There were a number of trades for the NYM region earlier this week including one for the NYMX17 (Nov 2017) contract. While traders would love to see tight markets 4+ years forward (5 when the Nov ’18 contract is rolled out in 2 months) how can one expect narrow bid-ask spreads when so much about possible impacts on forward home prices (e.g. GSEs, interest rates, etc.) is up in the air?
One answer might be that traders are more comfortable with relative prices than absolute levels. While calendar and intercity spreads can each be used for some element of relative value discussions, with this blog I hope to illustrate how using the two separate paths together might help traders get even more comfortable in discussions on distant home prices.
First, here’s a recent graph 
with prices for each of the 11 expirations (from Nov ’13 to Nov ’17) on the NYM contract. The graph shows higher forward bids and offers over the next four years with sharper gains over the next few months, followed by some seasonal declines (May contracts capture the worst of seasonal factors), and mid-market prices reaching 209.1 by Nov 2017. A 209.1 price is equal to about a 22.3% premium over spot levels (not shown) and a 29.1% premium over the 161.94 value for Dec ’12 as shown in the graph on the right axis. Bears take notice. (Graphs for other contracts are available upon request).
While the 22.3% premium over spot compares a non-actionable mid-market number one can also view and trade calendar spreads to get a sense of the timing of price increases.
The table below shows both mid-market to mid-market levels and percentage differences (highlighted in green) as well as the calendar spread markets for the CUS and NYM contracts in all the Nov/ Nov markets.
As a reminder, a calendar spread is where a trader enters a level where they agree to simultaneously buy and sell two different expirations for the same regional contract. While any calendar permutations are possible (and some allow traders to express views on seasonality) I’m only showing the Nov/ Nov series as a) the November contracts tend to be more liquid, and b) using one-year differences reduces the seasonal impacts that might dominate other spreads (e.g. May vs. November).
Furthermore, recall that the Nov ’13/Nov ’14 NYM calendar spread quotes of -10.8/-8.0 means that the bidder is willing to buy the Nov ’13 contract (X13) 10.8 points below where he would sell the Nov ’14 contract (X14), while the seller would “go the other way” but at an 8 point spread. Since calendar spreads are quoted as the front contract relative to the back, and since forward price are higher, calendar spreads are quoted as a negative number today (as opposed to 2009).
The columns under the calendar bids and offers are my attempt to convert prices into percentage changes. As such, the -10.8 bid is a price difference of 6.2%, while the -8.0 offer is a percentage difference of 4.6%. I will defer you to past blogs (need to post link) that goes into the arguments that these percentage numbers might reflects traders views on HPA (home price appreciation) over the relevant timeframe. Suffice it to say now that one can quote and trade these percentage differences either year-by-year, or over longer periods (e.g. Nov ’13 vs. Nov ’17) to come up percentage differences. As such, traders can debate (via better bids, lower offers, trades) their views on these percentage differences as time elapses.
While these one-year calendar spreads may be useful for viewing percent differences for one year intervals (implied HPA), and while these one-year spreads could be linked to create longer term percent changes, a better way to frame the debate about longer-term price changes is in the longer calendar spreads. 
(As an aside, the bid and ask for longer term contracts should be tighter than the sum of the annual calendar spreads; e.g. the -39.0 bid for then NYMX13/17 calendar spread is less than the sum (-45.0) of the annual spreads.)
For example, the table below shows the X13/X17 calendar spread markets and some analysis. (Again, I’ve focused on the Nov/ Nov spreads to avoid seasonal issues and to make the math of discounting longer-term percent differences into annual implied HPA.)
Similar to the analysis above once can convert the dollar spreads in the calendar spreads into outright and annualized percent changes.
The problem with distant contracts is that small HPA differences get compounded over time. For example the 1.8% implied difference in annual HPAs (for the CUS contract) translates over four years into a 15 point bid/ask differences in the four-year forward contracts. These wider dollar spreads are seen as an impediment to market liquidity (even though percentage HPA are small).
Bringing in intercity spreads to the analysis can sometimes help further narrow the bid-ask 
spread discussion.
The table to the right show one way of framing the discussion between a region, in this case NYM, and the CUS/HCI index. Similar to calendar spreads, an intercity (IC) spread is the level at which a trader is willing to simultaneously buy one contract while selling another. In this case the two trades are for the same expiration, but the regions differ. I’ve used NYM here as a) there was a recent trade, and b) since the NYM index represents ~27% of the CUS 10-city index there’s likely to be some strong correlation between it and the CUS index. (That said, one might expect even stronger correlations between like cities e.g. BOS v. NYM, or LAX v. SDG). (Like calendar spreads, all intercity trade permutations are allowable. Some just make more sense than others.)
The table is color-coordinated to help illustrate that intercity quotes (like calendar spreads) are quoted with the first contract relative to the second. So, one biding for the HCI/NYM X17 IC spread at 5.0 is committing to buy HCIX17 5 points higher than where they would sell the NYMX17 contract. (The seller would go the other way, in this example, at 9 points).
I’ve noted the percent change in various futures prices (index levels) relative to the spot Case Shiller index so that one can turn these numbers into a simple question “will the NYM index outperform the CUS/HCI index by Nov ’17, and if so, by how much?” (The answer from today’s quotes seems to be that NYM will mirror CUS performance. Until recently NYM was priced to underperform. The recent trade in NYMX17 seems to have re-focused that discussion).
There could be many ways to interpret this question (these numbers). One could take the above question directly and base their answer on a view toward foreclosure issues that apply to the NYM region (e.g. will inventory take longer to unload in judicial states). In buying NYM/selling CUS, one might alternatively express a view on the remaining portion of the CUS index (of which more than half is California) and express a view on its performance relative to the 10-city index.
Net, the NYMX17 market (of 207.0/213.2 today) has to be consistent with two views: a) that of HPA for NYM over the next 4+years, and b) that of the NYM index relative to the CUS/HCI index. A change in either view might result in a value outside the current range, resulting in a trade(!).
So, NYM traders, take a look at both your HPA assumptions for NYM and think about how you expect NYM to perform relative to CUS. If you disagree with the first there may be a calendar trade for you to do. If you disagree with the second an intercity trade. The tension of both sets of analysis, combined with the recent trade, might serve to keep the NYMX17 bid/ask spread , tight and lead to more trades.
This blog covers a lot of material, some of which is covered in more detail in past blogs. If you have any questions, or would like to discuss how this analysis might apply to other contracts, please feel free to contact me at johnhdolan@homepricefutures.com
for loosely defined geographic regions to try and estimate what the new weights might look like and how those changes might impact CUS values. I made the simplistic leap of faith that population growth is correlated to housing stock, which I believe, will be the denominator for assigning weights by region. The analysis is not intended to be precise but to illustrate some key moving parts in how a change in weighting might impact (if at all) the CUS index.
