The Impact of the Financial Crisis on UK Company Performance

Transcription

1 The Impact of the Financial Crisis on UK Company Performance Rebecca Riley*, Chiara Rosazza Bondibene* and Garry Young** *National Institute of Economic and Social Research & CFM **Bank of England & CFM 24 June 2014 Using Secondary Data Analysis to Research Economic Performance Economic Performance, Policy and Management Cluster Roundtable ESRC Secondary Data Analysis Initiative NIESR, London Disclaimers: Any views expressed cannot be taken to represent those of the Bank of England or to state Bank of England policy. This work contains statistical data which is Crown Copyright; it has been made available by the Office for National Statistics (ONS) through the Secure Data Service (SDS) and has been used by permission. Neither the ONS nor SDS bear any responsibility for the analysis or interpretation of the data reported here. This work uses research datasets which may not exactly reproduce National Statistics aggregates. Acknowledgements: The financial support of the Economic and Social Research Council grant reference ES/K00378X/1 is gratefully acknowledged.

2 Motivation Stylised facts UK labour productivity fell sharply during the recession of and recovered only sluggishly after that In level terms productivity is currently around 15% below a simple extrapolation of its pre-crisis trend In sharp contrast with the experience of other post-war recessions in the UK when the fall was less steep and the recovery was quicker By 2012 the stock of real bank debt held by UK corporations was more than 20% below its peak before the crisis, much of which reflected a tightening of credit supply

3 Productivity and bank lending following recessions Constant Price GDP Per worker ( 000, per quarter) Real stock of PNFC Bank debt outstanding ( bn)

4 What can we learn about the banking crisis and productivity by looking at firm level data? A banking crisis may inhibit efficient resource allocation across businesses Using firm level data we can examine whether aggregate productivity weakness arises because of resource misallocation between existing firms and/or a lack of creative destruction or cleansing effect of recession And may hinder investment growth Using firm level data we can measure these effects by comparing credit constrained bank dependent firms to similar firms that do not rely on bank finance or that were not credit constrained

5 Using firm-level data to break down productivity growth into different components Overall productivity growth = Average productivity growth within surviving businesses + Reallocation towards more productive surviving businesses + Reallocation towards new businesses + Reallocation from exiting businesses

6 Productivity dynamics: Related literature Other studies analyse the pattern of productivity dynamics in the wake of a financial crisis: Japan during the 1990s Griffin and Odaki (2009) US manufacturing firms during the Great Recession Foster, Grim and Haltiwanger (2013) UK firms during the Great Recession Barnett, Chiu, Franklin and Sebastia-Barriel (2014)

7 Data: the Annual Respondents Database (ARD) Establishment level survey which covers businesses in the non-financial nonfarm market sectors Data available for (now 2012) and for manufacturing back to 1974 A census of larger businesses and a stratified random sample of businesses with less than 250 employees It contains basic information for all businesses in the sampling frame (the IDBR, a list of all UK businesses registered for tax purposes) making it possible to: Determine business entry and exit Calculate grossing weights We focus on firms with 10 or more employees since the longitudinal sample is insufficient to support representative analysis of micro firms: The probability of observing a micro business in two separate years (conditional on being live) is only 1 in 10,000 We aggregate the data up to the enterprise level, as banking relationships are more likely to take place at this level of aggregation Financial information is published in current values GVA deflators available at 2- and 3-digit sector level are used to construct real values

8 Trends in labour productivity, UK ARD decomposition sample ABI & ABS Market sector Differences between the series expected because of differences in sector and size coverage and in cleaning and weighting procedures All these series exhibit broadly the same pattern over time, with labour productivity in 2011 a little more than 10% below a simple linear extrapolation of the trend

9 Decomposition of 4-year changes in labour productivity Growth components External Within Between Entry Exit Net entry Total Total Productivity growth(%) Productivity growth change (% points) to to Source: Annual Respondents Database, ONS, and authors' calculations. Notes: Growth components Within, Between, Entry and Exit sum to Growth Total. Entry and Exit sum to Net entry. Between, Entry and Exit sum to External Total. Non-farm non-financial market sectors excluding mining & quarrying, utilities and real estate activities. Britain. Firms are classified as live if they are active and have 10 or more persons employed.

10 Decomposition of 4-year changes in labour productivity (continued) Sub-group Growth components External Productivity growth change (% points) Within Between Entry Exit Net entry Total Total Manufacturing to to Services SMEs to to to to Large to to Source: Annual Respondents Database, ONS, and authors' calculations. Notes: Growth components Within, Between, Entry and Exit sum to Growth Total. Entry and Exit sum to Net entry. Between, Entry and Exit sum to External Total. Non-farm non-financial market sectors excluding mining & quarrying, utilities and real estate activities. Britain. Firms are classified as live if they are active and have 10 or more persons employed.

11 per annum 8% Decomposition of 1-year changes in labour productivity 6% 4% 2% 0% -2% % -6% -8% External Within Total

12 contribution per annum External restructuring and productivity growth, service sector SMEs, % 6% 4% 2% 0% -2% % -6% -8% More bank dependent sectors Other sectors Less bank dependent sectors Market services

13 Productivity dynamics: Conclusions The reduction in UK labour productivity between 2007 and 2011 was first and foremost the result of a broad-based decline in productivity within businesses Not a reduction in the contribution of business reallocation to aggregate productivity growth. The question of what has caused this productivity drop within firms remains (and banking sector collapse may be one reason). The recession does appear to have had some "cleansing effect" or been associated with creative destruction. Albeit not sufficient to offset fully the large drop in productivity within firms. We do observe patterns that suggest an empirical link between banking sector collapse and aggregate productivity via less efficient resource allocation Downward trend in the contribution of external restructuring due to SMEs in the more bank dependent sectors. Reduction in the productivity contribution of entering firms. Comparison of manufacturing firms in two different recessions suggests we might have expected a slightly higher productivity contribution from external restructuring (although key differences are due to the within component) Data limitations Exclusion of micro firms. What is the counterfactual? We cannot say with certainty what the productivity contribution of external restructuring would have been in the absence of a banking sector crisis.

14 A Quasi-Experiment Exploit exogenous variation induced by the financial crisis in credit availability to companies to investigate impacts of credit supply shocks Compare outcomes for companies who were subjected to tougher credit constraints to outcomes for companies that were less likely to be constrained Quasi-experimental approach Divide firm observations into treatment and control groups based on main bank lender Difficulty switching to a new lender during the crisis Provide direct estimates of the impact of credit constraints on UK firms Here we consider impacts on firm survival and the stock of fixed capital

15 Quasi-experimental evidence: Related literature Assessing the impact of credit constraints on real economic outcomes using variation in ease of access to external finance induced by the financial crisis in a natural experiment type approach Employment, unemployment and firm closure Chodorow-Reich, QJE, Bentolila, Jansen, Jiménez and Ruano, CEPR DP 9776, Duygan-Bump, Levkov and Montoriol-Garriga, Federal Reserve Bank of Boston QAU WP QAU10-6, Investment Almeida, Murillo, Laranjeira and Weisbenner, NBER WP 14990, Productivity (Krishnan, Nandy and Puri, NBER WP 20149, 2014.) On-going research at the Bank of England The importance of financial circumstances for firms investment decisions Difficulties in assessment due to issues of reverse causation See review by Bond, S. and J. Van Reenen (2007) in J. Heckman and E. Leamer, eds. Handbook of Econometrics

16 basis points The Different Experiences of UK Banks Bank funding costs (CDS premia) Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Jan 09 Jan 10 Jan 11 Jan 12 Jan From the Large Review: Having lent aggressively in the run-up to the crisis, RBS s lending volumes to SMEs have fallen faster than peers and its market share has contracted from an unsustainably high share in 2008, to a level more consistent with its customer base.

17 Data: Financial Analysis Made Easy (FAME) Company Accounts information held by Companies House provided by Bureau Van Dijk annual historical discs subsidiaries removed from the dataset Chargeholder recorded tells us which banks a company is borrowing from Data issues selective reporting of key accounts information reporting of employment and output is particularly sparse; better coverage of fixed assets decline over time in tendency to report detailed accounting information self-reporting of SIC codes

18 Difference-in-differences set-up Define Treatment (T) and Control (C) group T = Companies with an outstanding charge with RBS before the credit crunch and with above median short term leverage C = Companies with an outstanding charge with HSBC before the credit crunch or with RBS and with below median short term leverage Track difference in the development of outcomes between the T and C groups since the financial crisis FY 2007/8 - FY 2011/12 And compare this to differences in the development of outcomes between these two groups in a normal period (i.e. a pre-recession period) Formally Y Y RBS I ( t 2007 ) it 4 it it RBS it yr t ind 2 dig t X it it Where ɣ identifies the differential effect of credit constraints associated with being attached to a distressed bank Outcomes (Y) considered Log (1+Debt) Log (1+Fixed Assets) Indicator equal to one if a company is inactive (zero otherwise)

19 RBS borrowers relative to HSBC borrowers Debt held by companies with outstanding charges: RBS relative to HSBC Short term loans and overdrafts Long term debt Source: FAME BvD and authors calculations. Notes: Companies in the non-financial non-farm business sectors excluding the Mining, Utilities and Real Estate industries. Companies who do not have an outstanding charge with any other lender and who report their debt.

20 PNFC Switching Between Lenders Probability of changing lender after: 1 year 2 years 3 years 4 years All* Big Four HSBC RBS Source: FAME BvD and authors calculations. Notes: Companies in the non-financial non-farm business sectors excluding the Mining, Utilities and Real Estate industries. Companies who do not have an outstanding charge with any other lender. Big Four = RBS, Lloyds, HSBC, Barclays. Switching to another lender evaluated over the period Note that there was not a marked increase or decrease in this measure of switching around the time of the financial crisis. *Switching between 15 categories of lender

21 Sample Characteristics Control Treatment Control Treatment Exit rate Start-up Total asset distribution Young nd quintile Foreign owned rd quintile Exporter th quintile Foreign subsidiaries th quintile Count court judgment 0-24 mths Group accounts Normal credit score Full accounts Short term gearing Missing loans at time t Collateral Companies Notes: Companies in the non-financial non-farm business sectors excluding the Mining, Utilities and Real Estate industries. Companies who do not have an outstanding charge with any other lender and who report short term loans and overdrafts. Control group equals companies with an outstanding charge with HSBC before the credit crunch or with RBS and with below median short term leverage. Treatment group equals companies with an outstanding charge with HSBC before the credit crunch or with RBS and with below median short term leverage.

22 Short Term Loans and Overdrafts: Impact estimates DPV: 4-year change in log real loans Impact estimate Falsification test t=2004, 2007 t=2001, 2004 coeff pval coeff pval OLS all (0.109) (0.394) Robust regression all *** (0.001) * (0.056) OLS stayers ** (0.021) (0.526) Robust regression stayers *** (0.000) (0.225) Observations all stayers Notes: Sample includes firms that had an outstanding charge with either HSBC or RBS at times t and that did not have an outstanding charge with any other lender and for whom we observe short term loans. Treatment group includes firms with RBS with above median leverage. Companies in the non-financial non-farm business sectors excluding the Mining, Utilities and Real Estate industries. p-values in brackets. All regressions include the following controls, measured at time t: size (measured by the quintile in the distribution of total assets), group accounts, full accounts are filed, start-up (age 0-2 years), young (age 0-5 years), foreign owned, exporter, owns foreign subsidiaries, county court judgements in last 2 years, normal credit score, short term leverage (short term loans relative to total assets), collateral (fixed assets relative to total assets), filing quarter, 2-digit industry dummies and their interactions with times t, treatment group indicator. OLS standard errors clustered by chargeholder. Robust standard errors. Loans deflated by 2 and 3 digit industry GVA deflators. Stayers weighted to sample total by industry, year, size, and bank.

23 Fixed Assets: Impact estimates DPV: 4-year change in log real capital Impact estimate Falsification test t=2004, 2007 t=2001, 2004 coeff pval coeff pval OLS all (0.250) * (0.067) Robust regression all *** (0.000) (0.239) OLS stayers ** (0.014) (0.310) Robust regression stayers *** (0.000) (0.498) Observations all stayers Notes: Sample includes firms that had an outstanding charge with either HSBC or RBS at times t and that did not have an outstanding charge with any other lender and for whom we observe short term loans. Treatment group includes fimrs with RBS with above median leverage. Companies in the non-financial non-farm business sectors excluding the Mining, Utilities and Real Estate industries. p-values in brackets. All regressions include the following controls, measured at time t: size (measured by the quintile in the distribution of total assets), group accounts, full accounts are filed, start-up (age 0-2 years), young (age 0-5 years), foreign owned, exporter, owns foreign subsidiaries, county court judgements in last 2 years, normal credit score, short term leverage (short term loans relative to total assets), collateral (fixed assets relative to total assets), log capital, filing quarter, 2-digit industry dummies and their interactions with times t, treatment group indicator. OLS standard errors clustered by chargeholder. Robust standard errors. Capital deflated by 2 and 3 digit industry GVA deflators. Stayers weighted to sample total by industry, year, size, and bank.

24 Exit Rates: Impact estimate DPV: 4-year exit rate Impact estimate Falsification test t=2004, 2007 t=2001, 2004 coeff pval coeff pval probit *** (0.000) (0.767) Observations 48,058 51,997 Notes: Sample includes firms that had an outstanding charge with either HSBC or RBS at times t and that did not have an outstanding charge with any other lender and for whom we observe short term loans. Treatment group includes fimrs with RBS with above median leverage. Companies in the non-financial non-farm business sectors excluding the Mining, Utilities and Real Estate industries. p-values in brackets. All regressions include the following controls, measured at time t: size (measured by the quintile in the distribution of total assets), group accounts, full accounts are filed, start-up (age 0-2 years), young (age 0-5 years), foreign owned, exporter, owns foreign subsidiaries, county court judgements in last 2 years, normal credit score, short term leverage (short term loans relative to total assets), collateral (fixed assets relative to total assets), filing quarter, 2-digit industry dummies and their interactions with times t, treatment group indicator. Standard errors clustered by chargeholder. Marginal effects.

25 2007=100 Productivity amongst companies with outstanding charges: at LBG/RBS or HSBC/Barclays less distressed bank distressed bank Source: FAME BvD and authors calculations. Notes: Companies that report employment and turnover to Companies House and that have an outstanding charge with either LBG/RBS or HSBC/Barclays and with no other bank. Non-farm non-financial market sector excluding Real Estate, Utilities, Mining. Distressed banks LBG/RBS. Less distressed banks HSBC/Barclays. Productivity metric = median turnover per head.

26 Quasi-experimental evidence: Conclusions Did a credit supply shock reduce productivity? Not clear that forbearance is a major issue (outside real estate) Some evidence that an adverse supply shock has reduced growth in the capital stock, which in turn is likely to have reduced labour productivity growth Are credit constraints a key driver of recent productivity weakness? Our estimates measure only a partial effect Key explanations of recent productivity weakness need to be able to explain the weakness of productivity within companies Further research possibilities Direct estimates of the effects of credit constraints on productivity (and investment) The importance of capital-labour substitution in explaining productivity weakness Possibility of linking FAME to ONS datasets

27 References Riley, R., Rosazza-Bondibene, C., and G. Young (2014) The financial crisis, bank lending and UK productivity: sectoral and firm-level evidence, National Institute Economic Review, R17-R34. Riley, R., Rosazza-Bondibene, C., and G. Young (2014) Productivity Dynamics in the Great Stagnation: Evidence from British Businesses, Centre For Macroeconomics Discussion Paper No