Does Radian Guaranty have a liquidity problem?
A couple of weeks ago a Seeking Alpha article was published that highlighted some problems on the horizon for Radian Group (RDN). The article was excellent and it introduced me to the idea of liquidity risk at a mortgage insurance subsidiary. That led me into a much more detailed investigation of the Radian Guaranty insurance subsidiary, which I will discuss below.
The liquidity of an insurance sub
Before getting into the issues specific to Radian, let’s talk a bit about liquidity risk. For some reason liquidity is not at the forefront of discussion during conference calls and in brokerage reports on mortgage insurance companies. Questions and comments focus on risk to capital ratios and loan loss reserve methodologies, which, while providing important clues, do not in themselves allow you to conclude whether a company will have the cash available to pay the claims. The author of the SeekingAlpha article, Darren Oliver, suggested that this was because the mortgage insurance industry is not very well understood. This could be the case, I don’t know. I just find it surprising.
As a mortgage insurer, the bottom line is that you have the cash available to pay claims and that the regulator who watches over you believes that this is the case. Over time, the cash and short term investments on hand plus the premiums paid need to be enough to pay out claims made as well as operating expenses incurred. If there is a concern that the cash and future premiums will not be enough to cover the expected claims, the insurer will either be taken over by the regulator or put into run off.
Having the ability to pay claims out “eventually” is different than having the ability to pay claims promptly as they come in. While premiums trickle in every month, claims can be skewed disproportionately to the near term. This can lead to a situation where the cash immediately available is not sufficient to pay claims, even though over the long run there may be enough cash generated from premiums to meet all the claims over that time.
The argument that Darren Oliver makes is that something along these lines is going to happen at Radian Guaranty. While Radian Guaranty may have enough cash to pay claims in the long run, they will not have enough cash freed up and available to pay the claims that come due over the next couple of years. This shortfall will be caused by cash that is on-hand but is not immediately available because it is held in another insurance subsidiary, Radian Asset Assurance.
Radian Asset Assurance is a wholly owned subsidiary of Radian Guaranty. As a subsidiary, the capital at Radian Asset Assurance counts as capital available to Radian Guaranty. Thus, if one looks at Radian Guaranty claim paying resources, it includes cash and investments at Radian Asset Assurance.
But Radian Asset Assurance is also an insurance company. As an insurance company it has regulators that control the amount of cash it can release to its parent as a dividend. It’s exactly the same way that Radian Guaranty is regulated in the amount of capital it can dividend up to its parent, Radian Group. Even though there may be a lot of capital in the form of cash and investments on-hand at Radian Asset Assurance, getting that capital into the coffers of Radian Guaranty requires approvals of the insurance regulator.
Radian Asset Assurance has consistently been upstreaming capital to Radian Guaranty over the last 4 years. Below is a table of the capital transfers that have taken place.
While I expect that Radian Asset Assurance will continue to upstream capital going forward, the amounts are, at best, going to stay constant. And $40 million per year is not a lot of money in relation to the claims that Radian Guaranty will have to pay.
Therefore, the current cash and investments on the balance sheet of Radian Guaranty and the premiums received, need to be sufficient to pay claims over the next couple of years.
I really didn’t have a good idea of whether this would be the case. So I decided to create a spreadsheet that looks forward to the end of 2014, estimates the claims paid over that time offset by premiums collected, and comes up with a cash balance at RG in 2015.
Unfortunately, to come up with a forward looking model, I had to make a lot of assumptions. First I had to make assumptions with respect to the insurance business itself. I had to assume how much new insurance will be written (I extrapolated the Q3 numbers going forward), the margin on that business (I assumed 0.58% of primary insurance in force), and the run-off of insurance in force from the existing book of business (I used $3.8 million per quarter). All these assumptions were based on the historical trends.
Additionally, I had to make assumptions with respect to how the company wide business translates to the Radian Guaranty operating company. Radian discloses limited statutory data about each of their insurance subsidiaries. While it’s enough to come up with a gross picture of what’s going on at Radian Guaranty, it’s not enough to create a detailed model of how the RG operating unit will perform going forward. In order to do that, what is needed are some assumptions about how the company wide premiums and claims translate into the Radian Guaranty sub.
I used the historical data as a guide. In the last 4 available quarters of data (which is unfortunately only Q1 and Q2 2012 and Q1 and Q2 2011 because Radian doesn’t provide the quarterly statutory filings for prior years) are as follows:
The above table shows the premiums collected at Radian Guaranty as a percentage of insurance written within the Radian Group. Insurance written is the best proxy I have to premiums collected. In my projections I assumed 86% of insurance written would premiums collected at Radian Guaranty.
For claims data, for each of the 4 quarters where data is available I took the dollar value of claims paid at Radian Guaranty, assumed that the average amount paid per claim within Radian Group was the same as at Radian Guaranty, and from that backed out how many claims were incurred at Radian Guaranty. Since I have no real insight into the make up of claims at Radian Guaranty versus the other insurance subsidiaries this is the best I can do. After backing out the number of claims that were incurred at Radian Guaranty, I calculated the percentage of claims at Radian Guaranty versus at Radian Group.
In my forecast going forward I assumed that 88% of claims incurred by Radian Group would be attributable to Radian Guaranty.
As I mentioned, I’m going to assume that the amount paid per claim at Guaranty is the same as at Radian Group. Radian Group has paid the following amounts per claim each quarter for the last 6 quarters:
In my worst case scenario I assume that the amount paid per claim is $57,000. In my base case scenario I assume $52,000 per claim paid. The base case estimate would be consistent with what Radian has been paying on average over the last couple of quarters.
With respect to how premiums and claims at Radian Group translate to Radian Guaranty, I think that together my assumptions should be a little on the conservative side if anything. I am assuming the RG only receives 86% of the premiums, but is effectively paying 88% of the claims.
Next I had to come with the number of claims I expected to be paid at Radian Group. I broke up the claims into 3 buckets. First, to create a base claim rate forecast going forward I extrapolated the current claim paying trend. In the past year and a half claims have averaged 5.1% of delinquencies. In the worst case scenario I assumed that claims would average 6% of delinquencies going forward, while in the base case I assumed 5.5%.
Next I added claims that would result due to past denials coming back. Radian has said that servicers have 12 months to come up with the documentation on a denial. Therefore, to estimate denials on a quarterly basis, I simply added back a percentage of the denials from the quarter one year prior. In the worst case scenario I assumed 75% of denials would return as claims. In the base case I assumed 50%.
Finally, I wanted to account for the current elevated level of Radian’s unpaid claim bucket. This was a point made by Oliver. He noted the following:
The first fact to consider is that Radian has been slowing down claims. Because of this, “pending claims” are growing as a proportion of Radian’s reserves (see the quarterly slides). “Pending claims” almost certainly end up in a claim, which means Radian should have a higher loss reserve than MTG. We can adjust Radian’s reserve for comparison with MTG. Only 7.8% of MTG’s delinquent loans are in the pending claim category (see MTG quarterly release) vs. 17% for Radian. Radian should have already paid about 9,000 claims, which would result in a reserve per delinquency of $26,500 (and almost $500 million less cash, which would drain its mortgage insurance co to $1.7 million of cash and investments, which is dangerously low).
I accounted for this by adding a bucket called “Additional claims to bring unpaids down from 17% to 8.5%” (13% in the base case). This bucket represents the “extra” claims that Radian has perhaps been dragging its heels on and that are currently sitting as unpaid claims. In the worst case scenario I assume that the unpaid claim bucket is worked down to 8.5%, or a level consistent with MGIC. In the base case I assume the level is worked down to 13%, which is about where it was at during the first quarter of 2011. I assumed the claims would be worked off in 5 quarters ending in the third quarter 2013. My choice of 5 quarter is arbitrary; I did this so that after the end of 2013 I could see the first “normalized” quarter for the company, where they are simply paying off claims at anticipated rates, without any additional backlog due to elevated unpaid claims or denials.
Below are the results of both the worst case and the base case scenarios. In both cases I started off with the current cash and investments at Radian Guaranty ($2,274,820,287) and from there subtracted the cumulative cash flow out of the company due to claims being paid, expenses and less premiums collected over the period. I also add back $40 million to account for the likely capital addition from Radian Asset Assurance next year.
I limited my results to the end of 2013 because beyond that point there gets to be too many assumptions. However, as you can see from the actual spreadsheets posted at the end of the article, I did run the forecast out further, to the first quarter of 2015. I’ll discuss this in a second.
In both cases, Radian burns a lot of cash over the next year and a half. In the worst case scenario, it’s a whoooole lot of cash. Over $1.2 billion dollars in claims are paid by the end of 2013. This takes a huge chunk out of their cash reserves, which drop to a little over $1 billion dollars.
In my worst case scenario, I come up with results not that different than Darren Oliver. In his Seeking Alpha article he estimated that:
They are generating about $720mn in premiums per year and $680 million in pre-claim cash flow. If claims rise to $1.6 billion – as they should once the denial trick fades (they should actually temporarily rise higher as they “catch up” on delayed claims) – they will burn $900 million per year.
In the next four quarters, under the worst case scenario, I estimate that Radian will burn through a little over a billion dollars. So my analysis appears to be more pessimistic than his.
But the evil is in the details, and it’s not as bad as it appears. While Radian will have used up a big chunk of cash over the next year and a half, in doing so they will have gone a long way towards cleaning up their legacy book. They should not continue to lose money at this rate going forward.
At the end of 2013, in the worst case scenario, there would only be 64,000 delinquent loans remaining on their books. The number has been reduced because of all the claims that have been paid and because delinquencies coming in are falling. In addition, Radian will have brought their unpaid claim bucket down to the 8.5% level (so consistent with MGIC), and there will no longer be an overhang of denials. The result is that the claims the company will pay going forward will begin to decline substantially.
Meanwhile, as long as new insurance written continues to exceed the run-off of the existing book, cash flow into Radian Guaranty will continue to increase.
The result is that even with the worst case scenario, once Radian Guaranty gets to the normalized claims rate in Q4 2013, they begin to approach being cash flow positive. By this point Radian Guaranty is collecting almost enough from premiums to cover the claims and expenses. I don’t think the regulator will put Radian Guaranty into run-off with such an outlook.
In the base case scenario Radian Guaranty becomes cash flow positive on a consistent basis in 2014.
What to conclude
The point here is not so much the specific results that I have come up with. Whether or not Radian Guaranty will be at or near cash flow positive in 2013 or 2014 depends on a lot of things happening. The scenarios I have modeled basically assume that Radian gets all of the dirty laundry out right away and has a much cleaner insured loan book by Q4 2013. That may happen, or it may be extended over a longer time horizon. But if it is extended, that will be beneficial, not detrimental, to the cash flow position.
Rather, the point here is that even with the assumption that Radian Guaranty has to pay out a lot of cash in the coming 6 quarters, it is doubtful that the regulator will have to take action. The regulator will look at the claims and cash flow projections going forward and note that Radian Guaranty is moving towards a positive cash flow position. By 2014 Radian Guaranty will be generating positive cash in at least some quarters.
As an aside, while I haven’t gone through this sort of rigor with MGIC, the analysis of Radian gives me a lot of comfort with my position in MGIC. In fact, if MGIC continues to slide, I would consider buying more. As Oliver noted in his article, on a claims paying basis MGIC is actually in a better position than Radian in the short term. They have over $6 billion of cash and investments available to pay about 50% more claims. I don’t think the regulators are going to step in with MGIC even if MGIC has to pay a substantial amount to Freddie Mac in order to settle their dispute.
A final word on assumptions
This analysis is all about the assumptions that are being used. While my worst case scenario makes what I believe to be worst case assumptions about the existing delinquency and unpaid claim buckets at Radian, it is not making a worst case projection of how the housing market and the economy in general performs. I assumed that new delinquencies decline at 15% next year and the year after. I assume claims cure at the same rate as this year. If the economy does poorly, delinquencies rise and claims do not cure, then Radian Guaranty is going to be walking a fine line.
But in my opinion, that’s the risk it always has been. It’s the bet that I said it was originally, a bet on a bottoming of prices, a fall in delinquencies, an improvement in the US economy. What my analysis does help prove to me is that there is little risk of an internal implosion in the company due to a shortfall of liquidity. I just don’t see that happening.
I updated the spreadsheet to explicitly state what percentage of insurance in force was at RG as well as what percentage of rescissions and denials are denials. I had a comment that this wasn’t clear because I had embedded the multiple of each into the equation without explicitly saying where it was coming from.