Game hoarding in Europe: stock-price consequences of local bias?

Abstract

Local bias within a country and between countries is well established in the empirical literature. However, the underlying reasons are less well established. In a simple supply and demand framework, Hong, Kubik, and Stein (hereafter HKS) [2008. The only game in town: Stock-price consequences of local bias. Journal of Financial Economics 90, no. 1: 20–37.] find an ‘only-game-in-town’ effect in the USA – the stock price in a region decreases in the ratio of aggregate book value of listed firms to the aggregate personal income (‘RATIO’). We first replicate the HKS (2008) study using European data and find an opposite effect, a ‘game-hoarding’ effect. We then investigate the underlying factors of RATIO and find that after controlling for differences in origin of law, investor rights, corruption and Euro adoption, neither a game-hoarding effect nor an only-game-in-town effect is strongly supported in the European case. The results are important in understanding the concept of local bias in a cross-country framework.

Keywords:

• local bias
• only-game-in-town effect
• game-hoarding effect
• Europe

JEL Classification::

• G11
• G12

Acknowledgements

We are grateful for the comments and suggestions of two anonymous referees, Giovanna Zanotti, and participants at the FMA 2010 Annual Meeting and the 8th International Paris Finance Meeting.

Notes

The EU15 countries consist of the 15 ‘old’ EU countries (Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Sweden, Spain and U.K.). In 2004, 10 more countries (Cyprus, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Slovakia and Slovenia) joined the EU, and in 2007, two more countries (Romania and Bulgaria) joined the EU reaching the present number of 27 EU countries.

A dummy for cross-listing could be relevant in a European setting. Due to data constraints, we have only been able to obtain stock listing information from 2007, and because the European Commission changed the rules governing equity trade in 2007, this information will not be comparable to the previous years. However, in general, the stocks that do become cross-listed have a tendency to move more with the foreign countries’ prices of equity than prices in their home market (Baruch, Karolyi, and Lemmon 2007). Thus, the inclusion of stocks that are cross-listed will tend to bias the results against local bias. Furthermore, the stocks that are cross-listed tend to be stocks of large firms. In our robustness tests, we exclude large firms and find that this does not change our conclusions. Given the above, we feel confident that the exclusion of a dummy for cross-listing does not drive our results.

Sørensen et al. (2007) report the home country bias by providing percentage numbers of the equity bias in all but two of our EU15 sample countries for 2003. They are as follows: Austria 39%, Belgium 50%, Denmark 63%, Finland 65%, France 71%, Germany 54%, Greece 96%, Italy 57%, the Netherlands 37%, Portugal 68%, Spain 86%, Sweden 58% and the UK 68%. 0% represents no home country bias and 100% represents complete home country bias.

The distribution of the 22,407 firm-year observations over the years is 2001: 2340 (10%), 2002: 2,711 (12%), 2003: 2816 (13%), 2004: 3115 (14%), 2005: 3448 (15%), 2006: 3916 (17%), and 2007: 4061 (18%) adding up to 22,407 (100%) for the period of 2001–2007.

The distribution of the 4061 observations in 2007 across countries is Austria: 60 (1%), Belgium: 98 (2%), Denmark: 113 (3%), Finland: 111 (3%), France: 554 (14%), Germany: 631 (16%), Greece: 241 (6%), Ireland: 46 (1%), Italy: 213 (5%), Luxembourg: 9 (0%), the Netherlands: 95 (2%), Portugal: 44 (1%), Spain: 108 (3%), Sweden: 343 (8%) and the UK: 1395 (34%) adding up to 4061 (100%) for the 15 countries.

In Table 2, we find a positive but insignificant RATIO coefficient for 2007. The European Commission changed the rules governing equity trade in 2007 so that a stock exchange on its own initiative can decide to trade a firm's shares as opposed to the former situation where a listing abroad always happened on the (expensive) initiative of the firm. We rerun the regression excluding 2007 and find that the RATIO coefficients are statistically more significant in this restricted timeframe.

Austria: Erste Group AG; Belgium: Fortis Group SA; Denmark: A.P. Möller Maersk A/S; Finland: NOKIA OYJ; France: BNP PARIBAS; Germany: E-On AG; Greece: Public Power; Ireland: I Shares PLC; Italy: UNICREDIT SPA; the Netherlands: ING Group NV; Portugal: EDP SA; Spain: Banco Santander SA; Sweden: AstraZeneca; UK: Vodafone PLC.

In the HKS (2008) study, the RATIO variable is recalculated for each firm-year observation so that the numerator of RATIO omits the book value of the firm in question. This is done in order to ensure that especially observations from large firms do not drive a negative coefficient in a regression of log market-to-book against RATIO. HKS (2008) found this adjustment to be inconsequential. We indirectly address this concern in our sample by excluding the largest firm in each country.

Hardouvelis, Malliaropulos, and Priestley (2006) find that the EMU harmonization has made Euro countries more integrated leading up to the single currency implementation. They point out that this should lead to homogeneous valuation of equities in those counties. They show that the UK does not converge with the other countries because it did not adopt the euro. We restrict our sample to the 11 euro countries in our sample thus excluding Denmark, Sweden and the UK.

Data from World Federation of Exchanges on turnover velocity of domestic shares for the period 2001–2007 do not lend support to the argument that a low (high) RATIO is associated with a low (high) turnover velocity. As the structure of European exchanges has changed markedly in the period 2001–2007 through mergers and acquisitions, direct comparisons are not straightforward in all cases. The UK has the highest average RATIO of 0.86 (Table 1) but ‘only’ a turnover velocity of 113% (based on London SE). However, Italy and Germany with average RATIOs of 0.25 and 0.28 (Table 1) have turnover velocities of 147% and 151% (based on Borsa Italiana and Deutsche Börse, respectively). Also Spain with a RATIO of 0.31 (Table 1) has a high turnover velocity of 169% (based on BME Spanish Exchanges for the period 2002–2007). Greece and Austria have low RATIOs (0.28 and 0.23 in Table 1) as well as low turnover velocities (47% and 37% based on data for Athens Exchange and Wiener Börse). The OMX Nordic Exchange (Denmark, Finland and Sweden) has a velocity of shares equivalent to London SE (UK) for the period 2005–2007 but the RATIOs for the three Nordic countries are somewhat lower.

Apart from variables in line with work of La Porta et al. (1998) we also tried to integrate variables related to differences in home country bias (Sørensen et al. 2007), risk tolerance (Hofstede 1980) and public versus private equity (based on Orbis data) but none of these additional variables added significantly in explaining the variation in RATIO.

If we create a crude one-figure scale of investor protection and corruption by summing the three numerical variables in Table 3, we get the following result in ascending order: Greece (French) 7.6; Italy (French) 8.2; Belgium (French) 9.1; France (French) 10.3; Portugal (French) 10.5; Germany (German) 11.8; Ireland (Common) 12.5; Spain (French) 12.7; the Netherlands (French) 13.0; Austria (German) 13.1; Finland (Scandinavian) 13.4; Sweden (Scandinavian) 14.3; Denmark (Scandinavian) 14.4; the UK (Common) 17.4. The six lowest places are held by countries with French origin of law while the top four places are occupied by countries with Scandinavian origin of law and the UK with the latter as number 1.

We thank an anonymous referee for highlighting this point.

The relationship between the anti-director rights index of La Porta et al. (1998)/the anti-director rights index of Spamann (2010)/the difference between the ‘new index’ and the ‘old’ index is as follows for the 14 countries in the sample: Austria (2/4/+2); Belgium (0/2/+2); Denmark (2/4/+2); Finland (3/4/+1); France (3/5/+2); Germany (1/4/+3); Greece (2/3/+1); Ireland (4/4/0); Italy (1/4/+3); the Netherlands (2/4/+2); Portugal (3/4/+1); Spain (4/6/+2); Sweden (3/4/+1); the UK (5/5/0).