Nowcasting GDP and Inflation: The Real-Time Informational Content of Macroeconomic Data Releases

Finance and Economics Discussion Series Divisions of Research & Statistics and Monetary Affairs Federal Reserve Board, Washington, D.C. Nowcasting GDP and Inflation: The Real-Time Informational Content of Macroeconomic Data Releases Domenico Giannone, Lucrezia Reichlin, and David Small 2005-42 NOTE: Staff working papers in the Finance and Economics Discussion Series (FEDS) are preliminary materials circulated to stimulate discussion and critical comment. The analysis and conclusions set forth are those of the authors and do not indicate concurrence by other members of the research staff or the Board of Governors. References in publications to the Finance and Economics Discussion Series (other than acknowledgement) should be cleared with the author(s) to protect the tentative character of these papers.

Nowcasting GDP and Inflation: The Real-Time Informational Content of Macroeconomic Data Releases∗ Domenico Giannone, ECARES and European Central Bank, Lucrezia Reichlin, European Central Bank and CEPR David Small, Board of Governors, Federal Reserve This version: September 2005 Abstract Thispaperformalizestheprocessofupdatingthenowcastandforecastonoutputandinflationasnewreleasesofdatabecomeavailable. Themarginalcontribution of a particular release for the value of the signal and its precision is evaluated bycomputing“news”onthebasisofanevolvingconditioninginformationset. The marginal contribution is then split into what is due to timeliness of information and what is due to economic content. We find that the Federal Reserve Bank of Philadelphiasurveyshavealargemarginalimpactonthenowcastofbothinflation variables and real variables and this effect is larger than that of the Employment Report. When we control for timeliness of the releases, the effect of hard data becomes sizeable. Prices and quantities affect the precision of the estimates of inflation while GDP is only affected by real variables and interest rates. JEL Classification: E52, C33, C53 Keywords: Forecasting, Monetary Policy, Factor Model, Real Time Data, Large Data Sets, News ∗ We would like to thank the Division of Monetary Affairs of the Board of Governors of the Federal Reserve System for encouragement to pursue this project and providing financial support to Lucrezia Reichlin. We thank our research assistants at the Fed, Ryan Michaels and Claire Hausman, and Michele Modugno at ECARES, Univeriste Libre de Bruxelles. Thanks arealsoduetoDavidWilcoxandWilliamWascherfortheircomments,toseminarparticipants at the Fed in April 2004 and to our discussant Athanasios Orphanides at the EABCN conference in Brussels in June 2005. The opinions in this paper are those of the authors and do not necessarily reflect the views of the European Central Bank or the Federal Reserve System. Please address any comments to Domenico Giannone dgiannon@ulb.ac.be; Lucrezia Reichlin lucrezia.reichlin@ecb.int; or David Small, dsmall@frb.gov.

1 Introduction Monetary policy decisions in real time are based on assessments of current and future economicconditionsusingincompletedata. Sincemostdataarereleasedwithalagand aresubsequentlyrevised,thereconstructionofcurrent-quarterGDP,inflationandother key variables is an important task for central banks and one to which they devote a considerableamountofresources. Current-quarternumbersarealsoimportantbecause, in the short-run, there is a greater degree of forecastability than in the long run. For example, Giannone, Reichlin, and Sala (2004) (GRS from now on) document that, in forecasting GDP beyond the first quarter, the forecasts of the Federal Reserve staff and of standard statistical models do not perform better than that of a constant growth rate. Current-quarter estimates are particularly relevant because they are inputs for model-based longer term forecasting exercises. Nowcastsareconstructedatcentralbanksusingbothsimplemodelsandqualitative judgment. Those exercises involve the analysis of a large amount of information and a judgment on the relative weight to attribute to various data series. As new information becomesavailablethroughoutthemonth,thenowcastsandforecastsmaybeadjustedin responsetochangesinboththevaluesofthedataseriesandtheimplicitrelativeweights applied to those series. Typically, central banks and markets pay particular attention to certain data releases either because they arrive earlier, and can therefore convey news on key variables such as GDP, or because they are inputs in their estimates (e.g. industrial production or the Employment Report for GDP). In principle, however, any release, no matter at what frequency, may potentially affect current-quarter estimates and their precision. From the point of view of the short-term forecaster, there is no reason to throw away any information. This paper provides a framework that formalizes the updating of the nowcast and forecast of output and inflation as data are released throughout the month and that canbeusedtoevaluatethemarginalimpactofnewdatareleasesontheprecisionofthe now/forecast as well as the marginal contribution of different groups of variables. In the empirics, we focus on the nowcast and we use intra-month releases of monthly time series to construct (possibly) progressively more accurate current-quarter estimates. Our approach allows us to consider a large number of monthly time series (in principle all the potentially relevant ones) within the same forecasting model. Moreover, the model takes into account the non-synchronicity of the releases by exploiting vintages of panel data which are unbalanced at the end of the sample. The framework we propose is adapted from the parametric dynamic factor model proposed by Doz, Giannone, and Reichlin (2005) and applied by GRS to the same variables we are using here. It is similar in spirit to Evans (2005), but our focus is different since we exploit a large number of data series rather than just financial variables and we don’t consider information at frequencies lower than the month. Usingthisframework,weaskthreespecificempiricalquestions. Thefirstiswhether a large information set really helps to obtain an early and accurate estimate of current inflation and output. Several papers have made the point that a large information set helps in forecasting (cfr. Boivin and Ng (2005), Forni, Hallin, Lippi, and Reichlin (2003), Giannone, Reichlin, and Sala (2004), Marcellino, Stock, and Watson (2003), 1

Stock and Watson (2002)). This literature proposes and applies factor models adapted tohandlelargepanelsoftimeseries. Onthebasisofsuchmodels, BernankeandBoivin (2003) and GRS formalize the real-time application of large datasets to nowcasting and forecasting inflation and output in the United States. GRS in particular show that a specification of the model with two dynamic factors has a forecasting performance comparable to that of the Federal Reserve’s Greenbook. This paper builds on this literature, but instead of performing an out-of sample forecasting exercise, we compute measures of news and uncertainty and study their evolution as new information becomes available within the month. This is achieved by deriving explicitly the standard error of the nowcast or forecast as a function of the size of the information set. Changes in this standard error allow us to track the evolution of the uncertainty of the forecast and nowcast as the flow of information evolves within a month. Thesecondquestionistheassessmentofthemarginalcontributionofparticularsets of variables in constructing the nowcasts. What kind of information really matters? To provide an answer, we update our nowcasts and forecasts following each data release within the month and construct empirical measures of the “news” in each data block by conditioning on the data that was available in real time when the data was released and that is evolving within the month. Because the data are released in blocks and the releases follow a relatively stable calendar, each month the updates and news for each type of data release are conditional on the same (updated) set of data releases. Since blocks of releases typically correspond to an economic classification: money indicators, prices, industrial production series, labor market variables etc., our measure of news refers to aggregates of variables in a certain category rather than to a single indicator. The third question is whether the marginal contribution of a block of releases is due to its “timeliness” or to its “quality.” The distinction between timeliness and quality arises because the marginal value of a data release depends on the new information in the release; i.e. it depends on the difference between the data that are released and the values that were predicted by the model just before the release. The earlier a given seriesisreleased(timeliness), thesmallertheinformationsetforitspredictedvalueand the greater, ceteris paribus, is the news in the release. Its “quality” depends on the predictive power of an information block given the same conditioning information set as for other information blocks. Since data are very collinear, the order of the release matters and we may have a situation where high quality data such as GDP, have no marginal impact on GDP itself since they are released with a long lag. The paper is organized as follows. In Section 2, we describe the problem and the structure of the staggered releases in the United States. In Section 3 we introduce the model, our estimation technique, the computation of the standard errors, and the method for examining the “timeliness” of data. Section 4 describes the empirical analysis and comments on the results. Section 5 concludes. 2

2 The Problem and the Structure of the Data Sets 2.1 The Problem We will first describe the problem we are analyzing in a very stylized way. Our aim is to evaluate the current quarter nowcast of key indicators of real economic activity and price dynamics on the basis of the flow of information that becomes available during the quarter. Within each quarter, contemporaneous values of key macroeconomic variables like GDP are not available, but they can be estimated using higher frequencies variables which are recorded and published more timely. At month v we can define the relevant information set Ωn which includes the relevant n monthly time series and the relevant v sample up to month v and compute the following projection: Proj[GDP | Ωn]. t v Let us assume that Ωn is composed of two blocks [Ωn1 Ωn2] and that the variables v v v in Ωn2, say production, are released a month later than those in Ωn1, say surveys. v v This implies that, in month v, variables in Ωn1 are available up to month v, while v variables in Ωn2 are available up to month v−1. In order not to lose the information v in Ωn2 available up to the previous month, we will have to project on the basis of a v dataset which is unbalanced at the end of the month. Our forecasting problem is the generalization of this simple case. The conditioning set in the projection is a large panel of monthly time series, consisting of about 200 series for the US economy, broadly those examined closely by the staff of the Federal Reserve when making the forecasts. The data considered are published in thirty six releases per month. The blocks contain direct measures both of real economic activity and prices, and of aggregate and sectoralvariables. Moreover, theyinclude indirectmeasures ofeconomic developments, such as surveys, financial prices that may reflect current and expected future economic developments and measures of money and credit. To set the notation, we will denote the information set by: (cid:110) (cid:111) Ω = Y ; i = 1,...,n; t = 1,...,T vj it|vj ivj where v denotes the month of the release, andv the date of the jth data release within j themonth. Ateachpointintimev ,wewillrefertotheinformationsetasvintage. The j latter is composed n variables, Y , where i = 1,...,n identifies the individual time it|vj series and t = 1,...,T denotes time in months. Here, T indicates the last period ivj ivj for which series i in vintage v has an observed value. For example, when industrial j production is released in month v, the last available observation refers to the previous month T = v−1, while when surveys are released, the last values refer to the month ivj of the releases T = v. ivj Let us now track the flow of information within the quarter of interest. We will follow the convention that a quarter k is dated by its last month (for example, the first quarter of 2005, is dated by k =March05). Release j within each quarter k is given 3

by Ω where v = k −2,k −1,k, are, respectively, the first, the second and the third vj month of quarter k. At v , a set of variables Y ,i ∈ I is released and the information set expands j i,t vj from Ω to Ω . The new information set differs from the preceding one for two vj−1 vj reasons. First, there are new, more recent, observations: T ≥ T ,i ∈ I , while ivj ivj−1 vj T = T ,i ∈/ I . Second, old data are revised, and data revisions are given by ivj ivj−1 vj Y −Y ,i ∈ I . Notice that in absence of data revisions Ω ⊆ Ω , i.e. the it|vj it|vj−1 vj vj−1 vj information set is expanding as time passes by. Thetimingandtheorderofdatareleasescanvaryfrommonthtomonth,i.e. I can vj be different from I , for v (cid:54)= v˜. However, releases typically correspond to an economic v˜j classification: money indicators, prices, industrial productions, labor market variables etc. andwithfewexceptions,thedifferencesinthechronologicalorderofthereleasesare limited. This allows us to construct a stylized calendar in which we combine the series into fifteen data blocks so that, in most cases, they consist of roughly homogeneous variables, containing data released at roughly the same time in the month, roughly preserving the chronological order in which the data are released. We call pseudo vintages the releases which refer to our stylized calendar. We have: I = I ,j = vj j 0,1,··· ,J. We want to stress here that, abstracting from data revisions, due to the non synchronicity of data releases, the intra month flow of data is mainly reflected in the increase of cross-sectional information. In particular, at each release date v the inj formation set expands because of the inclusion of new information about a group of variables that corresponds to a particular economic classification. For each information set within the quarter of interest, we compute the nowcast for the variables of interest by simple projection. For a generic variable zq, e.g. GDP k growth rate, where the superscript q indicates that the variables is measured at quarterly frequency, we have: (cid:104) (cid:105) z(cid:98) q = Proj zq|Ω , v = k,k−1,k−2, j = 1,...,J. k|vj k vj Once we have obtained the projections, we can compute the news in block j as the change that the release of block j induces in the current estimates of the variable of interest: NEWS[zq,v ] = z(cid:98) q −z(cid:98) q . (2.1) k j k|vj k|vj−1 Notice that NEWS is not a standard Wold forecast error. First of all, the structure of the unbalancedeness changes with time so that the number of variables within the month is different from month to month. Second, it is affected by the order in which data arrive. The uncertainty associated with this projection, is estimated by Vzq = E[(z(cid:98) q −zq)2], v = k,k−1,k−2 k|vj k|vj k Since the dataset is expanding, Vzq ≤ Vzq and the uncertainty is expected k|vj k|vj−1 to decrease as time passes by. The evolution of this quantity across data releases 4

measures the extent to which each block of releases helps reduce uncertainty of the nowcast of the variables of interest: more informative releases are expected to produce larger reductions in uncertainty. The reduction of uncertainty provides a measure of the marginal information content of the jth data release and, in general, of the value of an increasingly larger information set. From the practical point of view, the computation of this projection is not simple. Due to the large number of data we are considering, Ω is very large. The basic vj idea of this paper is to exploit the collinearity of the series in our panel to summarize the information in Ω in a smaller space generated by the span of few common factors F . A projection on the space of the common factors F is able to capture the bulk t t of the covariance of the data and provides a parsimonious well performing forecast. Our problem is split in two steps. First, estimate the factors from the panel, Fˆ = (cid:104) (cid:105) t|vj Proj F |Ω . Second, project on the span of the estimated factors. Uncertainty of the t vj nowcast can hence be attributed to two components Vzq = Vχq +Vξq . k|vj z,k|vj z,k|vj The first component reflects uncertainty on the common component, i.e. the uncertainty arising from the estimation of the common factors; the second component reflects uncertainty on the idiosyncratic, i.e. the variance of that part of the variable not explained by the common factors. Onthebasisoftheframeworkoutlinedherewewillalsostudywhethertheimpactof a release depends on the fact that it is published early (timeliness) or by its economic content (quality). Quality of a block of release is defined as its marginal impact, controlling for the date of the release. To summarize, our objectives are: 1. Update the current quarter estimate and the forecast of the variables of interest, conditioning on a large set of information. 2. Update on the basis of a panel which at the end of the month is unbalanced. 3. Evaluate “news” in relation to the publication of data releases. 4. Evaluate uncertainty in relation to the flow of information. 5. Evaluate the impact of a release by distinguishing the effect due to timing and that due to quality. On the basis of this information, we want to evaluate the marginal contribution of different blocks of variables to the forecast and assess whether the latter is due of to the timeliness of the release or to its intrinsic quality. A model that is suitable to our objectives is defined in the next Section. 5

3 The Econometric Methodology Themethodologywewillproposehereistheparametricdynamicfactormodelproposed by Doz, Giannone, and Reichlin (2005) and applied by GRS to the same variables we are using here. In this framework, once the parameters of the model are estimated consistently through principal components, the Kalman filter is used to update the estimates of the signal and the forecast on the basis of the unbalanced panels. This parametric version of the factor model can also be used to derive explicit measures of data uncertainty across the vintages. The Kalman filter allows us to extract the innovation content of each data release (composed of several individual data series) and to identify the news – splitting it from the noise. The underlying signal is computed by the Kalman filter by weighting the innovation content of each variable according to its news to noise ratio. 3.1 The Model While in Section 2 we defined the problem for a generic quarterly variable zq , in k|vj describing the model, for simplicity, we will refer to monthly stationary variables. The appendixdescribesdatatransformationandtherelationbetweenquarterlyandmonthly quantities in detail. Here let us just say that the variable of interest, y is the corit|vj responding monthly series to zq , transformed so as to induce stationarity. Obviously k|vj different transformations will be required depending on the nature of the variable in question. We have: y = µ +λ F +ξ it|vj i i t it|vj whereµ isaconstantandχ ≡ λ F andξ aretwoorthogonalunobservedstochastic i it i t it|vj processes.1 In matrix notation we can write: y = µ+ΛF +ξ = µ+χ +ξ t|vj t t|vj t t|vj where y = (y ,...,y )(cid:48), ξ = (ξ ,...,ξ )(cid:48), Λ = (λ(cid:48),...,λ(cid:48) )(cid:48). We assume t|vj 1t|vj nt|vj t|vj 1t|vj nt|vj 1 n that the n×1 process χ (the common component) is a linear function of a few unt observed common factors F that capture “almost all” comovements in the economy, t while the n×1 stationary linear process ξ (the idiosyncratic component) is driven t|vj by n variable-specific shocks. Since data revision errors are typically series specific, we incorporate them in the idiosyncratic component. Additionally, the common factors are supposed to be the same across releases because they summarize the fundamental state of the economy underlying all data releases. The common and idiosyncratic components are identified under the methodology and assumptions used in estimating the model, as described in section A.3 of the Appendix. Thecommon factorscanbeconsistentlyestimatedbyprincipalcomponents (See Forni, Hallin, Lippi, and Reichlin (2000) and Stock and Watson (2002)) provided that the idiosyncratic shocks exhibit, at most, “weak” cross-correlations. 1The particular transformations that we use are discussed in Section C of the Appendix. 6

Our approach is to specify the the dynamics of the common factors as follows:2 F = AF +Bu (3.2) t t−1 t u ∼ WN(0,I ) (3.3) t q whereB isar×q matrixoffullrankq,Aisar×r matrixandallrootsofdet(I −Az)lie r outside the unit circle, and u is the shock to the common factor and is a white-noise t process. In such a model, a number of common factors (r) that is large relative to the number of common shocks (q) aims at capturing the lead and lag relations among variables along the business cycle (cfr. Forni, Giannone, Lippi, and Reichlin (2005) for details). In the empirical estimates, r and q will be set equal to ten and two, respectively. ThesechoicesarebasedonfindingsinGRSandcorrespondtotheideathattheeconomy can be described as being driven by q = 2 large pervasive shocks with heterogeneous dynamics captured by the parameter r. To estimate the factors on the basis of an unbalanced data set, for the idiosyncratic shock we assume: (cid:40) ψ if y is available E(ξ2 ) = ψ˜ = i it|vj (3.4) it|vj i ∞ if y is not available. it|vj The data generating process of the idiosyncratic components is parameterized by specifying, for available vintages, the following conditions: E(ξ ξ(cid:48) ) = diag(ψ˜ ,...,ψ˜ ) (3.5) t|vj t|vj 1 n E(ξ ξ(cid:48) ) = 0,s > 0. (3.6) t|vj t−s|vj We also assume that ξ is orthogonal to the common shocks u : it|vj t E(ξ u(cid:48) ) = 0,for all s. (3.7) t|vj t−s|vj Our model consists of equations 3.2 through 3.7, and we can use the Kalman filter to estimate the common factors F by assuming that errors are Gaussian. If we replace t the parameters of the model above by their consistent estimates (see section A.3 of the Appendix for details), we can estimate the common factors as: Fˆ = Proj[F | Ω ;Λˆ,Aˆ,Bˆ,Ψˆ]. (3.8) t|vj t vj In particular, imposing ψ˜ = ∞ when y is missing (see equation 3.4) implies it|vj it|vj that the filter, through its implicit signal extraction process, will put no weight on the missing variable in the computation of the factors at time t. 2The relation of our model to that used in estimating principal components is discussed in Section A.4 of the Appendix. 7

The Kalman filter is also used to evaluate the degree of precision of the factor estimates given the consistent parameter estimates, with the degree of precision reflecting that of the signal extraction process for estimating the factor: Vˆ = E[(F −Fˆ)(F −Fˆ )(cid:48);Λˆ,Aˆ,Bˆ,Ψˆ]. s|vj t t t−s t−s Our estimates of the signal and their degree of precision are given by: χˆ = Proj[χ | Ω ;Λˆ,Aˆ,Bˆ,Ψ˜] = Λˆ Fˆ it|vj it vj i t|vj E(χ −χˆ )2 = Λˆ(cid:48)Vˆ Λˆ it it|vj i 0|vj i A discussion of the assumptions is in the appendix. 3.2 Forecasts and Uncertainty Turning to the nowcast, notice that in the state space representation we assume that only the common component of each series is forecastable. Empirically, this restriction does not create any relevant loss of information because the common factors are able to capture not only most of the cross-sectional correlation, but also the bulk of the dynamics of the key aggregates (for evidence on this point, see GRS). Hence, if y is not available, because y has not been released yet at vintage v, it|vj it (this is always the case if t > v), then our estimates are given by yˆ = µˆ +χˆ . it|vj i it|vj On the other hand we assume that if an official estimate for y is available, so that it|vj y has been released at vintage v , then yˆ = y . More precisely: it j it|vj it|vj yˆ = Proj[y | Ω ;Λˆ,Aˆ,Bˆ,Ψˆ] (3.9) it|vj it|vj vj = (1−δ )y +δ (µˆ +χˆ ). it|vj it|vj it|vj i it|vj where: (cid:40) 0 if y is available δ = it|vj it|vj 1 if y is not available it|vj From these equations, as indicated in Section 1, we can compute the news induced by the release of block j to the nowcast of y : it NEWS[i,v ] = yˆ −yˆ (3.10) j it|vj it|vj−1 Because the projections by which these forecasts are calculated assume that the parameters are given, and thus the relative weights in the signal extraction process are unchanged, this measure of news reflects the updating of the factors due only to the new information in vintage v , conditional on the information in vintage v . This j j−1 measureofthenewsallowsustodeterminewhetherparticularreleasescontainrelevant 8

information in a real-time setting and thus whether it is worthwhile to estimate the signal at each intra-month data release. Also, for each vintage, the confidence bands for the forecast can be easily computed fromthestatespacerepresentation. Letusconsiderthedifferencebetweentheexpected value computed at vintage w and the official realized released in the future at date w˜ (w˜ > w). Our measure of uncertainty about this realized value is defined as: V(cid:99)y = E[(yˆ −y )2;Λˆ,Aˆ,Bˆ,Ψˆ]. (3.11) it|w it|w it|w˜ Alternatively, if y has not been released yet at vintage w, we have: it V(cid:99)y = E[(χˆ −χ )2;Λˆ,Aˆ,Bˆ,Ψˆ]+E(ξ2 ) it|w it|w it it|w˜ = V(cid:100)χ +V(cid:99)ξ it|w it|w where Vˆχ = Λˆ(cid:48)Vˆ Λˆ and V(cid:99)ξ = ψˆ . Notice that this measure of uncertainty is jt|w i 0|w i jt|w j independent of w˜ by assumption (cfr. section 3). On the other hand, if there is an official release of y at vintage w, we have it V(cid:99)y = E[(χˆ −χ )2 | y ,··· ,y ;Λˆ,Aˆ,Bˆ,Ψˆ]+E(ξˆ −ξ )2 it|w it|w it 1|w w|w it|w it|w2 where there is no covariance term due to the orthogonality of the factor and the idiosyncratic term. This quantity measures the size of the revision error between vintage w and vintage w˜. To estimate it, it is necessary to have an assessment on the evolution of the idiosyncratic component at each release E(ξˆ −ξ )2. In addition, notice that it|w it|w˜ E(χˆ −χ )2 will provide a lower bound for the variance of the revisions. For simplicit|w it ity, we will not measure uncertainty due to revision errors, hence we will assume that E[yˆ −y ]2 = 0 if there is an official release of y at vintage w.3 it|w it|w˜ it In summary, we have: (cid:179) (cid:180) V(cid:99)y = δ V(cid:100)χ +V(cid:99)ξ (3.12) it|vj it|vj it|vj it|vj Notice that there are two sources of uncertainty, one associated with the signal extraction problem (extraction of χ ), the other due to the presence of idiosyncratic t components (ξ ). t Theappendixdetailhowtoadaptthesemeasuresofnewsanduncertaintytoobtain the statistics described in Section 2.1 for the data of interest transformed in quarterly rates. 3Ananalysisofthedatarevisionprocesswillrequireaseparatediscussion,andisbeyondthescope of the paper. 9

4 Empirics The measures of news and uncertainty introduced in Section 2 will now be applied to the real-time vintages of data sets from June 2003 through March 2004 and to the pseudo real-time vintages we have constructed for each of those months, capturing the actual chronological order of the data releases (see again Section 2). We also present these measures in a way that controls for the timeliness of the data releases. 4.1 Data The dataset is described in Table 1. As anticipated in Section 2.1, it consists of about 200 macroeconomic indicators and the sample, in each vintage, starts in 1982. All variables are monthly, except for GDP and GDP deflator for which monthly measures are derived from linear interpolation.4 Details on data transformation are reported in Appendix C. Let us here stress that price variables are treated as I(2) in estimation, but results will be reported for the level of inflation. Table 1 describes the structure of the information within the quarter. Variables (releases) are indicated in Column 2 while Column 1 indicates the associated block. As described in Section 2, we have 15 blocks of releases.5 Different blocks of releases are published at different dates throughout a month (column 3) and may refer to different dates (column 4). Typically, surveys have very short publishing lags and often are forecasts for future months or quarters, while GDP, for example, is released with a relativelylongdelay.6 Industrialproduction,pricevariablesandothersareintermediate cases. In column 3, we start our “data month” with the Consumer Credit release on the 5th business day of the month and end it with the Employment Situation release on the first Friday of the following month. With this convention, the data set that we label as June, for example, only includes values for June and earlier, although the data in the latest Labor and Wages block contained in that data set were released in the first week of July. After the release of the Labor and Wages block, we track the flow of information within each month by exploiting the fact that our information blocks preserve the chronological ordering of the releases. As indicated in the third column of Table 1 and anticipated in the discussion of Section 2, the timing of releases varies somewhat from month to month. To overcome this problem, we construct pseudo intra-month vintages according to a stylized data release calendar, by assigning to the vintages the most common timing pattern and keeping that timing fixed across our 21 monthly data sets. The construction of the 4Although very simple, this transformation works because it is applied to only a small number of series and the distortion is expected to go into the idiosyncratic factor (See Altissimo, Bassanetti, Cristadoro,Forni,Hallin,andLippi(2001)). Infact,theresultsinGRSshowthatthemodelperforms quite well even with such a simple transformation. The procedure might be improved using more sophisticated types of interpolation, that is beyond the scope of the paper. 5Appendix C reports the source of each data release. The individual series in each release (and block) are reported in Appendix B. 6The releases of the GDP and Income block for the first, second and third months of the quarter containtheGDPandIncomedatafromthe“advance”,“preliminary”and“final”releases,respectively. 10

1 elbaT ycneuqerF gnihsilbuP gnimiT )2(esaeleR )1(emaNkcolB )5(atadfo )4(gaL )3( ).xorppa( ylhtnoM shtnomowt htnomfoyadssenisubht5 tiderCremusnoC91.G 1dexiM ylhtnoM htnomeno htnomfoht51-11 secivreSdooFdnaliateRroFselaSylhtnoMecnavdA 1dexiM ylhtnoM htnomeno htnomfoelddiM tnemnrevoG.S.UehtfosyaltuOdnastpieceRfotnemetatSyrusaerTylhtnoM 1dexiM ylhtnoM shtnomowt htnomfokeewllufdn2 5tibihxE :secivreSdnasdooGniedarTlanoitanretnI.S.U009TF 1dexiM ylhtnoM htnomeno htnomfoht71otht51 noitazilitUyticapaCdnanoitcudorPlairtsudnI71.G PI ylhtnoM htnomeno htnomfoht02ehtotht61 noitcurtsnoClaitnediseRweN 2dexiM ylhtnoM htnomtnerruc htnomfoyadsruhTdr3 yevruSkooltuOssenisuBaihpledalihPfoknaBevreseRlaredeF 2dexiM ylhtnoM htnomeno htnomfoelddiM secirPrecudorP IPP ylhtnoM htnomeno htnomfoelddiM secirPremusnoC IPC ylhtnoM shtnomowt esaeler-PDGretfayaD selasdnaseirotnevni :liated-PDG emocnI&PDG ylretrauQ retrauqeno htnomfokeewtsaL rotafledPDGdnaPDG :esaeler-PDG emocnI&PDG ylhtnoM htnomeno esaeler-PDGretfayaD syaltuOdnaemocnIlanosreP emocnI&PDG ylhtnoM htnomeno htnomfoyad .subtsalotdr3 yevruSsemoHderutcafunaM gnisuoH ylhtnoM htnomeno htnomfokeewtsaL selaSlaitnediseRweN gnisuoH ylhtnoM htnomeno htnomfokeewtsaL yevruSIMMdeFogacihC 1syevruS ylhtnoM htnomtnerruc htnomfo .seuTtsaL xednIecnedfinoCremusnoC 1syevruS ylhtnoM htnomtnerruc htnomehtfo .irFtsaL sremusnoCfoyevruSnagihciM 1syevruS ylkeeW htnomtnerruc .evaylhtnoM :htnomfo .sruhTtsaL tropeRsmialCylkeeWecnarusnItnemyolpmenU,smialC smialClaitinI ylkeeW htnomtnerruc .evaylhtnoM :htnomfo .deWtsaL yevruSegagtroMyramirPcaMeidderF setaRtseretnI yliaD htnomtnerruc .evaylhtnoM :htnomfoyadtsaL setaRtseretnIdetceleS51.H setaRtseretnI yliaD htnomtnerruc .evaylhtnoM :htnomfoyadtsaL setaRegnahcxEngieroF01.H laicnaniF yliaD htnomtnerruc .evaylhtnoM :htnomfoyadtsaL dlogfoecirP laicnaniF yliaD htnomtnerruc .evaylhtnoM :htnomfoyadtsaL ESYN laicnaniF yliaD htnomtnerruc .evaylhtnoM :htnomfoyadtsaL )ylkw(P&S laicnaniF yliaD htnomtnerruc .evaylhtnoM :htnomfoyadtsaL erihsliW laicnaniF ylhtnoM htnomtnerruc htnomfoyadssenisubts1 gnirutcafunaM-RGMP 2syevruS ylhtnoM htnomtnerruc htnomfoyad .subts1 repaPlaicremmoC 3dexiM ylhtnoM htnomeno htnomfoyad .subts1 ecalPnituPnoitcurtsnoC 3dexiM ylhtnoM htnomeno ht6-ht03/ht92-dr32 sredrOdnaseirotnevnI,stnempihSsrerutcafunaMsdooGelbaruDnotropeRecnavdA :3M 3dexiM ylhtnoM htnomeno selbaruDecnavdAretfasyad5 sredrOdnaseirotnevnI,stnempihSsrerutcafunaMsdooGelbaruDnotropeRlluF :3M 3dexiM ylhtnoM sretrauqowt )ylhtnomsiseires(ylretrauQ nitelluB .qnileDremusnoC tiderC&yenoM ylhtnoM htnomtnerruc htnomfo .sruhTts1 esaByratenoMehtdnasnoitutitsnIyrotisopeDfosevreseRetagerggA3.H tiderC&yenoM ylhtnoM htnomeno htnomfo .sruhTdn2 serusaeMkcotSyenoM6.H tiderC&yenoM ylhtnoM htnomeno htnomfo .irFts1 setatSdetinUehtnisknaBlaicremmoCfoseitilibaiLdnastessA8.H tiderC&yenoM ylhtnoM htnomtnerruc htnomfo .irFts1 noitautiStnemyolpmE segaW&robaL 11

vintages is discussed in more detail in Section A.1 of the Appendix. Following the notation introduced in Section 2, v indexes the vintage just before 0 the release of the first block (Mixed 1), while v indexes the vintage after the release 1 of Mixed 1 and before the release of the second block (IP). Just after the Labor and Wages release, we have the last vintage of the month, indexed by v . 15 Withthisconvention, thestartingvintageineachmonthisequaltothelastvintage of the subsequent month: so the vintages indexed by v and (v +1) are the same. 15 0 Because the data blocks defining the vintages are in the same order each month, we use v to index both the vintages and the time at which they are released. So, we will j say variables in the first block (Mixed 1) are updated in vintage v and are released at 1 time v . 1 The way we treat financial variables deserves a comment. Financial variables and interest rates are the most timely since they are available on a daily basis. In principle daily information could be used to update the estimates of GDP and inflation as, for example, in Evans (2005). Our approach is different. Since the bulk of our data is monthly, we disregard information from financial variables at frequencies lower than the month and let them enter the model as monthly averages. We make the arbitrary assumptionthattheybecomeavailableonlyattheendofthemonthwhichimpliesthat their effect is underestimated. 4.2 News and Uncertainty in Real-Time InthissectionwereportsummarystatisticsevaluatedusingrealtimevintagesfromJuly 2003toMarch2005. Thesemeasuresarederivedusingthe“real-time”and“pseudorealtime” vintages in their natural chronological order and thus correspond to the exercise inwhichtheforecasterupdateshernowcastsafterthereleaseofeachinformationblock. We report statistics on uncertainty around the current quarter nowcast of key variables and on the size of the news derived using real time vintages. For real variables, measures of news and uncertainty are constructed for quarterly quantities derived from monthlydata. Forinflation variables they are reported for annual inflation. The statistics used are based on formulas (3.10) and (3.12), modified so as to track the quarterly aggregates of interest. Themeasureofuncertaintyinformula(3.12)dependsontheestimatedparameters, which change over time because they are recomputed after each vintage of data. Below we report averages of the uncertainty measures across all the quarters considered in the real time exercise. We will refer to this measure as average uncertainty. Similarly, we measure the size of the news as the absolute value of the news measure (3.10) averaged across all the quarters considered in the real time exercise. Because the impact of the release of block j may differ according to whether the release is in the first, second or third month of the quarter, the average for both uncertainty and news is taken over the seven vintages in our sample and correspond to either to the first, second, or third months of the quarter. Chart 1, 2 and 3 focus on two key variables: quarterly growth of GDP (Charts 1a, 2a and 3a) and annual growth of GDP deflator (Charts 1b, 2b and 3b) while Charts 12

4a and 4b consider, respectively, additional real and nominal indicators.7 Measures of news for real growth and inflation are shown in Charts 1a and 1b, respectively. Charts 2a-2b, 3a-3b and 4a-4b report the evolution of the uncertainty on the signal (common) and the uncertainty on the variable itself (total). These charts complement the information in Chart 1a-1b by providing a systematic measure of how the accuracy of the nowcast evolves. Chart 2a-2b shows the evolution of the standard errors over the quarter: by understanding whether the marginal impact of a given release has a different effect in the first month than in later months, we can assess the importance of timing in explaining the impact of a particular release. Chart 3a-3b, on the other hand, overlays the three months of the quarter to allow for an easier comparison of the effects of a given block across the three months. Chart 4a-4b reports the same information as Chart 2a-2b, but for additional real and nominal series. These series are: employment on nonfarm payroll (NFP), unemployment rate (UR), personal consumption expenditure price index excluding food and energy (PCEX). Let us first concentrate on GDP growth. From Charts 1a, 2a and 3a we have three results: 1. Intra-month information matters. Data releases throughout the quarter convey news as can be seen by the fact that the estimates are generally updated as new releases are published (Chart 1a). Moreover, uncertainty decreases uniformly through the quarter (Chart 2a). 2. The release that has the largest impact on the nowcast and its precision in the first month is the “Mixed 2” block. Mixed 2 is composed of two series from the New Residential Construction Release and nine series from the Philadelphia Business Outlook Survey. By way of the Philadelphia survey, Mixed 2 is the most timely release since it is the first block to contain data or forecasts on the currentquarter. Thetwoprecedingreleasesinthemonth(Mixed1andIndustrial Production) convey information about earlier months only and have almost no impact since they are published relatively late. 3. Other important news for the nowcast of real GDP growth is contained in the blocksofLaborandWages(whichincludesthereleaseoftheEmploymentReport) and interest rates (the components of the block compose the yield curve). This emerges from both Chart 1a and 2a. In general, the striking result is that the surveys (Mixed 2) have a larger impact than the Employment Report (Labor and Wages) which is the news to which financial marketsreactmorestrongly. Thereasonisthat, bythetimethelaborblockisreleased, the information conveyed by the surveys has already been taken into account. This highlights the importance of timing. Noticeable is also the large effect of the interest rate block on both the nowcast and its uncertainty. The Interest Rate block is the end-of-month average of the weekly 30-year mortgage rate from Freddie Mac and of daily observations of nine interest rates 7All statistics are presented numerically in Section D of the Appendix. 13

from the Federal Reserve’s H.15 Release. The later include short and longer-term U.S. TreasuryratesandAAAandBAAcorporatebondyields. Likewise,theFinancialblock is composed of end-of-month averages of daily observations on foreign exchange rates, the price of gold, and U.S. stock prices. Chart 1a Average Size of News: Nowcasts of Real Growth 0.2 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL first month (m=1) second month (m=2) third month (m=3) Chart 2a Average Uncertainty: Nowcast of Real Growth 1.4 1.2 1 0.8 0.6 0.4 0.2 0 segaW dna robaL 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL segaW dna robaL 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL segaW dna robaL 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL total common First Month Second Month Third Month 14

Chart 3a Average Uncertainty: Nowcast of Real Growth (Common Component) 1.2 1.0 0.8 0.6 0.4 0.2 0.0 dna robaL segaW 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL first month (m=1) second month (m=2) third month (m=3) Chart 4a Average Uncertainty: Nowcast of Alternative Real Variables (Common Component) 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL GDP NFP UR First Month Second Month Third Month 15

We now turn to inflation. Let us first remark that, as mentioned, we focus on quarterly growth rate for GDP and on annual rate for what concerns price inflation. The latter is therefore smoother and less sensitive to the news by construction. This feature is evident from Chart 1b. From Chart 2b and 3b we can see that, as in the case of GDP, uncertainty decreases monotonically within the quarter as new information arrives. As for the importance of different blocks, two features are noticeable. First, looking at the evolution of the updates of the estimates (Chart 2a), we can see that a big jump occurs with the release of the GDP and Income block in the first month of the quarter. This release (the “advance” release) contains the first observation for the GDP deflator (and GDP) for the previous quarter and, thus, reveals information about the value of the idiosyncratic shock to the deflator in the previous quarter. This effect, however, is much less pronounced on the common component (the signal) and mainly affects the idiosyncratic component of inflation. This is explained by the fact that, since we have modelled inflation in first differences, the idiosyncratic component has a unit root (empirically it turns out to be well captured by a random walk) so that the nowcast reacts strongly to the information revealed about the idiosyncratic shock in the previous quarter. Moreinterestingly,animportantimpactontheprecisionoftheestimates(Chart2b) is due to the financial block release, containing data on exchange rates and the nominal prices of gold and equities, whereas, unlike in the case of GDP the interest rate block, has no effect. The Financial block, as we have seen, contributes to a noticeable decline in the uncertainty associated to GDP inflation but not for that associated to real GDP. Conversely, the Interest Rate block has an effect that is much more pronounced for real GDP than for inflation. Notice that the role of financial variables and interest rates is likely to be underevaluated since they are available from the markets on a daily basis but we assume that they become available only at the end of the month. To check for the robustness of these results for the Interest Rate and Financial blocks, we perform the same analysis as in Chart 3 but invert the order of these two blocks. This exercise is motivated by the fact that the order of these two blocks is arbitrary because we constructed them as month-end averages of weekly and daily observations which implies that they become available contemporaneously at the end of the calendar month. As shown in Chart 5, the relative impact of these two blocks are not sensitive to their ordering. WhilewehavefocusedonGDPinflationandgrowth,centralbankersandeconomists at large are also interested in other aggregate measures of inflation and real activity. Measures of uncertainty for the common factor of the nowcast for inflation based on the core deflator for personal consumption expenditures and for the growth rate of employment in nonfarm payrolls and the unemployment rate are presented in Chart 4a and 4b. Notice that the two measures for inflation move closely together, as do the three measures for real activity. Thus below we will continue to focus on the common factor for real GDP and for GDP inflation. Finally, let us remark that the size of news, unlike the measure of uncertainty, depends on the particular realization over the sample period we use for the out-ofsample exercise. This explains why results on the size of the news are some time different than results on average uncertainty. 16

Chart 1b Average Size of News: Nowcasts of Inflation 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL first month (m=1) second month (m=2) third month (m=3) Chart 2b Average Uncertainty: Nowcast of Inflation 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 segaW dna robaL 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL segaW dna robaL 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL segaW dna robaL 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL total common First Month Second Month Third Month 17

Chart 3b Average Uncertainty: Nowcast of Inflation (Common Component) 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 segaW & robaL 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL first month (m=1) second month (m=2) third month (m=3) Chart 4b Average Uncertainty: Nowcast of Alternative Inflation Measures (Common Component) 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL Deflator PCEX First Month Second Month Third Month 18

Chart 5a Average Uncertainty Under Alternative Ordering: Real Growth (Common Component) 1.2 1.0 0.8 0.6 0.4 0.2 0.0 segaW & robaL 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI laicnaniF setaR tseretnI 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL first month (m=1) second month (m=2) third month (m=3) Chart 5b Average Uncertainty Under Alternative Ordering: Inflation (Common Component) 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0.00 segaW & robaL 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI laicnaniF setaR tseretnI 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL first month (m=1) second month (m=2) third month (m=3) 19

4.3 The Information Content of the Blocks Conditional on Timeliness Themarginalimpactofablockisconditionalonthesetofpreviouslyreleaseddata. To control for the effect due to timeliness, we construct a counterfactual series of vintage data sets in which data don’t differ in their timing or lags of releases. In this way we can construct a measure of “quality” of the data independent of timeliness. Foreachofthefirstthreemonthsof2004,weconstruct16vintageswitheachvintage corresponding to one of the information blocks. These 48 vintages are denoted by y t|v˜j v˜ = 04m1,...,04m3; j = 0,...,15. In contrast to the real-time vintages, each of these counterfactual vintages is constructed from data in a single real-time vintage (which we choose to be y ,v = 05m3). We then truncate this data set at December 2003, t|v0 thereby producing a data set that is balanced because the truncation deletes periods which may have had missing observations due to lags in releasing data. We denote the series in this data set as y ,t = 83m1,...,03m11,v˜ = 04m1 and refer to measures of t,v˜0 uncertainty constructed from these series as “no release” measures. Startingwiththisbalanceddataset, weconstructpseudopanelsinwhicheachblock is the most timely. For the data set in which the Mixed 1 block is most timely, we add data for January 2004 but only for variables belonging to Mixed 1, obtaining the counterfactual vintage y .8 Similarly, we start anew with the balanced data set and t|v˜1 add data for January 2004 but only for variables belonging to the second block (IP), obtaining the counterfactual vintage y . In the end, we obtain the counterfactual t|v˜2 vintages y , for j = 0,...,15 and v˜= 03m1. tv˜j Then we do the same exercise with the balanced panel truncated at January 2004, y , v˜ = 04m2, and add February 2004 data for each block one by one to construct t|v˜0 y , forj = 0,...,15andv˜= 05m2andsoon, upthroughMarch. Intheend, weobtain tv˜j y , for v˜= 04m1,...,04m3, j = 0,...,15. t|v˜j Using these vintages, we construct measures of common-factor uncertainty for the nowcasts. They are reported in Chart 6a and 6b.9 The horizontal dashed lines are drawn at the level of the “no release” uncertainty. As it was expected, in each month, each block of information either leaves the average uncertainty of the nowcast unchanged, or reduces it, relatively to the “no release” value. In Chart 6a we report results for GDP. Industrial production has now become an important block and so has GDP & Income and Labor and Wages. The importance of surveys and interest rates is now reduced. In Chart 6b we report results for inflation. Compared with Chart 2b, where the main effect was due to surveys, GDP and income and financial variables, we now have a clear effect of the price blocks and of industrial production. The effect of financial variables remain sizeable while that of surveys is reduced. Ingeneral,harddatabecomeimportantwhiletheywerenotintherealtimeexercise, whilesoftdatahavealowerimpactwhichreflectsthefactthatpartoftheircontribution is mainly due to timeliness. 8The values for January 2004 that we use here, we use values for this month form the vintage, v=05m3. 9In computing these results, we run the Kalman filter over the various datasets but estimate the model parameters only once on the basis of the balanced panel (up to September 2004, in this case). Numerical details of these exercises are reported in Tables 4a and 4b. 20

We should stress, however, that the effect of financial variables on inflation uncertainty remains large and it is therefore independent of timeliness. Chart 6a Counterfactual Average Uncertainty : Real Growth 2004-Q1 (Common Component) 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 esaeler on 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC emocnI & PDG gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL Jan-04 Feb-04 Mar-04 Chart 6b Counterfactual Average Uncertainty: Inflation 2004-Q1 (Common Component) 0.06 0.05 0.04 0.03 0.02 0.01 0.00 esaeler on 1 dexiM noitcudorP .dnI 2 dexiM IPP IPC dna PDG emocnI gnisuoH 1 syevruS smialC laitinI setaR tseretnI laicnaniF 2 syevruS 3 dexiM tiderC & yenoM segaW & robaL Jan-04 Feb-04 Mar-04 21

5 Summary and Conclusion This paper has analysed the impact of the flow of information within the month on the estimate of current quarter GDP growth and inflation before these variables are published. We considered the unsynchronous release of about 200 monthly time series where releases are organized in groups of homogeneous variables. To this end, we have proposed a framework which is an adaptation of the parametric version of the large dynamic factor model proposed by GRS and Doz, Giannone, and Reichlin (2005). This model allows to analyze the flow of a large number of time series and update the signal on the basis of a panel which, due to the unsynchronous release of data, is unbalanced at the end of the sample. We find that information matters in the sense that the precision of the signal increases monotonically within the month as new data are released. We also find that both timeliness of the release and quality matter for decreasing uncertainty. Surveys have a large impact on both inflation and output in real time and their effect is larger thantheEmploymentReport. Harddatasuchaspriceandrealvariableshavenoeffect since they are released relatively late. When we control for timeliness, the contribution of hard data increases and we find a sizeable effect of both nominal and real variables on inflation while for GDP only real variables matter. Another finding is that interest rates affect the precision of the estimates of GDP, but not that of inflation while asset prices affect the precision of the nowcast of inflation, but not that of GDP. 22

References Altissimo, F., A. Bassanetti, R. Cristadoro, M. Forni, M. Hallin, and Lippi (2001): “EuroCOIN: A Real Time Coincident Indicator of the Euro Area Business Cycle,” CEPR Discussion Papers 3108. Bai, J. (2003): “Inferential Theory for Factor Models of Large Dimensions,” Econometrica, 71(1), 135–171. Bernanke, B., and J. Boivin (2003): “Monetary Policy in a Data-Rich Environment,” Journal of Monetary Economics, 50, 525–546. Boivin, J., and S. Ng (2003): “Are More Data Always Better for Factor Analysis?,” NBER Working Paper 9829, Journal of Econometrics, forthcoming. (2005): “Understanding and Comparing Factor-Based Forecasts,” NBER Working Paper 11285. Doz, C., D. Giannone, and L. Reichlin(2005): “AMaximumLikelihoodApproach for Large Approximate Dynamic Factor Models,” Unpublished manuscript. Engle, R. F., and M. Watson (1981): “A one-factor multivariate time series model of metropolitan wage rates,” Journal of the American Statistical Association, 76, 774–781. Evans, M. D. (2005): “Where Are We Now? Real-Time Estimates of the Macro Economy,” NBER Working Paper 11064, Internationa Journal of Central Banking, forthcoming. Forni, M., D. Giannone, M. Lippi, and L. Reichlin (2005): “Opening the Black Box: Structural Factor Models with large cross-sections,” Manuscript, (www.dynfactors.org). Forni, M., M. Hallin, M. Lippi, and L. Reichlin (2000): “The Generalized DynamicFactorModel: identificationandestimation,”ReviewofEconomicsandStatistics, 82, 540–554. (2003): “The Generalized Dynamic Factor Model: one-sided estimtion and forecasting,” CEPR Working Paper 9829, Journal of the American Statistical Association, forthcoming. Forni, M., and L. Reichlin (2001): “Federal Policies and Local Economies: Europe and the US,” European Economic Review, 45, 109–134. Giannone, D., L. Reichlin, and L. Sala (2004): “Monetary Policy in Real Time,” in NBER Macroeconomics Annual, ed. by M. Gertler, and K. Rogoff, pp. 161–200. MIT Press. Marcellino, M., J. H. Stock, and M. W. Watson (2003): “Macroeconomic Forecasting in the Euro Area: Country Specific versus Area-Wide Information,” European Economic Review, 47, 1–18. 23

Quah, D., and T. J. Sargent (2004): “A Dynamic Index Model for Large Cross- Section,” in Business Cycle, ed. by J. Stock, and M. Watson, pp. 161–200. Univeristy of Chicago Press. Stock, J. H., and M. W. Watson (1989): “New Indexes of Coincident and Leading Economic Indicators,” in NBER Macroeconomics Annual, ed. by O. J. Blanchard, and S. Fischer, pp. 351–393. MIT Press. (2002): “Macroeconomic Forecasting Using Diffusion Indexes,” Journal of Business and Economics Statistics, 20, 147–162. 24

A Appendix A.1 Construction of the Vintage Data Sets Weconstructthesequenceofvintagesv ,...,v foragivenmonthv fromtwodatasets: 1 15 theonescontainingalldatacollectedformonthsv−1andv (includingtheEmployment Report early in the following month). Because these data sets contain the releases of all 15 information blocks, they are denoted as (v−1) and v , respectively. The data 15 15 set (v−1) is also the initial data set for month v, so (v−1) = v . 15 15 0 Starting with v for month v, the data series in that data set are replaced and 0 updated recursively block-by-block with blocks that were released in month v (and that are contained in the data set indexed by v ). For example, v is constructed by 15 1 identifyingtheseriesinBlock1(Mixed1)andreplacingitsvaluesinv withthosefrom 0 v ,whileleavingthevaluesforseriesinallotherblocksunchanged. Whenmakingsuch 15 replacements, each series in the block is replaced by the new readings on its current and past values because new releases contain new values not only for the most recent dates, but also for past dates. We call v a “pseudo vintage”, because the data series 1 in it were not literally constructed in real time, they are constructed from information blocks that generally preserve the chronological order of the data releases. The pseudo vintage v is constructed from v by identifying all series in Block 2, taking their values 2 1 from v and using them to replace the values for the series reported in v for Block 2. 15 1 The pseudo vintages v ,v ,...,v are constructed in the analogous manner. 3 4 15 In sum, for each month (v= June, 2003; ... ; March 2005), we have 16 vintages indexed by (v−1) = v ,v ,...,v = (v+1) . 15 0 1 15 0 A.2 Transformations of the Data Series The transformations we apply to the raw data (Y ) so that the model estimation uses it data series that are stationary (y ) are: it Data transformations code transformation Description 0 y =Y no transformation it it 1 y =logY log it it 2 y =(1−L3) Y three-month difference it it 3 y =(1−L3)logY ×100 three-month growth rate it it 4 y =(1−L3)(1−L12)logY ×100 three-month difference of yearly growth rate it it The particular transformation that we apply to a series is reported in column 4 of the table in Section C of the Appendix. A.3 Estimation of Parameters In this section we do not consider the dependence of data on the vintage but instead work under the assumption that the data generating process of the idiosyncratic component is the same across different releases. In particular, we assume homoscedasticity 25

of the idiosyncratic component across vintages, Eξ ξ(cid:48) = Ψ for all v . However, ret|vj t|vj j laxing this assumption does not have major consequences for the results below because the principal component estimator is robust to a limited amount of heteroscedasticity, which could be induced by the data revision process (see e.g. Bai (2003)). Theassumptionsthatallowustoidentifythecommonandidiosyncraticcomponents of the model are: A1. Common factors are pervasive (cid:181) (cid:182) 1 liminf Λ(cid:48)Λ > 0, n→∞ n and A2. Idiosyncratic factors are non-pervasive (cid:181) (cid:182) 1 lim maxv(cid:48)Ψv = 0. n→∞ n v(cid:48)v=1 Assumption A1 implies that the common factors must be understood as sources of variation that remain pervasive as we increase the number of series in the dataset. In that sense, the common factors correspond to the notion of macroeconomic shocks. Assumption A.2 implies that idiosyncratic factors may affect more than one particular series (Ψ need not be diagonal, however the idiosyncratic shocks are assumed to be stationary), but the effects of an idiosyncratic shock are limited to a particular cluster and do not propagate throughout the macroeconomy. Next, we define: x = y −µˆ it it i 1 z = (y −µˆ ), it it it σˆ i (cid:113) (cid:80) (cid:80) where µˆ = 1 T y and σˆ = 1 T (y −µˆ )2. it T t=1 it i T t=1 it i Consider the following estimator of the common factors: (cid:88)T (cid:88)n (F˜,Λˆ) = argmin (z −λ F )2 t it i t Ft,Λ t=1i=1 To derive these estimators, define the sample correlation matrix of the observables (z ): t 1 (cid:88)T S = z z(cid:48) T t t t=1 Denote by D the r×r diagonal matrix with diagonal elements given the largest r eigenvalues of S and denote by V the n×r matrix of the corresponding eigenvectors subject to the normalization V(cid:48)V = I . We estimate the factors as: r 26

F˜ = V(cid:48)z t t The factor loadings, Λˆ, and the covariance matrix of the idiosyncratic components, Ψˆ, are estimated by regressing the variables on the estimated factors: (cid:195) (cid:33)−1 (cid:88)T (cid:88)T Λˆ = x F˜(cid:48) F˜F˜(cid:48) = V t t t t t=1 t=1 and Ψˆ = diag(S−VDV).10 The other parameters are estimated by running a VAR on the estimated factors, precisely: (cid:195) (cid:33)−1 (cid:88)T (cid:88)T Aˆ= F˜F˜(cid:48) F˜ F˜(cid:48) t t−1 t−1 t−1 t=2 t=2 (cid:195) (cid:33) 1 (cid:88)T 1 (cid:88)T Σˆ = F˜F˜(cid:48)−Aˆ F˜ F˜(cid:48) Aˆ(cid:48) T −1 t t T −1 t−1 t−1 t=2 t=2 Define P as the q × q diagonal matrix with the entries given by the largest q eigenvalues of Σˆ and by M the r×q matrix of the corresponding eigenvectors, then: Bˆ = MP1/2 The estimates µˆ, Λˆ, Ψˆ, Aˆ, Bˆ can be shown to be consistent as n,T → ∞. Under assumptions A1 and A2 this is proven in Forni et al. 2005 and, under slightly different assumptions by Stock and Watson(2002), Bai and Ng(2003) and Giannone, Reichlin and Sala(2003). For unbalanced panels the parameters of the model, µ,Λ,A,B,Ψ are estimated using data up to the last date when the balanced panel is available. Then we reestimate the factors through the Kalman filter as outlined above in section 3.1.11 Loosely speaking, the Kalman filter, computes the factors by weighting the innovation content of each variable (x −E[x |x ,...,x ;Λˆ,Aˆ,Bˆ,Ψˆ]) accordi,t+1 i,t+1 1 t ingly to its news (the part driven by common shocks u ) to noise (the part driven by t components ξ ) ratio. it 10For any square matrix A, diag(A) is the matrix A with off-diagonal elements set equal to zero. In estimating Ψ, we estimate only the diagonal elements and set the off-diagonal elements to zero. 11Notice that the parameters Λ,A,Ψ,B can reestimated by OLS on the new factors Fˆ using the t implied second order moments which can be computed by running the Kalman smoother. This is one step of the EM algorithm, hence by iterating until convergence, we obtain Maximum Likelihood estimatesunderGaussianassumptions. SuchaprocedurehasbeenusedbyEngleandWatson(1981)and StockandWatson(1989)withanhandfuloftimeseriestocomputecoincidentandlaggingindicators, and by Quah and Sargent (2004) with a larger panel of time series. On the development of this idea and some theoretical results, see Doz, Giannone, and Reichlin (2005). 27

A.4 Estimation of the common factors: relation to Principal Components and Weighted Principal Components Notice that principal components and weighted principal components are a particular case of the estimates of the common factors derived above. In fact, if we constrain (cid:80) Aˆ= 0andΨˆ = 1 n ψˆI = ψ¯I , thentheKalmanfilterisredundantsincethefactor n i=1 i n n estimated with the Kalman filter step will be proportional to the principal components estimates: Fˆ = (ψ¯I +Λˆ(cid:48)Λˆ)−1Λˆ(cid:48)x ∝ V(cid:48)z = F˜ t r t t t However, if only Aˆ= 0 is imposed, then Fˆ = (I +Λˆ(cid:48)Ψˆ−1Λˆ)−1Λ(cid:48)Ψˆ−1x , t r t so the estimated factors are proportional to the weighted principal components, i.e. principal components on the weighted data Ψ−1/2x .12 t With both principal components and generalized principal components, the estimates of the factors are computed by projecting only on the present observations and, thus, the dynamic properties of the factors are not taken into account. In our case, the Kalman filter performs the projection on present and past observations and, thus, takes into consideration the dynamics of the factors and the degree of commonality of each time series. However, when running the Kalman filter, we do not exploit the time series and cross-sectional correlations of the idiosyncratic shocks which are treated as uncorrelated both in time and in the cross section. Estimates are, however, still consistent under the approximate factor structure (Assumption A1 and A2), as shown in Doz, Giannone, and Reichlin (2005). A.5 Statistics for the Untransformed Data In general, the measures of news and uncertainty in equations 3.10 and 3.12 apply to measures of our data over which the model has been estimated: that is, they apply to monthly data and to data that has been transformed so as to be stationary. Here we derive such measures that apply to data expressed in ways more commonly used by economists. Series with native frequencies higher than monthly, such as financial and interest rates, are aggregated to monthly frequencies by taking simple within-month averages. And in general, to derive such measures from monthly variables, one or both of two adjustments need to be made to the measures: 1) to adjust from the model’s monthly forecaststoquarterlyforecastsand2)toadjustfromstationaryseriestonon-stationary series. This issues are discussed below. Case 1: Interpolations All the variables in our model are expressed as monthly series; for example monthly growth rates and monthly inflation. Accordingly, the measures of NEWS and uncertainty derived above in the text apply to series of this frequency. With most practitioners of monetary policy commonly interested in inflation 12DifferentversionsofsuchanestimatorwereproposedbyBoivinandNg(2003),ForniandReichlin (2001), Forni, Hallin, Lippi, and Reichlin (2003). 28

and growth at the quarterly frequency (in part because this is the highest frequency at which real GDP and the GDP deflator are published), we transform our measures of News and uncertainty to the quarterly frequency. To set notation, the quarterly measure of variable z will be denoted, as in section, 2.1, by: zq,k ∈ N. k As an example, consider the case of real GDP. Its quarterly growth rate, defined in the first equation below, can be expressed in terms of the measure y , over which the zt model was estimated: zq = (log(Y +Y +Y )−log(Y +Y +Y ))×400 k z,k z,k−1 z,k−2 z,k−3 z,k−4 z,k−5 Since variables enter our model as three-month annualized growth rates, y = (log(Y )−log(Y ))×400 z,k z,k z,k−3 Hence, we have: zq ∼ (y +y +y )/3 k z,k z,k−1 z,k−2 where, as stressed above, we have defined the quarter by its last month We aggregate the forecast accordingly: z(cid:98) q = y(cid:98) +y(cid:98) +y(cid:98) , k|vj z,k|vj z,k−1|vj z,k−2|vj and derive the measure of “NEWS” in a analogous manner to that of equation 3.10. For the construction of the corresponding uncertainty, we have to take into account theautocorrelationbetweentheextractedfactors, whichissummarizedinthefollowing matrix:   Vˆ ... Vˆ(cid:48)  0|vj s−1|vj  Vˆ s|vj =   . . . ... . . .   Vˆ ... Vˆ s−1|vj 0|vj Hence, uncertainty is given by: V(cid:100)Z q = E[(yˆq −yq )2 | y ,··· ,y ;Λˆ,Aˆ,Bˆ,Ψˆ] k|vj z,k|vj z,k|vj 1|vj vj|vj = (H ⊗Λˆ )Vˆ (H ⊗Λˆ )(cid:48)+ψˆ H H(cid:48) (A.13) z,k|vj z 2|vj z,k|vj z z z,k|vz z,k|vz = V(cid:100)χq +V(cid:100)ξq z,k|vz z,k|vj where H = [δ ,δ ,δ ] z,k|vj z,k|vj z,k−1|vj z,k−2|vj Case 2: Going from Stationary to Non-Stationary DataFor some variables, economists are interested in measures of them that are not stationary. For example, 29

the measure of GDP inflation used in this model is not stationary and was differenced to yield a stationary series with which the model could be estimated. In particular, GDP inflation enters the model as: y = ∆3mπ ≡ π −π π,t t t t−3 where π = (logP −logP )×100 and P is the level of the GDP deflator. We are t t t−12 t interested in forecasting annual inflation at a quarterly frequency: πq = π +π +π , k = 1,2,... k k k−1 k−2 AsdescribedaboveinGenericCase1,wecanfirstchangefrommonthlytoquarterly forecasts of the change of inflation: πq −πq = ∆qπq = ∆3mπ +∆3mπ +∆3mπ k k−3 k k k−1 k−2 Denoting the by ∆(cid:98)πq the estimates made at time v , our estimates for the level of k|vj j inflation are given by: (cid:88)k π(cid:98) q = πq + ∆(cid:100)qπ q k|vj 0|vj j|vj j=1 Uncertainty will be measured accordingly as: V(cid:100)πq = E[(πˆq −πq )2 | x ,··· ,x ;Λˆ,Aˆ,Bˆ,Ψˆ] k|vj k|vj k|vj 1|vj vj|vj = (H ⊗Λˆ )Vˆ (H ⊗Λˆ )(cid:48)+ψˆ H H(cid:48) (A.14) π,k|vj π s|vj π,k|vj π π π,k|vj π,k|vj = V(cid:100)χq +V(cid:100)ξq π,k|vj π,k|vj where H = [δ ,δ ,...,δ ] π,k|vj π,k|vj π,k−1|vj π,k−s|vj ands = k−v −lwherelisthemaximumdelayforthereleaseofπ ,asdefinedinsection j t 2. A similar treatment has been applied to recover the statistics for the unemployment rate which is treated as non stationary and hence enter our model in differences. 30

secruoS dna sesaeleR ataD B etisbeW emaN esaeleR emaN kcolB /91g/sesaeler/vog.evreserlaredef.www//:ptth tiderCremusnoC91.G 1dexiM fdp.buplluf/www/dsvs/vog.susnec.www//:ptth secivreSdooFdnaliateRroFselaSylhtnoMecnavdA 1dexiM /stm/vog.saert.smf.www//:ptth tnemnrevoG.S.UehtfotnemetatSyrusaerTylhtnoM 1dexiM /esaeleR-sserP/edart-ngierof/vog.susnec.www//:ptth edarTlanoitanretnI.S.U009TF 1dexiM /71G/sesaeler/vog.evreserlaredef.www//:ptth noitazilitUyticapaCdnanoitcudorPlairtsudnI71.G PI fdp.tsnocserwen/www/rotacidni/vog.susnec.www//:ptth noitcurtsnoClaitnediseRweN 2dexiM lmth.xedni/sob/noce/gro.brf.lihp.www//:ptth yevruSkooltuOssenisuB 2dexiM fdp.ipp/fdp/esaeler.swen/vog.slb.www//:ptth sexednIecirPrecudorP IPP fdp.ipc/fdp/esaeler.swen/vog.slb.www//:ptth xednIecirPremusnoC IPC psa.xednI/gniylrednuapin/bewapin/nd/aeb/vog.aeb.www//:ptth selbatliatedgniylrednumorfseiresdetceleS emocnI&PDG mth.1nd/aeb/vog.aeb.www//:ptth tcudorPcitsemoDssorG emocnI&PDG mth.esaelerswenip/lerswen/aeb/vog.aeb.www//:ptth syaltuOdnaemocnIlanosreP emocnI&PDG lmth.xednishm/www/tsnoc/vog.susnec.www//:ptth yevruSsemoHderutcafunaM gnisuoH fdp.selasserwen/tsnoc/vog.susnec.www//:ptth selaSlaitnediseRweN gnisuoH mfc.immfc/atad dna hcraesercimonoce/gro.defogacihc.www//:ptth xednIgnirutcafunaMtsewdiMdeFogacihC 1syevruS mth.remusnoc/moc.tropergnillop.www//:ptth xednIecnedfinoCremusnoC 1syevruS php.niam/ude.hcimu.rsi.acs.www//:ptth sremusnoCfoyevruS 1syevruS psa.hcrasmialc/yolpmenu/vog.atelod.swo//:ptth tropeRsmialCylkeeWecnarusnItnemyolpmenU smialClaitinI txt.mc/rw/atad/51h/sesaeler/vog.evreserlaredef//:ptth yevruSegagtroMyramirPcaMeidderF setaRtseretnI /etadpu/51h/sesaeler/vog.evreserlaredef.www//:ptth setaRtseretnIdetceleS51.H setaRtseretnI /rotaluclac/sexednI/moc.erihsliw.www//:ptth xednIerihsliW laicnaniF /hcnuleerf/moc.ymonoce.www//:ptth secidnIP&S laicnaniF /etadpu/01h/sesaeler/vog.evreserlaredef.www//:ptth setaregnahcxE laicnaniF lmth.dloglacirotsih/strahc/moc.octik.www//:ptth xiFMPdloGnodnoL laicnaniF /hcnuleerf/moc.ymonoce.www//:ptth egnahcxEkcotSkroYweN laicnaniF fdp.tnerruc/gro.ogacihc-mpan.www//:ptth tropeRogacihCehT 2syevruS fdp.dgrud/fdp/vda/3m/www/rotacidni/vog.susnec.www//:ptth srerutcafunaMsdooGelbaruDnotropeRecnavdA 3dexiM fdp.o-i-s/fdp/lerp/3m/www/rotacidni/vog.susnec.www//:ptth srerutcafunaMsdooGelbaruDnotropeRlluF 3dexiM mth.1elbat/pc/sesaeler/vog.evreserlaredef.www//:ptth gnidnatstuOrepaPlaicremmoC :repaPlaicremmoC 3dexiM fdp.esaeler/03C/tsnoc/vog.susnec.www//:ptth gnidnepSnoitcurtsnoC 3dexiM mth.ycneuqniledss/scitsitatS+dna+syevruS/moc.aba.www//:ptth noitaicossAsreknaBnaciremA tiderC&yenoM /3h/sesaeler/vog.evreserlaredef.www//:ptth sevreseRetagerggA3.H tiderC&yenoM /6h/sesaeler/vog.evreserlaredef.www//:ptth serusaeMkcotSyenoM6.H tiderC&yenoM /8h/sesaeler/vog.evreserlaredef.www//:ptth sknaBlaicremmoC.S.UfoseitilibaiLdnastessA8.H tiderC&yenoM fdp.tispme/fdp/esaeler.swen/vog.slb.www//:ptth noitautiStnemyolpmEehT segaW&robaL 31

seireS laudividnI dna skcolB C noitamrofsnarT seireS esaeleR emaN kcolB 3 oitareulavotnaol:)ASN(seinapmocecnanfiotuatasnaolracweN tiderCremusnoC 1dexiM 3 )$(decnanfitnuomA:)ASN(seinapmocecnanfiotuatasnaolracweN tiderCremusnoC 1dexiM 3 )$folim(latot,secivresdoof&liateR :selaS selaSliateR 1dexiM 3 )ASN()$folib(sulprusroticfiedtvoglaredeF tnemetatSyrusaerT 1dexiM 3 )$folim(sisabsusneclatot,stropxeesidnahcremlatoT edarTesidnahcreM.S.U 1dexiM 3 )$folim(sisabsusneclatot,stropmiesidnahcremlatoT edarTesidnahcreM.S.U 1dexiM 3 )ASN()$folim()eulavFIC(stropmiesidnahcremlatoT edarTesidnahcreM.S.U 1dexiM 3 latoT esaeleRPI PI 3 seilppuslairtsudni-nondnastcudorPlaniF esaeleRPI PI 3 stcudorplaniF esaeleRPI PI 3 sdoogremusnoC esaeleRPI PI 3 sdoogremusnocelbaruD esaeleRPI PI 3 sdoogremusnocelbarudnoN esaeleRPI PI 3 tnempiuqessenisuB esaeleRPI PI 3 slairetaM esaeleRPI PI 3 selbarud,ygrenenon,slairetaM esaeleRPI PI 3 selbarudnon,ygrenenon,slairetaM esaeleRPI PI 3 )SCIAN(gfM esaeleRPI PI 3 )SCIAN(selbarud,gfM esaeleRPI PI 3 )SCIAN(selbarudnon,gfM esaeleRPI PI 3 )SCIAN(gniniM esaeleRPI PI 3 )SCIAN(seitilitU esaeleRPI PI 3 )SCIAN(latot,ygrenE esaeleRPI PI 3 )SCIAN(latot,ygrene-noN esaeleRPI PI 3 )SCIAN()PVM(strapdnaselcihevrotoM esaeleRPI PI 3 )SCIAN()SCC(srotcudnocimes,.piuqe .mmoc,sretupmoC esaeleRPI PI 3 )SCIAN(SCClcxeygrene-noN esaeleRPI PI 3 )SCIAN(PVMdnaSCClcxeygrene-noN esaeleRPI PI 2 )SCIAN(latoT :noitazilitUyticapaC esaeleRPI PI 2 )SCIAN(gfM :noitazilitUyticapaC esaeleRPI PI 2 )SCIAN(selbarud,gfM :noitazilitUyticapaC esaeleRPI PI 2 )SCIAN(selbarudnon,gfM :noitazilitUyticapaC esaeleRPI PI 2 gniniM :noitazilitUyticapaC esaeleRPI PI 2 seitilitU :noitazilitUyticapaC esaeleRPI PI 2 srotcudnocimes,.piuqe .mmoc,sretupmoC :noitazilitUyticapaC esaeleRPI PI 2 SCClcxegfM :noitazilitUyticapaC esaeleRPI PI 3 )suoht(latoT :detrats,gnisuohdenwo-yletavirP noitcurtsnoClaitnediseRweN 2dexiM 3 )suoht(latoT :dezirohtuagnisuohdenwo-yletavirpweN noitcurtsnoClaitnediseRweN 2dexiM 2 ytivitcalareneG :kooltuO SOBaihpledalihP 2dexiM 32

noitamrofsnarT seireS esaeleR emaN kcolB 2 sredroweN :kooltuO SOBaihpledalihP 2dexiM 2 stnempihS :kooltuO SOBaihpledalihP 2dexiM 2 seirotnevnI :kooltuO SOBaihpledalihP 2dexiM 2 sredrodellfinU :kooltuO SOBaihpledalihP 2dexiM 2 diapsecirP :kooltuO SOBaihpledalihP 2dexiM 2 deviecersecirP :kooltuO SOBaihpledalihP 2dexiM 2 tnemyolpmEkooltuO SOBaihpledalihP 2dexiM 2 sruohkroW :kooltuO SOBaihpledalihP 2dexiM 4 )atadIPPllarof001=2891(sdoogdehsinfi:IPP secirPrecudorP IPP 4 ygrenednadoofsselsdoogdehsinfi:IPP secirPrecudorP IPP 4 sdoogremusnocdehsinfi:IPP secirPrecudorP IPP 4 slairetametaidemretni:IPP secirPrecudorP IPP 4 slairetamedurc:IPP secirPrecudorP IPP 4 dooflcxesdoogdehsinfi:IPP secirPrecudorP IPP 4 ygrenesselslairetamdoofnonedurc:IPP secirPrecudorP IPP 4 ygrenesselslairetamedurc:IPP secirPrecudorP IPP 4 )nabru(smetilla:IPC secirPremusnoC IPC 4 segarevebdnadoof:IPC secirPremusnoC IPC 4 gnisuoh:IPC secirPremusnoC IPC 4 lerappa:IPC secirPremusnoC IPC 4 noitatropsnart:IPC secirPremusnoC IPC 4 eraclacidem:IPC secirPremusnoC IPC 4 seitidommoc:IPC secirPremusnoC IPC 4 selbarud,seitidommoc:IPC secirPremusnoC IPC 4 secivres:IPC secirPremusnoC IPC 4 doofsselsmetilla:IPC secirPremusnoC IPC 4 ygrenednadoofsselsmetilla:IPC secirPremusnoC IPC 4 retlehssselsmetilla:IPC secirPremusnoC IPC 4 eraclacidemsselsmetilla:IPC secirPremusnoC IPC 0 )egnahcylretrauqdezilaunna(htworgPDGlaeR esaeler-PDG emocnI&PDG 4 xedniecirpPDG esaeler-PDG emocnI&PDG 3 )$69deniahcfolim(latoT :edarT&gfM :selaS liated-PDG emocnI&PDG 3 )$69deniahcfolim(latot,gfM :edarT&gfM :selaS liated-PDG emocnI&PDG 3 )$69deniahcfolim(selbarud,gfM :edarT&gfM :selaS liated-PDG emocnI&PDG 3 )$69deniahcfolim(selbarudnon,gfM :edarT&gfM :selaS liated-PDG emocnI&PDG 3 )$69deniahcfolim(elaselohwtnahcreM :edarT&gfM :selaS liated-PDG emocnI&PDG 3 )$69deniahcfolim(selbarud,elaselohwtnahcreM :edarT&gfM :selaS liated-PDG emocnI&PDG 3 )$69deniahclim(selbarudnon,elaselohwtnahcreM :edarT&gfM :selaS liated-PDG emocnI&PDG 3 )$69deniahcfolim(edartliateR :edarT&gfM :selaS liated-PDG emocnI&PDG 3 )$69deniahcfolim(latoT,edarT&gfM :seirotnevnI liated-PDG emocnI&PDG 3 )$69deniahcfolim(gfM,edarT&gfM :seirotnevnI liated-PDG emocnI&PDG 3 )$69deniahcfolim(selbarud,gfM,edarT&gfM :seirotnevnI liated-PDG emocnI&PDG 33

noitamrofsnarT seireS esaeleR emaN kcolB 3 )$69deniahcfolim(selbarudnon,gfM,edarT&gfM :seirotnevnI liated-PDG emocnI&PDG 3 )$69deniahcfolim(elaselohwtnahcreM,edarT&gfM :seirotnevnI liated-PDG emocnI&PDG 3 )$69deniahcfolim(edartliateR,edarT&gfM :seirotnevnI liated-PDG emocnI&PDG 3 emocnilanosrepelbasopsidlaeR emocnIlanosreP emocnI&PDG 3 )$69deniahcfolib(latoT:ECP emocnIlanosreP emocnI&PDG 3 )$69deniahcfolib(selbaruD:ECP emocnIlanosreP emocnI&PDG 3 )$69deniahcfolib(selbarudnoN:ECP emocnIlanosreP emocnI&PDG 3 )$69deniahcfolib(secivreS:ECP emocnIlanosreP emocnI&PDG 3 )$69deniahcfolib(sotuaweN-PVM-selbaruD:ECP emocnIlanosreP emocnI&PDG 4 latoT :xedniecirpthgiewniahcECP emocnIlanosreP emocnI&PDG 4 ygrenednadooflcxelatot :secirpECP emocnIlanosreP emocnI&PDG 4 selbarud :secirpECP emocnIlanosreP emocnI&PDG 4 selbarudnon :secirpECP emocnIlanosreP emocnI&PDG 4 secivres :secirpECP emocnIlanosreP emocnI&PDG 3 )AS()suoht(stnempihsgfm–semoheliboM semoHderutcafunaM gnisuoH 3 )suoht(latoT :dlossesuohylimaf-1weN selaSlaitnediseRweN gnisuoH 3 etartnerruc@ylppusshtnom–sesuohylimaf-1weN selaSlaitnediseRweN gnisuoH 3 )suoht(doirepfodnetaelasrofsesuohylimaf-1weN selaSlaitnediseRweN gnisuoH 3 ytivitcalareneG :yevruSgfMtsewdiMdeFogacihC yevruSIMMdeFogacihC 1syevruS 2 ecnedfinocremusnocfoxednI xednIecnedfinoCremusnoC 1syevruS 2 tnemitnesremusnocfoxednI :yevruSnagihciM yevruSnagihciM 1syevruS 3 smialclaitiniylkeewgvA ).sruhTylkw(smialC smialClaitinI 2 egagtromdexfiraey-03nodleiytekramyramirP ).deWylkw(caMeidderF setaRtseretnI 2 etarsdnuflaredef :etartseretnI )yliad(51.H setaRtseretnI 2 )tekraM .ces(yrusaerTom-3.S.U :etartseretnI )yliad(51.H setaRtseretnI 2 )tekraM .ces(yrusaerTom-6.S.U :etartseretnI )yliad(51.H setaRtseretnI 2 )ytirutamtnatsnoc(yrusaerTraey-1 :etartseretnI )yliad(51.H setaRtseretnI 2 )ytirutamtnatsnoc(yrusaerTraey-5 :etartseretnI )yliad(51.H setaRtseretnI 2 )ytirutamtnatsnoc(yrusaerTraey-7 :etartseretnI )yliad(51.H setaRtseretnI 2 )ytirutamtnatsnoc(yrusaerTraey-01 :etartseretnI )yliad(51.H setaRtseretnI 2 etaroprocAAAsydooM :dleiydnoB )yliad(51.H setaRtseretnI 2 etaroprocAABsydooM :dleiydnoB )yliad(51.H setaRtseretnI 3 etaregnahcxeevitceffelanimoN 01.H laicnaniF 3 )2(SU/oruEtopS 01.H laicnaniF 3 SU/ZStopS 01.H laicnaniF 3 SU/napaJtopS 01.H laicnaniF 3 SU/KUtopS 01.H laicnaniF 3 SU/ACtopS 01.H laicnaniF 4 ).m.pehtnidedrocer(tekramnodnoLehtno)zo/$(dlogfoecirP )yliad(xiFMPnodnoL laicnaniF 3 xednietisopmocESYN ESYN laicnaniF 3 lairtsudni:ESYN ESYN laicnaniF 3 seitilitu:ESYN ESYN laicnaniF 34

noitamrofsnarT seireS esaeleR emaN kcolB 3 etisopmocP&S P&S laicnaniF 3 dleiydnedividP&S )ylkw(P&S laicnaniF 3 oitarE/PP&S )ylkw(P&S laicnaniF 3 xednietisopmocerihsliW )yliad(erihsliW laicnaniF 2 )IMP(xednIsreganaMgnisahcruP gnirutcafunaM-RGMP 2syevruS 2 )tnemeganaMylppuSrofetutitsnI(noitcudorp :xednigfmMSI gnirutcafunaM-RGMP 2syevruS 2 tnemyolpmE :xednigfmMSI gnirutcafunaM-RGMP 2syevruS 2 seirotnevni :xednigfmMSI gnirutcafunaM-RGMP 2syevruS 2 sredrowen :xednigfmMSI gnirutcafunaM-RGMP 2syevruS 2 seireviledsreilppus :xednigfmMSI gnirutcafunaM-RGMP 2syevruS 3 )$folim(latoT :gnidnatstuodne-htnomrepaplaicremmoC repaPlaicremmoC 3dexiM 3 )$tnerrucfolim(latoT :ecalpnitupnoitcurtsnoC ecalPnituPnoitcurtsnoC 3dexiM 3 )$tnerrucfolim(etavirP :ecalpnitupnoitcurtsnoC ecalPnituPnoitcurtsnoC 3dexiM 3 )$folim(seirtsudnisdoogelbaruD :sredrOweN 3M/selbaruDecnavdA 3dexiM 3 )$folim(sdooglatipacesnefednoN :sredrOweN 3M/selbaruDecnavdA 3dexiM 3 )$folim(seirtsudnignirutcafunamllA :sredrOweN 3M 3dexiM 3 )$folim(sredrodellfinu/wseirtsudnignirutcarunamllA :sredrOweN 3M 3dexiM 3 )$folim(seirtsudnisdoogelbarudnoN :sredrOweN 3M 3dexiM 3 )$folim(seirtsudnignirutcafunamllA :sredrOdellfinU 3M 3dexiM 3 snaoltnemllatsniremusnocdleh-knabnoetarycneuqnileD nitelluB .qnileDremusnoC tiderC&yenoM 3 )$folim(esabyratenoM 3.H tiderC&yenoM 3 )$folim(latoT :sevresersnoitutitsniyrotisopeD 3.H tiderC&yenoM 3 )$folim(deworrobnon :snoitutitsniyrotisopeD 3.H tiderC&yenoM 3 )$folim(1M 6.H tiderC&yenoM 3 )$folim(2M 6.H tiderC&yenoM 3 )$folim(3M 6.H tiderC&yenoM 3 )$folim(latoT :sknablaicremmoclla@seitiruceSdnasnaoL 8.H tiderC&yenoM 3 )$folim(latot,seitiruceS :sknabmmoclla@seitiruceSdnasnaoL 8.H tiderC&yenoM 3 )$folim(tvog.S.U,seitiruceS :sknabmmoclla@seitiruceSdnasnaoL 8.H tiderC&yenoM 3 )$folim(snaoletatselaeR :sknabmmoclla@seitiruceSdnasnaoL 8.H tiderC&yenoM 3 )$folim(snaolsudnIdnammoC :sknabmmoclla@seitiruceSdnasnaoL 8.H tiderC&yenoM 3 )$folim(snaolremusnoC :sknabmmoclla@seitiruceSdnasnaoL 8.H tiderC&yenoM 2 etartnemyolpmenU noitautiStnemyolpmE segaW&robaL 2 etarnoitapicitraP noitautiStnemyolpmE segaW&robaL 3 tnemyolpmenufonoitarudnaeM noitautiStnemyolpmE segaW&robaL 3 skeew5nahtsseldeyolpmenusnosreP noitautiStnemyolpmE segaW&robaL 3 skeew41ot5deyolpmenusnosreP noitautiStnemyolpmE segaW&robaL 3 skeew62ot51deyolpmenusnosreP noitautiStnemyolpmE segaW&robaL 3 skeew+51deyolpmenusnosreP noitautiStnemyolpmE segaW&robaL 3 latoT :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 etavirplatoT :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 gnicudorp-sdooG :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 35

noitamrofsnarT seireS esaeleR emaN kcolB 3 gniniM :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 noitcurtsnoC :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 gnirutcafunaM :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 selbarud,gnirutcafunaM :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 selbarudnon,gnirutcafunaM :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 gnicudorp-ecivreS :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 gnisuoherawdnanoitatropsnarT :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 seitilitU :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 edartliateR :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 edartelaselohW :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 seitivitcalaicnaniF :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 secivresssenisubdnalanoisseforP :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 secivreshtlaehdnanoitacude :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 ytilatipsohdnaerusiel :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 secivresrehtO :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 tnemnrevoG :slloryapganonnotnemyolpmE noitautiStnemyolpmE segaW&robaL 3 etavirplatoT :srekrowyrosivrepusnonfonoitcudorpfo .srhylkeewgvA noitautiStnemyolpmE segaW&robaL 3 gfM:WNPfosrhylkeewgvA noitautiStnemyolpmE segaW&robaL 3 gfM:WNPfosrhemitrevoylkeewgvA noitautiStnemyolpmE segaW&robaL 4 )$(larutlucirganonlatoT :sgninraeylruohgvA noitautiStnemyolpmE segaW&robaL 4 )$(noitcurtsnoc :sgninraeylruohgvA noitautiStnemyolpmE segaW&robaL 4 )$(gfM :sgninraeylruohgvA noitautiStnemyolpmE segaW&robaL 4 )$(noitatropsnarT :sgninraeylruohgvA noitautiStnemyolpmE segaW&robaL 4 )$(edartliateR :sgninraeylruohgvA noitautiStnemyolpmE segaW&robaL 4 )$(edartelaselohw :sgninraeylruohgvA noitautiStnemyolpmE segaW&robaL 4 )$(etatselaerdna,ecnarusni,ecnanfi :sgninraeylruohgvA noitautiStnemyolpmE segaW&robaL 4 )$(secivresssenisubdnalanoisseforp :sgninraeylruohgvA noitautiStnemyolpmE segaW&robaL 4 )$(secivreshtlaehdnanoitacude :sgninraeylruohgvA noitautiStnemyolpmE segaW&robaL 4 )$(secivresrehto :sgninraeylruohgvA noitautiStnemyolpmE segaW&robaL 36

D Tables Table 2a: Average Size of the news for GDP growth rate Blocks v first month (m=1) second month (m=2) third month (m=3) b Mixed 1 0.104 0.081 0.081 Industrial Production 0.527 0.427 0.531 Mixed 2 0.676 0.179 0.127 PPI 0.073 0.038 0.050 CPI 0.100 0.064 0.056 GDP and Income 0.042 0.030 0.071 Housing 0.006 0.006 0.009 Surveys 1 0.414 0.205 0.135 Initial Claims 0.087 0.136 0.058 Interest Rates 0.489 0.764 0.583 Financial 0.166 0.067 0.076 Surveys 2 0.256 0.167 0.112 Mixed 3 0.007 0.010 0.004 Money & Credit 0.040 0.040 0.037 Labor and Wages 0.362 0.241 0.244 Table 2b: Average Size of the news for GDP Deflator inflation Blocks v first month (m=1) second month (m=2) third month (m=3) b Mixed 1 0.002 0.001 0.001 Industrial Production 0.029 0.027 0.023 Mixed 2 0.033 0.009 0.015 PPI 0.032 0.016 0.018 CPI 0.040 0.017 0.017 GDP and Income 0.160 0.015 0.032 Housing 0.001 0.001 0.001 Surveys 1 0.028 0.016 0.012 Initial Claims 0.002 0.003 0.002 Interest Rates 0.003 0.014 0.019 Financial 0.035 0.031 0.021 Surveys 2 0.008 0.008 0.006 Mixed 3 0.000 0.001 0.000 Money & Credit 0.002 0.001 0.001 Labor and Wages 0.006 0.010 0.009 37

Table 3a: Average uncertainty for GDP growth rate first month (m=1) second month (m=2) third month (m=3) Blocks total common total common total common Labor and Wages 1.305 1.004 1.067 0.669 0.902 0.351 (0.027) (0.027) (0.019) (0.018) (0.013) (0.010) Mixed 1 1.303 1.002 1.064 0.665 0.901 0.347 (0.027) (0.027) (0.019) (0.019) (0.013) (0.011) Industrial Production 1.290 0.985 1.043 0.631 0.888 0.311 (0.028) (0.028) (0.018) (0.017) (0.012) (0.009) Mixed 2 1.219 0.890 0.997 0.550 0.873 0.265 (0.024) (0.025) (0.016) (0.016) (0.010) (0.007) PPI 1.219 0.889 0.996 0.550 0.873 0.265 (0.024) (0.026) (0.016) (0.016) (0.010) (0.007) CPI 1.219 0.889 0.996 0.549 0.872 0.264 (0.025) (0.026) (0.017) (0.017) (0.011) (0.007) GDP and Income 1.218 0.889 0.995 0.548 0.872 0.263 (0.025) (0.026) (0.016) (0.016) (0.011) (0.007) Housing 1.218 0.889 0.995 0.548 0.872 0.263 (0.025) (0.026) (0.016) (0.016) (0.011 (0.007) Surveys 1 1.196 0.858 0.982 0.523 0.868 0.248 (0.024) (0.024) (0.016) (0.015) (0.011) (0.006) Initial Claims 1.179 0.834 0.969 0.499 0.863 0.232 (0.028) (0.030) (0.016) (0.014) (0.011) (0.006) Interest Rates 1.110 0.733 0.925 0.406 0.847 0.159 (0.022) (0.022) (0.013) (0.016) (0.011) (0.009) Financial 1.106 0.727 0.922 0.400 0.846 0.156 (0.023) (0.023) (0.014) (0.015) (0.011) (0.009) Surveys 2 1.096 0.712 0.916 0.387 0.844 0.147 (0.021) (0.021) (0.013) (0.012) (0.010) (0.006) Mixed 3 1.096 0.712 0.916 0.386 0.844 0.147 (0.021) (0.021) (0.013) (0.012) (0.010) (0.006) Money & Credit 1.095 0.711 0.916 0.386 0.844 0.146 (0.021) (0.021) (0.013) (0.012) (0.010) (0.006) Labor and Wages 1.072 0.675 0.902 0.351 0.840 0.121 (0.020) (0.019) (0.012) (0.009) (0.009) (0.012) 38

Table 3b: Average uncertainty for GDP deflators first month (m=1) second month (m=2) third month (m=3) Blocks total common total common total common Labor and Wages 0.156 0.062 0.110 0.042 0.105 0.024 (0.007) (0.009) (0.005) (0.006) (0.004) (0.004) Mixed 1 0.156 0.062 0.110 0.042 0.105 0.024 (0.007) (0.009) (0.005) (0.006) (0.004) (0.004) Industrial Production 0.155 0.061 0.109 0.041 0.104 0.023 (0.007) (0.009) (0.005) (0.006) (0.004) (0.004) Mixed 2 0.154 0.056 0.108 0.037 0.104 0.019 (0.007) (0.008) (0.004) (0.005) (0.004) (0.003) PPI 0.153 0.055 0.107 0.036 0.104 0.019 (0.007) (0.008) (0.004) (0.005) (0.004) (0.003) CPI 0.153 0.055 0.107 0.035 0.103 0.018 (0.007) (0.008) (0.004) (0.005) (0.004) (0.003) GDP and Income 0.115 0.054 0.107 0.035 0.103 0.018 (0.006) (0.008) (0.004) (0.005) (0.004) (0.003) Housing 0.115 0.054 0.107 0.035 0.103 0.018 (0.006) (0.008) (0.004) (0.005) (0.004) (0.003) Surveys 1 0.113 0.051 0.106 0.033 0.103 0.016 (0.006) (0.007) (0.004) (0.005) (0.004) (0.003) Initial Claims 0.113 0.051 0.106 0.032 0.103 0.015 (0.006) (0.007) (0.004) (0.004) (0.003) (0.002) Interest Rates 0.113 0.050 0.106 0.031 0.103 0.014 (0.005) (0.007) (0.004) (0.004) (0.003) (0.002) Financial 0.109 0.042 0.104 0.024 0.102 0.010 (0.005) (0.006) (0.004) (0.003) (0.003) (0.001) Surveys 2 0.109 0.041 0.104 0.024 0.102 0.010 (0.005) (0.006) (0.004) (0.003) (0.003) (0.001) Mixed 3 0.109 0.041 0.104 0.024 0.102 0.010 (0.005) (0.006) (0.004) (0.003) (0.003) (0.001) Money & Credit 0.109 0.041 0.104 0.024 0.102 0.010 (0.005) (0.006) (0.004) (0.003) (0.003) (0.001) Labor and Wages 0.109 0.041 0.104 0.023 0.102 0.009 (0.005) (0.006) (0.004) (0.003) (0.003) (0.001) 39

Table 4a: Uncertainty for GDP growth rate (04Q1) (counterfactual) v˜=04m1 v˜=04m2 v˜=04m3 Blocks common total common total common total no release 0.861 1.189 0.554 0.990 0.271 0.864 Mixed 1 0.784 1.135 0.488 0.954 0.230 0.852 Industrial Production 0.570 0.998 0.285 0.868 0.079 0.824 Mixed 2 0.733 1.100 0.446 0.933 0.205 0.845 PPI 0.837 1.172 0.536 0.980 0.261 0.860 CPI 0.820 1.159 0.522 0.972 0.253 0.858 GDP and Income 0.608 1.021 0.331 0.884 0.000 0.000 Housing 0.825 1.163 0.525 0.973 0.254 0.858 Surveys 1 0.734 1.101 0.446 0.934 0.204 0.845 Initial Claims 0.803 1.148 0.505 0.963 0.241 0.855 Financial 0.837 1.171 0.535 0.979 0.260 0.860 Interest Rates 0.715 1.088 0.422 0.922 0.184 0.840 Surveys 2 0.736 1.102 0.449 0.935 0.207 0.846 Mixed 3 0.750 1.111 0.456 0.938 0.209 0.846 Money & Credit 0.835 1.170 0.532 0.977 0.258 0.860 Labor and Wages 0.641 1.041 0.357 0.894 0.140 0.832 Table 4b: Uncertainty for GDP deflators (04Q1) (counterfactual) v˜=04m1 v˜=04m2 v˜=04m3 Blocks common total common total common total no release 0.055 0.110 0.035 0.101 0.018 0.097 Mixed 1 0.054 0.109 0.035 0.101 0.017 0.096 Industrial Production 0.043 0.104 0.025 0.098 0.011 0.096 Mixed 2 0.052 0.108 0.032 0.100 0.015 0.096 PPI 0.039 0.103 0.022 0.097 0.011 0.096 CPI 0.036 0.102 0.020 0.097 0.009 0.095 GDP and Income 0.040 0.103 0.023 0.098 0.000 0.000 Housing 0.053 0.109 0.034 0.101 0.017 0.096 Surveys 1 0.052 0.108 0.032 0.100 0.016 0.096 Initial Claims 0.055 0.109 0.035 0.101 0.017 0.096 Financial 0.041 0.103 0.024 0.098 0.012 0.096 Interest Rates 0.053 0.109 0.034 0.101 0.017 0.096 Surveys 2 0.052 0.108 0.032 0.100 0.015 0.096 Mixed 3 0.052 0.108 0.033 0.101 0.017 0.096 Money & Credit 0.054 0.109 0.034 0.101 0.017 0.096 Labor and Wages 0.049 0.107 0.030 0.100 0.014 0.096 40