Chair of Energy Economics and Public Sector Management Outsourcing in Local Public Transport: A Hidden Efficiency Determinant? Matthias Walter 15 May 2009 Motivation Data Facts Historically Organized Market Characterized by the Substantial Need for Transfers Several Measures to Increase Efficiency • High fragmentation of the market: several 100 operators • Unbalanced panel: 254 observations for 39 multioutput companies from 1997 until 2006 • Mergers and acquisitions, especially for companies operating on a network with connecting lines, e.g. Mannheim, Heidelberg and Ludwigshafen • Integrated operators of bus and tram or light railway or metro in medium-sized and larger cities • Physical data from VDV statistics, monetary data from annual reports • Monetary data in 2006 prices inflated by the German producer-priceindex (Destatis, 2008) • Tenders: so far only for bus services in Hesse and for regional bus services around Hamburg and Munich, maybe in the future also for other services • Very low level of cost coverage with mean of 73.8% (Verband Deutscher Verkehrsunternehmen, 2008) • Capital price including material costs, other operating expenses, depreciations, interests on borrowed capital and the opportunity cost of capital (equity base x interest rates for corporate bonds (Deutsche Bundesbank (2007) plus 2% risk premium) • Efficiency Analysis • for incentive regulation • Mostly municipal ownership with high degree of political intervention (Public service obligation, decision over new lines, …) Descriptive Statistics • for sunshine regulation (“naming and shaming”) Outsourcing (part of material costs) moves personnel costs to 3rd parties Data Correlations ID = (population in the supplied area) / (bus line and length rail-bound track length) not dependent on congestion Methodology Translog Cost Function Function Design Heteroscedastic Stochastic Cost Frontier Models • Local point of approximation: Mean • Use of random effects model to exploit the panel data structure and because of low within variation (at most 6% based on overall variation for costs, outputs and the remaining factor price) • Time dummies with neutral technical change following Farsi et al. (2005) and Farsi and Filippini (2009) • 1) Random Effects Model (ML estimation) as suggested by Pitt and Lee (1981): • No reliable results with linear time trend and time trend varying with input and output levels (Saal et al., 2007) • Heteroscedastic inefficiency determinants (ID, UR), revealing management quality following Bhattacharyya et al. (1995), Hadri et al. (2003) and Greene (2007): , • 2) True Random Effects Model (Simulated ML) based on Greene, 2004 and 2005: • 3) Random Parameter Model allowing for heterogeneity in the outputs • Heteroscedastic inefficiency term for the Random Parameter Models as for the Random Effects Model, except for the time-variant inefficiency determinants: Results Regression Results Descriptive Efficiencies Significant modeling results for unobserved heterogeneity and heterogeneous output characteristics Efficiency Rank Correlations Significant cost decreases over 10 years (up to 25%) Conclusions • Optimization of outsourcing as positive efficiency determinant should be in focus for the industry • Vast differences in the vehicle utilization rate for railcars determines efficiency predictions Tests on Variable Specification • Wald Tests on Random Parameter Models • LR-Tests on Random Effects Model - Model w UR better than model w/o OUT & UR (p = 0.014) - Model w OUT & UR better than model w OUT only (p = 0.000 each) - Model w UR better than model w OUT & UR (q = 0.009) - Model w OUT & UR better than model w UR only (p = 0.073, 0.067) - Improvement options can be related to enhancing speed through infrastructure measures (separate rail embankments, prioritization at traffic lights, tunnels in inner-city areas, new tracks, express trains, etc.) - Furthermore maintenance times could be reduced and procurement optimized • Profit and cost efficiency - Relatively high mean efficiencies suggest that the problem is likely to be not only on the cost side, where improvements through wage reductions have happened in the past Kernel Density of Efficiency Predictions - Cost saving potential for the dataset: 1.40 - 4.43 bn EUR based on 28.23 bn EUR total costs (in 2006 prices) - The revenue side should bear further optimization potential and should be analyzed in the future 16 • Negative skewness of all curves 14 • Similar distributions for True Random Effects and Random Parameter Models Kernel Density Estimate 12 10 • Bimodal distribution in the Random Effects Model against the expectation, Farsi and Filippinis’ (2009) explanation: “cost differences that are not due to inefficiencies but to other external factors” 8 6 4 2 0 0.5 0.6 0.7 0.8 0.9 1 Efficiency Random Effects Model True Random Effects Model Random Parameter Model 1.1 References (selected) Bhattacharyya, A., Kumbhakar, S., Bhattacharyya, A., 1995. Ownership structure and cost efficiency: A study of publicly owned passenger-bus transportation companies in India. Journal of Productivity Analysis 6 (1), 47–61. Destatis, 2008. Preise und Preisindizes für gewerbliche Produkte (Erzeugerpreise) – Juni 2008. Statistisches Bundesamt Fachserie 17 Reihe 2. Wiesbaden. Deutsche Bundesbank, 2007. Monatsbericht August 2007. On the Internet: http://www.bundesbank.de/download/volkswirtschaft/monatsberichte/2007/200708mb_bbk.pdf, retrieved 27 April 2009. Farsi, M., Filippini, M., Greene, W., 2005. Efficiency measurement in network industries: Application to the Swiss railway companies. Journal of Regulatory Economics 28 (1), 69–90. Farsi, M., Filippini, M., 2009. An analysis of cost efficiency in Swiss multi-utilities. Energy Economics 31 (2), 306–315. Greene, W., 2007. Limdep Version 9.0 Reference Guide. Econometric Software, Plainview. Greene, W. H., 2005. Reconsidering heterogeneity in panel data estimators of the Stochastic Frontier model. Journal of Econometrics 126 (2), 269–303. Greene, W., 2004. Distinguishing between heterogeneity and inefficiency: Stochastic Frontier Analysis of the World Health Organization’s panel data on national health care systems. Health Economics 13 (10), 959–980. Hadri, K., Guermat, C., Whittaker, J., 2003. Estimation of technical inefficiency effects using panel data and doubly heteroscedastic stochastic production frontiers. Empirical Economics 28 (1), 203–222. Pitt, M., Lee, L., 1981. The measurement and sources of technical inefficiency in Indonesian weaving industry. Journal of Development Economics 9 (1), 43–64. Saal, D. S., Parker, D., Weyman Jones, T., 2007. Determining the contribution of technical change, efficiency change and scale change to productivity growth in the privatized English and Welsh water and sewerage industry: 1985-2000. Journal of Productivity Analysis 28 (1-2), 127–139. Verband Deutscher Verkehrsunternehmen, 2008. VDV Statistik 2007. Köln.