Creating a 2000 IES-LFS Database in Stata



PROVIDE Project Technical Paper 2005:1
which may explain partly the large gap between SARB and IES 2000 as far as income from
property is concerned (see footnote 15).

February 2005


As far as income from transfers is concerned, the crude assumption that income from
transfers from other households minus transfer payments to other households is equal to net
transfers from the rest of the world presents a possible explanation for the large difference
here between the SARB 2000 and IES 2000 data. Note that transfers from the rest of the
world are not included in the IES 2000. Net inter-household transfers, which, according to the
assumption made earlier is equal to net income from the rest of the world, equals R930 on
average per household in the IES 2000, which is 2.8% of household income. The comparative
figures in the SARB 2000 data indicate that net foreign transfers only contributed 0.02% to
total household income.

Income tax is hugely underreported in the IES 2000 data. According to the IES 1995
figures the average income tax rate was 8.6% in 1995. This has dropped to 7.5% in 2000.
20 In
the SARB 2000 the comparative average tax rates, calculated here as tax on income and
wealth divided by current income, was 13.0% in 1995 and 13.8% in 2000. Although there
have been some reductions in marginal income tax rates between 1995 and 2000, improved
tax collections and a broadening of the tax base would have counteracted the impact of tax
relief on average tax rates. This suggests that the SARB 2000 is more likely to be correct and
that households underreported income tax in the IES 2000. Section 2.4.3 extends the
investigation into taxes by looking at tax rates by expenditure deciles.

The IES 2000 questionnaire asks respondents to indicate the amount of tax paid in the last
12 months, be it provisional payments or PAYE and SITE deductions from salary. The
financial year-end is at the end of February, while the IES 2000 was conducted around
September/October. At this stage only about six months’ worth of tax payments would have
appeared on the salary slips. It is likely that many respondents simply failed to include tax
payments made for the last six months of the previous tax year ending February 2000. It is
further unclear why Statistics South Africa chose to include income tax as a household-level
variable while clearly working individuals - who are required to provide other wage and
salary information in any case - would have been able to give a better indication of their tax
payments for the last 12 months. Furthermore, an unexpectedly large number of households
failed to report any income tax, causing average rates to be lower than expected.

The reported savings in the IES 2000 appears to be very high. The savings variable in the
IES 2000 is defined fairly broadly and is made up of various items including the capital

20 These figures vary slightly depending on the way average income tax is calculated. The figure here is the
average of all household tax rates rather than the rate calculated by dividing average income tax by
average income (as was the case in Table 5).

22

© PROVIDE Project



More intriguing information

1. The name is absent
2. THE USE OF EXTRANEOUS INFORMATION IN THE DEVELOPMENT OF A POLICY SIMULATION MODEL
3. The name is absent
4. THE CHANGING STRUCTURE OF AGRICULTURE
5. The name is absent
6. A novel selective 11b-hydroxysteroid dehydrogenase type 1 inhibitor prevents human adipogenesis
7. The name is absent
8. Problems of operationalizing the concept of a cost-of-living index
9. A simple enquiry on heterogeneous lending rates and lending behaviour
10. The name is absent
11. THE AUTONOMOUS SYSTEMS LABORATORY
12. The name is absent
13. Reconsidering the value of pupil attitudes to studying post-16: a caution for Paul Croll
14. The Economic Value of Basin Protection to Improve the Quality and Reliability of Potable Water Supply: Some Evidence from Ecuador
15. The name is absent
16. How much do Educational Outcomes Matter in OECD Countries?
17. The Clustering of Financial Services in London*
18. ‘Goodwill is not enough’
19. Naïve Bayes vs. Decision Trees vs. Neural Networks in the Classification of Training Web Pages
20. Comparison of Optimal Control Solutions in a Labor Market Model