In this analysis, we look at the macroeconomic factors which function as driving forces behind developments in banks' problem loans. Problem loans include non-performing loans and other particularly doubtful loans. Since the beginning of the 1990s, problem loans as a share of total loans have declined sharply and are now at a historically low level. However, the volume of problem loans is highly sensitive to cyclical developments and will usually increase during economic downturns. We have analysed banks' problem loans in the household and the enterprise sector respectively, using two empirical models. The analysis reveals that the declining share of problem loans in recent years is primarily attributable to developments in real interest rates and unemployment. We also project banks' problem loans based on two macroeconomic scenarios: A baseline scenario and a stress scenario which illustrates a deteriorating macroeconomic situation.
One of Norges Bank's key tasks is to monitor the financial system. The banks play a central part in the financial system as providers of credit and payment services. Norges Bank therefore closely monitors developments in the banking sector. Attention is particularly focused on developments which in the short or long term may weaken stability in the banking sector and prevent banks from discharging their responsibilities in a satisfactory manner. The experience of Norway and other countries shows that developments in problem loans along with losses on bank lending have a considerable impact on banks' ability to channel credit. (2)
Problem loans consist of both non-performing loans and non delinquent loans which the banks consider to be particularly doubtful. (3) Banks have to estimate their expected losses on problem loans if a borrower goes bankrupt or is, for other reasons, unable to service his debt. To a large extent, recorded losses consist of changes in these loss estimates (4). There will thus be a close connection between banks' problem loans and recorded losses. However, recorded losses are also affected by unexpected losses and reversals of previously recorded losses, and the lag between developments in problem loans and recorded losses may vary.
In this article, we look at the relationship between macroeconomic factors and banks' problem loans. Banks' problem loans serve as an important indicator of financial imbalances in the household and enterprise sectors. A high share of problem loans indicates that many borrowers are having problems in servicing their debt. This may result in higher loan losses for banks.
Developments in the volume of problem loans may primarily be directly related to borrowers' capacity and incentive to service debt. Relevant macroeconomic variables that affect the capacity and incentive to service debt may thus be included as explanatory variables in empirical models for banks' problem loans.
In general, a certain proportion of banks' loans will over time be non-performing and later be lost because borrowers cannot repay the loan as agreed. However, the volume of problem loans is sensitive to cyclical developments and will normally be higher in periods of contraction and lower in periods of expansion. Charts 1 and 2 show developments in problem loans in the household and the enterprise sector, respectively. In both sectors, the volume of problem loans peaked during the banking crisis, before falling sharply in the latter half of the 1990s. In recent years, the level of economic activity has been high and the volume of problem loans has thus been low. In addition, debt growth has been high in both sectors, resulting in a historically low level of problem loans as a share of total loans (see Chart 3).
In this article, we will try to answer the following questions:
--What are the most important macroeconomic explanatory factors for developments in banks' problem loans?
--How quickly and strongly do problem loans react to changes in these factors?
--What has driven developments in problem loans in recent years?
--How are problem loans expected to develop over the next few years?
In the next section, we discuss factors that may affect banks' problem loans. Section 3 presents two empirical models for problem loans in the household and the enterprise sector, respectively. In Section 4, we take a closer look at the contributions of each of the explanatory factors over the past few years. We further present projections of problem loans based on expected macroeconomic developments as outlined in Inflation Report 3/06. We also present projections based on a stress scenario which illustrates a deteriorating macroeconomic situation. Section 5 concludes.
2 What influences banks' problem loans?
Factors that determine developments in banks' non-performing loans and losses have previously been the subject of a number of analyses (5). Several of the analyses (6) have been based on an expression which indicates the expected level of non-performing loans:
(1) NPL = [SIGMA].sub.i-1 ..., n] [[rho].sub.i][L.sub.i]
where NPL = non-performing loans; [p.sub.i] = the probability that borrower i will default on his loan; [L.sub.i] = the borrower's debt i; i = 1, ..., n denotes the borrower.
According to relation (1), non-performing loans may be analysed based on the probability of the borrower defaulting on his loan and the size of the individual loan. However, we do not observe the probability of default for the individual borrower, [p.sub.i], but we may assume that this depends on the borrower's capacity and incentive to service his debt as agreed, i.e. in accordance with his contract with the bank.
Debt-servicing capacity depends on developments in borrowers' income, debt-servicing costs and other costs. Banks provide loans based on borrowers' expected future income and expenditure flows. If developments in these variables deviate from expected developments, the borrower's debt-servicing capacity may be reduced. In periods of weak cyclical developments, when unemployment is rising and corporate earnings are deteriorating, there may be an increase in both non-performing loans and banks' losses.
The incentive to service debt will normally depend on how the loan agreement is formulated, along with developments in collateral values and interest rate levels. Other studies have looked at how debt servicing is affected by borrowers' opportunity to submit false reports of their earnings, the bank's threat to foreclose the entire loan in the event of default, and the significance of the collateral. (7)
Households' capacity and incentive to service debt
Households' debt-servicing capacity generally depends on developments in their income, debt, borrowing rate and collateral values. Higher incomes are expected to contribute to reducing the volume of problem loans. However, incomes may be unevenly distributed across households. When unemployment is rising, many households may experience a substantial reduction in income. Thus, we expect that higher unemployment will lead to a higher volume of problem loans. When interest rates and/or debt are rising, a larger share of borrowers' current income will be used to service debt. In isolation, this will contribute to increasing the number of borrowers with debt-servicing problems and we expect the volume of problem loans to rise. It is reasonable to assume that an increase in the collateral value provides borrowers with greater opportunities to cope with a strained financial situation. Borrowers may achieve more favourable interest rate terms on their loans or possibly a deferral of principal payments. The volume of problem loans is therefore expected to decrease if house prices increase.
In general, households have a high incentive to service their debt, regardless of the collateral value. A large share of household debt is secured on dwellings. If the bank wishes to recover the collateral, households risk having to move. Moving costs can be substantial. Furthermore, in the event of a forced sale, prices for the collateral may be lower than the normal market value. In addition, households with defaulted debt may have negative credit information registered with credit information agencies, which may make it difficult to raise new loans. Even when the collateral value is lower than the debt, households will still have a high incentive to service their debt as most households will end up having outstanding debt after the collateral has been recovered.
Enterprises' capacity and incentive to service debt
Enterprises' capacity to service debt generally depends on their income and costs, borrowing rates and the size of the debt. Developments in corporate earnings will to a large extent follow business cycles. Unemployment is an indicator of the level of activity in the economy. If unemployment is low, domestic demand will be relatively high. This normally leads to solid corporate earnings and increased debt-servicing capacity. Therefore, lower unemployment is expected to lead to a reduction in problem loans. Oil prices also constitute an important cyclical variable in the Norwegian economy. This is primarily an important factor for the activity and investment level in the petroleum sector, but it also has spillover effects for suppliers to this industry. Norway's terms of trade also depends on oil prices. An increase in oil prices is expected to reduce the volume of problem loans. Developments in earnings for internationally exposed enterprises are affected by their competitiveness relative to foreign enterprises and by activity levels abroad in general. Deteriorating competitiveness and/or declining foreign demand are expected to lead to an increase in problem loans. Furthermore, a rise in costs is expected to increase the volume of...