first I think your equation is wrong

f(t) = s(t) + i(d) -i(foreign)

second, what is the domestic currency, obviously US dollar is assumed to be foreign currency

As we can see, if domestic currency is RMB, US dollar is depreciate

forget to say the above equation is under CIP

Under UIP

YOU WILL HAVE Et(S(t+1)) = s(t) + i(d) -i(foreign)

哦，对不起我忘记说了我这个用的是indirect exchange rate，1RMB=？USD公式没问题的，只是我的模型做出来和theory不一样

so if your equation is correct

then your conclusion is wrong

as you can see, forwrd rate is positively correlated with i(foreign)

is i(foreign) increase, that means your forward rate will increase

that means the domestic currency will depreciate

foreign currency will appreciate, which is contradictory with your conclusion foreign currency will depreciate

you need to check it.

By the way, the formula above I give is from mathematic finance, if you are doubted about it, you can check this book.

我用的是ARIMAX 对random walk process做经验学习，我觉得parity theory是neo-classical theory,所以我没有加varibles什么都没有动，就加了几个lag,对可能的multicolinerity进行了一下处理，结果与theory相反的，所以我想exchange rate的调整很可能是凯恩斯式的，所以可能达到一个partial equilibrium，而不是GE，所以我觉得这个适用于长期而不是短期。