In my recent book, Credit Crises: The Role of Excess Capital (Stevenson, 2024), I demonstrate that credit crises are a major force influencing the global economy and the commercial banking industry. Credit crises are defined as eruptions in defaults on credit agreements that result in extreme losses to lenders and investors and sometimes in bank failures. Specifically, a credit crisis is a period when charge-offs on loans held by banks meet or exceed the 95th percentile of historical loan charge-offs (Stevenson, 2024).
In particular, I show how these catastrophes are direct, though lagged, consequences of excess capital – debt that grows more rapidly than economic growth and that flows to subprime and non-investment grade borrowers. The high default probabilities of such borrowers emerge as concentrated periods of defaults on loans and bonds that cause extreme losses to lenders and, often, significant economic recessions.
The most recent example in the United States was the subprime mortgage crisis of 2007 to 2010, that produced the Financial Crisis of the same era. U.S. household net worth declined by nearly $11 trillion from its 3Q2007 peak, a fall of 16 percent.[1] U.S. housing prices fell almost 30 percent, on average, and the U.S. stock market dropped more than 50 percent by early 2009.[2] Simon Johnson of Bloomberg estimated that the lost output and income from the crisis was “at least 40 percent of 2007 gross domestic product.”[3]
The impact of the subprime mortgage crisis on the U.S. banking system was profound. Bank failures jumped dramatically in 2008, reaching 25 total failures for the year, up from 3 in 2007 and peaking at 157 failed banks in 2010 (Stevenson, 2024). Bank failures related to the crisis continued through 2015, even though the economy had been in recovery and growing for several years. Because the largest banks in the U.S. received capital infusions from the U.S. government via the Troubled Asset Relief Program, very few large banks failed.
Moreover, this crisis was not restricted to the United States. Mortgage markets collapsed throughout the developed world, especially in Spain and Ireland, after experiencing dramatic growth in the early- and mid-2000s.
This blog post is the second of two parts that examines the dynamics of Irish loan markets in the 2000s and 2010s in the context of the Excess Capital Hypothesis (ECH), a concept that I present Credit Crises (Stevenson 2024). In Part One of the blog, I show that the expansion and contraction of debt in Ireland, relative to the growth of the Irish economy, explains most, if not all, of the housing bubble that occurred in Ireland in the 2000s, as well as the severe economic contraction in Ireland that occurred after the housing bubble popped. In this part (Part Two), I examine how the predictions of the ECH fit the credit and banking crisis in Ireland during this period.
The Excess Capital Hypothesis
The ECH holds that credit crises are eruptions of losses that result from excessive lending occurring in discrete periods or waves. Excessive lending drives debt capital to increasingly risky borrowers, who default and generate losses. In response, lenders become risk-averse, cease to lend to risky customers, withdraw loans from the credit markets, and invest their capital in riskless securities. By withdrawing capital from the markets, banks precipitate more defaults.
The Excess Capital Hypothesis (ECH) makes 10 predictions that can be tested:
1. There is a temporal cycle of debt, including an expansive and ebullient period in which debt grows faster than GDP (“excess capital”) that is followed by a period of credit contraction during which debt grows more slowly than the economy.
2. In periods of excess capital, banks weaken lending and underwriting standards.
3. In these periods, lenders also expand risk selection by increasingly offering loans to non-investment grade and subprime borrowers.
4. Financial leverage increases for borrowers during periods of liquidity and excess capital, especially among weaker borrowers.
5. As leverage increases in the economy, demand for risky assets (e.g., real estate, equities) increases, and the values of these assets rise, giving impetus for more lending.[4]
6. Defaults emerge and they increase quickly; losses on defaulted loans build.
7. Banks become risk-averse and less tolerant of credit risk as defaults and losses increase. They narrow borrower selection, tighten underwriting standards, and invest more capital in low-risk borrowers and in low-risk and riskless assets.
8. Loans to the riskiest borrowers are not renewed, which, in turn, causes a capital crunch for those customers. There are further defaults.
9. Capital contraction causes demand for risky assets to fall; prices for these assets decline.
10. There is a lagged correlation between the rate of loan growth and the level of loan defaults and charge-offs.
The Credit Crisis of the Late 2000s in Ireland
Ireland’s largest economic recession began in 2008 when the national Gross Domestic Product (GDP) contracted for the first time for the first time since 1995. The country remained in recession in 2009: for the full year 2009, GDP shrank by 3.7 percent. Although GDP grew in 2010 (0.1 percent) and 2011 (0.7 percent), it shrank again in 2012 (-2.1 percent).
Before this collapse, dramatic growth in domestic debt[5] occurred in Ireland, particularly in relation to GDP (Figure 1). Domestic loans scaled to GDP grew steadily from 1960 to 1994. From 1994 through 2009, this measure of debt to GDP (“credit to GDP”, “debt to GDP”, “credit / GDP” or “debt / GDP”) nearly quadrupled (43.4 percent in 1994 and 169.3 percent in 2009, a 3.9x increase).
Figure 1. Domestic Credit Divided by Gross Domestic Product in Ireland, 1960 to 2022.
(Source: World Bank. Note: data for 1999 and 2000 are not available.)
Domestic debt to GDP collapsed, beginning in 2009, and continuing through 2022 (Figure 1). The Irish economy underwent a massive deleveraging during and following this recession.
In Ireland, the increase in excess capital in the 2000s was strongly correlated to a bubble in home prices (Figure 2; see also Part One of this blog). From 1Q1970 to 2Q1996, housing prices grew linearly, but from 3Q1996 though 2Q2007, they more than quadrupled. In the expansion phase of this bubble, home price inflation was three to seven times greater than inflation in other sectors of the Irish economy (Drudy & Collins, 2011).
From 2Q2007, prices crashed (-54.5%), reaching a bottom in the first quarter of 2013 (Figure 2).
Figure 2. Domestic Credit Divided by Gross Domestic Product Compared to the Residential Property Price Index in Ireland, 1990 to 2015.
(Sources: World Bank and Federal Reserve Bank of St. Louis. Note: data on Domestic Credit / GDP are not available for 1999 and 2000.)
Domestic credit divided by GDP was strongly correlated with the index of home prices in Ireland[6], especially during the 2000s. Growth in domestic credit (including financing for residential real estate) began to outpace economic growth in Ireland in 1995 and continued to exceed GDP growth through 2009. Growth in the Irish home price index began to accelerate in 1996 and continued through 2007, creating the growth phase of this financial bubble.
According to Storey (2011), a huge increase in household debt in Ireland, which is a subset of domestic credit, was the primary fuel for the property price bubble[7]. From 2003 to 2008, household debt nearly tripled, rising from €57 billion to €157 billion. The majority of this household debt was residential mortgages, which also tripled from 2003 (€44 billion) to 2008 (€128 billion) (also see Baudino et al., 2020). Kelly (2009) also argued that credit expansion in Ireland in the 2000s caused property values to rise and subsequent contraction in credit caused them to fall.
Loans to developers of residential properties played a significant role in this growth and the number of new homes built in Ireland in the 2000s was staggering. In 2000, there were 1.25 million households in Ireland. From 2000 to 2008, more than 0.6 million new homes were completed, representing nearly one-half of the households present in 2000 (48.8 percent; see Whelan, 2010; Quinn and Turner, 2020). Irish households numbered 1.67 million in 2010, implying that some, but not all, of the newly completed homes were absorbed by Ireland’s households. Kitchen et al. (2015) noted that, in the period 2002 to 2006, more housing units were built than households were formed in Ireland.
This heavy emphasis on real estate construction created two important results in Ireland. The first was a significant shift in employment to the construction sector, causing a reduction in the persistently high unemployment in Ireland. In 2007, construction accounted for 13.3 percent of all employment, the highest share in the OECD by almost five percentage points (Wheelan, 2010).
Second, the explosion of new homes in Ireland exacerbated a pre-existing situation: unoccupied homes. According to the Central Statistics Office in Ireland, the number of vacant dwellings doubled from 2002 to 2011 (143,418 to 289,451)[8].
Building new homes should have contributed to a decline in the price of homes in the 2000s, particularly since the number of unoccupied homes increased so dramatically. Yet, prices increased dramatically as seen in the financial bubble shown in Figure 2.
I believe that the ECH explains all of these dynamics, including why home prices rose in the face of over-building and why they fell when the bubble burst. In the rest of this article, I will examine the predictions of the ECH in light of the Irish credit crisis.
Debt Expands and Contracts in a Cyclical Pattern Relative to Gross Domestic Product
The ECH holds that there are temporal cycles of debt, including expansive and ebullient periods during which debt grows faster than GDP (excess capital) followed by periods of credit contraction during which debt grows more slowly than economic growth.
Figures 1 and 2 demonstrate that a pattern of expansion and contraction of domestic debt to GDP occurred in the 1990s and 2000s in Ireland. Because most of this domestic debt was loans used to finance the purchase of homes or to build new homes, it is not surprising that demand for homes should increase with the expansion of this debt capital, especially since the debt was growing much faster than the overall economy. When debt contracted and grew more slowly than GDP, prices fell, albeit with a temporal lag (Figure 2).
Lyons and Muelbauer (2013) examined credit conditions during the inflation and deflation of Irish housing prices in the 2000s. For these researchers, the ratio of mortgage credit to aggregate deposits at individual Irish banks was a proxy for the banks’ willingness to loosen or tighten underwriting standards on mortgage loans. When the ratio increased, mortgage portfolios (assets on a bank’s balance sheet) grew faster than deposits (a liability); when the ratio declined, the loan portfolios shrank relative to deposits. These researchers asserted that increases in this ratio resulted from loosened underwriting standards for mortgages and decreases in the ratio resulted from tightened standards.
Lyons and Muelbauer (2013) discovered that almost all of the price increase in Irish homes during the expansion of the bubble (2001 to 2007) was attributed to two measures of credit conditions: (1) the mortgage credit to deposits ratio (see above), which was growing and (2) loan-to-value ratios for first-time home buyers, which also was increasing.
In Periods of Excess Capital, Banks Weaken Lending and Underwriting Standards
The ECH also predicts that excess capital emerges in the economy when banks relax lending and underwriting standards on loans to borrowers that would not receive credit in periods of normal or lowered liquidity.
When underwriting standards are loosened, existing borrowers can receive more favorable terms on their loans, including larger borrowing limits. New borrowers, previously excluded from the debt markets, may be funded. In short, lenders take greater risk.
When underwriting standards are tightened, borrowers face more challenging terms on their loans, including lower limits, and thus, lenders take less risk and withdraw capital from the markets.
Stevenson (2014, 2024) reported a cyclical pattern of loose credit standards at U.S. banks associated with aggressive lending and loan growth exceeding economic growth, followed by tightened credit standards and lowered loan growth. There is also a strong correlation between the tightening of credit standards by U.S. banks and subsequent defaults by borrowers who received credit in a period of excess capital and lax lending standards.
This phenomenon of easing credit standards was associated with lending exceeding economic growth in other countries during the early 2000s, including Spain (Stevenson, 2016b, 2024). Doubtful loans exploded in the second half of the 2000s in Spain: beginning in 2007, peaking in 2014, and persisting at relatively high levels through 2020. Bankers tightened credit standards dramatically in response to these doubtful loans, beginning in early 2007, reaching historically high levels in the second half of 2007 and all of 2008 before relaxing somewhat in 2009 (Stevenson, 2016b, 2024).
The Central Bank of Ireland participates in the quarterly survey on lending and underwriting practices conducted by the European Central Bank with all national banks in the euro area. This Bank Lending Survey seeks to obtain information on lending conditions and the changes in banks’ supplies of loans.
Figure 3 presents the survey results on credit standards among Irish banks as reported for the period 4Q2002 to 4Q2023. Panel A of Figure 3 shows the tightening (positive values for the diffusion index) and loosening (negative values) of credit standards on loans made by Irish banks for purchase of a house. Panel B shows banks’ loosening and tightening of credit standards on loans to enterprises.
A striking cyclical pattern appears in these self-reported results in which credit standards on loans to purchase homes were loosened by Irish banks from late 2002 to the third quarter of 2007, and especially in 2004 (Figure 3 Panel A). In late 2007, there was a dramatic tightening of those standards that persisted through 2008, with somewhat lower tightening through mid-2012.
The temporal pattern of underwriting standards on loans to enterprises is broadly similar: easing in 2003 and 2004 and dramatic tightening from 2Q2007 to 2Q2010 (Figure 3 Panel B).
Figure 3 Panel A. Changes in Credit Standards for Approvals of Loans to Purchase Houses Made by Irish Banks, 4Q2002 to 4Q2023
(Source: European Central Bank)
Figure 3 Panel B. Changes in Credit Standards for Approvals for Loans to Enterprises Made by Irish Banks, 4Q2002 to 4Q2023
(Source: European Central Bank)
Several other researchers have highlighted the role of relaxed underwriting standards by Irish banks in the mid-2000s (Kelly, 2010; Drudy & Collins, 2011; Baudino et al., 2020; Quinn & Turner, 2020;). One example of these loosened standards was approvals of mortgages with increasingly larger loan-to-value (LTV) ratios. For example, Irish banks introduced mortgages with 100 percent LTV in 2004; by 2008, such loans made up 12 percent of all new mortgages (Norris and Coates, 2011).
A low value for LTV indicates that the borrower has invested a significant amount of his own capital (“equity”) in the purchase of the home. A high value for LTV implies low equity investment. An LTV of 100% means that the loan amount is equal to the home’s value and the equity investment is zero (0).
LTV is an important indicator of the likelihood that a borrower will default on his mortgage. Since the lender has a security interest in the home through the mortgage, it can foreclose on, and take possession of, the home after the borrower defaults. The lender typically avoids foreclosure since it increases its costs. The borrower also avoids foreclosure since it leads to loss of his home, increases his costs, and adversely impacts his credit history and credit score.
However, as LTV increases, the borrower’s incentive to default on the mortgage increases since his equity is shrinking. Accordingly, the borrower can “put” his loan back to the lender by defaulting on interest and principal payments[9]. Default has adverse impacts, as noted above, but it frees the borrower from further financial obligations on a home in which he has little or no equity capital.
Many studies demonstrate that, in the U.S. residential mortgage market, LTV is correlated with historical default rates on those mortgages (e.g., Kelly, 2009; Ozdemir, 2024); default is most likely when a borrower’s total debt exceeds the value of his home (“negative equity”) (Elul et al., 2010; Cambell & Coco, 2011).
The correlation of LTV with mortgage default also holds in Europe and the United Kingdom (e.g., Kelly 2009). For example, Gonzalez et al (2016) analyzed residential mortgages securitized in Spain from 2005 to 2008 and found that higher LTV ratios were correlated with greater default risk. The relation between the probability of default and LTV was nonlinear and there was a sharp increase in default when LTV was greater than 80 percent.
McCann (2014) examined those variables that were correlated with default on residential mortgages in the United Kingdom by banks headquartered in Ireland. LTV was an important predictor of the probability of default on these mortgages, especially on Buy-to-Let loans, over the period December 2009 to December 2013. As LTV increased, the probability of default increased.
LTV also is an important correlate with the loss incurred on a mortgage once a default has taken place (so-called “loss-given-default” [LGD]) (Qi and Yang, 2007; Greve and Hahnenstein, 2016; Zhang et al., 2010). In a foreclosure, a bank that has repossessed the home which is security on a defaulted mortgage attempts to sell the home. Once the home is sold, the bank uses the sale proceeds to offset the remaining unpaid balance on the mortgage.
If the home sells for less than the unpaid balance, the difference (loan balance minus proceeds of home sale) is the loss realized in the default (a positive value for LGD). If the home sells for more than the loan balance, the LGD is negative and if the sale price is equal to the loan value, the LGD is zero[10].
Not surprisingly, banks consider LTV to be an important criterion on which they underwrite a residential mortgage. Most U.S. lenders require an LTV of 80% or less when the mortgage is originated. That level is also typical of lenders in the United Kingdom and Europe.
When a lender approves a mortgage with an LTV greater than 80 percent, it has loosened the standard level for this underwriting criterion. When it requires an LTV of less than 80 on a newly originated mortgage, it has tightened relative to the standard level.
As noted above and in the next section, Irish banks loosened the LTV criterion dramatically in the mid-2000s, so much so that 12 percent of all residential mortgages originated in 2008 required no down payment by the borrower (i.e., LTV of 100 percent) (Norris and Coates, 2011).
In this “go-go” period, Irish banks also introduced new, increasingly risky types of residential property loans, including interest-only mortgages in which borrowers paid only interest in their regular payments, leaving repayment of borrowed principal to the very end of the loan’s life (Norris & Coates, 2011). In addition, “buy-to-let” transactions, in which an investor takes out a mortgage to purchase a rental property, became common.
During this time, Irish banks focused on growth and market share and this growth was facilitated by the inflow of cheap capital from overseas markets (Baudino et al., 2020). These banks lent heavily to property developers as well as individuals for home purchases (e.g., mortgages).
Relaxation of underwriting standards and introduction of risky types of loans in Ireland mirrored similar behavior of U.S. banks in the heyday of subprime mortgage lending in the mid-2000s. As I document in Credit Crises (Stevenson, 2024), loosened underwriting standards by U.S. banks permitted subprime borrowers to obtain mortgages at this time, even though there was clear evidence that, historically, subprime borrowers default at rates that were many multiples of the default rates of prime borrowers.[11] The credit risk of those subprime borrowers became evident when default and foreclosure rates exploded from 2007 to 2009 (the subprime mortgage crisis).
The cyclical waxing and waning of credit standards in Figure 3 is important for two other reasons. First, the pattern seen in Ireland is quite similar to that seen in U.S. banks and Spanish banks (Stevenson, 2016b, 2024), and likely is common among all banks in Western economies impacted by the mortgage crises that preceded the Global Financial Crisis.
Second, there is a strong association between the tightening of credit standards following a period of loose lending and excess capital and the dramatic increases in doubtful loans in Spain and in the U.S. (Stevenson, 2016b, 2024). The ECH holds that loans made during lax lending standards and excess capital become the most likely to default when banks cut back on lending. Those borrowers have the highest probability of default and, if external capital is restrained or their operating profits fall, they are likely the first to default, sometimes creating a credit crisis.
In Periods of Excess Capital, Lenders Expand Risk Selection
The ECH posits that in the ebullient and expansive phase of a bubble when excess capital is formed, lenders expand their choices of borrowers (“risk selection”) to include non-investment grade or subprime borrowers who would not receive credit in less expansive markets (Stevenson, 2024). For example, the subprime mortgage crisis in the United States grew out of the expansion of risk selection to include active lending by U.S. banks and mortgage companies to borrowers with credit scores below 620 (subprime). The elevated default probabilities of these borrowers became exceptionally high default rates in 2007 to 2009. Ultimately, numerous lenders failed and the subsequent Financial Crisis wrought havoc throughout Western economies.
I have been unable to find historical data on the credit quality of Irish borrowers. Therefore, I cannot test whether lenders offered credit to weaker borrowers during the expansive phase of this bubble and withdrew that credit in the contracting phase. Nevertheless, the literature provides evidence of expanding risk selection in Ireland in the 2000s.
As I documented in the previous section, the loosening of credit standards and the introduction of new loan products by Irish banks allowed them to lend to customers who did not previously have access to residential mortgages (Baudino et al., 2020). In particular, interest-only mortgages and mortgages with 100 percent LTV brought “marginal” retail borrowers into the market (Baudino et al., 2020).
First-time home buyers played a significant role in the housing bubble in Ireland. While there are no data concerning the proportion of first-time buyers (FTBs) in the Irish housing market in this period, there is evidence that the highly liquid nature of the market gave FTBs opportunities to purchase homes but made these individuals riskier than buyers making repeat purchases of homes.
FTBs have no history of repaying mortgages and, thus, may be considered riskier than borrowers who have repaid at least one previous mortgage. Repayment of past loans is considered positively by lenders. In the United States, for example, a history of repaying loans is a positive factor in a borrower’s credit score.
FTBs in Ireland during the 2000s had riskier mortgages than did the full population of buyers of Irish residential properties. Regling and Watson (2010)[12] found that FTBs took out mortgages with much higher loan-to-value ratios (LTVs) at the height of the Irish housing bubble (2004 to 2007) compared to the total pool of newly originated Irish mortgages in the same period. Specifically, more than half of mortgages issued to FTBs had LTVs of 91 percent or more (i.e., down payments of nine percent or less) whereas only approximately 30 percent of all mortgages issued in this period had LTVs of 91 percent or more. (See also Honohan, 2009.)
Conversely, one-third to 40 percent of all new mortgages in Ireland issued in this period had LTVs of 70 percent or less (i.e., down payments of 30 percent or more) (Honohan, 2009; Regling & Watson, 2010). However, less than 20 percent of the new mortgages issued to FTBs had LTV of 70 percent or less.
Kelly (2009, 2010) reported on the LTVs of mortgages taken out by FTBs purchasing new homes in Ireland in 2006, near the peak of the housing bubble (Figure 2). Only 24 percent of these mortgages had LTVs of 80 percent or less[13] and 64 percent had LTVs above 90 percent. Thirty percent of these 2006 mortgages used by FTBs to purchase new homes had LTVs of 100 percent (i.e., no downpayment at all).
Rising prices in the housing bubble created incentives for this risk-taking. According to Kelly (2009), the typical Irish FTB in 1995 took out a mortgage equal to three years of earnings. In 2006, the average FTB took a mortgage equal to eight times his average earnings. Given that the FTB had no equity capital earned from a previous investment housing, the down payment he made on his first home purchase likely came from cash savings. Younger FTBs likely had limited cash savings and, thus, probably took out a high LTV mortgage to afford a home in the face of rising home prices.
When prices began to fall in 2007, LTVs in Ireland skyrocketed. Initially in the housing boom, the number of Irish mortgages that had “negative equity” (i.e., when the loan amount is greater than the market value of the home) was relatively small but this number rapidly increased. For example, at the end of 2007, fewer than 25,000 Irish homeowners had negative equity (Duffy et al., 2014). However, the number of negative-equity mortgages increased geometrically as home prices fell and, by the end of 2012, the number exceeded 314,000.
Homeowners with negative equity mortgages are highly likely to default on those mortgages. Not surprisingly, the period in which the number of negative equity mortgages spiked was the same period in which delinquencies and defaults also spiked (see Figure 4). Specifically, FTBs who bought at the peak of the market prices in 2006 and 2007, using mortgages with high LTVs, felt the greatest pressure to default on their mortgages, and suffered the greatest loss, when prices collapsed from 2008 to 2011.
Financial Leverage Increases for Borrowers During Periods of Liquidity and Excess Capital, Especially Among Weaker Borrowers
In the previous section, I highlighted the widespread use of high LTV mortgages by FTBs, especially 100 percent LTV mortgages. By definition, high LTV mortgages mean high financial leverage since such loans imply little or no capital investment. While we have no data on whether FTBs were financially weaker than repeat borrowers, we know that repeat buyers typically have greater equity to invest in their next home. Specifically, in periods when house prices are rising, repeat buyers accumulate equity in their existing home that can reduce the LTV on their next home (Lyons & Muelbauer, 2013). The accumulated equity gives financial strength, and less leverage, to repeat buyers that FTBs lack.
As Leverage Increases in the Economy, Demand for Risky Assets Increases and the Values of These Assets Rise
I have shown a close correlation between the increase in domestic credit/ GDP and the prices of homes in Ireland from 1990 to 2015 (Figure 2). As the credit / GDP ratio increased, excess capital increased in the Irish economy and much of that excess capital was invested in homes or in their development. Credit-induced demand drove property prices higher.
Defaults Emerge and Increase Quickly
In 2007 in the United States, the subprime mortgage crisis began as defaults emerged in the residential mortgage markets. Further defaults quickly cascaded and, by late 2007, it became clear that securitizations of these mortgages would suffer significant losses, including at the AAA level. The mortgage markets froze as both lenders and investors became extremely risk-averse and abandoned the subprime market.
A similar increase in the level of non-performing loans occurred in Ireland, although the start occurred in 2008 and the rapid increase occurred from 2009 to 2013 (Figure 4).
Figure 4. Non-Performing Loans as a Percent of Gross Loans in Ireland, 2001 to 2021
(Source: World Bank)
Banks Become Risk-Averse and Less Tolerant of Credit Risk as Defaults and Losses Increase
The ECH holds that, as defaults increase and lenders experience losses, they become risk-averse and less tolerant of credit risk. They narrow borrower selection, tighten underwriting standards, and invest more capital in low-risk borrowers and in low-risk and riskless assets. Figure 3 Panel A shows that Irish bankers began tightening underwriting standards on loans to purchase homes in 2007 and 2008, precisely when delinquencies began to increase. This is evidence of risk aversion.
Risk aversion is a common reaction to market crises (Stevenson, 2016a, 2024). Human beings participating in the financial markets are dynamic, and their individual and collective behaviors play key roles in shaping the markets. Such behavior can appear erratic and irrational, with over-optimism leading to myopia about predictable, but low-probability, adverse events. Once such events occur, however, humans react strongly, typically withdrawing from the perceived source. If the adverse event is previously unknown, people can overreact, magnifying the perception of adverse outcomes into “disasters,” with adverse outcomes seen as more likely than their true probabilities.
Capital Contraction Causes Demand For, and Prices Of, Risky Assets to Fall
Stevenson (2024) demonstrated that, for U.S. banks, increasing delinquencies and charge-offs on loans signal bankers to tighten underwriting standards, which has the effect of withdrawing capital from the market. Demand, induced by the excess capital that results from loose underwriting standards, shrinks and asset prices, including home prices, contract.
Figure 5 shows the pattern of domestic credit provided to the private sector as a percent of GDP in the Irish economy from 1980 to 2022 compared to the incidence of doubtful loans. When scaled to GDP, the strikingly rapid growth in domestic credit becomes apparent. In 1980, domestic credit provided to the private sector was 40.6 percent of GDP; by 2009, it had more than quadrupled to 169.3 percent. Growth in indebtedness far outpaced economic growth during this period.
Figure 5. Domestic Credit Held by Irish Banks / Gross Domestic Product Compared to Non-Performing Loans / Gross Loans at Irish Banks, 1980 to 2022
(Source: World Bank)
The careful reader will note that 2009 was a pivotal year in the Irish credit crisis. In 2009, non-performing loans more than quadrupled to their peak value as a percent of gross loans (9.8 percent in 2009 versus 1.9 percent in 2008) (Figure 5). From 2009 onward, both debt / GDP and non-performing loans fell through 2021. Importantly, though, the Irish economy shed domestic debt faster than it shed non-performing loans. Specifically, domestic debt / GDP fell by 142 percentage points in this period (domestic debt / GDP in 2009 = 169.3 percent and domestic / debt in 2021 = 27.4 percent). In the 2010s, Ireland de-levered quickly.
In Figure 6, I plot Ireland's residential property price index with the ratio of doubtful loans to gross loans from 2000 to 2022. There is a clear negative association between the incidence of doubtful loans and the price index. When doubtful loans were low in the early- and mid-2000s and credit was extended liberally to borrowers, the demand induced by this excess capital drove home prices higher to a maximum in 2007 (Figure 6).
Figure 6. Home Price Index Compared to the Incidence of Non-Performing Loans Among Total Loans for Ireland, 2000 to 2022.
(Sources: Federal Reserve Bank of St. Louis and World Bank)
When doubtful loans started to increase as a percentage of total loans between 2007 and 2008, property prices began to fall (Figure 6). This decline continued through 2012 when home prices began to increase again. The incidence of non-performing loans declined, and the home price index increased, through 2021. From 2007 to 2021, the correlation between these two variables was almost perfectly negative[14].
The increase in delinquencies likely caused this increase in risk aversion. Tightened underwriting standards began in 2007 (loans to enterprises; Figure 3B) and 2008 (loans for home purchases; Figure 3A); they persisted through 2010.
Thus, tightened underwriting standards preceded the peak in domestic debt / GDP (2009) by one to two years indicating that excess capital was being withdrawn from the market very rapidly at this time. Since nominal GDP (the denominator of domestic debt / GDP) shrank in 2009 and 2010, the fact that the debt-to-GDP ratio began falling in 2010 (see Figure 1) meant that domestic debt was falling even faster due to significantly reduced loan originations resulting from the tight underwriting standards. Consequently, the demand induced by liberal underwriting shrank, and home prices fell.
There is a Lagged Correlation Between Loan Growth and Loan Defaults
The ECH holds that when the growth of outstanding loans exceeds economic growth, excess capital is formed and flows to increasingly risky borrowers who eventually default and produce losses to lenders. There are temporal lags between the emergence of excess capital and charge-offs taken by lenders for two reasons. First, loan defaults do not occur instantaneously with the emergence of excess capital because borrowers repay their loans until they cannot. Second, it takes time for lenders to recognize that repayment is doubtful for non-performing loans and to take charge-offs on those doubtful loans.
As noted previously, the ECH holds that loan growth is best understood when scaled to GDP: when the rate of loan growth exceeds economic growth, banks will lend to increasingly risky borrowers whose increasingly greater probabilities of default will eventually lead to loan losses (Stevenson, 1994, 1995, 2010, 2024).
The ECH holds also that there are correlations between loan growth rates and the level of non-performing loans and between loan growth rates and net loan charge-offs with temporal lags. That is, rapid growth in loans relative to GDP occurs earlier than growth in defaults and charge-offs because the loan growth exceeding economic growth means that loan capital likely flows to riskier borrowers. Default probabilities increase exponentially with this flow of capital and periodically manifest themselves in actual defaults that produce credit crises. The greater the excess capital lent to these non-investment grade borrowers, the more significant the crisis.
In the United States, the lags between loan growth and loan losses average 18 months to two years (Stevenson, 1994, 1995, 2010, 2024). In Ireland, there are long lags between rapid loan growth and the emergence of troubled and doubtful loans; the lags range from four to six years[15]
Similar patterns between loan growth and doubtful loans exist in Spain, including similar lag structures (Jiminez and Saurina, 2006; Salas and Saurina, 2002; Stevenson, 2016b, 2024).
Conclusions
The ECH explains well the dynamics of the 2007–2012 credit crisis in Ireland and the predictions of the ECH were borne out during this period. Lax credit standards in the first half of the 2000s gave rise to excessive lending, particularly in the markets for home mortgages and property development, and that growth in lending far outpaced the growth of the Irish economy. When banks finally became aware of the risk in the economy and in the housing market, they significantly tightened credit standards, effectively withdrawing excess capital from the market. This tightening was strongly correlated with a dramatic increase in the level of doubtful loans. The result was a lagged relationship between outstanding loans scaled to GDP and the incidence of doubtful loans, ranging from four to six years. These results are consistent with observations made by earlier analysts of the banking systems of the United States and Spain.
References
Baudino, P., Murphy, D. & Svoronos, J.-P. (2020). The banking crisis in Ireland. Bank for International Settlements, FSI Crisis Management Series No. 2
Cambell, J. Y. & J. F. Cocco. (2011). A model of mortgage default. National Bureau of Economic Research, Working Paper 17516. https://www. nber.org/papers/w17516
Drudy, P. J. & Collins, M. L. (2011). Ireland: From boon to austerity. Cambridge Journal of Regions, Economy and Society, 1-16. doi:10.1093/cjres/rsr021
Duffy, D., J. FitzGerald, K. McQuinn, D. Byrne, & C. Morley. (2014). Quarterly economic commentary. Economic and Social Research Institute, Dublin. 105 pp.
Elul, R., N. S. Souleles, S. Chomsisengphet, D. Glennon, & R. Hunt. (2010). What “triggers” mortgage default? Research Department, Federal Reserve Bank of Philadelphia. Working Paper 10-13. https:// finance.wharton.upenn.edu/~souleles/research/papers/PhilaFedwp10-13.pdf
González, L. O., P. D. Santomil, M. V. Búa & R. L. Sestayo. (2016). The impact of loan-to-value on the default rate of residential mortgage-backed securities. Journal of Credit Risk 12 (4): 1-13. DOI: 10.21314/JCR.2016.210
Greve, C. & L. Hahnenstein. (2016). Benchmarking the loss given default parameter for mortgage loan portfolios under stress. Journal of Credit Risk 12(4): 79-107. DOI: 10.21314/JCR.2016.217
Honohan, P. (2009). Resolving Ireland’s Banking Crisis, The Economic and Social Review 40(2), 207–231.
Jiminez, G. & Saurina, J. (2006). Credit cycles, credit risk and prudential regulation. International Journal of Central Banking, 2(2), 65–98. https://www.ijcb.org/journal/ijcb06q2a3.pdf
Kelly, M. (2009). The Irish credit bubble. University College Dublin, School of Economics Centre for Economic Research Working Paper Series WP09/32. 27 pp. https://www.ucd.ie/t4cms/wp09.32.pdf
Kelly, M. (2010). Whatever happened to Ireland? Centre for Economic Policy Research Discussion Paper No 7811.
Kitchen, R., Hearne, R. & O’Callaghan, C. (2015). Housing in Ireland: From crisis to crisis. National University of Ireland Maynooth, National Institute for Regional and Spatial Analysis. NIRSA Working Paper Series No. 77.
Lyons, R. C. & Muelbauer, J. (2013). Explaining the bubble: House prices, user-cost and credit conditions in Ireland, 1975-2012. Unpublished manuscript, Nuffield College, Oxford.
McCann, F. (2014). Modelling default transitions in the UK mortgage market. Research Technical Papers 18/RT/14, Central Bank of Ireland.
Norris, M. & Coates, D. (2011). Mortgage availability, qualifications and risks in Ireland, 2000-2009. Journal of Current Issues in Finance, Business and Economics 4 (3): 191-206.
Ozdemir, B. (2024). The blind spot in residential mortgages: Increasing default option value in the face of declining house prices. J. Risk Management in Financial Institutions 17 (3): 258-285. https://www.ingentaconnect.com/contentone/hsp/jrmfi/2024/00000017/00000003/art00004
Qi, M & Yang, X. (2007). Loss given default of high loan-to-value residential mortgages. Office of the Comptroller of the Currency, OCC Working Paper 2007-4. 36 pp.
Quinn, W. & Turner, J. D. (2020). Boom and Bust: A Global History of Financial Bubbles. Cambridge University Press, Cambridge, UK. 288 pp.
Regling, K & M Watson. (2010). A preliminary report on the sources of Ireland’s banking crisis, report commissioned by the Minister for Finance of Ireland, June.
Salas, V., & Saurina, J. (2002). Credit risk in two institutional regimes: Spanish commercial and savings banks. Journal of Financial Services Research, 22(3), 203–224. https://doi.org/10.1023/A:1019781109676
Stevenson, B. G. (1994). Research report: Capital flows and loan losses in commercial banking. Journal of Commercial Lending, 77(1), 18–26.
Stevenson, B. G. (1995). Capital flows and the cycles of losses in commercial real estate. Real Estate Review, 25(2), 43–49.
Stevenson, B. G. (2010). Credit crises: The excess capital hypothesis. Bank Accounting & Finance, (October-November), 39–51.
Stevenson, B. G. (2014). The excess capital hypothesis and cyclical changes in loan underwriting. RMA Journal, (April), 48–55.
Stevenson, B. G. (2016a). Stress testing, human behavior and the management of capital in the post-crisis world. Part I. Human behavior in financial crises. Alvarez & Marsal Readings in Quantitative Risk Management, 1-9. https:/www.alvarezandmarsal.com/sites/default/files/sidebar-callout/am-fias/stresstesthumanbehavior_05singles_final.pdf
Stevenson, B. G. (2016b). The excess capital hypothesis and the experience of Spanish banks from 1999 to 2013. Alvarez & Marsal Readings in Quantitative Risk Management, 1-12. https://www.alvarezandmarsal.com/sites/default/files/am_fias_excesscapitalhypothesis_final.pdf
Stevenson, B. G. (2024). Credit Crises: The Role of Excess Capital. Gatekeeper Press, Tampa, Florida. 460 pp.
Wheelan, K. (2010). Policy lessons from Ireland’s latest depression. The Economic and Social Review 41(2): 225-254.
Zhang, Y., Lu, J. & F. Liu (2010). Local housing market cycle and loss given default: Evidence from sub-prime residential mortgages. International Monetary Fund, Working Paper WP/10/167.
[1] Source: St. Louis Federal Reserve. fred.stlouisfed.org/series/BOGZ1FL192090005Q. Last accessed 28 September 2024.
[2] Source: Yahoo Finance. finance.yahoo.com/quote/%5EGSPC/history?period1=-1325635200&period2=1693267200&interval=1mo&filter=history&frequency=1mo&includeAdjustedClose=true. Last accessed 28 September 2024.
[3] www.bloomberg.com/view/articles/2013-09-26/three-unlearned-lessons-from-the-financial-crisis#xj4y7vzkg. Last accessed 28 September 2024
[4] As asset values increase, lenders may rely more on collateral values when making loans and less on a borrower’s fundamental ability to repay. This shift reflects an optimism fueled by the expansive nature of the market.
[5] Domestic or internal debt is debt incurred within a country and typically owed to lenders within the country.
[6] Over the period 1990 to 2015, the instantaneous Pearson correlation between the residential property index and domestic credit / GDP is 0.860.
[7] Andy Storey, “Ireland’s Debt Crisis Roots and Reactions”, November 2011. https://www.cadtm.org/Ireland-s-Debt-Crisis-Roots-and. Last accessed 18 June 2024.
[8] https://www.cso.ie/ en/releasesandpublications/ep/p-cp1hii/cp1hii/vac/. Last accessed 21 September 2024.
[9] Mortgage default is a put option that permits a homeowner to sell his or her house to the lender in exchange for mortgage elimination. The exercise price of this put is the value of the mortgage.
[10] In these comparisons, I exclude expenses required for managing a defaulted mortgage and for the foreclosure of that mortgage. These costs would be added to the loan value if a full accounting of the LGD were required.
[11] Delinquency and default rates on consumer loans, including mortgages, increase geometrically by declining credit score. Prime borrowers, whose credit scores typically are 660 or above, have default rates of 0.5 percent per year or less. Subprime customers, whose credit scores are less than 620, have default rates exceeding 5.0 percent per year and sometimes reaching 25.0 percent per year.
[12] As reported in Baudino et al. (2020).
[13] A mortgage that requires the borrower to make a down payment of 20 percent or more is traditional.
[14] Between 2000 and 2021, the instantaneous Pearson correlation of the Irish home price index and the ratio of doubtful loans to gross loans is -0.590. For the period 2007 to 2021, this correlation is -0.974.
[15] The lag structure for Pearson correlations between the ratio of domestic credit provided to the private sector to GDP and the ratio of doubtful loans to total loans is -0.023 in the current period (instantaneous correlation), 0.303 with a one-year lag, 0.488 with a two-year lag, 0.676 with a three-year lag, 0.829 with a four-year lag, 0.899 with a five-year lag, 0.856 with a six-year lag, 0.747 with a seven-year lag and 0.591 with an eight-year lag.
Comments