The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
orKing raper series Consumer Confidence and Economic Fluctuations John G. Matsusaka and Argia M. Sbordone Working Papers Series Macroeconomic Issues Research Department Federal Reserve Bank of Chicago November 1993 (WP-93- 13) FEDERAL RESERVE BANK OF CHICAGO CONSUMER CONFIDENCE AND ECONOMIC FLUCTUATIONS JOHN G. MATSUSAKA D e p a r tm e n t o f F in a n ce a n d B u sin e ss E c o n o m ic s S ch o o l o f B u sin e ss A d m in is tr a tio n U n iv e r s ity o f S o u th e rn C a lifo rn ia L o s A n g e le s, C a lifo rn ia 9 0 0 8 9 -1 4 2 1 ARGIA M. SBORDONE E c o n o m ic R esea rch D e p a r tm e n t F ed era l R e se rv e B a n k o f C h ica g o C h icago, Illin ois 6 0 6 0 4 -1 4 1 3 I f co n su m e rs b e c o m e p e s s im is tic a b o u t th e s t a t e o f th e eco n o m y, can th e r e b e a s lo w d o w n in o u tp u t, even i f th e ir p e s s im is m is n o t b a se d on e c o n o m ic fu n d a m e n ta ls ? A r e c e n t class o f m a c ro e c o n o m ic m o d e ls sh o w s th e a n sw e r is y e s , i f th e r e a re “s t r a t e g ic c o m p le m e n ta r itie s ” a n d m u ltip le eq u ilib ria . In th is p a p e r w e in v e s tig a te th e lin k b e tw e e n c o n su m e r co n fid en ce a n d e c o n o m ic flu c tu a tio n s fo r th e p e r io d 1 9 5 3 -1 9 8 8 u sin g v e c to r a u to re g re ssio n s. In all m o d e ls, a fte r c o n tro llin g fo r va ria b le s th a t p r o x y fo r e c o n o m ic fu n d a m e n ta ls, th e h y p o th e s is th a t c o n su m e r s e n tim e n t d o e s n o t cau se G N P (in th e G ra n g e r se n se ) can b e r e je c te d . V arian ce d e c o m p o s itio n s s u g g e st th a t c o n s u m e r s e n tim e n t a c c o u n ts fo r b e tw e e n 13 p e r c e n t a n d 2 6 p e r c e n t o f th e in n o v a tio n v a r ia n c e o f G N P . *We thank Joonmo Cho, Giorgio De Santis, Robert Erikson, Selahattin Imrohoroglu, Douglas Joines, Narayana Kocherlakota, Robert Kollman, Fulvio Ortu, Ian Parry, Christo pher Phelan, Pedro Teles, Mark Watson, Michael Woodford, anonymous referees, and work shop participants at The University of Chicago for many helpful comments. September 1993 I. INTRODUCTION The idea that consumer sentiment might cause output fluctuations is popular in the business press.1 It is easy to see where this belief comes from: in Figure 1 we plot an index of consumer sentiment against recessions for the postwar period.2 The pattern is striking; all recessions were preceded by a fall in confidence, and all major falls in consumer sentiment were followed by a recession (except in 1965 which, while not a recession, was the so-called “growth recession.”) Apparently either consumers were correctly forecasting output falls, or declines in consumer sentiment were inducing declines in output. It has been recognized for some time that business cycle and growth models can exhibit multiple (Nash) equilibria if they contain strategic complementarities.3 In a conventional single equilibrium model, output fluctuates only in response to changes in economic funda mentals, where the list of fundamentals includes technology, government purchases, money supply, the price of oil, and so on. In a multiple equilibria model output responds to funda mentals, but in addition there can be fluctuations as the economy shifts between equilibria. Perhaps the most intriguing feature of these models is that output can fluctuate simply because everyone expects it to. Put differently, expectations can be self-fulfilling in that if people expect bad times they get them. Thus, this class of models offers a theoretical rationalization of how consumer sentiment can affect GNP. Although these models intro duce dramatically different policy considerations than single equilibrium models, many of their observable implications are the same. As a result, their empirical relevance is largely unknown. In this paper, we study the behavior of the U.S. economy for the period 1953-1988. Our central purpose is to evaluate empirically how much truth there is, if any, to the idea that consumer sentiment causes fluctuations in GNP. Following Granger and others we take causality to imply temporal precedence, that is, if one series causes another, then movements in the forcing series precede movements in the forced series. We verify that the temporal ordering suggested in Figure 1, consumer sentiment leads GNP, is statistically significant. 1 Examples from the last recession are “Economy in the U.S. Isn’t Nearly as Sour As the Country’s Mood. But Pessimism Could Become A Self-Fulfilling Prophecy Further Stalling Recovery. Can Attitude Be Ev erything?” on the front page of The Wall Street Journal, November 4, 1991, and “Recession: Case of Bad Attitude? President Bush says people should be more confident—and that would restore economic vigor. But consumers cite layoffs and lost wages,” from the front page of the Los Angeles Times, November 9, 1991. 2 A recession as defined by the NBER is indicated with a vertical bar. 3 Roughly speaking, a model contains strategic complementarities if each agent’s optimal action is positively correlated with the action of other agents. A representative but incomplete list of the literature would include Ball and Romer [1991], Bryant [1983, 1987], Cass and Shell [1983], Kiyotaki [1985], Milgrom and Roberts [1990], Murphy, Shleifer, and Vishny [1989], Roberts [1987], Shleifer [1986], and Weil [1989]. Cooper and John [1988] provide a good overview. On theoretical objections see Woodford [1987]. 1 Temporal precedence is not enough for our purposes, however, because the evidence that movements in consumer confidence precede movements in output can be interpreted in two ways. Either consumer sentiment causes GNP or it simply a n tic ip a te s GNP. We attem pt to disentangle the two possibilities by estimating a series of vector au toregressions including GNP, consumer sentiment, and various other series that capture fundamentals or are good predictors of GNP. The working assumption is that if consumer sentiment can forecast GNP movements even after controlling for fundamentals and other publicly-available predictors of output, some support is provided for the idea that consumer sentiment causes output fluctuations. In our baseline regressions, the control variable is an index of leading indicators, composed of fundamentals and purely predictive variables. To check for robustness we consider a number of other controls, a “textbook” set of funda mentals that includes government spending, money supply, and sensitive materials prices; components of the leading indicators; and a default risk variable that has been identified as an excellent predictor of output movements. For each set of vector autoregressions we test whether consumer sentiment Granger causes GNP. Our main finding is that even after controlling for economic fundamentals and other good predictors of GNP, changes in consumer sentiment have a statistically significant effect on output fluctuations. In other words, we find evidence of Granger causality running from consumer sentiment to GNP. We also attempt to assess the economic significance of consumer sentiment by decomposing the variance of GNP innovations. Our second finding is that while sentiment is not the most important factor in GNP fluctuations it plays a quantitatively significant role: between 13 percent and 26 percent of GNP innovation variance can be attributed to variations in consumer sentiment. The evidence suggests that consumer confidence is an important independent factor in economic fluctuations, and more tentatively, it appears to provide some support for the class of macroeconomic models with strategic complementarities. The paper proceeds as follows. The next section provides a simple illustrative model that exhibits multiple equilibria for some parameter values. Its main features are consistent with the bulk of the macroeconomics literature on strategic complementarities, but we hope it is more transparent. The model also calls attention to some interesting features of actual economies that have been underexplored in this context. Section III discusses the related literature. In Section IV we present the evidence. Granger causality tests are reported that assess how important, if at all, consumer sentiment is in determining the level of aggregate production. Section V draws a few conclusions. 2 II. AN ILLUSTRATIVE MODEL It is easier to discuss the empirical procedures with a concrete example in mind. In this section we sketch a stylized model of output fluctuations that exhibits the basic features of multiple equilibria models. The model is amenable to generalization in a number of directions that we do not pursue; our goal is to illustrate the central issues in as simple a manner as possible. The model is motivated by two features of actual economies that have not received much attention in the recent macroeconomics literature, (i) many goods are produced to buyer specifications, and (ii) once goods are customized they become less valuable to other buyers. In fact, production to order is important in such industries as construction, machine tools, steel, women’s apparel, and durables manufactures in general [Zarnowitz, 1973]. What makes this interesting is that these are the industries where the business cycle primarily occurs. We model customization by assuming there is one consumption good, call it a house, that a person can have made to her specifications—she can choose the color of its paint. She can also buy a house on the resale market or from developers who build without having a buyer lined up. Because a resold house and a developer’s house are not specialized to her tastes—the paint is the wrong color—they give her lower utility. In our one period model consumers submit orders for custom-made goods at the start of the period, keeping their expected income in mind. If they order more than they can afford, they must resell part of their order, which is costly because other buyers place a relaitvely low value on the custom features. If they order too little, they must enter the resale market as buyers, which means they must buy goods whose custom features they dislike. Because consumers strive to match their orders with their income, expectations can be self-fulfilling: if, for some reason, people expect low incomes they submit low orders, leading the build-to-order firms to have low employment, and consumers to have low incomes. We assume there is a continuum of identical agents over the interval [0,1]. Agent i has a utility function of u ( i ) — — where l(i) is her labor and c(i ) is her consumption value. To capture the idea of specialized goods, her consumption value is written c(i) = co{i) + r c \ { i ) where co(i) is goods she has custom-made and ci(i) is goods that are not custom-made. She prefers goods that are customized: r < 1. There is a competitive production sector that produces Y goods where Y = A / q l ( i ) di. All integrals that follow are over [0,1]. Firms are jointly and equally owned and all profits are immediately returned as dividends (although in equilibrium profits are zero). The aggregate production function (we sometimes call it “the firm”) is completely described by A which represents economic fundamentals. The production technology is the same for custom goods 3 and generic goods so /co(z)di + /cx(z)di = Y . One way to think of this is that the cost of painting a house is the same for every color. If labor demand exceeds labor supply we assume that employment is rationed equally among workers.4 The sequence of actions in a period is as follows. First, each agent i submits orders for a quantity q ( i ) of custom-made goods taking as given the price of custom-made and generic goods. This is the upper bound on person i’s consumption of custom goods, that is, co(i ) < q( i ) . After orders are placed the firms produce and pay wages. Then workers are obligated to take delivery of their orders and a secondary market for consumption goods opens, which we call a resale market. Workers who do not have sufficient funds to pay for their orders must enter the resale market and sell part of their goods to pay off their debts. In addition, firms can sell whatever non-customized goods they produced. When trading is concluded consumption occurs. Let p o be the price of custom-made goods, p, be the price of goods in the resale market, and w be the wage rate. We normalize po = 1 and think of the wages paid by the firm as claims to a unit of the consumption good. These claims can be redeemed for one unit of the good which puts a ceiling on the equilibrium price in the resale market: p, < 1. In the resale period, if there are consumers who ordered more than they can afFord, they attem pt to raise money by selling some of their orders. They will be met by people who ordered less than they can afford, and are trying to spend their extra income (claims). The interactions of these buyers and sellers determines pt . It turns out that there is rational expectations equilibrium for any p, 6 [0, l].5 The rest of this example is confined to the case where px = r. This is an appealing case because goods in the resale market are discounted just enough to compensate for the lower utility they deliver. The next two propositions show that the behavior of the economy depends in an impor tant way on the value of r. In particular, when r = 1 there is a unique equilibrium where output is determined only by fundamentals. When r < 1 there is a continuum of equilibria and expectations have self-fulfilling effects on the level of output. Here and below an aster isk is used to denote equilibrium quantities, for example, equilibrium utility for agent i is «*(<) = *■(•■) - !*-(*?• PROPOSITION 1. If r = 1 there is a unique equilibrium with Y = A 2. One way this can be rationalized is by assuming that each worker produces a separate component of the consumption good. A more formal way to model this behavior would be to write the production technology as y=>lminj l(i). In keeping with the spirit of our model as an example, we eschew this additional formality because it makes the same point while introducing a heavier burden of notation. 0 We shall not prove this here, but note that it follows from the fact that no equilibrium has an active market in both customized and non-customized goods. In practice, exactly where the resale price settles will depend on the structure of the resale market and the nature of the bargaining that takes place. 4 4 If r = p i = 1 the firm finds it equally profitable to sell in the custom market and the resale market. Then the firm’s profit is A f l(i) di —w J l( i) di. Equilibrium requires w m = A and the firm hires as many as will work at that wage. From the consumer’s maximization problem, l m(i) = w ’ — A . Then Y m = A f /*(i) di = A 2. Q .E .D . P ro o f. The intuition behind the proposition is that when p x = r = 1 the price in the ordered goods market is the same as the price in the resale market so from the firm’s point of view a good produced to order is as valuable as a good produced to stock. Then it pays a wage equal to A and hires as much labor as is supplied at that wage. Given the utility function, each person supplies A units of labor. The equilibrium is Pareto optimal. In the situation described by Proposition 1 there is a unique equilibrium. There is no sense in which consumer expectations affect production. Output fluctuates only in response to changes in economic fundamentals, A . In these situations, when there are no customizing effects, the model behaves in what might be called the conventional way. The next proposition states that when r < 1 there is a continuum of equilibria. The argument has two parts. First, the firm produces only to order, that is, it produces only as many goods as are ordered in the customized market. The reason is that if the firm produces for the resale market the wage rate is p \ A , and after working enough to supply the custommade demand, agents will not supply additional labor at the wage p \ A . Second, given their income, which they take as exogenous, consumers order exactly as many customized goods as they can pay for. They lose if they enter the resale market either as buyers or sellers. In these cases a coordination failure occurs; if everyone could get together and jointly submit high orders efficiency would be attained. PROPOSITION 2. If r < 1 there is an equilibrium with Y = SA2 for any S € (r, 1). Consider an arbitrary 5 € (r, 1 ). The proof demonstrates that there is an equilibrium for S where q m(i) = S A , l* (i) = S'A, and w m — A . In the proposed equilibrium person i is willing to supply labor of l( i) = A but the demand is only l* (i) = S A , according to the rationing rule. At the margin if the firm adds a unit of labor it generates revenue p xA . But the marginal disutility of labor at l* {i) is S A > p xA . Agents are atomistic so they take /*(i) and w ‘ as given. If person i orders q ( i ) > q * (i) then she has to raise q (i) — q’ (i) in the resale market to cover her debts. This costs her P ro o f. {q (i) —9*(0)/Pi in consumption. Her utility is then “(0 = 9(0 “ (?(0 “ 9'(0)/Pi ~ ^"(O2 = 9“(0 “ ^ * (0 2 + (1 - l/? i)(9 (0 “ 9‘ (0) < «*(0» where the inequality follows because w * l‘ ( i) — q (i) = q ’ ( i) — q (i) px = r < 1 . If she orders q (i) < q~{i) to spend in the resale market. This buys Cx(z) = 5 then she has (q m(i) — q { i ) ) / p i - Her utility is then u(t) = q(i) + r ( q M(i) - q ( i ) ) / p i - = q*(i) - \ l ‘ ( i )2 = u*(i). Thus, person i has no incentive to place higher or lower orders. Q. E. D. The proposition implies that the level of output is determined not only by economic fundamentals, A , but also by an arbitrary parameter, 5, which can be thought of as consumer sentiment. In an important sense it can be said that consumer expectations c a u se output. If consumers believe the equilibrium will be S then it is individually rational for them to take actions that bring about the S equilibrium. The interesting conclusion is that aggregate production can fluctuate in response to waves of optimism and pessimism, even if the underlying technology does not change. Perhaps more intriguing, in view of the recent behavior of the U.S. economy, is that the model suggests GNP can become stalled at low levels even when there is nothing wrong with the fundamentals of the economy. III. RELATED LITERATURE How can we tell if strategic complementarities are a real concern, as the business press believes, or are merely of theoretical interest? One approach would be to try to measure r or p 0 — p i . Intuition suggests it is easy for parameter values to lie in the necessary range. For example, suppose a person orders a house built. She has many options, what color paint, what fixtures, how many rooms, what kind of carpeting or floors, and so on. If she has to sell the house, probably she will need to give the buyer a price break because her custom features are not the buyer’s first choice. It is plausible, then, that p i < p 0 for durable goods; if r < 1 there are multiple equilibria. An obvious problem with this way of examining the theory is that it is intimately linked to our model, which is quite stylized. Still, this example suggests that one way to test for multiple equilibria is to estimate directly the parameters of the model. Dagsvik and Jovanovic [1991] is a nice example of this approach. They estimate a fairly general Phillips curve capable of exhibiting multiple equilibria; then the parameters are examined to see if they lie in the multiple equilibria region of the parameter space. Their evidence indicates unique equilibria in all three time periods they investigate: 1901-1940, 1921-1940, and 1951-1986. The authors note that their approach is limited in that they are forced to posit an exogenous and constant equilibrium selection mechanism .6 Oh and Waldman [1991, 1993] take a completely different approach, one which is closer in spirit to ours. Rather than try to estimate a particular model, they identify a general 6 im r o h o r o g lu [1993] u se s a sim ila r a p p ro a ch to lo o k for n o n -fu n d a m e n ta ls in th e 1 923 G e r m a n h y p erin fla tio n , w ith la r g e ly n e g a tiv e r e su lts. 6 characteristic of coordination failure models and look for behavior consistent with that fea ture. Cooper and John [1988] show that what drives the multiple equilibria property of all these models is “strategic complementarities”: if everyone but person i takes the “high” action (in our model, submits high orders) then it is in person i ’s interest to take the high action as well. Oh and Waldman observe that when there are strategic complementarities people have an incentive to try to catch the waves: they want to invest when output is high and cut back when it is low. Therefore, in a world with strategic complementarities investment and output should respond to what people expect output to be. Their key insight is that if there is an announcement that the economy is about to boom and everyone believes it then future output should be high, even if the announcement is based on false information. In effect, people use the announcements to coordinate. As a measure of false forecasts, Oh and Waldman use revisions of the government’s Index of Leading Indicators (ILI). They look to see if these revisions had an effect on fu ture economic activity. Their evidence is mixed but more often than not it supports the idea that ILI revisions predict movements in industrial production, consistent with strategic complementarities. However, as they note, if the ILI provides new information to firms and consumers (that is, information other than what equilibrium is selected), then revisions can have effects even when there are no strategic complementarities In this paper we focus on another implication of multiple equilibria models: expecta tions are self-fulfilling. Specifically, we estimate a number of vector autoregressions and ask whether consumer sentiment about the future Granger causes GNP.' The basic intuition behind the test comes from Proposition 2, which shows that the level of output is Y = SA2. In contrast to models with a unique equilibrium, in a multiple equilibrium model consumer sentiment (5) has an effect on output (F) even after controlling for fundamentals (A). All coordination failure models generate an output relation analogous to Proposition 2 because consumers must expect to be in a particular equilibrium in order for the economy to move there. IV. EVIDENCE A. D e s c r ip tio n o f th e D a ta a n d V a riables To measure consumer sentiment, we use the Index of Consumer Sentiment (ICS) con structed by the Survey Research Center at the University of Michigan. The ICS captures7 7 O u r ap p ro a ch is r e la te d to an o ld er lite r a tu r e th a t in v e stig a te d th e a b ility o f co n su m er a tt itu d e s to p r e d ic t p u rch a ses o f d u ra b le g o o d s, for e x a m p le , F riend an d A d a m s [1964]. 7 consumer confidence about the economic present and future from survey responses to the five questions listed in Panel A of Table I .8 The output series is GNP. We also performed the tests substituting the value of goods output for GNP, and arrived at similar conclusions. The main control variable is the Index of Leading Indicators (ILI), compiled by the Bureau of Economic Analysis at the U.S. Department of Commerce. We use the version of the ILI that expired in 1989—after 1989, the ILI includes the Index of Consumer Expectations, which is based on the same questions as the ICS [Hertzberg and Beckman, 1989]. The series comprising the ILI are indicated in Panel B of Table I. In various regressions we also include as controls government expenditure, the components of the ILI, and the difference between the rate of interest on six-month commercial paper and the rate of interest on six-month Treasury bills, which we call “default risk.” The data run quarterly from 1953:1 to 1988:4, except for the default risk variable which runs from 1959:1. GNP and the ILI are seasonally adjusted at the source; the ICS does not appear to have a seasonal component. GNP and the ILI are expressed in log differences. This transformation makes the series approximately mean and variance stationary. The questions comprising the ICS ask about changes as well as about levels, suggesting that the ICS is a measure of change in sentiment. Given this and the fact that the ICS is trendless, the estimates use the ICS in levels. Default risk is also expressed in levels. Rather than take a strong stand on a particular number of lags, in most cases we report estimates from models with lags of 1, 2, 3, and 4 quarters, although we find that the 4 lags model tends to provide the best fit. The effects under consideration are weak with more than 4 lags. Ten observations are missing from the ICS for the 1950s and one for the 1960s. This makes for 133 observations. The data are published by the U.S. Department of Commerce in S u r v e y o f C u rre n t B u sin e ss. B. O u t p u t a n d C o n s u m e r S e n tim e n t Causality is a difficult concept in science and particularly so in economics where we seldom have the luxury of controlled experiments. The most common empirical approaches in economics to causality inference are based on the idea of “Granger causality.” Roughly speaking, series ments in y. x is said to Granger cause series y if movements in x help to predict move Granger causality has the look of “true” causality so we believe it provides a 8 T h e S u r v e y R e se a r c h C e n ter p u b lish e s a se c o n d in d e x c a lled th e In d e x o f C o n su m e r E x p e c ta tio n s (I C E ), b a s e d o n th e a n sw e r s to q u e stio n s 2, 3 , a n d 4 . W e u se th e IC S in s te a d o f th e IC E for tw o r e a so n s. F ir st, w e b e lie v e th e r e sp o n s e s to q u e stio n s 1 an d 5 g iv e a fu ller p ic tu r e o f c o n su m e r s e n tim e n t. S e c o n d , w e fo u n d t h a t th e IC E g a v e u n s ta b le r e su lts a cro ss su b sa m p le s. S ee C u r tin [1982] for a m o r e e x te n s iv e d isc u ssio n o f th e IC S . 8 natural stepping off point for empirical investigation. However, it should be kept in mind that Granger causality is neither a necessary or sufficient condition for true causality to exist.9 Our empirical strategy is to demonstrate first that Granger causality exists from the ICS to GNP, and second, that this relation is robust to inclusion of a large number of alternative specifications. We begin with a simple two variable system including only GNP and the ICS. Two types of evidence are reported to test if changes in consumer sentiment predict changes in output. The first test is the so-called “Granger test.” Bivariate vector autoregressions are estimated of the form 'Yt .St. '«(£) b(iy AL) <*(£). 'Yt .St. + 'ey/ .est. where Yt is GNP growth in quarter t , S t is the ICS, x ( L ) is a polynomial in the lag operator L, and t is an error.10 If the block of coefficients represented by b (L ) are not jointly equal to zero then we can reject the hypothesis that there is not Granger causality from consumer sentiment to GNP. The system also provides a statistical summary of the link apparent in Figure 1 . Table II presents the results for 1 , 2, 3, and 4 lags.11 In the table, each column is a regression. The number of lags and the dependent variable are indicated at the top of each column. The main entry for each variable is the p-value for the F-statistic testing whether the block of coefficients are jointly equal to zero. In all four models, the hypothesis that the ICS coefficients in the GNP equation are jointly equal to zero can be rejected at better than the 1 percent level. In other words, ICS innovations help to predict GNP movements. Table II also reports the sum of each block of coefficients and the standard error of the sum. The sum of ICS coefficients is positive in all GNP equations and significantly different from zero at better than the 1 percent level. There is a subtle difference between this information and that provided by the F-statistic. The F-statistic looks at the role of the ICS in explaining quarterly GNP fluctuations while the sum of coefficients represents the “long run” effect of the ICS on the level of GNP. The sum and block of GNP coefficients in the GNP equations are never significantly different from zero. The GNP coefficients in the ICS equations are always negative and marginally significant in the 3 and 4 lags models. This curious effect recurs throughout the estimates below. Finally, the ICS does an excellent job predicting its own movements. 9 O n ca u sa lity in e c o n o m e tr ic s se e G ew ek e, M eese a n d D e n t [1983] an d th e S e p te m b e r /O c to b e r 1988 sp e c ia l is su e o f th e Journal of Econometrics. 10 A ll r e p o r te d r esu lts are d raw n from e q u a tio n s th a t also in c lu d e a c o n s ta n t term . 11 H ere a n d b e lo w , th e IC S is sc a le d b y a fa cto r o f 1,000 for c la r ity o f p r e se n ta tio n . 9 A variant of the Granger test, described by Geweke, Meese, and Dent [1983], evaluates whether possible forcing variable x decreases the forecast variance of variable y . Let the residual variance of y in an autoregression be a2 . Let the residual variance when x is included as an explanatory variable be <t>2. Let the number of observations be N and the number of lags k. Geweke, Meese, and Dent show that m = ~ n xi asymptotically under the null hypothesis of no Granger causality. They argue that this type of test has better small sample properties than the Granger test above. The results for 77 are presented in Table III. In all four models the ICS decreases the forecast variance of GNP at better than the 1 percent level of significance. Because p ( k ) > T](k) in small samples, the ICS significantly decreases forecast variance using p as well. C. O u tp u t, C o n su m e r S e n tim e n t, a n d L e a d in g In d ic a to r s Tables II and III show that the ICS helps to predict GNP, and thus by definition that GNP is Granger caused by ICS. However, bivariate estimates do not make a compelling case for true causality between the ICS and GNP because they fail to address the obvious possibility that the correlation is driven by a third variable acting on both the ICS and GNP. Suppose that in reality GNP is caused entirely by a production shifter, say the price of oil, but with a lag of six months. How do we know that consumer sentiment is not simply a forecast of GNP based on observation of the price of oil? Because there are an infinite number of potential third variables, in general it is impossible to prove that a correlation is not caused by some omitted third variable. However, it is possible to evaluate if it is caused by specific variables. Our approach is to introduce into the vector autoregressions a succession of plausible variables that consumers might be using to forecast output and examine if they can eliminate the ICS-GNP correlation. To the extent that these third variables are unable to account for the correlation, the more confident we can be that the ICS-GNP relation is not spurious. The next step is then to estimate vector autoregressions including control variables that are correlated with economic fundamentals, that is, 'v r At .s t . = a (L ) m C(£)■ d (L ) e (i) m A L ) M i) i(i). 10 'Y t At .S t. ’ «y,' + €.4, . est . where A t represents the control variables. Series that are exogenous and have a good ability to forecast GNP are desirable. Inclusion of a large number of series in a vector autoregression consumes degrees of freedom and complicates statistical inference, so we choose a single series incorporating the effects of a number of different predictors, the ILI, which is explicitly constructed to forecast GNP movements. The ILI is a composite index that includes a number of series that might be considered forcing variables (money supply, sensitive materials prices). It also contains endogenous variables (inventories, unemployment claims). Because the ILI contains both exogenous and endogenous variables, one expects the relationship between the ILI and ICS to exhibit bidirectional causality if sentiment is causal. That is, the hypothesis that consumer sentiment causes output is not inconsistent with the possibility that it also reflects current information. The results are reported in Table IV. The table is formatted in the same way as Table II. There are three pieces of evidence on the ICS-GNP relation. First, p-values are presented corresponding to the F-statistic on the block of ICS coefficients in the GNP equations, that is, the test of c ( L ) = 0. The hypothesis that the ICS coefficients are jointly zero can be rejected at better than the 5 percent level in all models. The sum of ICS coefficients is positive and significantly different from zero at the 1 percent level in all models. The third relevant statistic, reported at the bottom of the table, tests for the block exogeneity of GNP and the ILI with respect to the ICS. A x2 is constructed by scaling the difference between the log determinant of the residual covariance matrix of the estimated system and that of a bivariate GNP-ILI system. This statistic allows for the possibility that the ICS affects GNP in d ir e c tly through the ILI. The hypothesis that GNP and the ILI are block exogenous can be strongly rejected in the 1, 2, and 4 lags models. The x2 from the 3 lags model does not attain statistical significance at conventional levels, but it is the sole exception to otherwise uniform evidence that the ICS predicts GNP, and as we discuss below, the 3 lags model can be statistically rejected in favor of the 4 lags model. Turning to other coefficients, it can be seen that GNP does a poor job predicting its own innovations, as above. Not surprisingly, the ILI predicts GNP. In the ILI equations, GNP appears to have a negative effect while the ICS has a positive effect after three or four quarters. The ICS equation indicates that consumer sentiment has a substantial autore gressive component. The leading indicators have a positive effect, suggesting that consumer sentiment is in part a forecast based on fundamentals. However, a comparison with Table II reveals that inclusion of ILI does not add much to the ICS R . GNP has a negative effect, suggesting a kind of regression to the mean in expectations; when times have been good for a while people believe all good things must end and when times have been bad they believe there is nowhere to go but up. 11 Consumer sentiment appears to have a statistically significant effect on GNP. We would like to know if it has a q u a n tita tiv e ly significant effect. Comparing the estimates in Table IV with those for bivariate vector autoregressions of GNP and the ILI (not included) shows that addition of the ICS adds only 2 percent to 4 percent to the R . The ILI and ICS move together so this crude estimate is probably a lower bound on the marginal contribution of sentiment to explaining GNP fluctuations. A better way to assess the quantitative importance of consumer confidence in GNP fluctuations is to use the vector autoregression residuals to decompose the forecast variance of GNP into contributions by each of the variables. The technique is briefly outlined below; a comprehensive discussion can be found in Sims [1980a] (see also Sims [1980b] and Litterman and Weiss [1985]). As a caveat to this sort of exercise we should repeat Schiller’s [1987] observation that variables which have very small effects in explaining long run variance nevertheless may be highly important in certain situations. Let u be a three element vector of forecast errors for a trivariate vector autoregression and E the covariance matrix associated with the u process. A corresponding orthonormal vector v and a lower triangular matrix G can be found such that G G ' = E and G v — u. Because v is orthonormal and G is lower triangular the variance of the forecast error of each of the endogenous variables can be decomposed into percentage contributions by innovations in each of the three equations. This procedure can be used to decompose variances of any arbitrary length forecast. The percentage contributions are sensitive to the decomposition method, specifically to the ordering of variables in the equation. For example, by construction the first variable listed contributes 100 percent of its own one-period-ahead forecast variance. Because G is lower triangular it is conventional to place forcing variables early in the order and endogenous variables later. For this reason the hypothesized driving variables, the ICS and ILI, are placed first in the first two decompositions. Consumer sentiment is placed last, somewhat atheoretically, in the third decomposition to describe the worst case. Decompositions of GNP and the ICS from one to eight quarters ahead are presented in Table V.12 The estimates are for the 4 lags model in Table IV. Here and below, when choosing which model to decompose we were guided by the following selection rule. First we computed the distance between the residual covariance matrixes of the 3 and 4 lags models. If they were significantly different, we chose the 4 lags model. If not, we compared the 2 and 3 lags models. If they were significantly different, we chose the 3 lags model, and so on. The variables are listed in the order they appear in the u vector. 12 R A T S v er sio n 4 .0 2 w a s u se d to p erfo rm th e v a ria n ce d e c o m p o s itio n s. T h e r o u tin e ig n o r e s th e sa m p lin g error c o m p o n e n t o f th e p ro c e ss an d co n sid ers o n ly th e e q u a tio n error. 12 In the top GNP decomposition, the variance of ICS explains 14.4 percent of the onequarter ahead forecast variance of GNP, rising to 25.8 percent eight quarters ahead. In the bottom decomposition, which is the worst case for finding an effect of consumer sentiment, the ICS explains 12.9 percent of the eight-quarter-ahead variance. Recall that the zero percent estimate of the one-quarter-ahead contribution in the last equation is entirely an artifact of the decomposition technique. Rough bounds on the mean effect of the ICS on GNP then would be between 13 percent and 26 percent. It seems reasonable to conclude that movements in consumer confidence are quantitatively important in explaining GNP fluctuations. The ICS decompositions provide evidence on the exogeneity of consumer sentiment. In the top panel, the ICS explains 83.6 percent of its own innovation variance eight quarters ahead. In the worst case for the ICS, the bottom panel, consumer sentiment still explains 72.1 percent of its eight-quarter-ahead innovation variance. Although consumer sentiment appears to be affected by GNP and ILI movements, neither of these variables can explain the majority of sentiment variation. This suggests that the ICS is to a large degree exogenous with respect to these variables, and tends to undermine the notion that sentiment is nothing more than a forecast of output. D. R o b u s tn e s s The results in the previous section support the hypothesis that consumer sentiment con tributes to fluctuations in aggregate output. However, like the test of Oh and Waldman [1990], the estimates can be interpreted in more than one way. In particular, there remains the possibility that we have omitted an economic fundamental which is forcing both the con sumer sentiment and GNP series. This section explores a number of different specifications and variables in pursuit of such a fundamental. First, we estimate a set of vector autoregressions with GNP, the ICS, and three “text book” control variables, money supply, government spending, and sensitive materials prices. We call these textbook controls because they are staples of almost every macroeconomics textbook. If one were to ask economists to name exogenous factors that cause GNP fluctu ations, these variables would probably appear on most lists. Table VI reports evidence from these vector autoregressions. To conserve space, only the GNP and ICS equations are reported, and only the coefficients on GNP and the ICS. The ICS coefficients in the GNP equation are of primary interest. The F-statistic on the block of ICS coefficients is significant at the 10 percent level in the 1 lag model and just beyond conventional significance levels in the 2, 3, and 4 lags models. Some fall in precision of estimates is to be expected, especially in the model with 4 lags, because of the loss of 13 degrees of freedom. The sum of ICS coefficients is positive and statistically significant in all models. The magnitude of the sum in the 1 lag and 2 lags models is quite a bit lower than in Table IV, but about the same for the 3 and 4 lags model, indicating that the controls rob consumer sentiment of some power in the first two quarters. The most important statistic is the x2 that tests for the block exogeneity of GNP and the controls with respect to the ICS. Recall that this indicates whether ICS has any effect on the system, direct or indirect. Block exogeneity can be rejected at approximately the 1 percent level in all models. Taken together, the estimates suggest that the textbook set of controls cannot account for the ICS-GNP relation. The bottom panel of Table VI reports variance decompositions. To maintain compara bility with other tables we choose to present the decomposition of the 4 lags model, although in this case our selection criteria suggested that the 2 and 3 lags models would be fine as well. In any case, the 2 and 3 lags decompositions are substantially the same. As in Table V we decompose GNP and the ICS. The sum of the percentage contribution of controls is indi cated in the column headed “X.” Again, to conserve space, only two orderings are reported. In the GNP equation, when the ICS is ordered before the controls, consumer sentiment can account for over 20 percent of the innovation variance of output, with its explanatory power peaking 3 and 4 quarters ahead at 23.7 percent. When the ICS is listed after the controls, it experiences only a minor decline—4 quarters ahead it still explains 18.6 percent of GNP innovation variance. In the ICS equation, the ICS accounts for no less than 65 percent of its own innovation variance when listed first, and no less than 56 percent when listed after all the controls. The decompositions from the textbook system do not differ in substance from the ILI system: the ICS has a quantitatively large effect on GNP, and the ICS appears to have an important exogenous component, suggesting it is not simply a forecast of GNP. The next set of vector autoregressions include as controls both the ILI and default risk, the spread between the interest rate on corporate debt and treasury bills. The default risk variable has attracted recent attention for its ability to predict GNP movements [Stock and Watson, 1989]. It is not clear whether this variable reflects an economic fundamental or is itself a non-fundamental, perhaps the animal spirits of investors. Table VII reports vector autoregressions with GNP, the ICS, the ILI, and default risk, indicated “DR” in the table. For space reasons, only the 4 lags model is reported. Again, our test for number of lags suggests that models with fewer than 4 lags can be rejected at about the 1 percent level. The key statistics, as usual, concern the ICS in the GNP equation. First, the hypothesis that the block of ICS coefficients are jointly zero can be rejected at almost the 5 percent level. Second, the sum of coefficients is positive and statistically different from zero at better than the 5 percent level. Addition of the default risk variable leads to a modest 14 decrease in the magnitude of the sum. The x 2 statistic indicates that block exogeneity of GNP, the ILI, and default risk with respect to the ICS can be rejected at roughly the 1 percent level. The evidence continues to point to a role for consumer confidence in economic fluctuations.13 For this model, we report all of the equations and coefficients because the default risk variable is of some interest in itself. This variable has a significant effect on the movement of all variables in the system according to the F-statistic. The sum of default risk coefficients is significant in the GNP, ILI, and default risk equations. Thus, default risk matters. This is interesting because default risk is not obviously an economic fundamental. It may be simply a good mirror of underlying fundamentals, but the failure of the ILI and GNP to achieve statistical significance in the default risk equation does not square well with this interpretation. The results were no better in the models with fewer lags. In fact, the only variable identified as driving default risk in this system is the ICS. Taken together, the evidence is not inconsistent with the idea that default risk represents investor sentiment, or captures some other aspect of consumer confidence. Table VIII reports the GNP and ICS variance decompositions for the model in Table VII. The evidence follows the pattern established above. The ICS accounts for a healthy fraction of the innovation variance of GNP, up to 21.8 percent eight quarters ahead in the worst case decomposition ordering. Default risk serves primarily to rob the ILI of its explanatory power. Furthermore, the ICS continues to appear largely exogenous. Even in the worst case decomposition, consumer sentiment accounts for no less than 70 percent of its own innovation variance. If anything, inclusion of default risk in the vector autoregressions suggests an even more important role for consumer confidence as a causal factor. In the final set of vector autoregressions, both default risk and the individual compo nents of the ILI are included as controls.14 The benefit of this is that it avoids the ad hoc parameter restrictions implied by construction of the ILI. The cost is that the large number of component series rapidly consumes the available degrees of freedom. The GNP and ICS coefficients for the GNP and ICS equations are reported in the upper panel of Table IX. Variance decompositions are reported in the bottom panel. Beginning as usual with the F-statistic on the block of ICS coefficients in the GNP equation, the hypothesis that consumer sentiment does not Granger cause GNP can be rejected at about 13 T h e x 2 s t a t is t ic an d su m o f c o efficien ts are a lso s ta tis tic a lly sig n ific a n t in th e u n rep o r te d 1, 2, an d 3 la g s m o d e ls, w h ile th e F s t a tis tic s lie ju s t o u ts id e s ta tis tic a l sig n ific a n c e . 14 T o b e p r e c ise , we use th e c o m p o n e n ts o f th e cu rren t ILI less th e IC E . T h e current ILI differs from th e v e rsio n liste d in T a b le I in th a t it in clu d es th e IC E an d u n filled ord ers an d ex c lu d e s th e c h a n g e in m a n u fa c tu r in g a n d tra d e in v en to ries. 15 the 5 percent level in the 1 lag and 3 lags models, but does not quite achieve statistical significance in the 2 and 4 lags models. Here again, this test loses a fair amount of power due to the large number of control variables, especially in the model with 4 lags. The sum of the ICS coefficients is positive and significantly different from zero in all models. Most important, the block exogeneity of GNP and the controls with respect to the ICS can be rejected at better than the 1 percent level in all four models. The bottom panel of Table IX reports variance decompositions for the 4 lags model. Here again the models with fewer than 4 lags can be rejected when comparing the log determinants of the residual covariance matrixes. Two decomposition orderings are reported, one that lists the ICS before the controls (indicated as “X”) and one that lists it after the controls; GNP is ordered last in both cases. In the first GNP decomposition, the ICS explains 21.6 percent of the one-quarter-ahead variance; the effect falls gradually to 16.3 percent eight quarters ahead. In the second GNP decomposition, the estimated ICS efFect is lower, peaking at 8.8 percent five quarters ahead. These effects are smaller than in previous tables, but not trivial. In the ICS decompositions, the evidence continues to support the idea that the ICS has a substanital exogenous component, that is, consumer sentiment is more than a forecast based on available information. In the first decomposition, the ICS explains no less than 52 percent of its own innovation variance, while in the second no less than 31 percent. The results in this section can be simply summarized. Even after controlling for a number of good predictors of GNP, the ICS continues to add predictive power to the model. These results cannot rule out the possibility that consumer sentiment is nothing more than a forecast of output. However, under this interpretation it appears that consumers are basing their forecasts on information that is unavailable to professional econometricians who specialize in forecasting GNP movements. The alternative interpretation that consumer confidence causes ouput movements seems no less plausible. V. CONCLUSION This paper explores the possibility that the economy’s total output occasionally varies not in response to a shift in fundamentals but in response to a shift in consumer sentiment. Specifically, the paper asks whether and to what extent exogenous declines in consumer confidence cause recessions and conversely whether and to what extent bullish consumers drive economic growth. There are two inspirations for this research. The first is the fact that something called “consumer confidence” plays an important role in popular explanations of business cycles and in the public statements of business and political leaders. There is a long tradition of confidence explanations for recessions. Keynes emphasized the “mass psychology of the 16 market” and “animal spirits” among investors, concluding that “(i)n estimating the prospects of investment, we must have regard, therefore, to the nerves and hysteria and even the digestions and reactions to the weather of those upon whose spontaneous activity it largely depends.” During the Great Depression President Franklin D. Roosevelt and General Motors Chairman Alfred P. Sloan, Jr., who disagreed in many ways on the role of government in helping the economy, seemed to be of the same mind on the nature of the problem and the primary impediment to recovery. Roosevelt’s famous inaugural speech contained the statement, “let me assert my firm belief that the only thing we have to fear is fear itself” while Sloan commented, “the main obstacle to the general revival of American enterprise is the fear the foundation of the economy is in jeopardy.”15 Our empirical estimates for the United States, 1953-1988, uniformly reject the hypothe sis that consumer sentiment does not cause GNP (in the Granger sense). This gives support to the idea that exogenous changes in consumer sentiment have real effects on output. Ac cording to our central vector autoregressions, between 13 percent and 26 percent of the variance of GNP innovations is the result of waves of consumer sentiment, which suggests a non-trivial role for expectations. Thus, the main contribution of the paper is a demonstra tion that consumer confidence is an important independent factor in economic fluctuations, a fact that did not previously rest on solid empirical ground [Leeper, 1991]. The second purpose of the paper is to provide some evidence on the rich collection of macroeconomic models with strategic complementarities that have been developed in recent years. All these multiple equilibria models have in common that expectations are selffulfilling—because agents must expect to be in a particular equilibrium before the economy can move to it, their expectations in a sense cause the movement to the equilibrium. An implication of these models, then, is that after controlling for movements in economic fun damentals, changes in consumer sentiment lead to changes in GNP. Our estimates appear to provide support for this key implication. However, we caution that our evidence does not prove that expectations cause output fluctuations. There remains the possibility that a missing third variable is forcing both series. Even so, the results have power because we could have found that expectations were completely insignificant. If that had been the case, it would have been safe to conclude that multiple equilibria are not empirically important. Like Oh and Waldman [1990], then, our results are open to an alternative interpretation. The two studies taken together constitute an initial case in favor of the class of macroeconomic models that view some recessions as coordination failures, but certainly additional research is called for. 15 Q u o ta tio n s in th e p a ra g ra p h are from K e y n e s [1964, p a g es 154, 161, 162], R o o se v e lt [1946, p a g e 13] an d C ray [1980, p a g e 311], r e sp e c tiv e ly . 17 REFEREN CES Ball, Laurence and David Romer, “Sticky Prices as Coordination Failure,” n o m ic R e v ie w , LXXXIII (June 1991), 539-552. Bryant, John, “A Simple Keynes-Type Model,” (August 1983), 525-528. A m e ric a n E c o T h e Q u a r te r ly J o u rn a l o f E c o n o m ic s, XCVIII Bryant, John, “The Paradox of Thrift, Liquidity Preference, and Animal Spirits,” m e tr ic a , LV (September 1987), 1231-1235. E co n o - Cass, David and Karl Shell, “Do Sunspots Matter?,” (April 1983), 193-227. Cray, Ed, C h r o m e C o lo ssu s: Book Company, 1980). J o u rn a l o f P o litic a l E c o n o m y , G e n e ra l M o to r s a n d I ts T im e s XCI (New York, NY: McGraw-Hill Cooper, Russell and Andrew John, “Coordinating Coordination Failures in Keynesian Mod els,” T h e Q u a r te r ly J o u rn a l o f E co n o m ics, CIII (August 1988), 441-463. Curtin, Richard T., “Indicators of Consumer Behavior: The University of Michigan Surveys of Consumers,” P u b lic O p in io n Q u a rte rly , XLVI (1982), 340-352. Dagsvik, John and Boyan Jovanovic, “Was the Great Depression a Low-Level Equilibrium?,” C. V. Starr Center for Applied Economics, R. R. 91-07, January 1991. Friend, Irvin and F. Gerard Adams, “The Predictive Ability of Consumer Attitudes, Stock Prices, and Non-Attitudinal Variables,” J o u rn a l o f th e A m e ric a n S t a ti s tic a l A s s o c ia tio n , LXXXIX (December 1964), 987-1005. Geweke, John, Richard Meese and Warren Dent, “Comparing Alternative Tests of Causality in Temporal Systems,” J o u rn a l o f E c o n o m e tric s, XXI (1983), 161-194. Hertzberg, Marie P. and Barry A. Beckman, “Business Cycle Indicators: Revised Composite Indexes,” B u sin e ss C o n d itio n s D ig e s t, XXIX (January 1989), 97-102. imrohoroglu, Selahattin, “Testing for Sunspot Equilibria in the German Hyperinflation,” J o u r n a l o f E c o n o m ic D y n a m ic s a n d C o n tro l, XVII (1993), 289-317. Keynes, John Maynard, T h e G en era l T h e o r y o f E m p lo y m e n t, York, NY: Harcourt Brace Jovanovich, 1964). I n te r e s t, a n d M o n e y (New Kiyotaki, Nobuhiro, “Multiple Expectational Equilibria Under Monopolistic Competition,” T h e Q u a r te r ly J o u rn a l o f E c o n o m ic s, CIII (November 1988), 695-713. Leeper, Eric M., “Consumer Attitudes and Business Cycles,” Federal Reserve Bank of At lanta, August 1991. Litterman, Robert B. and Lawrence Weiss, “Money, Real Interest Rates and Output: A Reinterpretation of Postwar U.S. Data,” E c o n o m e tric a , LIII (1985), 129-156. 18 Milgrom, Paul and John Roberts, “Rationalizability, Learning, and Equilibrium in Games with Strategic Complementarities,” E c o n o m e tric a , LVIII (November 1990), 1255-1277. Murphy, Kevin M., Andrei Shleifer, and Robert W. Vishny, “Industrialization and the Big Push,” J o u rn a l o f P o litic a l E c o n o m y , XCVII (October 1989), 1003-1026. Oh, Seonghwan and Michael Waldman, “The Macroeconomic Effects of False Announce ments,” T h e Q u a r te r ly J o u rn a l o f E co n o m ics, CV (November 1990), 1017-1034. Oh, Seonghwan and Michael Waldman, “The Leading Indicators as a Source of Expectational Shocks,” UCLA and Cornell, 1993. Roberts, John, “An Equilibrium Model with Involuntary Unemployment at Flexible, Com petitive Prices and Wages,” A m e ric a n E c o n o m ic R e v ie w , LXXVII (December 1987), 856-874. Roosevelt, Franklin D., N o th in g to Fear: T h e S e le c te d A d d r e s s e s o f F ran klin D. R o o s e v e lt 1 9 3 2 -1 9 4 5 , B.D. Zevin, editor (New York, NY: Houghton Mifflin Company, 1946). Schiller, Robert J., “Ultimate Sources of Aggregate Variability,” A m e ric a n LXXVII (May 1987), 77-92. Shleifer, Andrei, “Implementation Cycles,” 1986), 1163-1190. J o u rn a l o f P o litic a l E c o n o m y , Sims, Christopher A., “Macroeconomics and Reality,” 1-48. E c o n o m e tric a , E c o n o m ic R e v ie w , XCIV (December XLVIII (Jan. 1980a), Sims, Christopher A., “Comparison of Interwar and Postwar Business Cycles: Monetarism Reconsidered,” A m e ric a n E c o n o m ic R e v ie w , LXX (May 1980b), 250-257. Stock, James H. and Mark W. Watson, “New Indexes of Coincident and Leading Economic Indicators,” N B E R M a c ro e c o n o m ic s A n n u a l, edited by Olivier Jean Blanchard and Stan ley Fischer (1989), 351-394. Weil, Philippe, “Increasing Returns and Animal Spirits,” LXXIX (September 1989), 889-894. A m e ric a n E c o n o m ic R e v ie w , Woodford, Michael, “Three Questions about Sunspot Equilibria,” v ie w , LXXVII (May 1987), 93-98. A m e ric a n E c o n o m ic R e Zarnowitz, Victor, O rd ers, P r o d u c tio n , a n d I n v e s tm e n t —a C y c lic a l (New York, NY: NBER and Columbia University Press, 1973). a n d S tr u c tu r a l A n a ly s is 19 TABLE I Components of the Index of Consumer Sentiment and Index of Leading Indicators A. Index of Consumer Sentiment 1. We are interested in how people are getting along financially these days. Would you say that you (and your family living there) are better off or worse off financially than you were a year ago? Why do you say so? 2. Now looking ahead—do you think that a year from now you (and your family living there) will be better off financially, or worse off, or just about the same as now? 3. Now turning to business conditions in the country as a whole—do you think that during the next 12 months we’ll have good times financially, or bad times, or what? 4. Looking ahead, which would you say is more likely—that in the country as a whole we’ll have continuous good times during the next five years or so, or that we’ll have periods of widespread unemployment or depression, or what? 5. About the big things people buy for their homes—such as furniture, a refrigerator, a stove, television, and things like that. Generally speaking, so you think now is a good or a bad time for people to buy major household items? Why do you say so? B. Index of Leading Indicators (1988 Version) 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. Average weekly hours of production or nonsupervisory workers in manufacturing. Average weekly initial claims for unemployment insurance in state programs. Manufacturers’ new orders in consumer goods and materials industries. Contracts and orders for plant and equipment. Index of new private housing units authorized by local building permits. Index of stock prices of 500 common stocks. Money supply, M2. Percent of companies receiving slower deliveries from vendors. Change in sensitive materials prices. Change in business and consumer credit outstanding. Change in manufacturing and trade inventories on hand and on order. T A B L E II Vector Autoregressions with G N P and the Index of Consumer Sentiment 1 lag 2 lags 3 lags 4 lags D e p e n d e n t v a ria b le D e p e n d e n t va ria b le D e p e n d e n t va ria b le D e p e n d e n t va ria b le GNP ICS GNP ICS GNP ICS GNP ICS 0.118 0.206 0.563 0.314 0.761 0.086 0.808 0.063 Sum of coefficients Standard error 0.148 (0.094) -0.066 (0.052) 0.063 (0.124) -0.031 (0.072) -0.085 (0.158) -0.165+ (0.093) -0.066 (0.198) -0.263* (0.108) ICS F: p-value Sum of coefficients Standard error 0.002 0.265** (0.082) 0.000 0.946** (0.045) 0.003 0.272** (0.087) 0.000 0.936** (0.050) 0.000 0.314** (0.096) 0.000 0.979** (0.056) 0.001 0.327** (0.107) 0.000 0.995** (0.058) 0.162 0.825 0.157 0.831 0.215 0.838 0.202 0.845 GNP F: p-value R2 T h is ta b le rep o rts in fo rm a tio n from vector a u to reg ressio n s Z t = < b ( L ) Z t + e t w h ere Z t c o n ta in s G N P and th e IC S. T h e 1, 2, 3, and 4 lags m o d els con tain 121, 115, 110, and 109 o b serv a tio n s, resp ectiv ely . Each colu m n is a regression . T h e d e p en d en t variable is in d ica ted a t th e top o f each co lu m n . In th e m ain en tries, th e first n u m b er is th e p -valu e a sso c ia te d w ith th e F -sta tis tic for th e h y p o th e sis th a t the block o f coefficien ts are jo in tly eq u al to zero. B elow th is is th e su m o f th e block o f co efficien ts and its sta n d a rd error. S ig n ifica n ce lev els on th e su m o f coefficien ts are in d icated as follow s: “* * ” is sig n ifica n t rejection at 1 p e r c e n t, is sig n ifica n t rejection a t 5 p ercen t, and “- f ” is sig n ifica n t rejection at 10 p ercen t. T A B L E III Forecast Error Tests for G N P and the Index of Consumer Sentiment <72 <f>2 T](k) Observations 1 0.0000895 0.0000829 8.89 121 2 0.0000855 0.0000785 9.42 115 3 0.0000871 0.0000752 15.01 no 4 0.0000886 0.0000769 14.44 109 Lags (k) T h is ta b le r e p o r ts e v id e n c e o n w h e th e r th e IC S red u ces th e fo r e c a st v a ria n ce o f G N P . H ere a2 is th e resid u a l v a ria n ce o f G N P in a n a u to r e g r e ssio n , < t>2 is th e resid u a l v a ria n ce w h en IC S is in c lu d e d a s an e x p la n a to r y v a ria b le, a n d ri=N((<?2-<j>2)/<72). T h e n u ll h y p o th e s is is th a t e x p e c ta t io n s d o n o t h e lp to fo r e c a st G N P . T h e tj s t a t is t ic s a ll in d ic a te sig n ific a n t rejectio n a t b e tte r th a n th e 1 p e r c e n t le v e l. T A B L E IV Vector Autoregressions with GNP, the Index of Leading Indicators, and the Index of Consumer Sentiment 1 lag 2 Dependent variable lags 3 Dependent variable 4 lags Dependent variable lags Dependent variable GNP ILI ICS GNP I LI ICS GNP ILI ICS GNP I LI IC S 0 .1 4 9 0 .0 0 6 0 .0 0 2 0 .0 6 9 0 .0 4 6 0.011 0 .2 3 3 0.175 0.006 0.228 0.126 0.009 S u m o f coefficien ts - 0 .1 3 8 - 0 .5 9 2 * - 0 .1 8 2 * * -0 .2 5 7 + - 0 .9 1 3 * * - 0 .1 1 1 - 0 .3 2 7 + - 1 .1 6 1 * - 0 .2 5 9 * - 0 .3 9 7 - 1 .1 2 6 * - 0 .3 1 6 * S ta n d a rd error (0 .0 9 5 ) (0 .2 1 3 ) (0 .0 5 6 ) (0 .1 3 7 ) (0 .3 2 9 ) (0 .0 8 6 ) (0 .1 9 5 ) (0 .4 6 3 ) (0 .1 1 6 ) (0 .2 4 2 ) (0 .5 3 5 ) (0 .1 3 3 ) 0.000 0 .0 0 0 0 .0 0 0 0 .0 0 0 0 .0 0 0 0 .0 0 0 0 .0 0 0 0.000 0.000 0 .0 0 0 0.000 0.000 GNP F: p -v a lu e ILI F: p -v a lu e S u m o f coefficien ts 0.249** 0.656** 0.101** 0.298** 0.659** 0.101** 0.290** 0.700** 0.110* 0.318** 0 .3 6 9 + 0 .0 7 8 S ta n d a rd error (0 .0 4 1 ) (0 .0 9 2 ) (0 .0 2 4 ) (0 .0 5 4 ) (0 .1 3 0 ) (0 .0 3 3 ) (0 .0 7 5 ) (0 .1 7 9 ) (0 .0 4 4 ) (0 .0 9 2 ) (0 .2 0 5 ) (0 .0 5 1 ) 0 .0 0 4 0 .3 9 6 0 .0 0 0 0 .0 1 0 0 .3 8 7 0.000 0.042 0.544 0.000 0.045 0.052 0.000 IC S F: p -v a lu e S u m o f coefficien ts 0.214** 0 .1 3 8 0.925** 0.236** 0 .2 4 6 0.920** 0.260** 0.299 0.955** 0.280** 0.474* 0.980** S ta n d a rd error (0 .0 7 3 ) (0 .1 6 2 ) (0 .0 4 3 ) (0 .0 7 5 ) (0 .1 8 1 ) (0 .0 4 5 ) (0 .0 8 8 ) (0 .2 0 8 ) (0 .0 5 2 ) (0 .0 9 8 ) (0 .2 1 8 ) (0 .0 5 4 ) 0 .3 5 5 0 .3 1 3 0 .8 4 6 0 .3 6 3 0 .2 7 9 0.861 0 .3 4 7 0.258 0 .8 6 4 0 .3 3 9 0 .3 3 6 0.869 T? x 2 : p -v a lu e 0.003 0 .0 0 3 0 .2 7 3 0 .0 2 2 (B lo c k e x o g e n e ity ) T h is ta b le rep orts in fo rm a tio n from vector a u to reg ressio n s Zt=4>(L)Z(-He w here Zt c o n ta in s G N P , th e IC S, and th e ILI. T h e 1, 2, 3, and 4 lags m o d els co n ta in 121, 115, 110, and 109 o b se r v a tio n s, resp ectiv ely . E ach colu m n is a regression . T h e d e p e n d en t variable is in d icated a t th e to p o f each colu m n . In th e m ain en tr ie s, th e first num ber is th e p -value a sso c ia te d w ith th e F -sta tis tic for th e h y p o th e sis th a t th e block o f coefficien ts are jo in tly eq u al to zero. B elow th is is th e su m o f th e block o f coefficien ts and its sta n d a rd error. S ign ifican ce levels on th e su m o f coefficients are in d ica ted as follow s: “* * ” is sig n ifica n t rejection a t 1 p ercen t, is sig n ifica n t rejection a t 5 p erc e n t, and “- f ” is sig n ifica n t rejection at 10 percent. T h e x 2 s t a tis tic is for th e h y p o th e sis th a t G N P and th e ILI are block ex o g en o u s w ith resp ect to th e IC S. TABLE V Decomposition of the Innovation Variance of G N P and the Index of Consumer Sentiment from Vector Autoregressions with GNP, the ICS, and the ILI G N P d e c o m p o s itio n Quarter 1 4 8 ICS 14.4 21.1 25.8 ILI 10.8 21.8 21.7 GNP 74.8 57.1 52.5 G N P d e c o m p o s itio n Quarter 1 4 8 ILI 22.5 39.0 35.2 ICS 2.7 3.9 12.3 GNP 74.8 57.1 52.5 G N P d e c o m p o s itio n Quarter 1 4 8 GNP 100.0 71.5 62.9 ILI 0.0 25.S 24.3 ICS 0.0 2.7 12.9 IC S d e c o m p o s itio n ICS 100.0 90.1 83.6 ILI 0.0 2.8 3.2 GNP 0.0 7.1 13.2 IC S d e c o m p o s itio n ILI 25.2 32.5 22.4 ICS 74.8 60.4 64.5 GNP 0.0 7.1 13.2 IC S d e c o m p o s itio n GNP 14.4 5.9 4.1 ILI 13.3 29.7 23.8 ICS 72.2 64.4 72.1 T h is ta b le r e p o r ts th e p e r c e n ta g e o f G N P an d IC S in n o v a tio n v a ria n ce th a t c a n b e a ttr ib u te d to each v a ria b le in th e 4 la g s v e c to r a u to r e g r e ssio n . For ea ch m o d e l, th e d e c o m p o s itio n is c o m p u te d for th ree d ifferen t ord erin g s o f th e v a ria b les. T h e q u arter (a h e a d ) o f th e in n o v a tio n v a ria n ce b e in g d e c o m p o s e d is in d ic a te d in th e le ftm o s t c o lu m n . T h e r e m a in in g e n tries in d ic a te th e p e r c e n ta g e o f in n o v a tio n v a rian ce a ttr ib u ta b le to th e v a ria b le at th e to p o f th e co lu m n . T A B L E VI Vector Autoregressions and Variance Decompositions of GNP, the Index of Consumer Sentiment, and Textbook Controls 1 la g 2 la g s 3 la g s 4 la g s IC S GNP IC S GNP IC S GNP IC S 0 .7 2 2 0 .1 9 4 0 .7 9 2 0 .2 1 4 0 .7 5 1 0 .1 6 3 0 .8 7 0 0 .2 9 9 0 .0 3 5 -0 .0 7 0 -0 .0 6 8 0.051 - 0 .2 3 9 -0 .0 9 5 -0 .2 0 5 (0 .0 9 9 ) (0 .0 5 4 ) (0 .1 3 9 ) (0 .0 7 7 ) (0 .1 9 3 ) (0 .1 0 7 ) (0 .2 5 4 ) 0 .0 5 5 0 .0 0 0 0 .1 4 0 0 .0 0 0 0 .1 0 9 0 .0 0 0 GNP GNP F: p -v a lu e S u m o f co e ffic ie n ts S ta n d a r d error -0 .1 6 9 (0 .1 3 2 ) IC S F: p -v a lu e S u m o f co e ffic ie n ts 0 .1 6 1 + 0 .8 7 3 " 0.194* 0.850** 0 .2 6 7 * 0.876** S ta n d a r d error (0 .0 8 3 ) (0 .0 4 5 ) (0 .0 9 0 ) (0 .0 5 1 ) (0 .1 1 0 ) (0 .0 6 2 ) 0 .2 6 6 0 .8 5 0 0 .2 8 3 0.871 0 .2 8 5 0 .8 7 4 R2 0 .0 0 0 0 .0 0 0 x 2: p -v a lu e 0.000 0 .2 3 6 + 0.888** (0 .1 3 5 ) 0 .2 7 1 0 .0 0 0 IC S X GNP IC S 7 3 .6 X GNP 1 19.3 100.0 0 .0 0 .0 4 2 3 .7 18.7 5 7 .6 8 6 .1 11.4 2 .5 8 2 1 .7 3 0 .8 4 7 .5 65.1 2 8 .1 6 .8 X IC S GNP X IC S GNP 9.5 16.9 7 3 .6 6 .6 9 3 .4 0 .0 1 0.871 0 .0 1 5 7.1 Q u a rter (0 .0 7 0 ) ICS decomposition G N P decomposition Q u a rter 0 .1 5 8 4 2 3 .8 18.6 57.6 2 4 .0 7 3 .5 2.5 8 3 5 .0 17.5 4 7 .5 37.1 5 6 .1 6 .8 T h is ta b le r e p o r ts in fo r m a tio n fro m v e c to r a u to reg ressio n s Zt=$(L)Zt+et w h ere Ztc o n ta in s G N P , th e IC S , and con trol v a ria b les. T h e c o n tr o ls are g o v e r n m e n t sp e n d in g , m o n e y su p p ly M 2, an d se n s itiv e m a te r ia ls p rices. E ach colu m n in th e to p p a n e l c o r r e sp o n d s to a reg ressio n . O n ly th e G N P an d ICS r e g r e ssio n s are r e p o r te d , a n d o n ly th e G N P an d IC S co e ffic ie n ts. T h e 1, 2, 3, and 4 lags m o d els c o n ta in 121, 115, 110, a n d 109 o b se r v a tio n s, resp ectiv ely . T h e d e p e n d e n t v a ria b le is in d ic a te d a t th e top o f each co lu m n . In th e m a in e n tr ie s th e first n u m b er is th e p -v a lu e a ss o c ia te d w ith th e F -s t a tis tic for th e h y p o th e s is th a t th e b lo ck o f c o e ffic ie n ts are jo in tly e q u a l to zero. B elo w th is is th e su m o f th e b lo c k o f co e ffic ie n ts an d its sta n d a r d error. T h e x 2 s t a tis tic is for th e h y p o th e s is th a t G N P and th e co n tro ls are b lo c k e x o g e n o u s w ith r e sp e c t to th e IC S . S ig n ifica n ce le v e ls on th e su m o f co e ffic ie n ts are in d ic a te d as follow s: “* * ” is sig n ific a n t r e je c tio n a t th e 1 p ercen t level; is sig n ific a n t a t th e 5 p e r c e n t level; is sig n ifica n t a t 10 p er c e n t. T h e b o tt o m p a n e l rep o rts th e p e r c e n ta g e o f G N P an d IC S in n o v a tio n v a ria n ce th a t ca n b e a ttr ib u te d to each v a ria b le in th e 4 la g s v e c to r a u to r eg r e ssio n . T h e su m o f va ria n ce e x p la in e d b y th e co n tr o ls is rep o rted under th e h e a d in g “X .” T A B L E V II Vector Autoregressions with GNP, the Index of Leading Indicators, the Index of Consumer Sentiment, and Default Risk D e p e n d e n t v a r ia b le GNP ILI ICS DR 0.177 -0.441+ (0.242) 0.038 -1.316” (0.483) 0.020 -0.296* (0.130) 0.516 0.028 0.097) Sum of coefficients Standard error 0.118 0.205+ (0.111) 0.313 0.061 (0.221) 0.016 0.050 (0.060) 0.183 0.066 (0.044) ICS F: p-value Sum of coefficients Standard error 0.063 0.230(0.099) 0.369 0.272 (0.198) 0.000 0.971 (0.054) 0.001 0.039 (0.040) 0.060 -0.661+ (0.396) 0.000 -2.603” (0.791) 0.010 -0.117 (0.213) 0.000 0.754” (0.159) 0.374 0.489 0.881 0.578 GNP F: p -value Sum of coefficients Standard error ILI F: p -value DR F: p-value Sum of coefficients Standard error --9 R X2: p-value (Block exogeneity) 0.011 T h is ta b le r e p o r ts in fo r m a tio n from a v e c to r a u to r e g r e ssio n Z t = $ ( L ) Z t + u w h ere Z t c o n ta in s G N P , th e IC S , th e ILI, a n d d e fa u lt risk ( D R ) . T h e m o d e l has 4 la g s an d 109 o b se r v a tio n s. T h e d e p e n d e n t v a ria b le is in d ic a te d at th e to p o f e a c h c o lu m n . In th e m a in en tr ie s, th e F -s ta tis tic is for th e h y p o th e s is th a t th e b lo c k o f co e fficien ts are jo in tly e q u a l to zero . B e n e a th th e F is th e su m o f c o efficien ts a n d it s sta n d a r d error. T h e x 2 s t a t is t ic is for th e h y p o th e s is th a t G N P , th e IL I, a n d d e fa u lt risk are b lo ck e x o g e n o u s w ith r e sp e c t to th e IC S . S ig n ific a n c e le v e ls o n th e su m o f c o e ffic ie n ts are in d ic a te d a s fo llo w s: “* * ’ is sig n ific a n t r e je c tio n a t 1%, is sig n ific a n t r e je c tio n a t 5%, a n d * + ’ is sig n ific a n t r e je c tio n a t 10%. T A B L E VIII Decomposition of the Innovation Variance of G N P and the Index of Consumer Sentiment from Vector Autoregressions with GNP, the ICS, the ILI, and Default Risk G N P d e c o m p o sitio n Quarter 1 4 8 ICS 11.7 14.5 26.9 ILI 4.9 11.3 10.8 DR 0.0 18.1 15.0 GNP 83.3 56.1 47.2 IC S d e c o m p o sitio n ICS 100.0 86.3 85.9 G N P d e c o m p o s itio n Quarter 1 4 8 ILI 11.8 19.4 16.5 ICS 4.9 6.3 21.3 DR 0.0 18.1 15.0 GNP 83.3 56.1 47.2 DR 2.0 27.4 22.3 ILI 9.7 9.7 8.8 GNP 88.2 58.4 47.2 ICS 0.0 4.5 21.8 DR 0.0 7.3 4.5 GNP 0.0 4.9 8.3 IC S d e c o m p o sitio n ILI 16.7 21.0 14.9 G N P d e c o m p o sitio n Quarter 1 4 8 ILI 0.0 1.5 1.3 ICS 83.3 66.9 72.3 DR 0.0 7.3 4.5 GNP 0.0 4.9 8.3 IC S d e c o m p o sitio n DR 2.1 15.1 9.5 ILI 14.7 12.0 9.1 GNP 4.7 2.0 1.8 ICS 78.5 70.9 79.6 T h is ta b le r e p o r ts th e p e r c e n ta g e o f G N P an d IC S in n o v a tio n v a ria n ce th a t ca n b e a ttr ib u te d to each variab le in th e 4 lags v e c to r a u to r e g r e ssio n . T h e d e c o m p o s itio n is c o m p u te d for th ree differen t o rd erin g s o f th e v a ria b les. T h e q u a rter (a h e a d ) o f th e in n o v a tio n v a ria n ce b e in g d e c o m p o s e d is in d ic a te d in th e le ftm o s t co lu m n . T h e rem a in in g en tr ie s in d ic a te th e p e r c e n ta g e o f in n o v a tio n v a ria n ce a ttr ib u ta b le to th e v a riab le at th e to p o f th e c o lu m n . T A B L E IX Vector Autoregressions and Variance Decompositions of GNP, the Index of Consumer Sentiment, Default Risk, and Leading Indicators Components 1 la g GNP 2 la g s 3 la g s 4 la g s IC S GNP IC S GNP IC S GNP IC S 0 .0 1 4 0 .0 4 8 0 .0 3 1 0 .1 2 3 0 .0 3 6 0 .1 7 8 0 .3 2 4 0 .5 7 1 S u m o f co e ffic ie n ts - 0 .3 0 8 * - 0 .1 3 7 * -0 .4 8 2 * -0 .0 2 4 -0 .8 2 0 * - 0 .2 9 2 + - 0 .8 5 5 + -0 .2 6 3 S ta n d a r d error (0 .1 2 3 ) (0 .0 6 9 ) (0 .2 3 0 ) (0 .1 2 3 ) (0 .3 2 8 ) (0 .1 6 8 ) (0 .4 9 8 ) (0 .2 5 0 ) 0 .0 5 3 0 .0 0 0 0 .1 4 9 0 .0 0 0 0 .0 4 7 0 .0 0 0 GNP F: p -v a lu e IC S F: p -v a lu e S u m o f co e ffic ie n ts 0 .2 3 7 + 0 .781** 0 .2 7 8 + 0 .8 2 2 * “ 0.462** 0 .942** S ta n d a r d error (0 .1 2 1 ) (0 .0 6 7 ) (0 .1 6 0 ) (0 .0 8 9 ) (0 .2 0 5 ) (0 .1 1 0 ) 0 .3 3 6 0 .8 6 7 0 .3 3 0 0 .8 7 4 0 .3 6 9 0 .8 8 9 R2 0 .0 0 1 x 2: p -v a lu e 0 .0 0 0 IC S 0 .5 4 5 * (0 .2 7 2 ) 0 .3 1 5 0 .0 0 0 G N P decomposition Q u a r te r 0 .2 5 4 X GNP IC S 0.855** (0 .1 3 7 ) 0 .8 8 7 0 .0 0 0 ICS decomposition X GNP 1 2 1 .6 3 1 .1 4 7 .3 100.0 0 .0 0 .0 4 16.3 5 4 .9 2 8 .8 6 4 .8 3 4 .6 0 .6 8 16.3 6 1 .8 2 1 .9 5 2 .9 4 6 .6 0 .5 Q u a r te r 0 .0 0 0 X IC S GNP X IC S GNP 1 4 8 .0 4 .6 4 7 .4 3 9 .7 6 0 .3 0 .0 4 6 4 .1 7.1 2 8 .8 5 9 .6 3 9 .8 0 .6 8 7 0 .3 7 .8 2 1 .9 6 7 .8 3 1 .7 0 .5 T h is ta b le r e p o r ts in fo r m a tio n fro m v e c to r a u to r e g r e ssio n s Zt=$(L)Zt+ec w h ere Zt c o n ta in s G N P , th e IC S , an d co n tro l v a ria b les. T h e co n tr o ls are d e fa u lt risk a n d th e c o m p o n e n ts o f th e cu rren t IL I, e x c e p t for th e IC E . T h e 1, 2, 3, an d 4 la g s m o d e ls c o n ta in 114, 112, 110, a n d 109 o b se r v a tio n s, r e sp e c tiv e ly . E a ch c o lu m n in th e to p p a n e l co rresp o n d s to a reg ressio n . O n ly th e G N P an d IC S r e g ressio n s are r e p o r te d , a n d o n ly th e G N P a n d IC S c o e ffic ie n ts. T h e d e p e n d e n t va ria b le is in d ic a te d a t th e to p o f ea ch c o lu m n . In th e m a in en tries th e first n u m b e r is th e p -v a lu e a ss o c ia te d w ith th e F -s t a tis tic for th e h y p o th e s is th a t th e b lo ck o f co effic ie n ts are jo in tly e q u a l t o zero . B e lo w th is is th e su m o f th e b lo ck o f c o e ffic ie n ts an d its th e sta n d a r d error. T h e x 2 s t a tis tic is for th e h y p o th e s is th a t G N P an d th e con trols are b lo ck e x o g e n o u s w ith r e sp e c t to th e IC S . S ig n ifica n ce le v e ls for th e su m o f c o e ffic ie n ts are in d ic a te d as follow s: “* * ” is sig n ific a n t r e je c tio n a t th e 1 p e r c e n t level; is sig n ific a n t a t th e 5 p e r c e n t lev el; “+ ” is sig n ific a n t at 10 p e r c e n t. T h e b o t t o m p a n e l r e p o r ts th e p e r c e n ta g e o f G N P a n d IC S in n o v a tio n v a ria n ce th a t ca n b e a ttr ib u te d to e a c h v a riab le in th e v e c to r a u to r e g r e ssio n s. T h e su m o f v a ria n ce e x p la in e d b y th e c o n tr o ls is r e p o r te d u n d er th e h e a d in g “X .” Figure 1 - Index of Consumer Sentiment (1953-1988)