An award-winning research paper shows how alternative macroeconomic data can guide investment decisions
Economists and the broader investment community rely on government data to learn about key market indicators like oil inventories, agricultural production, and export volumes.
But access to new technology is changing that. Commercial satellites are loosening the authorities’ hold over such information and making it possible for investors to anticipate official announcements to stay ahead of the curve.
Recent research by Professors Abhiroop MUKHERJEE and George PANAYOTOV, and PhD student Janghoon SHON of the HKUST Business School (Department of Finance) focuses on such developments. Their paper “Eye in the Sky: Private Satellites and Government Macro Data” examines how alternative, satellitederived estimates can affect the perceived value of government announcements. It does so by relying on the asset price impact of such announcements, which measure the extent to which markets depend on officially sanctioned macro data.
(From left) Prof Abhiroop MUKHERJEE (Liwei Huang Associate Professor of Business) and George PANAYOTOV, and PhD student SHON Janghoon, Department of Finance, HKUST Business School
This topic has direct relevance for the asset management industry and the Asia-Pacific region. That helps to explain why it won the CFAM-ARX Best Paper Award, which is jointly sponsored by the Asia-Pacific Research Exchange of the CFA Institute (ARX) and the CFA Society of Melbourne (CFAM).
The main findings are significant for several reasons. Firstly, they point to a future where it might be easier to resolve uncertainties about macro trends, and where government control over this type of data will diminish. Secondly, they show it is possible to focus on a few locations -- such as manufacturing centers in China -- that are particularly important for estimating specific macro variables. Thirdly, the findings take account of factors like cloud cover above crude oil storage facilities, which can have a marked bearing on the accuracy of satellite images and the quality of interpretation.
Issues with government data
Historically, macro data has played a central role in the decision-making of individual investors, businesses and state entities. Markets have generally relied on governments to provide such data, because of the prohibitive costs a private company would have faced if trying to aggregate data from disparate sources. This raises a couple of major issues.
Macro information obtained in the usual way is also used to measure the government’s economic performance, and this could result in a conflict of interest. Even in advanced economies like the US, trust in government data is “far from absolute”, because it can vary significantly along partisan lines.
Government data is also released infrequently, and often comes with delays. That allows macro uncertainty to build around an issue before it is resolved on the day of an announcement. According to the authors, this “lumpiness” is often associated with significant price changes on those days.
But improvements in satellite imagery are changing the way markets obtain macro information. To show how this is occurring, the research team used a simple identification strategy. The aim was to assess the impact of independent satellite-based estimates on the value of a particular government announcement. This was achieved by comparing the asset price impact following cloudy weather, when commercial satellites cannot “see” key industrial hubs, with clear periods, when they can.
“We applied our approach in two different settings – US crude oil and Chinese manufacturing – where satellite-based macro estimates have drawn significant interest,” Professor Mukherjee says. “Our evidence points to such estimates substantially changing the market’s reliance on government macro data in both settings.”
While this source of alternative data may lack accuracy, and might only showcase a few sectors of the economy, it is already playing an important role. That’s because measuring economic activity at a few selected locations, such as production hubs or bottlenecks, is often sufficient to establish a clear idea of the general situation and make good overall estimates.
In the pipeline
In the US, crude oil is typically transported by pipelines. There are just a handful of places, such as the small towns of Cushing in Oklahoma and Patoka in Illinois, where several pipelines intersect, creating central hubs in the supply chain. Such hubs are home to a substantial proportion of the nation’s oil storage facilities. This made it possible for the research team to focus test design, and “randomize” the availability of satellite data over these hubs.
Since oil is usually stored in tanks with floating roofs, observing differences in the shadows cast inside each tank (when it’s sunny) is a way to estimate the amount of oil being stored.
The team found that in weeks with predominantly cloudy skies, when satellites were unable to get a good look, government announcements on inventory could move prices significantly. But in good weather, when some traders have been forewarned by the ability to monitor inventory levels from satellites, prices do not respond to these announcements in a similar way.
The same approach was used to assess macro data on manufacturing activity in China. The focus was on four key provinces – Guangdong, Jiangsu, Shandong and Zhejiang – which together account for 35 to 40 per cent of total production. It was found that when satellite-based data was available, news of the latest PMI (Purchasing Manager Index), a major monthly barometer, had much less effect on PMI sensitive stocks and the broader market index.
Despite these findings, the authors caution against making hurried assumptions. “Even if such satellite-based information becomes very accurate, governments may still have a role in validating these measures, or perhaps more importantly in disseminating macro information more broadly and in a more equitable fashion,” they conclude.