Is falling unemployment masking a broader deterioration in UK labour market performance? The ease with which a typical job seeker lands a job is a crucial indicator of the health of the labour market, which cannot be fully inferred from just a casual glance at the headline unemployment rate. It is true that unemployment has declined quite rapidly recently. But this is because job openings have been unusually abundant while the labour market’s capacity to match individual workers to available jobs quickly has actually worsened. This capacity is referred to as matching efficiency, and it started falling in the UK even before the 2008 recession.
A little theory
The concept of matching efficiency comes from a standard macroeconomic framework for thinking about the labour market called equilibrium unemployment theory. According to this approach, job formation reflects the outcome of a two-sided search process in which job seekers and firms trying to hire new employees search for one another in the labour market. The matching process that brings them together is not instantaneous. Rather, it is subject to frictions, meaning that it is time consuming and takes effort since not all vacancies and job seekers are perfect matches for each other. So at any given moment, there are always some unfilled vacancies and some people who are still searching for them. Granted, you don’t need a theory to tell you that much. The usefulness of the theory, however, stems from its ability to describe the nature of the frictions that characterise this costly matching process adequately.
This description is provided through the use of what macro-labour economists call the matching function. The matching function is a parsimonious mathematical tool which neatly summarises the endlessly complex interactions among all the different job vacancies and job seekers and the happenstance that brings them together. Leaving the technical details aside, the matching function is analogous to the (better known) idea of the production function for aggregate output. Instead of taking capital and labour as its inputs, the matching function takes job openings and the pool of job seekers as its inputs. From these inputs, it’s as if the labour market “produces” an ongoing flow of newly occupied jobs, or matches. Matching efficiency then simply refers to the productivity of the labour market in terms of creating newly occupied jobs from a given stock of vacancies and job seekers. It is affected by things such as how suitable the skills of job seekers are for the jobs on offer.
But matching efficiency is not the only factor that determines the probability that a typical job seeker becomes employed. Success at finding a job also depends on the business cycle state of the labour market. This is often summarised by looking at the ratio of vacancies to job seekers—commonly referred to as labour market tightness. The tighter are labour market conditions (i.e. the more vacancies there are relative to people seeking jobs), the higher is the probability that a typical job seeker becomes employed.
The flow of newly formed job matches is a key influence on the unemployment rate. Identifying whether changes in unemployment are mainly the result of movements in labour demand associated with the business cycle or more deep rooted structural disturbances in the labour market that affect matching efficiency is important for monetary policy. That is because monetary policy can only help to lower cyclical unemployment caused by a shortfall in labour demand. Structural movements in unemployment are much harder to deal with. An example of a shock that could affect matching efficiency is a large recession that impacts certain job types severely, leaving others unscathed. Since labour of a particular skillset cannot immediately switch to another occupation—construction workers do not tend to become healthcare workers overnight, say—such a reallocative disruption in the labour market creates an asymmetry between the types of labour demanded by firms and the skills distribution of potential recruits. This is called mismatch, and is one reason why measured matching efficiency might fall in a recession. To the extent that unemployment is high because of mismatch, there is little that monetary policy can do to address it.
Now that we’ve briefly sketched the theory, we can move on to the empirics. Three bits of data are needed in order to measure matching efficiency: (i) the stock of unfilled vacancies, (ii) the pool of job seekers, and (iii) the flow of job seekers who successfully become employed in a given period. Notice that the term “job seekers” is much broader than the headline unemployment measure. Many working age individuals who are classified as economically inactive in the Labour Force Survey, and are therefore not actively looking for a job, nevertheless end up becoming employed in the near future. Such transitions from out of the labour force to employment are roughly equivalent in magnitude to the flow of people from unemployment to employment in a given quarter. It is therefore important to adjust the headline unemployment rate for such passive job searchers when estimating matching efficiency.
To keep things simple for now, let’s just focus on headline unemployment as a proxy for the total measure of job seekers instead of the broader definition I just outlined. Labour market tightness fluctuates a lot over the business cycle. Figure 1 plots this ratio against the probability that an unemployed person becomes employed in the following quarter. Labour market tightness and the job finding probability tend to move closely, with both falling sharply during the 2008 recession. However, although the labour market is now even tighter than it was prior to the recession, the job finding probability is still about 5% below its pre-recession norm. The data therefore indicate that a wedge has opened up between the job-finding probability and labour market tightness, implying that matching efficiency may have worsened.
Figure 1: Labour market tightness and the job finding probability
That is corroborated by estimates from a formal statistical model—this time controlling for passive searchers, as described before. The model also controls for changes in the long-term unemployment share: the long-term unemployed find jobs at a very slow rate and so a higher incidence of long-term unemployment for a given level of market tightness would show up as a decline in matching efficiency unless properly accounted for.
The matching efficiency index from this model is shown in Figure 2. The first notable feature of this index is that it declined sharply prior to the 2008 recession. Interestingly, recent research on the US economy uncovered a similar finding. It’s hard to say with certainty what is driving the deterioration in aggregate labour market performance. One plausible hypothesis for the UK is that it is related to a phase of unusually rapid labour force growth around 2005. As the pool of available workers expands, it is reasonable to expect some—at least temporary—deterioration in the matching process as heterogeneity increases across labour force participants, particularly if new entrants are less productive and less attractive to hire.
There is another visible dip in matching efficiency right as the recession began in 2008, which reversed by the end of 2011. The decline in 2008 is likely to be related to dispersion in job destruction rates across occupational types in the immediate aftermath of the recession, causing a discrepancy between job losers and job openings, with the people leaving their jobs not necessarily being well suited to the jobs available. The 2008 recession had only a fleeting effect on matching efficiency, which has remained flat around its pre-crisis level since 2012.
Figure 2: Matching efficiency index
What the data suggest so far is that job seekers’ chances of finding a job in the UK are lower than would be expected given how tight the labour market has been. This conclusion is broadly consistent with other business intelligence gathered by the Bank’s network of Agents. How persistent the decline in matching performance turns out to be will have important implications for the evolution of unemployment. In general, low matching efficiency tends to slow the rate at which unemployment falls during cyclical recoveries. In the latest data, although unemployment did drop further, this appears to be due to a remarkably low job destruction rate rather than a significant improvement in exits from unemployment. Weak matching efficiency might be increasing the incentives for firms to raise employment through a fall in the firing rate rather than through hiring.
Most academic research on matching efficiency suggests caution about expecting a rapid improvement. One fairly robust (and somewhat counterintuitive) finding in the applied literature is that matching efficiency tends to trend downward over time (Petrongolo and Pissarides, 2001). This contradicts the notion that easier job search brought about by information technology has resulted in faster job finding rates, as a moment of casual introspection might otherwise initially suggest.
Detecting a trend is one thing. Explaining it is another. One possibility is that a gradual move towards increasingly specialised labour markets might actually be complicating the matching process. Maybe finding a match is increasingly becoming like looking for a needle in a haystack.
Bradley Speigner works in the Bank’s Structural Economic Analysis Division.
If you want to get in touch, please email us at email@example.com
Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.