Ameriški Fed je naredil še en velik korak k transparentnosti – po objavah zapisnikov sej OMC (Open Market Committee) je objavil tudi kateri makroekonomski model uporablja pri svojih simulacijah ukrepov monetarne politike, ki so nato podlaga za odločitve o konkretnih ukrepih. To je velik korak naprej v primerjavi z vsemi ostalimi centralnimi bankami.

Za mnoge pa je največje presenečenje pri objavi modela razkritje, da Fed ne uporablja DSGE modela, pač pa je njegov FRB/US model neke vrste Structural Econometric Model (SEM), ki je v bistvu ad hoc eklektična mešanica teoretskih predpostavk in ekonometričnih ocen. Nekaj, kar je za neoklasične fanatike bodisi absolutno zastarelo bodisi absolutno zgrešeno, ker ni – saj veste – mikro fundirano. Toda, kot komentirajo manj fanatični neoklasiki, je prav to največja prednost Fed modela – fleksibilnost in sposobnost prilagajanja dejanskemu dogajanju, ne pa dogmatsko sledenje že apriori napačnim predpostavkam, vgrajenim v DSGE modele.

Seveda je Fedov FRB/US model še vedno neoklasičen, uporablja produkcijsko funkcijo in optimizacijsko obnašanje gospodinjstev in podjetij. Toda pri tem je bistveno bolj življenjski in odstopa od dogmatičnosti DSGE modela. Prvič, Fedov model ne predpostavlja reprezentativnega gospodinjstva, pač pa upošteva različne strukturne tipe gospodinjstev in s tem njihove različne odzive na šoke. Denimo upokojenci ali osebe na socialnih transferjih bodo transferje porabili na drugačen način (hitreje) kot pa gospodinjstva z višjimi dohodki itd. Drugič, Fedov model odstopa od tega, da bi gospodinjstva in različni tipi ekonomskih subjektov uporabljali enako obrestno mero pri diskontiranju bodočih dohodkov, pač pa omogoča, da gospodinjstva uporabljajo bistveno višji diskontni faktor, kar odraža njihovo krajši časovni horizont (okrog 5 let) in večjo nagnjenost k porabi.

Tretjič, Fedov model pri številnih povezavah v modelu ne uporablja simulacij, pač pa – denimo pričakovanja – modelira z VAR (Vector Auto-Regression) modeli, kar omogoča, da preference posamičnih tipov subjektov ocenjujejo na podlagi preteklih izkušenj in ne na podlagi popolnoma informiranih racionalnih pričakovanj. Četrtič, v prejšnjih točkah navedeno implicira, da Fedov model odstopa od predpostavke Ricardijanske ekvivalence, ki je ena izmed predpostavk v DSGE modelu, ki eksplicitno predpostavlja, da dodatni transferji ali znižanja davkov nimajo bistvenega vpliva na povečanje porabe, čeprav življenjska dejstva kažejo, da je mejna nagnjenost k porabi posameznih tipov gospodinjstev zelo velika. Kar pomeni, da imajo dodatni socialni transferji lahko precejšen takojšen učinek na povečanje porabe in s tem na okrevanje gospodarstva. 

Petič, Fedov model se izogiba linearnostim v DSGE modelu in omogoča nelinearne povezave med endogenimi spremenljivkami v modelu. Tako denimo omogoča določanje kratkoročnih obrestnih mer v času ničelnih obrestnih mer (Zero Lower Bound) kot feedback zanke (gre za situacijo, ki se v neoklasični teoriji niti teoretično ne more zgoditi in je seveda v DSGE modelu ni mogoče modelirati).

Skratka, Fedov FRB/US model je skrajno eklektičen in na Fedu ga sproti prilagajajo konkretni gospodarski situaciji, kar mu omogoča precejšnjo fleksibilnost in življenjskost.

Nekaj osnovnih razlik med Fedovim FRB/US modelom in DSGE modelom, kot jih vidijo na Fedu:

• Because FRB/US is not built around a representative household paradigm, it is more generously parameterized than typical DSGE models and dispenses with many of the cross-equation restrictions imposed by the latter. Notably, future income is valued by different discount factors depending on whether it accrues to households or firms. Also, the marginal propensity of households to consume out of different types of income can vary, depending on which group of households receives the income. For example, transfer income is disproportionately received by retirees who are well-advanced in their lifecycles.

• Some optimization problems are specified in a different fashion in FRB/US than in many DSGE models. As noted earlier, the FRB/US specification of consumer spending bases the valuation of a large component of human wealth on a discount rate that is both fixed and quite large, implying that the effective planning horizon for many households in FRB/US is closer to the five years advocated by Friedman (1957) than to the much longer period embedded in a typical DSGE model (Carroll, 2001). In addition, the growth of consumer spending in FRB/US is not closely linked to the path of expected future short-term (risk-free) interest rates as it is in the Euler equation specification of consumption used in most DSGE models; rather, the level of spending in the model depends directly on intermediate-term consumer loan rates and indirectly on the long-term bond rates that influence the value of corporate equities.

• Another important dimension along which FRB/US is different from many DSGE models used in policy analysis is that the model allows for nonlinear interactions among endogenous variables, in contrast to the common practice of writing models as linear approximations around a steady state or balanced-growth path. For example, the model’s estimate of the average interest elasticity of aggregate demand has changed markedly over time as the composition of GDP has evolved; in particular, the aggregate elasticity fell sharply with the recent collapse of residential construction, because it is the most interest-sensitive sector of the economy. Another important nonlinearity concerns the zero lower bound on nominal interest rates, which has constrained the actual and expected future stance of monetary policy markedly since late 2008. It is straightforward in FRB/US to model the short-term policy rate as a feedback rule subject to the zero lower bound.⁷ 

• Broadly speaking, the eclectic approach to the specification of FRB/US permits the historical patterns in macroeconomic data to influence its structure more substantially than is the case for the typical DSGE model, whose structure is more tightly imposed by economic theory. Recognizing that this and other issues about the best design of a macroeconomic model are the subjects of ongoing debate, the staff at the Federal Reserve Board has also developed and uses the EDO and SIGMA DSGE models.⁸ 

Še dva odziva na Fedovo razkritje modela. Noah Smith špekulira, da si je Fed šele zdaj upal razkriti model, ker ga je bilo prej sram tega, saj so se mu top ugledni akademiki posmehovali. Toda glede na to, da so DSGE modeli klavrno pogrnili kot kakorkoli uporabna orodja za modeliranje ekonomske politike v času ekstremnih gospodarskih šokov, si je Fed končno drznil objaviti svoj model – ki grobo krši predpostavke, ki so jih zapovedali top ugledni neoklasični akademiki – ki pa deluje. Oziroma deluje bistveno bolje od interno konsistentnih mikro fundiranih DSGE modelov. Smith:

Why didn’t the Fed fully reveal FRB/US model before now? It always seemed to me that it was basically because of – for lack of a better word – embarrassment. Academic macroeconomists haven’t used or studied this type of model in decades (having abandoned everything else in favor of DSGE). In 2010, Chris Sims appeared to call models like FRB/US “something close to a spreadsheet“. Since most Fed employees are drawn from the same pool of people as academic macro (and interact with academic macroeconomists quite frequently), the fact that they use something like FRB/US must have seemed a bit awkward. In fact, I’ve heard academic macroeconomists make fun of FRB/US a number of times.

So if my guess is right, the Fed’s publication of FRB/US indicates that whatever embarrassment existed is now essentially gone. That is kind of interesting.

After all, FRB/US flies in the face of two key developments in academic macro. Since FRB/US has a huge number of parameters, all of which are assumed to be structural, it is a lot harder for this model to pass the intuitive test known as the “Lucas Critique“. Basically, more “structural” parameters = more assumptions = more chance to get some of the assumptions wrong = easier for any given economist to wrinkle his nose at the model.

Second, FRB/US does not force its users to use Rational Expectations (which the Fed more aptly calls “Model Consistent Expectations” or “MCE”). The model has an option that allows you to use it with MCE. But it also has an option to allow you to use it with non-model-consistent expectations. That flies in the face of what Robert Lucas told economists to do, and what most academic macroeconomists do in fact do.

Simon Wren-Lewis, ki sicer tudi sam uporablja DSGE modele, vendar kaže zdravo dozo akademske skepse do njih:

An alternative would be for the economist to build a DSGE model, and simulate that. This has a number of advantages over the reduced form estimation approach. The nature of the experiment can be precisely controlled: the fact that the tax cut is temporary, how it is financed, what monetary policy is doing etc. But any answer is only going to be as good as the model used to obtain it. A prerequisite for a DSGE model is that all relationships have to be microfounded in an internally consistent way, and there should be nothing ad hoc in the model. In practice that can preclude including things that we suspect are important, but that we do not know exactly how to model in a microfounded manner. We model what we can microfound, not what we can see.

A specific example that is likely to be critical to the impact of a temporary income tax cut is how the consumption function treats income discounting. If future income is discounted at the rate of interest, we get Ricardian Equivalence. Yet this same theory tells us that the marginal propensity to consume (mpc) out of windfall gains in income is very small, and yet there is a great deal of evidence to suggest the mpc lies somewhere around a third or more. (Here is a post discussing one study from today’s Mark Thoma links.) DSGE models can try and capture this by assuming a proportion of ‘income constrained’ consumers, but is that all that is going on? Another explanation is that unconstrained consumers discount future labour income at a much greater rate than the rate of interest. This could be because of income uncertainty and precautionary savings, but these are difficult to microfound, so DSGE models typically ignore this.

The Fed model does not. To quote: “future labor and transfer income is discounted at a rate substantially higher than the discount rate on future income from non-human wealth, reflecting uninsurable individual income risk.” My own SEM that I built 20+ years ago, Compact, did something similar. My colleague, John Muellbauer, has persistently pursued estimating consumption functions that use an eclectic mix of data and theory, and as a result has been incorporating the impact of financial frictions in his work long before it became fashionable.

So I suspect the Fed uses a SEM rather than a DSGE model not because they are old fashioned and out of date, but because they find it more useful. (Actually this is a little more than a suspicion.) Now that does not mean that academics should be using models of this type, but it should at least give pause to those academics who continue to suggest that SEMs are a thing of the past.