Capital Controls or Macroprudential Regulation?1 Anton Korinek 1 Johns
1
Damiano Sandri
2
Hopkins University and NBER
2 IMF
Research Department
Capital Flows, Systemic Risk, and Policy Responses Reykjavik April 2016
1 The views expressed herein are those of the authors and should not be attributed to the IMF, its Executive Board, or its management. Korinek and Sandri (JHU and IMF)
Capital Controls or Macroprudential?
October 2015
1 / 20
Introduction
Motivation
How to protect open economies against financial instability? Two instruments: Capital controls (CC) Macroprudential regulation (MP)
Both curb credit booms, but so far studied in isolation In this paper, we ask following questions What are the relative merits? Does MP eliminate the need for CC? Or vice versa? If not, what determines the optimal mix?
Korinek and Sandri (JHU and IMF)
Capital Controls or Macroprudential?
October 2015
2 / 20
Introduction
Definitions CC segment domestic and foreign capital markets MP places a wedge between domestic borrowers and all lenders
domestic borrowers
domestic borrowers
borrowers lenders international agents
domestic savers
international agents
Capital Controls
Korinek and Sandri (JHU and IMF)
domestic savers
Macroprudential Regulation
Capital Controls or Macroprudential?
October 2015
3 / 20
Introduction
Models of pecuniary externalities
We analyze CC and MP in models of pecuniary externalities Exchange rate externalities ⇒ both CC and MP are needed Asset price externalities ⇒ MP is sufficient, no need for CC
Korinek and Sandri (JHU and IMF)
Capital Controls or Macroprudential?
October 2015
4 / 20
Introduction
Literature review Ex-ante prudential policies motivated by pecuniary externalities CC due to RER externalities Korinek (2007, 2010), Bianchi (2011) MP due to asset price externalities Lorenzoni (2008), Jeanne and Korinek (2010), Bianchi and Mendoza (2010)
Ex-post policies to alleviate credit crunch Gertler and Kiyotaki (2010), Gertler and Karadi (2011,2013), Del Negro, Ferrero, Eggertsson and Kiyotaki (2011), Sandri and Valencia (2013)
Korinek and Sandri (JHU and IMF)
Capital Controls or Macroprudential?
October 2015
5 / 20
Model with RER externalities
Setup
Model with RER externalities Deterministic equilibrium Small open economy in three time periods t ∈ {0, 1, 2} Three classes of agents domestic borrowers B domestic savers S foreigners that borrow/lend at the risk-free rate
Discount factor and risk-free rate set to zero Domestic savers and borrowers maximize U i = u(ciT,0 ) + u(ciT,1 , ciN,1 ) + u(ciT,2 )
Korinek and Sandri (JHU and IMF)
Capital Controls or Macroprudential?
for
i = B, S
October 2015
6 / 20
Model with RER externalities
Setup
Budget constraints Domestic agents: i i receive endowments yT,t , yN,1
buy/issue bonds denominated in tradable goods bit
Budget constraints: i ciT,0 + bi1 = yT,0 + bi0 i i ciT,1 + pciN,1 + bi2 = yT,1 + pyN,1 + bi1 i ciT,2 = yT,2 + bi2
In period 1, borrowers face credit constraint: B B bB 2 ≥ −φ yT,1 + pyN,1
Korinek and Sandri (JHU and IMF)
Capital Controls or Macroprudential?
October 2015
7 / 20
Model with RER externalities
Laissez-faire equilibrium
Time 1 equilibrium
i , individual agents maximize Defining mi = bi1 + yT,1
i V i mi ; M B , M S = Log (ciT,1 )α (ciN,1 )1−α + Log yT,2 + bi2 i i i + µi mi + p(yN,1 − ciN,1 ) − ciT,1 − bi2 + λi bi2 + φ(yT,1 + pyN,1 )
The FOCs imply uiT,1 = uiT,2 + λi uiT,1 = uiN,1 /p
Korinek and Sandri (JHU and IMF)
Capital Controls or Macroprudential?
October 2015
8 / 20
Model with RER externalities
Laissez-faire equilibrium
Aggregate wealth effects Impact of aggregate wealth on individual utility ∂V j = ujT,1 · ∂M i
∂p j (y − cjN,1 ) i N,1 ∂M {z } |
+ λj ·
redistribution between agents Rij
∂p j φyN,1 i ∂M | {z }
relaxation of constraint Φji
Using market clearing in non-tradable goods ∂p = κ · M P Ci ∂M i where M P CB = 1 Korinek and Sandri (JHU and IMF)
, M P C S = 1/2
Capital Controls or Macroprudential?
October 2015
9 / 20
Model with RER externalities
Laissez-faire equilibrium
Time 0 equilibrium
At time 0 agents solve u(ciT,0 ) + V i mi ; M B , M S
max
subject to i i mi = bi0 + yT,0 − ciT,0 + yT,1
Individual agents take prices as given Standard Euler equation uiT,0 =
Korinek and Sandri (JHU and IMF)
∂V i = uiT,1 ∂mi
Capital Controls or Macroprudential?
October 2015
10 / 20
Model with RER externalities
Planner’s solution
Optimal Prudential Policy
Prudential planner: sets B1i but leaves laissez-faire for t ≥ 1 (as in Stiglitz, 1982, Geanakoplos-Polemarchakis, 1986) The planner sets γ i uiT,0 =
γ i uiT,1 | {z }
private benefit
+
∂V i ∂V j + γj γi i i | ∂M {z ∂M } social benefit
internalizing the effects of borrowing on future exchange rates
Korinek and Sandri (JHU and IMF)
Capital Controls or Macroprudential?
October 2015
11 / 20
Model with RER externalities
Planner’s solution
Implementation The planner’s solution can be implemented with borrowing taxes and saving subsidies uiT,1 uiT,0
= 1 − τi
Optimal taxes are
τ
B
∂p B λB ∂M B φYN,1 = B B − RB uT,0 1 + RB S
τB = Korinek and Sandri (JHU and IMF)
and
∂p B λB ∂M S φYN,1 τ = B B − RB uT,0 1 + RB S S
M P CB S ·τ >0 M P CS
Capital Controls or Macroprudential?
October 2015
12 / 20
Model with RER externalities
Planner’s solution
Capital controls or macroprudential regulation? Proposition In a model with RER externalities, both MP and CC are needed to achieve constrained efficiency. By segmenting domestic borrowers from capital markets ⇒ MP increases τ B without affecting τ S By segmenting domestic versus international markets ⇒ CC lead to an equal increase in both τ B and τ S The appropriate combination of MP and CC is given by 1 − τ CC 1 − τ MP Korinek and Sandri (JHU and IMF)
= 1 − τS 1 − τB = 1 − τS
Capital Controls or Macroprudential?
October 2015
13 / 20
Model with RER externalities
Planner’s solution
Stochastic setting
The results carry forward to a stochastic setting Without state contingent assets → Size of CC and MP depends on likelihood of constraints becoming binding With state contingent assets → Individual agents under-insure → CC and MP should be risk sensitive
Korinek and Sandri (JHU and IMF)
Capital Controls or Macroprudential?
October 2015
14 / 20
Model with RER externalities
Numerical illustration
Numerical illustration: 1997 East Asian Crisis Current account (percent of GDP)
Real exchange rate (1997=100) 105 100 95 90 85 80 75
10 5 0 -5 1995
1997
1999
2001
1995
1997
1999
2001
Balanced growth path if constraint does not bind Financial constraint φ tightens with 5% probability → to match CA surplus and REER depreciation Pre-crisis net foreign assets equal to −40% α 0.3
i YT,0
1
i YT,1
α
Korinek and Sandri (JHU and IMF)
i YN,1
1−α
i YT,2
1−
B0i
B0B
B0S
π
φ(L)
φ(H)
−0.8
0
5%
0.65
∞
Capital Controls or Macroprudential?
October 2015
15 / 20
Model with RER externalities
Numerical illustration
Wealth inequality and optimal taxes Under benchmark calibration, 2 percent CC and MP taxes Current account (percent of GDP) 20 15
42
Planner ↙
0.7 ↗ Laissez faire
0.6
First best 40
12 10 8 6 4 2 0
0.8
5 0
Optimal taxes
First best
0.9 ↖ Planner
10
Real exchange rate 1.0
Laissez faire ↘
0.5 44
46
48
50
40
-BB0 (percent of GDP)
42
44
46
48
50
τB τS
40
-BB0 (percent of GDP)
42
44
46
48
50
-BB0 (percent of GDP)
Greater wealth inequality, i.e. larger gross positions, ⇒ higher optimal taxes Korinek and Sandri (JHU and IMF)
Capital Controls or Macroprudential?
October 2015
16 / 20
Model with asset price externalities
Setup
Model with asset price externalities Domestic agents receive capital k1 that produces output at time 2 Borrowers have access to more efficient production technology F B (k2B ) = Ak2B
,
F S0 (0) = A
, F S00 (k2S ) < 0
Budget constraints: i + bi0 ciT,0 + bi1 = yT,0 i ciT,1 + bi2 = yT,1 + q(k1i − k2i ) + bi1 i ciT,2 = yT,2 + F i (k2i ) + bi2
In period 1, borrowers face credit constraint: B bB 2 ≥ −φqk2 Korinek and Sandri (JHU and IMF)
Capital Controls or Macroprudential?
October 2015
17 / 20
Model with asset price externalities
Laissez-faire equilibrium
Aggregate wealth and asset prices Laissez-faire FOCs uiT,1 = uiT,2 + λi
and
q=
F i0 (k2i ) φ + (1 − φ)uiT,1 /uiT,2
For unconstrained savers, uST,1 = uST,2 and ∂q ∂M S
= 0
⇒ Fisherian separation between consumption and investment B For constrained borrowers, uB T,1 > uT,2 and
∂q ∂M B Korinek and Sandri (JHU and IMF)
> 0
Capital Controls or Macroprudential?
October 2015
18 / 20
Model with asset price externalities
Constrained efficient equilibrium
Planner’s solution The planner reduces borrowing, but does not distort saving τ B = λB
∂q φk2B ∂M B B 1 + RB
τS = 0
Proposition In a model with asset price externalities, MP is sufficient to achieve constrained efficiency. No need for CC.
Korinek and Sandri (JHU and IMF)
Capital Controls or Macroprudential?
October 2015
19 / 20
Conclusions
Conclusions Contractionary RER depreciations ⇒ both CC and MP increase net worth of people who spend on domestic goods, i.e. both borrowers and savers but regulate borrowers more since higher M P C
τB =
M P CB S ·τ >0 M P CS
Fire sales of assets ⇒ MP is sufficient No need to increase savers’ wealth since no impact on asset prices
τB > 0 = τS
Korinek and Sandri (JHU and IMF)
Capital Controls or Macroprudential?
October 2015
20 / 20
