Ragdad Cani Miranti (p)

André Luis S Chagas

The use of night-time light data has been widely growing in socioeconomic studies over the past decades. This remote sensing approach is considered as a reliable proxy for local economic activity particularly in developing countries where traditional data is scarce. The prominent methods relying on official statistics are often time-consuming and experiencing time-lag delays in data availability. Furthermore, these traditional approaches struggle to provide reliable socioeconomic indicators at granular administrative levels, such as districts and sub-districts.
Our study examines the relationship between NTL intensity and sectoral Gross Domestic Regional Product (GRDP) across 514 Indonesian districts from 2019 to 2023. To address the limitation of global NTL in capturing rural agricultural production, we incorporate MODIS land cover data to distinguish between urban and rural NTL intensity. In this research, we apply Visible Infrared Imaging Radiometer Suite (VIIRS) NTL which offers advantages on higher spatial resolution and finer detail of geographical areas. We employ three estimators on regression models—OLS pooled estimator, Least Square Dummy Variable (LSDV) and between regression fixed effect—to assess the predictive power of NTL on sectoral GRDP. The LSDV regression shows significant and positive coefficient only for total and rural NTL on non-agriculture and agriculture GRDP, while OLS pooled and fixed effect between estimator reveal significant and positive coefficient of NTL data on all dependent variables. After controlling for district and year fixed effects, all NTL measures (urban, rural, and total) indicate significant and positive coefficients, suggesting the validity to explain annual changes in Indonesia's sectoral GRDP. A 10 % increase in global NTL implies an 8,27 % in economic activity in district’s all sectors, while a 10 % increase in urban and rural NTL correspond to 4,44 % and 4,49 % percent rise in non-agriculture and agriculture sector, respectively.
In addition, results from all three estimators indicate that urban and rural NTL intensity could explain sectoral GDP variations (agriculture, non-agriculture, and total GDP) well, with R-squared reaching over 80 percent in LSDV regression for all dependent variables. Higher R-squared values in the LSDV and between-estimator regression compared to the OLS pooled estimator suggest that VIIRS-NTL data more effectively predicts cross-sectional GDP variations between districts. Our findings confirm that VIIRS NTL data accurately predicts economic activity in both urban and rural areas.This research validates the use of remotely sensed data for evaluating long-term district-level GRDP patterns.

Eduardo Luzio, André Luis S Chagas (p), Pedro Barros

Dimitrios Tsiotas

This paper investigates how locational features influence bid values in Brazil’s oil and gas (O&G) concession auctions. Using a Spatial Autoregressive Model (SAR), we develop a novel approach incorporating both geographical proximity and the chronological order of auctioned blocks in a spatial weighting matrix. This methodological innovation allows us to assess how past bidding outcomes influence present bids, capturing spatial and temporal interdependencies overlooked in prior research. Additionally, we enhance our spatial econometric analysis with Geographic Information System (GIS) techniques, incorporating new variables such as distance to the coast, exploratory wells, producing fields, and seismic data availability. Our empirical analysis covers ten auction rounds (R1–R10) from 1999 to 2008, marked by significant regulatory and market changes. Brazil transitioned from a state monopoly (Petrobras) to a competitive auction system after the 1997 Oil Act established the National Petroleum Agency (ANP) as the sector’s regulator. These auctions allocated O&G exploration rights to private firms through a first-price sealed-bid mechanism, where bidders had limited geological information and faced high uncertainty, particularly in frontier basins. The results confirm that location matters in O&G auctions. Bidders do not rely solely on the inherent characteristics of individual blocks but also incorporate geospatial and historical bidding patterns. Our SAR estimates indicate a significant spatial effect (ρ ≈ 0.25–0.30), implying that past winning bids influence bid values in one block in neighboring blocks. Additionally, the proximity to previously leased blocks and producing fields positively correlates with bid values, as does the availability of 3D seismic data and pre-stack depth migration (PSDM) surveys. These findings suggest that bidders extract valuable information from the outcomes of past auctions, reinforcing the existence of information externalities. Beyond theoretical contributions, our findings offer policy implications for O&G auction design. The ANP’s minimum reserve price setting has historically underestimated block values, which could be adjusted to better reflect market and geological conditions. Moreover, public investment in seismic surveys before auctions could enhance market efficiency by reducing information asymmetry. This study advances the literature by integrating spatial econometrics and geospatial analysis to improve the understanding of O&G bidding behavior. Our approach has broader applications for other natural resource auctions, where spatial dependencies are crucial in price formation.

Dimitrios Tsiotas (p), Elias Giannakis, Christos Papadas

Christoph Hauser

An Input-Output (IO) Table is a model that captures financial exchanges among sectors at various macroeconomic levels, including international, national, multiregional, and regional. By recording the volume of sectorial interconnections, an IO-Table provides a useful model for understanding an economy’s structure, which is a major determinant of its functionality in response to disruptions. While traditional approaches to IO analysis examine the structure of the economy based on multiplier and linkage analysis, this paper conceives the square matrix of intermediate demand of an IO Table as a graph model and applies complex network and community detection analysis. Using the network paradigm to examine an IO economy allows the unveiling of structural information beyond the traditional approaches. Based on national sectorial data of the period 2005-2015 from Greece, this paper examines the topological features of the Greek IO economy prior to and after the 2008 economic crisis and detects sectors (IO network nodes) and markets (IO network communities) that were either resilient and vulnerable to this disturbance. The analysis provides empirical insights into the resilient performance of the Greek economy in terms of sectorial clustering at the neighborhood scale; the country’s tertiary specialization; the fixedness in the community membership of the trade and transportation industries; the inelastic demand in energy-related economic activities; and the commuting behavior in connectivity of the construction-related economic activities and the public sector. The analysis also provides evidence to consider tourism as a dynamic sector in Greece more due to its derived demand than its direct product. The community detection analysis revealed that the primary sector in the Greek economy is underperforming, while there is a significant level of integration between the secondary and tertiary sectors, particularly in tourism, transportation, and energy activities. Overall, this paper promotes the use of network analysis in input-output models and emphasizes the importance of interdisciplinary approaches in Regional Science.

Christoph Hauser (p), Miroslav Despotovic

Ragdad Cani Miranti

Cultural Ecosystem Services (CES) represent non-material benefits that humans derive from natural environments, encompassing aesthetics, recreation, inspiration, and cultural heritage. Traditional CES assessments rely on survey-based approaches, but the rise of user-generated content on social media offers a novel means to quantify CES spatially and temporally. Existing methods used to study landscape values using surveys are naturally ill suited to study dynamic-landscape scale processes. Social media data, by comparison, can be used to indirectly measure and identify valuable features of landscapes at various and arbitrary territorial levels.
This study integrates these two research streams, employing social media data (geotagged photographs from the photo sharing platform Flickr) and landscape metrics identified from survey-based conjoint analysis to measure CES. We replicate the identification of photographs containing CES based on geotagged keywords in the metadata pioneered by Van Zanten et al (2016)1 for the territories Bundesland Tirol (Austria) and the Autonomous Province of Bozen (Italy). Raster files containing information the count of photographs uploaded on Flickr in the area represent information on the distribution of CES in the two regions.
Survey based investigations by Schirpke et al (2019)2 have shown that a restricted number of landscape metrics can be used to explain the dominant variation in landscape preferences for CES in this area. Such metrics include number of patches as measures of landscape fragmentation, patch richness representing landscape diversity and shape complexity indices capturing variations in landscape form. Additional binary indicators (urban, forest, water, road) account for specific landscape characteristics.
By calculating measures for landscape metrics from the Corine database we test their impact on the geographical distribution of CES photographs derived from the Flickr plattform based on suitable spatial techniques such as spatial lag and spatial error models. The results provide important information on the suitability of crowd-sourced data from social media sites to locate the incidence of CES in a territory. In addition, the findings inform spatial planning policies and guide nature tourism initiatives.

1 Van Zanten, B. T., Van Berkel, D. B., Meentemeyer, R. K., Smith, J. W., Tieskens, K. F., & Verburg, P. H. (2016). Continental-scale quantification of landscape values using social media data. Proceedings of the National Academy of Sciences, 113(46), 12974-12979.

2 Schirpke, U., Tappeiner, G., Tasser, E., & Tappeiner, U. (2019). Using conjoint analysis to gain deeper insights into aesthetic landscape preferences. Ecological Indicators, 96, 202-212.