%%%%%%%%%%%%%%% description of key files %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% abbreviations: Ps = standard prior; Pql = quasi-likelihood prior, st st = steady state, theta_endo, theta_exo, theta_stst = groupings of DSGE model parameters, see paper IRFs: impulse responses marg lik = marginal likelihood %%%%%%%%%%%%%%% SETTINGS and DATA ---------------- varspec.m contains the main settings, such as: mspec: model mspec_ref: either [] for Ps, or NaN for Pql nlagsc: # VAR lags in quasi-lik (Pql only) nobsdrat: T* in quasi-lik (Pql only), expressed as ratio of actual sample mprior: prior choice stime = determines actual sample used in estimation nblocks, nsim: # blocks for MCMC, # draws for each block most other settings are self-explanatory ---------------- pathspec.m specifies path for reading data, saving stuff, etc ---------------- startdsswfed.m links to other directories containing subrutines ---------------- loaddata.m loads data (data6.txt) and contains a description of the series %%%%%%%%%%%%%%% MAIN FILES: PRIOR ---------------- mhpriostst.m generates prior for st st parameters (theta_stst) -- using (very fast) MCMC. you need to run this first for both Ps and Pql ---------------- prio.m generates prior for theta_endo in the standard way (drawing from independent marginals). also very fast. you need to run this next for both Ps and Pql ---------------- mhprio.m generates prior for theta_exo using quasi-lik -- using MCMC. takes some time you need to run this first for Pql only. ---------------- priosplice.m splices draws for theta_endo, theta_exo, theta_stst together, so we have one vector of DSGE parameters. in doing so rules out indeterminacy draws need to run this before computing wIRFs, marg lik ---------------- margdenprio.m computes marg lik for prior (eg, integrating constant), need to run this before computing marg lik for DSGE model ---------------- priomom.m computes and tabulates prior moments for DSGE parameters %%%%%%%%%%%%%%% MAIN FILES: POSTERIOR ---------------- mhvarsim.m maximizes posterior and runs MCMC. ---------------- margdensim.m computes marg lik for DSGE model ---------------- mom.m computes and tabulates posterior moments for DSGE parameters %%%%%%%%%%%%%%% IRFs and MOMENTS ---------------- Wprioirfs.m computes IRFs associated with either prior or posterior draws for DSGE parameters (in spite of 'prio' in name, does both prior and posterior draws) ---------------- priostats.m computes moments (variances, autocorrelations, cross-correlations) associated with either prior or posterior draws for DSGE parameters (in spite of 'prio' in name, does both prior and posterior draws) %%%%%%%%%%%%%%% DSGE ESTIMATION ---------------- dsgesolv.m solves DSGE model using gensys.m (See Sims 2002) ---------------- measur.m measurement equation ---------------- dsgelh.m measurement equation DSGE model likelihood, using kalman filter