The Analysis of Structural Breaks in Gasoline Demand Using a Bayesian Approach Koffi Akakpo a, Philippe Barla b and Stephen, Gordon c

Size: px
Start display at page:

Download "The Analysis of Structural Breaks in Gasoline Demand Using a Bayesian Approach Koffi Akakpo a, Philippe Barla b and Stephen, Gordon c"

Transcription

1 April 2017 The Analysis of Structural Breaks in Gasoline Demand Using a Bayesian Approach Koffi Akakpo a, Philippe Barla b and Stephen, Gordon c a Département d économique and CREATE, Université Laval, Quebec City, Canada b Département d économique and CREATE, Université Laval, Quebec City, Canada c Département d économique and CRREP, Université Laval, Quebec City, Canada Abstract: This paper focuses on the presence of structural breaks in the Canadian gasoline demand using a Bayesian approach. The analysis is carried on quarterly data from 1965 to Our methodology is based on Chib (1998) approach that allows for multiple and unknown breakpoints. The results suggest the existence of two breakpoints defining three regimes. In the first regime up to 1982, the price and income elasticities are in the range of values usually reported in the literature (-0.5 and 0.8 respectively). In the third regime after 1986, price reactiveness is close to zero while the income effect is much lower (0.2). The second regime provides results that are incoherent with economic theory and most likely corresponds to a transition phase disturbed by several shocks such as the oil crisis, a recession and fuel economy regulations. Keywords: Gasoline demand, price elasticity, income elasticity, structural breaks, Bayesian approach JEL-code: C11, Q41, Q Introduction Gasoline demand has been widely analyzed with hundreds of studies estimating price and income elasticities (Dahl and Sterner, 1991; Espey, 1998; Basso and Oum, 2007; Dahl, 2012). Based on a review of the literature, Basso and Oum (2007) conclude that the short run price elasticity of gasoline demand would be between -0.3 and -0.2 and the long run price elasticity between -0.8 and For income elasticity, the values would vary between 0.3 and 0.5 in the short run and between 0.9 and 1.3 in the long run. However, recent works suggest that the values of these parameters may have changed in recent decades. Hughes, Knittel and Sperling (2007) find a substantial decline in the short run price elasticities in the U.S. when comparing the 2001 to 2006 with the 1975 to 1980 periods. Small and Van Dender (2007) find a significant decline in the U.S. price and income elasticity over the 1966 to 2001 period. Park and Zhao (2010) estimate a smooth time-varying cointegration model using U.S. monthly data from 1976 to 2008 and conclude that price and income elasticities would have followed a cycle of increase-decrease ranging from to and 0.06 to 0.10, respectively. The authors link these changes to variation in the share of gasoline expenses in disposable income. Liu (2014) uses semi-parametric methods on a panel of U.S. states from 1994 to Contrary to the other studies, she finds that price and income elasticity would have slightly increased over this time period (from 0 to -0.1 and 0.12 to 0.2 respectively). Liu explains this time pattern by the dramatic fluctuations of gasoline prices as well as income growth. She also finds spatial variations in the elasticity values across states depending upon urbanization and income level. Overall, there are some evidence of changes in price and reactiveness of gasoline demand over time. However, none of the existing studies proceeds to formal structural break analysis. In this paper, we

2 address this gap by testing for the existence of structural breaks in the Canadian demand for gasoline from 1965 to This analysis is particularly relevant in the current context of the fight against climate changes. The transportation sector is the second most important source of GHGs in Canada with a share of 19% just behind the production of gas and petroleum. In the transportation sector, gasoline account for over 68% of the sector total emissions (WWF-Canada, 2012). Policies have been adopted by Canadian authorities in order to curb road transportation emissions. However, the effectiveness of several interventions such as carbon tax or cap-and-trade is in fact directly related to the values of the price and income elasticities. Moreover, the assessment of measures such as the strengthening of new light duty fuel economy standards requires predicting how gasoline consumption would have evolved in the business as usual scenario. Though, the accuracy of this scenario depends very much on using appropriate price and income elasticities. The methods for detecting structural breaks have considerately evolved in recent years both under the classical and Bayesian paradigm. While the classical approach may be somewhat easier to interpret as it relies on classical regression techniques, the Bayesian approach offers several important advantages (Western and Kleykamp, 2004 and Spirling, 2007). Firstly, the Bayesian inference does not rely on asymptotic theory like the frequentist methods. This is particularly advantageous with structural breaks limiting the sample size. Secondly, the Bayesian approach allows for taking into account priors on the parameters as well as on the number and timing of the breaks. Thirdly, the Bayesian method does not impose a precise date on the breaks but rather estimate probability distributions for each break at each point in time. Specifically, we adopt the Bayesian approach initially proposed by Chib (1998) for modeling multiple unknown structural break problems. In this setting, the determination of the number of breaks becomes an exercise of model selection based on the Bayes Factor criteria. As already mentioned, the main particularity of this method is that it estimates, for each observation, a set of probabilities of being in each regime thereby allowing for drawing inferences on the dates of the breaks. For comparison purposes, we also report the results obtained using Bai and Perron (2003) methodology in the frequentist context. Our main results are based on a static specification where the per capita gasoline consumption depends upon gasoline price, disposable income and quarterly dummies. The results show that the model without any structural breaks is very unlikely. In fact, the Bayes factor procedure identifies the model with two structural breaks and thus three regimes as the model that best describes the data. The first break occurs around 1982 and the second around During the first regime from 1965 to 1982, the expected value of the posterior distribution of the price elasticity is (standard deviation at 0.077). For the income elasticity, the expectation is (0.048). In the last regime, after 1985, the expected price elasticity is close to zero. The income elasticity is also much lower with an expected value of (0.121). The period covering the second regime between 1982 and 1986 is characterized by a major recession caused by the second oil shock, the gradual implementation of the CAFE standards in the U.S. and its equivalent in Canada. During the eighties, the fleet of vehicles drastically evolved resulting in substantial fuel economy improvements. 1 Our results are robust to specifications changes and our conclusion holds when using Bai and Perron method. The rest of paper is organised as follows. Section 2 discusses the econometric issues to model structural breaks and presents the Bayesian approach with its particularities. Section 3 describes the data and provides some basic descriptive statistics. We outline the results and the robustness analysis in Section 4. Finally, Section 5 discusses the main conclusions and policy implications. 1 For the US, the fuel economy rose by up to 25% during this period (Bureau of Transportation Statistics, 2017). While comparable statistics are unavailable for Canada, the same phenomenon occurred in Canada (Beauregard- Tellier, 2004). 2

3 2. Methodology Most empirical studies of gasoline demand adopt a log-log linear functional of the following form: lng t = β 0 + β 1 lnp t + β 2 lny t + β 3 X t + ε t [1] where G t is gasoline consumption per capita at time t, P t is the retail price of gasoline per liter, Y t is the real disposable income per capita, X t represents others potential covariates and ε t is the error term. β 1 and β 2 measure respectively the price and income elasticities. Specification [1] does however assume that the parameters are constant over time. Thus, we use instead specification [2] which allows for an undetermined m number of shifts or breakpoints in the parameters values. with: r = 1,2,, m + 1 τ r 1 < t τ r and ε t ~N(0, σ k 2 ) lng t = β 0,r + β 1,r lnp t + β 2,r lny t + β 3,r X t + ε t [2] The m structural breakpoints define m+1 regimes. The number of breakpoints m is assumed to be unknown and thus need to be determined. The timing of the breaks is characterized by the parameters τ r that also need to be estimated. 2 For example, if the data revealed one breakpoint this means two regimes with different gasoline demand parameters. The number and timing of structural breaks is treated as a problem of model selection based on some information criterion. The optimal number of breaks may be determined either through a sequential (Yao, 1988; Liu et al., 1997; Giron et al., 2007; Lai and Xing, 2011) or simultaneous process (Chib, 1998; Bai and Perron, 1998). In this paper, we adopt the latter which has been shown to be more consistent (Bai et Perron, 1998). The estimation can be carried either within the classical (frequentist) or Bayesian framework. We adopt the latter as it does not rely on asymptotic theory and does not face the infinite samples problem (Chib, 1998; Maheu and Gordon, 2008). These issues are particularly worrisome with structural breaks that limit the size of the estimation samples (Raftery, 1994). In the Bayesian framework, the estimation is based on the maximization of the likelihood function which depends upon the way breaks are modeled. Some authors impose a constant probability of break (Chernoff and Zacks, 1964) while other allows the probability to vary by regime (Chib, 1998). But the estimation of these models may be burdensome with a long time series and a high number of breaks. Chib (1998) proposes a more straightforward parametrization based on a Markov process. A discrete random variable s t is introduced which captures the state of the system at each period t. This latent variable, which takes values from 1 to m+1, indicates from which regime an observation has been drawn. It is assumed to follow a Markov process with the transition matrix: P = p 11 p p 22 p p mm p m,m+1 ( ) where p ij = Pr(s t = j s t-1 =i) is the probability of moving from state i to state j at time t given that the state at t-1 is i. 2 Note however that, by definition, τ 0 = 0 and τ m+1 = T 3

4 The structure of this matrix is such that there is only one unknown parameter per row (the sum of the probability in each row being one). Moreover, the transition is only possible to the next state. This excludes jump across different regime as may occur in a Markov switching model. This setting corresponds to what is referred as a Hidden Markov Model (Chib, 1996). The model selection (i.e. the optimal number of breaks) is based on the Bayes factors method which compares the marginal likelihoods across models. Chib (1995) has developed a simple computation technique for the estimation of the marginal likelihood. 3 For a model with m breaks (M m ), the marginal likelihood is calculated as follows: L(lnG 1:T Z 1:T, M m ) = f (lng 1:T Z 1:T, β, P, M m )π(β, P M m ) π(β, P lng 1:T, Z 1:T, M m ) where Z 1:T contain all the explanatory variables, (β, P) is any point in the parameter space, f (. ) is the likelihood, π(β, P M m ) is a prior density on the parameters and π(β, P lng 1:T, Z 1:T, M m ) is a posterior density. The Bayes factor is then defined as the ratio between likelihood of two models M m and M n : B m,n = L(lnG 1:T Z 1:T, M m ) L(lnG 1:T Z 1:T, M n ) The choice of model is then determined by a set of rules proposed by Jeffreys (1961). Essentially, a value of B m,n larger than one means that model M m better fits the data than M n. Moreover, the larger is this ratio the stronger is the conclusion. The evidence is considered decisive if log 10 (B m,n ) is greater than 2. The log-likelihood function is Now given Chib (1998) transition matrix, we have: Ln f(lng 1:T Z 1:T, M m ) = Ln f(lng t Z t, β, P) m Ln f(lng t Z t, β, P) = Ln f(lng t Z t, β, P, s t = k) Pr(s t = k lng 1:t 1, Z 1:t 1, β, P) k=1 where f(lng t Z t, β, P, s t = k) is the density of observation a time t conditional to the state s t = k, Pr(s t = k lng 1:t 1, Z 1:t 1, β, P) is the marginal probability to be in regime k which is defined through Bayes rule : Pr(s t = k lng 1:t 1, Z 1:t 1, β, P) = k T t=1 Pr(s t = k s t 1 = i)pr(s t 1 = i lng 1:t 1, Z 1:t 1, β, P) i=k 1 The estimation is carried by simulation of the posterior distribution of the parameters through an iterative process. The likelihood function is computed using the Expectation-Maximization (EM) algorithm. This iterative estimation technique is particularly efficient when the objective function contains latent variables. In fact, the EM algorithm generalizes the maximum likelihood principle. It takes place in two main stages: 1. Expectation which consists in simulating the latent variables given the priors; 2. Given the value of the latent variables, the likelihood is maximized with respect to the parameters. The estimated parameters are then used to update the value of the latent variables. 3 An alternative is the algorithm of Dufays et al. (2014) with stepping stone techniques. 4

5 For comparison purposes, we also estimate the model using a frequentist approach. Specifically, we use Bai and Perron (2003) techniques which are suitable for both pure and partial structural change models. This empirical method is inspired by their theoretical model on the estimation of multiple structural changes by least squares. In each regime, a minimal number h of observations is imposed. The optimal number and timing of breaks as well as the parameter values are derived by minimizing the following function: m+1 τ k k=1 t=τ k LS(β, Γ) = (lng β Z) (lng β Z) = [lng t β k Z t ] 2 Bai and Perron (2003) propose a new efficient algorithm to find β the vector of parameters and Γ the vector of breaks which minimize the function LS given m. The estimation is basically based dynamic programming and the use of the Bellman principle. The constraint on the minimum duration of a regime considerably decreases the number of potential regimes to be tested. 4 The optimal number and timing of the regimes is based on the comparison of the sum of the squares of the residuals (SSR), Bayesian Information Criteria (BIC) or log-likelihood. The BIC is generally considered to be the best criteria Data and descriptive analysis The main source of our dataset is Statistics Canada. Quarterly gasoline consumption is approximated by domestic gasoline sales. They include all type of gasoline designed for internal combustion engines other than aircraft and exclude exports and inter-company sales. They likely somewhat overestimate gasoline consumption for transportation but this is the only long term time series available. The price variable is constructed using the average retail prices for unleaded gasoline available starting in 2006 with a gasoline consumer price index which is available from The average price before 2006 is thus backtracked using the price index. Quarterly disposable income statistics are only available up to 2012Q2. Our dataset covers therefore the 1965Q1 to 2012Q2 period providing 190 observations. All dollars values are expressed in real Canadian dollar of Annex 1 details the sources of the variables used in the empirical analysis. Panel A of Figure 1 shows per capita gasoline consumption and real gasoline price over the analysis period. Panel B shows per capita gasoline consumption and real disposable income. From 1965 to about 1980, per capita gasoline consumption exhibits a strong positive trend with an average annual growth rate of about +2.8%. This is also a period of low gasoline price and high income growth. The second oil shock marks a drastic change in the trend of gasoline consumption: from 1980 to 1990, gasoline consumption declined by -1.9% per year on average. Obviously, this decline could be due to the large price hike as well as the deep recession that Canada experienced in the early 80 s. But this decline is also associated with the sharp improvement in the fuel economy of vehicles during this period in part because of the implementation of the U.S. Corporate Average Fuel Economy Standards (CAFE) and its equivalent in Canada the Company Average Fuel Consumption (CAFC) goals. Contrary the CAFE, the Canadian targets were not mandatory until 2011 (McCauley, 2011). However, given the deep integration of the US and Canadian automotive market, the Canadian targets have always been similar to the US standards and most manufacturers have voluntary met the targets. From 1979 to 1982, the average fuel consumption of new vehicles declined by about 25% due to technological advances but also a 20% decline in horsepower and a 15% drop in vehicle weight (BST, 2017). In fact, there was a rapid shift from large gas guzzlers to smaller fuel efficient automobiles. These events may have caused a break in gasoline demand. During that period, higher gasoline prices could have led to increase price reactiveness but improved energy efficiency could have had the opposite impact. 4 For more details see appendix 1. 5 The Bayesian results are obtained with Matlab while the classical results are derived by R. 6 A Canadian dollar was worth 0.87 U.S. dollar in

6 Panel C of Figure 1 shows the share of gasoline expenses in disposable income which may affect the price response. This share significantly increases after the second oil shocks but decline rapidly after 1985 before reaching historical lows in the nineties then trending up in the 2000s up to the financial crisis. During the last decade, gasoline prices have fluctuated with significant peaks in 2001, 2008 and The rise of disposable income has been slow in the 1990s and has picked up in the next decade up to the financial crisis. The increase in per capita gasoline during this period is often associated with the development of large, more powerful vehicles and especially the growing popularity of SUVs. Obviously, it is difficult to conclude based on these descriptive evidence if and when structural breaks have occurred in gasoline demand. We thus turn next to the econometric analysis results. Figure 1 : Gasoline consumption, disposable income and average retail price Panel A: Gasoline consumption and average retail price Panel B: Gasoline consumption and disposable Panel C: Share of gasoline expenses in relation to income 6

7 m breaks 4. Results First, we analyse the results of the most basic specification with gasoline consumption depending upon price, disposable income and quarterly dummies to account for seasonable effects. Table 1 shows the log 10 (B m,n ) namely the Bayes Factors in its log version for up to four breakpoints. In fact, beyond four breaks, the marginal likelihood function is continuously decreasing with m. The values in Table 1 allow comparing the likelihood of models with m breaks (M m ) with respect to models with n breaks (M n ). A positive value means that M m is more likely than M n while the opposite holds for negative values. Thus, it appears that M 2 fits best the data. However, Jeffreys rule indicates that the log 10 (B m,n ) should be higher than 2 in order to decisively conclude that M m is better than M n. In our setting, this means that we cannot clearly select M 2 over M 1 as log 10 (B 2,1 ) is well below two. Table 1 : Log of the Bayes Factor for comparing the basic specification model with different number of breakpoints Model n breaks Thus, we analyse the results of both models. Figure 2 and 3 plot the posterior density of regime change (Panel A) and the posterior probabilities of states (Panel B) for M 1 and M 2 respectively. Figure 2 : Posterior density and probabilities M 1 Panel A Panel B Figure 3 : Posterior density and probabilities M 2 Panel A Panel B 7

8 The posterior density of regime change for M 1 is concentrated on the period running from 1982 to 1985 with a maximum around 1984Q1. Panel B shows a transition from regime 1 to regime 2 in less than ten years. When considering two breakpoints, Figure 3 shows a first break around the last quarter of 1982 and a second break at the end of This pattern is coherent with the idea that the 1979 oil shock initiated a transition phase (i.e. regime 2) from regime 1 to regime 3. Figure 4 illustrates the posterior density of the price and income elasticities for M 1 and M 2. We also show the results for M 0 as it is the model most often estimated in the literature. Table 2 presents summary statistics on the distribution of these elasticities. We find that the average price elasticity obtained in the no break model is consistent with the findings in the literature (-0.33) while the income elasticity is somewhat on the low side (0.34). The model with one break shows an average price and income elasticity in the first regime that is very much in line the central values in the literature. In the second regime however, the average price elasticity is positive but very close to zero while the income effect is lower than in the first regime. In the model with two breaks, the results in the first regime and third regime are very close to those obtained with two breaks. In the second regime, the densities of the elasticities are very spread indicating that the price and income effect are very blurry. Recall that during this period of time several shocks affected the economy at large and the automobile market in particular. This second regime may thus be viewed as a transition phase. Figure 4 : Posterior density of price and income elasticities Panel A : No break Panel B : One break Panel C : Two breaks 8

9 Table 2 : Summary statistics on the posterior distribution of the elasticities Bayesian results - Basic Model Type of regime Statistics Price Income Intercept Q2 Q3 Q4 Mean Regime 1 St.dev Lower Upper One break Mean Regime 1 St.dev Lower Upper Mean Regime 2 St.dev Lower Upper Two breaks Mean Regime 1 St.dev Lower Upper Mean Regime 2 St.dev Lower Upper Mean Regime 3 St.dev Lower Upper Robustness analysis In order to access the robustness of our results, we examine several alternative specifications. First, we estimate a model with additional covariates namely the level of unemployment and the interest rate on Canadian treasury bills. These variables are commonly used in the literature and capture the macroeconomic conditions. The results on the elasticities (see Table 3) are very similar to those obtained without these covariates. Second, we estimate the basic model but with an interaction term between the price and income variable as an additional covariate. The idea is that the price reactiveness is decreasing with the level of income. Ignoring this possibility could eventually lead to the erroneous detection of structural breaks. Table 3 results shows that the inclusion of the interaction terms do not change drastically the value of the elasticities. Moreover, the data continues to indicate two structural breaks at around the same time than in the initial specification. 9

10 Table 3 : Summary statistics on the posterior distribution of the elasticities different specifications Model Basic Model Model with covariates 7 Price-Income 8 Interaction Model Type of regime Statistics Price Income Price Income Price Income Mean Regime 1 St.dev Lower Upper Mean Regime 2 St.dev Lower Upper Mean Regime 3 St.dev Lower Upper Third, we also estimate the basic specification with the classical approach. In this setting, we also find that the model with two breaks and three regimes fits best the data (see Table 4). The timing of the first break is very similar than in the Bayesian approach (i.e. in the early eighties). The second break is however placed later than in the Bayesian approach (1991 rather than 1985). This difference is however due to the imposition of a minimum number of observations by regime in this estimation approach. 9 For the values of the elasticities, the first regime results are very comparable to those obtained in the Bayesian setting. The values in the second regime are also inconsistent with economic theory. Once again, this regime likely reflects a transition phase. In the third regime, the price elasticity is negative and statistically significant but very small at while the average elasticity in the Bayesian setting is close to zero. The income effect is somewhat larger than in the Bayesian setting but remain smaller than in the first regime. These differences may also be explained by the constrained imposed on the sample size in the second regime that thereby also affects the third regime sample. Overall however, the results obtained in both approach lead to similar conclusions: the price and income reactiveness of gasoline demand has declined after the second oil crisis. Moreover, the transition seems to have been rapid i.e. about a decade. Table 4 : Comparison of models with Classical approach m breaks RSS BIC LogLik The covariates are rate of unemployment and interest rate of Treasury-Bill. 8 The elasticities are calculated on the average of each regime. The statistics are computed by evaluating the posterior distribution for both parameters on average point. 9 In the classical approach, we fixed the parameter h at 0.20 meaning that each regime should have at least 20% of the total number of observations in the sample. 10

11 Table 5 : Classical approach - results for basic model for m=2 m=2 {1982-Q1 ; 1991-Q3} Type of regime Price Income Intercept Q2 Q3 Q4 Regime *** *** *** *** *** *** (0.0446) (0.0221) (0.1817) (0.0101) (0.0106) (0.0101) Regime *** *** 01367*** *** *** (0.0705) (0.1360) (1.3776) (0.0145) (0.0196) (0.0165) Regime * *** *** *** 01122*** *** (0.0284) (0.0568) (0.3891) (0.0067) (0.0068) (0.0068) *** p< 0.01, ** p< 0.05 and * p< 0.1. Values in parenthesis are the standard errors of parameters. 5. Conclusion This paper proposed an investigation of Canadian gasoline demand from 1965Q1 to 2012Q2 using a Bayesian approach to capture breakpoints. Different specifications have been tested in order to check robustness. We also use a frequentist approach with multiple breakpoints. The analysis shows the present of two structural breaks one after the second oil shocks around 1982 and the other around The elasticity values before 1982 are consistent with the consensus values reported in the traditional literature. Our analysis also confirms that there has been a decline in price and income reactiveness in recent decade. We find a price effect that is essential null. Our analysis suggests that the shift occurs rapidly during a transition phase in the eighties. During this transition, the fuel economy of vehicles drastically improved because of high gasoline price and fuel economy regulations. The recent strengthening of the fuel economy standards in the US and Canada could therefore very well lead to an additional reduction in the price and income reactiveness of future gasoline demand. In future research, we would like to add dynamic in our model in order to be able to distinguished short and long term changes. It would also be valuable to apply the same approach on US data. 11

12 Appendix 1 The dynamic programming algorithm and the principle of Bellman We describe briefly the principle of Bellman used by Bai and Perron (2003). Let u(i,j) be the recursive residual at time j obtained with sample starting at date i ; SSR (i, j) as the sum of the squares of the residues between dates i and j with i <j. The recursive relation between residuals is SSR (i, j) = SSR (i, j-1) + u(i,j) 2. Let the function SSR(T r,n ) be the sum of the squares of the residuals for an optimal number of structural breaks on the first n observations. The recursive problem is defined as follows: SSR(T m,t ) = Min mh j T h { SSR(T m 1,j ) + SSR(j + 1, T)} Where T is size of data, m is optimal number of breaks and h is a parameter for limiting a minimum size of each regime. Annex 1 Table 6 : Source of data, Statistics Canada Name Table Frequency footnotes Gasoline consumption Monthly data from 1965:1 to 2012:6 Disposable income Quarterly data from 1965:1-2012:2 Average retail prices for Monthly data from 2006:1 unleaded gasoline to 2012:6 Treasury bills: 3 months Monthly data from 1965:1 to 2012:6 Government bonds Monthly data from 1965:1 to 2012:6 Population Quarterly data from 1965:1 to 2012:2 Consumer Price Index Monthly data from 1965:1 to 2012:6 Gasoline Consumer Index Monthly data from 1965:1 price to 2012:6 Sales by reporting companies, exclusive of exports and sales to other reporting companies, and adjusted for exports and imports by non-reporting companies. All gasoline type fuels for internal combustion engines other than aircraft. (Harmonized System code ,.14 and.15) Unadjusted disposable income for Canada from sectors accounts, persons and unincorporated business. For regular unleaded self-serve gasoline, average city prices are weighted by provincial volume supplied and cities population to calculate the Canada average retail price. Treasury bills at the last Wednesday unless otherwise stated, Bank of Canada. Government of Canada marketable bonds, average yield, over 10 years. Estimates of population, Canada, provinces and territories. Consumer Price Index for Canada, base(2002=100) Gasoline consumer price index for Canada, base(2002=100) References Bai J and Perron P Estimating and testing linear models with multiple structural changes. Econometrica, Vol. 66, pp Bai J and Perron P Computation and Analysis of Multiple Structural Change Models. Journal of Applied Econometrics, Vol. 18 (1), pp Basso LJ and Oum TH Automobile Fuel Demand: A Critical Assessment of Empirical Methodologies. Transport Reviews, Vol. 27(4), pp Chernoff H and Zacks S Estimating the current mean of a Normal distribution which is subject to changes in time. Annals of Mathematical Statistics, Vol. 35, pp Chib Siddhartha Marginal likelihood from the Gibbs output. Journal of the American Statistical Association, Vol. 90, pp

13 Chib Siddhartha Calculating posterior distributions and modal estimates in Markov mixture models. Journal of Econometrics, Vol. 75, pp Chib Siddhartha Estimation and comparison of multiple change-point models. Journal of Econometrics, Vol. 86 (2), pp Dahl C and Sterner T Analysing gasoline demand elasticities. Energy Economics, Vol. 13(3), pp Dahl Carol A Measuring Global Gasoline and Diesel Price and Income Elasticities. Energy Policy, Vol. 41, pp Espey Molly Gasoline demand revisited: an international meta-analysis of elasticities. Energy Economics, Vol. 20, pp Frédéric Beauregard-Tellier Fuel efficiency of motor vehicles in Canada. Library of Parliament, Parliamentary Research Branch. Jeffreys Harold Theory of Probability. Oxford University Press, Oxford. Jonathan E. Hughes, Christopher R. Knittel and Daniel Sperling Evidence of a Shift in the Short- Run Price Elasticity of Gasoline Demand. The Energy Journal, Vol. 29(1), pp Kenneth A. Small and Kurt Van Dender Fuel Efficiency and Motor Vehicle Travel: The Declining Rebound Effect. The Energy Journal, Vol. 28, No. 1, pp Liu Weiwei Modelling Gasoline Demand in the United States: A Flexible semiparametric approach. Energy Economics, Vol. 45, pp Liu J, Wu S and Zidek JV On segmented multivariate regressions. Statistica Sinica, Vol. 7, pp Maheu J.M. and Gordon S Learning, forecasting and structural breaks. Journal of Applied Econometrics, Vol. 23(5), pp McCauley S Canada s Passenger Automobile and Light Truck Greenhouse Gas Emission Regulations for Model Years Environment Canada, Informal document No. WP Raftery AE Changepoint and Change Curve Modelling in Stochastic Processes and Spatial Statistics. Journal of Applied Statistical Science, Vol. 1, pp Spirling Arthur Bayesian Approaches for Limited Dependent Variable Change Point Problems. Political Analysis, Vol. 15, No. 4 (Autumn 2007), pp Sung Y. Park and Guochang Zhao An estimation of U.S. gasoline demand: A smooth timevarying cointegration approach. Energy Economics, Vol. 32 (1), pp Western Bruce and Kleykamp Meredith A Bayesian Change Point Model for Historical Time Series Analysis. Political Analysis, Vol. 12, No. 4, pp WWF-Canada Road Transportation Emissions Reduction Strategies. Yao Y-C Estimating the number of change-points via Schwarz criterion. Statistics and Probability Letters, Vol. 6, pp

What Do We Know about Gasoline Demand Elasticities?

What Do We Know about Gasoline Demand Elasticities? What Do We Know about Gasoline Demand Elasticities? Carol A. Dahl Colorado School of Mines Working Paper http://dahl.mines.edu/courses/courses/dedd/dahl_g06.pdf November 2006 Numerous studies have been

More information

Evidence of a Shift in the Short-Run Price Elasticity of Gasoline Demand

Evidence of a Shift in the Short-Run Price Elasticity of Gasoline Demand Evidence of a Shift in the Short-Run Price Elasticity of Gasoline Demand Jonathan E. Hughes, Christopher R. Knittel, and Daniel Sperling ABSTRACT Understanding the sensitivity of gasoline demand to changes

More information

The impact of fiscal policies and standards on passenger car CO 2 emissions in EU countries

The impact of fiscal policies and standards on passenger car CO 2 emissions in EU countries Energy Production and Management in the 21st Century, Vol. 1 609 The impact of fiscal policies and standards on passenger car CO 2 emissions in EU countries A. Ajanovic & R. Haas Vienna University of Technology,

More information

FACTOR-AUGMENTED VAR MODEL FOR IMPULSE RESPONSE ANALYSIS

FACTOR-AUGMENTED VAR MODEL FOR IMPULSE RESPONSE ANALYSIS Nicolae DĂNILĂ, PhD The Bucharest Academy of Economic Studies E-mail: nicolae.danila@fin.ase.ro Andreea ROŞOIU, PhD Student The Bucharest Academy of Economic Studies E-mail: andreea.rosoiu@yahoo.com FACTOR-AUGMENTED

More information

FHWA Forecasts of Vehicle Miles Traveled (VMT): May 2014

FHWA Forecasts of Vehicle Miles Traveled (VMT): May 2014 FHWA Forecasts of Vehicle Miles Traveled (VMT): May 2014 Office of Highway Policy Information Federal Highway Administration May 22, 2014 Highlights Long-Term Economic Outlook Under the IHS Baseline economic

More information

Vehicle Controls and Petrol Demand in Hong Kong

Vehicle Controls and Petrol Demand in Hong Kong Vehicle Controls and Petrol Demand in Hong Kong W. David Walls School of Economics and Finance University of Hong Kong Pokfulam Road Hong Kong Tel: 852 2859 1049 Fax: 852 2548 1152 Email: wdwalls@hkusub.hku.hk

More information

Seminar Master Major Financial Economics : Quantitative Methods in Finance

Seminar Master Major Financial Economics : Quantitative Methods in Finance M. Sc. Theoplasti Kolaiti Leibniz University Hannover Seminar Master Major Financial Economics : Quantitative Methods in Finance Winter Term 2018/2019 Please note: The seminar paper should be 15 pages

More information

A Smart Approach to Analyzing Smart Meter Data

A Smart Approach to Analyzing Smart Meter Data A Smart Approach to Analyzing Smart Meter Data Ted Helvoigt, Evergreen Economics (Lead Author) Steve Grover, Evergreen Economics John Cornwell, Evergreen Economics Sarah Monohon, Evergreen Economics ABSTRACT

More information

The Implications of Heterogeneity for the Regulation of Energy-Consuming Durable Goods

The Implications of Heterogeneity for the Regulation of Energy-Consuming Durable Goods The Implications of Heterogeneity for the Regulation of Energy-Consuming Durable Goods Mark R. Jacobsen, 1,5 Christopher R. Knittel, 2,5 James M. Sallee, 3,5 Arthur A. van Benthem 4,5 December 8, 2014

More information

THREE LEVEL HIERARCHICAL BAYESIAN ESTIMATION IN CONJOINT PROCESS

THREE LEVEL HIERARCHICAL BAYESIAN ESTIMATION IN CONJOINT PROCESS Please cite this article as: Paweł Kopciuszewski, Three level hierarchical Bayesian estimation in conjoint process, Scientific Research of the Institute of Mathematics and Computer Science, 2006, Volume

More information

Econometric Analysis of Network Consumption and Economic Growth in China

Econometric Analysis of Network Consumption and Economic Growth in China Econometric Analysis of Network Consumption and Economic Growth in China Yuqi LI, Tingie LV, Xia Chen Beijing University of Posts and Telecommunications Beijing,China Abstract With the rapid development

More information

COORDINATING DEMAND FORECASTING AND OPERATIONAL DECISION-MAKING WITH ASYMMETRIC COSTS: THE TREND CASE

COORDINATING DEMAND FORECASTING AND OPERATIONAL DECISION-MAKING WITH ASYMMETRIC COSTS: THE TREND CASE COORDINATING DEMAND FORECASTING AND OPERATIONAL DECISION-MAKING WITH ASYMMETRIC COSTS: THE TREND CASE ABSTRACT Robert M. Saltzman, San Francisco State University This article presents two methods for coordinating

More information

Energy Economics 45 (2014) Contents lists available at ScienceDirect. Energy Economics. journal homepage:

Energy Economics 45 (2014) Contents lists available at ScienceDirect. Energy Economics. journal homepage: Energy Economics 45 (2014) 244 253 Contents lists available at ScienceDirect Energy Economics journal homepage: www.elsevier.com/locate/eneco Modeling gasoline demand in the United States: A flexible semiparametric

More information

Econ 366 Energy Economics. Fall 2012 Transportation Energy Demand

Econ 366 Energy Economics. Fall 2012 Transportation Energy Demand Econ 366 Energy Economics Fall 2012 Transportation Energy Demand Transportation Energy Demand Two major components: passenger (fuel consumption per passenger mile x passenger miles) and freight (fuel consumption

More information

LOSS DISTRIBUTION ESTIMATION, EXTERNAL DATA

LOSS DISTRIBUTION ESTIMATION, EXTERNAL DATA LOSS DISTRIBUTION ESTIMATION, EXTERNAL DATA AND MODEL AVERAGING Ethan Cohen-Cole Federal Reserve Bank of Boston Working Paper No. QAU07-8 Todd Prono Federal Reserve Bank of Boston This paper can be downloaded

More information

Is There an Environmental Kuznets Curve: Empirical Evidence in a Cross-section Country Data

Is There an Environmental Kuznets Curve: Empirical Evidence in a Cross-section Country Data Is There an Environmental Kuznets Curve: Empirical Evidence in a Cross-section Country Data Aleksandar Vasilev * Abstract: This paper tests the effect of gross domestic product (GDP) per capita on pollution,

More information

Taylor Rule Revisited: from an Econometric Point of View 1

Taylor Rule Revisited: from an Econometric Point of View 1 Submitted on 19/Jan./2011 Article ID: 1923-7529-2011-03-46-06 Claudia Kurz and Jeong-Ryeol Kurz-Kim Taylor Rule Revisited: from an Econometric Point of View 1 Claudia Kurz University of Applied Sciences

More information

An Analysis of Cointegration: Investigation of the Cost-Price Squeeze in Agriculture

An Analysis of Cointegration: Investigation of the Cost-Price Squeeze in Agriculture An Analysis of Cointegration: Investigation of the Cost-Price Squeeze in Agriculture Jody L. Campiche Henry L. Bryant Agricultural and Food Policy Center Agricultural and Food Policy Center Department

More information

Modeling the Safety Effect of Access and Signal Density on. Suburban Arterials: Using Macro Level Analysis Method

Modeling the Safety Effect of Access and Signal Density on. Suburban Arterials: Using Macro Level Analysis Method Modeling the Safety Effect of Access and Signal Density on Suburban Arterials: Using Macro Level Analysis Method Yuan Jinghui 1,2 and Wang Xuesong 1,2 (1. The Key Laboratory of Road and Traffic Engineering,

More information

Testing the Predictability of Consumption Growth: Evidence from China

Testing the Predictability of Consumption Growth: Evidence from China Auburn University Department of Economics Working Paper Series Testing the Predictability of Consumption Growth: Evidence from China Liping Gao and Hyeongwoo Kim Georgia Southern University; Auburn University

More information

DETECTING AND MEASURING SHIFTS IN THE DEMAND FOR DIRECT MAIL

DETECTING AND MEASURING SHIFTS IN THE DEMAND FOR DIRECT MAIL Chapter 3 DETECTING AND MEASURING SHIFTS IN THE DEMAND FOR DIRECT MAIL 3.1. Introduction This chapter evaluates the forecast accuracy of a structural econometric demand model for direct mail in Canada.

More information

AN ESTIMATION OF THE DEMAND FOR GASOLINE IN MONTANA, AND PROJECTIONS OF FUTURE GASOLINE CONSUMPTION. Aaron David McNay

AN ESTIMATION OF THE DEMAND FOR GASOLINE IN MONTANA, AND PROJECTIONS OF FUTURE GASOLINE CONSUMPTION. Aaron David McNay AN ESTIMATION OF THE DEMAND FOR GASOLINE IN MONTANA, AND PROJECTIONS OF FUTURE GASOLINE CONSUMPTION by Aaron David McNay A thesis submitted in partial fulfillment of the requirements for the degree of

More information

The Price and Income Elasticity Demand for Gasoline in the United States. Abstract

The Price and Income Elasticity Demand for Gasoline in the United States. Abstract The Price and Income Elasticity Demand for Gasoline in the United States By: Mark Groza Undergraduate: University of Akron (BSLE) May 5, 2006 Abstract Why do Americans consume so much gasoline? This is

More information

British Columbia s carbon tax: Greenhouse gas emission and economic trends since introduction

British Columbia s carbon tax: Greenhouse gas emission and economic trends since introduction British Columbia s carbon tax: Greenhouse gas emission and economic trends since introduction Sierra Rayne a,, Kaya Forest b a Chemologica Research, 318 Rose Street, PO Box 74, Mortlach, Saskatchewan,

More information

Journal of Asian Scientific Research

Journal of Asian Scientific Research Journal of Asian Scientific Research journal homepage: http://aessweb.com/journal-detail.php?id=5003 A METAFRONTIER PRODUCTION FUNCTION FOR ESTIMATION OF TECHNICAL EFFICIENCIES OF WHEAT FARMERS UNDER DIFFERENT

More information

Cost-Effective Policies to Reduce Vehicle Emissions

Cost-Effective Policies to Reduce Vehicle Emissions Cost-Effective Policies to Reduce Vehicle Emissions By DON FULLERTON AND LI GAN* To compare cost-effectiveness of different abatement methods, many studies estimate production or cost functions and plot

More information

The cyclicality of mark-ups and profit margins: some evidence for manufacturing and services

The cyclicality of mark-ups and profit margins: some evidence for manufacturing and services The cyclicality of mark-ups and profit margins: some evidence for manufacturing and services By Ian Small of the Bank s Structural Economic Analysis Division. This article (1) reviews how price-cost mark-ups

More information

Notes on Intertemporal Consumption Choice

Notes on Intertemporal Consumption Choice Paul A Johnson Economics 304 Advanced Topics in Macroeconomics Notes on Intertemporal Consumption Choice A: The Two-Period Model Consider an individual who faces the problem of allocating their available

More information

DETERMINANTS OF LABOUR PRODUCTIVITY IN MALTA FROM A FIRM-LEVEL SURVEY

DETERMINANTS OF LABOUR PRODUCTIVITY IN MALTA FROM A FIRM-LEVEL SURVEY DETERMINANTS OF LABOUR PRODUCTIVITY IN MALTA FROM A FIRM-LEVEL SURVEY Article published in the Quarterly Review 2018:3, pp. 43-49 BOX 3: DETERMINANTS OF LABOUR PRODUCTIVITY IN MALTA FROM A FIRM-LEVEL SURVEY

More information

CONSTRUCTING AN ECONOMIC COMPOSITE INDICATOR FOR THE UAE 1

CONSTRUCTING AN ECONOMIC COMPOSITE INDICATOR FOR THE UAE 1 CONSTRUCTING AN ECONOMIC COMPOSITE INDICATOR FOR THE UAE 1 ASSIL EL MAHMAH Research and Statistics Department June 2017 1 The views expressed in this paper are those of the author and should not be interpreted

More information

Heterogeneity in the Response to Gasoline Prices: Evidence from Pennsylvania and Implications for the Rebound Effect

Heterogeneity in the Response to Gasoline Prices: Evidence from Pennsylvania and Implications for the Rebound Effect 1 Heterogeneity in the Response to Gasoline Prices: Evidence from Pennsylvania and Implications for the Rebound Effect Alan Jenn Inês Azevedo Kenneth Gillingham * Carnegie-Mellon University Yale University

More information

HIGH FREQUENCY EVIDENCE ON THE DEMAND FOR GASOLINE

HIGH FREQUENCY EVIDENCE ON THE DEMAND FOR GASOLINE HIGH FREQUENCY EVIDENCE ON THE DEMAND FOR GASOLINE LAURENCE LEVIN MATTHEW S. LEWIS FRANK A. WOLAK Forthcoming in the American Economic Journal: Economic Policy Draft Date: October 20, 2016 Abstract Daily

More information

Taxing Motor Fuels. How Much is Enough?

Taxing Motor Fuels. How Much is Enough? A major element of climate change policy in Canada is the use of taxes and other forms of carbon pricing (i.e. through cap and trade systems) to raise the price of hydrocarbons. The economic theory behind

More information

DYNAMICS OF ELECTRICITY DEMAND IN LESOTHO: A KALMAN FILTER APPROACH

DYNAMICS OF ELECTRICITY DEMAND IN LESOTHO: A KALMAN FILTER APPROACH DYNAMICS OF ELECTRICITY DEMAND IN LESOTHO: A KALMAN FILTER APPROACH THAMAE Retselisitsoe Isaiah National University of Lesotho THAMAE Leboli Zachia National University of Lesotho THAMAE Thimothy Molefi

More information

Empirical Exercise Handout

Empirical Exercise Handout Empirical Exercise Handout Ec-970 International Corporate Governance Harvard University March 2, 2004 Due Date The new due date for empirical paper is Wednesday, March 24 at the beginning of class. Description

More information

The US dollar exchange rate and the demand for oil

The US dollar exchange rate and the demand for oil The US dollar exchange rate and the demand for oil Selien De Schryder Ghent University Gert Peersman Ghent University BoE, CAMA and MMF Workshop on Understanding Oil and Commodity Prices" 25 May 2012 Motivation

More information

The Combined Model of Gray Theory and Neural Network which is based Matlab Software for Forecasting of Oil Product Demand

The Combined Model of Gray Theory and Neural Network which is based Matlab Software for Forecasting of Oil Product Demand The Combined Model of Gray Theory and Neural Network which is based Matlab Software for Forecasting of Oil Product Demand Song Zhaozheng 1,Jiang Yanjun 2, Jiang Qingzhe 1 1State Key Laboratory of Heavy

More information

ARE MALAYSIAN EXPORTS AND IMPORTS COINTEGRATED? A COMMENT

ARE MALAYSIAN EXPORTS AND IMPORTS COINTEGRATED? A COMMENT Sunway Academic Journal 2, 101 107 (2005) ARE MALAYSIAN EXPORTS AND IMPORTS COINTEGRATED? A COMMENT TANG TUCK CHEONG a Monash University Malaysia ABSTRACT This commentary aims to provide an insight on

More information

HIGH-BREAKDOWN ROBUST REGRESSION IN ANALYSIS OF INTERNET USAGE IN EUROPEAN COUNTRIES HOUSEHOLDS

HIGH-BREAKDOWN ROBUST REGRESSION IN ANALYSIS OF INTERNET USAGE IN EUROPEAN COUNTRIES HOUSEHOLDS HIGH-BREAKDOWN ROBUST REGRESSION IN ANALYSIS OF INTERNET USAGE IN EUROPEAN COUNTRIES HOUSEHOLDS Dagmar Blatná Abstract Robust regression methods are acceptable and useful tools for analyzing dependences

More information

MARNA KEARNEY Energy Research Centre University of cape Town

MARNA KEARNEY Energy Research Centre University of cape Town Modelling the impact of CO 2 taxes in combination with the Long Term Mitigations Scenarios on Emissions in South Africa using a dynamic computable general equilibrium model MARNA KEARNEY 2008 Energy Research

More information

Measuring the Short-Run Impact of Fuel Efficiency on U.S. Automobile Industry

Measuring the Short-Run Impact of Fuel Efficiency on U.S. Automobile Industry 22 วารสารเศรษฐศาสตร ธรรมศาสตร Thammasat Economic Journal ป ท 23 ฉบ บท 1 ม นาคม 2548 Vol.23, No.1, March, 2005 Measuring the Short-Run Impact of Fuel Efficiency on U.S. Automobile Industry Abstract Supawat

More information

Chapter 3. Table of Contents. Introduction. Empirical Methods for Demand Analysis

Chapter 3. Table of Contents. Introduction. Empirical Methods for Demand Analysis Chapter 3 Empirical Methods for Demand Analysis Table of Contents 3.1 Elasticity 3.2 Regression Analysis 3.3 Properties & Significance of Coefficients 3.4 Regression Specification 3.5 Forecasting 3-2 Introduction

More information

Modeling Climate Change Policies in the U. S. and Canada: Preliminary Results

Modeling Climate Change Policies in the U. S. and Canada: Preliminary Results Modeling Climate Change Policies in the U. S. and Canada: Preliminary Results Joseph M. Roop, Pacific Northwest National Laboratory, Bill Tubbs and Chris Bataille, Simon Fraser University ABSTRACT Pacific

More information

Applications and Choice of IVs

Applications and Choice of IVs Applications and Choice of IVs NBER Methods Lectures Aviv Nevo Northwestern University and NBER July 2012 Introduction In the previous lecture we discussed the estimation of DC model using market level

More information

Consumer responses: how quick are they?

Consumer responses: how quick are they? Agenda Advancing economics in business Consumer responses: how quick are they? Ignoring the potential delayed reactions of consumers when thinking about investment, pricing, or competition cases, or when

More information

Analysis of the Chipty Report s conclusions regarding packaging changes and smoking prevalence in Australia

Analysis of the Chipty Report s conclusions regarding packaging changes and smoking prevalence in Australia Analysis of the Chipty Report s conclusions regarding packaging changes and smoking prevalence in Australia 30 August 2016 Europe Economics is registered in England No. 3477100. Registered offices at Chancery

More information

A Quantitative Approach to Detect Structural Breaks in the Trend of Bid Prices

A Quantitative Approach to Detect Structural Breaks in the Trend of Bid Prices A Quantitative Approach to Detect Structural Breaks in the Trend of Bid Prices Mohammad Ilbeigi, Baabak Ashuri, Ph.D., and Soheil Shayegh Georgia Institute of Technology Atlanta, Georgia Construction of

More information

Measuring long-term effects in marketing P.M Cain

Measuring long-term effects in marketing P.M Cain Measuring long-term effects in marketing P.M Cain Conventional marketing mix models are typically used to measure short-term marketing ROI and guide optimal budget allocation. However, this is only part

More information

Appendix A Mixed-Effects Models 1. LONGITUDINAL HIERARCHICAL LINEAR MODELS

Appendix A Mixed-Effects Models 1. LONGITUDINAL HIERARCHICAL LINEAR MODELS Appendix A Mixed-Effects Models 1. LONGITUDINAL HIERARCHICAL LINEAR MODELS Hierarchical Linear Models (HLM) provide a flexible and powerful approach when studying response effects that vary by groups.

More information

FOLLOW-UP NOTE ON MARKET STATE MODELS

FOLLOW-UP NOTE ON MARKET STATE MODELS FOLLOW-UP NOTE ON MARKET STATE MODELS In an earlier note I outlined some of the available techniques used for modeling market states. The following is an illustration of how these techniques can be applied

More information

The role of light duty vehicles in future air pollution: a case study of Sacramento

The role of light duty vehicles in future air pollution: a case study of Sacramento Transportation Planning and Technology Vol. 33, No. 6, August 2010, 541549 The role of light duty vehicles in future air pollution: a case study of Sacramento Guihua Wang* Institute of Transportation Studies,

More information

::Solutions:: Problem Set #1: Due end of class September 7, 2017

::Solutions:: Problem Set #1: Due end of class September 7, 2017 Multinationals and the Globalization of Production ::Solutions:: Problem Set #1: Due end of class September 7, 2017 You may discuss this problem set with your classmates, but everything you turn in must

More information

Economic Dynamics in Discrete Time. Jianjun Mia o/ The MIT Press Cambridge, Massachusetts London, England

Economic Dynamics in Discrete Time. Jianjun Mia o/ The MIT Press Cambridge, Massachusetts London, England Economic Dynamics in Discrete Time Jianjun Mia o/ The MIT Press Cambridge, Massachusetts London, England Contents Preface Acknowledgments xvii xxiii I Dynamical Systems 1 1 Deterministic Difference Equations

More information

Comparative study on demand forecasting by using Autoregressive Integrated Moving Average (ARIMA) and Response Surface Methodology (RSM)

Comparative study on demand forecasting by using Autoregressive Integrated Moving Average (ARIMA) and Response Surface Methodology (RSM) Comparative study on demand forecasting by using Autoregressive Integrated Moving Average (ARIMA) and Response Surface Methodology (RSM) Nummon Chimkeaw, Yonghee Lee, Hyunjeong Lee and Sangmun Shin Department

More information

DISCUSSION PAPER. The Rebound Effect for Passenger Vehicles. J o s h u a L i n n P St. NW Washington, DC

DISCUSSION PAPER. The Rebound Effect for Passenger Vehicles. J o s h u a L i n n P St. NW Washington, DC DISCUSSION PAPER Jul y 2013; revised November 2013 RFF DP 13-19-REV The Rebound Effect for Passenger Vehicles J o s h u a L i n n 1616 P St. NW Washington, DC 20036 202-328-5000 www.rff.org The Rebound

More information

Online Appendix Stuck in the Adoption Funnel: The Effect of Interruptions in the Adoption Process on Usage

Online Appendix Stuck in the Adoption Funnel: The Effect of Interruptions in the Adoption Process on Usage Online Appendix Stuck in the Adoption Funnel: The Effect of Interruptions in the Adoption Process on Usage Anja Lambrecht London Business School alambrecht@london.edu Catherine Tucker Katja Seim University

More information

DEPARTMENT OF ECONOMICS ISSN DISCUSSION PAPER 03/15. Conditional Convergence in US Disaggregated Petroleum Consumption at the Sector Level

DEPARTMENT OF ECONOMICS ISSN DISCUSSION PAPER 03/15. Conditional Convergence in US Disaggregated Petroleum Consumption at the Sector Level DEPARTMENT OF ECONOMICS ISSN 1441-5429 DISCUSSION PAPER 03/15 Conditional Convergence in US Disaggregated Petroleum Consumption at the Sector Level Hooi Hooi Lean a and Russell Smyth b Abstract: We test

More information

The Effects of Exchange Rate on Trade Balance in Vietnam: Evidence from Cointegration Analysis

The Effects of Exchange Rate on Trade Balance in Vietnam: Evidence from Cointegration Analysis (Research note) 地域経済研究第 27 号 2016 The Effects of Exchange Rate on Trade Balance in Vietnam: Evidence from Cointegration Analysis Abstract LE, Thuan Dong * ISHIDA, Miki There has been many researchers studied

More information

APPRAISAL OF POLICY APPROACHES FOR EFFECTIVELY INFLUENCING PRIVATE PASSENGER VEHICLE FUEL CONSUMPTION AND ASSOCIATED EMISSIONS.

APPRAISAL OF POLICY APPROACHES FOR EFFECTIVELY INFLUENCING PRIVATE PASSENGER VEHICLE FUEL CONSUMPTION AND ASSOCIATED EMISSIONS. Proceedings 28-29th August, ITRN2017 Dublin Dennehy: Evidence based analysis to guide private passenger vehicle policy decisions APPRAISAL OF POLICY APPROACHES FOR EFFECTIVELY INFLUENCING PRIVATE PASSENGER

More information

Choosing the Right Type of Forecasting Model: Introduction Statistics, Econometrics, and Forecasting Concept of Forecast Accuracy: Compared to What?

Choosing the Right Type of Forecasting Model: Introduction Statistics, Econometrics, and Forecasting Concept of Forecast Accuracy: Compared to What? Choosing the Right Type of Forecasting Model: Statistics, Econometrics, and Forecasting Concept of Forecast Accuracy: Compared to What? Structural Shifts in Parameters Model Misspecification Missing, Smoothed,

More information

No Bernd Hayo and Matthias Neuenkirch. Bank of Canada Communication, Media Coverage, and Financial Market Reactions

No Bernd Hayo and Matthias Neuenkirch. Bank of Canada Communication, Media Coverage, and Financial Market Reactions MAGKS Aachen Siegen Marburg Gießen Göttingen Kassel Joint Discussion Paper Series in Economics by the Universities of Aachen Gießen Göttingen Kassel Marburg Siegen ISSN 1867-3678 No. 20-2010 Bernd Hayo

More information

Differentiated Products: Applications

Differentiated Products: Applications Differentiated Products: Applications Commonly used instrumental variables BLP (1995) demand for autos using aggregate data Goldberg (1995) demand for autos using consumer level data Nevo (2001) testing

More information

The Role of Education for the Economic Growth of Bulgaria

The Role of Education for the Economic Growth of Bulgaria MPRA Munich Personal RePEc Archive The Role of Education for the Economic Growth of Bulgaria Mariya Neycheva Burgas Free University April 2014 Online at http://mpra.ub.uni-muenchen.de/55633/ MPRA Paper

More information

An Econometric Approach to Forecasting Vehicle Miles Traveled in Wisconsin

An Econometric Approach to Forecasting Vehicle Miles Traveled in Wisconsin 0 0 0 An Econometric Approach to Forecasting Vehicle Miles Traveled in Wisconsin Mike Sillence Traffic Forecasting Section Wisconsin Department of Transportation 0 Sheboygan Avenue, Room 0 Madison, WI

More information

The Implied Cost of Carbon Dioxide Under the Cash for Clunkers Program

The Implied Cost of Carbon Dioxide Under the Cash for Clunkers Program CSEM WP 189 The Implied Cost of Carbon Dioxide Under the Cash for Clunkers Program Christopher R. Knittel August 2009 This paper is part of the Center for the Study of Energy Markets (CSEM) Working Paper

More information

Gasoline Taxes and Consumer Behavior

Gasoline Taxes and Consumer Behavior Gasoline Taxes and Consumer Behavior Shanjun Li 1 Joshua Linn 2 Erich Muehlegger 3 1 Resources for the Future 2 Resources for the Future 3 Harvard Kennedy School and NBER Chicago-RFF Conference June 21,

More information

Technical Appendix. Resolution of the canonical RBC Model. Master EPP, 2010

Technical Appendix. Resolution of the canonical RBC Model. Master EPP, 2010 Technical Appendix Resolution of the canonical RBC Model Master EPP, 2010 Questions What are the causes of macroeconomic fluctuations? To what extent optimal intertemporal behavior of households in a walrasssian

More information

Examining the Short-Run Price Elasticity of Gasoline Demand in the United States

Examining the Short-Run Price Elasticity of Gasoline Demand in the United States Clemson University TigerPrints All Theses Theses 12-2012 Examining the Short-Run Price Elasticity of Gasoline Demand in the United States Michael Brannan Clemson University, michael.brannan4@gmail.com

More information

An Empirical Analysis of Demand for U.S. Soybeans in the Philippines

An Empirical Analysis of Demand for U.S. Soybeans in the Philippines An Empirical Analysis of Demand for U.S. Soybeans in the Philippines Jewelwayne S. Cain Graduate Research Assistant Department of Agricultural & Applied Economics University of Missouri 143-C Mumford Hall

More information

The United States alone has less than 5% of the

The United States alone has less than 5% of the The Effect of Prices on Oil Demand in the Transportation and Residential Sectors Amy Cline I. Introduction The United States alone has less than 5% of the world s population, but uses 25% of the world

More information

COMMISSION STAFF WORKING DOCUMENT EXECUTIVE SUMMARY OF THE IMPACT ASSESSMENT. Accompanying the documents

COMMISSION STAFF WORKING DOCUMENT EXECUTIVE SUMMARY OF THE IMPACT ASSESSMENT. Accompanying the documents EUROPEAN COMMISSION Brussels, 11.7.2012 SWD(2012) 214 final COMMISSION STAFF WORKING DOCUMENT EXECUTIVE SUMMARY OF THE IMPACT ASSESSMENT Accompanying the documents Proposal for a regulation of the European

More information

Okun s law and its validity in Egypt

Okun s law and its validity in Egypt Okun s law and its validity in Egypt Hany Elshamy The British University in Egypt (BUE) Email:hany.elshamy@bue.edu.eg Abstract Okun s law is a key relationship in microeconomics and finds that the relationship

More information

Retail Pricing under Contract Self-Selection: An Empirical Exploration

Retail Pricing under Contract Self-Selection: An Empirical Exploration Technology and Investment, 2013, 4, 31-35 Published Online February 2013 (http://www.scirp.org/journal/ti) Retail Pricing under Contract Self-Selection: An Empirical Exploration Yuanfang Lin, Lianhua Li

More information

Exploring the Determinants of Strategic Revenue Management with Idiosyncratic Room Rate Variations

Exploring the Determinants of Strategic Revenue Management with Idiosyncratic Room Rate Variations Exploring the Determinants of Strategic Revenue Management with Idiosyncratic Room Rate Variations Soohyang Noh a, Hee-Chan Lee a, and Seul Ki Lee a a College of Hospitality and Tourism Management Sejong

More information

CHAPTER THREE MARGINAL PRODCUTIVITY ANALYSIS OF GLOBAL SECTORAL WATER DEMAND 3.1 INTRODUCTION

CHAPTER THREE MARGINAL PRODCUTIVITY ANALYSIS OF GLOBAL SECTORAL WATER DEMAND 3.1 INTRODUCTION CHAPTER THREE MARGINAL PRODCUTIVITY ANALYSIS OF GLOBAL SECTORAL WATER DEMAND.1 INTRODUCTION Water use can be divided into two broad categories; residential and non-residential uses. Non-residential water

More information

Targeted Growth Rates for Long-Horizon Crude Oil Price Forecasts

Targeted Growth Rates for Long-Horizon Crude Oil Price Forecasts Targeted Growth Rates for Long-Horizon Crude Oil Price Forecasts Stephen Snudden Queen s University Department of Economics snudden@econ.queensu.ca July 2017 This paper proposes growth rate transformations

More information

Do U.S. Households Favor Higher Efficiency Vehicles When Fuel Prices Increase? Valerie J. Karplus

Do U.S. Households Favor Higher Efficiency Vehicles When Fuel Prices Increase? Valerie J. Karplus Do U.S. Households Favor Higher Efficiency Vehicles When Fuel Prices Increase? Valerie J. Karplus Submission for Dennis J. O'Brien USAEE/IAEE Best Student Paper Award July 8, 2010 Households owning multiple

More information

Volume 29, Issue 2. Asymmetric adjustment of retail gasoline prices in turkey to world crude oil price changes: the role of taxes

Volume 29, Issue 2. Asymmetric adjustment of retail gasoline prices in turkey to world crude oil price changes: the role of taxes Volume 29, Issue 2 Asymmetric adjustment of retail gasoline prices in turkey to world crude oil price changes: the role of taxes C. emre Alper Bogazici University Orhan Torul Bogazici University Abstract

More information

Ph.D. Defense: Resource Allocation Optimization in the Smart Grid and High-performance Computing Tim Hansen

Ph.D. Defense: Resource Allocation Optimization in the Smart Grid and High-performance Computing Tim Hansen Ph.D. Defense: Resource Allocation Optimization in the Smart Grid and High-performance Computing Tim Hansen Department of Electrical and Computer Engineering Colorado State University Fort Collins, Colorado,

More information

Comparison of Efficient Seasonal Indexes

Comparison of Efficient Seasonal Indexes JOURNAL OF APPLIED MATHEMATICS AND DECISION SCIENCES, 8(2), 87 105 Copyright c 2004, Lawrence Erlbaum Associates, Inc. Comparison of Efficient Seasonal Indexes PETER T. ITTIG Management Science and Information

More information

Study Plan Finance Agricultural Marketing Management International Markets and Agricultural 3 3 -

Study Plan Finance Agricultural Marketing Management International Markets and Agricultural 3 3 - Study Plan Faculty of Agriculture MASTER in Agricultural Economics and Agribusiness Management (Thesis Track) First: GENERAL RULES & CONDITIONS: Plan Number 2013 1. This plan confirms to the valid regulations

More information

7. The contribution of human capital to growth: some estimates

7. The contribution of human capital to growth: some estimates Bas van Leeuwen Human Capital and Economic Growth 7. The contribution of human capital to growth: some estimates 1. INTRODUCTION So far we have noted that human capital seems to affect economic growth

More information

PUBLIC GOODS PROVISION IN RURAL AREA: CASE FROM ZHEJIANGPROVINCE OF CHINA. Pan Weiguang 1 1. INTRODUCTION

PUBLIC GOODS PROVISION IN RURAL AREA: CASE FROM ZHEJIANGPROVINCE OF CHINA. Pan Weiguang 1 1. INTRODUCTION PUBLIC GOODS PROVISION IN RURAL AREA: CASE FROM ZHEJIANGPROVINCE OF CHINA Pan Weiguang 1 1. INTRODUCTION Rural public goods are relatively wide concept in economics. Samuelson defined pure public good

More information

Describing DSTs Analytics techniques

Describing DSTs Analytics techniques Describing DSTs Analytics techniques This document presents more detailed notes on the DST process and Analytics techniques 23/03/2015 1 SEAMS Copyright The contents of this document are subject to copyright

More information

Oil Supply and Demand in Canada s Energy Future: Current Context and Long Term Trends

Oil Supply and Demand in Canada s Energy Future: Current Context and Long Term Trends p.42 Oil Supply and Demand in Canada s Energy Future: Current Context and Long Term Trends By Matthew Hansen, Chris Doleman and Abha Bhargava Introduction The authors are with Canada is a large global

More information

Oral Capps, Jr. Gary W. Williams* TAMRC Commodity Market Research Report No. CM November 2007

Oral Capps, Jr. Gary W. Williams* TAMRC Commodity Market Research Report No. CM November 2007 HAS THE AMERICAN LAMB BOARD INCREASED THE DEMAND FOR LAMB? Oral Capps, Jr. Gary W. Williams* TAMRC Commodity Market Research Report No. CM-01-07 November 2007 * Capps is Professor of Agricultural Economics

More information

ARTICLE IN PRESS. Energy Policy

ARTICLE IN PRESS. Energy Policy Energy Policy ] (]]]]) ]]] ]]] Contents lists available at ScienceDirect Energy Policy journal homepage: www.elsevier.com/locate/enpol Rebound 2007: Analysis of U.S. light-duty vehicle travel statistics

More information

THE RESPONSE OF FINANCIAL AND GOODS MARKETS TO VELOCITY INNOVATIONS: AN EMPIRICAL INVESTIGATION FOR THE US*

THE RESPONSE OF FINANCIAL AND GOODS MARKETS TO VELOCITY INNOVATIONS: AN EMPIRICAL INVESTIGATION FOR THE US* THE RESPONSE OF FINANCIAL AND GOODS MARKETS TO VELOCITY INNOVATIONS: AN EMPIRICAL INVESTIGATION FOR THE US* by Flavio Padrini Ministry of the Treasury, Italy ABSTRACT It is commonly thought that interest

More information

Dynamics of Consumer Demand for New Durable Goods

Dynamics of Consumer Demand for New Durable Goods Dynamics of Consumer Demand for New Durable Goods Gautam Gowrisankaran Marc Rysman University of Arizona, HEC Montreal, and NBER Boston University December 15, 2012 Introduction If you don t need a new

More information

Joint Adoption of Conservation Agricultural Practices by Row Crop Producers in Alabama

Joint Adoption of Conservation Agricultural Practices by Row Crop Producers in Alabama Joint Adoption of Conservation Agricultural Practices by Row Crop Producers in Alabama Jason S. Bergtold, Agricultural Economist, USDA-ARS-NSDL, Auburn, AL Manik Anand, Graduate Student, Auburn University,

More information

EFFICACY OF ROBUST REGRESSION APPLIED TO FRACTIONAL FACTORIAL TREATMENT STRUCTURES MICHAEL MCCANTS

EFFICACY OF ROBUST REGRESSION APPLIED TO FRACTIONAL FACTORIAL TREATMENT STRUCTURES MICHAEL MCCANTS EFFICACY OF ROBUST REGRESSION APPLIED TO FRACTIONAL FACTORIAL TREATMENT STRUCTURES by MICHAEL MCCANTS B.A., WINONA STATE UNIVERSITY, 2007 B.S., WINONA STATE UNIVERSITY, 2008 A THESIS submitted in partial

More information

Design for Commodity-by-Industry Interregional Input-Output Models

Design for Commodity-by-Industry Interregional Input-Output Models Chapter Twelve Design for Commodity-by-Industry Interregional Input-Output Models R. B. Hoffman J. N. Kent In this paper we do not attempt to develop an interregional model. Rather, we propose a mathematical

More information

Financing Constraints and Firm Inventory Investment: A Reexamination

Financing Constraints and Firm Inventory Investment: A Reexamination Financing Constraints and Firm Inventory Investment: A Reexamination John D. Tsoukalas* Structural Economic Analysis Division Monetary Analysis Bank of England December 2004 Abstract This paper shows that

More information

Induced Innovation in Canadian Agriculture

Induced Innovation in Canadian Agriculture Induced Innovation in Canadian Agriculture J. S. Clark, Dalhousie University, Canada Lukas Cechura, Czech University of Life Sciences & S.J. Thompson, SJT Solutions, Canada Paper prepared for presentation

More information

What STIRPAT tells about effects of population and affluence on environmental impact?

What STIRPAT tells about effects of population and affluence on environmental impact? What STIRPAT tells about effects of population and affluence on environmental impact? Taoyuan Wei 1 July 21, 2010 Abstract In the literature of STIRPAT application to environmental impacts of population

More information

UNIT - II TIME SERIES

UNIT - II TIME SERIES UNIT - II TIME SERIES LEARNING OBJECTIVES At the end of this Chapter, you will be able to: Understand the components of Time Series Calculate the trend using graph, moving averages Calculate Seasonal variations

More information

A NEW COINCIDENT INDICATOR FOR THE PORTUGUESE ECONOMY*

A NEW COINCIDENT INDICATOR FOR THE PORTUGUESE ECONOMY* A NEW COINCIDENT INDICATOR FOR THE PORTUGUESE ECONOMY* António Rua** 1. INTRODUCTION * The views expressed in this article are those of the author and not necessarily those of the Banco de Portugal. The

More information

Technical Appendix. Resolution of the canonical RBC Model. Master EPP, 2011

Technical Appendix. Resolution of the canonical RBC Model. Master EPP, 2011 Technical Appendix Resolution of the canonical RBC Model Master EPP, 2011 1. Basic real business cycle model: Productivity shock and Consumption/saving Trade-off 1.1 Agents behavior Set-up Infinitively-lived

More information