ebook img

the greek banking case PDF

436 Pages·2014·4.04 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview the greek banking case

1 CARDIFF UNIVERSITY ESSAYS ON EFFICIENCY AND PRODUCTIVITY: THE GREEK BANKING CASE by Panagiotis Tziogkidis A Thesis Submitted in Fulfilment of the Requirements for the Degree of Doctor of Philosophy of Cardiff University 2014 E C O N O M I C S D E P A R T M E N T , C A R D I F F B U S I N E S S S C H O O L 2 This page was intentionally left blank 3 DECLARATION This work has not previously been accepted in substance for any degree and is not concurrently submitted in candidature for any degree. Signed …………………………………………………………. (candidate) Date ………………………… STATEMENT 1 This thesis is being submitted in partial fulfillment of the requirements for the degree of …………………………(insert MCh, Md, MPhil, PhD etc, as appropriate) Signed …………………………………………………………. (candidate) Date ………………………… STATEMENT 2 This thesis is the result of my own independent work/investigation, except where otherwise stated. Other sources are acknowledged by footnotes giving explicit references. Signed …………………………………………………………. (candidate) Date ………………………… STATEMENT 3 I hereby give consent for my thesis, if accepted, to be available for photocopying and for inter- library loan, and for the title and summary to be made available to outside organisations. Signed …………………………………………………………. (candidate) Date ………………………… 4 Abstract Bootstrap DEA is a valuable tool for gauging the sensitivity of DEA scores towards sampling variations, hence allowing for statistical inference. However, it is associated with generous assumptions while evidence on its performance is limited. This thesis begins with the evaluation of the performance of bootstrap DEA in small samples through a variety of Monte Carlo simulations. The results indicate cases under which bootstrap DEA may underperform and it shown how the violation of the fundamental assumption of equal bootstrap and DEA biases may affect confidence intervals and cause the evidenced underperformance. An alternative approach, which utilises the Pearson system random number generator, seems to perform well towards this respect. In particular, coverage probabilities converge to the nominal ones for samples as small as 120 observations and the bootstrap biases are very close to the DEA ones. In the presence of technological heterogeneity, though, poor performance is observed in all cases, which is not surprising as even the applicability of simple DEA is questionable. Using an illustrative example from the deregulation of the Greek banking sector during late 80s, potential differences arising from the various approaches are discussed. In particular, the theoretical explorations are extended to the case of the Global Malmquist productivity index, which is used to examine the productivity change of Greek banks during (de)regulation. Some differences are observed on the magnitudes of the estimated quantities of interest and on the probability masses at the tails of the relevant bootstrap distributions. Qualitatively, though, the overall conclusions are very similar; the provision of commercial freedoms enhanced the productivity of commercial banks whereas the imposition of prudential controls had the opposite effect. This result is of topical interest as the European Supervisory Mechanism, which recently assumed duties, will closely supervise “significant institutions” which includes the 4 biggest Greek banks and their banking subsidiaries. Keywords: efficiency, productivity, DEA, bootstrap DEA, Global Malmquist index, hypothesis testing, Monte Carlo simulations, banking, deregulation JEL Classifications: C12, C14, C15, C61, C67, G21, G28, L25 5 Acknowledgements I would like to thank my supervisors Prof Kent Matthews and Prof Patrick Minford for their support and guidance throughout the PhD, as well as the two examiners. Their comments have certainly improved the structure and the content of the thesis. I would also like to thank the staff and my colleagues at Cardiff Business School for their useful comments on my work presented at Cardiff Economics PhD Workshops. Research funding from the Economic and Social Research Council (ESRC), Cardiff Business School and the Julian Hodge Institute of Applied Macroeconomics is gratefully acknowledged. I have also benefited from comments from various participants at the International Data Envelopment Analysis Society Conference (Thessaloniki, 2011), the Quantitative Economics Doctorate Meeting (Copenhagen, 2011), the Financial Engineering and Banking Society Conference (Paris, 2013) and the European Workshop on Efficiency and Productivity Analysis Workshop (Helsinki, 2013). I would particularly like to thank Prof Leopold Simar and Prof Paul Wilson for their constructive feedback on my paper presented at the EWEPA conference. Their comments have significantly improved the quality of my work while their encouragement for my future research plans is deeply appreciated. Moreover, Prof Mike Tsionas, Prof Robin Sickles and Prof Kris Kerstens have also provided useful suggestions for my work. Thanks also go to Dr Yiannis Kouropalatis who has advised me on presentational and other peripheral, though important, aspects of my work. Special thanks also go to Prof Costas Siriopoulos who has been my mentor since my undergraduate studies at the University of Patras and who has also provided valuable comments and useful advice during my PhD. Last but not least, I would like to thank my family for supporting me all these years. My warmest gratitude, though, is reserved for my partner, Anna Ziouti, who has been patient, caring and understanding throughout my PhD life and to whom I devote this work. 6 List of Abbreviations AEC: Adjusted Efficiency Change AIC: Akaike information criterion AMISE: Asymptotic Mean Integrated Square Error ASE: Athens Stock Exchange BCV: Biased Cross-Validation CRS: Constant Returns to Scale DEA: Data Envelopment Analysis DGP: Data Generating Process DMU: Decision Making Unit DRS: Decreasing Returns to Scale FDH: Free Disposable Hull GAS: Greek Accounting Standards IAS: International Accounting Standards IRS: Increasing Returns to Scale ISE: Integrated Square Error LCV: Likelihood Cross-Validation LSCV: Least Squares Cross-Validation M&As: Mergers and Acquisitions MISE: Mean Integrated Square Error MPSS: Most Productive Scale Size OLS: Ordinary Least Squares RTS: Returns to Scale SFA: Stochastic Frontier Analysis SJPI or SJ: Sheather-Jones Plug-In method SW1998: Simar and Wilson’s (1998) confidence intervals SW2000: Simar and Wilson’s (2000a) confidence intervals TFA: Thick Frontier Approach VRS: Variable Returns to Scale 7 Contents 1 Introduction ............................................................................................................... 19 1.1 Purpose of study and preliminary results .......................................................... 21 1.2 Why Greece? ...................................................................................................... 24 1.3 Motivation and contribution .............................................................................. 27 1.4 Structure of the thesis ........................................................................................ 28 2 Small Samples and Bootstrap DEA: a Monte Carlo Analysis ..................................... 30 2.1 Introduction........................................................................................................ 30 2.2 General concepts................................................................................................ 35 2.3 Theoretical foundations ..................................................................................... 39 2.4 Estimation of technical efficiency ...................................................................... 44 2.4.1 Parametric approaches ............................................................................... 44 2.4.2 Non-parametric approaches ....................................................................... 46 2.4.3 Data envelopment analysis ......................................................................... 46 2.4.4 The DEA “estimators” ................................................................................. 51 2.5 General information about the bootstrap ......................................................... 53 2.6 Bootstrapping DEA efficiency scores ................................................................. 56 2.6.1 Bootstrap DEA: a practical consideration ................................................... 57 2.6.2 The Simar and Wilson’s (1998) bootstrap DEA algorithm .......................... 59 2.6.3 Bootstrap DEA: statistical inference and confidence intervals .................. 65 8 2.6.4 On smoothing the empirical distribution ................................................... 68 2.6.5 Developments and extensions .................................................................... 74 2.7 Monte Carlo simulations and previous results on bootstrap DEA ..................... 77 2.8 The Monte Carlo experiments ........................................................................... 82 2.8.1 The experiment outline .............................................................................. 82 2.8.2 The data generating process ....................................................................... 86 2.8.3 The economic interpretation of the DGPs .................................................. 90 2.8.4 Defining the fixed DMU .............................................................................. 92 2.8.5 Performing Monte Carlo simulations and associated issues ...................... 99 2.9 Monte Carlo Results: small samples ................................................................ 104 2.9.1 Identifying the population DGP from the data ......................................... 104 2.9.2 Bootstrap and DEA biases ......................................................................... 109 2.9.3 Coverage probabilities .............................................................................. 115 2.9.4 Bootstrap confidence intervals ................................................................. 120 2.9.5 Bootstrap distributions ............................................................................. 126 2.10 Monte Carlo Results: large samples ................................................................. 131 2.11 Conclusions....................................................................................................... 135 3 Testing hypotheses with bootstrap DEA ................................................................. 139 3.1 Introduction...................................................................................................... 139 3.2 Simar and Wilson’s intervals and implied tests ............................................... 142 9 3.2.1 Simar and Wilson’s (1998) intervals ......................................................... 143 3.2.2 Simar and Wilson’s (1998) implied tests .................................................. 145 3.2.3 Simar and Wilson’s (2000a) intervals ....................................................... 147 3.2.4 Simar and Wilson’s (2000a) implied tests ................................................ 149 3.3 Considerations and limitations ........................................................................ 150 3.3.1 Dealing with skewness .............................................................................. 151 3.3.2 Same-sample comparisons ....................................................................... 153 3.3.3 Cross-sample comparisons ....................................................................... 156 3.4 Can we “bypass” the issue of unequal biases? ................................................ 157 3.5 On testing returns to scale ............................................................................... 163 3.5.1 Measuring RTS in DEA ............................................................................... 163 3.5.2 Simar and Wilson’s (2002) approach of testing RTS ................................. 164 3.5.3 A proposed approach for testing RTS ....................................................... 168 3.6 Conclusions....................................................................................................... 174 4 A simple alternative to smoothing .......................................................................... 178 4.1 Introduction...................................................................................................... 178 4.2 Why use moments? .......................................................................................... 181 4.3 Method of moments ........................................................................................ 183 4.4 Pearson system random number generator .................................................... 184 4.5 The moments-bootstrap DEA ........................................................................... 188 4.6 Monte Carlo evidence ...................................................................................... 190 10 4.6.1 Population, sample and bootstrap moments ........................................... 190 4.6.2 Bootstrap and DEA biases ......................................................................... 193 4.6.3 Coverage probabilities - Small samples .................................................... 196 4.6.4 Confidence intervals ................................................................................. 198 4.7 Conclusions....................................................................................................... 202 5 Suggested guidelines on applying bootstrap DEA ................................................... 206 5.1 Assumptions ..................................................................................................... 206 5.2 Applying bootstrap DEA ................................................................................... 206 5.2.1 Step 1: Identify the underlying population ............................................... 207 5.2.2 Step 2: Enrich the empirical distribution .................................................. 209 5.2.3 Step 3: Apply the bootstrap ...................................................................... 210 5.3 Testing hypotheses........................................................................................... 211 5.3.1 Step 1: Define the null .............................................................................. 211 5.3.2 Step 2: Define the test statistic ................................................................. 212 5.3.3 Step 3: Confidence intervals and p-values ................................................ 212 5.3.4 Step 4: Accept or reject the null ............................................................... 213 6 An illustrative example: the Greek banking case .................................................... 215 6.1 Introduction...................................................................................................... 215 6.2 Contextual background .................................................................................... 218 6.3 Literature Review ............................................................................................. 224

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.