Outline

1. Does the law of one price holdin retail banking?An analysis of interest rates national differences in the euro areaMassimiliano Affinito and Fabio Farabullini(Banca d’Italia, Research Department)8th Conference of the ECB-CFS Research NetworkMadrid 30 November 2006

2. Outline • Motivations • Methodology • Data sources • Results

3. (1) Motivations of the work

4. a large stream of literature exists on the integration of national credit markets in the euro area • regarding price convergence, • a clear theoretical background seems to exist: the law of one price • a clear benchmark: full convergence once returns and risks are taken into account • recent harmonized data on euro area bank interest rates permit for the first time: • a consistent cross-country comparison • to assess, in a static sense, the current degree of similarity between national average rates (occurred convergence)

5. The law of one price identical goods  identical prices the law holds integration / no segmentation Conditions factors at work • competitive markets  deregulation programs • no transaction costs  intangibility of banking products • no barriers to trade  persistent national peculiarities: • different characteristics of the real economy: customer preferences and borrower risk profile • persistence of regulation barriers: legal system, fiscal framework, supervision authority • presence of specific banking practices: informational asymmetries, relationship banking, market power

6. The focus is twofold: at the market segments level: which rates are more similar? 14 types of bank interest rates: 5 categories of deposit interest rates 5 categories of lending interest rates to households 4 categories of lending interest rates to non-financial corporations (b) at national markets level: which countries are more similar? 12 euro area countries

7. (2) Methodology

8. The analysis is divided into three steps: • First step: unconditional test of the cross-country equality of interest rates (we use two different econometric approaches) • Second step: the cross-country equality is controlled for national demand side characteristics • Third step: the cross-country equality is controlled for national supply side characteristics, too

9. A benchmark of integration:the 12 euro area countriesversusthe 20 Italian regions • the results of euro area cross-country equality of interest rates are compared with Italian cross-region ones • the banking system of a country should be integrated (single legal system, similar customers, etc.), and therefore it should represent a benchmark for assessing the level of euro area integration

10. First step The homogeneity of interest rate is analyzed with two approaches: • First approach: tests of zero-mean stationarity of differentials based on ADF and KPSS tests, applied to the bilateral differentials δtij = ri, t – rj, t • Second approach: tests of equality of estimated country coefficients the second approach is used in all the steps of the analysis

11. Second approach • 14 regressions: one for each type of banking interest rate • first step: the independent variables are just time and country dummies • second step: national “demand side” characteristics • third step: national “supply side” characteristics • statistical tests of the significance between each pair of coefficients of country dummies • counting the number of cases in which the bilateral differentials are not significant (in other words the number of cases in which the interest rates are similar) • the percentage shares of the number of similar interest rates on the total number of pairs of interest rates

12. Why three steps? • the first step examines the existence of cross-country interest rate homogeneity/heterogeneity on the raw data • the further steps “clean up” the data, controlling for those factors that explain the differences in interest rate levels (segmentation of credit markets) • in fact, the law states “identical prices for identical goods”… the regressors allow to homogenize banking products and then to really verify the validity of the law

13. In formal terms 14 regressions: ri, t = α'tTi, t +β'i Di, t + γ'g Xi, t + δ'h Zi, t + εi, t, ri, t is the interest rate, specific to each regression, for country i in month t; t = 1, 2, …, 27 months (from January 2003 to March 2005); i = 1, 2, …, n countries; α, β, γ, δ are vectors (nt x 1) of coefficients; Ti, t is a matrix (nt x t) of time (monthly) dummies; Di, t is a matrix (nt x i) of country dummies; Xi, t is a matrix (nt x g) of demand side regressors; Zi, t is a matrix (nt x h) of supply side regressors; g and h indicate the number of regressors, different in each regression, respectively, in matrix Xi, t and in matrix Zi, t; εi, t is an error term.

14. Test of equality of estimated coefficients for each country(bilateral differentials): Null hypothesis: H0: βi = βj Hypothesis test based on statistic: F[1, 27n – k] Significance level: 5 %

15. Second step: + γ'g Xi, t “demand side” regressors factors influencing interest rate setting behavior related to the characteristics of bank depositors and borrowers: • GDP growth rate  all rate categories • Disposable income  rates to households • Risk exposure  lending rates • Alternative financing sources  lending rates to non-financial firms • Alternative forms of saving  deposit rates • Firms’ average size  lending rates to non-financial firms

16. Third step: + δ'h Zi, t “supply side” regressors those determinants of rates that depend on banking system characteristics • Bank balance sheet characteristics • Bank operating costs  all rate categories • Bank non-interest income  all rate categories • Bank liquidity  all rate categories • Bank capitalization  all rate categories • Bank liability structure  deposit rates • Bank asset structure  lending rates • Banking system structural characteristics • Banks' international presence  all rate categories • Banking market concentration  all rate categories • Bank average size  all rate categories • Bank M&As  all rate categories

17. Why do we distinguish the effect of the “demand” regressors from the “overall” effect ? • the law states “identical prices for identical goods”… • … we try to define homogeneous banking products… • … the demand side factors certainly define a product... (e.g. a loan is comparable if the risk profile of the borrower is comparable) • it is disputable whether the same goes for the supply side factors, but it is reasonable… (e.g. market power also differentiates the perception of goods)

18. (3) Data sources

20. (4) Results

21. (4) Results • the determinants of bank interest rates (interesting in itself and instrumental for the analysis of…) • the validity of the law

22. The determinants of national differences in euro area bank interest rates Summary econometric results

23. The determinants of national differences in euro area bank interest rates Summary econometric results

24. The determinants of national differences in euro area bank interest rates Summary econometric results

25. First step results: euro area countries versus Italian regions

26. First step results: • euro area banking market is still segmented • inter-country dispersion is greater than intra-country dispersion • even at the national level the homogeneity of interest rates is not full • the dispersion on loans is higher than on deposits

27. Percentage shares of statistically similar bank interest ratesFirst, Second and Third step Outline by market segments

28. Remarks by market segments • (as expected) the similarities progressively increase: • moving from the unconditional tests (first step) • to those based on demand regressors (second step) • up to those based on all the regressors (third step) • interest rates tend to be more homogeneous where the bank customer is likely to be stronger (greater market power or better information): • repos customers versus checking-account customers • corporations versus households • large versus small firms

29. Percentage shares of statistically similar bank interest ratesFirst, Second and Third step Outline by country

30. Remarks by country • the percentage share of non-significant differentials progressively grows for all countries • geographical proximity and other elements of natural and structural “closeness” between banking systems do not systematically influence the similarity of interest rates

31. SUMMING UP • prima facie, the law of one price does not hold in the euro area retail banking markets in spite of the long European integration process and single monetary policy, the euro area banking markets appear still segmented • nevertheless, some market segments are more integrated than others where bank customers are stronger • moreover, if we take into account “demand side” and “supply side” regressors many differences disappear • therefore, the euro area prices appear different, because national bank products appear different or because they are differentiated by national factors • there is scope for some further interest rate convergence… if banking services become more similar, the prices will become more similar as well