Comments for Terms of Trade Shocks and Investment in Commodity Exporter Economies by Fornero, Kirchner and Yany Alfonso Irarrazabal (Norges Bank) October, 2014
Interesting paper I Question: the e ect of exogenous commodity prices on the business cycle of commodity producers. I Empirical contribution: to study the e ect of investment in real activity for 5 commodity producer economies. I Theory contribution: to propose investment in commodity sector as the main driver for a boom in real activity (specially when commodity price is high and stays high.) I Calibrate the model for one country: Chile (I like the one country focus)
Two results I Empirical section: They find for five commodity producers an e ect of commodity prices on real activity. I DSGE application: they find a strong e ect of cooper prices on real total GDP. Also, cooper prices explain the expansion in investment in the cooper sector during the boom in prices. I Today, I will focus on two aspects that need to be addressed before we answer the main question of this paper.
Empirical Analysis I They estimate a VAR with block exogeneity to study the e ect of oil prices on real total GDP. I 6 countries: Australia, Canada, Chile, New Zeland, Peru, and South Africa. I Exogenous block (4 vars): real world GDP, annual US inflation, US fed interest rate, real commodity price index with the same ordering for exogeneity for identification. I Endogenous block (7 vars): real GDP, nominal non-mining investment/nominal GDP, nominal mining investment/nominal GDP, annual CPI inflation rate, annual nominal monetary policy rate, the real exchange rate, and CA/nominal GDP.
VAR Results I The e ect stronger for SA and Peru, then Chile and negative for Canada. Is this driven by mining GDP?
VAR Results I The e ect on non-mining investment larger in Canada, Chile and Peru.
The model I Standard monetary SOE model I Non-commodity tradable and imported sectors I Nominal rigidity in prices (domestically produced), imported goods. I Commodity sector: use capital, and investment produced partially with domestic good. I Fiscal and monetary policy rules.
Commodity sector I Production of commodity Y S,t = a S,t F S (T t, K S,t 1 ) I Capital adjustment K S,t =(1 S)K S,t +[1 S (X S,t n+1 /X S,t n )] I Composition of investment (CES with elasticity Is ) I S,t = F [I H,t (S), T F,t (S)] for given relative prices (P H,t /P IS, t)
Aggregate equilibrium I Market clearing condition for each domestic variety Y H,t (z H,t )=[P H,t (z H )/P H,t ] H Y H,t + PH,t(z H )/PH,t H YH,t with Y H,t = C H,t + I H,t + I H,t (S)+G H,t I Real GDP is Y t = C t + I t + I S,t + G t + X t M t I Notice here that a change in I s,t is likely to be absorbed by trade balance adjustments. This could be problematic to generate a large impact from investment in the commodity sector.
Main model results I 15 exog variables. Foreign exog variables from VAR from previous exercise. I Need to clarify how are re estimated the parameters (and shocks) for thw two models (with persistent and transitory) with di erent process for cooper prices. I For the exog VAR, why do all foreign GDP, foreign inflation and interest rate move with cooper prices? Need to check this!
Main model results I The more persistent the shock the more persistent the e ect (is the question how persistent? amplification?) I Better to show the e ect on non-commod GDP (as for investment) I Also, the issue of foreign variables (as foreign GDP moving) makes harder the comparison.
Minor issues I To understand mechanism, I would work with a smaller model. I What happens with violations of UIP? I I would be instructive to estimate/calibrate the model for Peru, a country with less active fiscal (and monetary?) policy.
Final remarks I Besides my issues with the VAR, it would be nice to understand (using a similar block exogeneity assumption) the e ect of commodity prices on investment and GDP non-commodity. I On the modelling, I would understand two aspects of this problem 1. the e ect of fiscal policy on the model (perform external validity tests). 2. the e ect of the investment channel using micro data if possible. I Keep up the good work!