Private Transfers

Intrahousehold transfers

(Updated by Gretchen 10/23/2008)

Households shift resources among their members. Those who are consuming more than their disposable income are receiving intra-household transfers from those who are consuming less than their disposable income. Disposable income is defined as labor income plus net public cash transfers (cash inflows less taxes) plus net inter-household transfers. If the disposable income of all household members combined exceeds the total consumption of all members combined, the surplus is transferred to the household head and saved. If total disposable income is less than total consumption, however, the household head supports the excess consumption using property income or by dis-saving.

We assume that intra-household transfers to support current consumption (non-durable consumption) are "financed" by imposing a household specific flat-rate tax on each member's surplus income. Within the household each member is taxed at the same rate. The tax rate does not vary by age. Moreover, we assume that the household specific tax rate for any sector is identical for each household member.

The consumption of durables, including the services from owner-occupied housing, are treated in a distinct fashion because, by assumption, the household head owns all household assets and all income generated by those assets flows to the head. The consumption of durables by any non-head household member is "financed" by an intra-household transfer from the head to the member equal to the value of member's durable consumption.

Once the current and durable consumption transfers are calculated at the individual level, profiles are constructed by aggregating by age group.

The following explanation of the methodology proceeds in five steps: A. Preparation of Data, B. Transfers for Current Consumption, C. Transfers of Unconsumed Surplus to Head, D. Transfers for Durable Consumption, and E. Construction of Age Profiles. A section at the very bottom discusses aspects of finalizing the profiles, such as smoothing and adjustment to control totals.

A. Preparation of Data

To calculate the intra-household variables, begin with unsmoothed variables adjusted to control totals. The necessary NTA variables, using indexes for person i in household j with consumption divided into sectors x, are as follows:

$Yl(i,j)\text{: labor income }$
$TGCash(i,j)\text{: public cash transfer inflows }$
$TGTax(i,j)\text{: taxes paid }$
$TFB(i,j)\text{: inter-household transfers }$
$CC(i,j,x)\text{: sector-specific current consumption}$
$CD(i,j,x)\text{: sector-specific durable consumption}$

In addition, household id and household head identifier variables are necessary.

There are three data cleaning considerations that come up in the intra-household transfer methodology. First, each household must have only one person designated as the household head (referred to here as i=1). Second, consumption values must be non-negative. Negative consumption values can arise when model-based estimation is used. Finally, although the best case is when all of the input variables listed above come from one survey, this is not always possible. If you need different surveys for different variables, additional steps are necessary to calculate intra-household transfers. See the section on "FINALIZING DATA" at the end of the Private Transfers methodology section for instructions.

In the notation used here, any variable written without all i, j, and x indices refers to a summation over the missing index. For example, total current consumption for individual i in household j is written as follows:

$CC(i,j)=\sum_x CC(i,j,x)$

B. Transfers for Current Consumption

B.1. Compute the current surplus or deficit for each household member and for the household.

A household member has a current surplus if his disposable income is higher than his current private consumption, where current private consumption excludes the services from housing and other consumer durables. A household member has a current deficit if his disposable income is less than his total current private consumption:

$X(i,j)= Yl(i,j)+TGCash(i,j)-TGTax(i,j)+TFB(i,j)-CC(i,j)$
$Surplus(i,j)= max[0,X(i,j)]$
$Deficit(i,j)= -min[0,X(i,j)]$

The total surplus and deficit for the household are calculated as:

$Surplus(j)=\sum_i Surplus(i,j)$
$Deficit(j)=\sum_i Deficit(i,j)$

B.2. Calculate the tax rate.

By assumption the tax rate on each individual's surplus varies across households but is independent of the age of the individual within the household. Thus, the rate at which the surplus is taxed in household j is equal to:

$tax(j)=min[1,Deficit(j)/Surplus(j)]$

If $Surplus(j)=0$ , then $tax(j)=0$ .

B.3. Calculate intra-household outflows for current consumption.

The current intra-household transfer outflow ( $TFWO_{c}(i,j)$ for person i in household j) for non-heads is equal to the tax rate times the surplus. For heads the outflow is the tax rate times the surplus, and any household shortfall that the head must finance through asset sales or dis-saving, but not including the head's own deficit, if any.

$Shortfall(j)=max[0,Deficit(j)-Surplus(j)]$
$TFWO_{c}(i,j)=-tax(j)Surplus(i,j) \text{ for } i\neq 1$
$TFWO_{c}(i,j)=min[0,-tax(j)Surplus(i,j)-Shortfall(j)+Deficit(i,j)] \text{ for } i=1$

B.4. Calculate the intra-household inflows and outflows for current consumption by sector.

Current transfer inflows to non-head i in each current consumption sector are proportional to that individual's current consumption in that sector. For heads, the calculation is similar except that the head may have to finance his or her own deficit through dis-saving or asset sales, which would not be recorded as a transfer. Thus for current consumption sector x:

$TFWxxxI_{c}(i,j,x)=\frac{CC(i,j,x)} {CC(i,j)} Deficit(i,j) \text{ for } i\neq 1$
$TFWxxxI_{c}(i,j,x)=\frac{CC(i,j,x)} {CC(i,j)} max[0,Deficit(i,j)-Shortfall(j)] \text{ for } i=1$

If $CC(i,j)=0$ , then $TFWxxxI_{c}(i,j,x)=0$ . If the only current consumption sectors being calculated are Education, Health and Other, then the "xxx" above will stand for E, H, or X, respectively.

Current transfer outflows from individual i in each current consumption sector are proportional to the total household inflows to each sector:

$TFWxxxO_{c}(i,j,x)= \frac {TFWxxxI_{c}(j,x)} {TFWI_{c}(j)} TFWO_{c}(i,j)$

If $TFWI_{c}(j)=0$ , then $TFWxxxO_{c}(i,j,x)=0$ .

C. Transfers of Remaining Surplus to Head

Any surplus held by non-heads that is not taxed for current consumption transfers is transfered to the head to be saved:

$TFWSO(i,j) = -Surplus(i,j) - TFWO_{c}(i,j) \text{ for } i\neq 1$
$TFWSO(i,j) = 0 \text{ for } i=1$

The head receives all of this excess surplus as an inflow:

$TFWSI(i,j) = 0 \text{ for } i\neq 1$
$TFWSI(i,j) = -\sum_i TFWSO(i,j) \text{ for } i=1$

D. Transfers for Durable Consumption

All transfers for durable consumption (i.e. asset consumption) come from the head and flow to non-heads. The inflows to non-heads by durable sector are equal to the consumption of each non-head in that sector, and non-heads have no outflows in that sector. The sector outflow for the head is equal to the total non-head sector consumption, and there is no inflow to the head for durable consumption:

$TFWxxxO_d(i,j,x)=CD(i,j,x)-CD(j,x) \text{ for }i=1$
$TFWxxxO_d(i,j,x)=0 \text{ for }i \neq 1$
$TFWxxxI_d(i,j,x)=0 \text{ for }i=1$
$TFWxxxI_d(i,j,x)=CD(i,j,x) \text{ for }i \neq 1$

If the only durable consumption sectors being calculated are Housing and Other, then the "xxx" above will stand for R or D, respectively.

E. Construction of Age Profiles

Once all current and durable consumption sector-specific inflow and outflow variables are constructed, total intra-household inflows and outflows are the sum of these sector-specific inflows and outflows, and the transfers of surplus to the head for saving:

$TFWO(i,j)=\sum_{xxx} TFWxxxO_c(i,j) + \sum_{xxx} TFWxxxO_d(i,j) + TFWSO(i,j)$
$TFWI(i,j)=\sum_{xxx} TFWxxxI_c(i,j) + \sum_{xxx} TFWxxxI_d(i,j) + TFWSI(i,j)$

Once the transfer variables have been constructed, a check on the calculations is that total inflows must equal total outflows for each sector, by household and in aggregate. After all checks have been performed, aggregate by age to construct the age profiles.

The file below shows a spreadsheet example of the transfer methodology described above for one household with current consumption sectors Education, Health and Other and durable consumption sectors Housing and Other Durable:

IntraHHExample

The next file contains STATA code to create the transfer variables and perform checks on unweighted micro-data.

IntraHHCode

FINALIZING ESTIMATES

This section discusses several specific issues in finalizing estimates:

Smoothing
Adjusting to control totals
Calculating intra-household profiles from several surveys (the "two survey problem").

Smoothing should be accomplished in the same way that you smooth all other profile. (The preferred NTA method is to use a local regression, or "lowess," smoother. See Smoothing.) Note however that transfers for education spending should be smoothed very little if at all, just as we use minimal or no smoothing on the education consumption profile to deal with the very choppy age-shape of education consumption.

For a discussion of control total adjustment (for inter- and intra-household transfers) and the two survey problem, see this document (UPDATED BY GRETCHEN ON 10/30/2009):

Private_Transfer_Notes

Inter-hh_Transfers_Mexico_pdf

Inter-hh_Transfers_Mexico_xls

Comments about the Private Transfers methodology:

Comments:

amason: (Thu Jun 22 01:45:13 2006 )

This is very clearly written and easy to follow. I have two suggestions. First, let's add a section at the beginning - preparation of data. This would highlight that we must use unsmoothed estimates adjusted to the control totals. It would also list the data needed so that researchers could tell quickly whether they had the data required to complete the work. Second, let's add a section at the end about smoothing. Of course, there is a problem in that I don't think we actually know yet how the age profiles should be smoothed. Also, I think we want to say something about not smoothing the education transfer inflows. (I assume that the consumption of education is not being smoothed.)

gstockma: (Sat Jun 24 00:14:13 2006 )

I have followed Andy's first suggestion, but will wait on the smoothing discussion because I'm still not familiar with all of the issues.

gstockma: (Fri Apr 20 17:05:30 2007 )

I've updated some of the methodology here and added a document that deals with some fine-tuning issues. Smoothing is discussed as well as per Andy's suggestions from June 2006.

For an earlier version of intra-household private transfer methods see Method Archives.