optimal simple fiscal rules for commodity exporters
TRANSCRIPT
Optimal Simple Fiscal Rules for CommodityExporters
Arthur Mendes Steven PenningsPUC-Rio World Bank
9 September 2016
Motivation
I Fiscal policy of commodity exporters often characterized asbeing needlessly pro-cyclical:
I spend in booms and cut in busts (balanced budget rule, BBR)
I Chile “poster child” of good fiscal policy: follows structuralsurplus rule (SSR) – save temporary copper revenues in a SWF
I Kumhof and Laxton (2013) (and others) find large welfaregains from SSR
I mostly because smooth consumption of Hand-to-Mouth HHs
Q: What is the optimal simple fiscal rule (OSR) for commodityexporters?
Motivation
I Fiscal policy of commodity exporters often characterized asbeing needlessly pro-cyclical:
I spend in booms and cut in busts (balanced budget rule, BBR)
I Chile “poster child” of good fiscal policy: follows structuralsurplus rule (SSR) – save temporary copper revenues in a SWF
I Kumhof and Laxton (2013) (and others) find large welfaregains from SSR
I mostly because smooth consumption of Hand-to-Mouth HHs
Q: What is the optimal simple fiscal rule (OSR) for commodityexporters?
This Paper
I Revisit this question in a model some key ingredients:
I Ricardian Households who can borrow/saveI Hand-to-Mouth (HtM) cannot borrow/saveI Government receives commodity revenues (and taxes) and
transfer them to HHs
I Main Result: The OSR is for commodity revenues is(surprisingly) procyclical.
I BBRs are often better than SSR for commodity revenues.
I Examine this question:
1. Analytically in a simple model with only HtM HHs2. Quantitatively in a model with both HtM HH & Ricardian
HHs and exogenous income (RBC results are similar)
Intuition: two observations
1. Permanent Income Hypothesis: HHs should consume out ofpermanent income
1.1 Spend permanent shocks and save temporary shocks
2. Commodity prices are highly persistent (close to random walk)
2.1 Oil/copper ρ =0.9-0.96 (annual); Half life of 7-17 years.
I Highly persistent commodity shocks don’t need smoothing.
I Just spend/transfer current income
I New: Few recent quantitative papers consider response tohighly persistent shocks (mostly calibrate to temporary shocks)
Other findings
1. Shock type is important:Non-commodity revenues are moretemporary and so should be saved
2. Reduce SWF volatility: govts should spend more thaninterest on SWF, otherwise SWF explode with persistentshocks
3. Share the burden: Entirely smoothing incomes of HtM HH ispossible but not optimal: big costs for Ricardian HHs
4. Same rule for all HHs: the optimal transfer rule forRicardians very similar to that for HtM HHs.
Model Overview
I Small open economy (exogenous world interest rate - annual)
I Exogenous non-resource income accrues to households (Y ).
I Exogenous resource income accrues to the govt (Q × P).
I Government able to save/borrow internationally (A)
I Fiscal policy aims to stabilize cons. over the business cycle.
I Find simple rules, where transfers are a function of Y , P , or A
Analytical Model (linear quadratic case)
I Assume only HtM HHs. Benevolent government’s problem:
max{Tr ′′} − E0
∞∑t=0
βt(c ′′t − γ)2 s.t.
such that :
[HH’s budget constraint] c ′′t = (1− τ)Yt + Tr ′′t
[Government’s budget constraint]At = (1 + r)At−1 + τYt + PtQ − Tr ′′t
[Exogenous shocks]Pt − P = ρP(Pt−1 − P) + ePtYt − Y = ρY (Yt−1 − Y ) + eYt
Optimal simple rule
Fiscal Rule : Tr ′′t = Tr ′′+θA(At−1−A)+θY(Yt − Y
)+θPQ(Pt−P)
I Response to oil price: θP =r
1 + r − ρPI As ρP → 0 then θP → r/(1 + r) ≈ 4% (SSR)I As ρP → 1 then θP → 1 (BBR)
I Response to SWF Assets: θA = r ≈ 4%
I Response to non-Res GDP θY = −[1− τ − r
1 + r − ρY
]
Some calculations
Fiscal Rule : Tr ′′t = Tr ′′+θA(At−1−A)+θY(Yt − Y
)+θPQ(Pt−P)
I Response to oil price: e.g . ρP = 0.96 and r = 0.04 then:
I Spend around half of movements in oil prices.
θP =r
1 + r − ρP= 0.5
I Response to non-Res GDP ρY = 0 and τ = 0.15 then
I Strongly counter cyclical rule for non-resource shocks.
θY = −[0.85− 0.04
1 + 0.04− 0
]≈ −0.8
Changes in the Quantitative ModelI Hand to Mouth Household. Same as in the analytical model with
population share 0.5 C ′′t = (1− τ)Yt + Tr ′′t
I The Ricardian household chooses foreign savings Bt and (perRicHH) consumption C ′
t to solve
max{Bt ,C ′t }
∞∑t=0
βtC ′1−σt
1− σs.t C ′t = Rt−1Bt−1 + (1− τ)Yt + Tr ′t − Bt
I Debt-elastic interest spread
I Government maximizes household welfare, which is equivalentto minimizing loss function: L = 0.5 {Var(c ′t) + Var(c ′′t )}
I Calibrate to Algeria (or Trinidad and Tobago)
Main result: BBR does surprisingly well
0.18 0.38 0.66 0.88 0.98persistence of oil-price shock (annual)
0
1
2
3
4
% o
f Css
Welfare Loss
BBRSSR
0.18 0.38 0.88 0.98persistence of oil-price shock (annual)
0
0.2
0.4
0.6
0.8
1
annu
al s
td d
evia
tion
Ric HH consumption
0.18 0.38 0.78 0.88 0.98persistence of oil-price shock (annual)
0
0.2
0.4
0.6
0.8
1
annu
al s
td d
evia
tion
HtM HH consumption
0.18 0.38 0.88 0.98persistence of oil-price shock (annual)
0
1
2
3
4
annu
al s
td d
evia
tion
Public Assets
Fiscal Rules
1. Full HtM Stabilization - govt completely smooths HtM’s cons
2. Balanced Budget Rule (BBR) - spend all income each period
3. Structural Surplus Rule (SSR) - only spend interest (save rest)
4. Hybrid BBR-CCY - Like BBR, but countercyclical w.r.t non-res Y
5. Hybrid SSR-CCY- Like SSR, but countercyclical w.r.t non-res Y
6. Optimal Simple Rule (OSR) chooses all parameters optimally minwelfare loss
7. OSR-Equal like OSR but where we must transfer the same to bothHHs
Large welfare loss from stabilizing HtM Cons
ψ = 0.01 (1) (2) (3) (4) (5) (6) (7)
ρP = 0.94 HtMHH BBR SSR BBR CCY SSR CCY OSR OSR Equal
θ′a 0.16 0.10 0.10 0.10 0.10 0.10 0.09
θ′y 0.02 0.16 0.00 -0.84 -0.84 -1.00 -0.71
θ′p 1.31 1.00 0.00 1.00 0.00 0.59 0.73
θ′′a 0.00 0.10 0.10 0.10 0.10 0.06 0.09
θ′′y -0.84 0.16 0.00 -0.84 -0.84 -0.75 -0.71
θ′′p 0.00 1.00 0.00 1.00 0.00 0.78 0.73
sd(c ′) 0.37 0.19 0.19 0.19 0.19 0.18 0.18
sd(c ′′) 0.00 0.19 0.22 0.19 0.22 0.19 0.18
sd(a) 1.37 0.04 3.00 0.13 3.00 1.21 0.95
sd(b) 0.69 1.78 3.13 1.77 3.13 0.47 0.24
Loss (% of Css) 6.96 3.59 4.26 3.54 4.23 3.38 3.38
I Full HtM Stabilization - govt completely smooths HtM’sconsumption
BBR performs better than SSR
ψ = 0.01 (1) (2) (3) (4) (5) (6) (7)
ρP = 0.94 HtMHH BBR SSR BBR CCY SSR CCY OSR OSR Equal
θ′a 0.16 0.10 0.10 0.10 0.10 0.10 0.09
θ′y 0.02 0.16 0.00 -0.84 -0.84 -1.00 -0.71
θ′p 1.31 1.00 0.00 1.00 0.00 0.59 0.73
θ′′a 0.00 0.10 0.10 0.10 0.10 0.06 0.09
θ′′y -0.84 0.16 0.00 -0.84 -0.84 -0.75 -0.71
θ′′p 0.00 1.00 0.00 1.00 0.00 0.78 0.73
sd(c ′) 0.37 0.19 0.19 0.19 0.19 0.18 0.18
sd(c ′′) 0.00 0.19 0.22 0.19 0.22 0.19 0.18
sd(a) 1.37 0.04 3.00 0.13 3.00 1.21 0.95
sd(b) 0.69 1.78 3.13 1.77 3.13 0.47 0.24
Loss (% of Css) 6.96 3.59 4.26 3.54 4.23 3.38 3.38
I Balanced Budget Rule (BBR) - spend all income each period;Structural Surplus Rule (SSR) - only spend interest (save rest)
Further improvements when countercyclical
ψ = 0.01 (1) (2) (3) (4) (5) (6) (7)
ρP = 0.94 HtMHH BBR SSR BBR CCY SSR CCY OSR OSR Equal
θ′a 0.16 0.10 0.10 0.10 0.10 0.10 0.09
θ′y 0.02 0.16 0.00 -0.84 -0.84 -1.00 -0.71
θ′p 1.31 1.00 0.00 1.00 0.00 0.59 0.73
θ′′a 0.00 0.10 0.10 0.10 0.10 0.06 0.09
θ′′y -0.84 0.16 0.00 -0.84 -0.84 -0.75 -0.71
θ′′p 0.00 1.00 0.00 1.00 0.00 0.78 0.73
sd(c ′) 0.37 0.19 0.19 0.19 0.19 0.18 0.18
sd(c ′′) 0.00 0.19 0.22 0.19 0.22 0.19 0.18
sd(a) 1.37 0.04 3.00 0.13 3.00 1.21 0.95
sd(b) 0.69 1.78 3.13 1.77 3.13 0.47 0.24
Loss (% of Css) 6.96 3.59 4.26 3.54 4.23 3.38 3.38
I Hybrid BBR/SSR-CCY - Like BBR/SSR, but countercyclical withrespect to non-resource GDP
OSR very similar to BBR_CCY
ψ = 0.01 (1) (2) (3) (4) (5) (6) (7)
ρP = 0.94 HtMHH BBR SSR BBR CCY SSR CCY OSR OSR Equal
θ′a 0.16 0.10 0.10 0.10 0.10 0.10 0.09
θ′y 0.02 0.16 0.00 -0.84 -0.84 -1.00 -0.71
θ′p 1.31 1.00 0.00 1.00 0.00 0.59 0.73
θ′′a 0.00 0.10 0.10 0.10 0.10 0.06 0.09
θ′′y -0.84 0.16 0.00 -0.84 -0.84 -0.75 -0.71
θ′′p 0.00 1.00 0.00 1.00 0.00 0.78 0.73
sd(c ′) 0.37 0.19 0.19 0.19 0.19 0.18 0.18
sd(c ′′) 0.00 0.19 0.22 0.19 0.22 0.19 0.18
sd(a) 1.37 0.04 3.00 0.13 3.00 1.21 0.95
sd(b) 0.69 1.78 3.13 1.77 3.13 0.47 0.24
Loss (% of Css) 6.96 3.59 4.26 3.54 4.23 3.38 3.38
I Optimal Simple Rule (OSR) chooses all parameters optimally minwelfare loss
Same rule for all HHs just as good
ψ = 0.01 (1) (2) (3) (4) (5) (6) (7)
ρP = 0.94 HtMHH BBR SSR BBR CCY SSR CCY OSR OSR Equal
θ′a 0.16 0.10 0.10 0.10 0.10 0.10 0.09
θ′y 0.02 0.16 0.00 -0.84 -0.84 -1.00 -0.71
θ′p 1.31 1.00 0.00 1.00 0.00 0.59 0.73
θ′′a 0.00 0.10 0.10 0.10 0.10 0.06 0.09
θ′′y -0.84 0.16 0.00 -0.84 -0.84 -0.75 -0.71
θ′′p 0.00 1.00 0.00 1.00 0.00 0.78 0.73
sd(c ′) 0.37 0.19 0.19 0.19 0.19 0.18 0.18
sd(c ′′) 0.00 0.19 0.22 0.19 0.22 0.19 0.18
sd(a) 1.37 0.04 3.00 0.13 3.00 1.21 0.95
sd(b) 0.69 1.78 3.13 1.77 3.13 0.47 0.24
Loss (% of Css) 6.96 3.59 4.26 3.54 4.23 3.38 3.38
I OSR-Equal like OSR but where we must transfer the same to bothHHs
Conclusions and Caveats
I This paper: BBR do surprisingly well for commodity exporters,because commodity prices are highly persistent.
I But governments should respond differently to different shocks.
I Caveat 1 (Scope): Only look at cons. volatility. Results mightbe different results for public investment. Political constraints.
I Caveat 2 (Linearities): We ignore non-linearities fromborrowing constraints and ignore precautionary savings.
I Caveat 3 (Multipliers): Limited interactions between fiscalpolicy and other parts of the economy.