The official Japanese intervention data consists of daily volumes of intervention operations in the JPY/USD foreign exchange market.
During the period under study, January 1999 to March 2004, all official interventions in the JPY/USD market are sales of JPY against purchases of USD.12

12:The U.S. government did not intervene in the JPY/USD exchange rate market during this period.

Table 1 shows intervention data summary statistics.
The table shows that Japan intervenes in the JPY/USD exchange rate market on a total of 159 days over the full sample period.

On most intervention days the magnitude of intervention is substantial, with purchases of over USD 1,000 million on 113 days and only 20 days with purchases of less than USD 250 million.

The table shows that only 30 of the intervention days occur between January 1999 and December 2002, 82 intervention days occur during 2003, while 47 intervention days occur during the first quarter of 2004.13  

13:Fatum and Hutchison (2006b) show that the described variation in intervention frequencies over the January 1999 to March 2004 time period is consistent with the existence of three different intervention reaction function regimes.

In order to compare the exchange rate effect of interventions that the market seem aware of to interventions that the market seem unaware of and, in turn, attempt to shed light on through which transmission channel intervention works, rumors of intervention are taken into account.

The analysis distinguishes between a rumor of intervention and a firm report of intervention, and only incorporates the former for the following reason.

A rumor of intervention occurs on the same day that the market thinks an intervention takes place, while a firm report of intervention typically occurs the day after the intervention has taken place.14

Therefore, whether or not an intervention is reported is generally a matter of “after-the-fact” information that can not play a role in the contemporaneous exchange rate response to intervention.

By contrast, whether or not an intervention coincides with a rumor seems a better indicator of whether the market is aware of the intervention operation the same day it is carried out.

Moreover, whether the market is aware of the intervention operation may affect the same-day market reaction to intervention as well as trigger a same-day market reaction to the rumor itself (whether or not intervention actually occurs.) 15

14: For example, a firm report of the January 12, 1999 official Japanese intervention operation is reported on the newswire on January 13, 1999.

15: For completeness, Factiva was also gleaned for firm reports of intervention.
For the full sample, a total of 31 firm reports of intervention were found. To compare, Chang (2006) finds 27 firm reports of intervention in the Wall Street Journal over the January 2000 to March 2003 time period.

While most Bank of Japan interventions are unreported, no false firm reports of intervention were found.
By contrast, more than half of the interventions under study coincide with a rumor and, furthermore, several “false” rumors of intervention (i.e. a rumor of intervention when no intervention takes place) were found.
This further illustrates the importance of distinguishing between reports of intervention and rumors of intervention.

The Factiva search engine and a comprehensive combination of various search words (e.g. Bank of Japan, intervention etc.) are used to find the days with a rumor of intervention.
The second row of Table 2 shows that a total of 269 days across the full sample are associated with a rumor of intervention.
Row three of Table 2 reports that 92 of the rumor days are also intervention days, i.e. 92 of the 269 rumors are “true”.

Row four shows that, accordingly, the remaining 67 of the 159 intervention days in the full sample do not coincide with a rumor of intervention. For the full sample, as many as 177 rumor days are, in fact, “false”.
The numbers of days associated with false rumors are reported in row five. 16

16: It is not surprising to find a large number of false rumors of intervention. Chang (2006) reports a total of 282 JiJi News (local Japanese newswire) and Wall Street Journal reports of rumors and speculation of Japanese intervention over the January 2000 to March 2003 time period.

Moreover, other studies question the accuracy of newswire reports of intervention (see Fischer 2006 and Osterberg and Wetmore Humes 1993).

Some studies suggest that official statements regarding threats of intervention or regarding the desired direction of future exchange rate movements (sometimes referred to as “oral interventions”) influence exchange rate.

In order to take into account this possibility, the analysis uses the Factiva search engine to find, respectively, newswire reports of official statements in support of intervention and/or a weaker JPY (“positive statements”), and newswire reports of official statements suggesting that further intervention in the JPY/USD rate is not recommended or unlikely (“negative statements”).

Rows six and seven of Table 2, respectively, report a total of 108 positive and 17 negative statements for the full sample period.