6 MIT LIBRARIES tf'^5| DEWEyij lllllllllllllllll */V>^ f^ 3 9080 03317 5958 Massachusetts Institute of Technology Department of Economics Working Paper Series Noisy Business Cycles George-Marios Angeletos Jennifer La'O Working Paper 09- 1 May 14, 2009 RoomE52-251 50 Memorial Drive Cambridge, MA 02142 This paper can be downloaded without charge from the Social Science Research Network Paper Collection http:/7ssrn.com/'abstract= 4 1 ! 4302 at Noisy Business Cycles* George-Marios Angeletos MIT and Jennifer La'O NBER May MIT 14, 2009 Abstract This paper investigates a real-business-cycle economy that features dispersed information about the underlying aggregate productivity shocks, taste shocks, and monopoly power. We show how the dispersion inertia in the response of of information can macroeconomic outcomes run response of employment to productivity shocks; to such shocks; (iii) only a small fraction of high-frequency fluctuations; business cycle; (v) formalize a certain type of (vi) generate cyclical variation in (i) contribute to significant (ii) induce a negative short- imply that productivity shocks explain (iv) contribute to significant noise in the demand shocks within an Finally, economy; and about the underlying fundamentals: they on the heterogeneity of information and the strength of trade linkages the level of uncertainty. RBC observed Solow residuals and labor wedges. Importantly, none of these properties requires significant uncertainty rest — potentially — shocks to in none of these properties are symptoms of the economy, not inefficiency: apart from undoing monopoly distortions or providing the agents with more information, no policy intervention can improve JEL upon the equilibrium allocations. codes: C7, D6, D8, Keywords: Business cycles, fluctuations, heterogeneous information, informational frictions, noise, strategic complementarity, higher-order beliefs. 'This paper was prepared for the 2009 NBER Macroeconomics Annual. We to our discussants, Christian Hellwig and Robert King, and the organizers, Michael Woodford. Emmanuel We also received useful Farhi, Mikhail Golosov, are grateful for their detailed feedback Daron Acemoglu, Kenneth Rogoff, and comments from Philippe Bacchetta, Ricardo Caballero, V,V. Chari, Guido Lorenzoni, Ellen McGrattan, Patrick Kehoe, Ricardo Lagos, Stephen Morris, Alessandro Pavan, Ricardo Reis, Robert Shimer, Robert Townsend, and seminar participants at University of Wisconsin at Madison, the Federal Reser-\'e Bank, the 2009 conference in honor of Truman Bewley Bank at the University of Texas, Austin, Macroeconomics Annual. Email: angeletOmit edu, jenlao9mit edu. . MIT, the of Minneapolis, the Federal Reserve Board, the . and the 2009 World NBER Digitized by the Internet Archive in 2011 with funding from Boston Library Consortium IVIember Libraries http://www.archive.org/details/noisybusinesscycOOange Introduction 1 There King (1982), and (1976), policy. a long tradition in macroeconomics, going back to Phelps (1970), Lucas (1972, 1975), Barro is others, to use information frictions to break the neutrality of This literature has recently been revived by Mackowiak and Wiederholt (2003a, 2008), Mankiw and and (2008), others. formalizations of the origins of informational frictions, most of theme monetary of breaking but frictions, shift focus to neutrality. In this paper, we Reis (2002), Sims (2003), While it this Woodford work has proposed new has remained focused on the old are also concerned with informational we study how informational a very different theme: monetary frictions impact the response of the economy to aggregate productivity shocks, and other real shocks, within the context of a micro-founded real-business-cycle model. This shift is motivated by two considerations. some type require significant lack of information, or policy is debatable. Indeed, the first the criticism that such information is First, the empirical relevance of theories that of unawareness, about the current monetary generation of the aforementioned literature succumbed to widely, readily, and cheaply available.^ that the dispersion of information about the real shocks hitting the than the one about the conduct of monetary policy. economy economic fundamentals matters not only how this heterogeneity for more severe far is far factors such as the value of certain assets, And the health of the financial system, or the broader economic fundamentals. how is In the ongoing crisis, for example, there more uncertainty, and disagreement, about non-monetary literature has little to say about Second, we contend yet, the pertinent the heterogeneity of information about the real underlying macroeconomic outcomes. FinaUy, we would may impact like to understand the positive properties of the business cycle but also normative properties. its Motivated by these considerations, RBC canonical first the paper introduces dispersed information model, where nominal prices are show that the dispersion RBC this paradigm — indeed in flexible ways that might that New-Keynesians have raised against the 'The new generation has attempted to escape or to process available, it may still plausible, it irrelevant. We impl}' that technology shocks explain only a small RBC this criticism same time helping overcome paradigm, We certain criticisms next show that this significant by postulating that, even if such information is readily be hard to update one's information sufficiently frequently (Mankiw and Reis, 2002) and absorb such information While these ideas are an otherwise of information can significantly alter certain positive properties of fraction of high-frequency business cycles, while at the and cheaply and monetary factors are in sufficiently well (Woodford, 2003a, 2008; Mackowiak and Wiederholt, 2008). seems hard either to gauge their quantitative importance or to reconcile them with the fact that financial markets respond nearly instantaneously to any news about monetary policy, or that a variety of economic agents appear to pay close attention to monetary policy. change in the positive properties of the RBC paradigm happens without affecting one important normative lesson: as long as there are no monopoly distortions, the equihbrium allocations coincide with the solution to a certain planning problem, leaving no room These results should not Our primary paradigm. be interpreted narrowly only as an attack against the New-Keynesian goal is to provide a clean theoretical models. And yet, our framework Our framework and results we analysts ahke. In this regard, enough rich is may benchmark for the positive and Abstracting from nominal frictions best serves normative implications of dispersed information. this purpose. for stabilization policies. to nest the real backbone RBC thus prove equally useful for makes not only a believe that our paper New-Keynesian of and New-Keynesian specific contribution into business-cycle theory but also a broader methodological contribution. Preview of model. The backbone of our from capital to simplify the analysis, but allow model for is a canonical RBC crucial role for aggregate fluctuations whether each of the goods is Keynesian models, facilitating in a it introduces a highlighted in Angeletos and La'O (2009b), play a when, and only when, information produced combined with monopoly power, first abstract a continuum of differentiated commodities. This multi-good (or multi-sector) specification serves two purposes. First and foremost, certain type of trade interactions that, as We economy. is dispersed; this competitive or monopolistic fashion. this specification Second, is true when permits us to nest the real backbone of New- a translation of our results to such models.^ Accordingly, while the core of our analysis focuses on shocks to technology (TFP), in principle we also allow for two other types of shocks to the fundamentals of the economy: taste shocks (shocks to the disutility of labor), and mark-up shocks (shocks to the elasticity of demand). However, none of our results rests on the presence of either monopoly power or these additional shocks. The only friction featured in our model is that certain economic decisions have to be made under heterogeneous information about the aggregate shocks hitting the economy. The challenge to incorporate this informational friction without an or the tractability of the analysis. Towards undue this goal, sacrifice in either the we formalize is micro-foundations this friction with a certain geographical segmentation, following similar lines as Lucas (1972), Barro (1976), Townsend (1983), and Angeletos and La'O (2008, 2009b). In particular, we assume that each period firms and workers meet in different "islands "and have to make their employment and production decisions while facing uncertainty about the shocks hitting other islands. At the same time, we assume that consumption choices take place in a centralized market, where information is homogenous, and that households are "big families", with fully diversified sources of income. This guarantees that our "Indeed, all the results we document in this economy admits paper directly extend to a New-Keynesian variant as long as monetary policy replicates flexible-price allocations, which in certain cases is the optimal thing to do (Angeletos and Lao, 2008), a representative consumer and maintains high tractabihty in analysis despite the fact that some key economic decisions take place under heterogeneous information. Positive results. As mentioned, the core on the special case where of our analysis focuses firms are competitive and the onlj' shocks hitting the fundamentals of the (TFP) shocks which makes the In standard (i) respond fast RBC models significant even in We show Some macroeconomic outcomes. the agents face if little that the dispersion of information induces Perhaps paradoxically, employment responds negatively the data; have pointed out that that this fact fact to paradigm. this inertia is Although to productivity shocks inconsistent with standard argue in favor of New-Keynesian models Kimball, 2006; Gali and Rabanal, 2004). this fact (e.g., RBC (iii) Many In the within the it RBC RBC remains debatable (e.g., We paradigm, technology shocks account is for the What of information can induce technology shocks to explain its is, monetary nature: drives the residual variation noise in available information, that of information bulk of short-run fluctuations. only a small fraction of the high-frequency variation in the business cycle. (iv) Christiano, empirically implausible and have favored New-Keynesian show that the dispersion business cycle remains neoclassical in and paradigm. economists have argued that this alternatives. models; and Gali, 1999; Basu, Fernald Eichenbaum and Vigfusson, 2003; McGrattan, 2004), we show that the dispersion can accommodate can be uncertainty about the underlying shocks. researchers have argued that have used this RBC Hansen, 1985; Prescott, 1986), macroeconomic outcomes (e.g., and strongly to technology shocks. inertia in the response of (ii) analysis directly comparable to the economy are technology m And yet, the entire factors play no role whatsoever. short-run fluctuations in our model is simply the correlated errors in the agents' expectations of the underlying technology shocks. Most interestingly, we show that the fraction of short-run volatility that to such noise can be arbitrarily high even is due the agents are nearly perfectly informed about the if underlying technology shocks. (v) These noise-driven fluctuations help formalize a certain type of RBC in setting. The associated errors in forecasting economic activity can be interpreted as variation expectations of "aggregate demand". while decreasing "demand shocks "within an its They correlation with output. help increase the relative volatility of emplo\'ment An identification strategy as in Blanchard and Quah (1989) or Gall (1999) would likely identify these shocks as ''demand "shocks. (vi) These noise-driven fluctuations involve countercyclical variation and procyclical variation in in Solow residuals, consistent with what observed these cyclical variations can be significant even the underlying technology shocks. if measured labor wedges, in the data. Once again, the agents are nearly perfectly informed about While we stop short of quantifying these heterogeneity of information has a very different certainty about fundamentals we hope that they results, is the mark on macroeconomics outcomes than the un- — a point that we further elaborate on Indeed, what drives our results how at least highlight in Angeletos and La'O (2009b). not per se the level of uncertainty about the underlying technology or other shocks, but rather the lack of common knowledge about them: our effects are consistent with an arbitrarily small level of uncertainty about the underlying fundamentals. At the same time, the lack of common knowledge does not alone explain the magnitude of our Rather, this depends crucially on the strength of trade linkages effects. our economy. This idea is linkages in our economy, among formahzed by our game-theoretic representation. namely the the firms and workers A measure elasticity of substitution across different goods, one to the degree of strategic complementarity in the game of the trade maps one-to- that represents our economy. One can then extrapolate from earlier more abstract work on games of strategic complementarity (Morris and Shin, 2002, Angeletos and Pavan, 2007a) that the strength may of trade linkages in our economy We play a crucial role in determining the equilibrium effects of heterogeneous information. conclude that our findings hinge on the combination of heterogeneous information with strong trade linkages We — but they do not hinge on the finally seek to level of uncertainty about the underlying fundamentals. understand the normative content of the aforementioned findings. that a planner could improve welfare by aggregating the information that or otherwise providing the agents with more information. But is We know dispersed in the economy, this provides no guidence on whether the government should stabilize the fluctuations that originate in noise, or otherwise interfere with the way the economy responds to available information. To address this issue, one has to ask whether a planner can improve upon the equilibrium allocations without changing the inform.ation structure. We show that the answer to this question is essentially negative. case of our model where firms are competitive, there raise welfare without changing the information that is is In particular, in the special indeed no way in which the planner can available to the economy. general case where firms have monopoly power, the best the planner can do monopoly We distortion, much alike what he is conclude that, insofar the information is supposed to do when information is As for the more merely to undo the is commonly shared. taken as exogenous, the key normative lessons of the pertinent business-cycle theory survive the introduction of dispersed information, no matter how severely the positive lessons might be affected. Layout. The remainder of the introduction discusses the related literature. Section 2 introduces the model. Section 3 characterizes the general equilibrium. Sections 4 and 5 explore the implications for business cycles. Section 6 studies efficiency. Section 7 concludes. Related literature. The macroeconomics tory, a revived present, in and literature — hopefully— a promising on informational future.'^ Among frictions has a this literature, most long his- influential our thinking have been Morris and Shin (2002), Woodford (2003a), and Angeletos and Pavan (2007, 2009). Morris and Shin (2002) were the first to highlight the potential implications of asym- metric information, and higher-order beliefs, for settings that feature strategic complementarity. Woodford (2003a) exploited the inertia of higher-order beliefs to generate inertia in the response of prices to nominal shocks in a stylized model of price setting. Finally, Angeletos and Pavan (2007a, 2009) provided a methodology for studying the positive and normative properties of a more general games with class of strategic complementarity The framework we use in this paper is and dispersed information. a variant of the ones we use in two companion papers Angeletos and La'O (2009b) we elaborate on the broader insight that the for different purposes. In heterogeneity of information introduces a distinct t>-pe of uncertainty about economic activity, and show how this can sustain sunspot-like fluctuations in a unique-equilibrium economy. In Angeletos and La'O (2008), on the other hand, we extend the analysis of the present paper by allowing information to get aggregated through certain price and quantity indicators and by mtroducing nominal we then explore the implications frictions; for optimal Part of the combined contribution of these papers is to fiscal and monetary policy. show how the equilibrium and effi- cient allocations of fully micro-founded business-cycle economies can be represented as the Perfect Bayesian equilibrium of a game similar to those considered in the more abstract setting of Morris and Shin (2002) and Angeletos and Pavan (2007a, 2009). This representation cilitates a translation of of business cycles. some of the more abstract At the same time, the insights of this earlier models. Indeed, it is of the welfare implications conjectured Finally, it is of is it fa- work within the context understanding complementarity featured only these micro-foundations that explain either complementarity turns out to be irrelevant when information none useful, as specific micro-foundations are crucial for both the positive and the normative implications of the particular form in business-cycle is commonly shared, or why why it this has by Morris and Shin (2002). worth noting how our approach differs from those of Beaudry and Portier (2004, 2006), Christiano, Motto, and Rostagno (2006), Gilchrist and Leahy (2002), Jaimovich and Rebelo (2008), and Lorenzoni (2008). These papers consider certain types ^See, e.g., Adam (2007), Amador and 2007a, 2007b, 2009), Bacchetta and Weill (2007, 2008), Wincoop of expectations-driven, or noise- Amato and Shin (2005), Coibion and (2006), Angeletos and Pavan (2004, Gorodnichenko (2008), Collard and Delias (2005), Hellwig (2002, 2005), Hellwig and Veldkamp (2008), Hellwig and Venkateswara (2008), Klenow and Willis (2007), Lorenzoni (2008, 2009), Luo (2008), Mackowiak and Wiederholt (2008, 2009), and Shin (2002, 2006), Moscarini (2004), Nimark (2008), Reis (2006, 2008), Mankiw and Rodma Reis (2002, 2006), Morris (2008), Sims (2003, 2006), Van Nieuwerburgh and Veldkamp (2006), Veldkamp (2006), Veldkamp and Woolfers (2007), and Woodford (2003a, 2008). driven, fluctuations. However, these fluctuations originate merely form uncertainty about future technological opportunities, obtain within representative-agent models, and do not rest on the het- erogeneity of information. Interestingly, Kydland and Prescott (1982) had also allowed for certain expectational shocks; but they, too, did not consider heterogeneous information. numerous papers that consider geographical and trading structures similar are model (e.g., Similarly, there to the one in our Lucas and Prescott, 1974; Rios-Rull and Prescott, 1992; Alvarez and Shimer, 2008), but once again rule out heterogeneous information about the aggregate economic fundamentals. To recap, it is this particular type of informational heterogeneity that is the distinctive feature companion papers. This of our approach in either the present paper or the aforementioned also distinguishes our approach from the Mirrless Uterature, which allows for heterogeneous information about idiosyncratic shocks but also rules out heterogeneous information about aggregate shocks. The model 2 There a is a (unit-measure) continuum of households, or "families", each consisting of a consumer and continuum of workers. There is a continuum of "islands", which define the boundaries of local labor markets as well as the "geography "of information: information but asymmetric across islands. Each island in the by symmetric within an = [0. 1]; firms and commodities by {i,j) & I x J: and periods by Each period has two stages. In stage 1, island, inhabited by a continuum of firms, which specialize production of differentiated commodities. Households are indexed hy h £ e / 2 is is i £ {0, H= [0, 1]; islands 1, 2, ...}. each household sends a worker to each of the islands. Local labor markets then open, workers decide how much labor to supply, firms decide labor to demand, and local wages adjust so as to clear the local labor market. At how much this point, workers and firms in each island have perfect information regarding local productivity, but imperfect information regarding the productivities in other islands. After employment and production choices are sunk, workers return home and the economy transits to stage that was previously dispersed becomes publicly known, and are now pre-determined by made during stage 1, At 2. commodity markets open. Quantities the exogenous productivities and the endogenous employment choices but prices adjust so as to clear product markets. Households. The this point, all information utility of household h is given by UiC^t)^ 1 S,,tVinh,,t)di, (=0 ith U{C) = ^ 1 — and 7 V{n) = -— 1 6 -l- e Here, 7 > parametrizes the income elasticity of labor supply,'' elasticity of labor supply, n^j.t 1 of period Sfi,t is i, is who the labor of the worker e > parameterizes the Frisch gets located on island an island-specific shock to the disutility of labor, and Ch,t the commodities that the household purchases and consumes during stage This composite, which also defines the numeraire used CES by the following nested is durmg stage a composite of all 2. wages and commodity for i prices, structure: is given • p-i = Ch.t Ui ^klt d^ where and where j on island by any the quantity household h consumes in period c/„j,( is Here, i. is rj^t islands. Letting the within-island elasticity of trade linkages by the is < differ 77 is commodity produced by firm will a case of special interest that (77 — » the elasticity of substitution across different from the across-islands we and mark-up shocks, thus will be determined much of our analysis complementarity (which will concentrate on for 00) while the strategic complementarity remains RBC 00); this case nests a canonical, competitive letting the within-island elasticity to be finite p permits us to elasticity be determined by the former) from the strength of strategic the limit where monopoly power vanishes non-trivial (p while p i, monopoly power (which and the associated degree latter). In fact, of the a random variable that determines the period-i elasticity of demand faced indi\'idual firm within a given island distinguish the degree of t and random permits us At the same time, economy. to introduce monopoly power facilitating a translation/extension of our results to the New-Keynesian framework. Households own equal shares of is firms in the economy. all The budget constraint of household h thus given by the following; Pij,tChzj.td{j,k) / + < Bh,t+i JlxJ where pij^t is is B/,_( is objective of each household is and informational constraints faced by ""Note that risk aversion is wunht,tdk no the amount of i, is RtBk,t, in period is the period-f t. utility subject to the members. Here, one should think role in i, tt,jj the period-f nominal gross rate of bonds held and intertemporal substitution play no capital. Therefore, Rt firm j on island simply to maximize expected its + Jl the period-f wage on island return on the riskless bond, and insurable and there TTtjjd{iJ)+ commodity produced by the period-i price of the profit of that firm, Wit The / Jjx! of the our setting because all budget worker-members idiosyncratic risk is 7 only controls the sensitivity of labor supply to income for given wage. of each family as solving a different information sets off during stage a function of 1 its (i) team problem: they share the same objective (family when making their labor-supply choices. market. In stage The is the 2, consumer-member how much to consume in to supply labor as (ii) the wage that collects all the income each of the commodities to save (or borrow) in the riskless bond. Asset markets. Asset markets operate information how the information that will be available to them at that stage and that the worker-member has collected and decides and how much Formally, the household sends workers to different islands with bidding instructions on will prevail in their local labor but have utility) commonly sole role of the in stage 2, when along with commodity markets, all shared. This .guarantees that asset prices do not convey any information. bond market in the model then to price the risk-free rate. Moreover, because is our economy admits a representative consumer, allowing households to trade risky assets in stage 2 would not affect any of the Firms. The output where A^^t is results. ' ! ' ' of firm j on island the productivity in island z, i nijj ' , during period is t The maximize valuation of expectation of U'{Ct)'Kij^f its Labor and product markets. operate in stage markets are as 2. its is (0, 1) parameterizes given by • Labor markets operate in stage 1, while product markets Because labor cannot move across islands, the clearing conditions follows: - . for labor . / n.j^tdj = n.hi.tdh Vi / Jh JJ On " / expectation of the representative consumer's is namely, given by firm's realized profit Finally, the objective of the f^rm its profit, is the firm's employment, and 9 G the degree of diminishing returns in production. to -_., ' the other hand, because commodities are traded beyond the geographical boundaries of islands, the clearing conditions for the product markets are as follows: : / , chij,tdh = q^j^t V(i,j) _ Jh ' Fundamentals and information. Each island in our shocks: shocks to the technology used by local firms (TFP economy causing variation in their monopoly power (mark-up shocks). components to these shocks. subject to three types of shocks); shocks to the disutility of labor faced by local workers (taste shocks); and shocks to the elasticity of idiosyncratic is We demand faced by local firms, allow for both aggregate and The aggregate fundamentals of the shocks (.4^;, 5i(,7/,,() standard practice period t. we consider In contrast, signals about ^^ choices. meet m On in stage 1, to assume that assume that when they have 'I'f m is commodity and "if and others is the beginning of imperfect and, most employment and production becomes common known Alternatively, make any in stage 2, when agents special assumptions about the For example, we can impose a Gaussian structure as we could allow some be imperfectly informed, mimicking the idea to 'it in financial markets. available to each island. Morris and Shin (2002). commonly known is The denote this distribution. 4'( their decentralized For our main theoretical results we do not need to information that Let different islands observe only noisy private (local) make to we assume that the other hand, the centralized is by the joint distribution are identified t situations where information about We thus importantly, heterogeneous. in period the cross-section of islands.^ in macroeconomics in economy of the in islands to be perfectly informed Mankiw and Reis (2002) that only a fraction of the agents update their information sets in any given point of time. To some extent, we could even (2003) and interpret the noise in these signals as the product of rational inattention, as in Woodford More (2003a). generally, we do not expect Sims the details of the origins of noise to be crucial for our positive results. We thus start by allowing a rather arbitrary information structure, as in the more abstract work of Angeletos and Pavan (2009). This variable encodes all First, we let ujt denote the "type "of an island during period the information available to an island about the local shocks as well as about the cross-sectional distribution of shocks and information denote the distribution of W( state of the (signals). Finally, we distributions over S^^, We in 'it let t; note that the aggregate state S^ denote the set of possible types for then uses Qt to make independent draws of make their current-period In special cases (as with Assumption mean 1 cvt now Ut includes not only the cross- each island, Sq the set of probability € S.^, Sq one In the beginning of period using the measure V{Qt\^t-\)-' for later on), this distribution f, and Nature each island. In the beginning of period employment and production choices, agents in any given might be conveniently parameterized by the values of the shocks; but in general the aggregate fundamentals are identified by the entire distribution. ^To avoid getting distracted by purely technical However, none of our results hinges on this ' let and P(|-) a probability measure over Sq.^ then formalize the information structure as follows. before they economy. Next, we of the shocks but also the cross-sectional distributions of the information conditional on Qt-i, Nature draws a distribution Qt G t, in the the cross-section of islands. This variable identifies the aggregate economy during period sectional distribution t. issues, if we let Su, and Sn as if they were finite sets. restriction. Note that we have imposed that the aggregate state notation, nothing changes our proofs treat fit follows a Markov process; apart from complicating the the aforementioned probability measure depend on all past aggregate states. island get to see only their own cot', informs them perfectly about their local shocks, in general, this but only imperfectly about the underlying aggregate state Hj. In the end of the period, however, Qt becomes commonly known (ensuring that To '^t also model recap, the key informational friction in our the underlying aggregate state 0(. immaterial Whether they type of effects we analyze for the that the agents of an island learn their we denote with as functions of w^: own becomes commonly known). is that agents face uncertainty about about their own face uncertainty in this local paper. Merely for convenience, then, local shocks in stage We 1. shocks is we assume can thus express the shocks A{uJi) the local productivity shock, with S(ujt) the local taste shock, and with 77(^4) the local mark-up shock. Equilibrium 3 In this section we characterize the equilibrium by providing a game-theoretic representation that turns out to be instrumental for our subsequent analysis. Definition 3.1 Because each family sends workers to every island and receives each family's income is fully diversified during stage 2. profits from every firm This guarantees that our model admits a To simplify the representative consumer and that no trading takes place in the financial market. we thus exposition, of the any symmetry set Bt — and abstract from the of preferences, technologies loss of generality to impose symmetry Finally, because of the absence of capital summarizes all and information within each in the choices of and the Markov aggregate outcomes, do depend on An 1. strategy q S^ x Sq : -^ M.+ , fit. the other hand, the We -^ R+, a wage function c : R^ — » M.+ , is without t. state, ^(-i Recall then only Ut. As a result, depend on 0(_i and all prices in stage 2, Wf, and thus define an equilibrium as follows. w an aggregate employment function and a consumption strategy 1 is commodity equilibrium consists of an employment strategy Definition Q Sq On it on the aggregate restriction the local levels of labor supply, labor demand, wage, and output can : island, workers and firms within each island. that the additional information that becomes available to an island in stage all Furthermore, because financial market. the payoff-relevant public information as of the beginning of period but not the current aggregate state 0(. the economy, in : N S^ x Sq -^ : Sq -^ n IR+, : S^. x a production an aggregate output function R+. a price function p such that the following are true: 10 Sq —* R+ : S^, x Sq ^ R+, The price function (i) normalized so that is = P{Qt,^t-i) for = fp{u,nunt-i?-''dnt{uj) 1 all (fit: ^(.-i)(ii) The quantity c{p,p',Q) whose price is p when the price of output (mcom.e) (Hi) When clears the of that firm are. levels : = and / I The quantities firm in island u>t , Qt wage is n{uJt.D,t-i) 'i'S the aggregate p{u}t,V.t,flt-\): the o-nd q{ijJt.,flt--\), and employment indices r ' and employment with q{ujt.^t-i) m — are, respectively, N{nt,^t-\) = n(w, f](_i)(iQ,(w). / and g(w(,fi(_i) are optimal from the perspective of taking into account that firms the typical other islands are behaving according to the same given by w{ujt- Q,t-i), that prices will be determined in stage 2 so as and that aggregate income The local wage w{u>t,^t-i) Note that condition (i) m I'S will be given by Q{D,t^^t-i)- such that the quantity n{uJt,^t-i) for the The preferences. for our economy rest of the conditions among different islands in stage Let us expand on what we return to labor. is the mean by CES composite then represent a hybrid of a for the in stage 2, once incomplete-hiformation 1. this. When employ and how much to produce during stage will sell their also the optimal labor complete-information exchange economy that obtains production choices are fixed, and a subgame-perfect equilibrium plaj'ed ^s an island of type iOf simply means that the numeraire when we introduced Walrasian equihbrium they and the past aggregate state is fi(-i, the price that o.'^^d respectively. g(w,f2(_i)^dr2,(cj) supply of the typical worker game is p' product markets, that the representative consumer will behave according to consumption to clear all defined is the aggregate output n(u)t,flt~i) strategies, that the local (v) other commodities from the same island the current aggregate state Q(Qi,f](_i) c, all demand for any commodity Q. is A{ii>t)n{iOt,flt-i)^ strategy the representative consumer's optimal market for the product of the typical firm from islanduJt and output (iv) is 1, firms in an island decide how much labor to they face uncertainty about the prices at which product during stage 2 and hence they face uncertainty about the marginal Similarly, uncertainty about the real when workers mcome their in an island decide how household will have much in stage 2 labor to supply, they face and hence face uncertainty about the marginal value of the wealth that they can generate by working more. But then note that firms and workers in each island can anticipate that the prices that clear the and the realized level of real income are, in commodity markets equilibrmm, determined by the level of employment 11 and production in other islands. economy by reducing depend on This suggests that we can solve game, where the incentives of firms and workers to a certain it for the general equilibrium of the and workers their expectations of the choices of firms We other islands. in an island in implement this solution strategy in the following. Remark. To simplify notation, we often use qu as a short-cut a short-cut for for Q{Q.t,Qt-i), Ejt as and so on; E[-|a;(, f](_i], Qt as a short-cut for q{uit,Q.t-i)^ we drop the also, indices h and j, because we know that allocations are identical across households, or across firms within an island. Characterization 3.2 Towards solving demand for the equilibrium, consider first of the representative price of other where Pt — I commodities in consumer the same island is given by c{p,p',Q) function faced by a firm during period Consider now stage is a commodity from island is pj( is t = U'{Ct) and that Ct = (1), we conclude the = the above objective firm's problem sufficient. ^^ ^itn^f is is by a particular follows that the equilibrium It of nominal income for the representative equihbrium, the objective of the firm - is simply Witritt)] ' ' maximizes the following objective: 1 > (1 — ^)6' > (2) (which we assume to be always the case), unique and that the corresponding first-order condition is is both necessary and simply given by equating the expected marginal cost and revenue of this condition, note that island; the ,, U'{Qt)[{p'J~'^'Qt"'<l,r-' -^rtnu ^^'^S as the a strictly concave function of n,, which guarantees that the solution to the This condition To understand Qt- . typical firm on island ut E,t where qu = ,;, El, [U' {Qt) {pitqu Using when is p,t ', is in Qt whose price i The optimal 2. p~^ {p'Y'~^ Q. Equivalently, the inverse demand Given that the marginal value 1. stage in given by the following: by our choice of numeraire.^ In equilibrium. Ct consumption strategy household for how the economy behaves demand for the cj, = I -^ commodity ) Ci is the demand for the busket of commodities produced of a particular firm in that islands 12 is then c,i = ( ^ ) c',,. labor, evaluated under local expectation of the equilibrium pricing kernel: {^^^]Eu Eu[U'{Q^t)]wu= Next, note that, since Pit — Pit- Along with all '9A,K-'). firms within an island set the ^. (1), this gives same price in equilibrium, it (3) must be that ^.r— This simply states that the equilibrium price of the typical commodity of an island relative to the numeraire is equal to the MRS between that commodity and the numeraire. Fmally, note that the optimal labor supply of the typical worker on island MRS between the numeraire and i is given by equating the local wage with the leisure: ^^^ ""''-E^tlU'iQt)] Conditions cation. and (4) (5) give the equilibrium prices and wages as functions of the equilibrium allo- Using these conditions into condition we conclude that the equilibrium (3), allocation is pinned down by the following condition: ..,„,. (-_i).„U,„ This condition has a simple interpretation: island. To U' {Qt) [q^] i; as for the right-hand side, (6) . simply the marginal disutility of an extra unit of is '^"~' is the reciprocal of the local monopolistic mark-up, I. ''is the marginal utility of an extra unit of the typical local commodity, and OAitrt^^ the corresponding marginal product of labor. Note that condition Qit OAun^-^ equates the private cost and benefit of effort in each see this, note that the left-hand side labor in island _ is it (I) m (6) expresses the equilibrium levels of local employment n,( and local output relation to the local shocks production function, q,t = and so on), local expectations of aggregate Aun^f, to eliminate precise notation of Definition A{ijJt), and the 1 we reach the (i.e., replacing n,f q^t in this condition, output Qt- and reverting l-e + e + TB 13 to the more with q(iJt,^t-i), Qt with Q(f]t,ilt_i), Ait with following result. 1\ Using the 9 1+e be a composite of all the local shocks hitting an island of type u> and define the coefficient '--1 p+ The e and aggregate output are equilibrium, levels of local the solution to the following fixed-point problem: log q {ut, nt-i) = (1 - a) fi'^t) Q[Qt,^t-i) +a = log { E Qi^t, f^t-i)^ q{uj,Qt-i) V(tuf,n,_i) (7) V(Q,,a_i). dnt{uj) '' ^-M fit-l iOt, (8) This result establishes that the general equilibrium of our economy reduces to a simple fixed-point relation between local and aggregate output. In so doing, it of our economy, similar to the one established in Angeletos with capital. To see in IR+. Identify' this, consider a game with a game with an a "player "in this output of that island as the "action "of the offers and La'O (2009b) number large a game-theoretic representation for a variant economy of players, each choosing an action economy and island in our interpret the level of corresponding player. Next, identify the "types "of these players with ut, which encodes the local shocks and local information sets in our economy. Finally, let their "best responses "be given by condition (7). equilibrium of this game It is then evident that the Perfect Bayesian identifies the general equilibrium of our Note then that the variable /(wf) conveniently summarizes all economy. the local economic fundamentals, To while the coefficient a identifies the degree of strategic complementarity in our economy. more this :• clearly, consider a log-linear log , g^f = approximation to conditions const logQf -K (1 = - const + a)/j, + / E,, [log Q(] (8): :': -^ ; , ' logqifd?', where const capture second- and higher-order terms. ^ identifies the slope of Q' and (7) It is see (9) (10) then e\'ident that the coefficient a an island's best response to the activity of other islands —which is the standard definition of the degree of strategic complementarity. Finally, note that Proposition While much ture (e.g. 1 holds no matter the information structure. This of the recent literature has focused Mankiw and on Reis, 2002; Sims, 2003; ^In general, these second- and higher-order terms is important. specific formalizations of the information struc- Woodford, 2003a), our may depend on the underlying result .state mdicates that the and the above is only an approximation. However, when the underlying shocks and signals are jointly log-normal with fixed second moments (as imposed by Assumption 1 in the next section), these terms are invariant, the approximation error vanishes, and conditions (9) and (10) are exact. 14 down information structure typically matters only by pinning We activity. would thus the agents' forecasts of economic pay more attention on the theoretical and invite future researchers to empirical properties of these forecasts as opposed to the details of the information structure. and strategic complementarity Ti-ade links 3.3 As evident from Proposition the degree of complementarity, a, 1, elasticity of substitution across the commodities is a monotone function of the what In of different islands, p. follows, the convention that variation in q represents variation in p for given other parameters. a interpret as a measure of the strength motivated by the following observations. household lives Proposition — First, if we but also on e,-), and also These choices are consider a variant of our model where each one island and consumes only the products of that island, then 0; in this sense, it is islands that introduces strategic interdependence (q p, We in and works only holds with a 1 of trade linkages in our economy. we adopt precisely the trade linkages across different ^ Second, while 0). q- depends, not only on these other parameters affect the composite shock / and matter for 9. equilibrium allocations whether islands (agents) are linked or not; in contrast, p affects only q. For these reasons, we henceforth use the notions of strategic complementarity, elasticity of substitution across islands, and strength of trade linkages, as synonymous to one another. However, we also note that strong complementarity in our model does not strictly require low labor supply (9 — > 1), The is small (7 ^0), the Frisch elasticity then the degree of complementarity is is high high (q — > (e ^ 1) no matter what p 0), the wealth effect of and production is nearly linear is. insight that trade introduces a form of strategic complementarity even in neoclassical, perfectly-competively settings is extend well beyond the boundaries of the model we have likely to considered here or the variant in Angeletos and La'O (2009b). under-appreciated m have thought us that as games. And prior it work on business rarely helps, and it when information is To understand what we mean by the symmetric aggregate shock (i.e,, We believe that this insight has been cycles for two reasons. First, the commonly last point, we highlight here is simply irrelevant shared. consider the response of the economy to a a shock that keeps the level of heterogeneity invariant). Formally, denote the cross-sectional average of the composite fundamental that varies the average fundamental, idiosyncratic two welfare theorems can often be misleading, to think of Walrasian settings second, the type of strategic complementarity for the business cycle let ft if p: /(, /,t and consider any shock without varying the cross-sectional distribution of the components of the fundamentals, ^,( 15 = J^t — ft- When all information is commonly shared, aggregate output also is commonh' known ^ logq.t It is {I - equilibrium. Condition (7) then reduces to in + a){ft ^it) + alogQt. (11) then immediate that the entire cross-sectional distribution of logq-u moves one-to-one with ft, which establishes the following. Proposition 2. Then invariant. Suppose that information is commonly shared and the equilibrium levels of aggregate output log Qt Recall that, by its definition, the = const + is that the level of heterogeneity is given by ft. composite shock depends on e and 7 but not on p. It is then evident that the response of the economy to the underlying aggregate productivity, taste, or mark- up shocks of strategic The is independent of is p. In this sense, the business cycle complementarity that is intuition behind this result is indeed independent of the degree induced by trade. is further explained in Angeletos that the strength of trade linkages matters only for how much and La'O (2009b). The key agents care about forecasting the level of economic activity relatively to forecasting the underlying economic fundamentals. But when information of is symmetric (commonly shared), any uncertainty the agents face about the economic activity reduces to the one that they face about the underlying economic fundamentals, which renders the degree of strategic complementarity irrelevant. In contrast, asymmetric (dispersed), agents can face additional uncertainty about the The bej'ond the one they face about the fundamentals. precisely the impact on equilibrium This distinct is important. It is level outcome precisely the aforementioned property that — for it level of economic is activity, strength of trade linkages then dictates of this residual uncertainty from uncertainty about the fundamentals when information is about economic activity. makes dispersed information only the heterogeneity of information that breaks the coincidence of forecasts of economic activity with the forecasts of the underlying fundamentals when the equilibrium is unique. We further elaborate on this point in Angeletos La'O (2009b), showing how dispersed information can open the door fluctuations. We refer the reader to that paper for a and to a certain type of sunspot-like more thorough discussion of this important, broader insight. In what follows, we concentrate on how this broader insight helps understand why the combination of dispersed information with the aforementioned type of complementarity can have a significant impact on the positive properties of the 16 RBC paradigm. 3.4 . The Relation to complementarity in New-Keynesian models (e.g., Woodford, 2003b) looks like the following: ^ Pr,t where 2,.( pi^t is {I - oyt + the target price of a firm (in logs), yt with condition (9) in complementarity is + z,,t, shocks, and ^ is striking. is pt is the aggregate price level, a coefficient that we compare If The only interpreted is the above condition noticeable diiTerence seems New-Keynesian model, while a price in the (12) .,. in pricing decisions. our model, the resemblance to be that the rele\'ant choice variable ipt nominal GDP, is demand captures idios\-ncratic productivity or as the degree of strategic in New-Keynesian paradigm familiar condition that characterizes optimal target prices in the it is a quantity our model. However, there are some crucial differences behind this resemblance. First, condition (12) does not alone pin down other conditions regarding the determination of (9) offers a the equilibrmm. Rather, 3^^, GDP the nominal it must be combined with In contrast, condition level. complete, self-contained, representation of the equilibrium in our model. Second, the endogeneity of yt undermines the meaning of condition (12). For example, letting denote real GDP and using yt = pt+yt, condition (12) can also be restated as p^j but then the degree of complementarity appears to be is more informative when money is neutral, because yt condition determines only relative prices. But even degree of complementarity at the very least is 1, in pricing decisions not is In contrast, in our when money is non-neutral, ^ simply because nominal is model the degree only by preferences and technologies, and is + (l— Oyf+^i,*! In fact, this alternative representation because monetary policy responds to variation to policy parameters. P( then exogenous to nominal factors and this incorporated, the complementarity in pricing decisions down ^. — yt in pt different GDP and is yt- fails to far identify the from exogenous Once this endogeneity from ^ and becomes sensitive of strategic complementarity is pinned completely invariant to monetary policy. Third, the comparative statics of the complementarity in our model (q) with respect to deeper preference and technology parameters are different from those of In particular, note that decreases with e a the labor and production side, as what contributes in to strong is to do with is important to bear either the degree of complementarity crucial, along in mind that our notion monopoly power S, of in the in our model is with high substitutability Hansen (1986) and King and Rebelo discussants highlighted, the opposite comparative statics hold for it (^). (the inverse of the Firsch elasticity of labor supply), and increases with 6 (the degree low substitutability in the commodity side, so that trade Nevertheless, New-Keynesian counterpart decreases with p (the elasticity of substitution across different goods), of diminishing returns to labor). Hence, in its (2000). As one of our New-Keynesian paradigm. complementarity may have little or the price elasticities of individual demands. In 17 our model, that latter are pinned down by degree of strategic complementarity is (the within- island elasticity of substitution), while the 77 pined down by p (the across-island elasticity). Last, but not least, the complementarity highlighted in the vanish in is New-Keyenesian framework would New-Keyenesian complementarity firms were setting real (indexed) prices. In this sense, the if a nominal phenomenon, whereas ours a real phenomenon. is Dispersed information and the business cycle 4 we seek In this section to illustrate positive properties of the RBC how the introduction of dispersed information can impact the paradigm. To facilitate this task, we impose a Gaussian specification on the shocks and the information structure, similar to the one Pavan (2003a), Angeletos and Assumption 1. The shocks and The aggregate shock (i) (2007), ft and many and variance al The (ii) follows a Gaussian = local shock = I/k/, ft is i.i.d. ft, and i.i.d. sian sufficient statistic where Qt is £,it: all ^ ut is and — fit + <^tt r . 1 Normally distributed with m.ean zero and variance a^, x^ such ft is summarized m a Gaus- that = ft + ^it, , . Normally distributed with mean zero and variance across islands.^^ i.i.d. Normal innovation, with a across islands. The public information about statistic yt where noise, process: i^t, - ^It (iv) + random walk The private information of an island about the aggregate shock (ill) and ipft-i or given by a purely idiosyncratic shock, is ^-u orthogonal to ft AR(1) over time. ft where others. parameterizes the persistence of the composite shock and ip mean Morris and Shin (2002), Woodford the available information satisfy the following properties: ft where in ; cj^ = I/k^,, orthogonal to both ' • , the aggregate shock ft is summarized in a - „ Gaussian sufficient such that --.'• St is noise, yt^ ft + ^t, Normally distributed with mean zero and variance a^ = l/^y, and orthogonal to other variables. Note that the local fundamental /,t is itself a private signal of /(. However, by the sufficient statistic of all the local private information, the informational content of /,( 18' is fact that we define x,, as already included in Xn. a This specification imposes a certain correlation up shocks: composite shock fu to follow a univariate process as above, for the three type of shocks are We simplicity. underlying productivity, taste and mark- in the moved bj' a single underlying factor. However, this it is must be that all the only for expositional can easily extend our results to a situation where each of the shocks follows an independent Gaussian process, or consider a more general correlation structure among the shocks. Closed-form solution 4.1 Under Assumption 1, we can identify W( with the vector {ft,xt.yt)- normal distribution with mean can guess and verify that there and \ogQt familiar is linear in (ft-i, ft,yt)- method of undetermined out any other equilibrium. Proposition We log9j( We then a joint which logg,( in is (ft,yt)- Next, we linear in {ft-i, fit, x%t, Vt) then find the coefficients of these linear functions by the we can use an independent argument to coefficients. Finally, rule thereby reach the following result. Under Assumption 3. from n^ to the more convenient vector always an equilibrium is is and an invariant variance-autocovariance matrix, we can ift,ft,yt) also reduce the aggregate state variable Because Qt = the equilibrium, 1, const + + (yJ_i/f_i of local output level, <fffit + fxXit + is given by (13) 'Pyyt, where the coefficients {ip-i,tpf,ipx,tpy) are given by fx - { [ (1 ^T - a)Ki + Ky + ^v ) Oi Kf = } ^ { S + ^/ 1(1- Q)^x + This result gives a closed-form solution of the equilibrium level of output ^ » (14) J in each island as a log- linear function of the past aggregate fundamental ft-i, the current local fundamental the local (private) signal output IS Xit, and the public signal i/(. Note then that the equilibrium necessarily an increasing function of the local fundamental interpret this sign, note that higher / a lower monopolistic distortion. means a higher But whether and how f,t'- VJ/ > fit, level of necessarily. To productivity, a lower disutility of labor, or local output depends on ft-i, x^t and yt is determined by the degree of strategic complementarity a. To understand this, note that local output depends on these variables only because these vari- ables contain mformation about the current aggregate shocks and, in so doing, help agents forecast the aggregate level of output. But when a = 0, the demand- and supply earlier perfectly offset each other, so that at the local incentives depend only the local side effects that we discussed end economic decisions are not interdependent: fundamentals and not on expectations of aggregate 19 activity. follows that the It hand, if a ^ 0, dependence of local > 0), output to /t-i, xn and output depends on ft-i, help predict aggregate output. (a local Xit In particular, and yt j/t vanishes when a = On 0. because, and only because, these variables when economic complements decisions are strategic the equilibrium level of output in each island responds positively to expectations of aggre- gate output; in this case, the coefficients decisions are strategic substitutes (q > and (/?_i,y)j., 0), <py are all positive. When instead economic the equilibrium level of output in each island responds negatively to expectations of aggregate output; in this case, the coefficients ^p-\,i^x: and negative. As mentioned earlier, more rather, it yt, is or 0, and hence in all which economic as the empirically most interpretation of noise and comparative statics we proceed, we would the signal a > are generally. Remark on Before the case in which ipy For this reason, our subsequent discussion will focus on this case; however, our relevant scenario. results apply we view good news about aggregate fundamentals, activity responds positively to 4.2 the other its noise like to emphasize that one should not give a narrow interpretation to This signal £(. is not meant to capture only purely public information; a convenient modeling device for introducing correlated errors in beliefs of aggregate fundamentals. Indeed, the results we document below can easily be re-casted with a more general information structure, one that allows agents to observe multiple private signals and introduce imperfect cross-sectional correlation in the errors of these private signals; the origin of noise, then, is not only the public signal, but also the correlated errors in the private signals of the agents. more general interpretation invite the reader to keep this it is a acronym Similarly, of what We "noise "stands for in our model; for all sources of correlated errors in expectations of the fundamentals.^' we would like to warn the reader not to focus on the comparative equilibrium with respect to the precisions of private and public information, discussants suggests. These comparative statics fail to isolate the distinct Kj- and statics of the K.y, as one of our impact of the heterogeneity of information, simply because they confound a change in the heterogeneity of information with a change in the overall precision of information.'- signals with correlated errors, ''in fact, one could go further do not move at all it Furthermore, if we had allowed would be unclear whether an increase and interpret "noise 'as a certain cis robustly once information dispersed. See Angeletos and is ours, such shocks cannot exist in the precision of a certain type of sentiment shocks, namely shocks that the agents' beliefs about the fundamentals and nevertheless a unique-equilibrium model for multiple private move equilibrium outcomes. With when information is commonly shared; but emerge La'O (2009b). '^For example, an increase in Ki would increase the heterogeneity of information, but would also increase the overall precision of information; and while the former effect would tend to amplifv' the volatility effects here, the latter eifect would work in the opposite direction. 20 we have documented signal raises or reduces the heterogeneity of information. on the comparative statics how follows we focus of information.-^^ and noise to fundamentals the dispersion of information and the strength of trade linkages affect aggregate Towards fluctuations. what in simply because the degree of complementarity matters for aggregate Macroeconomic responses study mind, model only by regulating the impact of the heterogeneity fluctuations in our We now this in with respect to a. These comparative statics best isolate the distinct role of dispersed information, 4.3 With we aggregate condition this goal, (13) and use the fact that ft = 4'ft-i + i^t to obtain the following characterization of aggregate output. Corollary 1. Under Assumption the equilibrium level of aggregate output 1, = \ogQi const + i' ft-\ + LpuVt + is given by (15) ^E^t, where = V3^ <^/ and where + Ut fundamentals, + V^a; = St ft = Condition (15) (/^y — yt = 1 - i'ft-i — aKf T; ^p^ = Lpy aKy = is 4-' £t- by the aggregate noise. the equilibrium level of aggregate output as a log-linear function of the Consider the impact coefficient ipi,. effect of This same essentially the effect of Because the have that the impact is latter an innovation is (e.g., underlying shocks, the an innovation fundamentals. in fundamentals decreases with the — is measured Reis, 2002): the less informed we level of noise. monetary shocks frictions (e.g., Lucas, 1972; Barro, 1976) they respond to these shocks. agents interact with one another and the current This effect insight as the one that drives the real effects of Mankiw and less in i^t, a decreasing function of the precisions k^ and Ky, both the older macro models with informational recent descendants (16) the persistence in the past aggregate fundamentals, ft-\, the current innovation in the fundamentals, noise, , f ^ \ the innovation in the fundamentals, is ft ts the gi\'es and \ ^; in and their economic agents are about the Clearly, this is true no matter whether true even in a single-agent decision problem. it is ''Angeletos and Pavan (2007a) propose that a good measure of the "commonahty" of information (an inverse measure of the heterogeneity of information) is the cross-sectional correlation of the errors in the agents' forecasts of the fundamentals: holding constant the variance of these forecast errors, an increase in the correlation implies that agents can better forecast one another's actions, even though they cannot better forecast the fundamentals. Following this alternative route would deliver similar insights as the ones 21 we document here. More interestingly, we find that ip^, a decreasing function of a. That is is, the more economic agents care about aggregate economic activity, the weaker the response of the economy to innova- At the same time, we tions in the underlying fundamentals. of a. That find that f^ is an increasing function the more economic agents care about aggregate economic activity, the stronger the is, equilibrium impact of noise. These properties originate from the interaction of strategic complementarity with dispersed information. Indeed, if the underlying shock was common knowledge here can be nested by taking the limit as the public signal becomes infinitely precise, Hy would cease to depend on then both y?^ and (p^ a reduces ^Pi, and raises (p^- business cycle once information Corollary 2. When how This highlights is information But a. as long as information strategic complementarity is (which — > oo), dispersed, a higher becomes crucial for the dispersed. is dispersed, and only then, stronger complementarity dampens the impact of innovations in the fundamentals on eqmlibrimn output arid employment, while amplifying the impact of noise. The key ...__. intuition behind this result is the same as the one in the more abstract work of Morris and Shin (2002) and Angeletos and Pavan (2007a). Public information and past fundamentals (which here determine the prior about the current fundamentals) help forecast the aggregate level of output relativel}' better in than private information. The higher a any given island depends on the more anchored why information. The anchoring why similar anchoring effect of the common more less it depends on the in of each island sensitive to public information, effect of past aggregate output responds less to any innovation sensitivity to noisy public information explains output and the a induces the equilibrium output to the past aggregate fundamentals, less sensitive to private the more the equilibrium level of output local forecasts of aggregate local current fundamentals. It follows that a higher to be is, and aggregate fundamentals explains the fundamentals, while the heightened aggregate output responds more to noise. prior underlies the inertia effects in A Woodford (2003a), Morris and Shin (2006), and Angeletos and Pavan (2007a), while the heightened sensitivity to public information we favor a is the same as the one more general interpretation in Morris and Shin (2002). However, as mentioned before, of the signal j/(, not as a public signal, but rather as a source of correlated noise in forecasts of economic fundamentals. As another way to appreciate the aforementioned result, consider following variance-decomposition exercise. Let logQt be the projection of logQt on past fundamentals. by logQt = logQt — logQt aggregate output. Its total — The residual, which is given puVt +p}^et, can be interpreted as the "high-frequency component variance is of the innovation in the fundamentals Var(log Q; — ) plal+'pfa'^, where a^ (= 1/k/) and a^ (= I/k^) 22 is the variance of the noise, is ''of the variance The fraction of the high-frequency variation in output that originates in noise Var {log Qtli^t) _ _ — ^Mioise Var(\ogQt) Since a higher Q' raises about the aggregate is and reduces i/?£ level of economic i/Pj/, it is thus given by the following ratio 2^2 fjcr] E" E ^Wu + ^l^Y more agents care necessarily raises this fraction: the more the high-frequency activity, the output that volatility in driven by noise. We can then further highlight the distinct nature of dispersed information by showing that, as long as Q is high enough, the contribution of noise to short-run fluctuations can be large even level of noise small. is Note that the fundamentals is given by Proposition 4. When k, = k.o + overall precision of an agent's posterior + Hx information is K,y We w a is not be possible when information tion of noise on the business cycle as this precision becomes infinite. this possible is is can be arbitrarily sufficiently high, agents is ~ 1). commonly shared. In that case, the contribu- when information In contrast, noise in the business cycle can be high even What makes about the underlying tightly connected to the precision of information is when the is and vanishes dispersed, the contribution of precision of information arbitrarily high. is the combination of heterogeneous information with a sufficiently strong degree of strategic complementarity induced by trade linkages. Note then how this result also con- would have trasts with our earlier observation that this particular type of strategic complementarily been irrelevant Finally, it is for the business cycle had information been commonly shared. worth noting how the dispersion of information and trade hnkages behavior of aggregate employment. The latter = logA^( where 0( is of employment proportional to that of output. is const the aggregate productivity shock immediate that the response to The same + (i.e., lilogQt - at), the cross-sectional average of log/1, an aggregate shock is common in either tastes or > 0. When information an innovation to aggregate productivity "This fraction equals may now 1 minus the R-square is is simply commonly /3, 23 It is then interestingly, turn from a positive To see this, let shared, the sensitivity of output to and that of employment of the regression of log (). monopoly power More true for the response to noise. information to a negative sign under dispersed information. P = -^ = ^_g^^,g affect the cyclical given by the response of employment to an aggregate productivity shock sign under the co) and, yet, the high-frequency variation in aggregate output can be driven almost exclusively by noise (Rnmse Clearly, this if can then show the following. dispersed and well informed about the fundamentals {k is 14 Q, on the innovation is u,. g(/? — 1). When, instead, information employment, for dispersed, the corresponding sensitivities are is witli Suppose as in (16). ipi, common positively to a productivity shock under RBC framework. As noted that, when information and may actually turn is it earlier, ipi, /3 > 1, for output and ^{ip^(3 — 1) which means that employment responds information, as in any plausible calibration of the necessarily lower than is </3^/3 1 and dampen dispersed, stronger trade linkages is decreasing in a. It follows employment the response of negative. Slow learning 5 The preceding has focused on common knowledge a setting where the underlying shocks become within a period. Although this permitted a sharp theoretical analysis of the distinct implications of dispersed information, and of its interaction with trade linkages, results to either empirical business cycles or calibrated RBC models. it map our seek to illustrate how makes We now hard to it incorporating slower learning can facilitate a better mapping between our analysis and the data. Towards licly this goal, to relax the assumption that the aggregate state, fif, becomes pub- revealed at the end of each period, and instead allow for more interesting learning dynamics. Accommodating ized we need commodity micro-founded way would require that there this possibility in a fully no central- is trading: with centralized trading, equilibrium prices are likely to reveal the state. However, allowing for decentralized trading would complicate the analysis by introducing informa- tional externalities and/or by letting the relevant state space explode as in are currently exploring some possibilities along these lines. However, (1983). We purposes, we Townsend for the current opt for tractability and expositional simplicity. In particular, we assume that period, nor do they learn about Rather, they type as in onl}' it firms and workers do not observe the state from observing past aggregate economic outcomes or past prices, 1, and they use these signals to update each period underlying state. Think of this as follows. Each firm has two managers: one employment and production; and another who realized profits to the firm's shareholders. —but Similarl}', the consumers, this information to the solely sells on the exogenous who their beliefs who The two managers share the same He about the decides the level of the product, receives the revenue, and sends the do not communicate with one another. receives any signals on economic activity. signals. end of each keep receiving exogenous signals about the current fundamentals, of the same Assumption firm valuation at the Moreover, the objective first — maximize manager never only observes the exogenous local private and public observe all the prices in the economy, fail to communicate workers in their respective famihes. The workers also base their decisions signals. 24 Needless to say, this specification of the learning process would also be naive to take to receive each period are that these agents knowledge too may it too literally: meant have. not particularly elegant. However, it the exogenous signals that we allow firms and workers more generally the multiple sources to capture of information To the extent that the underlying shocks do not become common more plausible formalizations fast, is of the learning process, albeit highly desirable, need not impact the qualitative properties we wish highlight here.^^ Under the aforementioned the same specification, equilibrium behavior continues to be characterized best-response-like condition as in the basehne model: logq,j = (1 where we have normalized the constant - + Q)fi,t Q'E,,t The to zero. [logQf] (17) only difference is in the information that underlies the expectation operator in this condition. Fnially, for concreteness, on productivity shocks as the only shock to fundamentals: The procedure we is by summarized in a vector Woodford (2003a). Xt comprised We = /SlogA,,;, with dynamics follow to soh'e for the equilibrium similar to the one in /,,( we henceforth focus is based on /3 = ^_qI^^,q Kalman filtering and guess and verify that the aggregate state can be fundamental and aggregate output: of the aggregate s, Xt (18) logQt Firms and workers in any given island never observe the state, but instead receive the following vector of signals each period: ft + (;it It + et before, yt should not be taken too literally — (19) Zit Vt As emphasized introducing and common is a 2 x 2 matrix, while equilibrmm values of Finally, we guess Xt follows a simple law of motion: A'; M a convenient modeling device for noise in the agents' forecasts of the state of the economy. verify that the state vector where it is M,m^, and - MXt-i + m^ and m,. are m.iyVt + rn^Et 2x1 vectors. (20) We then seek to characterize the m.^. ^^The learning process we assume here is similar to the one in Woodford (2003a). We refer the reader to Amador and Weill (2008), Angeletos and La'O (2008), Angeletos and Pavan (2009), Hellwig (2002), and Lorenzoni (2008) some alternative formalizations of the learning process. the positive results we document to the extent that they make the in this section. None of these alternative formalizations would for crucially affect However, some of them would have distinct normative implications, learning endogenous to the actions of other agents. See Angeletos and La'O (2008) and Angeletos and Pavan (2009) on this issue. 25 In each period t, firms and workers start with that they receive in the beginning of period prior about Xt and use the new signals update their behefs about Xt- to t some Local output is then determined Condition (17) then givens local output as a function of the local belief about Xt- Aggregating across islands, we obtain the aggregate of motion that aggregate output follows the equilibrium is We must match the one believed by the a fixed point between the law of motion believed by agents firms. making following decisions that firms and workers are characterize the fixed point of this problem in the and used to form their signal extraction problem. Appendix and use its solution to numerically simulate the impulse responses of output and employment to positive innovations in Vt For our numerical simulations, we interpret a period as a quarter. Accordingly, we for the we standard deviation of the productivity innovation and set 6 = .60 and e = Less standard implying that labor income is irrelevant, is is our choice of 7. period is = 0.02 Next, persistence. is no capital, the only source of wealth, the elasticity of intertemporal substitution bay. level of specific value for We elasticity of labor supply. ^^ = accordingly set 7 .2 to Next, we set the standard deviations These values are arbitrary, but they are not implausible: when the a economic activity is likely to (equivalently, p). Rather, be very limited. Finally, we do not pick we study how the variance decomposition of the high-frequency components of output and employment varies as we vary q from mind = interpreted as a quarter, the information about the current innovations to fundamentals and/or the current any (Ty its Recall that in our setting there ensure an empiricallj' plausible income effect on labor supply. = 0.99 for let a^, £(. These parameter values are broadly consistent with 2. and 7 only controls the income of the noises as a^ = -i/' and which correspond to an income share of labor equal to 60% and a .5, Frisch elasticity of labor supply equal to the literature. Therefore and the law of motion induced by the optimal output and their forecasts of the aggregate state, employment In equilibrium, the law level of output. that a higher a means to 1 (keeping in stronger trade linkagess or, equivalentlj', a lower p). Impulse responses to productivity and noise shocks 5.1 Figure 1 plots the impulse responses of aggregate output of productivity, for various degrees of a. responses we report, common is Size of the innovation here, equal to one standard deviation.) Clearly, if and in all other nnpulse aggregate productivity were knowledge, then output would follow the same AR(1) process as aggregate productivity This is simply because there information is dispersed but there itself. (The and employment to a positive innovation Woodford (2003a) uses a similar is is number no capital in our model. The same thing happens when no strategic complementarity for calibrating 26 in output decisions (a the New-Keynesian model in = the absence of capital. 0). Impulse Response ol Output to Productivity Shock 0.03 0,025 " * "> •.-,,« 0.02- .^^^^ ** 0.015- - - a=0, common knowledge ' u=.5 0.01 0.005 10 5 ( Impulse Response M 15 periods B of Employment to Productivity Shock ' ^^^^ (• - " a=0, common knowledge / - - - -a=.5 -0.04 20 10 periods Figure This is island if Impulse responses to a positive innovation 1: a = simply because when knows perfectly its own In contrast, and output islands. when information result, is knowledge. dispersed but islands are interconnected (a less to the shock than what it ^ 0), employment in other would have done had the shock been knowledge, precisely because each island expects output Note then that the key for the response of each island is tangle an aggregate shock from an idiosyncratic shock. informed about the aggregate shock, as long as q than under to the aggregate shock as even though each island remains perfectly informed about their local funda- mentals, each island responds common common economy responds one island depends crucially on expectations of employment and output in As a islands are effectively isolated from one another; but as each productivity, the entire the aggregate shock had been in productivity. common knowledge if it > is other islands to respond less. not per se whether the island can disen- Even if a particular island was perfectly the island will respond less to this shock expects the other island to respond the other island has imperfect information about the shock. response of aggregate outcomes in less, presumably because Thus, the key for the inertia m the the uncertainty islands face about one another's response, not necessarily the uncertainty they themselves face about the aggregate shock. As evident in Figure complementaritj'. This less is 1, the equilibrium inertia because of two reasons. is higher the higher the degree of strategic First, there is a direct effect: the higher a is, the the incentive of each island to respond to the underlying shock for any given expectation of the response of other islands. But then there is also an indirect, multiplier-like, 27 efTect: as all islands are Impulse Response of Output to Noise Shock 0.02 0.01 10 periods Impulse Response of Employment Figure expected to respond less to At the same time, the to the shock is 2: to Noise Shock Impulse responses to noise. the underlying shock, each island finds inertia vanishes in the long-run: the same as with common it optimal to respond even the long-run response of the is First, note that agents are always perfectly informed about their no learning in this is that's only part of the own fundamentals, so there dimension. Second, recall that agents do not care per se about the aggregate fundamentals, so the fact that they are learning more about them the key economy knowledge. This seems intuitive: as time passes, agents become better informed about the underlying aggregate shock. However, stor3'. is per se inconsequential. Rather, that agents in each island are revising their forecasts of the output of other islands. then drives the result that inertia vanishes in the long-run is What merely that forecasts of aggregate output eventually converge their common-knowledge counterpart.'^^ Finally, a salient property of the Q, the short-run happens for dynamic response of ' employment is that, for sufficiently high impact of a productivity shock on employment turns from positive to negative; this parameters values for had information been symmetric. ^^It less. may be hard which the model would have generate a strong positive response We find this striking. to fully appreciate this point, because icnowledge counterpart is itself how The fast baseline RBC paradigm has long been output forecasts converge to their common- pinned down by the speed of learning about the underlying aggregate productivity shock. However, with richer information structures, one can disentangle the speed of adjustment in output forecasts from the speed of learning about the fundamentals. Angeletos and La'O (2009a) for a related It is then only the former that matters for the result. example within the context of a Calvo-like monetary model. 28 See criticized for generating a near perfect correlation data this correlation negative q if is Some authors near zero. In our setting, this correlation could be close to zero or even turn is paradigm information VAR the structural shocks lead to a reduction RBC Of sufficiently high. in (e.g., may between employment and output, whereas in the in course, correlations confound the effects of multiple shocks. literature have thus sought to show that employment and have then argue that Gah, 1999; Gah and Rabanal, 2004). potentially help the RBC identified technolog}' this as a clear rejection of the we have shown that dispersed Here, paradigm accommodate this fact without any need to invoke sticky prices. Indeed, note that as to fall. If agents had been imperfectly informed about the produc- shock but information had been common, then they could much as they employment — is for how could they respond in the first to the shock place? Thus, Turning to the in to increase their effects of noise, in by reducing employment employment well informed about the shock but the shock ployment fail employment would have done with perfect information, but they would not have reduced their aware of the shock agent the dispersion of information, not the uncertainty about the productivity employment shock, that causes tivity it is is falls not m if they were not our model precisely because each common knowledge. Figure 2 we consider the impulse responses of output and em- response to a positive innovation in £(. As emphasized before, this should be interpreted as a positive error in expectations of aggregate output, rather than as an error in expectations of When a = aggregate fundamentals. 0, such forecast errors are irrelevant, simply because individ- ual incentives do not depend on forecasts of aggregate activity. But when a a positive response in output and employment, thus becoming partly — 0, thej' self-fulfilling. generate Furthermore, the stronger the complementarity, the more pronounced the impact of these errors on aggregate employment and output. The figure considers a positive noise shock, economic activity. The impact these shocks occur, output, movement type of in TFP. The demand shocks. of a negative shift in expectations employment and consumption move resulting We not. This 5.2 is is in the shift in expectations about symmetric. same Note that when direction, without any recessions could thus be (mis)mterpreted as a certain a moment. Finally, note that the impact of employment can be quite persistent, even though the noise itself simply because the associated forecast errors are themselves persistent. Variance decomposition and forecast errors Comparing the responses former booms and will return to this point in these noise shocks on output and is which means a positive is of employment with those of output to the two shocks, we see that the smaller than the latter in the case of productivity shocks but quite larger in the case of 29 Fraction ot Outpul Variance due to Noise Fraction ot Employment Variance Oue to Noise 10 Figure This noise. is 12 Variance decomposition. 3; simply because productivity siiocks have a double effect on output, both directly and indirectly through employment, while the noise impacts output only through employment. But then the response of employment to noise there are diminishing returns to labor [6 < is bound 1), to be stronger than that of output as long as and the more show the lower 6. It follows that noise contributes to a higher relative volatility for employment, while productivity shocks contribute in the opposite direction. In the standard RBC framework, employment may exhibit a higher volatility than output to the extent that there are powerful intertemporal substitution have ruled out since we have also ruled out in this RBC capital). dimension. Our results here indicate framework in this Comparing Figures how RBC However, the effects framework (which here is known we to lack noise could help improve the performance of the dimension. 1 and 2, it is evident that low- frequency movements in employment and output are dominated by the productivity shocks, while noise contributes relatively more to highfrequency movements. To further illustrate this property, in Figure 3 we plot the variance de- composition of output and employment at different time horizons. For sufficiently strong strategic complementarity, productivity shocks explain only a small fraction of the high-frequency variation in output of noise —short-run fluctuations are driven mostly by is is As for employment, the contribution quite dramatic. Finally, Figure 4 plots the level of noise. dynamics of the average forecast of aggregate output and the true aggregate output in response to a productivity or noise shock. The average forecast error the distance between the two aforementioned variables. forecast errors are smallest when A salient feature of this figure the degree of strategic complementarity 30 is highest. is that Impulse Responses to Produclivily Impulse Responses Shock, tt=.5 0,025 0.02 0,02 016 0,015 0,01 to Noise Shock, a= 5 - - - average 0,01 .. torecasi aciual output : \ 005 average lotecasi • s actual output 0.005 5 ( 15 10 10 5 ( Impulse Responses 15 periods periods to Productivity Shock, a= 9 Impulse Responses 0.025 0,025 0.02 0,02 0015 0015 lo Noise Shock, a=,9 - - - average 001 Z^'^.. 005 , This is of output - - - average crucial. 4: We lorecast - Q, the 10 10 periods is earlier showed that a higher a leads to both more to productivity shock, a guarantees and to a bigger is inversely related to the magnitude the impact of noise become nearly is highest inertia in the response impact of when a is noise. In this sense, highest. However, one in large forecast errors. more driven by To the contrary, forecasts of economic activity, so that the forecast errors are smaller. of the deviations of actual Combined, these -• ;-" Forecast errors in response to producti\'ity and noise shocks. implies that actual economic activity magnitude ... ... periods and employment that at the end a higher - ^^ should not expect that these large deviations will show up o lorecast ,;. . 005 the deviation from the common-knowledge benchmark a higher . \ok 01 . Figure •. \. It follows that, as we vary outcomes from their common-knowledge counterparts of the associated forecast errors. Indeed, both the inertia self-fulfilling as results illustrate the distinct q gets closer to and 1. and powerful mark that the dispersed information can have on macroeconomic outcomes once combined with strong strategic complementarity. Not only can the macroeconomic effects we have documented be quite significant, but they are also consistent with small errors in the agents' forecasts of either the underlying economic fundamentals or the level of economic activity. VARs 5.3 Demand Many economists have found the idea that short-run fluctuations are driven primarily by technology shocks, new-Keynesian models, and structural shocks implausible either on a priori grounds or on the basis of certain structural VARs. Blanchard and Quah (1989) were the first to attempt to provide some evidence that short-run fluctuations are 31 "demand driven by "rather than "supply "shocks, albeit with the caveat that one cannot know what the shocks they identify really capture. Subsequent contributions by Gali (1999), Basu, Fernald and Kimball (2006), Gali and Rabanal (2004), and others have tried to improve way dimension. in that One or another, though, this basic view that business cycles are not driven by technology shocks appears to underly the entire New-Keynesian literature. Our iindings here are consistent with this view. In our environment, technology shocks may explain only a small fraction of the high-frequency volatility in macroeconomic outcomes. However, the residual fluctuations have nothing to do with monetary shocks. Rather, they are the product of the noise in the agents' information. Importantly, to the extent that information dispersed and is trade linkages are important, this noise might be quite small and nevertheless explain a big fraction macroeconomic outcomes. of the high-frequency volatility in Furthermore, the noise-driven fluctuations we have documented here, albeit being purely neoclassical in their nature, they could well shocks in the following sense. This is be interpreted as some kind of "demand" or "monetarj'" because they share many of the features often associated with such shocks: they contribute to positive co-movement in employment, output and consumption; they are orthogonal to the underlying productivity shocks; they are closely related to shifts in expectations of aggregate demand; and they explain a large portion of the high-frequency variation in employment and output while vanishing To at low frequencies.-'^ we generate data from our model using better appreciate this, suppose that specification for the productivity shock keynesian type would then — to run a structural correctlj' identify the and VAR let an applied macroeconomist as in — preferably of the new- like; is One allowed and the underlymg noise shocks by the and Quah, the productivity shocks would be interpreted as "supply shocks "and the noise shocks as and the random- walk underlying innovations to productivity by the shock that In the language of Blanchard relation to sticky prices a ' Blanchard and Quah (1989) or Gali (1999). to have a long-run effect on output or labor productivity, residual. ^^ ' "demand shocks", however, the to the contrary, both type of shocks latter would have no emerge from a purely supply-side mechanism. In the language of Gali (1999), on the other hand, the productivity shocks would be interpreted as "technology shocks". Furthermore, of employment to these identified shocks as already noted, the short-run response would be negative for high enough a; but this would no favor a sticky-price interpretation. '*A priori, our predictions regarding real quantities seems consistent with any predictions about nominal prices. Further exploring under what conditions our noise-driven fluctuations could be associated also with procyclical nominal prices requires a monetary extension of the model. "incidentally, note that the econometric issues studied in Blanchard, L'Huillier, and Lorenzoni (2009) do not apply to the type of noise fluctuations obtained in our model. 32 As mentioned Gilchrist Beaudry and Portier (2004, 2006), Christiano in the introduction, et al. (2008), and Leahy (2002), Jaimovich and Rebelo (2008), and Lorenzoni (2008) have explored the idea that noisy news about future productivity contribute to short-run fluctuations. Furthermore, how Lorenzoni (2008) interprets the resulting fluctuations as "demand shocks" and discusses help match related facts. However, they these papers focus on fluctuations that originate from un- all certainty about a certain type of fundamentals (namely future productivity), not on the distinct type of uncertainty that emerges when information is paper. ""^ Second, as often the case with new-keynesian models, Lorenzoni's found real shocks in his model (as in ours), expectations of future productivity policy away from the one that would shocks "obtain in RBC an setting would have been irrelevant is and are completely unrelated shock in no capital for to current news about that they cause an expansion in replicate flexible-price allocations. Finally, note that a positive productivity In contrast, our monetary "demand monetary pohcy. our model induces a small impact on output foUowed by a large persistence response at high frequencies, is prices been flexible; the only reason then that future productivity cause demand-like fluctuations this "demand shocks" con- with monetary shocks. By this we mean the following. Since there macroeconomic outcomes had nominal m heterogeneous and that we highlight at lower frequencies,-^ Again these properties are consistent with the estimated dynamics of "technology" shocks. More prices generally, dampen is it many interesting to note that in empirical new-keynesian models sticky the response of output to productivity shocks relative to the RBC framework and help get a negative response for employment. According to some researchers, these properties seem to be more consistent with the data than these models is their RBC is that monetary policy optimal thing to do. sticky properties Perhaps as fails to model a success for policy often reacts to approach may have is typically the is based on the idea that monetary error in the level of aggregate economic activity (Bernanke is more plausible of a producti\ity shocks mean-reverting. becomes If instead we assume that at positive, while the rest of the results 33 is a random In an calibration of the model, ^^In our numerical exercises, the impact of the productivity shock vanishes asymptotically, only because Oi is (slowly) and that measurement model, Lorenzoni considers a representative-agent model with symmetric information. extension, he allows for dispersed information, but only to facilitate a assumed that RBC intriguing implications for the identification of Mihov, 1995: Christiano, Eichenbaum and Evans, 1999), In particular, the idea his baseline in the baseline for policy. of the standard identification strategies measurement from that economy same empirical properties without introducing and without presuming any suboptimality interestingly, our differs replicate flexible-price allocations, which Here, instead, we obtain the monetary shocks. One ^'^In is only a failure for monetary policy: the only reason that the response of the to productivit}' shocks in the baseline new-keynesian model counterparts. However, what we have walk, then the long-run impact remain unaffected. error justifies the existence of random shocks true underlying state of the economy. If monetary pohcy, which are orthogonal to the to one then traces the impact of these particular shocks on subsequent aggregate outcomes, one can escape the endogeneity problem and identify the impact of monetary shocks. However, these measurement errors, or more generally any forecast errors that the central bank makes about current and future economic activity, are likely to be correlated with the corresponding forecast errors of the private sector. But then the so-identified monetary shocks may actually be proxying for the real effects of the forecast errors of the private sector, which unfortunately are not observed by the econometrician. 5.4 Labor wedges and Solow residuals Many authors have argued that a good theory of the business cycle must explain the observed variation in the labor wedge and the Solow residual (e.g.. We now 1999; Chari, Kehoe, and McGrattan, 2007; Shimer, 2009). model for these two key characteristics of the business Following the literature, we define the labor wedge The left Hall, 1997; Rotemberg and Woodford, consider the implications of our cycle. r„,f implicitly by panel of Figure 5 plots the impulse response of the labor wedge to a positive productivity and a positive noise shock. two types The labor wedge follows very different dynamics in response to the of shocks. In particular, a positive productivity shock induces a positive response in the labor wedge, implying positive comovement of the labor wedge with output. On the other hand, a positive noise shock produces a negative response in the observed labor wedge, implying a negative comovement with output. Multiple authors have documented that variation in the labor wedge plays a large role in ac- counting for business-cycle fluctuations during the post-war period. Importantly, the labor wedge is highly countercyclical, exhibiting sharp increases during recessions. facts and the multiple explanations that have been proposed the labor. These include taxes, shocks to the for the disutility of labor, observed countercyclicality of mark-up shocks, fluctuations wage-setting power, and Shimer's preferred explanation, search frictions we have found that We wards this goal, this input Solow noise offers another possible explanation for the finally consider the potential implications of we now introduce responds to shocks, but residual. As in Shimer (2009) surveys the the labor market. Here, in same in fact. our results for observed Solow residuals. To- a variable input in the production function; the optimal use of is unobserved by the econometrician and is King and Rebelo (2000), our preferred interpretation of 34 thus absorbed in the this input is capital Impulse Response ol Labor Wedge to Productivity Impulse Response Stiock ol Solow Residual to Productivity Shock 0.02 ----.«'-,_-, f " • 0,01 - - a=0 I 10 5 0^ ^^^ ' - - ' -0.005 Wedge to Impulse Response ol Solow Residual to Noise Shock Noise Shock :^ 0,01 .005 -0.01 -0.015 15 10 periods periods Impulse Response of Labor - -a=5 - - y*^ - - / - \ - a=0 - - - cz=.5 . ^^ if ,.=0 - - - n=.5 . ,.= ,9 , -0.02 10 10 5 periods periods Figure The only utilization. caveat is 5: Labor wedges and Solow residuals. that in our model capital exogenously fixed. However, we could introduce capital following the same approach as Angeletos and La'O (2009b), without affecting the quahtative points we seek to We make here. denote the unobserved input by and we specify the cost of this input net product of a firm then The optimal (1 — 6)^ = is input level of this (5 (1 -I- (J) Xif q^t — Qit we x,(; m terms of — -i^ ) 9 and given by equating is = i its -r^ ] A","' . Our = King and Rebelo, 2000), which implies , where marginal product with S,,5 > 0. The marginal cost: its 9 = (21) analysis then remains intact, provided we We .88. set ^ of the model continues = .6 and ^ = .1 we reinterpret (a preferred value also re-calibrate the underl3'ing aggregate = — 6'logA^t) implied by the to have a standard deviation of 0.02 and a productivity shocks so that the observed Solow residual {SRt common-knowledge version (5\jj -4,;,n„ the production function in the above way. Accordingly, in product as y4j(\'j~ nf,; obtain obtain the following reduced-form production function: ^^'^ ^'^"^s A,t final = Solving out for the optimal level of this input, ^Xut q,t where 9 the gross product of a firm be qu let logQt persistence of 0.99. The right panel of Figure 5 plots the noise shock. Both shocks has a persistent effect. raise the dynamic response measured Solow of the Solow residual to a productivity or a residual, but only the innovation in productivity Moreover, these responses of the Solow residual mirror those of output. follows that the Solow residual and output move tightly together, much 35 alike in a standard It RBC model, although employment has the more distinct behavior we mentioned Finally, it from variation worth noting that additional variation is in surveys is 5.5 is left Note that the observed heterogeneity in in particular highly countercyclical, suggesting that the dispersion of information how such countercyclical. Exploring cycle measured Solow residuals could obtain the dispersion of information, simply because the dispersion of information affects the cross-sectional allocations to resources. foreccist in earlier. may also be variation in the dispersion of information affects the business work. for future Discussion While the characterization of equilibrium in Section 3 allowed for arbitrary information structures, the more concrete positive results that information structure (Assumption emphasized We to we documented thereafter presumed a However, we do not expect any of the predictions we have 1). be unduly sensitive to the details of the information structure. build this expectation on the following observations. Proposition economy Gaussian specific, to a class of games with hnear best responses, (2002) and Angeletos and Pavan (2007, 2009). permits us to 1 map those studied in Morris and Shin like In this class of games, one can show under arbitrary information structures that a stronger strategic complementarity makes equilibrium outcomes sensitive to first-order beliefs (the forecasts of the fundamentals) that higher-order beliefs are more sensitive to the initial less and more sensitive to higher-order One can then proceed beliefs (the forecasts of the forecasts of others). our common to prior, show quite generally to public signals, and to signals with strongly correlated errors, than lower-order beliefs, simply because these pieces of information are relatively better predictors of the forecasts of others. beliefs are less sensitive to innovations in the noise than lower-order beliefs. dampens the response of noise We —which of the It follows that higher-order fundamentals and more sensitive to common sources of Combined, these observations explain why stronger complementarity economy to innovations in fundamentals while amplifj'ing the impact we documented are the key properties that drive the results in Sections 4 and 5. conclude that these results are not unduly sensitive to the details of the underlying information structure; rather, they obtain from robust properties of higher-order beliefs and the very nature of the general-equilibrium interactions in our economy. Our analysis has implications, not only for aggregate fluctuations, but also for the cross-sectional dispersion of prices and quantities. As evident from condition (24), a higher a necessarily reduces the sensitivitj' of local output to local fundamentals, while increasing the sensitivity to expectations of aggregate output. When of aggregate output, and hence heterogeneity information is commonl}- shared, in all agents share the same expectation output (and thereby 36 in prices) can originate only from heterogeneity in fundamentals (productivities, tastes, etc). necessarily reduces cross-sectional dispersion in output and only source of heterogeneity. However, once information prices, simply because dispersed, there is higher tlren follows that a It it dampens a the an additional source is of heterogeneity: different firms have different expectations of aggregate economic activitj'. It then follows that a higher We a dampens the former source of heterogeneity while amplifying the latter. conclude that, once information dispersion is ambiguous and quantities may dispersed, the impact of complementarity on cross-sectional —which also implies that evidence on the cross-sectional dispersion of prices guidance pro\'ide little Similarly, evidence is on the for a quantitative assessment of our results."^ monopolistic mark-ups, or the elasticity of demands faced by size of individual firms, do not necessarily discipline the magnitude of our results. This First, in our it is is for two reasons. model, the mark-up and the elasticity of individual demands identify only p that matters for complementarity. complementarity in our model is And second, evident from the definition of q, a high cis consistent with any value of p, provided that there is a sufficiently small wealth effect on labor supply in the short run, a sufficiently high Frisch elasticity (as 1985), and nearly linear returns to labor in the short run (as in Finally, it is New-Keynesian model which firms cannot in demand room tell King and Rebelo, 2000). own analysis. That paper considers a apart aggregate monetary shocks from idiosyncratic monetary shocks even when these shocks are unobserved, thus lea^ing These findings are interesting on for real effects. our motivation for focusing on However, real shocks. their it own right — and indeed complement was unfortunate for our discussant to extrapolate from that paper to the likely quantitative importance of our results. mechanism for local paper does not apply to our context: of that m ones if number ^^One of our discussants made the opposite argument. But necessarily reduces cross-sectional dispersion. introduced in Assumption 1 his First, the core firms were to confuse aggregate shocks our model, this confusion would only reinforce our results. quantitative findings of that paper are based on a a Hansen, shocks. For a particular calibration, this confusion induces firms to adjust their prices a lot in response to little in worth noting that, unlike what suggested by one of our discussants, the findings of Hellwig and Venkateswaran (2009) bear httle relevance to our productivity or whereas 77, '^'^ of heroic assumptions, And second, the which might serve argument was based on the premise that a higher This happens to be true under the specific signal structure we but, as just explained, is not true in general. We refer the reader to Angeletos and La'O (2007) for a stark example where higher complementarity increases both the non-fundamental component of the business cycle and the cross-sectional dispersion of quantities and prices. ^''To see this, recall shock Q ,j_^ in from Proposition 3 and Corollary 3 that the response of equilibrium output to an idiosyncratic fundamentals +,t +^ • is ^^ long given as b)' a > idiosyncratic shock only helps v^/ 0, ipf dampen = 1 is smaller than — Q, while its yPi,, response to an aggregate shock given hy ipi, which means that mistaking an aggregate shock the response of the economy to the aggregate shock. 37 is = for \ — an certain purposes but are completely out of place in our With these observations we own context.^'' are not trying to escape the need for a serious quantitative exercise, nor are we ready to speculate on the outcome of such an exercise. guidance in this and (ii) for any future quantitative exploration of our paper hinge only on (i) results. We are only trying to provide The key effects some we have documented the sensitivity of individual output to forecasts of aggregate output the sensitivity of these forecasts to the underlying shocks. We are thus skeptical that micro evidence on prices or quantities can alone provide enough guidance on the quantitative importance of our results. We instead propose that a quantitative assessment of our results should rely more heavily on sur^'ey evidence about the agents' forecasts of economic activity. Indeed, these forecasts summarize concisely all the informational effects in our model, and their joint stochastic behavior with actual outcomes speaks to the heart of our we In this regard, find the approach taken results. Coibion and Gorodnichenko (2008) particularly in promising. This paper uses survey evidence to study how the agents' forecasts of certain macroe- conomic outcomes respond to certain structural shocks (with the structural VARs). latter being identified In effect, the exercises conducted in that paper are empirical analogues of the we conducted theoretical exercise in Figure 4 for the case of productivity shocks. paper focuses on how the stochastic properties of the forecasts alone could help cific However, that apart spe- tell formalizations of the informational fictions and ignores the fact that the forecasts and actual Here, instead, we propose that the emphasis should be shifted outcomes are jomtly determined. from the details of the underlying informational forecasts '^''in frictions to the joint stochastic properties of these and the actual macroeconomic outcomes. particular, Hellwig and Venkateswaran (2009) assume that workers are perfectly informed about the monetary shocks, so that nominal wages adjust one-to-one with them. elastic demands, firms cannot tell this see no good reason hitting the economy. their for firms face constant real marginal costs will and iso- move one-to-one with monetary shocks even if nominal wages have moved because of nominal or idiosyncratic reasons. Clearly, the is assuming a questionable even within the context of that paper. As for our priori that own context, workers are perfectly informed about the aggregate real shocks Furthermore, Hellwig and Venkateswaran (2009) assume that firms are free to adjust their action at no cost and at a daily or weekly frequency. When paper), this assumption serves a useful pedagogical purpose: But once that action When assumption can alone guarantee that prices whether empirical relevance of this assumption we by specific is interpreted as a real employment that action it is interpreted as a nominal price (as in that helps isolate information frictions from sticky prices. or investment choice (as in our model), this assumption stops making sense even from pedagogical perspective: the "stickiness "of real employment and investment decisions is merely a matter of technology in the model. 38 6 Efficiency The positi^'e properties we have documented However, their normative content are intriguing. is unclear. Is the potentially high contribution of noise to business-cycle fluctuations, or the potentially high inertia in the response of the economy to innovations in productivity, a More generally, the information that But information. is it is this obvious that a planner could improve welfare symptom if of inefficiency? he could centralize dispersed in society and then dictate allocations on the basis of would endow the planner with a power that seems powers that policy makers have in reality. far all this all remote from the Furthermore, the resulting superiority of centralized allocations over their decentralized equilibrium counterparts would not be particularly insightful, since it would be driven mostly by the assumption that the planner has the superior power overcome the information frictions (2007a. 2009) and Angeletos and to imposed on the market. Thus, following Angeletos and Pavan La'O both practical and conceptual grounds (2008), — is we contend that a more interesting question to understand whether a planner could improve —on upon the equilibrium while being subject to the same informational frictions as the equilibrium. This motivates us to consider a constrained efficiency' concept that permits the planner to choose am- resource-feasible allocation that respects the geographical segmentation of information economy in the — by which we simply mean that the planner cannot make the production and employment choices of firms and workers in one island contingent on the private information of another island. A formal definition of this efficiency concept and a detailed analysis of efficient allocations can be found, for a variant model, in Angeletos and La'O (2008). Here we focus on the essence. Because of the concavity of preferences and technologies, sumption across households, Using these facts, as well as : 5^, x [ 5n JSn — M+ » , of local production and employment Q Sq : ^> strategies, q R+ , so as to j^^s(u)n{iv,nt-i)'+'dnt{ij)\dp{nt\n,^i Js^ I in con- are interested in as follows. and an aggregate output function, \u(Q(nt,nt-i))- f symmetry across firms and workers within any given island. we can represent the planning problem we Planner's problem. Choose a pair R+ and n symmetry efficiency dictates S^ x Sq ^ maximize (221 J subject to q{u.',nt-i) Q(n,.n,_i) = A{oj)n{u,nt-if q{ijj,rtt-i) p dQt(^) where P(Q(jf](_i) denotes the probability distribution of 39 y[u.'Mt-i) fit V(Q,,n,_i) conditional on ^t-i- (23) (24) This problem has a simple interpretation. the representative household; jS{Lo)n{u>,flt-iY worker in U{Q{rit,^t-i) is the utility of consumption for the marginal disutility of labor for the typical is a given island; and the corresponding integral is the overall disutility of labor for the representative household. Furthermore, note that, once the planner picks the production strategy q, the employment strategy n down by (23), is pnmed down by The reduced-form (23) objective in (22) is and the aggregate output function Q is pinned thus a functional that gives the level of welfare implied by any arbitrary production strategy that the planner dictates to the economy. Because down by this problem strictly concave, is it has a unique solution and this solution is pinned the following first-order condition:"^ SitiT-lt = git OA^n^r)- U'iQt) Ej, (25) This condition simply states that the planner dictates the agents to equate the social cost of employ- ment in their island employment. with the Essentially the expectation of the social value of the marginal product of that local same condition characterizes The only sj'mmetric-information paradigm. on the commonly-available information information As with equilibrium, we can use = q^t is standard, that there expectations are conditional while here they are conditional on the locallj'-available ,, ..,;. .,- sets. set, difference (first-best) efficiency in the ,/?..;;.:,,' Aun^^t ^° eliminate • .:.'-' nu in the above condition, thereby reaching the following result. Proposition Let 5. r(u;) = log{0i^ g+^-l ( A{lj)\ f+T-i A(w)\ SH S{^^) be a the fixed point to the following: \ogq{u,t,nt-i) IS = (l - "^{iOtMt-i (26) A number of . a)f*{u;t) + a]oglE\Q(ntMt-i)'"' uJt,^t-i]^'] role for the efficient allocation as the simplicity, p~\ q{Lj,nt^}) remarks are worth making. Because of the continuum, the we bypass the almost : ' Q(a,a_i) = " S^ x Sq composite of the local productivity and taste shocks. The efficient strategy q First, p is ft (27) /,* played for the equilibrium: determined only qualification throughout the paper. 40 V(Q,,Q,_i). note that the composite shock composite shock efficient allocation dQt{^) for almost every uj. plays a similar it identifies the For expositional fundamentals that are relevant from the planner's point of view. This above result, but also directly from the planner's problem: using expression for welfare given of the production strategy m qt = we can express the planner's problem, and the composite shock f* is evident, not only Afuf to eliminate much what Proposition alike 1 alone. did for equilibrium: coincides with the Bayes-Nash equilibrium of a economy and islands of the game in the rit welfare as a simple function Second, note that Proposition 5 permits a game-theoretic interpretation of the tion, from the in efficient alloca- economy the efficient allocation of the which the different players are the different their best responses are given by (26). Third, note that, apart from the different composite shock, the structure of the fixed point that characterizes the efficient and the equilibrium allocation with f{u.'t), f*{wt) = condition (26) coincides with for every f{ijJt) Corollary ujt if and only if is the same: once equilibrium counterpart, condition its there is is replace efficient, f*{iOt') And because (7). no monopoly power, the following In the absence of m.onopoly distortions, the equilibrium 3. we immediate. is no matter the information structure. This result estabhshes that neither the presence of noise nor the dispersion of information are per se sources of inefficiency. This result might sound bizarre in light of our earlier results that the economy can feature extreme amplification elTects, with a tiny aggregate fluctuations. effects is However, it amount of noise contributing to large What should be ex post obvious. causes these large positive the combination of dispersed information and strong complementarity. But neither one introduces a wedge between the equilibrium and the planner. Indeed, the geographical segmentation of information is similar to a technological constraint that impacts equilibrium allocations in a completely symmetric way. and technologies, not any type of market As for the inefficiency, complementarity, We can generalize this result of noise, for situations it origin is It follows that, efficient preferences guaranteeing that private motives nating economic activity are perfectly aligned with social motives. complementarity amplifies the impact it's and in coordi- when stronger does so without causing any inefficiency.^^ where firms have monopoly power, to the extent that there are no aggregate shocks to monopol}' power, as follows. Corollary 4. Suppose that information mark-up shocks (f* — ft is fixed). is Gaussian (Assumption Then, the the business cycle logQt — log Q* between the equilibrium and If But we allow this is for mark-up shocks, then earlier, this is is efficient in the sense the the efficient level of output is gap invariant. clearh' the equilibrium business cycle ceases to be efficient. true irrespectively of whether information As mentioned holds) and. there are no aggregate 1 is dispersed or commonly the opposite of what happens in Morris and Shin (2002). 41 shared. We conclude that the dispersion of information per se competitive RBC for New-Keynesian model. We further discuss the implications of optimal pohcy and the social value of information this result for We or a monopohstic not a source of inefficiency, whether one considers a is in Angeletos and La'O (2008). conclude this section with an important qualification. While our efficiency results allowed an arbitrary information structure, they restricted the information structure to be exogenous to the underlying allocations. This ignores the possibility that information gets endogenously ag- gregated through prices, macro indicators, and other channels of social learning an important omission. We — which who clearly address this issue, too, in Angeletos and La'O (2008), by allowing information to get partly aggregated through certain price and quantity indicators. that a planner is We first show internalizes the endogeneity of the information contained in these indicators will choose a different allocation than the equiUbrium. This typically means that the planner likes to increase the sensitivity of allocations to private information, so as to increase the precision of the information that gets revealed by the available macroeconomic indicators. We then explore policies that could help in this direction. Concluding remarks 7 The pertinent macroeconomics literature has used informational frictions to motivate agents may happen, or choose, to be partly times the informational friction main modeling is why economic unaware about the shocks hitting the economy. Some- exogenous, sometimes role of informational frictions it is endogenized. Invariably, though, the seems to remain a simple and basic one: to limit the knowledge that agents have about the underlying shocks to economic fundamentals. Our approach, very distinct instead, seeks to highlight that the heterogeneity of information mark on macroeconomic outcomes than highlighted this in this paper by showing inertia how the uncertainty about fundamentals. In Angeletos when We the heterogeneity of information can induce significant m the response of the economy to productivity shocks, driven fluctuations, even may have a and can also generate significant noise- the agents are well informed about the underlying fundamentals. and La'O (2009b), we further show that the heterogeneity the door to a novel type of sentiment shocks — namely shocks of information can open that are independent of either the underlying fundamentals or the agents' expectations of the fundamentals and nevertheless cause variation in the agents' forecasts of economic activity and thereby in actual economic activity, despite the uniqueness of equilibrium. This in turn permits a broader interpretation of stood for in what noise the present paper: noise could be interpreted more generally as any variation in the forecasts of economic activity that is orthogonal by fundamentals. 42 In this paper, we focused on the dispersion of information omy, ruhng out sticky prices and dismissing any lack of monetary however, does not policy. This, persed information and nominal to assume common knowledge mean frictions. It we monetary monetary policy real shocks hitting the econ- common knowledge about we policy. is innovations to no interesting interaction between see only means that of the current interaction between our approach and that about the find it a good modeling Where we dis- benchmark instead see an intriguing when there the following dimension: is dispersed information about the underlying real shocks hitting the economy and nominal prices are rigid, the may be response of monetary policy to any information that becomes available about these shocks crucial for emphasized at a how the economy responds more abstract level to these shocks in the first place. This point by Angeletos and Pavan (2007b, 2009) and is was first further explored by Angeletos and La"0 (2008) and Lorenzoni (2009) wdthin new-Keynesian variants of the economy we have studied We in this conclude with a paper. comment on the alternative formalizations of informational frictions. For certain questions, one formalization might be preferable to another."' For other questions, however, the specifics of any particular formalization we have emphasized on the in this may prove unnecessary, or even distracting. tion structure matters for economic We To highlight outcomes only through would thus its this, we showed that the informa- impact on the agents' forecasts of invite other researchers not to commit to any partic- ular formalization of the information structure (includmg ours), but rather to take a approach to the modeling results paper appear to hinge only on the heterogeneity of information, not specific details of the information structure. aggregate economic activity. The of informational frictions. After all, more flexible the data cannot possibly inform us about the details of the information structure. What, instead, the data can do the stochastic properties of the agents' forecasts of economic activity is to inform us about — which, as mentioned, is the only channel through which the dispersion of information matters of economic behavior. Thus, in our view, it is ^For example, only this evidence that should help discipline the theory. if one wishes to understand which particular pieces of information agents are attention to, Sims (2003) offers an elegant methodology. 43 likely to pay more Appendix Proof of Proposition sion in the main 1. The characterization of equilibrium follows directly from the discus- tlie and uniqueness can be obtained by showing that the equilibrium text. Its existence coincides with the solution to a concave planning problem. For the case that there power result [t] = from our analysis oo), this follows directly can be obtained for the case and in Section 6 in no monopoly is Proposition A 5. similar with monopoly power. Proof of Proposition 2. This follows from the discussion Proof of Proposition 3. Suppose main text. on Uf and ^t-i, Qi^t,^t-i) that, conditional with variance independent of wt; that this in the log-normal, is true under the log-normal structure for the underlying is shocks and signals we will prove shortly. Using log-normality of Q in condition (7), we infer that the equilibrium production strategy must satisfy condition (9) with consi = - f - - Var[logQ(Q(,n(_i)|cj£,nt_i] 7J = and Var[logQ(r!f,nt_i)K,Q(_]] We now output is '^'^ fi,_i)|fi,_i]. = ipo + f-ift-i +ipfft Suppose the equilibrmm production + ipx'Xt + ^yyt^ for some strateg)' takes a log-linear coefficients {(f-i,ipf,Lpx,'fiy). Aggregate then given by logQ{ntMt-i) = ^o + where .. guess and verify a log-linear equilibrium under the log-normal specification for the shock and information structure. form: logqt Var [logQ(n,, \\\._ = ipo + ^ '^-ift-i It (^^ j + i'-Pf + ^x}ft + ^yyt follows that Q{Qt,^t-i) (28) is indeed log-normal. with E[logQ(n,,fit-i)|wt,nf-i] = ^'Q Var[\ogQ{^t,^t-i)\^t,^t-i] = {^f where E [/t|wi,nt-i] = ^^j+^fqr^^V'/t-i + + i^-Jt-i + {'^f + ^S~{ \Kf + 'f,)^[ft\iOt,^t-i]+Vyyt T"^-+ -r fix l^y (29) (30) ) / ~^xt + ^j^^^^y,- Substitutmg these expressions into (9) gives us \ogq{ujt,^t-i) = const+ +a{ipf {I + - a) f Lpx) [uj) + a{tpQ' —+ +— \ Kf Kx K.y 44 + 'fi-ift-i+ipyyt) iift-i + —+ K.f Kx — + Ky X, + —+ Kf ' Kx — + Hy yt For this to coincide with \ogq and (a;) = ipo + ip-ift-i+fff + fxX + ifyy sufficient that the coefficients {(fo, f-i, (po — const ipf = I <Px = aiff + aifQ + Px = iPy a{ipf !</ + + ipx and Kx + . Kf V' ifx) ^j solution to this system for {ip-\,ipf,px^Py) "t~ ^x ^y "r , the one given in the proposition; is lpq is equation of this system along with the definition of const first p'q. Proof of Proposition and 0; /?. necessary K.7, + a{ipf + Q(Py then uniquely determined from the y), it is Py) solve the following system: , The unique every {f.x, —a = Q(p-i+ p-i 'Pf, ^Px, for — 1. * That finally, is, The 4. take by a result follows 00. It is First, take triple limit. a — > 1; easy to check that this triple hmit implies k the precision of the agents posterior about the fundamentals (the next, take ^ .do and mean squared forecast error) converges to zero, while the fraction of the high-frequency variation in output that is due to noise converges to Kalman filtering for similar to that found in State Vector and of the economy dynamic extension. The method we t of Motion. are Claim. The dynamics of the /( We is guess and verify that the relevant aggregate state variables and logQ( and thus define state vector economy Xt are given by the following law of — MXt-\ + m.„i't with + p M i' A'( in (18) accordingly. motion m^St (31) -1 -1 r- 1 ,m„ s A/21 The use in solving this equilibrium Woodford (2003b). Law at time 100%. ,m£ M, = m„2 (32) nie2 coefficients {M2i,M22,my2,'rn.^2) cire given by M21 = ^-(^'21 M22 = ^ (1 rnu2 - 1 - ^2 = aK:. Q' + A'22) (33) - A'22) (34) - (35) /\2] (1 - A'21 A'22) (36) 45 and K= K22 K2\ is the matrix of K = E [{Xt We kalman - gains, defined by E,,(_i [Xt]) - {zi,t E,,t-i In each period Observation Equation. and public E [(z,,f - E,,t_i [z^,t]) (^,,t and describe the procedure verify this claim in the following in (19), of private [^^,t])'] firms and workers on island i, ~' E,,t_i (37) [z,,t])'] for finding the fixed point. i observe vector z^f as and error terms, island In terms of the aggregate state signals. - i's observation equation takes the form ^\' ' 1 Xt + Zi,t + Qf where Cj is defined as a column vector of length two where the j-th entry • are - 0. ^ : Forecasting and Inference. Island E,; where E^^t-i [Xt] is island's = follows that Ei,t_i [X,] i's t — ME,,f_i 1 i's t i-1 . — 1 forecast of — l-t.tj kalman island use the filter to is the 2x2 .; zl is E^,i-l forecast of Xt. given by . all other entries ,••,- , ,. , . , [A^t] Combining this with the law of motion (31), it '• [X,._i]. update their = E,:, f^i of the current state, firms forecasts. [Xt] K + matrix of Kalman gams, defined Updating [z,,t - is E,,f_i and workers on each done via [z,^t]) (39) , in (37). Substitution of island i's i — 1 forecast ' , E,,t [Xt] = and • oi z] into (39) gives us Let Et [Xt] 1 ' E,,, [Xt] K is ' To form minimum mean-squared-error estimates where (38) £( 1 .''^ Jj Ej_< [Xt] di ^\I^K be the time t ME,,(_i [Xt-i] + A'2,,t average expectation of the current state. Aggregation over (40) implies ^ E, [A', I - K (40) ^ e' MEt-i 46 [Xt-i] + K / z,,tdi Finally, using the fact that aggregration over signals yields / Xt 2,,fdz + £(, it follows that the average expectation evolves according to K % \Xt MXt^i + +K VlyVt + I \ K - ME,_i [AVi] "e'l" K + nir (41) £( 1 e'l _ where AI.viy,m^ are given by (32). Characterizing Aggregate Output. Local output hke condition in (17), over this condition, we which may be rewritten as find that aggregate log = Qt [{l-a)i> + +aM22Et_i ai) (A'2i = - (1 determined by the best-responsea) ft + Qe2Ej,t [Xt]. Aggregating output must satisfy from [A'(] log(j,,t is = [l~a)ft+ae'2Et[Xt] \ogQt Substituting our expression for E( each island in + [logQ,_i] (41) into (42), gives us A'22)] It-\ + (42) [(1 - + +Q q) - [oMsi (A'21 + ail' {K21 A'22)] vt = Moreover, rearranging condition (42), we find that E( [logQj] + + K22)] Et-i [ft-i] QK22et ^ (logQt — (1 — cv) ft). Finally, using this condition in the above equation gives us logQt = [(1 -a)^' + + [aM2i - aV'(Ar2i atp (A'21 +A'22)-^/22(l-Q)]/t-l +-A/22logQ(-l + A'22)] Et-i [ft-i] For this to coincide with the law of motion conjectured it is necessary and sufficient tliat Af2i = (1 mv2 — 1 m^2 = cvA'22 - Q')^ + - Q +Q a'!^'(A'2i (A'21 + in (31) - a+a and (A'21 + A"22)l Vt + aK22£t (32) for every [ft-\..\ogQt^i.vt,et), atl' (A'21 Therefore, given the Icalman gains matrix A', + A'22) - Af22 (1 - a) A'22) + solution to this system for (M21, A/22, motion of Xt [I the coefficients [M2i,M22,rny2,m.e2) solve the following system: = QM21 — The unique + A'22) '^1^2^ "^£2) we can uniquely . 47 is the one given in the proposition. identify the coefficients of the law of Kalman Filtering. Let us define the variance-covariance matrices of forecast errors as V = E[{Xt-E,A^t]){Xt-E,,t\Xt]y] These matrices be the same for will islands all i, since their observation errors are have the same stochastic properties. Using these matrices, we may write K assumed to two as the product of components: - Ea-i E, [{Xt [Xt]) iz,,t - Kt-i [z^,t]y] =£ 6] CTrTTT-E ei 1 and E^ - [{Z,^t 1 " lz^,t]) (z«,( E^,t-l E,,t_i [z,,,])'] ei + ei (43) crl ' ' e'l + C7t m.; + 1 m'. ei ei + 1 K Therefore, is given by K where a^ = Finally, and V E, 1 - [(z,,( E,,f_i what remains = E ei [zi^t]) {zi,t - to determine a'^m.^ ei 1 )k?)" (44) ' E,,f_i [2i,(])'] is the matrix S. is given by (43). The law of motion implies that matrices S satisfy E = MVM' + alm.ym'.^, + a'^'iriem'^, In addition, the forecasting equation (40) imply these matrices V=E ei ei \ + ON cx^m^ must further satisfy -1 + 1 (^; 1 Combining the above two equations, we obtain the stationary E = MT.M' - M +almyTn'^ where M, ruy, m^ E ei ei a'^m^ 1 Ricatti Equation for E: {^'X' rn _- + a^m^m'^ are functions of the (45) kalman gains matrix K, and K is itself a function of and m^. The variance-covariance matrix E, the kalman gains matrix K, and the law matrices A/, m„, by (33)-(36), m^ (44), M' of E motion are thus obtained by solving the large non-linear system of equations described and (45). This system is too complicated to allow further analytical results; thus solve for the fixed point numerically, 48 we Proof of Proposition is unique and is 5. The planner's problem pinned down by its is strictly convex, first-order conditions. guaranteeing that The Lagrangian of this written as its solution problem can be • , , . , = A Js„ + The A(a) / Q{^) and first-order conditions with respect to q{uj) are given bj' the following: i7'(Q(Qt,nt-i)) + A(fi()(^^)Q(Q(,Qt_i)-p = 0(46) = (47) P -^5(w)e-'^°9(c..,n,_i)^'Sn where p-l x{n, q{uj.nt-i) p J^{9.t\uj.nt^i) L J^ {Qt\^,Qt-i) denotes the posterior Restating condition (46) as Xi^t) (^j = into condition (47), gives condition (26), about Qt (or. equivalently, about ft -U' {Q(9.tMt-i)) Qi^tMt-i)" and and yt) given wt- substituting this which concludes the proof. References [1] Adam. Klaus etary [2| (2007), "Optimal Economics monetary policy with imperfect common knowledge?, "Journal of Mon- 54, 276-301. Amador, Manuel, and Pierre-Olivier Weill (2007), "Learning from Private and Public Observations of Others' Actions," Stanford University/UCLA mimeo. |3| Amador, Manuel, and Welfare,'' Stanford |4| Amato, Jeffery. Pierre-Olivier Weill (2008). "Learning from Prices: Public Communication and University/UCLA mimeo. and Hyun Song Shin (2006), "Imperfect Common Knowledge and the Information \'alue of Prices, " Econom.ic Theory 27, 213-241. [5| Angeletos, George-Marios, and Jennifer La'O (2009a), "Incomplete Information, Higher-Order Beliefs, and Price Inertia," MIT mimeo, "MIT mimeo. (6] Angeletos, George-Marios, and Jennifer La'O (2009b), "Sentiments, |7] Angeletos, George-Marios, and Jennifer La'O (2008), "Dispersed Information over the Business Cycle: Optimal Fiscal and Monetary Policy with Dispersed Information, "MIT mimeo. 18| Angeletos, George-Marios, Guido Lorenzoni, and Alessandro Pavan (2008), "W'all Street and Silicon Valley: A Delicate Interaction," MIT/Northwestern mimeo. 49 |9| Angeletos, George-Marios, and Alessandro Pavan (2009), "Policy with Dispersed Information, "Journal Economic Association of the European |10| Econometrica 75:4, 1103-1142. and Policy Implications, "Journal of the European Economic Association Bacchetta, Philippe, and Eric van Wincoop (2005), Baisu, "Can Information Heterogeneity Explain the Ex- 2, 1-32. "American Economic Review 51, 1183-1216. Beaudry, Paul, and Franck Portier (2006), "Stock Prices, News, and Economic Fluctuations," American Economic Review |18| 96, 1418-1448. Beaudry, Paul, and Franck Portier (2004), "Exploring Pigou?s Theory of Cycles," Journal of Monetary Economics |17| Benhabib, Jess, 96, 1293-1307. Richard Rogerson, and Randall Wright (1991), "Homework hold Production and Aggregate Fluctuations, "Journal of Political |19| Bernanke, Ben, and nomics |20| Ilian Mihov (1995), in Economy Macroeconomics: House99, 1166-1197. "Measuring Monetary Policy," The Quarterly Journal of Eco- 113, 869-902. Blanchard, Olivier, Jean-Paul L'Huillier, and Guido Lorenzoni (2009), "News, Noise, and Fluctuations: An |21| 95. Susanto, John Fernald, and Miles Kimball (2006), "Are Technology Improvements Contrac- tionary?, |16| 94, 91-98. Barro, Robert (1976), "Rational Expectations and the Role of Monetary Policy," Journal of Monetary Economics [15| 585-593. Economies with Investment Complementarities," American Economic Review change Rate Determination Puzzle?, "American Economic Review [14| 5, Angeletos, George-Marios, and Alessandro Pavan (2004), "Transparency of Information and Coordination in |13| " Angeletos, George-Marios, and Alessandro Pavan (2007b), "Socially Optimal Coordination: Characterization |12| 11-60. Angeletos, George-Marios, and Alessandro Pavan (2007a), "Efficient Use of Information and Social Value of Information, |11| 7, empirical exploration, Blanchard, Olivier, and "MIT mimeo. Danny Quah (1989), "The Dynamic Effects of Aggregate Demand and Supply Disturbances," American Economic Review 79, 655-673. [22| Chari, V. V., Patrick Kehoe, and Ellen McGrattan (2007), "Business Cycle Accounting," Econometrica 75, 781-836. |23] Christiano, Lawrence, Martin Eichenbaum, and Charles Evans (1999), "Monetary Policy Shocks: Have We Learned and to What End?," in J. B. Taylor & M. Woodford , eds., Handbook What of Macroeco- nomics, Elsevier. |24| Christiano, Lawrence, Martin Eichenbaum, and Robert Vigfusson (2003), "What Happens After A Technology Shock?" Board of Governors of the Federal Reserve System, International Finance Discussion Papers No. 768. 50 |25| Cosmin Hut, Roberto Motto, and Massimo Rostagno Christiano, Lawrence, and stock market boom-bust |26| UC Working Paper No. 955. "What can survey of "New Perspectives on Monetary Cambridge University Gall. Jordi, Gilchrist, Synthesis," CNRS- S. Turnovsky (eds.), Policy, Inflation, Advances m and the Business Cycle, Economic Theory, vol. "in M. HI, 151-197, Press. and Pau Rabanal (2004), "Technology Shocks and Aggregate Fluctuations: "NBER How Well Does Macroeconomics 2004. Simon, and John Leahy (2002), "Monetary Policy and Asset Prices, "Journal of Monetary Economics |32| A 89, 249-271. the Real Business Cycle Model Fit Postwar U.S. Data?, |31| Money: Employment, and the Business Cycle: Do Technology Shocks Explain Gall, Jordi (1999), "Technology', Gall, Jordi (2003), us about informa- tell Bern mimeo. Dewatripont, L. Hansen, and 130| forecasts Collard, Fabrice, and Harris Delias (2005), "Misperceived vs Unanticipated Aggregate Fluctuations?," American Economic Review |29| "Monetary policy Berkeley mimeo. GREMAQ/University |28l ECB Coibion, Olivier, and Yuriy Gorodnichenko (2008), tional rigidities?" |27| cycles," (2008), 49, 75-97. Goodfnend, Marvin, and Robert King (2001), "The Case for Price Stability, "NBER Working Paper 8423. |33| Goodfriend, Marvin, and Robert King (1997), "The tary Policy, |34] Hall, Neocla.ssical Synthesis 1997, Cambridge, MIT and the Role of Mone- Press. Robert E. (1997), "Macroeconomic Fluctuations and the Allocation of Time, "Journal of Labor Economics, |35| ''NBER Macroeconom.ics Annual New 15, S223-S250. Hansen, Gary (1985), "Indivisible Labor and the Business Cycle, "Journal of Monetary Economics 16, 309-337. |36| Hellwig, Christian (2002), "Public Announcements, Adjustment Delays, and the Business Cycle, "UCLA mimeo. |37| Hellwig, Christian (2005), "Heterogeneous Information and the Welfare Effects of Public Information Disclosures, |38| Hellwig, Christian, and Laura in |39| Veldkamp (2008), "Knowing What Others Know: Coordination Motives Information Acquisition, "Review of Economic Studies, forthcoming. Hellwig, Christian, and "UCLA |40| "UCLA mimeo. Venky Venkateswara (2008), "Setting The Right Prices For The Wrong Reasons, mimeo. Jaimovich, Nir, and Sergio Rebelo (2006), "Can News About the Future Drive the Business Cycle?," Stanford /Northwestern mimeo. 51 |41| King, Robert (1982), "Monetary Policy and the Information Content of Prices, "Journal of Political Economy |42| 90, 247-279. King, Robert and Sergio Rebelo (2000), Woodford, |43| eds.. ''Resusitating Real Business Cycles," in J. B. Taylor Handbook of Macroeconomics, & M. Elsevier. Kydland. Finn, and Edward Prescott (1982), "Time to Build and Aggregate Fluctuations, " Econometrica 50, 1345-70. Demand Shocks, "American Economic Review, forthcoming. |44| Lorenzoni, Guido (2008), "A Theory of |45] Lorenzoni, Guido (2009), "Optimal Monetary Policy with Uncertain Fundamentals and Dispersed Information, "Review of Economic Studies, forthcoming. |46| Lucas, Robert E., 4, [47| Lucas, Robert E., Jr. (1975), "An Equilibrium Model 11, 366-385. Mackowiak, Bartosz, and Mirko Wiederholt (2008), "Optimal Sticky Prices under Rational Inattention, Mankiw, N. Gregory and Ricardo Reis New (2002), "Sticky Information Versus Sticky Prices: A Proposal to Keynesian Phillips Curve," Quarterly Journal of Economics 117, 1295-1328. Mankiw, N. Gregory and Ricardo Reis 164-169. [53| ; "European Central Bank/Nothwestern University mimeo. Replace the |52] . . Mackowiak, Bartosz, and Mirko Wiederholt (2009), "Business Cycle Dynamics under Rational Inattention, |51| Econ- Luo, Yulei (2008), "Consumption Dynamics under Information Processing Constraints, "Review of Eco- "American Economic Review, forthcoming. |50| of the Business Cycle," Journal of Political 83, 1113-1144. nomic Dynamics |49| "Expectations and the Neutrality of Money," Journal of Economic Theory 103-124. omy (48| Jr. (1972), .- 96, - • . Economic Review (2006), "Pervasive Stickiness, " Am,erican " McGrattan. Ellen (2004), "Comment on Gali and Rabanal," Federal Reserve Bank of Minneapolis, Research Department Staff Report 338. |54| Moscarini, Giuseppe (2004), "Limited Information Capacity as a Source of Inertia, "Journal of Economic ' Dynamics and Control |55| Morris, Stephen, and Economic Review |56| |57| Hyun Song Shin (2002), "The Social Value of Public Information, "American Hyun Song Shin (2006), "The Inertia of Forward-Looking Expectations, "American 96. 1521-1534. Nimark, Kristoffer (2008), "Dynamic Pricing and Imperfect Economics ' ., 92, 1521-1534. Morris, Stephen, and Economic Review 28, 2003-2035. 55, 365-382. 52 Common Knowledge, "Journal of Monetary Edmund Phelps. in (1970), "Introduction: A. A. Alchian, C.C. Holt et New The New Microeconomics in Employment and Inflation Theory," Microeconomic Foundations of Employment and Inflation Theory, al., York: Norton. Prescott, Edward (1986). "Theory Ahead of Business Cycle Measurement, '^Federal Reserve Bank Min- neapolis Quarterly Review 10, 9-22. "Review of Economic Studies Reis, Ricardo (2006), "Inattentive Producers, Reis, Ricardo (2008), "Optimal Monetary Policy Rules in 73, 793-821. an Estimated Sticky-Information Model, "American Econom.ic Journal: Macroeconomics, forthcoming. Rodina, Giacomo (2008), "Incomplete Information and Informative Pricing, "UCSD mimeo. Rogerson, Richard (1988), "Indivisible Labor, Lotteries and Equilibrium, "Journal of Monetary Eco- nomics 21, 3-16. Rotemberg, and S. Julio, and Michael Woodford (1991), "Markups and the Business Cycle," Fischer (eds.), Rotemberg, Julio, NBER Macroeconomics Annual 1991, Cambridge, in O.J. Blanchard and 1997, Cambridge, MIT Rotemberg, and Michael Woodford Taylor & O.J. Blanchard Press. and Michael Woodford (1997), "An Optimization-Based Econometric Model Evaluation of Monetary Policy," Julio, MIT in M. Woodford S. Fischer (eds.), NBER for the Macroeconomics Annual Press. (eds.), Handbook Shimer, Robert (2009), "Convergence Journal: Macroeconomics 1, (1999), "The Cyclical Behavior of Prices and Costs.," in J. B. of Macroeconomics. Elsevier. in Macroeconomics: The Labor Wedge, "American Economic 280-297. Sims, Christopher (2003), "Implications of Rational Inattention," Journal of Monetary Economics 50, 665-690. Sims, Christopher (2006), "Rational Inattention: Beyond the Linear-Quadratic Case, "American Eco- nomic Review 96, 158-163. Townsend, Robert M. (1983). "Forecasting the Forecasts of Others, "Journal of Political Economy 91, 546-588. Van Nieuwerburgh, Diversification, "NYU Van Nieuwerburgh, Stijn, and Laura Veldkamp (2008), "Information Acquisition and Llnder- Stern mimeo. Stijn, and Laura Veldkamp (2006), "Learning Asymmetries in Real Business Cycles, "Journal of Monetary Economics 53, 753-772. Veldkamp, Laura (2006), "Media Frenzies Review in Markets 96, 577-601. 53 for Financial Information, "American Economic |74| Veldkamp, Laura, and Justin Woolfers (2006), "Aggregate Shocks or Aggregate Information? Information and Business Cycle Comoveraent, |75l Woodford, Michael (2003a), "Imperfect Aghion, R. Frydman, Modem |76] J. Stiglitz, ^^ Journal of Monetary Economics 54, 37-55. Common Knowledge and M. Woodford, Macroeconomics: In Honor of Edmund Costly eds., and the Effects of Monetary Policy," in P. Knowledge, Information, and Expectations in S. Phelps, Princeton University Press. Woodford, Michael (2003b), Interest and Prices: Foundations of a Theory of Monetary Policy, Prince- ton: Princeton University Press. (77| Woodford, Michael (2008), "Information-Constrained State Dependent Pricing," Columbia mimeo. 54