2  Data Description

The dataset used in this paper is the prototype Longitudinal Business Database (LBD), developed by Statistics New Zealand (SNZ).^[6] This dataset is a combination of several resources which are shown in Appendix A and includes longitudinal data on all economically active firms in New Zealand.

Economically active enterprises included in the dataset are defined as enterprises that met at least one of the following criteria in a particular year:

• LEED rolling mean employment (RME) greater than zero
• GST sales greater than zero
• GST purchases greater than zero
• IR 10 total income greater than zero
• IR 10 total expenditure greater than zero
• IR 10 total fixed assets greater than zero.

Most important sources in the prototype LBD are:^[7]

GST Returns

Definition: Goods and services tax (GST) is a tax on most goods and services in New Zealand, most imported goods, and certain imported services.

Eligibility: Businesses need to register for GST if their annual turnover is more than $40,000. Below this threshold they may, but not have to, register for GST.

Collection: GST data is collected on a monthly, bi-monthly or six-monthly basis by IRD, depending on the annual turnover of the business.

Variables: GST data include information on sales/income & purchases/expenses.

IR10 Accounts Information

Definition: IR10 data is essentially a general summary of information relating to the customer’s business and operations.

Variables: This form contains information on sales and other income, purchases, a breakdown of expenses including but not limited to depreciation, research and development, and salaries and wages. Balance sheet items also include: current assets, fixed assets (broken down into vehicles; plant and machinery; furniture and fittings; land & buildings; and other), other assets, liabilities (broken down into current and term), and shareholders funds.

Eligibility: IR10 is designed to collect information for statistical purposes and is not a compulsory form.

IR4 Income Tax Return for Companies

Eligibility: All active New Zealand resident companies must file an income tax return every year by completing IR4.

Variables: This form includes variables on New Zealand and overseas income (including interest, dividends and income from “business or rental activities”) and losses. They also contain a binary foreign-ownership indicator.

LEED

Definition: LEED data is constructed by Statistics NZ from IRD tax data, notably Pay-As-You-Earn (PAYE) returns for employees. To protect the confidentiality of individuals, LEED variables available in the LBD dataset have been aggregated to the enterprise-level (allowing the data to be accessed through the Data Lab).

Variables: LEED data includes counts of employers (on an annual basis) and employees (on a monthly basis) with matching data on income. Summary characteristics of individuals also include gender and banded age breakdowns, tenure distributions of employees, and summary measures of the dispersion of wages within the firm.

Customs Data

The Customs data is recorded monthly at the shipment level and contains information on products at the 10 digit Harmonised System classification, value, volume and currency of the transaction, whether it has been hedged, means of transport and the country of destination and origin.

The definitions of “sector”, “industry” and “product” are based on New Zealand Harmonised system classifications.^[8]The international HS follows a hierarchical structure, comprising 21 sections, 98 chapters (2 digit), 1228 headings (4 digit), and 5,059 sub-headings (6 digit). This structure is further broken down into approximately 13,500 statistical keys (10 digit) for New Zealand's purposes. Throughout the paper, sections are referred to as sectors, chapters as industries and 10 digit codes as products.

In the sample period from 1996 to 2007, there have been various revisions to these 10 digit classifications such as introduction of new goods and changes in the tariff system. Since the analysis is based on the product mix of firms, it is crucial to get a consistent definition of products, so the dataset is manipulated such that if two or more products have been merged or if a single product has split into several products, these have been aggregated into a single product.^[9]