Learning-from-Prices: A Research Agenda

Most of my research builds on the idea that economic agents interpret fluctuations in market prices as signals of underlying changes in the economy that they cannot directly observe. Hence, price changes may impact agents more on their expectations than on their budget sets.

In my papers, I have explored different mechanisms:

• in Gaballo (AEJ:Macro, 2016) households learn from current prices about others' noisy foresight;

• in Gaballo (ReStud, 2018) firms learn about overall inflation from local input prices;

• in Ehrmann, Gaballo, Hoffmann and Strasser (JME, 2019) learning from prices lets public information increase uncertainty;

• in Gaballo and Marimon (JME, 2021) ABS investors learn about underlying risk from equilibrium market outcomes;

• in Chahrour and Gaballo (ReStud, 2021)  consumers learn about the prospects on economic activity from changes in the real price of housing.

My ERC project LearnInCycle naturally lies in my research agenda in that I explore the implications of consumers' learning from realised prices about overall current and future inflation. An overview of it follows below.

 LearnInCycle   in a nutshell

The Eurozone is currently facing a sudden and sharp increase in inflation. Whether inflationary pressures will persist and what will be the cost of an eventual monetary tightening crucially depends on inflation expectations. 

The aim of LearnInCycle is to shed new light on how market prices influence consumers’ perceived inflation and how this may allow for the transmission of monetary policy even in the absence of rigidities in firms’ pricing.  

It shifts the focus from a traditional firm-centric to a (complementary) household-centric view of money non-neutrality. It shares the spirit of a recent trend in central banks’ efforts in getting a better appraisal of the importance of consumers’ expectations, after decades of a strict empirical focus on firms’ pricing.

The key novelty of the project is studying, not only how expectations move market outcomes, as typical in the literature, but also how these feed-back to expectations as agents observe realized prices, in a full circle of general equilibrium implications. 

LearnInCycle  consists of five working packages – two theoretical and three empirical - sharing the idea that business cycles fluctuations maybe led by waves of optimism and pessimism in consumers’ inflation perceptions driven by actual market prices. 

WP1. The first package proposes a new theory of business fluctuations where, even if the distribution of posted markups is fixed at the firm level, a counter-cyclical average paid markup may originate from aggregate movements in spending re-allocation: this is because households react to local prices hikes by hunting for lower prices (for ongoing preparatory work see Gaballo and Paciello (2021)). 

WP2. The second package aims at an empirical validation of the main testable implication of the outlined model, i.e. that there exists a counter-cyclical behavior of the gap between posted and paid inflation due to consumers’ price hunting in reaction to higher perceived inflation. 

WP3. The objective of the third package designs an empirical test to tease apart different types of shopping behavior in the data, and to look for direct empirical support to the learning-from-prices mechanism. 

WP4. The fourth WP provides a new dynamic model of inflation where consumers learn from current inflation about future economic prospects on which all have dispersed foresight. Differently from WP1, here households’ learning from prices influences consumption-saving choices, rather than spending allocation. 

WP5. The fifth package will estimate the model developed in WP4 by using available survey data on individual and market expectations. 

Contributions. Overall, the learning-from-prices view of inflation and business cycles brings three ground-breaking contributions:

LearnInCycle may reverse the typical firm-centered view on the source of output-inflation comovement, showing that households’ inflation expectations may drive realistic business fluctuations even with frictionless firms’ pricing. This channel complements the one emphasized by workhorse monetary models models that rely on firms’ expectations and nominal rigidity, through the so-called New-Keynesian Phillips curve.

Assuming frictions on the side of firms has until now been the state-of-the-art in monetary policy analysis. This is intuitive as wages must be rising quicker than firms’ posted prices for households to increase demand and labor supply in response to an inflationary shock. LearnInCycle completely reverses the focus of the traditional literature, giving to households’ expectations a central role in the propagation of monetary shocks without requiring any rigidity on the firms’ side.

LearnInCycle  allows for full general equilibrium microfoundations of incomplete information in macroeconomic models. By looking at market prices as imperfect signals, LearnInCycle sheds light on the impact of market outcomes on expectations, complementing the partial equilibrium view typically adopted in the literature looking at the impact of expectations on market outcomes.

Typical in the literature, the existence of noisy signals about fundamentals is assumed rather than derived; in particular, information frictions are modeled as completely disconnected from market outcomes. This is because most of attention has been devoted to the study of the effect of incomplete information on market outcomes, without looking at the inverse feedback. A notable exception is the literature on Rational Inattention which, however, has modeled the existence of a "market" for information rather than microfounding information as by-product of good markets outcomes.   

LearnInCycle  provides empirical counterparts to incomplete information sets by assuming that actual market prices are the main signals used in expectation formation. This will greatly enlarge the scope for empirical investigation of informational frictions models and their practical relevance for policy making.

The absence of market micro-foundations has largely limited the impact of the information frictions literature. The effort of finding external validation has been frustrated by the problem of finding an empirical counterpart to what is tagged “signals” and “noise” in these models. What is informational noise? How to measure it? Whereas it is possible in principle to measure a decrease of productivity or quantify a financial friction independently from any particular model, the identification of informational noise seems inherently elusive.