Dreary what SS is saying that there are three types of spreads: nominal, Z, and OAS. Don’t confuse nominal with Z or use them interchangeaby (which it seems like you did)
Nominal: nobody really pays attention to in this case. Nominal is just YTM - benchmark yield. It doesn’t tell us much. YTM just gives us a single yied rate
Z spread - we must assume that each cash flow is discounted at its appropriate spot rate. If we use a binomial tree, we are assuming some volatility around those benchmark rates. However, if we discount a bond at the observed spot rates with the benchmark spot rates (or using the binomial tree with forward rates), then we will probably come upon a price that is different than the current market price (ie, if treasury spot rates are the appropriate benchmark rates, these are most likely lower than the rerquired rates given that the risk of the bond > risk of treasury bond). Ie, we will come up with a higher price. We know that these rates can’t be appropriate, however, because the risk of a bond (before assuming options here) from Ford is much different than a Treasury bond. So, we need to add some amount to each of those rates to make sure that the market price is equal to the price that our binomial tree spits out. How much do we add to each rate? The z spread.
Imagine FORD issues a noncallable bond. And we discount each cash flow at the benchmark rates, to come up with a price of 120. But, the market price is more like 102, so we should probably adjust those benchmark rates up to a level that will produce the market price. The z spread is a constant spread added to all of the benchmark rates so that our model spits out 102 instead of 120. Higher rate = lower price.
=============================
OAS - with OAS, we are accounting for the fact that the CASH FLOW will change due to the embedded option. Ie, if in our tree rates fall to a certain point with a callable bond, the issuer will probably call the bond. In which case the bond price is limited by the par value (issuer will pay par to redeem). So, if our falling interest rate in our binomial tree is such that the price at a node becomes 115 when the rate falls below the call rate, we dont discount the value of 115 anymore because we expect the bond to be called at 100.
So, the CASH flow will be lower at that time period (remember, the embedded option has no bearing on the benchmark rates) - so now if we discount the cash flows using the benchmark rates, we will get a lower valuation, right?
Assume a similar example as above, except now FORD has a callable bond that is identical to the one they issued above. (Assume the volatility in this example is set high enough that the bond will be called in the down scenario). We can expect the cash flows to be lower if the bond is called, because we are expecting the price to be 115 in a down scenario, but in reality we will only get 100. So, the value will be lower than the first example, so now we don’t need to add as much spread to the benchmark rates for valuation because the decrease in cash flows is already reducing the value closer to the market price
the option cost is relevant when you are comparing two bonds with embedded options. If two bonds have the same OAS but one has a lower option cost (meaning value of noncallable bond = value of callable + option cost), take the one with the lower option cost.
OAS vs z spread is relevant when comparing a bond without an embedded option to a bond with an embedded option (I’m 95% sure)