Why are there different numbers of gene_id and exon_id in a txdb database from UCSC?

[Dbuch]

I'm trying to run DEXseq but am having trouble with the newExonCountSet function because my gene_id list and exon_id list are different sizes. When I unlist the mm10 gene_id dataset from UCSC annotated with Ensembl Genes (ensGene) it yields 351,862 gene_ids even though the corresponding exon_id dataset contains only 350,630 IDs. I had a similar problem when trying to annotate the UCSC database with RefSeq Genes (tablename = 'refGene'). Any thoughts on why there is a difference and how I can get the Gene_ID and Exon_ID datasets to match up? Thanks!

library("GenomicRanges")
library("Rsamtools")
library("GenomicFeatures")
library("rtracklayer")

txdb <- makeTranscriptDbFromUCSC(genome="mm10", tablename = "ensGene")

#Getting rid of strand information
exonRangesList <- exons(txdb)
strand(exonRangesList) <- '*'
exonRangesNoStrand <- split(exonRangesList)

ExonIDs <- exons(txdb, vals=NULL, columns=c('gene_id', 'exon_id'))
GeneIDList <- elementMetadata(ExonIDs)
GeneIDVector <- GeneIDList$gene_id
ExonIDVector <- GeneIDList$exon_id
GeneIDVector2 <-unlist(GeneIDVector)

[Dbuch]

Update: When I make the txdb using the code below (getting the data from Ensembl instead of UCSC) you get equal numbers of gene_id and exon_id.

 txdb <- makeTranscriptDbFromBiomart(biomart="ensembl", dataset="mmusculus_gene_ensembl")