Forecasting monthly and quarterly time series using STL decomposition

This paper is a re-examination of the benefits and limitations of decomposition and combination techniques in the area of forecasting, and also a contribution to the field, offering a new forecasting method. The new method is based on the disaggregation of time series components through the STL decomposition procedure, the extrapolation of linear combinations of the disaggregated sub-series, and the reaggregation of the extrapolations to obtain estimates for the global series. Applying the forecasting method to data from the NN3 and M1 Competition series, the results suggest that it can perform well relative to four other standard statistical techniques from the literature, namely the ARIMA, Theta, Holt-Winters’ and Holt’s Damped Trend methods. The relative advantages of the new method are then investigated further relative to a simple combination of the four statistical methods and a Classical Decomposition forecasting method. The strength of the method lies in its ability to predict long lead times with relatively high levels of accuracy, and to perform consistently well for a wide range of time series, irrespective of the characteristics, underlying structure and level of noise of the data.     

Booking horizon forecasting with dynamic updating: A case study of hotel reservation data

A highly accurate demand forecast is fundamental to the success of every revenue management model. As is often required in both practice and theory, we aim to forecast the accumulated booking curve, as well as the number of reservations expected for each day in the booking horizon. To reduce the dimensionality of this problem, we apply singular value decomposition to the historical booking profiles. The forecast of the remaining part of the booking horizon is dynamically adjusted to the earlier observations using the penalized least squares and historical proportion methods. Our proposed updating procedure considers the correlation and dynamics of bookings both within the booking horizon and between successive product instances. The approach is tested on real hotel reservation data and shows a significant improvement in forecast accuracy.     

Firm heterogeneity, productivity and spatially bounded knowledge externalities

In this paper, we argue that conceptually disentangling the ‘context versus composition’ aspects of regional growth is a multilevel issue. By applying multilevel models (also called random-effects models), we show (1) the importance of considering firm-specific characteristics simultaneously with region-specific characteristics, as we find that a large part of what is traditionally assigned to the impact of the region should be assigned to firm-specific characteristics and (2) that existing single-level methodologies can be problematic, as they are vulnerable to the charge of estimating significance levels that are too liberally assigned and promote exaggerations. This is illustrated empirically by showing that single-level approaches would lead to the conclusion that innovation spillovers are highly significant in a setting of Dutch urban growth differentials, while multilevel analyses shows less liberally assigned significance levels. We conclude that multilevel-effect models better fit research questions that combine firm and spatial characteristics simultaneously, especially because they allow firm-specific characteristics to be differently linked to their regional contexts.     

Making unilateral trade liberalisation beneficial to the poor

In this paper, we investigate the impacts of unilateral import liberalisation by a representative South Asian developing economy, Nepal, and demonstrate those conditions required to make the impacts ‘pro-poor.’ Applying the Computable General Equilibrium model to Social Accounting Matrix data, we conclude that import liberalisation is growth-enhancing but that, unfortunately, the rich benefit more than do the poor. We envisage a restructured but plausible model economy that requires a transformational period of ten years, and simulate unilateral trade liberalisation but, in the context of a dynamic model. We conclude that improvement in efficiency parameters, reorganisation of investment patterns, along with reallocation of factors of production by both household group and activity type are required to make growth accrued by import liberalisation ‘pro-poor’ in developing economies such as that of Nepal.     

Foreign direct investment: A genetic algorithm approach

This paper proposes a genetic algorithm (GA) approach as an analytical tool with a carefully defined fitness function for variable selection. Discriminant analysis will be used as a parameter evaluation method for the analysis of inward foreign direct investment (FDI) in China. Results indicate that the proposed GA method is more efficient in classifying “successful or unsuccessful” inward FDI by providing higher accuracy rates while using fewer variables than previous efforts. An implication of this result is that, given a scarcity of resources and the need to promote FDI, the proposed GA may provide a more efficient way to concentrate on those fewer variables found to be important determinants of “successful” FDI inflow.     

Integrating expected coverage and local reliability for emergency medical services location problems

Daskin's MEXCLP model [Daskin M. A maximum expected covering location model: formulation, properties, and heuristic solution. Transportation Science 1983;17:48-70] was one of the first efforts to capture the stochastic nature of emergency medical services (EMS) location problems within a mixed-integer formulation. With their subsequent introduction of MALP, ReVelle and Hogan [The maximum availability location problem. Transportation Science 1989;23:192-200] offered two key advances, local vehicle busyness estimates and the α-reliability objective. While these constructs have influenced many subsequent EMS location models, they have been subjected to relatively little empirical analysis. To address this, we introduce the LR-MEXCLP, a hybrid model combining the local busyness estimates of MALP with the maximum coverage objective of MEXCLP. We then solve a series of problems with all three models and employ simulation to estimate aggregate service levels. We find that LR-MEXCLP leads to modest but consistent service gains over both MALP and MEXCLP. These results support the merits of local busyness estimates, but they also suggest that the α-reliability objective may be inappropriate when seeking to maximize aggregate system response capabilities. More generally, our research underscores the utility of (a) linking modeling assumptions and goals with real-world application contexts, and (b) employing simulation or other techniques to validate theoretical results.     

Group decision-making: Head-count versus intensity of preference

This paper puts forth a framework for reshaping the group decision-making process. The proposed framework extends from the usual one-issue-at-a-time decision-making to one that involves several related issues simultaneously. Weaknesses of the traditional majority voting mechanism are first identified, and then a different voting method that takes each individual voter's sentiment into account is discussed. Specifically, a decision-maker is asked to express his/her intensity of preference for the issues encountered. Three hierarchical structures—benefits, costs, and risks—are developed to evaluate the alternatives. Due to the nature of pairwise comparisons and synthesis, the proposed method is amenable to consensus building and has higher reliability and consistency. It can be used for candidate selection, e.g. governmental election, when a large population is involved. It is also effective for resource allocation and prioritization when a small group or business is concerned. We believe the proposed approach has potential for resolving deficiencies of the conventional voting mechanism, and can be applied to many real-world problems. Its implementation on the Internet is also discussed.     

Economic development and resource degradation: Conflicts and policies

The paper develops a model that shares common features with computable general equilibrium (CGE) models. It is used to address two questions. First, what are the future prospects of a green gross domestic product (GDP); should we be concerned with resource degradation or not; and, to what extent, and under which conditions? Secondly, which policies are more effective than others? Model closures are simulated using different specifications of exogenous variables. Further, alternative policies are treated: human capital, price incentives, property rights and poverty reduction. In the African context, we show that while the prospects of environment-friendly economic development, i.e. a rising green GDP, are weak in the medium-run, under certain structural conditions there is a range of effective policies that resolves the conflict between economic growth and resource degradation, thus contributing to a rising green GDP.