Aims and Objectives
This paper will be examining the scale and the technical efficiencies of 35 large Japanese financial institutions that will be chosen from the different financial sub-sectors, e.g. City banks, Regional Banks, Regional II Banks and Trust Banks, all types of service providers such as Holdings firms and commercial banks and a prominent proportion of them whom have undertaken consolidation activities. They will be selected in terms of their total assets sizes for the period of 2002-2008. During this period, the Japanese banking sector was in the prime of its restructuring and reorganisation brought about from changes in reforms and deregulation in order to deal with the dire state of the economy mainly due to the Asian crisis and its non-performing loans problem. The main impact of these changes was the vast number of mergers and acquisitions by the Japanese institutions. Technical and scale efficiencies are both affected by mergers, therefore this paper will inspect the improvements or worsening in the institutes' technical and scale efficiencies after the optimal year (2001) of mergers taking place and with more reorganisations within the banks, as well as whether they helped Japan cope with the start of the global financial crisis. Therefore it will be interesting to observe whether the major restructuring activities have caused changes to the banks efficiency rates and if these changes helped the institutes in dealing with the current crisis. As well as this, it will tell us what type and size of banks gained or lost the utmost amount of efficiency over the data period. In order to test the efficiencies, the paper will use both parametric, Stochastic Frontier Approach (SFA), and non-parametric, Data Envelopment Analysis (DEA) models; and will compare the results found in both models.
Furthermore, the paper will implement a Tobit second stage regression to omit external environmental factors out of the results (such as those carried out by (F-M. Hsu and C-C Hsueh) in 2008.) This will then give a more accurate indication of efficiencies because the environmental factors, such as GDP growth and land prices, will no longer have an effect on the results.
This paper will be organised in the following manner. Chapter 2 will consist of the changes within the Japanese financial market since the beginning of the Asian Crisis, as well as a literature review of previous studies in this area. Chapter 3 will be the methodology, analysing the models which will be used in this paper. Chapter 4 will give the results of the models and compare the findings. Finally, Chapter 5 will conclude with a summary and implications of the findings.
Chapter 2 - Background of Japanese Banking Industry
2.1 Japan's significant events
Japanese banks were once part of one of the strongest and most sophisticated financial systems in the world until two decades ago, when its system began to crumble. Since 1980s the Japanese banking system has undergone numerous changes. The main causes for the economy's downfall being the bursting of the asset price bubble in the late 1980s, which led to substantial amount of non-performing loans (bad loans) being created, increased competition internationally and the governments poor response in addressing these issues. These resulted in a large number of banks taking part in mergers within the industry in order to survive within the system. This was then worsened by the Asian Financial Crisis in 1997. Just as Japan began to work its way out of its troubles, the Global Financial Crisis in 2007 hit the sector once again. The table 1 below shows the total asset value changes in Japanese banks between 2003-2008.
2.2 Japan's ongoing non-performing loans problem
Over the last two decades, Japan's biggest problem leading to its fall, has been its non-performing loans (NPLs) brought about by a sharp decline in house prices after the asset price bubble bursting in the 1980s. These NPLs not only led to a large part of the financial sectors meltdown, but also the systemic consequences of decline in economic activity, investment and total factor production within the whole economy (L.Barseghyan, 2006), which all led to the contraction of Japan's economy. Along with this came a period of stagnation and deflation, which meant far more NPLs were created. Post-price bubble bursting, it was found that a substantial number of loans were in fact defaulted. Along with this, the Ministry of Finance in Japan were slow to address the issue, delaying the bailout programs which were later put in place. Had the government stepped in sooner, the problem may not have been as severe. For example, the Bank of Japan was slow at dropping the interest rate to help ease the problem. Only in 1995 did they reduce interest rate to 0.5%, but in 1992 the interest rate was still too high at 4%, showing the slow reaction of the government. But once the Bank of Japan (BoJ) finally reduced this interest rate it has remained close to zero since for a number of years.
The write-off of NPLs only really became apparent in 1995, when Sumitomo Bank finally took control as the first bank to start writing down their NPLs (K.E.Uerr and G.Miyazaki, 1999). After this, other banks followed suit, leading to major restructuring of Japan's banking sector. A substantial number of NPLs were disposed of while Heizo Takenaka was heading the FSA with his financial Revival program in 2001. Over 10 years on, Japanese banks have disposed of over 90 trillion yen worth of NPLs (Bank of Japan, 2002). Which were estimated to be 7% of its GDP (Anderson & Campbell, 2000). Figure 2 on the following page shows the decline of NPLs over the years for major banks and regional banks; it can be seen that 2001 onwards there has been a substantial fall, but there is still more that can be done.
2.3 Overview of developments in the Japanese banking sector
The systemic Asian financial crisis in 1997 and undesirable economic conditions that it brought to Japan, along with the problems of NPLs, came to a head in this period. The worst affect on Japan's financial sector was some of its largest banks becoming insolvent, such as the failure of its largest securities firm, Yamaichi Securities and its tenth largest bank at the time, Hokkaido Takushoku. At this time the government allowed banks to set up financial holding groups, and increased competition meant that a wave of mergers and consolidations between Japanese banks increased substantially in order for the banks to become more competitive and survive. This wave of consolidation led to the creation of Japans fives major financial groups. Along which, this the government finally began to put better reforms in place to help deal with a lot of the problems, such as loan loss provision tightening and prompt corrective action scheme as well as the nationalisation of a number of banks, such as Long-term credit bank and Nippon credit bank. This led to the restructuring of the Japanese financial system. This all contributed to Japan heading in the right direction to getting itself out of the problems of the past decade. During this time the ‘Big Bang' package was implemented to help reorganise the sector with a 5 year plan, consisting, among other factors, increased number of mergers within Japans financial market.
In 2001 Japans banking sector began to really see great improvements with new reforms being put into place. It was thought this was partly due to the Financial Revival plans put into place by the head of the FSA at the time, which substantially reduced the number of NPLs. With the number of mergers in Japan becoming overwhelming due to the reforms, making it easier for banks to integrate.
The years to follow saw Japans economy recover and the financial sector moving from strength to strength. In 2006 Japan finally no longer faced deflationary rates. With positive income for banks for the first time in years Japan began finally on the right track to recovery.
The Global crisis of 2007 put all the world's economies on halt, including Japan. Although the impacts felt in Japan are not as strong as other countries, they still find that their economic growth becoming sluggish, stock prices plummeting and the comeback of Japan's deflation. The final impacts on Japans financial market from the crisis are not yet apparent at this early stage, but the government has stepped in early this time to resolve the negative impacts as soon as possible.
2.4 Restructuring and Reorganising Japan
In 2000/01 came a time of major reorganisation and restructuring changes, brought about by the ‘Big Bang' and ‘Financial Revival' packages designed by the MoF. The objectives of which were to dispose of the NPLs and to restructure the securities sector. The main way this was done was through deregulation and changes in reforms to open the market. As a result there was a period of substantial mergers and integration within Japan's banking sector, which were carried out by the FSA. Overall there were an estimated 900 mergers in Japan from 1998 to 2001 (Anderson et al, 2000:200). The changes in the number of banks in each sector between 2002 to 2008 can be seen in Table 2:
The biggest of these mergers was the creation of Japans major banks and financial groups, known as the ‘Mega Banking Groups'. The largest of these were; Mizuho Financial Group, Resona Holdings, Mitsubishi UFJ Financial Group and Chuo Mitsui Trust Holdings.
The wave of mergers carried on through from 2002 along with the packages being implemented into the market. At this time Japans economy began to recover from the crisis and showed signs of the worst of the impact being behind Japan. The first half of 2002 still saw the Japanese economy struggling, this is mainly due to the short-run effects of the restructuring of banks and the uncertainty this brought with it. Along with this unemployment increased not only from banks cutting workforce, but also ongoing bankruptcies. Although in the long-run the changes of the packages outweighed these short-run negative implementation effects, in 2002 the negative effect was felt at the beginning of that year. By the end of 2002 the economy began to turn more favourable, with increased asset prices, export, consumption and economic growth increased overall by 0.26% (World Bank-World Development Indicators, 2009). 2003 and 2004 saw steady pace of recovery in the first time in years. Furthermore, stock prices increased which also showed the economy was finally in its recovery period. The disposal of NPLs was still increasing, which helped the banks positions as well as the economy's growth. The government also set up the ‘Industrial Revitalization Corporation of Japan' (IRCJ) in 2003, which had the objective to assist with the restructuring and disposal of NPLs within the banks. Most of Japan was feeling optimistic about its new position, except a few regional banks, who were not feeling as optimistic. On 29th November 2003 it was announced, after years of injecting capital, that the government would undertake another nationalisation of a bank, which was Ashikaga Bank, Japans tenth largest regional bank. The bank was unable to once again manage its bad loans and weak balance sheets. This time the temporary nationalisation caused depositors to be protected as usual, but the banks equities were cleared, which meant shareholders lost out this time. But overall these years show that the implementation of the packages and mergers, the early impacts of them and signs of recovery in the economy, with the exceptions of a few banks still feeling the strain.
2005 was yet another successful year for the country, with increased economic growth and public confidence returning, banks began to see positive incomes for the first time in years. This could have been due to the disposal of a vast amount of bad loans by the banks in the recent years, and still reducing in 2005. Another reason for this could have been that banks were still restructuring and looking for ways to lower expenses further so they could become more efficient. Some of the ways the banks were doing this was by reducing the amount of branches they had and cutting their workforce. Although in this period most of the Japanese banks were seeing positive changes within their banks, some banks were still facing difficulties. Regional banks were still not feeling the success like the rest of the market due to the lack of disposal of their bad loans and inadequate capital ratios. Along with this, international competitors were gaining far higher income compared to that of Japanese banks, showing the banks still had a fair way to go in terms of matching international standards. Much of 2006 was the same, with the economy recovering nicely, the banks saw further growth in profits. Furthermore, the recovery also resulted in positive growth for bank loans for the first time in years. Overall it seemed that Japan was on the right track to making a steady recovery.
2.5 Japan and the Global Financial Crisis
In 2007, what started as U.S. sub-prime loans defaulting turned into a catastrophic Global Financial Crisis, which could be felt in Japan's economy. Having spent years disposing its NPLs, Japanese banks now faces the prospect of enduring new defaulting loans due to the current crisis in the following few years. For the early part of the crisis the impact was not felt strongly, with Japans banks still recording positive income growth; it was only at the end of the year the world began to feel the impacts. Early 2008 there were substantial losses made on securities around the world, and the global economy came to a halt. The impacts at this time were spread unevenly around the world, with places such as U.S. and Europe feeling the impacts substantially more than the likes of Japan. Japans economy was less affected and their economy still relatively stable, with their NPLs still reducing, but at a slower rate. Although it posted negative economic activity in 2008, seen in Figure 4, one of the reasons for this was the decrease in its exports due to the decline in demand overseas as other countries feel the effects of the crisis.
Also Japan's credit risk became a problem for the country, after the failure of Lehman Brothers on 15th September 2008 in U.S, Japan's stock exchange plummeted yet again (Japan Financial System Report (JFSR), 2009:7).
This has increased the losses on Japan's securities further and capital adequacy ratios remaining low. After the news of Lehman Brothers, days later Nomura Holdings announced its plans to take over Lehman Brothers offices in the Asia Pacific area. Furthermore, the following day they also announced their takeover plans of the banks in Europe and Middle Eastern equities and investment operations as well (Bank of Japan, 2009). Mitsubishi UFJ Financial Group announced at the same time that they will take over the common stocks of Morgan Stanley. By the end of 2008, Japanese banks really started to feel the losses from their securities. The losses from securities for large banks estimated at 5.3 trillion Yen and losses for regional banks at 2.7 trillion Yen (JFSR,2009:6) . This time the government have acted promptly to help Japan get out of the crisis by itself. Interest rates had been cut to 0.1% and are thought to stay there for awhile (The Economist, 2009). The government are also setting up a large fiscal stimulus emergency package with their 2009 budget to help ease the economy in the crisis (JFSR, 2009). BoJ also faced a blow in 2009, after ending years of deflation in 2006, it returned to Japan. The ability of individual banks to stay afloat during this period depends on their capital buffers put in place by bank management and the state of their credit portfolio. The sector has overcome most of its NPLs problems, but it is not sure if this issue will again surface in the near future. Japan's banking sector faces further uncertainty in the near future much like the rest of the world, although this will test the strength of how successful the restructuring taken place in Japan's financial system has been. Overall the impact of the crisis has not been felt by Japan as much as many other countries as of yet.
(Table 3) shows the different inputs and outputs used by each author and a brief summary of their findings from the Literature Review below.
Table 3 Summary of Literature Review
Over the years there have been a number of different approaches to estimate the efficiency of different banking sectors. Drake, Hall and Simper (2009) conducted a comparative analysis of non-parametric methods by using both the intermediation and production approaches, along with a profit/revenue-based approach, as proposed by Berger and Mester (2003), in the slacks-based measure for DEA, to calculate the efficiency of the Japanese Banking system. Liu and Tone (2006) also used DEA and SFA to conduct a multi-stage regression, but went one stage further to carry out the second stage regression to omit external environmental factors which could affect efficiency. In previous studies before, Drake and Hall (2003) examined the efficiency of Japanese banking using only the non-parametric technique DEA. Fukuama and Weber (2002) uses quasi-distance functions to measure allocative efficiency and productivity change for Japanese banks during this period 1992-1996. What made this study different was that it compared the use of Farrell radial efficiency measure (which includes slacks onto the input constraint) and Russell non-redial efficiency measure (which does not allow slacks into the input constraint). The results showed that the banks efficiency rate was greater when using the Farrell measure and both measures showed a reduction in productivity growth within the period of about 2% each year. A different approach other than DEA or SFA was carried out by McKillop et al (1996) using four cost function models, which are composite, separable quadratic, generalised translog and translog of the five largest Japanese banks (which were all city banks), between 1978-91, in which all the models showed the banks having scale efficiencies. In contrast, Drake, Hall and Simper (2009) ran three different methods on 1109 banks between the period of 1995 to 2002, which all showed different results for efficiency rates. The intermediation approach showed that efficiency was at its highest in 1996 at 77.60 and at its lowest in 1998 at 67.46. For the production method the highest efficiency score was in 2002 at 33.40 and lowest in 1999 at 24.47. Finally profit/revenue approach showed that efficiency peaked in 2002 at 68.52 and slumped to its lowest in 1998 at 55.16. The efficiency results varied widely between year to year within each method which implies that conclusions on efficiency could be biased depending on the methodology used.
Fukuyama and Weber (2008) suggest that non-performing loans are more accurately modelled as an undesirable or bad output, rather than as an input. They use the directional output function of Fare et al (2005) which studied the bad effects of pollution as an output for environmental efficiency so they adapted it to apply to non-performing loans. It is quite an appropriate approach considering the well-known problems of NPLs in Japanese banking. To contrast, for example Altunbas et al (2000) use RQCF cost function which takes into account risk and quality factors as they suggest previous efficiencies for Japanese banks had been overstated as they did not include quality and risk factors (such as NPLs) into account when measuring the banks efficiencies, this is particularly important in the case of Japan given their asset quality and capitalisation problems since 1990s. Although, it should also be noted that Berger and Humphrey (1997) argue that it is sometimes not needed to account for NPL that are shown through risk and quality measures. They argue that if the dominating explanation is ‘bad luck' then problem loans should be accounted for, but if the dominating explanation for the relationship between NPLs and efficiency is ‘bad management' then they shouldn't be controlled when measuring the efficiency of banks.
The different studies in Japanese banking not only use varied models, but also adopt different approaches of choosing inputs and outputs. The table already outlines the inputs and outputs used by the different studies which shows similarities but in particular there is a notable difference in the treatment of how bad loans are included for example, Azad (2009) uses NPLs as an input i.e. cost, in contrast to Fukuyama and Weber (2008) who states it as a ‘bad' output. Furthermore, total provisions included as an input for all approaches by Drake, Hall and Simper (2009) consists of loan loss provisions which can be a proxy of NPLs. It should be noted that only the more recent studies appear to explicitly take the NPLs into account which suggests that the previous studies such as McKillop et al (1996) could have overestimated the efficiency of Japanese banking. Another difference that can be observed is that deposits can be treated as an input such as in the intermediation approach or an output in the production approach. This is significant because Drake, Hall and Simper (2009) found that the intermediation approach almost always produces the highest relative efficiency scores while the production methodology nearly always produces the lowest scores, which is the reverse of what Tortosa-Ausina (2002) find in their results. The different inputs and outputs could explain the reason why there are varied efficiency results between the studies in Japanese banking.
As a way to start comparing efficiencies, Amel et al. (2004) reviewed the international evidence of mergers and acquisitions and efficiency in the financial sector along industry and country lines in order to shed light on common features and understand the main features. It attempted to reach a level of generality by covering most industrialised. They noted that difference in regulations, institutions and market structure across countries meant that conclusions drawn from the analysis of one country should be generalised to others only very carefully. In particular for Japan they found that the gap between the best and the average practice institution (the average cost inefficiency) was around 5-7% which means that consolidations were likely to bring about smaller efficiency improvements in Japanese banks compared to banking systems with a higher dispersion of efficiency scores. They suggested that after the crisis of the 1990s Japanese banks operated closer to their efficient frontier because the gap had reduced from around 14% in the 1980s. They also found that consolidation in the financial sector was beneficial up to a relatively small size, but there was little evidence that mergers yielded economies of scope or gains in managerial efficiency. This can be compared to the findings of Azad (2009) who found that for technical efficiency, pure technical efficiency and scale efficiency the average scores were 0.83, 0.92 and 0.90 respectively for 73 financial institutions in the years 2002, 2003, 2005 and 2006. Their results showed that changes in the policies had positive effects on the efficiency of Japanese banks. From 2003 to 2006 their results for TE increased from 82% to 85% which shows an improvement for the financial institutions. Although during this period it was found that scale efficiency remained the same at 90% which suggests that mergers may not improve efficiency for the financial institutions; or could be explained because the effects of the mergers have not yet been felt. This appears to be supported by Tadesse (2005) who found that there were diseconomies of scale in large banks, which challenged the need for consolidation of the Japanese banking system. At the same time there was underlying technological progress which produced economies in larger banks hence justified the ongoing trend in consolidation of Japanese banks. Instead of testing the impact of mergers and acquisitions, Montgomery and Shimizutani (2007) use panel data to calculate whether the Japanese government capital injections into banks were effective or not. Their results show that the first round of capital injections in 1997 did not meet their policy objectives whereas in 1998 they were more successful at improving the banks' positions by increasing bank capital ratios and increasing the write-offs of NPLs.
Drake and Hall (2003) examined the efficiency of Japanese banking with regard to the size efficiencies and the efficiencies between the sub-sectors within the Japanese Banking Industry, using the non-parametric technique DEA with 149 banks in 1997. They also examined efficiencies with and without controlling problem loans within the model. The results found that within the sub-sectors trust banks and Long-term credit banks had the best efficiency rates for both scale and technical efficiencies with 100% efficiency levels before and after taking account of risk and quality factors. The high efficiency rates may be due to specialisation within these two sub-sectors, it is therefore thought that scale efficiency could be strongly influenced by bank types. Most other banks showed increased returns, although large City banks shows decreasing returns with mean SE of 91.27. Overall it was found that as the bank size increased so did the scale inefficiency. In relation to size efficiency the results show that scale efficiency potential is greater in smaller banks. With technical efficiency it was found that larger banks were most efficient, with a mean OE (overall efficiency) of 88.99 in group 6, than smaller banks, with a mean OE of 68.43 in group 1. Moreover, with the controlling of risk factors it is found that the mean of pure technical efficiency for all banks increases from 78.11 to 89.38 and the mean of scale efficiency increases from 92.78 to 96.59. Finally the results show that small banks will gain more from X-efficiency by mergers than large banks. This study also shows a contradiction with Altunbas et al (2000) findings in that it shows technical efficiency as being more sensitive to the addition of risk factors than scale efficiency, although both agree that scale efficiency is overestimated when risk factors are not included.
Uchida and Satake (2009) investigated whether depositors have an influence in the way that banks are managed, therefore how important depositors are as a market discipline. The data used consists of both city and regional banks in the period of 2000-2005. There are 692 bank observations within the model. They use cost efficiencies and profit efficiencies to test their hypothesis. The results for cost efficiency show that more depositors within a bank means the bank will be more cost efficient. This is thought to be because the depositors put pressure on bank management, particularly on the inputs they use. On the other hand the profit-efficiency results show that more outstanding deposits do not mean the bank is more cost efficient. But the models show that the result for cost efficiency is more significant than those of profit efficiency in this case. This means that in this study it was found that depositors play a role in disciplining bank management.
Fukuyama and Weber (2008) used a directional output distance function, in which banks are said to be efficient if they can no longer increase their amount of ‘good' outputs and reduce their ‘bad' outputs, while holding inputs constant. They extend Fare et al work by using shadow prices of problem loans instead. In this study both DEA and a parametric linear programming framework are used for the period 2002-2004, with the sample size of Japanese banks ranging from 118-126 between the periods. Their results showed that both methods had similar outcome of bank inefficiencies, although DEA framework showed shadow prices to be 13% of the banks, whether as the alternative method gave the shadow price of 99% of the banks. Therefore overall the study showed that NPLs should be controlled as an undesirable by-product when testing banks' efficiency rates.
A study undertaken by Harimaya (2008) was to test whether there were scale and scope efficiencies for regional banks entering the trust banking sector. The model used a generalized translog function. The study consisted of 64 regional banks between 1994-2003; with an intermediation approach been taken. In the initial results showed the RSE (Ray Scale Economies) of overall mean showing economies of scale increasing over the years. The rate of technical change showed that cost-reducing technological change was found to be 2.07% per year. The cost complementarities showed that there was no cost saving benefits for regional banks entering the trust sector. Having found these results, Harimaya carried out a second test to include bad loans. This is because Japan's bad loan problem could have made the initial results unreliable, therefore the model was re-estimated using revenue as the output measure. The results on the second model showed similar findings to the initial model. As a result, it study showed that there does not seem to be cost reducing advantages for regional banks when entering the trust banking sector, shown by the cost complementarities. Overall, the regional banks exhibit economies of scale, although non-entrants have higher economies of scale than entrants.
Liu and Tone (2006) initial-stage results showed that banks had performance during difficult economy environment conditions, such as 1997/98 and 2000/01, and gave good performance during booms. But the final-stage results, which reduced impacts of exogenous factors and statistic noise on the model, showed a more stable pattern with an increase beginning in 2000. The final results are adequate at explaining credit risk management in Japan, as they invested in IT infrastructure, therefore became more efficient over time which can be seen as the ‘information age' shown in final results. Therefore this paper shows that by using multistage methods to measure efficiency the results become more accurate by omitting the external factors within the data which implies that all previous studies which do not conduct the second stage regression, their efficiency results could be affected by these factors. This is supported by Fried et al. (2002) who argued that poor bank management does not necessarily mean that this is the cause of the bank's inefficiency. Furthermore, they suggest that the banks' inefficiency could be caused by environmental factors which are not within the control of bank mangers. All of the findings in these studies therefore have different significant aspects to take into consideration when choosing what approach to use in the model.