GDPLive (gdplive.net) uses large amounts of daily transaction data from New Zealand companies and a machine-learning algorithm to track GDP in real-time for the country, regions and various business sectors.
When the official values are announced, we compare them with our live values from three to six months ago. We put a red dot in our graphs to ensure everyone can see how good (and rarely bad) our live values are. A red dot represents the mean live value for each day during a quarter. We also use the official value to re-train our algorithm and analyse why we might have over or underestimated GDP.
We tracked quarterly growth at 0.97 per cent during the year's second quarter. This was far more optimistic than the Reserve Bank of New Zealand forecast of 0.5 per cent. The major banks also predicted quarterly growth to be around the half per cent mark, so our value raised a few eyebrows.
So, how did we do when Statistics New Zealand recently computed the second-quarter GDP growth rate? The officially announced value is 0.9 per cent. Yet again, GDPLive outperformed the forecasts of the RBNZ and the major banks. This is not a fluke, as in many quarters since we launched GDPLive, we got closer to the official figure than anyone else.
As government agencies collect more data, official GDP values get revised. So far, revisions have always reduced the difference between our live value and the first-announced value, which reassures us of the accuracy of our tracking model. Combining daily data with an advanced machine-learning algorithm gives us an edge. With the high quality of our real-time data, we simply get a great snapshot of daily economic activity.
Given the power and capabilities of artificial intelligence (AI) algorithms, it is surprising that government agencies have not added modern technology to their traditional forecasting models. GDPLive produces reliable real-time GDP and inflation values, which could complement the current process. Having access to real-time figures between official announcements would benefit decision-making and reduce the anxiety of official announcements.
Christoph is a Professor of Innovation and Economics at Massey University, the Director of the Knowledge Exchange Hub (Massey University's big data-driven research centre) and the ambassador for Australia and New Zealand of the Kronberg Academy (Germany). This article was first published HERE
We tracked quarterly growth at 0.97 per cent during the year's second quarter. This was far more optimistic than the Reserve Bank of New Zealand forecast of 0.5 per cent. The major banks also predicted quarterly growth to be around the half per cent mark, so our value raised a few eyebrows.
So, how did we do when Statistics New Zealand recently computed the second-quarter GDP growth rate? The officially announced value is 0.9 per cent. Yet again, GDPLive outperformed the forecasts of the RBNZ and the major banks. This is not a fluke, as in many quarters since we launched GDPLive, we got closer to the official figure than anyone else.
As government agencies collect more data, official GDP values get revised. So far, revisions have always reduced the difference between our live value and the first-announced value, which reassures us of the accuracy of our tracking model. Combining daily data with an advanced machine-learning algorithm gives us an edge. With the high quality of our real-time data, we simply get a great snapshot of daily economic activity.
Given the power and capabilities of artificial intelligence (AI) algorithms, it is surprising that government agencies have not added modern technology to their traditional forecasting models. GDPLive produces reliable real-time GDP and inflation values, which could complement the current process. Having access to real-time figures between official announcements would benefit decision-making and reduce the anxiety of official announcements.
Christoph is a Professor of Innovation and Economics at Massey University, the Director of the Knowledge Exchange Hub (Massey University's big data-driven research centre) and the ambassador for Australia and New Zealand of the Kronberg Academy (Germany). This article was first published HERE
No comments:
Post a Comment