According to this paper
http://www.arxiv.org/abs/hep-th/0401160
as of last December they hadn't. But check out the last part where they try to show that supermassive strings can change into black holes - smoothly!
In my Zwiebach book, chapter 16 discusses how they do the extremal black hole (this is, he says , the simplest example) in 5 dimensions. They start in supersymmetric type IIB superstring theory and compact five of its 10 dimensions on tiny circles. This leaves a 5 dimensional Minkowski space uncompacted and the black hole sits in it. The theory assumes zero string coupling constant, but the supersymmetry will preserve the physical effects calculated with zero coupling even at finite coupling, which is a main reason for choosing this particular configuration.
They wrap D1 and D5 branes around the compacted circles in a particular way, these branes, being compacted, look to an observer in the 5-space like points. They are to be placed at the center of the black hole. This will reproduce the physical properties of the black hole (I left out some byplay with charges), but they do it in only one way, and the point of the excercise is to count the many ways of building the black hole. The total momentum is an integer N, and what they have to do is to count the number of ways N can be partitioned into momenta from physical states.
The key is that there are open strings between the branes, with their ends fixed on them. Zwiebach cites three facts about this configuration:
(1) The D1/D5 brane system is a bound state. Strings between two D1 branes or two D5 branes become massive and idle and do not contribute to the count.
(2) A string from a D1 brane to a D5 brane has an opposite string going the other way, and between them these two strings have eight modes: four bosonic and four fermionic.
(3) The Q
1 D1 branes can join to form a brane wrapped Q
1 times around one of the compact circles and the Q
5 D5 branes can join to wrap Q
5 times around the whole compact space.
According to (2) there are 4Q
1Q
5 ways to assign bosonic string states, and an equal number of sdifferent fermionic state. But according to (3) the string endpoints can be assigned to any of the wrapppings. With this extra fact included, the count partition count comes out correct.