تاثیر بحران بدهی های مستقل در فعالیت بانک های ایتالیایی
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|24089||2014||44 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Banking & Finance, Available online 21 May 2014
We examine the implications of the sovereign debt market tensions on the Italian credit market by estimating the effect of the 10-year BTP-Bund spread on a wide array of bank interest rates, categories of loans and income statement variables. We exploit heterogeneity between large and small intermediaries to assess to what extent the transmission of sovereign risk differed in relation with different banks’ balance-sheet characteristics and business strategies. Regarding the cost of funding, we find that changes in the BTP-Bund spread have a sizeable effect on the interest rates on term deposits and newly issued bonds but virtually no effect on overnight deposits. Furthermore, the sovereign spread significantly affects the cost of credit for firms and households and exerts a negative effect on loan growth. All these results are magnified when considering alone the five largest banks, which are typically less capitalized, have a larger funding gap and incidence of bad loans and rely more on non-traditional banking activities. Sovereign tensions also affect the main items of banks’ income statement.
Throughout the summer of 2011 the spread between the yield on the 10-year Italian government bond and the corresponding German one (henceforth, BTP-Bund spread) rapidly increased, reaching a peak of 550 basis points (b.p.) in November (Fig. 1). As shown by the sharp increase recorded by banks’ credit default swap spreads, the tensions in the sovereign debt market swiftly transmitted to Italian financial intermediaries. Funding cost and availability, especially on wholesale markets, deteriorated significantly. Furthermore, during the financial crisis Italian banks considerably tightened the lending standards to households and firms, not only by increasing the margins charged on new business loans but also, particularly in the most acute phases of the crisis, reducing the availability of credit to the private sector (Bank of Italy, 2012 and Del Giovane et al., 2013).Italy is an especially good case for assessing the effects of sovereign risk on the banking sector. First, the high level of public debt and the heavy exposure of Italian banks to the public sector suggest that sovereign tensions are likely to have a strong impact on banking sector. Second, Italy experienced periods of tensions on its sovereign debt market also during the 1990s, which helps to better identify the effects of the sovereign tensions in the empirical analysis (e.g. Favero et al., 1997). Third, while in some European countries the sovereign crisis started out when government borrowing conditions deteriorated following substantial public intervention to support weak banking sectors, in Italy the initial increase in government borrowing costs was related to weakness of the public sector itself in the context of a relatively healthy domestic banking sector. This distinct pattern of contagion allows us to treat the sovereign spread as an exogenous determinant of banking activity in Italy.1 In this paper we seek to quantify the effect of the sovereign debt market tensions – proxied by the 10-year BTP-Bund spread – on the cost of funding for Italian intermediaries, on the cost and availability of lending to firms and households, and on the main items of banks’ income statement. In addition to gauging the overall impact of sovereign tensions on the Italian banking activity, we also investigate the presence of heterogeneity in the transmission of these tensions across financial intermediaries. Sovereign risk transmits to banks’ funding and lending conditions via a number of different channels, reflecting banks’ high exposure to domestic sovereign debt, the role of government securities as collateral in secured transactions and the connections between sovereign and banks’ credit ratings.2 In principle, the intensity of each of these channels is likely to depend on banks’ balance-sheet characteristics – such as the level of capitalization, the reliance on (un)stable funding sources, the quality of the loan portfolio – as well as on business models and lending strategies. For example, banks more reliant on wholesale funding are likely to be more severely affected by a dry-up of international capital markets following a sovereign crisis; banks with lower levels of capital and/or a riskier loan portfolio might be forced to a stronger tightening of lending conditions to induce de-leveraging; banks with stronger lending relationships are likely to attenuate the transmission of negative sovereign shocks. For the Italian banks, a convenient way to explore the heterogeneity in the transmission of sovereign tensions is to distinguish between the five largest banking groups and the rest of the system, since these two groups of financial intermediaries differ significantly in terms of balance-sheet characteristics as well as business models and lending strategies. In general, the largest banks tend to be less capitalized; they tend to fund a larger share of their loans through wholesale financing; and, at the onset of the sovereign debt crisis they had a greater incidence of bad loans. Moreover, while smaller banks’ activity relies almost entirely on traditional banking operations, the largest banks gain a greater share of income from non-traditional activities and operations, such as trading and investment banking, insurance and other financial services. In addition, while small banks typically have a local presence, large banks are geographically diversified, with some of them having significant foreign operations. A corollary is that small banks typically rely more on soft information for borrowers’ selection and engage relatively more in relationship lending. Given these structural differences, one may expect the transmission of sovereign debt tensions to the banking activity to be stronger for larger banks. Our empirical approach is to estimate regression equations for various banking activity variables (borrowing and lending rates, loan developments and main income statement items) where the 10-year BTP-Bund spread shows up as an additional explanatory variable next to the standard determinants identified in the literature. For the borrowing and lending rates we also check for potential non-linear effects of the 10-year BTP-Bund spread by estimating equations in which the spread is interacted with two dummy variables that identify periods when the sovereign tensions were high: the pre-EMU period and the recent sovereign debt crisis. We draw on two data sources. The first dataset contains quarterly aggregate information for all Italian banks in the period 1991Q1-2011Q4, including variables such as interest rates on deposits and bonds; interest rates and growth rates of loans; net interest income, non-interest income and loan loss provisions. An advantage of this dataset is that the long time span allows us to exploit the episodes of tension in the Italian sovereign debt recorded in the 1990s to more precisely estimate the effect of sovereign risk. The second dataset, available for a shorter time period (since 2003) contains the same information, for bank rates and lending growth, at a higher (monthly) frequency and with breakdowns by size of the banks (five largest groups versus other banks), by size of the loans to firms (below versus above 1 million euro) and by type of mortgage loans to households (fixed- versus variable-rate). Our main results for the banking system as a whole can be summarized as follows. First, regarding the cost of funding, our estimates suggest changes in the BTP-Bund spread have a sizeable effect on the interest rates on term deposit, repurchase agreements and newly issued bonds. The pass-through increases during the sovereign debt crisis of 2010-11: during that period, a temporary (one-quarter) 100 b.p. rise in the sovereign spread is associated – within one quarter – with an increase of around 40 b.p. for deposits with agreed maturity and repurchase agreements and 100 b.p. for bonds. The sovereign spread appears not to exert any significant effect on the remuneration on households’ overnight deposits, consistently with the typically very sluggish adjustment behavior of this type of interest rates to current market conditions. Second, for interest rates on loans to firms and on mortgages to households we find a significant effect of the spread, which is stronger during crises as opposed to normal times: during the sovereign debt crisis a temporary (one-quarter) 100 b.p. increase in the sovereign spread is associated – with a one-quarter lag – with a rise of 70 and 30 b.p., respectively for business loans and households’ mortgages. Third, as regards the dynamics of lending to both firms and households for house purchases, we find that changes in the BTP-Bund spread exert a significant direct effect in addition to the indirect effect occurring through higher interest rates and the consequent lower demand for credit. Estimates suggest that a 100 b.p. increase in the spread is associated with a reduction of roughly one percentage point of the annual growth for loans to households and of 0.7 for loans to firms. Finally, when we estimate the impact of the sovereign spread on income statement items, we find a mildly positive coefficient for the net interest income, suggesting that the positive effect on the loan-deposit rate differential – calculated by weighting the lending and funding rates by their respective amounts – generally prevails over the negative effect on loan amounts. The estimated impact of sovereign tensions on loan loss provisions is positive and substantial, while we find no effect on the other revenues. These results are consistent with the idea that sovereign tensions increased risk premia that banks charge on borrowers: this justifies the stronger pass-through of the sovereign spread for lending as compared to deposit rates, which drives the result for the net interest margin, and is compatible with the deterioration in borrowers’ quality proxied by the increase in loan loss provisions. When we replicate our analysis distinguishing between the five largest banks and the rest of the system, we find that banks’ balance-sheet characteristics do matter for the transmission of the sovereign debt tensions and that, as expected, the effect is more pronounced for larger banks. As an example, the pass-through to firm loan rate after one quarter is almost double that of smaller firms; the estimated effect on loan amount is around one third smaller for the smaller banks. Moreover, when we consider also the breakdown by the size of the loan, which can be considered a proxy for the size of the borrower, we find that the largest difference in the loan rate pass-through between large and small banks is attributable to the low responsiveness of small banks towards small firms. These quantitative differences translate into an opposite effect on the net interest income, which is negatively affected by an increase in the spread for the five largest banks while the relation is positive for the other banks. For non-interest and other income, we find that the spread has a negative effect for the large banks while the impact is positive but only marginally significant for the smaller intermediaries. We find no difference across banks for loan loss provisions. Overall, the results by bank size are consistent with the fact that large bank’s lower capitalization, greater dependence on wholesale funding and greater role for non-traditional banking activity contributed to a stronger transmission of sovereign tensions to funding and lending conditions. On the other hand, the impact was somewhat attenuated for smaller banks and their borrowers, thanks to stronger capital and funding positions, greater reliance on traditional banking activity and more intense engagement in relationship lending. Over the last few years the empirical literature on the linkages between sovereign risk and banking activity has grown substantially. Angeloni and Wolff (2012) and Chan-Lau et al. (2013) find that European banks’ stock market performance was adversely affected by exposures to impaired sovereign debt. Correa et al. (2012) find that sovereign rating changes, especially in the case of downgrades, exerted a negative impact on bank stock returns. Gennaioli et al. (2013), analyzing 18 sovereign debt crises between 1998 and 2012, find that the correlation between a bank’s holdings of public bonds and its future loans is positive in normal times, but turns negative during defaults. Other studies have specifically examined the effect of sovereign risk on bank interest rates and lending developments. Studying the effect of the recent crisis on the main euro-area countries, Neri (2013) finds that bank lending rates significantly increased in the peripheral countries; De Marco (2013) finds that banks more exposed to the sovereign tensions tightened credit supply by more than banks that were less exposed. Similarly, Popov and Van Horen (2013), using syndicated loan data, find that lending by European banks with sizeable balance sheet exposures to impaired sovereign debt was negatively affected after the start of the sovereign debt crisis. Neri and Ropele (2013), using a FAVAR model, provide a country-specific quantification of the macroeconomic effects of the sovereign debt crisis and confirm that sovereign tensions, proxied by the Greek sovereign spread, have led to a significant deterioration in credit access for firms and households in peripheral countries. Zoli, 2013 and Bofondi et al., 2013 and Del Giovane et al. (2013) are particularly close to our study as they provide a quantification of the effects of the crisis on the Italian credit market, using different methodologies. Zoli (2013) estimates a VAR to study the pass-through of the sovereign spread on lending rates, finding it in the order of 30-40 percent within a quarter. Bofondi et al. (2013) use microeconomic data from the Credit Registry between December 2010 and December 2011 and estimate that tensions in the sovereign debt market reduced lending growth by Italian banks by 3 percentage points and increased interest rates by between 15 and 20 basis points. Del Giovane et al. (2013), using bank-level responses to the Euro area Bank Lending Survey to identify credit demand and supply, estimate that between 2010Q2 and 2012Q2 the restriction of credit standards (cumulatively) increased loan rates by around 150 bps and reduced loan growth by almost 5 percentage points. They show that the bulk of this effect can be attributed to the tensions in the sovereign debt market. Differently to these papers, our study focuses on the impact of the Italian sovereign debt crisis on a much wider array of banking (borrowing and lending) interest rates and breakdowns of loans to firms and households. We also provide a disaggregated analysis by bank size. Lastly, to the best of our knowledge, we are the first to consider the impact of the sovereign debt crisis on the main items of banks’ income statement. The rest of the paper is organized as follows. Section 2 briefly describes the sources of data and provides descriptive evidence of the variables used. Sections 3, 4 and 5 present the results of the analysis for banks’ funding and lending interest rates, for lending volumes, and for the main items of banks’ profit and loss statements, respectively. Each of these sections looks first at the results for the system as an aggregate and then distinguishes between small and large banks. Section 6 presents a concluding discussion of our main results.
نتیجه گیری انگلیسی
We have presented a comprehensive analysis of the quantitative effects of sovereign debt tensions, proxied by the 10-year BTP-Bund spread, on banking activity in Italy, focusing in particular on the crisis that started in 2010. Based on aggregate data for Italian banks, we have first estimated the pass-through of changes in the sovereign spread to bank funding and lending rates. Our estimation results clearly show that the evolution of the 10-year BTP-Bund spread significantly raises the interest rates on term deposits and newly issued bonds, and even more so on lending rates firms and households. Moreover, there is evidence of non-linearity in the transmission, with the estimated pass-through increasing during periods characterized by a high level of the spread. We have also found evidence of a direct effect of the sovereign spread on loans developments: a 100 b.p. increase in the sovereign spread reduces annual lending growth by around 1 percentage point for households and 0.7 for firms. Finally, we find a mildly positive relation between the spread and the net interest income as well as on loan loss provisions. These results suggest that the sovereign spread transmits to lending condition also through an increase in risk premia and not just the through an increase in bank funding costs. In order to study potential heterogeneity in the transmission of sovereign debt tensions to the banking sector we have separately analyzed the five largest banks and the other, smaller, intermediaries. This distinction is relevant because the two groups of intermediaries differ along various balance-sheet characteristics, business models and lending strategies. We find that banks’ balance-sheet characteristics do matter for the transmission of the sovereign debt tensions and that, as expected, the effect is more pronounced for larger banks consistently with the fact that these intermediaries are less capitalized, have a larger funding gap and rely more on non-traditional banking activities. A convenient way to assess the overall impact of the 2010-11 sovereign crisis on credit markets is to run a counterfactual exercise to assess how interest rates and lending growth would have behaved if the BTP-Bund spread had remained unchanged at the level observed before the sovereign crisis. Fig. 5 illustrates the counterfactual evolution of lending rates and annual growth of loans to firms and households for house purchase (together with the actual developments), under the assumption that the sovereign spread remains unchanged at 70 b.p. (i.e. the level observed in 2010Q1).29 The results indicate that in 2011Q4, under the hypothetical scenario, the cost of lending would have been about 170 and 120 b.p. lower, respectively for firms and households, than its actual value (Panels A and B in Fig. 5). In both cases, around a half of the final effect is attributable to the cumulative tensions between 2010Q2 and 2011Q3. The other half of the increase reflect the strong tensions recorded in the final quarter of 2011, during which the BTP-Bund spread increased about 160 b.p.30 As for the amount of lending (Panels C and D), the exercise indicates that the sovereign tensions reduced the (annual) growth of loans to both firms and households by about 2 percentage points in 2011Q4. As shown by the decomposition in the figure, for loans to firms this reduction can be attributed for nearly three quarters to the direct effect of the rise in the sovereign spread and for one quarter to the indirect effect (i.e., via the increase in loan rate); for loans to households the contribution of the indirect effect is very small.31In terms of policy implications, our results warrant support to the actions undertaken by the European Central Bank during the sovereign debt crisis: the sharp reduction in official interest rates helped containing the increase in borrowing cost following the tensions in sovereign debt markets; the implementation of unconventional operations such as the 1-year LTROs, as well as the Securities Markets Programme and the Covered Bonds Purchase Programme, helped alleviating the negative effect of sovereign risk on banks’ funding conditions thus attenuating in turn the impact on the flow of credit to the non-financial private sector.