RTransProb (version 0.1.0)

cohort.TTC: Cohort - Data Weighting and "TTC" Calculation

Description

Calculate Through-the-Cycle transition rates using raw transition counts and starting obligor counts using the cohort method.

Usage

cohort.TTC(transCount, initCount, gYear, gHorizon, transtype)

Arguments

transCount

monthly "raw" data transitions

initCount

monthly "raw" data counts

gYear

number of relevant years

gHorizon

number of months in the horizon

transtype

averaging method. The valid values

• averageDataTransAnnual - "Raw" Data Average

• averageMonthlyTransAnnual - Average of Monthly Transitions

• transAnnual - Average of Annual Transitions

• averageMonthlyTransAnnualGM - Geometric Average of Monthly Transitions

Value

One of the output options corresponding to the 'transtype'value selected as input.

averageDataTransAnnual

"Raw" Data Average - weighted average of separately scaled transition and obligor counts.

averageMonthlyTransAnnual

Average Monthly Transitions - compute monthly transition matrices then average over years, e.g., average January matrices, then February matrices,...

transAnnual

Average Annual Transitions - compute annual transition matrices then average over years

averageMonthlyTransAnnualGM

Geometric Average Monthly Transitions - similar to Average Monthly Transitions except the geometric average of the monthly transition matrices, on an element-byelement basis, is used

Details

Many credit risk models require a long-run average (Through-the-Cycle) PD estimate. This has been interpreted as meaning the data from multiple years should be combined and the method capable of supporting some form of weighting of samples.

The three methods of weighting considered for data generated via the cohort method are:

1. Scale the number of transitions and firm counts/years using the a single year count to preserve dynamics, then average transitions and firms counts/years separately

2. Estimate the single-year quantities (estimate with monthly transition matrices), then average across years

3. Average annual transition matrices

The Markov property allows for direct weighting as each year can be regarded as distinct.

Examples

# NOT RUN {
# }
# NOT RUN {
#Get a sequence of years from "2000-01-01" to "2005-01-01"
TotalDateRange <- seq(as.Date("2000-01-01"), as.Date("2005-01-01"), "mon")
TotalDateRange_Yr <- seq(as.Date("2000-01-01"), as.Date("2005-01-01"), "years")

#Set parameters
snapshots <- 12    #monthly transition matrices
interval <- 1/12    #1 month transition matrices

#define list to hold initial counts and transition counts
lstCnt <-rep(list(list()), length(TotalDateRange_Yr))
lstInit <-rep(list(list()), length(TotalDateRange_Yr))

#initialize counters
nn <- 1
k <- 1
kk <- 1

#Create transition and initial counts (start vector) used as inputs to the function
for (l in 1:(length(TotalDateRange)-1)){

  istartDate = POSIXTomatlab(as.POSIXlt(as.Date(TotalDateRange[l],format = "%Y-%m-%d")))
  iendDate = POSIXTomatlab(as.POSIXlt(as.Date(TotalDateRange[l+1],format = "%Y-%m-%d")))
  DateRange        <- as.Date(matlabToPOSIX(cfdates(istartDate,iendDate,snapshots)))

  for(i in 1:(length(DateRange)-1)){

    sDate  <- DateRange[i]
    eDate    <- DateRange[i+1]

    Sample1<-TransitionProb(data,sDate, eDate, 'cohort', snapshots, interval)

    lstCnt[[k]][[kk]] <- Sample1$sampleTotals$totalsMat
    lstInit[[k]][[kk]]  <- Sample1$sampleTotals$totalsVec
    #print(lstCnt)

    if(kk>=snapshots){
      kk <- 1
      k <- k+1
    } else {
      kk <- kk+1
    }

  }

}



#average of monthly transition matrices using the monthly initial portfolio
#counts  (start vector) and transition counts.

#remove empty elements from the lists (lstCnt and lstInit)
lstInit <- lstInit[lapply(lstInit,length)>0]
lstCnt <- lstCnt[lapply(lstCnt,length)>0]

transtype <- "AverageMonthlyTransitions"   #averaging method
gYear <- length(lapply(lstCnt,length))     #Add Error Checking for this
gHorizon <- snapshots                      #number of months in the horizon (default= 12)

transMatTest <- cohort.TTC(lstCnt,lstInit,gYear,gHorizon,transtype)
# }