【文章內(nèi)容簡(jiǎn)介】 e disaggregated approach has the advantage of assessing the impact of stress tests on different ponents of banks’ loan portfolios, it suffers from the drawback that data are available, at the earliest, only from 1993 and thus do not cover a plete economic remainder of the paper is anised as follows. Section 2 reviews the literature on stress testing,Section 3 discusses the choice of variables in our VAR, Section 4 discusses data and estimation issues, and Section 5 contains the main results of the estimations. In Section 6 a number of robustness checks on the results are discussed while in Section 7 the key factors affecting writeoffs are deposed. Section 8 concludes. Literature review A number of approaches have been used in the past to stress test banks for credit risk. The most mon approach used in IMF country FSAPs are single factor sensitivity tests. These look at the impact of a marked change in one variable, such as the exchange rate or the policy interest rate, on banks’ balance sheets. However, these stress tests do not allow for the interaction between macroeconomic variables (‘scenarios’) such as, in the example above, the impact of changes in the interest rate on real activity and thus on banks’ loan portfolio. Scenarios can be developed through a number of methods. One approach is to use a structural macroeconomic model. This was done, for example, in a number of IMF FSAPs on developed countries. An alternative avenue is pursued by Boss (2021) to stress test the Austrian credit portfolio. His analysis is based on CreditPortfolioView174。, which models the default probability of certain industrial sectors as a logistic function of a sectorspecific index, which, in turn, depends on the current value of a number of macroeconomic variables. The parameter estimates derived from this model are then used to assess the future losses on Austrian banks’ loan portfolios. A different methodology to assess the impact on the Austrian banking sector of credit and market risk is applied in Elsinger, Lehar and Summer(2021). Their paper analyses the effect of macroeconomic shocks on a matrix of Austrian interbank positions. Specifically, the authors are able to assess the probability of individual bank failures in response to a series of macroeconomic factors while at the same time taking into account the effect that these failures have on the rest of the banking system. This model thus deposes bank defaults into those that arise directly and those that are a consequence of contagion. The interaction between banks’ financial conditions and the macroeconomy is modelled by assuming that macroeconomic scenarios are drawn from a joint probability distribution of interest rate shocks, exchange rate and stock market movements, as well as shocks related to the business cycle. Pesaran et al (2021) and Alves (2021) use a VAR model to assess the impact of macroeconomic variables on firms’ probabilities of default. In Pesaran et al the VAR includes GDP, consumer prices, the nominal money supply, equity prices, exchange rates vis224。vis the dollar and nominal interest rates for eleven countries/regions over the 197999 period. The global VAR is used as an input into simulations for firms’ equity returns, which are then linked to the loss distribution of a corporate loan portfolio. A clear advantage of this approach is that it links the credit risk of internationally diversified loan portfolios in a detailed macroeconomic model that allows for differences across country and region. Alves (2021) constructs a cointegrated VAR, using KMV’s corporate expected default frequencies (EDFs) as endogenous variables and macroeconomic factors (the twelvemonth change in industrial output, the threemonth interest rate, the oil price, and the twelvemonth change in a broad stock market index) as exogenous variables. The expected default frequency (EDFs) of each EU industrial sector is modelled based on exogenous macroeconomic factors together with the EDFs of other in