Credit conditions in a boom and bust property market: Insights for macro-prudential policy

Accepted Manuscript Title: Credit Conditions in a Boom and Bust Property Market: Insights for Macro-Prudential Policy Author: Yvonne McCarthy Kieran McQuinn PII: DOI: Reference:

S1062-9769(16)30065-5 http://dx.doi.org/doi:10.1016/j.qref.2016.08.002 QUAECO 968

To appear in:

The

Received date: Revised date: Accepted date:

28-5-2015 1-7-2016 10-8-2016

Quarterly

Review

of

Economics

and

Finance

Please cite this article as: Yvonne McCarthy, Kieran McQuinn, Credit Conditions in a Boom and Bust Property Market: Insights for Macro-Prudential Policy, (2016), http://dx.doi.org/10.1016/j.qref.2016.08.002 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ip t

cr

Credit Conditions in a Boom and Bust Property Market: Insights for Macro-Prudential Policy

Kieran McQuinn

us

Yvonne McCarthy

Central Bank of Ireland

an

Economic and Social Research Institutek September 14, 2016 – Version 2.0 Abstract

Ac ce p

te

d

M

The substantial role played by greater credit provision in the international house price boom between 1995 to 2007 has lead to an increase in the use of macro-prudential measures as a means of preventing future credit bubbles. Calibration of these measures can prove difficult, however, given the relative ‘newness’ of the macro-prudential field and the limited experience with the use of these measures. Micro-loan level data can contribute to this debate by providing detail on how these measures tend to evolve across different buyer types and at different points in the business cycle. This paper uses a sample of 400,000 loans to assess the evolution of mortgage credit conditions across borrower groups in Ireland during the 2000s. The Irish case is particularly interesting given the severity of its housing boom and bust during this period. Furthermore, controlling for demand-side factors, we use the granular loan level data to provide estimates of an index of mortgage credit availability over the same period for the Irish mortgage market.

JEL classification: E31, R32, R38. Keywords: Credit, House Prices, Income, Interest Rates, Macro-prudential.

k The authors are economists at the Central Bank of Ireland (CBI) and the Economic and Social Research Institute (ESRI) respectively. The views expressed are not necessarily those held by the CBI or the ESRI. Any errors and omissions are the responsibility of the authors who can be contacted at [email protected] and [email protected].

Page 1 of 35

1

Introduction

Ac ce p

te

d

M

an

us

cr

ip t

Given the significant contribution of credit to the international house price boom preceding the financial crisis of 2007/08, a considerable amount of attention now focusses on the use of macro-prudential policy as an effective tool in preventing future credit bubbles. In particular, many authorities have proposed the use of limits on loan-to-value (LTV) and loan-to-income (LTI) ratios as a means of tempering excessive credit growth and house price inflation.1 The calibration of these instruments can be difficult, however, given the relative ’newness’ of the macro-prudential field and the relative lack of experience with the use of the relevant instruments. Micro-loan level data can contribute to the debate on the calibration of macro-prudential loan instruments by providing detail on how these instruments tend to evolve across different buyer types and at different points in the business cycle. However, the existing research on these issues is relatively limited. This paper contributes to the debate by examining credit conditions in the Irish mortgage market during the 2000s, an example of a country that experienced a severe house price boom and bust during this period. The analysis is based on a loan-level dataset that covers 70 per cent of the Irish mortgage market. We examine how credit conditions evolved in Ireland during the 2000s, and focus, in particular on developments across different buyer groups. In general, we observe a substantial increase in credit levels across all borrower types during the boom phase (2000 - 2007) and an equally significant contraction during the subsequent decline period (2007 - 2011). In accordance with the standard life-cycle hypothesis, younger households, in particular, appear to have availed of more liberal conditions by assuming increased amounts of mortgage credit over the period 2000 2007. In examining trends for the three most popular credit instruments, loan-to-value ratios, income fractions and mortgage duration, we are able to categorise changes based on the income quintile of the mortgaged household. As suggested by other studies of borrowing constrained households, it would appear that relatively poorer families were also very sensitive to changes in the credit regime. Given the general finding of significant variation in credit levels, we also, using a unique combination of regulatory prudential and survey data, estimate an indicator of mortgage credit supply in the Irish market, which 1

For a general overview of macroprudential policy see Galati and Moessner (2012) and Lim et al. (2011).

2

Page 2 of 35

Ac ce p

te

d

M

an

us

cr

ip t

we label the mortgage market credit indicator (MMCI). Controlling for household characteristics as well as market conditions, we are able to quantify the impact of changing credit standards in the Irish mortgage market. While studies such as Fernandez-Corugedo and Muellbauer (2006), Aron and Muellbauer (2011) and Jansen and Krogh (2011) have undertaken a similar type analysis at an aggregate, macro level, to our understanding, no study has examined credit conditions, in this manner, at the micro, household level. Understanding the determinants of credit conditions is particularly pertinent at a time when a growing debate is centred on the relative contribution of financial innovation and fundamental economic factors to house price movements.2 Furthermore, an understanding of the relative contribution of supply and demand side factors to credit developments is important when choosing macro-prudential instruments targeting credit growth. Amongst countries which experienced house price booms, the Irish case stands out. On average from 1995 until 2007 real Irish house prices grew by nearly 9 per cent per annum - the next highest country growth rate in the OECD was 7.6 per cent. During this time, the Irish economy experienced a period of sustained growth with living standards increasing substantially. Ireland, in the 1980s, witnessed negligible economic growth, an average unemployment rate of 14 per cent and high levels of personal taxation. The emergence of the so-called Celtic Tiger in the mid-1990s saw the size of the economy double over the period 1995 to 2005 with the total number of people employed in the country increasing by almost 50 per cent. Almost simultaneously, the Irish credit market experienced a considerable degree of financial liberalisation. This easing of credit standards was compounded by the ability of Irish credit institutions post-2003 to access international wholesale markets, thereby increasing significantly the supply of credit available. The already buoyant nature of the residential and commercial property markets provided considerable demand for this increased source of funding. Therefore, accurately delineating the effects of supply versus demand-side factors on credit conditions would appear to be particularly warranted in the Irish case.3 Furthermore, understanding the determinants of credit conditions is likely to be informative when considering housing market developments in countries such as Spain, the UK and the US where financial markets are also considered to have been relatively liberal and innovative in the provision of credit. 2

See, for example, Borio and McGuire (2004), Tsatsaronis and Zhu (2004), or the debate between Taylor (2009) and Dokko et al. (2011) 3 For examples of studies linking deregulation to housing booms, see Agnello and Schuknecht (2011) or Roy and Kemme (2012).

3

Page 3 of 35

d

M

an

us

cr

ip t

The costs to the Irish economy of the aftermath of the credit-fuelled property boom have been truly severe. Since 2007, Irish house price declines have been the most severe across the OECD.4 Given that nearly 40 per cent of the total stock of Irish mortgages was issued between 2004 and 2007, when prices were at their highest, the sharp subsequent decline has given rise to a significant degree of negative equity being experienced by Irish households. When coupled with the sharp increase in unemployment experienced by the Irish economy post-2008, the possibility of substantial credit risk in the mortgage books of Irish banks was one of the main reasons for the financial crisis which engulfed the Irish banking sector. Consequently, one of the major cornerstones of the 2010 program of support agreed between Ireland, the EU and the IMF was a significant commitment to deal with the degree of loan impairment on the mortgage books of Irish financial institutions. The remainder of this paper is structured as follows; Section 2 provides background information on aggregate changes in credit provision in the Irish mortgage market. In Section 3, we discuss the dataset used in the current study, the channels through which banks can provide mortgage credit to Irish households and the micro-level evidence on credit developments among Irish borrowers. In Section 4, we present our model of credit supply. Finally, Section 5 offers some concluding comments.

The availability of credit and the Irish mortgage market

Ac ce p

te

2

The Irish credit market had, since the mid-1980s, been experiencing considerable financial deregulation and liberalisation, involving the removal of credit and interest-rate controls.5 However, the most profound development in the provision of credit, from an Irish perspective, was the increased ability of Irish banks, post-euro, to attract deposits from non-residents. Given the build-up in demand side pressures in the Irish economy throughout the late 1990s, Irish financial institutions availed substantially of the increased funding available within the euro area. Overall, the combined effect was to increase the elasticity of the supply of credit to the household sector. The consequence of such a flatter supply curve was that financial institutions were able to increase the amount of lending to the household sector 4

By mid-2013, house prices had fallen by 50 per cent from the peak in mid-2007. Table A1 in the Appendix summarises the main developments from the mid-1980s to the present. Further detail is available in Kelly and Everett (2004); see, in particular, Box 1 on page 96. 5

4

Page 4 of 35

Credit liberalisation in Ireland: micro data and stylised facts

Ac ce p

3

te

d

M

an

us

cr

ip t

with little upward pressure on the interest rate. However, this flatter supply curve, inevitably, lead to a substantial increase in debt levels within the Irish economy. Figure 1 details the source of funding for Irish resident credit institutions from 2001 onwards along with the difference between credit extended and the deposit base in the Irish financial system. The rapid increase in debt securities issued by Irish credit institutions post-2003 can, especially, be observed. The resulting substantial gap between lending and deposits underpinned the vulnerability of the Irish banking sector to the severe distress observed in these markets during the financial crisis.6 The net consequence of the increase in mortgage credit can be observed in Table 1. The total value of mortgages issued increased threefold between 2000 and 2005. The total number of new mortgages went from just under 50,000 in 1995, to 80,000 in 2000 and to over 120,000 mortgages by 2005. The average size of a mortgage also increased considerably over the period. In 1995 the average mortgage extended by an Irish credit institution was e54,094, by 2005, this had climbed to e231,206. Inevitably with such an expansion in credit, house prices increased substantially over the period. Between 2000 and 2007, prices rose by almost 65 per cent. The peak in house prices occurred in mid-2007 and since then the residential market has witnessed a substantial decline in activity as both housing supply and prices have fallen considerably. In the next section, we discuss the micro elements of these aggregate developments.

3.1

The dataset

To examine the aggregate developments in credit in further detail, we rely on a new loan-level dataset that covers three Irish banks, which account for about 70 per cent of the volume of mortgage lending in Ireland.7 This dataset was collected by the Central Bank of Ireland as part of a Prudential Capital Assessment Review exercise that was designed to assess the potential 6

Lane (2014) provides a thorough exploration of funding by Irish banks during the economic boom period in the 2000s. He notes that the euro-share of external liabilities of domestic banks fell between 2004 and 2008 while the US dollar share increased dramatically from 2005. 7 The included banks are Allied Irish Bank (including the Educational Building Society portfolio), Bank of Ireland and Irish Life and Permanent.

5

Page 5 of 35

3.2

us

cr

ip t

capital requirements of the Irish banks under various stress scenarios. The dataset includes a snapshot of the entire residential mortgage books of the three banks at June 2012, and provides a wealth of information on the stock of loans outstanding in the Irish economy. Table 2 provides an overview of the information contained in the dataset. Crucially, for the purposes of this paper, the dataset includes information from the point of loan origination on the mortgage amount, the terms applying to the loan (interest rate, LTV, etc.) and borrower characteristics at the point of loan application. The analysis in this section is based on almost 400,000 loans from this dataset.8

How did banks extend credit?

Ac ce p

te

d

M

an

There are three main channels through which banks can affect the level of mortgage credit in an economy; the loan-to-value (LTV) ratio, the mortgage term and the income fraction (κ), or the proportion of gross monthly income accounted for by mortgage repayments. Using the loan-level dataset, in Figure 2 we plot the median values of the credit instruments across the three institutions for loans originating over the period 2000 to 2011. From the charts it is clear that the previously observed aggregate expansion in credit (from Section 2) was reflected in each mortgage credit instrument, with all three registering a marked increase for loans originating between 2000 and 2007. On average, income fractions increased from a low of 16 percent in 2000 to a peak of about 25 percent at the height of the housing boom in 2007. Thereafter, κ dropped to a sample low. In the middle panel of Figure 2 it can be observed that Irish banks also eased credit by offering higher LTV ratios to mortgages over time. For loans originating in 2001, the median LTV value was only 59 percent. At the height of the housing boom, this figure had increased to a high of almost 80 percent. Finally, in the right panel of Figure 2 developments in the term of the mortgage by year of loan origination can be observed. The median mortgage term increased from 20 years in 2000 to 26 years in 2007.

3.3

3.3.1

Who received the expanded credit? Evidence from the existing literature

Before assessing how credit in the Irish economy was extended across cohorts of the population, it is first useful to set out the expected impact of credit liberalisation on different cohorts, as hypothesised in the existing literature. 8

In the Appendix, we discuss the sample of interest to this study.

6

Page 6 of 35

Ac ce p

te

d

M

an

us

cr

ip t

A good theoretical starting point for this is the standard life cycle hypothesis (Modigliani and Brumberg (1954) and Modigliani and Brumberg (1990)). In the life cycle framework, credit liberalisation should facilitate increased borrowing by younger households as they seek to smooth housing consumption, where such households believe their present income is below that of its permanent counterpart. Recent theoretical and empirical contributions support the finding that younger and poorer families tend to benefit more from relaxing credit constraints. Chambers et al. ((2007a), (2007b) and (2008)), for example, show that loan products which relax down payment levels or that facilitate increasing repayment schedules, tend to increase the participation of younger and lower income households in the mortgage market. Similarly, Ortalo-Magn´e and Rady (1999) and (2006) indicate that the financial liberalisation experienced in the United Kingdom in the early 1980s was associated with an increase in owner occupancy rates amongst young households. Doms and Krainer (2007) provide evidence that younger and constrained households in the United States used newly designed mortgage products to finance increased expenditure on housing. Furthermore, they show that lower income households increased their spending on housing at a greater pace than higher income households. Bover et al (2016) also find higher mortgage market participation rates among households in those Euro Area countries with relatively less restrictive lending rules. In summary, the existing theoretical and empirical literature suggests that credit liberalisation should lead to increased credit usage among the more marginal house purchasers, and those who are most credit constrained; the younger cohorts, first-time buyers, and the relatively less well-off cohorts of society. 3.3.2

The Irish case

To identify who benefitted from the easing of credit standards observed in the Irish case, we divide the sample of loans into five groups, ranked according to their position in the borrower income distribution.9 The group labelled “Lowest” captures borrowers in the bottom 20 percent of the income distribution and the group labelled “Highest” captures those in the top 20 percent of the distribution. The results in Figure 3 show that, while credit conditions were eased for all borrowers in the sample, the expansion in credit was most dramatic for those borrowers at the lower end of the 9

Income refers to gross income at the point of loan origination. We express all income values in 2006 terms.

7

Page 7 of 35

Ac ce p

te

d

M

an

us

cr

ip t

income distribution. For this category of borrowers, the income fraction for the median case rose by almost 14 percentage points between 2000 and 2008, to reach a peak of 32 percent (left panel of Figure 3). The corresponding increase for borrowers in the highest income quintile was closer to 8 percentage points. Similarly, in the second and third panels of Figure 3, while we see an increase in the average LTV and mortgage term for all categories of borrowers, the increase is most pronounced for those borrowers in the lower income quintiles.10,11 Next we assess developments in the credit instruments according to the type of buyer involved (Figure 4). We focus on three categories of borrower - first-time buyers, second-time buyers (or trade-up/down) and those who took out a mortgage to purchase a property for rental purposes (buy-tolets). While credit conditions were eased for all borrower types, first-time buyers experienced the most dramatic change over the period, registering a 12 percentage point increase in κ between 2000 and 2008, a 20 percentage point increase in the median LTV ratio between 2000 and 2006, and a 10 year increase in the median mortgage term between 2000 and 2008. Our micro-loan level dataset also allows for an exploration of credit conditions by current arrears status and mortgage originating institution. Insights on such issues can also inform the calibration of lending based macroprudential instruments. In Figure 5 we examine trends in credit conditions according to the current arrears status of the mortgage account. We use information from June 2012 to define the arrears status and define three outcomes: no arrears, arrears outstanding at end-June 2012 to the tune of 90 or more days worth of mortgage repayments, and arrears amounting to less than 90 days worth of mortgage repayments at end-June. While there are some slight differences in the median κ value and the median mortgage term between accounts in arrears (of any amount) and those that are performing, the middle panel of Figure 5 points to more pronounced dif10

It can be argued that changes in the Irish tax code during the early part of the 2000s reduced net taxes so disposable income actually rose more than gross. Therefore, an increase in κ may not have had as significant an impact on household affordability as what the gross figures suggest. Of course, the opposite is the case in the aftermath of the financial crisis when personal tax takes were increased. 11 We also explore the degree of correlation across the different credit channels through scatter plots, i.e., we examine if borrowers tended to experience an easing of credit conditions across more than one mortgage channel. Fernandez-Corugedo and Muellbauer (2006) cite the possibility of a trade-off between these concepts so that the overall risk exposure to each borrower remains constant. In general, we find an upward sloping relationship between income fractions and LTVs, suggesting that the easing of these credit conditions is correlated in the Irish case. These charts are available from the authors, on request.

8

Page 8 of 35

Ac ce p

te

d

M

an

us

cr

ip t

ferences in the average originating LTV ratio of accounts by arrears status. The average originating LTV for the group of accounts that were in arrears by 90 or more days at end-June 2012 reached a peak of 80 percent, while those accounts that were performing at end-June 2012 had a peak average originating LTV of 76 percent in 2006. One issue which attracted some comment during the growth phase of Irish prices was the more liberal loan amounts secured for individuals by mortgage brokers compared to traditional branch banking. Consequently, in Figure 6 credit conditions according to the institution responsible for granting the mortgage are plotted i.e. agents, institutional branches and mortgage brokers. The data in this chart are only available for one institution in the dataset and only for those loans originated by the end of 2010. Conforming with a priori expectations the graph suggests that broker originated loans were associated with a greater easing of credit conditions during the housing boom than loans obtained from a bank branch or mortgage agent. For example, broker originated loans had a median κ value of 19 percent for loans originated in 2000 while loans originated in 2007 had an average κ value of 28 percent. The respective figures for bank branch originated loans are 16 percent (2000) and 25 percent (2007). This difference in broker and non-broker originated loans is not unique to Ireland. Jiang et al (2014) observe that broker originated loans in the U.S. are of a lower credit quality than non-broker originated loans and that broker originated loans are associated with a higher probability of default than loans originated by other institution types. In summary, the loan level data show that credit conditions in Ireland eased considerably over the period leading up to the collapse in Irish house prices. Further exploration reveals that this easing of credit conditions occurred for all borrower types and income levels, though first time buyers and borrowers in lower income quintiles benefitted to a greater extent. These findings are in accordance with results from similar studies in the international literature. An examination of credit conditions by current arrears status and across mortgage granting institutions points to some important differences among the groups, highlighting the need to monitor such developments when calibrating macro-prudential instruments.

4

A model of credit conditions

Given the substantial changes in credit provision over the period, an inevitable question that arises is the extent to which these developments re9

Page 9 of 35

Ac ce p

te

d

M

an

us

cr

ip t

flect movements in underlying fundamental variables in the Irish economy or whether the changes in credit levels are due to conscious variations in lending standards by financial institutions. As noted earlier, between 2000 and 2007 due to the increases in living standards and numbers in employment, prospective Irish mortgagees were presenting to financial institutions with significant improvements in their borrowing capabilities. In such a context, changes in credit levels could be due to increased demand-side pressures. However, the revolution in funding sources available to credit institutions outlined in section 2 would suggest that supply-side issues are also a likely cause of change. Therefore, we examine the determinants of the income fraction and the loan-to-value ratio and estimate the extent to which these credit channels were impacted by changes in credit standards in the Irish market over the period 2000 - 2011. In doing this, we draw on a growing body of empirical research which has sought to account for changes in credit standards across mortgage markets. Much of this work was instigated by the development of the credit conditions indicator (CCI) by Fernandez-Corugedo and Muellbauer (2006) and has been further developed in Aron and Muellbauer (2011) and Muellbauer and Williams (2012). The approach mainly involves treating changes in credit conditions as a common unobserved factor which determines intercept and parameter shifts in equations for consumption, house prices and mortgage credit. In the cited literature, this is represented as a spline function consisting of smoothed step dummies, where credit conditions enter the differing equations as a common intercept term and through their interaction with interest rates, income growth expectations and housing collateral. A related stream of work in Duca et al. (2011) and (2012) involves modelling changing credit conditions by cyclically adjusting the loan-to-value ratio (LTV). The adjusted LTV measure, it is argued, captures exogenous shifts in credit standards. Our approach is influenced by this literature in that we model the two credit condition channels as a function of demandside factors such as income levels, interest rates, and other household specific characteristics. We include year dummies and argue that the remaining variation in these channels captured by the dummies reflects exogenous changes in credit conditions. Our approach in using micro-level data as opposed to time-series aggregate information is relatively unique. While the loan-level dataset alluded to earlier includes much of the information of interest for this model, the dataset does not include much detail on the characteristics of the borrower that might impact loan-demand. For example, the loan-level dataset does not indicate the size of the household applying for a loan, the marital status of the borrower or their age. 10

Page 10 of 35

Ac ce p

te

d

M

an

us

cr

ip t

At this point, we therefore rely on an additional source of information a household survey that was undertaken on a representative subset of the loan-level dataset. This survey was designed to supplement the information in the loan-level dataset with both current information on the economic circumstances of Irish mortgagees and originating information on borrowers at the time of their loan application. The survey was undertaken on a representative sub-sample of 2,000 borrowers from the loan-level dataset. Each individual’s survey responses can be linked back to their corresponding mortgage information in the loan-level dataset, where the respondent gave permission for this linking to take place.12 To model credit supply, we focus on the portion of the sample that allowed their survey responses to be linked to their loan-level data, so the sample size at this stage is 1,837.13 The main demand-side variables specified for both credit channels are the household-specific house price-to-income ratio (pi /yi ) and the mortgage interest rate (ri ), where both are recorded at the point of loan origination. In the case of the income fraction, the house price-to-income ratio is expected to exercise a positive impact on the size of κ as a high ratio puts pressure on potential homeowners to get the highest possible loan amount. However, in the case of the loan-to-value ratio, Fernandez-Corugedo and Muellbauer (2006) contend that the ratio should act in the opposite direction. Two reasons are advanced for this; one, a high house price, ceteris paribus, indicates a greater probability of a decline in house prices; secondly, in cases where high house price-to-income ratios exist, households are more likely to have higher income fractions. In such a case, lenders may be more cautious about offering a high loan-to-value ratio. The persistent improvement in both economic conditions and house prices is likely to have impacted on individual households’ expectations both in terms of price and income levels. Therefore, in our application, we take the expected house price to expected income ratio (pei /yie ). While adequately capturing households expectations can be quite difficult, we follow the recent literature such as Case et al. (2012), who, in documenting the inefficiency of the housing market in the US, highlight how backward-looking house price expectations can be. Similarly, both Himmelberg et al. (2005) and Duca et al. (2011) argue that expected house price appreciation may be tracked by lagged house price appreciation over the prior five and four years respectively. Accordingly, we take both house price and income expectations to 12

Almost 9 out of 10 households agreed to have their two data sources linked. We provide further information on the survey in the appendix, along with an overview of the characteristics of the sample used at this stage of the analysis. 13

11

Page 11 of 35

Ac ce p

te

d

M

an

us

cr

ip t

be the average of the lagged levels over the previous four year period.14 For mortgage interest rates, in the case of both credit channels, a rise in rates is expected to increase the probability of mortgage default, so a negative sign is hypothesised. Drawing on the income survey, we are also in a position to control for a number of household characteristics, which may also impact on the level of credit extended. The choice of control variables is motivated by the existing literature on micro-level datasets, which provide for a link between lending outcomes and personal / household characteristics. See, for example, Bover et al. (2016) or Gomez-Salvador et al. (2011). Specifically, we include dummies for the following considerations; whether the house purchased is a primary dwelling property (pdh), the number of contributors to the mortgage at the point of loan origination (contribs), whether the head of the household is a female (f emale), whether they are employed in the public sector (public), whether they are self-employed (self − employed), whether the household has savings or investments (saving), the education level of the head (ed − medium and ed − high), whether the household has an additional mortgage (addition) and the age of the head at the point of loan origination (age). In the case of the age variable, this effect should exercise a negative impact as households’ attainment of the financial assets very often necessary to acquire a housing downpayment, typically, tend to increase with age. For the employment variables, anecdotal evidence about the Irish market suggests that borrowers who are self-employed and working in the private sector face greater difficulty in securing mortgage credit than those operating in the public sector. Also, one would expect that a higher level of educational attainment would lead to a greater amount of credit, via the two channels, being secured. To capture the impact of changing credit standards, we specify year dummies for 2001 to 2010.15 For both credit channels, the following model is estimated: 22 2010 10 X X X pei βj controli,j + βm yearn + ²i . (1) cci = β0 + β1 e + β2 ri + yi m=13

j=3

n=2001

where cci is either the loan-to-value ratio or the κ for household i, controli is the vector of controls and year is the vector of year dummies for 2001 to 14

This involved backcasting the income level and house price reported at the point of loan origination with data on regional and county-level information on house prices and income levels for the four year period prior to this. 15 The dummy for the year 2000 is excluded.

12

Page 12 of 35

us

cr

ip t

2010. We use the nonlinear least squares system estimator to estimate the models on a cross-sectional basis. We use this estimator as we wish to test certain cross-equation restrictions.16 Results for both channels are shown in Table 3. In each case, the house price-to-income ratio is significant and has the expected sign. Amongst the household controls, in the case of both credit channels, the attainment of a higher level of education increases the level of credit secured, while the age of the head of household is, as expected, negatively related to the amount of credit obtained. For the κ channel, where the household has an additional mortgage and where the mortgage is for the primary dwelling house, this is related to a lower amount of credit being secured.

te

d

M

an

Turning to the year dummies, while the results are broadly similar across the two channels, there are some notable differences. For the LTV ratio, there appears to have been a significant increase in the credit supply function for the entire period, while the function appears to have moved on a significant basis for κ from 2008 onwards. In Figure 7, based on the dummy coefficients in both regressions, the resulting mortgage market credit indicators (MMCI) for the two credit channels are presented. The difference between both channels from 2007 to 2010 is apparent; the MMCI estimated for κ indicates that the contraction in credit standards is every bit as large as the liberalisation of standards during the boom phase.17

Ac ce p

From a policy perspective, this difference across the two channels is quite notable and reinforces the case made by Campbell and Cocco (2015) that regulators should pay attention to both the loan-to-value ratio and the income fraction in moderating the exposure of individual institutions to credit risk. As a robustness check on the sensitivity of the coefficient estimates, we also use the related, but alternative GMM systems estimator and full information maximum likelihood (FIML) estimator to estimate (1). The results may be observed from Table 4. Overall, there is very little difference in the coefficients; this is particularly the case for the dummy variable estimates, which are, ultimately, the important estimates for the MMCI.

16

The programs in RATS are available, upon request, from the authors. As a robustness check, we also estimate the system where the indicator is constrained to be the same across the two channels, however, the restriction is rejected at the 1 per cent level. 17

13

Page 13 of 35

5

Conclusions

Ac ce p

te

d

M

an

us

cr

ip t

Up to the recent financial crisis, many countries experienced a sustained period of house price growth. In general, this was facilitated by a combination of benign macroeconomic conditions and greater availability of mortgage credit. The Irish case is a profound example of this; a small open economy enjoying sustained economic growth with a financial system to the fore in exploiting new international funding conditions. It is the extent of these changes which makes analysis of the Irish property market over the period 2000 - 2011 such a compelling study. Courtesy of granular, micro level stress-testing of the three main Irish financial institutions conducted by the Central Bank of Ireland, mortgage loan level data is now available for a significant proportion of the Irish residential property market. Using this information we conduct a detailed analysis of the trends in the different credit instruments over the period 2000 - 2011. Many of the stylized facts which emerge in the Irish market are in keeping with both the empirical and theoretical literature; younger and less well-off households would appear to be the main beneficiaries of the more liberal credit regime. Younger households, in particular, appeared to conform to the life-cycle hypothesis by borrowing considerably to smooth out housing consumption. The relaxation of credit constraints within the financial sector most likely enabled many marginal households to enter the Irish housing market leading to a sizeable increase in demand. We also estimate an indicator of mortgage credit supply in the Irish market (MMCI). Controlling for demand-side factors amongst households, we estimate the changes in both the income fraction and the loan-to-value ratio. Our analysis suggests significant movements in the mortgage credit supply function during both the boom and subsequent downturn in the Irish market. The greater availability of detailed loan level data offers a number of policy opportunities, particularly from a macro prudential perspective. Given the role played by property markets in the lead up to the recent financial crisis, a renewed debate has centred on the appropriate choice of options available to policy makers and more specifically, on the relative contribution of financial innovation and fundamental economic factors to house price movements. A more efficient and precise calibration of influential macro prudential policy levers countering swings in the property market is now a much greater possibility with the availability of micro loan data.

14

Page 14 of 35

References

ip t

[1] Agnello L. and L. Schuknecht (2011). “Booms and busts in housing markets: Determinants and implications”, Journal of Housing Economics, Vol. 20, Issue 3, pp. 171-190.

cr

[2] Aron J. and J. Muellbauer (2011). “Wealth, credit conditions and consumption: Evidence from South Africa”, Review of Income and Wealth, Vol.59, Issue Supplement S1, pp. 161-196, October.

us

[3] Borio C. and P. McGuire (2004). Twin peaks in equity and housing prices?, BIS Quarterly Review, March.

an

[4] Bover O., Casado J.M., Costa S., Du Caju P., McCarthy, Y., Sierminska E., Tzamourina P., Villanueva E., and T. Zavadil, (2016). “The Distribution of Debt Across Euro Area Countries: The Role of Individual Characteristics, Institutions and Credit Conditions”, International Journal of Central Banking, (forthcoming).

M

[5] Campbell J.Y. and J.F. Cocco (2015). “A model of mortgage default”, Journal of Finance, Vol.70(4), pp. 1495-1554.

d

[6] Case, K. E., Shiller R.J. and A. K. Thomson (2012). What have they been thinking? Home buyer behavior in hot and cold markets, Brookings Papers on Economic Activity Fall 2012, pp. 265-98.

te

[7] Chambers S. M., Carriga C. and D. Schlagenhauf (2007a). Mortgage contracts and housing tenure decisions, Federal Reserve Bank of St. Louis, Working Paper 2007-040A, September.

Ac ce p

[8] Chambers S. M., Carriga C. and D. Schlagenhauf (2007b). Equilibrium mortgage choice and housing tenure decisions with refinancing, Federal Reserve Bank of Atlanta, Working Paper 2007-25, December. [9] Chambers S. M., Carriga C. and D. Schlagenhauf (2008). The loan structure and housing tenure decisions in an equilibrium model of mortgage choice, Federal Reserve Bank of St. Louis, Working Paper 2008-024B, July. [10] Dokko J., Doyle M. B., Kiley T. M., Kim J., Sherlund S., Sim J. and S. Van Den Heuval (2011). “Monetary policy and the global housing bubble”, Economic Policy, Vol.26, Issue 66, pp. 237-287, April. [11] Doms M.S. and J. Krainer (2007). Innovations in mortgage markets and increased spending on housing, Federal Reserve Bank of San Francisco Working Paper Series No. 2007/05. [12] Duca J.V., Muellbauer J. and A. Murphy (2011). “House prices and credit constraints: Making sense of the U.S. experience”, Economic Journal, Vol.121, pp. 533-551.

15

Page 15 of 35

ip t

[13] Duca J.V., Muellbauer J. and A. Murphy (2012). Credit standards and the bubble in US house prices: New econometric evidence, BIS Papers No 64 Property markets and financial stability, Proceedings of a joint workshop organised by the BIS and the Monetary Authority of Singapore in Singapore on 5 September 2011 pp. 83-89.

cr

[14] Fernandez-Corugedo, E. and J. Muellbauer (2006). “Consumer credit conditions in the UK”, Bank of England working paper 314.

us

[15] Galati G. and R. Moessner (2012). “Macroprudential policy - A literature review,” Journal of Economic Surveys, published online May 11.

an

[16] Gomez-Salvador R., Lojschova A. and T. Westermann (2011). “Household sector borrowing in the euro area: A micro data perspective”, European Central Bank Occasional Paper Series, No. 125, April.

M

[17] Himmelberg, C., Mayer C. and T. Sinai (2005). Assessing high house prices: Bubbles, fundamentals and misperceptions, Journal of Economic Perspectives, 19, Winter, pp. 67-92.

d

[18] Jansen S. and T. Krogh, 2011. Credit conditions indices: Controlling for regime shifts in the Norwegian credit market, Discussion Papers 646, Statistics Norway, Research Department.

te

[19] Jiang, W., Nelson, A. and E. Vytlacil (2014). “Liar’s loan? Effects of origination channel and information falsification on mortgage delinquency”, The Review of Economics and Statistics, Vol.96(1), pp.1-18.

Ac ce p

[20] Kelly, J. and M. Everett (2004). “Financial liberalisation and economic growth in Ireland”, Central Bank of Ireland, Quarterly Bulletin Article, Autumn pp. 91-112. [21] Kelly R., McCarthy Y. and K. McQuinn (2012). “Impairment and negative equity in the Irish mortgage market”, Journal of Housing Economics, Vol.21, pp. 256-268. [22] Lane, P. (2015). “The funding of the Irish domestic banking system during the boom?”, Journal of the Statistical and Social Inquiry of Ireland, Vol.XLIV, pp. 40-65. [23] Lim C., Columba F., Costa A., Kongsamut P., Otani A., Saiyid M., Wezel T. and X. Wu (2011). Macroprudential policy: What instruments and how to use them? International Monetary Fund (IMF) working paper WP/11/238. [24] Modigliani, F. and R. H. Brumberg (1954). “Utility analysis and the consumption function: an interpretation of cross-section data”, in Kenneth K. Kurihara, ed., PostKeynesian Economics, New Brunswick, NJ. Rutgers University Press, pp. 388-436.

16

Page 16 of 35

ip t

[25] Modigliani, F. and R. H. Brumberg (1990). “Utility analysis and aggregate consumption functions: an attempt at integration”, in Andrew Abel, ed., The Collected Papers of Franco Modigliani: Volume 2, The Life Cycle Hypothesis of Saving, Cambridge, MA. The MIT Press, pp. 128-197.

us

cr

[26] Muellbauer J. and D. Williams (2012). Credit conditions and the real economy: The elephant in the room, BIS Papers No 64 Property markets and financial stability, Proceedings of a joint workshop organised by the BIS and the Monetary Authority of Singapore in Singapore on 5 September 2011, pp. 95-101. [27] Ortalo-Magn´e F. and S. Rady (1999). “Boom in, bust out: Young households and the housing price cycle”, European Economic Review, Vol. 43, pp. 755-766.

an

[28] Ortalo-Magn´e F. and S. Rady (2006). “Housing market dynamics: On the contribution of income shocks and credit constraints”, The Review of Economic Studies, 73, No. 2, pp. 459-485.

M

[29] Roy S. and D. Kemme (2012). “Causes of banking crises: Deregulation, credit booms and assets bubbles then and now”, International Review of Economics and Finance, Vol 24, October, pp. 270-294.

d

[30] Taylor, J. (2009). “The financial crisis and the policy responses: an empirical analysis of what went wrong”, NBER working paper series 14631, January. http://www.nber.org/papers/w14631.

Ac ce p

te

[31] Tsatsaronis K. and H. Zhu (2004). “What drives housing price dynamics: crosscountry evidence”. BIS Quarterly Review, March.

17

Page 17 of 35

Taxonomy of factors influencing Irish credit supply

ip t

A

cr

Table A1 summarises the main developments from the mid-1980s in financial deregulation and liberalisation in the Irish market. [INSERT TABLE A1 HERE]

B.1

us

The loan-level dataset: Sample of interest and creation of additional variables Data Cleaning

an

B

Ac ce p

te

d

M

At the end of June-2012, the three institutions in the loan level dataset had over 660,000 mortgage loans outstanding on their books, and accounted for 70 percent of the Irish mortgage market. However, some of the outstanding loans are equity releases which were taken out some time after the original mortgage was secured and the property purchased. Since our aim is to examine credit conditions at the time of mortgage drawdown, equity releases are excluded from the analysis. Furthermore, we drop those loans which have little or no information on the geographic location of the property on which the loan is secured. Since we do not have a full year of data for 2012, we also drop loans originated in 2012 from the dataset. A final cohort of loans removed are cases where information is missing on borrower income or the house price at the point of loan origination.18 Table B1 shows the number of observations removed at each step of the cleaning exercise. The final dataset comprises 388,969 loan observations. [INSERT TABLE B1 HERE]

B.2

Generating the income fraction

While the mortgage term and LTV ratio are readily available in the loan-level dataset, the income fraction must be calculated for each loan. To compute the income fraction, we use the standard definition of a mortgage annuity (Bt ) as shown below: ¶ µ 1 − (1 + Rt )−τ . (B.1) Bt = κYt Rt 18

In general, this applies to mortgages that were subsequently ‘topped-up’ after the house was purchased. In these cases, the responding bank reported only the latest value for these variables at the point of the subsequent loan rather than also reporting the relevant values at the point of origination of the main mortgage.

18

Page 18 of 35

LT I 1−(1+Rt )−τ Rt

´.

(B.2)

cr

κ= ³

ip t

Where Yt is borrower income at origination, Rt is the relevant mortgage interest rate, τ is the mortgage term and κ is the income fraction.19 Solving for κ, we can re-write equation B.1 as follows:

an

us

An estimate of κ (the income fraction) is calculated for each loan in the dataset. As it incorporates both interest rates and the loan tenure, κ is a more precise measure of mortgage affordability than the related and more popular loan-to-income ratio. The concept is generally used as an indicator of mortgage market repayment stress (see Kelly et al. (2012) for example).

C

The survey of mortgage holders

Ac ce p

te

d

M

The survey used in the present study was conducted by ipsos MRBI on behalf of the Central Bank of Ireland. The primary purpose of the survey was to collect up-to-date information on a mortgage holder’s financial position and additional data on borrowers from the point of loan origination, which could be appended to the mortgage loan level information held by the Central Bank for the three main Irish financial institutions (AIB, BOI and ILP). The survey was designed to be representative of the loan books of the three main institutions along five dimensions: lender type, borrower type, interest rate type, arrears and county of residence. A two-stage sampling approach was used for the selection of cases for interview. In the first stage, representative clusters were formed from the loan-level data. In the second stage, clusters were randomly selected for interview. The total sample size achieved was 2,086 households, while the linked sample (those cases that permitted for their survey information to be linked back to their loan-level data at the Central Bank of Ireland) accounted for 88 per cent of this. The survey was conducted over the period May 2012 to February 2013 with 97 questions, in total, being asked of participating households. The survey included questions in the following categories: 1. Mortgage background, including questions on the contributors to the mortgage repayment, the educational and employment characteristics of such contributors and details of unemployment where relevant. 2. Income and finances, including detailed questions on household income, recent income changes, details on household expenditures and questions on repayment difficulties where relevant. 19

The interest rate at point of loan origination is not available in the loan-level dataset. As a proxy, we use the CSO’s quarterly average mortgage interest rate series.

19

Page 19 of 35

ip t

3. Buy-to-lets and other financial holdings, details of institutions where borrowings and savings are held and questions on credit applications and rejections, and future expectations.

cr

4. The mortgage arrears resolutions process (MARP), including questions on participation in the MARP process and the degree and nature of contact with the mortgage lender.

an

us

Table C1 provides an overview of the characteristics of the survey sample used in this study. We focus on the portion of the sample that allowed their survey responses to be linked to their loan-level data, so the sample size at this stage is 1,837. Among the sample, the largest portion of respondents are in the 35 to 44 year age group. The majority of respondents are married (83 percent), employed (85 per cent) and are relatively well educated, with over 40 per cent of respondents having a third level degree or higher. In terms of household composition, the average household in the sample comprises three persons (usually two adults and one child).

Ac ce p

te

d

M

[INSERT TABLE C1 HERE]

20

Page 20 of 35

1990

1995

2000

2005

Outstanding Level of Residential Lending

million

6,563

11,938

32,546

94,259

Total Value of Mortgages Issued

million

1,492

2,666

9,004

27,753

24,064

3,412

Average Mortgage Issued

42,856

54,094

111,355

231,206

271,154

184,113

Total Number of Mortgages Issued

34,812

49,288

80,856

120,037

88,747

18,532

House Prices

65,541

77,994

169,191

276,221

322,634

227,376

Housing Supply

M

2007

19,539

30,575

49,812

80,957

78,027

8,428

cr

Unit

an

us

123,002

2012

84,973

d

Variable

ip t

Table 1: Summary of Irish Residential Mortgage Market Statistics: 1990 2012

Ac ce p

te

Note: For all data except the outstanding level of residential lending, the observation for 2012 is quarter 2

21

Page 21 of 35

d

us

an M

cr

ip t Property

Loan

Interest Rate

Performance

te

Borrower

Current Interest Rate Interest Rate Type Interest Rate Margin Rate Revision Date

Ac ce p

Unit Identifier Borrower Type (FTB, BTL, etc.) Income Income Verified Credit Quality

Arrears Balance (June-2012) Arrears Balance for Past 12 months Collection Status Modification / Forbearance Flag

Page 22 of 35

Origination Date Original Loan Balance Current Loan Balance Loan Term Loan Purpose Current Repayment Payment Type Interest Rate Info. Performance Info.

Table 2: Loan-Level Data Fields / Information Content

Bank Borrower Property Loan

Geographic Location Property Type New or Existing Original Valuation (and date) Original LTV Construction Year

Notes: The above fields are not always populated in full.

22

ip t

Table 3: Regression estimate results

cr

-1.32 0.43 -1.15 -2.37 -0.44 -0.47 -0.03 1.72 -2.18 -6.81

us

-0.03 0.01 -0.02 -0.07 -0.01 -0.01 0.00 0.06 -0.07 -0.01

ltv Coefficient T-Stat 0.69 4.32 -0.43 -17.89 -1.98 -0.88 -0.02 0.03 -0.02 -0.05 -0.04 -0.02 0.02 0.12 -0.04 -0.02

M

an

T-Stat -16.65 25.87 3.18

te -0.07 0.00 -0.04 -0.03 0.09 0.06 0.05 -0.08 -0.20 -0.16

Ac ce p

Constant (p/y)i ri Controls: num f emale public pdh self saving ed − medium ed − high addition age Year dummies: 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

κ Coefficient -2.43 0.56 6.50

d

Dependent Variable

-1.28 0.19 -0.08 0.18 -0.63 0.14 -0.37 0.14 1.36 0.28 1.03 0.27 1.11 0.34 -1.75 0.32 -2.89 0.36 -2.07 0.28 N = 1,381

-1.08 1.41 -1.17 -1.62 -1.46 -0.94 0.71 3.27 -1.10 -15.38 3.01 2.79 1.99 1.85 3.91 4.11 6.61 6.20 4.71 3.36

Note: The dependent variables and the house price-to-income ratios are logged, interest rates are in actual rates, while the rest of the variables are 0/1 dummies.

23

Page 23 of 35

ip t

Table 4: Regression estimate results - Alternative estimators GMM

FIML

Coeff. 0.69 -0.43 -1.98

ltv T-Stat 4.58 -14.59 -0.98

-0.03 0.01 -0.02 -0.07 -0.01 -0.01 0.00 0.06 -0.07 -0.01

-1.36 0.43 -1.15 -2.07 -0.44 -0.47 -0.03 1.78 -2.10 -5.72

-0.02 0.03 -0.02 -0.05 -0.04 -0.02 0.02 0.12 -0.04 -0.02

-1.08 1.42 -1.15 -1.57 -1.46 -0.97 0.68 3.20 -1.12 -13.38

M

d

te

Ac ce p

-0.07 0.00 -0.04 -0.03 0.09 0.06 0.05 -0.08 -0.20 -0.16

-1.59 -0.09 -0.68 -0.43 1.58 1.17 1.11 -1.66 -3.19 -2.19

Coeff. -2.43 0.56 6.50

0.19 0.18 0.14 0.14 0.28 0.27 0.34 0.32 0.36 0.28

κ

T-Stat -16.51 25.66 3.15

Coeff. 0.69 -0.43 -1.98

ltv T-Stat 4.29 -17.74 -0.88

-0.03 0.01 -0.02 -0.07 -0.01 -0.01 0.00 0.06 -0.07 -0.01

-1.30 0.42 -1.14 -2.35 -0.44 -0.46 -0.03 1.71 -2.16 -6.75

-0.02 0.03 -0.02 -0.05 -0.04 -0.02 0.02 0.12 -0.04 -0.02

-1.07 1.40 -1.16 -1.61 -1.45 -0.94 0.70 3.24 -1.10 -15.26

3.71 -0.07 3.03 0.01 2.18 -0.04 2.06 -0.03 4.55 0.09 4.63 0.06 7.27 0.05 5.95 -0.08 5.04 -0.20 3.41 -0.16 N = 1,381

-1.27 -0.08 -0.63 -0.37 1.34 1.02 1.10 -1.73 -2.87 -2.05

0.19 0.18 0.14 0.14 0.28 0.27 0.34 0.32 0.36 0.28

2.98 2.77 1.98 1.84 3.88 4.08 6.55 6.15 4.67 3.33

us

T-Stat -18.57 20.66 3.58

an

κ

Constant (p/y)i ri Controls: num f emale public pdh self saving ed − medium ed − high addition age Year dummies: 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Coeff. -2.43 0.56 6.50

cr

Estimator Variable

Note: The dependent variables and the house price-to-income ratios are logged, interest rates are in actual rates, while the rest of the variables are 0/1 dummies.

24

Page 24 of 35

ip t cr

us

Table A1: Taxonomy of factors influencing Irish credit supply

2000 - present

Major relaxation of exchange controls.

Introduction of 100 per cent loan to value ratio (LTV) mortgages (2005). Introduction of tracker mortgages into the Irish market (1999/2000).

M

Formal trigger mechanism for changes in retail interest rates suspended.

an

1988 - 1999

Growing use of derivatives to manage risk.

Fixed-rate mortgages introduced by some banks for first time.

Mortgage securitisation.

Secondary liquidity requirement abolished.

Equity withdrawal and loan consolidation.

Ac ce p

te

d

Limitations on FX borrowing by residents and Irish Pound borrowing by non-residents removed.

Reduction in primary liquidity ratio 8 to 2 per cent.

25

Page 25 of 35

ip t cr us

M

Number of Loans

an

Table B1: Loan-Level Data: Overview of Cleaning Exercise

Total 654,959 140,494 820 8,600 38,708 7,243 56,987 13,138

Final Dataset

388,969

Ac ce p

te

d

Initial Observations Non-Primary Loan Missing Geographic Info. Top/Bottom 1% of House P. Distribution Missing LTV at Origination Top/Bottom 1% of Income Distribution Missing Income at Origination Original LTV¿100%

26

Page 26 of 35

18-34 35-44 45-54 55-64 65+

14.6 39.9 29.8 12.6 2.7

Marital Status

Married / Couple Widowed/Separated Single

83.3 6.1 10.5

Work Status

Employed Unemployed Inactive

84.5 6.1 9.2

Education Status

Low Medium High

13.1 43.6 42.5

1 Adult, 0 kids 2 Adults, 0 kids 3+ Adults, 0 kids 1+ Adults, with kids Undefined

9.4 16.0 7.4 60.0 7.2

te

d

M

an

us

cr

Age Group (years)

Ac ce p

Household Composition

ip t

Table C1: Demographic and economic characteristics of the survey sample, % of respondents unless otherwise stated Variable %

Median Financial Data (e)

Income Current House Price Mortgage Outstanding

Negative Equity Any Arrears Has Savings/Investments

% of Group % of Group % of Group

55,000 181,428 144,554 39.0 19.8 56.7

Note: Where group totals do not equal 100%, the residual is accounted for by “don’t know” or “refused” responses. Sample size is 1,837 except in the case of the current house price and negative equity; the sample sizes here are 1,808 and 1,795 respectively.

27

Page 27 of 35

i cr us

Figure 7

Ac

ce

pt

ed

M

an

Irish Mortgage Market Credit Indicator (MMCI)

2001

2002

2003

2004 LTV

2005

2006

2007

Income Fraction

2008 Page 28 of 35

2009

2010

ip t cr us an M d te

Figure 6

Ac ce p

Credit Conditions by Source of Loan: 2000 - 2011

Income Fractions

27.5

22.5

Mortgage Term 37.5

90

35.0

85

32.5

80

30.0

%

%

25.0

Loan to Value Ratios

95

Years

30.0

75

27.5

20.0

17.5

15.0

70

25.0

65

22.5

60 2000

2002

2004

Agent Branch

2006

2008

2010

Broker

20.0 2000

2002

2004

Agent Branch

2006

2008

2010

Broker

2000

2002

2004

Agent Branch

2006

2008

2010

Broker

Page 29 of 35

ip t cr us an M d te

Figure 5

Ac ce p

Credit Conditions by Arrears Status: 2000 - 2011

Income Fractions

25.0

Mortgage Term 32

80

30

75

28

70

26

%

%

22.5

Loan to Value Ratios

85

Years

27.5

65

24

20.0

60

22

55

20

17.5

15.0

50 2000 2002 2004 2006 2008 2010 90+DPD Arrears <90DPD Arrears No Arrears

18 2000

2002

2004

2006

2008

90+DPD Arrears <90DPD Arrears No Arrears

2010

2000

2002

2004

2006

2008

90+DPD Arrears <90DPD Arrears No Arrears

2010

Page 30 of 35

ip t cr us an M d te

Figure 4

Ac ce p

Credit Conditions by Buyer Type: 2000 - 2011

Income Fractions 30.0

Loan to Value Ratios

Mortgage Term

100

27.5

36

34

90

25.0

22.5

32

80

30

Years

%

28

%

20.0 70

17.5

15.0

60

26

24

12.5

22 50

10.0

20

7.5

40 2000

2003 FTB STB

2006

2009 BTL

18 2000 2002 2004 2006 2008 2010 FTB STB

BTL

2000 2002

2004 FTB STB

2006

2008 2010 BTL

Page 31 of 35

ip t cr us an M d te

Figure 3

Ac ce p

Easing of Credit Conditions by Income Quintile: 2000 - 2011

Income Fractions 32.5

Loan to Value Ratios

Mortgage Term

85

30.0

32

80

27.5

30

75

25.0

28

70

26

%

Years

65

%

22.5

20.0

60

17.5

55

24

22 15.0

50

12.5

45

20

10.0

40 2000 2002 2004 2006 2008 2010 Lowest 2 3

4 Highest

18 2000

2002

2004

Lowest 2 3

2006

2008

2010 4 Highest

2000

2002

2004

Lowest 2 3

2006

2008

2010 4 Highest

Page 32 of 35

ip t cr us an M d te

Figure 2

Ac ce p

Easing of Credit Conditions: 2000 - 2011

Income Fractions 26

Loan to Value Ratios

Mortgage Term

77.5

30

75.0

24

28

72.5

22

26

Years

20

%

%

70.0

67.5

24

65.0 18

22 62.5

16

20 60.0

14

57.5 2000

2003

2006

2009

18 2000

2003

2006

2009

2000

2003

2006

2009

Page 33 of 35

i cr us

Figure 1

an

Funding of Irish Financial Institutions: 2001- 2010 Issuance of Debt Securities by Irish Mortgage Lenders

300

60

euros billions

Ac

ce

120

euros billions

350

pt

140

80

400

ed

160

100

Private Sector Credit and Deposit Levels in the Irish Financial System

M

180

250

200

150

40

20

100

0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

50 Irish counterparties Other Monetary Union counterparties Rest of the world counterparties

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Private sector credit

Page 34 of 35

Deposits

Credit Conditions in a Boom and Bust Property Market: Insights for Macro-Prudential Policy Highlights

ip t

 The significant contribution of credit to the international house price boom preceding the financial crisis of 2007/08, has led to a considerable amount of attention on the use of macroprudential policy as an effective tool in preventing future credit bubbles. A number of countries, for example, have introduced loan-to-value and loan-to-income limits.

us

cr

 The calibration of these instruments can be difficult, however, given the relative ’newness’ of the macro-prudential field and the relative lack of experience with the use of the relevant instruments.

an

 Exploiting a new dataset for the Irish mortgage market, this paper shows the value of micro loanlevel data which provides detail on how these instruments tend to evolve across different buyer types and at different points in the business cycle.

M

 The Irish case is of particular note; between 1995 until 2007 real Irish house prices grew by nearly 9 per cent per annum - the next highest country growth rate in the OECD was 7.6 per cent. Since 2007, Irish house price declines have been the most severe across the OECD.

ce pt

ed

 During the boom, the Irish credit market experienced a considerable degree of financial liberalisation. This easing of credit standards was compounded by the ability of Irish credit institutions post-2003 to access international wholesale markets, thereby increasing significantly the supply of credit available.  The micro-level analysis shows that, in keeping with the existing literature, the boom period easing of credit conditions occurred for all borrower types and income levels, though first time buyers and borrowers in lower income quintiles benefitted to a greater extent.

Ac

 An examination of credit conditions by current arrears status and across mortgage granting institutions points to some important differences among the groups, highlighting the need to monitor such developments.  An indicator of mortgage credit supply highlights the significant movements in the supply of mortgages to Irish households over the period of examination.  The paper highlights the importance of detailed loan level data for monitoring housing credit and for assessing the potential impact of macroprudential policies.

Page 35 of 35