You have been given the ARDL(1,2) model below in the context of panel data

jQuery22405579869907500219_1628098363063? = ?1???1 + ?2??? + ?3???1 + ????2 +?+ ???(1)

Where ??? is an idiosyncratic error term and ? is a time-invariant and case-specific effect.

In this case, what is the Nickell bias? Is there an upward or downward bias, and if so, why?

What estimators are available for eliminating the fixed effect and achieving consistent coefficient estimates in Model 1? Which one is better in which situation?

Let X represent the entire amount of money (cash) kept at home by a family. With a sample size of n = 100, we get a sample mean of 800 euros and a standard error of 50 euros. 500 euro represents the sample standard deviation.

(a) 95% confidence E[X].

(b) 99% confidence E[X].

(c) interval that will approximately 95% of the outcomes of X.

(d) Do you think that the distribution of X is normal?

(e) Would it be a good idea to construct my sample by interviewing 100 persons waiting at the railway station? Would you suggest another way of selecting a sample? Which one?

3. In solving the bivariate regression model

yi = xi + i,

with i i.i.d. with mean 0 and variance 2, you face the problem that you cannot perfectly observe xi. You have access to a random sample of information {yi, xi}Ni=1, where

xi = xi + vi,

where vi is i.i.d. with mean 0 and variance 2v (x, , and v can all be considered independently distributed with respect to one another). The variables yi and xi are expressed as deviations from their sample means. If you have information that 2v = 1 (from an earlier study), can you form a consistent estimator of ? If so, what is it?