Working out the unemployment figures

Most economists are taught in their cradles never to overestimate the importance of one month's data. A better rule, I believe, is never to underestimate the importance of one month's data, especially when a 'shock' figure is published at or near a major cyclical turning point.

There have been many occasions in the past when a sudden shock has been dismissed as an aberration at the time, but has later been seen as the start of an important new trend. I remember in particular the deterioration in the balance of payments in the mid 1980s, and in the public sector borrowing requirement in the early 1990s, being signalled by a sudden atrocious figure coming apparently out of the blue.

It follows that when two shock figures are published back-to-back, it is even more important to take them seriously.

This has now happened with the monthly declines of about 26,000 in unemployment seen in both February and March. There is no question that this decline in unemployment has come surprisingly early in the economic recovery, and it needs to be explained.

First, is it 'genuine'? Clearly, we know that the economy is now recovering quite rapidly (and this should be confirmed by a rise in non-oil gross domestic product of about 0.8 per cent when the first-quarter figures are published later today).

Furthermore, there has been a confirmatory rise in unfilled vacancies. These have risen by about 6,000 a month on the official data so far this year, which allowing for the usual under-recording in the official figures implies that the overall total of vacancies in the economy may be rising by about 20,000 per month. Vacancies have been a good coincident indicator of economic activity in recent cycles, and this is a relatively strong rate of increase for the early stages of an upswing.

Business survey data are also beginning to suggest that the number of jobs in the economy may be stabilising. The Dun & Bradstreet survey for March was the first to report that a positive balance of firms, albeit a tiny one of plus 4 per cent, is now expecting to take on extra labour. But this was not fully confirmed by last week's Chambers of Commerce survey, which showed a negative balance of minus 8 per cent in the manufacturing sector, offsetting a positive balance of plus 1 per cent in services. It will be interesting to watch the CBI quarterly trends survey tomorrow for further evidence of a recovery in employment expectations.

If unemployment has indeed started to fall, why is this happening at a time that seems so early in the economic recovery? One factor is that some form of recovery (slow at first, and now much faster) has been progressing for much longer than people realise - at least for a year, possibly more. But there are two more important factors.

First, perhaps because of the long- term effects of union reform, firms seem now to be treating labour as a much more variable input into the production process than they have done in previous cycles.

There is good evidence in manufacturing, though much less in services, that companies have maintained productivity growth at unusually high levels throughout the recession by shedding more labour than past relationships would have implied. About 2 million jobs have disappeared in this recession, roughly twice as many as in the 1979- 81 downswing. (Even more dramatically, the decline in jobs during the 1974-75 recession was only 93,000).

It is correspondingly possible that firms will add to their labour forces more quickly than usual during the upswing. Furthermore, from a shorter-term perspective, the economic crisis last autumn produced a virtually unprecedented burst of job shedding, with 398,000 people (1.6 per cent of the workforce) losing their employment in a single quarter. This left firms unprepared for the recent acceleration in demand. As the graph shows, if we take an average over the whole of the past nine months, the change in unemployment has not been far out of line with the pattern at this stage of the last two cycles.

The second main reason is that the underlying labour supply is increasing far less rapidly now than it was in the last two economic recoveries, so any job creation is likely to feed through to falling unemployment more rapidly.

The population of working age was rising at about 250,000 per annum as the economy recovered in 1983-84. Now it is rising at about 50,000 a year, and demographics will keep the number of school leavers low for many years.

Economic forecasters have known about these two factors for some time, and have attempted to take account of them in their projections. Nevertheless, most forecasters thought that unemployment would rise to at least 3.2 million before peaking some time in 1994. The recent unexpected improvement in the unemployment statistics will probably cause a wholesale downward revision of these projections, since they suggest that the two favourable forces mentioned above may be having an earlier impact than anticipated.

This in turn may have further consequences. As employment projections are revised upwards, the fundamental under-pinnings for the recovery in domestic demand will appear more soundly based. As everyone is now aware, the American recovery has been held back in recent quarters by the absence of job creation, which has severely retarded the growth of real incomes and has made the consumer recovery dependent on the shaky foundation of declining savings. If employment growth begins to strengthen relatively early in the British upswing, the opposite effect could apply here.

Naturally, extra jobs will tend to dampen the usual strong pick-up in productivity at this stage of the cycle, which will in turn lead to faster growth in wage costs than has been anticipated.

Furthermore, wage settlements might themselves begin to firm as unemployment drops - all of which raises the crucial question of how far, and how fast, unemployment can drop without triggering a significant rise in inflation.

It is all very well to sit back and enjoy the first stages of a demand-induced drop in unemployment, but the long-term determinant of unemployment is on the supply side of the economy.

As explained here two weeks ago, I am far from persuaded that the 'equilibrium' level of unemployment - that consistent with stable inflation - has been brought down during the 1980s by labour market reform. Indeed, it is possible that the recent recession may actually have seen it rise further. Nevertheless, it would be surprising if the 'equilibrium' male unemployment rate were much above 8-9 per cent of the labour force, compared with the present rate of more than 14 per cent.

If this is indeed the case, then the fundamental disinflationary forces in the economy should remain strong for quite a while yet, even if the level of unemployment starts to decline reasonably rapidly. Although the Government might soon be thinking about tightening policy so that it can dampen the economy over the next 12-18 months, and thus elongate the recovery until the general election is due, I doubt if this will actually be justified on inflation grounds for quite a while.

The balance of payments, even after Friday's encouraging figures, is another matter. Britain is enjoying a demand-led recovery while the rest of Europe is still in deep recession. The sustainability of recovery in 1993 and 1994 depends more on the market's willingness to finance the resulting trade deficit than on anything else.

(Graph omitted)