The fourth conference of the Centre’s European Network on the Japanese Economy was held jointly with the Suntory and Toyota International Centre for Economics and Related Disciplines (STICERD) and the Centre for Economic Performance (CEP) at the LSE on 13/14 December 1996. The conference, which was organized by Jenny Corbett (Nissan Institute of Japanese Studies, Oxford, and CEPR), brought together a wide range of research on Japan. Although not thematically organized, the papers concentrated mainly on monetary policy and finance, internal labour markets, and corporate structure and behaviour. In each area, stereotypes of, and conventional wisdoms about, the ‘Japanese model’ were challenged or illuminated. Financial support for the conference was received from STICERD and Mr David Atkinson, Goldman Sachs (Tokyo).

The first paper, by Jun Naqayasu and Ronald MacDonald (Strathclyde University), on ‘Structural Econometric Exchange Rate Models: Japan and Germany’, applied the modern general-to-specific approach to model selection to a model of the exchange rate based on real interest rate differentials. The authors argued that recent econometric techniques might overcome the poor performance of earlier structural exchange-rate models. This would go some way to address the Meese-Rogoff challenge – that a random-walk model outperforms structural models. The paper demonstrated that a simplified real interest-rate model, using relatively few variables, can beat a random-walk model for the Japanese exchange rate, although this is not the case for Germany. The model provides good medium-term forecasts of exchange rates and suggests that, in the long term, purchasing power parity does hold. Discussion centred on the technical merits of the model, on the validity of its predicted values for the yen and the role played by risk premia.

Adrian van Rixtel (Netherlands Central Bank) and Wolter Hassink (Vrije Universiteit Amsterdam) presented ‘Amakudari and the Monitoring of Private Banks. Which Determinants and Has Performance Improved?’. They investigated whether the presence of former Ministry of Finance and Bank of Japan officials on the boards of Japanese private banks was related to the banks’ financial situation. The paper asked whether these appointments were an element of prudential policy or intended to strengthen management via external monitoring.

The authors tested a model of ‘hierarchical delegated monitoring’, using data for 77 banks from 1977 to 1993. They assumed (1) that Amakudari officials retained their loyalty primarily to their former employers, (2) that this determined their incentive structure, and (3) that they represented the interests of small depositors. They also assumed that several levels of monitoring exist: main banks monitor small, non-main banks; non-financial keiretsu firms monitor small keiretsu banks; these, in turn, are monitored by keiretsu main banks; and the MOF and BOJ monitor the major banks. A probit analysis sought to relate the appointment of officials to a number of performance and relationship variables for individual banks.

The key finding was that retired officials are most likely to be appointed to the boards of banks that had a main bank, and to banks with deteriorating solvency. There was also a positive relationship between the appointment of officials and the subsequent development of profitability and lending to risky industries. This casts doubt on the idea that Amakudari are being used as an instrument of prudential policy, i.e. this is not a mechanism for ‘monitoring the monitors’.

Costas Lapavitsas (SOAS), questioned both the assumption that banks monitor other banks with which they are in direct competition, and the conclusion that retired officials remain the instruments of their former ministries. An alternative interpretation might be that banks were using the retired officials to acquire information about ministry policies and attitudes, e.g. in relation to lending to risky industries. Econometrically, he suggested that a model with firm-specific effects would be an obvious next step. Another interpretation of the findings, namely that firms brought in Amakudari as a strategic response to their own poor performances, emerged in discussion.

Robert Aliber (Chicago University, visiting the Bank of England) spoke about ‘The Outlook for Japan as a Creditor Nation’. He raised two questions. First, why had the current account surplus persisted in the 1990s despite the yen’s real appreciation? Second, was the trade surplus likely to continue? Paradoxically, Japan’s capital outflow had persisted despite apparently very low real returns on Japanese investments in the United States. The driving mechanism behind both the capital outflow and the current account surplus was excess domestic savings, the role of which – contrary to the conventional view – had been to create a search for alternative investments, particularly in the United States, because savers had exhausted the attractive domestic investment opportunities. The result had been the capital outflow and the depreciation of the yen, thereby creating the trade surpluses and the matching fiscal balances. Cost-cutting by Japanese firms had kept the real exchange rate low, despite the yen’s nominal appreciation in the 1990s. Eventually, however, low returns on investments in the United States would engender a reversal of the capital flows, creating a real yen appreciation that would slow growth.

In ‘Identifying Monetary Policy in Japan’, Etsuro Shioji (Universitat Pompeu Fabra, Barcelona) used two competing models to describe Japanese monetary policy: the H-model, used by most Japanese scholars, and the BL-model, favoured by the Bank of Japan. In the H-model, high powered money (H) is controlled by the Bank; in the BL-model, the quantity of H is given by private-sector activities, and the Bank can only change the supply of various components, particularly the quantity of Bank loans (BL). The crucial assumption is the increasing marginal cost of BL, occasioned, for example, by closer surveillance by the Bank and loss of reputation in financial markets. This feature allows the Bank of Japan to control inter-bank market interest rates. Shioji showed the appropriateness of a non- recursive ‘structural VAR’ model, and his test results supported only the H-model. Examination of the slope of the supply curve of H also showed that the Japanese monetary policy rule was one of partial accommodation, rather than strict interest-rate or quantity targeting.

Geoffrey Wood (City University Business School, London) noted that Shioji’s method was effective not only for characterizing Japanese monetary policy, but also for resolving the well-known puzzle that an increase in price level sometimes follows tightening of monetary policy. But Wood questioned the use of monthly data as well as the use of import prices as a leading indicator of inflation. He also provided a useful summary of related studies. Yukinobu Kitamura (Bank of Japan) pointed out that data frequency could significantly affect Shioji’s test results, e.g. weekly or daily, instead of monthly, data would favour the BL-model.

In ‘Comparative Analysis of Payment Systems: Implications for BOJ-NET’, Shuji Kobayakawa (University of Oxford) discussed the properties of different payment systems. Two settlement Systems, namely the Designated Time Net Settlement (DTNS) and the Real Time Gross Settlement (RTGS) systems, are in use. The paper showed that DTNS systems are efficient, in that they reduce necessary reserves for transactions, but they expose participants to credit risk. RTGS systems have the reverse properties. In Japan, more than 99% of all transactions are through DTNS schemes whereas the international trend is towards RTGS. The two major RTGS systems are the European type, which requires collateral but no fee, and the US type, which requires a fee but no collateral. The Bank of Japan’s system, BOJ-NET, is to be redesigned along European lines.

Kobayakawa compared both these RTGS types using a game-theoretic model which allowed choice of settlement period. He derived results showing the possibility of delay of payment and the profitability for participants in equilibrium in each type. In the ‘European’ system there is a unique non-delay equilibrium; in the US system, a delay equilibrium is possible depending on the parameter values. The paper proved that either system could be the more profitable depending on parameter values. Richard Dale (Southampton University) described complementary developments in the international debate – spurred by the failure of Barings – about a derivatives clearing house, where the major issues are financial safeguards, regulatory policy issues and international regulatory responses. Dale stressed that measures for enhancing a robust financial-payment structure, rather than strengthening supervision of individual financial institutions, constituted the right direction for policy. Torn Vosa (Bank of England) suggested that Kobayakawa’s model could be usefully extended to explicitly consider the negative externality of systemic risk in the payment system. Treatment of the relationship between participants inside the RTGS system, and those outside it, was particularly important.

‘Why Do Japanese Firms Choose Joint Ventures? Evidence from Manufacturing Affiliates in China’ was presented by Roger Strange (King’s College, London). Strange surveyed the post-1979 liberalizations of foreign investment in the People’s Republic of China (PRC), and analysed the ownership structure of Japanese manufacturing affiliates established there between 1979 and 1994. The 1990 changes had made equity joint ventures more attractive. After reviewing the relative advantages of such ventures. Strange demonstrated that the ownership structures of some 500 Japanese manufacturing affiliates were almost evenly divided between minority, majority and wholly owned shareholdings. The shareholding ratio had increased after the 1986 liberalization measures but not, apparently, after the 1990 measures. Strange also claimed that the theoretical prediction that external shareholdings would be larger in high-technology sectors was borne out by the data. He further suggested that, in principle, the ownership structures of Japanese affiliates would be affected by variables such as R&D expenditure/sales ratios, the size of the Chinese affiliate’s industrial sector and the total value of investment in the affiliate.

Geoffrey Owen (CEP, LSE) was interested in how far the effects of relevant variables had changed. For example, accumulating experience would diminish the importance of Chinese partners’ local knowledge, and increased global operation by Japanese affiliates might reduce the attractions of minority shareholding. Japanese investments in the United Kingdom had usually started as 50/50 equity joint ventures and ended as 100% Japanese-owned. He questioned whether Japanese companies’ experiences in Korea and Taiwan, where local companies eventually became competitors, would influence attitudes in China. Further research was needed on shareholding structures in Japanese, US and European firms and on the specific role of Japanese trading companies. Sudipto Battacharya (LSE) suggested that, contrary to theoretical expectations, Chinese partners in high-tech sectors might want higher stakes to help them absorb technology effectively.

Fumio Ohtake (Osaka University) presented ‘Labour Demand and the Structure of Adjustment Costs in Japan’, written with Andrew Hildreth (Osaka and Essex Universities). They addressed the empirical puzzle, reported in US plant-level studies, suggesting that labour adjustments to demand shocks were ‘lumpy’ or discontinuous. When aggregated to larger units, however, adjustments can appear smooth. Studies comparing Japan and the United States have reported varying results, but typically adjustment in Japan has been no faster, and usually slower, than in other major economies. The authors argued that firm data under report the extent of short-run adjustment. Using data for a multi-plant motor car manufacturer in Japan, they examined adjustment costs at different levels of aggregation, and found the assumption of smooth adjustment to be valid at both the plant and aggregated-company levels. Adjustment came both in the form of employment and in working hours.

David Soskice (WZB, Berlin) raised four points: (1) the role of prices as a mechanism for smoothing adjustment to exogenous demand shocks; (2) the potential effects on plant- and company-level adjustment patterns of transfers of labour outside the firm, e.g. within keiretsu; (3) the likelihood that in-company training, and generally higher skill levels, might make intra-firm transfers easier in Japan; and (4) the possibility, suggested by some managerial models, that centralized and decentralized decision-making may have different impacts on labour-adjustment patterns. Kenn Ariga (Kyoto University) argued that a demonstrable link between plant- and company-level data did not necessarily permit extrapolation to the macro level. Therefore, whether adjustment in Japan was smoother than in the United States was still arguable. Alan Manning (LSE) suggested that it is the US response which is atypical: US firms tend to use temporary layoffs for reasons related to the nature of social security benefits rather than adjustment costs.

The next paper, by Yukinobu Kitarmura (Keio University and Bank of Japan), Hiroshi Yoshida (Meikai University) and Noriyuki Takayama (Hitotsubashi), was on ‘Generational Accounting in Japan’. The authors applied the Kotlikoff method to Japan as part of an international project to compare the intergenerational transfer implications of existing social security programmes. According to their figures, future generations will have to pay a tax burden nearly double the current generation’s to sustain present levels of benefits and public-sector deficits. The implications of a wide range of policy options and simulations, involving changes in government purchases, tax revenues and transfer payments, were considered. Most of these, however, led to scenarios requiring heavy tax increases or painfully large cuts in government spending. The paper also considered alternative measures of intergenerational fairness and developed a measure of the Musgrave criterion. Taking into account the benefit to existing generations of government investment in infrastructure, the results indicated that annual budget balance could be achieved with ‘Musgrave intergenerational fairness’ with an increase in tax revenues of 12–14% and a cut in transfer payments of 13–17%.

For Jenny Corbett (Nissan Institute of Japanese Studies, Oxford, and CEPR), the results were heavily dependent on the assumptions made about productivity growth and discount rates. Furthermore, the macroeconomic implications of the pension problem might be less serious than the paper implied, since they depended on the responses of savings and consumption of the young and the old, which would not be known in advance. The links between aggregate savings, investment and productivity growth also would be crucially important, and several scenarios were possible in seeking to determine whether countries with ageing populations could ‘afford’ their pension systems. Fumio Ohtake queried whether Musgrave-type intergenerational risk-sharing analysis was appropriate where demographic change was not uncertain. David Soskice suggested the problem could be solved by incorporating individual behavioural choices, such as the choice of working hours.

The paper by Kenn Ariga (Kyoto University), Giorgio Brunello (Udine) and Yasushi Ohkusa (Osaka University), entitled ‘Fast Track: Is it in the Genes? A Case Study of Promotion Policy of a Large Japanese Firm’, considered the common view that Japanese firms typically select employees for more rapid promotion only late in their careers. US firms, by contrast, are generally assumed to pick ‘stars’ at an early stage. Hitherto, data availability has limited empirical testing of the hypothesis. The authors’ data for a large, high technology firm in Japan suggest several results: there is little evidence to support the stereotypical late-selection approach; there are multiple entry points; a substantial number of new employees are hired with significant previous job experience; and there is evidence of significant and persistent ‘fast-track effects’ for employees previously promoted more rapidly than their cohorts.

Alan Manning (LSE) noted that the results were difficult to interpret because there were too few similar studies. For example, the authors found little evidence of ‘late selection’, but what would be ‘early selection’? Similarly, interpretation of the evidence in their transition matrix table for 1994–6 is hampered by the lack of longer time-series information, which would allow for a more accurate picture of points of entry and speeds of promotion. Evidence of fast tracking was not borne out by the survivor curves, which showed most employees being promoted together. Paolo Rebioni (LSE) asked whether competing models of internal labour markets would give different predictions about fast tracking.

David Soskice and Mari Sako raised the possibility of firm-specific growth-rate effects and different skill- or occupation-specific effects.

The penultimate paper, ‘Inventory Behaviour: A Comparative Study of UK and Japanese Firms’, was presented by Jenny Corbett (Nissan Institute of Japanese Studies, Oxford, and CEPR), and written with Donald Hay (Jesus College, Oxford) and Helen Louri (Athens University of Economics and Business). The authors analysed the determinants of the inventory-to-asset ratio in panel-data sets for Japanese and UK firms in the period 1960–85. They treated firms’ inventory decisions in a portfolio framework, which permitted inclusion of financial rates of return in the inventory-demand equations. Their results showed that inventories in Japan were related positively to sector-specific inflation rates and expected sales, negatively to expected profits and were not much affected by interest rates. By contrast, UK inventories generally were related positively to profit rates and responded to short-term interest rates.

Earlier empirical studies of Japanese firms’ inventory behaviour presented conflicting results on whether inventories smoothed production and on the importance of just-in-time techniques. Their study did not address the smoothing debate, but it had interesting implications for just-in-time techniques and for the observed secular reductions in inventory ratios. Once other variables were taken into account, the time-trend coefficient for Japan was positive, rather than the expected negative. For the United Kingdom, the trend was negative, though not highly significant. A possible explanation was that the variables in the model were able to explain long-term downward movements in inventories in Japan, leaving little to be explained by just-in-time techniques.

Alessandra Guariglia (Essex University) acknowledged the merits of the underlying theoretical model, but had reservations about the econometrics. Better instruments were needed for the lagged variables and a GMM model would handle the unbalanced panel data better. It would be useful – and methodologically feasible – also to have separate results for large and small firms. Etsuro Shioji noted that a secular downward trend in the share of inventories in assets must be matched by some other (unidentified) upward-trending item or items. Fumio Ohtake suggested checking for different behaviour in firms in structurally depressed industries.

In the final paper, Yishay Yafeh (Hebrew University, Jerusalem) presented a joint paper with Tarun Khanna (Harvard Business School) on ‘Corporate Groups in Developing Economies: Paragons or Parasites?’. The paper investigated the performance of diversified business groups in Korea and India. Why are diversified corporate groups so prevalent in emerging markets? And how do they respond to structural changes and economic reforms? The authors used firm-level data to consider the effects of diversification on performance. Contrary to the stylized facts for the United States, they found that extreme diversification is associated with superior performance in terms of both profit and growth rates. The average cost of capital for diversified groups is one contributory factor, but some performance advantage is derived from state regulation and favours. Thus, the internal capital market story is not the main feature. They also found that firms became more diversified in response to economic reforms.

Michel Habib (LBS) noted that the paper found a hierarchy of performance results, with large corporate groups doing best, followed by non-group firms, then small- and medium-sized group firms. This produced a U-shaped relationship between performance and diversification – the reverse of the US case where the mid-range of diversification, built around core capacity, reportedly gives the best performance. Explanations for diversification are three-fold: to achieve market power; because managers seek expansion (agency view); and to acquire resources. In developing countries the resource view seems the most likely, with political influence viewed as another resource. This might explain the Indian, but not the Korean, result. In Korea, firms are more likely to be engaged in related, value-enhancing diversification. The idea of focus in diversification – so important in the United States – would be a fruitful research area. Fumio Ohtake suggested that internal labour markets might be as important as internal capital markets. Yukinobu Kitarmura noted that stages of development and economy-wide economies of scale might also help to explain the results.