Roberto Garcia-Saltos
Transcription
Roberto Garcia-Saltos
Roberto Garcia-‐Saltos Research Department: Economic Modeling Division International Monetary Fund March 2015 1 Key Models used for Produc/on The Global Integrated Monetary and Fiscal Model (GIMF) The Global Economy Model (GEM) The Flexible System of Global Models (FSGM) comprised of three core modules (G20MOD, EUROMOD, EMERGMOD) The Global Projection Model (GPM) 2 Global Integrated Monetary and Fiscal Model (GIMF) Multi-‐country structural DSGE model Derived fully from the optimizing behavior of households and firms Full intertemporal stock-‐flow accounting Important stabilization role for policy owing to: Frictions from sticky prices and wages and real adjustment costs Households with finite planning horizons Liquidity constrained households 3 Households Two types of households Over lapping generations (OLG) households Optimize their consumption-‐savings decisions over a 20 year planning horizon OLG households treat government debt as wealth since they may not live to pay the associated tax liabilities – consequently the path for public debt has important economic implications Liquidity constrained (LIQ)households Do not save, do not have access to credit, consume their disposable income each period 4 Firms Firms produce tradable and nontradable goods Firms are owned by OLG households and managed according to their preferences Firms face adjustment costs in: Prices Investment Labor hiring 5 Interna/onal Linkages All bilateral trade flows explicitly modeled Export and import of intermediate and final goods All relative prices for each region are also explicitly modeled Uncovered interest parity determines exchange rates Global savings and investment decisions (a function of household’s finite planning horizons) determine the global real interest rate 6 Fiscal Policy The fiscal authority: purchases final goods directly for both consumption and investment, makes transfers to households, and funds these with a full range of tax instrument The fiscal authority can impose: indirect taxes on consumption, direct taxes on labor and corporate income, as well as lump-‐sum taxes on households, and tariffs on imported goods Fiscal policy is implemented via a simple policy rule that ensures a stable debt-‐to-‐GDP ratio in the long run, but allows for a countercyclical fiscal stance in the short run 7 Monetary Policy Monetary policy is characterized as inflation-‐forecast targeting The policy rate adjusts to the gap between the one-‐year ahead model-‐consistent inflation forecast and the target rate 8 GIMF’s Strengths Because of GIMF’s overlapping-‐generations structure fiscal policy actions that alter the path of public debt have important short-‐run macroeconomic consequences This make GIMF particularly useful for examining fiscal policy issues from short-‐run temporary stimulus to long-‐term consolidations Because of its detailed structure, the implications of using the full range of fiscal policy instruments can be analyzed 9 The Global Economy Model (GEM) Key difference from GIMF is that households live forever The important implication is that fiscal policy has far less powerful short-‐run effects – households do not treat government debt as wealth and thus the path for public debt has less significant implications for economic outcomes GEM is quarterly whereas as GIMF is annual Thus GEM is slightly more appropriate for monetary policy issues since monetary policy decision are taken more frequently than fiscal policy decisions GEM includes full production sectors for oil and non-‐oil commodities Useful for analyzing the impact of shale gas in the US for example 10 Both GIMF and GEM Have Size Constraints Because GIMF and GEM have multiple goods and all bilateral trade flows are tracked, there is a limit on the number of countries/regions that can be modeled at any time Largest production versions have six regions, typically the US, the euro area, Japan, emerging Asian, Latin America, and the rest of the world However, there is high demand for more individual country results, for example in the work for the G20MAP there is a need to be able to model all the G20 countries individually This brings us to FSGM, the newest system of models 11 FSGM (The Flexible System of Global Models) Comprises 3 core models – G20MOD, EUROMOD, and EMERGMOD Each model captures the whole world: G20MOD has each individual G20 country plus several other blocks capturing the rest of the world EUROMOD merges some G20 countries into large blocks so it can include individual blocks for 11 major euro area countries EMERGMOD also merges some G20 countries into large blocks to it can individually model more emerging market economies 12 FSGM (The Flexible System of Global Models) Semi-‐structural with single good rather than fully structural with multiple goods like GIMF and GEM Private consumption and investment structural (micro-‐ founded) Trade, labor supply, and inflation have reduced-‐form representations Supply is determined by an aggregate Cobb Douglas production function Full stock-‐flow consistency with rational expectations Monetary and fiscal policy are endogenous -‐ pinned down with simple rules (same as GIMF and GEM) 13 Why a Semi-‐Structural Model For many exercises it is desirable to have results for a large number of countries G20+ Fully structural models like GIMF and GEM with multiple goods and full tracking of bilateral trade flows are to hard to solve with more than 6 countries or regions (memory limits in simulation software) Collapsing to a single good with reduced-‐form trade dramatically reduces the complexity of the problem making it feasible to have 24 blocks in FSGM modules 14 Minimizing the Tradeoffs Giving up structure comes at a cost, but the model has been developed to minimize those costs Development process strives to ensure that FSGM maintains insights from the fully structural models Further because of FSGM’s simpler structure, more empirical validation is feasible At this point parameters come largely from single equation estimation (GIMF and GEM are calibrated) Bayesian Simulated Method of Moments with System Priors (BSMMSP) technique is being developed to do full system estimation 15 The Applica/ons of GIMF, GEM and FSGM GIMF, GEM and FSGM are used to generate scenarios and policy analysis for a wide range of Fund outputs Scenarios are generated for use in the WEO, REOs, the vulnerability exercises, FSAPs, and the spillover report Policy analysis generated with the models is used in Staff Reports and Selected Issues Papers to support policy recommendations in ART IV consultations as well as in the WEO 16 The Global Projec/on Model (GPM) GPM is primarily a forecasting model whereas GIMF, GEM and FSGM are used for scenarios and policy analysis GPM is the simplest model in terms of structure It is a reduced-‐form model with only a handful of key behavioral equations Its smaller size makes full system estimation of many model’s parameters feasible The main production version contains seven regions: the United States, the euro area, Japan, China, Emerging Asia-‐Ex China, Latin America, and the rest of the world 17 The Global Projec/on Model (GPM) Several other versions of GPM have been developed A euro area version has been built for the European Department which models individually Germany, France, Italy and Spain An eleven region version is being developed that includes more individual Asian economies Several country specific versions that take the results of the GPM as given. 18 The Global Projec/on Model (GPM) GPM is used to generate quarterly economic projections The GPM projections are used as an input into the WEO forecasting process The GPM projections are prepared jointly by RES and area departments 19 Other Models The Modeling Division also has a range of other small models that are developed for special applications. Some examples include: Small models that include expanded financial sectors are used for macro prudential analysis Small models with detailed public debt maturity structures are used for debt sustainability analysis Small model of the oil market used to evaluate scenarios for world GDP growth, oil prices and oil production. 20 Training and Support The Modeling Division shares its models both internally and externally GIMF internal and external use is supported with semi annual training provided by the Division GPM is provided to a network of policy institutions around the world, primarily central banks that use it to generate globally consistent forecasts for use in their domestic forecasting exercise GEM has been distributed to many policy institutions world wide, but is not supported with regular training Once FSGM has been completed, it too will be shared 21 Summary The Modeling Division in Research builds and supports a wide range of macroeconomic models Different models have different strengths and weakness and are thus used for different purposes The Division’s models are constantly being refined and extended with current focus on expanded country coverage and macro-‐financial linkages 22