New Data and Methods Used for Policy Analysis in the Czech National Bank

Transcription

1 New Data and Methods Used for Policy Analysis in the Czech National Bank Kamil Galuščák Chief Economists workshop Bank of England, 19 May 2016

2 Wealth of information available on Internet Micro-data at disposal at Czech National Bank General agreement with Czech Statistical Office allowing exchange of data collected CNB supervises Czech financial system Outline of presentation Use of on-line data for near-term forecast Individual-level wage data: estimation of tax optimisation effects Financial stability analysis 2

3 On-line data for near-term forecast Internet allows valuable data to be gathered for near-term forecast, particularly from e-shops, but government data are also available There is vast number of e-shops operating in Czech Republic, focusing mainly on retail and food In addition, several comparison web servers gathering data from these e-shops and offering comparison of prices operate as well (food, retail, fuel, real estate prices) 3

4 On-line data for near-term forecast Publicly available government databases are rare in Czech Republic, but open data initiatives create pressure to publish government data Public procurement database is available online and updated daily Currently we web-scrap data from Tesco, Metro Cash & Carry, biggest Czech retail e-shop, fuel card company, real-estate web server and government public procurement database 4

5 On-line data for near-term forecast Tesco index Data are obtained daily by webscrapping Tesco e-shop E-shop prices are linked to one Prague store, so price changes cannot be considered representative of Czech food inflation Index computed from data gives at least qualitative assessment of near-term food inflation 5

6 On-line data for near-term forecast Fuel price index Fuel price index is computed from data from fuel card company (CCS), which publishes fuel prices at pumps its cards were used at Since company fuel card coverage is quite broad, fuel price index matches official fuel price inflation very well In this case, therefore, fuel price index serves not just as qualitative, but also as quantitative indicator of near-term fuel price inflation 6

7 On-line data for near-term forecast Public procurement vs government investment Public procurement data are obtained quarterly at the time new forecast is being made Public procurement data fit government consumption with 3- quarter lead well. This allows us to forecast government investment 3 quarters ahead, which is important especially at times of high volatility such as in See box in Inflation Report I/2016 7

8 On-line data for near-term forecast Decomposition of investment In current CNB forecasting practice, external demand is used for private investment forecast, while public procurement data are used for government investment forecast This distinction allowed us to explain strong overall investment growth in 2015 and forecast its sharp decline in

9 Individual-level wage data Average Earnings Information System provided by Czech Ministry of Labour and Social Affairs It contains information on level and structure of remuneration as well as on working hours paid Firms with 250 or more employees are all included; smaller firms with up to 249 employees are sampled 9

10 Individual-level wage data Income taxation change (effective since 2013): incentives for high-earning employees to draw bonuses in advance Income tax rate for employees earning more than four times average wage was increased by 7 p.p. Ceiling on health insurance was abolished These legislation changes were known sufficiently in advance and thus led to tax optimisation behaviour Observed wage growth in business sector in 2012Q4 was not in line with economic fundamentals 10

11 Individual-level wage data We used data on irregular bonuses among firms with 250 or more employees Substantial differences across firms and in timing Part of observed changes in behaviour could be due to cyclical reasons => 20% cut-off was used During period, irregular bonuses declined only in

12 Individual-level wage data Wage growth in business sector (annual percentage changes; seasonally adjusted) Tax optimisation influenced wage growth in business sector throughout 2013 Highest effect was estimated in manufacturing and banking sectors Adjusted time series is much more in line with other economic fundamentals Sensitivity analysis revealed robustness of our results 12

13 Financial stability analysis Chart II.45 Share of financially distressed households by income quintile (%) st quintile 2nd quintile 3rd quintile 4th quintile 5th quintile Total Baseline Scenario 2015 Adverse Scenario Source: Household Budget Statistics, CNB calculation Note: The shares are given for households that have some sort of loan. However, the income quintiles are based on the full sample of households. Micro data provided by Czech Statistical Office Household budget statistics for household stress tests (see CNB WP 2/2014) Use of micro data on financial indicators of non-financial enterprises 13

14 Financial stability analysis Chart IV.25 Relationship between LTV and LTI for new loans (x-axis: LTV in %; y-axis: share of loans in %) Source: CNB < > 100 LTI < 3 LTI LTI LTI LTI LTI LTI > 5.5 CNB is integrated supervisor (both micro and macroprudential functions) Access to monetary statistics and supervisory reporting Individual supervisory data and information available to macroprudential part of CNB Conduct of Bank Lending Survey (specific questions) Possibility to conduct ad hoc data surveys (e.g. survey of characteristics of housing loans, survey of commercial real estate lending standards, survey of recovery rates) 14

15 Financial stability analysis Use of data from Central Credit Register run at CNB Non-financial enterprises only (households are covered by separate private registers) Linking with other data sources (financial indicators of firms) Credit risk monitoring, calculation of default rate, lending standards indicator, credit risk vintage analysis, differences in client risk classification across banks, ad hoc analysis (developers, solar business) See Financial Stability Report 2014/2015 article Use of the Czech central credit register for financial stability purposes 15

16 Financial stability analysis More granular data required recently for macroprudential purposes Bank-by-bank and loan-by-loan information on mortgages from 2015H2 (LTV, LTI, DSTI, maturity, ) before AnaCredit starts Selected information from private credit registers for assessment of default rates in consumer credit segments (structure of loans by region, age groups, etc.) 16

17 Financial stability analysis Database of merged transaction and price data collected by Czech cadastre (ongoing CNB research project) Calculation of own aggregate property price index, price indices for different market segments, and number of transactions realised Aim is to solve existing data limitations and use database in supervisory practice Use of price index model for assessment of banks own housing appraisal processes (introduction of LTV limits by CNB in June 2015) 17

18 Thank you for your attention Kamil Galuščák Executive Director Economic Research Department 18