|
INTERVIEWS
WITH ECONOMISTS
PETER
HARPER
Peter Harper
is Head of the National Accounts Branch of the Australian Bureau
of Statistics. He has an economics degree from the Australian
National University, and over 15 years experience in the compilation
of economic statistics.
What are
the major problems in constructing the national accounts?
There is no
single source of information for constructing the national accounts.
Therefore, construction involves assembling information from a
wide variety of sources within the national accounting framework.
Not all of these sources provide information on an appropriate
basis, and adjustments are often necessary. Furthermore, key information
is not always available with either the preferred frequency or
timeliness, so indicator information is used to extrapolate and
interpolate basic source data. For some parts of the accounts,
more than one source of information is available. Sometimes these
sources provide conflicting information, so these conflicts have
to be resolved. When source information is derived from sample
surveys, the information is subject to sample error, which impacts
on the quality of the information. Non-sample errors (e.g. reporting
errors) may be present in any data source. The national accountant
needs to be aware of these problems and to make allowance for
them in compiling the national accounts.
While the
conceptual framework that underpins the national accounts is generally
well established, economic behaviour is constantly evolving. The
national accountant must be aware of changes in the economy and
ensure that they are reflected properly in the statistics. It
is also important that the national accounts are compiled in a
way that makes them comparable over time and across countries.
To assist the latter there exists an international standard for
compiling national accounts statistics called the System of National
Accounts (SNA).
How useful
is GDP as a measure of national wellbeing? Are there alternative
measures available?
GDP is an
unduplicated measure of the value of production of an economy
in a particular period. It includes all production that is sold
in the market, as well as goods and services that are provided
free of charge or at prices that are economically insignificant.
Also included are goods produced by households for their own consumption
and housing services from owner-occupied dwellings. However, other
services (e.g. cooking and cleaning) that are produced by households
for their own consumption are not included, and this exclusion
has attracted criticism from some quarters. The measurement of
GDP excludes income attributable to residents from productive
activity undertaken in other countries and includes income from
domestic production that is attributable to non-residents. The
depreciation of fixed capital is also not taken into account.
However, information on income transactions with non-residents
and depreciation is available elsewhere in the national accounts.
GDP does not
take into account the depletion of non-produced assets (such as
native forests and sub-soil assets) nor does it reflect the impact
of environmental degradation. On the social side, GDP provides
no information about the impacts on national well-being of changes
in income distribution, longevity, crime rates, divorce rates,
etc. Information about many of these environmental and social
issues is available from statistics published by national statistical
agencies and other sources.
Some commentators
have attempted to adjust GDP for environmental and social effects
to arrive at a single measure of wellbeing. However, in the absence
of a generally agreed framework for measuring wellbeing, the nature
of the adjustments is typically subjective and assigning appropriate
values to the adjustments is often problematic. Furthermore, having
arrived at a measure of wellbeing, it is often difficult to interpret
changes in the measure over time.
Economists
have suggested using purchasing power parity (PPP) conversion
factors for the purposes of comparing the GDPs of various countries.
What are some of the issues associated with the compilation of
PPPs?
PPPs are a
form of price index and, as in the construction of other types
of price indexes, decisions need to be made on the most appropriate
index number formula for weighting the volume information underlying
their construction. For PPPs, it is generally accepted that the
Fisher formula, which involves using a geometric mean of volume
relativities in the two countries being compared, is the most
appropriate. However, Fisher indexes are not transitive. That
is, using Fisher indexes that compare country A with country B
and country B with country C to compare prices in country A with
those in country C generally provides a different result from
comparing countries A and C directly using a Fisher index. To
overcome this problem, an approach is used where, for a group
of countries with similar characteristics, Fisher indexes are
compiled for each possible pair of countries. These indexes are
then adjusted to achieve transitivity. Least-square techniques
are used to minimise the damage to the original bilateral Fisher
indexes. Comparisons between two countries in different groups
are achieved through ‘link’ countries that are present
in both groups.
The compilation
of high quality PPPs requires the provision of good quality price
and national accounts information, the latter being used to provide
the price weights, from all countries involved in the exercise.
If a large number of countries are involved, ‘quality assuring’
the information provided can be a difficult exercise. Furthermore,
the regimen of goods and services being priced must be sufficiently
common across countries to provide meaningful comparisons. To
overcome this type of problem, pricing is often based on goods
or services with similar characteristics, rather than specific
types of goods or services. However, this often leads to practical
problems in price collection.
Services provided
free of charge or at economically insignificant prices, which
include many services provided by governments and non-profit institutions
serving households, present particular problems. As no meaningful
output prices can be observed for these services, comparisons
are typically based on input costs. Comparing these costs, especially
wage costs, at ‘constant quality’ across countries is
difficult and some commentators have advocated that these non-market
services should be excluded from the compilation of PPPs.
GDP can
be measured using three approaches, the production, income and
expenditure approaches. Normally, the measures produce different
results. Why is this so, and what can be done about it?
Conceptually,
each of the three approaches should produce the same estimate
of GDP. In practice, however, when different data sources and
methods are used for each approach, there will be differences
in results. The national accountant is generally required to use
different data sources and methods because, as described above,
there is no single source of data for national accounts statistics.
However, the
size of the differences can be reduced if the data sources used
produce information that is fully consistent with national accounts
requirements. Often the national accountant can influence the
nature of the data sources to try to achieve this. Nonetheless,
given that there will be errors in any data source (including
sample error in sample survey data sources), ensuring that the
source data are consistent with national accounts requirements
will not fully eliminate differences. Using common data sources
for each of the approaches can also reduce differences, but there
are limits to the extent to which this can be done.
Another way
around the problem is to confront the differences in results and
make decisions about where changes need to be made to eliminate
the differences. The best method for doing this is through a supply
and use table, which is a form of input-output table. A supply
and use table shows the production and distribution of all goods
and services within an economy. Through a supply and use table,
the compiler can systematically confront imbalances in supply
and use and in the value of production and incomes earned from
production. By eliminating these imbalances, which is typically
an iterative process involving judgements on the likely source
of imbalance, the compiler derives a ‘balanced’ table.
From this balanced table, the same estimate of GDP is derived
regardless of whether the production, income or expenditure approach
is used.
|