The sudden recession and its impact on credit quality has been unprecedented in speed and depth--a deterioration not triggered by an asset pricing bubble or fundamental economic factors, but by a global health pandemic. The impact of COVID-19 containment measures, plus disruptions in the oil markets, has caused a sharp drop in revenues and earnings for companies in consumer-facing sectors, and the resulting recession has affected companies across the board. This has hurt creditworthiness, as evidenced by the high levels of corporate downgrades; 11% of corporate issuers are now rated in the 'CCC' category as of July 14, 2020, compared to 5% at the same time in 2019, while corporate defaults have reached 83 in the first half of 2020 compared to 78 for the entire 2019 year and we anticipate more defaults to follow. S&P Global Ratings' baseline forecast for the forward 12-month (through June 2021) U.S. speculative-grade default rates is expected to be 12.5%.

From a recovery standpoint, we do not expect the COVID-19 pandemic to have an immediate impact on recovery ratings as our recovery analysis looks through distressed paths to default, although there are circumstances where we do revise recovery ratings, including secular shifts and incurrences of additional debt. The market has also continued to raise concerns about issues from covenant-lite structures to more involved factors such as flexibility in credit agreements (e.g. being primed by other senior lenders, the ability to transfer assets out of a company's collateral group into unrestricted subsidiaries and to raise capital against that) and how this affects ultimate recoveries if and when a default eventually occurs.

In this study, we looked at historical U.S. corporate debt recoveries from January 2008--the beginning of the financial crisis--to June 2020--the midst of the current COVID-19 crisis. To start, we focused on companies that emerged from bankruptcy in the last 12.5 years (from 2008 to the six months ended June 30, 2020), the past 2.5-year period since our last recovery study (which covered up to the end of 2017). Then, we reviewed our recovery ratings on the defaulted debt to gauge recoveries and examine how they compare to actual realized recoveries, which are based on information from bankruptcy documents (see the Appendix for how we calculated actual recoveries for this analysis). Our lookback also considers the key factors that affected recoveries and how different debt classes fared from an absolute recovery standpoint.

Since we published our 10-year study in February 2019, we have added 50 defaulted (rated) companies with $124 billion in total debt outstanding at default to our dataset. These companies exited bankruptcy from 2018 through the first half of 2020. In total, our dataset comprises 292 rated companies with total debt at default of over half a trillion dollars.

Dataset Profile

The dataset includes companies for which we could obtain reliable and clear recovery data, primarily from the reorganization plans and disclosure statements that debtors filed when they prepared to exit bankruptcy. While we rated all of these companies at the entity level, at the instrument level we rated about 75% of the total number of issuances. The dataset represents about 16.8% of the current $3 trillion in total rated secured and unsecured debt of U.S. corporate speculative-grade rated issuers (as of April 1, 2020) and 20% of the current 1,600 speculative-grade rated companies (as of May 14, 2020).

Table 1

In this review, consistent with our earlier recovery studies, we excluded distressed exchange transactions (which we classify as "selective defaults") as issuers often prefer to restructure debt outside of the more involved and costly bankruptcy proceedings in anticipation that the revised debt structure will reduce leverage and relieve debt service requirements on cash flow and that sponsors still retain control and ownership. However, based on what we observed, companies may undertake a series of distressed exchanges and/or subsequently file for bankruptcy (at which point we would include it in our dataset). As a result, the interim recoveries implied by the exchange terms are not a complete indicator of the eventual actual recovery rates the instrument experienced coming out of a bankruptcy.

The dataset includes the following characteristics:

The time period includes the great recession, the oil and gas downturn, and the beginning of the COVID-19 pandemic. 

There were two points when the number of bankruptcy filings spiked in the period reviewed (see chart 1): during the height of the great recession period between 2008-2010 when 122 companies (in our dataset) filed for bankruptcy, and then between 2015 and 2017 during the oil and gas downturn when oil prices--which had been in the $90-100 per barrel range for a long time--dropped to as low as $26 per barrel before remaining in the $40-60 per barrel range starting 2018. During this time of low oil prices, another 86 companies defaulted, of which 45 were oil and gas companies. Subsequently, 14 oil and gas companies defaulted in 2018 and later. More recently, oil prices reached unprecedented lows in second-quarter 2020, at one point even reaching below zero (because of storage costs), but they've since bounced back to the $40 level in the third quarter. Currently, approximately 43% of the upstream U.S. portfolio is in the 'CCC+' rating category and below.

Chart 1

While our dataset captured companies that emerged from bankruptcy during the first six months of 2020--which includes the beginning of the COVID-19 pandemic and recession--it does not reflect the full impact of these events because they're still ongoing. However, there were 83 defaults during the first half of 2020, already surpassing the 78 total defaults in 2019.

Shorter bankruptcy durations.  Chapter 11 bankruptcies have shortened to an average of seven months in the past five years. Borrowers predominantly enter into proceedings with prepackaged or pre-negotiated bankruptcy plans backed by restructuring support agreements (RSA), which fully or partially outline the restructuring terms and help ensure an expeditious emergence from bankruptcy. While the time in bankruptcy for 2018 emergences was in line with the seven-month average (as shown in chart 1), 2019 emergences were shorter, averaging five months as more bankruptcies already had RSAs before they filed. In some cases, such as McDermott Industries and NPC International, super-priority credit facilities were obtained and RSAs signed, followed by a Chapter 11 bankruptcy filing. There are of course still cases that are administered for a longer time in bankruptcy due to unresolved complexities among various reasons, such as with FirstEnergy Solutions Corp., which took 22 months to emerge in February 2020.

During the financial crisis, the average time entities remained in bankruptcy was 10 months due to the number of bankruptcies that were backed up in courts and the capacity of the courts to process them. If the number of defaults reach or exceed that of the financial crisis, we would likely anticipate proceedings to take longer. Customarily, Chapter 11 cases are predominantly filed in the Third Circuit in Delaware, Second Circuit in the Southern District of New York, and Fifth Circuit in the Southern District of Texas. As of June 30, 2020, according to the Federal Judiciary about 57% of the 2,042 Chapter 11 bankruptcy cases that began in the U.S. were filed in these three districts (29% Delaware and 14% each in Texas and New York).

Exits from bankruptcy were predominantly reorganizations.  Of the 292 companies, 245 exited from bankruptcy predominantly under reorganization plans. Of these, 62 were prepackaged plans, which meant that before the debtor filed for bankruptcy, it pre-negotiated and agreed with its creditors to an RSA that laid out its plan of reorganization along with any post-petition debtor-in-possession (DIP) financing commitments and exit financing arrangements upon emergence. Typically, a debtor would have an RSA agreed upon with the majority of its creditors, including across debt classes, such that the debtor had enough creditors on board to ratify and approve the plan of reorganization, and if not all creditors then enough to cram down the junior debt classes and approve the plan.

There were 32 "363" asset sale transactions, which included credit bids and sales of the core business in whole or in part. Fifteen companies exited under liquidation plans that were conducted under Chapter 11. There were no Chapter 7 filings, although in some instances the cases eventually converted to a Chapter 7 after the debtor's assets were sold in a 363 asset sale transaction and there were residual assets to wind down and remaining matters to resolve.

In the past 2.5 years, 27 companies have exited bankruptcy via plans of reorganization, 17 were pre-packs, five were 363 asset sales, and one via a plan of liquidation. This reflects the bankruptcy trend toward a speedier and more efficient proceeding when it is organized and negotiated before a company enters into a bankruptcy filing. During the 12.5-year period, except for 13 months in 2011, the average time in bankruptcy was around five to 10 months.

Debt mix was predominantly first-lien debt.  In our full dataset, first-lien debt represented the largest debt category, at 45% of total debt outstanding at default. This excludes asset-based lending (ABL) facilities and similar structures (such as securitization facilities and reserve-based facilities), and which we classified as priority claims. These facilities have different collateral and debt structures than first-lien term loans and notes. Most of these priority facilities, which are not usually rated, have historically realized full recoveries, as they typically have stronger structural and collateral protections, including borrowing base restrictions. In addition, ABLs usually roll up into a DIP facility that provides the DIP lenders with a super-priority claim that is senior to prepetition debt claims.

Under our classification, first-lien (non-ABL) debt includes pari passu debt facilities consisting of revolving loans, term loans, and secured notes. These debt instruments typically have a blanket first-lien security interest in either all of the company's assets, or if there is an ABL facility in the structure, on all non-working-capital assets and a second-lien on the ABL collateral. For securitization assets, lenders do not have direct recourse to these assets because they would be held by a special-purpose entity to support the securitization debt, although any residual proceeds would typically flow back to the company and benefit first-lien lenders.

For the full data set (in dollars), priority debt made up 8%, first-lien debt made up 45%, junior secured debt made up 11%, unsecured debt made up 28%, and subordinated debt made up 8%. Of course, individual company debt structures can differ significantly from these averages and this can fundamentally affect the recovery outcomes for different debt classes. In addition, a company's debt structure often depends invariably on its sector, business model, and asset base. For example, a retail company generally has an asset-based revolver to finance its working capital (primarily inventory) and a term loan, while larger companies' debt structures may include unsecured note issuances.

Debt cushion matters.  Over the 12.5-year period, there is a correlation between first-lien loans and bonds that have debt cushion (junior debt that is subject to loss before the first-lien debt is affected) and their respective recoveries (up to a certain level of cushion behind the first lien debt) (see chart 2). First-lien debt recoveries increase with higher levels of cushion and peak at an average 90% recovery when there is a 60%-70% debt cushion. Recoveries are lowest and average about 40% for entities that have zero to minimal debt cushion. Company-specific factors such as size, leverage, credit rating, and its ability to access the capital markets matter when determining the capital structure composition and debt cushion. Larger entities more likely can issue unsecured bond debt, as opposed to smaller companies and those that operate in certain sectors that may have limited access.

Chart 2

Actual Recoveries

Overall recoveries (12.5 years)

The average recovery for priority debt over the 12.5-year study, mainly consisting of revolving ABL and reserve-based lending facilities, was 98%. Average recoveries for first-lien (non-ABL) debt was 79%. Similarly, with a rank order effect, junior-lien and unsecured debt recoveries average in the 28%-30% range, while the subordinated debt average was lower at 15%. The overall weighted average recovery of 56% is in line with historical averages.

The 98% average recovery for priority claims fell short of full recovery primarily due to outlier results related to two oil and gas companies. One recovered 4% because the lien was not properly perfected and therefore became unsecured, and in the other case, ABL lenders to an offshore oil and gas helicopter services provider recovered 39% because the debtor abandoned all of its helicopters, which were financed by and secured the ABL; however, the oil and gas downturn hurt the value of those assets. Substantially all of the other priority claim facilities included in our dataset attained 100% recovery.

Table 2

For the 50 companies we added in the 2.5 years since our last update, first-lien debt recoveries declined to an average of 71% compared to the 12.5-year average of 79%. There were low debt recoveries, particularly in the retail (eight companies) and consumer (five companies) sectors, which averaged 56% and 58%, respectively, because they were affected by ongoing distress in the brick and mortar retail segment. Most of these defaulted companies, as predominant in these sectors, had relatively large ABL facilities ahead of the first-lien debt. In contrast, the 71% 2.5-year average recovery was positively offset by very high 94% first-lien recoveries in the oil and gas sector (14 companies). Junior debt recoveries remained around historical averages, although subordinated debt recoveries were higher in the 50%-70% range for Weatherford International (63%) and American Tire Distributors (55%).

We further broke down recoveries into percentage ranges (commensurate with our recovery rating ranges) by debt class in chart 3 below, which shows that 51% of first-lien debt recoveries (excluding ABL revolvers with a different collateral package) were in the 90%-100% recovery range, with 16% and 12%, respectively, in the 70%-90% and 50%-70% recovery ranges.

Actual recoveries for junior debt classes were mostly in the bottom brackets because most of the recovery accrues to the benefit of first-lien lenders. Still, there were instances of 50% or greater recoveries for about 25% of junior-lien debt classes, 24% of unsecured debt, and 12% of subordinated debt. Recovery for these junior debt classes endure high volatility given the thickness (or lack of) of these instruments, making them sensitive to even slight differences between the going-concern value and the amount of priority debt ahead of it.

Chart 3

Sector Recoveries

In table 3, we broke down the average recoveries based on 15 corporate sectors. Several sectors have a relatively small number of defaults in our dataset, so the average recoveries may not be as significant compared to a sector with a more representative sample of defaults.

Table 3

The top three sectors with the most companies in the dataset were oil and gas (67 companies), media and publishing (41 companies), and natural resources (38 companies). Of media and entertainment companies, 28 filed and emerged in or before 2012, while 13 companies were after 2012. This sector's defaults were mainly driven by secular changes in the publishing and media segments during the great recession period.

Fifteen of the 67 oil and gas companies in the dataset defaulted and emerged in the past 2.5 years, while 43 were during the oil and gas downturn and the remainder were before 2015. Exploration and production (E&P) companies were affected by volatile oil prices, and this had a knock-on effect on oilfield services companies as well. Furthermore, volatile oil prices affected how creditors negotiated with debtors, in particular the valuation for E&P companies, which depended on the companies' oil reserves and oil prices at that time.

Gauging Recovery Levels And Ratings

Average recovery comparisons by debt class

There is a rank order of the debt class priority of claims position and actual recoveries. Our recovery estimates at default were 10% below the actual average for first-lien debt class, while we were closer to recoveries based on estimate at origination (2 percentage points above the 79% actual recovery average, although further away from the 96% median; see table 4 and chart 4).

Table 4

Similarly, with junior debt classes from second-/third-liens to unsecured debt to subordinated debt, our recovery estimates continued to track to actual averages and reflect the rank order of the debt classes, and were within about 10 percentage points from actual averages and aligned with the median.

Chart 4

In the past 2.5 years, average recovery estimates, both at origination and at default, were within 10 percentage points of the average actual recoveries (see table 5 and chart 5). The first-lien debt average recovery of entities that emerged during this period was 71% versus an average of 63% for our recovery estimates at origination and 62% at default. Our recovery estimates at origination and at default for both junior-lien and unsecured debt classes also track within 10 percentage points of the actual average recoveries. There is a higher range for the subordinated debt, but this is because recoveries are sensitive to the thinness of the debt and can drive volatile recoveries.

Chart 5

Table 5

Average recovery comparisons by recovery rating scale for the last 12.5 years

In table 6 and chart 6 we show the recovery range for our recovery rating scale from '1+' to '6' and in chart 6 we plotted the average recovery to see whether they were within the range or outside of it. Average recoveries continue to reflect a rank order within the recovery range associated with each recovery rating. With the exception of recovery ratings of '1' and '6', the average recovery fell within the ranges for all other recovery ratings issued at origination. The average actual recovery for the '1' recovery rating (90%-100% issued at origination) was 80%, and for a '6' recovery rating (0%-10%) was 15%.

A comparison of the average actual recoveries to recovery ratings at default shows that except for '5' and '6' recovery ratings, the actual recoveries fell within the range. The average recovery at default for '5' recovery ratings was 47%, which is 17% better than the top end of the 10%-30% recovery range. The average recovery at default for '6' was 17%, which is again 7% higher than the top end of the 0%-10% range. The key factors here are that debt instruments at the '5' recovery rating mostly comprise junior debt, which is more volatile because it depends on residual recovery value after senior creditors are repaid first, and the debt size is sensitive.

Chart 6

Table 6

Average recovery comparisons by recovery rating scale for the past 2.5 years

To look at the more recent average recoveries to our recovery ratings scale, chart 7 and table 7 focus on the past two-and-a-half years. We observed narrower results as actual average recoveries were within range for recovery ratings '1' to '3' and slightly better (within 7 percentage points) for '4' and '6' recovery ratings. Similarly with the 12.5-year comparison, actual average recoveries were better (13-15 percentage points) than our range on '5' recovery ratings.

Chart 7

Table 7

Actual recovery dispersion from S&P Global Ratings' recovery estimates

In the charts below, we show a distribution of the percentage of the actual recoveries for each rated debt class and their dispersion from our recovery estimates both at origination and at default. So for all actual recoveries for first-lien debt in our dataset (see table 8), 57% of the debt classes were within 20 percentage points from our recovery estimate at origination, 70% were within 30 percentage points, and 81% were within 40 percentage points of our recovery estimates. It is important to note that recoveries are characterized by a significantly wide range of dispersion with a multitude of factors that affect recoveries.

Table 8

As an example, if we estimated a first-lien term loan recovery of 50% when we assigned the recovery rating and the actual recovery was 75%, then the dispersion is 25% higher than our estimate (on the right side of charts 8-10 below). So this term loan would be included in the above 70% of debt issues within 30 percentage points of our recovery estimate. This reflects each actual debt recovery's percentage point difference from each distinct recovery estimate on a debt instrument or debt class.

Table 9

The average dispersion for first-lien, second-lien, and unsecured debt was in the low- to mid-20% range, while subordinated debt dispersion was lower at 17%-18%. The dispersion between actual recovery and estimated recoveries at default was about the same as at origination, except for unsecured debt, which was a three to six percentage points higher at default, perhaps due to sensitivity to the level of debt ahead of it. The proximity between the origination point dispersion and the default point dispersion reflects the stability of our recovery ratings and our estimation of recoveries as distressed companies endure a path to default, likely with changes to their debt structures and operating prospects.

As we can see in the first-lien chart 6 below, the outliers on the left side of the curve were primarily from actual recoveries that were below our recovery ratings (and recovery estimate) assigned at origination, while the outliers on the right side of the curve were primarily actual recoveries that were better than our recovery ratings (and recovery estimate) at or near the time of default. Other than that, 69%-70% of actual recoveries were within 30 percentage points from our recovery estimates.

As we move down to the junior debt classes, it was less likely that actual recoveries were below our recovery estimate because recovery estimates, especially for unsecured and subordinated classes, are lower down the scale. As such, there is more upside potential, which is why for the junior debt classes there is a slightly higher percentage of actual recoveries that are higher than our recovery estimates.

Chart 8

Chart 9

Chart 10

Chart 11

Volatility

The recovery (as a percentage) for the junior debt classes (consisting of both unsecured and subordinated) experiences volatility because recovery depends on the amount of senior debt ahead of these classes, as well as the thickness of the tranche. For example, a $100 recovery value, after $90 in first-lien debt, yields $10 in net recovery for the next debt class. If the next debt class is unsecured and there was $20 in unsecured debt, then it is a 50% recovery, but if the unsecured debt was $10, then it is a 100% recovery. If net recovery for unsecured debt were just $5, then recovery would be 25% if debt outstanding were $20 and 50% for the $10. This scenario has been particularly true in recent years because senior debt, namely first-lien institutional B term loans, has made up a growing share of total debt.

Looking Ahead

While we do not expect the COVID-19 pandemic broadly speaking to have any immediate or first-order impact on recovery ratings as we experienced with credit ratings, because our recovery ratings already envision a distressed scenario where the company defaults, we think there are indirect factors from the pandemic that have contributed to lower recovery ratings and will continue to drive them downward. These include companies raising more debt to tide through the crisis, putting more pressure on the existing debt in the capital structure and their recovery ratings. Further, secular shifts accelerated by the crisis arising in the sector or industry could drive their valuations down and reduce recoveries, such as in the nonessential brick-and-mortar, leisure, and travel sectors. In addition, continued economic uncertainty could lower merger and acquisition activity and buyers could seek lower valuations especially for distressed transactions.

We believe recoveries will generally be lower this time around than historically due to the fundamental factors that drive recovery ratings, including increased leverage, increased senior leverage, and reduced levels of junior debt available as cushion or subordination for the senior pieces, a larger portion of the capital structure being institutional first-lien term loans, and to a lesser extent, the absence of financial maintenance covenants in loan agreements. We have already seen this phenomena with companies that emerged in the past 2.5 years.

Appendix

Framework of our study

Our study assesses the actual recoveries of rated debt instruments of issuers that defaulted in 2008 or later and exited bankruptcy in or before June 30, 2020. We compared them to recovery ratings assigned at the time of origination and to the recovery ratings that stood at the time closest to the default date. By doing so, we aimed to capture our performance against what actually happened and what the differences were between these two recovery rating points in time.

Determination of actual recovery

Determining the best approach to calculate actual recovery is difficult because there is no true method. One approach is to base it on trading prices, which are subject to stale pricing and the lack of a secondary market for smaller companies. For this study, however, we decided that the most appropriate method was to use available recovery data in the bankruptcy documents. These documents provided us with direct information regarding plans of liquidation and asset sales and how they affected recoveries (because they may or may not be represented in trading prices). In addition, we captured recovery from equity values based on plan valuations, which also may or may not be reflected in trading prices.

Definitions

Actual recovery:   Actual recovery is an estimated amount that we derive directly from our review of the bankruptcy documents--particularly the disclosure statements (including the plans of reorganization or liquidation), exhibits, and other filings. In our actual recovery calculation, we include the following forms of payment:

• Any amounts of prepetition debt outstanding that were used to make a credit bid;
• Any amounts of prepetition debt outstanding rolled up into a DIP facility;
• Cash payments;
• Reinstatement of debt or new debt issued as part of the exit financing; and
• Equity ownership in the form of warrants, preferred stock, and/or common stock (assuming the equity value assessed in the bankruptcy documents).

Origination recovery rating:   For origination recovery ratings, if the origination date preceded our June 2007 roll-out of recovery ratings on secured debt under the revised recovery rating scale (March 2008 for unsecured debt), we used the recovery rating first assigned under the revised scale in this study. For example, if a rated company was originally assigned recovery ratings in 2006, we would then use the recovery ratings assigned in June 2007 under the revised scale. Alternatively, if a company completed a major recapitalization in September 2009, we considered the recovery ratings assigned at that time as the origination ratings because the new capital structure requires a full reassessment of recovery. If the same company completed relatively small add-on debt issuances, we would continue to use September 2009 as the origination point.

Default recovery rating:   For default recovery ratings, we used the most recent recovery ratings on the debt instruments at the time of default--preferably the ratings that were on the debt immediately preceding the default, but no later than a week subsequent to the default date.

S&P Global Ratings' revised recovery rating scale:  Our revised recovery methodology for secured debt, implemented in June 2007, expanded our previous recovery rating scale to seven recovery rating levels from five, while narrowing the recovery range band (see table 8). In March 2008, we rolled out recovery ratings for unsecured debt instruments, which use the same scale. In December 2016, we further revised our recovery methodology and provided guidance surrounding our recovery assumptions.

Table 10

Related Research

• Corporate Defaults Tripled In The Second Quarter, Sept. 29, 2020
• Default, Transition, and Recovery: The Gap Between Market Expectations And Credit-Based Indications Of U.S. Defaults Is Growing, Aug. 27, 2020
• Credit Trends: A Round-Trip Ticket: Some Companies Downgraded To 'CCC+' Could Be Headed To 'B-' As The Economy Recovers, Aug. 7, 2020
• U.S. Leveraged Finance Q2 2020 Update: Recovery Ratings Maintain Social Distance From Credit Impact Of COVID-19 Pandemic, July 23, 2020
• Industry Top Trends: Oil And Gas, July 16, 2020
• Default, Transition, and Recovery: More Than One-Quarter Of Speculative-Grade Issuers Are Weakest Links, May 14, 2020