New York Occasions on HANK, and questions

By the requirements of mainstream media protection of technical economics, Peter Coy’s protection of HANK (Heterogeneous Agent New Keynesian) fashions within the New York Occasions was really fairly good.

1) Consultant brokers and distributions.

Sure, it begins with the standard misunderstanding about “consultant brokers,” that fashions assume we’re all the identical. A few of that is the usual journalist’s response to all financial fashions: we’ve got simplified the assumptions, we’d like extra basic assumptions. They do not perceive that the genius of financial principle lies exactly to find simplified however tractable assumptions that inform the primary story. Progress by no means comes from placing extra components and stirring the pot to see what comes out. (I imply you, third yr graduate college students on the lookout for a thesis subject.)

However on this case many economists are additionally confused on this subject. I have been to fairly just a few HANK seminars during which distinguished teachers waste 10 minutes or so dumping on the “assumption that everybody is similar.”

There’s a stunning outdated theorem, referred to as the “social welfare perform.” (I realized this in graduate college in fall 1979, from Hal Varian’s wonderful textbook.) Individuals can have virtually arbitrarily totally different preferences (utility features), incomes and shocks, corporations can have virtually arbitrarily totally different traits (manufacturing features), but the mixture financial system behaves as if there’s a single consultant client and consultant agency. The equilibrium path of combination consumption, output, funding, employment, and the costs and rates of interest of that equilibrium are the identical as these of an financial system the place everybody and each agency is identical, with a “consultant agent” consumption perform and “consultant agency” manufacturing perform. Furthermore, the consultant agent utility perform and consultant agency manufacturing perform needn’t look something like these of any explicit particular person particular person and agency. If I’ve energy utility and you’ve got quadratic utility, the financial system behaves as if there’s a single client with one thing in between.

Defining the job of macroeconomics to grasp the motion over time of aggregates — how do GDP, consumption, funding, employment, worth stage, rates of interest, inventory costs and so forth. transfer over time, and the way do insurance policies have an effect on these actions — macroeconomics can ignore microeconomics. (We’ll get again to that definition in a second.)

Now uniting macro and micro is vital. Macro estimation being what it’s, it could be awfully good to make use of micro proof. This system kicked off by Kydland and Prescott to “calibrate” macro fashions from micro proof can be very helpful. Kydland and Prescott could have had a little bit of grass-is-greener optimism about simply how a lot exact proof macroeconomists have on companies and other people, however it’s a good suggestion. Including up micro proof to macro is difficult, nevertheless. Right here “aggregation principle,” typically confused with the “social welfare perform” theorem comes up, extra as a nightmare from graduate college. The situations underneath which the consultant agent preferences appear like particular person individuals are rather more restricted.

Like all good theorems, this one rests on assumptions, and the assumptions are false. The essential assumption is full markets, and particularly full threat sharing: There may be an insurance coverage market in which you’ll be compensated for each threat, particularly dropping your job.

A generalized kind nonetheless works, nevertheless. There may be nonetheless a consultant agent, however it cares about distributions. The consultant agent utility perform will depend on combination consumption, combination labor provide however now additionally statistics concerning the distribution of consumption throughout individuals. In asset pricing, the Constantinides-Duffie mannequin is a superb instance: the cross-sectional variance of consumption turns into a vital state variable for the worth of the inventory market, not simply combination consumption.

All financial theorems are false after all, in that the assumptions are usually not actually true. The query is, how false? Standard macroeconomics comes all the way down to an outline of how aggregates evolve over time, based mostly on previous aggregates:

[aggregate income, consumption, employment, inflation… next year ] = perform of [aggregate income, consumption, employment, inflation… this year ] + unforecastable shocks.

That is it. That is what macroeconomics is. Principle, estimation and calibration to determine the perform. If HANK is helpful to macroeconomics, then, it should be that including distributional statistics helps to explain combination dynamics. Actuality should be

[aggregate income, consumption, employment, inflation… next year ] = perform of [combination revenue, consumption, employment, inflation, distribution of consumption, employment, and so forth. … this yr ] + unforecastable shocks.

So here’s a central query I’ve for HANK modelers: Is that true? Do statistics on the distribution throughout individuals of financial variables actually assist us to forecast or perceive combination dynamics? Thus far, my impression is, not a lot. The social welfare perform theorem may be mistaken in its assumptions, but nonetheless a reasonably good approximation. And “heterogeneity” has been round macro for a very long time, however by no means has appeared to matter a lot in the long run. (The funding literature of the early 1990s is a superb instance.) However I’d be comfortable to be proved mistaken. This publish is as a lot a suggestion for HANK modelers as a critique.

One other risk: Perhaps HANK is about aggregation in any case. Can we really use micro proof, and add it up constructively, to be taught what the consultant agent – social welfare perform is? Even earlier than HANK, there have been good examples. For instance, the literature on labor provide: Macro fashions need individuals to work extra in response to briefly greater wages. Most particular person individuals work eight hours a day or zero, so micro proof finds a small response. However a small variety of individuals transfer from non-work to work as wages rise. So the consultant agent can have a a lot bigger elasticity than particular person individuals. And, you must perceive labor market construction, and the distribution of who is on the market to work so as to add up from micro to macro proof. Right here, I want to know the fundamental useful kind — how a lot does the SWF care about at present vs. tomorrow, threat, work vs leisure, in addition to any distributional impact?

2) Earnings results

Coy additionally goes on with the standard New York Occasions schtick about how dumb and irrational all of the little hoi polloi are. (In fact we of the elite and the federal authorities handing out nudges would by no means be behavioral.) However you do not want HANK to imagine that the consultant investor is dumb both. He goes on to explain fairly nicely the place the present literature is.

Behind that is, nevertheless, one of many main options of HANK fashions to date. One among its most vital makes use of has been to place present revenue within the IS equation.

(Economists speak amongst yourselves for a bit whereas I clarify this to common individuals. Thus far, the central description of demand in new Keynesian fashions is predicated on “intertemporal substitution:” When the actual rate of interest is greater, you devour a bit much less at present, save a bit extra, with the intention to devour much more tomorrow. That’s the essential mechanism by which greater actual rates of interest (say, induced by the Fed) decrease demand at present. Outdated Keynesian fashions did not have individuals in them in any respect, however hypothesized that consumption merely follows revenue. That provides a extra highly effective mechanism, the “multiplier:” an preliminary revenue drop lowers consumption, which lowers revenue and round we go. )

HANK fashions typically add some “hand to mouth” shoppers. Some individuals take into consideration at present vs. the long run, however others simply eat what revenue they make at present. You may get this out of “rational, liquidity constrained” individuals, however that is usually not sufficient. To get important results, you want individuals who simply behave that means. So, there’s this little little bit of behaviorism in lots of HANK fashions. However it’s a little bit spice within the in any other case Lucas soup.

In equations, the usual mannequin says

consumption at present = anticipated consumption tomorrow – (quantity) x actual rate of interest

After an immense quantity of algebra and laptop time, HANK fashions permit you to write

consumption at present = (quantity) x revenue at present + (quantity) x anticipated consumption tomorrow – (quantity) x actual rate of interest

New Keynesian fashions have been invented on the hope they might transform holy water sprinkled on old-Keynesian considering, for instance justifying massive spending multipliers and powerful financial coverage. They turned out to be nothing on the type when you learn the equations. A motion is underway to change (torture?) new-Keynesian fashions to appear like old-Keynesian fashions, to deliver macro again to roughly the 1976 version of Dornbush and Fisher’s textbook. Advanced expectation formation theories and this facet of HANK may be digested that means.

So right here is my second query for HANK modelers: Is that this it? After we boil all of it all the way down to the linearized equations of the mannequin you are taking to information, to elucidate aggregates and financial and financial coverage, is there a giant backside line past an excuse to revive bits of the Keynesian consumption perform? That too is an sincere query, and maybe a suggestion–show us the textbook again of the envelope backside line mannequin. (It could be awfully good if distributions mattered right here too, theoretically, empirically, and quantitatively.)

3) Micro implications of macro

Perhaps you disagreed just a few paragraphs in the past with my definition of macroeconomics, as solely involved with the motion of aggregates over time. Speaking with a few of my HANK colleagues, a distinct goal is at work — determining the results of macroeconomics on totally different individuals. Recessions fall more durable on those that lose jobs, and sure revenue and different teams; more durable on some industries and areas than others. Right here HANK dovetails with issues over revenue range and “fairness.”

That is a superbly good motive to check it, however let’s then be clear. If that is the case, HANK actually does not change our understanding of how insurance policies and occasions transfer aggregates round, it’s actually nearly understanding how these aggregates have an effect on totally different individuals otherwise.

Which will change calculations of optimum financial coverage. If the target perform cares negatively about revenue range, then including HANK could produce a mannequin that makes no distinction in any respect for the impact of financial coverage on aggregates, however provides a higher weight to employment vs. inflation. Many fashions have observationally equal predictions for aggregates however totally different welfare implications, and the identical mannequin can have totally different welfare implications should you put in several preferences for distributions throughout individuals. However certainly HANK has extra to supply than a long-winded excuse for dovishness in direction of tolerating inflation rather than unemployment.

Additionally, within the massive image this looks as if a traditional reply looking for a query. Should you care concerning the much less lucky, you begin with the massive points: crime, terrible colleges, household breakdown, alternative. The extra profit for the much less lucky from the extent of the in a single day federal funds fee could be enjoyable to isolate in a mannequin, however we’re actually gazing a caterpillar on a leaf of a tree and lacking the forest of financial misfortune.

4) Final ideas

I hesitate to put in writing, as I’m a client not a producer of HANK analysis, and thus will in all probability get issues mistaken or present my restricted information of the literature. Please fill the feedback with corrections, amplifications, tips that could good papers, and so forth.

There’s a tendency in economics to pursue a brand new technical risk with out actually realizing the place it is going or why. That is not unhealthy; determine what you are able to do first, and what to do later. The why at all times does come later. This was true of rational expectations, actual enterprise cycles, new-Keynesian fashions and extra. Now that HANK is fairly nicely developed and is popping out in public, with admiring New York Occasions articles, it’s value assessing the why, the underside line, what it does.

I am additionally hesitant to put in writing and particularly too critically. I vividly recall being in grad college, and a few speaker (I mercifully forgot who) went on a tirade about all these younger whippersnappers utilizing an excessive amount of math and never sufficient instinct and simply being in love with constructing fashions. I vowed if I ever thought that I’d retire. What do we are saying to the angel of outdated age? Not at present. Convey it on, and let’s all determine what it means.

M	T	W	T	F	S	S
					1	2
3	4	5	6	7	8	9
10	11	12	13	14	15	16
17	18	19	20	21	22	23
24	25	26	27	28	29	30
31

Leave a Reply Cancel reply

Related News

You may have missed