Units of observation

Owned by Philip Stevens
Last updated: Aug 25, 2022
3 min read

The basic unit of observation is the enterprise. This roughly corresponds to the economic concept of the firm. Below that are ‘Kind of Activity Units’ (KAU) and Geographic Units (GEOs).

KAUs are less relevant to the LBD. They are units which engage in one predominant economic activity, and the concept is used for calculating industry level national accounts (via the Annual Enterprise Survey).

Geographic Units (GEOs) are normally a physical location from which predominantly one kind of activity takes place on a permanent basis. They correspond to what economists call the ‘plant’, and others might call a ‘branch’.

In single-location, single activity entities, the enterprise (ENT) equals the Kind of Activity Unit (KAU) equals the Geographic Unit (GEO), as depicted in the leftmost diagram below.

An enterprise may have many KAUs, or GEOs. It may also have a number of GEOs under each KAU. This is shown on the right in the diagram below.

 

More complex still is the multiple enterprise entity, where a number of firms are owned by one entity. This is commonly called a group, and that topmost entity is unsurprisingly called a group top enterprise.

Enterprises with group tops and parents
Source: A Rough Guide to New Zealand's LBD

While the LBF shows links between enterprises at each point in time, the practical definition of an enterprise as a legal unit leads to some challenges in identifying individual firms across time as changes in legal structure (eg, a shift from an individual proprietorship to a limited liability company) can lead to firms being assigned a new enterprise number despite being essentially the same economic unit as before. Fabling, R. (2011) 'Keeping it Together: Tracking Firms in New Zealand’s Longitudinal Business Database' provides a method for repairing these longitudinal links, summarised in Bringing it all together.

Give me time!

The fundamental unit of time in the LBD is the financial year. This is because some of the most important data come from annual tax or survey data. Of course, not all firms have the same financial year, so we have to find a way of bringing them together.

The core time variables have a specific form in the LBD and IDI. The form the data in these variables take is the year followed by the month. For example, March 2000 is 200003. Perhaps the most important date variable, which is used to construct fact tables and is usually the unit of time for most analysis is called dim_year_key.

Tax filing rules mean that most firms tend to have a 31st March year end balance date. However, some industries, such as in the public sector, have 30 June balance date. Sub annual data are annualised to each financial year of the firm. Firms' financial years are then allocated to the 31st March year-end that has the greatest overlap with their financial year.

Because of this, the year format of dim_year_key is that of the March at year end, i.e. 200003. To convert other dates into this format, therefore, one can usually multiply the year by 100 and add the month. StatsNZ also provides a handy set of tables and tools for converting dates to dim_year format and for working with the resulting variables using the tables dim_bal_date_year, dim_mar_year and dim_month and functions such as fn_month_get_next and fn_month_span (in Programmability/Scalra-valued Functions).

For more detail see