EE Efficiency Analysis of German Public Transit Is Big Beautiful? - - PowerPoint PPT Presentation

• 1 -

EE²

Efficiency Analysis of German Public Transit – Is Big Beautiful?

Christian von Hirschhausen and Astrid Cullmann

5th INFRADAY Berlin 07.10. 2006

Dresden University of Technology, Energy Economics and Public Sector Management, and DIW Berlin

EE²

• 2 -

EE²

Agenda

• 1. Issues, Motivation, Literature
• 2. Methods
• 3. Data and Model Specification
• 4. Empirical Results
• 5. Conclusions

• 3 -

EE²

ÖPNV - Verkehrsverbünde

Quelle: http://www.oepnv-info.de/dkarte/index.php

1 Aachener Verkehrsverbund (AVV) 33 Verkehrsgemeinschaft Landkreis Cham (VLC) 2 Augsburger Tarif- und Verkehrsverbund (AVV) 34 Verkehrsgemeinschaft Landkreis Passau (VLP) 3 Bodensee-Oberschwaben Verkehrsverbund (BOD 35 Verkehrsgemeinschaft Münsterland (VGM) 4 Donau-Iller-Nahverkehrsverbund (DING) 36 Verkehrsgemeinschaft Niederrhein (VGN) 5 Gemeinschaftstarif Vorpommern (GTV) 37 Verkehrsgemeinschaft Rottal-Inn (VGRI) 6 Großraum-Verkehr Hannover (GVH) 38 Verkehrsgemeinschaft Ruhr-Lippe (VRL) 7 Hamburger Verkehrsverbund (HVV) 39 Verkehrsgemeinschaft Westfalen-Süd (VGWS) 8 Heidenheimer Tarifverbund (HTV) 40 Verkehrsunternehmensverbund Mainfranken (VVM) 9 Heilbronner-Hohenloher-Haller Nahverkehr (H3NV 41 Verkehrsverbund Berlin-Brandenburg (VBB) 10 Karlsruher Verkehrsverbund (KVV) 42 Verkehrsverbund Bremen/Niedersachsen (VBN) 11 Kitzinger Nahverkehrs Gemeinschaft (KING) 43 Verkehrsverbund Großraum Nürnberg (VGN) 12 KreisVerkehr Schwäbisch-Hall (KVSH) 44 Verkehrsverbund Hegau-Bodensee (VHB) 13 Ludwigsluster Tarifverbund (LTV) 45 Verkehrsverbund Mittelsachsen (VMS) 14 Mitteldeutscher Verkehrsverbund (MDV) 46 Verkehrsverbund Neckar-Alb-Donau (NALDO) 15 Münchner Verkehrs- und Tarifverbund (MVV) 47 Verkehrsverbund Oberelbe (VVO) 16 Nordhessischer Verkehrsverbund (NVV) 48 Verkehrsverbund Ostwestfalen-Lippe/Der Sechser (VVOWL) 17 Regensburger Verkehrsverbund (RVV) 49 Verkehrsverbund Pforzheim-Enzkreis (VPE) 18 Regio Verkehrsverbund Lörrach (RVL) 50 Verkehrsverbund Region Kiel (VRK) 19 Regio-Verkehrsverbund Freiburg (RVF) 51 Verkehrsverbund Region Trier (VRT) 20 Rhein-Main-Verkehrsverbund (RMV) 52 Verkehrsverbund Rhein-Mosel (VRM) 21 Rhein-Nahe-Nahverkehrsverbund (RNN) 53 Verkehrsverbund Rhein-Neckar (VRN) 22 Saarländischer Verkehrsverbund (saarVV) 54 Verkehrsverbund Rhein-Ruhr (VRR) 23 Schleswig-Holstein-Tarif (SH-Tarif) 55 Verkehrsverbund Rhein-Sieg (VRS) 24 Tarifgemeinschaft Lübeck (TGL) 56 Verkehrsverbund Rottweil (VVR) 25 Tarifverbund Ortenau (TGO) 57 Verkehrsverbund Schwarzwald-Baar (VSB) 26 Tarifverbund Schaffhausen (TVSH) 58 Verkehrsverbund Süd-Niedersachsen (VSN) 27 Verbundtarif Mittelthüringen/Voll-Mobil-Ticket (VM 59 Verkehrsverbund Tuttlingen/TuTicket (VTU) 28 Verbundtarif Region Braunschweig (VRB) 60 Verkehrsverbund Vogtland (VTV) 29 Verkehrs- und Tarifverbund Stuttgart (VVS) 61 Verkehrsverbund Warnow (VVW) 30 Verkehrs-Gemeinschaft Freudenstadt (VGF) 62 Waldshuter Tarifverbund (WTV) 31 Verkehrsgemeinschaft am Bayerischen Untermai 63 Westpfalz Verkehrsverbund (WVV) 32 Verkehrsgemeinschaft Bäderkreis Calw (VGC) 64 Zweckverband Verkehrsverbund Oberlausitz-Niederschlesien (ZVON

• 4 -

EE²

Motivation: (German) Public Transit in Turmoil

„Wir wollen Wettbewerb, und wir haben bereits einen funktionierenden Wettbewerb im deutschen ÖPNV. Was wir aber nicht wollen, sind unfaire Konkurrenzbedingungen zwischen einem kleinen Busunternehmer und einem europäischen Mobilitätsgroßkonzern. Das hätte nicht unseren Vorstellungen eines fairen Wettbewerbs entsprochen, der in Deutschland die Existenz von mehr als 1000 gut aufgestellten mittelständischen Unternehmen gefährdet hätte.“

Bundesverkehrsminister Tiefensee: Pressemitteilung zum EU- Verkehrsministerrat mit dem Thema ÖPNV-Verordnung 1191 (Bonn, 9. Juni 2006) (Hervorhebung zugefügt)

=> „We want competition, … but not if it endangers our 1,000 small and medium enterprises“ Minister of Transport Wolfgang Tiefensee

• 5 -

EE²

State of the Literature (I): No Clear Evidence

Berechman (1993): „Results concerning economies of scale are rather inconclusive“:

• Bus industry as a whole: constant scale economies
• Small firms (less than 100 buses) likely to experience increasing scale efficiencies
• Medium-sized firms (100-500 buses) facing very small or constant scale economies
• Large-scale bus systems (> 500 buses) most likely decreasing returns to scale (in

particular Chicago: 2,500 buses, New York MTA: 3,000 buses)

Related literature on public transit efficiency measurement

• Farsi/Fillipini/Kuenzle (2005, 2006): on stochastic frontiers and average cost functions,

indicating first falling, then rather constant average costs

• 6 -

EE²

State of the Literature (II): No Clear Evidence

Brons et al. (2005) overview of different aspects and applications; explain the variation in efficiency findings reported in the literature Viton (1981) specify and estimate flexible cost functions for 54 US bus transit companies; advantages of translog cost functions Several country studies except for Germany Mizutani and Urakami (2002) efficiency between private and public bus

• perators in Japan; apply econometric cost functions

Matas and Raymond (1998) Spain during the period 1983–1995; econometric cost function Filippini and Prioni (1994), Filippini and Prioni (2003) Swiss regional bus companies; cost frontier approach; question if inefficiencies are due to a regulatory problem. Tulkens (1993) apply the methodology of free disposal hull (FDH) to measure of productive efficiency in urban transit.

• 7 -

EE²

Agenda

• 1. Issues, Motivation, Literature
• 2. Methods
• 3. Data and Model Specification
• 4. Empirical Results
• 5. Conclusions

• 8 -

EE²

Benchmarking Methods – Survey

Frontier Analysis Parametric Deterministic (COLS) Stochastic (SFA) Extensions for Panel Data Fixed Effects Model GLS MLE True Random Effects Non-parametric DEA FDH Quality Efficiency

• Cost Efficiency
• Technical Efficiency

Productivity Total Factor Productivity Partial Indicators Malmquist Indices Order-m

• 9 -

EE²

Y e.g. units sold X e.g. labour, network size

C B A

Efficiency Frontier DEA CRS Efficiency Frontier DEA VRS

,

max ( ´ / ´ ), ´ / ´ 1, 1,2,... ,

u v i i i i

u y v x u y v x j N u v ≤ = ≥

,

max ( ´, ), ´ 1 ´ ´ 0, 1,2..., , 0,

i i i i

y v x y x j N

µ ν µ

µ ν µ ν = − ≤ = ≥

,

m in ,

i i

y Y x X

θ λ θ

λ θ λ λ − + ≥ − ≥ ≥

Data Envelopment Analysis (DEA)

• 10 -

EE²

Methods

Y X B A

Data Envelopment Analysis Out put Inpu t True Production Frontier True order-m Frontier Estimated Order-m Frontier A D G C B E F Free Disposal Hull Order-m

• 11 -

EE²

Agenda

• 1. Issues, Motivation, Literature
• 2. Methods
• 3. Data and Model Specification
• 4. Empirical Results
• 5. Conclusions

• 12 -

EE²

Model Specification (I)

Empirical analysis of the technical efficiency: Look in detail at 200 German public transit bus companies (including companies operating exclusively in the public bus transit, not included companies operating in different transit sectors (multi-output including metro)) Observation period (1990-2004) Different nonparametric approaches (DEA, FDH, Order-m) Sensitivity Analysis by means of Bootstrapping

• 13 -

EE²

Model Specification (II) – Data Description

• Physical and geographical data
• Technical efficiency only (no cost and input factor price data available at

this time)

• Cannot consider allocative efficiency
• Data taken from VDV “Verband deutscher Verkehrsunternehemen”
• Sorted out missing data – balanced panel
• Problem of outsourcing: sorted out utilities with less than 10 employees
• Companies including all sizes operating in urban and rural service areas

• 14 -

EE²

Model Specification (III) – Base Model

Production Frontier Models Inputs: Labour: number of workers Number of busses approximation for capital input Outputs: Seat kilometers

• 15 -

EE²

Model Variation (Sample 1990-2004)

I (non- dis) I Input Density Index I I I Output Passengers km I I I I I Input Busses I Model 5 I I Model 1 Model 4 Model 3 Model 2 Output Bus km Output Seat km Input Length of Lines in km Input Labor I I I I

• 16 -

EE²

Agenda

• 1. Issues, Motivation, Literature
• 2. Methods
• 3. Data and Model Specification
• 4. Empirical Results
• 5. Conclusions

• 17 -

EE²

DEA Model 1 Pooled Regression

DEA Model 1 Pooled Regression Difference Results VRS-CRS

0,00% 30,00% 60,00% 90,00% 1 1 23 245 367 489 61 1 733 855 977 1 099 1 221 1 343 1 465 1 587 1 709 1 831 1 953 2075 21 97 231 9 2441 C ompanies ordered by size

Difference increases when firm size decreases, scale inefficiency (IRS)

• 18 -

EE²

DEA Model 5 (CRS) Including Density

DEA Model 5 (CRS) Desnity as non-discretionary input

0,00% 10,00% 20,00% 30,00% 40,00% 50,00% 60,00% 70,00% 80,00% 90,00% 1 69 137 205 273 341 409 477 545 613 681 749 817 885 953 1021 1089 1157 1225 1293 1361 1429 1497 1565 1633 1701 1769 1837 1905 1973 2041 2109 2177 2245 2313 2381 2449

C ompanies ordered by size

Small utilities operating in less densely settled areas are

• 19 -

EE²

Most Efficient Companies

2003 Künzelsau (NVH) 1,25 1,33 1,00 2004 Ulm (RAB) 1,13 1,39 1,00 2002 Burg(NJL) 1,08 1,08 1,00 1998 Burg (NJL) 1,00 1,03 1,00 1999 Burg(NJL) 1,00 1,01 1,00 1991 Celle (OHE) 0,96 0,97 1,00 1991 Bad Pyrmont 0,95 1,03 1,00 1999 Weimar (Verwaltungsges) 0,93 1,02 1,00 1990 Ulm (RAB) 0,93 1,17 1,00 2003 Burg(NJL) 0,92 0,92 1,00 1990 Bad Pyrmont 0,91 0,99 1,00 2002 Weimar (Verwaltungsgesells 0,91 0,95 1,00 2003 Weimar (VWG des ÖPNV) ' 0,91 0,95 1,00 2003 Ulm (RAB) 0,86 0,86 0,86 1990 Stuttgart (RBS) 0,85 0,98 1,00 2001 Groß-Gerau (RWGG) 0,85 0,88 1,00 1990 Bielefeld (BVO) 0,80 1,03 1,00 2004 Frankfurt/ Main (VU) 0,80 0,80 1,00 1990 Kassel (RKH) 0,79 0,98 1,00 1999 Lübeck (SL) 0,78 0,78 1,00

Including Super-efficiency Problem of outsourcing, Künzelsau

• 20 -

EE²

Technical Efficiency over time 1990-2004 DEA Model 1

Technical Efficiency over time 1990-2004 DEA Model 1

0,32 0,34 0,36 0,38 0,4 0,42 0,44 1 2 3 4 5 6 7 8 9 1 1 1 1 2 1 3 1 4

• Technical Efficiency in the bus sector increases through the observation

period (1990-2004) under the CRS assumption

• 1995 = 0,39
• 2004 = 0,42
• Results can be confirmed by the VRS and FDH assumption

• 21 -

EE²

Comparison – Seat Utilization

DEA Model 1-Dea Model 3

• 80,00%
• 60,00%
• 40,00%
• 20,00%

0,00% 20,00% 40,00% 60,00% 80,00% 1 175 349 523 697 871 1045 1219 1393 1567 1741 1915 2089 2263 2437

• Technical Efficiency in the bus sector higher when we define seat km as
• utput
• inefficient use of seats (Auslastungsgrad!!)

• 22 -

EE²

Test of Robustness

• Correlation analysis (deterministic nonparametric DEA and stochastic

Order-m estimation) 0,78

• FDH also confirms efficiency ranking of the most efficiency bus companies
• Bootstrapping: Mean Bias 0,19, Mean Variance 0,013.

• 23 -

EE²

Conclusions

• Contrary to the international literature, German public transit seems

characterized by increasing scale economies

• Uncertainty about the large enterprises
• Scale efficiency does not seem to have changed much recently
• This would imply high pressure on mergers and acquisitions (in fact, this is

what one observes in reality)

• Extension with cost function necessary; more research necessary on

allocative issues and cost functions

• 24 -

EE²

Efficiency Analysis of German Public Transit – Is Big Beautiful?

Christian von Hirschhausen and Astrid Cullmann

5th INFRADAY Berlin 07.10. 2006

Dresden University of Technology, Energy Economics and Public Sector Management, and DIW Berlin

EE²