Load a loan tape and simulate the book one year at a time. Exposure, probability of default and loss given default are all random, costs are charged, capital is read off the simulated loss distribution, and RAROC comes out as a distribution you can cut by any column in the file.
1 · Loan tapeⓘ
One row per credit. CSV, semicolon or tab-separated, including a block pasted straight from Excel. The file stays in your browser.
✓Loan tape loaded into the model: Demo portfolio · 5,000 loans
Guessed from the headers. Balance is required; PD comes from a score or a PD column, LGD from a guarantee type or an LGD column.
Dimensions to report byⓘ
Every column switched on here becomes a breakdown in the results.
2 · Loan-level risk inputs
How each row gets its PD, its LGD and its cost of funds.
Probability of defaultⓘ
One-year transition probabilities between states. Each row is normalised to 1; the last state is default. PD of a state = its column to default after the horizon.ⓘ
Horizonⓘ years
from \ to
1
2
3
4
5
6
PD
1 · CURRENT
%
%
%
%
%
%
2.91%
2 · DELAY 1-30 DAYS
%
%
%
%
%
%
7.04%
3 · DELAY 31-60 DAYS
%
%
%
%
%
%
11.77%
4 · DELAY 61-90 DAYS
%
%
%
%
%
%
19.56%
5 · DELAY +90 DAYS
%
%
%
%
%
%
24.17%
6 · DEFAULTED
%
%
%
%
%
%
100.00%
Loss given defaultⓘ
The expected LGD of a PERT is (min + 4·most likely + max) / 6, not the most likely value.
Guarantee
Min
Most likely
Max
Mean
%
%
%
89.8%
%
%
%
27.5%
%
%
%
38.3%
%
%
%
45.8%
LGD for types not in the tableⓘ%
Cost of fundsⓘ
Currency
Rate
%
%
3 · Systematic risk by segment
The part of the risk that does not diversify away. Each segment’s factors scale every loan in it; their means are 1, so expected PD and LGD stay what the tape says.
Segment
loans
balance
PD
LGD
EAD multiplierⓘ
Parameters
PD volatility (CV)ⓘ
LGD volatility (sd)ⓘ
Beta(α, β) PD · LGDⓘ
CORPORATE
227
12.7M
5.46%
67.8%
CV %
%
%
1.1, 19.3 · 14.1, 6.7
SME
756
4.1M
8.63%
54.2%
CV %
%
%
1.5, 16.3 · 20.5, 17.3
MORTGAGE
633
3.6M
4.85%
27.5%
CV %
%
%
3.8, 73.7 · 15.0, 39.4
VEHICLES
865
1.6M
5.97%
38.3%
min % max %
%
%
3.7, 58.3 · 18.1, 29.1
CONSUMER
1,514
931.3K
7.32%
87.7%
min % max %
%
%
3.0, 37.9 · 37.0, 5.2
CREDIT CARD
1,005
272.7K
8.05%
89.8%
min % max %
%
%
2.5, 28.3 · 31.9, 3.6
Correlations (Spearman)ⓘ
Between the segment factors. Cross-segment coefficients are the within-segment ones times the macro coefficient, so the matrix is always consistent. Each coefficient is 0 or above: a negative one would let one factor hedge another, which a capital figure should not count on.
Correlation used by the simulation: independent
EAD
PD
LGD
EAD*
PD*
LGD*
EAD
1.00
0.00
0.00
0.00
0.00
0.00
PD
0.00
1.00
0.00
0.00
0.00
0.00
LGD
0.00
0.00
1.00
0.00
0.00
0.00
EAD* (other segment)
0.00
0.00
0.00
1.00
0.00
0.00
PD* (other segment)
0.00
0.00
0.00
0.00
1.00
0.00
LGD* (other segment)
0.00
0.00
0.00
0.00
0.00
1.00
Correlation is off: every factor is simulated independently (ρ = 0). Tick “Correlate factors” to see the coefficients the simulation would use.
4 · Income, costs and capital
Operating costs are charged per loan; capital is either simulated or a fixed ratio.