PROVIDE Project Technical Paper 2005:1
February 2005
P2102Total | |
26835 |
1152.961 |
4247.576 |
0 |
360000 |
P2102Total~K | |
26835 |
1152.961 |
4247.576 |
0 |
360000 |
P2103Total | |
26835 |
2603.309 |
27987.51 |
-20000 |
3000000 |
P2103Total~K | |
26835 |
2603.309 |
27987.51 |
-20000 |
3000000 |
P2104Total | |
26835 |
4470.326 |
39426.95 |
0 |
3105800 |
P2104Total~K | |
26835 |
4470.326 |
39426.95 |
0 |
3105800 |
P2105Total | |
26835 |
432.705 |
7858.462 |
0 |
874000 |
P2105Total~K | |
26835 |
432.705 |
7858.462 |
0 |
874000 |
3.2.6. Annualising and creating control totals (annualise.do and totals.do)
Some of the income and expenditures reported in the IES 2000 questionnaire is weekly or
monthly, and should be converted to annual figures. This do-file simply changes all the
weekly or monthly figures to annual figures. Do-file totals.do recalculates the income and
expenditure sub-totals and also creates variables totincCHECK and totexpCHECK, which can
be used to make sure that the mapping of income and expenditure in do-files mapinc.do and
mapexp.do is done correctly. At the end of do-file totals.do food expenditure values that are
missing or zero are imputed. A similar process if followed for missing or zero tax expenditure
values when the total income level of the household creates the expectation that the household
should have reported tax expenditure. A discussion of food and tax expenditure imputations
follows in section 3.2.7.
3.2.7. Imputing ‘missing’ food and tax expenditure values
The total food expenditure variable of all households reporting zero food expenditure was
changed to missing based on the assumption that each household has to at least report some
expenditure on food. These ‘missing’ food expenditure values were then imputed by
estimating a double-log Engel equation of the form
ln(Yi)=a+b.ln(Xi)+c.ln(Hi)+εi
where a, b and c are constants, Yi is the food expenditure (logfoodexp), Xi the total household
expenditure (logtotexp), Hi the household size (logH) and εi the error term. This double-log
formulation ensures that the share of total expenditure spent on food declines as total
expenditure increases, while larger households benefit from scale economies (see Van der
Berg et al., 2003b). The following regression results were obtained (sampling weights used):
Source | SS df MS
Number of obs = 25944
F( 2, 25941) =30994.49
Prob > F = 0.0000
R-squared = 0.7050
Adj R-squared = 0.7050
Root MSE = .47803
---------+------------------------------
Model | 14165.074 2 7082.537
Residual | 5927.7668 25941 .228509572
---------+------------------------------
Total | 20092.8408 25943 .77449951
48
--------------- logfoodexp | |
Coef. |
Std. Err. |
t |
P>|t| |
[95% Conf. |
---------- Interval] |
logtotexp | |
.5992633 |
.0025101 |
238.74 |
0.000 |
.5943434 |
.6041832 |
logH | |
.2185103 |
.0041317 |
52.89 |
0.000 |
.210412 |
.2266087 |
_cons | |
2.439185 |
.0247711 |
98.47 |
0.000 |
2.390632 |
2.487738 _ _ _ _ _ _ _ _ _ _ |
© PROVIDE Project
More intriguing information
1. Feeling Good about Giving: The Benefits (and Costs) of Self-Interested Charitable Behavior2. Yield curve analysis
3. The name is absent
4. The Folklore of Sorting Algorithms
5. A MARKOVIAN APPROXIMATED SOLUTION TO A PORTFOLIO MANAGEMENT PROBLEM
6. The name is absent
7. Restructuring of industrial economies in countries in transition: Experience of Ukraine
8. The name is absent
9. Importing Feminist Criticism
10. PROJECTED COSTS FOR SELECTED LOUISIANA VEGETABLE CROPS - 1997 SEASON