LALR parsers have an equal number of states as an LR(0) parser. LR(1) parsers have a number of states dependent on the lookahead. Often different states are identical, except they have a different lookahead. i.e.
Sample LR(1) item sets:
1)
A -> B.C,'x'
A -> E.C,'y'
2)
A -> B.C,'g'
A -> E.C,'h'
Equivalent LALR(1) item set:
1)
A -> B.C,'x'
A -> E.C,'y'
A -> B.C,'g'
A -> E.C,'h'
Fewer states means fewer shifts and reduces. And at least as importantly, a much smaller parse table.