Geospatial Data, GIS and Spatial Analytics

Introduction

International business is conducted in space. Markets are distributed across territories; suppliers, ports, labour pools, customers and competitors occupy identifiable locations; and distance, connectivity, regulation and physical geography shape the cost and feasibility of commercial activity. Geospatial analysis provides the methods for converting these spatial relationships into evidence for market selection, supply-chain design, cluster analysis and facility-location decisions.

A modern geospatial information system (GIS) does more than display maps. It integrates spatial data with economic, demographic, transport, environmental and organizational information; supports measurement and modelling; and communicates results through maps, dashboards and decision-support tools. For international managers, the central question is not simply “Where is something?” but “Why does location matter, how confidently can it be measured, and what decision follows?”

This chapter examines maps, GIS, remote sensing, spatial databases, location quotients, accessibility measures, gravity analysis, spatial interaction, clustering and dashboards. It also emphasizes data quality, scale, uncertainty, ethics and the danger of treating analytical outputs as objective facts independent of their construction.

Learning Objectives

By the end of this chapter, readers should be able to:

1. Distinguish major forms of geospatial data and explain how coordinate systems and scale affect interpretation.
2. Design a GIS workflow for market, cluster, trade-flow and site-selection analysis.
3. Calculate and interpret location quotients, accessibility indicators and gravity-model estimates.
4. Evaluate remote sensing, spatial databases, dashboards and emerging AI-enabled tools.
5. Recognize uncertainty, bias, privacy, governance and regulatory constraints in geospatial decision-making.
6. Translate spatial evidence into defensible international-business recommendations.

1. The Spatial Perspective in International Business

Geography influences business through at least five mechanisms:

• Market potential: population, income, purchasing power and demand are geographically uneven.
• Access: roads, railways, ports, airports, digital networks and border crossings affect effective distance.
• Agglomeration: firms benefit from proximity to suppliers, workers, knowledge institutions and customers.
• Risk and constraints: hazards, climate, water availability, land-use regulation and political boundaries affect operations.
• Institutional context: tax jurisdictions, customs territories, trade agreements and administrative units structure economic activity.

Spatial analysis differs from ordinary tabular analysis because observations may be related by location. Nearby regions can share infrastructure, labour markets, environmental conditions or consumer cultures. This creates spatial dependence. It also means that an apparently strong statistical relationship may reflect omitted geography, common regional trends or the way boundaries were drawn.

A map is therefore an analytical model, not a neutral mirror of reality. Every map selects variables, boundaries, projection, classification and visual hierarchy. A choropleth map of average income by country can conceal inequality within countries, while a map of points can overrepresent places with better reporting systems. Responsible analysis makes these choices explicit.

2. Geospatial Data: Structures, Sources and Quality

2.1 Vector and raster data

Vector data represent discrete features using points, lines and polygons. A port may be represented as a point, a highway as a line, and a customs region as a polygon. Each feature is linked to an attribute table containing information such as throughput, capacity or population.

Raster data represent space as a regular grid of cells. Each cell contains a value, such as land-surface temperature, elevation, vegetation index, night-time illumination or satellite reflectance. Raster resolution is the ground dimension represented by one cell. A 30 m raster is more spatially detailed than a 1 km raster, but it may require more storage and may not imply greater accuracy.

2.2 Coordinate reference systems

A coordinate reference system (CRS) specifies how locations are represented on the Earth. Geographic coordinates use latitude and longitude, usually in degrees. Projected coordinates transform the curved Earth onto a plane, usually in metres. Distances and areas should generally be calculated in an appropriate projected CRS or on a geodesic model, rather than by treating degrees as constant-distance units.

A poor CRS choice can distort area comparisons, buffers and distances. Global studies often require a geodesic calculation or a projection suited to the region of interest. Analysts should record the CRS, datum, units and transformation process as part of metadata.

2.3 Data quality and metadata

Important dimensions of geospatial quality include:

• Positional accuracy: closeness of recorded coordinates to true locations.
• Attribute accuracy: correctness of associated values.
• Completeness: presence of all relevant features and records.
• Temporal currency: whether the data describe the period under study.
• Resolution: the spatial and temporal granularity of observation.
• Consistency: compatibility across sources, definitions and boundaries.
• Lineage: documentation of origin, processing and transformations.

Common sources include national statistical agencies, customs authorities, port authorities, OpenStreetMap, commercial location providers, satellite programmes, mobile-data vendors, logistics companies and firm-level surveys. Open data are not necessarily complete or unbiased, and commercial data are not necessarily valid merely because they are expensive.

3. GIS as an Analytical Workflow

A professional GIS project normally follows these stages:

1. Decision framing: define the business question, geography, time period and decision criterion.
2. Data acquisition: identify relevant demographic, economic, network, environmental and institutional layers.
3. Data preparation: geocode records, harmonize boundaries, transform CRS, remove duplicates and document missing values.
4. Exploratory analysis: map distributions, inspect outliers, compare scales and identify spatial patterns.
5. Modelling: calculate indicators, perform network analysis, estimate flows or evaluate candidate sites.
6. Validation: compare outputs with field knowledge, alternative datasets and sensitivity tests.
7. Communication: present assumptions, uncertainty and recommendations through maps, reports and dashboards.
8. Governance: control access, preserve lineage and establish update and audit procedures.

GIS operations include spatial joins, buffering, overlay, interpolation, geocoding, aggregation, network routing and raster algebra. Each operation encodes assumptions. For example, a 10-kilometre buffer may represent geographic proximity, but not necessarily a 10-minute service area. A spatial join may assign a point to an administrative region even when the relevant economic market crosses that boundary.

4. Remote Sensing and Earth Observation

Remote sensing obtains information about the Earth without direct physical contact, commonly through satellites, aircraft or unmanned aerial systems. Sensors may measure visible, infrared, microwave or thermal energy. Applications in international business include:

• estimating crop conditions and commodity supply;
• monitoring construction, industrial expansion and land-use change;
• detecting road, port and warehouse development;
• assessing flood, wildfire and drought exposure;
• estimating urban expansion and economic activity using night-time lights;
• evaluating environmental compliance and supply-chain risk.

A frequently used vegetation indicator is the Normalized Difference Vegetation Index (NDVI):

Here, NIR is near-infrared reflectance and Red is red-band reflectance. The ratio is dimensionless because both inputs are reflectance values measured on the same scale. NDVI is useful as an indicator of vegetation greenness, but it is not a direct measure of crop yield, income or firm productivity. Clouds, atmospheric conditions, sensor differences, soil background and seasonal timing affect interpretation.

Remote sensing is particularly valuable where official statistics are delayed or unavailable. Nevertheless, imagery requires classification and validation. A model may confuse a logistics yard with an industrial site, or night-time brightness with economic welfare. Analysts should report accuracy assessments and avoid presenting inferred variables as direct observations.

5. Spatial Databases and Data Architecture

A spatial database stores geometries together with attributes and supports spatial queries. Relational systems commonly use tables linked by keys, while spatial extensions provide operations such as intersection, containment, distance and nearest-neighbour searches. A business architecture may include:

• a raw-data layer preserving source files;
• a standardized warehouse with common identifiers and CRS;
• analytical tables containing derived indicators;
• a publication layer for dashboards and application programming interfaces;
• metadata, access controls, versioning and audit logs.

A spatial database should distinguish location identifiers from business entities. A company can operate several facilities; a facility can change ownership; and an administrative boundary can change over time. Time-enabled data are therefore essential for longitudinal analysis.

Data integration is often more difficult than the statistical modelling. Names may differ across languages, addresses may be incomplete, postal codes may be reused, and country or region codes may change. Geocoding produces a location estimate, not certainty. A confidence score and manual review process are appropriate for important decisions.

6. Measuring Economic Concentration with Location Quotients

A location quotient (LQ) measures whether an industry or occupation is relatively concentrated in a region compared with a reference economy. The standard employment formulation is:

Where:

• Eᵢᵣ = employment in industry i in region r;
• Eᵣ = total employment in region r;
• Eᵢ𝑛 = employment in industry i in the reference economy;
• E𝑛 = total employment in the reference economy.

The ratio is dimensionless. An LQ above 1 indicates greater regional specialization than the reference economy; an LQ below 1 indicates lower relative concentration. An LQ of 2 does not mean that the industry is twice as large in absolute terms. It means its employment share is twice the reference share.

Worked example. Suppose Region A has 12,000 manufacturing workers out of total employment of 100,000. The reference country has 600,000 manufacturing workers out of 10,000,000 total employment.

Regional share = 12,000 ÷ 100,000 = 0.12.

National share = 600,000 ÷ 10,000,000 = 0.06.

LQ = 0.12 ÷ 0.06 = 2.0.

Manufacturing is therefore twice as concentrated in Region A as in the reference country. This does not establish a competitive cluster by itself. Cluster analysis should also consider firm counts, productivity, inter-firm linkages, innovation, wages, export performance and persistence over time.

LQ results are sensitive to the reference geography, industry classification, employment measure and spatial scale. A very small region can produce a high LQ from a single specialized employer. Analysts should report absolute employment alongside the LQ and use confidentiality-protected data responsibly.

7. Accessibility and Site Opportunity

Accessibility measures the ease with which a location can reach opportunities or be reached by them. A simple cumulative-opportunity measure counts opportunities within a threshold travel time:

Here, Aᵢ(T) is accessibility from location i within threshold T; Oⱼ is the quantity of opportunity at destination j (such as population, jobs or retail demand); tᵢⱼ is travel time from i to j; and I(·) equals 1 when the condition is true and 0 otherwise. The measure assumes that all opportunities within the threshold are equally accessible and all those outside it are inaccessible, which is often unrealistic.

A continuous gravity-based accessibility measure reduces the abrupt threshold:

where f(tᵢⱼ) is a non-negative impedance function that decreases as travel time or cost increases. A common example is exponential decay, f(t) = e^(−βt), where β > 0 controls sensitivity to travel time. The units of accessibility depend on the units of Oⱼ; if opportunity is population, accessibility is population-equivalent after weighting.

For facility selection, accessibility should be combined with land cost, labour availability, taxes, reliability, environmental risk and service constraints. A site with high population access may be unsuitable because of congestion, zoning or limited freight access.

8. Gravity Models and Spatial Interaction

Gravity models explain or predict flows between origins and destinations using the size of places and the friction of distance. A basic form is:

Where Fᵢⱼ is flow from origin i to destination j; Mᵢ and Mⱼ are masses such as population, income, production or market demand; dᵢⱼ is distance, travel time or generalized cost; G is a scaling constant; and α, γ and β are parameters. Usually, α, γ and β are estimated from data rather than assumed. The model requires positive masses and a meaningful, consistently measured impedance variable.

The model is useful for preliminary estimates of trade, migration, shopping trips and logistics demand. It should not be interpreted as a causal law without a carefully specified empirical design. Trade flows also depend on tariffs, common language, exchange rates, trade agreements, product composition, border delays, firm networks and political relations.

A calibrated model can be evaluated using historical data. Analysts should avoid fitting and evaluating on the same observations, assess prediction errors, and test whether results change under alternative distance measures and border assumptions. When trade data contain many zero flows, log-linear estimation can be problematic; count or multiplicative estimators may be more appropriate depending on the data-generating process.

9. Cluster, Hot-Spot and Network Analysis

Spatial clusters are not simply places that look dense on a map. A cluster may involve geographic concentration, specialized capabilities, knowledge exchange, supplier relationships and institutional support. LQ identifies relative specialization; firm directories and input-output data help identify linkages; patent and collaboration data may reveal knowledge networks.

Hot-spot analysis tests whether high or low values are spatially patterned rather than randomly distributed. Global measures assess overall spatial autocorrelation, while local measures identify concentrations. Such methods require attention to the spatial-weight matrix, which defines which areas are neighbours and how strongly they interact. Results can change when contiguity, distance bands or k-nearest-neighbour definitions are used.

Network analysis represents locations as nodes and connections as edges. In logistics, nodes may be ports, warehouses and distribution centres, while edges represent routes with travel time, cost or capacity. Important concepts include degree, betweenness, connectivity, redundancy and bottleneck exposure. A geographically central facility may still be operationally peripheral if it lacks reliable network connections.

10. Dashboards and Spatial Decision Support

A spatial dashboard combines maps, indicators, filters, charts and alerts. Effective dashboards answer a defined decision question rather than displaying every available layer. A market-entry dashboard might show demand, purchasing power, competitor locations, travel-time catchments, regulatory zones and uncertainty indicators.

Good dashboard design follows several principles:

• state the decision purpose and date of data refresh;
• use consistent definitions across countries;
• distinguish observed, estimated and forecast values;
• show denominators and units;
• avoid misleading colour scales and unnecessary three-dimensional effects;
• permit drill-down without hiding the overall pattern;
• expose data sources, confidence levels and missingness;
• protect confidential or personally identifiable information;
• provide downloadable methodology and audit information.

A dashboard can create false confidence because interactive graphics appear authoritative. A well-designed system makes uncertainty visible and encourages users to compare scenarios rather than accept a single ranking.

11. Illustrative Case: Selecting a Regional Distribution Centre

Background

The following is a fictional illustrative case. Meridian Home Systems, a multinational consumer-electronics company, wants a distribution centre to serve three neighbouring countries. Management initially proposes the capital city because it has the largest population.

Problem

The capital has high demand but severe congestion and expensive land. Two secondary cities have lower populations but better highway and border access. The firm must compare market reach, labour, land cost and disruption risk.

Analysis

The project team assembled population and income data, competitor locations, road and border networks, industrial parcels, flood-risk raster data and customs-processing times. It harmonized all layers to a common CRS and used travel-time routing rather than straight-line distance.

For each candidate site, the team calculated population accessible within four hours, using the cumulative-opportunity measure. It also estimated expected shipments with a gravity-style model in which destination demand was represented by household expenditure and impedance by travel time plus border delay. A sensitivity analysis varied border delays by ±30 percent and tested a scenario with a major highway closure.

The capital reached 8.1 million consumers within four hours, but its annualized land and congestion cost was highest. City B reached 7.4 million and had the lowest disruption exposure because it had two independent highway corridors. City C reached 6.8 million but had a flood-prone access road and weaker labour availability. When border delays increased, City B became the preferred location by a wider margin.

Approach and solution

The team used a multi-criteria decision analysis. Scores were standardized separately for access, operating cost, labour, resilience and environmental risk. Management assigned weights after reviewing scenarios rather than allowing the GIS team to choose them unilaterally. City B was selected provisionally, subject to a site survey, legal due diligence and confirmation of customs-service capacity.

Outcome and learning points

The analysis changed the decision from “largest city” to “best regional accessibility under realistic constraints.” The result depended on network travel times, not merely population; on scenario testing, not a single forecast; and on governance, because decision-makers had to approve the weights. The case also shows that spatial outputs support judgment but do not replace commercial negotiations or engineering, legal and environmental assessments.

12. Limitations, Ethics and Governance

The modifiable areal unit problem

The modifiable areal unit problem (MAUP) occurs when results change because data are aggregated into different zones or because boundaries are drawn differently. A relationship visible at the country level may disappear at the district level. Analysts should test multiple scales and avoid treating administrative units as natural markets.

Ecological fallacy and individual privacy

The ecological fallacy arises when relationships observed for areas are assumed to apply to individuals. A high-income district does not mean every resident is affluent. Conversely, individual-level mobility or transaction data can create privacy risks. Appropriate safeguards include aggregation, de-identification, access controls, purpose limitation, retention limits and disclosure testing.

Bias and unequal visibility

Places with weak statistical capacity may appear less economically active simply because fewer records exist. Geocoding services can perform poorly for rural or informal settlements. Mobile-phone data may underrepresent people without smartphones or with limited network access. AI-generated classifications can reproduce historical bias and should be independently validated.

Legal and institutional issues

Geospatial work may involve data-protection laws, cross-border transfer restrictions, intellectual-property rights, satellite licensing, export controls and national-security rules. Firms should establish data ownership, permitted uses, audit trails and responsibility for errors. Maps used in investment decisions should be reproducible and retain archived versions of source data and code.

13. Contemporary Developments and Best Practice

Cloud platforms now permit large-scale processing of satellite imagery, mobility data and trade records. Digital twins can combine physical infrastructure, sensors and simulation. Machine learning supports image classification, address matching, demand forecasting and anomaly detection. Generative AI can assist with query construction and explanation, but it should not be trusted to invent sources, infer sensitive attributes or make unreviewed location decisions.

A mature international-business geospatial practice should:

1. begin with a decision and theory of change, not with available data;
2. use the finest appropriate spatial and temporal scale without implying unsupported precision;
3. document CRS, definitions, transformations, missing data and model parameters;
4. triangulate official statistics, field evidence, network data and remote sensing;
5. validate outputs against known locations and alternative specifications;
6. show uncertainty, sensitivity and confidence alongside rankings;
7. separate descriptive maps from predictive or causal claims;
8. involve domain experts and affected stakeholders;
9. protect privacy and commercially sensitive information;
10. establish procedures for updating, auditing and retiring obsolete layers.

Chapter Summary

Geospatial data and GIS enable international-business decisions to incorporate location, distance, networks, boundaries and environmental conditions. Vector, raster, remote-sensing and network data each represent different aspects of reality and require appropriate coordinate systems, metadata and validation. GIS is a complete workflow involving decision framing, data preparation, modelling, validation, communication and governance.

Location quotients identify relative economic specialization but should be combined with absolute scale and evidence of linkages. Accessibility measures quantify opportunities reachable within thresholds or through distance-decay functions. Gravity models provide useful approximations of spatial interaction, although trade and business flows also depend on institutions, borders and firm networks. Dashboards can make evidence accessible, but they must disclose units, dates, uncertainty and data limitations.

The principal professional lesson is that spatial analysis is conditional. Results depend on scale, boundaries, definitions, network assumptions, data quality and model specification. Responsible managers therefore use GIS not as an automatic answer generator, but as a transparent and testable system for improving judgment.

Key Terms

Accessibility: The ease with which a location can reach opportunities or be reached by them.

Choropleth map: A map that shades areas according to an attribute value.

Coordinate reference system: A framework for representing locations on or near the Earth.

Geocoding: Converting an address or place description into geographic coordinates.

Geospatial data: Data that describe the location, shape or spatial relationship of phenomena.

GIS: A system for storing, analysing and communicating geographically referenced information.

Gravity model: A spatial-interaction model relating flows to the size of places and friction of distance.

Location quotient: A ratio comparing an industry’s regional share with its share in a reference economy.

MAUP: The problem that analytical results change with the spatial units or scale selected.

Network analysis: Analysis of connected nodes and edges such as roads, routes or facilities.

NDVI: A normalized satellite-derived indicator based on near-infrared and red reflectance.

Positional accuracy: The closeness of a recorded location to its true location.

Raster: A grid of cells containing spatially referenced values.

Remote sensing: Collecting information about the Earth using sensors without direct contact.

Spatial autocorrelation: The degree to which nearby observations have similar or dissimilar values.

Spatial database: A database capable of storing and querying geometries and spatial relationships.

Spatial interaction: Movement or exchange between geographically separated origins and destinations.

Vector data: Spatial data represented as points, lines and polygons.

Zoning: The division of land into areas with specified permitted uses.

Review Questions

Short Answer Questions

1. What is the difference between vector and raster geospatial data?
2. Why is the choice of coordinate reference system important for distance and area calculations?
3. What does a location quotient greater than 1 indicate?
4. How does a cumulative-opportunity accessibility measure work?
5. Identify two ethical risks associated with mobility or location data.

Long Answer Questions

1. Explain how a GIS workflow can support international market-entry analysis, from data acquisition to managerial recommendation.
2. Discuss the assumptions and limitations of gravity models in analysing international trade or distribution flows.
3. Compare remote sensing with official economic statistics as sources for international-business intelligence.
4. Explain how the MAUP and ecological fallacy can distort regional market analysis.
5. Design principles for a dashboard that communicates site-selection evidence to senior executives.

Analytical / Application Questions

1. A region has 18,000 logistics employees out of 150,000 total employees. The reference economy has 400,000 logistics employees out of 8,000,000 total employees. Calculate the location quotient and interpret it alongside the importance of absolute employment.
2. Two potential sites have equal land costs. Site A reaches 1.2 million consumers within four hours but depends on one border crossing; Site B reaches 1.0 million consumers and has two independent crossings. Explain what additional spatial and non-spatial evidence you would require before recommending a site.
3. Propose a spatial-analysis design for identifying an electronics-manufacturing cluster in a developing country. Specify datasets, indicators, validation procedures, ethical safeguards and at least two sensitivity tests.

Multiple Choice Questions with Answers

Question 1

Which statement best describes a location quotient of 2.0?

A. The industry employs twice as many people in absolute terms as the reference economy.

B. The industry’s employment share is twice the share observed in the reference economy.

C. The region has twice as many firms as the reference economy.

D. The industry has grown by 200 percent during the study period.

Correct Answer: B. The industry’s employment share is twice the share observed in the reference economy.

Question 2

Which data structure is most appropriate for representing a continuous satellite-derived land-surface temperature field?

A. Raster

B. Point vector

C. Line vector

D. Relational text table without coordinates

Correct Answer: A. Raster

Question 3

In the accessibility expression Aᵢ(T) = ∑ⱼ Oⱼ · I(tᵢⱼ ≤ T), what does the indicator function do?

A. Converts travel time into monetary cost

B. Assigns a value of 1 when the threshold condition is met and 0 otherwise

C. Estimates the population of the origin

D. Standardizes all destination opportunities to zero

Correct Answer: B. Assigns a value of 1 when the threshold condition is met and 0 otherwise

Question 4

A region has 12,000 manufacturing workers out of 100,000 total workers. The reference economy has 600,000 manufacturing workers out of 10,000,000 total workers. What is the manufacturing location quotient?

A. 0.5

B. 1.0

C. 2.0

D. 12.0

Correct Answer: C. 2.0

Question 5

Which statement is a valid interpretation of a gravity model?

A. Distance has no effect when market size is large.

B. Flows are generally modelled as increasing with place size and decreasing with spatial impedance.

C. It proves that distance is the only cause of trade.

D. It can be used without defining the units of distance or flow.

Correct Answer: B. Flows are generally modelled as increasing with place size and decreasing with spatial impedance.

Question 6

Why can a geographic buffer be misleading as a measure of market access?

A. Buffers cannot contain points.

B. A fixed radius does not account for travel networks, congestion or barriers.

C. Buffers are only valid for raster data.

D. A buffer automatically removes all spatial error.

Correct Answer: B. A fixed radius does not account for travel networks, congestion or barriers.

Question 7

Which is an example of positional accuracy?

A. Whether a dataset includes every relevant business

B. Whether a recorded warehouse coordinate is close to the warehouse’s true location

C. Whether employment values use the same industry classification

D. Whether a dataset has been updated this year

Correct Answer: B. Whether a recorded warehouse coordinate is close to the warehouse’s true location

Question 8

What is the main concern represented by the modifiable areal unit problem?

A. Results may change when spatial units or scales change.

B. Satellite sensors cannot record any numerical values.

C. All administrative boundaries are legally invalid.

D. Network data cannot represent travel time.

Correct Answer: A. Results may change when spatial units or scales change.

Question 9

Which practice most improves the credibility of a site-selection dashboard?

A. Hide missing data to preserve visual simplicity.

B. Show data dates, definitions, uncertainty and sensitivity to assumptions.

C. Use a single ranking without revealing weights.

D. Replace field validation with interactive graphics.

Correct Answer: B. Show data dates, definitions, uncertainty and sensitivity to assumptions.

Question 10

NDVI uses which two reflectance bands in its standard formulation?

A. Blue and green

B. Thermal infrared and microwave

C. Near-infrared and red

D. Ultraviolet and blue

Correct Answer: C. Near-infrared and red