Structural Time Series Model Research Articles

Bayesian Modeling of Labor Earnings in Construction

Labor costs constitute a significant portion of construction costs. Reliable forecasts provide insight into the movements of labor costs and are critical to the success of projects. Past studies have primarily focused on forecasting construction cost indexes or material costs. Only a few studies have concentrated on forecasting construction labor costs. This study presents a multivariate Bayesian structural time series (MBSTS) model to characterize the future values of construction labor’s average hourly earnings (AHE) using a set of candidate predictors. The methodologies commonly used by past studies do not adequately address the uncertainties associated with the modeling process. In contrast, MBSTS recognizes the uncertainty in its modeling process, which enables practitioners to quantify and account for future labor cost risks in decisions. Furthermore, the MBSTS method results in a transparent model, helping analysts investigate the rationality of the parameters. This article trains MBSTS models under four different data subset lengths (i.e., 150, 144, 138, and 132 months) to study the consistency of the explanatory variables and their corresponding coefficients. The analysis results indicate that the gross domestic product (GDP), housing starts (HS), number of building permits (BP), Construction Cost Index (CCI), Dow Jones Industrial Average (DJI), and Standard and Poor’s 500 index (SPI) are the most frequently used predictors in the regression component of the MBSTS models. The results indicate an inverse relationship between AHE from one side and HS and BP from the other. Likewise, a direct relationship exists between AHE and GDP, CCI, DJI, and SPI. The MBSTS model performed well on the validation subset in the midrange prediction intervals (i.e., 12- and 18-month periods). However, it was outperformed by conventional time series models [i.e., seasonal autoregressive integrated moving average (SARIMA)] when used for short-term forecasting. The proposed framework can be applied to facilitate monetary resource allocation in projects.

Forecasting Euro Banknotes in Circulation with Structural Time Series Models in Times of the COVID-19 Pandemic

Forecasting Euro Banknotes in Circulation with Structural Time Series Models in Times of the COVID-19 Pandemic

Forecasting Banknote Circulation During the COVID-19 Pandemic Using Structural Time Series Models

Forecasting Banknote Circulation During the COVID-19 Pandemic Using Structural Time Series Models

Forecasting Banknote Circulation During the Covid-19 Pandemic Using Structural Time Series Models

Forecasting Banknote Circulation During the Covid-19 Pandemic Using Structural Time Series Models

Evaluation of the short- and long-term impacts of the COVID-19 pandemic on bus ridership in Miyazaki City, Japan

We used a Bayesian structural time series (BSTS) model to evaluate the short- and long-term impacts of the coronavirus disease 2019 (COVID-19) pandemic on transit ridership. We accessed smart-card data from Miyazaki City, Japan. We defined attributes based on card types (commuters, students and elders) and aggregated attributes (high-frequency users and “frequently used bus-stop pairs”) and analyzed the differences between all users and the extracted groups. Among card types, the short-term impact on elders was almost identical to that of all users, however, the short-term impact of the pandemic on commuters was much smaller and that of students was much larger than that of all users. The long-term trend of commuters was less fluctuated than that of all users. The long-term ridership recovery of students was higher than that of all users. Among aggregated attributes, the short-term impact was smaller on “high-frequency users” than on all users: the decrease in ridership immediately after the appearance of COVID-19 was smaller among “high-frequency users” than among all users. The long-term recoveries in the riderships of the extracted subsets were slower than the recoveries of riderships of all users.

Estimation of Lockdowns' Impact on Well-Being in Selected Countries: An Application of Novel Bayesian Methods and Google Search Queries Data.

Lockdowns introduced in connection with the COVID-19 pandemic have had a significant impact on societies from an economic, psychological, and health perspective. This paper presents estimations of their impact on well-being, understood both from the perspective of mental health and considering economic security and similar factors. This is not an easy task because well-being is influenced by numerous factors and the changes happen dynamically. Moreover, there are some obstacles when using the control group. However, other studies show that in certain cases it is possible to approximate selected phenomena with Google search queries data. Secondly, the econometric issues related to the suitable modeling of such a problem can be solved, for example, by using Bayesian methods. In particular, herein the recently gaining in popularity Bayesian structural time series and Bayesian dynamic mixture models are used. Indeed, these methods have not been used in social sciences extensively. However, in the fields where they have been used, they have been very efficient. Especially, they are useful when short time series are analyzed and when there are many variables that potentially have a significant explanatory impact on the response variable. Finally, 15 culturally different and geographically widely scattered countries are analyzed (i.e., Belgium, Brazil, Canada, Chile, Colombia, Denmark, France, Germany, Italy, Japan, Mexico, the Netherlands, Spain, Sweden, and the United Kingdom). Little evidence of any substantial changes in the Internet search intensity on terms connected with negative aspects of well-being and mental health issues is found. For example, in Mexico, some evidence of a decrease in well-being after lockdown was found. However, in Italy, there was weak evidence of an increase in well-being. Nevertheless, the Bayesian structural time series method has been found to fit the data most accurately. Indeed, it was found to be a superior method for causal analysis over the commonly used difference-in-differences method or Bayesian dynamic mixture models.

A Structural Time Series Analysis of the Effect of Quantitative Easing on Stock Prices

In this paper, a structural time series model is estimated to analyse the effect of quantitative easing (QE) on stock prices for the US, UK and Japan. The model is estimated by maximum likelihood in a time-varying parametric framework, using the DJIA, S&P500, NASDAQ, FTSE100 and the NIKKEI225 as the dependent variable and the balance sheet of the respective Central Bank as an explanatory variable, along with the unobserved components that account for the behaviour of other explanatory variables that are not explicitly specified in the model. The results show that QE had a significant but not exclusive effect on the DJIA, S&P500 and the NASDAQ, suggesting that these stock prices are also affected by other missing variables and cyclical movements. However, for the UK and Japan, no effect of QE on the FTSE100 and the NIKKEI225 is found, suggesting that variables other than QE are important for the rise in these stock prices. One plausible explanation for this result is that perhaps QE becomes effective only after a certain threshold level is met.

Extreme Rainfall Forecast and Flood Prediction in Equatorial Zone of West Africa: Port Harcourt, Nigeria in Focus

Purpose: Rainfall is the most important element of climate in the tropics, it dictate human outdoor activities, recharge underground water and streams, dictate agricultural calendar and many cultural activities. More recently the incidence of extreme rainfall leading to urban flood in the city of Port Harcourt is increasingly becoming a looming disaster. Despite the importance of rainfall in the tropic, the fact that it varies significantly is worrisome to both scholars and the stakeholders. This study focused on extreme rainfall forecast and flood prediction in equatorial zone of West Africa especially in a humid tropic environment like Port Harcourt.
 Methodology: Statistical Time Series Model performed very well in rainfall forecast, but did better in flood prediction. Secondary data extracted from Nigeria Meteorological Agency, Oshodi, Lagos form 100/3 of the agency record servers as main source of data for this work. Descriptive statistics and inferential statistics were used, among which are mean, standard deviation, and coefficient of variation; Pearson product moment correlation and structural time series model.
 Findings: The study discovered that rainfall distribution has reduced over the years, besides the dry spell in the month of August has deduced gradually but steadily over the years. It also shown that flood cannot occur in the first 6 months of the year. Rainfall distribution varies significantly over the climatic period.
 Conclusion: There is a significant variation in rainfall distribution over Port Harcourt. August dry spell is steadily disappearing, giving raise to excessive run-off that triggers flood during late September to October. Flood will likely occur whenever the forecasted rainfall exceeds the generated level output during the seasonal peak. Structural Time Series Model performed well in flood prediction than in rainfall forecast.
 Recommendations: Extreme RF forecast by the authorities should be taken serious by inhabitants and warning be given and adhered to by residences. Also, structural time series model should be used for long time forecast since it performed very well, but this should be done with large data.

Causal impacts of the COVID-19 pandemic on daily ridership of public bicycle sharing in Seoul

Public bicycle can be a disease-resilient travel mode during the coronavirus disease 2019 (COVID-19) pandemic. Nonetheless, its evidence on public bicycle sharing is still inconclusive. This study used Bayesian structural time series models and causal impact inference for the data on the daily ridership of public bicycles in Seoul, South Korea, for 1826 days from January 1, 2017, to December 31, 2021. The study found that the usage of public bicycles was robust against the COVID-19 pandemic even in densely populated Seoul. Compared with the pre-pandemic period, public bicycles’ usage was unaffected on days when weather conditions, such as snow, rain, and wind speed were not as severe, as well as on days with non-seasonal event factors, such as weekdays, public holidays, and traditional Korean holidays. In addition, its robustness against the pandemic became more pronounced as the number of bicycle racks increased and the intensity of social distancing increased. However, public bicycles were in demand primarily for leisure and exercise, not for travel, during the pandemic. Public bicycle sharing can be a disease-resilient travel mode. Continuous investment in infrastructure such as bicycle paths and public bicycle is required to become a more resilient travel mode against infectious diseases.

Monitoring urban dynamics using time-series satellite imagery in Dodoma, Tanzania

ABSTRACT This study examines the changes in Dodoma between 2014 and 2021 using Landsat and night-time light (NTL) data. The results indicate that the built-up areas have grown by 47.1 km² between 2014 and 2021. The change point detection results of NTL time series data show that at one period (location 43) of the time series, the probability of a change is very high (> 0.70). This location corresponds to the year 2017. The Bayesian structural time series model results show that the relocation to Dodoma increased NTL by a statistically significant 81% (p < 0.001) from 2017.

Spatiotemporal impact of non-pharmaceutical interventions against COVID-19 on the incidence of infectious diarrhea in Xi'an, China.

Non-pharmaceutical interventions (NPIs) against COVID-19 may prevent the spread of other infectious diseases. Our purpose was to assess the effects of NPIs against COVID-19 on infectious diarrhea in Xi'an, China. Based on the surveillance data of infectious diarrhea, and the different periods of emergence responses for COVID-19 in Xi'an from 2011 to 2021, we applied Bayesian structural time series model and interrupted time series model to evaluate the effects of NPIs against COVID-19 on the epidemiological characteristics and the causative pathogens of infectious diarrhea. A total of 102,051 cases of infectious diarrhea were reported in Xi'an from 2011 to 2021. The Bayesian structural time series model results demonstrated that the cases of infectious diarrhea during the emergency response period was 40.38% lower than predicted, corresponding to 3,211 fewer cases, during the COVID-19 epidemic period of 2020-2021. The reduction exhibited significant variations in the demography, temporal and geographical distribution. The decline in incidence was especially evident in children under 5-years-old, with decreases of 34.09% in 2020 and 33.99% in 2021, relative to the 2017-2019 average. Meanwhile, the incidence decreased more significantly in industrial areas. NPIs against COVID-19 were associated with short- and long-term reductions in the incidence of infectious diarrhea, and this effect exhibited significant variations in epidemiological characteristics.

Policy impact on microplastic reduction in China: Observation and prediction using statistical model in an intensive mariculture bay

Plastic pollution in the environment has spurred debate among scientists, policymakers, and the general public over how industrialization and consumerism are wreaking havoc on our ecosystem, but some policies might assist to ameliorate the problem in the near future. In this study, the decision tree classifier and Bayesian Structural Time Series (BSTS) model was used to anticipate the possible sources of microplastics and their near future state in 26 surface sediment and a sediment core, respectively in Sansha Bay, which has been criticized for its intensive mariculture applications. An inventory of microplastics in the sediment core was estimated, and it was discovered that during the previous six decades, an average of 181.95 tons of microplastics were deposited, with an average deposition (by a layer of sediment) of 179.44 tons/cm. According to the DT classifier, mariculture was the primary source of microplastics, whereas urban and industrial areas were the primary sources of POPs. The Bayesian Structural Time Series (BSTS) model revealed a microplastic downward slope, indicating that regional and national strategies implemented might successfully reduce microplastic pollution regionally.

Has EU Accession Boosted Patent Performance in the EU-13? A Critical Evaluation Using Causal Impact Analysis with Bayesian Structural Time-Series Models

This paper provides new insights into the causal effects of the enlargement of the European Union (EU) on patent performance. The study focuses on the new EU member states (EU-13) and accession is considered as an intervention whose causal effect is estimated by the causal impact method using a Bayesian structural time-series model (proposed by Google). The empirical results based on data collected from the OECD database from 1985–2017 point towards a conclusion that joining the EU has had a significant impact on patent performance in Romania, Estonia, Poland, the Czech Republic, Croatia and Lithuania, although in the latter two countries, the impact was negative. For the rest of the EU-13 countries, there is no significant effect on patent performance. Whether the EU accession effect is significant or not, the EU-13 are far behind the EU-15 (countries which entered the EU before 2004) in terms of patent performance. The majority of patents (98.66%) are assigned to the EU-15, with just 1.34% of assignees belonging to the EU-13.

Non-renewable resource extraction over the long term: empirical evidence from global copper production

Global mine production of copper has risen more than 80 times over the last 135 years. What were the main drivers? I examine this question based on copper market data from 1880 to 2020. I employ a structural time series model with sign restrictions to identify demand and supply shocks. I find that a deterministic trend drives most of the output growth. At the same time, unpredictable demand and supply shocks caused substantial fluctuations around the trend. A global commodity demand shock that is, for example, linked to a 3% unexpected expansion of the global economy due to rapid industrialization causes a 10% rise in the real copper price, incentivizing a 5% increase in global copper production. This provides empirical evidence for the feedback control cycle of mineral supply.

Go Wide or Go Deep? Assortment Strategy and Order Fulfillment in Online Retail

Problem definition: Expansions in product assortment by online retailers often engender operational challenges. In undertaking such expansions, retailers exercise a strategic choice between expanding assortment width or depth. Our understanding of how this choice affects the order fulfillment process is limited. Thus, we examine the impact of these dimensions of assortment strategy on order delivery timeliness. Academic/practical relevance: Order delivery timeliness is a critical measure of operational success in online retail. We contribute to theory and practice by adopting a multidimensional perspective of retailer assortment strategy and studying the relative impact of assortment width and depth on order delivery timeliness. Methodology: Employing a data set comprising more than 200 million orders, we study the effects of assortment strategy on delivery timeliness using an instrumental variable approach. We then utilize a two-stage model to estimate the impact of delivery performance on sales. Further, we employ a matched difference-in-differences and a novel Bayesian structural time-series model to confirm this relationship. Results: We find that assortment width has a greater negative impact on order delivery timeliness compared with assortment depth. A one-standard-deviation increase in assortment width increases average delivery times by 0.55 days. Further, we find this effect to be positively moderated (i.e., worsened) by the average size of orders and to be negatively moderated (i.e., improved) by the logistic service provider’s (LSP) experience. Finally, a one-day increase in delivery times for 10% of the orders results in a 2.7% reduction in sales. Managerial implications: Our findings suggest that online retailers focused on ensuring timely deliveries should be wary of widening product assortments, especially when facing larger average order sizes. We also find that experienced logistic service providers can help mitigate the dilatory effects of assortment width expansions. However, the benefits of experienced LSPs are limited for retailers deepening their assortments.History: This paper has been accepted as part of the 2018 MSOM Data Driven Research Challenge.Supplemental Material: The online appendices are available at https://doi.org/10.1287/msom.2022.1156 .

Are quantitative easing effects transitory? Evidence from out-of-sample forecasts

Purpose Advocates of quantitative easing (QE) policies have emphasized some evidence that structural models do not predict long-term asset yields as well as naive forecasts, implying that predictions of price reversals cannot be profitable and that QE effects are not transitory. The purpose of this study is to reconsider the out-of-sample forecasting performance of structural time series processes relative to that of a random walk with or without drift. Design/methodology/approach This study uses bivariate vector autoregression and Markov switching representations to generate out-of-sample forecasts of ten-year sovereign bond yields, when the information set is augmented by including the growth rate of the monetary base, and the estimation relies on monthly data from countries that have pursued unconventional policies over the last decade. Findings The results show that naive forecasts are not better than those of structural time series models, based on root mean squared errors, while the Markov model provides additional information on price reversals, through probabilistic inferences regarding policy regime switches, which can induce agents to counteract QE interventions and reduce their effectiveness. Originality/value The novel features of this work are the use of a large information set including the instrument of unconventional monetary policy, the use of a structural model (Markov process) that can really inform about potential asset price reversals and the use of a large sample over which QE policies have been pursued.

Trend of hand, foot, and mouth disease from 2010 to 2021 and estimation of the reduction in enterovirus 71 infection after vaccine use in Zhejiang Province, China.

BackgroundZhejiang, ranked in the top three in HFMD (hand, foot, and mouth disease) incidence, is located in the Yangtze River Delta region of southeast China. Since 2016, the EV71 vaccine has been promoted in Zhejiang Province. This study aimed to investigate the trend and seasonal variation characteristics of HFMD from 2010 to 2021 and estimate the reduction in enterovirus 71 infection after vaccine use.MethodsThe data on HFMD cases in Zhejiang Province from January 2010 to December 2021 were obtained from this network system. Individual information on cases and deaths was imported, and surveillance information, including demographic characteristics and temporal distributions, was computed by the system. The Joinpoint regression model was used to describe continuous changes in the incidence trend. The BSTS (Bayesian structural time-series models) model was used to estimate the monthly number of cases from 2017 to 2021 based on the observed monthly incidence during 2010–2016 by accounting for seasonality and long-term trends. The seasonal variation characteristics of HFMD pathogens were detected by wavelet analysis.ResultsFrom 2010 to 2021, the annual incidence rate fluctuated between 98.81 cases per 100,000 in 2020 and 435.63 cases per 100,000 in 2018, and 1711 severe HFMD cases and 106 fatal cases were reported in Zhejiang Province, China. The annual percent change (APC) in EV71 cases was -30.72% (95% CI: -45.10 to -12.50) from 2016 to 2021. The wavelet transform of total incidence and number of cases of the three pathogens all showed significant periodicity on the 1-year scale. The average 2-year scale periodicity was significant for the total incidence, EV71 cases and Cox A16 cases, but the other enterovirus cases showed significant periodicity on the 30-month scale. The 6-month scale periodicity was significant for the total incidence, EV71 case and Cox A16 case but not for the other enteroviruses case. The relative error percentage of the performance of the BSTS model was 0.3%. The estimated number of cases from 2017 to 2021 after the EV-A71 vaccines were used was 9422, and the reduction in the number of cases infected with the EV71 virus was 73.43% compared to 70.80% when the impact of the COVID-19 epidemic in 2020 was excluded.ConclusionsSince 2010, the incidence of EV71 infections has shown an obvious downward trend. All types of viruses showed significant periodicity on the 1-year scale. The periodicity of the biennial peak is mainly related to EV71 and Cox A16 before 2017 and other enteroviruses since 2018. The half-year peak cycle of HFMD was mainly caused by EV71 and Cox A6 infection. The expected incidence will be 2.76 times(include the cases of 2020) and 2.43 times(exclude the cases of 2020) higher than the actual value assuming that the measures of vaccination are not taken. EV71 vaccines are very effective and should be administered in the age window between 5 months and 5 years.

Assessing the effectiveness of quarantine measures during the COVID-19 pandemic in Chile using Bayesian structural time series models.

BackgroundWith the emergence of the COVID-19 pandemic, all existing health protocols were tested under the worst health crisis humanity has experienced since the Black Death in the 14th century. Countries in Latin America have been the epicenter of the COVID-19 pandemic, with more than 1.5 million people killed. Worldwide health measures have included quarantines, border closures, social distancing, and mask use, among others. In particular, Chile implemented total or partial quarantine measures depending on the number of infections in each region of the country. Therefore, it is necessary to study the effectiveness of these quarantines in relation to the public health measures implemented by government entities at the national level.ObjectiveThe main objective of this study is to analyze the effectiveness of national- and region-level quarantines in Chile during the pandemic based on information published by the Chilean Ministry of Health, and answers to the following question are sought: Were quarantine measures in Chile effective during the COVID-19 pandemic?MethodsThe causal effect between the rates of COVID-19 infections and the population rates in Phase 1 and Phase 2 quarantines in the period from March 2020 to March 2021 in different regions of Chile were evaluated using intervention analyses obtained through Bayesian structural time series models. In addition, the Kendall correlation coefficient obtained through the copula approach was used to evaluate the comovement between these rates.ResultsIn 75% of the Chilean regions under study (12 regions out of a total of 16), an effective Phase 1 quarantine, which was implemented to control and reduce the number of cases of COVID-19 infection, was observed. The main regions that experienced a decrease in cases were those located in the north and center of Chile. Regarding Phase 2, the COVID-19 pandemic was effectively managed in 31% (5 out of 16) of the regions. In the south-central and extreme southern regions of Chile, the effectiveness of these phases was null.ConclusionThe findings indicate that in the northern and central regions of Chile, the Phase 1 quarantine application period was an effective strategy to prevent an increase in COVID-19 infections. The same observation was made with respect to Phase 2, which was effective in five regions of northern Chile; in the rest of the regions, the effectiveness of these phases was weak or null.

Mortality, incidence and seasonality of canine and feline patients treated with tick antiserum in three far North Queensland veterinary clinics from 2000 to 2020.

Tick paralysis is a paralysis caused by bites from Ixodes holocyclus, affecting an estimated 10,000 companion animals in Australia annually. Despite tick antiserum being the cornerstone of treatment, there are no large-scale general practice studies that examine survival outcomes in tick antiserum-treated animals. In this retrospective study, clinical records from three far north Queensland general practice veterinary clinics were searched for tick antiserum-treated canine and feline patients were seen between 2000 and 2020. Patient records were assessed for survival outcomes,then logistic regression and Bayesian structural time-series model were used to assess trends in incidence and mortalityand the relationship between these and time of year, rainfall, and species. The study included 2019 dog and 953 cat records. When patients with unknown outcomes were removed, canine mortality was 11.8% (213/1799) and feline mortality was 5.3% (46/872). Dogs were found to have 2.41 odds of dying following treatment than cats. August and September had the highest mean number of monthly treatments, and rainfall in the previous 5-8months was positively correlated with the number of patients treated in each month. The odds of mortality did not vary significantly by month or season, and from 2015 onwards, there was a significant decrease in the proportion of dogs treated by the clinics. Overall, this study provides new information on tick antiserum treatment outcomes in general practice as well as new information on tick paralysis incidence in far north Queensland.

Impact of COVID-19 on the endoscopy department since the early phase of the pandemic in 2020: A questionnaire study among patients with canceled examinations at a single Japanese institution.

IntroductionIn early 2020, the Japanese government declared a nationwide state of emergency for the COVID‐19 pandemic. We investigated the impact of the emergency declaration on endoscopy adherence and conducted a follow‐up study of patients with canceled examinations at a tertiary endoscopy facility in Japan in 2020.MethodsWe compared the number of endoscopies performed, and cancelations at the endoscopy unit between 2019 and 2020 and used the Bayesian structural time series (BSTS) model to estimate the decrease in the number of endoscopies in 2020. We administered a questionnaire to those who had not undergone a scheduled endoscopy.ResultsOf 14 146 and 13 338 scheduled examinations, 1233 (8.7%) and 1403 (10.5%) were canceled in 2019 and 2020, respectively. During both years, age < 50 years, age > 80 years, upper endoscopy, and experience of endoscopy in the past 5 years were significantly associated with cancelations. In 2020, cancelations in the 14th–26th week of the year, including the period of state of emergency, increased significantly, and more women canceled. Of the 409 questionnaire‐respondents, 174 (42.5%) indicated that COVID‐19 had influenced their cancelation, and 315 (77.0%) had not undergone similar endoscopic examinations since then. The BSTS model predicted a decrease of 957 (95% CI −1213 to −708, P = .003) examinations.ConclusionIn 2020, despite low numbers of COVID‐19 cases in the study site, the number of endoscopies decreased, and cancelation increased. Further research is needed on the future impact of a decrease in the number of endoscopies during a COVID‐19 pandemic.