Exchange Rate Exposure and Firm Dynamics Juliana Salomao University of Minnesota

Liliana Varela University of Warwick and University of Houston

October 2017

Motivation

− Capital inflows associate with increases in foreign currency borrowing in developing economies.

− This paper argues that FC borrowing arises from a trade-off between exposure to currency risk and firms’ growth.

Stylized Facts

→ Foreign currency borrowing in developing economies:

1. Firms hold high shares of foreign currency loans.

−

In 2014: Hungary 48%, Romania, 52%, Serbia 75%, Bulgaria 67%, Croatia 66%, Peru 55%...

2. Foreign currency borrowing associates with deviations from the risk-free UIP. 3. Large cross-sectional heterogeneity in the share of FC borrowing across firms.

Stylized Facts in Hungary 1. Share of foreign currency loans: 48% on total loans in 2014. 2. Correlation between deviations from risk-free UIP and firms’ FC loans. 3. Cross-sectional variation in firms’ FC borrowing: - 33% of firms borrowed in FC. - heterogeneity in the share of FC loans (66% of firms did not expo/impo). Distribution of Share of Foreign Currency Loans (2005) 50 40 In % 30

.45

0

1

10

1.05 1.1 Deviations from Risk-Free UIP

Ratio FC Loans on Total Loans .5 .55 .6

Deviations UIP (right)

20

FC/Total Loans (left)

1.15

.65

Foreign Currency Loans and Deviations from Risk-Free UIP

2005q4

E (st+1 ) st

0 2008q4

2011q4

2014q4

s

(1+r )

t (1 + rt∗ ) 6= (1 + rt ) → Devt = E (s t ) ∗ >1 t+1 (1+rt )

More

20 40 60 80 Share of Foreign Currency Loans on Total Loans

100

Contribution I

1) Build a firm dynamics model and proposes mechanism leading to FC borrowing: • Aggregate deviations from the risk-free UIP → but FC loans increase idiosyncratic risk of default.

Contribution I

1) Build a firm dynamics model and proposes mechanism leading to FC borrowing: • Aggregate deviations from the risk-free UIP → but FC loans increase idiosyncratic risk of default.

• Two sources of heterogeneity:

− Selection: productivity threshold to borrow in FC. − Heterogeneous share of FC loans across firms, driven by idiosyncratic risk. Trade-off between exposure to currency risk and firm’s growth.

Contribution II

2) Test this mechanism using firm-level census data on Hungarian firms: • Why Hungary?

− High levels of foreign currency borrowing. − Firm-level census data on all economic activities over 1996-2010. • Test model’s firm-level implications using simulated and Hungarian data.

Contribution III

3) Conduct numerical exercises: − Quantify the impact of FC borrowing on the aggregate. − Explore the role of selection, exchange rate volatility, and ER pass-though.

Preview of Main Results

→ Firm-level results 1. Deregulation of FC loans: firms using in FC are more productive and grow faster. 2. Deviations from risk-free UIP:

− increase FC borrowing and investment. − productive firms with low capital borrow more in FC and grow faster.

Preview of Main Results

→ Firm-level results 1. Deregulation of FC loans: firms using in FC are more productive and grow faster. 2. Deviations from risk-free UIP:

− increase FC borrowing and investment. − productive firms with low capital borrow more in FC and grow faster.

→ Quantitative analysis 1. Economies with FC borrowing see higher capital and sales, but are more volatile. 2. Selection is critical to generate gains from FC borrowing. 3. Less volatile exchange rates deepen FC borrowing.

Related Literature

• Balance sheet effects and crises: Krugman (1999), Aghion, Banerjee & Bacchetta (2001), Jeanne (2003), Caballero & Krishnamurthy (2003), Schneider & Tornell (2004), Cespedes, Chang & Velazco (2004), Eichengreen & Hausmann (2005), Kalantzis (2015)... • Country studies: Pratap & Urrutia (2004), Aguiar (2005), Beakley & Cowan (2008), Ranciere, Tornell & Vamvakidis (2010), Kim, Tesar & Zhang (2015), Kalemi-Ozcan et al (2013)... • Capital inflows and credit: Baskaya et al (2017), Maggiori, Neiman & Schreger (2017)...

Outline

1. Data 2. Model 3. Firm-Level Analysis: Model vs Data 4. Aggregate Implications: Numerical Exercises 5. Extensions − Model: Sensitivity Analysis − Empirics: Currency Depreciation during Great Recession

Data We use two datasets: 1. APEH: census data on all firms in the economy (1996-2010). 2. Credit Register: census data on loans by currency denomination (2005-2010). → Panel on all economic activities. Number of firms Sector

All

Borrowing in FC

(1)

(2)

A

Agriculture, forestry and fishing

7,511

748

B

Mining and quarrying

351

30

C

Manufacturing

22,656

3,083 50

D

Electricity, gas steam and air conditioning supply

357

E

Water supply, sewerage, waste management and remediation activities

1,099

119

F

Construction

19,334

1,738

G

Wholesale and retail trade, repair or motor vehicles and motorcycles

48,198

H

Transportation and storage

6,291

631

I

Accommodation and food service activities

9,305

611

J

Information and communication

8,153

351

M

Professional, scientific and technical activities

18,522

814

N

4,485

Administrative and support service activities

10,014

525

R

Arts, entertainment and recreation

3,933

97

S

Other service activities

4,935

211

160,659

13,493

Total

Notes: Nace Rev.2 Industry Classification. Source: APEH.

Data

• Financial liberalization in 2001 deregulated for FC borrowing. • In 2005: − 33% of firms had FC loans, accounted for 40% VA & 34% of employment. − SME accounted for 63% of FC loans, 14% of VA & 18% employment. − Firms had a large FC exposure: - 66% firms didn’t expo or impo, and had 66% of FC share. - Only 4% of firms use financial hedges (MNB 2006).

Firm Optimization Model

• DRTS, idiosyncratic productivity and exchange rate shocks. • Capital adjustment costs, fixed cost of operation. • External financing with debt:

− Local and foreign currency debt. − Debt is non-enforceable, firms can default. − Deviations from the risk-free UIP. − Fixed credit costs. • Endogenous firm entry and exit. • Partial equilibrium analysis.

Technology and Shocks

• Production:

→ Firms produce with F (z, k) = zk α where log z 0 = ρz log z + σz 0z ;

z ∼ N (0, 1)

• Exchange Rate & UIP: log s 0 = ρs log s + σs 0s ;

s ∼ N (0, 1) ;

θE (s 0 /s)(1 + r ∗ ) = s(1 + r )

s(LC/ FC) θ>1

Firms’ Problem → Incumbent Firms: V = max

V R (s, z, k, b, b ∗ ) =

max





V R, V D = 0




k 0 ,b 0 ,b 0∗

[e + β Ez 0 ,s 0 V s 0 , z 0 , k 0 , b 0 , b 0∗ ]

e = p[zk α −i(k, k 0 )−ψ(k, k 0 )−cf ]−[b +sb ∗ ]+[qb 0 +q ∗ sb 0∗ −pcI(b0 +b0∗ >0) −pcI∗ 0∗ (b

where

p = p∗ s η

(p ∗ = 1)

and

ψ(k, k 0 ) = c0



k 0 − (1 − δ)k k

→ Entrant Firms:

Ve (s, ν) =

max [−pk 0 + βEz 0 ,s 0 V (s 0 , z 0 , k 0 , b 0 , b 0∗ )]

k 0 ,b 0 ,b 0∗

if Ve (s, ν) ≥ pce .

2

k

>0)

]

Financing

• Debt: qb + sq ∗ b ∗ • Bond prices:

 q=



1 − Pz,s (∆(k,b,b ∗ ) )

where

1+r

∆k,b,b ∗ =



 and

q∗ =



1 − Pz,s (∆k,b,b ∗ ) 1 + r∗



(s, z) s.t. V R (s, z, k, b, b ∗ ) ≤ 0 .

,

Equilibrium Definition A recursive equilibrium is a set of functions for (i) V (s, z, k, b, b ∗ ) and Ve (s, ν), k 0 (s, z, k, b, b ∗ ), b 0 (s, z, k, b, b ∗ ), b 0∗ (s, z, k, b, b ∗ ) and ∆(k,b,b ∗ ) , and (ii) q(s, z, k 0 , b 0 , b 0∗ ) and q ∗ (s, z, k 0 , b 0 , b 0∗ ) and ∞ (iii) bounded sequences of incumbents’ measure {Γt }∞ t=1 and entrants’ measure {Ωt }t=0 such that:

1. given the bond price (q(s, z, k 0 , b 0 , b 0∗ ) and q ∗ (s, z, k 0 , b 0 , b 0∗ )), the value function V (s, z, k, b, b ∗ ), capital holdings k 0 (s, z, k, b, b ∗ ), debt choices b 0 (s, z, k, b, b ∗ ) and b 0∗ (s, z, k, b, b ∗ ), and default set ∆(k,b,b ∗ ) satisfy the firm’s optimization problem; 2. the bond price (q(s, z, k 0 , b 0 , b 0∗ ) and q ∗ (s, z, k 0 , b 0 , b 0∗ )) satisfy the zero expected profit condition, 3. for all Borel sets Z × K ⊂ <+ × B × B ∗ × <+ and ∀t ≥ 0

∗

Z Z

0

Ωt+1 (Z × K × B × B ) = M

dΥ(ν)dH(z /ν), Z

Be (K ,B,B ∗ ,s)

where Be (K , B, B ∗ , s) = {ν s.t.k 0 (s, ν) ∈ K , b 0 (s, ν) ∈ B, b 0∗ (s, ν) ∈ B ∗ andVe (s, ν) ≥ ce } 4. for all Borel sets Z × K ⊂ <+ × B ⊂ <+ × B ∗ ⊂ <+ × <+ and∀t ≥ 0

∗

Z Z

Γt+1 (Z × K × B × B ) =

∗

0

∗

dΓtt (k, b, b , z)dH(z /z) + Ωt+1 (Z × K × B × B ), Z

B(K ,B,B ∗ ,s)

where B(K , B, B ∗ , s) = {(k, b, b ∗ , s) s.t. V (s, z, k, b, b ∗ ) > 0, k ∈ K , b ∈ B and b ∗ ∈ B ∗ }

Mechanism: Bonds’ Prices

Trade-off between deviations from risk-free UIP and idiosyncratic cost of funds:

 q=

1 − Pz,s (∆(k,b,b ∗ ) )

where

1+r ∆k,b,b ∗ =







∗

q =



1 − Pz,s (∆k,b,b ∗ ) 1 + r∗



(s, z) s.t. V R (s, z, k, b, b ∗ ) ≤ 0 .

• Given a (k, b) choice:

− for b ∗ > 0, if Pz,s (∆(k,b,b∗ ) ) ∼ 0, then q < q ∗ . − FC borrowing exposes firms to s shocks and rises Pz,s → lowers q and q ∗ . − This defines the optimal level of FC debt for each firm.

Simulation Strategy

To simulate the years following the deregulation of FC loans in Hungary: 1. Solve the model without FC borrowing and find a stationary distribution. 2. Solve the model with foreign currency borrowing. 3. We simulate 160.000 firms from distribution in (2) using:

− policies of the model with foreign currency and − realized ER shock between 2001-2010.

Calibration Parameter Values Value

Target

Foreign currency risk-free rate

r ∗ = 1.76%

German Bund, 1 year rate

Domestic currency risk-free rate

r = 7.35%

Hungarian Government Bond, 1 year rate

Parameters selected independently

Exchange rate shock

ρs = 0.86

Euro-HUF Forint rate

σs = 0.3 Firm productivity

ρz = 0.63

Hungarian firms

σz = 0.57 Return to scale

α = 0.6

Depreciation rate

δ = 10%

Exchange rate pass-through

η=0

Hungarian firms

Jointly calibrated parameters Fixed cost of credit

c = 0.7

Share of firms borrowing (30%)

Fixed cost of FC debt

c ∗ = 0.12

FC share of borrowing firms (19%)

Fixed operational costs

cf = 2

Default rate (2%)

Investment adjustment cost

c0 = 0.2

Investment of firms borrowing (12%)

Discount factor

β = 0.85

Share of firms with only LC debt (21%)

Shocks

Non-Targeted Moments Moment

Group

Model

Data

(1)

(2)

6

6

FC debt only

2

3

LC debt only

0.97

0.99

LC & FC debt

1.02

1.02

FC debt only

1.07

1.05

LC debt only

1

0.97

LC & FC debt

1.02

1.06

FC debt only

0.91

0.99

LC debt only

9

9

LC & FC debt

18

18

FC debt only

22

19

LC & FC debt

59

50

FC debt only

100

100

LC & FC debt Firm share (%)

Relative productivity*

Relative capital*

Investment rate (%)

FC Share (%)

Notes: 2005-2006. We simulate approximately 160,000 firms from the stationary distribution of no foreign currency. In this simulation, we use the realized exchange rate shocks between 2001 to 2010 and the optimal policies of the model with foreign currency borrowing to obtain the moments for 2001-2010.*Relative productivity and capital are considered with respect to firms with credit.

Mechanism: Bonds’ Prices



1 − Pz,s (∆(k,b,b ∗ ) )

q=

1+r



 q

∗

1 − Pz,s (∆k,b,b ∗ )

=

1 + r∗



Mechanism: Bonds’ Prices for Low and High Productive Firms



1 − Pz,s (∆(k,b,b ∗ ) )

q=

1+r



 q

∗

1 − Pz,s (∆k,b,b ∗ )

=

1 + r∗



Model’s Implications I

-Lemma 1. Only high productive firms borrow in foreign currency. These firms have higher investment rates and grow faster.

Model’s Implications II -Lemma 2. Higher deviations from the risk-free UIP increase foreign currency borrowing and decrease the productivity level to use this financing. Importantly, these deviations raise investment and sales for firms issuing foreign bonds.

FC Debt Policy 1 Theta=1.05 Theta=1.07

0.9 0.8

FC Share

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1

1.5

2

Firm Productivity (z)

2.5

3

Outline

1. Data 2. Model 3. Firm-Level Analysis: Model vs Data 4. Aggregate Implications: Numerical Exercises 5. Extensions − Model: Sensitivity Analysis − Empirics: Currency Depreciation during Great Recession Firm-level results

Lemma 1: Access to FC Loans and Firms’ Growth

1) Test if firms borrowing in FC are more productive and invest more

• Use model simulated data and Hungarian data. • Exploit deregulation of FC loans for domestic firms in 2001. • Test Lemma 1 in two steps:

− probability of borrowing in FC and share of FC loans, and − investment rate and sales.

Lemma 1: FC Dummy and Share of FC Loans

• Using the simulated and Hungarian data, we estimate:

yi = β log zi + εi

− yi : FC Dummyi , log FC Sharei in 2005. − Productivity: zi and RTFPi in productivity pre-reform (2000). − For Hungarian data: restrict sample of firms that gain access to FC loans in 2001: domestic firms. We add µj : four-digit sector FE, and cluster standard errors at four-digit sector.

Lemma 1: FC Dummy and Share of FC Loans con’t

Foreign Currency Loan Dummy Model

Log productivity

Log Share of Foreign Currency Loans

Data

Model

Data

(1)

(2)

(3)

(4)

(5)

(6)

(7)

0.027***

0.024***

0.024***

0.013***

0.012***

0.011***

0.006***

0.002*

(0.001)

(0.001)

(0.002)

(0.002)

(0.000)

(0.000)

(0.001)

(0.001)

Log capital

0.007***

0.036***

0.002***

(0.001)

(0.002)

(0.000)

Sector FE

Yes

Yes

(8)

0.013*** (0.001) Yes

Yes

R2

0.008

0.009

0.029

0.060

0.006

0.006

0.026

0.042

N

156,806

156,806

37,051

37,051

156,806

156,806

37,051

37,051

Notes: *, **, *** significant at the 10, 5, and 1 percent level. Standard errors in parentheses. Source: APEH and Credit Register.

Lemma 1: FC Dummy and Share of FC Loans con’t

Foreign Currency Loan Dummy Model

Log productivity

Log Share of Foreign Currency Loans

Data

Model

Data

(1)

(2)

(3)

(4)

(5)

(6)

(7)

0.027***

0.024***

0.024***

0.013***

0.012***

0.011***

0.006***

0.002*

(0.001)

(0.001)

(0.002)

(0.002)

(0.000)

(0.000)

(0.001)

(0.001)

Log capital

(8)

0.007***

0.036***

0.002***

0.013***

(0.001)

(0.002)

(0.000)

(0.001)

Sector FE

Yes

Yes

Yes

Yes

R2

0.008

0.009

0.029

0.060

0.006

0.006

0.026

0.042

N

156,806

156,806

37,051

37,051

156,806

156,806

37,051

37,051

Notes: *, **, *** significant at the 10, 5, and 1 percent level. Standard errors in parentheses. Source: APEH and Credit Register.

More

Lemma 1: Investment Rate and Sales log yit = β (FLt x FC Dummyi ) + FLt + φi + εit

Log Investment Rate Model

FL* FC dummy

FL

Firm FE

Log Sales Data

Model

Data

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

0.319***

0.139**

0.214***

0.046*

0.639***

0.055***

0.280***

0.038*** (0.014)

(0.032)

(0.061)

(0.018)

(0.025)

(0.005)

(0.010)

(0.013)

0.002

0.117***

0.180***

0.207***

-0.190***

0.002

0.222***

0.013**

(0.007)

(0.013)

(0.031)

(0.026)

(0.001)

(0.002)

(0.010)

(0.006)

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Time trend

Yes

Yes

Yes

Yes

FC d.* Time t.

Yes

Yes

Yes

Yes

R2

0.218

0.218

0.651

0.489

0.571

0.570

0.834

0.836

N

1,568,060

1,568,060

432,864

432,864

1,568,060

1,568,060

436,062

436,062

Notes: *, **, *** significant at the 10, 5, and 1 percent level. Standard errors in parentheses. FL is a dummy for the period 2001-2005. Period 1996-2005. Source: APEH and Credit Register.

Lemma 1: Investment Rate and Sales log yit = β (FLt x FC Dummyi ) + FLt + φi + εit

Log Investment Rate Model

FL* FC dummy

FL

Firm FE

Log Sales Data

Model

Data

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

0.319***

0.139**

0.214***

0.046*

0.639***

0.055***

0.280***

0.038*** (0.014)

(0.032)

(0.061)

(0.018)

(0.025)

(0.005)

(0.010)

(0.013)

0.002

0.117***

0.180***

0.207***

-0.190***

0.002

0.222***

0.013**

(0.007)

(0.013)

(0.031)

(0.026)

(0.001)

(0.002)

(0.010)

(0.006)

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Time trend

Yes

Yes

Yes

Yes

FC d.* Time t.

Yes

Yes

Yes

Yes

R2

0.218

0.218

0.651

0.489

0.571

0.570

0.834

0.836

N

1,568,060

1,568,060

432,864

432,864

1,568,060

1,568,060

436,062

436,062

Notes: *, **, *** significant at the 10, 5, and 1 percent level. Standard errors in parentheses. FL is a dummy for the period 2001-2005. Period 1996-2005. Source: APEH and Credit Register.

Lemma 2: Deviations from Risk-Free UIP: FC Borrowing and Growth

2) Test if deviations from UIP correlate with increases in FC borrowing and growth • Use data for all firms between 2005-2010. • Exploit firm-level variations in terms of productivity and capital stock. • Test Lemma 2 in two steps:

− probability of borrowing in FC and share of FC loans, and − investment rate and sales.

Lemma 2: Deviations from Risk-Free UIP: FC Borrowing and Growth • Using the simulated and the Hungarian data, we estimate

yit = β log UIPt + φi + εit

(1)

yit = β log(UIPt x zi ) + ιt + φi + εit

(2)

yit = β1 log(UIPt x QHLi ) + β2 log(UIPt x QHHi ) + β3 log(UIPt x QLLi ) +β4 log(UIPt x QLHi ) + ιt + φi + εit

(3)

− yit : FC Dummy, log FC Share, log investment rate, log sales. − log UIPt : deviation from risk-free UIP. − QHLi = 1, if z H and k L ; QHHi = 1, if z H and k H ; QLHi = 1, if z L and k H ; and QLLi = 1, if z L and k L (2005).

Lemma 2: FC Dummy FC Dummy Model (1) Log Dev. UIP

(2)

Data (3)

0.071**

0.035***

(0.014)

(0.008)

Log (Dev. UIP x QHH )

Log (Dev. UIP x QLL )

Log (Dev. UIP x QLH )

Yes

Year FE

(6)

(0.018) 0.055***

Log (Dev. UIP x QHL )

Firm FE

(5)

0.139***

(0.028) Log (Dev. UIP x Productivity)

(4)

0.246***

0.189***

(0.029)

(0.032)

0.230***

0.080**

(0.025)

(0.041)

0.180***

0.050*

(0.025)

(0.030)

0.177***

0.089**

(0.016)

(0.041)

Yes

Yes

Yes

Yes

Yes

Sector* Year FE

Yes

Yes

Yes

Yes

Yes

Yes

R2

0.419

0.501

0.51

0.741

0.696

0.742

N

940,836

940,836

940,836

1,019,461

1,019,461

1,019,461

Notes: *, **, *** significant at the 10, 5, and 1 percent level. Standard errors in parentheses. Period 2005-2010. Source: APEH and Credit Register.

More

Lemma 2: Share of FC Loans Log Share of Foreign Currency Loans Model (1) Log Dev. UIP

(2)

Data (3)

0.063***

0.022***

(0.008)

(0.004)

Log (Dev. UIP x QHH )

Log (Dev. UIP x QLL )

Log (Dev. UIP x QLH )

Yes

Year FE

(6)

(0.010) 0.022***

Log (Dev. UIP x QHL )

Firm FE

(5)

0.076***

(0.015) Log (Dev. UIP x Productivity)

(4)

0.177***

0.085***

(0.018)

(0.016)

0.148***

0.059***

(0.015)

(0.021)

0.170***

-0.000

(0.015)

(0.017)

0.117***

0.053**

(0.010)

(0.023)

Yes

Yes

Yes

Yes

Yes

Sector* Year FE

Yes

Yes

Yes

Yes

Yes

Yes

R2

0.402

0.515

0.508

0.716

0.664

0.717

N

940,836

940,836

940,836

1,019,461

1,019,461

1,019,461

Notes: *, **, *** significant at the 10, 5, and 1 percent level. Standard errors in parentheses. Period 2005-2010. Source: APEH and Credit Register.

Lemma 2: Investment Rate Log Investment Rate Model (1) Log Dev. UIP

(2)

Data (3)

(4)

0.099***

0.079*

(0.027)

(0.042)

Log (Dev. UIP x Productivity)

(5)

0.190***

0.328***

(0.031)

(0.017)

Log (Dev. UIP x QHL )

Log (Dev. UIP x QHH )

Log (Dev. UIP x QLL )

Log (Dev. UIP x QLH )

4.708***

0.235***

(0.026)

(0.071)

1.032***

0.188**

(0.027)

(0.082)

0.079***

0.129**

(0.025)

(0.060)

-5.598***

-0.019

(0.027) Firm FE

Yes

Year FE

(6)

Yes

Yes

Yes

Yes

(0.071) Yes

Sector* Year FE

Yes

Yes

Yes

Yes

Yes

Yes

R2

0.42

0.412

0.706

0.033

0.042

0.625

N

940,836

940,836

940,836

513,116

513,116

513,116

Notes: *, **, *** significant at the 10, 5, and 1 percent level. Standard errors in parentheses. Period 2005-2010. Source: APEH and Credit Register.

Lemma 2: Sales Log Sales Model (1) Log Dev. UIP

(2)

Data (3)

0.152**

(5)

(0.031) 0.210**

0.145***

(0.087) Log (Dev. UIP x QHL )

Log (Dev. UIP x QHH )

Log (Dev. UIP x QLL )

Log (Dev. UIP x QLH )

(0.041) 9.963***

0.250*

(0.079)

(0.146)

6.218***

0.248*

(0.077)

(0.145)

-4.649***

0.193

(0.077)

(0.176)

-6.584***

0.297

(0.078) Firm FE

Yes

Year FE

(6)

0.059*

(0.072) Log (Dev. UIP x Productivity)

(4)

Yes

Yes

Yes

Yes

(0.191) Yes

Sector* Year FE

Yes

Yes

Yes

Yes

Yes

Yes

R2

0.722

0.722

0.873

0.877

0.852

0.905

N

940,836

940,836

940,836

765,611

765,611

765,611

Notes: *, **, *** significant at the 10, 5, and 1 percent level. Standard errors in parentheses. Period 2005-2010. Source: APEH and Credit Register.

Lemma 2: Productivity Level to Borrow in Foreign Currency Test if deviations from UIP lower the productivity level to employ FC loans • For firms borrowing in foreign currency, we test :

log yit = β log UIPt + φi + εit

(if FC Dummyit = 1)

Log Productivity

Log Dev. UIP

Model

Data

(1)

(2)

-5.435***

-0.655***

(0.482)

(0.212)

Firm FE

Yes

Yes

R2

0.883

0.821

N

119,663

64,556

Notes: *, **, *** significant at 10, 5, and 1 percent. Standard errors in parentheses. Period 2005-2010. Source: APEH and Credit register.

Mechanism: Cost of Funds Test whether firms borrowing in FC pay lower interest rate • Use firm-level data on interest rates for Hungary (2005) from BEEPS • We estimate yit = β FC Dummyi + εi

LC Bond Price

FC Bond Price

Interest Rate

Model

FC Dummy

Data

(1)

(2)

(3)

(4)

0.075***

0.592***

-1.068**

-0.896*

-0.921*

(0.002)

(0.005)

(0.531)

(0.427)

(0.555)

Firm Level Controls

(5)

Yes

Sector FE

Yes

Yes

R2

0.005

0.070

0.014

0.033

0.042

N

156,806

156,806

291

291

291

Notes: *, **, *** significant at 10, 5, and 1 percent. Standard errors in parentheses. The survey asks firms the interest rate charged for the most recent loan obtained from a local financial institution and whether this loan was in local or foreign currency. Firm-level controls are age, employment, export status and dummy for foreign-owned firm. Source: BEEPS 2005, Hungary, the World Bank and the EBRD.

Summary of Firm-Level Analysis 1. Access to FC loans and growth (reform)

− Initially more productive firms have a higher probability of borrowing in FC and share of FC loans after the reform.

− Firms borrowing in FC associate with higher investment rates and sales. 2. Deviations from risk-free UIP

− Correlate with prob. of FC borrowing, FC share, investment and sales. − Larger expansion for productive firms; among them, firms with low capital. − Additional evidence on productive and young firms. − Lower the productivity level to borrow in FC. 3. Mechanism

− Firms borrowing in FC pay lower interest rates. Return

Outline

1. Data 2. Model 3. Firm-Level Analysis: Model vs Data 4. Aggregate Implications: Numerical Exercises 5. Extensions − Model: Sensitivity Analysis − Empirics: Currency Depreciation during Great Recession

Numerical Exercises

1. FC vs non FC borrowing

2. Selection into FC Borrowing

3. Exchange Rate Volatility

Numerical Exercises: FC vs non FC borrowing

Benchmark

No FC Borrowing

(1)

(2)

FC debt share

12.1

-

Investment rate

10.6

8.3

E(K)

20.0

17.8

Default rate

2.7

3.3

Productivity threshold

1.2

-

Sales (level)

100.0

87.8

Panel A. Firm-level results

Panel B. Aggregate results

Capital (level)

100.0

82.8

Std. dev. sales

100.0

0.0

Std. dev. capital

100.0

0.0

Notes: Rows 1, 3, and 4 are in percentage, rows 6-9 are with respect to column 1. Period 2001-2010.

Numerical Exercises: Selection into FC Borrowing

Benchmark

No FC

No Selection

Borrowing (1)

(2)

(3) Panel A. Firm-level results

FC debt share

12.1

-

12.1

Investment rate

10.6

8.3

7.6

E(K)

20.0

17.8

23.1

Default rate

2.7

3.3

7.7

Productivity threshold

1.2

-

0.0

Sales (level)

100.0

87.8

79.6

Capital (level)

100.0

82.8

78.6

Std. dev. sales

100.0

0.0

84.2

Std. dev. capital

100.0

0.0

39.7

Panel B. Aggregate results

Notes: Rows 1, 3, and 4 are in percentage, rows 6-9 are with respect to column 1. Period 2001-2010.

Conclusion

Numerical Exercises: Exchange Rate Volatility

Benchmark

No FC

No Selection

Borrowing

(1)

(2)

(3)

ER Volatility Low

High

(σs = 0.15)

(σs = 0.45)

(4)

(5)

Panel A. Firm-level results FC debt share

12.1

-

12.1

42.9

0.0

Investment rate

10.6

8.3

7.6

11.9

9.5

E(K)

20.0

17.8

23.1

21.6

18.8

Default rate

2.7

3.3

7.7

2.5

3.1

Productivity threshold

1.2

-

0.0

1.0

1.3

Sales (level)

100.0

87.8

79.6

111.9

98.9

Capital (level)

100.0

82.8

78.6

115.3

95.2

Std. dev. sales

100.0

0.0

84.2

68.1

22.6

Std. dev. capital

100.0

0.0

39.7

93.3

7.5

Panel B. Aggregate results

Notes: Rows 1, 3, and 4 are in percentage, rows 6-9 are with respect to column 1. Period 2001-2010.

Outline

1. Data 2. Model 3. Firm-Level Analysis: Model vs Data 4. Aggregate Implications: Numerical Exercises 5. Extensions − Model: Sensitivity Analysis − Empirics: Currency Depreciation during Great Recession

Model: Sensitivity Analysis

1. Exchange rate pass-through:

− high pass-through lowers the risk of FC debt. − higher FC share and investment, and lower default and selection. 2. Endogenous interest rates:

− a stochastic discount factor, adding a premium during depreciations. − lower FC share and investment, and higher default and selection. 3. Aggregate shock:

− productivity shock increases uncertainty. − lower FC share and investment, and higher default and selection. More

Empirical Exercise: Depreciation during the Great Recession

• Use the depreciation after 2008 to study the performance of FC borrowing firms. • After a full set of controls, we find that:

− firms in foreign currency see negative balance sheet effects, as they differentially decrease their FC share (20%), leverage (18%), investment (50%), but...

− they outperform their industry counterparts in terms of sales and do not exit more. • These results are robust for non-exporters and exporters. More

Conclusions

→ Cross-sectional heterogeneity in FC borrowing in two dimensions:

− Selection into FC borrowing, and heterogeneous FC share across firms. − Conditional on productivity, low capital firms benefit more. → Economies with FC borrowing see higher income, but higher volatility. → Selection plays a key role to generate gains from FC borrowing.

Extra Slides

Stylized Facts I

g

m

ur

do

Ki

ng

Ita ly

bo

m xe

d

Lu

Un

ite

st ria nm G ark er m an y G re ec e

Au

De

ia

ia

rb

an

Se

m

y

nd la

Po

Ro

ic

ar

ng

Hu

bl

oa

pu

lg

Re

h ec Cz

Bu

Cr

ar

ia

tia

0

20

40

60

80

- Fact 1: Firms hold high shares of foreign currency loans in developing countries.

Developing

Developed Year 2014. In % Source: CHF Lending Monitor.

Stylized Facts II - Fact 2: Deviations from the risk-free UIP make FC loans attractive. Poland

.95

1

1

Deviations from UIP 1.05 1.1

Deviations from UIP 1.05 1.1 1.15

1.2

1.15

Hungary

2005q4

2008q4

2011q4

2014q4

2005q4

2008q4

2005q4

2014q4

Deviations from UIP .95 .975 1 1.025 1.05 1.075 1.1

Croatia

Deviations from UIP .95 .975 1 1.025 1.05 1.075 1.1

Romania

2011q4

2008q4

→ UIP doesn’t hold:

2011q4

E (st+1 ) st

2014q4

2005q4

2008q4

2011q4

(1 + rt∗ ) 6= (1 + rt ) → Devt =

2014q4

(1+rt ) st >1 E (st+1 ) (1+rt∗ )

Dev

Stylized Facts III - Fact 3: FC loans and deviations from the risk-free UIP are highly correlated. FC/Total Loans (left)

Deviations UIP (right)

1.05 1.1 1.15 Deviations from UIP

.34 FC/Total Loans .28 .3 .32

2005q4

2008q4

2011q4

2014q4

2005q4

.95

.45

.24

1

1

.26

1.15 1.05 1.1 Deviations from UIP

FC/Total Loans .55 .6 .5

1.2

Poland

.65

Hungary

2008q4

2008q4

→ UIP doesn’t hold:

2011q4

E (st+1 ) st

2014q4

.95 .975 1 1.025 1.05 1.075 1.1 Deviations from UIP

.76 FC/Total Loans .7 .72 .74 .68 .66

.65 .6 .55

FC/Total Loans

.5 .45

2005q4

2014q4

Croatia .95 .975 1 1.025 1.05 1.075 1.1 Deviations from UIP

Romania

2011q4

2005q4

2008q4

2011q4

(1 + rt∗ ) 6= (1 + rt ) → Devt =

2014q4

(1+rt ) st >1 E (st+1 ) (1+rt∗ )

Stylized Facts IV - Fact 4: Cross-sectional heterogeneity in the share of foreign currency loans. 1. Not all firms borrow in foreign currency. 2. The share of FC loans is heterogeneous across firms.

− Hungary: 33% of firms borrowed in FC.

0

10

20

In % 30

40

50

Hungary: Distribution of Share of Foreign Currency Loans (2005)

0

Return

20 40 60 80 Share of Foreign Currency Loans on Total Loans

100

→ IR Difft = 1

FC/Total Loans .55 .6

2005q4 2008q4

(1+rt ) >1 (1+rt∗ ) 2011q4 2014q4

Return

2005q4

2005q4

1.08

2014q4

1.02 1.04 1.06 Int. Rate Differential

2011q4

.76

2008q4

FC/Total Loans .7 .72 .74

1.08

.65

2005q4

.68

1.02 1.04 1.06 Int. Rate Differential

.5

1

.24

1

.45

FC/Total Loans .55 .6

FC/Total Loans .28 .3 .32

1.02 1.04 Int. Rate Differential

.26

1.03 1.06 1.09 Int. Rate Differential

.5

1.06

.34

1.12

.65

FC/Total Loans (Left)

.66

1

.45

FC Loans and Interest Rate Differential in Developing Countries IR Differential (Right) Hungary Poland

2008q4

Romania

2008q4

2011q4

2011q4

2014q4

Croatia

2014q4

→ IR Difft = 2005q4 2008q4

(1+rt ) >1 (1+rt∗ ) 2011q4 2014q4 .08

2005q4

1.06

.09

2005q4

.98 1 1.02 1.04 Int. Rate Differential

.07

2014q4

.06

2011q4

.05

2008q4

FC/Total Loans

1.06

.02

2005q4

.96

.018

.98 1 1.02 1.04 Int. Rate Differential

.016

FC/Total Loans .014

.96

.04

.96

.08

.06

.07

.98 1 1.02 1.04 Int. Rate Differential

.05

FC/Total Loans

.98 1 1.02 1.04 Int. Rate Differential

FC/Total Loans .1 .12 .14

1.06

.08

1.06

.16

FC/Total Loans (Left)

.04

.96

.012

FC Loans and Interest Rate Differential in Developed Countries IR Differential (Right) UK Germany

2008q4

Italy

2008q4

2011q4

2011q4

2014q4

Austria

2014q4

Deviations from Risk-Free UIP in Developed Countries

2005q4

Deviations from UIP .95 1 1.05 1.1 .9

.9

Deviations from UIP .95 1 1.05 1.1

1.15

Germany

1.15

UK

2008q4

2011q4

2014q4

2005q4

2008q4

→ Devt =

1.15 Deviations from UIP .95 1 1.05 1.1 .9

Deviations from UIP .95 1 1.05 1.1 .9

2005q4

2014q4

Austria

1.15

Italy

2011q4

2008q4

2011q4

(1+rt ) st >1 E (st+1 ) (1+rt∗ )

2014q4

2005q4

2008q4

2011q4

2014q4

2005q4

→ Devt = .04

FC/Total Loans .014 .016 .018 .02

2008q4 2011q4

(1+rt ) st >1 E (st+1 ) (1+rt∗ ) 2014q4

2005q4

2005q4

.95 .975 1 1.0251.051.075 1.1 Deviations from UIP

2014q4

.09

2011q4

FC/Total Loans .06 .07 .08

2008q4

.05

2005q4

.95 .975 1 1.0251.051.075 1.1 Deviations from UIP

.012

.04

.08

.16

.08

.95 .975 1 1.0251.051.075 1.1 Deviations from UIP

FC/Total Loans .05 .06 .07

.95 .975 1 1.0251.051.0751.1 Deviations from UIP

FC/Total Loans .1 .12 .14

Deviations from Risk-Free UIP in Developed Countries and FC Loans

UK Germany

2008q4

Italy

2008q4

2011q4

2011q4

2014q4

Austria

2014q4

Foreign Currency Loans and Deviations from the UIP (3M & 2Y)

2011q4

2013q4

FC/Total Loans .5 .55 .6 .65

2015q4

2005q4

2007q4

2009q4

2011q4

2013q4

2009q4

2011q4

2013q4

2013q4

2015q4

2005q4

2007q4

2009q4

2011q4

2013q4

2015q4

FC Loans and Deviations from the UIP (US Dollars) (2 Years)

2015q4

FC/Total Loans .5 .55 .6 .65 .45

1 1.05 1.1 Deviations from UIP

FC/Total Loans .5 .55 .6 .65 .45

2007q4

2011q4

FC/Total Loans .5 .55 .6 .65

2015q4

FC Loans and Deviations from the UIP (US Dollars) (3 Months)

2005q4

2009q4

.45

.95 1 1.05 1.1 Deviations from UIP

FC/Total Loans .5 .55 .6 .65 .45

2005q4

2007q4

FC Loans and Deviations from the UIP (Swiss Franc) (2 Years) 1 1.1 1.2 Deviations from UIP

2009q4

FC Loans and Deviations from the UIP (Swiss Franc) (3 Months)

1 1.1 1.2 Deviations from UIP

2007q4

.45

1 1.05 1.1 Deviations from UIP

.45

2005q4

1 1.05 1.1 1.15 Deviations from UIP

FC Loans and Deviations from the UIP (Euro) (2 Years)

FC/Total Loans .5 .55 .6 .65

FC Loans and Deviations from the UIP (Euro) (3 Months)

2005q4

2007q4

2009q4

2011q4

FC/Total Loans (left)

2013q4

2015q4

Deviations UIP (right)

Calibration: Shocks

− Firms’ productivity process:

log zijt = ρz log zijtt−1 + φi + µjt + εijt ,

where φi and µjt are firm and four-digit NACE industries-year FE.

− Exchange rate process: log st = ρs log st−1 + εt . Return

Robustness: Decision into Foreign Currency Borrowing

Foreign Currency Loan Dummy (1) Log productivity

Log capital

Log age

(2)

(3)

(4)

(5)

0.014***

0.010***

0.009***

0.012***

0.017***

(0.002)

(0.003)

(0.002)

(0.002)

(0.002)

0.037***

0.036***

0.036***

0.035***

0.034***

(0.002)

(0.002)

(0.002)

(0.002)

(0.002)

-0.019***

-0.015***

-0.018***

-0.020***

-0.020***

(0.003)

(0.003)

(0.003)

(0.003)

(0.003)

Exporter

0.048*** (0.007)

Sector FE

Yes

Yes

Yes

Yes

Yes

R2

0.061

0.060

0.063

0.063

0.055

N

37,051

37,051

38,237

37,051

39,740

Note: *, **, *** significant at the 10, 5, and 1 percent level. Standard errors in parentheses. Column 1 includes age as a regressor. Column 2 uses RTFP measured with the Olley and Pakes (1996) methodology. Column 3 employs labor productivity as a proxy for firms’ RTFP. Column 4 includes a dummy for exporter. Column 5 employs the average for 1998-2000 as initial conditions. Source: APEH and Credit Register.

Robustness: Decision into Foreign Currency Borrowing Log Share of FC Loans (1) Log productivity

Log capital

Log age

(2)

(3)

(4)

(5)

0.003**

0.006***

0.003***

0.004***

0.004***

(0.001)

(0.002)

(0.001)

(0.001)

(0.001)

0.014***

0.015***

0.013***

0.012***

0.014***

(0.001)

(0.001)

(0.001)

(0.001)

(0.001)

-0.010***

-0.011***

-0.010***

-0.011***

-0.010***

(0.002)

(0.001)

(0.002)

(0.002)

(0.002)

Exporter

0.025*** (0.004)

Sector FE

Yes

Yes

Yes

Yes

Yes

R2

0.043

0.039

0.048

0.050

0.044

N

37,051

37,051

38,237

37,051

39,740

Note: *, **, *** significant at the 10, 5, and 1 percent level. Standard errors in parentheses. Column 1 includes age as a regressor. Column 2 uses RTFP measured with the Olley and Pakes’ (1996) methodology. Column 3 employs labor productivity as a proxy for firms’ RTFP. Column 4 includes a dummy for exporter. Column 5 employs the average for 1998-2000 as initial conditions. Source: APEH and Credit Register.

Return

Robustness: Deviations from Risk-Free UIP

FC Dummy

Log (Dev. UIP x QHY )

Log (Dev. UIP x QHO )

Log (Dev. UIP x QLY )

Log (Dev. UIP x QLO )

Firm FE

Log FC Share

Log Investment Rate

Log Sales

Model

Data

Model

Data

Model

Data

Model

Data

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

0.435***

0.315***

0.258***

0.096***

4.054***

0.209***

6.186***

0.700***

(0.040)

(0.047)

(0.021)

(0.025)

(0.033)

(0.062)

(0.091)

(0.165)

0.177***

0.032

0.119***

0.054***

3.463***

0.183**

5.979***

0.047

(0.040)

(0.034)

(0.021)

(0.017)

(0.033)

(0.073)

(0.088)

(0.169)

0.137***

0.172***

0.108***

0.056***

-1.649***

0.008

-5.142***

0.567***

(0.040)

(0.039)

(0.021)

(0.021)

(0.032)

(0.065)

(0.082)

(0.163)

-0.231***

-0.024

-0.081***

-0.003

-2.156***

0.161***

-5.370***

-0.125

(0.039)

(0.034)

(0.020)

(0.019)

(0.032)

(0.054)

(0.079)

(0.185)

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Sector*Year FE

Yes

Yes

Yes

Yes

R2

0.42

0.742

0.403

0.717

0.623

0.700

0.748

0.919

N

940,836

1,019,461

940,836

1,019,461

940,836

513,116

940,836

765,611

Notes: *, **, *** significant at the 10, 5, and 1 percent level. Standard errors in parentheses. Period 2005-2010. Source: APEH and Credit Register.

Return

Model’s Extensions: Exchange Rate Pass-Through

Firms’ equity become:

e = s η [zk α −i(k, k 0 )−ψ(k, k 0 )−cf ]−[b+sb ∗ ]+[qb 0 +q ∗ sb 0∗ −s η cI(b0 +b0∗ >0) −s η cI∗ 0∗ (b

Benchmark

η = 0.2

η=1

(1)

(2)

(3)

FC debt share

12.1

27.3

Investment rate

10.6

12.0

12.5

E(K)

20.0

19.5

21.5

Default rate

2.7

2.5

2.3

Productivity threshold

1.2

1.1

1.0

Firm-level results

Notes: Rows 1, 3, and 4 are in percentage. Period 2001-2010.

42.3

>0)

]

Model’s Extensions: Endogenous Interest Rates - Investors’ discount factors:

m0 =

1 (1 + r )

 0 γ s s

m0∗ =

and

1 (1 + r ∗ )

 0 γ ∗ s s

(m and m0 : baseline rates r and r ∗ adjusted by ER with sensitivity γ and γ ∗ ).

- Define risk-free rates as :

1+˜ r (s) =

1 E (m0 |s)

1+˜ r ∗ (s) =

and

- Re-write the UIP condition as:

θ

E (s 0 |s) (1 + r ) = s (1 + r ∗ )



E (s 0γ∗−γ |s) s γ∗−γ

 .

1 . E (m0∗ |s)

Model’s Extensions: Endogenous Interest Rates

Benchmark

γ = γ∗ = 1

γ = 1 and γ ∗ = 0

(1)

(2)

(3) Firm-level results

FC debt share

12.1

11.4

Investment rate

10.6

10.2

11.4

E(K)

20.0

23.8

25.0

Default rate

2.7

3.2

3.0

Productivity threshold

1.2

1.2

1.2

˜ r

7.4

6.8

6.8

˜ r∗

1.8

1.2

1.8

θ

1.05

1.05

1.04

Notes: Rows 1, 3, 4, 6 and 7 are in percentage. Period 2001-2010.

10.5

Model’s Extensions: Aggregate Shock

Let the aggregate shock be a function of the exchange rate shock

Z = s −ν

F (s, z, k) = s −ν zk α

and

Benchmark

ν = 0.05

ν = 0.1

ν = 0.2

(1)

(2)

(3)

(4)

FC debt share

12.1

11.5

9.5

Investment rate

10.6

10.5

10.3

9.8

E(K)

20.0

19.8

19.2

18.2

Default rate

2.7

3.0

3.1

3.4

Productivity threshold

1.2

1.2

1.2

1.3

Firm-level results

Notes: Rows 1, 3, and 4 are in percentage. Period 2001-2010.

Return

0.1

Empirical Exercise: Depreciation during the Great Recession

Exploit the ER depreciation to test how firms borrowing in FC perform: • Empirical Strategy

Yijt = β(Dt x FC Ratioi ) + φi + (Tt x FC Dummyi ) + (µi x Dt ) + Xjt + εijt

− Yijt log FC Share, log leverage, log investment, log sales & exit (2005-10). − Dt : dummy for the depreciation years (2008-10). − FC Ratioi : firm’s foreign currency debt ratio in 2005. − Tt x FC Dummyi : time trend interacted with FC debt dummy. − µi x Dt : firm’s initial productivity and import in initial year − φi : firm fixed-effects. Xjt : sector-year FE.

Empirical Exercise: Depreciation during the Great Recession con’t → FC borrowing firms reduce their share of FC debt, leverage and investment, but...

Log FC Share (1)

(2)

Log Leverage (3)

(4)

(5)

Log Investment (6)

(7)

(8)

(9)

All Firms D*FC Ratio

D

Firm FE

-0.330***

-0.253***

-0.204***

-0.129***

-0.146***

-0.179***

-0.673***

-0.455***

-0.478***

(0.017)

(0.016)

(0.013)

(0.015)

(0.015)

(0.013)

(0.135)

(0.094)

(0.094)

0.009***

0.009***

(0.001)

(0.001)

yes

yes

yes

yes

-0.568*** (0.024) yes

yes

yes

yes

yes

Firm-Time controls

yes

yes

yes

yes

yes

yes

Year FE

yes

yes

yes

yes

yes

yes

Sector*Year FE

yes

yes

yes

R2

0.701

0.704

0.705

0.572

0.572

0.478

0.800

0.826

0.827

N

843,545

843,545

843,545

843,545

843,545

843,545

441,685

441,685

441,685

Notes: *, **, *** significant at 10, 5, and 1 percent. Std. errors in parenthesis. Standard errors are cluster at year and four-digit NACE industries. FC ratio is the firm’s foreign currency debt over total assets in the initial year (2005). Firm-time varying controls include the interaction of foreign currency borrowing with a time trend, and the interaction of firms’ initial productivity and import share with the depreciation dummy. Sector-year fixed effects are estimated at two-digit NACE industries.

Empirical Exercise: Depreciation during the Great Recession con’t →... they see their sales less affected and don’t exit more. Log Sales (1)

(2)

Exit (3)

(4)

(5)

(6)

All Firms D*FC Ratio

D

0.079**

0.100***

0.073***

-0.036***

0.018*

-0.001

(0.034)

(0.027)

(0.027)

(0.009)

(0.006)

(0.009)

yes

yes

-0.122***

0.082***

(0.010) Firm FE

yes

(0.002) yes

yes

Firm-Time controls

yes

yes

yes

yes

Year FE

yes

yes

yes

yes

Sector*Year FE

yes

yes

yes

R2

0.936

0.936

0.937

0.586

0.602

0.614

N

655,996

655,996

655,996

725,501

725,501

725,501

Notes: *, **, *** significant at 10, 5, and 1 percent. Std. errors in parenthesis. Standard errors are cluster at year and four-digit NACE industries. FC ratio is the firm’s foreign currency debt over total assets in the initial year (2005). Firm-time varying controls include the interaction of foreign currency borrowing with a time trend, and the interaction of firms’ initial productivity and import share with the depreciation dummy. Sector-year fixed effects are estimated at two-digit NACE industries.

Return

Empirical Exercise: Depreciation during the Great Recession con’t

→ Non Exporters: Log FC Share (1)

(2)

Log Leverage (3)

(4)

(5)

Log Investment (6)

(7)

(8)

(9)

Non-Exporters D*FC Ratio

D

-0.333***

-0.252***

-0.255***

-0.118***

-0.138***

-0.137***

-0.777***

-0.454***

-0.541***

(0.019)

(0.017)

(0.017)

(0.016)

(0.017)

(0.017)

(0.157)

(0.106)

(0.144)

0.008***

0.009***

(0.001) Firm FE

yes

-0.587***

(0.001) yes

(0.025)

yes

yes

yes

yes

yes

yes

Firm-Time controls

yes

yes

yes

yes

yes

yes

yes

Year FE

yes

yes

yes

yes

yes

yes

R2

0.693

0.695

0.696

0.577

0.578

0.578

0.788

0.815

0.791

N

771,756

771,756

771,756

771,756

771,756

771,756

324,963

324,963

324,963

Notes: *, **, *** significant at 10, 5, and 1 percent. Std. errors in parenthesis. Standard errors are cluster at year and four-digit NACE industries. FC ratio is the firm’s foreign currency debt over total assets in the initial year (2005). Firm-time varying controls include the interaction of foreign currency borrowing with a time trend, and the interaction of firms’ initial productivity and import share with the depreciation dummy. Sector-year fixed effects are estimated at two-digit NACE industries.

Empirical Exercise: (3) Depreciation during the Great Recession con’t

→ Non-Exporters: Log Sales (1)

(2)

Exit (3)

(4)

(5)

(6)

Non-Exporters D*FC Ratio

D

0.075***

0.104***

0.070**

-0.046***

0.017

-0.011

(0.028)

(0.029)

(0.029)

(0.010)

(0.012)

(0.013)

-0.115***

0.084***

(0.002) Firm FE

yes

(0.002) yes

yes

yes

yes

yes

Firm-Time controls

yes

yes

yes

yes

Year FE

yes

yes

yes

yes

Sector*Year FE

yes

yes

R2

0.931

0.931

0.932

0.591

0.607

0.619

N

587,140

587,140

587,140

664,290

664,290

664,290

Notes: *, **, *** significant at 10, 5, and 1 percent. Std. errors in parenthesis. Standard errors are cluster at year and four-digit NACE industries. FC ratio is the firm’s foreign currency debt over total assets in the initial year (2005). Firm-time varying controls include the interaction of foreign currency borrowing with a time trend, and the interaction of firms’ initial productivity and import share with the depreciation dummy. Sector-year fixed effects are estimated at two-digit NACE industries.

Depreciation during the Great Recession: Short vs Long Term Loans

Log FC Share (1)

(2)

Log Leverage (3)

(4)

(5)

Log Investment (6)

(7)

(8)

(9)

All Firms D*ST Ratio

-0.343***

-0.231***

-0.250***

-0.071*

-0.081**

0.022

0.064

(0.057)

(0.060)

(0.060)

(0.037)

(0.038)

(0.068)

(0.034)

(0.149)

(0.166)

D* LT Ratio

-0.421***

-0.354***

-0.355***

-0.138***

-0.143***

-0.195***

-0.199***

-0.122***

-0.091**

-0.084***

-0.062

(0.020)

(0.020)

(0.020)

(0.018)

(0.020)

(0.016)

(0.029)

(0.036)

(0.030)

D* ST < Ratio

-0.267***

-0.196***

-0.206***

-0.130***

-0.137***

-0.141***

0.049

0.104

0.107

(0.024)

(0.024)

(0.024)

(0.023)

(0.025)

(0.027)

(0.094)

(0.092)

(0.093)

D

0.009***

0.008***

(0.001) Firm FE

yes

-0.579***

(0.001) yes

yes

yes

yes

yes

yes

yes

yes

yes

yes

yes

Year FE

yes

yes

yes

yes

yes

yes

Sector*Year FE

yes

(0.007)

yes

Firm-Time controls

yes

yes

yes

yes

R2

0.716

0.719

0.719

0.600

0.600

0.510

0.805

0.808

0.829

N

843,545

843,545

843,545

843,545

843,545

843,545

441,685

441,685

441,685

Notes: *, **, *** significant at 10, 5, and 1 percent. Std. errors in parenthesis. Standard errors are cluster at year and four-digit NACE industries. FC ratio is the firm’s foreign currency debt over total assets in the initial year (2005). Firm-time varying controls include the interaction of foreign currency borrowing with a time trend, and the interaction of firms’ initial productivity and import share with the depreciation dummy. Sector-year fixed effects are estimated at two-digit NACE industries.

Depreciation during the Great Recession: Short vs Long Term Loans con’t Log Sales (1)

(2)

Exit (3)

(4)

(5)

(6)

All Firms D*ST Ratio

D* LT Ratio

D* ST < Ratio

D

0.078

0.095

0.003

-0.069

0.019

0.024

(0.132)

(0.071)

(0.071)

(0.045)

(0.027)

(0.045)

0.077***

0.088***

0.044**

-0.048***

-0.001

-0.012

(0.029)

(0.023)

(0.023)

(0.009)

(0.007)

(0.012)

0.077

0.092*

0.046

-0.044*

0.011

0.015

(0.058)

(0.051)

(0.051)

(0.023)

(0.013)

(0.023)

-0.122***

0.086***

(0.010) Firm FE

yes

(0.002) yes

yes

yes

yes

Firm-Time controls

yes

yes

yes

yes

Year FE

yes

yes

yes

yes

Sector*Year FE

yes

yes

yes

R2

0.936

0.936

0.937

0.639

0.602

0.661

N

655,996

655,996

655,996

725,501

725,501

725,501

Notes: *, **, *** significant at 10, 5, and 1 percent. Std. errors in parenthesis. Standard errors are cluster at year and four-digit NACE industries. FC ratio is the firm’s foreign currency debt over total assets in the initial year (2005). Firm-time varying controls include the interaction of foreign currency borrowing with a time trend, and the interaction of firms’ initial productivity and import share with the depreciation dummy. Sector-year fixed effects are estimated at two-digit NACE industries.

Euro vs Swiss Franc Loans

Swiss Franc Loan Dummy (1) RTFP

Employment

Age

Exporters

(2)

-0.033***

-0.008*

(0.008)

(0.004)

-0.041***

-0.044***

(0.009)

(0.004)

0.022*

0.023***

(0.013)

(0.008)

-0.151***

-0.155***

(0.022)

(0.009)

Sector FE

yes

yes

R2

0.171

0.209

N

4,409

11,938

Notes: *, **, *** significant at 10, 5, and 1 percent. Std. errors in parenthesis. This regression only includes firms that have either Swiss Franc or Euro Loans. The dependent variable is a dummy of whether the firm hold a Swiss Franc loan in 2005. Firm-level controls include size, age and export status. All regressions include four-digit NACE industry fixed effects. Column 1 is estimated for the year 2000, and Column 2 estimates the average over 1996 and 2000. Firm-level controls in Column 2 are considered in the initial year (1996).

Euro vs Swiss Franc Loans con’t

Log Sales

Log Investment

(1)

(2)

(3)

(4)

FL*Swiss Dummy

0.073**

0.090***

0.077***

0.053**

(0.037)

(0.033)

(0.021)

(0.026)

FL

0.257***

0.063*

0.451***

0.039*

(0.032)

(0.036)

(0.021)

(0.021)

yes

yes

yes

Firm FE Time Trend

yes

Swiss Dummy*Time Trend

yes yes

yes

yes

R2

0.620

0.622

0.817

0.842

N

49,249

49,249

49,472

49,472

Notes: *, **, *** significant at 10, 5, and 1 percent. Std. errors in parenthesis. This regression only includes firms that have either Swiss Franc or Euro Loans. Swiss dummy is a binary variable of whether the firm hold a Swiss Franc loan in 2005. FL is a dummy for the period 2001-05. Sample period 1996-2005.

Euro vs Swiss Franc Loans con’t Log FC Share (1)

(2)

Log Leverage (3)

(4)

(5)

Log Investment (6)

(7)

(8)

(9)

-0.301***

All Firms D*Euro Ratio

-0.281***

-0.218***

-0.178***

-0.160***

-0.176***

-0.200***

-0.278**

-0.282**

(0.016)

(0.017)

(0.015)

(0.022)

(0.023)

(0.021)

(0.140)

(0.119)

(0.100)

D*SF Ratio

-0.362***

-0.288***

-0.232***

-0.103***

-0.120***

-0.168***

-0.938***

-0.504***

-0.503***

(0.025)

(0.022)

(0.018)

(0.019)

(0.020)

(0.015)

(0.113)

(0.095)

(0.090)

D

0.010***

yes

yes

0.009***

(0.001) Firm FE

yes

-0.568***

(0.001) yes

yes

Firm-Time controls

yes

Year FE

(0.007) yes

yes

yes

yes

yes

yes

yes

yes

yes

yes

yes

yes

yes

Sector*Year FE

yes

yes

R2

0.698

0.701

0.702

N

840,196

840,196

840,196

yes

yes

yes

0.574

0.575

0.480

0.799

0.802

0.824

840,196

840,196

840,196

439,937

439,937

439,937

Panel B: F-Test on Equality of Coefficients F-stat

128.76

88.88

16.99

P-value

0.0000

0.0000

0.0000

Notes: *, **, *** significant at 10, 5, and 1 percent. Std. errors in parenthesis. Standard errors are cluster at year and four-digit NACE industries. FC ratio Euro and SF are the firm’s foreign currency debt in Euros or Swiss Francs over total assets in the initial year (2005). Firm-time varying controls include the interaction of foreign currency borrowing with a time trend, and the interaction of firms’ initial productivity and import share with the depreciation dummy. Sector-year fixed effects are estimated at two-digit NACE industries. These regressions exclude firms that had U.S. denominated loans and both Euro and Swiss Franc loans in 2005.

Euro vs Swiss Franc Loans con’t Log Sales (1)

(2)

Exit (3)

(4)

(5)

(6)

Panel A : All Firms D*Euro Ratio

D*SF Ratio

D

Firm FE

0.098*

0.071***

0.057**

-0.029***

0.022**

0.008

(0.051)

(0.013)

(0.023)

(0.011)

(0.010)

(0.015)

0.113***

0.102***

0.080***

-0.030***

0.014*

0.006

(0.036)

(0.012)

(0.015)

(0.009)

(0.008)

(0.014)

yes

yes

-0.122***

0.082***

(0.010)

(0.002) yes

yes

Firm-Time controls

yes

yes

yes

yes

yes

Year FE

yes

yes

yes

yes

Sector*Year FE

yes

yes

yes

R2

0.935

0.936

0.936

0.587

0.603

0.614

N

652,658

652,658

652,658

722,152

722,152

722,152

Panel B: F-Test on Equality of Coefficients F-stat

14.92

0.18

P-value

0.0000

0.8362

Notes: *, **, *** significant at 10, 5, and 1 percent. Std. errors in parenthesis. Standard errors are cluster at year and four-digit NACE industries. FC ratio Euro and SF are the firm’s foreign currency debt in Euros or Swiss Francs over total assets in the initial year (2005). Firm-time varying controls include the interaction of foreign currency borrowing with a time trend, and the interaction of firms’ initial productivity and import share with the depreciation dummy. Sector-year fixed effects are estimated at two-digit NACE industries. These regressions exclude firms that had U.S. denominated loans and both Euro and Swiss Franc loans in 2005.