Introduction The model Calibration Steady State Transition dynamics Conclusion

Optimal Capital Account Liberalization in China1

Zheng Liu1 Mark M. Spiegel1 Jingyi Zhang2

1Federal Reserve Bank of San Francisco

2Shanghai University of Finance and Economics

Reforms and Liberalization ofChina’s Capital Market Conference

September 21-22, 2018

1The views expressed herein are those of the authors and do not necessarilyreflect the views of the Federal Reserve Bank of San Francisco or the FederalReserve System.

Introduction The model Calibration Steady State Transition dynamics Conclusion

Domestic financial repression

• Government-favored firms (SOEs) can obtain directed lendingat below-market rates (Brandt and Zhu, 2000; Lardy, 2014)

• Non-favored firms (POEs) can borrow only at higher marketrates

• Preferential lending to SOEs impacts on bank profits

• Banks avoid loss by• raising market rates: hurt private firms• reducing deposit rates: hurt households

Introduction The model Calibration Steady State Transition dynamics Conclusion

Historic examples

0

2

4

6

8

10

12

14

1995 2000 2005 2010 2015Source: CEIC/IMF

1995-presentChina deposit and lending rates

Percent

Deposit rate

Lending rate

• Late 1990s: SOE restructuring; bank losses passed tohouseholds

Introduction The model Calibration Steady State Transition dynamics Conclusion

Strict capital account controls in China

• Domestic investors restricted from investing abroad (QDIIvery small)

• Foreign investors restricted from investing in China• FDI allowed, but share in total investment declined to < 2%

since 2009• Financial investment more restrictive (B shares and QFII)

• Capital controls led to persistent deviations of domestic assetreturns from (FX-adjusted) foreign returns: UIP wedge

Introduction The model Calibration Steady State Transition dynamics Conclusion

Order of liberalization question: Ease capital accountrestrictions or financial repression first?

• Literature: Benefits of capital account liberalization not clearunder distorted domestic financial system

• Opening capital account could increase leverage and prob offinancial crisis (Eichengreen, et al. (2011); Enchengreen andLeblang (2003); Chinn and Ito (2006))

• Countries with more developed financial system benefit morethan those less developed (Ju and Wei, 2010)

• Benefits work through “secondary improvements” or“discipline effects” for domestic financial institutions (Kose, etal., 2009; Wei and Tytell, 2004)

Introduction The model Calibration Steady State Transition dynamics Conclusion

Abundant policy discussions, but few formal theories

• Plausible link: capital account liberalization can exacerbatemisallocation under distorted domestic financial system

• However, “there is a lack of formal theories that articulate thislink” (Wei, 2018)

• We present such a theory to study optimal capital accountliberalization under financial repression in China

Introduction The model Calibration Steady State Transition dynamics Conclusion

OLG model

• Households live for two periods: young and old

• The young consumes, works, and saves; and the old consumesaccumulated assets

• Goods produced using labor and capital in two sectors: SOEs(monopolistic competition) and POEs (perfect competition)

• SOEs less productive than POEs

• All firms need to borrow to pay their working capital.

Introduction The model Calibration Steady State Transition dynamics Conclusion

Financial repression and capital controls

• Financial repression: domestic banks are required to lendminimum fraction γ of loans to SOEs at below-market rate

• Capital controls: capital inflows and outflows both taxed(two-way capital controls, τl and τd )

Introduction The model Calibration Steady State Transition dynamics Conclusion

Financial repression raises tradeoffs in capital accountliberalization

• Capital outflow liberalization

• Access to foreign asset market raises returns to householdsavings → improved intertemporal allocations

• Banks pass through higher deposit rate to market lending rates• POE funding costs higher → more misallocation and lower

aggregate TFP

• Capital inflow liberalization

• Competition from foreign investors reduces domestic marketloan rate, benefiting POEs → aggregate TFP ↑

• Banks cut deposit rate to remain solvent → reducingintertemporal allocation efficiency

• Tradeoff between intertemporal allocation and cross-sectorallocation in both cases• Capital account and financial liberalization complementary

Introduction The model Calibration Steady State Transition dynamics Conclusion

The Model

Introduction The model Calibration Steady State Transition dynamics Conclusion

Households• The utility function for household born in period t

E

ln(C y

t )−ΨhH

1+ηt

1 + η+ β ln(C o

t+1)

• Budget constraints

C yt +Dt +Bd

ft + qkt Kot + It +

Ωk

2(ItKot− I

Ko)2Ko

t = wtHt +Tt + Γt

Cot+1 = RtDt + (1− τd )R

∗t B

dft + dt+1 + [qkt+1(1− δ) + rkt+1](K

ot + It )− Γt+1

where Γt+1 denotes bequest

• Capital stock law of motion

Kt = K ot + It = (1− δ)Kt−1 + It

• Capital outflow tax drives wedge b/n domestic deposit rate Rand world rate R∗

Rt = (1− τd )R∗t

Introduction The model Calibration Steady State Transition dynamics Conclusion

Firms

• Final goods production requires intermediate inputs from twosectors: SOE and POE

Yt = Yφtst Y

1−φtpt

• Intermediate goods production requires labor and capital ineach sector j ∈ s, p

Yjt = Aj (Kjt)1−α(Hjt)

α

• Firms face working capital constraint of wage bill and capitalrents

Bjt ≥ (wtHjt + rkt Kjt)

• Cost-minimizing condition

pjt =εj

εj − 1αw α

t (rk)1−α(Rjt)/Ajt

Introduction The model Calibration Steady State Transition dynamics Conclusion

Financial intermediaries (banks)

• Competitive banks take deposits from HHs and lend to firms

• Directed lending: a fraction γ of loans lent to SOEs atbelow-market rates (normalized to 0), 1− γ fraction lent atmarket rate Rlt > 1

Bgt ≥ γ(Bgt + Bt)

• Bank’s break-even condition

Rt = γ + (1− γ)Rlt

• Financial repression creates interest rate wedges:

Rlt > Rt > 1

where interest on ”directed lending” loans is 1.

Introduction The model Calibration Steady State Transition dynamics Conclusion

Capital inflow wedges

• Two wedges on foreign inflows:

• Capital inflow tax τlt : returns for foreign investors subject totaxes

• Risk premia Φ(B lft

Yt): foreign funds supply upward sloping, and

increasing in size of foreign debt to domestic output

• No arbitrage for capital inflows

(1− τl )Rlt = R∗t Φ(B lft

Yt

)

• Risk premium → spillover externality from foreign debt

Introduction The model Calibration Steady State Transition dynamics Conclusion

Reallocation effect of market lending rate

• SOEs’ and POEs’ funding costs:• POEs’ funding cost

Rpt = Rlt

• SOEs’ funding cost

Rst =Bgt

Bst× 1 +

Bst − Bgt

BstRlt

• SOEs’ funding cost is less sensitive to changes in marketlending rate than POEs’.

• An increase in market lending rate reallocates resources fromPOEs to SOEs.

Introduction The model Calibration Steady State Transition dynamics Conclusion

Market clearing and equilibrium

• Final goods market clearing

NXt = Yt − C yt − C o

t − It −Ωk

2(ItK ot− I

K ot

)2K ot

• Labor market and capital market clearing

Ht = Hst +Hpt Kt−1 = Kst +Kpt

• Loanable funds market clearing

Bst + Bpt = Bgt + Bt + B lft

• Balance of payments condition

NXt + (R∗t−1 − 1)Bdf ,t−1 −

[R∗t−1Φ

(B lf ,t−1

Yt−1

)− 1

]B lf ,t−1

= (Bdft −B l

ft )− (Bdf ,t−1 −B l

f ,t−1) + ∆t

Introduction The model Calibration Steady State Transition dynamics Conclusion

Calibration

Introduction The model Calibration Steady State Transition dynamics Conclusion

Calibrate to Chinese data where possible

• Productivity gap: As = 1 and Ap = 1.42 [Hsieh and Klenow(2009)]

• SOE share: φ declines from 0.5 to 0.3 (structural changes)

• Financial repression: γ = 0.5 to match SOE share in bankloans in 2000

• Capital control taxes: τd = 16.62% and τl = 5.09%, to matchratios of privately held foreign assets and liabilities to output

Introduction The model Calibration Steady State Transition dynamics Conclusion

Steady State

Introduction The model Calibration Steady State Transition dynamics Conclusion

One-way liberalization of capital outflows (↓ τd , inflowsnot allowed)

0 0.2 0.4 0.6 0.80

0.01

0.02Capital outlfow

0 0.2 0.4 0.6 0.8-1

0

1Capital inflow

0 0.2 0.4 0.6 0.8

1.4

1.5

1.6

Domestic deposit rate

0 0.2 0.4 0.6 0.8

1.8

2

2.2

POE funding cost

0 0.2 0.4 0.6 0.8Outflow tax d

0.575

0.58

0.585Aggregate TFP

0 0.2 0.4 0.6 0.8-9

-8.8

-8.6Utility

Introduction The model Calibration Steady State Transition dynamics Conclusion

One-way outflow liberalization (no inflows)

• Sufficient reduction in τd leads to capital outflows

• No arbitrage raises deposit rate → HH returns on savingsincreased

• Financial repression → Banks ↑ market rate, hurting POEs

• Reallocation to SOEs reduces aggregate TFP

• Tradeoff between higher returns on HH savings and lowerTFP → interior welfare optimum

Introduction The model Calibration Steady State Transition dynamics Conclusion

One-way liberalization of capital outflows (inflows allowed)

0 0.2 0.4 0.6 0.80

0.01

0.02

Capital outlfow

0 0.2 0.4 0.6 0.8

1.4

1.6

1.8

210 -3 Capital inflow

0 0.2 0.4 0.6 0.8

1.4

1.5

1.6

Domestic deposit rate

0 0.2 0.4 0.6 0.8

1.8

2

2.2

POE funding cost

0 0.2 0.4 0.6 0.8Outflow tax d

0.584

0.585

0.586

0.587Aggregate TFP

0 0.2 0.4 0.6 0.8-9

-8.8

-8.6

Utility

Introduction The model Calibration Steady State Transition dynamics Conclusion

One-way liberalization of capital inflows (↓ τl , outflows notallowed)

0 0.2 0.4 0.6 0.8-1

0

1Capital outlfow

0 0.2 0.4 0.6 0.80

1

210 -3 Capital inflow

0 0.2 0.4 0.6 0.81.35

1.36

1.37

Domestic deposit rate

0 0.2 0.4 0.6 0.81.7

1.72

1.74

POE funding cost

0 0.2 0.4 0.6 0.8

Inflow tax l

0.584

0.585

0.586

0.587Aggregate TFP

0 0.2 0.4 0.6 0.8-9.02

-9

-8.98

-8.96Utility

Introduction The model Calibration Steady State Transition dynamics Conclusion

One-way liberalization of capital inflows (no outflows)

• Sufficient reduction in τl attracts foreign capital inflows

• Competition from foreign investors reduces market loan rate,benefiting POEs → ↑ aggregate TFP

• Financial repression → banks lower deposit rate, reducing HHinterest earnings

• Risk premium → exacerbate over-borrowing externality

• Tradeoff between higher TFP, lower HH asset returns andover-borrowing externality → interior optimum of capitalinflow controls

Introduction The model Calibration Steady State Transition dynamics Conclusion

One-way liberalization of capital inflows (outflows allowed)

0 0.2 0.4 0.6 0.80

1

2

310 -3 Capital outlfow

0 0.2 0.4 0.6 0.80

1

210 -3 Capital inflow

0 0.2 0.4 0.6 0.8

1.36

1.365

1.37

Domestic deposit rate

0 0.2 0.4 0.6 0.8

1.72

1.73

1.74

POE funding cost

0 0.2 0.4 0.6 0.8Inflow tax l

0.584

0.585

0.586

0.587Aggregate TFP

0 0.2 0.4 0.6 0.8

-8.96

-8.94

-8.92

-8.9Utility

Introduction The model Calibration Steady State Transition dynamics Conclusion

Optimal capital controls given financial repression (γ)

0 0.2 0.4 0.60.08

0.09

0.1

0.11Capital outflow tax

0 0.2 0.4 0.60

0.1

0.2Capital inflow tax

0 0.2 0.4 0.61.44

1.46

1.48

1.5Domestic deposit rate

0 0.2 0.4 0.61.5

2

2.5Market loan rate

0 0.2 0.4 0.6

Financial repression .

0.57

0.58

0.59

0.6Aggregate TFP

0 0.2 0.4 0.6-8.55

-8.5

-8.45

-8.4Utility

Introduction The model Calibration Steady State Transition dynamics Conclusion

Liberalizations of capital controls and financial repressionare complementary

• Higher γ implies higher optimal capital control taxes onoutflows• ↑ γ → ↑ market rate, ↑ misallocation and ↓ TFP• Planner ↑ τd to lower market rate and to partly undo

misallocation

• Higher γ implies higher optimal capital control taxes oninflows• ↑ γ raises market rate and attracts more capital inflows →

more external borrowing• To mitigate over-borrowing externality, planner raise τl

• Note interior optimal γ > 0 to partly undo monopoly power ofSOEs

Introduction The model Calibration Steady State Transition dynamics Conclusion

Transition Dynamics

Introduction The model Calibration Steady State Transition dynamics Conclusion

Transition dynamics

• SOE share falls from initial steady state φ0 = 0.5 to newsteady state φ1 = 0.3

• Examine optimal magnitude and speed of capital accountliberalization along transition path

• Notation (Example for τd)

• τdt = τd0 + (τd1 − τd0)[1− (1− αd )t ] if t ≥ 1

• τd0 initial SS value of τd• τd1 final SS value of τd• αd ∈ [0, 1] pace of liberalization of τd• Larger values of αd → faster pace of liberalization

• Similar for τl and γ• τlt = τl0 + (τl1 − τl0)[1− (1− αl )

t ] if t ≥ 1• γt = γ0 + (γ1 − γ0)[1− (1− αγ)t ] if t ≥ 1

Introduction The model Calibration Steady State Transition dynamics Conclusion

Transition dynamics

• Welfare along transition path (starting in period 1)

V1(τd1, αd , τl1, αl , γ1, αγ) =∞

∑t=1

βt

(ln(C y

t )−ΨhH

1+ηt

1 + η+ ln(C o

t )

)

Introduction The model Calibration Steady State Transition dynamics Conclusion

Liberalizing capital outflows and financial repression alongtransition paths

Case 0 1 2Description Benchmark Optimal τd1, αd Optimal τd1, αd , γ1, αγ

τd1 16.62% 5.84% 0.00%αd - 15.11% 58.47%τl1 5.09% 5.09% 5.09%αl - - -γ1 50.00% 50.00% 3.09%αγ - - 88.43%

Welfare gainsalong transition 0.00% 1.94% 11.26%Welfare gains

at new SS 0.00% 36.84% 58.91%

Case 0: benchmark (initial SS);Case 1: optimal liberalization of outflow controls for

transition; Case 2: optimal liberalization of both outflow controls and financial

repression for transition.

• Financial liberalization calls for more aggressive capitaloutflow liberalization

• Optimal (second-best) policy calls for fasterfinancial liberalization than capital outflow liberalization

Introduction The model Calibration Steady State Transition dynamics Conclusion

Liberalizing capital inflows and financial repression alongtransition paths

Case 0 1 2Description Benchmark Optimal τl1, αl Optimal τl1, αl , γ1, αγ

τd1 16.62% 16.62% 16.62%αd - - -τl1 5.09% 14.65% 0.01%αl - 36.52% 55.42%γ1 50.00% 50.00% 15.21%αγ - - 100.00%

Welfare gainsalong transition 0.00% 0.36% 7.73%Welfare gains

at new steady state 0.00% 0.69% 17.82%

Case 0: benchmark (initial SS); Case 1: optimal liberalization of inflow controls for

transition; Case 2: optimal liberalization of both inflow controls and financial

repression for transition.

• Financial liberalization again calls for more complete andfaster capital inflow liberalization.

• Financial liberalization should be instantaneous.

Introduction The model Calibration Steady State Transition dynamics Conclusion

Liberalizing capital account and financial repression alongtransition paths

Case 0 1 2τd1 16.62% 2.37% 1.35%αd - 69.69% 59.94%τl1 5.09% 2.37% 2.83%αl - 69.69% 100.00%γ1 50.00% 2.39% 2.23%αγ - 91.06% 89.57%

Welfare gainsalong transition 0.00% 11.27% 11.29%Welfare gains

at new steady state 0.00% 69.33% 64.88%

Case 0: benchmark (initial SS); Case 1: choose τd = τl , αd = αl , γ, and αγ to

maximize transition welfare; Case 2: optimal liberalization policy with no restrictions

on policy parameters.

• Financial liberalization should be radical and rapid

• Faster liberalization of capital inflows than outflows:accelerating transition by helping productive private firms

Introduction The model Calibration Steady State Transition dynamics Conclusion

Conclusion

• We present a OLG model to study implications of capitalaccount liberalization under financial repression• Liberalizing capital controls incurs tradeoff between aggregate

productivity and intertemporal allocation• Easing capital inflows reduces funding costs for productive

POEs and improves TFP, but depresses deposit rate on HHsavings

• Easing capital outflows improves returns on HH savings, butraising funding costs for POEs and reduces TFP

• Along transition path, liberalizing financial repression andeasing capital controls are complementary reforms

• Liberalization of financial repression calls for more aggressivecapital account liberalization

• Second-best policy calls for gradual liberalization of bothcapital inflows and outflows ...

• ... but fast liberalization of financial repression

Introduction The model Calibration Steady State Transition dynamics Conclusion

Other calibrated parameters

Parameter Description Valueβ Household discount rate 0.665η Inverse of labor supply elasticity 2

Ψh Utility weight of labor 38δ Capital depreciation rate 0.651

Ωk Capital adjustment cost 1r ∗ Foreign interest rate 1.629Γ Transfer from old to young 0.75θ Fraction of working capital 1α Labor income share 0.5ε Elasticity of substitution between SOE products 20

Φb Elasticity of risk preimum to external debt-to-GDP ratio 3