The drivers of higher unemployment across countries

Unemployment is rising in nearly every advanced economy, sometimes sharply and almost exclusively because of a weaker demand for labour. Further large rises are possible given that job vacancies are broadly back at pre-COVID levels in most countries, which would place additional pressure on central banks to cut rates. Falling vacancies are yet to return to pre-pandemic levels in the euro area and Australia, which could delay the rise in euro area unemployment from its multi-decade low and slow the increase in Australia’s unemployment rate, which, unusually, has so far been driven by an increased supply of labour (higher unemployment in NZ has also been driven by increased supply, although weak demand should take over when Q2 statistics are released this week).

After the largest co-ordinated tightening in over two decades, several central banks, mostly European, have started to lower their policy rates.

These central banks have become more confident that inflation will return to target over the next year or two, but are also responding to higher unemployment.  

Central banks are sensitive to higher unemployment because it places downward pressure on inflation in the traditional Phillips curve framework, where some banks give it additional weight if required by government.

The typical central bank reaction function involves setting the policy rate relative to the estimated neutral level in response to the gap between inflation and its target and the output gap.

The output gap is often approximated by the gap between the unemployment rate and the estimated NAIRU. Unemployment is timely, but can be volatile, with volatility exacerbated by lower survey response rates in the wake of COVID in some countries, most strikingly the UK.

In a modified Taylor rule, the weight on the unemployment gap is regularly assumed to be two, such that an unexpected 0.5pp increase in the unemployment rate would see the central bank consider cutting the policy rate by 100bp, other things being equal.

However, in practice, central banks often smooth interest rates, gradually adjusting the policy rate by a fraction of the recommendation from a reaction function. For example, Federal Reserve analysis puts this smoothing weight at 15% in the US, such that a 0.5pp increase in the unemployment rate would see FOMC consider an initial 25bp rate cut in rounded terms (the RBA Martin model has a 30% weight in its policy rule, which would also point to a 25bp rate cut in rounded terms).

Central banks generally believe that the NAIRU changes slowly over time, which means they would view the rise in the unemployment rate to date as representing a later-than-usual response to higher interest rates that alters the gap with the NAIRU.

Comparing labour markets across the advanced economies, the median increase in the unemployment rate from its recent low point has been about ½pp to date, although there is a wide variation across countries.

• Unemployment in the Nordics and Canada has increased by a very large 1½pp, with New Zealand not far behind at just over 1pp and the US at nearly 1pp;
• The UK has seen a ¾pp increase in unemployment to date, while Australia and a couple of small advanced economies have experienced a more modest increase of about ½pp, in line with the median increase;
• Unemployment in Korea and Japan is little changed, up a marginal ¼pp, which is within the bounds of survey volatility in most countries; and
• Euro area unemployment is the outlier, having yet to increase from a multi-decade low.  

The longer response time of unemployment partly reflects the impact of COVID, where the initial impact of higher interest rates on the demand for labour has been met by falling job vacancies, which reached their highest level in decades across the advanced economies during the pandemic.

Focusing on a handful of countries, job vacancies are now below pre-pandemic levels in New Zealand and broadly in line with them in the US, UK and Canada. This suggests that the usual negative relationship between unemployment and vacancies – known by economists as the Beveridge curve – is likely to reassert itself, such that further weakness in the demand for labour is more likely to be reflected in a sharper rise in unemployment in these countries.

In contrast, vacancies are still falling in the euro area and Australia, but are yet to return to pre-COVID levels. This suggests that a further decline in vacancies could slow the increase in unemployment in Australia for a time and delay the increase in the euro area, where unemployment is still at its lowest level since the early 1980s.

    

Comparing the rise in unemployment to date with other periods of labour market weakness, each country has averaged about ten episodes of rising unemployment in the period from 1970 onwards.

Small rises in unemployment have tended to occur over short periods, but unemployment can often increase for two or three years, with Europe experiencing long stretches of rising unemployment.

The rise in unemployment to date is at the smaller end of the historical experience for the US, Australia, and the UK, while it is approaching the middle of the distribution for Canada and, most likely, New Zealand, where Q2 data are due this week. The euro area, as already highlighted, is the odd one out with no increase in unemployment to date.   

Given the range of recent outcomes, we tried to get a better understanding of the role of higher interest rates in driving unemployment higher by considering the relative importance of supply and demand in the labour market.

Ordinarily, the supply of labour does not matter much for central banks, with higher interest rates restraining the demand for labour and pushing up the unemployment rate.

However, COVID triggered extreme swings in both demand and supply across labour markets and the supply of labour is yet to return to pre-pandemic trends in several countries.

For example, at the broadest level, growth in the working-age population is yet to return to pre-COVID rates in many countries, with Canada, New Zealand, and Australia still posting extremely large increases of 3% or more over the past year.

  

In our analysis, CCI used a simple technique to split the increase in each country’s unemployment rate – covering both past episodes and the increase to date – into contributions from the demand and supply of labour.

This involved using the identity that expresses the unemployment rate as a function of two key ratios, namely the ratio of employment to the working-age population and the participation rate, which is the ratio of the workforce to the working-age population.

Exploiting this identity, demand was measured using the weighted change in the employment-population ratio and supply was approximated using the weighted change in the participation rate.[1]

Analysing the periods of rising unemployment using this framework, higher interest rates play a consistent and normally dominant role in increasing unemployment by reducing the demand for labour. Supply often partly offsets the boost to unemployment from weaker demand when some people give up looking for work and drop out of the workforce.

Focusing on one extreme of the cross-country experience, the US stood out, in that every past period of rising unemployment involved a large increase in the jobless rate that was driven by weaker demand for labour, signalling a recession in each case.

At the other extreme, there were a small number of instances where an increased supply of labour drove the increase in unemployment. These instances generally involved a smaller increase in unemployment, most notably in the early 2000s in countries other than the US and in the 1970s in Canada, New Zealand and (once in) Australia. 

In the latest episode, all, or nearly all, the increase in unemployment to date can be attributed to weaker demand in the US, the UK, and Canada. This suggests that higher interest rates are working to cool the labour market, placing additional pressure on forward-looking central banks to cut rates as inflation slows.

In Australia and New Zealand, the picture is different and the increase in unemployment has so far been driven by an increased supply of labour, rather than by higher interest rates curbing demand.

• In Australia, the employment-population ratio is marginally below the level from when unemployment was at its most recent low point, while the participation rate is up slightly, near its record high.
• In New Zealand, the story seems out of date given the calculations are based on Q1 data, with Q2 statistics due this week likely to show weaker demand taking over as the driver of higher unemployment given the employment-population ratio fell sharply at the start of this year.

The difference with other countries suggests that the RBA might be a little less concerned than usual in worrying about a gradual supply-driven increase in unemployment, particularly when there still seems scope for job vacancies to absorb further weakness in demand.

The RBNZ faces a different, more pressing challenge, with pressure to cut rates from an already large rise in unemployment that should be boosted by weakness in demand in Q2 given business surveys suggest there could have been a large contraction in GDP in the quarter.

Note:

[1] The unemployment rate = 1- (employment/working age population)/participation rate. The linear approximation of the change in the unemployment rate includes a small interaction term that was added to the contribution from the supply.