fagan2888 / can-uncertainty-promote-trade

Code repository for "Can Global Uncertainty Promote International Trade?" (JIE)

"Can Global Uncertainty Promote International Trade?"

Code repository for "Can Global Uncertainty Promote International Trade?" By Isaac Baley, Laura Veldkamp, and Michael Waugh

Journal of International Economics, May 2020

---

PRELIMINARIES

• Some routines use functions of the compecon library by Miranda and Fackler. http://www4.ncsu.edu/~pfackler/compecon/toolbox.html

• See Appendix C for details on computation and parameterization.

---

Figure 1: Trade Policy Uncertainty and Exports

The folder Figure-1 contains:

• TPUandExports.xlsx Excel file with two data series.

1. US exports from NIPA (quarterly frequency).

2. US Trade Policy Uncertainty (monthly frequency). Categorical Economic Policy Uncertainty Index for the U.S. constructed by Baker, Bloom, and Davis and downloaded from http://www.policyuncertainty.com. Variables are normalized to 100 for 2014Q1.

• Figure1.m

• Reads Excel data from TPUandExports.xlsx

• Plots fig1_tpuexports.eps

---

Figure 2: Export Coordination under Perfect and Imperfect Information

Figure 3: Uncertainty Increases Terms of Trade Volatility and Average

The folder Figures-2-and-3 contains:

• Simu_twoprecisions.m: Solves the model with CES preferences under perfect (zero signal noise) and imperfect information (limit to infinite signal noise) and simulates for T=20 periods.

  • Calls PI.m to guess initial coefficients from solution of perfect information case.

  • Calls modelsolve.m to solve for imperfect information policy.

  • Calls simulation.m to generate series of outcomes (exports and terms of trade).

• Saves the results from the simulations in results_twoprecisions.mat

• Figure23.m

  • Reads results_twoprecisions.mat

  • Plots fig2_simutradelevel.eps

  • Plots fig3_simutermstrade.eps

---

Figure 4: Effect of Uncertainty in Trade Depends on the Elasticity of Substitution

The folder Figure 4 contains:

• Solve_twoelasticities.m Solves the model for two elasticities of substitution and various levels of signal noise.

  • Calls PI.m to guess initial coefficients from solution of perfect information case.

  • Calls modelsolve.m to solve for imperfect information policy.

  • Saves the results in results_twoelasticities.mat

• Figure4.m

  • Reads results_twoelasticities.mat

  • Plots fig4_elasticity.eps

---

Figure 5: Higher Uncertainty Brings Average Foreign Beliefs Closer to the Prior

The folder Figure-5 contains:

• Figure5.m Computes average foreign beliefs for low and high realizations of domestic endowment, for various levels of signal noise.

  • Plots fig5_priorbeliefs.eps

---

Figure 6: State-dependent Responses to Uncertainty

The folder Figure-6 contains:

• Solve_statedep.m c Solves the model for low elasticities of substitution and various levels of signal noise.

  • Calls PI.m c to guess initial coefficients from solution of perfect information case.

  • Calls modelsolve.m c to solve for imperfect information policy.

  • Saves the results in results_statedep.mat

• Figure6.m c

  • Reads results_statedep.mat

  • Plots fig6_statedep.eps

---

Figure 7: Uncertainty Decreases Trade Coordination and Increases Utility Correlation

The folder Figure-7 contains:

• Solve_risksharing.m Solves the model for a low elasticity of substitution and various levels of signal noise.

  • Calls PI.m to guess initial coefficients from solution of perfect information case.

  • Calls modelsolve.m to solve for imperfect information policy.

  • Calls simulation.m to compute trade and utility correlation across T=100,000 draws.

  • Saves the results in results_statedep.mat

• Figure7.m

  • Reads results_risksharing.mat

  • Plots fig7_risksharing.eps

---

Figure 8: Completing the Market Reduces Exports

The folder Figure-8 contains:

• Solve_financial.m Solves the model with two types of agents: a fraction with perfect information (complete markets) and a fraction with imperfect information. Considers low elasticity of substitution.

  • Calls PI.m to guess initial coefficients from solution of perfect information case.

  • Calls modelsolve_financial.m to compute equilibrium policies.

  • Calls simulation_financial.m to simulate T=1,000 draws.

  • Saves the results in results_financial.mat

• Figure8.m

  • Reads results_financial.mat

  • Plots fig8_financial.eps

---

Figure 9 (Appendix): Comparative statistics for risk aversion

The folder Figure-9 contains:

• Solve_twoelasticities_riskaversion.m Solves the model three times with different levels of risk aversion. For each level of risk aversion, solves the model for low and high elasticities of substitution and various levels of signal noise.

  • Calls PI.m to guess initial coefficients from solution of perfect information case.

  • Calls modelsolve.m to solve for imperfect information policy.

  • For sigma = 1-theta (benchmark), saves the results in results_ra_bench.mat

  • For sigma = 0 (neutral), saves the results in results_ra_neutral.mat

  • For sigma = 1.5 (averse), saves the results in results_ra_averse.mat

• Figure9.m

  • Reads the three files: results_ra_bench.mat, results_ra_neutral.mat, and results_ra_averse.mat

  • Plots fig9_riskaversion.eps