##### what is the marginal effect of the age of wine on price based on the estimation outputs (1)-(4) given below.

In a small essay up to 1,500 words provide a clear answer, that is what is the marginal effect of the age of wine on price based on the estimation outputs (1)-(4) given below. Support your answer by a thorough discussion of the advantages and disadvantages of the estimated models and the applied sequential modelling procedure. In particular, you may wish to:

- compare outputs (1) and (2) and comment on what mistake would the researcher commit by

not including the interaction term France

age

in equation (1);

analyse the estimation outputs (1)-(4) to explain which one is your preferred specification; explain your choice by comparing the statistical characteristics of these models as well as their economic interpretation; comment on what other test(s) could be performed further that would help in improving the reliability of models’ estimation;

use your preferred model specification selected from (1)-(4) to explain whether its functional form requires improvement and, if so, explain why do you think so and what variables you would add to it;

criticise or justify the approach of adding more variables to the model that has been initially chosen by a researcher (that is estimating model (1) in the first instance and then adding more and more variables sequentially);

use your preferred model specification to decide whether it allows advising a wine investor who looks for a profit in a long-run to invest in French or Spanish wines.

In your essay, you may decide to address some, or all, points raised above, or concentrate on other issues that are not listed if you think it is important to address them.

The modelling procedure and estimated outputs are presented below.

The data are from randomly selected 230 wineries. Four following regression models have been estimated sequentially. That is, model (1) had been estimated first; then, after inspecting its estimates, model (2) was estimated next and so on.

ln(Price ) 3.41 0.04671age 0.0019temp

0.0211France ,

i (0.512) (0.015)

i (0.009)

i (0.020) i

R2 0.3537, R2

0.3451,

(1)

AIC 2013.313, BIC 2001.067, p _ RESET 0.000, i 1,…, 230,

ln(Price ) 2.36 0.0250 age 0.0021temp

0.0109 France 0.0130(France age ),

i (0.412) (0.015)

i (0.009)

i (0.020)

i (0.004) i i

R2 0.3652, R2

0.3601,

(2)

AIC 2154.513, BIC 2132.067, p _ RESET 0.000, i 1,…, 230,

ln(Price ) 3.01 0.03618age 0.0022temp

0.0081Spain 0.0079(Spain age ),

i (0.301) (0.021)

i (0.005)

i (0.017)

i (0.210) i i

R2 0.3650, R2

0.3537,

(3)

AIC 2150.016, BIC 2167, p _ RESET, 0.000 , i 1,…, 230,

ln(Price ) 1.83 0.0185age 0.0015temp 0.0101France 0.0054 Spain

i (0.165) (0.028)

i (0.075)

i (0.031)

i (0.073) i

0.0064(Spaini agei ) 0.0111(Francei agei ),

(0.312) (0.110)

(4)

R2 0.4127, R2 0.3969,

AIC 2166.114, BIC 2168.031, p _ RESET 0.000, i 1,…, 230.

where Price is the pre-tax price of a 750 ml bottle of wine (expressed in pounds); age is the age of the wine (measured in years); temp is the average temperature over the growing season for the grapes (expressed in Celsius degrees); France and is a dummy variable which equals 1 if the bottle of wine was French and equals 0 if the wine came from another country; Spain is a similar dummy

variable for Spain. Spain age

and France age

are products of the respective variables. AIC and

BIC are values of Akaike and Schwartz Information Criteria, respectively. The p-values for RESET test are denoted as p_RESET. The p-values for testing the hypothesis that particular regression coefficients in equations (1)-(4) are equal to zero against the two-sided alternative are given in

parentheses beneath the parameters’ estimates, R2

and

2

adj

stand for the coefficient of

determination and adjusted coefficient of determination correspondingly.