First, etymological arguments are inconclusive.
Both -ant and -ent, as suffixes in English, are used to refer to personal agents. All of these etymologies will derive from the Oxford English Dictionary:
regent (from Latin present participle regent-, regens, from verb regere)
claimant (from French derived verb claim + -ant added to signal personal agency, similar to "defendant" or "appellant")
appellant (from French present participle appellant, from verb appeller)
defendant (from French present participle defendant, from verb defendre)
One would be tempted to conclude that -ant forms always correspond to French derivations and -ent forms always correspond to Latin derivations. However, words that have both possibilities make a decision between them:
assistant (From French present participle assistant, from verb assister, but also Latin assistent-, assistens, from verb assistere)
confidant (from French noun confident and overlapping with English adjective/noun confident, and also Latin present participle confident-, confidens, from verb confidere; was confident, but the personal noun form switched to confidant in the 18th century)
Cedent/cedant has both possibilities. It could derive from the Latin present participle form cedent-, cedens, from the verb cedere, to withdraw or give up. It could also derive from the French present participle form cédant, from the verb céder. This could depend on whether cedent comes from Roman or French legal jargon. However, determining that would provide no firm answer. No matter what etymological argument can be used for its spelling, later usage could switch the -ent to an -ant (as happened to the noun confident -> confidant). So etymology is not reliable here.
In modern use individual organizations make a decision about which spelling to favor.
To better determine spelling, I used three modern corpuses of English speech and writing: the Corpus of Contemporary American English, the British National Corpus, and the Corpus of US Supreme Court Opinions. These tools would give me an intersection of Englishes in both media and official contexts. (I had a hunch that cedent/cedant might differ based on discourse community.)
- COCA: no hits for either word. (One proper noun hit seemed incidental.)
- British National Corpus: no hits for either word.
- Corpus of US Supreme Court Opinions: no hits for either word.
These results indicate that these terms aren't used much in their respective contexts.
So I added in the largest BYU corpus, the iWeb Corpus, to see if I could find results. I did: 101 for cedent, and 65 for cedant. Focusing further on individual subdomains:
The actuarial forum at acted.co.uk had users using both cedent and cedant. In at least one case the same poster used both spellings.
Domains conning.com, asianlii.org, stikeman.com, lloyds.com, and qbe.com have multiple results attesting the cedant spelling. They appear to be a mix of private firms and legal resource pages.
Domains hoganlovells.com, austlii.edu.au, crowell.com, and irmi.com have multiple results attesting the cedent spelling. Again, they appear to be a mix of private firms and legal resource pages.
A final point of interest: Eugene Wollan, JD, in a Reinsurance Interest Group newsletter article titled "Cedent or Cedant: Which Is Proper?" deliberately avoids answering the question:
My online dictionary defines cedent (or cedant, take your choice; as far as I’m concerned, it makes no difference) as a party who passes a financial obligation to an insurer or a reinsurer.
Unless the regulations in your community or organization clearly prefer a specific spelling, it looks like you have a choice.