I had the same issue where I just needed to delete folders & subfolders after a document has been deleted.  Louis-Simon Binette (see below) had a good suggestion to created a list of folders from the starting entry, sort them by longest first, then check if emtpy.

Here is the workflow details.  You can run it manually or can use "Invoke Workflow" to call it with a parameter:

<Assign Token Value> Titled, "Create a Folderlist Token"
 Create the following tokens:
 • %(mFolderList) (multi-value)  (token value is blank)
 • %(mEntryPath) = %(Entry Path)

<Conditional Decision> Titled, "Is the starting entry a folder?"
  Yes - a folder:
  If all of these conditions are true
    Starting Entry : Type equals Folder

    If entry is a folder - add the entryname to the path:
    Modify the following tokens:
    • %(mEntryPath) = %(Entry Path)\%(Entry Name)

  No - not a folder:
  continue

<Search Repository> Titled, "Search for folders starting at entry path"
 Search Query:
  {LF:LOOKIN="%(mEntryPath)"} & {LF:Name="*", Type="F"}
 Return: All results

<For Each Entry>  Titled, "For Each Entry" 
    --- loops through the search results to get a list of folders then we'll see if empty -
 
 Iterate through the entries in:
 (Search for folders starting at entry path::Output Entries)

  Assign a Token - append to a token for each folder found
   Modify the following tokens:
   • %(mFolderList): Append %(ForEachEntry_CurrentEntry_FullPath) (multi-value)

--- after the For Each Entry loop ---   
<Assign Token Value> Titled, "Create a Token for Sorting"
 Create the following tokens:
 --- notice the use of the function to sort decending ---
 • %(mSorted) = %(mFolderList#@SortDescending@#) (multi-value)
   
--- let's loop through starting with the longest path name to the shortest --- 
<Repeat> Titled, "Repeat through for each path found"
 If all of these conditions are true
  %(Repeatthroughforeachpathfound_Iteration) is less than or
    equal to %(Searchforfoldersstartingatentrypath_Result Count)
 • Evaluate condition before running contained activities
 • Start the Iteration token at 1
 • Uncheck the "limit the number of repetitions...."

<Find Entry> Titled, "Find Entry"
  Finds the entry at:
  %(mSorted#[%(Repeatthroughforeachpathfound_Iteration)]#)

  --- to prevent a warning in WF results, add a Conditional Decision to check that the folder is empty ---
  <Conditional Decision> Titled, "Is the folder empty?"
  
    Yes - empty folder:
    If all of these conditions are true
      (Find Entry::Output Entry) : Document Count equals 0
      (Find Entry::Output Entry) : Folder Count equals 0

      <Delete Entry> Titled, "Delete Entry"
      Entry to delete: (Find Entry::Output Entry)
      Only delete empty folders

--- note: you could use the "Move Entry" instead of the "Delete Entry" for tracking --
    
    
    No - don't delete:
    do nothing
    
--- end of workflow ---

The shorter the depth of the path, the quicker it runs. 
So to run with only a few folders deep and not many sub-folders, it will take seconds. 

Hope this helps!

Chuck