Browser issue

It looks like the browser you're using doesn’t work well with our website. For a better experience, please update to the latest version of Chrome, Edge, Firefox or Safari.

Deriving indicators of economic activity from traffic sensor data

Christopher Ball and Scott Graham

Key findings

  • We develop monthly indicators of economic activity in New Zealand from granular data measuring traffic counts for both heavy and light traffic. Our indicators are highly correlated with New Zealand's official measure of aggregate economic activity - Gross Domestic Product.  
  • Our indicators can be disaggregated into regional components at a daily frequency, highlighting variation that would remain masked in aggregate measures. 
  • These traffic indices provide an independent check on other high-frequency economic indicators, better monitoring of regional disparities in economic activity, and support timely policy advice in response to economic shocks. However, the higher volatility of these traffic indices requires careful interpretation, and these traffic indices should be used as part of a broader suite of economic indicators. 


Why we did this research

Timely information about economic activity is crucial for effective monetary policy decision-making. Official statistics like GDP are published with significant delays and are subject to revisions, creating challenges for real-time economic assessment. This research develops high-frequency traffic indicators that can provide early signals of economic changes before official data becomes available. 

The project directly supports our forecasting and modelling research theme, which emphasises the importance of developing new data sources to enhance our ability to monitor the economy in real-time. This need was highlighted as a priority in our most recent Review and Assessment of the Formulation and Implementation of Monetary Policy (RAFIMP). 

Our monetary policy research

Transportation activity has strong theoretical links to economic performance, serving both as an input to production processes and reflecting final demand through the movement of goods. By leveraging newly available open traffic sensor data from the New Zealand Transport Agency Waka Kotahi (NZTA), we have developed novel indicators that track economic activity at daily and regional levels. 

For monetary policy, these high-frequency indicators complement our existing suite of economic monitoring tools, improving our ability to assess current economic conditions and respond appropriately. The regional dimension may also provide insights into how economic activity varies across different parts of New Zealand, which can be valuable when evaluating the effects of monetary policy and other economic shocks. 


What data have we used?

Our traffic indices are constructed using data from January 2010 to the third quarter of 2024. They will be updated on an ongoing basis. The data comes from both historical records and the current Traffic Monitoring System (TMS) data portal maintained by the NZTA. 

Data source Description Time period Update frequency
NZTA Traffic Monitoring System (TMS) Portal  Daily traffic counts from 586 light traffic monitoring sites and 333 heavy traffic monitoring sites across New Zealand state highways  January 2018 to Q3 2024 Daily, with an update lag of 1 to 2 weeks after month-end 
NZTA Historical Traffic Data  Historical traffic counts from monitoring sites across New Zealand state highways  January 2010 to January 2018 Historical data (not updated) 
 GDP data Quarterly real GDP figures used for comparison with traffic indices   Q1 2010 to Q3 2024  Quarterly, with an update lag of around 3 months after quarter-end 
 ANZ Truckometer  Monthly heavy and light traffic indices used for comparison  January 2010 to Q3 2024 Monthly, with an update lag of around 2 weeks after month-end 

The traffic data includes vehicle classifications (light or heavy) based on vehicle length. Vehicles under 5.5m are classified as light, those over 11m as heavy, and those between 5.5m and 11m are split equally between the two categories. Data processing includes filtering for site references with sufficient data availability and imputing missing observations using appropriate statistical methods.