Kleven et al. (2013)

• Abstract: We analyze the effects of top tax rates on international migra-tion of football players in 14 European countries since 1985. Both coun-try case studies and multinomial regressions show evidence of strongmobility responses to tax rates, with an elasticity of the number of for-eign (domestic) players to the net-of-tax rate around one (around 0.15).We also find evidence of sorting effects (low taxes attract high-abilityplayers who displace low-ability players) and displacement effects (lowtaxes on foreigners displace domestic players). Those results can be ra-tionalized in a simple model of migration and taxation with rigid labordemand.

1

Background

• There is very little empirical work on the effect of taxation on the spatialmobility of individuals, especially among high-skilled workers.

– Vast empirical literatures on labor supply and taxable income re-sponses to taxation within countries (see, e.g., Saez et al. (2012)).

– Also many studies on the effects of capital taxation on multinationalcorporations and international capital mobility that find substantialmobility effects (see, e.g., Devereux and Griffith (2003)).

– While a small literature (e.g., Young and Varner (2011)) has consid-ered the mobility of people across local jurisdictions within coun-tries, empirical work on the effect of taxation on international mo-bility appears to be virtually nonexistent partly due to a lack of mi-cro data with citizenship information and challenges in identifyingcausal tax effects on migration.

2

Data• Data on the first-league football players in 14 European countries (Aus-

tria, Belgium, Denmark, England, France, Germany, Greece, Italy, theNetherlands, Norway, Portugal, Spain, Sweden, and Switzerland) since1985 from online sources.

– The main online source is the website playerhistory.com, with de-tailed information available since the 1970s.

– Individual player information in the data include name, nationality,date of birth, club affiliation, and national team selection of each per-son in each first-league club in all 14 countries from 1985 to present.

• Top earnings tax rate data across countries and over time, taking intoaccount special tax rules applying to immigrant workers and sometimesto athletes specifically.

– Top football players’ average tax rate is well approximated by the top

3

marginal tax rate when combining (i) the top individual income taxrate, (ii) uncapped social security contributions, and (iii) value-addedtaxes.

• We restrict our sample to players who are citizens of one of the countries.

– We also exclude players with multiple nationalities.

∗ A number of scandals (especially in Italy) revealed that some play-ers listed with multiple nationalities had fake European passportsin order to get around the quotas.

• Furthermore, we have obtained information on individual player salariesfor about half of our sample for the years 1999–2000 and 2004–2008.

– We use those actual earnings to impute individual earnings and cal-culate average tax rates in our full sample for years 1996–2008.

4

Bosman Ruling• The European Court of Justice made the Bosman ruling in December

1995, which eliminated the three-player rule and the transfer-fee rulefor European (UEFA member) clubs.

– The three-player rule stipulated that no more than three foreign play-ers could be aligned in any game in the UEFA club (and in mostnational) competitions.

∗ Non-European players are still subject to the foreign-player quotasthat were imposed on all players in the pre-Bosman era.

– The transfer-fee rule allowed clubs to require a transfer fee when aplayer wanted to move to another club even at the end of the player’scontract.

– Bosman was a Belgian player who sued his club, which was refusingto let him go at the end of his contract.

5

6

• Country-level club performance is measured by the total points (calcu-lated according to UEFA’s formula) earned by all clubs in a given coun-try in the UEFA competitions.

7

Spanish Tax Reform in 2004: Beckham Law

• Special tax scheme passed in 2005, applicable to foreign workers (notjust football players) moving to Spain after January 1, 2004.

– Flat tax of 24% in lieu of the regular progressive income tax with atop rate of 43% in 2008 (45% when the Beckham law was passed).

– Eligibility requires not having been a tax resident in Spain at anypoint during the preceding 10 years.

– David Beckham moved from Manchester United to Real Madrid, andbecame one of the first foreigners to take advantage of it.

8

9

• Top-ability players are here defined as those who have been selected atleast once for the national team of their home country over the course oftheir career to date, while lower-quality players are those who have not.

• The weights on different countries in the construction of a synthetic con-trol country are nonnegative and chosen to minimize the pre-reform dis-tance between treatment and control in terms of the outcome of interestand indexes of football league quality.

10

Danish Reform in 1992: Researcher’s Tax Scheme

• In 1992, Denmark enacted a preferential tax scheme for foreign re-searchers and high-income foreigners in all other professions, who signcontracts for employment in Denmark after June 1, 1991.

– Flat tax of 25% (30% from 1991 to 1995) in lieu of the regularprogressive income tax with a top rate above 60% (68% when thescheme was introduced).

– Maximum period of 36 months after which the taxpayer becomessubject to the ordinary schedule.

– To become eligible, (i) the taxpayer cannot have been tax liable inDenmark in the 3 years prior to going on the scheme, and (ii) for non-researchers, eligibility requires an annual income of at least 765,600Danish kroner—about 103,000 euros—as of 2009, where the thresh-old is indexed to average nominal wage growth in Denmark.

11

• The impact of the scheme on the football sector has been the subject ofmuch public debate over the past 10–15 years.

• The widespread use of the scheme by football players was an unintendedconsequence of the reform that has been criticized subsequently by someof the political parties responsible for passing the law in the first place.

– Recently, the Danish Minister of Taxation has been working on areform proposal that would abolish the tax scheme specifically forathletes.

12

13

14

• Each panel plots the number of foreign workers from 1980 to 2005 (nor-malized to one in 1990, just before the scheme introduction) for for-eigners with earnings above the scheme eligibility threshold (treatment)and foreigners with earnings between 80 percent and 99.5 percent of thescheme eligibility threshold (control).

15

Fixed effect

• yi,t = a+bxi,t + czi,t + ei,t , zi,t = zi +∆zi,t

⇐⇒ yi,t = ai +bxi,t + c∆zi,t + ei,t , ai = a+ czi (fixed effect)

i y x (z) (e)

1 20 1 (2) (0)

1 20 2 (2) (0)

2 10 2 (1) (0)

→ y = 25−5x+ e?

–

y D1 D2 x (c∆z+ e)

20 1 0 1 (0)

20 1 0 2 (0)

10 0 1 2 (0)

→ b = 0 (a1 − a2 = 10)

16

Estimated Model• Multinomial logit model:

lnPi

nt

Pimt

(= lnPint − lnPi

mt) = α ln(1− τ int)+b ·homei

nt +xitβn + γn

− (α ln(1− τ imt)+b ·homei

mt +xitβm + γm)

– Pint : player i’s probability of locating in country n at time t.

∗ ∑m Pimt = 1 ⇐⇒ Pi

nt = eα ln(1−τint )+b·homei

nt +xit βn+γn

∑m eα ln(1−τimt )+b·homei

mt +xit βm+γm

.

– τ int : the average tax rate on i in n at t.

– homeint : home country dummy (equal to 1 if n is i’s native country).

• Estimated by maximum likelihood: with pit = Pi

nt if i located in n at t,

maxα,b,β ,γ

∏i,t

pit

(⇐⇒ max

α,b,β ,γ∑i,t

ln pit

)⇒ α, b, β , γ.

17

Elasticity• The individual elasticity of Pi

nt with respect to 1− τ int :

ε int ≡

d lnPint

d ln(1− τ int)

= α · (1−Pint).

• The country n level elasticities:

εndomestic =

d ln(∑i∈In Pint)

d ln(1− τnd)= α ∑

i∈In

Pint

∑ j∈In P jnt

(1−Pint),

εnf oreign =

d ln(∑i∈ICn

Pint)

d ln(1− τn f )= α ∑

i∈ICn

Pint

∑ j∈ICn

P jnt

(1−Pint).

– In (ICn ) is the set of all natives (nonnatives) from country n.

• εdomestic and ε f oreign are the weighted averages across all countries.

– The country n weight for εndomestic (εn

f oreign) is ∑i∈In Pint (∑i∈IC

nPi

nt ).

18

19

• We construct a quality index, and include a set of 4 dummy variablescorresponding to each quartile of the quality distribution in the currentyear.

– Alternative measure of player quality is based on the average qualityof a player’s clubs over just the 3 preceding seasons, as opposed toover all preceding years.

• The top marginal tax rate is a less accurate proxy for the average taxrate in the pre-Bosman era where football wages were generally lower,which may create downward bias.

• As most players play at home, Pint is small for foreign countries and

therefore ε f oreigner ≈ α . Pint is large for home countries (around 90 per-

cent) and therefore εdomestic ≈ α/10 ≪ α .

20

Demand Rigidity

• The number of teams per league tends to be fixed and the number ofplayers per team is constrained by the fact that the game involves exactly11 players on the field and a maximum of three substitutions per game(although picked from a pool of potential substitutes that can be larger).

– On the other hand, clubs play many games over a season, and there-fore require a much larger number of players to insure themselvesagainst injuries and fluctuations in player performance over time.

• We add the second leagues from the top-five countries (England, France,Germany, Italy, and Spain).

– Information on second leagues is only available since 1999.

21

22

References

Devereux, Michael P. and Rachel Griffith (2003) “Evaluating Tax Policy forLocation Decisions,” International Tax and Public Finance, Vol. 10, No.2, pp. 107–126.

Kleven, Henrik Jacobsen, Camille Landais, and Emmanuel Saez (2013)“Taxation and International Migration of Superstars: Evidence from theEuropean Football Market,” American Economic Review, Vol. 103, No. 5,pp. 1892–1924.

Saez, Emmanuel, Joel Slemrod, and Seth H. Giertz (2012) “The Elasticity ofTaxable Income with Respect to Marginal Tax Rates: A Critical Review,”Journal of Economic Literature, Vol. 50, No. 1, pp. 3–50.

Young, Cristobal and Charles Varner (2011) “Millionaire Migration and

23

State Taxation of Top Incomes: Evidence from a Natural Experiment,”National Tax Journal, Vol. 64, pp. 255–83.

24