Power in Standardisation: The Case of CEN

Homo Oeconomicus 23(3/4): 327 345 (2006) www.accedoverlag.de Power in Standardisation: The Case of CEN Jörg Gröndahl Institute of SocioEconomics, University of Hamburg, Germany (email: groendahl@econ.uni-hamburg.de) Abstract This paper emphasizes the importance of voting power in standardisation, especially for public standardisation committees. The voting game in committee standardisation is described by the example of CEN. Different power indices are applied to express the distribution of power in European standardisation organisations with regards to the changes in the design of these institutions that came with EU-enlargement. All indices considered yield fairness as implied by the criterion population share only for The Netherlands. Larger countries are under- and smaller over-represented. It is shown that adopting voting rules and weights from another institution does not necessarily yield the same result. The public good property of standards makes a difference in this respect, compared to voting on private goods. The increase in the decision rule is working in favour of CEN s effort to unanimity, but at the cost of a considerable decrease of CEN s ability to act. Keywords power to standardise, committee standardisation, voting power 1. Introduction The debate about the new Nice-treaty voting weights for the European Council of Ministers, associated with the debate about the European Conventions draft European Constitution, shows that the question of voting power is of paramount importance in public institutions. Instead of concentrating on the Council of Ministers, the focus in this article is on European standard development organisations (SDOs) that vote about formal standards, i.e. CEN (Comité Européen de Normalisation), CENELEC (Comité Européen de Normalisation Electrotechnique) and ETSI (European Telecommunications Standards Institute). The distinction of a formal standard from a market or de-facto standard is that the latter evolves over time on a given market by market forces, while the former i.e. a committee standard is decided upon by 2006 Accedo Verlagsgesellschaft, München. ISBN 3-89265-063-2 ISSN 0943-0180

328 Homo Oeconomicus 23(3/4) some SDO. Generally, governments require committee standards for health and safety issues. For reasons of compatibility they are needed in case of market failure or if the market result is inefficient from a welfare perspective. It is a well-established result that compatibility standards create positive network effects. Substantial work has been done to examine network effects and positive externalities due to standardization in general, e.g. by Farrell and Saloner (1985, 1986, 1988), Katz and Shapiro (1985, 1986), and Economides (1996). The importance of formal or committee standards is still underestimated and receives less attention than it deserves in the literature. However, its relevance is growing. The European unification process brought with it enormous efforts to unify the diverse national standards and create European standards via the European SDOs. According to CEN Statistics (2004) their output of European Standards and approved documents has increased to 9406 by January 2004 and over the last ten years the number of European Standards available has risen from 2000 to more than 14 000 and... continues to grow by more than 1000 annually according to the European Commission (2003). So far, with the exception of Farrell and Saloner (1988) and Goerke and Holler (1995), not much work has been invested into examining committee standardisation. Especially the power to standardise has not been a key subject of earlier studies. Generally SDOs are national or international committees, e.g. ISO (International Organization for Standardization), ANSI (American National Standards Institute), or DIN (Deutsches Institut für Normung). These are private non-profit organisations, developing national standards on behalf of countries (national parliaments) or supranational institutions (e.g. The European Union [EU]). European norms are developed by the three European SDOs: CEN, CENELEC and ETSI. Standardisation in electro technical and information and communication technologies in Europe are the subjects of CENELEC and ETSI, respectively. The most important standardisation organisation of the EU, measured by the number of European norms developed, 1 CEN, promotes voluntary technical harmonisation in Europe in conjunction with world-wide bodies and its partners in Europe. [CEN] was founded in 1961 by the national standards bodies in the European Economic Community and EFTA [European Free Trade Association] countries. Now CEN is contributing to the objectives of the European Union and European Economic Area with voluntary technical standards which promote free trade, the safety of workers and consumers, inter operability of networks, environmental protection, exploitation of research and devel- 1 By early 2004 CEN developed about 2/3 of all European norms.

J. Gröndahl: Power and Standardisation 329 opment programmes, and public procurement (CEN, 2005). Although CEN is the main subject of this study, due to the common rules for standardisation work for CEN and CENELEC, the results derived are in principle the same for both organisations. CEN generally adopts voting weights and voting rules of the EU Council of Ministers and is thus affected by the above-mentioned changes. In addition there are votes for EFTA members, who are also members of CEN, and for some possible EU entrants. The voting weights of these non-eu countries are comparable to weights for EU countries with a similar population size. Furthermore, the decision rule is slightly altered (71 instead of 73.9 percent after enlargement) and there is no triple-majority threshold as in the EU Council of Ministers, i.e. in CEN there is only weighted voting of the members and no threshold of member countries population or member countries has to be met. This article examines the power implications of the recent changes in voting weights and decision rule in CEN. It furthermore provides an attempt to evaluate the implications on the representation of countries, i.e. whether e.g. Germany is under- or Spain and Poland are over-represented. In the following chapter power in standardisation is examined in greater depths. Particular attention is paid to the difference between public and private good interpretations of power in standardisation. Chapter 3 illustrates that power is important in committee standardisation. Chapter 4 is concerned with modelling the voting game according to the statutes of CEN. Chapter 5 gives a short overview of the applied power indices and shows why the public good property of standards is vital for the selection of a specific index. Chapter 6 examines the changes in the distribution of power in CEN. The analysis demonstrates how the balance of power in a standardisation body changes with the composition of the body and the decision rule on one hand, and with the applied power index on the other hand. Concluding remarks follow in chapter 7. 2. On power in standardisation The efforts of measuring voting power have mainly been dedicated to investigating decision making in legislative bodies, mostly within the UN, national parliaments, EU, WTO, and IMF. Power indices assign values (power) to resources (e.g. voting weights) and are thus a valuable tool for analysing institutional changes or comparing alternative institutional frameworks. A priori voting power in these organisations was measured in order to evaluate the distribution of power in these organisations that results from the institutional design. A priori voting power is the question of effect on the outcome, i.e. the question whether changing the own vote will change the outcome of the group decision. This means that the ability to

330 Homo Oeconomicus 23(3/4) influence the outcome is measured and not the outcome of an explicit decision process. The preferences of the respective members of the committee and the issue at hand are therefore not considered. Before investigating committee standards and the related power to standardise, the question must be answered, what power means in this context. Among the general properties of formal standards are non-excludability and non-rivalry in consumption, the so-called public good property of standards. A formal standard is not excludable, since it is publicly available, 2 and the standard-specifics as such are always non-rival in consumption. Non-rivalry in consumption in this context means that the application of a standard by agent i does not forestall the use of the same standard by agent j. In most studies about the consequences of the Nice treaty a policy-outcome is treated as a private good as implied by the use of private good power indices such as Shapley-Shubik or Banzhaf indices, e.g. Leech (2002) and Felsenthal and Machover (2001). It remains an open question, whether the outcome of a policy decision is a private or a public good. Since a policy-outcome (e.g. a law) will be applied to the entire country, it seems to be appropriate to treat it as a public good. Alternatively, one could argue that policymakers sometimes have huge stakes in their decisions and will therefore regard the outcome as a private good. This reasoning could also apply to standardisers, although we claim standards to be a public good. Since there are valid arguments for both perspectives, voting power in standardisation is evaluated with regard to power indices appropriate for public goods and for private goods. Therefore the Public Good Index will be applied in addition to the above-mentioned indices. The national members of CEN are the standardisation bodies of the member countries, i.e. national SDOs like DIN, AFNOR (Association Française de Normalisation), or BSI (British Standards Institute). They generally have two different reasons to vote for a proposed standard: the first is to agree on a common European standard and the second is to agree on a standard that fits a country s interests. Concerning the first reason we have to view the standard as a public good, since it will be a standard for the entire EU. As to the second reason, of course a member country that wants a common standard might not be satisfied with the standards proposal put forth by CEN-workgroups (technical committees). But to forestall this explicit standard, it has to either vote against the standard. This decision might not be in its interest, since this could translate into no standard at 2 The costs of buying standards might be considerable, but not prohibitively high.

J. Gröndahl: Power and Standardisation 331 all. Or the country s representatives have to take part in the creation of the proposed standard in the workgroup by e.g. devoting some resources to this workgroup. Furthermore the country could decide to free ride on the resources that other countries might contribute, especially if some other countries have corresponding interests. But since these countries could also decide to free ride they all face the risk of an unfavourable standard or no standard at all. A full analysis of voting power that derives from participation in standard creation by devoting resources to a workgroup, i.e. agenda setting power, is beyond the scope of this paper. The following is restricted to a priori voting power. The focus is on the question how the distribution of voting power among the members of a standardisation body influences the outcome. Knowledge about the power implications of voting rules and voting weights is needed in order to design appropriate institutions. Concerning fair representation, in many studies about voting power in the EU (compare e.g. Leech, 2002) it is assumed that the influence of a voter shall be the same in all countries. Due to the two-stage voting procedure, first in the respective country and then in the EU Council of Ministers, this could be achieved by the square root rule. 3 In standardisation, e.g. in CEN, there is not necessarily a second voting process involved. In the member countries there are not generally voting decisions taken within the national SDOs. Either their stance on an issue is given by their agenda or regulations, or they decide on the subject matter. But even if a final vote takes place within the national SDO prior to the decision in CEN, the respective country s population is not involved in the SDOs decision via voting. Therefore it is assumed here that the national members (SDOs) should have the influence of the sum of the respective voters. Smaller countries could be given more power in order to protect minority rights. In the EU Council of Ministers this shall be accomplished by further majority thresholds of 62 percent of EU population and simple majority of member countries. The question comes up, whether membership should rather be based on a regional level in order to protect minorities within countries. Or, whether all countries should be given the same power by attributing the same voting weight to every country, although analysing this case would be less interesting and trivial. Since there is no sufficient reason for any of these possibilities, in this paper, countries power shares are regarded as fair, if they correspond to 3 The square root rule yields the same voting power for every citizen in e.g. the EU. It states that the power of a member country should be proportional to the square root of this member country s population. For an accessible explanation why, see Baldwin and Widgrén (2004).

332 Homo Oeconomicus 23(3/4) the share of countries populations. Thus the power shares should be identical to the population shares. Correspondingly, fair representation is defined here as the distribution of power between member countries that gives every member country its fair share. 4 Fair representation cannot be accomplished, if the voting body gives members voting weights that imply power shares that do not correspond to population shares. Therefore we have to consider these power implications when constructing new or changing existing institutions. Accordingly, fairness in European standardisation decision-making is tested, applying correspondency of power and population shares as measure. 3. Committee standards It has been said that in the context of European standardisation the objective of a standard is to increase welfare. Sometimes, however, standards can have negative implications on welfare. Goerke and Holler (1995) examined the theoretical possibilities of committees to standardise in one out of two possible standards or to choose variety. They conclude that standardisation is efficient, if the network effects of the standard are superadditive in that they are more than sufficient to compensate the outvoted group [of consumers of a standardised product] for the utility reduction it incurred due to the loss of choice (p. 340). If standards increase welfare we need formal standards for the case that the market outcome is inefficient from a welfare perspective, i.e. if the market outcome would be a suboptimal standard or no standard at all. But even if a de-facto market standard creates positive network effects, a formal standard can still be more efficient. Farrell and Saloner (1988) present a game-theoretical approach to committee standardisation, modelling the coordination process within a committee. Analyzing the mixed-strategy equilibrium they show that the committee system outperforms the market outcome (although a committee system is slower than the bandwagon system), but that a (mixed) hybrid system would do even better. Of course the standard proposed to a committee does not fall out of the blue, but evolves over some unspecified period of time, coined in negotiations by the vested interests of all the persons and groups that take part in 4 This is not to say that population share, whether measured by the relative share of the population or by the square root rule is the only relevant criteria for fairness in this respect, nor that fairness is an ultimate goal of standardisation. In EU voting procedures there are e.g. double-majority rules that require a simple majority of members and 62 percent of population to be in favour of a proposal for this proposal to pass. The second majority criterion can in principle compensate for unfair representation that follows from applying the first.

J. Gröndahl: Power and Standardisation 333 its development. In CEN these are national members (national standardisation bodies; they are the only participants in voting decisions), counsellors (European Commission and EFTA secretariat), affiliates (standardisation bodies of countries that can in principle become EU and/or EFTA members that are not already national members, e.g. Bulgaria, Romania, and others), corresponding organisations, and associates, which are pressure groups representatives of industry and small and medium-sized enterprises, consumers, environmentalists and trade unions. This process is essentially the same for all committee standards, although different groups might participate, different rules might apply, and the objectives might differ. A de-facto market standard may also become an official intra-industry, regional, national, or international standard, if the need for a formal standard has been discovered by the respective standardisation organisation and after a decision within this body has been made. Therefore the theoretical reasoning provided is applicable to a wide range of decision-making bodies. On the one hand, there are public standards organisations that issue legal or free formal standards. On the other hand, there are private bodies within an industry. These are, again, mostly committees, e.g. ZVEI [Zentralverband der Elektroindustrie]. The feature that establishes a link between all these standardisation bodies is that their decision is about a public good, or perhaps a club good, and not about a private good. The differentiation between public goods and club goods on one hand and private goods on the other hand is the possibility of free riding, which exists for public goods and club goods. It is not possible to exclude anyone anyone that is a member of the club or, in the case of a national standard, anyone in that country from the standard. Every member in the group of decision makers will have to accept the standard and cannot be excluded or exclude himself from the standard, even if he voted against the standard. Examples of legally binding standards are some European norms that have become part of national legislature. This is the case with quality standards of health and safety. According to the statutes of CEN, articles 6.4 and 6.1.2, if the standardisation bodies of the EU countries have approved a standard, conflicting standards have to be withdrawn in the respective countries, unless this is outside the standardisation body s competence 5 (CEN 2002, 2004). 4. Modelling committee standardization Formal standardisation can be modelled as a simple voting game Γ. A given standard-proposal is decided upon by players i with respective vot- 5 An exception from this rule is described in the next chapter.

334 Homo Oeconomicus 23(3/4) ing weights w i and a decision rule d. In general, decision rules d > 0.5 will be regarded only. Otherwise multiple majorities could occur. There are n individual players within the voting body N. W is the set of winning coalitions. Then, the coalition value of coalition S, vs ( ), is either 1, if S W, or vs ( ) = 0 otherwise. The understanding of simple games, represented by the decision rule d, the voting weights w i, and the players i ( i= 1,..., n), Γ = ( d; w1,..., wi,..., w n), is essential for institutional design, i.e. implementing efficient and effective voting rules and voting weights in voting bodies. Voting in CEN is generally single majority (of those voting), every member (organisation) having one vote. If more than 50% of the votes are in favour of a given proposal, it is approved. Article 6.1.1. of the statutes of CEN (2002, 2004) states that [i]n all cases where a decision is required, every effort shall be made [by the national representatives] to reach unanimity. However, according to article 6.1.4., in four cases there is weighted voting: a) formal approval of European norms (EN) or documents of harmonization (HD), 6 b) formal approval of technical specifications (TS), c) any initiation of a work item to become an EN or TS within a CEN technical committee, or d) approving B-Differentials (for HDs only). According to the statutes of CEN (2002, 2004), article 6.2.1., the threshold level for weighted voting is, [...] 71% or more of the weighted votes cast (abstentions not counted) [...] and, according to 6.2.2. [i]f the proposal is not adopted in accordance with 6.2.1, the votes of the members from EEA [European Economic Area] countries shall be counted separately. The proposal shall be adopted, if 71% or more of the weighted votes cast by these countries (abstentions not counted) are in favour [...] If a standards proposal is adopted, the members have to implement the standard and withdraw conflicting standards. A standard is non-mandatory for EFTA countries, if they vote against the standard. But it is mandatory for the club of all EEA countries and the EFTA countries in favour of the standard. In the remainder of this study, this rather special case is not taken into consideration. Article 6.1.3. states, among other things, that abstentions shall not be counted as a vote. Of course, this makes computations of voting power messy, since it could happen that the power could change due to an abstention. See e.g. Braham and Steffen (2002). In the calculations below, abstentions will not be taken care of in order to not further complicate the results. Also, the question of agenda setting will be disregarded. In CEN draft standards (proposals) are prepared by technical (sub)committees. Since a 6 HDs are special cases of ENs.

J. Gröndahl: Power and Standardisation 335 voting decision in a committee is for or against a proposed standard, the group that creates the proposal has considerable power concerning the outcome. But its power is limited by the fact that if it does not come up with a proposal that the voting body finds acceptable, they might be replaced by another group or be forced to revise their draft. Therefore the proposal is taken as given. Furthermore, only voting for or against a single standard is considered. Decisions about the selection of one out of two or several possible standards is not considered, since in CEN this selection is done in the technical committees while preparing the corresponding draft standard. 5. Power indices In this study the concept of a priori voting power is applied. This implies that only the weighted votes per country and the decision rule are considered when measuring power. Ex post power, which would include measuring the expected outcome of a specific decision process this involves assumptions concerning players preferences, derived from previous decisions, opinion polls, etc. is not taken into account. According to Holler and Widgrén (unfinished manuscript) the a priori analysis is justified, since neither the preferences of the players nor the state of the world at any given time are known, when the decision rules of a voting body are determined. There is a wide selection of power-indices to measure a priori voting power. The qualities of the different power indices that will be applied are discussed shortly. The Shapley-Shubik Index (SSI; Shapley and Shubik, 1954) gives an answer to the question, what player i s share of the coalition value vs ( ) of a coalition S, consisting of s members, voting in favour of the proposal will be, given that all permutations (i.e. orderings) of voters are equally possible. With the SSI the pivotal player i entering subcoalition S\{} i, receives vs ( ) vs ( \{ i }) the value that player i adds to subcoalition S\{} i. This value is multiplied by the number of possible orderings, ( s 1)!( n s )!, as a share of all possible n! orderings. The (absolute) Banzhaf Index (BI; Banzhaf, 1965, 1968 and Coleman, 1971) answers the question, in how many of the possible coalitions player i has a swing. A swing is the ability to turn a winning coalition S into a losing coalition S\{} i by leaving it (and vice versa). With the BI the critical player i who has a swing in S receives the additional coalition value i contributed. The normalized version of the BI (NBI), yields the relative number of swings of player i. As a consequence of the public good property of standards, every member of the voting body has to receive the entire coalition value, if a

336 Homo Oeconomicus 23(3/4) standard has been decided, since no member can be excluded from its use and application. This makes free riding, i.e. consuming the public good without contributing, possible. In order to make free riding unfavourable, coalitions containing possible free riders should not be attributed power. This can be taken care of, when measuring power, if only strict minimum winning coalitions, also called decisive sets, are accounted for. A decisive set is a winning coalition that contains no dummies or surplus players. That is if S W, then S\{i} W for all i in S. The set Mv ( ) is the set of all decisive sets of N. The Public Good Index (PGI; Holler, 1982 and Holler and Packel, 1983) is appropriate for measuring power in the case of public goods. It regards only decisive sets and every member of the coalition that is critical for the acceptance of the standard is ascribed to the entire additional coalition value. Note, that if only decisive sets would be considered with the NBI, it would yield the same value as the PGI, as every member of a decisive set is a swinger and thus receives the coalition value. With the SSI, the pivotal (i.e. critical) player receives the marginal coalition value. It is often argued, that the public good index is not appropriate, since it does not satisfy the axiom of local monotonicity. 7 Local monotonicity means that power should be monotonic in voting weights more voting weight should not imply less power. But if we want to assure local monotonicity in institutional design, should we use an index for the construction of the institution that always yields monotonic results? It seems to be more reasonable to use an index that does not yield monotonic results in all cases and create rules and weights that yield monotonic results even for this index. Furthermore, the Coleman collectivity power index (CC; Coleman, 1971) will be calculated. Unlike the indices introduced above, it does not measure the power of single players. Instead it expresses the ability of a collectivity (committee) to act by measuring the number of all possible winning coalitions over the number of all possible coalitions, providing the probability that a winning coalition will form. This measure is also called passage probability. It expresses the probability that any random proposal will pass behind a veil of ignorance (Baldwin and Widgrén, 2004). It enables an overall comparison of different voting games with reference to the possibility to accept a proposal, without referring to the power of the members. 7 For a reply to this critique see Holler and Napel (2005). They argue that if the results of a specific game violate local monotonicity this should give us some information about this game.

J. Gröndahl: Power and Standardisation 337 6. Power to standardise in CEN In the following the distribution of power and foremost the changes in this distribution is analysed with respect to the latest revisions of 2002 and 2004 in decision rule and voting weights in CEN. All indices introduced in the preceding chapter are applied. We will see that the outcome hinges critically on the applied power index. The less interesting case of single majority voting with every member having one vote is neglected, since this distribution of votes yields the same power for every member of CEN. In the case of weighted voting the member states have the voting weights and the threshold level (before and after the 2002 changes and after the 2004 changes) displayed in table 1 in the appendix. Concerning the changes in 2002 two countries, Austria and Sweden, suffered a decrease in voting weight, 4 instead of 5 votes each. Supposedly this happened because their population is considerably less than that of other 5-vote-countries. In contrast to EFTA country Switzerland, Austria and Sweden already accepted a decrease in weights for the EU Council of Ministers. In 2004 changes of voting weights were introduced to take care of the possible EU enlargement. All voting weights were changed according to the Nice treaty. The result was that all countries gained votes in absolute terms. However, some members of CEN (e.g., Switzerland) lost votes in relative terms, compared to other 5-vote-countries, while other countries gained relative votes (e.g. The Netherlands). With some exceptions these changes brought about the identical voting weights for countries with similar sizes of population. All incumbent countries, except for Spain, lost relative votes in the 2002 and 2004 changes due to the entry of new member states. Spain and the 2002 entrants Czech Republic, Hungary, and Slovakia gained relative votes in the 2004 changes. The power implications of these changes according to the different indices are displayed in table 2 in the appendix. The ability of CEN to act, as measured by CC, decreased only slightly as a result of the 2002 changes in voting weights. Before the 2002 changes, 48.22 percent of the possible coalitions were winning, while after these changes only 45.57 percent were winning. But CEN s ability to act decreased substantially due to the 2002 increase in the decision rule from 51 to 71 percent, which reduced the percentage of winning coalitions to a mere 4.32. This is perfectly in line with CEN s efforts towards unanimity. The new (2004) voting weights lowered CC further, but not considerable: 3.57 percent of the possible coalitions are winning. Taking SSI as power measure, all incumbent countries lost power in the 2002 changes of CEN voting weights. For some countries part of the losses due to new members was compensated by gains due to the changes in the

338 Homo Oeconomicus 23(3/4) decision rule. Austria, Sweden, and Iceland lost considerably more power in CEN than the other incumbent countries, albeit Iceland s voting weight remained unchanged. In the realm of the 2004 changes all countries lost power, except for Spain, Czech Republic, Hungary, and Slovakia, who gained considerably. Concerning the incumbent countries and all changes (2002 and 2004), Spain lost least power. Larger countries (with more than 5 votes before the 2002 changes), The Netherlands and Iceland lost less power than smaller (2- to 5-vote-) countries. The results for the BI show that in the wake of the 2002 changes every member country lost power due to the considerable increase in the threshold level. Due to the new members only Austria, Sweden and Iceland lost considerable power (although Iceland s voting weight was not altered). Concerning the 2004 changes, only Czech Republic and Hungary gained power. Considering all changes, all incumbent countries lost about the same amount of power in relative terms. Applying NBI to both 2002 changes, all countries except for Luxembourg lost power. Austria and Sweden lost most, followed by the big (10- vote) countries, as well as Iceland, and Spain. The change in the decision rule implied an increase in power for all except the larger countries (with more than 5 votes each). Iceland s power increased most, which partly compensated for their considerable decrease in power due to the new members. Concerning the 2004 changes, as with SSI, Spain, Czech Republic, Hungary, and Slovakia gained power (Czech Republic and Hungary considerably), while all others lost power. Again, The Netherlands (who gained a vote and thus naturally lost less power), Iceland and larger incumbent countries lost less power (Spain lost considerably less) than smaller countries. With the PGI, as a result of the 2002 change of the decision rule, all countries with at least 5 votes (after the 2002 changes) gained power, while all others lost power. The power of 5-vote-countries remained approximately constant. Needless to mention, that concerning the new members all countries lost power. However, Luxembourg s power remained nearly constant. If we consider both 2002 changes, only the big 10-vote-countries France, Germany, Italy, and The United Kingdom gained power. Concerning the 2004 change the only country gaining power was Iceland. Considering all changes, the big countries and Iceland lost least power. These results show that the effects of a change in the voting weights and voting rule differ significantly with the applied power index. The power of incumbent members could be expected to decline on average, when new members were introduced into the game in 2002. Since they lost 20 percent of their votes, Austria and Sweden could be expected to lose significantly more power than the other incumbent member states. According to the private good indices, Austria and Sweden lost most

J. Gröndahl: Power and Standardisation 339 power in relative terms. However, according to PGI they lost less power than Denmark, Finland, Ireland, and Norway. Although Iceland s voting weight was not altered, it lost nearly as much power in relative terms as Austria and Sweden, according to all indices applied. Luxembourg lost less power than the other members according to the private good indices (even gained power according to NBI) and lost least power except for the larger countries according to PGI. These results show that the intention to decrease some countries power might well lead to decreasing another country s power. The changes in the power of Luxembourg and, foremost, Iceland were supposedly unintended, since their voting weights have not been altered in the 2002 changes. Taking into account the public good property of standards, the changes in possible decisive sets that were brought about by the changes in voting weights and voting rule, implied the probably most striking result. The non-monotonicity in voting weights that gave Spain, Denmark, Finland, Ireland, and Norway more power than France, Germany, Italy, and The United Kingdom, disappeared. With the old (pre-2002) voting weights and decision rule, the private good indices power ranking was significantly different from the PGI power ranking, where Spain had most power, followed by Denmark, Finland, Ireland, and Norway. After the 2004 changes, the ranking of countries according to voting weights corresponds to their power ranking according to all indices considered (SSI, BI, NBI, and PGI). Concerning fairness, if we apply correspondency of population shares and power shares as a fairness criterion (compare table 2), the results do also depend on the applied power index. Only The Netherlands approximately receive their fair share. According to the private good indices, especially according to SSI, Poland and Spain also receive approximately their fair share. The new voting rule and voting weights favour smaller countries (smaller than The Netherlands), which are generally over-represented, and yield a disadvantage for big countries (larger than Poland and Spain). Among the group of big countries, Germany s under-representation is most noticeable. According to PGI the over- and under-representation is most striking, since the power of member states is within a much smaller spectrum than the power according to private good indices and the population shares. 7. Conclusions It has been demonstrated how the power-concept of game theory and political science can be applied to standardisation bodies. The relevance of committee standardisation is not limited to public or private standardisa-

340 Homo Oeconomicus 23(3/4) tion organisations that develop international, national, or intra-industry standards. In the globalization process in formal standardisation the bias of decisions can be influenced in favour of the one that sets up the game. This is also of paramount importance from a European perspective, if the vested interests of all countries involved are to be regarded in a fair way, as the analysis of a priori power in CEN shows. If representation according to population share were the only relevant fairness criterion and voting power were the only criterion for assessing a countries power, the game in European standardisation has to be changed, if a fairer distribution of power should be accomplished. As it is, larger countries are under- and smaller countries are over-represented in European standardisation. The Netherlands are the dividing line from a public good perspective. The private good indices yield fairness as implied by the criterion of population shares for Spain, Poland and The Netherlands. Otherwise, the other relevant criteria must yield more power for larger countries than for smaller countries in order to correct for the unfair distribution of voting power. This could be the case, if we take into account agenda setting power in workgroups. Larger countries tend to have better founded standardisation bodies that should have more resources at their disposal for contributing to workgroups. Thus larger, or rather economically better equipped countries could compensate for less power in voting in CEN. Or the over-representation of smaller and the under-representation of larger countries are intended. The examples of Iceland and Luxembourg in the 2002 revisions show that the power implications of (changes in) voting weights and decision rule might not be in line with the intentions. Lowering the voting weights of Austria and Sweden when new members entered the game in 2002 lead to a reduction of Iceland s voting power of the same magnitude, while Luxembourg lost less power than the average incumbent member. Thus the resulting implications for the power distribution within the body have to be taken into consideration very carefully. According to the private good indices Spain is the overall winner from the changes considered, since Spain lost less power than all other countries. This result is contrasted by the results from a public good perspective, with big and very small countries losing significantly less power than others. Since different power indices yield different distributions of power, due to their different perspective on the goods character, we may argue that the public good property of standards should be taken into account. The numerical results show that adopting voting rules and weights from another institution does not necessarily yield the same results, if the relevant context public or private good perspective differs and the difference is taken into consideration.

J. Gröndahl: Power and Standardisation 341 Applying the Coleman collectivity power index the increase in the decision rule in CEN is working in favour of the desired effort to unanimity at the cost of a considerable decrease of the ability to act. The substantial decrease of CEN s ability to act, as measured by A, might lead to kind of rubber-stamping. Quality of CEN standards might decrease, leading to a loss of reputation and, in turn, of revenue. Countries that feel under-represented might try to increase their influence on decision making in technical committees and working groups. This should increase the agenda setting power of working groups at the expense of the power of the actual voting decision. Furthermore, the decrease in the ability to act might shift standardisation activities out of CEN towards independent standardisation bodies, consortia, and the like. As a result CEN could move further towards rubber-stamping and facilitate outsourcing of standardisation activities via fast track processes such as the CEN workshop agreement. Acknowledgements The idea of this article derives from an unfinished manuscript by Manfred J. Holler and Mika Widgrèn, The Power to Standardize with Applications to the European Union. Earlier versions of this paper have been presented at the Adam Smith Seminar at the University of Hamburg, the EPCS-conference 2004 and the EURAS-conference 2005. I am indebted to Matthew Braham, Heide Coenen, Ruxandra Haradau, Manfred J. Holler, Tobias Langenberg, Marc van Wegberg and Mika Widgrèn for valuable insights and fruitful discussions. Of course I claim responsibility for all remaining mistakes.

342 Homo Oeconomicus 23(3/4) Appendix Table 1 Voting weights and decision rules in CEN before (old) and after the 2002 changes (intermediate [int]), after the 2004 changes (new), and absolute and relative population of the member states Country Votes (old) Share (%) (old) Votes (int) Share (%) (int) Votes (new) Share (%) (new) Pop. (mio.) Pop. (%) France 10 10.20 10 9.43 29 8.50 58.966 12.75 Germany 10 10.20 10 9.43 29 8.50 82.038 17.74 Italy 10 10.20 10 9.43 29 8.50 57.612 12.46 UK 10 10.20 10 9.43 29 8.50 59.247 12.81 Poland 27 7.92 38.667 8.36 Spain 8 8.16 8 7.55 27 7.92 39.394 8.52 Netherlands 5 5.10 5 4.72 13 3.81 15.760 3.41 Belgium 5 5.10 5 4.72 12 3.52 10.213 2.21 Greece 5 5.10 5 4.72 12 3.52 10.533 2.28 Czech Rep. 3 2.83 12 3.52 10.290 2.22 Portugal 5 5.10 5 4.72 12 3.52 9.980 2.16 Hungary 3 2.83 12 3.52 10.092 2.18 Switzerland 5 5.10 5 4.72 10 2.93 7.261 1.57 Austria 5 5.10 4 3.77 10 2.93 8.082 1.75 Sweden 5 5.10 4 3.77 10 2.93 8.854 1.91 Denmark 3 3.06 3 2.83 7 2.05 5.313 1.15 Finland 3 3.06 3 2.83 7 2.05 5.160 1.12 Ireland 3 3.06 3 2.83 7 2.05 3.744 0.81 Norway 3 3.06 3 2.83 7 2.05 4.525 0.98 Lithuania 7 2.05 3.701 0.80 Slovakia 2 1.89 7 2.05 5.393 1.17 Luxembourg 2 2.04 2 1.89 4 1.17 0.429 0.09 Cyprus 4 1.17 0.752 0.16 Estonia 4 1.17 1.446 0.31 Latvia 4 1.17 2.439 0.53 Slovenia 4 1.17 1.978 0.43 Malta 2 1.89 3 0.88 0.379 0.08 Iceland 1 1.02 1 0.94 3 0.88 0.290 0.06 Computations using Powerslave (Pajala et.al., 2002).

J. Gröndahl: Power and Standardisation 343 Table 2

344 Homo Oeconomicus 23(3/4) References Baldwin, R. and Widgrén, M. (2004), Winners and losers under various dual majority rules for the EU s Council of Ministers, CEPS Policy Brief 50. Banzhaf, J.F.III (1965), Weighted Voting Doesn t Work: A Mathematical Analysis, Rutgers Law Review 19: 317 343. Banzhaf, J.F.III (1968), One Man, 3.312 Votes: A Mathematical Analysis of the Electoral College, Villanova Law Review 13: 304 332. Braham, M. and Steffen, F. (2002), Voting power in games with abstentions, in: Holler et al (eds.), Power and Fairness, Jahrbuch für Neue Politische Ökonomie 20: 333 348. CEN (2002), CEN/CENELEC Internal Regulations, Part 2: Common Rules for Standardization Work (2002). CEN-Homepage in December 2003, http://cenorm.be. CEN (2004), CEN/CENELEC Internal Regulations, Part 2: Common Rules for Standardization Work (2004). CEN-Homepage in March 2004, http://cenorm.be. CEN (2005), Information about CEN on the CEN-Homepage. http://cenorm.be/cenorm/index.htm, http://cenorm.be/cenorm/aboutus/index.asp. CEN Statistics (2004). http://cenorm.be/cenorm/aboutus/information/statistics/index.asp. Coleman, J. (1971), Control of Collectivities and the Power of a Collectivity to Act, in: B. Liebermann (ed.), Social Choice, Gordon and Breach: New York. Economides, N. (1996), Network Externalities, Complementarities, and Invitations to Enter, European Journal of Political Economy 12: 211 232. European Commission (2003), http://europa.eu.int/comm/enterprise/standards_policy/document/guidelines/ joint_news_release.pdf. Farrell, J. and Saloner, G. (1985), Standardization, Compatibility, and Innovation, The RAND Journal of Economics 16: 70 83. Farrell, J. and Saloner, G. (1986), Installed Base and Compatibility: Innovation, Product Preannouncements, and Predation, American Economic Review 76: 940 955. Farrell, J. and Saloner, G. (1988), Coordination through Committees and Markets, The RAND Journal of Economics 19: 235 252. Felsenthal, D. S. and Machover, M. (2001), The Treaty of Nice and qualified majority voting, Social Choice and Welfare 19: 465 483. Goerke, L. and Holler, M.J. (1995), Voting on Standardization, Public Choice 83: 337 351. Holler, M.J. (1982), Forming Coalitions and Measuring Voting Power, Political Studies 30: 262 271. Holler, M.J. and Napel, S. (2005), Local monotonicity of power: Axiom or just a property? Quality and Quantity 38: 637 647. Holler, M.J. and Packel, E.W. (1983), Power, Luck and the Right Index, Zeitschrift für Nationalökonomie (Journal of Economics) 43: 21 29. Holler, M.J. and Widgrén, M. (unfinished manuscript), The Power to Standardize

J. Gröndahl: Power and Standardisation 345 with Applications to the European Union. Katz, M.L. and Shapiro, C. (1985), Network Externalities, Competition, and Compatibility, American Economic Review 75: 424 440. Katz, M.L. and Shapiro, C. (1986), Technology Adoption in the presence of Network Externalities, Journal of Political Economy 94: 822 841. Leech, D. (2002), Designing the voting system for the Council of the European Union, Public Choice 113: 437 464. Pajala, A., Meskanen, T. and Kause, T. (2002), POWERSLAVE Power Index Calculator, http://powerslave.val.utu.fi/. Shapley, L.S. and Shubik, M. (1954), A Method for Evaluating the Distribution of Power in a Committee System, American Political Science Review 48: 787 792.