![Black and Gray Laptop Computer Turned on Doing Computer Codes](/_next/image?url=https%3A%2F%2Fproduction.infometrics.co.nz%2Fmedia%2Fimages%2Fpexels-christina-morillo-1181271.width-1024.format-avif.avif&w=2048&q=75)
Nowcasting employment to improve our dashboards
The Monthly Employment Indicators published by Stats NZ are a valuable addition to the suite of employment measures in New Zealand. They are extremely up to date, being published approximately four weeks after the reference month. They also provide a lot of geographical and industry detail. This article outlines how Infometrics is utilising this valuable data source to enhance our regional products.
Employment is a key economic indicator
Employment levels, and changes in employment over time, are key indicators of industry and regional economic performance. For example, increases in healthcare employment in the past year indicate that the recent surge in net inward migration to New Zealand has alleviated chronic skill shortages in the healthcare sector. Falls in agricultural employment are indicative of rising farming costs, falling international commodity prices, and falls in agricultural production resulting from adverse weather events in early 2023. Forecasts of employment are equally important indicators of future economic performance, and are used by regional and sector agencies, central government, tertiary education providers and businesses across a range of sectors.
More timely employment data
During the COVID-19 pandemic, Stats NZ started to the publish the Monthly Employment Indicator (MEI) series. MEI offers detailed industry employment counts for TAs and is published about one month following data collection, which is timely compared with other detailed employment series such as the Linked Employee Employer Dataset (LEED). LEED also counts industry employment for TAs but is published around 9 months after the data the data is collected. Because of its timeliness, we have been using MEI in our Quarterly Economic Monitor.
However, the drawback of MEI from our point of view is that it counts employment based on the place where the employee resides (a residence-based count). A residence-based count can give a distorted picture of local labour markets in TAs where a lot of commuting occurs into and out of that TA. LEED on the other hand counts detailed employment based on the location of the business where the person is employed (a workplace-based count). Both MEI and LEED give us detailed employment data. How do we combine the timeliness of MEI with the workplace-based element of LEED?
The workplace-residence transformation matrix
We have been looking at ways to use MEI to bring our LEED employment counts more up to date – a process called ‘nowcasting’ which currently means spanning the gap between the September 2022 quarter and the September 2023 quarter. However, for MEI to be useful we need to convert MEI from being a residence-based count to being a workplace-based count.
We converted MEI from a residence-based employment count (which we call MEI-Res) to a workplace-based employment count (MEI-WP) using a ‘workplace-residence transformation matrix’. We built the matrix using data extracted from LEED which has details of both workplace address (from quarterly LEED) and residence address (from annual LEED) of individuals in employment. The matrix shows, for every level 4 ANZSIC industry, the TA where employees work and the TA where they live.
Chart 1 shows an example of the output of the process described above. The example uses a single industry (C111300 - Cured Meat and Smallgoods Manufacturing) in a single district (Carterton) which has a bacon factory. The chart shows that of the 370 persons employed in the industry in Carterton in September 2023, the largest proportion of workers (175) live in the much larger adjoining district – Masterton – while only 95 live in the district of Carterton.
Nowcasting
We use the MEI-WP data series, which extends to the September 2023 quarter, to nowcast LEED from the September 2022 quarter to the September 2023 quarter. In turn, we use the nowcasted LEED data series to nowcast our Regional Industry Employment Model (RIEM) to the September 2023 quarter. RIEM sits at the core of many of our online, interactive, economic data dashboards, delivering quarterly historical counts of employment by 500 level 4 ANZSIC industry for every Territorial Authority (TA) in New Zealand from year 1999. ANZSIC is the Australia and New Zealand Standard Classification of New Zealand.
All this means that we now have a continuous quarterly time series of employment by 500 level 4 ANZSIC industries for every TA from June 1999 to September 2023, which we will shortly start using in our Quarterly Economic Monitor.
Nowcasting employment will improve our dashboards and forecasts
Bringing our detailed workplace-based employment counts up to date enables several significant improvements in our online, interactive economic data dashboards.
- Employment counts in our Quarterly Economic Monitor will soon be based on the location of the business at which a person is employed, rather than where the employee lives, as it is now. This change will provide a better measure of an area’s economic performance, particularly for those areas in which a lot of commuting between areas occurs. It will also mean our measurement of employment in the Quarterly Economic Monitor will be consistent with the Regional Economic Profile.
- Our updated employment counts will be used to inform our provisional quarterly regional GDP estimates, which we use in our Quarterly Economic Monitor.
- Our employment nowcasts will inform our industry employment forecasts (available through our Sector Profiles and Te Matapae) as we will have a very up-to-date base from which to forecast.
Annual commuting flows in the pipeline
Having detailed residence based and workplace-based employment counts means we can develop annual estimates of changes in commuting patterns between TAs. Currently, the Census is the only data collection that offers reliable commuting estimates and that happens only every five years. The annual commuting estimates will take some time to bring up to our quality standards, after which we will incorporate them into our Regional Economic Profile.