Expiration day effects and market manipulation: evidence from Taiwan

For example, in February 2004, the US Securities & Exchange Commission (SEC) filed a civil action in the United States District Court citing the case of the Southern District of New York against Moises Saba Masri and his broker, Albert Sutton. The SEC’s complaint stated that Saba had sold a large number of TZA (NYSE-listed Azteca American Depository Receipts) put options with strike prices of $5.00 and $7.50 and an expiration date of 20 August 1999. On that date, the price of TZA ranged between $4.9375 and $5.1875; however, in order to avoid his put options being exercised at the strike price of $5.00, Saba instructed Sutton to begin buying TZA in the open market during the last ten minutes of trading so as to cause the price to close above $5.00. The SEC alleged that both Saba and Sutton had violated the anti-fraud provisions of the federal securities laws by unlawfully ‘marking the close’ (SEC 2004).

Examples include Stoll and Whaley (1987, 1990, 1991), Chamberlain et al. (1989), Herbst and Maberly (1990), Karolyi (1996), Chung and Hseu (2008) and Hsieh (2008).

These two hypotheses are not mutually exclusive, since arbitrageurs and manipulators can coexist within the futures market.

The contract months in the TAIEX are the nearest 2 months plus three quarter-months, making up a total of 5 months.

From April 2006 to November 2008, the average daily settlement price of TAIEX futures for the nearest month contract (also the most actively traded contract) was 7,613, ranging from 3,902 to 9,862.

There are two reasons why the sample period for our analysis starts in April 2006. Firstly, the trade-by-trade data are only available from July 2005 onwards, which allows us to calculate the open positions of traders by tracing their transaction records for the quarterly month contracts from June 2006, and for non-quarterly month contracts from September 2005. Secondly, foreign investors were not allowed to trade futures (with the exception of their hedging needs) prior to March 2006, after which this restriction was lifted.

The first index price in the database is provided at 9:01, reflecting the non-synchronized trades for all stocks during the open session. Therefore, \( P_{t}^{open} \) represents the index price at 9:01 on day t.

It is worth noting that the cash index in the closing and opening sessions are more volatile than the index in the midday sessions. Chang et al. (2011) argue that the high volatility in the closing and opening sessions is attributable to the new information regarding common market factors or firm-specific fundamentals. Their explanation does not contradict our theory because our examination focuses on the difference in close-open volatilities between settlement days and non-settlement days.

Since the sample size in the ‘new settlement procedures’ period (37 settlement days and 734 non-settlement days) is smaller than the sample size in the ‘previous settlement procedures’ period (77 settlement days and 1,543 non-settlement days), the insignificance of the difference in the volatility between settlement days and non-settlement days could be due to the low power of the test resulting from the small sample size.

As a comparison, we calculate the daily trading volume and open interest in the nearest contract for the TAIEX futures, and find that the average daily trading volume is 50,921 contracts, whereas the average daily open interest is 41,495 contracts. Open interest for final settlement therefore accounts for approximately 48 percent of the daily open interest, or 39 percent of the daily trading volume.

The differences observed between the long and short positions imply that traders offset their open interest positions in TAIEX futures with their opposite open interest positions in Mini-TAIEX futures. Our summary statistics from Panels A and B of Table 3 suggest that the net effect is that traders offset their long positions in TAIEX futures with short positions in Mini-TAIEX futures.

The aim of the categorization used in this study is to divide all institutional investors (including foreigners, proprietary traders, and other institutions) evenly into four groups. The results are quantitatively similar if we use other cut-points to group traders.

Bris et al. (2007).