The Australian spatial econometrics and statistics workshop

02/17/2023 09:00 am 02/17/2023 06:00 pm Australia/Melbourne The Australian spatial econometrics and statistics workshop

The Department of Econometrics and Business Statistics at Monash Business School is pleased to host the Australian Spatial Econometrics and Statistics workshop.

The aim of the workshop is to bring together academic scientists, researchers and research scholars who are interested in or are working on spatial econometric and statistical issues. The program will consist of invited and contributed papers that cover theoretical, methodological and empirical areas in Spatial Econometrics and Statistics.

This workshop will also be preceded by a three-day course on Spatial Econometrics which will be held from Tuesday 14 to Thursday 16 February 2023. It is not required that you attend the course in order to participate in this workshop. For more information on the course please visit our website.

There is no participation fee to attend this workshop. This is a face-to-face workshop being held in Melbourne, Australia, and will not be streamed online.

Invited guest speakers

Professor Tomohiro Ando, Melbourne Business School, University of Melbourne

Prof Ando is Full Professor of Management (Business Administration) at Melbourne Business School, University of Melbourne. His research relates to Big Data, AI and Analytics, develops methodology for panel and spatial data, and is connected but not limited to Accounting, Economics, Human Resource Management, Finance, Health, Marketing, Operations Management, Public policy, Strategy and Sustainability. His contributions have appeared in the Annals of Statistics, Biometrika, Journal of the American Statistical Association, Journal of Econometrics, Management Science, among other distinguished journals. He is director of the Centre of Excellence: Big Data, AI and Analytics, and Co-organiser of the online seminar series "Frontiers of Big Data, AI and Analytics". He is an academic convener of the Melbourne Centre of Data Science, which is built out of a joint collaboration between Statistics and Computer Science at University of Melbourne.

Professor Giuseppe Arbia, Universita' Cattolica del Sacro Cuore

Prof Arbia is Full Professor of Economic Statistics at the Faculty of Economics, Rome office. He holds a PhD from the University of Cambridge (UK). His extended research contributions lie in the area of Spatial Statistics and Spatial Econometrics. He has published 8 books (by Palgrave-Macmillan, Kluwer, Springer-Verlag, Routledge and CEDAM), 8 book chapters and around 200 articles which are highly cited across the discipline. He is founder and Editor-in-Chief of the "Journal of Spatial Econometrics" (Springer-Verlag), as well as founder and Chairman of the Spatial Econometrics Association since 2006. He is a member of the editorial board of "Empirical Economics" and "Letters in Spatial and Resource Sciences" journals, and is Director of the book series "Spatial Econometrics and Spatial Statistics" (Elsevier).

Professor Noel Cressie, University of Wollongong

Prof Cressie is Distinguished Professor in the Faculty of Engineering and Information Sciences, Director of the Centre for Environmental Informatics in the National Institute for Applied Statistics Research Australia (NIASRA), and Professor of Statistics in the School of Mathematics and Applied Statistics, at the University of Wollongong. His research focus is on developing statistical methodology for analysing spatial and spatio-temporal data with applications in the environmental sciences. He develops Bayesian and empirical-Bayesian methodology for big, complex, hierarchical statistical models. Noel has authored 4 books and has published over 350 articles/chapters/reviews/discussions which are highly cited across the discipline. He has received numerous honours, awards, and fellowships. Among these are from the Committee of Presidents of Statistical Societies (Fisher Award and Lecture - 2009) and from the Statistical Society of Australia (Pitman Medal - 2014) for his outstanding achievements in the discipline of Statistics. He was elected a Fellow of the Australian Academy of Science in 2018.

Dr Petra Kuhnert, CSIRO

Dr Petra Kuhnert is the Group Leader of the Statistical Machine Learning group in the Analytics and Decision Sciences Program within CSIRO's Data61 and works on a wide range of environmental problems. She has a PhD in applied statistics, focussing more recently on the development of methods at the intersection between statistics and machine learning to develop more informative predictions for risk based decision-making. Petra also leads the Hybrid Prediction Activity in the Machine Learning and Artificial Intelligence Future Science Platform in CSIRO, where the focus is on the development of predictive models that are a hybrid of data driven statistical machine learning and domain knowledge. Petra has a passion for science communication and developing methodologies that are fit for purpose, have relevance to the problem being solved and can aid in the decision-making process to stakeholders. She is interested in developing data workflows with trusted science outputs and uses visualisations as a medium for communicating outputs from data and models to facilitate decision-making.

Program

View the full program

TIME	SESSION
8.50 – 9.00am	Opening address by G. Athanasopoulos, Head of Department, Econometrics and Business Statistics, Monash University
9.00 – 10.45am	SESSION 1
9.00 – 9.45am	G. Arbia, Universita' Cattolica del Sacro Cuore (joint with Nardelli, V. and Semerikova, E.) Title: 'Some Robust Approaches to Spatial Econometrics' Abstract: This presentation aims at discussing some robust alternatives to the classical spatial econometric methods. The first topic relates to spatial autocorrelation testing in econometric models. In this area the most widely used measure (the Moran coefficient) is based on the covariance between a variable and its spatially lagged values, a measure which suggests reliable hypothesis testing procedures when the data-generating process is stationary over space and the variable under is, at least approximately, Gaussian. However, many empirical situations are characterized by non-normality. We introduce a robust version of the spatially lagged value and derive a class of alternative robust measures of spatial correlation following the seminal ideas of Peter Huber. The power of the Moran-Huber test is studied through a series of Monte Carlo experiments and results are corroborated by a series of illustrative examples. The second theme refers to robust alternatives to linear spatial econometric model specifications. In this respect, conditional quantile (and specifically median) regression has been recently considered in the spatial econometric literature as a robust alternative when data are characterized by non-normality. However, so far, the literature concentrated on conditional quantile spatial lag specifications which incorporate spatial correlation in the form of a spatially lagged variable included in the list of predictors. We specify conditional spatial quantile models in the form of a spatial error correlation and by suggesting feasible estimators of the parameters involved. We propose a procedure analogous to the feasible GLS suggested for the standard linear models, we examine its small sample properties through a set of Monte Carlo experiments and illustrate its use on real data.
9.45 – 10.15am	W. Wei, EBS, Monash University (joint with Anderson, H., Gao, J., Vahid, F. and Yang, Y.) Title: 'Does Climate Sensitivity Differ Across Regions? A Varying–Coefficient Approach' Abstract: Global mean surface temperature has been increasing in response to growing greenhouse gas concentrations (IPCC, 2021). While Earth is getting warmer overall, regions that differ in local geological features experience unequal increases in temperature. In this paper, we measure climate sensitivity, defined as the increase in temperature following a doubling of CO2 concentrations, in various land regions with a dynamic varying-coefficient panel data model. The inference method proposed in this paper is capable of accommodating heterogeneous co-integrating relationships between global and local variables, and it allows co-moving climate time series to possess both stochastic and deterministic trending components. Using observational data of mean surface temperatures, solar radiation, and carbon dioxide concentrations between 1959–2017, our model provides an estimate of 3.7 C for the average climate sensitivity. Moreover, our estimates indicate that high-latitude regions in the Northern Hemisphere are most vulnerable to global warming.
10.15 – 10.45am	B.H. Nguyen, The University of Queensland (joint with Zelenyuk, V. and Wang, Z.) Title: 'Spatial Stochastic Frontier Analysis: A Case of Queensland Public Hospitals' Abstract: In the vast literature on the productivity and efficiency of hospitals, there is a growing interest in the interaction between spatial contexts and hospital performance. As a complement to the literature, we employ a range of main stream spatial models in a stochastic frontier framework introduced by Cornwell et al. (1990) to estimate the time-variant inefficiency of 97 Queensland public hospitals in a four-year period and to investigate different channels of spatial effects in hospital performance. Meanwhile, when different groups of hospitals perform diversely, we propose a local maximum approach to obtain more reasonable estimates with the spatial stochastic frontier models.
10.45 – 11.05am	BREAK
11.05 – 12.50pm	SESSION 2
11.05 – 11.50am	T. Ando, Melbourne Business School, University of Melbourne (joint with Song, Y.) Title: 'A Bayesian Approach for the Analysis of Large-Scale Nonlinear Dynamic Spatial Panel Data' Abstract: This talk explores a Bayesian approach for the analysis of large-scale nonlinear dynamic spatial panel data where individuals are associated through spatial effects as well as unobserved heterogeneity. After addressing a theoretical challenge associated with the dynamics, the spatial effects and the unobserved heterogeneity, an estimation procedure and asymptotic theory are studied. Markov chain Monte Carlo computational approach is developed. Due to large-scale, nonlinearity, spatial effects, and unobserved heterogeneity of panel as well as a potentially non-smoothness of the objective function, it is not straightforward task to theoretically analyse the model. Here, the Bayesian consistency will be established. Numerical results on a real dataset of gasoline price in Australia are also reported.
11.50 – 12.20pm	B. Peng, EBS, Monash University (joint with Gao, J. and Yan, Y.) Title: 'Higher-Order Expansions and Inference for Panel Data Models' Abstract: In this paper, we propose a simple inferential method for a wide class of panel data models, with a focus on such cases with both serial correlation and cross-sectional dependence. In order to establish an asymptotic theory to support the inferential method, we develop some new and useful higher-order expansions, such as Berry-Esseen bound and Edgeworth Expansion, under a set of simple and general conditions. Interestingly, these expansion results bridge the literature on bootstrap methods and that of heteroskedasticity and autocorrelation consistent (HAC) approaches for panel data models. The new findings fill gaps left by the bulk literature of the block bootstrap for panel data models. Finally, we show the superiority of our approach over several natural competitors using extensive numerical studies.
12.20 – 12.50pm	L. Li, Department of Economics, Monash University (joint with Bailey, N. and Caggiano, G.) Title: 'Measuring Cross-Country Spillovers of Growth-at-Risk' Abstract: This paper assesses the existence and magnitude of spill-over effects of downside risk to GDP growth across a selection of 40 countries worldwide using a spatial autoregressive model with heterogeneous lagged spatial coefficients. We proceed in two steps. First, following recent contributions on measuring downside risk, we estimate country-specific quantile regressions to construct the measure of downside risk to GDP growth, Growth-at-Risk (GaR). Then, we model our estimated GaR using a heterogeneous spatial autoregressive (HSAR) framework to quantify the spillovers across countries. Our empirical results point to the existence of significant spill-over effects in downside risks across countries as well as a high degree of heterogeneity in their strength.
12.50 – 2.00pm	LUNCH
2.00 – 4.00pm	SESSION 3
2.00 – 2.45pm	N. Cressie, University of Wollongong (partly joint with Burden, S., Steel, D. and Wikle, C.) Title: 'Inferring SAR Dependence in Big, Noisy, and Incomplete Data' Abstract: The SAR model is widely used in spatial econometrics to model Gaussian processes on a discrete spatial lattice but, for large datasets, fitting it can become computationally prohibitive. In applications, spatial-econometric data can be noisy and missing in some parts of the lattice; in this talk, I shall modify the SAR model to incorporate these features and derive the predictive distribution. To deal with computational bottlenecks, the computationally efficient spatial random effects (SRE) model with noisy and missing data is calibrated to a SAR model that allows for heteroskedasticity and an asymmetric SAR spatial-dependence matrix. Then restricted maximum likelihood (REML) estimation of the SRE model’s covariance parameters, as well as optimal spatial prediction of the latent Gaussian SAR process have computation times that are only the order of the size of the dataset. An application is given to data on mean usual weekly income from the Australian Census.
2.45 – 3.15pm	C. Lei, EBS, Monash University (joint with Forbes, C. and Bailey, N.) Title: 'Local Case Influence for Spatial Data' Abstract: This paper develops a general methodological framework for Bayesian case influence analysis for spatial autoregressive model. An algebraic representation for the importance sampling weights associated with case-deletion is derived for use with sampled draws from a full data posterior produced via Markov chain Monte Carlo. Once these weights are obtained, a principal components analysis is used on the covariance matrix of log-case deletion weights to produce low dimensional case influence summary plots. The methodology is then applied to artificial data sets, each with a single influential observation. Influential observations are detected in the plots.
3.15 – 3.45pm	L. Rodriguez Pineda, EBS, Monash University Title: 'Distance Matters. Modelling Higher Education Enrolment Decisions in Colombia' Abstract: In the last decade, there has been a special effort in the literature to understand the determinants of higher education participation. There is a consistent observation that the probability of student enrolments in higher education is unequal in many countries and that these inequalities are generally driven by difference in financial capacity and the distance effects. Though, several empirical approaches have been used to find the impact of these determinant factors in Higher Education Enrolments (HEE), we consider that the gravity model may offer some beneficial insights. We also include the academic performance as another determinant of HEE. Our preferred model is the zero-inflated negative binomial as we consider that there are two processes generating the data. We estimate a gravity model for attending a specific HEP and estimate the effect that distance alongside push variables over different types of HEP enrolments to provide relevant information for policy-makers. We find that distance still plays a strong deterrence effect in higher education enrolments. Also, our data suggests that students take distance into consideration in the decision of which type of HEP they enrol into, with more probability of enrolment in geographically closer types of HEP. The role of the academic performance of students in high-school also plays an important role in the choice of which type of HEP students attend.
3.45 – 4.05pm	BREAK
4.05 – 5.50pm	SESSION 4
4.05 – 4.50pm	P. Kuhnert, CSIRO’s Data 61 (joint with Nelson, S., Lucchesi, L., Chin, B.) Title: 'VizumApp: A Tool for Visualising Predictions and Uncertainty from Spatial Models' Abstract:* With spatial modelling, we can make predictions about the future and our environment that inform decisions today. Maps play an important role in understanding and communicating these predictions. In this presentation, I introduce four mapping methods in the VizumApp Shiny app that incorporate prediction uncertainties into the picture, for a more complete story. I present the methods using real-world environmental examples from the Bowen and Burdekin catchments in the Great Barrier Reef. The presentation will include an app demo, touring various customisation features and options to export the Vizumap R code behind the VizumApp maps. *The Vizumap R package and VizumApp Shiny app recently won the “Venables Award” for “New Developers of Open Source Software for Data Analytics”, which was sponsored by the Australian Research Data Commons (ARDC).
4.50 – 5.20pm	S. Kandanaarachchi, CSIRO (joint with Wong, S., Flegg, J.A., Golding, N.) Title: 'Four, fast geostatistical methods -- a comparison' Abstract: Geospatial data is used to model critical applications including disease spread and weather patterns. Traditional methods for geospatial modelling generally take a longer time compared to newer methods. We present a comparison of four, fast geostatistical methods -- INLA, FRK, GPBoost and SpRF -- on malaria prevalence data for Kenya in 2009.
5.20 – 5.50pm	H. Sherry Zhang, Monash University (joint with Cook, D., Menendez, P., Laa, U., Langrene, N.) Title: 'Switching Between Space and Time: Spatio-temporal Analysis with Cubble' Abstract: Spatio-temporal data refer to measurements taken across space and time. In practice, spatio-temporal data can be decomposed into spatial and temporal components: at one time, we would select a spatial location and inspect the temporal trend; at another time, we might select one or multiple time value(s) and explore the spatial distribution. Ideally, we could make multiple maps and multiple time series to explore these together; however, doing all of these actions is complicated when data arrive fragmented in multiple objects. To make it easy to do all these tasks, ideally spatial and temporal variables are in a single data object that we can slice and dice in different ways to conduct different visualisations. In this talk, I will discuss a new data structure, `cubble`, to organise spatio-temporal data so that different types of information can be easily accessed for exploratory data analysis. `cubble` is also capable of handling data with a hierarchical structure, matching data from multiple sources, constructing interactive graphics, and performing spatio-temporal transformation. Data from Australian climate weather stations will be used to demonstrate `cubble`.
5.50 – 6.00pm	Closing remarks

Note: All times are in AEDT

Important dates:

Deadline for paper submissions: 8 December 2022
Decisions notified by: 5 January 2023
Registrations close: 31 January 2023
Workshop date: February 17 2023

Organised by

Dr Natalia Bailey, Department of Econometrics and Business Statistics (EBS), Monash Business School

The workshop is sponsored by the Department of Econometrics and Business Statistics (EBS), Monash Business School and the Spatial Econometrics Association.

Contact

For any queries regarding the Australian Spatial Econometrics and Statistics workshop, please contact us at: ses2023@monash.edu

Event Details

Date:: 17 February 2023 at 9:00 am – 6:00 pm
Venue:: The Pavilion - Level 8, Building H, Monash University Caulfield campus
Campus:
Open to:
Cost:
Categories:: Econometrics and Business Statistics; General

Description

The Department of Econometrics and Business Statistics at Monash Business School is pleased to host the Australian Spatial Econometrics and Statistics workshop.

There is no participation fee to attend this workshop. This is a face-to-face workshop being held in Melbourne, Australia, and will not be streamed online.

Invited guest speakers

Professor Tomohiro Ando, Melbourne Business School, University of Melbourne

Professor Giuseppe Arbia, Universita' Cattolica del Sacro Cuore

Professor Noel Cressie, University of Wollongong

Dr Petra Kuhnert, CSIRO

Program

View the full program

TIME	SESSION
8.50 – 9.00am	Opening address by G. Athanasopoulos, Head of Department, Econometrics and Business Statistics, Monash University
9.00 – 10.45am	SESSION 1
9.00 – 9.45am	G. Arbia, Universita' Cattolica del Sacro Cuore (joint with Nardelli, V. and Semerikova, E.) Title: 'Some Robust Approaches to Spatial Econometrics' Abstract: This presentation aims at discussing some robust alternatives to the classical spatial econometric methods. The first topic relates to spatial autocorrelation testing in econometric models. In this area the most widely used measure (the Moran coefficient) is based on the covariance between a variable and its spatially lagged values, a measure which suggests reliable hypothesis testing procedures when the data-generating process is stationary over space and the variable under is, at least approximately, Gaussian. However, many empirical situations are characterized by non-normality. We introduce a robust version of the spatially lagged value and derive a class of alternative robust measures of spatial correlation following the seminal ideas of Peter Huber. The power of the Moran-Huber test is studied through a series of Monte Carlo experiments and results are corroborated by a series of illustrative examples. The second theme refers to robust alternatives to linear spatial econometric model specifications. In this respect, conditional quantile (and specifically median) regression has been recently considered in the spatial econometric literature as a robust alternative when data are characterized by non-normality. However, so far, the literature concentrated on conditional quantile spatial lag specifications which incorporate spatial correlation in the form of a spatially lagged variable included in the list of predictors. We specify conditional spatial quantile models in the form of a spatial error correlation and by suggesting feasible estimators of the parameters involved. We propose a procedure analogous to the feasible GLS suggested for the standard linear models, we examine its small sample properties through a set of Monte Carlo experiments and illustrate its use on real data.
9.45 – 10.15am	W. Wei, EBS, Monash University (joint with Anderson, H., Gao, J., Vahid, F. and Yang, Y.) Title: 'Does Climate Sensitivity Differ Across Regions? A Varying–Coefficient Approach' Abstract: Global mean surface temperature has been increasing in response to growing greenhouse gas concentrations (IPCC, 2021). While Earth is getting warmer overall, regions that differ in local geological features experience unequal increases in temperature. In this paper, we measure climate sensitivity, defined as the increase in temperature following a doubling of CO2 concentrations, in various land regions with a dynamic varying-coefficient panel data model. The inference method proposed in this paper is capable of accommodating heterogeneous co-integrating relationships between global and local variables, and it allows co-moving climate time series to possess both stochastic and deterministic trending components. Using observational data of mean surface temperatures, solar radiation, and carbon dioxide concentrations between 1959–2017, our model provides an estimate of 3.7 C for the average climate sensitivity. Moreover, our estimates indicate that high-latitude regions in the Northern Hemisphere are most vulnerable to global warming.
10.15 – 10.45am	B.H. Nguyen, The University of Queensland (joint with Zelenyuk, V. and Wang, Z.) Title: 'Spatial Stochastic Frontier Analysis: A Case of Queensland Public Hospitals' Abstract: In the vast literature on the productivity and efficiency of hospitals, there is a growing interest in the interaction between spatial contexts and hospital performance. As a complement to the literature, we employ a range of main stream spatial models in a stochastic frontier framework introduced by Cornwell et al. (1990) to estimate the time-variant inefficiency of 97 Queensland public hospitals in a four-year period and to investigate different channels of spatial effects in hospital performance. Meanwhile, when different groups of hospitals perform diversely, we propose a local maximum approach to obtain more reasonable estimates with the spatial stochastic frontier models.
10.45 – 11.05am	BREAK
11.05 – 12.50pm	SESSION 2
11.05 – 11.50am	T. Ando, Melbourne Business School, University of Melbourne (joint with Song, Y.) Title: 'A Bayesian Approach for the Analysis of Large-Scale Nonlinear Dynamic Spatial Panel Data' Abstract: This talk explores a Bayesian approach for the analysis of large-scale nonlinear dynamic spatial panel data where individuals are associated through spatial effects as well as unobserved heterogeneity. After addressing a theoretical challenge associated with the dynamics, the spatial effects and the unobserved heterogeneity, an estimation procedure and asymptotic theory are studied. Markov chain Monte Carlo computational approach is developed. Due to large-scale, nonlinearity, spatial effects, and unobserved heterogeneity of panel as well as a potentially non-smoothness of the objective function, it is not straightforward task to theoretically analyse the model. Here, the Bayesian consistency will be established. Numerical results on a real dataset of gasoline price in Australia are also reported.
11.50 – 12.20pm	B. Peng, EBS, Monash University (joint with Gao, J. and Yan, Y.) Title: 'Higher-Order Expansions and Inference for Panel Data Models' Abstract: In this paper, we propose a simple inferential method for a wide class of panel data models, with a focus on such cases with both serial correlation and cross-sectional dependence. In order to establish an asymptotic theory to support the inferential method, we develop some new and useful higher-order expansions, such as Berry-Esseen bound and Edgeworth Expansion, under a set of simple and general conditions. Interestingly, these expansion results bridge the literature on bootstrap methods and that of heteroskedasticity and autocorrelation consistent (HAC) approaches for panel data models. The new findings fill gaps left by the bulk literature of the block bootstrap for panel data models. Finally, we show the superiority of our approach over several natural competitors using extensive numerical studies.
12.20 – 12.50pm	L. Li, Department of Economics, Monash University (joint with Bailey, N. and Caggiano, G.) Title: 'Measuring Cross-Country Spillovers of Growth-at-Risk' Abstract: This paper assesses the existence and magnitude of spill-over effects of downside risk to GDP growth across a selection of 40 countries worldwide using a spatial autoregressive model with heterogeneous lagged spatial coefficients. We proceed in two steps. First, following recent contributions on measuring downside risk, we estimate country-specific quantile regressions to construct the measure of downside risk to GDP growth, Growth-at-Risk (GaR). Then, we model our estimated GaR using a heterogeneous spatial autoregressive (HSAR) framework to quantify the spillovers across countries. Our empirical results point to the existence of significant spill-over effects in downside risks across countries as well as a high degree of heterogeneity in their strength.
12.50 – 2.00pm	LUNCH
2.00 – 4.00pm	SESSION 3
2.00 – 2.45pm	N. Cressie, University of Wollongong (partly joint with Burden, S., Steel, D. and Wikle, C.) Title: 'Inferring SAR Dependence in Big, Noisy, and Incomplete Data' Abstract: The SAR model is widely used in spatial econometrics to model Gaussian processes on a discrete spatial lattice but, for large datasets, fitting it can become computationally prohibitive. In applications, spatial-econometric data can be noisy and missing in some parts of the lattice; in this talk, I shall modify the SAR model to incorporate these features and derive the predictive distribution. To deal with computational bottlenecks, the computationally efficient spatial random effects (SRE) model with noisy and missing data is calibrated to a SAR model that allows for heteroskedasticity and an asymmetric SAR spatial-dependence matrix. Then restricted maximum likelihood (REML) estimation of the SRE model’s covariance parameters, as well as optimal spatial prediction of the latent Gaussian SAR process have computation times that are only the order of the size of the dataset. An application is given to data on mean usual weekly income from the Australian Census.
2.45 – 3.15pm	C. Lei, EBS, Monash University (joint with Forbes, C. and Bailey, N.) Title: 'Local Case Influence for Spatial Data' Abstract: This paper develops a general methodological framework for Bayesian case influence analysis for spatial autoregressive model. An algebraic representation for the importance sampling weights associated with case-deletion is derived for use with sampled draws from a full data posterior produced via Markov chain Monte Carlo. Once these weights are obtained, a principal components analysis is used on the covariance matrix of log-case deletion weights to produce low dimensional case influence summary plots. The methodology is then applied to artificial data sets, each with a single influential observation. Influential observations are detected in the plots.
3.15 – 3.45pm	L. Rodriguez Pineda, EBS, Monash University Title: 'Distance Matters. Modelling Higher Education Enrolment Decisions in Colombia' Abstract: In the last decade, there has been a special effort in the literature to understand the determinants of higher education participation. There is a consistent observation that the probability of student enrolments in higher education is unequal in many countries and that these inequalities are generally driven by difference in financial capacity and the distance effects. Though, several empirical approaches have been used to find the impact of these determinant factors in Higher Education Enrolments (HEE), we consider that the gravity model may offer some beneficial insights. We also include the academic performance as another determinant of HEE. Our preferred model is the zero-inflated negative binomial as we consider that there are two processes generating the data. We estimate a gravity model for attending a specific HEP and estimate the effect that distance alongside push variables over different types of HEP enrolments to provide relevant information for policy-makers. We find that distance still plays a strong deterrence effect in higher education enrolments. Also, our data suggests that students take distance into consideration in the decision of which type of HEP they enrol into, with more probability of enrolment in geographically closer types of HEP. The role of the academic performance of students in high-school also plays an important role in the choice of which type of HEP students attend.
3.45 – 4.05pm	BREAK
4.05 – 5.50pm	SESSION 4
4.05 – 4.50pm	P. Kuhnert, CSIRO’s Data 61 (joint with Nelson, S., Lucchesi, L., Chin, B.) Title: 'VizumApp: A Tool for Visualising Predictions and Uncertainty from Spatial Models' Abstract:* With spatial modelling, we can make predictions about the future and our environment that inform decisions today. Maps play an important role in understanding and communicating these predictions. In this presentation, I introduce four mapping methods in the VizumApp Shiny app that incorporate prediction uncertainties into the picture, for a more complete story. I present the methods using real-world environmental examples from the Bowen and Burdekin catchments in the Great Barrier Reef. The presentation will include an app demo, touring various customisation features and options to export the Vizumap R code behind the VizumApp maps. *The Vizumap R package and VizumApp Shiny app recently won the “Venables Award” for “New Developers of Open Source Software for Data Analytics”, which was sponsored by the Australian Research Data Commons (ARDC).
4.50 – 5.20pm	S. Kandanaarachchi, CSIRO (joint with Wong, S., Flegg, J.A., Golding, N.) Title: 'Four, fast geostatistical methods -- a comparison' Abstract: Geospatial data is used to model critical applications including disease spread and weather patterns. Traditional methods for geospatial modelling generally take a longer time compared to newer methods. We present a comparison of four, fast geostatistical methods -- INLA, FRK, GPBoost and SpRF -- on malaria prevalence data for Kenya in 2009.
5.20 – 5.50pm	H. Sherry Zhang, Monash University (joint with Cook, D., Menendez, P., Laa, U., Langrene, N.) Title: 'Switching Between Space and Time: Spatio-temporal Analysis with Cubble' Abstract: Spatio-temporal data refer to measurements taken across space and time. In practice, spatio-temporal data can be decomposed into spatial and temporal components: at one time, we would select a spatial location and inspect the temporal trend; at another time, we might select one or multiple time value(s) and explore the spatial distribution. Ideally, we could make multiple maps and multiple time series to explore these together; however, doing all of these actions is complicated when data arrive fragmented in multiple objects. To make it easy to do all these tasks, ideally spatial and temporal variables are in a single data object that we can slice and dice in different ways to conduct different visualisations. In this talk, I will discuss a new data structure, `cubble`, to organise spatio-temporal data so that different types of information can be easily accessed for exploratory data analysis. `cubble` is also capable of handling data with a hierarchical structure, matching data from multiple sources, constructing interactive graphics, and performing spatio-temporal transformation. Data from Australian climate weather stations will be used to demonstrate `cubble`.
5.50 – 6.00pm	Closing remarks

Note: All times are in AEDT

Important dates:

Deadline for paper submissions: 8 December 2022
Decisions notified by: 5 January 2023
Registrations close: 31 January 2023
Workshop date: February 17 2023

Organised by

Dr Natalia Bailey, Department of Econometrics and Business Statistics (EBS), Monash Business School

The workshop is sponsored by the Department of Econometrics and Business Statistics (EBS), Monash Business School and the Spatial Econometrics Association.

Contact

For any queries regarding the Australian Spatial Econometrics and Statistics workshop, please contact us at: ses2023@monash.edu

Name
E-Mail
Phone
Organisation

The Australian spatial econometrics and statistics workshop

Invited guest speakers

Professor Tomohiro Ando, Melbourne Business School, University of Melbourne

Professor Giuseppe Arbia, Universita' Cattolica del Sacro Cuore

Professor Noel Cressie, University of Wollongong

Dr Petra Kuhnert, CSIRO

Program

Important dates:

Organised by

Contact

Event Details

Description

Invited guest speakers

Professor Tomohiro Ando, Melbourne Business School, University of Melbourne

Professor Giuseppe Arbia, Universita' Cattolica del Sacro Cuore

Professor Noel Cressie, University of Wollongong

Dr Petra Kuhnert, CSIRO

Program

Important dates:

Organised by

Contact

Event Contact