ディスクロージャー研究学会

（青空に物事を晒すと虫干しされ綺麗になる）

文書No.
970306e

The Sixth Osaka International Finance Conference

January 21-22, 1997

Contact: Jun Uno

Chief Researcher of Financial Engineering Research Dept.

Quick Research Institute, Inc.

Shinkawa Sanko Bldg.

1-3-17 Shinkawa, Chuo-ku, Tokyo

Tel. (03) 5541-7162

Fax (03) 5541-7154

E-mail: juno@qri.co.jp

Cross transactions for large orders have been actively executed since 1994 in conjunction with writing-off of non-performing debts by financial institutions. Through a detailed examination of how large orders are filled under the present system, this paper addresses the constraints and issues that the current trading systems have.

As a result of partial deregulation of commission rates, the commission rate for large orders is now down to half or a quarter of that of the fixed rate period. In the execution of cross transaction, if a broker becomes a counter party of the trade, an incurring the securities transfer tax places the largest component of the costs for the broker. Realized commission rates went below the break-even rates which we computed in case of broker's participation of cross transaction. The results suggest that brokers has reduced their participation in cross transactions to accept clients' demand for cutting commission rates.

The second purpose is to explain why the majority of cross transactions are executed at stock exchanges other than the Tokyo Stock Exchange. Since cross orders are still subject to price and time priority, execution prices on first and second day of cross transactions were exactly same about 92% of our samples. If a stock has multiple listings and the satellite exchanges are used for execution of cross, execution prices are not affected by overnight price fluctuation. Price mechanism between primary and satellite exchanges provides price buffer for turn round cross executions.

This system may not be appropriate to apply for orders from institutional investors. It allows too much deviation of pricing on the satellite exchanges from that on the primary exchange. Considering the development of the upstairs market in the U.S., the need to establish alternative system for large orders in Japan is now high and the review of the existing price mechanism is necessary.

The execution of large orders is always a major issue for any securities trading system. It is difficult to avoid a price impact in executing large orders considering the inherent liquidity that exists in the securities market. However, clients with large orders are often very important customers for securities firms and, therefore, usually optimum efforts are made to accommodate the needs and the expectations of these clients. It is possible to clarify the constraints and issues that the current trading systems have through a detailed examination of how large orders are filled under the present system.

In the Japanese stock market, the cross transaction method is often applied to large orders which is a way for corporations to settle their accounting issues. Yet there are certain common elements with large order executions in general in the market where institutionalized. This paper is intended to review the cross transactions from the following two standpoints.

First is an analysis of the current status of large order executions called "cross transactions". Cross transactions have been actively executed since 1994 in conjunction with writing-off of non-performing debts by financial institutions. The practice has gained so much momentum that it has now reached the level of 20% of the total volume of all the stock trading, a level which is difficult to be dismissed simply as a special type of trading.

The reason for investors to be engaged in cross transactions is clear, but what about the profit level for securities firms that execute these transactions? This paper intends to examine the trend of commission rates and the profit level of cross transactions from the securities firms' perspective in conjunction with the deregulation of commission rates of large transactions.

Due to the deregulation of commission rates of large orders, the commission level is now down to half or a quarter of that of the fixed rate period. However, study of the change of the commission level on a quarterly basis shows no correlation with the increase of the average size of trades during the same period. (Table 1) This seems to be based on the fact that trades over 1 billion yen are mainly "cross transactions" and that when execution cross transaction, an incurring the securities transfer tax places the most significant impact on the profit level of cross transactions for broker.

The second purpose is to explain the relationship between the pricing rules of securities with multiple listings in Japan and cross transactions focusing on the fact that the majority of cross transactions are executed at stock exchanges other than the Tokyo Stock Exchange. These large orders tend to be heavily concentrated at the Osaka and the Nagoya Stock Exchanges rather than at the Tokyo Stock Exchange, and it is not uncommon for these transactions to occupy more than half of all the buys and sells at these two exchanges. Why does this kind of concentration occur? Is it due to an economic factor? Or is it due to the inherent nature of trading mechanisms? This paper intends to clarify these issues. The Japanese stock price information does not include the concept of composite pricing like in the United States. Rather, each exchange determine and disseminate its own pricing. In Japan the system is structured in such a way that a consistency in price formation between different exchanges during trading hours is maintained within a certain range which is much larger than bid and ask spreads. This paper examines execution prices on turn around cross transactions and how this factor affects which exchanges these cross transactions are executed at. Approximately 90% of all the cross transactions executed at the Osaka and Nagoya Stock Exchanges use the same price for the first and the second transactions. However, the price of matched trades in cross transactions at the Tokyo Stock Exchange tends to deviate from the original transaction price due to the effect of price fluctuation.

Considering how the upstairs market developed in the U.S., the need to establish a system that responds to large orders from institutional investors in Japan is now being recognized. However, as you will see from the result of our examination, it will not be appropriate to apply the same rules that are used in current cross transactions in Japan without any change for the Japanese block trading market.

The subsequent sections in this article are as follows. In Section 2 the relationship between large trading and cross transactions is investigated. Section 3 explains the close relationship between the liability of securities transfer taxes paid by securities companies and the commission rates, examined from the standpoint of profitability of cross transactions. We compute minimum commission rate to cover broker's cost in section4. Section 5 analyzes the relationship between the risk of price fluctuation overnight and the execution rules, and Section 6 examines the price impact of cross transactions. Lastly, we will summarize the current situation and the future prospects of the large transactions market in Japan.

One transactions per stock over 1 billion yen which are included in Table 1 are considered to be quite large if these transactions belong to institutional investors' diversified portfolios. An issue in a portfolio whose market value is over 1 billion yen is a type of issue that comprises the core portion of a portfolio with a large total market value. In the usual course of practice, it is hard to imagine this type of issue being transacted all at one time. Therefore, a transaction with a size over 1 billion yen is possible only in a market in which a unique and special type of format such as cross transactions is applied. It is our belief that we can better understand the fluctuation of commission rates by fully understanding how this trading is practiced in the market.

As far as the data related to the trend of large transactions is concerned, there is aggregate data of three markets collected by major securities firms in addition to Table 1 which was quoted earlier. (Hereafter referred to as "Brokers' Estimates".) Figure 1 shows the trend of the past five years (from April 1991 to December 1995) using this data. When the aggregate total volume of the three markets, the Tokyo, Osaka and Nagoya Stock Exchanges, is contrasted against the trend of the Brokers' Estimates on a monthly basis, you will notice that in 1991, more specifically after April '91, large transactions occupied 15% of the aggregate total volume of the three markets. After that point the volume declined to less than 10% level and stayed there quite a while. Large transactions picked up again after January 1994, and in the fiscal year 1994 (April 1994 to March 1995) the large transactions level reached as high as 22% of the aggregate volume of the three markets. This trend coincided in timing with the deregulation of commission rates. However, this trend was not triggered by the deregulation of commission rates itself, rather, according to our analysis, it was the result of the need on the corporations side to generate profits on the books due to the poor performance of the companies. According to the estimate made by the Japan Economic Journal (Nihon Keizai Shimbun), the total earnings realized by selling off stocks on the part of listed financial institutions in the fiscal year ending 1994 reached as high as 3 trillion yen approximately. The increase of the large transactions ratio can be attributed to a reflection of the increase of cross transactions for the purpose of generating profits when the stock market as a whole did not recover well.

However, the transactions over 300,000 shares which are included in the "Brokers' Estimates" do not necessarily exceed 1 billion yen threshold in yen amounts. Therefore, it is necessary to supplement the analysis based on the aggregate values with the analysis based on individual cross transaction data. As for the information source of individual cross transactions, there is "Major Cross Transactions in Stock Market" in the market information column of the morning edition of the Japan Economic Journal. (Hereafter referred to as "Individual Cross Data".) Using this column one can abstract information daily on twelve issues out of those listed under cross transactions executed at major stock exchanges pertaining to their numbers of shares, prices, locations of exchanges which executed those trades, and names of member securities firms. There were 3,666 transactions listed as cross transaction data in the morning editions of the Japan Economic Journal during the 15 months between January 1994 through March 1995. Out of these transactions, 2,541 transactions exceeded the 1 billion yen threshold per transaction. If you classify the 3,666 transactions by the exchanges where these transactions were executed, 893 transactions belonged to the Tokyo Stock Exchange, 2,128 to the Osaka Stock Exchange and 644 to the Nagoya Stock Exchange. Osaka has a high concentration of these transactions. The aggregate number of shares on a monthly basis based on "Individual Cross Data" occupied 25% to 71% of the total shares of large transactions listed in "Brokers' Estimates". During the period specified above, the average volume of such transactions reached 40%. Figure 2 shows the number of individual cross transactions on a monthly basis and, as you can see, transactions less than 3 billion yen occupy 60 to 70% of all the transactions in the months of April and May and this ratio drops down to 20% in September and March. In other words, it is clear that large cross transactions frequently occurred during these months.

The purpose of these individual transactions is not disclosed in the column of the newspaper. However, you can make a deduction if you focus on the fact that, unlike regular investment activities, cross transactions for the purpose of generating profits occur on a turn around basis in a very short time. Therefore, it is fair to assume that the purpose of these cross transactions was to generate profits, if large transactions occur as a turn around transaction with the same issue, with the same number of shares handled by the same securities firm within a few days. Using this method, we attempted to select the transactions that were seemingly executed for the purpose of generating profits out of these 3,666 cross transactions. The result was that approximately 50% of these transactions, namely, 1,812 transactions, fell into this category. These 1,812 cross transactions were traded in a turn around manner with the same number of shares. Therefore, they can be considered as 906 pairs of corresponding transactions and will be further analyzed in Section 4 and in the subsequent sections. The number of actual cross transactions fluctuates daily, yet the number of transactions quoted in the newspaper daily was fixed in number. This means that not all the turn around trades were covered by the newspaper. We will work on cross transactions which we can confirmed as turn around trades.

During the fiscal years '94 and '95 in order to write off bad loans, huge amounts of cross transactions were executed by financial institutions to generate profits by selling off stocks being held and bought back. In cross transactions there are basically two types of arrangements in one type a securities firm serves as the counter party for a particular transaction, and in the other type a securities firm matches the client with another corporation which holds the same issue and executes the trade. Whether or not the securities firm bears the cost of the securities transfer tax at the time of selling off seems to make a big impact on the rates of negotiated commissions between the client and the broker.

Let's calculate the balance sheet of a securities firm when it becomes a temporary counter party buyer in a cross order transaction using a specific example. Let's take a case of a cross transaction with 1 billion yen in value of an issue which is priced at 1,000 yen with 1 million shares.

The income for this securities firm is generated from the commission due to this firm and the regulation dictates that it should not fall below the 1.535 million yen range. The commission income for turn around transactions represents 3.07 million yen(0.307%) and this amount is the same as when the fixed commission rates were applied.

On the other hand, as far as the expenses are concerned, the securities firm has to pay the fixed membership fee and the trading fees to on execution exchange. According to our conversation with securities firms, estimated calculations of expenses (on a turn around basis) associated with exchanges differ from exchange to exchange due to the variance in what each exchange charges as the fixed rate membership fees. The Tokyo Stock Exchange (TSE) charges approximately 0.028% of the transaction amount, the Osaka Stock Exchange (OSE) charges approximately 0.017% and the Nagoya Stock Exchange (NSE), 0.024%. When the transaction amount is 1 billion yen, the TSE charges 280,000 yen, the highest amount, and the OSE charges 170,000 yen, the least amount.

The second expense for securities firms is the securities transfer tax which is charged against the second trade of the paired cross transaction. The rate charged is 0.12% of the transaction amount. If the transaction amount is 1 billion yen, the charge will be 1.2 million yen which represents the biggest expense item. In addition, securities firms have to take into consideration obtaining the capital to buy the shares in the first transaction of the cross transaction. Supposing the interest rate is 2%, the interest cost per day will be approximately 54,000 yen. Thus, the total direct cost of executing two cross transaction will be calculated as 1.53 million yen at the TSE which is the high end (0.153% of the transaction amount) and 1.42 million yen at the OSE which is the lowest end (0.142% of the transaction amount). The gross profit represented by the spread between the commission income and the expenses ranges from 1.54 million to 1.65 million yen and the gross profit ratio is approximately 50%.

The profit margin declines as the amount of a cross transaction increases. Figure 3 shows the relationship between the income and the expenses of a securities firm as a percentage against the amount of the cross transaction which incrementally increases from 100 million yen to 20 billion yen. In the case of a securities firm becoming the counter party of a cross transaction, since the direct cost is determined by a sliding scale against the amount of the transaction, the profitability will degenerate as the amount of the transaction increases under the fixed commission rate environment. For example, the turn around commission income of a cross transaction of 10 billion yen would have been about 17 million yen (0.17%) even before the deregulation of commission rates on large transactions. This ratio is about half of that of a 1 billion yen transaction. On the other hand, the expenses occur at a constant rate against the transaction amount, which will range from 15.3 million to 14.2 million yen and the gross profit will range from 1.86 million to 2.76 million yen and the gross profit level will be 10 - 15%.

According to the quarterly reports compiled by the National Conference of Stock Exchanges (in Japan) and the Japan Securities Dealers Association (Table 1), actual commission income charged on the portion of the amount exceeding 1 billion yen declined to 0.02% to 0.04% after April 1994 due to the deregulation of commission rates charged against large transactions. Assuming 0.038% as the commission rate charged for an amount exceeding 1 billion yen during April and June 1994, any cross transaction over 3.5 billion yen would have offset the income, if the securities firm had had to bear the cost of the transaction tax. A close examination of 256 cases of large cross transactions over 1 billion yen executed during the same period which are selected from "Individual Cross Data" leads us to the conclusion that approximately 21% , namely 55 cases, were in theory losing money. During October through December of 1994 when the commission rate was as low as 0.021%, any cross transactions over 2.5 billion yen were, in theory, in the red. In other words, 226 cases out of 428 (52.8%) should be showing negative results. During January to March 1995 when the actual commission rate rallied up to 0.044%, the break even point reached as high as about the 4 billion yen level. Yet since during this time frame cross transactions overall increased in size, more than half, 358 cases out of 618, were losing money.

The reason why the theoretical calculation produces more cases in the red than in actuality was because securities firms engaged themselves more often in the second type of executions in which they were not obligated to bear the cost of the transaction tax. If the securities firms had engaged themselves in this type of executions and as long as the companies generated 0.017 to 0.011% commission income, other direct expenses would have been covered. The minimum commission for a transaction over 1 billion yen is 1.535 million yen which, in theory, covers all the direct expenses associated with transactions between 9 to 14 billion yen.

However, it is not always possible to find a counter party company that is interested in a trade of the same issue immediately. Particularly toward the end of a term when cross transactions increase, it is difficult to match cross transactions among client companies. It appears during the fiscal year end periods there are more cases in which securities firms become the counter party. In these instances securities firms have to hold the bottom line by earning at least enough income from the commission to cover the cost of involving the transaction.

As a next step, let us calculate the necessary commission rate for involving a "cross" transaction based on the formula listed below and compare it with the trend of the actual commission rates. "V" represents the amount of the cross transactions. We assumed that the necessary cost for a turn around of a cross transaction was an aggregate amount of the transfer tax (0.12%), the fixed membership fee specifically set for the Osaka Stock Exchange, the trading fee, and the interest rate cost for raising capital. We assumed the percentage of the aggregate amount against the first traded amount to be 0.142%. The securities firm needs to recoup this amount with turn around transactions, therefore, the necessary commission rate for a one way transaction is calculated as 0.071% which is half.

Comm(%)=(V x 0.00071 - 1,535,000)/(V - 1,000,000,000) x 100 (1)

Figure 4 shows the relationship between the break even commission (Comm) imposed on a trade exceeding 1 billion yen based on the calculation formula and the amount of the cross transaction. When the cross transaction amount is below 2.16 billion yen, part of the commission income remains as their profit after paying the securities transfer tax, even if their commission income is minimal based on the bracket of 1 billion. (3.07 million yen for the turn around). However, there is no discount of the transfer tax regardless of the traded amount, the bigger the amount gets, the smaller the effect of the minimum commission will be. The necessary commission rate converges at 0.07%, the level that reflects the average cost of a turn around transaction cost that the securities firm has to bear which consists of the transfer tax rate plus trading fees, etc.

Next, let us calculate individually the necessary commission rate of 2,541 cross transactions over 1 billion yen and see what kind of relationship exists between the trend of the average amount and the actual values of commissions. (Table 2) Since the actual values are expressed by the rate of aggregate commissions for the portions over 1 billion yen, it is more appropriate to compare the necessary commission rate with order value weighted averages. However, Individual Cross Data tends to be selected focusing on large transactions. Therefore, if this is compared with the actual values which are expressed as the weighted average of all the transactions over 1 billion yen, the rate will appear higher during the periods when cross transaction amounts increase. Taking this type of bias into consideration, our examination also refers to simple averages of necessary commission rates on an individual transaction basis.

During the January to March 1994 period prior to the deregulation, the break even commission (with a weighted average) was 0.052%. The actual commission income during this period was 0.075% and, therefore, cross transactions were financially viable trades. During April to June 1994 right after the deregulation, the break even commission rate was 0.034%, while the actual performance was 0.038%. During this period the rate has come down to a level where, if a firm were to underwrite all the necessary transactions, there would have been hardly any profit left. During the following months of July to September, the cross transactions became larger and the break even commission rate on a weighted average basis was higher than the actual amount, indicating that if securities firms underwrote the transactions by becoming the counter party, the balance would have been in the red. During the October to December period, the actual commission income further declined to the all time low. However, the break even commission rate on a weighted average ended up being on a plateau due to the impact of the 28 cross transactions over 10 billion yen that occurred during December.

If a securities firm does not underwrite a cross transaction but can find a counter party company, it can bring down the break even commission rate to the lowest possible level. The fact that there was a drastic decline in actual commission income seems to be the result of securities firms actively pursuing this type of execution. The tendency of the weighted average of the break even commission rate to exceed the actual performance became a chronic situation after the July to September 1994 period. This seems to indicate that securities firms began to adopt an on-going strategy of maintaining profits by executing cross transactions with third parties. During the January to March 1995 period, the actual commission rate became 0.044%, highest ever since the deregulation. This can be explained by the assumption that the demand for cross transactions experienced rapid growth, the size of transactions got larger, and due to the time constraint securities firms increased the cross transactions in which they became the counter parties. The result suggests that the commission rate of large transactions is determined as a reflection of the relationship between the cross execution formats and the cost of the securities transfer tax.

When we summarize the status of large transactions in 1994 depicted in figure 5, the total amount of large transactions is estimated as 2.4 trillion yen which is a 2.7 time increase from the previous year based on the number of shares of large transactions compiled by Brokers' estimates. However, commission income for securities firms was at 16.4 billion yen, which was about double that of the previous year. On the other hand, the securities transfer tax paid by investors is estimated as 36.2 billion yen and by the securities firms as 14.5 billion yen. This means at least over 50 billion yen was paid in total as transfer tax. This represents three times more than the amount of the commission income. This amount is based on the assumption that all the large transactions were cross transactions and that securities firms underwrote all the transactions. If transactions had been executed between two investors, the amount of tax would have been much higher. When we examine the cost structure of the Japanese large transactions market, it is clear that unless deregulation of the commission rates and lowering or elimination of the securities transfer tax go simultaneously, the cost structure of stock trading will be even more distorted and the country will not be able to compete in the international arena.

So far the calculation of profits for securities firms was based on the assumption that the turn around cross transactions were priced equally on both transactions. The turn around cross transactions for generating profits need to be executed on two separate days and there is always an inherent risk of price movements between the two dates. If there is a tick difference between the two prices and if the stock price level is 1,000 yen, there is a plus, minus 1% of profit or loss against the transaction price. If the cross transaction is 1 billion yen, the difference will be 10 million yen and the loss (increase of expenses) would be intolerable for both parties. We suspect that securities firms as agents have a strong motive to execute both transactions with the identical price.

Table 3 is a summary of Price Differential associated with the executions of 906 cross transactions that have been confirmed as paired transactions.

Execution Price Differential = (2nd trade price - 1st trade price)/ 1st trade price (2)

Of all the applicable transactions, 833 cases were traded with the same price for both buy and sell (92% of the total) and represent the majority of transactions. There were only 72 cases (8%) in which trades prices were different between the first and the second trade. Furthermore, when we examine the execution Price Differential ratio of cross transactions by stock exchanges, Differential of transactions between Osaka and Osaka or Nagoya and Nagoya are minus 0.021% and 0.001% respectively, both of which are very close to zero. But between Tokyo and Tokyo, it is a 0.192% increase which is higher than the other exchanges by a quite a margin. This indicates that there is a Differential depending on where a trade is executed.

Next, let us analyze the relationship between the Differential of executed trades and the stock price fluctuation of that trade day incorporating the execution rules of cross transactions.

There are no specific rules established for executing cross transactions in Japanese stock exchanges. Therefore, all cross transactions are required to be ordered and traded according to the exchange-centered trading principle and existing trading practice rules Securities firms, in order to execute a cross order, communicate the limit orders of a buy and sell with the same number of shares to a jobber ("Saitori" or specialist) and have them execute these trades. In this circumstance, the floor is obligated to include any other limit orders that satisfy the priority in terms of the price and timing. Therefore, the securities firms are better positioned in executing cross transactions if they use Osaka or Nagoya which receive less limit orders. Since less limit orders means a higher probability of executing the desired matched trades, Osaka and Nagoya tend to be chosen as the execution floor for cross transactions. It is a common misunderstanding that since these markets receive few limit orders on the floor, it will be easier to execute a transaction with the desired price without being affected by the price movement at the Tokyo Stock Exchange. However, the reality is not that simple. There are inter-market regulations established that regulate how to determine pricing of stocks with multiple listings and trade prices are determined accordingly.

There are many issues that are cross listed among the eight domestic stock exchanges in Japan. Price setting mechanism of these issues is made in such a way to avoid any disproportionate deviation among different markets. Each issue designates its primary market and when the issue is traded at a non-primary market without any recent reference price, "Adjusted Quotation Prices" are quoted which will ensure that the next trading price will fall within the limited range of the trading price of the primary market. This price spread is defined as "Maximum Price Variation" and is issued inter-markets. In the execution of cross transactions, this serves as a "Price Fluctuation Buffer" which eases the price volatility that can occur between the first trade and the second.

Let's take an example of a cross transaction at the Osaka Stock Exchange. The first trade of a cross transaction is usually executed at the closing of the market. The closing price of the OSE is determined as follows. If an order comes into the OSE at the closing and is executed, naturally, the traded price becomes the closing price of the day. If the designated primary market of this issue is not the OSE, the Adjusted Quotation Prices plays an important role in setting the closing price or final quotation of the day. If the issue is traded on the system trading at the OSE and if it designates the TSE as its primary market, there is a system linkage between the TSE and the OSE to make the closing quotation at the OSE automatically coincide with the closing price of the TSE. As far as issues with common primary markets are concerned, the Adjusted Quotation Prices at the time of closing is used as the closing quotation and, therefore, the closing quotation will be near the closing price of the TSE. This way the final quotation which is the de facto closing price of the OSE is linked to the price of the primary market at the closing through the existence of Adjusted Quotation Prices. This is an important element in matching the price of the second trade with that of the first trade.

Let us show you an example using Figure 6. Let us assume the closing price at both the TSE and the OSE on the first day is 800 yen, and the first trade of a cross transaction was executed at this price. The base opening price of the next day is set as 800 yen and the updated price spread on this price level is 5 yen. In this case, if the opening price at the Tokyo Stock Exchange is between 795 yen and 805 yen, it is possible to execute the second trade at the OSE at 800 yen right after the trade opens at the TSE. Thus, it becomes possible to execute round transactions at the same trade price.

As was shown, "Maximum Price Variation" has the effect of either eliminating or easing the overnight volatility of the price of targeted issues in cross transactions. The effectiveness of Price Fluctuation Buffer generated from Maximum Price Variation is determined depending on the price level of the issue. In the example of Figure 7, the effectiveness of the buffer against the stock price of 800 yen can be expressed as 0.625%. The buffer effect can protect against volatility between a maximum of 3% (equivalent of 100 yen with the stock price) and a minimum of 0.5% (999 yen with the stock price). Figure 6 shows the relationship among the effectiveness of the buffer, Maximum Price Variation Range at the OSE and the stock price level.

We measured to what degree this Maximum Price Variation eases up possible overnight price volatility. Table 4 used as references individual cross transactions that were confirmed as pairs, and compared the actual price fluctuations, realized profits and losses, and the effectiveness of the price buffer. Through this process we measured how effective the Maximum Price Variation are in easing up price fluctuations of the issues with multiple listings. The size of "The Price Buffer" for each sampled issue can be defined as follows:

Price Fluctuation Buffer =

Maximum Price Variation of OSE(NSE) / Cross Price (1st trade) (3)

Overnight Price Change after the First Cross =

TSE Opening Price on the following day / Cross Price(1st day) (4)

Next, we compared the overnight price change rate after the first cross transaction with the size of the fluctuation buffer. The difference is calculated as "Excessive Overnight Price Change". All the cross transacted issues that are under examination designate either the TSE as the primary market or common primary market. The price formation, in reality, is based on the prices at the TSE and, therefore, the comparison between the cross traded price and the opening price of the issue at the TSE on the following day leads us to calculate the price fluctuation that occurred overnight.

Excess Fluctuation Rate = (in case of Price Increase)

TSE Opening Price / Cross Price - Price Buffer (5a)

Excess Fluctuation Rate = (in case of Price Decrease)

TSE Opening Price / Cross Price + Price Buffer (5b)

In Table 4, when the Excess Fluctuation Rate is negative (when the overnight price fluctuation is smaller than the buffer), the Excessive Fluctuation Rate is considered zero.

When the opening price exceeded the Updated Price Spread, the timing of the execution could be delayed to minimize the deviation of the second trade. This can be called "Execution Timing Effect" and can be calculated from the difference between the Excessive Fluctuation Rate and the actual execution price differential.

Execution Timing Effect = Excessive Fluctuation Rate - Execution Price Change (6)

When the 2nd transaction was executed at the TSE, the Price Fluctuation Buffer effect does not apply. Therefore, in Table 4 only the execution price differential and the actual price fluctuation rate are indicated and the total average used only the samples whose second trades were traded at exchanges other than the TSE.

In the case of price increases, the average overnight price increase after the cross transaction was 0.88% and this exceeded the Price Fluctuation Buffer Rate (average 0.82%) of each issue by 0.27% on average. The effectiveness of easing the overnight fluctuation by the price buffer is 69.3%. The actual execution profit/loss is 0.05% which is smaller than the Excessive Fluctuation Rate and this indicates that by selecting the right timing of the execution 25% of the overnight fluctuation ratio was hedged.

Similarly, in the case of price decreases, the actual price fluctuation after the cross transaction was -0.96%, but the execution price differential was -0.12%. We can see that 65.6% of the overnight price fluctuation was offset by the Price Fluctuation Buffer effect and 21.9% by Execution Timing Effect leaving -0.12%.

Examined exchange by exchange, 70% of the overnight price change is covered (in the case of decrease, 67%) when both transactions were executed at the OSE and the Execution Timing Effect was 23% (in the case of decline, 20%). When both transactions were executed at the NSE, the Price Buffer Effect served for 72% cases (for decline, 65%) and the Execution Timing Effect worked for 24% cases (for decline, 30%)

On the other hand, the executed Price Differential issues executed at the TSE tend to reflect the increase in price from the execution of the cross transaction to the opening price the next day. The stock price increased 2.01% overnight after the cross transaction and the price differential of the execution was 1.87%. When the price declined or stayed the same, the price fluctuation rate and the executed price differential coincided with each other and it appears that the second trade was executed at the time of the opening.

The results of measuring the actual data indicate that the Price Buffer Effect in fact was at work for cross transactions that were executed at the OSE and the NSE and 70% of all the price fluctuations were covered by this effect. This effect is only possible with stocks with multiple listings including the markets where less orders come in. This explains why non-TSE exchanges are used primarily for cross transactions.

Next, we would like to explain what kind of influence cross transactions have in forming the price of that particular issue. We will focus on the findings on the impact on price formation after the execution is completed.

In order to examine whether there is a price impact after the cross transaction is executed, we compiled the calculation data of daily positive returns against TOPIX for six days after the first cross transaction was executed. (Table 5). During the process of examining when the second trades were executed on a daily basis and to see if there was any impact, we noticed that of all the 905 cross transactions (We excluded one issue which is listed at the OSE.) 750 transactions were executed on the next business day and there was no significant positive return statistically. When the first transaction was executed at the close of one trading day and the second transaction was executed at the opening of the next day, the purpose of the cross transaction would have been clear to other participants in the market and, therefore, the cross transaction did not bring any impact on the pricing.

On the other hand, when the second trades were not executed on the following day, the second trades were executed sometime between the second day and the fifth day after the day the first trade was executed. In each case there was no negative return recorded on the following day after the trade, and checking the three days following the day after the trade, we notice that some show positive returns. This phenomenon can be interpreted as indicating that the stock price rose after the cross transaction. But the prices are downwardly corrected on the third day (+3) after the second trades are executed and, therefore, the executed profit/loss of cross transactions for these cases are also close to zero. Taking these points into consideration, it is more natural to interpret that due to the price increase right after the first trade, they waited until the prices settled down and then executed the second trades. In 1995 the media picked up the issue that certain companies moved toward releasing cross holdings and some viewed that cross transactions are related to selling off activities of held stocks. But as far as the analysis of the cross transactions in '94 indicates, there were no signs to hint of that linkage.

The Japanese accounting system is based on the acquisition cost method. Cross transactions are uniquely Japanese transactions which were born out of the necessity to reflect unrealized gains of cross holding securities in the accounting and at the same time to maintain the traditional cross holdings relationship. Normally the turn around transactions of large orders inherently bear the risk of overnight price change, but the methods of price determination for stocks with multiple listings have a built-in buffer function which absorbed the risk of price fluctuations. With these elements in place within the framework of the existing system, we can say it was possible to execute these transactions. What the analysis of price formation of cross trade tells us is that there is no evidence that cross transactions are having an impact on the pricing of ordinary course of trading in exchanges. Examining it this way, it is difficult to find the need for keeping these unique transactions within the framework of exchange-centralizing policy.

The current structure of cross transactions is not suited for use as a way of effecting large orders from investment portfolios of institutional investors. If the discretionary nature of the maximum price variation built into the current system is used for executing large orders from institutional investors, there is possibility for those involved in the trading to determine the price far from the current bid and ask spread. As we saw in Figure 6, the maximum price variation which accounts for two to five ticks is much larger than usual bid-ask spread for most stocks and the maximum percentage deviation is 0.8% on average and this represents 70% of the overnight price fluctuation. This discretionary portion is too large from the view points of maintaining the fair price formation. In addition, this may allow price on satellite exchange far from that on the primary exchange, it may create discrimination for small orders and large orders in the same stock. In summary, the current rules allow satellite markets to match price discovery occurred in the primary market only loosely.

In the United States, upstairs market for block trades has similar component of trade arrangement. Counter part of large orders are searched by block traders and submitted to the exchange floor after arranging same amount of buy and sell orders. Block traders may use an electric bulletin board such as AutEx to search trade interests on the same stock and negotiate price and quantities through an electric trading system such as Instinet. Those pre-arranged cross orders are effected by member firms at the prevailing quote or within best quotations . There has been severe competition for block orders not only between NYSE and regional exchanges but also between NYSE and electronic trading systems called crossing network which facilitates the crossing of large orders from institutional investors without intervene of brokers.

Contrast to the case of the U.S., lack of alternative trading systems and high charge of the securities transfer tax becomes a cause for losing the cost competitive edge in the investment management business in Japan. Protecting traditional exchange centralizing policy and applying one trading rule for all types of orders must have negative impact on development of large transactions market in Japan. Even though the commission rates of large transactions have been deregulated, there are still unsolved issues pending which bring disadvantages to portfolio managers in competing in the world market in terms of transaction tax and execution rules. It will be necessary for the trading system to be reconstructed to be able to meet the needs of the new requirements for order executions.

Materials in Japanese:

(1)宇野淳・大村敬一[1996]「クロス取引と価格変動リスク」法政大学経済志林、1996年7月

(2)証券取引審議会作業部会[1993]「大口取引に係る株式委託手数料の自由化について」、1993年3月23日

(3)「大口取引に株券委託手数料率の公表について」東京証券取引所「証券」1994年8月,94年11月,95年2月,95年5月,95年8月、95年11月,96年2月

(4)東京証券取引所調査部編 [1994]「東京証券取引所　その機能と仕組みについて」1994年

(5)日本経済新聞 1996年1月27日,2月2日,3月9日,4月18日,4月21日各朝刊

1. 日経金融新聞 1996年2月5日,2月8日

Materials in English:

(7)New York Stock Exchange "Agency Block Cross Transaction of 25,000 Shares or More" October 23,27 1992

(8) Schwartz, R.A. "Equity Markets: Structure, Trading and Performance" Harper & Row、1988

(9)Torres, C. "Big Board Set To Allow Bypass Of Floor Trades" Wall Street Journal, October 26,1992

Table.1 Quarterly Block Trade Activities and Commission Rates

Year/Month Number Total Average Average

of trades Value Value Commission rates

94:4ー6 218 - 0.038

94:7ー9 2,278 - 0.038

94:10ー12 1,271 - 0.021

95:1ー3 4,127 - 0.044

95:4ー6 138 3,159 22.9 0.036

95:7ー9 1,173 39,751 33.9 0.044

95:10-12 1,197 53,076 44.3 0.020

96:1ー3 3,490 155,984 44.7 0.035

Note: The value is 100 million yen. The rates are percentage. Total value represents aggregated value of block trades which are equal to or more than one billion yen. Average commission rates are one which is applied to value exceeding one billion yen.

Source: National Conference of Stock Exchanges and the Japan Securities Dealers Association

Table.2 Actual Commission Rates and Break-even Rates for Large Orders

Quarter Average Break-even Rates Number of Rates (a) (b) of Cross Trade

94:1-3 0.075 0.034 　0.052 613

94:4ー6 0.038 0.016 　0.034 256

94:7ー9 0.038 0.029 　0.048 626

94:10ー12 0.021 0.024 　0.048 428

95:1ー3 0.044 0.038 　0.058 618

Note: Average commission rates are value weighted average. For break-even rates which is computed by formula (1), (a) is an arithmetic average of all cross transaction and (b) is a value-weighted average. The number of cross trade is individual cross trade which value is at least 1 billion yen.

Note: In each column, the top row shows in case of price rise, the middle does in case of no price change, and the bottom does in case of price decline. Price changes are measured by overnight price change(formula(b)).

(a) Execution Price Differential = (2nd trade price - 1st trade price)/ 1st trade price

(b) Overnight Price Change after the First Cross =TSE Opening Price on the following day / Cross Price(1st day)

(c) Price Change between the First Cross and the Next Day's Close =TSE Closing Price on the following day / Cross Price(1st day)

(d) Price Fluctuation Buffer = Maximum Price Variation of OSE(NSE)/Cross Price (1st trade)

(e) Excess Fluctuation Rate = (in case of Price Increase) TSE Opening Price/Cross Price - Price Fluctuation Buffer

Excess Fluctuation Rate = (in case of Price Decrease) the TSE Opening Price / Cross Price + Price Buffer

(g) Execution Timing Effect = Excessive Fluctuation Rate - Execution Price Change (6)

Note: Negotiated commission rate is average rates at First Quarter of 1995.

お問い合わせ ik8m-ysmr@asahi-net.or.jp