Non-profit Institutions Satellite Account

Non-profit institutions satellite account (NPISA) measures and analyses the contribution of non-profit institutions to New Zealand’s economy and includes an estimate of the contribution volunteers make to their activities.

The development of the NPISA is part of a wider international comparative study of the non-profit sector that New Zealand is participating in. It is being led by the Centre for Civil Society Studies at Johns Hopkins University, Baltimore. The Johns Hopkins University project compares information on the non-profit sector around the world. Started in 1990, the project now extends to more than 40 countries and involves a common framework for analysis, including information-gathering strategies and a set of definitions for determining which institutions fit within the project’s scope. The work entails commissioning each country in the project to research the history of the non-profit sector, the legal and policy environments, and the impact of the sector on society.

Classification of non-profit institutions

The United Nations Handbook on Non-Profit Institutions in the System of National Accounts recommends the International Classification of Non-Profit Organizations (ICNPO), as a tool to differentiate between the various types of institutions that have been defined as non-profit institutions. This is primarily done based on their ‘economic activity’, as is the International Standard Industrial Classification (ISIC) on which ICNPO is based. Some purpose criteria have been included where activities are similar. By grouping together institutions in this way, a basis for meaningful data analysis is formed.

The main ICNPO groups are as follows:

• 01 Culture and recreation
• 02 Education
• 03 Health
• 04 Social services
• 05 Environment
• 06 Development and housing
• 07 Law, advocacy and politics
• 08 Philanthropic intermediaries and voluntarism promotion
• 09 International
• 10 Religion
• 11 Business and professional associations, unions
• 99 Not elsewhere classified

As the ICNPO is mainly an activity based classification, there are no specific categories for population groups such as women or people with disabilities. Many Mäori groups fit within the criteria of a non-profit institution and will be included in the NPISA. Institutions targeting these populations are categorised based on the predominant activity of the institution. For example, if an institution provides medical treatment of a disability then it is coded to the health group. If, instead, it provides social assistance for people with disabilities then it comes under the social services group.

Many non-profit institutions have multiple activities, each falling under separate ICNPO groups, but they can only be classified in one group. In these cases the institution’s ‘primary economic activity’ is used to assign an appropriate ICNPO category. It is usually measured as the activity with the largest share of:

1. value-added – a measure of the institution’s contribution to gross domestic product
2. gross output, if value-added is not available
3. employment, if neither value-added nor gross output is available.

Classifying non-profit institutions in the satellite account

For the Statistics NZ Business Frame population the initial classification was done using a concordance with the Australian and New Zealand Standard Industrial Classification (ANZSIC) 1996. For example, preschool education (ANZSIC N8410) is concorded with early childhood education (NZSCNPO 2 110), hospitals (ANZSIC O8611) with hospitals and rehabilitation (NZSCNPO 3 100) and residential property owners (ANZSIC L7711) with housing (NZSCNPO 6 200). It should also be noted that where a non-profit institution carries out two or more distinct activities, its ANZSIC code is that of the majority activity.

For unallocated institutions from this first analysis, extensive keyword search lists were applied. Although several hundred keywords (for example ‘tennis’ and ‘church’) have been used, the list is still not exhaustive.

Finally, manual classifications were made for the largest institutions (in terms of sales of goods and services) that remained, and for those units where their activity was known. After all these steps were completed, the number of non-profit institutions from the Business Frame remaining in the not elsewhere classified group is below 3 percent.

For those institutions added from the Inland Revenue administrative database and the Companies Office registers the overall method was similar. The ANZSIC-NZSCNPO concordance was applied if an Inland Revenue ANZSIC tag was available. For the remainder the keyword analysis was applied.

A further allocation was then based on the results of the manual classification of a 10 percent sample of not elsewhere classified non-profit institutions, using the name of the institution. A further two-thirds of institutions in this group could be coded. The institutions that remain uncoded are those where either their activity is truly different to those included under any of the other 11 main groups or where insufficient information about them is available to allow coding.

Keyword coding was applied where no industrial code was available, that is for the whole of the non- Business Frame population and some of the Business Frame population. Where multiple keywords are in the name, a precedence list was applied. For example, ‘The Church of XYZ Tennis Club’ would be classified to sports institutions because in the use of keywords ‘tennis’ is given a higher classification ranking than ‘church’.

It is difficult to identify the sometimes very specific non-profit institutions in some of the NZSCNPO subgroups, for example the income support and maintenance subgroup or the employment and training subgroup. An undercount in those subgroups is likely, because some institutions may have been coded to the overarching groups of social services or development and housing instead, based on their industry code or keywords.

Therefore the results of the subgroup counts need to be interpreted with caution. In contrast, the results at the main group level are far more robust. A further point to note is precisely what institutions and activities are included under each subgroup. For example, the fundraising subgroup includes large, nationally active institutions with fundraising as their main activity, whereas institutions fundraising to support a specific activity covered under one of the other main groups will be coded to ’support and ancillary services’ under this respective group. A second example is that rest homes and other aged residential care (except nursing homes that provide first and foremost medical services) are under social services, not under the health subgroup of nursing homes.

Market and non-market non-profit institutions

A ‘market’ producer is an institution that sells its goods or services at competitive market (or ‘economically significant’) prices. Most non-profit institutions are classified as ‘non-market’ because they provide their services for free or below market prices. A significant number of non-profit institutions are market producers however, and are included in the non-profit institution population as long as they meet the structural operational definition.

Examples of non-profit institutions that are classified as market producers include:

• most business associations
• gaming trusts
• most non-profit institution hospitals
• racing clubs.

These units are distinctive within the non-profit institution population. To generate a positive operating surplus, these institutions need to sell their output at market prices. At the same time they meet the legal requirements of being not for profit because the surplus is generally not retained within the institution, but is distributed to another non-profit institution or charitable purpose.

Business associations are classified as market producers where they are financed by dues and subscriptions rather than by government. The rationale is that business associations are non-profit institutions that support market producers.

Gaming trusts need to produce large operating surpluses in order to make community grants. Because of this, they are implicitly assumed to be charging economically significant prices, even though the market in which they operate is heavily regulated. Similar comments apply to racing clubs, although in their case any operating surplus is provided back to industry participants (through such measures as increased stakes).

Most hospitals are classified as market producers because they charge what are considered to be market, or close to market, prices. They would be classed as non-market where they are clearly charging below market rates, for example, free hospitals where staff are volunteers.

Many non-profit institutions rely on government funding that is contested. Non-profit institutions receiving funding from government, on the basis of winning contested contracts with government, are not necessarily market producers. If they have a significant volunteer labour component, they may be able to charge out their services at below market rates.

From these examples it should be apparent that, in practice, whether or not a non-profit institution should be classified as market is not always immediately obvious. If it is unclear whether economically significant prices are being charged by the unit, but it is making a positive operating surplus over time, then that is usually taken as implying that the unit is operating on a market basis. By applying these principles it is possible to come up with an acceptable set of market and non-market classifications for the non-profit institution population.

A market and non-market split also allows for better comparisons with non-profit institution satellite accounts of other countries. Countries such as Canada have a ‘core’ set of non-profit institution satellite accounts and a ‘non-core’ set of accounts. The non-core set includes hospitals, universities and colleges. The noncore estimates were provided separately because hospitals were very significant within the Canadian non-profit institution account. The core Canadian accounts appear to be reasonably comparable with the New Zealand non-market estimates. In the Australian non-profit institution estimates, hospitality clubs and business and professional associations are treated as market producers.

Data sources and methods

Counting the number of non-profit institutions

Business Frame population

The primary source of information for counting the number of non-profit institutions is Statistics New Zealand’s Business Frame, which identifies enterprises. For an enterprise to be on the Business Frame it must meet any of certain criteria, the most relevant of which, for the count of non-profit institutions, are:

• annual goods and services tax (GST) expenses or sales of more than $30,000
• an employment count greater than zero
• IR10 income (rent received, interest and dividends and total income) greater than $40,000.

A snapshot of the Business Frame was obtained for the first week of October 2005. This date was chosen to match with the other major data sources used, being the most recent date for which consistent data was available across all three sources.

The Business Frame classifications used to identify non-profit institutions included:

• business type
• institutional sector (NZISC)
• industrial activity (ANZSIC 96).

At the highest level the NZISC recognises five distinct sectors:

1. non-financial producer enterprises
2. financial enterprises
3. general government
4. non-profit institutions serving households
5. households.

All of sector 4 was in scope for the count in this report by default. Incorporated societies in other sectors (for example racing clubs, business associations and industry training organisations) were, unless under government control, also included by definition. Unincorporated associations in other sectors were assessed by industry to determine whether they were non-profit institutions. Also included were charitable companies but trading or family trusts, which made up the majority of trusts, were excluded.

Non-Business Frame population

Because the Business Frame only includes non-profit institutions that pass one of the size thresholds listed above, other sources were required for the thousands of institutions that do not meet these criteria. The two main sources of information used for these smaller institutions were administrative databases maintained by Inland Revenue and the Companies Office lists of incorporated societies and charitable trusts.

From the administrative databases maintained by Inland Revenue, institutions potentially in scope were those classified as qualifying trusts, incorporated societies and unincorporated associations. Added to these, the Companies Office provided a list of charitable trusts and incorporated societies on their registers as at the end of September 2005.

Reconciling the populations

The institutions from the three (overlapping) population sources were integrated and reconciled. A hierarchical approach was initially taken to remove institutions duplicated in the overall list. For example, those institutions that were in the Business Frame population were removed from all other population subsets because the Business Frame provides the most information. Secondly, the Companies Office registers of incorporated societies and charitable trusts took precedence over the administrative databases maintained by Inland Revenue.

Beyond this, other methods employed included name matching and ‘fuzzy’ matching where only high probability matches were implemented. Fuzzy matching is a search function enabling names of institutions to be grouped according to whether they have a high probability match, a medium probability match or some other possibility. The need for such matches arises from the name of an institution being on two or more lists but with different formatting, spelling or completeness.

Limitations of the data

As the study of the Masterton District Council revealed (refer to chapter 4), the number of non-profit institutions identified may still be under-counted. On the other hand, it may also be overstated through the failure to remove all duplicates from the various registers and the inclusion, because of not having full information, of institutions that do not meet the full definition of a non-profit institution.

Moreover, the Inland Revenue and Companies Office registers are maintained for non-statistical purposes, therefore registered institutions may not be ceased at the same time as they are on Statistics New Zealand’s Business Frame. Furthermore, the administrative databases maintained by Inland Revenue do not easily allow charitable trusts to be distinguished from trading or family trusts. To identify charitable trusts, the analysis therefore relied on the Companies Office register.

The administrative databases maintained by Inland Revenue have many thousands of institutions listed as unincorporated associations. For a large number of these, no associated tax data exists, such as GST sales or purchases. It was possible to verify that a small number of non-profit institutions report GST as part of a group return. Therefore, although the individual non-profit institution is recorded with zero GST, it is nevertheless still active. For many other small non-profit institutions, however, it is not known if they are actively operating or if they simply retain a listing on the administrative databases maintained by Inland Revenue because they have an Inland Revenue number. It is therefore possible that the number of unincorporated associations is overstated.

Counting employment numbers

The head count of salary and wage earners is sourced from taxation data on a monthly basis. The employment count is obtained primarily from administrative databases maintained by Inland Revenue.

Annual Enterprise Survey

The Annual Enterprise Survey provides financial information by industry and sector groups. This includes measures of financial performance and financial position. Output variables include income, expenditure, profit, purchases of fixed assets, and equity. The Annual Enterprise Survey data also forms the basis of national accounting variables such as value-added, gross output and gross fixed capital formation.

Population

The target population for the Annual Enterprise Survey is all economically significant businesses operating within New Zealand. The population for this survey is selected from the Business Frame.

In total, the Annual Enterprise Survey is estimated to cover approximately 90 percent of New Zealand’s gross domestic product (GDP). Some industries are excluded; the Australian and New Zealand Standard Industrial Classification 1996 (ANZSIC) industry exclusions are:

• residential property operators not elsewhere classified (L771100-90)
• foreign government representation (M813000)
• religious institutions (Q961000)
• private household employing staff (Q970000).

Design of the Annual Enterprise Survey

The current design of the Annual Enterprise Survey was introduced in the 1999 financial year. The Annual Enterprise Survey was designed as the principal collection vehicle of data used in the compilation of New Zealand’s national accounts. The data collected feeds into the calculation of the economy’s GDP, via the current price annual industry accounts, which are compiled within an input-output framework.

The Annual Enterprise Survey collects financial data for most of the industries operating in the New Zealand economy. The Annual Enterprise Survey industries are based on the ANZSIC. The Annual Enterprise Survey is designed at approximately the four-digit ANZSIC level, or 107 industries.

Sample design

The Annual Enterprise Survey is a stratified sample. Each industry contains between one and four strata, defined by size of turnover (sourced from GST information) and rolling mean employment. Each industry has a full coverage stratum made up of large units with significant economic activity within their industry group. Most industries also have a tax strata where IR10 information is used for self-employed individuals and partnerships up to a level of $10 million turnover. The remaining strata contain a sample of medium sized units.

Religious institutions

Religious institutions have been excluded from the Annual Enterprise Survey in the past. Given that the Annual Enterprise Survey is the primary data source for compilation of the NPISA, an alternative method of estimating the contribution of religious institutions was required.

Accordingly, the estimate was based on a sample of annual accounts supplemented by dated reports collected from previous studies and a number of reports that were available on the registers held by the Companies Office.

Statistics NZ also utilised the Annual Enterprise Survey data that was available for institutions from the activity subgroup religious institutions not elsewhere classified. This subgroup represents institutions that do not have administration or worship services as their primary activity, and that do not belong in another activity group, but that do have religion as their purpose. This might include a retail store or a record label whose information would be collected as part of an industry study.

Compiling the data

Religious institutions, especially Christian ones, have complex structures of administration. Many denominations include regional and national bodies and there are transfers of money and labour between them. Furthermore, the legal structures of financial accounts mean there are often separate legal entities that exist for the purposes of insurance, investment and employment.

Consequently, principles applied in the processing of the accounts included:

• the number of units was determined from other sources instead of from the legal entities on the Business Frame
• the entities that existed within the control of clearly recognised bodies were considered part of that body and their financial flows netted out and recorded as part of the controlling body’s.

Statistics NZ acknowledges the Vision Network directory of churches for its usefulness in assessing the overall number of churches.

Estimating the contribution of non-economically significant units

The non-economically significant units consist of some institutions on the Business Frame and all institutions from other administrative databases or registers. These two groups were estimated independently and then added together to produce an estimate for the financial contribution of non-economically significant units.

Those institutions on the Business Frame were represented by units that had participated in the Annual Enterprise Survey. There were not enough units to create an average estimate for each activity group and only enough for the estimation of a single homogenous institution. For all other institutions the estimates were based on a sample of annual accounts of charitable trusts and incorporated societies on the Companies Office registers.

Calculating and valuing hours of formal unpaid work

Any activity that was identified as formal volunteer labour in the Time Use Survey 1998/1999 was extracted. Rules were then set up to exclude ‘out of scope’ activities, such as sleeping and socialising. This was an iterative process. All activities were then analysed according to the eight organisation codes used in the Time Use Survey.

These were:

1. Mäori-based committee, organisation, grouping, etc
2. Disability support and health-related services
3. Social support and assistance
4. Education
5. Community safety and protection
6. Leisure and recreation
7. Member benefit groups
8. Other.

The total number of hours volunteered for these organisation groups was then calculated and totalled, giving the total number of hours volunteered for the period 1 July 1998 to 30 June 1999. It also provided the average hours volunteered per person for all New Zealanders over the age of 12.

To extrapolate the Time Use Survey, total hours to March 2001 (the census period), the average number of hours volunteered per person in the population aged over 12 years, was multiplied by the population aged over 12 years at 31 March 2001. This assumes that the number of hours volunteered was constant, and that the number of volunteers grew at the rate of the population. This was done because there was no adequate data available for 1999 to 2001. This had only a negligible effect on the overall estimate.

This average number of hours worked per volunteer is also assumed to have remained constant between 2001 and 2006. However, the number of volunteers has increased by 2.1 percent. Using a simple straight line equation, the growth of 2.1 percent between 2001 and 2006 for the number of volunteers was modelled to 2004. The modelled increase in the number of volunteers was then multiplied by the average number of hours worked per volunteer, thus giving the total hours volunteered per year. It was assumed that the types of activities remained the same as in 1998/99.

Key assumptions for volunteer labour

• The average hours worked per volunteer has not changed since the 1998/99 Time Use Survey.
• The types of activities (and therefore their proportional representation in each activity grouping) have not changed since the Time Use Survey 1998/99.
• The monetary value of one hour of work for any activity in the paid sector equals the implicit monetary value of one hour of work for the same activity in the unpaid sector.
• The Time Use Survey 1998/99 has provision for ancillary activities (that is, doing two or more activities at once). It has been assumed that these activities have the same productivity as primary activities, since adjusting for any productivity reduction due to multi-tasking (and therefore a reduced wage rate) would be purely on an arbitrary basis.

Differences between the census and the Time Use Survey

The Census of Population and Dwellings is a self-administered questionnaire that asks a range of questions. The census has the advantage of collecting data from the whole population of New Zealand, the data can be broken down to a regional level, and it is also relatively timely for this report. Although the census collects data about unpaid work, this is not its primary aim. It is generally agreed that the census is a lower-end estimate for the number of people involved in informal and formal unpaid work outside of the home.

The Time Use Survey for 1998/99 collected data about formal unpaid work in two modes: through a personal questionnaire, which collected demographic and activity data from the four weeks prior to completing the questionnaire, and a 48-hour diary, which recorded all activities for a 48-hour period. The Time Use Survey is the most finely-tuned survey instrument available for getting a good estimate of volunteer labour, both for the number of volunteers and the hours that they work. However, the main limitation is that the data is now some years out of date. The data from the next Time Use Survey is expected to be available in 2010.

Both the census and the Time Use Survey collect data about the number of volunteers in New Zealand. However, their estimates differ quite significantly for several reasons.

The census is conducted as a self-administered survey and the Time Use Survey is an interviewer administered survey. This affects the response rates because in self-administered surveys there is no interviewer present to probe for further answers or to explain the question if it is misunderstood. Interviewer administered surveys lead to fewer misunderstood questions and fewer inappropriate responses.

The Time Use Survey has a further advantage in that it is able to use diaries to pick up on the respondents’ unpaid activities that have been overlooked or misunderstood in the survey questions. This may lead to an increase in response rates.

While the census is a key part of a wider integrated population and social statistics system, it cannot provide the depth of information of a targeted social survey such as the Time Use Survey. In the census, unpaid work is a third tier subject and greater emphasis is placed on other questions. This is evident in that the term ‘voluntary work’ is placed at the bottom of the unpaid work question and has few behavioural prompts due to the limited space on the census form. The Final Report on Content for the 2006 Census concluded that although unpaid work would be better suited to a specialist time use survey or similar, it had important value in measuring social capital, providing recognition of the importance of this work, and in allowing people who do not undertake paid work to record their participation in unpaid activities.

Classification of non-profit institutions

The United Nations Handbook on Non-Profit Institutions in the System of National Accounts recommends the International Classification of Non-Profit Organizations (ICNPO) as a tool to differentiate between the types of institutions defined as NPIs. This is primarily based on their ‘economic activity’, as is the International Standard Industrial Classification (ISIC) on which NZSCNPO is based. Some purpose criteria are included where activities are similar. By grouping together institutions in this way, we form a basis for meaningful data analysis.

The main ICNPO groups are:

• 01 culture and recreation
• 02 education
• 03 health
• 04 social services
• 05 environment
• 06 development and housing
• 07 law, advocacy and politics
• 08 philanthropic intermediaries and voluntarism promotion
• 09 international
• 10 religion
• 11 business and professional associations, unions
• 99 not elsewhere classified.

See Appendix 2 in Non-profit Institutions Satellite Account: 2004 for the full New Zealand Standard Classification of Non-profit Organisations (groups and subgroups).

As the ICNPO is mainly an activity-based classification, it has no specific categories for population groups such as women or people with disabilities. Many Māori groups fit within the NPI criteria and are included in the NPISA. Categories for institutions targeting a population subgroup are classified on the predominant activity of the institution. For example, if an institution provides medical treatment for a disability then it is coded to the ‘health’ group. If it provides social assistance for people with disabilities, it comes under the ‘social services’ group.

Many NPIs have multiple activities, each activity falling under a separate ICNPO group, but the institution can only be classified in one group. In these cases, we use the institution’s ‘primary economic activity’ to assign an appropriate ICNPO category.

It is usually measured as the activity with the largest share of:

1. value-added – a measure of the institution’s contribution to gross domestic product
2. gross output, if value-added is not available
3. employment, if neither value-added nor gross output is available.

Classifying non-profit institutions in the satellite account

Coding NPIs from the Business Register

For the Statistics NZ Business Register population we made the initial classification using a concordance with the Australian and New Zealand Standard Industrial Classification 2006 (ANZSIC06). For example, preschool education (ANZSIC P8010) is concorded with early childhood education (NZSCNPO 2 110), hospitals (ANZSIC Q8401) with hospitals and rehabilitation (NZSCNPO 3 100), and residential property operators (ANZSIC L6711) with housing (NZSCNPO 6 200).

Note that where an NPI carries out two or more distinct activities, its ANZSIC06 code is that of the majority activity.

For unallocated institutions from this first analysis, we applied extensive keyword search lists. Although several hundred keywords (eg ‘tennis’ and ‘church’) were used, the list is still not exhaustive.

Finally, we made manual classifications for many institutions that remained, and for units where their activity was known.

Coding outside the Business Register

For the institutions we added from Inland Revenue’s administrative database, the Companies Office register, and the Charities Services’ register, the overall method was similar. We applied the ANZSIC-NZSCNPO concordance if an Inland Revenue ANZSIC tag was available. For the remainder, we used the keyword analysis. A further refinement was based on the results of analysis to eliminate duplicates from the data, and after analysing lists of donee organisations and organisations receiving government contracts.

The NPIs we could not code were those where either their activity was truly different to those we included under any other main group or where insufficient information was available to allow coding.

We used keyword coding where no industrial code was available. Where multiple keywords were in the name, we applied a precedence list. For example, ‘The Church of XYZ Tennis Club’ would be classified to sports institutions because the keyword ‘tennis’ has a higher classification ranking than ‘church’.

It is difficult to identify the sometimes very specific NPIs in some NZSCNPO subgroups; for example, the ‘income support and maintenance’ subgroup or the ‘employment and training’ subgroup. Institutions may have been coded to the overarching ‘social services’ or ‘development and housing’ groups instead, based on their industry code or keywords. However, the results at the main group level are robust.

Another difficulty revolves around precisely what NPIs and activities are included under each subgroup. For example, the ‘fundraising’ subgroup includes large, nationally active NPIs with fundraising as their main activity. However, institutions fundraising to support a specific activity covered under another main group will be coded to ’support and ancillary services’ under their main group.

A second example is that rest homes and other aged residential care (except nursing homes providing first and foremost medical services) are under ‘social services’, not under the health subgroup ‘nursing homes’.

Distinguishing market and non-market non-profit institutions

A ‘market’ producer is an institution that sells its goods or services at competitive market (or ‘economically significant’) prices. Most NPIs are classified as ‘non-market’ because they provide their services for free or below market prices. However, a significant number of NPIs are market producers, and are included in the NPI population as long as they meet the structural/operational definition.

Examples of NPIs classified as market producers include:

• most business associations
• gaming trusts
• most NPI hospitals
• racing clubs.

These units are distinctive within the NPI population. To generate a positive operating surplus, these NPIs need to sell their output at market prices. At the same time, they do meet the legal requirements of being not for profit because the surplus is generally not retained within the institution, but is distributed to another NPI or for a charitable purpose. Business associations are classified as market producers where they are financed by dues and subscriptions rather than by government. The rationale is that business associations are NPIs that support market producers.

Gaming trusts need to produce large operating surpluses in order to make community grants. Because of this, they are implicitly assumed to be charging economically significant prices, even though the market in which they operate is heavily regulated. Similar comments apply to racing clubs, although for them any operating surplus goes back to industry participants (through measures such as increased stakes).

Most hospitals are classified as market producers because they charge market, or close to market, prices. They are classed as non-market where they clearly charge below-market rates (eg charity hospitals where staff are volunteers).

Many NPIs rely on government funding that is contested. NPIs that receive funding from government, on the basis of winning contested contracts, are not necessarily market producers. If they have a significant volunteer labour component, they may charge out their services at below-market rates.

From these examples it is apparent that it is not immediately obvious whether or not an NPI should be classified as market. If it is unclear whether economically significant prices are being charged by the NPI but it is making a positive operating surplus over time it usually implies it is operating on a market basis. By applying these principles we can establish an acceptable set of market and non-market classifications for the NPI population.

Making international comparisons

A market and non-market split also allows for better comparisons with the NPI satellite accounts of other countries. Canada has both a ‘core’ and a ‘non-core’ set of NPI satellite accounts. The non-core set includes hospitals, universities, and colleges. The non-core estimates are provided separately because hospitals are very significant within the Canadian NPI account. The Australian NPI estimates treat hospitality clubs and business and professional associations as market producers.

Data sources and methods

Counting the number of non-profit institutions

Although many NPIs are found on the Business Register, we need to include other information sources to get a clear picture on the number of NPIs.

Business Register population

The primary source of information for counting NPIs is Statistics NZ’s Business Register, which identifies enterprises. For an enterprise to be on the Business Register it must meet any one of certain criteria; for counting NPIs, the most relevant criteria are:

• annual goods and services tax (GST) expenses or sales of more than $60,000
• an employment count greater than zero
• IR10 income (rent received, interest and dividends, and total income) greater than $40,000.

The Business Register classifications used to identify NPIs included:

• business type
• institutional sector (NZISC)
• industrial activity (ANZSIC 06).

At the highest level, the NZISC recognises five distinct sectors:

• non-financial producer enterprises
• financial enterprises
• general government
• NPIs serving households
• households.

All NPIs serving households were in scope for the count in this report by default. Incorporated societies in other sectors (eg racing clubs, business associations, and industry training organisations) were included by definition unless under government control. We assessed unincorporated associations in other sectors by industry to determine whether they were NPIs. We included charitable companies but excluded trading or family trusts, which made up the majority of trusts.

Non-Business Register population

Because the Business Register only includes NPIs that pass one of the size thresholds listed above, we needed other sources for the thousands of institutions that do not meet these criteria. The three main information sources we used for these smaller institutions were administrative databases maintained by Inland Revenue, the Companies Office, and the database maintained by Charities Services.

Institutions potentially in scope from the Inland Revenue and Companies Office databases were those classified as qualifying trusts, incorporated societies, and unincorporated associations. The Charities Services register provided a list of registered charities, which we added to these.

Reconciling the populations

We integrated and reconciled the NPIs from the three (overlapping) sources. We took a hierarchical approach to remove institutions duplicated in the overall list. For example, we first removed NPIs in the Business Register population from all other population subsets – because the Business Register provides the most information. Second, the Companies Office and Charities Services registers of incorporated societies and charitable trusts took precedence over Inland Revenue’s administrative databases.

Beyond this, other methods we employed included name matching and ‘fuzzy’ matching. Fuzzy matching is a search function that enables names of institutions to be grouped according to whether they have a high probability match, a medium probability match, or some other possibility. We need these matches because an institution’s name may be on two or more lists but with different formatting, spelling, or completeness.

Limitations of the data

The number of NPIs identified may still be undercounted due to being unable to identify institutions of a more ‘informal’ nature. For example, groups with large memberships but which are organised on a relatively informal basis (eg local walking, gardening, or tree planting groups; groups that are organised online).

In contrast, the population may be overstated through failing to remove all duplicates from the various registers and by including (due to a lack of information) institutions that do not meet the full definition of an NPI.

Overstatement could also result from the Inland Revenue and Companies Office registers being maintained for non-statistical purposes, which means registered NPIs may not be ‘ceased’ at the same time as they are on the Business Register. The administrative databases that Inland Revenue maintains do not easily allow charitable trusts to be distinguished from trading or family trusts – to identify charitable trusts, our analysis relied on the Charities Services register.

Inland Revenue’s administrative databases list many thousands of institutions as unincorporated associations. For a large number of these, no associated tax data exists (eg GST sales or purchases). We verified that a small number of NPIs reported GST as part of a group return. Therefore, although the individual NPI was recorded with zero GST, it was still active.

However, we cannot know if many other small NPIs are actively operating or if they are still on the administrative databases maintained by Inland Revenue because they have an Inland Revenue number. It is therefore possible that the number of unincorporated associations is overstated.

Counting the number of employees

The count of salary and wage earners is sourced from taxation data on a monthly basis. The employee count comes primarily from administrative databases maintained by Inland Revenue and from the Charities Services databases.

Annual Enterprise Survey

The Annual Enterprise Survey (AES) provides financial information by industry and sector groups. This includes measures of financial performance and financial position. Output variables include income, expenditure, profit, purchases of fixed assets, and equity. AES data is also the basis of national accounting variables such as value-added, gross output, and gross fixed capital formation.

Population

The target population for AES is all economically significant businesses operating within New Zealand. The population for this survey is selected from the Business Register. In total, we estimate AES covers approximately 90 percent of New Zealand’s gross domestic product (GDP). We exclude some industries; the ANZSIC06 industry exclusions are:

• residential property operators not elsewhere classified (L671100)
• foreign government representation (O752200)
• religious institutions (S954000)
• private household employing staff (S960).

Design of the Annual Enterprise Survey

The current AES design was introduced in the 2009 financial year. AES was designed to be the principal collection vehicle for data used in compiling New Zealand’s national accounts. The data collected feeds into calculating the economy’s GDP, through the current-price annual industry accounts, which are compiled within an input-output framework. AES collects financial data for most industries operating in New Zealand’s economy. The survey is designed at approximately the four-digit ANZSIC level (it has 107 industries).

Sample design

AES is a stratified sample. Each industry contains one to four strata, defined by size of turnover (sourced from GST information) and rolling mean employment. Each industry has a full-coverage stratum made up of large units with significant economic activity within their industry group. This includes non-profit units sampled from units collected by the Charities Services survey of charitable trusts. Most industries also have a tax stratum, where IR10 information is used for self-employed individuals and partnerships up to a level of $10 million turnover. The remaining strata contain a sample of medium-sized units.

Religious institutions

We exclude religious institutions from AES. Since AES is the primary data source for compiling the NPISA, we needed an alternative method of estimating the contribution of religious institutions. The estimate we used was based on income and expenditure data from the Charities Services data collection, supplemented by annual financial accounts information for larger institutions. This is a much more comprehensive and consistent collection than we used for the 2004 NPISA, which had a sample of annual accounts supplemented by reports collected from previous studies, plus reports available on registers held by the Companies Office.

Estimating the contribution of non-economically significant units

The non-economically significant units consist of some NPIs on the Business Register and all NPIs from other administrative databases or registers. We estimated these two groups independently, then added them to produce an estimate for the financial contribution of non-economically significant units. NPIs on the Business Register were represented by units in AES, supplemented by data from sources such as government departments, crown entities, and funding agencies. For all other NPIs we based the estimates on data collected by Charities Services, which surveys most registered charities.

Calculating and valuing hours of formal unpaid work

Based on the Activity Classification for Time Use Surveys (ACTUS), we identified formal volunteer labour within the ‘committed activity’ from the Time Use Survey 2009/10. We extracted information about all committed time activity that was worked for an organisation. We analysed these activities according to the 12 NZSCNPO codes used in the Time Use Survey.

These were:

• 01 Culture, sport, and recreation
• 02 Education and research
• 03 Health
• 04 Social services
• 05 Environment
• 06 Development and housing
• 07 Law, advocacy and politics
• 08 Grant making, fundraising, and voluntarism promotion
• 09 International
• 10 Religion
• 11 Business and professional association, unions support, and services
• 99 Not elsewhere classified (residual category).

The number of hours volunteered for these organisation groups was calculated and totalled, which gave us the total number of hours volunteered for the period 1 September 2009 to 31 August 2010. It also provided the average hours volunteered for each New Zealander over the age of 12 years.

We extrapolated from the Time Use Survey to find the total hours for the year to March 2013 (the reference period and coinciding with the census). We multiplied the average number of hours volunteered for each person in the population aged over 12 years, by the over-12-years population (at 31 March 2013). Doing this assumes the number of hours volunteered was constant, and that the number of volunteers grew at the same rate as the population between 2010 and 2013. We did it this way because no adequate time-use data was available for 2013 – the process had a negligible effect on the overall estimate. We also assumed that the types of activities remained the same as in 2009/10.

Key assumptions for volunteer labour

Because the Time Use Survey year did not coincide with the NPISA reference year, and because it did not place a monetary value on time spent volunteering, we made the following assumptions:

• The average hours worked per volunteer did not change between the 2009/10 Time Use Survey and March 2013.
• The types of activities (and therefore their proportional representation in each organisation group) have not changed since the 2009/10 Time Use Survey.
• The monetary value of one hour of work for any activity in the paid sector equals the implicit monetary value of one hour of work for the same activity in the unpaid sector.
• The 2009/10 Time Use Survey provided for ancillary activities (ie doing two or more activities at once). We assume these activities have the same productivity as primary activities, since adjusting for any productivity reduction due to multi-tasking (and therefore a reduced wage rate) would be arbitrary.

Differences between the census and the Time Use Survey

The Census of Population and Dwellings is a self-administered questionnaire that asks a range of questions. The census has advantages: it collects data from the whole population of New Zealand, the data can be broken down to regional level, and it is timely for these NPISA statistics. Although the census collects data about unpaid work, this is not its primary aim. The census provides a lower-end estimate for the number of people involved in informal and formal unpaid work outside the home.

The 2009/10 Time Use Survey collected data about formal unpaid work in two modes: through a personal questionnaire, which collected demographic and activity data for the four weeks before completing the questionnaire; and a 48-hour diary, which recorded all activities for a 48-hour period. A Time Use Survey is the most finely-tuned survey instrument available for getting a good estimate of volunteer labour, both for the number of volunteers and the hours they work. However, its limitation is that the data is now some years out of date.

Both the census and the Time Use Survey collect data about the number of volunteers in New Zealand. However, their estimates differ quite significantly for several reasons. The census is a self-administered survey and the Time Use Survey is interviewer administered. This affects the response rates; when self-administered, no interviewer is present to probe for further answers or to explain a misunderstood question. Interviewer-administered surveys have fewer misunderstood questions and fewer inappropriate responses.

An advantage of the Time Use Survey is using diaries that can pick up respondents’ unpaid activities that may be overlooked or misunderstood in the survey questions. This may increase response rates.

While the census is a key part of a wider integrated population and social statistics system, it cannot provide the depth of information of a targeted social survey such as the Time Use Survey.