KEY TAKEAWAYS

• In this report, we analyzed default and recovery performance data for 7,064 unrated global project bank loans for the period between Jan. 1, 1995, and Dec. 31, 2016, which based on external data sources and our own analysis we believe accounts for approximately two-thirds of total global project finance loans generated during this timeframe.
• For the performance of rated projects and rated infrastructure, see the study "2017 Inaugural Infrastructure Default Study And Rating Transitions," published on Nov. 20, 2018, on RatingsDirect.
• The 10-year cumulative default rate for all unrated project finance bank loans stood at 6.3%. However, the rate drops to 5.85% just for the core infrastructure sectors of power, transport/social infrastructure, and oil & gas (core infra). The percentage further drops to 5.6% for projects in the public sphere (PPP or PFI), illustrating their comparatively lower-risk nature.
• The unrated project finance loans had a 1.5% average annual default rate over 1995-2016 (1.3% for core infra), but this rate was much lower in the past 10 years at 0.9% (0.8% for core infra).
• The annual default gap between Organization for Economic Cooperation and Development and emerging markets has fallen over the past decade. OECD project financings still showed a slightly lower average annual default rate (1.4%) over 1995-2016 than in emerging market and developing economies (1.5% for EMDE A and 1.6% for EMDE B). The difference is less than we expected, probably because emerging markets were less affected by the 2008-2009 financial crisis and the prolonged 2011-2015 crisis in southern Europe.
• Sovereign and economic-related stress, as well as industry-related crises (notably the U.S. utilities crisis in 2001-2003) have triggered the lion's share of project defaults in our view. Emerging-market countries with more severe past sovereign crises or higher country risks, such as Argentina, Thailand, Colombia, Indonesia, and Egypt, display the weakest project annual default performance. While not a sovereign default, Spain's economic downturn over 2011-2015 also led to above-average default rates. Finally, we believe the U.S. also showed comparatively above-average default rates, caused by a spike in 2001-2003 due to the dot-com bust and the U.S. utilities crisis.
• By contrast, we find comparatively low annual defaults across some OECD as well as EMDE countries: These include Canada, Japan, and Netherlands, most Middle Eastern countries, and, maybe somewhat unexpected, Turkey, Russia, and Mexico.
• Annual defaults by industry showed that transport/social infrastructure projects had the lowest average annual default rate over 1995-2016, and power projects the highest rate among core infra. However, the picture is different if the starting point is 2008, marking the start of the global financial crisis: over the last decade default rates for transport/social infrastructure projects, impacted by weaker economic conditions, rose comparatively more.
• Project finance loan recovery rates averaged a high 84% on a nominal basis (77% on a discounted basis). This is slightly higher than our rated senior secured recoveries of 80% for nonfinancial corporates and much higher than the 50% average for rated senior unsecured debt of nonfinancial corporates.
• Notably, we see no meaningful difference in recoveries for projects in OECD, EMDE A, or EMDE B countries, with discounted average recoveries of 78%, 77%, and 75%. However, we observed a meaningful difference in average time to recovery of 1.7 years for projects in the OECD, versus 2.6-2.8 years in EMDE-A and EMDE-B countries, which we think reflects the better predictability of insolvency proceedings and legislation in most OECD countries.

Data Access, Scope And Definitional Limitations

S&P Global Ratings prepared this report based solely on aggregated and anonymized project finance loan performance data provided by S&P Global Market Intelligence, a division of S&P Global. A consortium of project finance bank lenders contributed underlying project finance loan portfolio information to S&P Global Market Intelligence. The data set analyzed as part of our study consisted of 7,064 unrated global project loans banks for the period between Jan. 1, 1995, and Dec. 31, 2016 (the "Data Set"). In analyzing the Data Set, we also applied the World Bank's geographical classifications and country groupings. The scope of our analysis is a function of granularity we were able to derive from the Data Set.

This report focuses on the historical performance of unrated project finance loans. The concept of default applied for the purposes of this study to the Data Set is assumed to be consistent with the Basel II definition of default. Please note that for the purposes of credit ratings, S&P Global Ratings uses a different approach to analyzing default (See the Appendix).

Consequently, we caution that default statistics associated with the Data Set may not be fully comparable with default statistics for rated project finance loans.

For recovery statistics, we believe that the distinction between Basel II and S&P Global Ratings' definitions of default is less consequential and hence we have, in this report, drawn a more explicit comparison between recoveries for unrated and rated project finance debt.

For our just published default and recovery study for rated infrastructure, which includes rated project finance, see "2017 Inaugural Infrastructure Default Study And Rating Transitions," published on Nov. 20, 2018, on RatingsDirect.

Scope

Of the 7,064 project loans originated between Jan. 1, 1995 to Dec. 31, 2016, that were part of our report (see chart 1), 564 were reported to have defaulted over the period. Of the defaulted projects, we have recovery data for 376 projects that emerged from default, corresponding to 1,338 tranches or debt instruments.

For the purposes of this study we defined regions (see chart 2) according to the definition of region used by S&P Global Market Intelligence. We note that S&P Global Market Intelligence includes Russia and some other countries in the Commonwealth of Independent States (CIS) in Eastern Europe.

Furthermore, we have split the Data Set into three subsets according to the classifications the World Bank uses (see chart 3). Advanced economies are high-income countries that are members of either OECD or EEA. As for emerging-market and developing countries (EMDE), EMDE-A countries are all non-high income countries, while EMDE-B countries are the same excluding non-high income OECD and EEA countries such as Mexico, Turkey, Chile, Czech Republic, Estonia, Latvia, Hungary, Poland, Slovakia, as well as Croatia, Bulgaria, Lithuania, and Romania.

Broken down by industry (see chart 4), power accounts for 40% of total loans, of which 25% are non-renewables and 15% are renewable projects. Public-private partnerships (PPP) and private finance initiatives (PFI) account for 24% of the total loans, largely concentrated in the transport/social infrastructure space.

Data sourcing

We believe the conclusions in this report should fairly well reflect the performance of global project finance loan activity, as based on external data sources and our own analysis, we believe that the Data Set accounts for roughly two-thirds of total global project finance loans over the period. However, project data available for this report from consortium banks have declined in relative terms to less than 50% of estimated global project financings for the 2012-2016 period.

Data samples and concentration

We caution that statistical conclusions stemming from smaller data samples, such as for specific countries, may not be as reliable or illuminating as those based on larger data samples. It should be noted that the number of projects by country varies widely: the U.K. and U.S. each account for more than 1,000 projects; followed by Spain (more than 600), Australia (about 400), and Germany, Italy, and France with about 300 each; and Brazil, Canada, and Mexico, with close to 200 each. Other major project finance countries with approximately 100 projects included each are Japan, Netherlands, Chile, Indonesia, Portugal, Turkey, Russia, and China.

KEY DEFAULT STATISTICS

Average annual default performance

We have found, over the period 1995-2016, an average annual default rate of 1.3% for core infrastructure projects (versus 1.5% for all projects). For projects in the public sphere, that is, PPP or PFI, the average annual default rate falls to a mere 0.7%, illustrating their distinctly lower risk profile relative to the Data Set overall. We believe that focusing on "core infrastructure" projects (core infra) makes this report more comparable with the rated project financings in our "2017 Inaugural Infrastructure Default Study And Rating Transitions" study. As such, we view core infrastructure (core infra) as comprising the larger segments of power (both merchant and regulated utilities); transport/social (which includes airports, toll roads, tunnels, ports and terminals; water, waste utilities, and other); and oil & gas projects. Further, we view core infra assets as excluding project finance activity in the fields of media, leisure, metals & mining, and chemicals, among others. We would normally also exclude exploration and production (E&P)-exposed oil & gas projects, but such subsector details were not available from the Data Set.

What's more, we see a sharp decline in bank loan project default rates over the past 10 years (compared with 1995-2007), with annual average default rates falling to 0.8% for core infrastructure and since 2015 back to the lows seen before the financial crisis.

Finally, we note large and smaller spikes in default rates over the 20 years. These coincided with specific sovereign crises, industry-driven downturns (such as the dot-com bust and the U.S. utilities crisis following the demise of Enron at the start of the decade), and economic fallout from the global financial crisis in 2008. Keep in mind that the effect on project finance default rates may lag because of the use of debt service reserve accounts (DSRAs, see chart 5).

Project finance cumulative and marginal default rates

We calculate a 10-year cumulative default rate for all (unrated) bank loan projects of 6.3% (see chart 7 and table 1). When narrowing the scope of projects to core infrastructure sectors, this rate drop to 5.85% and when limited to PPP/PFI projects the rate is 5.6%.

Table 1

Cumulative Default Rates For All Project Finance Projects
		--Year--
(%)	Issuers (no.)	1	2	3	4	5	6	7	8	9	10
1995	371	2.16	1.93	1.69	0.86	2.31	0.89	1.79	1.21	0.31	0.31
1996	489	1.84	2.08	2.34	2.61	0.89	1.81	1.15	0.23	0.23	0.70
1997	658	1.52	2.78	3.33	1.31	2.00	2.04	0.35	0.35	0.87	0.35
1998	855	2.81	2.77	1.24	2.76	2.58	0.79	0.53	0.67	0.40	0.00
1999	1,040	2.60	1.28	3.20	2.79	1.38	0.54	0.54	0.44	0.22	0.11
2000	1,276	1.25	3.02	3.52	1.70	0.95	0.52	0.61	0.18	0.09	0.00
2001	1,540	3.05	4.49	1.89	1.00	0.58	0.51	0.15	0.15	0.00	0.07
2002	1,733	4.73	2.73	1.00	0.75	0.57	0.19	0.13	0.13	0.06	0.13
2003	1,839	2.50	1.00	0.68	0.62	0.29	0.11	0.11	0.06	0.11	0.17
2004	1,944	1.08	0.73	0.63	0.26	0.26	0.21	0.11	0.16	0.37	0.11
2005	2,134	0.75	0.61	0.29	0.24	0.33	0.10	0.14	0.34	0.10	0.19
2006	2,279	0.61	0.26	0.35	0.40	0.22	0.18	0.36	0.13	0.32	0.18
2007	2,578	0.27	0.51	0.55	0.51	0.55	0.40	0.20	0.44	0.24	0.12
2008	2,956	0.61	1.06	0.76	0.76	0.52	0.49	0.56	0.32	0.11
2009	2,603	1.38	1.01	0.94	0.79	0.80	1.01	0.49	0.25
2010	2,639	0.99	0.96	0.81	0.82	1.02	0.48	0.40
2011	2,606	1.04	0.81	0.86	1.18	0.56	0.40
2012	2,793	0.82	0.97	1.28	0.66	0.37
2013	2,874	1.01	1.23	0.64	0.47
2014	2,907	1.27	0.66	0.53
2015	3,049	0.72	0.53
2016	3,133	0.51
Marginal default rates		1.27	1.19	1.01	0.84	0.67	0.48	0.36	0.27	0.20	0.15
Cumulative default rates		1.27	2.45	3.43	4.24	4.88	5.33	5.67	5.93	6.12	6.26
Source: S&P Global Market Intelligence.

In terms of marginal default rates (see chart 8 and table 1), all project financings show higher levels in the first three years, coming down to very low levels after year 5 when projects typically have established an operational track record and have dealt with construction risks.

Key recovery statistics

The average discounted recovery rate at the debt instrument level was 77%, compared with a nominal 84%. What's notable about project recovery rates is that they have been fairly constant across cycles and largely delinked from default performance years (see chart 9). In contrast, corporate recoveries tend to be negatively correlated to spikes in default rates, reflecting the more direct impact of adverse economic conditions on corporate valuations. The main exception relates to 2010-2011 when recovery rates dropped to 64% after the financial crisis, down from 80% previously. Since 2012, recovery rates have hovered around 71%.

Compared with rated corporate debt, we found that recoveries for projects emerging from default are more resilient. As mentioned, the average nominal recovery rate for the Data Set was 84% (or even 87% for core infra). This compares with only 59% for rated project recoveries: As mentioned above, the lower recovery rate than for unrated bank loans may result from the relatively small rated sample size (46 recovery tranches only), essentially of U.S. projects many with merchant power exposure, and possibly differences between bonds and bank recoveries.

The high nominal recovery level of (unrated) projects (of 84%) is nearly the same as that of rated senior secured corporate debt of close to 80%. Recoveries for all rated debt instruments for rated infrastructure corporate entities was observed to be equally high at 75%, well ahead of rated nonfinancial corporates (at 59% and 50% when considering unsecured bonds only, see table 3).

Table 3

Nominal Recovery Rates By Debt Instrument And Asset Class (Rated And Unrated)
(%)	Average nominal recovery	Median nominal recovery
Unrated project finance (bank consortium)	84	100
Unrated core infrastructure project finance (bank consortium)	87	100
Rated project finance	59	65
Rated corporate infrastructure	75	91
Rated nonfinancial corporates*	59	60
Of which senior unsecured	50	41
*From 1995-2016. Sources: S&P Global Market Intelligence, S&P Global Ratings.

Certain types of project finance loans may experience relatively high recoveries as:

• Certain projects may fund the provision of essential services such as power, water, and gas often have little or no practical substitute.
• Low industry risk and stable cash flows derived from long-term offtake contracts, such as power purchase agreements or availability-based payments under a concession, operations and maintenance contracts, or fuel supply contracts, among others.
• The ability to add debt as well as to distribute profits is typically limited by covenants and debt is predominately secured.

Finally, when looking at the recovery profiles for various rated and unrated asset classes (see chart 10), we observe a barbell distribution with 52% of unrated project finance having a recovery more than 90% and a similarly elevated percentage of 53% for rated infrastructure corporates, compared with 34% for nonfinancial corporates. The rated project universe (predominately U.S.) also shows a barbell but with more variance among the 46 tranches.

DEFAULT ANALYSIS BREAKDOWN BY REGION AND INDUSTRY

Defaults by region and by advanced versus developing country

While project financing typically includes structural features such as pledged collateral and contracted off-take agreements, which can mitigate debt service coverage volatility and exposure to parents and other counterparties, analysis of historical time series suggest that it does not completely insulate a project from shocks within economic cycles.

Over the past 20 years, we have seen increases in project finance defaults both across sovereign stress and specific industry-related market downturns (see box).

Recovery rates by region   As can be seen in chart 11, Latin American annual default rates have been outliers (averaging 2.2% over 1995-2016), followed by North America (1.8%), compared with the global average of 1.5%. Average annual default levels for Asia-Pacific have dropped over the past decade, now standing at 1.6% over 1995-2016, not far from the Western and Eastern Europe averages of 1.4%-1.5%. The region with the lowest average annual default statistics by far has been the Middle East.

Breakdown by advanced versus developing country  We found that EMDEs still have a higher average annual default rate over the 1995-2016 period at about 1.8% for EMDE-B countries and 1.6% for EMDE-A. This compares with 1.4% on average for OECD (and 1.3% for the European Economic Area; see chart 12.) However, the gap between the OECD's and EMDEs' performances has continued to narrow, as annual default rates converged to below 1% for both country groups over the last decade, explained by the fact that projects in Latin American and in particular Asia-Pacific economies were comparatively less affected over the past decade by the financial crisis.

We nevertheless found notable outliers in both OECD countries and EMDEs. Understandably, emerging markets that experienced a major sovereign crisis have shown above-average defaults, but so did Spain, while the comparatively weaker performance in the U.S. is partly explained by the utilities crisis in 2001-2003. By contrast, we uncovered a comparatively strong project performance by most countries in the Middle East, developing economies such as Turkey, and Russia, as well as in Mexico, with the lowest defaults in advanced economies observed for Canada, Japan, and the Netherlands, among others.

Default by industry

Largely reflecting the industry distribution of the project finance data, most defaults occurred in the power segment (39%), followed by infrastructure (23%), and oil & gas (10%; see table 4).

Table 4

Defaults By Industry
(No.)	1995-2003	2004-2016	Total defaults (from 1995)	Total bank loan defaults (%)
Power	125	94	219	39
Transport/social infrastructure	21	106	127	23
Oil & gas	19	38	57	10
Media & telecom	58	12	70	12
Metals & mining	27	17	44	8
Chemicals production	8	7	15	3
Leisure & recreation	3	8	11	2
Manufacturing	6	4	10	2
Other	2	5	8	1
Source: S&P Global Market Intelligence.

We have already highlighted that the three core infrastructure sectors (see chart 13) show a lower cumulative default rate and thus a somewhat lower risk profile than the other segments. While power projects have shown the highest (1.6%) average annual default rate over 1995-2016 for these three sectors (that averaged 1.3%), it actually became the lowest over the past 10 years. Explained differently, transport/social projects displayed the least volatile performance over the 20-year period, but nevertheless suffered more from the economic downturn following the financial crisis.

Power: renewables versus non-renewables  The performance of power projects has differed depending on what underpins revenues. Renewable projects, as of today, tend to have benefited from predictable payments from power purchase agreements, contracts, or tariffs that are often subsidized, and therefore should have greater revenue stability than merchant power (so far mainly thermal) projects, which are exposed to energy price fluctuations. However, the annual default performance of renewables projects has not been lower than for nonrenewable (thermal) projects. This is in our view because of regulatory changes in Spain, with substantial downward revisions of remuneration in 2014, which particularly affected renewable projects, as well as due to contractual issues or deterioration of counterparty creditworthiness during the U.S. utilities crisis.

For completion's sake, we show the performance of the other industries below: Annual default rates for each of them exceeded 2% on average over 1995-2016 (see chart 14).

Time to default by region and advanced versus developing country

The median time to default was observed to be about three years. Differences among regions and World Bank country groups appear quite limited (see table 5).

Table 5

Time To Default By Region And Advanced Versus Developing Countries
(Years)	10th percentile	25th percentile	Median	75th percentile	90th percentile
Western Europe	1-2	2-3	3-4	5-6	8-9
North America	0-1	1-2	2-3	4-5	6-7
Asia-Pacific	0-1	1-2	2-3	4-5	6-7
Latin America	0-1	1-2	3-4	4-5	6-7
European Economic Area (EEA)	1-2	2-3	3-4	5-6	8-9
OECD (or EEA)	0-1	1-2	3-4	5-6	7-8
EMDE-A	0-1	1-2	3-4	4-5	6-7
EMDE-B	0-1	1-2	2-3	4-5	6-7
EMDE--Emerging market and developing countries. EMDE-A countries are non-high income countries, plus East European countries in transition such as the Baltics, Poland, Hungary, and Slovakia. EMDE-B countries exclude EMDE-A countries, as well as non-high income OECD countries such as Mexico, Turkey, Chile, and Croatia, Bulgaria, and Romania. Source: S&P Global Market Intelligence.

RECOVERY ANALYSIS BY REGION AND INDUSTRY

In terms of the Data Set of the 564 projects that defaulted, we have recovery data for 376 projects that emerged from default, corresponding to 1,338 tranches. Based on the available recovery information, S&P Global Market Intelligence categorized each of the 1,338 recovered debt instruments as one of eight resolution types (see chart 15). Adjusted for projects for which data were not provided, we find that most (82%) of the recovered debt instruments were restructured, sold, acquired, or returned to performing. This compares with just over 10% of projects that did not find a solution and were liquidated.

In analyzing recoveries by emergence year, we find that the period 2002-2005 represents approximately 61% of recoveries. North America and Western Europe accounted for 71% of those recoveries.

Recovery rates by region, advanced versus developing country, and jurisdiction

Here, we focus on the regions--Western Europe, North American, Latin America, and the Asia-Pacific (excluding Oceania) for which we believed we had enough data to develop meaningful conclusions.

Recovery rates by region   Interestingly, differences among the above-mentioned regions is limited, with all showing average discounted recovery rates close to the global average of 77%. Generally, distributions are comparable, with the exception of North America, which shows a slightly weaker recovery for the 10% decile (see table 6).

Table 6

Discounted Recovery Rates Distribution By Region
	10th percentile	25th percentile	Median	75th percentile	90th percentile	Average
Western Europe	30-40	56	94	99	100	77
North America	20-30	65	88	99	100	76
Asia-Pacific	30-40	64	85	100	100	76
Latin America	30-40	64	93	100	100	80
World total	30-40	61	91	99	100	77
Source: S&P Global Market Intelligence.

Recovery rates by advanced versus developing country   We observed only minor divergences in recovery rates between advanced economies and emerging and developing markets, with average recovery rates of 78% and 77%/75% respectively (see table 7) for the Data Set. However, recoveries for the lowest 10% quartile were somewhat lower for EMDE-B countries.

Table 7

Discounted Recovery Rates Distribution By Advanced Versus Developing Countries
	10th percentile	25th percentile	Median	75th percentile	90th percentile	Average
European Economic Area (EEA)	30-40	56	94	99	100	78
OECD (or EEA)	30-40	61	92	99	100	78
EMDE-A	30-40	63	90	100	100	77
EMDE-B	20-30	58	87	100	100	75
EMDE--Emerging market and developing countries. EMDE-A countries are non-high income countries, plus East European countries in transition such as the Baltics, Poland, Hungary, and Slovakia. EMDE-B countries exclude EMDE-A countries, as well as non-high income OECD countries such as Mexico, Turkey, Chile, and Croatia, Bulgaria, and Romania. Source: S&P Global Market Intelligence

Recovery rates by jurisdiction  For this breakdown, we used our own classification of jurisdiction groupings, developed according to a specific methodology (see "Criteria | Corporates | Recovery: Methodology: Jurisdiction Ranking Assessments," published on Jan. 20, 2016.) This methodology is used for recovery analysis of nonfinancial corporate issuers, project finance recovery ratings, and for certain nonbank financial services companies. It classifies insolvency regimes into three jurisdiction groupings: Group A (comprising mainly OECD countries), Group B, and Group C (higher-risk emerging markets; see table 8). For this article only, we also added countries without any jurisdiction assessment to Group C.

The jurisdiction groupings are an indicator of our view of the relative degree of protection that a country's insolvency laws and practices afford creditors' interests and of the predictability of those proceedings. This assessment reflects, on a relative basis how insolvency proceedings and rule-of-law considerations in a given jurisdiction are likely to affect postdefault recovery prospects for creditors subject to insolvency proceedings in that jurisdiction.

Table 8

Country Ranking By S&P Global Ratings Jurisdiction*
Countries	Jurisdiction	Countries	Jurisdiction	Countries	Jurisdiction
Australia	A	Brazil	B	India	C
Canada	A	Czech Republic	B	Indonesia	C
France	A	Dubai International Finance Centre	B	Kazakhstan	C
Germany	A	Greece	B	Romania	C
Japan	A	Italy	B	Russia	C
Poland	A	Mexico	B	Ukraine	C
Portugal	A	South Africa	B	Vietnam	C
Spain	A	Turkey	B
U.K.	A	United Arab Emirates	B
U.S.	A
*Non-exhaustive list. Source: S&P Global Ratings.

Analyzing the recovery data for projects according to these three groupings, our conclusions are similar to those we made for advanced versus developing countries above. Although one would expect higher outcomes for group A, the difference in recovery values between groups A and C is very narrow, with an average level of 77% and 78% (see table 9-10) for the Data Set. However, the median for group C is 10% lower. While this may indicate that project finance security structures are effective in emerging markets, outcomes are likely to be influenced as well by type of project, not just creditor laws.

We view group B data as less representative, given the limited debt instrument sample and concentration across four countries (Mexico, Italy, Greece, and Brazil) with high recovery values stemming from Mexican and Italian projects.

Table 9-10

Discounted Recovery Rates Distribution By S&P Global Ratings' Jurisdiction
(No.)	Debt instruments (no.)	25th percentile	Median	75th percentile	Average
Jurisdiction A	936	67	90	98	78
Jurisdiction B	42	79	87	96	88
Jurisdiction C and countries without a jurisdiction assessment	321	63	81	95	77
Discounted recovery rates distribution by recovery rate
(No.)	0%-10%	10%-20%	20-30%	30%-40%	40%-50%	50%-60%	60%-70%	70%-80%	80%-90%	90%-100%	100%
Jurisdiction A	39	11	28	23	33	68	48	63	84	125	414
Jurisdiction B	0	0	0	0	2	3	1	2	0	15	18
Jurisdiction C and countries without a jurisdiction assessment	12	9	9	6	12	17	21	25	41	38	131
Total	51	21	37	29	47	88	69	90	125	178	563
Note: On the basis of 1,299 total recovery data points (instead of 1,338, as countries with less than five debt instruments on recovery were excluded. Source: S&P Global Market Intelligence.

Recovery rates by industry

Discounted recovery rates by industry show limited variance on average (see chart 16). Average discounted recovery rates for core infra were about 80%, slightly higher than the overall average of 77% for the Data Set. Equally relevant we believe is the lower observed downside recovery risk for core infra, with the 25% percentile of lowest recoveries averaging 68%, compared with 40% for other, higher-risk project finance sectors, such as media and metals & mining.

Time to resolution by region, advanced versus developing country, and jurisdiction

Time to resolution (TTR) is defined as the length of time between the reported loan default date and the reported resolution date.

TTR By region  In contrast to average recovery levels, we see a clear distinction in TTR by region for the Data Set (see table 11), with advanced economies showing a faster workout process. Western Europe and North America show an average TTR of about 1.5 years versus 2.5 years for Asia-Pacific and Latin America.

Table 11

Time To Recovery By Region
(Years)	10th percentile	25th percentile	Median	75th percentile	Average
Western Europe	4-5	2.0	0.7	0.3	1.4
North America	3-4	2.4	1.6	0.6	1.7
Asia-Pacific	5-6	4.2	2.2	1.0	2.7
Latin America	6-7	4.2	1.4	0.2	2.4
Source: S&P Global Market Intelligence

TTR by advanced versus developing countries  Fully in line with TTR by region, we find longer resolution times for both EMDE-A and B countries (on average 2.7 years) versus 1.7 years for OECD countries. Moreover, the 10% decile shows the period to resolution can extend to five or six years in EMDEs, whereas the 10% of outlier OECD projects are resolved within three to four years (see table 12).

Table 12

Time To Recovery By Country
(Years)	10th percentile	25th percentile	Median	75th percentile	Average
European Economic Area (EEA)	3-4	2.2	1.0	0.4	1.6
OECD (or EEA)	3-4	2.4	1.2	0.5	1.7
EMDE-A	5-6	3.8	2.1	0.7	2.6
EMDE-B	5-6	4.1	2.2	0.8	2.8
EMDE--Emerging market and developing countries. EMDE-A countries are non-high income countries, plus East European countries in transition such as the Baltics, Poland, Hungary, and Slovakia. EMDE-B countries exclude EMDE-A countries, as well as non-high income OECD countries such as Mexico, Turkey, Chile, and Croatia, Bulgaria, and Romania. Source: S&P Global Market Intelligence.

TTR by jurisdiction  Not surprisingly, we find a strong correlation between predictability of insolvency proceedings and time to recovery, with group A countries showing an average time to resolution of 1.2 years versus 2.2 years for group C (see tables 13 and 14) for the Data Set.

Table 13-14

Time To Recovery (TTR) By S&P Global Ratings Jurisdiction
(Years)	Debt instruments (no.)	10th percentile	25th percentile	Median	75th percentile	Average
Jurisdiction A	936	3-4	2-3	1-2	0-1	1.2
Jurisdiction B	42	2-3	1-2	1-2	0-1	1.0
Jurisdiction C and countries without jurisdiction assessment	320	5-6	3-4	2-3	0-1	2.2
Number of projects by S&P Global Ratings jurisdiction and TTR
(No.)	Debt instruments (no.)	<1 year	1 to <2 years	2 to <3 years	3 to <4 years	4 to <5 years	5 to <6 years	6 to <7 years	7 to <8 years	8 to <9 years	9 to <10yrs	10 to <11yrs	11 to <12yrs	12 to <13yrs
Jurisdiction A	936	413	224	155	73	35	8	18	0	4	1	2	0	3
Jurisdiction B	42	20	13	4	3	0	2	0	0	0	0	0	0	0
Jurisdiction C and countries without jurisdiction assessment	320	95	48	58	39	36	19	10	6	4	4	0	1	0
Total	1,298	528	285	217	115	71	29	28	6	8	5	2	1	3
Note: On the basis of 1,298 total recovery data points (instead of 1,338, as countries with less than five debt instruments on recovery were excluded. Source: S&P Global Market Intelligence.

TTR versus recovery

An analysis of resolved project loans by TTR versus the corresponding average discounted recovery rate shows an inverse relationship (see chart 17). That is, the longer the time to emergence from default, the lower the recovery rate. Longer recovery processes typically incur higher costs, but equally very complex workouts are more likely to relate to highly distressed projects. Somewhat surprisingly and albeit statistically not representative, we note that a very small number of project resolutions in year 10 and beyond yielded stronger recoveries again.

APPENDIX: DEFAULT AND RECOVERY DEFINITIONS AND CALCULATIONS

Basel II definition of default

The definition of default that we use in this article is based on the standard Basel II definition of default, which captures a wider range of defaults, including circumstances wherein a reporting bank considers the obligor is unlikely to pay its credit obligation in full. According to this definition, a project is in default if:

• A payment is past due more than 90 days on any material credit obligation.
• The lender takes a charge or an account-specific provision because of a perceived deterioration in credit quality of the project exposure.
• The lender sells the project instrument at a material credit-related loss.
• The lender consents to a distressed restructuring likely to result in a diminished financial obligation caused by the material forgiveness of principal or interest.
• The obligor has sought or has been placed in bankruptcy protection.

How the Basel II definition compares with S&P Global Ratings' definition of default:

• Our definition is stricter in that any payment beyond 30 days would be considered a default. However, we would not expect this to yield material differences as, unlike corporates, projects typically enjoy a six-month funded DSRA.
• Our definition of a distressed exchange tends to encompass broader situations where a project's credit quality is under pressure forcing lenders to extend maturities.
• Whereas our default definition focuses on missing a cash payment, Basel II requires a different approach in that it considers a default whenever the lender takes a charge or an account-specific provision following a perceived deterioration in credit quality of the project exposure.

Our definition of recovery

S&P Global Ratings' concept of recovery is not materially different than that used by the bank consortium with respect to the Data Set. The recovery definition used for the purposes of the Data Set considered whether:

• After default a project loan resumes scheduled payments on a regular basis (that is, returns to performing).
• Following a restructuring, scheduled payments resume based on restructured debt service.
• The lender sells or transfers the defaulted debt instruments.
• Liquidation proceeds have been distributed to creditors.
• The bankrupcty process is completed.
• The guarantor provides additional capital support covering some portion of scheduled debt service.

Static pool methodology

S&P Global Market Intelligence conducted the calculations on the basis of groupings called static pools, which they formed by grouping projects by performance status at the beginning of each year the study covered. They then follow each static pool from that point forward. Their analysis assigned all deals included in the study to one or more static pools. The pools are static in that their membership is constant, similar to a buy-and-hold portfolio.

When a project defaults, they assign that default back to all of the static pools to which the project belonged. S&P Global Market Intelligence's use of the static pool methodology is intended to mitigate certain issues that may arise in estimating default rates and allow calculations across multiple time horizons.

Annual default rate calculation

Annual default rates were calculated for each static pool as percentages of projects that defaulted during the year with respect to the number of projects at the start of the year in each category.

Marginal default rate calculation

S&P Global Market Intelligence defines the percentage of projects that default in nth year of performance history as the marginal default rate for that year.

Cumulative average default rate calculation

Cumulative default rates that average the experience of all static pools were determined by calculating marginal default rates, conditional on survival (survivors being nondefaulters) for each possible time horizon and for each static pool, weight-averaging the conditional marginal default rates, and accumulating the average conditional marginal default rates. Conditional default rates are calculated by dividing the number of projects in a static pool that default at a specific time horizon, by the number of projects that survived (did not default) to that point in time. Weights are based on the number of projects in each static pool. Cumulative default rates are one minus the product of the proportion of survivors (non-defaulters).

Primary Credit Analysts:	Trevor J D'Olier-Lees, New York (1) 212-438-7985; trevor.dolier-lees@spglobal.com
	Karl Nietvelt, Paris (33) 1-4420-6751; karl.nietvelt@spglobal.com
	Gonzalo Cantabrana Fernandez, Madrid (34) 91 389 6955; gonzalo.cantabrana@spglobal.com
Secondary Credit Analyst:	Ben L Macdonald, CFA, Centennial (1) 303-721-4723; ben.macdonald@spglobal.com
Research Contributors:	Massimo Schiavo, Paris + 33 14 420 6718; Massimo.Schiavo@spglobal.com
	Yogesh Balasubramanian, Mumbai; yogesh.subramanian@standardandpoors.com