You can read Part 1 of this series here.
Labour Market Insight (LMI), as the name suggests, is simply insight, information and intelligence about labour markets. On the most basic level, it is information relating to occupations and industries, such as job numbers and salaries, but it can also include a range of more specific details, such as educational levels for occupations, workforce demographics, and the number of establishments in particular sectors.
In terms of where LMI comes from, most of it is freely available from publicly available data sources, which are mainly collated by the Government. Some of the major sources of LMI in the UK include Working Futures (WF); Business Register Employment Survey (BRES); Workforce Jobs Series (WJS); and Labour Force Survey (LFS); Annual Survey of Hours and Earnings (ASHE); and Employer Skills Survey Working (ESSW).
The purpose of LMI is really to use current and historic industry and occupation data, in order to facilitate a better understanding of the trends, structure and demands in an economy. The information is also useful in that it can be used to project into the future for particular industries, occupations or geographical areas, based on past and current trends. Such information can be of inestimable value to any organisation that is seeking to understand the structure and the possible future of a particular labour market.
If we wanted to sum all that up in a very simple definition, we might say that LMI is:
Data taken from publicly available sources, which gives information on jobs and industries, that helps us understand the structural trends and demands of an economy.
The problem with public datasets
From that definition alone, it might seem that LMI is very straightforward, and that all an organisation has to do to understand the structural trends and demands in its focus labour market is to get its hands on those publicly available sources. However, in practice it is not nearly so simple.
The biggest reason for this is that data which is available from public sources is what we call “raw data”, by which we mean two things: Firstly, each of these datasets contains inherent strengths and weaknesses, and none of them can give a complete picture of the labour market. Secondly, within each one of these publicly available datasets, some of the data is subject to suppressions. For instance, for specific industries or occupations in a given geography, the data will often be suppressed to avoid revealing sensitive data. This means that it is often impossible to identify details of a local labour market at a granular level.
In order for these raw datasets to be turned into useful data, two things need to happen:
1. The weaknesses of particular datasets need to be discarded and the strengths retained.
2. Any suppressions in the data need to be removed.
There is a further issue with the use of such data, which is simply one of practicality. Most such datasets are delivered on huge excel spreadsheets, and unless the user is a data scientist, attempting to navigate around them can be a complex and very frustrating business.
Solving these problems
Since its foundation in 2000, Emsi has been developing solutions to these problems. To overcome the first problem – the differing strengths and weaknesses in each dataset – what we do is to bring a variety of sources together to create a robust composite dataset, taking the best of each set and discarding any weaknesses. This means that we can describe the labour market in far more detail than if we are just relying on data from one or two sources. For instance, by combining Business Register Employment Survey (BRES) with Workforce Jobs Series (WJS), which describe detailed local industry jobs and much more recent regional jobs figures respectively, we are able to provide estimates down to local industry level with much greater recency.
However, as we mentioned above, along with having various strengths and weaknesses, each of these datasets also comes with some of the data subject to suppressions. Most of these occur at the most specific industry, occupation, or geographic levels, and the reason they are applied is to prevent too much information about specific groups of workers being revealed. So for example, if there are only a small number of establishments in a particular industry in a region, the data might be suppressed in order to prevent one firm finding out details about their competitors.
Leaving these suppressions in the data means that you are going to end up making assumptions about the more granular details you don’t have, based on the broader data that you do have. But the problem with this is that it will inevitably produce erroneous figures. If you have data on a 2-digit Standard Occupation Classification (SOC) – say Leisure, travel and related personal service occupations – but you don’t have data on the occupations that fall within this group at the 3 and 4-digit SOC code level, you might well assume that because the 2-digit SOC classification has grown by a certain percentage, the more specific occupations within the classification have also grown at the same rate. But this is far from being the case.
For instance, between 2013-2018, the 2-digit SOC category, Leisure, travel and related personal service occupations grew by 3% in the GFirst LEP region. However, if we look at the 4-digit SOC occupations within that classification, we can see that they are not only extremely disparate jobs, but they have grown or declined at very different rates:
Making the assumptions about change at the 4-digit level based on data for the 2-digit level is clearly going to lead to some very inaccurate data. This is why we have spent a lot of time and resources on creating a robust data modelling technique to unsuppress the data, in order to reveal the true figures for industries and jobs at the most specific SIC and SOC levels (this is all perfectly legal by the way). What this means is that we are not only able to combine a variety of different datasets to produce a more holistic view of the labour market, but we can do so at the most specific levels of industry and occupation. What is more, we can do this for any geography in the country, including national, Government Office Region, LEP region, and Local Authority area.
From industries to occupations and back again
The modelling technique that we use to unsuppress those public data sources to end up with our own holistic and granular dataset also has another huge benefit to the end user. Whereas those sources by themselves describe aspects of industries or occupations, but not both, by combining them and then “filling in the blanks”, we are not only able to give you granular data about industries by themselves, or occupations by themselves, but we are able to link the two together to give a much deeper and richer picture.
For example, if you wanted to understand more about a certain industry in your area, our data will enable you to very quickly see how many people are employed in it, what the median salary is, how much the job numbers have changed over a given period, and how much the sector is set to grow in the coming years. However, this doesn’t actually tell you anything about which jobs are employed in the industry, and how much each of those are paid, have grown, and are set to grow. But because of the way our data has been modelled, we can run a Staffing Pattern to identify the occupations that are employed in the sector, the percentage of the sector that is made up by each occupation, and a number of other metrics. The example in the table below is for the Top 10 occupations within the Professional Services cluster in the Swindon and Wiltshire LEP region:
As you can see from the table, this process enables us to get a much better understanding of the make-up of an industry. Although most of the jobs listed are high paying occupations, requiring a degree, which we might naturally expect in Professional Services, there are other lower paid more administrative roles at Level 2 and 3, reminding us that industries don’t just employ occupations that we might directly associate with them – they also employ a number of other more generic jobs.
What this function means is that we can very quickly get a window on the occupation and skills demand for very specific sectors that are of interest. And because the process can also be run in reverse, where instead of beginning with an industry and unpicking the occupations it employs, we start with an occupation and run an Inverse Staffing Pattern to identify which industries it is found in, we can also quickly begin to understand in which sectors particular occupations and skillsets in a region are found.
Job Postings Analytics
Another part of the data that we need to consider is Job Postings Analytics (JPA). As a company, we have always cautioned against using job postings by itself as a measure of labour market demand, for a number of reasons. Firstly, employers do not always advertise their jobs, and so it is only ever a partial view of employment demand. Secondly, a single job vacancy can refer to multiple jobs, and so there is the issue of actual job counts. Thirdly, real-time sweeps can never pick up every job on the internet. Fourthly, many jobs are seasonal, and so a snapshot of jobs postings at certain times — say in the run up to Christmas — will not give a true indication of labour market trends. And finally, whereas the structured LMI we have discussed in this paper so far all refers to actual jobs, JPA does not represent actual jobs, but rather job vacancies, which may or may not be filled.
That said, JPA is indicative of what employers are looking for right now, and can therefore offer some great insights as to current skills demand, particularly if used carefully in conjunction with the more structured LMI. Emsi’s JPA data contains over 800,000 unique job postings every month, and because we have integrated it alongside our LMI, we can map the job titles that appear in the postings across to 4-digit SOC codes. What this means is that the data can be used to answer the following questions about employer demand in your local labour market:
• What jobs are currently in demand in our region?
• Which companies are looking to hire in our area and what industry are they in?
• Which skills are most needed in our region?
• What are the most relevant soft and hard skills to a specific job title?
• Which jobs share the same or similar skillsets as each other?
Being able to answer these questions gives you a vital window on your labour market, and when combined with the more structural LMI, can really help your organisation understand a number of crucial aspects about your local economy.
How we deliver the data
The careful modelling of different datasets to produce our LMI as described above, combined with JPA, leaves us with something entirely unique: employment trends dating back to 2003, projections out to nine years in the future, and a window on employer skills demand, all at a level of detail that makes the data useful to a broad range of local, regional, and national organisations. But without an effective method of deploying the data, it would still remain largely inaccessible to all but the most data savvy.
We have two main ways of delivering our data to you. The first of these is through Analyst, which is a highly accessible and easy-to-use online tool that enables users to find out answers to questions on industries, occupations and skills for their area simply and quickly. With instant access to our complete dataset, an intuitive interface, and presentation-ready reports that can be downloaded, Analyst users can very quickly become self-sufficient in being able to answer complex labour market queries in high volume and at high speed.
In addition to delivering the data through software, we also employ a team of expert economists who are able to answer very specific questions through custom consulting reports, a couple of which we mentioned in the Foreword. The aim of these consulting pieces is to develop a tailored analysis that addresses particular strategic questions LEPs are wrestling with, both in regards to general challenges around skills and growth, and also more specific issues such as Skills Advisory Panels and Local Industrial Strategy. Fundamentally, these pieces are about how skills and growth factors interact with the wider environment, including technological change; wider labour market issues; the spatial distribution of economic development activity; and how employers respond to skills shortages.
What we have done at Emsi is to create one holistic dataset from a number of sources; one which is greater than the sum of its parts, which reveals the granular details that were suppressed in the original data, and which allows users to see the occupational make-up of industries. Furthermore, in addition to this detailed structural view of the labour market, we have also incorporated real-time Job Postings Analytics to give a much richer picture of current employment trends, including who is hiring, what roles they are looking to fill, and what skills they are looking for. This insight is available through a single software system – Analyst – which has been designed to allow users to investigate their labour market at the touch of a button, or through in-depth consulting reports, which seek to answer very specific questions asked by the LEP.
Get in touch to find out how we can help you build a solid evidence-base for your skills and industrial strategy.