Third Meeting of the Working Group on Macroeconomic Aspects of Intergenerational Transfers,
East-West Center and the Center for the Economics and Demography of Aging, University of California-Berkeley,
Honolulu, Hawaii, January 20-22, 2006.
Summary of Issues Raised
Private Consumption Allocation Methods
Education: Methodology appears to work well, as evidenced by a comparison between estimated and actual individual expenditure for Brazil (Figure 5 in ws3.8 presentation). Specific ages of enrolled and not-enrolled to include in the regression depends on the particular data set. It is recommended that the upper open age group is used when observations are falling off and mostly noise.
Health: Several countries, including Brazil, have reported receiving negative health expenditures for children when allocating household health using the regression method. Furthermore, results obtained through the regression method appear to be inconsistent with those obtained through individual self-reported expenditures in the case of Chile. This may be due to decreased per child health expenditures with increased numbers of children, large numbers of children being born into poorer families, or increased expenditures with a second parent in the household versus single-parent households.
- Include additional variables in the regression (e.g. gender, income, etc.). A problem that may arise with this would be the treatment of additional coefficients obtained.
- Use of a health indicator to obtain the "J" curve observed in the self-reported data.
- Constrain estimation to positive coefficients. A working paper by Ray Fair may provide insights, as it uses frontier methods to constrain coefficients.
Sang-Hyop to review papers by Mankiw and Weil, and McFadden, which uses similar regression methods, and to report on any suggestions that may have been made on how to deal with the problem of negative coefficients.
Cassio to explore different approaches and to suggest an improved methodology by which to allocate consumption, since Brazil's data set includes individual self-reported expenditures by which to compare estimated. He is also willing to post his data on the website so that others may work on this.
Other: Methodology has been decided upon using the a priori equivalence scales (Flow Account Methods). Countries may use other methods to create a second set of estimates, but will create a set using the a priori equivalence scales for the NTA database.
Public goods in private consumption may be treated separately. These include housing, heating, utilities, durables, etc. Allocation of these goods may be done on another basis. One possible approach is to treat these goods as public, and allocate the full amount to each member.
- Jorge and Cassio will consult with each other and work on this.
Elder support by family may introduce "waves" into the consumption profiles for several countries (most pronounced in Japan and Korea).
- Hiro to look into modeling the effects of co-residence. How is the decision to co-reside made? What to the elderly bring when they move in with their children (and vice versa)? Does this exceed what they consume?
The importance of accounting for time transfers was raised by both the Japan and France teams. In co-resident homes time is a convenient form of transfer, and time transfers may have effects on labor force participation, and the value of household production is significant, and ignoring such transfers will tend to lead to an underestimate of private transfers and income. Estimates done by both teams indicate that time transfers are significant, though data requirements for estimating time transfers are numerous.
- Given project limitations, Francois-Charles will lead the comparative and methodological work to incorporate time effects, without each country having to create a separate set of estimates. He will inform others who are interested in the methodology he has used (may also consider writing something for the methodology page), so that comparative work can be done.
The most recent method of estimating intrahousehold transfers was discussed, which allows sector-specific and saving allocations. However, the method requires a single survey with all necessary data. In some cases, such as the United States, data is obtained through multiple survey sources and this method is not applicable as currently estimated. Due to data limitations, an alternate methodology of estimatng intrahousehold transfers is being used by the United States, which does not yield sector-specific flows, but net intrahousehold flows.
- Comfort to work with Tim to test the method being used by the United States using aggregated Taiwan data to determine whether results obtained are consistent with those using the most recent method.
- Terminology for distinguishing between consumption components which are consumed privately, but publicly subsidized (e.g. Medicare) and the share paid out of pocket needs to be developed. This would allow consistent treatment of these in the calculation of public and private consumption.
- A concern raised by several teams was the availability of multiple micro data sources, none of which provide all data necessary to calculate the variables in the NTA accounts. When using several data sets, problems may arise with systematic differences between data sets, and may require adjustments in methodology. One possible approach is to use the data from one survey to adjust the data from another. Another proposed approach is to create a data set with synthetic people, and impute data from available data sets for a country to create a master data set with all necessary variables. Concerns were raised whether imputing data to create consistency between data sets will result in incorrect covariances, which may affect further calculations.
- Adjustments made to match estimates with aggregate control totals were of concern, as aggregate controls may not be consistent over time (e.g. in the case of Taiwan ws3.4 presentation), or may be quite large. It may be beneficial to keep track of the magnitude of adjustments, and their changes over time as large adjustments may suggest adjustments in methodology or data.
- Using reported values for estimated asset income, rather than the defining it as a residual leads to notable differences in estimates for Korea (ws3.9 presentation), and the use of reported values is proposed as an alternative. Use of a residual to measure household saving ("residual of a residual) was raised as another difficulty encountered in estimation.
- The smoothing method that is currently being used has displayed sensitivity to bandwidth and survey weights. The lowess command in Stata does not incorporate survey weights. Thus, when the application of weights significantly change the shape of age profiles, smoothed estimates are no longer representative. Pre-weighting data yielded data sets too large to be smoothed within a reasonable amount of time.
- Cohort analyses are limited by the number of years for which cross-sectional data is available. Assuming constant age profiles, and using population data and adjustment factors, indirect estimation methods can be applied to fill in the gaps and follow the cohort experience. Indirect methods can be applied to historical data to backcast previous years age profiles.
Other issues that were raised in brief include: data problems (changes in survey formats, coding errors, inconsistent variable definitions), treatment of incarcerated individuals, and "proper" discount rates.
Selected sections of the webpage have been opened to the public, including working papers. It is the responsibility of individual authors to remove public access, should they feel their papers are not ready for public viewing. Recent additions to the webpage include the Recent Developments and Issues pages. The Recent Developments page is intended for dissemination of information on recent changes to methodology pages, while the Issues page is to be used to raise concerns and issues related to NTA and open them to discussion.
Several suggestions were made for website improvement:
- Addition of a link to "Recent Changes" on the main menu once logged in.
- Creation of a topic-oriented menu, which would allow one to access information from all countries concering a specific topic.
- Possibly accomplish this through improvements to the existing Methods webpage, by including a table of contents with links to specific topics.
Questions that arise while estimating can be raised on the issues page, e-mailed, or links to a webpage describing the issues may be sent.
Future Plans for 2006
UN Population Division is interested in using NTA profilies as a basis for imputing profiles for all countries. The data to be provided include age profiles of labor income, public consumption, and private consumption. No objections to providing the UN Population Division with their country's data were made.
Deadline for Uploading Data
It is recommended that a target date be made by which countries upload specific data. In particular, the data requested by the UN Population Division. As individual countries are at various stages in the estimation process, no consensus was reached on when to set this deadline.
- Andy will send e-mails with a list of requested variables and a target date for uploading. Individual countries can then respond with their concerns.
The successful use of teleconferencing led to the suggestion of a monthly teleconference, where a specific issue could be addressed with each meeting. This would also allow countries to better synchronize their work. A PowerPoint presentation would be available for download, and each meeting would be approximately an hour. Concerns raised were the coordination of meeting times given the multiple time zones; and technological issues, such as phone capabilites.
- Individual countries are to consider the proposal and e-mail their thoughts and concerns to either Ron or Andy.
The first location option for the fourth workshop is the University of California, Berkeley. The option of holding the workshop in other locations, such as Paris, France is being considered.
|All meetings will take place John A. Burns Hall Room 2012|
|Rapporteur: Comfort Sumida - firstname.lastname@example.org|
Friday, January 20, 2006
8:15-9:00 Registration and Continental Breakfast
Welcome and Overview
9:00-9:15 Welcome - Ronald Lee and Andrew Mason
9:15-10:00 An Overview of Age Reallocations: Ronald Lee (ws3.1)
10:00-10:45 Coffee break
10:45-11:30 National Transfer Accounts: Andrew Mason (ws3.2)
Country reports (30 minute presentations and 15 minute discussion)
Chair: Jorge Bravo
11:30-12:15 US: Tim Miller and Avi Ebenstein (ws3.3 presentation)
2:00-2:45 Japan: Naohiro Ogawa and Rikiya Matsukura (ws3.5 paper ws3.5 presentation)
2:45-3:00 Break and group photo
3:00-3:45 Indonesia: Maliki, Suahasil Nazara, and Turro Wongkaren (ws3.6 presentation)
- Young Jun Chun, "Fiscal Burden of Korean Public Pension System: A Generational Accounting Apporoach" (ws3.7 presentation)
- Chong-Bum An and Eul Sik Gim, Country Report (ws3.7a) (ws3.7a presentation)
6:15 Pick-up at Lincoln Hall for Reception 6:30-9:00 Reception at Mason residence
Saturday, January 21, 2006
8:30-9:00 Continental Breakfast
Country reports (30 minute presentations and 15 minute discussion)
Chair: Naohiro Ogawa
9:00-9:45 Brazil: Cassio Turra and Bernardo Queiroz via telephone
9:45-10:30 France: Francois-Charles Wolff (ws3.10 presentation)
11:00-11:45 Constructing a base-line NTA for Chile: preliminary findings, Jorge Bravo
(ws3.9 paper ws3.9 presentation)
Recent Developments, Methodological Issues, Future Plans
Chair: Ron Lee
1:00-3:00 Recent Developments and Methodological Issues
- Recent Developments: Discussion of methodological developments since the last NTA workshop: Andy Mason (ws3.12)
- Consumption allocation
- Intra-household transfers
- Asset-based reallocations
- NT Flow account
- Unresolved methodological issues
3:15-4:30 Plans for 2006 and other administrative matters
- UNFPA Project
- Uruguay Project (Jorge Bravo)
- Website: Andy Mason(ws3.14)
- Suggestions for improvement
- Administrative issues
- Rules of the game
- NIA requirements
- Plans for the coming year
- Overall goals
- Country plans
- UN Pop Division
- PAA meeting
- 4th Workshop
Sunday, January 22, 2006 (optional)
8:30-9:00 Continental breakfast (for out-of-town participants) Morning Country group meetings Afternoon Country group meetings