Tuesday, December 8, 2009

Economics 08/12/2009: Irish businesses ICT use

Per CSO release yesterday: “in 2009, 95% of all enterprises had a computer connected to the internet while 66% had a website or homepage. Access to the internet using broadband remained high in 2009 with 84% of all enterprises having a broadband connection. High-speed DSL broadband was used by 45% of enterprises compared with 41% in 2008. The use of mobile broadband reached a level of 27% in 2009 compared with 24% in 2008. As a consequence there has been a decline in the use of lower-speed DSL broadband and other fixed connections (e.g. cable, leased line etc).”

Sounds good? Well, actually… More detailed data shows the following worrisome trends:

  • Use of computers has actually fallen from 98% of enterprises surveyed in 2008 to 97% in 2009. It has remained static at 98% in Manufacturing sector, fallen from 99% to 96% in Construction sector (possibly a function of inactive enterprises, or in the opposite direction – a surprising result if the survivourship bias applies to the sample). In Services sector there was a decline from 98% to 97% between 2008 and 2009. These results are rather strange. On one hand, if the sample included inactive firms, then one can expect the declines due to companies folding operations in Construction and lower value-added Services sectors. But if only actively trading firms were included, then this suggests that survivourship bias was actually selecting against the ICT-using firms for some unexplainable reason.
  • Interestingly, the proportion of firms with a written ICT strategy has increase overall from 20% in 2008 to 21% in 2009, and in no sub-sector was there a decline in proportion. Construction sector firms led here with an increase from 8% in 2008 to 11% in 2009, which suggests that CSO sample incorporates survivourship bias. And this is really bizarre – on one hand, sample selection clearly favoured surviving firms that have ICT policies, but on the other hand the same sample favoured survivor firms with lower penetration of ICT… Hmmm…
  • Proportion of firms using internet fell from 96% to 95% between 2008 and 2009. The declines were showing in all three broader sectors, with Construction firms registering the largest drop from 99% to 96%. This is again inconsistent with survivourship bias apparently present in the data. But even more strange were the results for the percentages of firms having their own websites. In Manufacturing, the number of firms with their own websites rose strongly from 72% a year ago to 77% in 2009. In Construction sector, surviving firms actually dramatically increased their websites presence from 48% to 58% despite having shown a decline in internet use in general. There was a decline in proportion of companies with their websites in Services sector – from 65% to 64%.
  • Overall use of e-services (interfacing either with the public sector or private sector clients) has declined across the board except for Manufacturing sector.
  • E-commerce is growing strongly with percentages of purchases and sales via e-commerce pathways as a share of total purchase costs and turnover, respectively, rose strongly between 2008 and 2009. But as in 2008, most of e-Commerce appears to be driven by purchases, not sales. And in volumes, e-Commerce has declined in line with overall economic activities. There is only tentative evidence that e-Commerce has taken up some of the traditional purchasing and sales activities share during this recession.
  • Another interesting and surprising feature of the data shows that enterprises with access to broadband have reduced their e-Commerce-based purchases from 60% to 54%, and also reduced their e-Commerce-based sales from 28% to 23%. Enterprises with no connection to broadband have lowered their purchases via e-Commerce vehicles from 29% to 24% and their sales from 14% to 9%. This seem to show that access to broadband does not result in more resilience to the recessionary contraction in enterprise activities. But, enterprises with broadband connection have retained their propensity to employ workers who e-work at 37%, while enterprises with no broadband connection have increased this share from 9% to 10%. Rising workforce mobility and flexibility for those with no broadband connection while static workforce mobility / flexibility for those with broadband connection? Clearly this can’t be happening…

5 comments:

Myles Duffy said...

The percentage of Irish households with internet access has increased from 57% in 2007 to 67% in 2009 and the percentage with broadband connectivity has increased from 31% in 2007 to 54% in 2009. 65% of EU-27 households have internet access while 56% have broadband connectivity. Myles

TrueEconomics said...

Agree, Myles, but I am focusing on businesses. Another issue to consider is the proportion of quasi-broadband <2MB and also the quality of broadband connection. My own service provider - Eircom - delivers nothing close to 2MB even though I paid for the premium package. Their technician came out on several occasions and was unable to fix the issue. Would I fall into 54% or 67% by connectivity?

Eamonn said...

This has to be a result of problems with their sample set. For niche surveys, the CSO do not employ proper statistical methods. For example for their rent reports they survey less than 30 landlords/agents.

Eamonn said...

Another problem could be the questionnaire method. In good times companies will overstate figures in questionnaires like investment plans, employee figures etc. in a recession they will understate. It would be a much more robust report if it was based on a live dataset, like the eircom subscription data.

Umthatguy said...

The number of companies using computers has gone down, but the number of companies with an ICT strategy has gone up. Internet usage has gone down, but the percentage of companies with websites has gone up.
The survivorship bias appears to be against companies using computers "by default" and in favour of those who made conscious decisions about how ICT was to be used - or ignored them as a distraction.
Does this make a plausible hypothesis? More to the point, does it say anything useful?