iassist quarterly - 3

Public Data
In Use:

A Case Study

of Ireland

by John Blackwell'

Resource and Environmental Policy centre.

University College, Dublin

Introduction

This paper is concerned with the policy issues

which arise in enabling the public data which

are most useful for research purposes to be

disseminated in a timely fashion. Case study

material from Ireland is drawn on, outlining

problems which have arisen in identifying and

in meeting the needs of users, and the extent to

which it is possible to meet the requirements at

a lime when there is no prospect of an increase

in real Government expenditure on the

provision of statistics. Particular examples in

the areas of earnings, employment and social

'Presented at lASSIST/IFDO International

Conference May 1985, Amsterdam

welfare statistics will be given. The extent to

which administrative records can be a fruitful

source of data, and can substitute for

purpose-built surveys and censuses, especially in

view of the cost constraints, is assessed. At the

end of the paper, some policy issues are posed.

The background - data provision in Ireland

In Ireland there is a centralized Central

Statistics OfTice (CSO), whose output ranges

from population and vital statistics to labour

force, employment and unemployment statistics,

industrial production, prices, earnings and hours

worked in industry, trade, transpon and

distribution. Many of these statistics are

produced by means of censuses and surveys.

The methods of dissemination used by the CSO
vary from published volumes to regular

mimeographed series (issued usually monthly or

quarteriy); data on Ireland which appear only in

publications of international organizations such

as Eurosiat and OECD; small area data such as

from the Census of Population and from the

Census of Agriculture which are disseminated

directly to the users; and special analyses which

the CSO does in answer to specific requests

e.g., from the labour Force Survey and the

Household Budget Survey, which are usually not

charged for.

However, in recent years there has been an

increase in the amount of data available from

public sector bodies outside the CSO, which in

the main complement the existing range of data

from the CSO. The monetary and public

expenditure data, produced by the Central Bank

and the Department of Finance respectively, can

be put with the national accounts as the basic

raw material for macroeconomic analysis. Much
of the information on education and manpower
fiows and science and technology, on activities

such as transport, energy, and tourism, on

Fdl/Winter 1985



4 - iassist quarterly

education, health, housing: and social services,

and on taxation arc produced by a variety of

bodies in the public sector outside the CSO. in

a number of cases, these data are produced by

government departments (e.g. Education and

Social Welfare); in other cases the data comes

from government agencies or state bodies (e.g.,

in the case of science and technology statistics).

Problems of resource allocation

There has been no increase in the volume of

resources provided to CSO in recent years,

rather, a decline. While there is no coherent

account of the amounts of resources which are

put into provision of statistics outside CSO,
these are unlikely to have increased. The
budgetary problems in Ireland rule out any

prospects of an increase in resources for

statistics. Indeed, one of the elements which

worked to the advantage of data users in

Ireland in recent years was Ireland's entry into

the European Economic Community. This

meant that EEC regulations and directives on

statistics became binding. This provided, for

example, a much needed set of statistics on the

distribution of earnings tlirough the 1979

Structure of Earnings Survey. As a reflection of

EEC budgetary pressures, a proposed 1986

survey will not now take place, and there is no

domestic survey which can fill the gap.

Especially at a time of pressure on the

government budget some means of ranking the

statistical output by reference to the intensity of

demand, is required. This is very difficult to

achieve, for the following reasons:

There is a heterogeneous body of users,

ranging from government to research

institutes, individual researchers and

private sector firms. The implied value

of attributes such as timeliness, frequency,

accuracy of estimate, level of

disaggregation, will vary among these

groups of users. Many private firms, and

those in the financial sector, put a

premium on timeliness. Hence, as a

result of timeliness problems in the case

of the Irish Census of Industrial

Productions (problems which are currently

being righted, slowly), there has been

minimal use of industrial statistics by

business firms. There are other reasons

for this neglect of industrial statistics on

the part of the private sector. Firms will

typically want highly disaggregated data in

order to answer questions such as market

share or productivity change for a

particular product In a country such as

Ireland, where one or two firms may
dominate some sub-sectors, there often

there can be no possibility that CSO
would be able to provide the data for

reasons of confidentiality. Hence, the

popularity among business users of the

foreign uade statistics, which are quite

disaggregated and are more timely than

production data.

Related to the preceding point, it is

difficult to establish the nature of the

"trade-offs" which exist between

improvements along these different

dimensions of statistics. For example,

timeliness could be improved by

sacrificing some accuracy of estimate. For

some firms, such as those in the financial

sector, timeliness is crucial. For them, it

is often more important to have an early

estimate which is subject to later revision,

than to have a purer number which

arrives after a long interval. By contrast

in the case of social policy, the time-scale

Fall/Winter 1985



iassist quarterly - 5

is different Here, the analysis of large

data sets, gathered at longer intervals,

concerned more with details at household

or family level, is often needed. With

regard to frequency, there is a danger

that in a perverse way, an

over-conceniiation on frequency could

slow up decision-making, especially if

estimates are subsequently revised to a

marked extent In the case of national

accounts data - where there is already a

time-lag due in part to slowness in

provision of local authorities' data - there

has in Ireland been a good deal of data

revision long after the event

The price mechanism is not used to

ration demand so that those uses of

highest value (or which use up least

scarce resources) are effected. That is

partly because information has many of

the attributes of a "pure public good".

This is, information which is made
available to one person (through, for

example published sources, in the widest

sense) can be made available to the

community as a whole, and one person's

consumption of information does not

"take away" from the consumption of

anoiiier person. (Of course, this must be

qualified: information can be withheld

from certain consumers and can be

bought and sold on the market)

Moreover, a number of the statistics are

provided at close to zero prices, which

gives rise to difficulties in assessing the

intensity of preferences among actual and

potential users.

In summary, therefore, a key issue is how many
more, or how many fewer, resources should be

devoted, at the margin, to producing particular

data series. This is, in a sense, a problem of

providing a surrogate for the market system in

guiding resource allocation to the provision of

public statistics. Difficulties abound in assessing

the intensity of user preferences, partly because

of the various preferences of the different users.

Furthermore, it is difficult to get people to

reveal their preferences, a refiection of the

'public good' aspect of statistics. Users will be

tempted to overstate their preferences, in the

knowledge that if information is provided by

somebody else, they in turn can benefit In

other words, people can obtain information at

no cost to themselves, even though, if they were

confronted with the stark alternative of "pay up

or do without the information" they would be

prepared to pay something.

The "classic" perception of the way in which

information is gathered and organized has often

centred on the following sequence by which

decisions are made: identification of a problem;

collection of data to illuminate that problem;

identification of feasible solutions; choice of the

best solution from the alternatives. However, in

practice, this logical sequence is not always

followed. Information may be gathered only

after a proposal has been decided on, either to

monitor the implementation of policy, or to

co-ordinate activities across Departments, or to

serve as a back-up for particular cases which a

public body may wish to make.

Why does the classic model not apply? In pan
it reflects the fluctuating pressures for decisions,

often made under time pressure. There is

inevitably a time lag before information can be

gathered, by which time decisions may have

already been made. In pan it reflects the lack

of a clear set of signals from the users of data

to the producers. In Ireland, the CSO has

found it difficult to get users to articulate their

requirements an unambiguous way. While there

has been a generalized feeling of dissatisfaction

among users, there has been no formalized

method whereby the reactions of users can be

channelled to the producers. At mid-1984, for

the first Lime, a one-day statistical users'

seminar was held in order to help

communications between users and producers.

Even here, there was a disappointing turnout

Fall/Winter 1985



6 - iassisl quarterly

from private sector users. CSO does consult

with major users, but this consultation gives

little weight to non-institutional users.

It is almost trite to observe that the demand for

statistics is a derived demand. Yet it should be

noted that these derived demands are governed

by the ecomic structure and by the way in

which policy analysis and research occurs.

Thus, for example, most social science research

in Ireland occurs in large research institutes. In

the absence of a funding body for social science

research, and given a weak tradition in team

research in the social sciences, the countervailing

power of the lone researcher, not to speak of

ability to use large data sets, may be weak.

Indeed, for some uses such as macro-economic

models, the data supply/demand position is

more or less one of one monopoly producer

(CSO) and one or two large users such as the

Economic and Social Research Institute.

Another example is the use, in other countries

with more decentralized of government, of

sub-national statistics to govern the spatial

allocation of public resources: with the degree

of centralization in Ireland, this use is virtually

absent A final example: the more rapid the

pace of social change (such as women's

participation in the labour force), the greater

the likelihood that the full censuses will (again,

given current methods) lag behind what would

be needed to monitor change. This has

implications for the potential use of sample

surveys or else of simpler censuses.

There will tend to be a chronic excess of user

demands over the supply of statistics (given

existing resources and methods). The question

arises whether it would be possible to decide on

allocating resources to statistics by using formal

criteria such as cost-benefit analysis. The
answer is "no". While it is possible to estimate

costs, the diffuse nature of the uses of statistics

would make the estimation of benefits extremely

difficult Except in a few particular cases, the

monetary value of statistics will be almost

impossible to compute. Moreover, given the

links between information, policy and

"background" knowledge of the worid, the

potential returns from statistics may be greater

than the actual returns if there is potential to

improve the linkage between public

policy-making and the provision of statistics.

This can be illustrated by reference to the Irish

Census of Population. The cancellation of the

1976 Census had some direct effects on other

data provision - it led to increasingly inaccurate

estimates of population and of the labour force

and consequently of employment and of national

output At the same time, there seems to be

almost no accounting for the individual

decisions, including those in physical planning,

which have been affected. But the measure of

these effects would be difTerent with a difTerent

type of public administration, one that was more

responsive to changes in economic and social

information. In this sense there are interaction

effects which complicate any attempt to

ascertain a pure return to statistics.

It is difficult in fact to point to particular

policies in Ireland which were critically

infiuenced by the provision or absence of

certain data. There arc a number of reasons for

this. First the provision of data per se is best

seen as contributing to the understanding and

explanation of economic change and social

processes. Second, even if a starkly empirical

view were (mistakenly) taken - that the matter

begins and ends with gathering "the facts" -

the dissemination of data would work its effects

in a diffused way, with time lags. Third, there

are usually time pressures on policy-makers -

research findings and associated data must be

speedily available, and this is not always

possible. Fourth, it is difficult to predict data

needs ahead of Lime - the focus of policy

attention tends to change over time, while the

provision of statistics lags behind. Finally, there

are insufficient contacts between researchers and

policy-makers (including lack of effective

dissemination of findings) and there is certainly

too little interaction between the three groups of

Fall/Winter 1985



iassist quarterly - 7

policy-makers, researchers and producers of

data.

However, it is possible to get a feel for the

value of statistics by asking in any particular

case:

what are these statistics for?

who is using them?

In anwering these questions, due account should

be taken of the nature of the benefits, many of

which are indirect, rather than relating to

specific uses by those who decide on public

policy. And a number of benefits relate to

users outside Government

In particular, there is need to concentrate on

decisions at the margin - what additional

information should be provided, and what

should be dispensed with? This would enable

the identification of those data which are most

crucial to policy decisions. It would also enable

the recognition of those pieces of information

which are redundant, or where a lowering in

frequency or a degeneration in timeliness could

be tolerated, especially if this meant that

resources would be freed to devote to the

provision of information for which there would

be a high return. In Ireland it has, however,

been extremely difficult to identify "redundant"

data series. This may in part reflect a system

which is producing the "minimum critical mass"

of statistics. It may also reflect a "ratchet

effect" whereby, once a series is put in place,

any attempts to discard it are met with

opposition.

Scope for Co-ordination and Rationalizaton

While the collection of statistics by a wide

variety of public bodies outside the CSO has

filled a number of gaps in knowledge, it has

given rise to a number of problems.

No explicit co-ordination of this output

of statistics occurs, nor is there any

indication of the amount of resources

which go into this activity. There is no

"statistics budget" for the public sector as

a whole, with an associated set of outputs.

CSO itself has no "leverage" over the

gathering of statistics by other public

sector bodies. The activities of CSO
could be greatly influenced by the data

gathering of other bodies, either because

they complemented the work of the CSO,
or because the CSO could use the data

for purposes such as national accounts

compilation, or because data could be

used either as a means of verification, a

sampling frame, or a register for coding

purposes.

There is no quality control with regard to

non-CSO statistical outpuL This has in

part reflected the way in which data

collection began - sometimes as a

by-product of filling internal plaiming

needs and the need for co-ordination,

sometimes as a result of implementing

administrative schemes. Unpublished data

are available from an employment survey

conducted annual by the Industrial

Development Authority, from a quarterly

consumer survey undertaken by the

Agricultural Institute for the EEC, and in

an information system on public sector

employment in the Department of the

Public Service. The first and last cases

exemplify data sets which have been

compiled primarily for internal

Fall/Winter 1985



iassisl quarterly

managemenl purposes.

There are no principles of access lo

non-CSO data. While there has been

some dissemination of the Industrial

Development Authority and Department

of the Public Service data, this has been

as a by-product of secondary analysis by

"outside" workers. The Industrial

Development Authority data have been

used as a source for information which,

in the absence of CSO industrial

employment data on a "component of

change" (job gains, job losses) basis,

cannot be obtained from other sources.

They could, in part, be substituted by

CSO data if the basic CSO data were

analysed in a form which facilitated the

"component of change" work.

Government departments hold a great

deal of information which is of interest to

policy analysts, some of it in the form of

raw data. A certain furtiveness in even

disclosing the presence of data seems to

be endemic. The ultimate banier is when
a Department or agency does not release

a body of data. For example, twice since

1973, the results of housing conditions

surveys, undertaken on behalf of the

Department of the Environment have

been kept within the Department This is

a notable case, because the information

would complement the census data on

housing and, in the areas of unfitness and

obsolescence, provide information for

policy which is more relevant than could

ever be obtained from the census.

While there are many ad hoc surveys

conducted by Departments and other

bodies, surveys conducted by the

Economic and Social Research Institute

for the Institute work, and surveys

conducted by other research institutes and

by individual researchers, none of this

body of material is pulled together in one

place. Nor is it even catalogued. There

is no central user's guide to the body of

data.

There is danger that the current round of

computerization which is occurring in

public sector bodies will result in further

fragmentation. There is, for example, a

danger that computerization among local

authorities will miss out on the possibility

of providing the means of a data bank.

To take another example: the Department

of the Public Service is to expand its

personnel information system. Ideally, in

this process, the CSO should be consulted

and compatible codes for occupations and

other entities should be used.

This leads to the issue of whether there would

be potential for rationalizing the data-gathering

on Irish by State-sponsored bodies. A number
of these bodie^ - national training agency, the

Irish Export Board, the Industrial Development

Authority, the Institute for Industrial Research

and Standards, and the National Board for

Science and Technology - compile data on Irish

firms, as does the CSO itself. On the surface,

rationalization could contribute to the problems

of non-response and of slowness in response,

which in part account for the lime lags in the

issuing of CSO industrial statistics. This may be

the case up to a point, but it has to be

qualified in view of the client relationship

which exists between many of these firms and

the Stale body in quesuon, which is likely to

facilitate response. A more compelling reason

for seeking some rationalization is the richer

potential for cross-tabulation which would

ensue, as at the moment this is limited to the

variables on each individual inquiry. Moreover,

there would arise a common set of standards

with regard to methods of compilation and of

analysis, which would ideally be in accord with

CSO categories as regards coding. There would,

of course, remain a sub-set of data which

would be confidential to individual agencies.

Fall/Winter J 985



iassisl quarterly - 9

Some co-ordinating role is needed in relation to

the output of statistics Trom sources other than

the Central Statistics Office. There are a

number of ways in which the desired

co-ordination of non-CSO sources could occur.

One would be through expanding the

responsibilities of CSO to include the

co-ordination of all public statistics. Another

would be a co-ordinator of non-CSO. The
co-ordinator could be in a Government
department, or outside the departmental system,

or could work through an inter-departmental

committee.

The main objectives of the co-ordinating

function would be as follows:

this could occur: through an "umbrella"

survey, or through the linking of records.

The latter would be much more difficult

to achieve. Some move towards an

"umbrella" survey would be the most

fruitful route.

To consider the use of these statistics as a

means of verfication of sampling frames

or registers for coding purposes.

To take part in a revised and explicit

system of resource allocation for statistics,

infiuencing the allocation of resources and

pointing up the areas where attention

needs to be focused.

To encourage public sector bodies to

consider the statistical possibilities of

administrative records and to consult on

this with bodies such as the CSO. This

is especially important at a time when
new computer facilities are being installed.

To ensure that statistics which are of

importance to policy-makers but which

are of little or no importance to the

collecting agency, are produced. This may
at times require that such a subordinate

objective be given explicitly to the agency.

To ensure quality control, i.e., some
minimum standards and compatibility

between individual data sets. The more
standard classifications and definitions are

used, the lesser the likely burden on

respondents as their records would be in

conformity with all requirements.

To examine the potential for

rationalization in the data-gathering

activities of State bodies. This could lead

to richer possibilities of data analysis, as

it might be possible across the headings

which are contained in the existing

surveys. There are two ways by which

To fulfill some gatekeeping functions with

regard to the increasing volume of

surveys of Irish firms by those in the

public sector.

Among the cases which are candidates for

action are manpower and energy. Another

instance is the road vehicle file of the

Department of the Environment which is

computerized but which is not being used to

produce data on the vehicle fieet by age, make,

type of fuel used. Perhaps the prime candidates

are the Department of the Public Service data

on public sector employment (mentioned above)

and Revenue Commissioner's data derived as a

by-product of tax collection. This leads to the

next section, on the potential for using

administrative records.

Using Administrative Records

In part because of the budgetary problems

which face us, the use of administrative records

as sources of statistics is an attractive option.

Limited resources can be devoted to the

Fall/Winter 1985



10 - iassist quarterly

gathering of stalisiics by surveys; siatislics based

on administiaiive records can be collected at

relatively low cost, in contrast with surveys. In

addition, there has been increasing resistance on

the part of respondents to provide information

to the statistical office. (Indeed, this was the

main reason the pre-1968 survey on the

distribution of industrial earnings by the CSO
was dropped). In addition, income data derived

from administrative records have potential for

being more accurate than income data derived

from surveys, which are subject both to

sampling enor and to understatement of income.

This potential is particulariy evident in the case

of the service sector where relatively little data

are available on earnings and employment -

even for the public sector, where the only

readily available information is on civil service

employment

The potential for mak-ing greater use of

administrative records as a data source, or as a

basis for sampling, will increase as existing

records become more computerized, thereby

enhancing the potential for cross-tabulations. A
number of departments and agencies are now
apprised of the potential for decentralized

collection and dissemination of data via

micro-computers. The idealized picture, thus, is

one where the contraints on resources which

cunently bind us are loosened through a

combination of the use of administrative records

allied with computer collection and analysis.

There are a number of potential snags which

could disturb this pleasant sequence of events.

First, the definitions and categories used in

administrative records refiect the administrative

process (e.g., entitlement of benefits, being in

the tax net). They may not be ideal for

statistical analysis, and many change over time.

Ideally, statisticians would be involved at an

early stage, either when new systems are being

put in place or when manual records are

computerized. One feels apprehensive that the

current wave of computerization could lead to a

rash of incompatible variables and codes.

Experience to date with some of non-CSO
sources-with their lack of compatibility with

CSO data and changes over time in categories

which impede time series analysis - does not

inspire confidence that this can be avoided

without some policy initiatives. The Department

of Social Welfare data, which have a good deal

of potential for providing data on social welfare

recipients, are tied to benefit recipiency. Hence,

it is difficult to explore problems of lack of

take-up of benefit And in the case of

unemployment data, in Ireland (as in other EEC
countries) a substantial minority of the

unemployed (from the labour Force Survey)

receive neither unemployment benefit nor

unemployment assistance.

Second, much of policy analysis requires

linkages - either to CSO data or to

departmental data. An example would be the

relationship between social welfare payments

and patterns of work. Such linkages are either

quite difficult to effect or can raise problems of

confidentiality.

Third, the only aspects of administrative records

which are reliable are those which are regularly

used by the authority in question and are,

therefore, kept accurate and up to date.

Considerable clerical resources may be required

to maintain administrative records for statistical

analysis purposes (i.e., keeping designatory

details up to date, elimination of duplication,

etc.) and authorities have other priorities in

mind. If administrative records are to be

fruitful, they need constant attention. Important

administrative details required for analytic

purposes (e.g., designatory details such as

business descriptions, occupation, age) are not

usually complete nor up-to-date because they

are not frequently used by the authority. This

is the biggest handicap to the use of these

records, and arises in particular in the case of

the Irish Revenue Commissioners data which

have the potential to provide earnings and

employment data and to fill notable gaps in

coverage, especially on the service sector.

Fall/Winter 1985



iassist quarterly - 11

Fourth, there is need for close coordination

between siausticians and administrators in order

that the needs of both administration and

provision of statistics are meL Decisions have

to be made about classifying data in cases

where there may well be conflicts between the

desires of different users. At times when
schemes are designed, changed or computerized,

there is need to be sensitive to the possibilities

of using these records to provide statistical

tabulations. This would mean consulting with

statisticians in the design stage. If consultation

were to occur too late, a costly redesign of

systems could be required. In the past in the

case of the Irish health sector, where there is a

certain ceniralizaton of the functions of Health

Boards, computer systems which were

incompatible across Health Boards were

acquired. While there has since been a rethink

of information requirements within the

Department of Health, this has involved too

little contact with those with a planning function

in government

Fifth, a major constraint to using the Revenue
Commissioners' data relates to objectives. This

is linked both to the issue of resources and the

need to use clerical resources. Understandably,

the Revenue Commissioners regard their main
purpose as getting in money from the tax

system and would regard information provision

as, at best a subsidiary objective, or at worst

having no jusdficaiion at all. In the case of a

tax with high collection costs per yield, such as

the residential property tax, the minimum
amount of information would be provided.

From the point of view of the public weal, and
from the point of view of cost-minimization for

the State as a whole, there is need to use the

Revenue Commissioners data. But there is no
incentive for the Revenue Commissioners

themselves to take account of this. Rather the

opposite: if people are switched from

revenue-raising in order to provide more
information, the Revenue Commissioners will be

in danger of receiving criticism for failure to

maximize tax revenue.

If the basic data were "clean", it would take

litUe resources to produce the required data, by

using the automatic data processing facilities

which Revenue has. The biggest obstacle to

producing tables which would help

policy-making outside the specific remit of

Revenue is the need to deploy clerical workers

to "clean" data with regard to designatory

codes. This would especially be the case as

Revenue is prepared to consider the provision

of a subset of basic data on magnetic tape to

outside agencies - with the individual

identifying characteristics removed. Users could

then analyse these data using either a software

package or a custom-built program. This would

be a particularly valuable facility.

There will of course, always remain areas of

inquiry where there is no substitute for survey

or census data. An example arises in the case

of labour force statistics.

It is manifest that existing data are often

inadequate to answer the analytical and policy

questions which arise. This is partly because

most of the labour force data are given in

terms of stocks. Many of the questions raised

need an analysis of flows between states.

Another reason for difficulty is that existing

data come, in part from administrative records

which are tied to benefit criteria, which differ

over time, and in part from self-description of

states which are inherently ambiguous, such as

"unable to work due to disability". An instance

of this from the Irish Labour Force Survey is

that in 1977 a category "unable to work due to

permanent sickness or disability" was introduced,

but led some people who described themselves

as "unemployed" in 1975 to opt for this new
category. There is an interplay between

disability, unemployment and disability, and

between retirement and disability. If one is to

capture the impact of social security payments

on the labour market there is need to

complement the existing stock data with fiow

data.

Fall/Winter 1985



12 - iassist quarterly

Ideally, flow data (shown in Table 1) by sex

would be required. In this table, the

intersection of a row and a column gives the

flow between states. From this table, a table

showing transition rates (for, say, one or two

years) between labour force stales could be

constructed. This would show entry rates into a

particular state from all types of states, together

with exit rates, and continuation rates (the

proportions of people who remain in a

particular state between one period and

another).

In Ireland there is already one survey which in

part can be used to provide flow data: an

annual survey of school-leavers undertaken for

the Department of Labour. This is an example

of a survey which would provide the answer to

a number of policy questions if the basic data

were accessible.

Policy Issues

In summary some of the policy issues which

arise are:

There is need for some form of

coordination (with no coercion*^) of

non-CSO data with CSO data, in order to

ensure quality control, i.e.. some minimum
standards and compatibility between

individual data sets. There is also scope

for some gatekeeping functions with

regard to the increasing volume of

surveys, including surveys of Irish firms,

by those in the public sector. And there

could be some rationalization of the

existing surveys of Irish firms, leading to

richer possibilities for data analysis.

Should some minimum standards be set

in relation to data which are collected as

a by-product of government contracts,

including their deposit in a central

location?

What forms of surrogates for the market

system to identify user demands
("needs"?) can be developed - e.g.. some
mechanism for feedback and complaints,

for wide consultation and receipt of

suggestions at an early stage of the design

of inquiries? For example, could there be

greater use of charges in order to get an

indication of the implicit value which

different users put on different types of

statistics?

What forms of users' guide are required

for the Existing mass of data which public

bodies hold? And what ground rules on

access to public sector data by outside

users should be established?

There is need for public sector bodies to

consult with CSO on the statistical

possibilities of using administrative

records. Related to this, how can the

danger that the increased diffusion of

computers among public bodies, including

the local authorities, will lead to the onset

of incompatible data sets, which are in

turn incompatible with CSO data, be

avoided?

Despite the above remarks on the

potential for using administrative records,

there are many issues of economic and

social policy where there is no substitute

for survey or census statistics at the level

of the individual or of the household,

especially where it comes to establishing

relationships between variables. This is

especially true in view of the increasing

Fall/Winter J985



iassist quarterly 13

incidence of "non-standard" households

such as lone-parent families and the

many family policy issues which result

To what extent could existing surveys be

used with additional questions being asked

at marginal cost? In Ireland, there are

many possible uses of a general

household survey with a central base of

questions and a varying sub-set of

questions from year to year. This could

provide information on education and on

use of health services, for example.

There is no possibility of adding to the

Household Budget Survey (i.e., family -

expenditure type), which is already

overloaded.

To what extent should a simpler Census

of Population be conducted, with a

shorter time lag for dissemination, leaving

matters such as housing to sample

surveys?

To what extent should one try to force

the development of intermediaries who

would take primary industrial data and

engage in "packaging" and marketing to

industrial customers; are there economies

of scale in this activity?

Underlying the paper as a whole, there

are the difficulties in deciding how much
public resources to put into sialislics. For

example, non-CSO data will reflect

producer preferences as to internal

management needs or inter-departmental

and inter-agency coordination: this may
not be optimal from the point of view of

society as a whole, n

Schcmalit outline ol lypc of labour markcl How dala whicli arc required

Initial State Final State

(numbered as in rows)

li) lii) 2 3a) 4 5 6 7

1. Employed

a. full time

b. pan time

2. Unemployed

3. In partial retirement

a. In the same job as when
last fully employed

b. In a different job

4. Fully retired

5. In labour force, sick

6. Unable to work due to

disability

7. Not in labour force due to

"home duues"

1

Fall/Winter 1985