Forthcoming in Journal of Banking & Finance Sources of Bank Productivity Growth in China: A Disaggregation View Tzu-Pu Chang Jin-Li Hu (http://jinlihu.tripod.com) Ray Yutien Chou Lei Sun July 15, 2011 Outlines Introduction Methodology Empirical results Conclusions 2 Introduction (1/4) In the past three decades, the Chinese banking system has reformed gradually and gained remarkable successes in many respects. The total assets of Chinese banking industry have been more than 60 trillion RMB and increased about 300 times than that in 1978. On November, 2009, the capital adequacy ratio and the provision coverage of Chinese banking industry has been larger than 10% and 150%, respectively. Industrial and Commercial Bank of China (ICBC), China Construction Bank (CCB), and Bank of China (BOC) are the largest three listed banks in the world. 3 Introduction (2/4) The financial reforms also make efficiency and productivity improvements in the banking sector (Chen et al., 2005; Matthews et al., 2009). This application aims to investigate how the total-factor productivity (TFP) changes and to disaggregate the sources of productivity change in Chinese banking industry from 2005 to 2009. It is worthy that the ‘Big Four’ state-owned banks (SOBs) have partially privatized to take on minority foreign ownership since 2005. 4 Introduction (3/4) Previous studies focused on productivity change in banking typically adopt the Malmquist TFP index or Luenberger TFP index approach. However, these indices are aggregative indices, meaning that it might lack some insights if we want to see the change of each factor. 5 Introduction (4/4) Therefore, this paper tries to overcome the disadvantage of total factor productivity index and traditional partial productivity measures . This paper proposes a total-factor input productivity index (TIPI) to deal with the above concerned. The productivity growth of each factor under total factor concerns can be calculated. 6 Methodology (1/4) We first assumed that the production technology Ft models the transformation of multiple t M inputs, x R , into multiple outputs, yt RS , for each time period t The Luenberger productivity index relies on directional distance functions. Following Chambers et al. (1998), the directional distance functions could be defined at t as: D(t ) (xt , yt ; g x , g y ) max{ : (xt g x , yt g y ) Ft }. S M R R where (gx, gy) is a nonzero vector in × . 7 Methodology (2/4) The Luenberger productivity index is measured as follows: L(xt 1 , y t 1 , xt , y t ) 1 D (t ) (xt , y t ) D (t ) (xt 1 , y t 1 ) 2 D (t 1) (xt , y t ) D (t 1) (xt 1 , y t 1 ) , if the Luenberger productivity index is less than, equal to, or greater than zero, then it respectively stand for productivity regress, no change, or progress between period t and t+1. 8 Methodology (3/4) Briec (2000) introduces a Färe-Lovell efficiency measure that has the advantage to select a strong efficient vector onto the frontier. 1 D (t ) (x , y ) max ( 1 M t t M ) N s.t. t t x x j ij io (1 i ), i 1,..., M , j 1 N t t y y j rj ro , r 1,..., S , j 1 j 0, i 0, j 1,..., N ; i 1,..., M ; r 1,..., S . 9 Methodology (4/4) The TFP change can be decomposed into the productivity change of the M individual inputs as follows: 1 TFPCH TIPI1 TIPI 2 TIPI M M 1 1 EFFCH1 EFFCH M TECHCH1 TECHCH M M M EFFCH TECHCH 10 Data and variables’ descriptions Based on intermediation approach, this article specifies two outputs and three inputs to investigate the total-factor input productivity change of banks in China. The output variables include total loans, and other earning assets. In input variables, labors, capitals, and funds are the conventional inputs in previous researches. Funds defined as the total deposits; capital is measured by the total fixed assets; labor is the total number of employees of a bank. 11 Data and variables’ descriptions This application collects a balanced panel data covering 2005-2009 from 21 Chinese commercial banks, including Big Four state-owned banks, national shareholding commercial banks, and major city commercial banks in China. The financial data, including the items of the balance sheets and income statements, are taken from Bankscope. The information on the numbers of employees is quite incomplete in Bankscope. Therefore, this variable is complemented through each bank’s annual report. All nominal prices are transferred using GDP deflator with 2009 as the base year. 12 13 Productivity analysis at the industry level 14 Figure 1 Cumulative changes of TFP and its components 15 Figure 2 Annually change of TFP and total-factor input productivity 16 17 Productivity analysis at the group level 18 Note: SOB is state-owned banks; JSB is joint-stock banks; CCB is city commercial banks. 19 Change of total-factor input productivity 20 Empirical results - firm level Innovators: 5 banks; only Bank of Beijing (#2) is the innovator which shifts the frontiers of all inputs. Negative growth of TFP: Bank of China TFP growth is driven by efficiency improvement: 2 banks TFP growth is driven by technical progress: 4 banks Technological gains transcend the efficiency regressions and results in TFP growth: 9 banks 21 Conclusion There’re two advantages of TIPI TIPI can calculate productivity change of particular input factor under total-factor framework TIPI measures TFP change as the arithmetic mean of the productivity change of each input Thus, we can find out what is the main driver for TFP growth In China’s bank industry, the TFP gains are principally driven by technical progress, especially for capital usage. 22 The End 23