I. Non-Parametric Models
Author/s / Country / Time / DMUs / Inputs / Outputs / Method / Policy ImplicationsLewin et al.
(1982) / U.S.A.
(North Carolina) / 1982 / · 30 criminal district courts
· 100 criminal superior country courts / · Number of district attorneys and assistants
· Days of criminal courts held (proxy for number of judges)
· Caseload
· Number of misdemeanours in the caseload
· Size of the white population / · Total number of dispositions
· Cases pending less than 90 days / · DEA (regression is used as an intermediate step to determine the relevant input factors for the DEA analysis) / · 63 out of the 97 superior country courts analyzed are inefficient.
· The sum of the inefficiencies observed at the county level are more than double those observed at the judicial district level, indicating that most inefficiencies may be due to administrative policies or practices at the judicial district level.
Kittelsen & Førsund (1992) / Norway / 1983-1988 / · 107 district courts
(comprising 91 diversified courts, 6 general city courts, 10 specialized city courts) / · Judges
· Admin staff / · Civil cases
· B-cases
· Examination and summary jurisdiction cases
· Ordinary criminal cases
· Registry cases
· Cases of duress
· Probate and bankruptcy cases / · DEA ( input and output oriented, CRS)
· Pure scale efficiency & scale indicator
· Malmquist productivity index (catching-up & frontier shift) / · Smaller courts are predominant among the overall scale inefficient courts, while larger courts tend to have a score of 1.
· Total efficiency loss is estimated to be 8-10% in the aggregate. Most of this loss is due to non-optimal scale rather than technical inefficiency.
· The optimal scale size lies not far above the average court, and for a wide range of court sizes one can reject the presence of VRS.
· Provided the severity of cases is not much greater in the cities than in rural areas and small towns, the optimal scope of the courts seem to be fully diversified.
Tulkens (1993) / Belgium / 1983-1985 / · 187 Courts on Justices of the Peace (courts of one judge) / · Staff (non-judge) working in each Court: e.g. clerks / · Settled civil and commercial cases
· Family arbitration sessions held
· Minor offense cases settled / · FDH (free disposal hull) / · 82% to 87% of the courts studied are inefficient according to the FDH results.
· Most inefficient units are middle size courts as measured by clerical staff.
· 35% of available backlog can be reduced by productivity improvements. This percentage reaches 40% in middle size courts.
· Still 70% of the backlog cannot be reduced. Thus personnel increases seem to be justified.
Pedraja & Salinas (1996) / Spain / 1991 / · 21 Spanish supreme courts (administrative litigation division; e.g. personnel cases, taxes, ..) / · Judges
· Office staff / · Cases resolved through full legal process (sentences)
· Other resolved cases (e.g. dismissals, withdrawals, conciliations, etc.) / · Fractional DEA (output oriented, CRS[1]) / · From the 21 units analyzed, five are relatively efficient.
· The mean efficiency of the 21 courts is 77.38%, i.e. there appears to be significant scope for improvement.
Sampaio and Schwengber
(2005) / Brazil, Rio Grande do Soul / 2002 and 2003 / · 161 Lower courts
(only 16 for the analysis of the m-frontier) / · Judges
· Admin staff
· Case stock / · 6 different types of cases (e.g. civil, criminal, children) / · FDH and order-m-frontier (output oriented) / · Small courts tend to be more inefficient than its larger counterparts.
· Reducible backlog decreases with the size of the courts.
· Due to their size, smaller courts do not explore economies of scale in relatively more resources, as they do not benefit from specialization, found in larger courts.
· Mean Efficiency: 101.82%.
· Standard Deviation:14.39%.
Hagstedt and Proos (2008) / Sweden / 1998/99 (before reform) and 2006/07
(after reform) / · District courts
(civil and criminal:
21 courts in the first time period and 17 courts in the second time period) / · Labour costs / · Number of Resolved cases / · DEA (output oriented, VRS and CRS) / · The reform program (reducing the number of existing courts) has improved the relative efficiency of most district courts.
· Mean Efficiency increased by 4% after reform (from 73.55% to 77.6%).
· Standard Deviation: 15.86% (1998-1999), 15.47% (2006-2007)
García- Rubio, Angel and Rosales-López (2010) / Spain (Andalusia) / 2008 / · 65 Civil FICs / · Caseload (filed + pending)
· Admin staff / · Sentences and warrants / · DEA (output oriented, VRS) / · Efficiency improvements of Andalusian civil FICs can help reduce the pending cases by only 9.38%, which means that efficiency improvements won’t entirely resolve the problem of backlogs of Andalusian Civil Jurisdictions.
El-Bialy and García- Rubio (2011) / Egypt / 2010 / · FICs (21 civil and 21 criminal districts) / · Judges
· Admin staff
· Computers / · Number of resolved cases / · DEA (output oriented, VRS)
· Mann- Whitney tests testing the relative significance of efficiency differences within courts of the same type and among the two different court types: civil and criminal. / · There is no significant difference between the relative efficiency levels of courts belonging to the same group; however there is a significant difference in the relative efficiency among the two court types.
· Criminal district courts are more efficient (mean efficiency 68%) than their corresponding civil district courts (mean efficiency 64%) belonging to the same FICs even after the reform program.
· Mean efficiency 132% criminal district courts136% civil
· Standard Deviation: Civil 37.568%, Criminal 56.213%
Yeung and Azevedo
(2011) / Brazil / 2008 / · 27 first and second degree State Courts / · Judges/100.000 inhabitants
· Admin. Staff/100.000 inhabitants
· Number of computers per users / · Resolution rate (output/caseload), where caseload = filed +pending / · DEA (output oriented, CRS) / · Relative efficiency levels vary substantially across different courts.
· Court efficiency should rather be linked to the organizational climate, staff motivation and management quality within a court rather than the lack of material and human resources.
· Standard Deviation: 0.2532 (2006), 0.2401(2007), 0.2129 (2008)
· Mean: 0.660 (2006), 0.590(2007), 0.676(2008)
Peyrache, and Zago (2012). / Italy / 2003-2008 / 165 courts / Professional judges
Non-professional judges
Non-judge staff
Pending civil cases
Pending criminal cases / -Civil resolved cases
-Criminal resolved cases / Output oriented directional distance function (DDF) / · 35% of total inefficiency of the judiciary is due to size.
· Splitting large courts leads to higher efficiency scores.
Santos and Amado (2014) / Portugal / 2007-2011 / 223 FICs / 2 input variables, judges and support staff / -Civil resolved cases-Criminal resolved cases / DEA / · Courts with a higher proportion of support staff perform better than courts with a higher proportion of judges
· Smaller courts are less efficient than larger courts.
Schneider (2005) / All of these rely on the so-called 2 stage DEA, which is a combination of DEA and regression analysis, hence see the Parametric Model Table
Deyneli
(2011)
Finnocchiaro Castro, and Guccio (2014).
Melcarne and Ramello (2015)
Falavigna et al., , , (2015).
II. Parametric Models
Luskin and Luskin
(1986) / USA (Detroit) / Apr. 1976-March 1978 / · 2000 felony cases that reached disposition in the criminal court tribunal of Detroit
· Microdata, case-level data / · 20 explanatory variables :
Case-specific intcentives:
· Attorney Type
· Pretrial Release (+)
· Seriousness (+)
· Defendant’s prior record (+)
· Regular judge vs visiting judge
Case complexity:
· Number of Defendants (+)
Case events:
· Early Dismissal before hearing (-)
· Trial (+)
· Pretrial Motions (+)
· Psychiatric Hiatus (+)
· Defendant absence (+)
· Number of continuances (+)
· Late or Second preliminaryhearing
· Mistrial
(Delay reduction project)
Structural incentives:
· Docket Type (central vs individual)
· Case-Track mechanism
· Decentralized Plea Bargaining (-)
· The crash program(^^ reform)(-)
Caseload:
· Court caseload
· Individual caseload / · Duration of trial: (time between arrival and disposition) micro-level-case-data) / · GLS / · Case load: under the central docket, the court caseload has no significant effect, whereas under the individual docket, the judge's caseload has a negative significant effect
· Docket type, late or second preliminary hearing, and mistrial are statistically insignificant.
· For other significant variables, see indep variable column; the variables with signs.
Ostrom and
Hanson (2000) / USA / 1994 / · US states criminal trial
courts (9)
· Using case-level data: (400) felonies per court / · Severity of charge at indictment
· Procedural aspects
· Manner of resolution
· Defendant
· Resources / · Number of days required to resolve each case: as measured from the time of indictment or bind over to final disposition, adjusted for a criminal defendant’s time out on bench / · Questionnaires:
(A three-page questionnaire was distributed to each judge, prosecutor, and full-time public defender who handled felony cases in 1994.Response rates varied, but at least 15 prosecutors and 15 defenders completed questionnaires in each jurisdiction)
· Regression / · The combined influence of a most violent felony charge, the issuance of a bench warrant, pre-trial release on bond, and resolution by trial tended to produce a significant increase in the time to resolution in all courts studied.
· The nine court systems handled their common caseloads with the same relative degree of timeliness.
Murrell
(2001) / Romania / 1999 / · 42 Tribunals
· (FICs)
· Commercial cases (patrimonial vs non-patrimonial) / Supply:
· Court resources: Number of judges in a tribunal (serving commercial as well as non-commercial trials[2])
· Caseload: (1) Number of filed commercial cases; (2) Number of cases of all types in the Tribunal other than those included in the model
· Local legal culture: criminal court performance as a proxy for court's performance outside its commercial section
· Socioeconomic environment: dummy variable capturing whether a judet is in Transylvania (innovative ara in Romania)
· Percentage of votes for the major left-of-center party in the 1996 presidential election
Demand:
Level of congestion
· Tribunal congestion
· Judicatorial congestion
Appeal:
· Percentage of appeals that are successful in the pertinent appeals court (Appeal success rate)
· Enterprises
· Number of enterprises in each jurisdiction
· Level of eco. Activity ( Total revenues of small and medium enterprises, 1998)
· Percentage of enterprises in the judet classified as large or medium, 1998
Urbanization:
· Percentage of the population in urban areas, 1998
· Industry:Percentage of the work force in industry, 1998 / Supply:
· Cappelletti-Clark index = [(pending+filed)/disposed]. Differentiating between 3 categories: (1) patrimonial cases, (2) non-patrimonial cases and (3) an aggregation of these two
Demand:
· Filed cases / · 3SLS / Supply:
· Case load: higher caseload leads to higher levels of congestion.
· Local legal culture: criminal court congestion variable has the strongest results.
· Socioeconomic environment: positive sign and the significance of the variable measuring local support for Iliescu begs interpretation.
Demand:
· Congestion : court congestion statistics are statistically significant.
· Appeals: Enterprises react to the probability of success in the appeals court, but the variable is only statistically significant for non-patrimonial cases.
· OLS estimation can be misleading. OLS : an increase in caseload decreases court congestion.
· 3SLS: although with weak statistical Significance, an increase in caseload increases the degree of court congestion
· Large firms impose much less burden per unit of economic activity on the court system than do small firms.
Djankov et al. (2003) / Cross country / NA / · Lower courts of (109) countries / OLS1:
· GNP per capita
· Socialist legal origin
· French legal origin
· German legal origin
· Scandinavian legal origin
OLS2:
· Log GNP per capita
Formalism index / OLS1: identifying the determinants of formalism:
· Formalism Index (7 components of the index)
OLS2: determine the consequences of formalism:
· Duration of trial
· Efficiency of the
· Judicial system
· Access to justice
· Enforceability of contracts
· Corruption
· Human rights
· Legal system is fair and impartial
· Legal system is honest and uncorrupt
· Legal system is quick
· Legal system is affordable
· Legal system is
· Consistent
· Court decisions are enforced
· Confidence in legal system / · Questionnaires:
(to construct an index of procedural formalism in dispute resolution).
· Regression (conducting regression analysis first: to test the impact of several institutional and legal factors on judicial formalism. Second: to measure the impact of formalism on the quality of the legal system). / · Being a civil law country has a positive significant impact on formalism.
· The expected duration is systematically higher in countries with more formalized proceedings, but is independent of the level of development. Perhaps more surprisingly, formalism is nearly universally associated with lower survey measures of the quality of legal system, including judicial efficiency.
· Access to justice, honesty, consistency, impartiality, fairness, and even human rights.
Beenstock and Heitovsky
(2004) / Israel / 1965-1995
1976-1995
1971-1995 1981-1993 / · M1: High courts (1)
· M2: District courts (5)
· M3: Magistrates (3)
· M4: Magistrates (16) / · Newly filed cases
· Pending cases[3]
· Judges / · Number of case dispositions / Regression
· M1: time series
· M2-M4: Panel(FE) / · Judges complete more cases under pressure and less when new judges are appointed
Micevska and Hazra (2004)[4] / India / 1995-1999 / · Indian lower courts
· Belonging to (27) different states and union territories