HGRENT is a measure of gross rent, i.e. it equal to the base rent plus utility costs (HGUTIL) but does not include heating (HGHEAT), electricity costs (HGELECTR) and additional gas costs (HGGAS). HGRENT is converted into Euro values for all waves including those prior to Wave S.

The questions for HGRENT and utility costs have changed considerably since the first SOEP questionnaire in Wave A. From Waves A to Wave G, the amount of rent stated by the households in SOEP is in principle the desired concept of gross rent, i.e. basic rent excluding heating and electricity costs but including utility costs. However, in these waves the information on utility costs was not collected.

From Wave H onwards, households simply state the amount of rent they pay. Following this question, it is asked whether heating and utility costs are included in that amount of rent and what the exact costs for heating and utilities eventually are (in the latter case only if they are included).

In Wave BE and Wave BF, the questionnaire design was different to all other waves. In a first step, households were asked for their overall monthly housing costs (sum of base rent, heating costs, additional utility costs, electricity and gas). In a second step, they stated the respective expenses. In order to obtain RENT$$, heating, gas, and electricity costs have to be deducted from these overall monthly housing costs. In Wave BE, some households state monthly housing costs that presumably do not include all possible expenses. They might do so because they do not know all their expenses or because they misunderstood the question. We attribute the drop of rent from Wave BD to Wave BE and Wave BF to the different questionnaire design and do no longer adjust the values as we did in data release v31. Therefore, the average rent of Wave BE (data release v31) is about 10 Euros lower in data release v32 and v33 than it was in v31. Both, in Wave BE and Wave BF, implausible values of rent (e.g. if utility costs are above total living costs) have been set to missing to be imputed in the following step.

The questionnaire design in Wave BG was set back to the format of Wave BD.

Missing values of HGRENT are treated as follows: In a first step, past values of the two last years are carried forward and adjusted for inflation given that the household still lives at the same address and the dwelling's size has remained the same. The type of rent (heating/utility costs included or not) is then also carried forward. In a second step, if HGRENT is still missing, values of the two subsequent years are carried backwards in the same manner. In a third step, the remaining missing values of HGRENT are imputed by Stata's chained imputation procedure. Imputed values are not carried forward.

General Information on the Imputation Procedure

Since data release v31, missing values of HGROOM, HGSIZE, HGRENT, HGHEAT, HGUTIL, and HGELECTR are imputed with Stata's chained imputation procedure (mi impute chained). We impute separately for each wave, for East and West and for owners and tenants. The regression specifications vary between these groups but in general they include:

• For owner:
  • HGROOM, HGSIZE, HGCONDIT, HGCNSTYRMAX
  • Number of persons in household
  • Household net income as stated in questionnaire
  • Type of dwelling (family house, apartment, etc.)
• For tenants:
  • HGROOM, HGSIZE, HGRENT, HGHEAT, HGUTIL, HGELECTR, HGCONDIT, HGCNSTYRMAX, HGEQPHEA, HGEQPFHEA, HGEQPTER, HGEQPGAR, HGEQPSOL, HGEQPAIR, HGCNSTYRMAX, HGREVAL, HGREDUC
  • HGSIZE squared
  • Number of persons in household
  • Household net income as stated in questionnaire
  • Type of dwelling (family house, apartment, etc., $$wum1)
  • Type of tenant
  • Dummy: HGHEAT included in HGRENT (Wave G to Wave BD)
  • Dummy: HGUTIL included in HGRENT (Wave H to Wave BD)
  • Dummy: HGUTIL partly included in HGRENT (Wave H to Wave BD)
  • Dummy: HGHEAT includes electricity costs (since Wave BE)
  • Dummy: household pays only utility costs (since Wave BE)
  • Household has children younger than 16

In addition, the regression of HGRENT contains HGRSUBS, a dummy for children under 16, information on the residential area ($$wum3) and BIK regions. The regression of HGHEAT includes a dummy for children under 16. The regression of HGUTIL includes HGEQPLIF and omits HGRENT, HGCONDIT, the type of dwelling and the dummy for whether HGHEAT is included in HGRENT. The regression of HGSIZE omits HGSIZE squared.

HGRENT is not normalized across households before imputation and may include heating and utility costs or not. Solely partially included utility costs are subtracted. We use dummies for HGHEAT/HGUTIL included in HGRENT to account for the different types of rent. Note that these dummies are also imputed if they are missing. If HGUTIL contains a partial amount, it is set to missing and then also imputed.

Residents of homes (HGOWNER==5) are excluded from the imputation. Values larger than four times the 99th percentile in one of the target variables are also excluded.

For all six target variables, the predictive mean matching imputation method is applied. We use 175 iterations as burn-in period. Furthermore, we only distribute one imputed value for each observation.

Gross warm rent including electricity (Bruttowarmmiete inkl. Strom) could be computed from this data as follows: HGRENT + HGHEAT + HGELECTR.

HGRENT is set missing (=-5) for all M3 and M4 respondents although the migration household questionnaire surveyed rent information. However, the surveyed information for M3 and M4 is not comparable with the information for the other samples as information on HGHEAT, HGUTIL, HGELECTR, etc. is not surveyed.