Censuses and the development of official statistics for Jersey

STATES OF JERSEY

r

CENSUSES AND THE DEVELOPMENT OF OFFICIAL STATISTICS FOR JERSEY

Lodged au Greffe on 29th November 2005 by the Policy and Resources Committee

STATES GREFFE

PROPOSITION

THE STATES are asked to decide whether they are of opinion

to r e fer to their Act dated 6th July 2005, in which they agreed that a census should be held in 2006, to

rescind that decision and to agree that the next census should be held in 2011.

POLICY AND RESOURCES COMMITTEE

REPORT

1. I n t roduction

1. In bringing this Reportand Proposition to the States now, the Policy and ResourcesCommittee is fully aware that it cannotbe debated until 2006, when it will be for the Chief Minister and Council ofMinisters to decide whetherto proceed with the proposition. However, the Committeewishtoprovide a clear recommendationto the Council ofMinistersand a statementof its position to the House.TheCommittee also believes that, given the urgency for a decision tobemadeby the States in the light oftheextensive planning period requiredinordertoundertake a census, the early tabling of the Report and Proposition will allowthe full time as set out in the House's new Standing Orders should the Chief Minister decide to proceed with the proposition. It will also allow those interested in the topic to discuss it with the Statistics Unit and propose any amendments.
2. O v er the past 18 months Jersey hasseen a welcomeimprovementandexpansion in the amountof high quality data available to inform debate. We now have, for the first time, accurate data on thesizeof the economy and more timely data on the finance sector. At the same time, data are now more accessible via the Statistics Unit website and its new publications.
3. W  e also now have an annual estimate of the size of our population. By bringing together data on employment,schoolageandpre-school children from education and health (allofwhich are censuses in their ownright) and other data, the Statistics Unit now produces anannual estimate of the Island's resident population. This annualmeasure is asreliableas the census measure, because it is simply the change in the population (a numberofhundreds) that needstobe measured and not the whole number (some tens of thousands). Thus, far from hiding information aboutchanges in the sizeofthepopulation, wearenowmoreinformed than previously.
4. B u t we still needmoretimelyinformation, specifically on social issuesandon key economic factors such as trade, investmentand short-term indicators ofeconomic health. Theissue the Headof Statistics faces, therefore, is how best to use the finite professional statistical resourceswe have in order to further expand statistics.
5. R u nningeachcensuscostsaround half a million pounds.By increasing the interval between censuses to 10  years the States of Jersey could save itself that amountevery10 years. However, simply notdoing census isnot what isbeingproposed.Ratherbyreinvestingsomeof the moneyset aside forcensuses into more efficient waysof collecting and compiling data we will have more information, and moretimely updates ofcensus variables, and still achieve a saving totheStatesof £300,000.
6. It i s becausewe need todevelopmore statistics, have better informed debate and use ourresourcesmore efficiently that the Policy and Resources Committee fully supports andrecommends to the States the proposal from theHead of Statistics, supported by the independent Statistics Users Group (astatement from SUGisat the Annex attached), tomoveto a 10 year census cycle, withthe next census occurring in 2011.
7. T h e position held by the Committee is based on the independent advice of the Head of Statistics, supported bythe Statistics UsersGroup.Thisindependent advice will notchangesimply because wehave a different structure to our government.

2. B a c kground

1. C e nsuses provide, and will continue to provide, a very useful benchmark covering a smallnumberof topics. Butcensuses, even at a frequencyof 5  years, cannotprovide data that can quickly identifykey changes in our society oreconomy.Todo this we need more timely andmorecomprehensive data. For example via a censuswe can knowhowmanycars are owned in Jersey, but through a moredetailed annual surveywe can understand howandwhen they areused and what factors influence their use, to help improve our transportplanning. Similarly a census will tell ushowmanypeople live in a Parish but

an annual survey can inform us about issues that affect them including the determinants of their health, the use

they make of public facilities and their views on public services, all of which mean we are better informed to make policy.

2. T h atiswhy the moveto10 year censusisneeded,asit will enable the re-prioritisation of professional statistical  resources towards the  development of new annual surveys, greater  use of existing administrative data and an enhanced ability to undertake analysis with the data collected, including greater utilisation of census data, than is currently possible.
3. M  oving to 10 yearlycensuses means that wewill have more information, notless.Such a change will allow the development and annual production ofnew data sources such asthe Jersey AnnualSocial Survey (JASS) and a newBusiness Inquiry. JASS will cover the keyvariablesincludedin the census on the demographicmake-upofthepopulation (including age, placeof birth, and length ofresidency), economic activity and household structure. A Business Inquiry will provide vital data on trade and investment. Aswefacechange driven by recently adopted policies, including theEconomicGrowthPlan, Fiscal  reform  and  Migration,  the  advantage of having  more  information  on an annual basis  is a compelling reason whynow is the time to make the change to a 10 yearly censuscycle.
4. F o llowing the decisiontakenby the States on 6th July 2005, the Statistics Unit held a briefing meeting for all States Memberson 3rd October2005to discuss thefrequencyofcensuses and the need todevelop statistics. At this meeting itwasalso explained whyitwas not now possible torun a census in 2006, the earliest feasible date being spring 2007which,given the need and desire to run a censusin 2011, would not beanoptimal use of resources. Themeetingwas unfortunately only attendedby 8 States Members, but did include those whohadvoted in favour and against the proposition to run a census in 2006. However, no-one present raised any arguments against the case as set outby the Head of Statistics. The case, which the Policy and Resources Committee believes provides a compelling casetomove to 10 year censuses, isoutlinedbelow.

3. F r e quencyofcensusesand the developmentof statistics

1. In considering the frequencyofcensuses it is importanttorememberhow the provisionof statistics has developed in Jersey. Looking back 20 years orso, there is little doubt that,given the lackofother sources of information, a 5  yearlycensuswas vital to have even the smallest insight into social issues. But, through better use of technology, utilising administrative data and the application ofsound statistical techniques, there are now better meansofprovidingcomprehensive data than relying onexpensiveand resource intensive censuses.
2. T h e developmentofsurveymethodology over the past few decades has been one of the biggest changes in the field of statistics. Ithasmeant that rather than havingtoask everyone the same question, it is possible to getaccurate results byasking a sampleofthe population. Samplesurveys have theadvantage of being easier and quicker to run and can leadto results in a much shorter spaceoftime.Forexample, the Statistics Unit plan to publish results of the 2005JASS survey in February 2006(4 months after running the survey), compared to the last census report being published over 18  monthsafter census day. Moving to 10 yearcensuses will allow resourcesto be better focussed into samplesurveys, to provide the timely information that is required.
3. O f coursethe results of a survey are onlyas good as the sample from which they arederived; this is why the Island has a professional teamof statisticians to undertake such work. By running surveys such as the Household Expenditure Survey the Statistics Unit has developed a rangeoftoolsandapproacheswhich mean that hard to reach elements ofourcommunity, such as live-infarmand hotel workers, have been included. Survey results will thereforebe more comprehensive and future surveys even more inclusive.
4. T h e central argument in proposing that wemove to 10 year censusis to ensure weare better informed; this means ensuring that ourprofessional statisticians can provide us with the information that weneed. The SocialSurveyisprovingto be a great success with a responserate of over 50%; a ratewhich the

Committee is informed is very good for a voluntary postal survey. But the real value from the Social Survey

comes from the fact that it can be repeated: to update topics covered this year (for example, providing us with a full internationally comparable measurement of unemployment and economic activity annually rather  that  every  5  years);  to  cover  new  topics;  and,  crucially,  that  the  data  collected  are  used  for cross-analyses so as to better understand the effect of policy intervention and of policy interactions. If the Statistics Unit were to run a census prior to 2011 such work would not be possible and the Island will remain in the situation where, too often, decisions are taken in an information vacuum.

5. C o upled to the developmentof social statistics bysurvey will be a parallel initiative tomakemoreuseof the wealth of administrative data already existing in the Island. As indicated previously, health and education data already play animportant role in the annualestimateof the total residentpopulation, whilst  aggregate  data  held  by the  Employment and Social  Security Department can be used for monitoring structural changesinage and genderprofilesand of migration flows.
6. W  hilst muchof the focusof the debateabout the frequency ofcensuses is focused onsocialdata, it is vital to acknowledge that Jersey still needs far more economic data if we are to really understand developments andchanges in our economy. We are a trading island but haveno trade data;weknow next to nothing about investment; and wehaveno information onshort-termeconomicvariables such as retail sales.
7. F r om the  perspective of economic  statistics,  moving to a  10 year  census  cycle will  also  enable professional statistical resources tobe devoted to addressing current gaps in this area.Lastyear, the Statistics Unitdeveloped a methodology,basedon international standards, for determining the magnitude of economic activity in Jersey, overall and by sector. The next step in this regard willbe the introduction of a Business Inquiry to facilitate a deeperunderstanding of the Island'seconomy and the developmentof a morecomprehensiveframeworkof National  Accounts.
8. T h e developmentofeconomic statistics is the mechanism for the ongoing monitoring and evaluation of the policiesfor economic growthencapsulated by the Strategic PlananddetailedintheEconomicGrowth Plan.
9. O n e way of meeting the need for additional and more timely data would be to devote additional resources to statistics. However,since the debate of 6th July 2005 no-one has put forward a suggestionasto from where any additional resources shouldcome. Therefore, we have toface the fact that in ordertoimprove our understandingof the issues described abovewe need to re-prioritiseour main statistical resource, i.e. the statisticians weemploy,and allow them tomove to a 10 year census cycle and to usethe resulting freed resources to develop new sources ofinformation.

4. P r a cticalities ofundertaking a census

1. R u nning a census is not a simpletask.Itisnot merely a caseof taking the form used last time andre- issuing it. Following such anapproachwithout proper planning wouldsimply result in a very incomplete picture and half a million poundsbeing wasted.
2. R u nning a census means collecting information from everyone in the Island. Toachieve this requires accurate  information  on where everyone  lives,  forms  which are clear  and easily  understood by all members of  our population, which will  include translations,  a  large  team of around  250 trained enumerators,as well as appropriate IT infrastructure to capture and analyse the data. Inshort,ittakes around 18 months to plan and run an effective census.
3. A n other requirement, especially due to the seasonal nature ofJersey'semployment and thus population, is that censusesare run each time atabout the same time of the year, normally inMarch.Starting in July 2005 it would not have been possible to run a census in March2006.The earliest a census cannowberun is March 2007. However, given theneed to run a census in 2011, whenthere will be a needto provide an updated benchmark from 2001 and, furthermore, it will be when the UK,Guernsey,otherEUcountries and most otherOECD countries will be running censuses, it is hard to seehow a 2007 census will be

anything other than a non-optimal use of money and resources. Indeed running a census in 2007 will simply mean

resources will not be available to produce the timely economic and social data that we need.

5. R  ec ommendations

1. T h e Policy and ResourcesCommitteerecommends that the next Jersey Censusshould take place in 2011. This move will not result inany loss ofaccuracy in the recordingofpopulation statistics. However,it will allow statistical resources to be devoted to gathering more meaningful and up-to-date information which will improveourunderstandingoffundamental social and economic issues. ConductingthenextJersey Census in 2011, coupled with newannualsurveys,makesrealthe principle ofrobust data providing the basis for evidence-based policy development.
2. C o nducting a census in 2006 is now nolongerpossibledueto the substantial amountofwork that is required to plan andundertake a census, whilst a censusin2007 will result inthenew initiatives ceasing and hence will result inthe loss of the timely information necessary for informedpolicymaking.The Policy andResourcesCommittee, therefore, fully supports and recommends the proposition to conduct a Censusin2011.

6. F i n ancialandmanpower implications

1. R u nning a census in both 2007 and in 2011 will require additional fundingof £500,000over the period 2010 to2012.Bynot running a census until 2011,but refocusing £200,000on statistical work over 5  years,toensure that many census variables are updated annually and that the totality of information available is vastly improved, will result in a real saving to the States of £300,000. This amountcould then be carried forward togo a long way to meeting the costsof the 2011 census and thus substantially reduce the need foradditionalfunding in the future. The latter proposal was approved as part ofthe Policy and Resources carry forward by the  Finance and Economics Committee on 14th April 2005.  However, following the States decisionof 6th July 2005 this agreement has not been implemented andnoneof the money allocated to the 2006 censushas been used for other statistical work.
2. R u nning a censusrequiresaround250 staff. Around230 enumerators and supervisors would beemployed for around a month,18office staff for a periodof 3 to 4  monthsand 2 full-time census officers for a period of18 monthsto 2 years.

ANNEX August 2005

STATISTICS USER GROUP Statement on the Proposed Census

The Statistics User Group (the independent advisory group on official statistics) wishes to express its concern about the recent decision of the States to proceed with a five-yearly census against the recommendation of the States of Jersey Statistics Unit which is to move to a ten-yearly cycle for censuses. This matter has been considered in some detail by the User Group.

By way of background the Statistics Unit, like all States departments, undertook an exercise earlier this year to identify possible areas for efficiency improvements in the context of tight budgetary controls and increasing demands from users for new series and for more comprehensive and timely data. It identified the census planned for 2006 as a potential saving. The cost of running the census is around £500,000 and nowadays much of the information it provides can be derived from less onerous surveys – indeed, such surveys can provide more detailed and relevant information than the census itself. Because of their universal scope, census responses inevitably provide relatively low-level information rather than sophisticated data. Targeted surveys can also be conducted more frequently, so that socio-economic trend changes (for example in population profile) and other developments can be identified swiftly.

In particular, to supplement the information yielded by the census, the Statistics Unit has developed an annual household survey and a bi-annual business survey. The data derived from these surveys will be of major importance to States departments as well as to users in the business community, but the funding is dependant on moving to ten-yearly from five-yearly censuses and thus freeing up the resources previously committed to collecting and collating census data.

Ten-yearly censuses are now the norm in almost all jurisdictions. Few run five-yearly censuses because of the heavy resource commitment and the corresponding diversion from other work, and because nowadays the relevant information can be extrapolated from other sources. The Statistics User Group represents a broad range of business and other interests committed to achieving the best possible range and quality of statistics within the resources available. Having considered the various arguments, the User Group unanimously agreed at its 5th April meeting to back  the Statistics  Unit  proposal (minutes of this meeting can  be  found on the  Unit's  website www.gov.je/statistics). At a further meeting on 26th July the members expressed their unanimous concern about the States decision to hold a census next year, and its potential impact on the resourcing of other statistical work.

We therefore urge the States to reconsider its position on the timing of the next census. We understand that in any case it would not now be possible because of other commitments to run the census until 2007 at the earliest but even that would cause considerable disruption to the planned work of the Statistics Unit. We therefore strongly support the proposed deferral of the census to 2011.