter, the processes of liberalization will differ. Internationalization, made possible by liberalization, will be pursued via different strategies, according to institutional circumstances so, national and sectoral responses to liberalization will result in various internationalization responses, explained by institutional difference; even if different paths are taken towards a similar end point.
Building on a body of scholarship on telecommunications and electricity reform (Börsch 2004, Eising 2002, Haar and Jones 2008, Héritier 2002, Murillo 2009, Thatcher 2001, Van Kranenburg and Hagedoorn 2008 this article analyses the role of liberalization policy in explaining incumbent internationalization outcomes in telecommunications and electricity in the EU. Correlation and cluster analysis methodology is deployed to analyse all major telecoms (12) and electricity (17) Multinationals in the EU plus Norway. 3 Analysis of these two sectors is justified because: of their role in economic growth; they provide critical networks for the movement of knowledge and energy required by the Single Market; they still constitute important instruments of the State; and, finally, they constituted key sectors in the privatization and liberalization 'wave' during the 1990s. After multiple rounds of liberalization, it appears much work is left to be accomplished in telecommunications and, particularly, electricity, before the Single Market could be understood to be nearing completion (Ilzkovitz et al., 2008). In July 2009, the EC ruled E.ON and Gaz de France-Suez had participated in 'market sharing', fining them 553 million each. Previously, in 2007, Telefónica was forced to pay 152 million euros when the EC ruled it had set unfair prices. There are, of course, many other issues that do not end up in highly publicised sanctions.
Deeper insight into the role played by liberalization policy in the internationalization of incumbents can shed new light on the political economy of market integration. It is found here that no causal relationship exists between incumbent internationalization and liberalization. Liberalization and internationalization changed the opportunity sets available for EU incumbents and their governments, but 'policy space' matters. Policy space is a fast-moving zone where States and firms 'embrace', 'baulk' or 'limp forward'. Some of the larger players moved aggressively to 'swallow up' smaller or less-convinced market players, in a West-East and North-South direction. Diversity is encountered, at the country, sectoral and, particularly, firm level. Decisions taken inside policy space can have long-lasting consequences on the ways in which the economy is structured.
The rest of the article is organized as follows. The second section presents the three main arguments on the relationship between incumbent internationalization and liberalization and derives hypotheses for competitive testing. The third section operationalizes the hypotheses and synthesizes the research design. The fourth section provides data on the major EU telecoms and electricity incumbents including internationalization. The fifth part contains the analysis divided into two sub-sections, telecommunications and electricity. Conclusions follow.

Three Hypotheses on the Role of Liberalization on Incumbent Internationalization
There is a vast literature in the social sciences on why firms internationalize. It is a daunting task to summarise this literature; here, three main points will be made about the state-of-the-art literature in order to contextualise the more specific political economy literature which deals with the role of policy on internationalization. It is first important to remember that most research on why firms went abroad focused on the manufacturing, oil and financial sectors, reflecting the profile of most twentieth-century Multinationals. Logically, much less attention has been paid to why firms in network industries go abroad, so their recent internationalization is presenting new research challenges (UNCTAD 2008, Jakopin 2008. Secondly, the reasons why a firm goes abroad are complex and interwoven, and cannot usually be reduced to a single factor. Theories or paradigms developed to explain firm internationalization take in multiple variables. Despite differences across schools of thought on international business, one particularly influential perspective was developed by John Dunning (1989). Briefly, the 'OLI' paradigm locates reasons for internationalization in 'O' (firm-specific advantages), 'L' (country specific advantages) and 'I' (internalization). Thirdly, scholars are increasingly recognising the role of policy and other institutional factors as variables in the internationalization decision, after having been rather neglected (Dunning 2009;Spar 2001). Policy considerations would fit broadly into 'L', since differences in the timing, extent and quality of policies such as liberalization implemented in the home and host country constitute part of the business environment in which firms operate. Policy is arguably an even more important factor influencing internationalization in the so-called 'heavily regulated' network industries. Telecoms and electricity incumbents did not have international presence to speak of at the beginning of the 1990s, and regulatory change, including liberalization, 'enabled' this internationalization to occur. Attention is now turned to how the relationship between liberalization policy and incumbent internationalization is conceptualised in different strands of political economy literature in order to derive the hypotheses.
The first two hypotheses are based on political economy arguments. The first argument underlies the logic of the Single Market project, as detectable in thousands of EC policy documents. 4 It is also the view expressed by network industry managers in a world-wide survey on internationalization drivers (UNCTAD 2008) as well as other academic accounts (Stienstra et al., 2004). Liberalization at the sectoral level forces incumbents to react and readjust. They increasingly notice how their monopolistic markets are being challenged by new entrants, resulting in a decline of their business. They also fear that a failure to liberalize domestically will compromise any potential outward expansion, due to reciprocity demands, meaning delay could prevent them from entering the internationalization 'race'. Firms that embrace 4 Neelie Kroes (2005), Commissioner for Competition explained: 'Companies that face strong competition at home are more likely to become successful on a global scale'. liberalization will be freer to seek out better, more profitable business abroad, exploiting their economies of scale. Macro policy reform has a direct impact on firm behaviour, therefore, which is assumed to be rational, uniform and profit-seeking. This perspective is 'generalistic' since attention is focused on the transformative power of policy and anticipates a common response from firms. Little attention is paid to institutional or firm-based differences. It is also 'optimistic', since it anticipates liberalization will have a uniform, lineal path, from design to outcome. If firms, States or both attempt to oppose liberalization, supervisory and disciplinary instruments can be used to ensure compliance by the EC. This narrative represents the 'hope' of policy-makers: competitive markets will drive down prices thus providing consumers with better services at lower cost. Thus, hypothesis 1 claims that the greater a firm is exposed to earlier and deeper liberalization, the more that firm will respond to increased pressure on its domestic market by increasingly going abroad in search of markets.
Another reading from political economy predicts a different outcome from which the second hypothesis is derived (Bonardi 2004;Chari and Gupta 2008, Haar and Jones 2008, Sarkar et al. 1999. In common with the first approach, liberalization is understood as being an important factor when explaining internationalization patterns of network industries; firms and States are understood to behave rationally and uniformly; institutional aspects are downplayed. The crucial difference is in the direction of the linkage between internationalization and liberalization. In a battle for survival, as liberalization quickens and deepens, firms, sometimes supported by their States, will seek to avoid or restrict liberalization at home. 'National champion' policies are a case in point: governments may opt to 'cushion' national players from the onset of a potentially damaging policy in order to shore up valued political support. Highly publicised examples include Italian Prime Minister Berlusconi's stated preference to keep Telecom Italia in 'Italian hands' and France's former Prime Minister Dominque de Villepin's 'patriotisme economique' pledge to protect eleven 'strategic' industries -including casinosfrom foreign takeover. 5 States can deliberately implement liberalization incorrectly, partially or slowly, giving 'breathing time' to domestic players to readjust and exploit other markets which opt to open up earlier. State protection of industry may be even more likely to emerge in industries such as networks, associated historically with the nation in economic, political, strategic and social terms. Protection provides a firm with 'safe' financial resources derived from monopoly rents which can be used to undertake risky international operations. Thus, hypothesis 2 argues that greater firm internationalization is associated with relatively slower and limited implementation of liberalization. The third hypothesis is influenced by the comparative political economy and institutionalist literature (Hall and Soskice 2001) which focuses on reform in these sectors (Börsch 2004, Levi-Faur 2006, Murillo 2009, Thatcher and Héritier 2002. The most nuanced of these accounts is Thatcher (2007). This adopts a policy analysis approach and argues different paths to reform, explained by institutional differences, may eventually lead to relatively similar outcomes. In common with the other two perspectives, it is assumed that liberalization is important and firms and States act rationally. Institutional and geopolitical differences, however, matter, and significantly shape processes and outcomes, hence countries and sectors may embark on different paths towards a similar reform direction. So, hypothesis 3 claims that governments and firms responded in various rational ways to liberalization, incumbent internationalization being one of those responses, and these differences of timing and extent can be explained by institutional differences even if some convergence is finally attained.
Testing these three hypotheses is the central aim of this article. However, there are two secondary questions that require brief attention: ownership and firm size. Liberalization has often been confused with privatization. These two policies are conceptually quite different, since liberalization entails introducing competition, whilst privatization means more private 5 Times 18 April 2007 ownership. Whilst the EC has competence in liberalization policy, it is up to national governments to implement privatization (Clifton, Comín and Díaz-Fuentes 2006). Did privatization influence internationalization? It could be argued that more privatization makes a company more visible to its stock-holders, forcing it to be efficient and maximise profits, whilst cutting its political ties make the firm more agile to move (Megginson and Netter 2001). Thus, the greater a firm is privatised, the more likely that firm is to respond to increased pressure on its domestic market by seeking out more profitable markets abroad (hypothesis 4). In many ways, hypothesis 4 is the corollary of hypothesis 1, in that greater liberalization and deeper privatization form part of the reform of the network industries so are seen as going hand in hand. The opposite of this argument is the corollary to hypothesis 2. Mergers and Acquisitions are often once-off, risky and politically complex operations: board-room politics often becomes transformed into 'high politics' when potential gains are significant. Incumbents with significant political involvement may be at an advantage in that they could have access to more information as well as to politicians who could 'smooth the way' for the operation to take place. Following this logic, firms with significant public ownership may be more likely to internationalize more strongly, thanks to interference from a political 'visible' hand. Thus, less privatization should be correlated to more internationalization (hypothesis 5). Finally, firm size could be an important factor enabling firm internationalization. For instance, there may be a minimum size that firms need to reach before internationalization becomes possible. Hence, firm size is a control variable throughout the analysis.

Research Design
Four hypotheses predict a lineal and continuous relationship between internationalization and liberalization policy (1 and 2), and internationalization and privatization (4 and 5), albeit in different directions. If hypothesis 1 is correct, we would expect to find correlations between higher levels of firm internationalization with deeper and faster implementation of liberalization, in its multiple forms. Hypothesis 2 is correct if restricted and more sluggish liberalization implementation was correlated with greater incumbent internationalization. As regards internationalization and ownership, hypothesis 4 predicts that more privatization will be correlated with greater internationalization, and hypothesis 5, less privatization would be correlated with greater internationalization. Incumbent size is controlled for throughout. Correlational analysis measures the strength of the associations between the independent and dependent variables, thus is appropriate to test these hypotheses. Hypothesis 3, in contrast, predicts that there is no fixed relationship between the variables; rather, there will be multiple paths in terms of the timing and extent towards incumbent internationalization and liberalization, which can be explained by institutional differences. Cluster analysis is ideal for testing this, since patterns of incumbent behaviour are made visible.

A 'Snapshot' of Recent Internationalization of EU Telecoms and Electricity Incumbents
Before proceeding to competitively test the hypotheses on the relationship between incumbent internationalization and liberalization, a sketch of the internationalization of major EU telecoms and electricity incumbents is provided. Tables 1 and 2 respectively show the major telecoms and electricity Multinationals between 1999 and 2006, ranked by revenue in 2006. The 'Western bias' of the integration process can be seen since only major Western European Multinationals emerged, whilst Eastern Member States were generally recipients of this process. Data is provided on the timing and extent of internationalization, liberalization and ownership, revenue and employees. Definitions and measurements of internationalization, liberalization and privatization require explanation. International activity by firms takes two main forms: global alliances or the physical extension of the firms' sales, assets and/or employees abroad. It is this second activity that has been of greater importance in telecommunications and electricity, so it is this 'physical' internationalization that is considered here. Internationalization is quantified as foreign revenues as a percentage of overall revenues. 6 Data on foreign operating revenues is derived from annual company reports and Amadeus (2009). Liberalization is complex to quantify. The OECD (2009) is perhaps the most comprehensive quantitative database of regulation and is used here. OECD methodology constructs different sets of indicators for liberalization in telecommunications and electricity (Conway and Nicoletti 2006). For telecommunications, liberalization is measured in two ways. Firstly, an indicator is constructed for 'Entry Regulation', meaning to what extent legal systems allow for new entrants, 0 being they do not, and 1 being completely. 7 The second indicator, 'Market Structure', indicates what market share new entrants enjoy, as a means of gauging the extent to which liberalization leads to actual competition. Zero means none and 1 means the total market. For electricity, 'Entry Regulation' measures the terms and conditions for third party access, the extent to which consumers can chose supplier, and the existence of a liberalized wholesale market for power. Zero means none, 1 means this is fully liberalized. The second electricity liberalization indicator is 'Vertical Integration', or the extent to which the industry has been unbundled, 1 meaning the industry is integrated, 0 meaning it is fully unbundled (Conway and Nicoletti 2006). Indicators for ownership are also included: 0 means full public ownership; 1 means total private ownership. 6 Methodology deployed by the United Nations Conference on Trade and Development (UNCTAD) constructs a composite proxy of internationalization combining data on sales, employees and assets abroad. When this information is not available, however, data on sales, employment or assets abroad only is used. To avoid possible bias that the UNCTAD approach could cause, we use here 'only' data on sales abroad. 7 This is a composite indicator including mobile, trunk and international long distance telecommunications services.  (Table 2). Technological, economic and ideological factors influencing reform and EU responses have been documented elsewhere (Domanico 2007). Again, all incumbent Multinationals are from Western Europe. Unlike telecoms, where there is one major national incumbent, in electricity, there may be several, due to the organization of the sector as regional monopolies or else as a result of unbundling, as in the case of National Grid. This fact should not cause sample bias because firms in the same policy environment may -indeed do -behave differently. Thomas (2003) predicted that the outcome of the Single Market in electricity would be a 'seven sisters' oligopoly. Along similar lines but for the case of France, Bauby and Varone (2007) argued that one of the paradoxes of European market integration was the successful 'engineering' of national energy giants, EDF and the multi-utility GDF-Suez. By 2006, the EU only had five energy giants left: E.ON, RWE, EDF, GDF-Suez 8 and ENEL. A pessimistic 'seven sisters' oligopoly now seems over-optimistic. Examining internationalization patterns, E.ON recalls Telefónica's behaviour in that it went international very early on (48% of sales were earned abroad in 1999) whilst enjoying monopoly conditions at home. Internationalization of incumbents was, on average, 11% in 1999, and 39% by 2006, a strikingly similar outcome to the extent of internationalization in telecoms incumbents over the same period. International patterns are uneven but it is notable how some of the fastest growing incumbents during this period were the medium-sized firms, namely Vattenfall, EnBW, National Grid and EDP. Comparison of the data on progress towards liberalization for telecoms and electricity shows how, whilst in telecoms, Entry Regulation was virtually in place by 1999, progress was slower in electricity. One convincing reason for the slowness of transposition of electricity vis-à-vis telecoms reform has been provided by the argument that the first was an intergovernmental process whilst the second was supranational (Levi-Faur 1999). So, if by 1999, nearly all countries had liberalized telecommunications, there were a number of laggards in electricity. As in telecoms, the timing and extent of liberalization was very uneven. The UK was uniquely early in its pre-emption of EC Directives: full liberalization and unbundling were reached as early as 1995. The Nordic countries were also early movers to liberalise Entry Regulation, though Denmark was the only one to fully unbundle by 2002. These countries had historically traded electricity with each other to balance their systems and, in the late 1990s, they established the Nordic Power Exchange for a single electricity market for the four countries. After the UK lead, the path to unbundling was uneven; Spain (2002), Italy (2003) and the Netherlands (2004) responded to EC Directives. Other countries, particularly France, Belgium, Germany and Portugal moved more slowly. As regards ownership, privatization was quite slow during the period, increasing on average from 44% to 51% of these incumbents. Here, there was huge diversity: in 2006, incumbents from Germany and the UK were fully privatised whilst public ownership still dominated in Sweden, the Netherlands, France, Switzerland, Austria and Denmark. The privatization of incumbent Multinationals went much further in telecoms than in electricity.

Analysis
The five hypotheses are now tested using correlation and cluster analysis techniques. Results are divided into telecoms and electricity.

Telecoms
Correlation between variables using Pearson bivariate correlation, Kendall rank and Spearman rank correlation were used to detect the strength of association between internationalization and entry regulation, market structure, ownership, size (revenue and employees) for 1999, 2003 and 2006. Results are shown in Annex 1. Using Pearson's correlation, in 1999, there is a negative relationship between internationalization and public ownership (privately-owned incumbents were more likely to go abroad), though this correlation is not apparent applying Kendall and Spearman correlations. However, this correlation did not reappear using Pearson in 2003 or 2006. No other significant variables were detected which correlated with internationalization, including the two indicators for liberalization. So, over this period, indicators on liberalization, ownership or size do not explain the extent of incumbent internationalization. No evidence on hypotheses 1, 2, 4 or 5 is obtained.
Next, cluster analysis is deployed to search for groups that are found to be similar in one or more sets of variables, to test hypothesis 3. All 12 telecoms incumbent Multinationals were considered for extent of internationalization, entry regulation and market structure for 1999, 2003 and 2006. Results are shown in Table 3. Since all Member States had attained complete entry regulation from 1999, this variable is no longer of use and is excluded from the analysis henceforth.

High internationalisation and high liberalisation
The cluster analysis reveals some interesting patterns. Starting with 1999, there are two sets of findings: internationalization and entry regulation, and internationalization and market structure. Regarding the former, two incumbents -TDC and Telefónica -set the pace to internationalise, and which constitute cluster 4. Both incumbents underwent significant internationalization and were based in countries where entry regulation had been liberalised. The vast majority of incumbents, however, fell into cluster 3; here, internationalization is rather slow, whilst entry regulation is liberalised. Portugal and Greece predictably fall into a fourth category, cluster 1; where incumbent internationalization is slow and entry regulation is officially delayed.
Analysis of internationalization and market structure throws a more nuanced light on these results, particularly as regards the strategies of TDC and Telefónica. TDC is left alone in Cluster 4, since market share is quite liberalized in Denmark. Telefónica uniquely comprises Cluster 2, having embarked on an ambitious internationalization programme whilst enjoying a relatively high share of its domestic market. Hence, TDC and Telefónica emerge as opposites: the two most international of companies pursued this expansion based on different shares of the domestic market. Again, the vast majority of incumbents fell into the same category, cluster 1; here, internationalization is low, as is market structure liberalization. Exceptions are BT and Deutsche Telekom (cluster 3), where internationalization is quite low but market structure has been highly liberalized. The clusters in 1999 show clearly that there are no automatic relationships between the variables under study, rather, in similar situations incumbents pursued different paths toward internationalization.
A number of patterns emerge over the next seven years. Firstly, Telefónica is joined by Telenor in cluster 2. As mentioned in section four, Telefónica and Telenor were both internationalization 'stars' as regards their aggressive pursuit of internationalization. Both did so in a context of slower liberalization of market structure. Telenor emulates Telefónica's strategy from 2003. Secondly, ambitious internationalization is now pursued by other incumbents this time in the context of a liberalised market structure. TDC's strategy is adopted by TeliaSonera, France Télécom, Deutsche Telekom and, to a lesser extent, Telekom Austria, which comprise cluster 4. There is a third group of incumbents (cluster 1) which internationalized more slowly, based in countries where market structure was less liberalised: Telecom Italia, KPN, Portugal Telecom and OTE. Finally, BT alone forms cluster 3, as incumbent internationalization was relatively low and where market structure had been highly liberalization. BT's lower international level was due to the fact that much of its initial international activity was sold off after its abandonment during firm re-organization.

Electricity
Using the same correlation techniques and periods of time as for telecoms, the extent of incumbent electricity Multinationals was analysed, considering entry regulation, vertical integration, ownership, revenue and employees. No correlations were detected between incumbent internationalization and entry or vertical integration (see Annex). In 1999, there is a significant correlation between incumbent size and internationalization, though this is not seen in 2003 and 2006. It appears that larger firms had the edge when embarking on internationalization strategies in the earlier period. However, since no correlations were found between liberalization and internationalization, no evidence for hypotheses 1, 2, 4 and 5 was detected.
Next, cluster analysis is applied in order to detect any patterns in incumbent internationalization, considering the same variables and time period as previously. Results are shown in Table 4. The relationship between internationalization and entry regulation is first analysed, shown on the left part of Table 4. In 1999, the most internationalized of incumbents fell into two clusters. On the one hand were those incumbents which internationalized strongly whilst entry regulation was also liberalized, Fortum and Endesa, forming cluster 4. E.ON, in contrast, stands out for its aggressive internationalization in the context of low entry regulation liberalization. As in telecoms, the leader incumbent internationalizers emerged from contexts where liberalization is both less and more advanced. E.ON could be compared to Telefónica in its pursuit of ambitious internationalization from a relatively closed market. Most incumbents pursued relatively cautious internationalization programmes in 1999. There were two similarly-sized clusters of incumbents here: cluster 3 where liberalization was more advanced, and cluster 1 where this was delayed. Included in cluster 3 were Spanish regional incumbents (Iberdrola and Unión Fenosa) and in cluster 1 were German regional incumbents (RWE and EnBW). These Spanish and German incumbents had much lower internationalization levels than Endesa and E.ON respectively. This suggests that even a nationalsectoral approach cannot account for the variety of firm strategies adopted. Rather, varieties of response are located at the firm level. Finally, most incumbents pursued internationalization slowly; only five of seventeen incumbents were pursuing internationalization with great enthusiasm in 1999. based in home markets where entry regulation is liberalised. The second largest cluster, 3, is composed of four incumbents, grouping incumbents which internationalised less though from liberalized environments (Enel, Iberdrola, EnBW and Essent). Interestingly, two of these more 'hesitant' internationalizers, Iberdrola and Enel have, since 2006, completed huge acquisitions. In 2007, Iberdrola took over Scottish Power, whilst in 2009, Enel took over Endesa. It seems that the 'hesitant' internationalizers 'bided their time' until they made their move to acquire more internationalised incumbents, absorbing all their international business. Chronologically, the UK was the first of the three to liberalise entry, followed by Spain with Italy trailing behind. From this perspective, a 'waitand-see' logic may have proved advantageous: slower liberalisers took advantage of incumbents in countries had had liberalised previously. The wave of massive Mergers and Acquisitions reflects the fact that the EU electricity market is characterised by monopolistic competition, and that market integration is resulting in domination by a small number of huge Multinationals. Belgium's traditionally private Electrabel was the main exception to the rule. Here, a defensive strategy was at work. Electrabel pursued an ambitious internationalization programme between 1999 and 2006 as the government delayed market opening. Fears about the incumbent's vulnerability were proved correct when immediately, on opening the market, Electrabel was snapped up by Suez, after which both were merged with Gaz de France to form one of Europe's largest multi-utilities (Bauby and Varone 2007).
As regards internationalization and vertical integration, the most internationalized of electricity incumbents are divided up nearly equally into two clusters, since, whilst entry regulation liberalization was nearly complete in 2006, progress on unbundling was mixed. First, there was a group of seven highly internationalised incumbents based in countries where unbundling was more advanced (cluster 4). This included National Grid, Scottish Power, Endesa, Unión Fenosa, Dong Energy and EVN. These incumbents were able to internationalise as both entry regulation and unbundling were implemented. Second, a group of six incumbents (cluster 1) pursued significant internationalisation expansion, in a context of liberalised entry but delayed unbundling (E.ON, EDF, RWE, Vattenfall, EDP, and Fortum). The main exception was Electrabel, which had delayed both forms of liberalization, as previously explained. So, Finland's Fortum, which in 1999 seemed to be setting the pace for internationalization in the context of advanced liberalization, and saw its foreign revenues increase over seven years from 32% 73%, did so whilst unbundling remained stagnant. A similar observation can be made of the other cluster members. A third cluster, 3, comprises three incumbents whose internationalization was somewhat slower in a context of greater progress unbundling. EnBW is alone in cluster 2, enjoying higher vertical integration but less internationalization. Here, it can be seen how Iberdrola and EnBW, operating in the same policy environment as their other highly internationalized Spanish and German peers, were both much slower to internationalise. Again, diversity is beyond national and sectoral patterns, it is ultimately located at the firm level.

Conclusions: Explaining Internationalization Patterns in Telecoms and Electricity
Regulatory reforms defined broadly as liberalization were a prerequisite for the rise of telecoms and energy Multinationals. The internationalization of EU incumbents could not have taken place without liberalization of entry regulation and would have been difficult without progress on unbundling and privatization. However, on the ground, it is highly unlikely that all countries implement liberalization in precisely the same way at the same moment. The perception that some countries behave asymmetrically, by delaying or restricting liberalization whilst promoting 'national champions' to takeover other countries' strategic 'jewels in the crown', is a considerable source of tension in the EU. A clearer understanding of State and firm response to liberalization helps shed light on the political economy of market integration.
Three main hypotheses on the relationship between internationalization and liberalization were established. The first hypothesis predicted that those incumbents most exposed to earlier and deeper liberalization would be those which internationalized most. The second argument predicted that incumbents would pressurise States to restrict or delay liberalization, so those with secure financial and political resources would be most able to embark on high-risk adventures abroad. Correlation was used to test these hypotheses, and it was confirmed that no evidence existed on a direct relationship between internationalization and liberalization or ownership. Hypotheses 1 and 2 (and secondary hypotheses 4 and 5) were rejected.
Attention was then turned to hypothesis 3, which, asserting the importance of institutional diversity, anticipated that countries and sectors would pursue various routes toward liberalization and internationalization, though often moving towards a similar point. Cluster analysis was used to reveal a diversity of responses to liberalization and internationalization. In general, this diversity can be organised at the country level, with modifications for sectors and, also, for firms. Countries and firms can be organised into larger and smaller ones.
The Single Market led to the emergence of Multinationals in telecommunications and electricity from Western Europe; Eastern Europe was a recipient. The large continental countries, particularly France and Germany, dominated the battle for precedence in assuring their respective national incumbents would dominate European Multinationals in both sectors. Neither were liberalization 'pace-setters' nor were they consistent 'laggards': rather, they were 'middle-of-the-roaders'. France was slower than average to liberalize electricity, whilst E.ON's early internationalization occurred in near monopolistic conditions. In telecoms, France moved to liberalise at an average pace; Germany was somewhat faster. Spain and Italy took strides to join them: Telefónica emerged as a leading world Multinational in near monopoly conditions, though Spain was among the 'pace-setters' liberalizing electricity. Relatively faster liberalization did not prevent Endesa from emerging as a leading European Multinational, though Iberdrola and Unión Fenosa were more hesitant to internationalize. Italy was somewhat slower to implement liberalization and its incumbents were slower to go abroad, nevertheless, Enel and Telecom Italia occupied positions in the top five by 2006. The most international of the EU's Multinational telecoms and electricity incumbents emerged from the larger continental economies: France, Germany, Italy and Spain. Though there were no automatic relations between the timing and extent of liberalization and incumbent internationalization, the large part of these Multinationals came about more thanks to a slower or middle-of-the-road approach to liberalization than a faster one. The UK took a different path: it embraced liberalization enthusiastically, acting prior to EC directives in both sectors. Today, UK incumbents do not dominate the EU Multinational rankings in these sectors. In telecommunications, BT sacrificed its domination of the rankings, de-internationalising in order to prioritise its home market. The UK now presents itself as a highly attractive site for investment: Telefónica's O2 has already overtaken Vodafone in the UK, and proposed mergers between Orange and T-Mobile, and France Télécom and Deutsche Telekom would put Vodafone further down the UK ranking (Parker 2009). In electricity, Scottish Power was taken over by Iberdrola.
Among the smaller economies, the Nordic countries constituted a close group. In general, these countries liberalised earlier, whilst incumbents responded enthusiastically to internationalization options, though on a sub-regional basis. In electricity, this was because a trade pooling system existed; in telecommunications, the 'star' internationaliser, Telenor, earned 23% of foreign revenue from other Nordic countries, 12% from Eastern Europe and 30% from beyond Europe. In telecommunications, Norway liberalised more slowly than Sweden, Finland and Denmark, and its investments beyond Europe were high, hence Telenor's comparison with Telefónica. Defensive patterns dominated behaviour of many of the other smaller economies. In Greece, the Netherlands and Portugal, liberalization was implemented relatively slowly, and incumbents internationalised though quite cautiously. In Belgium, the efforts to protect Electrabel via slow liberalization ultimately failed. Austria was bolder to liberalize and its incumbents to internationalise, particularly focusing on the markets in East Europe.
Internationalization patterns of EU incumbents in telecoms and electricity are best explained using comparative political economy lenses, whereby country and sectoral trends, interwoven with the firmlevel, provide a superior explanation for the outcome of market integration in these critical sectors.

Annex I
Correlation techniques were used to test for the relationship between incumbent internationalization, liberalization and privatization, controlling for incumbent size in 1999, 2003 and 2006. Bivariate Pearson's correlation coefficients were run to determine the degree of relationships among the predictors and incumbent internationalisation (Table A.1 tests electricity, Table A.3, telecommunications). We also conducted Kendall rank correlation (a non-parametric test) and Spearman rank correlation (a non-parametric test to measure the degree of association between variables). In general, Kendall and Spearman correlations show similar results, so, only Kendall correlations are shown (Table A.2 for  electricity, Table A.4 for telecommunications). Notes: N=16, * Correlation is significant at the *p<0.05 level (2-tailed). ** Correlation is significant at the **p<0.01 level (2-tailed).  Notes: N=12, * Correlation is significant at the *p<0.05 level (2-tailed). ** Correlation is significant at the **p<0.01 level (2-tailed).
a. Cannot be computed because at least one of the variables is constant.