Looking under the hood at US life insurers’ CML portfolios
- The U.S. life insurance industry has large exposure to commercial real estate, about two-thirds of which is through direct privately placed commercial mortgage loans (CMLs).
- Typical life insurance CMLs are extremely high quality and have experienced insignificant losses in the past.
- A minority of life insurers who specialize in higher risk loans or are overweight to retail or hotel CMLs may face problematic impairment losses or significantly increased capital charges.
- Insurance industry commercial mortgage lending is concentrated in major cities, which could lead to long-term problems with these loans if COVID-19 is a catalyst for increased remote work in the future.
- More conservative underwriting on new loans has created an opportunity for insurers to make relatively low leverage loans at attractive yields.
- We define a severe CML rating stress and show how this might impact RBC ratios.
1. Insurers and commercial real estate lending
The largest sources of commercial real estate financing include banks, the securitized market, alternative lenders (such as debt funds), U.S. government agencies and life insurers. The latter are the focus of this paper. Exposure to commercial real estate financing is obtained via a variety of asset classes, but in the U.S. life insurance industry the most significant sources are commercial mortgage-backed securities (CMBS), real estate investment trust (REIT) debt and commercial mortgage loans. Respectively, these account for 4.3%, 2.41% and 12.56% of total assets in the U.S. life insurance industry, as shown in Exhibit 1A.
Exhibit 1A: U.S. life insurance industry average asset weighted allocation to real estate debt
CMBS are securitizations of CMLs. Agency CMBS are bonds that are guaranteed by a government agency, while non-agency CMBS are not, and have credit risk to the borrower. Non-agency CMBS are typically tranched to generate securities that span a wider range of yield and risk. CMBS are typically groups of loans, but single asset single borrower (SASB) loans are common on very large properties. As shown in Exhibit 1A, the majority of U.S. life insurers’ CMBS holdings are agency, AAA or AA.
Unlike CMBS and REIT debt, CMLs do not trade regularly; thus, there are no market prices to indicate when a loan is likely to experience credit deterioration or impairment losses. This causes concern regarding potential losses that may not yet have surfaced. In this paper, we will discuss U.S. life insurers’ exposure to CMLs and attempt to identify their risks, which are heightened by COVID-19 and its potential impacts. For this reason primarily, the focus of this paper is CMLs. U.S. statutory insurance financials provide a great amount of detail on underwriting quality, geographic location, property type, interest rate and maturity for CMLs, and we use this information as the basis of our analysis.
CML allocations for U.S. life insurers
In this paper we separate life insurers into cohorts based on primary business, ownership structure and investment focus. These cohorts are summarized in Exhibit 1B.
Exhibit 1B: U.S. life insurance cohorts
CMLs are a significant asset class in the U.S. life insurance industry. The average asset weighted allocation to CMLs in the industry is 12.6% of total assets; however, this obscures a fairly wide variation in the usage of CMLs, as evidenced in Exhibit 1C. This exhibit displays the U.S. life insurance industry’s CML allocations as a percent of total assets (y-axis), and separates the allocations into percentiles (x-axis), further broken down by the cohorts described previously.
What is the U.S. life insurance industry?
In this paper, when we refer to the U.S. life insurance industry, we are including all insurers with more than $1 billion of statutory invested assets, as of year-end 2019. Companies with less than $1 billion of statutory assets were excluded because they typically have no or very small allocations to CMLs. This paper discusses the general account assets of U.S. statutory life insurance entities. We exclude non-U.S. insurance companies owned by U.S. insurers as well as assets in holding companies outside of regulated insurance companies. All separate account assets, such as those associated with variable annuities and certain pension risk transfer deals, are excluded from the statistics shown in this paper.
Exhibit 1C: U.S. life insurance industry CMLs as a percent of total assets
The outliers tend to be small companies with less than $10 billion in assets and non-traditional insurers. The former often lack in-house origination capabilities, and the latter, given their usual preference for higher spread assets, typically focus on higher yielding loans (including hotels, construction and mezzanine loans) when investing in CMLs. As seen in Exhibit 1C, most other large life insurers, whether mutual or stock companies, have similar CML allocations, and a minority of insurers have a small allocation (or none at all). The significant variation in CML investing that we observe across cohorts is in property types and degree of concentration in large cities.
Exhibit 1D: Rating categories for office, industrial, retail and multi-family CMLS
CML quality types and associated capital charges
Using the type of mortgage loan, debt service coverage ratio (DSCR) and loan-to-value (LTV) ratio as inputs, the rating for a mortgage loan is formulaic. In Exhibit 1D, we show how the NAIC ratings are determined for office, industrial, retail and multi-family CMLs1 . The rating, from CM1 to CM5, determines the regulatory capital charge.
Nearly 95% of CMLs held by U.S. life insurers are rated either CM1 or CM2. As a percentage of assets, the allocation to CM3 or lower CMLs is less than 1%. Additionally, CMLs in the life insurance industry have traditionally been very conservatively underwritten and, as a result, have had modest impairment losses. Because CMLs are private direct loans and not securitizations, historically, even when a borrower encountered difficulties, insurers (or asset managers working on behalf of insurers) were often able to restructure the loan. This could provide the underlying borrower with the ability to avoid foreclosure, and the insurer the ability to avoid impairment losses. Capital charges on non-performing loans, and those the insurer intends to dispose of, are very punitive. Thus, insurers tend to be reluctant to foreclose and are much more comfortable modifying the terms of the loan.
CML accounting and historical impairments
From an accounting perspective, CMLs are carried at amortized cost on statutory balance sheets unless an impairment is recognized. This insulates insurers’ balance sheets from short-term mark-to-market movements. The aftermath of the financial crisis is instructive for understanding the length of time that may pass before impairments are fully recognized on CMLs after a downturn. As shown in Exhibit 1E, during the five-year period from 2008 to 2012, impairments on CMLs averaged 26bps per year and total impairments were nearly $3.9 billion. Since 2015, annual industry impairments have been under 5bps per annum.
Exhibit 1E: U.S. life insurance cml holdings book value, yield, underwriting and impairments
As noted in our recent NAIC updates in response to COVID-19 paper, the NAIC has provided a limited time exemption from assessing impairments on mortgage loans (among other investments) due to short-term deferral or modifications of interest or principal payment in response to COVID-19. However, this does not delay the recognition of other than temporary impairments (OTTI) if the investment is sold or if the insurer identifies that it will not recover its carrying value.
Many commercial mortgage lenders are willing to work with troubled borrowers to avoid impairments. This, coupled with U.S. life insurers’ historically conservative underwriting on CMLs, book value accounting and temporary NAIC relief, will likely result in impairments and capital charge increases being deferred into 2021 or further, if they end up occurring at all. Prospects in the long term will ultimately be driven by the length and severity of the COVID-19–related downturn. At the end of this paper we provide an analysis on how CML ratings might evolve in a severe stress scenario.
2. Current CML market environment
At the start of the downturn in March, most lenders pulled back from the CML market, causing liquidity to evaporate. Banks and insurance companies alone account for a significant portion of outstanding commercial mortgage debt and, while insurers are starting to come back to the market, liquidity is still relatively scarce. Short term, collections have been trending lower in all sectors, and rent relief requests have been increasing. Long term, it is simply too early to assess what will happen in the market.
The lenders that have returned have done so with caution. However, it is a lender-friendly environment. Underwriting standards have tightened with reductions in leverage and stricter covenants.
Given the economic impacts of COVID-19, there are looming concerns that the CML portfolios of insurers are going to experience income interruptions, rating deterioration that will trigger increased capital requirements and impairments. These outcomes may be more or less acute, depending on portfolios’ sector allocations, LTV levels, and metropolitan statistical area (MSA) exposures. In the following sections, we discuss the U.S. life insurance industry’s exposure to these considerations and the possible implications of such.
Exhibit 2A displays how the different life insurance cohorts we previously defined have allocated their CML portfolios across various sectors. Life insurers with less than $10 billion in assets tend to overweight the small retail market; 27% of their CMLs are allocated to the retail sector. Non-traditional insurers are overweight the hotel and office sectors, while mutual companies are overweight the apartment/multi-family sector.
Exhibit 2A: CML allocations by cohort and sector as a percent of total CMLs
Exhibit 2B shows that insurers with less than $10 billion in assets and non-traditional insurers, on average, have smaller allocations to CMLs than the rest of the industry. In the event of a commercial real estate downturn, this smaller balance sheet allocation to CMLs may mitigate the associated capital impacts for these insurers.
Exhibit 2B: CML allocations by cohort and sector as a percent of total assets
On average, total exposures to the currently more problematic sectors of retail, hotel and office, are typically only around 5%-6% of assets. While this number may seem small, a typical life insurer has statutory invested assets between 8 and 14 times statutory surplus. At 8x leverage, 5% of assets is 40% of statutory surplus, and at 14x leverage, 5% of assets is 70% of statutory surplus. This leverage in life insurance company balance sheets explains why problems in asset classes with seemingly small allocations can still lead to large balance sheet issues. After this inflammatory sentence, we hasten to point to Exhibit 2C which shows that most of these loans were at less than 60% LTV before the downturn started and the probability of significant losses on most will be quite small. As a share of total CMLs, the non-traditional and small insurers both had larger allocations to CMLs with LTVs greater than 60%, but given lower allocations to CMLs in general, their overall balance sheet exposure is less than what other insurers have.
Exhibit 2C: CML allocations by cohort and LTV band as a percent of total assets
The retail sector has been hit particularly hard by COVID-19. The tenants that have been able to remain open are generally still paying rent, but many non-essential businesses have experienced prolonged closures or extreme decreases in revenue. Income from essential business may not be sufficient to support the properties as a whole. Large tenant essential businesses like grocery store anchors typically pay less per square foot and this could lead to problems for the property owners if the other higher margin tenants do not survive the crisis. There has been a sharp uptick in retail bankruptcies and many of these tenants are in malls. There is a concern that even what were considered above average malls before the downturn may not survive.
In a post-COVID-19 retail world, pent up demand may be a boon, but this will have to offset likely reductions in discretionary spending that are a consequence of the decrease in wealth and economic security that many consumers have felt. Should the latter effect be more significant, the retailers who had weak balance sheets before the crisis could encounter significant stress, resulting in lower ratings and impairments on retail CMLs.
Although at the cohort average level retail CML exposure is small, Exhibit 2D shows the distribution of allocations to retail CMLs. At the 90th percentile, the Public Life and Annuity cohort has 4.6% of total assets allocated to retail CMLs. The Annuity cohort contains only 10 insurers, so the very large allocation (13.2%) at the 90th percentile is driven by one particular insurer.
Exhibit 2D: U.S. Life insurance retail CMLs as a percent of total assets
Exhibit 2E shows that high LTV (60–70 and 70+) retail CMLs are disproportionately owned by non-traditional and small insurers. These are the loans most likely to experience impairment losses and rating deterioration.
Exhibit 2E: Retail CML allocations by cohort and LTV band as a percent of total retail CMLs
With occupancy rates falling off and lenders considering forbearance, the lodging/hotel sector is likely to be the most deeply impacted by COVID-19. Given the conservative nature of insurance companies, very few insurers make hotel loans even in good times. The median allocation amongst U.S. life insurers to hotel CMLs is 0% of total assets (though this industry median includes a lot of very small companies) and only about .5% of life insurance assets are in hotel CMLs (Exhibit 2F).
Exhibit 2F: U.S. life insurance lodging/hotel CMLs as a percent of total assets
The U.S. life insurers who have invested in hotel CMLs have underwritten conservatively. The insurer with the largest allocation to hotel CMLs with LTVs greater than 70 holds these exposures at less than 1% of total assets. Exhibit 2G shows that hotel loans tend to be underwritten at slightly more conservative LTVs than retail, but there are certainly hotel loans at greater than 70% LTV, which could cause insurers to have to moonlight as innkeepers in the near future.
Exhibit 2G: Lodging/hotel CML allocations by cohort and LTV band as a percent of total lodging/hotel CMLs
With rent being relatively high on the priority list of expenditures, the outlook for multi-family CMLs is more positive than retail and hotels. Life insurers also tend to lend against more upscale multifamily housing, where we infer that a greater number of tenants will be able to maintain employment via remote work. Additionally, the short commitment nature of the leases in the multi-family sector can be seen as an advantage in times of uncertainty. However, while collection rates have been higher than expected, they do fall as asset quality decreases. If unemployment claims are slow to be processed and the potential longer-term economic downturn does come to fruition, this sector is not completely protected from experiencing stress.
We note, based on Exhibit 2H, that smaller insurers have a significantly different LTV profile in the multi-family loan space than the other cohorts, as they hold a much larger percentage of loans that are greater than 70% LTV.
Exhibit 2H: Multi-family CML allocations by cohort and LTV band as a percent of total multi-family CMLs
In Exhibit 3C later in this document, we show that 46% of multi-family lending is done in the top 10 MSAs in the United States. This skew could prove problematic if COVID-19 results in a fundamental shift in demand away from densely populated areas or enables people with high paying jobs currently centered in large and expensive MSAs to work remotely. In urban areas, many new lease trade outs (current tenant leaves and is backfilled by a new tenant) are showing negative rental rate trends, meaning the new tenant is paying less than the previous tenant. Lastly, though on average insurers tend to lend against higher quality multi-family properties than the market as a whole, that is only on average. Some insurers will have exposure to multi-family properties where the tenants are more likely to experience income interruptions.
The industrial sector has also fared relatively well in terms of rent collections, and the importance of logistics based, especially last mile and infill properties, has become increasingly evident. Industrial properties also tend to be more sparsely populated, good for social distancing, and are reasonably feasible to modify for different use cases.
However, many industrial tenants are exposed to global trade. The combination of the trade war with China and the effects of pandemic-related shortages could create pain for some suppliers. Those associated with brick-and-mortar retailers may also experience stress trickling down from the retail sector.
While no one can predict exactly what the post-COVID-19 world may look like, for the insurance industry, one key question relates to the office sector. Remote work could lead to a reduction in demand for office space overall, or crucially for life insurers, a reduction in demand for office space in expensive large cities. The countervailing trend is the need to socially distance and spread out employees in office buildings. However, the latter is likely to only be temporary, while a shift to remote work could be permanent. There is consensus building that many companies will move toward flexible arrangements with employees coming to a centralized location two to three days a week. If this holds true, companies will need to plan to be able to accommodate peak occupancy on any given day.
Exhibit 2I shows that LTVs for office properties tend to be lower on average than other types of loans. However, smaller insurers and non-traditional insurers do have slightly more exposure to high LTV (>60%) loans. Smaller insurers also have less exposure to large cities.
Exhibit 2I: Office CML allocations by cohort and LTV band as a percent of total office CMLs
3. Key concerns for U.S. life insurers
There is a wide range of concerns about U.S. life insurers’ CML exposures in the current environment. Each month, a large number of loans need to be refinanced and insurers are still originating new loans. Traditionally, most life insurers have avoided some problematic exposures, such as loans to hotels, but have large allocations to potentially troubled sectors like office and retail. Large life insurers have historically heavily concentrated CMLs in the largest MSAs in the country, especially in the office sector.
Upcoming re-financings: Cause for concern or opportunity?
$29.8bn of life insurers’ current outstanding CMLs are due to refinance in 2021. Exhibit 3A shows the dollar amount of CMLs that U.S. life insurers hold that will need refinancing through 2024.
Exhibit 3A: Distribution of maturities for CML outstanding loan balances ($BN)
The impacts of COVID-19 have created uncertainty regarding the proper valuations for commercial real estate. We know anecdotally that as a consequence of this, life insurance commercial mortgage lenders are lending at lower LTVs than they have in the past. The cause for concern comes from insurers who have made loans on properties that may have declined in value. If borrowers are unable to refinance, insurers may be forced to modify existing loans or foreclose. However, maturing CMLs are less than 1% of the industry’s assets in any given year so this seems to be a manageable problem, though one that may generate some impairment losses.
On the positive side, refinancing needs will also create an opportunity for some insurers. In today’s market for new loans and refinancing, first lien lenders are often only willing to lend up to approximately 60% LTV for commercial loans and approximately 65% for multi-family loans, with some unwilling to surpass even lower LTV limits of about 55%. This has created an opportunity for mezzanine lenders. However, the LTV of the combined first lien and mezzanine loans is typically in the range of where borrowers used to obtain first lien financing from life insurers, so the overall underwriting may not be a concern for many well-capitalized insurers2. Because first lien lenders have reduced LTVs where they’re willing to lend, life insurers who are willing to provide mezzanine financing can do so at very attractive yields. In Exhibit 3B we show the distribution of LTVs for loans originated in 2019. Nine percent of loans held had LTVs greater than 70% and nearly 40% had LTVs greater than 60%.
Exhibit 3B: Distribution of LTVs on new 2019 loans held by life insurers
In addition to the positive market dynamic, life insurance regulatory capital is very favorable toward commercial real estate mezzanine lending. This is true for separately managed accounts held on Schedule B as well as funds on Schedule BA, which may be structured to receive beneficial capital treatment on pools of mezzanine loans.
Exposure to the largest MSAs in the United States
In Exhibit 3C we focus on exposures to the largest MSAs3 in the United States. The bottom right corner of the table shows that 46% of the industry’s CMLs are located in the top 10 MSAs. All insurer cohorts have large allocations to the top 10 MSAs, though those with assets less than $10bn have a much smaller allocation (only 26%). We previously noted that smaller insurers tend to specialize in retail loans outside of major MSAs. This is clear in Exhibit 3C, which shows that only 20% of retail CMLs held by insurers with less than $10 billion in assets are in the top 10 MSAs. Another striking observation is that 61% of office lending (by loan value) is found in the top 10 MSAs, and for mutual insurers specifically it is 70%. Life insurers could be seriously challenged if there is a structural change in the need for office space in large MSAs.
Exhibit 3C: Percent of CMLs in industry’s top 10 MSAs, shown by insurer cohort and property type
The concentration in the top MSAs is driven by the need for large loans, but also by the fact that most of the 3rd party CML managers for the life insurance industry are large life insurers themselves and focus on lending in major cities. The reliance of life insurers on very large MSAs for nearly half of their CMLs is starkly different from the typical philosophy around corporate bond portfolio construction, where life insurers often impose issuer caps much lower than market benchmarks and have broadly diversified portfolios with very little idiosyncratic risk.
Severe stress test
To the extent that CMLs become impaired, the impairments will likely emerge slowly, as was the case after the financial crisis. However, if the capital relief described in Section 1 is rescinded, ratings for CMLs will be based on revised DSCR4 and LTV metrics. This is extremely unlikely to happen in the near term, as there are already advanced talks to extend the relief and it doesn’t seem like there is any appetite to rescind the relief as long as the COVID crisis is ongoing. Nonetheless, we have created a severe CML stress test, as described below, which reflects the capital impacts from extremely large reductions in DSCR and modest reductions in LTV.
Our assumptions for the severe stress test were as follows:
1. For all non-hotel and non-retail loans:
- Increase RBC by 10% for loans with LTV <60%
- Increase RBC by 30% for loans with LTV between 60% and 70%
- Increase RBC by 50% for loans with LTV greater than 70%
2. Assume all hotel loans go to CM5
3. Assume retail goes to a 50/50 blend of CM3/CM4
In Exhibit 3D we show the impact of our stress scenarios for life insurers in our analysis (some outliers are not shown in the plot). For most insurers, average RBC for CMLs increases by about 50%.
If the stress scenario we highlighted above actually took place, we would see a wide range of impacts on life insurance RBC ratios, but again we emphasize this is a hypothetical scenario showing what could happen without COVID related capital relief.
Exhibit 3D: Average CML RBC under the stress test vs. Starting CML RBC for various life insurers
How COVID-19 will impact the CMLs held by insurers is highly uncertain and will depend on the extent of the downturn and whether it leads to transformational changes in the economy. In the near term, those insurers that may experience significant impairment losses will likely not be the insurers with the largest allocations to CMLs, but those that have higher risk loans, such as mezzanine and high LTV first lien loans, or exposure to the most troubled sectors including hotels and non-essential retail. In the longer term, the potential for impairment losses on high value office properties in large cities and multi-family is certainly a looming concern, but given the modest LTVs the industry as a whole has underwritten to, it is likely that only a few aggressive insurers will be severely impacted, but the industry as a whole will be fine. As we mentioned previously, we see opportunities for yield enhancement and improved risk diversification. To enhance yield going forward, well-capitalized insurers can take advantage of the attractive yields and capital efficient structures in commercial real estate mezzanine lending. To reduce geographic concentration risk, insurers should consider diversifying by reducing exposure to large cities.
1 Hotels and specialty commercial properties require more conservative underwriting to achieve the same rating categories.
2 Though the mezzanine pieces of course have much more embedded leverage than if the financing were done as a single first lien loan.
3 These top 10 MSAs are 1. New York-Newark-Jersey City, NY-NJ-PA. 2. Los AngelesLong Beach-Anaheim, CA. 3. Washington-Arlington-Alexandria, DC-VA-MD-WV. 4. Chicago-Naperville-Elgin, IL-IN-WI. 5. Seattle-Tacoma-Bellevue, WA. 6. San Francisco-Oakland-Hayward, CA. 7. Dallas-Fort Worth-Arlington, TX. 8. Boston-Cambridge-Newton, MA-NH. 9. Houston-The Woodlands-Sugar Land, TX. 10. Miami-Fort Lauderdale-West Palm Beach, FL.
4 Even though the net operating income (NOI) used for this calculation is based on a rolling three-year weighted average, the most recent year is given a 50% weight. The large reductions in NOI for many properties in 2020 and potentially in 20121 would results in substantial declines in DSCR.
For more information
Contact your J.P. Morgan Asset Management Client Advisor or email to discuss real estate mezzanine lending strategies structured to be capital-efficient for insurers and other CML strategies that focus on MSAs outside of the top 10.