Board of Governors of the Federal Reserve System
International Finance Discussion Papers
Number 810, July 2004Screen Reader Version*
NOTE: International Finance Discussion Papers are preliminary materials circulated to stimulate discussion and critical comment. References in publications to International Finance Discussion Papers (other than an acknowledgment that the writer has had access to unpublished material) should be cleared with the author or authors. Recent IFDPs are available on the Web at http://www.federalreserve.gov/pubs/ifdp/. This paper can be downloaded without charge from the Social Science Research Network electronic library at http://www.ssrn.com/.
Abstract:
I present empirical evidence of how the U.S. economy, including percapita hours worked, responds to a technology shock. In particular, I present results based on permanent changes to a constructed direct measure of technological change for U.S. manufacturing industries.
Based on empirical evidence, some claim that hours worked declines and never recovers in response to a positive technology shock. This paper's empirical evidence suggests that emphasizing the drop in hours worked is misdirected. Because the sharp drop in hours is not present here, the emphasis rather should be on the small (perhaps negative) initial response followed by a subsequent large positive response. Investment, consumption, and output have similar dynamic responses.
In response to a positive technology shock, a standard flexible price model would have an immediate increase in hours worked. Therefore, such a model is inconsistent with the empirical dynamic responses. I show, however, that a flexible price model with habit persistence in consumption and certain kinds of capital adjustment costs can better match the empirical responses.
Some recent papers have critiqued the use of long run VARs to identify the dynamic responses to a technology shock. In particular they report that, when long run VARs are applied to data simulated from particular economic models, the point estimates of the impulse responses may be imprecisely estimated. However, based on additional simulation evidence, I find that, although the impact response may be imprecisely estimated, a finding of a delayed response is much more likely when the true model response also has a delayed response.
Keywords: macroeconomic models, vector autoregressions, impulse responses, weak instruments, longrun identification assumption
JEL Classifications: D24 E24 E32 O47
Recent papers by Galí (1999), Basu, Fernald, and Kimball (2004), and Francis and Ramey (2003) have claimed that in response to an unexpected improvement in technology, hours spent working declines. This finding challenges the standard macroeconomic flexibleprice model because that model predicts a strong positive correlation between employment and technology.
This paper's empirical evidence suggests that emphasizing the drop in hours worked is misdirected. Because the sharp drop in hours is not present in several of the data series examined here, the emphasis rather should be on the small (perhaps negative) initial response followed by a subsequent large positive responses. The paper also presents a flexibleprice model that is broadly consistent with the empirical dynamic responses. The empirical evidence describes how the economy responds to shocks to a productivity measure that has been corrected for utilization and reallocation. For the quantities considered here (output, consumption, investment, and hours worked), a consistent pattern emerges. When the technology shock occurs, the variables respond only slightly. Over time, the variables' dynamic responses gain strength.
To be consistent with the empirical responses, the standard quantitative dynamic flexibleprice model must be modified. The modifications pursued here are to make utility depend on past consumption (habit persistence) and to have capital adjustment costs.
Galí (1999) and Francis and Ramey (2003) measure productivity using aggregate labor productivity. Aggregate labor productivity, however, is a poor measure of technology because it can change for many reasons besides technological growth. In particular, in response to changes in the economy, workers sometimes vary their effort and, hence, output. Because this variation in the utilization of inputs is unobservable, changes in the utilization rate will appear as changes in productivity. To obtain an accurate measure of how technology increases productivity, one must control for these changes in utilization. In addition, reallocation of labor from an industry with low labor productivity to an industry with high labor productivity will also show up as in improvement in aggregate labor productivity that may not be due to increases in technology.
Combining the methods of Burnside, Eichenbaum, and Rebelo (1996) and Basu, Fernald, and Kimball (2004), the current paper constructs a quarterly measure of productivity that has been corrected for variations in utilization and also for reallocation between industries. In spite of the steps taken to remove endogenous influences, this productivity measure still might be influenced by other economic variables. To control for this endogeneity, I use the longrun approach of Galí (1999) to consider how the economy responds to exogenous shocks to the productivity series. The empirical results support the view that the immediate response to a technology shock is either negative or small. Although the initial responses are small, this paper shows that, within six quarters, the responses are positive and large.
The empirical work is presented with an emphasis on robustness. The paper considers responses to a technology shock identified under different identification assumptions. In addition, the paper reports confidence intervals for the impact response of a technology shock that are valid under the assumption of weak instruments. Although the resulting confidence intervals are wide, compared to the possible range associated with an unidentified shock, the identification scheme does restrict the possible responses.
Having documented how the economy responds to a technology shock, the challenge is to construct a model that has the same dynamic responses. The empirical responses are incompatible with a standard quantitative dynamic flexibleprice model; because, in the standard model, the period of the technology shock is when the variables respond most. The model needs to be modified to generate a more realistic delayed dynamic response. Galí (1999) and Basu, Fernald, and Kimball (2004) propose to resolve the model's problem by including sticky prices. Basu (1998) presents a sticky price model that successfully matches the immediate small empirical responses to a technology shock but fails to match the mediumterm empirical responses. With both sticky prices and wages, Altig, Christiano, Eichenbaum and Linde (2004) better match these responses. The current paper does not use sticky prices. Rather it uses a flexibleprice model with modifications to both preferences and technology. Preferences are modified such that today's utility depends on the previous level of consumption, habit persistence. Habit persistence implies that consumption responds more slowly to an increase in technology. Just habit persistence, however, is not enough to match the data. The technology to transform investment into capital must be modified to dampen the responses by investment and output. The paper presents two specifications for the investment technology: timetoplan and convex capital adjustment costs. When the capital adjustment costs depend on the ratio of new investment to capital, the model can match the initial period's responses but fails to match the subsequent increases. Having the adjustment costs depend on the growth rate of investment results in a better match to the longrun response. A timetoplan model also matches but, with only one kind of investment good, its responses are somewhat too jagged. These last two specifications work better because although they constrain the initial response by investment they allow the subsequent responses to be strong.
Although these models can match the consumption, investment and hours worked responses, the models presented here have a difficult time matching the response of the real interest rate. Although there is some uncertainty with respect to the true real interest rate response, the empirical impulse responses, in general, indicate that the real interest rate increases in response to a positive technology shock. Chiefly because of habit persistence, in the reported economic models, the real interest rate instead falls in response to a positive technology shock. I show that a model with consumption adjustment costs can better match the real interest rate response. Even this model, however, does not completely capture the empirical response by the real interest rate.
Recent papers by Erceg, Guerrieri and Gust (2004) and Chari Kehoe and McGrattan (2004) have critiqued the use of longrun VARs to construct impulse responses. Using simulated data from particular economic models, they show that, for these particular models, the point estimates of long run VARs are imprecisely estimated. However, when I adopt their approach of simulating data from a benchmark model, I find further evidence in favor of studying the shape of the impulse responses. Although the impact response may be imprecisely estimated, a finding of a delayed response is much more likely when the true model response also has a delayed response.
This paper's empirical work combines the two approaches taken in the literature to study productivity shocks. As in Basu Fernald and Kimball (2004), industrylevel data is used to construct a utilizationcorrected aggregate technology series. This series is then used as a variable in a vector autoregression, where, as in Galí (1998), exogenous technology shocks are identified under the assumption that only exogenous technology shocks affect the permanent level of productivity. In the first subsection, I describe how to use fluctuations in electricity usage (as in Burnside, Eichenbaum and Rebelo (1996)) and fluctuations in average hours (as in Basu Fernald and Kimball (2004)) to construct a quarterly utilizationcorrected technology series. In the second subsection, I then use the same vector autoregression approach that Galí (1999) and Francis and Ramey (2003) used for labor productivity to calculate impulse responses to a permanent shock to my constructed technology series. The main findings are that the response by per capita hours worked is initially small but ,within two years, hours worked experiences a large increase.
Forces besides technological progress affect labor productivity. In particular, the effort expended by workers and machines can vary endogenously over time. One cannot always observe this timevarying utilization of inputs, which can be a serious issue in measuring technology growth because a change in utilization can be mistaken for a change in technology.
This section uses methods proposed by Basu, Fernald, and Kimball (2004) [BFK] and Burnside Eichenbaum and Rebelo (1996) [BER] to approximate the changes in utilization with changes in observable variables. BFK approximate changes in utilization by using changes in average hours worked. The intuition for this approximation is that a firm would choose to vary both workers' hours and utilization until the costs and the benefits are the same.^{3} BER approximate changes in capital services by changes in electricity usage. In addition, because reallocation can increase aggregate productivity without requiring an increase in technology, one should also control for reallocation between industries. As is done in BFK, the aggregate productivity series is constructed by aggregating industrylevel productivity series.
To calculate the productivity series at the quarterly frequency requires a few strong assumptions. Because the industrylevel capital stock is unobservable at the quarterly frequencies, changes in electricity usage are used to approximate the changes in capital services. Because data on material usage is unavailable at the quarterly frequency, one must assume that there is very limited substitutability between materials and a mix of capital services and labor. BFK have been critical of estimating productivity without data on materials usage. In trying to measure quarterly productivity growth, it, however, is likely better to implement partially their methods than not use them at all.
The productivity equation is estimated using quarterly data at the industry level between 1972 to 2001. As in BFK, these industrylevel productivity series are then aggregated to generate an economywide productivity series.^{4} The industries used here are the eighteen twodigit SIC manufacturing industries. Their names and SIC codes are reported in Table 1.
Applying the methods of BER and BFK results in the following equation whose residual is a measure of productivity growth. Output and all the inputs are expressed in logged first differences of quarterly data. For each industry , the growth rate at quarter of the productivity series will be the residual from the following estimated equation^{5}:
(1) 
This output data will be measured by the Federal Reserve's measures of industrial production. The capital services variable will be approximated by data on electricity usage.^{6} The average hours and employment data are taken from the corresponding BLS measure. The capital and labor shares are calculated as the average valueadded shares from the BLS KLEMS database.^{7}The values of and are estimated. The values of and are constrained to be the same for all durable good sectors and the same for all nondurable good sectors.
In aggregating these industry level estimates, the aggregation equation is the same as in BFK.^{8} Aggregate productivity is calculated as
(2) 
The weight is the share of value added by industry . The share of materials is calculated using the materials share of grossoutput reported in the KLEMS dataset.
The parameters are estimated using twostep GMM under the assumption that the values of and are correlated but that there is no serial correlation. I use the three instruments that are commonly used in estimating this kind of production function. One of the instruments is the previous quarter's value of the Federal Funds shock resulting from the monetary VAR estimated in Christiano, Eichenbaum, and Evans (2001). The second instrument is the current and previous quarters' values of the difference between the aggregate GDP price deflator and the growth rate of the price of oil.^{9} The third instrument is the current and previous quarters' growth rate in real defense spending.^{10}
The GMM estimator requires an estimate of the variancecovariance matrix. Because, the estimation exercise is a system of equations for eighteen industries, the unconstrained version of the variancecovariance matrix requires a large (171) number of parameters. To reduce this number, one could place structure on the correlation between sectors. Conley and Dupor (2003) assume that the industries that use similar inputs have similar patterns of productivity. Given my interest in aggregated quantities, this correction has only a small effect on the aggregated results. ^{11}
The coefficient estimates are reported in Table 2. For the nondurable goods producing sectors, the estimate of the markup parameter is less than one. The difference between the estimate and constant returns, however, is not statistically significant. Some of the estimates reported in BFK were also less than one. They argue (p.29) that one possible explanation for these low estimates of is the omitted variable bias that would result from not including an estimate of reallocation effects.
The estimates of are positive and significantly different from zero. Positive coefficients imply that when utilization growth and, therefore, average hours growth are high, the growth in the utilizationcorrected measure of productivity is less than the growth in the measure of uncorrected productivity.
Although the work here has tried to construct a technology series that corrects for utilization and reallocation, one may be concerned that the series is still affected by measurement error. A such, this estimated technology series will be combined with a longrun identification assumption to identify the part of this series that seems to have a longrun affect on the level of technology. The next section describes how to implement such a longrun identification assumption.
Here, as in Galí (1999), a technology shock will be identified as a permanent shock to productivity.^{12} These shocks and resulting impulse responses are computed in the following manner. Consider the following structural vector autoregression (VAR) for a vector of variables
(3) 
(4) 
The shocks and (where has elements) are assumed to be independent. Galí identifies the technology shock by assuming that only the technology shock can have a permanent effect on the level of productivity . All other shocks are assumed to have no longrun effect. To impose this restriction is to impose a restriction on the moving average representation of the data. Denote the moving average representation by:
(5) 
To actually estimate the structural VAR with this longrun restriction requires a restriction on the structural VAR coefficients. To make the notation clear, I rewrite the structural VAR as
(6) 
for all  (7) 
(8) 
Estimating this equation results in a time series of the technology shocks. To calculate the impulse responses requires the moving average representation of in terms of and . One approach to calculate this moving average representations starts with the reduced form VAR^{16}
(9) 
(10) 
To summarize, the impulse responses can be calculated using the following four steps.
(11)  
The next section reports the dynamic responses to an exogenous shock to productivity. The measure of productivity is the result of using Equation 2 to aggregate the industrylevel estimates calculated using Equation 1.
Two sets of results are reported. The first results correspond exactly to the methods used by Galí (1999) and Francis and Ramey (2003) except that I replace their use of labor productivity with my productivity measure. Compared to the initial response by hours when using laborproductivity, the initial response here is much more positive. Either hours declines less or it does not decline at all. The result reported here do not overturn earlier critiques of the standard quantitative dynamic flexibleprice model. Hours still do not respond positively to the shock for the first year.
The second set of results report the dynamic responses by investment, output, and consumption as well as hours. These results are useful because they provide the basis upon which to characterize the ability of the macroeconomic model to match the data. These variables do not respond much in the period of the shock. In subsequent periods, the variables start to increase in response to the technology shock.
As in Galí (1999) and Francis and Ramey (2003), the impulse response are from a twovariable VAR on the growth rates of a productivity series and the growth rates of hours worked. Productivity is measured using two different time series: labor productivity and my constructed aggregate productivity series. In addition, impulse responses are calculated from a VAR of the constructed productivity series and the levels of percapita hours worked. Christiano, Eichenbaum and Vigfusson (2003) argue that estimating the longrun VAR with hours in first differences is misspecified relative to estimating a VAR with percapita hours in levels. This section, however, reports both sets of results to maximize comparability with the earlier literature. Confidence intervals around the estimate are calculated using a bootstrapped approach. Sampling from the residuals with replacement, the estimated VAR is used as a data generating process to simulate time series. Using the simulated data, the VAR is estimated and the responses to a permanent positive increase in technology are calculated. After simulating the data 500 times, the variance of each period's impulse response is calculated. The resulting confidence interval is the estimated response plus or minus 1.96 times the standard deviation.^{17}
Figure 1 reports the response of the three different VARs. The three different specifications have similar responses. Nothing much happens on impact. Subsequently, hours worked begins to increase. For all three VARs, the response on impact is much greater than that reported in Galí or in Francis and Ramey. These results, however, are not large enough to overturn the conclusions of previous work. Hours still do not respond much initially to a positive productivity shock. Therefore, the evidence is still against models that predict a large initial response to increases in productivity.
To compare the model to the data, one needs to know how other variables in the economy respond to a technology shock. This section describes how other variables (output , consumption, investment , and the real interest rate )^{18}respond to a technology shock. The approach taken here is to estimate the set of equations described in Section 2.2 . The vector of other variables is constructed as follows
(12) 
Figure 2 reports the data used in the VAR. The log of percapita hours worked increases over this time period. This increase in hours worked is the result of two opposite trends. An increase in the labor force participation rate (from 60 to 67 percent over this time period) offsets the decline in average hours worked, (for production workers, the average work week has declined from 36 hours to 34 hours).
Figure 3 reports the implications for the technology growth series by applying the longrun identification assumption. The constructed technology series is presented along with the technology series implied by the VAR and allowing only permanent shocks to technology. In order to emphasize the role of the long run identification assumption, both series are presented with mean zero. Perhaps not surprising the series that results from the long run identification series is much less volatile than the original series.
Impulse responses for the baseline longrun VAR are reported in Figure 4. In general, quantities take time to respond to the technology shock. Hours worked responds only slightly in the impact period of the shock. It takes several quarters before hours worked has a strong response. Consumption only responds gradually over time. In the impact period, the response is only about half of what it will be 10 quarters later. In percentage terms, the investment response is stronger on impact but the strongest investment response is about four quarters later. The real interest rate, however, has a very different response. The real rate jumps 80 basis points on impact. It then steadily declines but it does take six years before the real rate returns to normal.
The results for the two other VARs are similar. Identify technology shocks with a shortrun recursiveness assumption produces a few small differences. The largest difference is that the real interest rate response is much weaker. Also, the consumption response is not as strong but the investment response is of greater size and duration. For this dataset, identifying a technology shock as a permanent shock to labor productivity again produces a very similar set of responses.
I have claimed that the delayed response to a positive technology shock is a robust feature of the data. The following section characterizes this robustness. Using the baseline VAR as the DGP process, Figure 5 reports that in the majority of cases the response by hours six quarters after a shock is greater than the response on impact.
Table 3 reports that for the majority of simulations, the impact period response for consumption, investment, or hours worked is much smaller than responses several periods later.
To identify a permanent change in technology requires estimating an instrumental variables regression. One may be concerned about whether the above conclusions concerning the shape of the impulse responses are robust to weak instruments. Although the actual instrumental variables regression may have problems with weak instruments, the evidence is that the conclusions concerning shape are more robust.^{20}
When instruments are only weakly correlated with the explanatory variables, confidence intervals can often be much wider than those calculated using standard methods. In the equation estimated here, the lagged level of hours worked is used to instrument for the growth rate of hours worked. If hours worked has either a unit root or else approaches a unit root asymptotically, then hour worked would be a weak instrument.^{21} Valid confidence intervals for estimation using weak instruments, however, were established by Anderson and Rubin (1949). In the present context, their method can be implemented as follows. Begin with the IV regression where any dependence on lagged values has been removed by a linear projection. All that is left is to estimate , in the following equation,
These confidence intervals are constructed holding as fixed the values of the reduced form VAR coefficients. Hence if we combine these estimates of the possible values of with the reduced form VAR, the resulting hours response six quarters later is between 0.05 and 0.27.
Similar calculations can be done for the six variable system. A grid search on all the possible values of would be particularly laborious for the larger system. For example, a five dimensional space with 100 grid points per dimension would be ten billion points. However, one can approximate the grid search by instead sampling over the parameter space. A random sample of the same space should be sufficiently informative.^{23} Table 4 reports confidence intervals for that result from those values of that belong to the AR confidence interval. Going from the bivariate autoregression to the multivariate regression seems to have both tightened and shifted upwards the confidence intervals on the hours worked response.
There appears to be a great deal of uncertainty of investment's response on impact. This uncertainty does not seem to be reflected in the standard bootstrap confidence intervals reported above. Perhaps the most surprising thing is the improvement of identification that results from using the constructed technology series.
The emphasis on the shape of the hours worked response can also be studied in the context of weak instruments. Consider Figure 7, which shows the connection between the response by hours on impact combined with its response six quarters later. The oval indicates all observed responses, holding the reduced form VAR coefficients fixed. The grey area indicates those values that are associated with a value of the Anderson Rubin statistic less than the 95 percent critical value of 9.488. Although this oval does not take into account the sampling uncertainty of , the figure is supportive of placing a greater emphasis on the shape of responses.
The evidence from the weak instruments reinforces the view that the robust finding is that the hours response is initially small but grows over time. Models therefore should be constructed to attempt to match this finding.
The following sections report on the sensitivity to using different data and also to using different sample periods.
To check for data sensitivity, the empirical VAR is estimated using an ex ante measure of the real interest rate in place of the ex post measure. The ex ante real interest rate is based on the difference between the three month treasury rate and the forecasted inflation in the GDP deflator for the next quarter. The forecasted inflation is the median forecast from the Survey of Professional Forecasters. The results are presented in Figure 8. On impact, the ex ante real interest rate only increase 50 basis points rather than the ex post real interest rate increase of 100 basis points. However, all of the quantities experience responses that are similar to those reported in the baseline VAR.
One might wonder if the stronginterest rate response is stable over time. In particular, Galí, LópezSalido and Vallés (2003) argue that monetary policy changed in the United States with Paul Volker and that therefore the responses to a technology shock look very different after 1983 than they did earlier. There are two ways to test the stability. The first is to calculate from a regression of just a subsample of the identified technology series on the reduced form residuals, estimated from the fullsample VAR. This method holds the VAR fixed but sees whether a particular episode drove the estimation results. These results, although not reported, are almost identical to the results in Figure 8. Therefore, we have evidence against any particular technology shock episode driving the results.
The second approach, which was used by Galí LópezSalido and Vallés, is to reestimate the entire VAR but begin in 1983. The impulse responses from this VAR are reported in Figure 8. For the first year after impact, these responses are quite different from the responses in the baseline VAR. For the shorter sample VAR, the variables are less responsive on impact. After two years, the responses by quantities are quite similar. These results emphasizes the robustness of describing the response to a technology shock as being a delayed response. One important difference between the two sets of results is that, with the shorter sample, the real interest rate does not respond to the technology shock.
The shocks identified using the post1982 VAR are very different from the same shocks identified using the full sample of data. The full sample shocks are much more volatile. In addition, they are not closely related to the post1982 shocks. The post1982 shocks are the same sign as the full sample shocks only about 50 percent of the time.
The question is whether the full sample or the truncated sample correctly identifies the true technology shocks. Determining which is the correct set of responses is a difficult question. A simple likelihood ratio test would support using the truncated sample. However, to exclude three of the four recessions in the covered time period throws away a lot of information. As such, the analysis here will continue to use the benchmark responses, but with the caveat that other responses are possible.
Having described the empirical responses, the next step is to develop a model that can match the data. The model presented here has two features that are different from a standard quantitative dynamic flexibleprice model. The first is that the economic agent has habit persistence in the utility function. Thus, the previous period's level of consumption affects current utility. Habit persistence results in a slower response by consumption. (In order to match the real interest rate response, a model with consumption adjustment costs is also presented.) The second feature and the focus of the paper is how investment is transformed into capital. I consider two different specifications of this transformation: timetobuild and capitaladjustmentcost models. Both specifications have the property of preventing capital from adjusting quickly.^{24} The timetobuild model has a lag between the decision to increase the capital stock and the actual increase in the capital stock. Likewise in the capital adjustment model, increasing investment is expensive and therefore an economic agent will have an incentive to smooth out investment.
All of these features have been used previously to explain other economic phenomena. In particular, Christiano and Todd (1995) and Christiano and Vigfusson (2003) document the properties of a particular parameterization of the timetobuild model, the timetoplan model, where investment cannot respond much in the first period of a shock. Christiano and Vigfusson show that this model is much better than a standard quantitative dynamic flexibleprice model in matching the output growth dynamics and the leadlag relationship between output and business investment.^{25} Models with capital adjustment costs that depend on the ratio of investment to capital have been used extensively in the Tobin's Q literature. (See Chirinko (1993) for a survey.) Boldrin, Christiano, and Fisher (2000) and Beaudry and Guay: (1996) document how adding capital adjustment costs and habit persistence allows a macroeconomic model to explain both business cycle facts and asset pricing issues. Topel and Rosen (1988) make capital adjustment costs depend on the growth rate of investment to explain housing investment. Christiano, Eichenbaum, and Evans (2001) use a similar specification to generate improved dynamics in a sticky price model. Francis and Ramey (2003) also consider a capital adjustment model to explain the low correlations between productivity and employment. In their model, capital adjustment costs depend on the ratio of investment to capital.
The model has a representative agent who chooses consumption and the fraction of time spent working to maximize utility, where utility is defined as
(13) 
(14) 
As an alternative to habit persistence, I also consider a model that features consumption adjustment costs. In this model, the habit persistence coefficient is set to zero and the resource constraint is the following.
(15) 
Two different production functions are considered here. As is common in the macroeconomic literature, I use a Cobb Douglas technology function.
The second constraint specifies how investment is transformed into capital and will be described in Section 3.3 .
One difference between this model and Francis and Ramey is the specification of the utility for leisure. They used an indivisible labor model; whereas, here, the model has the standard divisible labor utility function. The two specifications imply very different values for the labor supply elasticity. The labor supply elasticity is much greater for the indivisible labor model, where the labor supply is infinitely elastic. With divisible labor, the labor supply would have an elasticity of about three.^{28} Therefore, the labor supply will be less responsive in the specification studied here.
In order to match the data, the model should have growth and therefore the level of technology should be nonstationary. The level of technology has the following functional form
(16) 
(17) 
This section describes how investment is transformed into capital. Three different specifications are considered. The first is the timetobuild model of Kydland and Prescott (1982). In this model, several quarters pass before a desired increase in the capital stock is realized. The second is the convex capital adjustment costs where the costs are a function of the ratio of investment to capital. The third has adjustment costs that depend on the growth rate of investment.^{29}
In the timetobuild model (Kydland and Prescott 1982), the investment technology has two features. The first is that the time between the decision to increase the capital stock and the actual increase is greater than a quarter. In the current application, four quarters pass between making a decision to increase the capital stock and the actual increase. The second feature is that the increase in the capital stock is paid for over time. In other words, a project initiated at quarter results in an increase in the capital stock four quarter later. The total cost of the project equals the increase , but the project is paid for in installments. Thus, investment consists of several different projects that are at various stages of completion. In particular, period investment equals:
(18) 
Two versions of capital adjustment costs are described here. The first, (which I will refer to as CIK), is the more common version where the cost of capital adjustment is a function of the ratio of current investment to capital
(19) 
(20) 
At calibrated values, the standard RBC model fails to match the main facts discussed in the empirical section. In particular, all the variables respond to a positive technology shock most strongly on impact. Therefore, they completely miss the delayed response observed in the data.
Recent model estimation has often tried to match impulse responses by using a GMM weighting function. However, the goal of this paper is to show the flexibility of these macroeconomic models to match estimated technology impulse responses. As such, rather than attempting to match any particular set of responses, I will present a comparative analysis that will clarify how the flexibleprice models are compatible with the responses observed both in this paper and in many of the other papers in the literature.
Figure 9 reports results for just the response of hours on impact and how the different models are able to capture the response of hours to a positive technology shock. For each of the different models, some of the model's coefficients were fixed and others were allowed to vary. The coefficients that were allowed to vary were the ones that characterize the newer features of the model. In particular, results are reported for different values of both the degree of habit persistence and the coefficients related to how investment is translated into capital, (i.e. the degree of investment adjustment costs or timetobuild weights .). The other coefficients were held constant.^{31}The models were loglinearized and solved using the undetermined coefficients method of Christiano (2001).^{32}
In Figure 9, there are several things to note. First, all three models have parameterization that are consistent with hours worked falling in response to a positive technology shock. Many, including Galí and Basu, Fernald, and Kimball, have argued that the earlier empirical claims of negative hours responses to a technology shock is evidence against flexibleprice models. Clearly, the flexibleprice models reported here can imply that hours fall on impact of a positive technology shock. Therefore, these previous criticism of flexibleprice models, although valid for the standard RBC model, do not apply to these models. Furthermore, for all three models, different combinations of investment adjustment costs and habit persistence can generate a given hours response. Habit persistence and investment adjustment costs actually work against each other. Investment adjustment costs decrease the investment response and increase the consumption response, and habit persistence increases the investment response and decreases the consumption response.
Figure 10 illustrates these trade offs for the CII model and how consumption and investment impact responses depend on the degree of habit persistence and investment adjustment costs. The shaded areas indicate the bootstrap confidence intervals around the empirical impact effect. For investment, the width of the area suggest that the investment response is not very informative about the best values of habit persistence and the investment adjustment cost.
Figure 11 reports on the ability of all three models to capture the dynamic response of hours worked, consumption, and investment. The benchmark empirical results from Figure 2 are reproduced here.
The first panel reports responses for a model with habit persistence and investment adjustment costs that depend on the growth rate of investment, the CII model. As seen in Figure 10, it is fairly easy to find model parameterization that match the consumption response. The investment response is somewhat more difficult with the response being too strong compared to the empirical response. The hours worked response is delayed. However, hours actually do not respond enough on impact and then responds too strongly afterwards. The differences in responses, however, are within the standard bootstrapped confidence intervals, indicated by the shaded regions.
The second panel reports results from the timetobuild and CIK models. The CIK model can match the initial responses. However the CIK model can not match the increases in responses because the ratio of investment to capital does not quickly change. Without a change in the ratio of investment to capital, the investment adjustment costs remain high. The timetobuild model does better at creating responses that are small initially but then increase. The main problem with the timetobuild model is that the responses are too jagged. The jaggedness results from there being only one kind of capital with only a four period building period. A timetobuild model with much smoother responses can be found in Edge (2000) where there are many different kinds of capital goods, with each kind of capital requiring a different number of periods to build.^{33}
Although both the investment growth rate model and the timetoplan model have done well in fitting the responses of consumption, investment and hours worked, this section makes clear that the models have a much harder time matching the strong interest rate response observed in the benchmark results. Figure 12 reports the real interest rate results from the empirical VAR and from the economic models. Given that consumption only slowly increases in both the CII and timetobuild models, it may seem puzzling that the real interest rate actually falls on impact. The explanation, however, involves the definition of marginal utility with habit persistence. First, the real interest rate can be expressed in terms of a ratio of marginal utilities. In particular, ignoring uncertainty, the real interest rate can be written as
For further intuition, consider the real interest rate in a model of internal habit which drops the forward looking part of the external habit specification used here.^{34}
Modeling the production function as a CES production function, rather than CobbDouglas, does strengthen the growth rates of investment. However, this feature does not overcome the negative interest rate response generated by the habit persistence. Another option would be to have habit persistence depend on the difference between current consumption and a habit stock that only slowly evolved with current consumption. In other words, the new utility function is
One partial remedy to the falling real rate is to replace habit persistence with consumption adjustment costs. With consumption adjustment costs, one continues to delay the consumption response but, unlike habit persistence, consumption adjustment costs do not introduce the growth rate of consumption into marginal utility. As such, the real interest rate increases on impact. However, as seen in Figure 12, the rise is not as much or as persistent as the empirical point estimates.
Consumption adjustment costs may seem particularly adhoc. However, like investment adjustment costs, they help macroeconomic models match the empirical impulse response functions. Future work will be required to find a more structural mechanism to reduce how quickly both the level and the marginal utility of consumption increases in response to a positive technology.
Another possible solution would be to have serially correlated technology shocks. Although serially correlated shocks can result in a model with a stronger real interest rate response. Figure 13 reports results with the assumption that the autoregressive coefficient on the growth rate of technology equals 0.7. As shown in Figure 13, in models with investment adjustment costs, the real interest rate increases almost 50 basis points on impact. ^{35} However, this real interest rate response is less than the estimated response. In addition, the responses of the other variables particularly investment and hours worked are now too weak compared to the estimated responses. So any gain in better real interest rate fit is lost in worse fit of hours worked and the real interest rate.
The real interest rate response is a problem for these models. As mentioned in the empirical section, the interest rate response is somewhat uncertain with both a wide confidence interval and being sensitive to the sample period. Therefore, some may question the seriousness of a failure to match this response. In fact, many researchers have felt that a fall in real interest rates followed by a positive increase in output is a desirable characteristic in a technology driven model.
Recent papers by Erceg, Guerrieri and Gust (2004) and Chari Kehoe and McGrattan (2004) have criticized the use of longrun VARs to construct impulse response. Both papers have shown that, for particular model parameterizations, the point estimates of the impulse responses may be estimated imprecisely. This section provides some additional simulation evidence concerning the responses. I show for a particular set of models that, although, these authors do have a valid concern about the possible imprecision of the point estimates, the impulse responses are informative about the shape of the responses. Of specific relevance to the current paper, simulation evidence shows that these impulse responses can distinguish between a model where hours responds most on impact and a model where hours have a delayed (hump shaped) response. However, sign restrictions are not as informative. For the models considered here, the finding of a positive response is unlikely to allow us to discriminate between a model that has a positive hours response on impact and a model that has a negative hours response on impact.
The simulation evidence presented here comes from three models: a standard quantitative dynamic flexibleprice model (where hours responds positively and with the largest response being immediate), a model with CII investment adjustment costs (where hours responds positively and with a delayed response), and a model with CII investment adjustment costs and habit persistence. In order to estimate a nontrivial bivariate VAR, each model has an additional shock that affects the labor preference parameter . Hence the utility function becomes
(21) 
Given these estimates, I then used each model to generate 500 data sets of 200 observations each on labor productivity and hours worked. For each simulated data set, I then estimated a long run VAR on the growth rate of labor productivity and the log level of hours worked. Figure 14 reports the theoretical model response for hours worked and also the average response estimated from the simulated data. For all models, the average responses are biased upwards, but the average responses are reasonably close. However, an interval that contains 90 percent of the simulated responses is very large. Figure 14 plots such an interval for the CII model. Given these wide intervals, one may be concerned that one could not distinguish between the models.
Some measures but not others are able to distinguish between the models. Figure 15 reports the probability of observing two results in each model. The top panel reports the probability of observing a positive impact response. Because of the wide confidence intervals and the upward bias in the hours response, all three models, including the model where the true response is negative, have a high probability of observing a positive value. Therefore, a finding of a positive response is not very informative about the true data generating model. As can be seen in the bottom panel of Figure 15, the shape results are much more informative. In the standard RBC model, the downward trend in the hours response is apparent in the small fraction of responses that are greater than the impact response. Likewise, in the model with investment adjustment costs, the humpshaped response is readily apparent by the large fraction of responses that are greater than the impact response.
One way to quantify the difference between models would be with a posterior odds ratio. In particular given the observed data , one would calculate the odds of model one being preferred over model two as follows.
These calculations should be taken as only a guideline. Other models with more features or other shocks might give different results for both the identification of point estimates and response shapes. However, at least for these simple models where the criticisms of EGG and CKM are valid, studying the shape of the responses seems to be a valid way to use longrun VARS to learn about the economy.^{38}
The main empirical conclusion of this paper is that, for the U.S. economy, the response by percapita hours worked to a technology shock is initially small but subsequently increases. The small initial response is evidence against any model, including the standard quantitative dynamic flexibleprice model, that predicts a large immediate response by employment to a technology shock. These results do not, however, completely invalidate the use of real technology shocks to explain business cycles since variables do respond in the medium term to these shocks. Therefore, the task is to develop models that can explain both the shortterm and longterm responses to technology shocks.
The current paper presents quantitative dynamic flexibleprice models that can be reconciled with the observed responses by quantities to a technology shock. Of course, this reconciliation is not a rejection of other possible explanations. These other possible explanations might include the examples provided by Basu, Fernald, and Kimball (2004): stickyprice models, multisector reallocation models, and cleansing models of recessions. All of these explanations should be scrutinized further to determine their relative merits.
The current paper has done three things. First, it has presented new empirical dynamic responses for models to match. Second, it has shown that the estimated shape of these dynamic responses may be more informative than the sign of the impact response. Third, it has put forward flexibleprice models that better explain these delayed responses. In effect, it has raised the standard for criticisms of the flexibleprice model. It was easy to show that the small initial response by hours worked was inconsistent with the standard flexibleprice model. With the additional features discussed here, a flexibleprice model can be made consistent with these observations concerning hours worked. The new challenge will be to build upon the improved fits described here.
In order to make the paper somewhat more selfcontained, this appendix summarizes how to estimate a model by maximum likelihood in the frequency domain. For more details and application,. see Christiano and Vigfusson (2003).
Begin with a time series of data, where is a finitedimensional column vector with zero mean. In this paper's analysis, the vector is defined as
(22) 
It is well known (Harvey, 1989, p. 193) that for large, the Gaussian likelihood for such a time series of data is well approximated by:
(23) 
(24) 
To estimate a model by frequency domain maximum likelihood, one needs the mapping from the model's parameters, to the spectral density matrix of the data, . The following describes this mapping.
The first step is to solve a linearized version of the macroeconomic model. One can then use the linearized solution to write a linear approximation of the process
(25) 
In the twoshock model, is defined in (22), is matrix polynomial in and
In all cases, I restrict so that
The spectral density of at frequency is
Altig, David, Lawrence J. Christiano, Martin Eichenbaum and Jesper Linde, 2004, `` An Estimated Dynamic, General Equilibrium Model for Monetary Policy Analysis,'' Manuscript.
Anderson, T.W. and H. Rubin 1949 `` Estimation of the Parameters of A Single Equation in A Complete System of Stochastic Equations'', Annals of Mathematical Statistics, 20, pp. 4663
Basu Susanto 1998 `` Technology and Business Cycles: How Well Do Standard Models Explain the Facts?'' Beyond Shocks: What Causes Business Cycles? Federal Reserve Bank of Boston, Conference Series No. 42
Basu Susanto John G. Fernald 2000 `` Why is Productivity Procyclical? Why do we care?'' NBER Working Paper 7940
Basu Susanto, John G. Fernald, and Miles S. Kimball. 2004. `` Are Technology Improvements Contractionary'' NBER Working Paper 10592
Basu Susanto John G. Fernald and Matthew D. Shapiro 2001 `` Productivity Growth in the 1990s Technology Utilization or Adjustment'' NBER Working Paper 8359
Beaudry Paul, Alain Guay. 1996 `` What Do Interest Rates Reveal about the Functioning of Real Business Cycle Models?'', Journal of Economic Dynamics and Control, 20,.16611682
Boldrin, Michele, Lawrence J. Christiano and Jonas D.M. Fisher, 2001 `` Habit Persistence, Asset Returns and the Business Cycle'' American Economic Review 91 149166.
Burnside Craig 1996 `` Production Function Regressions, Returns to Scale and Externalities,'' Journal of Monetary Economics, 37, 177201.
Burnside Craig, Martin Eichenbaum and Sergio Rebelo, 1996 `` Sectoral Solow Residuals,'' European Economic Review, 40, 861869.
Carroll Christopher D. , Jody Overland, and David N. Weil 2000 `` Saving and Growth with Habit Formation'' American Economic Review, 90(3), 341355
Chari, V.V., Patrick Kehoe, and Ellen McGrattan 2004 `` An Economic Test of Structural VARs'' Federal Reserve Bank of Minneapolis Research Department Working Paper Number 631
Chirinko Robert S. 1993 `` Business Fixed Investment Spending'' Journal of Economic Literature 31 18751911.
Christiano, Lawrence J., 2001, `` Solving Dynamic Equilibrium Models by a Method of Undetermined Coefficients'', forthcoming. See also, NBER Technical Working paper 225.
Christiano Lawrence J. Martin Eichenbaum and Charles Evans 2001 `` Nominal Rigidities and the Dynamic Effects of a Shock to Monetary Policy'' manuscript Northwestern University
Christiano, Lawrence J. and Martin Eichenbaum, 1992, `` Current RealBusinessCycle Theories and Aggregate LaborMarket Fluctuations'', American Economic Review, 82(3), June, 43050.
Christiano Lawrence J. and Jonas D. M. Fisher 1998 `` Stock Market and Investment Good Prices: Implications for Macroeconomics,'' Federal Reserve Bank of Chicago Working Paper
Christiano, Lawrence J. and Richard M. Todd, 1996, `Time to Plan and Aggregate Fluctuations', Federal Reserve Bank of Minneapolis Quarterly Review, Winter, 1427.
Christiano,Lawrence J. and Robert J. Vigfusson Maximum likelihood in the frequency domain: the importance of timetoplan Journal of Monetary Economics 50(4),789815
Conley Timothy G. William D. Dupor 2003 `` A Spatial Analysis of Sectoral Complementarity'' Journal of Political Economy 111(2) p 31152
Cooper, Russell W. and John C. Haltiwanger 2000 `` On the Nature of Capital Adjustment Costs'' manuscript
DeJong David N., Beth F. Ingram, and Charles H. Whiteman, 2000 `` Keynesian Impulses versus Solow Residuals: Identifying Sources of Business Cycle Fluctuations'', Journal of Applied Econometrics, 15(3), pp. 275287.
Edge, Rochelle M. 2000 `` Timetobuild, timetoplan, habitpersistence, and the liquidity effect,'' Board of Governors International Finance Discussion Paper 673
Erceg,Christopher, Luca Guerrieri and Christopher Gust. 2004 `` Can LongRun Restrictions Identify Technology Shocks?'', Federal Reserve Board of Governors International Finance Discussion Paper 792
Faust, Jon and Eric Leeper 1997 `` When Do LongRun Identifying Restrictions Give Reliable Results?'' Journal of Business and Economic Statistics 15(3) 34553
Francis, Neville and Valerie A. Ramey 2003 `` Is the TechnologyDriven Real Business Cycle Hypothesis Dead? Shocks and Aggregate Fluctuations Revisited'' manuscript October
Fuhrer, Jeffrey C. 2000 `` Habit Formation in Consumption and Its Implications for MonetaryPolicy Models'' American Economic Review 90(3) 371390
Galí, Jordi, 1999 `` Technology, Employment, and the Business Cycle: Do Technology Shocks Explain Aggregate Fluctuations,'' American Economic Review, 89 249271.
Galí, Jord, J. David LópezSalido and Javier Vallés 2003 `` Technology shocks and monetary policy: assessing the Fed's performance'' Journal of Monetary Economics, 50(4), 723743
Harvey, Andrew, 1989, Forecasting Structural Time Series Models and the Kalman Filter, Cambridge University Press.
Jermann, Urban J., 1998 `` Asset Pricing in Production Economies,'' Journal of Monetary Economics 41 257276.
King, Robert G., Charles. I. Plosser, James H. Stock, and Mark W. Watson. 1991. `` Stochastic Trends and Economic Fluctuations.'' American Economic Review. 81(4) 819840.
King Robert G. and Mark W. Watson 1997 `` Testing longrun neutrality'' Federal Reserve Bank of Richmond Economic Quarterly 69101
Koeva, Petya 2000 `` The Facts About TimetoBuild'' IMF Working Paper August
Kydland, Finn E. and Edward C. Prescott, 1982, `` Time to Build and Aggregate Fluctuations', Econometrica, 50, November, 134570.
Lum Sherlene K.S. and Brian C. Moyer 1998 `` Gross Product by Industry 199597'' Survey of Current Business November
Pesavento, Elena and Barbara Rossi 2003 `` Do Technology Shocks Drive Hours Up or Down? A Little Evidence From an Agnostic Procedure'' Duke University manuscript
Shleifer, Andrei 1986 `` Implementation Cycles'',Journal of Political Economy, 94(6) 116390
Topel, Robert and Sherwin Rosen 1988 `` Housing Investment in the United States'', Journal of Political Economy 96:4 718740.
Vigfusson, Robert J. 2003 `` How Does the Border Affect Productivity? Evidence from American and Canadian Manufacturing Industries '', Federal Reserve Board of Governors, International Finance Discussion Papers 788
Whelan, Karl 2000 `` A Guide to the Use of Chain Aggregated NIPA Data'' Federal Reserve Board of Governors Finance and Economics Discussion Series 200035 (June)
Wright, Jonathan H. 2002 `` Testing the Null of Identification in GMM'', Federal Reserve Board of Governors, International Finance Discussion Papers 732
Durable Good Producers  SIC Code  Nondurable Good Producers  SIC Code 

Lumber  24  Food  20 
Furniture  25  Textiles  22 
Glass Stone & Clay  32  Apparel  23 
Primary Metals  33  Paper  26 
Fabricated Metals  34  Printing  27 
Industrial Machinery  35  Chemicals  28 
Electrical Machinery  36  Petroleum  29 
Transportation  37  Rubber and Plastics  30 
Instruments  38  
Miscellaneous  39 
Coef NonDurable 
Tstat NonDurable 
Coef Durable 
Tstat Durable 


0.82  1.13  1.04  0.28  
0.31  2.45  0.17  1.65 
Results: 18 Manufacturing
Industries 19722001
GMM Estimation with Asymptotic
Standard Errors
Tstat for is test of equal to one.
Tstat for is test of equal to zero.
Degrees of Freedom Equal to
167
Periods After Shock  H  C  I 

1  90.1  77.4  94.1 
2  90.7  68.0  92.0 
3  91.4  77.0  93.4 
4  87.8  74.4  88.4 
5  88.5  79.4  88.2 
6  90.3  84.7  87.6 
7  89.7  83.8  86.3 
8  89.0  85.6  85.1 
9  88.6  86.0  83.1 
Estimate Confidence Intervals Technology 
Estimate Confidence Intervals Labor Productivity 
Theoretical Bounds Technology VAR 
Theoretical Bounds Labor Productivity VAR 


Labor Productivity  (0.06, 0.38)  (0.22, 0.36)  0.518  0.55 
Hours Worked  (0.007, 0.084)  (0.09, 0.11)  0.108  0.124 
Real Interest Rate  (28, 98)  (38, 100)  101.1  102.6 
Output  (0.05, 0.42)  (0.31, 0.45)  0.564  0.62 
Consumption  (0.07, 0.32)  (0.03, 0.20)  0.469  0.46 
Investment  (0.69, 1.08)  (1.09, 1.30)  1.32  1.41 
Notes: Confidence Intervals Constructed using 95 percent critical value
of 11.07 for tech and 9.488 for labor
productivity
Thick Line VAR with Constructed Productivity Series and Hours In Levels
'X's: VAR with Constructed Productivity Series and Hours In Difference
Stars VAR with Labor Productivity and Hours in Differences
Grey Area: 95 Percent Bootstrapped Confidence Interval Centered Around VAR with Constructed Productivity Series and Hours In Levels
The xaxis represents time after the shock and ranges from zero to forty in increments of five.
The yaxis ranges from 0 to 0.45 in increments of 0.05.
Figure 1 reports the response of three different VARs.
Figure 2 is titled ``Data Used in the VAR Analysis''. It consists of six graphs presented in a threebytwo matrix.
The values on the xaxis are years ranging from 1972 to 2001.
The upper left graph is titled ``Productivity Growth''.
The yaxis ranges from 0.045 to 0.035.
The line appears as jagged oscillations around 0.01.
The upper right graph is titled ``Log of Per capita Hours''.
The yaxis ranges from 1.505 to 1.545 in increments of 0.05
The line begins at 1.52. By 1975, it has decreased to about 1.505. It gradually increases to about 1.525 in 1980. It decreases to 1.505 in 1983. Afterwards, it increases fairly quickly and achieves a value of 1.535 in 1987. It then begins to drop, and has a value of 1.52 in 1992. Afterwards, it increases sharply to 1.545 in 2000. In the remaining year of the graph, it decreases slightly and ends with a value of 1.535 in 2001.
The middle left graph is titled ``Real Interest Rate''.
The yaxis ranges from 4 to 8 in increments of 2.
The real interest appears to follow a general trend. The actual values appear to oscillate around this trend. The trend remains close to 1 from 1973 to 1980. Afterwards, it increases to about 6 by 1983. From there, the trend appears to decreases to about 2 in 1993. Afterwards, it increases slightly to 3 in 1995 and remains roughly there for the remaining periods.
The middle right graph is titled ``Output Growth''.
The yaxis ranges from 0.03 to 0.025 in increments of 0.01.
The line appears as jagged oscillations around 0.005. There are three noticeable spikes in this time frame. In 1975, output growth feel to about 0.025. In 1978, output growth grew at about 0.025. In 1981, output growth fell to about 0.03.
The lower left graph is titled ``The Consumption Share of Output''.
The yaxis ranges from 0.78 to 0.83 in increments of 0.01.
The line begins at 0.805. It increases to 0.815 in 1975. It then decreases to 0.78 in 1980. From there it increases to 0.81 in 1983 before falling down again to 0.795 in 1984. Afterwards, it trends upwards to about 0.83 in 1992. From there, it trends downwards to 0.80 in 1999. Afterwards, it increases up to 0.815 in 2001.
The lower right graph is titled ``Investment Share of Output''.
The yaxis ranges from 0.17 to 0.22 in increments of 0.01.
The line begins at 0.195. It decreases to about 0.185 in 1975. From there it trends upwards to 0.22 in 1978. It decreases to 0.185 in 1983. From there it increases slightly until 0.205 in 1985 before falling to 0.170 in 1992. It trends upwards from there and has a value of 0.20 in 199. It decreases until it has a value of 0.185 in 2001.
Thin Line Constructed Technology Growth Rate Series (Demeanded)
Thick Line: Technology Growth Rate Series (Demeanded) implied by the Long Run Identification Assumption
Figure 3 is titled ``The Result of Applying the LongRun Identification Assumption''.
The xaxis ranges from 1972 to 2001.
The yaxis ranges from 0.03 to 0.04.
The graph consists of two lines.
Thick Line VAR with Constructed Productivity Series and Hours In Levels, with Long Run Identification Assumption
Grey Area: 95 Percent Bootstrapped Confidence Interval Centered Around VAR with Constructed Productivity Series and Hours In Levels
Circles VAR with Constructed Productivity Series and Hours In Levels, with Short Run Identification Assumption
'Stars's VAR with Labor Productivity and Hours in Differences with Long Run Identification Assumption
Figure 4 consists of six graphs presented in a threebytwo matrix.
In each of these six graphs, the xaxis ranges from zero to 20 in increments of five.
Each of the six graphs reports the response of three different VARs.
The first response was from the VAR with constructed productivity series and hours in levels.
A gray area surrounds this line corresponding to the 95 percent bootstrapped confidence interval.
The second response was constructed from the VAR with constructed productivity series and hours in levels, with shortrun identification assumption.
The third response was constructed from the VAR with labor productivity and hours in differences, with longrun identification assumptions.
The upper left graph shows ``Labor productivity''. The yaxis ranges from 0.2 to 1.2.
The upper right graph shows ``Hours Worked''. The yaxis ranges from 0.05 to 0.25.
The middle left graph shows ``Real Interest Rate''.
The middle right graph shows ``Output''.
The lower left graph shows ``Consumption''.
The lower right graph shows ``Investment''.
Figure 5 is titled ``Shape of the Response of Hours''.
The xaxis is labeled ``Response on Impact'' and ranges from 0.25 to 0.25; a vertical line ranges through zero.
The yaxis is labeled ``Response Six Quarters later'' and ranges from 0.25 to 0.25; a horizontal line ranges through zero.
A diagonal line runs through the points where y equals x.
Thus, there are four quadrants. In the upper right quadrant and the lower left quadrant, a diagonal line runs from the bottom left point to the upper right point.
The upper left quadrant contains the label ``18 percent''.
The top left half of the upper right quadrant contains the label ``71 percent''.
The bottom right half of the upper right quadrant contains the label ``4%''.
The top left half of the bottom left quadrant contains the label ``1 percent''.
The bottom right half of the bottom left quadrant contains the label ``3%''.
The bottom right quadrant contains the label ``3%''.
The graph's points are densely located in an oval region where the xvalue is between 0.02 to 0.08, and the yvalue is between 0.02 and 0.17. Points outside this region do exist, but they are less densely packed.
Figure 6 is titled ``Confidence Intervals for a0, Impact Response on Hours, and the Response Six Periods Later''. It consists of four graphs presented in a twobytwo matrix.
The upper left graph titled ``AndersonRubin Confidence Set for a0''.
The xaxis is labeled ``Coefficient Estimate'' and ranges from 35 to 25.
The yaxis is labeled ``AR Statistic'' and ranges from 0 to 10.
A dashed line exists where the AR statistic is about 3.9.
The line begins at 7.5. It slowly decreases to 5 when the coefficient estimate is 10. It then quickly decreases to 0 when the coefficient estimate is 2. From there it quickly increases to 8.5 when the coefficient estimate is 12. It finally ends at 9.2 when the coefficient estimate is 25.
The upper right graph is titled ``Mapping from a0 to Gamma''.
The xaxis is labeled ``Coefficient Estimate'' and ranges form 40 to 30.
The yaxis is labeled ``Gamma'' and ranges from 0.2 to 0.15.
The line has a negative logistic shape. It begins at 0.12 and slowly decreases to 0.1 when the coefficient estimate is 5. It then quickly decrease to 0.12 when the coefficient estimate is 8. From there, it slowly decreases to 0.12 when the coefficient estimate is 22.
The lower left graph is titled ``AndersonRubin Confidence Set for Hours Response on Impact''.
The xaxis is labeled ``Hours Response on Impact'' and ranges from 0.125 to 0.125.
The yaxis is labeled ``AR Statistic'' and goes from 0 to 10.
A dashed line exists where the AR statistic is about 3.9.
The line has a parabolic shape. It begins at 9.2 when the response on hours is about 0.125. It decreases to 0 when the hours response on impact is 0.04. Afterwards, it increases to 7.5 when hours response on impact is 0.125.
The lower right graph is titled ``AndersonRubin Confidence Set for Hours Response Six Periods Later''.
The xaxis is labeled ``Response Six Periods after Impact'' and goes from 0.15 to 0.25.
The yaxis is labeled ``AR Statistic'' and goes from 0 to 10.
A dashed line exists where the AR statistic is about 3.9.
The line begins at 9.2. It decreases to 0 when the xvalue is 0.2. It then increases to 3.9 when the xvalue is 0.25. Afterwards, the line continues to increase and arch leftwards. It ends at 8 when the xvalue is 0.22.
Figure 7 is titled ``AndersonRubin Confidence Set.
The xaxis is labeled ``Response on Impact'' and ranges from 0.8 to 0.10.
The yaxis is labeled ``response six Quarters Later. It ranges from 0.04 to 0.14.
The chart contains an ovalshaped figure whose ends extend from the lowerleft region of the graph to the upper right region. These lines of this ovallike figure indicated all possible combinations of gamma sub zero with H superscript and gamma sub six with H superscript.
The area of this oval where the xvalues are between 0 and 0.06, and where the yvalues are between 0.06 and 0.14 are shaded in a gray color.
The gray area indicates those combinations with an AndersonRubin statistics less than the 95 percent critical value.
Thick Line VAR and Grey Area: Baseline Results replicated from Figure 4
'X's VAR with exante Real Interest Rate, full sample 19732001.
Squares VAR with exante Real Interest Rate, short sample 19832001.Figure 8 is titled ``Robustness of Impulse Responses to Different Subsamples''.
It consists of six graphs presented in a threebytwo matrix.
In each of these graphs, the xaxis is labeled from 0 to 15 in increments of 5.
The graphs each contain the results from three responses.
The first line represents the baseline results replicated from figure 4
The second line is from the VAR with ex ante Real Interest Rate, full sample 1973 to 2001.
The third line is from the VAR with ex ante Real Interest Rate, short sample 1983 to 2001.
The upper left graph is titled ``Labor Productivity''.
The upper right graph is titled ``Hours Worked''.
The middle left graph is titled ``Real Interest Rate''.
The middle right graph is titled ``Output''.
The lower left graph is titled ``Consumption''.
The lower right graph is titled ``Investment''.
Notes
Thick Black Line: Point Estimate of Response of Hours on Impact
Grey Area: 95 percent weakinstrument confidence interval as reported in Table 4.
Figure 9 is titled ``How Hours Responds on Impact''.
It consists of three contour plots presented in a twobytwo matrix with the lower right graph deleted.
The upper left graph is titled ``InvestmentdividedbyCapital Adjustment''.
The xaxis is labeled ``Investment Adjustment Costs, gamma'', and ranges from 0 to 100 in increments of 20.
The yaxis is labeled ``Habit persistence b'', and ranges from 0 to 0.9 in increments of 0.1.
The plot shows the combinations of investment adjustment costs and habit persistence which produce no change in the response of hours. A line runs from investment adjustment costs of 0 and a habit persistence of 0.9 to the point where investment adjustment costs are about 5 and habit persistence is 0. A confidence region surrounds this line.
The plot also shows that as investment adjustment costs increase and habit persistence increase, there is a more negative effect on how hours respond.
The upper right graph is titled ``Change in Investment Adjustment''.
The xaxis is labeled ``Investment Adjustment Costs, gamma'', and ranges from 0 to 1 in increments of 0.2.
The yaxis is labeled ``Habit Persistence b'' and ranges from 0 to 0.9.
The plot shows the combinations of investment adjustment costs and habit persistence which produce no change in the response of hours. A line runs from investment adjustment costs of 0 and a habit persistence of 0.9 to the point where the investment adjustment costs are 0.2 and habit persistence is 0. A confidence interval surrounds this line.
The plot also shows that as investment adjustment costs increase and habit persistence increase, there is a more negative effect on how hours respond.
The lower left graph is titled ``Time to Build''.
The xaxis is labeled ``fire period timetobuild weight'', and ranges from 0 to 0.4 in increments of 0.1.
The yaxis is labeled ``Habit persistence b'', and ranges from 0 to 0.9 in increments of 0.01.
The plot shows the combinations of first period timetobuild weight and habit persistence which produce no change in the response of hours. A line runs from first period timetobuild weight of 0.15 and habit persistence of 0 to the point where the first period timetobuild weight of 0.25 and habit persistence of 0.9. A confidence interval surrounds this line.
The plot also shows that as habit persistence increases and first period timetobuild weight decreases, there is a more negative effect on how hours respond.
Figure 10 is titled ``The Tradeoff between Investment Adjustment Costs and Habit Persistence''.
It consists of two contour graphs, sidebyside.
The left graph is titled ``Consumption''.
The xaxis is labeled ``Investment Adjustment Costs, gamma''. It ranges from 0 to 1 in increments of 0.2.
The yaxis is labeled ``Habit Persistence, rho''. It ranges from 0 to 0.9 in increments of 0.1.
As habit persistence increases, there is a smaller effect on consumption.
The right graph is titled ``Investment Response''.
The xaxis is labeled ``Investment Adjustment Costs, gamma''. It ranges from 0 to 1 in increments of 0.2.
The yaxis is labeled ``Habit Persistence, rho''. It ranges from 0 to 0.9 in increments of 0.1.
As investment adjustment costs decrease and habit persistence increase, there is a larger effect on the Investment Response.
Notes on first row of figures in Figure 11:
Thick Solid Line: Empirical Response, Thin Line, Standard flexibleprice model
Squares: Model with habit
persistence and investment adjustment Costs
'*'s Model with investment
adjustment Costs
Notes on second row of figures in Figure 11:
Thick Solid Line: Empirical Response, Stars: Model with habit persistence and Time to PlanFigure 11 is titled ``Dynamic Responses Comparing Models and Data Investment Adjustment Costs as Function of Investment Growth''. It consists of six graphs presented in a twobythree matrix.
The two graphs on the left are titled ``Consumption''.
The two middle graphs are titled ``Investment''.
The two graphs on the right are titled ``Hours Worked''.
The xaxis of each of the three graphs ranges from zero to ten in increments of five.
In the top three graphs, each contains four lines.
The first line is the empirical response.
The second line is the response from the standard flexibleprice model with b and phi are zero.
The third line is the response from the model with habit persistence and investment adjustment costs where b equals 0.4 and phi equals 0.09.
The fourth line is the response from the model with investment adjustment costs where b equals 0 and phi equals 0.09.
Recall that the upper left graph is labeled ``Consumption''.
Recall that the upper middle graph is labeled ``Investment''.
Recall that the upper right graph is labeled ``Hours Worked''.
In the bottom three graphs, each graph contains four lines.
The first line is the empirical response.
The second line is the model with habit persistence and time to plan where b equals 0.5 and phi equals 0.05.
The third line is the response from the model with habit persistence and time to build where b equals 0.5 and phi equals 0.20.
The fourth line is the response from the CIK model with habit persistence where b equals 0.28 and phi equals 5.
Recall that the lower left graphs is labeled ``Consumption''.
Recall that the lower middle graph is labeled ``Investment''.
Recall that the lower right graph is labeled ``Hours Worked''.
Thick Solid Line: Empirical Response, Thin Line, Standard flexibleprice model
Squares: Model with habit persistence and investment adjustment Costs
Stars: Model with habit persistence and Time to Plan
Diamonds: Model with consumption adjustment costs and investment adjustment costs.
Figure 12 is titled ``Real Interest Rate Response''.
The xaxis ranges from zero to ten in increments of two.
The yaxis ranges from 2.5 to 1.5 in increment of 0.5.
The graph contains five lines.
The first line shows the empirical response.
The second line shows the response of the standard flexibleprice model when b and phi equals 0.
The third line shows the response of the model with habit persistence and investment adjustment costs when b equals 0.4 and phi equals 0.09.
The fourth line shows the response from the model with habit persistence and Time to Plan when b equals 0.5 and phi equals 0.05.
The fifth line shows the response of the model with consumption adjustment costs and investment adjustment costs when b equals 0.4 and phi equals 0.2.
Thick Solid Line: Empirical Response, Thin Line, Standard flexibleprice model with
Squares: Model with habit persistence and investment adjustment Costs with
Figure 13 is titled ``Model Responses with Correlated Shocks''.
It presents four graphs, presented in a twobytwo matrix.
Each graph contains three lines.
The first line is the empirical response.
The second line is the response from the standard flexibleprice model when b and phi equal zero and rho sub z equals 0.7.
The third line is the response of the model with habit persistence and investment adjustment costs where b equals 0.4, phi equals 0.09, and rho sub z equals 0.7.
The upper left graph is titled ``Consumption''.
The xaxis ranges from zero to ten in increments of five.
The yaxis ranges from 0 to 1.2.
The upper right graph is titled ``Investment''.
The xaxis ranges from zero to ten in increments of five.
The yaxis ranges from 1 to 3.
The lower left graph is titled ``Hours Worked''.
The xaxis ranges from zero to twelve in increments of five.
The yaxis ranges from 0.2 to 0.3.
The lower right graph is titled ``Real Interest Rate''.
The xaxis ranges from 0 to 10 in increments of 10.
The yaxis ranges from 0 to 150 in increments of 50.
Thick Lines, Model Responses, Plain Standard RBC, Diamonds CII Adjustment Cost, Squares,
CII Adjustment and Habit
Thin Lines, Average Estimated Response, Plain Standard RBC, Diamonds CII Adjustment Cost,
Squares, CII Adjustment and Habit
Dashed Lines, 90 Percent of all estimated responses for the CII Model are contained within dashed lines
Figure 14 is titled ``Hours Response to a Technology Shock''.
The xaxis ranges from 0 to 10 in increments of 2.
The yaxis ranges from 0.5 to 1 in increments of 0.5.
It graph contains six lines.
The three lines correspond to model responses.
The first line is the response from the standard RBC.
The second line is the response from the CII Adjustment Cost.
The third line is CII Adjustment and Habit.
The last three lines correspond to the average estimate response.
The fourth line is the response from the standard RBC.
The fifth line is the response from the CII Adjustment Cost.
The sixth line is from the CII Adjustment and Habit.
The seventh and eight lines show the 90 percent interval of all estimated responses for the CII Model.
No Adj Costs: Standard flexibleprice model with no investment adjustment costs or habit persistence.
Inv Adj Costs: flexibleprice model with CIItype investment adjustment costs (
Plus Habit
flexibleprice model with CIItype investment adjustment costs (
and with habit persistence coefficient
Figure 15 is titled ``Simulations Results''.
It contains two bar graphs presented in a twobyone matrix.
In both graphs, the xaxis is labeled ``Periods after Shock'' and ranges from zero to nine, and
The yaxis is labeled ``Percent'' and ranges from 0 to 100 in increments of 20.
At each time period, there exist three bars corresponding to three different scenarios.
The first bar is labeled ``No adjustment costs'' and comes from the standard flexibleprice model with no investment adjustment costs or habit persistence.
The second bar is labeled ``Investment Adjustment Costs'' and comes from the flexibleprice model with CIItype investment adjustment costs (gamma equals 0.01).
The third bar is labeled ``Plus Habit'' and comes from the flexibleprice model with CII type investment adjustment costs (gamma equals 0.01) and with habit persistence coefficient (b equals 0.6).
The top bar graph is titled ``Responses that are Positive''.
For the bars corresponding to ``No adjustment costs'', the value remains at roughly 83 percent for all the periods.
For the bars corresponding to ``investment adjustment costs'', the bars begin at 78 percent and increases to 83 percent by period 3. It remains there for the remaining periods.
For the bars corresponding to ``Plus Habit'', the bars begin at about 67 percent and increase to 74 percent by period 2. It remains there for the remaining periods.
The bottom bar graph is titled ``Responses that are larger than Impact Response''.
For the bars corresponding to ``No adjustment costs'', the bars begin at 40 percent. They gradually decrease to about 25 by period 7 and stay there for the remaining periods.
For the bars corresponding to ``investment adjustment costs'', the bars begin at 90 percent. They decrease almost linearly until period 9, where the value is about 60.
For the bars corresponding to ``plus habit'', the bars begin at 90 percent. They decrease almost linearly until period 9, where the value is about 67.
1. Board of Governors of the Federal Reserve System. (Email robert.j.vigfusson@frb.gov) Return to text
2. Thanks to Martin Eichenbaum, Larry Christiano, Tim Conley, and John Fernald for helpful comments in early drafts of this paper. Additional thanks for comments from Rochelle Edge, Chris Erceg, Chris Gust, William McCausland, and seminar participants at the Bank of Canada, the Bank of England, Northwestern University, the University of Alberta, UQAM, the Université de Montréal, and the Board of Governors of the Federal Reserve. An earlier draft of this paper was circulated as The negative response to a positive technology shock, a neoclassical explanation. The views in this paper are solely the responsibility of the authors and should not be interpreted as reflecting the views of the Board of Governors of the Federal Reserve System or of any person associated with the Federal Reserve System. Return to text
3. In more formal terms, using hours to approximate for utilization implies an assumption that, at least for empirically relevant values, the output expansion path is an upward sloping line. The assumption would be satisfied if output were produced, for example, by a homothetic production function (such as CobbDouglas) using hours and utilization with constant costs of increasing either input. Return to text
4. A lack of data requires the use of this manufacturingbased technology measure to approximate the economywide fluctuations in technology. Return to text
5. This equation is similar to that estimated in Burnside Eichenbaum and Rebelo (1996) and in Conley and Dupor (2003) except for two differences. First, these authors don't use the average hours correction, represented in Equation 2 by . Second, the value of is estimated as the sum of the shares of capital services and labor which are estimated separately. In other words, they don't use the information on KLEMS shares but rather just estimate the shares. Impulse responses generated using this approach were similar to the ones reported in the paper. Return to text
6. Some might be concerned that electricity usage might be overly sensitive to weather fluctuations. Because the electricity usage data is on a national basis, geographic diversification should limit any dependence on weather. Return to text
7. The KLEMS dataset is used to measure multifactor productivity as the annual frequency. It and the Canadian equivalent are featured in Vigfusson (2003). Return to text
8. This aggregation equation calculates the increase in aggregate technology resulting from industrylevel technology growth holding the distribution of inputs among industries fixed. Return to text
9. The results reported here use the IMF world price of oil, an average of international wellhead prices. Other researchers including Basu, Fernald, and Shapiro (2001) have used the Producer Price Index for crude oil as an instrument. The Producer Price Index, however, only measures the price paid for domestic but not foreign oil. Not including foreign oil results in a very misleading price series because, between the summer of 1973 and January 1981, American produced oil was under governmentimposed price controls. Because of these price controls, the PPI does not measure the true cost of oil. In particular, it does not capture the dramatic effect of the oil shocks of 1973 and 1979 that are observable in other series such as the IMF oil series or the Department of Energy's series, Refiner Acquisition Nominal Cost of Imported Crude Oil. Return to text
10. To specify these variables as instruments requires an assumption that these variables are uncorrelated with the technology shocks. There are models in which this assumption would be violated. For example, if implementation cycles models (Shleifer 1986) were empirically relevant, then these variables may be correlated. In such a model, reductions in oil prices could result in an economic expansion that would cause people to implement their ideas increasing productivity. Return to text
11. The aggregation, however, does obscure the important role played by industryspecific productivity growth that is found in US industrial production data in Conley and Dupor (2003) and again in KLEMS data for both Canada and the United States in Vigfusson (2003). Return to text
12. Several recent papers have identified a technology shock using a longrun restriction. These papers include Gali (1999); Francis and Ramey (2003); Altig, Christiano, Eichenbaum, and Linde (2004); and Fisher (2002). Although these methods have proven popular, some researches have critiqued these methods. In particular, Faust and Leeper (1997) describe problems associated with constructing confidence intervals (see footnote 15 for more discussion), and the perils of excluding variables from the VAR. Erceg, Guerrieri,and Gust (2004) present modelbased Monte Carlo evidence on the longrun identification assumption. Their work will be discussed in section 4. Return to text
13. The equivalency can be seen by examination of the cofactor formula for the inverse of a matrix. Return to text
14. This specification is the `` double difference'' approach described in King and Watson (1997). Return to text
15. If the VAR is correctly specified, lagged 's cannot be used as instruments for For a VAR with lags, the first values of will also be in the equation. Hence, they cannot serve as instruments. Other values are not valid instruments by the definition of the VAR being correctly specified. Being correctly specified implies that these values cannot contain any information about beyond that contained in . Return to text
16. Because there are no other restriction on the equation, the results calculated using the reduced form VAR are identical to results calculated using the structural VAR with coefficients and . Imposing additional restrictions on such as also identifying a monetary policy shock, as in Altig, Christiano, Eichenbaum, and Linde (2003), would require using the structural VAR to calculate impulse responses. Return to text
17. Faust and Leeper (1997) note that the standard confidence intervals for impulse responses are not valid unless restrictive assumptions are made concerning the data generating process. For the bootstrapped confidence intervals considere here, the implicit assumption is the true data generating process actually is a six lag bivariate vector autoregression. Under such a restrictive assumption, the confidence intervals reported here would be valid. Return to text
18. All variables are expressed in logs. The variables span the second quarter of 1972 to the end of 2001 and are taken from the DRI BASIC Economics (nee Citibase) database. The series with their mnemonics are as follows: real consumption (the sum of consumption of services GCS, nondurables GCN and government consumption divided by the gross output price deflator GDPD and consumption of durables GCD), real investment (gross private investment (GPI) and government investment divided by the output price deflator GDPD), output (nominal consumption and investment divided by GDPD), real interest rate (3 month Treasury Bill rate minus the growth rate of the GDP price deflator), and hours worked (nonfarm business hours LBMNU). All quantities are expressed in percapita term by dividing by the population over 16. Deflating consumption and investment by the same price measure rather than using the separate published deflators for investment and consumption is deliberate. See Whelan (2000) for a discussion of using chainweighted data. Return to text
19. The labor productivity output series considers total output divided by business hours. As such, this definition of labor productivity is more like the work by Gali (1999) than the definition used in Francis and Ramey and Christiano, Eichenbaum, and Vigfusson (2003). As in CEV, I could have defined labor productivity as the ratio of business output to business hours worked. This has the advantage of avoiding the problem of increases in government spending being misconstrued as a technology shock. However, it comes at the cost of being less transparent concerning the imposed cointegration results. Return to text
20. Complementary to this work is a paper by Pesavento and Rossi (2003) where they calculate confidence intervals for impulse responses at long horizons. Return to text
21. Addition detail is provided in Christiano, Eichenbaum and Vigfusson for the weak instruments problem that occurs when hours has a unit root. Return to text
22. The proof of this claim is straightforward. By definition the part of that does not depend on lagged is Given the previously described identity,
the response by to a onestandard deviation shock to is . But because and are uncorrelated, the variance of is the following23. The following table is based on 20,000 samples over the parameter space. These results were then checked by making an additional 30,000 simulations. In no case did these additional samples change the results reported here. Return to text
24. Models with capital adjustment costs do have a similarity with models of marginal efficiency shocks (Dejong Ingram and Whiteman 2000). In both models, the amount of output required to produce a unit of capital (the marginal efficiency of investment) varies over time. The models differ in how they determine the marginal efficiency. In models with capital adjustment costs, the marginal efficiency varies endogenously as agents vary how much they invest. In Dejong Ingram and Whiteman (2000), the marginal efficiency is an exogenous random variable. Return to text
25. Additional support for timetobuild models comes from Kovea (2000), which presents firmlevel evidence of timetobuild being a feature of investment in structures. Based on reports from newspapers and trade journals on 106 randomly chosen firms, she estimates an average timetobuild for structures of about two years. Furthermore, she reports that very few of the projects that she examines were cancelled. Return to text
26. The specification of habit persistence used here is standard in the literature. One drawback of this specification is that it requires that the level of consumption must always be greater than the habit stock to avoid marginal utility being infinite. Carroll, Overland, and Weil (2000) discuss a different specification, where what matters is the ratio of current consumption to the lagged habit stock. Using the ratio avoids the problems of infinite marginal utility as long as consumption is positive. Return to text
27. For the CES production function, technology is labor augmenting. This assumption is important for the normalization required when technology is a random walk. Return to text
28. Microeconomic evidence suggests that the labor supply elasticities are much smaller. The labor literature reports labor supply elasticities near zero for males and about 1 for females. Most macroeconomic models, however, require these larger estimates in order to match the data. Return to text
29. The emphasis in this paper is on convex adjustment costs. Cooper and Haltiwanger (2000) emphasize that including firmlevel nonconvex costs (such as fixed costs) also matters for aggregate investment. Including such nonconvexities is a goal for future work. Return to text
30. The difference in the two specifications can be seen by examining the cost of the investment adjustments costs when investment is times greater than the steady state value. For the investment to capital ratio case, the adjustment costs would equal For the investment growth rate case, the adjustment costs would equal . Hence, the values of must be different for the two specifications. Return to text
31. The value of is 1.5.The following coefficients were taken from Christiano and Eichenbaum (1992). The vector equals . Return to text
32. Because the model is assumed to have a unit root in the technology, in calculating a solution, the variables will be normalized by dividing through by the level of technology. Hence the model will be solved for { where equals . Return to text
33. As is evident in Christiano and Vigfusson (2003), this jaggedness is not a problem when examining the spectrum implied by the timetoplan model. Return to text
34. With external habit, the real interest rate is a somewhat less tractable
35. With serially correlated shocks,the standard flexible price model has a delayed response to the onset of a technology shock. The response on impact however seem to be too negative with both investment and hours worked falling sharply. Return to text
36. Appendix A describes briefly the estimation procedure. Return to text
37. These values were calculated for the version of the flexible price model without any adjustment costs. As such, a richer structure might result in different parameter estimates. However, the goal of this section is to provide some evidence on the usefulness of long run VARs to discriminate between models, not a full maximum likelihood estimation. Return to text
38. In the above results, I did not consider the effect of using the growth rate of hours in the longrun VAR. In these models, hours worked is stationary. As is discussed in CEV (2003), estimating a VAR after first differencing a stationary variable is a form of specification error. Hence using the first difference VAR on modelsimulated data would be using a misspecficied VAR. As was found in CEV and CKM, I also found that applying the first difference VARs would result in a very high probability of a negative response by hours on impact. Return to text
39. Let for integer values of Then,
This version is optimized for use by screen readers. A printable pdf version is available.