TAIL FORECASTING WITH MULTIVARIATE BAYESIAN ADDITIVE REGRESSION TREES

We develop multivariate time-series models using Bayesian additive regression trees that posit nonlinearities among macroeconomic variables, their lags, and possibly their lagged errors. The error variances can be stable, feature stochastic volatility, or follow a nonparametric specification. We evaluate density and tail forecast performance for a set of U.S. macroeconomic and financial indicators. Our results suggest that the proposed models improve forecast accuracy both overall and in the tails. Another finding is that when allowing for nonlinearities in the conditional mean, heteroskedasticity becomes less important. A scenario analysis reveals nonlinear relations between predictive distributions and financial conditions.     

Seasonal models of peak electric load demand

Energy consumption in a pilgrim city belonging to a Gulf Cooperation Council (GCC) country exhibits strong seasonal pattern due to higher demand in summer season and additional load during the pilgrimage months. The pilgrimage month's timing is not fixed in the Gregorian calendar. The event varies according to the lunar calendar called the Hegira calendar, which lags behind the former by approximately 14 days in a year. Ten seasonal demand models are developed to model energy estimate for a GCC pilgrimage city. Among the long-range forecast models, three trigonometric models, a multiplicative model, and a multivariate model using categorical variables are considered. Further, a composite nonlinear model whose coefficients are nonlinear is suggested. This model combines the seasonality extracted from a multivariate regression model and a model that represents the peak electric load pattern. Adopting least square fit of a chi-square error function expanded by parabolic expansion, the parameters of the nonlinear model are identified. Moreover, smoothing-based techniques, such as moving average, double exponential smoothing, Winter's, and a multiplicative seasonal model, are suggested. The peak electric load model on lunar and solar calendars is closely related, and the deference in fitting error can be attributed to the magnitude of data. Computational results and statistical tests are presented to analyze the models. It is observed that the multiplicative model performs better to predict the peak electric load demand.     

Realization and accumulation in a Marxian model of the circuit of capital

A mathematical model of Marx's circuit of capital is presented, whose elements, the composition of capital, the rate of surplus value, the rate of capitalization of surplus value, and production, sales and financial time lags, are measurable from the accounts of capitalist firms. The model is solved for exponential paths. A solution with a positive rate of accumulation generally exists. On paths with positive rates of accumulation and positive time lags in the recommital of realized value to production an expansion of credit is required to permit the sale of commodities. A model of demand pull inflation is incorporated in the framework.     

Generalized Spectral Tests for Conditional Mean Models in Time Series with Conditional Heteroscedasticity of Unknown Form

Economic theories in time series contexts usually have implications on and only on the conditional mean dynamics of underlying economic variables. We propose a new class of specification tests for time series conditional mean models, where the dimension of the conditioning information set may be infinite. Both linear and nonlinear conditional mean specifications are covered. The tests can detect a wide range of model misspecifications in mean while being robust to conditional heteroscedasticity and higher order time-varying moments of unknown form. They check a large number of lags, but naturally discount higher order lags, which is consistent with the stylized fact that economic behaviours are more affected by the recent past events than by the remote past events. No specific estimation method is required, and the tests have the appealing „nuisance parameter free” property that parameter estimation uncertainty has no impact on the limit distribution of the tests. A simulation study shows that it is important to take into account the impact of conditional heteroscedasticity; failure to do so will cause overrejection of a correct conditional mean model. In a horse race competition on testing linearity in mean, our tests have omnibus and robust power against a variety of alternatives relative to some existing tests. In an application, we find that after removing significant but possibly spurious autocorrelations due to nonsynchronous trading, there still exists significant predictable nonlinearity in mean for S&P 500 and NASDAQ daily returns.     

Lag length selection in vector autoregressive models: symmetric and asymmetric lags

This study used Monte Carlo simulations to study the performance of alternative lag selection criterion for symmetric lag and asymmetric lag vector autoregressive models. Lag models with short lags and with long lags were considered. The alternative criteria considered were the AIC, SIC, Phillips' Posterior Information Criterion, and Keating's modification of the AIC and SIC. The alternative criteria were evaluated by computing the frequency distribution of lags selected, by computing the out-of-sample forecasting performance of models with lags selected using each criterion, and by comparing the ability of models with lags selected using each criterion to mimic the ‘true’ impulse response functions for the lag model.     

The Lags of Monetary Policy

The length of the transmission lags from monetary pblicy to output has been the subject of much research over the years, but there are serious problems in isolating the lags with any precision. This paper uses a simple model of Australian output to estimate the length of the lags, and then examines how attempts to grapple with the estimation problems might change the results.
We estimate that output growth falls by about one-third of one per cent in both the first and second years after a one percentage point rise in the short-term real interest rate, and by about one-sixth of one per cent in the third year. This implies an average lag of about five or six quarters in monetary policy's impact on output growth. Each of these estimates is, however, subject to considerable uncertainty. We discuss the implications for policy of these relatively long and uncertain lags. Finally, we find no evidence that the average lag from monetary policy to output growth has become any shorter in the 1990s.     

An inference about the length of the time-to-build period

This study attempts to infer the length of aggregate time-to-build period by estimating DSGE models with different investment lags and comparing their fits to the data. The models considered in this study use two, four, six, and eight quarters of investment lags. The Bayesian estimation result indicates that the model with six quarters of investment lags fits the data significantly better than the others.     

How predictable are environmental compliance inspections?

The goal of this paper is to examine the timing of environmental compliance inspections and determine the extent to which such inspections can be predicted. The paper focuses on modeling the inspections at hazardous waste facilities in the U.S. using detailed data on individual inspections and facilities. The paper uses a number of parametric and semi-parametric duration models to predict the timing of inspections and finds that the exponential model provides the best balance in terms of the explanatory power and simplicity of the model. However, even with this model it is difficult to accurately predict the timing of most compliance inspections. The paper also examines the extent to which using data on individual inspections can improve empirical predictions about aggregate inspections. If the goal is to estimate the annual number of inspections at hazardous waste facilities, neither the exponential model or a Poisson model is clearly superior. Which model is more appropriate depends on the question the researcher wants to answer. Similarly, if the focus is on whether any inspection occurred in a given time period, the benefits of using the exponential model depend on the nature of the questions to be answered. While the exponential model performs better than a probit model in predicting which entities will be inspected, it also results in a higher number of “false positives,” that is predicting an inspection when no inspection actually occurs.     

Conserving biodiversity with tradable permits under changing conservation costs and habitat restoration time lags

Tradable permits are a common environmental policy instrument that has recently been applied also to the conservation of biodiversity. Biodiversity conservation differs in many respects to the classical applications of tradable permits like emissions control. One particularity is that, even if the permit system maintains a constant total amount of species habitat, habitat turnover (the destruction of a habitat and restoration elsewhere) affects the ecosystem. Another particularity is that the restoration of habitats often takes much time, leading to time lags between the initiation of restoration activities and the time when restored habitat is available for trading. We use an agent-based model of a tradable permit market to study the influence of heterogeneous and dynamic conservation costs and habitat restoration time lags on key variables of the market, such as the costs incurred to the market participants and the amount of habitat turnover. Our results show that there may be trade-offs between these key variables. We also find that restoration time lags can lead to fluctuations in permit prices that reduce the efficiency of the permit market. We conclude that temporal lags deserve a careful analysis when implementing tradable permit systems for the preservation of natural habitats and biodiversity.     

Beyond the Linearity Critique: The Knife-edge Assumption of Steady-state Growth

The “linearity critique” of endogenous growth models is presented in a general context of an arbitrary growth model and reassessed. It is argued that presence of linearities is not a valid criterion for rejecting growth models. Existence of exponential/geometrical steady-state growth (i.e. of a balanced growth path with strictly positive growth rates) necessarily requires some knife-edge condition which is not satisfied by typical parameter values. Hence, balanced growth paths are fragile and sensitive to smallest disturbances in parameter values. Adding higher order differential/difference equations to a model does not change the knife-edge character of steady-state growth.     

Some results about the exponential ordering of inactivity time

The purpose of this paper is to study new notions of stochastic comparisons and aging classes based on the exponential order. We provide some preservation properties of the exponential order of inactivity time under the reliability operations of convolution and mixture. Some applications to shock models are discussed.     

Competing auctions with endogenous quantities

Two sellers decide on their discrete supply of a homogenous good. There is a finite number of buyers with unit demand and privately known valuations. In the first model, there is a centralized market place where a uniform auction takes place. In the second, there are two distinct auction sites, each with one seller, and buyers decide where to bid. Using the theory of potential games, we show that in the one-site auction model there is always an equilibrium in pure-strategies. In contrast, if the distribution of buyers values has an increasing failure rate, and if the marginal cost of production is relatively low, there is no pure-strategy equilibrium where both sellers make positive profits in the competing sites model. We also identify conditions under which an equilibrium with a unique active site exists. We deal with the finite and discrete models by using several results about order statistics developed by Richard Barlow and Frank Proschan [R. Barlow, F. Proschan, Mathematical Theory of Reliability, Wiley, New York, 1965; R. Barlow, F. Proschan, Inequalities for linear combinations of order statistics from restricted families, Ann. Math. Statist. 37 (1966) 1593-1601; R. Barlow, F. Proschan, Statistical Theory of Reliability and Life Testing, McArdle Press, Silver Spring, 1975].     

On the backward-forward procedure

It is shown that backward-forward procedure which selects solutions in dynamic models with rational expectations may fail to give a solution when the lags are of order two.     

Population dynamics and utilitarian criteria in the Lucas–Uzawa Model

This paper introduces population growth in the Uzawa–Lucas model, analyzing the implications of the choice of the welfare criterion on the model's outcome. Traditional growth theory assumes population growth to be exponential, but this is not a realistic assumption (see Brida and Accinelli, 2007). We model exogenous population change by a generic function of population size. We show that a unique non-trivial equilibrium exists and the economy converges towards it along a saddle path, independently of population dynamics. What is affected by the type of population dynamics is the dimension of the stable manifold, which can be one or two, and when the equilibrium is reached, which can happen in finite time or asymptotically. Moreover, we show that the choice of the utilitarian criterion will be irrelevant on the equilibrium of the model, if the steady state growth rate of population is null, as in the case of logistic population growth. Then, we show that a closed-form solution for the transitional dynamics of the economy (both in the case population dynamics is deterministic and stochastic) can be found for a certain parameter restriction.     

Dynamics of self-rated health and selective mortality

Self-rated health status (SRHS) is one of the most frequently used health measures in empirical health economics. This article analyzes the first seven waves of the Health and Retirement Study (HRS) and finds that (1) all available lags have decreasing but significant predictive power for current SRHS and (2) SRHS and future mortality are strongly related which leads to a specific selection problem known as survivorship bias. A parsimonious joint model with an autocorrelated latent health component in both the SRHS and the mortality equation is suggested. It is better able to capture the empirical facts than commonly used models including random effects and/or state dependence and better able to correct the survivorship bias than commonly used strategies such as inverse probability weighting.     

MONTHLY GDP FORECASTING USING BRIDGE MODELS: APPLICATION FOR THE FRENCH ECONOMY

This paper presents a model to predict French gross domestic product (GDP) quarterly growth rate. The model is designed to be used on a monthly basis by integrating monthly economic information through bridge models, for both supply and demand sides, allowing thus economic interpretations. For each GDP component, bridge equations are specified by using a general‐to‐specific approach implemented in an automated way by Hoover and Perez and improved by Krolzig and Hendry. This approach allows to select explanatory variables among a large data set of hard and soft data. A rolling forecast study is carried out to assess the forecasting performance in the prediction of aggregated GDP, by taking publication lags into account in order to run pseudo real‐time forecasts. It turns out that the model outperforms benchmark models. The results show that changing the set of equations over the quarter is superior to keeping the same equations over time. In addition, GDP growth seems to be more precisely predicted from a supply‐side approach rather than a demand‐side approach.     

An Evaluation of Inertia Models of Unemployment

Two approaches have been used to model unemployment. The first, conventional, approach involves linking the unemployment outcome to observed indices of productivity, structural factors and discrimination such as educational attainment, location and birthplace. The second approach, the inertia model, involves using a person's labour market history as a way of including in unemployment models information on the 'unobservables' that influence employability. This paper evaluates the performance of both models of unemployment. The results provide unambiguous support for the inertia model when modelling unemployment. The inertia model has higher explanatory power, higher within-sample prediction rate success and fewer out-of-sample forecasting errors than the conventional model. The estimates from the inertia model can be used to provide quite accurate predictions of the risk of becoming unemployed. This is important if individuals at high risk of becoming unemployed are to be targeted for labour market assistance.     

Trust in triads: effects of exit, control, and learning

This paper provides theoretical background for some effects of social networks on trust. We study the implications of a model with rational actors in two settings with three actors. In the first setting, there are two trustees who are involved in transactions with one truster implying that the truster has an exit option. In the second setting, two trusters play with one trustee, which gives the trusters options for voice, i.e., complaining and informing each other about the trustee's behavior. We compare these models with a baseline model in which there is only one truster and one trustee. It turns out that the opportunities for placing and honoring trust do not change for the exit model compared to the baseline model. The opportunities for trust in the voice model differ from the baseline model only if both trusters inform each other at a rate that is high enough. Only if the possibilities for receiving information and transmitting information are large enough for both trusters, trust will increase due to the information exchange possibilities in the voice model.     

Cuba’s reform and economic growth: a comparative perspective with Vietnam

Cuban reform process lags behind the GDP growth reached by the Vietnamese. When comparing the evolution of the different sectors and demand components of GDP, Vietnam has had higher growth rates in all cases, highlighting exports first and investment second. Once the Balance of Payments Constrained Growth model has been estimated, the significant effect of the foreign exchange constraints on growth for both countries is confirmed. However, the Vietnam growth can be explained not only by its export success. International openness, which included the end of the US embargo, and institutional factors also explains the differential of results.     

Endogenous growth cycles in an overlapping generations model with investment gestation lags

Summary. A simple overlapping generations model with investment gestation lags is constructed. The model shows that, if the technology is of the AK type with capital-deepening externalities, the existence of investment gestation lags always generates permanent cyclical fluctuations in the economic growth rate. The mean growth rate is shown to be positive if the external effect is strong. The model also shows that, if the production technology takes the Cobb-Douglas form, there exists a unique steady state in which the economy exhibits neither cyclical fluctuations nor long-run growth.Received: 3 July 2003, Revised: 3 December 2003, JEL Classification Numbers: E32, B13. Correspondence to: Akiomi KitagawaAkiomi Kitagawa, Akihisa Shibata: The authors would like to thank Yasushi Iwamoto, Kazuo Mino, an anonymous referee and seminar participants at the Macroeconomics Workshop for their helpful comments and suggestions. Financial support from the Ministry of Education, Culture, Sports, Science and Technology of Japan is gratefully acknowledged.