PRODUCTIVITY CHANGE IN BANKING: A COMPARISON OF PARAMETRIC AND NON-PARAMETRIC APPROACHES

1. PRODUCTIVITY CHANGE IN BANKING: A COMPARISON OF PARAMETRIC AND NON-PARAMETRIC APPROACHES Barbara Casu* Claudia Girardone Philip Molyneux Correspondence to: Department of Economics, The University of Reading – b.casu@reading.ac.uk

2. AIMS OF THE STUDY • To compare parametric and non-parametric estimates of productivity change in European banking during the 1990s. • To extend the comparison to the decomposition of productivity change

3. LITERATURE ON PRODUCTIVITY GROWTH IN BANKING • NON-PARAMETRIC STUDIES • Based on the ideas of Malmquist (1953) and the distance function approach. Cave et al. (1982) provide the theoretical framework for the measurement of productivity. • PARAMETRIC STUDIES • Generally represent technical change by including a time trend in the estimated cost or profit functions.

4. LITERATURE ON PRODUCTIVITY GROWTH IN BANKING • MALMQUIST PRODUCTIVITY INDEX

5. LITERATURE ON PRODUCTIVITY GROWTH IN BANKING • PARAMETRIC APPROACH

6. LITERATURE ON PRODUCTIVITY GROWTH IN BANKING • PARAMETRIC APPROACH • All parametric studies revised use variations of the time trend approach • Berger and Mester (1999, 2001) introduce a decomposition of total cost changes into a portion due to a change in business conditions and a portion due to changes in productivity. Productivity change is then decomposed into the change in best practice and change in inefficiency components. • The value of the decomposition is that it provides information on the sources of the overall productivity change.

7. METHODOLOGY • MALMQUIST TFP INDEX Several different methods can be used to compute the distance functions which compose the Malmquist TFP index. We follow the DEA-like method suggested by Fare et al.(1994). Efficiency Change Technical Change

8. METHODOLOGY • PRODUCTIVITY CHANGE FROM A DECOMPSITION OF COST CHANGES The cost of theindustry at time t is represented by the predicted costs of a bank with average business conditions, average inefficiency for the period and zero random error. The total gross change in cost between period t and period t+k is measured by the ratio of the predicted costs in the two periods.

9. METHODOLOGY • PRODUCTIVITY CHANGE FROM A DECOMPSITION OF COST CHANGES ΔTOTALC is decomposed into the gross changes in best practice, inefficiency and business conditions. ΔPROD C t, t+k

10. METHODOLOGY • The study employs a standard translog functional form:

11. DATA Annual information for a balanced panel of over 2000 EU large banks (>€450 million) during 1994-2000. Estimations are carried out on individual banking markets (France, Germany, Italy, Spain and UK). INPUTS AND OUTPUTS DEFINITON INPUTS: AVERAGE PRICE OF LABOUR , DEPOSITS AND CAPITAL OUTPUTS: TOTAL LOANS, OTHER EARNING ASSETS, OFF-BALANCE SHEET ACTIVITIES

12. RESULTS – MALMQUIST TFP M0 < 1 INDICATES DECLINE; M0 > 1 INDICATES GROWTH

13. RESULTS - TOTAL COSTS DECOMPOSITION ΔPROD < 1 INDICATES GROWTH; ΔPROD > 1 INDICATES DECLINE

14. PRODUCTIVITY CHANGE TRENDS

15. PRODUCTIVITY CHANGE DECOMPOSITION Note: For the Malmquist FTP Index decomposition [technological change (TC) and efficiency change (TEC)], a number <1 indicates decline; a number >1 indicates growth. For the Cost Productivity estimate decomposition [change in best practice (BESTPR) and change in inefficiency (INEFF)], a number >1 indicates an adverse shift toward higher costs; a number <1 indicates a favourable shift. We would therefore expect mirror trends in the two graphs.

16. RESULTS • Both approaches consistently identify those systems that have benefited most (and least) from productivity change during the 1990s. • Results also suggest that (where found) productivity growth has mainly been brought about by improvements in the performance in the best practice banks. • There does not appear to have been catch-up by non best-practice institutions. • Both approaches reveal a decreasing trend in the performance of best-practice banks towards the end of the 1990s.

17. CONCLUSIONS In general the findings reported in this study illustrate that different methodologies can help corroborate productivity change estimates for European banking.