Testofcontrols
Itiscommonfortheengagementteamandclienttoholdapreliminarymeetingbeforeanauditbegins. Oneoftheitemsdiscussedatthismeetingiswhethertheclient'scontrolsystemsareeffective.Ifso, theauditorwillplantotestcontrols.Otherwise,theengagementapproachwillbepurelysubstantive. Controltestingismostlydoneforauditsofpubliccompanies.Publiccompanieshavedepartments dedicatedtotheimplementationandevaluationofcontrolsystems,suchasinternalaudit.Also,public companiesarelargeandprocessmanytransactions,makingitverydifficulttoachievesufficientdollar coverageforsubstantivepurposes.Controltestingisrarelydoneforownermanagedorprivate companiesbecauseasubstantiveauditapproachismoreefficientforsmallerclients. BeforetheauditorcantestmanualandITapplicationcontrols,theITauditgroupmusttestITgeneral controls.ITgeneralcontrolsincludelogicalandphysicalaccesscontrols,changemanagementcontrols andsecuritycontrols.Applicationandmanualcontrolsdependonenvironmentalcontrols,soifIT generalcontrolsareineffective,theauditorisunabletoplacerelianceonthesystems.Mostprivate companiesdonothaveITgeneralcontrolsystemssuitablefortheauditor'sreliance. Theauditortestscontrolsinthesamewayawalkthroughisperformed,byreviewingthehistoryof documentationthatconnectsthetransactiontotheclient'sgeneralledger.Normally,2436transactions arebetested.Twoorthreetransactionsareselectedfromeachmonthtoensurethatcontrolsoperated effectivelythroughouttheyear. Unlikesubstantivetesting,transactionsareselectedforcontroltestingatrandomregardlessofthedollar value.Also,anequalnumberoftransactionsareperformedtoverifyoccurrenceaswellas completeness.Occurrencetestinginvolvesselectingtransactionsfromthegeneralledgerandtracing themtosourcedocuments.Thiswastheapproachusedforthewalkthroughsinthepreviousarticle. Completenesstestinginvolvesselectingtransactionsfromsourcedocumentsandandtracingthemto thegeneralledger. Iftheauditorencountersanerrorincontroltesting,theywillfirstconsiderwhethertheerroris systematicorisolated.Systematicerrorsnormallyhaveapervasiveimpactonthecontrolsystem,so theauditorwillnotbeabletorelyoncontrolsunlessacompensatingcontrolisidentifiedandtested.If theerrorisisolated,theauditorwillexpandtheirtestingsampleby35transactions.Ifthelarger samplesizedoesnotyieldanyadditionalerrors,thecontrolsystemisconsideredeffective.
2009myCAsite.com