Recession watch, UK

[David Smith]

Today's figures were remarkably good, though you wouldn't necessarily think so from the way they were reported. But I take your point about the most recent flows.

Quite interesting that deep in the figures the difference between UK born and non-UK born appears to have disappeared in the latest data. Both seem to have fallen by proportionally similar amounts (1.5% and 1.2%) over the past 12 months.

[cmilas]

Hi,
I am not sure if Paul has his own forecasts but one way (of the very many ways) of looking at this is via an Okun-type regression of changes in the number of unemployed people in (i) annual gdp growth relative to "normal growth" and (ii) lags, that is:

u-u(-1)=a+b*(gdp_growth-2.5%)+c*u(-1)+d*(u(-1)-u(-2)).

Using the GDP projections from the August inflation report, I forecast the number of unemployed people to rise to 2.52million in 2009q4 and continue rising up to 2.63m in 2010q4. I will avoid forecasting beyond 2010. Obviously, with 200bn of QE and more optimistic growth rates reported by the Bank yesterday (exact numerical values will be available next Wednesday), "my" unemployment forecasts reported above will be slightly lower.
Thanks
Costas

[paulbiv]

I haven't done forecasts recently for ILO unemployment recently, but Costas' forecasts above would come within what I would call 'broadly flat', and re-estimation using the later figures might accentuate that with a definable risk of reduction towards the latter end of the period.

Claimant count is at the moment affected upwards by administrative factors, so a consistent series may well have fallen. Of course it will probably take the next five years to persuade ONS they need to do that work.

The administrative factors are:

Flexible New Deal (phase 1) effect - the youth claimant count for longer term claimants (ONS codes AGNC and AGND) are affected by young people staying on count to a greater extent starting from April and accelerating in July as New Deal Option starts terminate before Flexible New Deal started in October - this is temporary as a policy change starting in October and coming into force on January 2010 will remove most of this group from the count again, and as this policy change is temporary, later on they come back.

Lone parents - ONS Table 10(1) - which is one month further delayed - a monthly increase of 4,260 lone parents with youngest child aged 12-15 NSA (code JDA5). As from October, those with youngest child aged 10-11 (code JDA4) will start to accelerate as they are moved into the count. October 2010 on current schedules code JDA3 are affected in the same way (with much bigger underlying numbers).

Employment and Support Allowance refusals - unknown number but may be 17-18,000 a month (that's assuming that all the 50,000 off-flows from ESA in quarter to May are refusals and go straight to the unemployed count and that it continues at that level). There's some Parliamentary question numbers on the proportion of cases refused but no proper stats yet.

Paul

[cmilas]

Doing my Christmas shopping over the last 2 weeks or so, I noticed that things are worryingly much more quiet comapared with the same time last year. [The alternative, internet purchases, are not my cup of tea. That said, I tried to buy a laptop on-line to avoid the VAT rise but the websites of all major computer retailers were extremely slow; presumably everybody else was doing the same].

Weak retail sales shouldn't be surprising given the latest saving rate figures . However, Marks and Spencer appears to be doing quite well this year.
Last but not least, very best wishes for a Merry Christmas and a very healthy 2010 to David and the list members.
Costas

[paulbiv]

A bit difficult to judge - last year a lot of people came out for the Woolworths firesale - and Borders isn't quite at the same scale, though I did see a book on "how to make money in Dubai" in Victoria Books etc.... If it had been pennies I'd have got it as a souvenir.

Retailers seem to have been expecting a marginally better year than last year - Jobcentre new vacancies for sales occupations (not seasonally adjusted) slightly up on the year. Data at http://spreadsheets.google.com/ccc?key=0AoS2tWC7FmCFdGNGbkdFVXpJS0NmVjVkcy1xWDhLRVE&hl=en_GB - Don't know if Google spreadsheets links will work.

The Internet shopping would come under Transport and mobile machine operators (drivers and forklift operators) and elementary plant and storage (warehouse and post people). Customer services, the beloved call centres, are back up. The other elementary group include shelf stackers and catering assistants, so equally seasonal.

These are all vacancies through Jobcentres, including agency ones placed there. Weighted to bottom end of the market. If you look at the spreadsheet and wonder why Jobcentres have lots of vacancies for corporate managers, it's where they put managers of shops.

[cmilas]

The latest OECD predictions about the UK output gap for 2010-2011 (available from:
http://www.oecd.org/dataoecd/26/40/38785295.htm) look "rather" unrealistic. OECD expects the UK output gap to be –6.2% in 2010 and –5.5% in 2011. How these square with annual growth rates of 1.2% for 2010 and 2.2% for 2011 (available from the same link) is a mystery to me. Let’s hope they do not attract prominent attention by the media. [Incidentally, OECD expects the policy rate to rise to 2.1% at the end of 2011].
Costas

[David Smith]

Interesting - hadn't looked at that. The implication, and not just for the UK, is that the OECD seems to be assuming very weak trend growth after the crisis. Ireland's output gap moves in a similar way.

[paulbiv]

Agree it's interesting.

One does wonder if they are forecasting the series individually or constraining them so they make sense. Surely an output gap forecast should be the result of a a calculation involving potential growth (based on population etc. and GDP forecasts, with output gaps falling out of the calculation naturally. Maybe they are following the ONS pre-recession population forecasts of continued high net in-migration. I'd guess that also applies to Republic of Ireland.

The OECD's Business Cycle Clock has most UK indicators either in recovery or expansion phases (the exception being Industrial Production).

There's probably an inbuilt delay in international indicators so the information in the model might be a couple of months old, say, just after the first release of UK GDP data for the third quarter showing continued recession when the surveys had suggested recovery.

[cmilas]

Using the approximate relationship:
annual_growth(t)-trend_growth(t)=output_gap(t)-output_gap(t-1),

OECD thinks that trend growth has been 2.4% until 2008 and then dropped sharply to 1.5% in 2009. It will drop further to 1.2% in 2010 and go back up to 1.5% in 2011. Difficult times ahead (also for unemployment) if these predictions are taken seriously.
Costas

[paulbiv]

A pretty good set of labour market figures, much as I expected, though I was surprised that the youth figures were so good.

I don't think the number of subsidised jobs through the Future Jobs Fund are as yet big enough (month by month) to make much of a dent in the figures (and Government continues to add extra people diverting from lone parent benefits and from Incapacity benefits).

Jobcentre vacancies are particularly strong, with particularly notable upward trends for elementary trades, plant and storage (aka warehouse etc), transport and mobile machine drivers, customer service jobs, process, plant and machine operatives, leisure and other personal service jobs. Skilled construction seems to have come off the bottom, but skilled metal and electronics still languishing. Sales jobs going a bit haywire, but possibly up. Jobcentre vacancies not reliable at all for professional and managerial jobs.

The charts and spreadsheets as usual are up on the work website.
http://www.cesi.org.uk/statistics/

[cmilas]

The difference between real-time and final-time data is quite striking and is reflected in measures of the output gap; the latter is of course used by the MPC in their policy decisions made in real time.

Using real-time data, the output gap is twice (!!!) as volatile as the one with final-time data. Equally important, real-time output gap is consistently below final-time output gap by an average of 2 percentage points, suggesting that the economy was consistently performing below capacity since 2001(!!). The implication (via policy reaction function calculations) is that if real-time data was taken seriously into account by the MPC, UK interest rates would have been (on average) lower by as much as 1 percentage point since 2001.
In light of the above, I will conclude there is no reason for the MPC to place more weight on unemployment (as Blanchflower suggested) unless of course the ONS manages to reduce measurement errors.
Thanks
Costas

[paulbiv]

I suspect that explicitly adding in unemployment to MPC remit would add greater political pressures, which are large enough already.

However, there is a case that some (not the headline) unemployment-related measures may be concurrent with cycle rather than lagged and less subject to measurement errors than GDP and output gap.

Blanchflower has highlighted the involuntary part-time figures (working, part-time but want to work full-time). These have had a bigger effect this recession - partly due to the Tax Credit effect discussed in the SMPC minutes. Equally, the inflows to either claimant count or ILO unemployment look pretty concurrent to me.

So are the unemployment figures affected by measurement errors - well, of course, but if we know about the sources of error, they can be accounted for.

I've said here, repeatedly, that the change in claimant count is higher than trend, due to policy changes on lone parents, on new forms of employment programme and on new benefits for those with health problems. We have official measures on the lone parent effect and can, with a bit of work, identify the employment programme effect. The Incapacity Benefits effect is likely to be larger than the combination of the other two but I'd have difficulty in estimating its effect to any degree of accuracy.

Claimant count figures just get revised once, the following month. The factors I've outlined wouldn't be called measurement errors by the ONS, but possibly discontinuities of a small but possibly significant size.

ILO unemployment figures will get revised subsequently, due to reweighting of the population base. However, the survey returns remain intact, it's the weighting that changes. In particular, the population change is based on ONS population projections, which in turn largely depend (absent flu pandemics with significant mortality) on the migration projections. Migration projections are the dodgiest part of the whole population projection business, so there is a source of measurement error there.

It's probably a question of how explicitly you refer to the overall thing you are trying to measure, or to a set of indicators that may be correlated with what you want to measure, but the degree of corrrelation may change over time.

[paulbiv]

Hello again

My analysis of the labour market figures, including downloadable spreadsheet with data (the Download link) is now online at http://www.cesi.org.uk/statistics/

I was rather surprised by last month's claimant count figures, but these have now been revised down very substantially. It looks as though previous trends have resumed after the snow.

Some of the January oddity appears to have been related to people leaving the count to go onto schemes like New Deal options. These are declining over time due to changes in programme design.

While long-term unemployment is rising (as would be expected to follow approx. 12 months after unemployment inflows rocketed), it looks as if the flows into long-term unemployment are falling quite sharply. It may be that some of these are moves into subsidised Future Jobs Fund jobs rather than market jobs, but it is interesting.

These positive trends for longer-term unemployed (if they are not simply subsidised jobs) would show something about the hysteresis assumptions in most of the models. If long-term unemployed are not increasingly isolated from the labour market, then greater flexibility in jobsearch and job matching could have broader effects (e.g. on average weekly earnings).

[David Smith]

Many thanks. I find these points particularly interesting:

"In previous recessions the proportion of the unemployed not claiming benefit has shrunk to around zero. This recession, we started with a historically high proportion of unemployed not claiming JSA. The proportion not claiming fell, as expected. In March–June the pattern changed in an unexpected way, with the proportion of the unemployed who are not claiming JSA rising. The latest data shows the proportion not claiming JSA staying at 34% (837,000)."

"The youth unemployment rate has now dropped slightly to 17.5%. The rate for those aged 25–49 is now 6.2% and for those aged 50 and over is 4.7%. The rise in unemployment rates since the recession started has been 5.3 percentage points for young people, 2.3 points for those aged 25-49 and 1.9 points for those aged over-50."

[paulbiv]

Yes, those points either do or don't match expectations.

For the unemployed (ILO) not claiming JSA we just compare the numbers. There are some odd overlaps as well (can claim JSA if working under 16 hours, but as both Policy Exchange and Centre for Social Justice point out, the disincentives are punitive).

There's a lot of chitchat in comments about JSA being limited to six months, which rather takes aback those of us who are long in the tooth. The six-month limit applies to contributory JSA - so those not eligible for income-based JSA dropped out in 1995-6 when that changed. This applies mostly to second earners in couples and those with savings worth counting. However, since then the flows past the six month barrier have shown much the same responses to economic change as any other flow rates, there's nothing new.

What seems to be going on is something related to the increasing proportion of women in the labour force (including the above). The availability of a substantial in-work benefit system (Tax Credits) calculated on a household basis, means that unemployed second earners have choices - claim JSA or change the Tax Credit claim. Since the recession started roughly 0.5 billion a month extra has been going onto Tax Credits http://www.hmrc.gov.uk/stats/tax_receipts/menu.htm. Government has been known to say that Tax credits have helped x families in the recession.

In addition (and for the youth) we also have students. This is not a mutually exclusive group from the labour market. When DWP examined the numbers of unemployed not claiming JSA when the pattern started to behave in unexpected ways they discovered that lots of them were students, at which point they stopped looking. You can see this in their press releases.

We do try and break down the young people statistics by whether or not in full-time education (Chart 6). The gap between JSA and unemployed not in FT Ed has closed a bit, but this is where the rise is. The unemployment rate as quoted is a bit distorted by silly denominators - unpacking what labour market participation means for young people is more than usually complicated, given that employment is 1 hour plus per week, and so unemployment can be looking for 1 hour jobs.

Government response has been to print a guide for parents to help boomerang youngsters who are out of work.

Would probably be helpful if the over-50s lobby would check what was going on before lobbying, but I guess this is too much to ask. There probably is something about over-50s who lose jobs finding it more difficult to get back in, but this is outweighed by people staying in work who last recession would have been pensioned off (regulatory changes in discrimination law).