Nowcasts
UNCTAD’s nowcasts are data and model-based predictions of global trade and GDP period over period growth. The latest data available and data revisions are fed into models on a weekly basis to update and revise the nowcasts and provide insight on current economic and trade conditions well before final figures are published with several months’ delay.
The trade nowcasts represent quarter-over-quarter growth rates of seasonally adjusted series while the GDP nowcast represents year-over-year growth rates. The nowcast is one of multiple approaches and tools UNCTAD employs to provide insight to the development of global trade and economic output, in this case solely based on a set of high frequency statistics available at each point in time.
The points on the graph represent what growth rate the model predicted for the selected time period with data available on the date indicated on the X-axis. The bars represent that week’s data releases’ impact on the evolution of the nowcast. See the “METHODOLOGY” tab for more information on how to read this plot.
For more details on the methodology and underlying indicators, see the research paper on Economic Nowcasting with Long Short-term Memory Artificial Neural Networks (LSTM).
Loading nowcasts...
- trade in merchandise (volumes)
- trade in merchandise (values)
- trade in services
- gross domestic product
- methodology
- metadata
The nowcast stands at % for as of .
News from this week’s data releases this nowcast by percentage points.
What are UNCTAD’s nowcasts?
UNCTAD’s nowcasts are data and model-based predictions of global trade and GDP period over period growth. The latest data available are fed into models on a weekly basis to update and revise the nowcasts and provide insight on current economic and trade conditions.
How to interpret the charts?
UNCTAD’s nowcasts are updated on a weekly basis to reflect the impact of new data releases. As a result, the nowcasts are presented not only as a single number representing the latest and best estimate, but also as a line, representing the nowcasts’ evolution over time as more information becomes available. The line chart for a selected target period and series reflects the date the nowcast was made on the X-axis, and the nowcasted growth with data available as of that date on the Y-axis. The dips and rises in the line over time reflect how the outlook economic outlook changed over time in response to signals in the data being published. The bars represent the impacts of those new data on the nowcast. The sum of the negative and positive portions of the bar sum to the revision of the nowcast from the prior week. See the documentation of the multi-programming language nowcast_lstm library for more information on how these impacts are calculated.
What is the modelling approach?
The nowcasts are produced using long short-term memory artificial neural networks (LSTM). LSTMs have been shown to attain high performance in economic nowcasting applications in UNCTAD research, during both stable periods and periods of great economic shock, such as the COVID crisis. For more information on LSTMs’ use in nowcasting, see either this, this, or this UNCTAD research paper, or this benchmarking analysis, in which LSTMs’ performance is contrasted with that of other popular nowcasting approaches.
What data are used in the model?
The nowcasts take as inputs numerous economy and trade-related variables, obtained from a variety of data sources. See the “Metadata” tab for more information on specific input variables. The variables were chosen from a large pool of over 200 potential explanatory variables via a process of recursive feature addition (RFA). See the documentation of the nowcast_lstm library, and more specifically the variable_selection function of that library, for more information on this process.
