The economic crisis has exacerbated the problem of risk evaluation in mortgage loans. Before 2015, banks used to give mortgage loans to almost all applicants, but as they’ve faced a growing amount of overdue debts and mortgage defaults, they’ve toughened the conditions of lending. This, together with growing interest rates and decreasing purchasing power, has led to a market meltdown.
The banks granted mortgage loans for a total of 460.7 billion roubles during the first six months of 2015, which turned out to be 40% less when compared with the same period of 2014 (last year, annual growth was 30%). According to preliminary expert evaluations, the mortgage market by the end of 2015 will shrink by 30-40% and will total 800-900 billion roubles.
The government launched a public subsidy programme for mortgage loans in March 2015 in order to support the construction industry, which reanimated the mortgage market a little bit. But the problem of creating an effective risk evaluation system still exists. The crisis demonstrated the importance of studying the key factors of default and the flaws in the existing methods of credit risk evaluation; the decision on granting a loan is often based on the planned indicators for granting mortgage loans, not on the borrower’s specifics.
Alexander Karminsky, Professor at the HSE Faculty of Economics Department of Finance, and Agata Lozinskaya, Junior Research Fellow at the Laboratory of Interdisciplinary Empirical Studies, HSE Campus in Perm, built an econometric model of mortgage default (a two-dimensional probit model with the use of a two-stage Heckman method to evaluate its parameters). It allowed them to determine the key risk factors related to borrowers’ demographics, mortgage loan parameters, and macroeconomic indicators. The model can be used in a system of evaluating mortgage borrower’s credit risk.
The model was described in the paper ‘Credit Risk Evaluation in Housing Mortgage Loans’ (in Russian) and trialed on some unique micro- and macro-data of mortgage borrowers. Karminsky and Lozinskaya gathered aggregated regional monthly data on mortgage loans and macroeconomic indicators that determined the demand and offers in this market in 2008-2012, as well as data from one of the public Housing Mortgage Agency offices on 4,298 Russian borrowers who applied for a mortgage between 2008 and 2012. The sample included information on both approved and declined mortgage applications.
The probability of mortgage default depends mostly on the borrower’s revenue, which is the main source of mortgage repayment, the authors discovered. The higher the income, the less the risk of arrears, and vice versa. Obviously, low revenues can be unstable and insufficient for covering mortgage liabilities. An income of 30,000 roubles a months could make the bankers expect timely and unproblematic loan repayments.
‘The results we received show that it’s necessary to develop special loan programmes for people with low incomes and the borrowers who can’t officially confirm their income’, Karminsky and Lozinskaya said. They believe that loans from building-and-loan associations could be an alternative to bank mortgages. Saving and loan associations usually bring together citizens who don’t have access to the formal credit market due to the lack of reliable information on their income and their ability to repay loans over a long term.
Paradoxically, informal incomes turned out to be more reliable than formal salaries, the researchers reported. Borrowers with unconfirmed incomes showed minimal risk of default. ‘This can be explained by the considerable gap between the declared and the real incomes of borrowers, which can’t always be confirmed on paper’, Karminsky and Lozinskaya explained. Informal incomes are higher and they often turn out to be higher than those confirmed on paper.
The data analysis showed the highest credit risks are specific for male borrowers, single men and women or those who don’t indicate their marital status, as well as public officials. ‘This is largely due to bad payment discipline, low life expectancy, and higher risks of divorce, illness, or losing their job’, the authors commented.
These empirical results didn’t confirm the influence of the level of education on the probability of default. Accordingly, there’s no correlation between the probability of mortgage default and people’s financial awareness.
In addition to income, credit discipline is also considerably influenced by the cost of the mortgage, Karminsky and Lozinskaya found out. The probability of borrowers’ default grows on average, and under otherwise equal conditions, by 3% when the interest grows by 1%.
The researchers failed to prove any influence of the length of the loan on the probability of default. According to them, this is because most of mortgage loans have been issued for a period of over 15 years. Taking into account that the data covers a shorter period of time, it is quite possible that defaulting hasn’t occured yet in these loans. The same thing could explain the higher probability of defaults given for a term of less than 15 years.