PREZENTĀCIJAS NOSAUKUMS PROGRAMMES FOR FIRM PERFORMANCE*

Transcription

1 THE IMPORTANCE OF THE EU REGIONAL SUPPORT PREZENTĀCIJAS NOSAUKUMS PROGRAMMES FOR FIRM PERFORMANCE* Konstantīns Beņkovskis (LB, SSE Riga), Oļegs Tkačevs (LB), Naomitsu Yashiro (OECD) First Annual Workshop of the ESCB Research Cluster, November 16, 017 * The views expressed are those of the authors and do not necessarily reflect the views of Latvijas Banka or the OECD

2 EU FUNDING : MAIN OBJECTIVES Convergence ERDF Innovation Environment/ risk prevention Accessibility Infrastructure ESF Human resources Administrative capacity Cohesion Fund Transport (TENs) Sustainable transport Environment Renewable energy Regional Competitiveness and Employment ERDF Innovation Environment/risk prevention Accessibility ESF European Employment Strategy European territorial Cooperation ERDF Innovation Environment/risk prevention Accessibility Culture, education Source: European Commission

3 SOME RELATED STUDIES A positive effect of participation in EU farm programs on individual farm sales in Germany in , with no effect on farm productivity (Pufahl and Weiss, 008) The EU support of R&D is found to have a positive effect on Spanish companies' internal investment in R&D and employment (Arce and San Martin, 016) The effect of the EU regional policy (ESF+ERDF+KF) on regional firm performance in almost all EU countries in was found to be insignificant by de Zwaan and Merlevede (014) 3

4 DATA USED IN THE ANALYSIS (1) EU funding in by Fund, mln EUR,408 1,989 1,540 1, ERDF ESF KF Number of entities receiving EU funds ERDF ESF KF Allocated EU funds Database Used in the analysis EU funds database Used in the analysis Source: esfondi.lv; EU funds database 4

5 DATA USED IN THE ANALYSIS () Firms' indicators comprehensive database (Statistical office, State Revenue service) Balance sheet data, profit and loss statements Employment, value added Sector, two-digit NACE classification Merchandise trade database (Statistical office) Value and volume Product, eight-digit Combined Nomenclature (CN8) Destination Service trade database (Bank of Latvia) Value 47 types of services External assets and liabilities (Bank of Latvia) FDI Countries of origin Employee-employer database (Statistical office, State Revenue service) Gross wage Employment status, gender and age EU funds database (Ministry of Finance) Total number of firms in the dataset All firms Excluding small firms (<10 empl.)

6 FIRMS RECEIVING EU FUNDS EXHIBIT HIGHER PRODUCTIVITY log of labour productivity distribution, log of TFP distribution, Firms receiving EU funding appear to be more productive as compared to other Latvia's firms We don't however know the causality of this relationship 6

7 HOW DO WE DEFINE TREATMENT IN THIS STUDY? Treatment group: Firms that had not received ERDF funding until the period t 1 and started an ERDF funded project in period t Control group: Firms that did not take part in the ERDF support until t+3. Treated No ERDF Embark on the ERDF funded project?? Control No ERDF No ERDF No ERDF No ERDF t 1 t t+1 t+ 7

8 METHODOLOGICAL FRAMEWORK OF THE STUDY (1) We use propensity score matching (PSM) provides methods to match similar firms and create a counterfactual control group STEP 1: Define a control group In order to identify the probability of starting an ERDF funded project (propensity score) we estimate a probit model: Pr EU i,t = 1 = F(Lprod i,t 1, size i,t 1,, year t, sector i ) 8

9 METHODOLOGICAL FRAMEWORK OF THE STUDY () STEP : Define a control group Based on the propensity scores (PS) we match each firm in each treatment group to 1, or 5 nearest neighbours among the potential controls (with and without a caliper) We ensure that both come from the same -digit NACE sector and same year We apply the common support condition: dropping entrants whose PS is higher than max/lower than min than control group. Calculate the Difference-in-difference estimator (DiD) treated control DiD s = 1 Y N i,t+s w ij Y j,t+s T i,t T j,t C Y i,t+s = ln y i.t+s ln y i.t 1 ; s 0,1, 9

10 METHODOLOGICAL FRAMEWORK OF THE STUDY (3) STEP 3: Estimate the heterogenous effects on firm performance Estimate the equation stating the DiD estimator s years after the start of a project as a function of pre-treatment characteristics and project parameters: Y i,t+s j,t C F i are firm characteristics Z i are project parameters w ij Y j,t+s = α 0 + α 1 F i + α Z i + α 3 Macsec i + α 4 Year i + e i,t 10

11 LARGE, YOUNG, PRODUCTIVE, EXPORTING FIRMS HAVE A HIGHER PROBABILITY TO ENGAGE IN ERDF Probit model Log(labour productivity) t Log(labour productivity) t *** Age t *** Age t *** Log(employment) t *** Log(employment) t Log(capital to labour ratio) t *** Log(capital to labour ratio) t *** Liquidity ratio t Indebtedness ratio t Exports of goods t *** Exports of services t FDI to OECD t *** FDI to non-oecd t Employees with EU funds experience t Managers with EU funds experience t Number of observations Pseudo R 0.5 Probability to start receiving ERDF funds is higher for firms: with higher labour productivity that are larger and recently established with higher capital-to-labour ratios that do not belong to foreign companies from OECD goods exporting firms 11

12 MATCHING QUALITY Indicators Unmatched NN(1) NN() NN(5) NN(1) NN() NN(5) without caliper caliper=0.05 Log(labour productivity) t *** Log(labour productivity) t *** Age t *** Age t-1.0*** Log(employment) t *** * Log(employment) t *** ** Log(capital to labour ratio) t-1 9.7*** Log(capital to labour ratio) t Log(liquidity ratio) t Log(indebtedness ratio) t Exports of goods t *** Exports of services t-1 8.8* Log(FDI to OECD) t-1 3.6*** Log(FDI to non-oecd) t *** Employees_eu funds exper t Managers_eu funds exper t *** Setting a caliper to 0.05 drops 10 observations but ensures a more successful matching 1

13 t t+1 t+ t t+1 t+ t t+1 t+ t t+1 t+ t t+1 t+ t t+1 t+ IT TAKES TIME FOR PRODUCTIVITY PREMIA TO MATERIALIZE Total factor productivity Capital to labour ratio 3-15% 5% Labour productivity Number of employees 8% 6% 4% % 6% 4% % Nominal wage Exports to turnover Firms that obtain ERDF funding: increase capital-to-labour ratio immediately increase the number of employees manage to eventually raise both TFP and labour productivity as well as employees' remuneration * nearest neighbours with caliper 13

14 WHY IT TAKES TIME? before EU funds first year of EU funded project afterwards 14

15 baseline 3 months lag level and growth baseline 3 months lag level and growth baseline 3 months lag level and growth baseline 3 months lag level and growth baseline 3 months lag level and growth baseline 3 months lag level and growth ROBUSTNESS CHECK: DIFFERENT MATCHING STRATEGIES 5% 15% 5% 4 3 Total factor productivity Capital to labour ratio 5% 15% 5% 19% 19% 18% 18% Labour productivity Number of employees 11% 9% 9% 8% 6% 4% % Nominal wage Exports to turnover Two robustness checks: matching ERDF beneficiaries that start a project during the last 3 months of a year with non-beneficiaries in the next year; matching with nonbeneficiaries also on the back of productivity growth (not just level) * nearest neighbours with caliper 15

16 Science and innovation Enterpreneurship support Infrastructure and human capital Other 1st quintile nd quintile 3rd quintile 4th quintile 5th quintile Rīga Rīga region Kurzeme Latgale Vidzeme Zemgale Various regions 1st quintile nd quintile 3rd quintile 4th quintile 5th quintile THE EFFECT ON LABOUR PRODUCTIVITY DIFFERS 75% 5 5% -5% -5 Region Initial labour productivity 4 3 Activity Initial size (employment) * nearest neighbours with caliper: effect on labour productivity at t+ period 16

17 HETEROGENEITY OF DID: REGRESSION ANALYSIS Factors affecting treatment effect on treated in t+ TFP Labour Capital to Exports to Wage Employment productivity labour ratio turnover ratio log(tfp) t *** *** ** 0.03 log(employment) t *** 0.56*** * Age t * log(k/l ratio) t *** Risk of the project log(project size) Science and innovation ** Entrepreneurship support Environment * Year effects Yes Yes Yes Yes Yes Yes Macroeconomic sectors effects Yes Yes Yes Yes Yes Yes Region effects Yes Yes Yes Yes Yes Yes Number of observations R The effect on productivity is larger for initially bigger and less productive firms; the opposite holds for employment The effect on wages is more pronounced for younger firms 17

18 THE IMPACT OF THE FUNDING SOURCE Ederveen et al (003) shows that EU funding replaces a private one to a certain degree Czarnitzki and Lopes Bento (014) look at the effect of national subsidies for innovation in Germany compared to, or in combination with, the effect of European subsidies on innovation and R&D intensity. The study finds that EU subsidies have smaller impact on firms sales In order to test for the impact of funding source we ensured that a paired control firm has experienced a similar increase in capital-to-labour ratio (rough proxy for similar investments) as a treated firm during the threeyear period (comparing t+ with t 1) 18

19 MATCHING QUALITY neighbours Indicators Unmatched 5 groups of K/L growth 10 groups of K/L growth no caliper caliper=0.05 no caliper caliper=0.05 Log(lprod) t *** Log(lprod) t *** 1.4* * 8.6 Age t *** Age t-1.0*** Log(employment) t *** 5.9*** *** 19.1** Log(employment) t *** 3.7*** 15.* 43.1*** 1.3** Log(capital to labour ratio) t-1 9.7*** 11.6* * 10.1 Log(capital to labour ratio) t Log(liquidity ratio) t Log(indebtedness ratio) t Exports of goods t *** ** -.7 Exports of services t-1 8.8* Log(FDI to OECD) t-1 3.6*** Log(FDI to non-oecd) t *** Employees_eu funds exper t Managers_eu funds exper t *** Capital to labour growth (between t+ and t-1) 39.6*** Setting a caliper drops around 40 observations but ensures a more successful matching in particular when matching is implemented within 5 quintiles of capital growth 19

20 EU FUNDED PROJECTS RESULT IN EXPANDING FIRMS' SIZE Difference-in-difference (DiD) neighbours 5 groups of K/L growth 10 groups of K/L growth no caliper caliper=0.05 no caliper caliper=0.05 Log(TFP) t t t ** 0.19** Log(labour productivity) t t t * Log(wage) t t ** 0.064** * t+ 0.07** 0.088** Log(capital to labor ratio) t t t Log(employment) t 0.105*** 0.103*** 0.096*** 0.091*** t *** 0.154*** 0.159*** 0.151*** t+ 0.18*** 0.19*** 0.196*** 0.195*** Exports to turnover ratio t t * 0.06 t * 0.08** 0.034** Number of treated Number of control Controlling for investments size EU funding leads to a: larger increase in labour productivity and TFP in the third year larger increase in wages and the number of employees 0

21 DECOMPOSITION OF PRIVATE FINANCING INTO LOANS AND OWN RESOURCES Difference-in-difference (DiD) for various sources of capital financing (5 groups of K/L growth; caliper=0.05) ERDF financing vs predominantly loans ERDF financing vs predominantly own resources Log(TFP) t t t * Log(labour productivity) t t t Log(wage) t t t ** Log(capital to labor ratio) t t t * 0.00 Log(employment) t 0.17*** 0.090*** t *** 0.13*** t+ 0.94*** 0.0*** Exports to turnover ratio t *** t ** t ** Number of treated 76 3 Number of control

22 CONCLUSIONS EU support is clearly associated with an increase in company's size and capital, bringing about larger output and sales The effect on productivity is not clearly evident in the first two periods, while companies manage to catch up starting from period 3 This effect on productivity is larger for initially bigger and less productive firms; the opposite is the case for the effect on employment These findings point out at lags in capital utilization due to several possible reasons: knowledge gaps insufficient local market size lack of export expansion When firms undertake investment activities using EU funds they tend to grow bigger as compared to privately financed investments.