Restricted Discussion of “General equilibrium model of Bank Indonesia ((GEMBI)) 2007” Feng Zhu Bank for International Settlements 1 Restricted Overview of the paper z BI experience: p practical p DSGE modelling g ¾ Oriented towards policy issues: theory & reality ¾ Emerging g g economy y features: complexity y ¾ Large & frequent changes in economic structure ¾ Transition dynamics vs. steady state z Great efforts & progress ¾ Well-conceived & developed p ¾ Impulse responses to important shocks ¾ Very interesting findings 2 Restricted DSGE Modeling in Central Banks z Canonical models • • • Christiano, Eichenbaum & Evans (2005) Laxton & Pesenti (2003) S t & Wouter Smets W t (2003) z Why DSGEs at central banks? • • Better policy analysis Better forecasts, or both z Some examples • • BEQM, MAS, NEMO, RAMSES, ToTEM Many CBs have DSGE(s), some use a DSGE as benchmark 3 Restricted GEMBI: features I z Six sectors ((complex p GEMBI)) ¾ Household; Firms (intermediate/importers & final): agriculture; Banking; Government (FA & BI); External sector (ROW) ¾ Complex: 33 equations, 43 parameters z Rich & interesting features ¾ Optimisation-based model, hybrid expectations ¾ General equilibrium with clear transmission ¾ Heterogeneous households (complex version) ¾ Agriculture; Commodity producers 4 Restricted GEMBI: features II z Core inflation ¾ Unconventional definition: interpreted as PPI ¾ But really: less volatile component here ¾ Permanent-transitory decomposition z Separate budget constraints for CB & FA ¾ CB b balance l sheet h t concerns can llead d tto iindeterminacy d t i or instability ¾ Issuing CB bills without fiscal backup can lead to balance sheet constraints on policy: BG≠ BCB ¾ Definition of fiscal rule (spending): taxation ¾ Analysis of zone of indeterminacy & instability 5 Restricted Rigor-fit tradeoff: DSGE free lunch? z Theoretical rigor & goodness of fit ¾ Theoretical consistency: compelling story? ¾ Data coherency: good fit? ¾ Parsimony (theory) & complexity (reality): good compromise? z Theoretical consistency? ¾ Completely specified, economy-wide, probabilistic model for observed time series z Good fit to data & superior forecasts? ¾ Smets and Wouters (2003), Edge, Kiley & Laforte (2006), Adolfson, Andersson, Lindé, Villani & Vredin (2006), Adolfson, Lindé & Villani (2007); 6 Restricted Structural GEMBI? z Is GEMBI model structural? ¾ Immune from Lucas Critique? z Shocks & frictions: “wedgeology” g gy ¾ Examples: habit formation, adjustment costs, CalvoYun price & wage devices, judgment ¾ Shocks “structural” or statistical necessity? Microfounded frictions? Why 1st-order AR process? Why independent source of variations in DSGE? ¾ Wedgeology (Hansen & Sargent 2008): adjustment for risk or uncertainty? 7 Restricted Taking GEMBI to data I z Is GEMBI model data-consistent? ¾ Beyond calibration: estimated GEMBI ¾ Immune from Sims Critique? z Shocks & frictions: add factors ¾ Remedy for empirical failures of baseline DSGE? more interesting dynamics? Better fit? ¾ Sims (2008): “add enough sources of friction and inertia … so that it can fit the complex p behavior we see in data” but this “strains the credibility of the claims to complete behavioral interpretation” ¾ Does the model detect structural changes? 8 Restricted Taking GEMBI to data II z Bayesian estimation of GEMBI ¾ Posterior too close to prior z Usefulness - focus ¾ More on medium-size estimable GEMBI ¾ Less on model forecasts ¾ DSGE advantage: understanding the mechanism ¾ Weak identification & small samples, data quality ¾ Great uncertainties with model, parameter estimates & data: other models & data analysis 9 Restricted Taking GEMBI to reality I z A changing world? ¾ Large, persistent shocks to commodity prices ¾ Shocks from some advanced industrial economies ¾ Possibility of stagflation & frequent crises, 1970s style z LPHI events & uncertainty: avoid large errors ¾ Uncertainty typical in EMEs (model, data, parameter) ¾ Weak financial-real linkages in models ¾ Pervasive nonlinearity: linearisation insufficient ¾ Robust monetary policy at turning points: direction of monetary policy? 10 Restricted Taking GEMBI to reality II z Assessing GEMBI as a FPAS ¾ Impulse responses (vs. VAR empirics?) ¾ Statistical fit (& forecasting performance) ¾ Assessment of current state of economy ¾ Usefulness for policy decisions z GEMBI and policymaking ¾ IT & clarity in monetary transmission ¾ Shaping expectations ¾ Financial stability as a goal? ¾ Policymaking under uncertainty 11 Restricted Making GEMBI work z Do we trust our model? ¾ Too much is missing in any model (incl GEMBI) ¾ Assuming g stationarity y ((detrending)? g) DSGEs often not trusted in lowest frequencies & misbehave in very high frequencies ¾ Credible & practical model z Use of GEMBI & communications ¾ How to form and substantiate model forecasts? ¾ Is the model too complex for decision makers? ¾ How to communicate model output to public? 12
© Copyright 2026 Paperzz