DISCLOSURE OF INDIVIDUALIZED EXECUTIVE COMPENSATION FIGURES: AN EMPIRICAL ANALYSIS OF COMPLIANCE WITH THE GERMAN CORPORATE GOVERNANCE CODE

Transcription

1 Corporate Ownershp & Control / Volume 4, Issue 1, Fall 2006 DISCLOSURE OF INDIVIDUALIZED EXECUTIVE COMPENSATION FIGURES: AN EMPIRICAL ANALYSIS OF COMPLIANCE WITH THE GERMAN CORPORATE GOVERNANCE CODE Joerg Rchard Werner*, Jochen Zmmermann** Abstract From 2002 to 2005, the German Corporate Governance Code advsed that stock lsted companes should (voluntarly) dsclose ndvdualzed executve compensaton fgures. In a sample of bg publcly traded German companes, we examne whch determnants drve frms to comply wth that soft law requrement. Usng a probt model, we consder 15 explanatory varables. We fnd that block-holdngs, average executve remuneraton, book-to-market rato and the percentage of unon representatves n the supervsory board sgnfcantly decrease the lkelhood of dsclosures on ndvdualzed executve compensaton numbers. Frm sze, the absolute number of supervsory board members and the presence of takeover actvty turn out as havng a sgnfcant postve nfluence on the dsclosure behavor. Addtonally, we fnd that t was less lkely n 2002 that ndvdualzed remuneraton fgures were publshed, ndcatng that Code recommendatons are consdered as more bndng than Code suggestons. Keywords: corporae governance, German governance code, board * Correspondng author. Unversty of Bremen, Department of Accountng and Control. Hochschulrng 4, Bremen, phone , fax , e-mal jrwerner@un-bremen.de. **Unversty of Bremen, Department of Accountng and Control. Hochschulrng 4, Bremen. Executve Compensaton and The German Governance Code Regulaton concernng dsclosures of executve compensaton fgures has undergone a sgnfcant change n Germany durng the last decade. Tradtonally, Art. 285 No. 9 of the German Commercal Law (Handelsgesetzbuch, HGB) demanded dsclosures about the total amount of fxed and varable compensaton of the management board members. Accordng to Art. 286 (4) HGB, such numbers were not allowed to be dsclosed n cases where conclusons about the remuneraton of sngle board members were possble. Ths regulaton was explaned by the legslator s fears that such dsclosures would volate prvacy protecton, a base rght deductble from the German consttuton. However, n 2005, ths approach changed wth an amendment that made dsclosures of ndvdualzed executve compensaton fgures a legal requrement for fnancal years begnnng after December 31 th 2005 (German Parlament 2005). Ths decson was caused by a longstandng publc debate about the amount of management compensaton, whch was frequently perceved as exaggerated. Executve compensaton made headlnes n the mass meda 106 especally n context of the hostle takeover of Mannesmann by Vodafone n 2000, and n a followng lawsut courts had to judge whether settlements pad to the former company offcals were approprate. 1 Ths made (some) compensaton practces known to the general publc. Wth the German economy facng an overall downward trend, the mpresson spread that a lot of board members behave n a self-servng fashon whle at the same tme they lay off ther staff. However, the amendment of 2005 has a precursor: From 2002 to 2005, the German Corporate Governance Code (GCGC) advsed companes to dsclose ndvdualzed executve compensaton numbers on a voluntary bass. The Code s not law tself but Artcle 161 of the Stock Corporaton Act forms ts legal base (Netsch 2005: 371). The GCGC ams to defne nternatonally recognzed standards for good and responsble governance that are formulated as recommendatons or suggestons (Kühne & Fuss 2003: 229). 2 1 See e.g. Germany s fat cats on tral (Economst, Sep 25 th 2003) and Corporate Germany on tral (Economst, Jan 23 rd 2004). 2 The actual and prevous versons of the Code can be found on

2 Corporate Ownershp & Control / Volume 4, Issue 1, Fall 2006 Companes have to dsclose annually any devaton from the Code s recommendatons n a Declaraton of Conformty. Ths s sometmes labeled as a comply or explan approach (Cromme 2005: 363). The frst Declaratons had to be publshed for the fnancal year The Code s suggestons are weaker than recommendatons; they do not have to be publshed. Dsclosures of ndvdualzed executve compensaton fgures ntally were a suggeston n the frst verson of the Code before t turned nto the followng recommendaton n 2003: Compensaton of the members of the Management Board shall be reported n the Notes of the Consoldated Fnancal Statements subdvded accordng to fxed, performance-related and longterm ncentve components. The fgures shall be ndvdualzed. (GCGC 2005, Paragraph 4.2.4). Snce 2006, such dsclosures are mandatory. Voluntary remuneraton dsclosures correspondng to the GCGC are observable for the fnancal years 2002, 2003 and In our analyss we rase the queston whch determnants or frm characterstcs drve companes (not) to comply wth the Code s advce to publsh ndvdualzed executve compensaton fgures. The remander of the paper s organzed as follows: In chapter 2, we outlne our hypotheses and specfy the model for an emprcal analyss. In chapter 3, the sample and data sources are descrbed. In chapter 4, the man fndngs of the study wll be dscussed. The artcle ends wth a summary and dscusson n chapter 5. Hypothess Development and Model Specfcaton In our analyss, we assess the nfluence of 15 ndependent varables on a company s dsclosure behavor wth respect to the GCGC s advce to publsh ndvdualzed executve compensaton fgures. The varables proxy for mportant frm characterstcs. The complance behavor of a partcular company s measured by the dummy varable COMPDISC ( compensaton dsclosure ) whch takes a value of 1 f a company dscloses ndvdualzed compensaton fgures and thus follows the Code s advce and zero otherwse. As the dependent varable s categorcal, we use a multvarate probt model for the analyss of the nfluence of the ndependent varables. It s not that obvous whch frm characterstcs mght have a strong bearng on the dsclosure behavor. Thus, we use a relatvely hgh number of ndependent varables selected accordng to the framework outlned by Börsch-Supan/Köke (2002). The authors argue that mssng explanatory varables are a general problem of emprcal studes on corporate governance and suggest the complete use 3 Fnancal reports for 2005 are not yet avalable for (most of) the sample frms. of specfc nformaton concernng ownershp, fnancal and board structure, competton stuaton and takeover actvtes. Table 1 provdes an overvew over the varables used n our analyss and the predcted sgns of the nfluence on complance wth the Code s advce. To account for ownershp structure, we frst consder block-holdngs (blckhol). Block-holdngs by non-bankng companes are measured as the fracton of common stock held by the three largest shareholders. We assume that the presence of blockholdngs decreases the lkelhood of complance as the Code s an nstrument ntended to strengthen external (.e., market) control. Block-holders are, n fact, nsders because they have access to a company s supervsory board (Hackethal et al. 2005: 398). The code requres dsclosure of ndvdual compensaton for management and supervsory board members. Wth the strong captal connecton of bg German companes t s lkely that supervsory board members of one frm are management board members n another. (Schllng 2001: 149) Ths reduces the ncentves of block holder representatves to press for ndvdual dsclosures of management board remuneraton. Inversely, we hypothesze that the same arguments lead to an ncreasng lkelhood of Code complance f the level of block-holdngs s low (and market pressure hgh). In the German nsder corporate governance system, unversal banks have tradtonally played an mportant role (Hackethal et al. 2005: 399 ff). Thus, we use bank ownershp (bnkown) as a proxy for montorng actvtes of unversal banks (bank ownershp s defned as a percentage of common stock owned by unversal banks). Insder control systems typcally provde large control benefts for the major shareholders who are n our case banks (see Hart 1995, for some crtcal comments). Bank domnance has consttuted a stable and effectve element of corporate governance. Gorton/Schmd (2000), among others, provde evdence of the postve nfluence of bank block holdngs on the performance of German companes. However, the relatonshp between bank ownershp and the degree of complance wth the Code s not straghtforward. On the one hand, t s possble that the abovementoned block-holder arguments apply as well. On the other hand, banks mght have nterests dfferent than those of (ndustry) block-holders. However, Edwards/Nbler (2000) argue that bank nfluence on company polces should not be over-estmated. We therefore make no predctons on the expected sgn. We do not have data to drectly observe the fracton of a company s shares that are held by foregners. However, f a company s shares are lsted at foregn exchanges, ths ndcates that ths company tres to attract foregn nvestors. If dsclosures about ndvdualzed executve compensaton fgures are nternatonal best practce (as the Code clams), the lkelhood of dsclosng these numbers should be 107

3 Corporate Ownershp & Control / Volume 4, Issue 1, Fall 2006 hgher f a company s lsted at least at one foregn stock exchange. The dummy varable xlst takes a value of 1 for a foregn lstng, and we expect a postve sgn for ths varable. Table 1. Governance Issues and Proxy Varables Varable Abbrevaton Measurement Pred. Sgn Block-holdngs blckhol Fracton of common stock held by the three largest shareholders Bank Ownershp bnkown Percentage of common stock owned by unversal banks? Crosslstngs xlst Dummy varable takng a value of 1 f a company s cross-lsted at one or more foregn exchanges (and 0 otherwse) Leverage levrge Rato between total debt and total assets + Free Cash Flow fcfps Free cash flow per share (Compustat tem) + Book-to-Market Rato btm Rato between book and market values of equty? Board Sze nbm Total number of the Management Board members? Supervsory Board Sze nsbm Total number of the Supervsory Board members? + Unon Functonares n Supervsory Board umpctg Percentage of unon functonares n the supervsory board + Executve Compensaton avgep Average Executve Compensaton per member of the Management Board Frm Sze logbs Natural logarthm of total assets + Proftablty profblty Return on Equty? Takeover Announcements tkover Takeover-relevant legal announcements pursuant to the WpÜG (dummy varable) Dummy varable for year 2002 y2002 Dummy varable takng a value of 1 for 2002 and 0 otherwse + Dummy varable for year 2003 y2003 Dummy varable takng a value of 1 for 2003 and 0 otherwse? Leverage depcts a company s fnancal structure. We defne leverage (levrge) as the rato between total debt (numerator) and total assets (denomnator). Debt can also be a lever of control (Jensen 1986; Cable 1985). Debt holders are expected to montor frm performance more actvely f ther share n the captal structure ncreases (e.g., Hutchnson/Gul 2004). Fnancal leverage can thus be used as a proxy for external control through debt holders. We expect a postve sgn for ths. Addtonally, we use free cash flow per share (fcfps) and the book-to-market rato (btm) as proxes for a company s fnancal structure. Wth free cash flow ncreasng, shareholders face a hgher rsk of management proflgacy, whch should lead to tghter controls and therefore pressures for addtonal dsclosure. The btm stands for growth opportuntes. Growth opportuntes may on the one hand lead to more controls whether there are exploted, on the other hand optmstc shareholders may slacken n ther wshes for dsclosure. We therefore do not predct a sgn for ths varable. Management board sze (nbm), supervsory board sze (nsbm), the percentage of unon representatves n the supervsory board (umpctg), and average remuneraton of the management board members (avgep) are used to account for the board structure. The absolute number of varables consdered s relatvely hgh here, to cover the peculartes of the German two-ter Board Structure. On one hand, the (large) sze of the supervsory board s often consdered as a problem of the German corporate governance system (Wooldrdge/Panner 2005: 230). On the other hand, a large number of supervsory board members mght decrease the nfluence of partcular nterest groups lke block-holders, banks or unonsts, whch mght work n favor for the control actvtes of mnorty shareholders. Remuneraton of executves (but also of supervsory board members) plays an mportant role n provdng ncentves and attenuatng prncpal-agent problems. Evdently, hgher average executve compensaton ncreases the lkelhood that payments are hgher than approprate and nclude large CEO remuneraton, whch, n turn, nduces the recevers of those payments not to gve full partculars on them. Thus, we hypothesze a negatve nfluence on complance. Frm sze (logbs) stands as an explanatory factor for a range of governance ssues,.e., publc montorng, poltcal costs or the ablty to draw up market barrers. Frm sze s quantfed here as the natural logarthm of total assets. We expect a postve nfluence of frm sze on complance 108

4 Corporate Ownershp & Control / Volume 4, Issue 1, Fall 2006 behavor because of the ncreasng publc montorng and poltcal costs. Ths becomes obvous f one recalls that CEO remuneraton of DAX30- consttuents s frequently publshed n German busness magaznes. Smaller frms (.e., Non-Dax30 frms) are not of nterest there. Fnally, a measure for takeover actvtes s needed. In Germany, takeover actvtes are assocated wth mandatory announcements pursuant to the Securtes and Takeover Act (Wertpapererwerbs- und Übernahmegesetz, WpÜG). Ths law regulates offers to purchase securtes that have been ssued by a lsted company that s target of takeover actvtes. Any announcement concernng a partcular target company wll sgnfy that at least for ths company a market for control exsts. To proxy for takeover actvtes, we use a dummy varable (tkover) takng a value of 1 f there s a takeover relevant legal announcements pursuant to the WpÜG. We presume a postve relatonshp between takeover actvtes and the wllngness to publsh remuneraton partculartes, because managers facng a takeover am to sgnal that the company s well-governed and that benefts from a takeover are low for shareholders. We use two addtonal dummy varables for the respectve years. The frst one (y2002) takes a value of 1 for the fnancal year 2002 (and 0 otherwse), the second one a value of 1 for the fnancal year 2003 (and 0 otherwse). The dummy varable for year 2002 s of partcular nterest, because the Code s advse to dsclose ndvdualzed numbers was a suggeston durng ths very year (whle t was a recommendaton n the followng years). As recommendatons are more bndng than suggestons, we suppose that the lkelhood of complance was lower n 2002 than n 2003 and We generally perform two-taled tests of the zero hypotheses that the varables n total (.e., the model) and each sngle varable as well, provde sgnfcant explanatory power on the value the dependent varable takes. The model s as follows: COMPDISC = α + α y α y α blckhol + α bnkown + (1) α profblty + α xlst + α tkover + α fcfps + α btm + ξ α levrge + α nbm + α nsbm + α umpctg + α avgp + α log bs We control for a possble presence of heteroscedastcty by usng Whte-adjusted standard errors n the multvarate analyss (Whte 1980). Data sources and Sample descrpton In our study, we analyze German companes lsted n DAX 30 and MDAX n the begnnng of These are the bggest German companes. Our ntal sample ncluded 80 companes and 240 frm years. As outlned n Table 2, 16 bank, fnancal servce and nsurance companes were excluded from the sample because of ther dfferent fnancal structure. Two frms n the legal form of a German partnershp lmted by shares were excluded as well, because they had a dfferent legal and corporate governance structure. Companes not domcled n Germany are also excluded from the sample, because they are not (drectly) subject to German regulaton and thus mght follow dfferent corporate governance practces. We fnally arrve at a test sample of 60 companes wth observatons for the years 2002, 2003 and That makes, n total, 180 frm years. Fnancal data s obtaned from the HOPPENSTEDT BILANZDATENBANK and COMPUSTAT databases. Declaratons of conformty, as publshed by the companes, were analyzed qualtatvely. The ownershp structure of the sample companes was obtaned from the HOPPENSTEDT AKTIENFUEHRER. Other stock nformaton s taken from the DATASTREAM databank. Informaton about takeover actvtes s obtaned from the webste of the BUNDESANSTALT FÜR FINANZDIENSTLEISTUNGSAUFSICHT (BAFIN) Table 2. Sample Selecton Item Frms Frm years DAX companes MDAX companes = Sum Insurance companes Banks Fnancal servce companes Legal form of partnershp lmted by shares (KGaA) Foregn domcled companes 2 6 = Total number of frms / frm years

5 Corporate Ownershp & Control / Volume 4, Issue 1, Fall 2006 Dscusson of the Man Fndngs Companes n our sample are very dverse n many respects, e.g. proftablty, ownershp structure, or number of board and supervsory board members. The standard devaton of bank ownershp, for example, s more than twce the mean. Ths s true for free cash flow per share. Descrptve statstcs of the varables can be found n Table 3. Table 4 shows that eght of the ndependent varables have a statstcally sgnfcant nfluence on the dsclosure of ndvdualzed executve compensaton fgures. Whle bank ownershp and cross-lstngs do not seem to have sgnfcant nfluence on the complance behavor, t turns out that the lkelhood of dsclosng ndvdualzed executve compensaton fgures decreases consderably wth ncreasng block-holdngs. Ths (expected) fndng can theoretcally be justfed by two arguments: Frst, block-holders are typcally represented n the supervsory board. Thus, they do not need dsclosures to gather nformaton about a company or, n partcular, about executve compensaton. As dsclosures on ndvdualzed executve compensaton fgures typcally go n hand wth dsclosures on ndvdualzed supervsory board member compensaton, they ndeed have negatve ncentves n forcng managers to follow the code. Second, block-holdngs are negatvely correlated wth free float and thus wth captal market pressure. In contrast to block-holders, mnorty shareholders do not have the access to supervsory board and to nternal nformaton. Thus, they have to rely on the dsclosed nformaton. Therefore, t does not surprse that lower block-holdngs,.e., hgher free float, ncreases the lkelhood of complance wth the suggeston/recommendaton to dsclose ndvdualzed executve compensaton fgures Table 3. Descrptve Statstcs Varable Observatons Mean Std. Dev. Mnmum Maxmum compdsc y y blckhol bnkown levrge nbm nsbm umpctg avgep logbs proft xlst tkover fcfps btm Table 4. Probt Results Varable Coeffcent Rob. Std. Err. z-value Prob (P>z) y2002 * y blckhol *** bnkown levrge nbm nsbm * umpctg ** avgep * logbs * proft xlst tkover * fcfps btm ** Intercept n = 180; Wald-Ch-Sqr. = 36.99***; Pseudo-R-Sqr = *, **, *** ndcate the level of sgnfcance (10%, 5% and 1%). 110

6 Corporate Ownershp & Control / Volume 4, Issue 1, Fall 2006 The coeffcent on the book-to-market-rato (btm) also turns out as beng negatve and sgnfcant whle the other proxes for fnancal structure do not have sgnfcant nfluence on the complance behavor. A possble nterpretaton of ths fndng s that frms wth hgher book-to-market rato have lower growth opportuntes, ncreasng the lkelhood of overnvestng surplus cash. Ths gves managers ncentves to sgnal that they do not use surplus cash to exproprate nvestors. Three of the varables used to proxy for board structure have a sgnfcant nfluence on the complance behavor. Frst, we fnd evdence that each addtonal supervsory board member ncreases the lkelhood of the dependent varable takng a value of 1.e., dsclosng ndvdualzed compensaton numbers. Ths can be explaned by an argument that an ncreasng total number of supervsory board members decreases the nfluence of partcular nterest groups. Independent supervsory board members, then, are more lkely to act n the nterest of mnorty shareholders. Second, we fnd that the lkelhood of dsclosng ndvdualzed remuneraton fgures decreases wth an ncreasng percentage of unon representatves n the supervsory board. Ths s surprsng only at a frst glance: As unonsts often complan about excessve compensaton for board members, one would expect that a larger nfluence of unon offcals n supervsory boards ncreases the lkelhood that companes do publsh ndvdualzed compensaton fgures. However, the actual nterrelatonshp between dsclosure behavor and unon representaton s the reverse. A possble explanaton for ths fndng s that unonsts mght use questons related to executve remuneraton n negotatons about other employee related ssues. Thus, they do not want to lose ths pressursng nstrument. The above-mentoned lawsut aganst former Mannesmann offcals fts wth our fndng: A former trade unon leader who was a member of the supervsory board, dd not exert any actons to prevent the payments n queston. The average remuneraton of board members tself has a negatve mpact on the lkelhood of ndvdualzed compensaton dsclosures. Ths s an appealng fndng: The more board members earn, the less they are wllng to publsh these numbers. However, there s a lmtaton to ths nterpretaton because of mssng data wth respect to the varance n board compensaton. For example, t mght be that the heght of CEO pay has an nfluence on the reportng behavor, but we cannot examne ths explctly because there s no possblty to obtan data on CEO remuneraton from those frms that dd not publsh ndvdualzed fgures. Frm sze also helps to explan the complance behavor. The coeffcent on the frm sze varable s sgnfcantly dfferent from zero and has a postve sgn, ndcatng that bgger companes are more lkely to dsclose ndvdualzed remuneraton fgures. Ths can be traced to the fact that there s greater publc montorng and nterest n bgger companes. If a company s a target of takeover actvtes, the lkelhood of ndvdualzed dsclosures also ncreases. Ths s not surprsng ether, because the management of a takeover target frm has ncentves to sgnal that the frm s well-governed. Ths mght ncrease the level of total complance wth the Code and n partcular wth the recommendaton of dsclosures about remuneraton, because not reportng these numbers mght lead the shareholders to conclude that the management s overpad and thus neffcent. Fnally, the coeffcent on the dummy varable for the year 2002 s, as expected, negatve and sgnfcantly dfferent from zero. Ths ndcates that the ndvdualzed executve compensaton fgures were less lkely to be dsclosed n 2002 than n 2003 or Ths allows us to draw the concluson that the Code s recommendatons are ndeed consdered as more bndng than ts suggestons. Summary Usng a probt model, we nvestgated the nfluence that 15 dfferent ndependent varables have on the complance level wth the GCGC advce to publcze ndvdualzed compensaton fgures. It turned out that the ndependent varables under consderaton sgnfcantly help to explan the complance behavor. We conclude that durng 2002 through 2004, complance wth the code was nformatve about mportant frm characterstcs and thus nterestng under a sgnalng perspectve. Whle ths questons the new and tougher regulaton at least to some extent, the legal amendment mght decrease agency costs for mnorty shareholders. As outlned, block-holders have an access to nternal nformaton. Obvously, they do not consder t as necessary to dsclose such nformaton to mnorty shareholders. Hence, the amendment strengthens the poston of mnorty shareholders as t provdes them wth more nformaton than they would otherwse dspose of. However, t s hard to predct whether ths advantage wll compensate for the negatve consequences of the tougher regulaton. References 1. Börsch-Supan, Axel/Köke, Jens (2002): An Appled Econometrcans Vew of Emprcal Cor-porate Governance Studes, n: German Economc Revew, Vol. 3 (2002), Cable, J. (1985): Captal Market Informaton and Industral Performance - The Role of West German Banks, n: Economc Journal, Vol. 95 (1985), Cromme, Gerhard (2005): Corporate Governance n Germany and the German 111

7 Corporate Ownershp & Control / Volume 4, Issue 1, Fall 2006 Corporate Governance Code, n: Corporate Governance An Internatonal Revew, Vol. 13, Issue 3, Edwards, Jeremy/Nbler, Marcus (2000): Corporate governance n Germany: the role of banks and ownershp concentraton, n: Economc Polcy, Vol. 15 (2000), Issue 31, German Parlament (2005): Beschlussempfeh lung und Bercht zum Entwurf enes Gesetzes über de Offenlegung der Vorstandsvergütungen (Vorstandsvergütungs-Offenlegungsgesetz, VorstOG), n: Bundestagsdrucksache 15/5860, pdf (Jan 18 th 2006). 6. GCGC (2005): German Corporate Governance Code, Verson as amended on June 02, 2005, (convenence translaton), lnk: corporate-governance-code.de/eng/kodex/1.html (Jan 18 th 2006). 7. Gorton, Gary/Schmd, Frank A. (2000): Unversal Bankng and the Performance of German Frms, n: Journal of Fnancal Economcs, Vol. 37 (2000), Hackethal, Andreas/Schmdt, Renhard H./Tyrell, Marcel (2005): Banks and German Corporate Governance: on the way to a captal market-based system?, n: Corporate Governance An Internatonal Revew, Vol. 13 (2005), No. 3, Hart, Olver (1995), Corporate Governance Some Theory and Implcatons, n: The Economc Journal, Vol. 105 (1995), Hutchnson, Maron/Gul, Ferdnand A. (2004): Investment opportunty set, corporate governance practces and frm performance, n: Journal of Corporate Fnance, Vol. 10 (2004), Iss. 4, Jensen, Mchael C. (1986): Agency Costs of Free Cash Flow, Corporate Fnance, and Takeovers, n: The Amercan Economc Revew, Vol. 76 (1986), Kühne, Eberhardt/Fuss, Josef (2003): Corporate Governance n Germany, n: Busness Law Revew, Vol. 24 (2003), No. 10, Netsch, Mchael (2005): Corporate Governance and Company Law Reform: a German perspectve, n: Corporate Governance: An Internatonal Revew, Vol. 13 (2005), Iss. 3, Schllng, Floran (2001): Corporate Governance n Germany: the move to shareholder value, n: Corporate Governance An Internatonal Revew, Vol. 9 (2001), No. 3, Whte, Halbert (1980): A Heteroskedastcty- Consstent Covarance Matrx Estmator and a Drect Test for Heteroskedastcty, n: Econometrca, Vol. 48 (1980), No. 4, Wooldrdge, Frank/Panner, Matthas (2005): The German Corporate Governance Code Status and Development, n: European Busness Law Revew, Vol. 16 (2005), No. 2,