Maximum Likelihood Estimators when $F(x) = 1 - (b/x)^a$

63 Views Asked by At

Suppose that random variables $X_i, \ldots, X_n$ are independent and identically distributed with the CDF:

$$F(x|a,b) = 1-\left(\frac{b}{x}\right)^a$$

With the conditions $x \ge b, b \gt 0$ and $a \gt 0$. And for $x \lt b$, the distribution is $0$.

What are the maximum likelihood estimators for $a$ and $b$?

I approached this by taking the derivative of $F(x)$ to get the PDF, $f(x)$. Then, I did the product from $i=1$ to $i=n$ of $f(x_i)$ to get the likelihood functoin. I took the log of the likelihood, then partial derivatives with respect to $a$ and $b$, set the partials to $0$, and ended up with this MLE for $a$:

$$a = \frac{n}{\sum_{i=0}^n (x_i) - n\ln(b)}$$

For $b$, I got this as the partial derivative w/respect to $b$:

$$\frac{an}{b}$$

So if we set that to $0$, there is no solution and I can't get the MLE of $b$. Where do I go from here?

Thank you!