Copying a reference creates a copy of the original variable (original and the new one are on different areas), moving a local variable casts to a rvalue your local variable (and again, original and the new one are on different areas).
From the compiler point of view, move may be (and is) faster:
#include <string>
void MyClass(std::string title){
  std::string title2 = std::move(title);
}
translates to:
MyClass(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >): # @MyClass(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >)
        sub     rsp, 40
        mov     rax, rdi
        lea     rcx, [rsp + 24]
        mov     qword ptr [rsp + 8], rcx
        mov     rdi, qword ptr [rdi]
        lea     rdx, [rax + 16]
        cmp     rdi, rdx
        je      .LBB0_1
        mov     qword ptr [rsp + 8], rdi
        mov     rsi, qword ptr [rax + 16]
        mov     qword ptr [rsp + 24], rsi
        jmp     .LBB0_3
.LBB0_1:
        movups  xmm0, xmmword ptr [rdi]
        movups  xmmword ptr [rcx], xmm0
        mov     rdi, rcx
.LBB0_3:
        mov     rsi, qword ptr [rax + 8]
        mov     qword ptr [rsp + 16], rsi
        mov     qword ptr [rax], rdx
        mov     qword ptr [rax + 8], 0
        mov     byte ptr [rax + 16], 0
        cmp     rdi, rcx
        je      .LBB0_5
        call    operator delete(void*)
.LBB0_5:
        add     rsp, 40
        ret
However,
void MyClass(std::string& title){
  std::string title = title;
}
generates a bigger code (similar for GCC):
MyClass(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >&): # @MyClass(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >&)
        push    r15
        push    r14
        push    rbx
        sub     rsp, 48
        lea     r15, [rsp + 32]
        mov     qword ptr [rsp + 16], r15
        mov     r14, qword ptr [rdi]
        mov     rbx, qword ptr [rdi + 8]
        test    r14, r14
        jne     .LBB0_2
        test    rbx, rbx
        jne     .LBB0_11
.LBB0_2:
        mov     qword ptr [rsp + 8], rbx
        mov     rax, r15
        cmp     rbx, 16
        jb      .LBB0_4
        lea     rdi, [rsp + 16]
        lea     rsi, [rsp + 8]
        xor     edx, edx
        call    std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >::_M_create(unsigned long&, unsigned long)
        mov     qword ptr [rsp + 16], rax
        mov     rcx, qword ptr [rsp + 8]
        mov     qword ptr [rsp + 32], rcx
.LBB0_4:
        test    rbx, rbx
        je      .LBB0_8
        cmp     rbx, 1
        jne     .LBB0_7
        mov     cl, byte ptr [r14]
        mov     byte ptr [rax], cl
        jmp     .LBB0_8
.LBB0_7:
        mov     rdi, rax
        mov     rsi, r14
        mov     rdx, rbx
        call    memcpy
.LBB0_8:
        mov     rax, qword ptr [rsp + 8]
        mov     qword ptr [rsp + 24], rax
        mov     rcx, qword ptr [rsp + 16]
        mov     byte ptr [rcx + rax], 0
        mov     rdi, qword ptr [rsp + 16]
        cmp     rdi, r15
        je      .LBB0_10
        call    operator delete(void*)
.LBB0_10:
        add     rsp, 48
        pop     rbx
        pop     r14
        pop     r15
        ret
.LBB0_11:
        mov     edi, offset .L.str
        call    std::__throw_logic_error(char const*)
.L.str:
        .asciz  "basic_string::_M_construct null not valid"
So yes, std::move is better (under these circumstances).