Assume some bacteria have capsular antigen. Do these bacteria always have a capsule? I think not, I think the antigen only makes it possible.
In short: It can depend on your definition of encapsulation.
According to the Oxford Dictionary of Biochemistry, a capsular antigen is defined as
any of the antigens, usually polysaccharide in nature, that are carried on the surface of bacterial capsules. The term K antigen is used for those that mask somatic (O) antigens.
In general, capsule polysaccharides have complex structures and are produced by a the action of several enzymes, with the genes necessary for production of the capsule typically making up a gene cluster in the genome (see for example Chen et al.: http://www.biomedcentral.com/1471-2180/10/274). On the face of it, it would seem somewhat unlikely that a bacterium would produce a known capsular polysaccharide antigen without retaining the ability to produce a capsule, as the selective advantage of the capsule then would be lost.
In actuality, the answer might depend on how you define "have capsular antigen" and "capsule". Historically, several methods have been used for demonstrating encapsulation. Methods based on staining microscopy or colony morphology observations tend to identify fewer strains as encapsulaed than does serotyping (reaction of serum with antigen). With this in mind, strains that can be shown to be encapsulated without serotyping are sometimes referred to as "highly encapsulated" to distinguish them from strains that test positive for encapsulation through serotyping only.
For discussions of capsule detection methods as applied to Staphylococcus aureus, see Sutra et al. (http://jcm.asm.org/content/28/3/447.full.pdf) and O'riordan and Lee (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC321462/). O'riordan and Lee writes
Isolates belonging to the remaning serotypes produce nonmucoid colonies on solid medium, and their colony morphology is indistinguishable from that of strains lacking a capsule. Some investigators have referred to nonmucoid, encapsulated S. aureus isolates as microencapsulated to distinguish them from the atypical mucoid strains.
Thus, if both "have capsular antigen" and "have a capsule" is defined by a positive serotyping result for a capsular antigen, then all strains with a capsular antigen as detected by serotyping have a capsule per definition. If on the other hand you define encapsulation based on ink staining microscopy or colony morphology, there are definitely strains that produce K-antigens and don't qualify as "encapsulated".
Note that while bacterial capsules consist mostly of polysaccharides, capsule proteins make up a second class of capsule antigens. One example is the Caf1 capsule antigen protein produced by Yersinia pestis: http://en.wikipedia.org/wiki/Caf1_capsule_antigen