Presentation is loading. Please wait.

Presentation is loading. Please wait.

Once a Loser, Always a Loser? The effect of relegation on future career development in the England Soccer League Ang Sun Zhichao Wei Oct 29, 2012.

Similar presentations


Presentation on theme: "Once a Loser, Always a Loser? The effect of relegation on future career development in the England Soccer League Ang Sun Zhichao Wei Oct 29, 2012."— Presentation transcript:

1 Once a Loser, Always a Loser? The effect of relegation on future career development in the England Soccer League Ang Sun Zhichao Wei Oct 29, 2012

2 Research Question 1 The effect of job assignment on long-term career development A signal for future employers Opportunities for human-capital accumulation in the work place In the context of economic crisis Oreopoulos, Wacher and Heisz (2012) Little empirical research on state dependence of job quality

3 Research Question 2 Human-capital accumulation in the work place Empirical work on evaluating on-the-job training programs Learning by doing: Argote and Epple (1990) and Thompson (2010) We show dominant importance of practice in the soccer industry

4 The closely-relegated versus closely- non-relegated research design Study the state dependence of job quality by exploiting the setting of relegation of the league in England. Compare the teams ranked 3 rd -last and 4 th -last. Identifying assumption: The team on the margin but that succeeded in maintaining its position in the Premier League forms a valid counterfactual for the team on the margin but that is relegated to the second-tier league.

5 Main Findings Main findings: No difference in turnover (dropout or transfer) The players in the relegated teams played in lower-ranked clubs, had more appearances in the first three years after relegation, but eventually played in higher-ranked clubs four years after relegation. This pattern is driven by players below 30. Older players are worse off in both the short run and long run.

6 Interpretations The relegated team invests less in the transfer market, and players have more appearances in which to practice. This helps the players accumulate human capital (or slows down human-capital depreciation) and land relatively good teams in the future.

7 External Validity Methodological advantages Interesting and surprising results The extent of practice, future career development are easier to measure, and these measures closely mirror that in many immediately relevant domains The growing tradition of research that exploits the wealth of data and well-defined incentives in sports to investigate more general economic phenomena( Kotchen and Potoski, 2011)

8 Background on English football league system---Hierarchical format

9 Hierarchical revenues Television revenues shared exclusively among the Premier League members since 92 season Huge gap in revenues between the 1 st -and 2 nd –tier leagues The average Premier League team receives £45 million, while the average Football League Championship club receives £1 million.

10 The big effect of relegation on clubs and players For clubs Consequent financial problems including, in some cases, liquidation. For relegated team players Immediate transfers are restricted Less and lower-quality on-the-job training (-) Poorer facilities and limited access to other resources (-) Face weaker external and internal competition (?)

11

12

13 Data Focus on the whole career path of those who were once on the margin of relegation Step 1 : names of the clubs once on the margin Step 2 : the quad of each team in the season when on the margin of relegation Step 3 : based on the player roster, look for the player-level information in two relevant datasets to construct a panel by player and by time

14 Player-level info: data source 1 Online dataset Soccer Base Advantage: Rich info Date of birth, birth place, nationality, height, weight, position, all teams on which player played, the exact period played on each team (including the team he has contract with and the teams he was loaned or swapped to), transfer fee for each transfer he experienced, appearances and goals on each team for which he played. Disadvantage Appearances are aggregated to club level

15 Records precise time interval for experiences in each club for each player Match the data with yearly rankings Use this data set for the empirical analysis about the long-term placement of players, including the ranking of the club on which he played and signed up with at each point of change in contract.

16 Player-level info: data source 2 Online dataset Player History Advantage: aggregates players' performance to each season instead of club Disadvantage: No info about a loan or swap No record for transfers in the middle of a season. No info about transfer fee.

17 Use data source 2 to examine the change in appearances after relegation. The changes can be detected even if a player stays in the same club after relegation.

18 Summary statistics Sample clubs ranked third- and fourth-last during 1992-2002 #clubs16 10 On the margin once 6 On the margin twice #players670 By position 36.10%Defense 21.9%Forward 11.03%Goal Keeper 30.66%Midfield By nationality 64.24%England 35.76%Others Career length (yrs)17.036 (8.425) #transfers6.272 (3.521)

19 Third- last versus fourth-last teams

20 Player characteristicsTeams closely relegated Teams closely not relegated Difference (1)-(2) (1)(2)(3) Height 1.808 [.0036] 1.803 [.0038] -.0047 [.0052] Weight 76.73 [.507] 76.43 [.421] -0.296 [.657] Foreigner.377 [.0266].346 [.0265] -.019 [.038] Age at the relegation season 26.05 [.27] 26.05 [.29].005 [.395] Years of experience before 7.357576 [.3688691] 7.377778 [.4017467] -.020202 [.5462223] loaned or swapped to other clubs before.230303 [.0328767].2666667 [.0382017] -.0363636 [.0501499] Player’s maximum transfer fee 244059.7 [31947.88] 194463.5 [28311.36] 49596.27 [43469.08] #players 340330

21 Empirical Strategy The relegated versus non-relegated research design suggests the following specification:

22 Heterogeneous effect across players Direct implication: Younger players should “benefit more” in long-term job placement A Triple-Difference Strategy will not work! Categorize the sample into two sub-groups according to the age at relegation season

23 Discussing different measures (1) Two measures for career development Team ranking  job quality Transfer fee  market value Use rank as a measure for job quality Higher ranking also means higher wage wage=basic pay + appearance money + bonuses

24 Measure job quality

25 Average UK Workers wage, £ Season Premier League First Division Second Division Third Division (1)(2)(3)(4)(5)(6) £18,3561992-93£77,083£40,728£21,840£16,640 £18,8241993-94£93,968£47,358£23,745£17,190 £19,5521994-95£116,448£51,480£24,076£19,760 £20,3321995-96£130,896£52,000£26,000£21,840 £21,6321996-97£175,066£56,680£28,600£20,540 £22,8281997-98£244,908£59,280£34,060£24,492 £23,6601998-99£313,959£62,608£39,104£23,244 £24,5961999-00£383,835£71,500£38,532£25,272 £26,0002000-01£451,274£94,640£52,416£28,704 £27,3002001-02£566,932£115,700£50,336£28,600 £28,1322002-03£611,068£90,948£59,488£30,056 £28,6002003-04£651,222£92,300£55,052£42,472 The basic average annual pay of England’s professional players Sources: from internal PFA union files.

26 Measure market value : transfer fee

27 Discussing different measures (2) Use transfer fee as a measure for market value Selected sample? Relegated team players: 3.03 transfers Non-relegated team players: 2.94 transfers Over 90% of the players experienced at least one transfer after season 0 Measurement error in the dependent variable : As long as the error is not correlated with relegation assignment, it only results in a larger error variance.

28 Measure the degree of practice Total annual appearances (including substitute appearances) Appearances in starting line-up Issue of measurement Appearances may not be able to capture the exact minutes Intensity

29 Graphical presentation on the primary findings

30

31

32

33

34 Understand the mechanism

35

36 Other outcomes

37

38

39

40 Results of the statistical models Preview All results from fitting the Difference-In- Difference regression are consistent with the graphical patterns shown using raw data The regression results are robust to player fixed-effect and calendar-year fixed-effect

41 Table 3 Difference-In-Difference estimation on the effect of relegation on the ranking of the club signed up VARIABLESLeague LevelOverall Ranking (1)(2)(3)(4)(5)(6) Relegate*Season -4-0.094-0.121-0.067-2.150-2.724-1.584 (0.112)(0.109)(0.102)(2.566)(2.505)(2.306) Relegate*Season -3-0.128-0.164-0.129-3.595-4.400*-3.692 (0.118)(0.116)(0.112)(2.688)(2.625)(2.535) Relegate*Season -2-0.100-0.155-0.080-0.158-1.4110.107 (0.116)(0.113)(0.114)(2.639)(2.559)(2.537) Relegate*Season -10.1750.1040.1712.0250.4601.853 (0.111)(0.110)(0.112)(2.522)(2.483)(2.495) Relegate*Season 00.057-0.0260.0400.790-0.9990.421 (0.106)(0.105)(0.112)(2.409)(2.385)(2.505) Relegate*Season 10.487***0.413***0.510***5.078**3.5165.634** (0.111)(0.112)(0.118)(2.561)(2.552)(2.657) Relegate*Season 20.1800.1230.235*1.2880.1062.404 (0.120) (0.125)(2.803)(2.781)(2.854) Relegate*Season 3-0.065-0.1070.044-2.464-3.360-0.268 (0.132)(0.131)(0.135)(3.034)(3.002)(3.078) Relegate*Season 4-0.179-0.211-0.067-4.376-5.087-1.918 (0.148)(0.146)(0.151)(3.371)(3.337)(3.427) Relegate*Season 5-0.235-0.267*-0.232-5.570-6.266*-5.505 (0.156) (0.160)(3.514)(3.494)(3.590) Relegate*Season 6-0.333**-0.359**-0.303*-7.296*-7.865**-6.460* (0.166) (0.172)(3.732)(3.723)(3.832) Relegate*Season 7-0.378**-0.384**-0.345**-8.498**-8.559**-7.450* (0.172) (0.173)(3.848)(3.831)(3.860) Relegate*Season 8-0.437**-0.452**-0.384**-11.514***-11.793***-10.389** (0.181)(0.180)(0.182)(4.060)(4.032)(4.071) Relegate*Season 9-0.406**-0.437**-0.342*-10.321**-10.924***-8.735* (0.194)(0.191)(0.190)(4.287)(4.226)(4.189) Relegate*Season 10-0.210-0.230-0.166-5.261-5.575-4.093 (0.212)(0.208)(0.228)(4.761)(4.651)(4.503) Calendar Year F.E.NoYes.YesNoYes.Yes Player F.E.No YesNo Yes Se clustered at player levelYes Observations86401 R-squared0.1240.1450.4150.1100.1360.411

42 Table 4- DID estimation on the effect of relegation on the ranking of club played (Including the cases of being loaned or swapped to other clubs) VARIABLESLeague levelOverall ranking (1)(2)(3)(4)(5)(6) Relegate*Season -4-0.093-0.118-0.056-2.096-2.631-1.352 (0.108)(0.106)(0.100)(2.481)(2.431)(2.271) Relegate*Season -3-0.139-0.172-0.133-3.909-4.643*-3.847 (0.114)(0.113)(0.110)(2.606)(2.559)(2.497) Relegate*Season -2-0.121-0.173-0.081-0.717-1.890-0.015 (0.112)(0.110)(0.111)(2.543)(2.482)(2.475) Relegate*Season -10.1160.0480.1220.615-0.8830.642 (0.108)(0.107)(0.109)(2.438)(2.416)(2.430) Relegate*Season 00.007-0.0730.000-0.383-2.117-0.492 (0.103) (0.109)(2.354)(2.335)(2.434) Relegate*Season 10.402***0.332***0.431***3.3461.8884.047 (0.108)(0.109)(0.114)(2.493)(2.491)(2.563) Relegate*Season 20.1650.1110.225*0.963-0.1232.229 (0.117) (0.121)(2.728)(2.716)(2.775) Relegate*Season 3-0.080-0.1190.033-2.908-3.708-0.618 (0.128) (0.131)(2.956)(2.938)(2.987) Relegate*Season 4-0.233-0.262*-.210-5.637*-6.254*-2.982 (0.144)(0.143)(.153)(3.285)(3.265)(3.337) Relegate*Season 5-0.292*-0.322**-0.303*-6.826**-7.474**-7.014* (0.153) (0.164)(3.444)(3.437)(3.505) Relegate*Season 6-0.374**-0.399**-0.363**-8.280**-8.801**-7.822** (0.164) (0.175)(3.671)(3.670)(3.754) Relegate*Season 7-0.398**-0.403**-0.381**-9.250**-9.289**-8.556** (0.171) (0.181)(3.813)(3.802)(3.796) Relegate*Season 8-0.467***-0.480***-0.427**-12.218***-12.446***-11.354*** (0.179) (0.193)(4.014)(3.989)(3.997) Relegate*Season 9-0.449**-0.479**-0. 404*-11.293***-11.867***-10.114 (0.191)(0.189)(0.207)(4.224)(4.161)(4.667)** Relegate*Season 10-0.247-0.268-0.221-6.202-6.515-5.343 (0.213)(0.209)(0.228)(4.775)(4.663)(5.220) Player F.E.No YesNo Yes Calendar year F.E.NoYesNo YesNo Se clustered at player levelYes Observations86401 R-squared0.1180.1390.4110.1060.1300.407

43 Table 5- Difference in difference estimation on turnover VARIABLESDropoutTransfer (1)(2)(3)(4)(5)(6) Relegate*Season -40.007 0.008 (0.006) Relegate*Season -3-0.011* -0.010* (0.006) Relegate*Season -2-0.0000.0000.001 (0.006) Relegate*Season -1-0.007 -0.006 (0.005) (0.006) Relegate*Season 00.001 0.002 (0.005) (0.006) Relegate*Season 10.0020.0040.0170.005 0.006 (0.003)(0.006)(0.017)(0.005) Relegate*Season 2-0.005-0.0030.011-0.004 -0.003 (0.014)(0.015)(0.024)(0.005) Relegate*Season 3-0.006-0.0040.0090.000 0.002 (0.020)(0.021)(0.028)(0.006) Relegate*Season 40.0010.0020.016-0.001 -0.000 (0.024)(0.025)(0.033)(0.006) Relegate*Season 5-0.007-0.0050.0080.001 0.002 (0.029)(0.030)(0.037)(0.006) Relegate*Season 60.0100.0120.025-0.001 (0.032)(0.033)(0.039)(0.006) Relegate*Season 70.0210.0230.0360.0010.002 (0.035) (0.040)(0.006) Relegate*Season 80.0100.0120.025-0.000 0.001 (0.036)(0.037)(0.040)(0.006) Relegate*Season 90.0200.0210.0350.011* 0.012* (0.037) (0.040)(0.007) Relegate*Season 10-0.018-0.016-0.0030.012* 0.014* (0.037) (0.039)(0.007) (0.008) Player F.E.No YesNo Yes Calendar year F.E.NoYesNo YesNo Se clustered at player levelYes Observations8965 8655586,555

44 Transfer fee VARIABLESTransfer fee (1)(2)(3) Relegate*Season -4-48,407.177-54,363.095-295,124.153 (140,078.780)(140,172.362)(213,075.683) Relegate*Season -3-82,677.053-76,472.486-265,909.875 (131,331.597)(127,038.489)(191,687.705) Relegate*Season -214,541.05416,912.936-46,487.679 (113,227.529)(111,555.667)(131,505.001) Relegate*Season -1-37,341.910-48,576.166-138,731.819 (174,583.523)(172,264.787)(247,725.277) Relegate*Season 0-59,308.761-211.932-147,310.248 (112,229.513)(110,831.593)(155,562.698) Relegate*Season 1210,060.860260,744.03383,967.831 (184,266.562)(177,518.608)(181,085.527) Relegate*Season 2-180,997.077-233,821.726*-354,786.711** (123,433.656)(127,034.759)(161,678.502) Relegate*Season 3-59,203.310-70,007.515-63,641.260 (111,532.839)(113,629.212)(122,012.637) Relegate*Season 4-111,566.301-120,516.895-323,435.679 (217,302.434)(215,477.001)(236,702.864) Relegate*Season 5257,807.959239,246.635180,373.068 (242,884.018)(237,414.380)(228,568.433) Relegate*Season 6146,866.165143,371.901142,849.853 (283,689.014)(272,994.684)(238,768.606) Relegate*Season 7260,721.252246,985.656131,414.907 (242,084.795)(242,659.982)(213,617.193) Relegate*Season 8172,471.458190,129.559124,178.822 (185,333.221)(177,373.157)(137,994.897) Relegate*Season 9248,341.556291,105.84425,735.687 (242,833.961)(239,398.718)(186,480.490) Relegate*Season 10384,006.926316,833.964112,553.754 (250,832.246)(255,371.252)(194,334.784) F.E.NoCalendar year f.e.Player f.e. Se clustered at player levelYes Observations4,165 R-squared0.0230.0530.424

45 Appearances VARIABLESAppearances (1)(2)(3) Relegate*Season -41.1531.0981.252 (2.115)(2.127)(2.167) Relegate*Season -3-2.817-3.013-2.801 (2.222)(2.204)(2.250) Relegate*Season -2-0.626-0.744-0.571 (2.285)(2.289)(2.333) Relegate*Season -14.101*3.863*3.801 (2.270)(2.276)(2.322) Relegate*Season 02.1401.9991.382 (2.097)(2.119)(2.217) Relegate*Season 15.315**5.046**4.852** (2.223)(2.239)(2.338) Relegate*Season 24.395*4.132*3.998* (2.243)(2.238)(2.398) Relegate*Season 33.3973.1272.874 (2.338)(2.334)(2.526) Relegate*Season 41.7181.5761.923 (2.344)(2.367)(2.544) Relegate*Season 50.9060.6871.662 (2.169)(2.174)(2.365) Relegate*Season 60.8400.6530.899 (2.333)(2.331)(2.492) Relegate*Season 7-1.369-1.537-1.378 (2.522)(2.531)(2.703) Relegate*Season 80.1780.0060.240 (2.559)(2.561)(2.834) Relegate*Season 91.2520.8862.469 (2.521)(2.510)(2.780) Relegate*Season 100.5340.2820.816 (2.660)(2.689)(2.957) Observations5,402 R-squared0.0270.0390.218

46 Dropout rate top panel =30

47 Transfer rate top panel =30

48 Ranking of the team on which one played top panel =30

49 Appearances top panel =30

50 Concluding remarks A negative causal relationship between current job quality and future job quality. The big impact of practice or, more generally, learning-by-doing on human-capital accumulation in the specific industry of the English soccer league


Download ppt "Once a Loser, Always a Loser? The effect of relegation on future career development in the England Soccer League Ang Sun Zhichao Wei Oct 29, 2012."

Similar presentations


Ads by Google