Voting with their feet

Not literally, of course, as no one since Fionn mac Cumhail has left the island on foot. Most people fly. While reliable and timely data on migration are hard to get, we do know how many people fly out of country, and how many fly in. This data can be had from the CSO for every month since January 2006. Unfortunately, the data are released with a six-month delay (courtesy of Dublin Airport Authority). They are instructive nonetheless.

Between Jan 2006 and Oct 2008, 30,282 more people arrived at one of the seven airports than left. This trend has reversed however. While 35,453 entered the Republic of Ireland between Jan 06 and Oct 06, and 72,936 between Nov 06 and Oct 07, 78,106 people left between Nov 07 and Oct 08.

That is, eighty thousand people took a one-way flight, before the scale of the recession was clear.

Of those 78,106 people, 58,980 left for Great Britain, 11,018 for Germany, 6,750 for Poland, 6,409 for the Czech Republic, and 4,161 for the Baltic countries. Between Nov 07 and Oct 08, 4,164 people came from Italy, 3,966 from France, and 2,456 from Hungary.

33 replies on “Voting with their feet”

Analysis of the PPS numbers as provided by the Dept. Social & family Affairs reveal that for the first 3 months of 2009 we have reached an inflection point – where total number of PPS no.s issued to Irish nationals is as great as the total issued to non-nationals.

Indeed, the 12-month running total of Irish PPS numbers has been pretty constant at around 90,000 p.a. since Feb’08.

The data also shows that the number of Poles coming here as halved, now running at annual rate of just 35,000 having peaked at 94,000 in Jan’07 (i.e. 12-months Feb’06 to Jan’07). Whereas there has been no let up in the number of UK citizens coming to ireland to work – still an around 1300 to 1400 per month or annual clip of 19,000 (same as 2006-2008).

Analysis the numbers in greater detail – to show ‘flows’ of migrants coupled with Census Data of ‘stock’ of migrants as well as estimates of present population of Poles in Ireland suggests anecdottaly that half of them stay for a period of no more than 3 years then leave.

Anecdotally then, since the big influx of Eastern European migrants occurred in 2005 -2008 it is no surprise then to learn that the outflow has already begun.

I would expect the next batch of DAA data released through the CSO to confirm this hypothesis.

I am not sure how much this might impact on the numbers but I have noticed a difference in how people book flights. As the competition gets more and more aggressive I have booked, and friends tell me they have booked, trips by booking an outbound flight with say Ryanair and a return with Aer Lingus and visa versa. Or sometimes even booking outbound and inbound flights separately with the same airline can get you a better deal. This might only account for a small influence on the numbers of “one-way” flights, but worth thinking about perhaps.

That should not matter. Airlines are obliged to report the number of passengers on each flights. Flying with two different companies on a return trip should not affect aggregate data.

Flying out and taking the ferry back, or flying out of Dublin and flying into Belfast does contaminate the data but the sign of the bias is not obvious.

Totally anecdotal, but end of may I will leave this country by plane, to pickup a motorbike in France, and take the ferry LeHavre-Rosslare back. I might be counted a one of those emigrants.
In the same way, I know bikers coming with the Fishgard-Rosslare ferry, touring Ireland, and heading back to Great Britain from NI.

Those 2 examples might not be very significant, but just monitoring one way of entering the country might be biased.

Some wishful thinking in the comments. Don’t like the implication of the figures so try to think of ways to undermine them. One thing I hope we’ve learnt from the last 5 years is “this time it’s not different”. Our economy is on a downturn, there are less jobs. It is a less attractive place to come for foreign nationals and less easy to stay for Irish ones.

There will be net emigration over the next x years, the population may even start to decline again. This has implications for housing demand, consumer spend and government income. Ireland got a boost from the net immigration in the last 5 years, we’re now about to see the opposite effect.

Well it could be people flying out to pick up their new second hand UK car to take the ferry on the return trip back to Ireland. Based on anecdotal evidence that is a factor.

@ Stuart. Totally agree.

During the 12 Thatcher years 1979-1990 almost 200,000 young Irish emigrated (including myself in 1988 – 41000 that year following 40,000 in 1987). But more than half of those returned between 1991-1994. Two-thirds of the Irish that I lived in London with returned during that period. They sold expensive 2-bed flats in Clapham & Battersea and bought 4-bed detached houses in Stillorgan & Clonskeagh for half what they sold their London properties for.

That was a precursor to the Celtic Tiger era, along with a confluence of other factors.

Emigration remains an economic safety-valve against rising jobless especially if the global economy recovers before Ireland.

Housing demand has not been seriously debated since the Bacon Reports.

My understanding is that the six-month delay in publishing the airport data is due not to the DAA but to the CSO because of what it considers to be the commercial sensitivity (i.e. value) of the numbers. The airport data excludes traffic within the Republic of Ireland and Republic-NI data and sea data is not published in the same detail. The format for the published data is detailed to the point of being quite unuser friendly (6,067 airport-pairs, most of them zero, by month: no option for city- or country-pairs, no accumulation of annual or seasonal totals). The airport-listings are in sequence of IATA three-letter codes rather than city or country (with some errors: Toronto appears under two codes). This was all pointed out to the CSO the first month the data was published but there has been no useful response.

The UK CAA publishes similar air travel data (domestic and international) about 20 days after the end on the month in much more accessible form.

Of course, the CSO also includes everyone aged 15 years and above in the labour force and the Census treats a family with a mother aged 70 and a 40 year old son as being a lone parent with children …

The conclusions expressed in some posts regarding net migration up to the end of 2008 are wrong. Air transport passenger figures are not a reliable guide to net migration. If air transport passenger figures only are considered, there is nearly always an apparent large net outflow from Ireland (ROI). The following are the CSO figures for (a) net air transport passenger outflow from Ireland (ROI) (b) net migration into Ireland (ROI) each year from 1996 to 2008. The air transport figures are for the 12-month periods from April of one year to March of the following year inclusive. The migration figures are for the years from mid-April of one year to mid-April of the following year (this is how the CSO reports net migration). For both sets of figures, a + indicates a net inflow and a – indicates a net outflow. All figures are in 1,000s. All figures are calculated from the CSO website (data dissemination service/tourism section).

96/97 (a) -84.5 (b) +19.2 , difference +103.7
97/98 (a) -103.8 (b) +17.4 , difference +121.2
98/99 (a) -70.9 (b) +17.3 , difference +88.2
99/00 (a) -51.6 (b) +26.0 , difference +77.6
00/01 (a) -58.7 (b) +32.8 , difference +91.5
01/02 (a) -47.3 (b) +41.3 , difference +88.6
02/03 (a) -53.0 (b) +30.7 , difference +83.7
03/04 (a) -44.1 (b) +32.0 , difference +76.1
04/05 (a) -7.9 (b) +55.1 , difference +63.0
05/06 (a) +34.5 (b) +71.8 , difference +37.3
06/07 (a) +5.8 (b) +67.3 , difference +61.5
07/08 (a) -57.8 (b) +38.5 , difference +96.3

also calendar year 2008 figure is available for net air transport passenger outflow, but not for net migration – this is:

calendar 2008 (a) -79.3 (b) not known

As can be seen, there was net immigration every year from 1996 to 2008, while there was a net air transport passenger outflow every year from 1996 to 2008 apart from 05/06 and 06/07. The net migration inflow was on averation 82.4 thousand greater than what the air transport passenger flow would have indicated, ranging from 37.3 thousand net immigration under-estimation in 05/06 to 121.2 thousand net immigration under-estimation in 97/98. If the under-estimation for year 07/08 (i.e. April 07 to April 08) was repeated for calendar year 2008 (whch obviously can not be
known), then that would indicate net immigration in calendar year 2008 of
aproximately 17.0 thousand.

The only other official CSO figures available that give an indication of net migration in calendar year 2008 are in the 2008 Q4 Quarterly National Household Survey. This gives figures for the population aged 15 plus. This showed an increased of 31.0 thousand compared with 2007 Q3, implying net immigration in calendar year 2008 of around 15.0 thousand (on the reasonable assumption that the migration trend for the population aged under 15 did not deviate from that for the population aged 15 plus).

While it is impossible to be certain, both sets of figures seem to indicate that net immigration continued up to the end of 2008, albeit at a lower level than in recent years.

The above is intended to be an analysis of published figures covering the period up to the end of 2008 only. It is not intended to be a forecast of net migration in 2009 and beyond. That is in the lap of the Gods and will obviously depend not just on how the Irish economy performs, but on how other economies (esp. U.K, U.S. and eastern Europe) perform.

Historically, migration forecasting in Ireland has a very poor track record. A famous report circa 1990 forecast that net emigration would continue indefinitely at close to 1980 levels and that the population would fall to 3.3 million by 2010. Many of the infrastructure problems Ireland faced in the late 1990s and early 2000s (and to some extent still does) resulted from the fact almost all forecasts in the early 1990s were for substantial net emigration and negligible population growth, when in fact the reverse occurred.

A mistake in the fourth last paragraph of my previous post – it should be:

“This showed an increased of 31.0 thousand compared with 2007 Q4, …”

@John: That is very interesting that the figures you give are in such a big discrepancy — why is this? Any suggestions for the cause of the discrepancy?


No first-hand knowledge. I can only guess.

First, one would need to add sea transport figures to get a better overall picture of net passenger flows. I only gave air transport figures in my post because the opening post by Richard Tol was also dealing exclusively with air transport figures. I was trying to point out that air transport figures alone were not a reliable indicator of migration (as the figures for 1996 to 2008 showed).

Second, there is the border between N. Ireland and R. Ireland. Remember all the figures both Richard and I gave were for the Republic only. Figures are available for bus and rail-transport across the border with N. Ireland, but not for private car transport, which is the vast majority. I’d say 99.99 per cent of cross-border movement in Ireland is unrecorded because of this.

So, say an American family of six come on holiday to Europe. They first fly to Scotland and spend a week there. They they get the boat to Belfast and spend a few days there. Then their cousin, Billy, in Belfast drives them to Dublin. They spend a week in the Republic and fly back to America from Shannon. There would be no record of their entry into the Republic, so they would count as a net outflow of six from the Republic. Likewise, Irish rugby fans from Belfast, who get a lift to Dublin airport, fly from Dublin to Cardiff to see Ireland win the Grand Slam, but fly back direct to Belfast. All counted as a net outflow from the Republic. These are just examples, but they illustrate the point that, with a border where only a tiny proportion of movement across it is recorded, figures for air and sea transport alone are not reliable migration indicators.

Third, a purely statistical point. We are dealing with huge flows. In 2008 there were approximately 15 million air passengers into the Republic and approximately 15 million air passengers out of the Republic. Then, a few more million for sea passengers. The difference between them is minute in comparison with the totals. In statistics generally, its allways risky to estimate some quantity by getting the difference between two vastly larger quantities. Even a minuscule error (and, in real life there always are recording errors) in measuring one of the quantities will completely destroy the accuracy of the value for the difference between them.

@John: Very good point about the forecasting errors on demographic projections. However, the Irish government itself has taken a giant leap of faith in population projections in “Regional Population Projections 2011-2026” published by the CSO in early-December 2008. It predicts that the population of Ireland will reach 5.7 million people by 2026 whilst the population of Dublin is expected to fall by 100,000.

It indicates the folly of flat-line assumptions in a stochastic process. In order words the asymptotic fallacy trap.

Im more interested in how, given that Dublin is an effective monopoly, the figures can be commercially sensitive. Who is going to use them against DAA? Weston airfield? Gormanstown?

Thanks for that. The discrepancy between the numbers is indeed worrying. Commercial airlines have a duty under IATA regulations to account for any passenger, so the raw data should be fine. That does not mean, of course, that the reported data is without errors.

The point I was making, though, was not about the exact numbers, although the precision with which I reported may have given that impression. If you look at air passenger flows, there is a sharp turn from immigration to emigration starting in summer 2008.

I look at the same thing for all passenger flows. There is also a sharp turn in bus and rail, although I would not put much faith in those numbers, and a slight turn in sea passengers.

All of which suggests that emigration had started already last summer.

@Antoin Daltún
I am sure that, should the DAA write to the CSO that these data are not commercially sensitive, the data can be released with a few weeks delay only. That would really help us understand migration flows almost in real time, which would be a real boon in times like these.

@Derek. I take your point but its not quite true.

First, the CSO isn’t the government. The CSO is composed (hopefully, I don’t actually know any of them) impartial statisticians from all parties and none. But, that’s just a minor point.

Second, more importantly, the CSO don’t actually do forecasts as such. What they do are statistical projections, which are often wrongly reported in the media as forecasts, but are actually just a range of estimates based on different assumptions, but without any claim by them to know which, if any, of the assumptions will turn out to be what actually occurs. Its a purely statistical exercise on their part, not a claim by them to know the future. Other organisations do make forecasts (e.g. ESRI, Banks, Stockbrokers etc), but not the CSO.

If you have a look at the document you referred to, you’ll see the CSO make different estimates depending on what happens to migration (M) and fertility (F). They label these M0, M1, M2 and F1, F2. They don’t claim to know which of M0, M1, M2 or F1, F2 will ocur. None of them might occur. They simply say that, IF its M0F1 that occurs, the population will be x in 2020, but IF its M1F2 that occurs, the population will be y in 2020, and so on. As I say, its purely a number-crunching effort on their part.

For example, no one, not even the CSO, knows what the level of migration will be between now and 2020 and no one knows what the fertility rate will be between now and 2020. So, to that extent, no one knows what the population will be in 2020. But, what CSO statisticians can do is calculate that, IF there is net immigration of 20,000 between now and 2020 and IF the fertility rate is 1.85 between now and 2020, then the population will be 5.2 million or whatever (I’m just making these figures up to illustrate) – but, IF there is net emigration of 20,000 between now and 2020 and IF the fertility rate is 1.65 between now and 2020, then the population will be 4.6 million or whatever. As I say, they do this purely statistical exercise for a range of assumptions. But, they don’t predict which of the assumptions, if any, will be what actually occurs. The media are quite wrong in presenting these types of statistical projections as forecasts. Although, as I say, other organisations do make forecasts.

@Richard Tol
As I understand it, it is airlines rather than the DAA which are sensitive about data publication. The DAA data ultimately comes from airlines which pay fees on the basis of the same numbers and should therefore be reliable. It should include fairly small and unidentified elements of transit traffic, such as US-UK via Ireland. Some of the CSO data, e.g. on residence of visitors, is based on quite small samples, which ought to reflect the spread of times that people travel, a high proportion at unsocial hours. It sometimes looks very odd, especially on a monthly basis. Some CSO data gets seasonally adjusted which probably obscures more than it illuminates.

@ John. Take your point re. range of possible outcomes by CSO and their statistical independence from the political birds of passage. It still didn’t stop the government from using similar projections though e.g. population >5 million by 2021 in NDP & Transport 21

In the course of my duties as a vampire I would ask to see the ESB bills for properties that were said to be vacant. A meter reading is also available from the ESB (VAT is charged so we can access it if Revenue wish) so pretending they had lost the bills was no avail … Sad to say, all seemed to have steady use of substantial amounts of power over the claimed vacant period. Most dwellings have electric power.
Recent claimed growth figures for PRC, China as some call it, are contradicted by their power figures and the reduction in power outages as the industrial use of power declined substantially. Oooops!

Power use is a good indicator for the number of people, although not as good as water use. While data on water use are very poor, data on power use are much better but inaccessible to third parties — and definitely not in a timely manner and at the level of detail needed to discern residential electricity use while correcting for such things as weather and sunlight.

Sorry, I was having so much fun, I forgot to make the point which may be obvious, but should be said:
The ESB figures will be clear on housing vacancies and indirectly on population. They should be asked for these as a measure of trends etc. Should keep the Aspergers contingent busy?

@ Richard
VAT is levied on power so theoretically, and in the case of other businesses, Revenue have used their powers to access the data. This is usually tied to a case by case basis, but the ESB is also state controlled, last time I looked, although their pay increases seem to suggest they think otherwise!
The data can be made available but may not be, justifications may abound but it would be sensitive and might be restricted to certain circles of a golden hue…..

The data is there, indeed, it just cannot be accessed by the ESRI for such purposes as estimating a demand function for electricity or counting the number of people.

Comments are closed.