left outer join을 사용해야 되지 않을까요?

drop table dbo.tSurprise
create table dbo.tSurprise
(	cDate		char(8)	not null
,	cNumber		int		not null
,	cSubType	int		not null
,	cIsOpen		int		not null
,	cOpenDate	char(8)	not null
)

truncate table dbo.tSurprise
insert into dbo.tSurprise
select	convert(char(8),dateadd(day,cGroup,getdate()),112)
	,	row_number() over(partition by cGroup order by (select 1))
	,	1
	,	abs(cRand) % 2
	,	convert(char(8),dateadd(day,cRand%5,getdate()),112)
from	(	select	top 3000
					checksum(newid()) as cRand
				,	checksum(newid()) % 2 as cGroup
			from	syscolumns as a
				,	syscolumns as b
		) as a
go

select	a.*, b.*
	,	case
			when a.cIsOpen != b.cIsOpen then N'공개변경'
			else ''
		end as cChanged
from	dbo.tSurprise as a
		left outer join
		dbo.tSurprise as b
	on	b.cNumber = a.cNumber
	and	b.cSubType = a.cSubType
	and	b.cDate = '20160619'
where	a.cDate = '20160620'
go