Statistics Exercises for Economists
Uploaded by
lattut50Statistics Exercises for Economists
Uploaded by
lattut50Common questions
Powered by AIIf X is log-normally distributed with parameters mu and sigma^2, then X = e^(Y) where Y is N(mu, sigma^2). If Y is shifted by ln(k) for any real number k, then the product kX = k * e^(Y) = e^(ln(k) + Y) maintains the log-normal form since Y retains normality under affine transformations. Thus, kX is also log-normally distributed.
To show that Y = -2 ln X has a chi-squared distribution with 2 degrees of freedom, start with the property that if U ~ Uniform(0,1), then - ln U follows an exponential distribution with mean 1. Thus, Y being -2 times ln U gives Y an exponential distribution with mean 0.5, which is equivalent to a chi-squared distribution with 2 df, since a chi-squared variable with 2 df is just twice an exponential(1) random variable.
The covariance between two independent random variables is zero because their joint density function can be expressed as the product of their individual densities, affirming independence. To verify independence, check if the joint density f(x1, x2) equals the product f1(x1)f2(x2) for all x1, x2 in their domains. For example, if f(x1, x2) = (x1 + x2) on (0, 1), both random variables are independent since there is no interaction. Therefore, their covariance is zero.
The negative exponential utility function implies risk aversion, with the individual preferring less uncertainty in wealth W. If W is N(mu, sigma^2) distributed, the expected utility E[U(W)] is influenced by the mean and variance of W. The individual maximizes E[U(W)] by choosing options that reduce variance while maintaining or increasing expected returns, adjusting their choice variable X to optimize mu = f(X) and sigma^2 = g(X). The optimal behavior involves trading off variance and expected return to maximize utility.
The variance of Y = exp(X1 + X2) can be calculated using the properties of log-normal distributions. Since X1 + X2 is normally distributed with mean (1 - 1) = 0 and variance 1 + 1 = 2, Y follows a log-normal distribution. The variance of Y, given the parameters, can be computed using Var(Y) = (exp(sigma^2) - 1) * exp(2mu + sigma^2), resulting in Var(Y) = (exp(2) - 1) * exp(0) = exp(2) - 1.
If Y = ln X is N(mu, sigma^2), then X is log-normally distributed and the density function is f_X(x) = (1/(x sigma sqrt(2pi))) * exp(-((ln x - mu)^2)/(2 sigma^2)) for x > 0. To find E[X^k], leverage the property that when X is log-normal, E[X^k] = exp(k mu + (k^2 sigma^2)/2). This uses the moment-generating characteristic of the log-normal distribution.
The moment generating function (MGF) M_X(t) of a random variable X is defined as E[e^(tX)]. For X as in question 4 (Bernoulli), M_X(t) = (1-p) + p * e^t. To derive the mean, differentiate M_X(t) with respect to t and evaluate at t=0, yielding E[X] = M_X'(0) = p. For variance, derive the second central moment M_X''(0) - (M_X'(0))^2 = p(1-p). Using MGF for distribution derivation highlights characteristic function analysis.
To find the joint discrete density of X1 and X2, we consider all possible outcomes when drawing two balls without replacement from an urn of four balls numbered 1 to 4. The joint discrete density function is determined by calculating the probability of each outcome where X1 is the first ball drawn and X2 is the maximum of the two numbers drawn. For example, if the first ball is 1 and the second is 2, then (X1, X2) = (1, 2) with probability 1/12, and so on for all possible draws, considering permutations of ball numbers.
To determine the efficiency level Z = Y/f(K, L), you compute E[Z] assuming X follows N(0,1). Since U = exp(-X^2), the mean efficiency is calculated by taking E[U] = E[exp(-X^2)]. Integration over X provides Z = 3^(-0.5) as derived from probability transformations and expectations.
For n independent random variables, each having a Bernoulli distribution with parameter p, the sum Y of the variables is binomially distributed with parameters n and p. The mean of Y is given by E(Y) = np, and the variance is given by Var(Y) = np(1 − p). This is derived because each Xi has expectation p and variance p(1 − p), and independence allows the sum of variances.