Estimates of GDP growth are published with a considerable lag – even in some major economies we still only have partial data on what GDP growth was in Q1 2018. So ‘nowcasting’ GDP using more timely indicators of economic activity is an important way of assessing the strength of the world economy in real time. Good indicators are timely, correlated with measures of world activity and should outperform simple benchmarks. Unlike other global indicators such as business surveys or trade data, metals prices are available minute by minute. They also tend to move closely with world GDP. This post assesses how well they perform at nowcasting world GDP.
Why nowcast world GDP?
Assessing the strength of the world GDP in real time is difficult. Official GDP releases are generally published with a significant lag. For example, having entered summer, we are now the third quarter of the year. But won’t have any estimates for growth in Q2 in any economy until the end of this month. And getting a full set of global data will take even longer (we still only have partial data for GDP growth in Q1 this year). And even these are early estimates are based on incomplete data and subject to significant revision. More timely indicators of activity may also help to identify turning points in the world economy before they are reflected in GDP statistics. So using other indicators is crucial to assess what is happening in the world economy in real time or ‘nowcast’ economic activity.
Metals prices are highly correlated with world activity…
‘Nowcasting’ GDP is very common across professional economists, and involves using indicators to estimate GDP growth in advance of official estimates. There are lots of examples of nowcasting techniques for individual economies, such as Reichlin, the NY Fed tracker and Fulcrum. And global indicators can also provide a signal of the strength of the world economy in aggregate. Although rarely used to nowcast individual country’s GDP growth, metals prices are widely considered to be a good indicator of the strength of the world economy as a whole. They are significantly more time than other indicators such as business surveys or global trade data. And as shown in Chart 1, metals consumption moves very closely with world GDP growth. As a result metals prices are highly correlated with indicators of world economic activity. The red and blue lines in Chart 2 show a high correlation between the S&P GSCI Industrial Metals price index and world GDP and trade. The orange and grey lines show that this correlation is high across both advanced and emerging market economies. As a result, a number of studies have found that world GDP can explain a significant proportion of movements in commodity prices. That’s not surprising – commodities account for around a quarter of world good exports, so would be expected to be strong when world trade is growing strongly. Metals prices rose 30% over 2017, reflecting the continued and surprising strength of the global economy, which grew by its fastest rate since 2011.
Chart 1: Global metals consumption vs global GDP growth
Sources: Bloomberg, Thomson Reuters and Author calculations
Chart 2: 3 year rolling correlation between metals prices and world activity measures
Sources: Thomson Reuters, Author Calculations
That isn’t to say global demand is always the key driver of metals prices. As you can see in Chart 2, there are times when the correlation between growth and metals prices falls quite substantially (and even turns negative). That’s because other factors can also be important. For example, movements in the dollar can have a significant impact on metals prices. Commodities are traded in dollars, but the vast majority of consumption and production occurs outside the US. So metals prices have a strong negative correlation with changes in the dollar. Country-specific factors can drive prices as well. China accounts for around 50% of global consumption of many metals, and is also a significant producer of commodities like copper and aluminium. So news to either supply or demand in China can have a material impact on prices. For example plant closures relating to environmental policy last year pushed up on metals prices. Other idiosyncratic factors include the effect of supply disruptions which can have a significant impact on prices. A example includes the recent US sanctions on Russia, which cut off significant aluminium supply from the global market and boosted prices considerably.
… and perform well at predicting world GDP in the near-term
So we know that metals prices can be a good indicator of global GDP growth. But how well do they perform as a predictor? To test that I have used a simple model which tries to predict GDP growth in the current quarter using monthly changes in metals prices. The model regresses quarterly world GDP growth on one lag of GDP growth (an AR term) and monthly changes in metals prices in each month of the quarter (so there are different models for the first, second and third month of the quarter). You can see the performance of this model in Chart 3 below. This shows the out of sample nowcast for world GDP based on movements in metals prices over the same quarter. To test how well the model performs I calculate the root mean square error (RMSE), a typical metric for assessing forecast performance and compare it to a simple benchmark AR (1) model, which is common in the literature. A simple autoregressive model, in which GDP growth depends linearly on its previous values, is often used as benchmark to assess model performance. The model including metals prices performs well, and has a RMSE of around 0.4 since 2003, outperforming an AR model (which has a RMSE of 0.5). This model also outperforms AR models one quarter ahead, but not beyond that horizon. It’s worth noting that the AR term is still useful, for example without it the nowcast from this model under-predicts growth in 2014/15 (when the correlation between metals and world GDP falls in chart 2). World nowcasting models that include more variables, such as surveys of activity or measures of world trade, outperform this simple metals model. But metals prices have the advantage of being incredibly timely – available at a minute-by-minute frequency.
It’s also striking how well metals prices predicted the surprise increase in global GDP growth in early 2016. Few forecasters at the time expected a rise in global growth, but the blue line picks up quite sharply in early 2016 – indeed the IMF’s April 2016 WEO was rather ominously called ‘Too Slow for Too Long’ referencing the disappointing pace of growth at the time. So it’s likely that the near 12% rise in metals prices in the first half of 2016 was a good signal that global growth was picking up, before we saw it in the official GDP statistics.
Chart 3: Metals prices nowcast vs world GDP
Sources: Thomson Reuter, Author Calculations
Metals prices are timely, highly correlated with world economic activity and perform well at predicting short-term movements in GDP. And unlike a number of other indicators, they are available at incredibly high frequency in real time. So, despite also responding to other factors, they are a robust nowcasting indicator and corroborated the strength in the global economy over 2017.
Tom Wise works in the Bank’s Global Analysis Division
If you want to get in touch, please email us at email@example.com or leave a comment below.
Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.