Reallocation, Competition and Productivity: [1.2ex ... - Google Sites

Reallocation, Competition and Productivity: Evidence from a Financial Liberalization Episode Liliana Varela

November 2016

−

Capital market distortions can be a source of misallocation and reduce TFP.

−

This paper: 1. shows that effect can be amplified through product market competition. 2. tests this empirically by focusing on distortions in the access to international capital markets.

Financial Liberalization and Productivity Cross-country studies associate financial liberalization with increases in aggregate TFP. 1+1 .96 .92 TFP 1.08 1.04 t-3 t-2 tt-1 t+1 t+2 t+3 Year Note: TFP .04 TFP Growth level is normalized Following to the year Financial of the reform Liberalization

.92

.96

TFP 1

1.04

1.08

TFP Growth Following Financial Liberalization

t-3

t-2

t-1

t Year

t+1

t+2

t+3

Note: TFP level is normalized to the year of the reform

70 countries, 1970-2004. Chinn-Ito Index of Financial Liberalization

Contribution I

1) This paper proposes a particular mechanism:

→

Capital controls distort competition and all firms’ investments in technology.

→

Financial liberalization can foster productivity growth through two forces: 1. Previously policy-discriminated firms increase investment in technology. 2. Pro-competitive forces lead non-discriminated firms to do the same.

Contribution II

2) This paper tests this mechanism using a particular liberalization episode:

→

Liberalization of international borrowing in Hungary in 2001.

→

Why Hungary?

− No other reform occurred at that time. − Capital controls created asymmetries in international borrowing: → Home & Foreign firms.

→

Micro data: firm-level census data.

Preview Model: − Asymmetric access to international borrowing between Home & Foreign firms. − Financial liberalization fosters productivity growth through two forces: 1. Better financing terms leads Home firms to increase their innovation. 2. Deeper competition leads Foreign firms to do the same.

Main Empirical Results: 1. Reallocation towards Home firms: − Better financial terms and increased leverage. − Capital intensity, productivity and innovation. 2. Presence of pro-competitive forces on Foreign firms: − Markups decrease. − Reductions in industry concentration and TFP dispersion. 3. Expansion in aggregate TFP growth, driven by within-firm productivity.

Related Literature Finance and Growth: • Financial Liberalization: Bekaert, Harvey & Lundblad (2005 & 2011), Bonfiglioli (2008), Chari, Henry & Sasson (2012), Levine (2001), Levchenko, Rancière & Thoenig (2009), Kose, Prasad, Rogoff & Wei (2006)... • Financial Development: King & Levine (1993), Rajan & Zingales (1998), Beck et al (2005), Townsend & Ueda (2007),... • Capital Controls: Forbes (2003 & 2007), Devereux & Yetman (2016), Schmitt-Grohe & Uribe (2016), Blanchard et al (2016)...

Misallocation: • Restuccia & Rogerson (2008), Hsieh & Klenow (2009), Buera, Kaboski & Shin (2011), Midrigan & Xu (2012), Peters (2013), Bollard, Klenow & Sharma (2013), Gopinath et al (2016)...

This Talk

1. Model 2. Liberalization of International Borrowing in Hungary 3. Empirics I. Data & Identification Strategy II. Empirical Results

Model

→ Develop a small economy model with three main ingredients: 1. Endogenous investment in technology, and direct competition (Bertrand). 2. Local banking sector (low financial development). 3. Capital controls create asymmetries in external finance across firms.

Capital controls distort competition and all firms’ innovation efforts.

Mechanism

• One-period, one sector and two equally productive firms (H & F). • If successful innovation → the firm becomes the industry leader.

Capital controls distort profits...

→ If Foreign firms successfully innovate their profits are higher: - Markups are higher (cost advantage): ξ F =

p MC F

=

MC H MC F

= λ τ.

where λ: technology advantage, τ : difference in financing terms b. H& F (>1)

→ If Home firms successfully innovate their profits are lower: - Markups are lower (cost disadvantage): ξ H =

...and their optimal innovation efforts.

p MC H

=

MC F MC H

=

λ τ

Mechanism con’t

→

In the symmetric economy, firms’ optimal innovation intensities, x(F ) and x(H) :

x(F ) = •

Y w

1 τ

(1 − λ−1 ) Y w

x(H) = (1 − τ λ−1 ) Y w

: market size in labor units.

• λ: increment in technology. τ : wedge between the domestic and the foreign interest rates.

Distortions in capital markets reduce Home and Foreign firms’ innovation.

Mechanism con’t

→

In the asymmetric economy, firms’ optimal innovation intensities, x(F ) and x(H) :

x(F ) = •

Y w

1 τ

(1 − λ−1 ) Y w

x(H) = (1 − τ λ−1 ) Y w

: market size in labor units.

• λ: increment in technology. • τ : wedge between the domestic and the foreign interest rates (>1).

→

Distortions in capital markets reduce Home and Foreign firms’ innovation.

Model’s Implications

Financial Liberalization can lead to two different outcomes: → If financial development is sufficiently high: − capital inflows and increase in TFP growth.

→ Instead, if financial development is low: − capital outflows and decrease in TFP growth.

Model’s Implications con’t Financial Liberalization can lead to two different outcomes: → If financial development is sufficiently high (capital inflows): 1. All firms’ innovation efforts increase, relatively more for Home firms: ∂x(H) ∂τ

< 0,

∂x(F ) ∂τ

∂x(F )

∂x(H)

< 0, and | ∂τ | < | ∂τ |.

2. Home firms increase their leverage: 3. Foreign firms’ markups decline:

∂ξ(F ) ∂τ

∂L(H) ∂τ

< 0.

> 0.

4. The productivity gap between Home and Foreign firms decreases: 5. Aggregate productivity growth increases:

∂gq ∂τ

< 0.

→ Instead, if financial development is low (capital outflows): − All these predictions are reversed. Capital Controls

∂∆ ∂τ

>0

This Talk


Capital Controls in Hungary before 2001 → Foreign exchange (FX) market regulations were the main tool of capital controls.

1. Restrict banks’ ability to intermediate foreign funds: → Spot and Forward FX markets: − Forward: banned all instruments to hedge the currency risk. − Spot: made very costly and difficult to acquire foreign currency. → Critical: costly and illiquid spot market and inexistent forward market. → Banks relied their credit supply on local savings, leading to low credit.

Reform

In 2000

Hungary

OECD

Credit-to-GDP Ratio

0.27

0.86

Credit-to-Deposit Ratio

0.83

1.20

Capital Controls in Hungary before 2001 con’t

2. Regulate firms’ international borrowing: → The regulations divided firms into 2 groups:

− Home firms: only take credit locally in national currency. − Foreign firms: directly access to international funds. IMF on Hungary (1998): Foreign firms enjoyed 2 sources of foreign funds: - "foreign credit ... facilitated by the relationship b/ the parent company and its bank". - Internal capital markets with parent companies: 35% of total credit.

→ Difference in credit conditions between Home and Foreign Firms.

− Home Firms: - paid higher interest rates (4pp more). - faced a higher value for the required collateral (61pp more). - had a lower level of leverage (44% less).

Deregulation of Capital Controls (2001) -4 RoW: 3 Local 6 4 2 0 -2 1 5 25 20 15 10 1997 1999 2001 2003 2005 Banks: Note: Assets Liabilities Total Swaps 997 inDerivatives 0 FX billions External Net Market: Capital of USDebt dollars. Daily Inflows Turnover Source: NBH and IMF

Local FX Market: Daily Turnover 4

4

RoW: Derivatives Assets

Total

2

3

Swaps

0

-4

1

-2

0

2

Liabilities

1997

1999

2001

2003

2005

1997

2001

2003

2005

Banks: External Debt

-2

5

0

10

2

15

4

20

6

25

Banks: Net Capital Inflows

1999

1997

1999

2001

2003

2005

1997

1999

2001

2003

2005

Note: in billions of US dollars. Source: NBH and IMF

Impact on the Credit Market

Aggregate Economy (in %)

Before

After

Credit-to-GDP Ratio

27

44

Credit-to-Deposit Ratio

83

113

Lending interest rate

12.8

7.5

Credits to SME

34

51

SME debt in FX

0

33

Interest rate differential b. Home and Foreign

4.2

0.38

Differential in collateral b. Home and Foreign

61

23

Firms

Notes: For rows 1-5 the source is the National Bank of Hungary, and data corresponds to December 2000 and December 2004. Rows 5-6 come from Business Environment and Enterprise Performance Survey of the World Bank and EBRD, 2001 and 2004.

Data and Identification Strategy

→ Two databases: − APEH (National Bank of Hungary): census data on manufacturing firms (1992-2008). − BEEPS (World Bank and EBRD): representative surveys on R&D and innovation, financing terms.

→ Identification Strategy: − Firm-level analysis, 3 sources of variation: 1. Time: reform (2001). 2. Cross-sectional: Home vs Foreign firms. 3. Cross-sectional: Home vs Foreign firms across sectors (external finance). − Similar trends before the reform.

Growth Trends of Home and Foreign Firms

1.1

1.06

1.09

1.15

1.08 1.06 1

1

1

1.03

1.05

1.04 1.02

Capital Intensity

1.12

RTFP 1.2

Labor Productivity

1996

1999

2002

2005

2008

1996

Note: 1996=1

Markup

2002

2005

2008

1996

1999

2002

2005

Note: 1996=1

Leverage 1.6 1.4

Home Foreign

.85

1

.9

1.2

1 .95

1.8

1.05

1999

Note: 1996=1

1996

1999

Note: 1996=1

2002

2005

2008

1999 2001 2003 2005 2007 Note: 1999=1

2008

This Talk


Test Prediction 1: Investments in Technology

All firms’ investments in technology increase, but Home firms relatively more.

→ Test it in two steps: I. Differential impact on Home firms: 1. Capital intensity, labor productivity and RTFP. 2. R&D and innovation activities.

II. Financial channel: 3. Capital intensity, labor productivity, RTFP and skill-intensity. 4. Leverage and financing terms.

1. Capital Intensity, Labor Productivity, RTFP Regress: • Differential impact on Home firms: δ2

yit = δ0 Hi + δ1 Tt + δ2 (Hi x Tt ) + εit where T = 1 if year>2001, 0 otherwise; H dummy for home firms

• Estimate first-differences at firm-level:

∆yi = δ1 + δ2 Hi + ∆εi where ∆ yi = log( 13

P2004 2002

yit ) − log( 13

P2000 1998

yit )

• Controls: • Firm-level: size, age, productivity in the initial year (1998). • Industry-level: -Local trends: pre-growth trends in productivity and capital intensity (1996-97) in Hungary at 4-digit level; -Global trends: growth rate in productivity and capital intensity in the US at 4-digit level. • Cluster the standard errors at 4-digit industries.

1. Results on Capital Intensity, Labor Productivity, RTFP

Greater expansion of Home firms.

∆ Capital Intensity (1) Home

(2)

(3)

∆ Labor Productivity (4)

(5)

(6)

∆ RTFP (7)

(8)

0.239*** 0.253*** 0.252*** 0.074*** 0.051*** 0.053*** 0.098*** 0.032** (0.023) (0.025) (0.025) (0.017) (0.017) (0.016) (0.015) (0.014)

Firm controls

yes

Local trends

yes

yes

yes

Global trends

yes

yes

yes

yes

(9) 0.032** (0.016) yes yes

yes

yes

R2

0.019

0.030

0.030

0.004

0.027

0.040

0.008

0.075

0.088

N

5,448

5,448

5,448

5,448

5,448

5,448

5,448

5,448

5,448

Notes: *, **, *** significant at 10, 5, and 1 percent. Std errors are clustered at 4-digit NACE industries. Global industry controls include capital intensity and TFP growth rate of the 4-digit NACE industries in the US between 1998-04. Local industry controls are capital intensity and RTFP average growth rate at 4-digit level in Hungary in the late 90s. Firm-level controls are age, employment and RTFP in the initial year (1998). Source: APEH




(2)

(3)


(5)

(6)

∆ RTFP (7)

(8)

0.239*** 0.253*** 0.252*** 0.074*** 0.051*** 0.053*** 0.098*** 0.032** (0.023) (0.025) (0.025) (0.017) (0.017) (0.016) (0.015) (0.014)

Firm controls

yes

Local trends

yes

yes

yes

Global trends

yes

yes

yes

yes

(9) 0.032** (0.016) yes yes

yes

yes

R2

0.019

0.030

0.030

0.004

0.027

0.040

0.008

0.075

0.088

N

5,448

5,448

5,448

5,448

5,448

5,448

5,448

5,448

5,448





(2)

(3)


(5)

(6)

∆ RTFP (7)

(8)

0.239*** 0.253*** 0.252*** 0.074*** 0.051*** 0.053*** 0.098*** 0.032** (0.023) (0.025) (0.025) (0.017) (0.017) (0.016) (0.015) (0.014)

Firm controls

yes

Local trends

yes

yes

yes

Global trends

yes

yes

yes

yes

(9) 0.032** (0.016) yes yes

yes

yes

R2

0.019

0.030

0.030

0.004

0.027

0.040

0.008

0.075

0.088

N

5,448

5,448

5,448

5,448

5,448

5,448

5,448

5,448

5,448


2. R&D and Innovation Activities

Regress: • Differential impact on Home firms: δ2

yit = δ0 Hit + δ1 Tt + δ2 (Hit x Tt ) + εit where T = 1 if year=2004, 0 if year=2001, y dummy if the firm undertook R&D/ innovation activities.

• Controls: • Firm-level: size and age. • Sector-fixed effects. • Cluster the standard errors at sector level.

2. Results on R&D and Innovation Activities Greater expansion of Home firms.

R&D Activities

Innovation Activities

(1)

(2)

(3)

(4)

(5)

(6)

Home

-0.153*** (0.028)

-0.058 (0.032)

-0.032 (0.030)

-0.242*** (0.057)

-0.158** (0.054)

-0.090 (0.056)

Home*Reform

0.107* (0.048)

0.083** (0.033)

0.090* (0.044)

0.176** (0.066)

0.167** (0.055)

0.122* (0.056)

Reform

0.023 (0.055)

0.046 (0.052)

0.023 (0.043)

-0.084 (0.063)

-0.071 (0.075)

-0.099 (0.081)

yes

yes

yes

yes

Firm-level controls Sector-fixed effects

yes

yes

R2

0.019

0.064

0.081

0.014

0.037

0.069

N

774

774

774

774

774

774

Notes: *, **, *** significant at 10, 5, and 1 percent. Std errors are clustered sector level. All regressions include a constant term. R&D is a dummy if the firm reports positive R&D expenditures. Innovation is a dummy if the firm reports any of the following activities: developed successfully a major product line, upgraded an existing product line, acquired a new production technology, obtained a new licensing agreement, and obtained a new quality accreditation. Firm-level controls are age and size. Source: BEEPS.

2. Results on R&D and Innovation Activities Greater expansion of Home firms.

R&D Activities

Innovation Activities

(1)

(2)

(3)

(4)

(5)

(6)

Home

-0.153*** (0.028)

-0.058 (0.032)

-0.032 (0.030)

-0.242*** (0.057)

-0.158** (0.054)

-0.090 (0.056)

Home*Reform

0.107* (0.048)

0.083** (0.033)

0.090* (0.044)

0.176** (0.066)

0.167** (0.055)

0.122* (0.056)

Reform

0.023 (0.055)

0.046 (0.052)

0.023 (0.043)

-0.084 (0.063)

-0.071 (0.075)

-0.099 (0.081)

yes

yes

yes

yes


yes

yes

R2

0.019

0.064

0.081

0.014

0.037

0.069

N

774

774

774

774

774

774

Notes: *, **, *** significant at 10, 5, and 1 percent. Std errors are clustered sector level. All regressions include a constant term. R&D is a dummy if the firm reports positive R&D expenditures. Innovation is a dummy if the firm reports any of the following activities: developed successfully a major product line, upgraded an existing product line, acquired a new production technology, obtained a new licensing agreement, and obtained a new quality accreditation. Firm-level controls are age and size. Source: BEEPS.

3. Financial Channel: Capital Intensity, Labor Productivity and RTFP

Exploit cross-sectional variation in terms of: • Sector financial needs (Rajan and Zingales, 1998).

Regress: • Pro-competitive forces: F firms in accordance with sectors’ financial needs: δ3 . • Differential impact on H firms in accordance with sectors’ financial needs: δ4 .

∆yij = δ1 + δ2 Hi + δ3 FDj + δ4 (Hi x FDj ) + ∆εij where FDj is the financial dependence index at 4-digit industries.

3. Financial Channel: Capital Intensity, Labor Productivity and RTFP

Exploit cross-sectional variation in terms of: • Sector financial needs (Rajan and Zingales, 1998).

Regress: • Pro-competitive forces: F firms in accordance with sectors’ financial needs: δ3 . • Differential impact on H firms in accordance with sectors’ financial needs: δ4 .

∆yij = δ1 + δ2 Hi + δ3 FDj + δ4 (Hi x FDj ) + ∆εij where FDj is the financial dependence index at 4-digit industries.

3. Results on the Financial Channel: Capital Intensity, Labor Productivity and RTFP

∆ Capital Intensity (1)

(2)

(3)


(5)

∆ RTFP

(6)

(7)

0.015 (0.017)

0.083*** -0.010 (0.018) (0.023)

(8)

(9)

Home

0.210*** 0.221*** 0.219*** 0.058*** 0.017 (0.021) (0.024) (0.024) (0.017) (0.015)

Home * FD

0.142* (0.080)

0.156* (0.076)

0.155* (0.077)

0.093* (0.053)

0.155*** 0.147*** 0.087 (0.045) (0.046) (0.072)

0.181** (0.080)

0.167** (0.067)

Fin. Dep.

-0.084 (0.064)

-0.061 (0.070)

-0.053 (0.077)

0.276** (0.124)

0.320** (0.124)

0.222** (0.10)

0.277*** (0.092)

yes

yes

Firm controls Local trends

yes

yes

Global trends

0.334** (0.134)

0.162 (0.107)

yes

yes

yes

yes

-0.016 (0.023)

yes yes

yes

yes

R2

0.020

0.031

0.031

0.034

0.074

0.081

0.022

0.111

0.120

N

5,143

5,143

5,143

5,143

5,143

5,143

5,143

5,143

5,143

Notes: *, **, *** significant at 10, 5, and 1 percent. Std errors are clustered at 4-digit NACE industries. Financial Dependence is Rajan and Zingales index (1998). Global industry controls include capital intensity and TFP growth rate of the 4-digit NACE industries in the US between 1998-04. Local industry controls are capital intensity and RTFP average growth rate at 4-digit level in Hungary in the late 90s. Firm-level controls are age, employment and RTFP in the initial year (1998). Source: APEH.



(2)

(3)


(5)

∆ RTFP

(6)

(7)

0.015 (0.017)

0.083*** -0.010 (0.018) (0.023)

(8)

(9)

Home

0.210*** 0.221*** 0.219*** 0.058*** 0.017 (0.021) (0.024) (0.024) (0.017) (0.015)

Home * FD

0.142* (0.080)

0.156* (0.076)

0.155* (0.077)

0.093* (0.053)

0.155*** 0.147*** 0.087 (0.045) (0.046) (0.072)

0.181** (0.080)

0.167** (0.067)

Fin. Dep.

-0.084 (0.064)

-0.061 (0.070)

-0.053 (0.077)

0.276** (0.124)

0.320** (0.124)

0.222** (0.10)

0.277*** (0.092)

yes

yes


yes

yes

Global trends

0.334** (0.134)

0.162 (0.107)

yes

yes

yes

yes

-0.016 (0.023)

yes yes

yes

yes

R2

0.020

0.031

0.031

0.034

0.074

0.081

0.022

0.111

0.120

N

5,143

5,143

5,143

5,143

5,143

5,143

5,143

5,143

5,143




(2)

(3)


(5)

∆ RTFP

(6)

(7)

0.015 (0.017)

0.083*** -0.010 (0.018) (0.023)

(8)

(9)

Home

0.210*** 0.221*** 0.219*** 0.058*** 0.017 (0.021) (0.024) (0.024) (0.017) (0.015)

Home * FD

0.142* (0.080)

0.156* (0.076)

0.155* (0.077)

0.093* (0.053)

0.155*** 0.147*** 0.087 (0.045) (0.046) (0.072)

0.181** (0.080)

0.167** (0.067)

Fin. Dep.

-0.084 (0.064)

-0.061 (0.070)

-0.053 (0.077)

0.276** (0.124)

0.320** (0.124)

0.222** (0.10)

0.277*** (0.092)

yes

yes


yes

yes

Global trends

0.334** (0.134)

0.162 (0.107)

yes

yes

yes

yes

-0.016 (0.023)

yes yes

yes

yes

R2

0.020

0.031

0.031

0.034

0.074

0.081

0.022

0.111

0.120

N

5,143

5,143

5,143

5,143

5,143

5,143

5,143

5,143

5,143




(2)

(3)


(5)

∆ RTFP

(6)

(7)

0.015 (0.017)

0.083*** -0.010 (0.018) (0.023)

(8)

(9)

Home

0.210*** 0.221*** 0.219*** 0.058*** 0.017 (0.021) (0.024) (0.024) (0.017) (0.015)

Home * FD

0.142* (0.080)

0.156* (0.076)

0.155* (0.077)

0.093* (0.053)

0.155*** 0.147*** 0.087 (0.045) (0.046) (0.072)

0.181** (0.080)

0.167** (0.067)

Fin. Dep.

-0.084 (0.064)

-0.061 (0.070)

-0.053 (0.077)

0.276** (0.124)

0.320** (0.124)

0.222** (0.10)

0.277*** (0.092)

yes

yes


yes

yes

Global trends

0.334** (0.134)

0.162 (0.107)

yes

yes

yes

yes

-0.016 (0.023)

yes yes

yes

yes

R2

0.020

0.031

0.031

0.034

0.074

0.081

0.022

0.111

0.120

N

5,143

5,143

5,143

5,143

5,143

5,143

5,143

5,143

5,143




(2)

(3)


(5)

∆ RTFP

(6)

(7)

0.015 (0.017)

0.083*** -0.010 (0.018) (0.023)

(8)

(9)

Home

0.210*** 0.221*** 0.219*** 0.058*** 0.017 (0.021) (0.024) (0.024) (0.017) (0.015)

Home * FD

0.142* (0.080)

0.156* (0.076)

0.155* (0.077)

0.093* (0.053)

0.155*** 0.147*** 0.087 (0.045) (0.046) (0.072)

0.181** (0.080)

0.167** (0.067)

Fin. Dep.

-0.084 (0.064)

-0.061 (0.070)

-0.053 (0.077)

0.276** (0.124)

0.320** (0.124)

0.222** (0.10)

0.277*** (0.092)

yes

yes


yes

yes

Global trends

0.334** (0.134)

0.162 (0.107)

yes

yes

yes

yes

-0.016 (0.023)

yes yes

yes

yes

R2

0.020

0.031

0.031

0.034

0.074

0.081

0.022

0.111

0.120

N

5,143

5,143

5,143

5,143

5,143

5,143

5,143

5,143

5,143

Notes: *, **, *** significant at 10, 5, and 1 percent. Std errors are clustered at 4-digit NACE industries. Financial Dependence is Rajan and Zingales index (1998). Global industry controls include capital intensity and TFP growth rate of the 4-digit NACE industries in the US between 1998-04. Local industry controls are capital intensity and RTFP average growth rate at 4-digit level in Hungary in the late 90s. Firm-level controls are age, employment and RTFP in the initial year (1998). Source: APEH



(2)

(3)


(5)

∆ RTFP

(6)

(7)

0.015 (0.017)

0.083*** -0.010 (0.018) (0.023)

(8)

(9)

Home

0.210*** 0.221*** 0.219*** 0.058*** 0.017 (0.021) (0.024) (0.024) (0.017) (0.015)

Home * FD

0.142* (0.080)

0.156* (0.076)

0.155* (0.077)

0.093* (0.053)

0.155*** 0.147*** 0.087 (0.045) (0.046) (0.072)

0.181** (0.080)

0.167** (0.067)

Fin. Dep.

-0.084 (0.064)

-0.061 (0.070)

-0.053 (0.077)

0.276** (0.124)

0.320** (0.124)

0.222** (0.10)

0.277*** (0.092)

yes

yes


yes

yes

Global trends

0.334** (0.134)

0.162 (0.107)

yes

yes

yes

yes

-0.016 (0.023)

yes yes

yes

yes

R2

0.020

0.031

0.031

0.034

0.074

0.081

0.022

0.111

0.120

N

5,143

5,143

5,143

5,143

5,143

5,143

5,143

5,143

5,143


3. Results on the Financial Channel: Skill-Intensity Differential increase in skill-intensity across firms and sectors.

∆ Skill-Intensity

Home

(1)

(2)

(3)

(4)

0.003* (0.002)

0.014** (0.006)

0.013** (0.006)

-0.014** (0.006)

Home * FD

0.041** (0.016)

Fin. Dep.

0.021** (0.009)

Firm-level control

yes

yes

Local trend

yes

yes

Global trends

yes

yes

0.038 1,221

0.053 1,179

R N

2

yes

0.004 1,221

0.028 1,221

Notes: *, **, *** significant at 10, 5, and 1 percent. Std. errors are clustered at 4-digit NACE industries. All regressions include a constant term. Financial Dependence is the Rajan and Zingales index (1998). Global industry controls include capital intensity and TFP growth rates of the 4-digit NACE industries in the United States between 1998 and 2004. Local industry controls are capital intensity and RTFP average growth rates at 4-digit level in Hungary in the late 90s. Firm-level controls are age, employment and RTFP in the initial year (1998). Source: APEH.

3. Results on the Financial Channel: Skill-Intensity Differential increase in skill-intensity across firms and sectors.

∆ Skill-Intensity

Home

(1)

(2)

(3)

(4)

0.003* (0.002)

0.014** (0.006)

0.013** (0.006)

-0.014** (0.006)

Home * FD

0.041** (0.016)

Fin. Dep.

0.021** (0.009)

Firm-level control

yes

yes

Local trend

yes

yes

Global trends

yes

yes

0.038 1,221

0.053 1,179

R N

2

yes

0.004 1,221

0.028 1,221

Notes: *, **, *** significant at 10, 5, and 1 percent. Std. errors are clustered at 4-digit NACE industries. All regressions include a constant term. Financial Dependence is the Rajan and Zingales index (1998). Global industry controls include capital intensity and TFP growth rates of the 4-digit NACE industries in the United States between 1998 and 2004. Local industry controls are capital intensity and RTFP average growth rates at 4-digit level in Hungary in the late 90s. Firm-level controls are age, employment and RTFP in the initial year (1998). Source: APEH.

Test Prediction 2: Financing Terms Financing terms decreased for Home firms.

Interest Rate

Value of Collateral

(1)

(2)

(3)

(4)

(5)

(6)

Home

4.253*** (1.132)

3.707*** (1.027)

3.729*** (1.051)

60.789*** (15.391)

49.174** (15.727)

52.106*** (11.263)

Home*Reform

-3.879** (1.134)

-3.858*** (1.018)

-3.947*** (1.076)

-37.653* (17.130)

-35.438* (17.104)

-31.170** (10.911)

Reform

-0.026 (0.951)

-0.159 (0.830)

-0.221 (0.890)

20.968 (12.571)

19.574 (13.192)

13.368 (11.635)

yes

yes

yes

yes


yes

yes

R2

0.175

0.202

0.217

0.035

0.045

0.103

N

415

415

415

399

399

399

Notes: *, **, *** significant at 10, 5, and 1 percent. Std errors are clustered industry level. All regressions include a constant term. Firm-level controls are age and size. Source: BEEPS.

Test Prediction 2: Firms’ Leverage Home firms increased their leverage.

∆ Leverage

Home

(1)

(2)

(3)

(4)

0.160** (0.073)

0.239*** (0.085)

0.230*** (0.088)

0.238** (0.100)

Home* FD

0.526** (0.266)

Fin. Dep.

-0.595** (0.234)

Firm-level controls

yes

Local trends Global trends

yes

yes

yes

yes

yes

yes

R2

0.002

0.006

0.007

0.015

N

2,742

2,742

2,742

2,457

Notes: *, **, *** significant at 10, 5, and 1 percent. Std errors are clustered at 4-digit NACE industries. Financial Dependence is the Rajan and Zingales index (1998). Global industry controls include capital intensity and TFP growth rate of the 4-digit NACE industries in the US between 1998-04. Local industry controls are capital intensity and RTFP average growth rate at 4-digit level in Hungary in the late 90s. Firm-level controls are age, employment and RTFP in the initial year (1998). Source: APEH.

Test Prediction 2: Firms’ Leverage Home firms increased their leverage.

∆ Leverage

Home

(1)

(2)

(3)

(4)

0.160** (0.073)

0.239*** (0.085)

0.230*** (0.088)

0.238** (0.100)

Home* FD

0.526** (0.266)

Fin. Dep.

-0.595** (0.234)

Firm-level controls

yes

Local trends Global trends

yes

yes

yes

yes

yes

yes

R2

0.002

0.006

0.007

0.015

N

2,742

2,742

2,742

2,457


Test Prediction 3: Foreign Firms’ Markups

Regress: • The model predicts larger declines for F firms: δ2

∆ξij = δ1 + δ2 Fi + ∆εij where Fi is a dummy for foreign firm.

• Markup: as a wedge between firm’s labor share (θijt ) and labor elasticity (βj ). ξijt =

1 βj θijt

Results on Foreign Firms’ Markups Foreign firms’ markups declined.

∆ Markups

Foreign

(1)

(2)

(3)

(4)

-0.017* (0.009)

-0.025** (0.011)

-0.026** (0.012)

0.030* (0.016)

Foreign*FD

-0.205*** (0.043)

Fin. Dep.

0.212*** (0.069)

Firm-level control

yes

Local trend Global trends

yes

yes

yes

yes

yes

yes

R2

0.000

0.023

0.024

0.057

N

5,376

5,376

5,376

5,086


Results on Foreign Firms’ Markups Foreign firms’ markups declined.

∆ Markups

Foreign

(1)

(2)

(3)

(4)

-0.017* (0.009)

-0.025** (0.011)

-0.026** (0.012)

0.030* (0.016)

Foreign*FD

-0.205*** (0.043)

Fin. Dep.

0.212*** (0.069)

Firm-level control

yes

Local trend Global trends

yes

yes

yes

yes

yes

yes

R2

0.000

0.023

0.024

0.057

N

5,376

5,376

5,376

5,086


Robustness Tests

-

Alternatives estimations of RTFP and markups (Wooldridge, Levinsohn and Petrin 2011; De Loecker and Warzynski 2012; and PCM).

-

Effect by Year.

-

Falsification Test: 1998.

-

Unbalanced sample: panel regressions.

-

Sector and Sector-Year-fixed effects (4-digit).

-

Export status.

-

Ownership status.

-

Export platforms.

-

Change in ownership status.

Taking Stock

→ Test two forces: 1. Reallocation towards Home firms: capital intensity, labor productivity, RTFP and R&D and innovation activities (prediction 1).

− Financial channel is key: leverage increased (prediction 2), particularly in sectors with greater financial needs.

2. Evidence of pro-competitive forces: Foreign firms increased their productivity, RTFP, skilled intensity, and decreased markups in sectors where the distortion was larger (prediction 3).

− Reduction in RTFP and markup gaps, and concentration (prediction 4). − Acceleration of RTFP growth (prediction 5), driven by within-firm RTFP. Concentration Results

Sources of Aggregate Productivity Growth 1. Aggregate productivity growth increases. 2. Change in the source of aggregate productivity growth:

− Driven by increases in within-firm productivity (82%).

Total Sample ∆RTFP

(1)

Reallocation

(2)

Balanced Panel Within-

Within-

Firm

Firm

(3)

(4)

A- Mean Growth Rate Before

5.8

4.8

1.0

0.9

After

9.7

1.7

7.9

7.3

B- Contribution to Aggregate RTFP Growth (column 1) Before

100.0

83.4

16.5

16.5

After

100.0

18.0

82.0

75.4

Sources of Aggregate Productivity Growth 1. Aggregate productivity growth increases. 2. Change in the source of aggregate productivity growth:

− Driven by increases in within-firm productivity (82%).

Total Sample ∆RTFP

(1)

Reallocation

(2)

Balanced Panel Within-

Within-

Firm

Firm

(3)

(4)

A- Mean Growth Rate Before

5.8

4.8

1.0

0.9

After

9.7

1.7

7.9

7.3

B- Contribution to Aggregate RTFP Growth (column 1) Before

100.0

83.4

16.5

16.5

After

100.0

18.0

82.0

75.4

Conclusions

→

The effect capital market distortions can be magnified through competition.

→

I tested this mechanism in presence of asymmetries to international borrowing.

→

Financial Liberalization can foster TFP growth through two forces: 1. Reallocation towards discriminated firms → invest in technology. 2. Pro-competitive forces lead non-discriminated firms to do the same.

Extra Slides

Banks and the Direction of Capital Flows

− Consider a perfectly competitive financial sector of risk-neutral banks: D L R(t+1) = R(t+1) − µ.

− Banks need to pay a tax τ˜ per unit of foreign transaction. − The direction of capital flows depends on the level of financial development (µ):

Condition on R D D R(t+1)

>

∗ R(t+1)

International Capital Flows

Use foreign savings to lend locally

Capital Inflows

D ∗ R(t+1) < R(t+1) − τ˜

Use local savings to lend abroad

Capital Outflows

∗ D ∗ R(t+1) − τ˜ < R(t+1) < R(t+1) + τ˜

Use local savings to lend locally

Closed Economy

Return

+ τ˜

Banks’ Optimal Behavior

Hungary Prior to the Reform (2001)

Major reforms had already taken place by mid-1990s:

• Trade and FDI liberalization: fully achieved by 1995 • Bank deregulation: fully achieved in 1997 • Privatization of public companies: in 1997 the share of public firms in the VA of the manufacture sector was 2%

• Market competition: Competition Act 1997

Evolution of Trade with the EU Return

.85 .8 .75 .7 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Manufacturing Exports Total Trade Sector: Trade with the EU

.7

.75

.8

.85

Manufacturing Sector: Trade with the EU

1995

1996 1997 1998

1999 2000 2001 Exports

2002 2003 2004

Total Trade

2005

FDI Evolution

10 8 6 4 2 0 12 10 1997 1998 1999 2000 2001 2002 2003 2004 2005 Hungary: FDI to GDP

0

2

4

6

8

10

12

Hungary: FDI to GDP

1997

1998

1999

2000

2001

2002

2003

2004

2005

Volume of Trade Evolution

300 100 2 0 Exports 800 600 400 200 1994 1996 1998 2000 2002 Imports 1995=100 Volume Developed Developing 00 of Imports and Exports Volume of Imports and Exports

1995=100 800

Imports

0

100

200

200

400

300

600

400

Exports

1994

1996

1998

Developed

2000

2002 Developing

1994

1996

1998

Developed

2000

2002 Developing

Capital Inflows Evolution

6Portugal 4 2 -2 -4 Transition 1 5 0 -5 -10 10 1997 1999 2001 2003 2005 SOE Note: Deregulated Net Czech Poland Hungary Brazil 997 inRepublic 0 Capital billions Economies US Flows dollars. Inflows: IMF Financial Institutions Net Capital Inflows: Financial Institutions

Deregulated Flows SOE

-4

-10

-2

-5

0

0

2

5

4

6

10

Transition Economies

1997

1999

2001

Hungary Poland

Note: in billions US dollars. IMF

2003

2005

Czech Republic

1997

1999

2001

Hungary Portugal

2003

2005 Brazil

Test Prediction 4: Change in the Productivity Gap and its Initial Level

Return

Definitions:

→ RTFP and Markup Gap between Foreign and Home Firms: • Compute the median of F and H firms in each 3-digit sector j • Gap: κjt = p50Fjt − p50Hjt , ∆κjt = κjt − κjt−1

→ To control for pre-trends, I use 3 periods and regress ∆κjt = β1 κjt + β2 T + β3 (κjt ∗ T ) + εjt

→ Concentration: • Concentration: Cjt =

P ij

Dijt ∗ Lernerijt , ∆Cjt = Cjt − Cjt−1

Regression Results

∆ RTFP Gap

Initial Value

∆ Markup Gap

∆ Concentration

Reform

Incl. Pre-trends

Reform

Incl. Pre-trends

Reform

Incl. Pre-trends

(1)

(2)

(3)

(4)

(5)

(6)

-0.202** (0.079)

-0.076 (0.077)

-0.730*** (0.135)

-0.419*** (0.079)

-0.317*** (0.085)

-0.177*** (0.060)

Initial Value* T

-0.222** (0.107)

-0.310** (0.140)

-0.245*** (0.091)

T

0.186 (0.128)

0.134** (0.054)

0.211*** (0.072)

R2

0.074

0.100

0.280

0.325

0.145

0.223

N

82

164

78

156

82

164

Notes: all regressions include a constant. *, **, ***significant at 10, 5, and 1 percent. Std errors in parenthesis. 3-digit NACE industry correlations. Source: APEH.