In the previous article, we argued that economists and analysts alike should give more attention to an unassuming component of GDP – the inventory. Despite its tiny size, merely 1.6% of total GDP, we showed that it does have a profound effects on both seasonality and swings, hence on all GDP growth forecasting exercise.
Inventory explains as high as 65% of seasonal swings in the total GDP – and this is just on average. In fact, the seasonal component usually reaches its peak in Q1 where inventory explains more than 50% of the movement, compared with tourism which accounts only 32%.
What this means is that one cannot forecast GDP growth accurately without forecasting inventory correctly. For example, in 2014, inventory contributed -2.4% to GDP growth. So, if a forecaster ignored it, he will come up with 3.3% GDP growth, assuming he could perfectly forecast other components. Actual GDP growth came in at 0.9%. Far off all forecasts, wasn’t it?
Thus, given its significance, we will investigate the behavior of inventory deeper with one chief question in mind. How could we explain and forecast the inventory?
Basics of Inventory movement
If we plot an inventory in a line chart, one could see that it is quite similar to wave. That is it has peaks and troughs at a regular interval. Moreover, it seems like the variance of the series increases over time. One could see that inventory run roughly between -50 billion to 50 billion baht in real term before 2009. But after 2009, the range doubled to -100 to 100 billion baht. This characteristic is actually understandable. Inventory should increase in accordance with the size of the economy. However, if we choose to divide inventory by the GDP, the ratio of inventory lies within -5% to 5% of the GDP throughout the sample. In other words, we could use the ratio of inventory for further analysis, which is more stable for both comprehension and forecasting exercises.
Going behind inventory movements
As inventory possesses strong seasonality, seasonal component should clearly be emphasized. According to our calculation, seasonality could explain 36% of the movement in inventory.
However, this simple calculation suggests that the ratio of inventory to GDP should be 1.9% and 2.1% in Q1 and Q4, respectively. This result is in contrast with the stylized facts that the peak is usually reached in Q1. Hence, there must be something missing!
So, what else could affect inventory? The first couple of candidates are imports and exports of goods. Normally, Thailand imports a large chunk of raw materials to produce goods for re-exports. Therefore, we should observe higher inventory during a quarter with high imports, and lower inventory with high exports.
Accordingly, we found significant impact from both exports and imports. When exports and imports increase by 1% quarter-on-quarter, ratio of inventory to GDP decreases and increases by 0.2 percentage point, respectively. The result is indeed in-line with expectation.
Another candidate is consumption. The rationale is as follow. Since it takes time to produce goods, when consumers consume more, part of inventory might be used up, thus reducing the inventory. Consequently, we divided consumption into 2 regimes – increasing and decreasing cycles. Then, we find the significant relationships!
During increasing consumption cycle, a 1% increase in consumption will decrease ratio of inventory by 0.4%, but increase by 0.4% during decreasing consumption cycle! Totally opposite outcomes! Weird, isn’t it?
What might explain these asymmetric responses of inventory to consumption? One possible explanation is that producers react too slowly to changing environment. For example, suppose the consumption just enters increasing period where there is a growing demand for goods. However, producers have not yet realized the change, thus they do not feel the need to accelerate their production. Once the demand accelerates, producers could only sell what are in their inventories. Thus, inventories decrease. The opposite logic could be applied for the case when consumption enters a recessionary period.
In addition, the effect of business sentiment on inventory supports our conjecture that it takes sometime before there is an adjustment to production. Naturally, if producers have good business sentiment, the companies should expect their sale to increase, thus building up this inventory is an appropriate preparation. This is very intuitive. Our analysis also supported the intuition. However, there is a huge trick. It is the sentiment in previous quarter sentiment which affect inventory, NOT the concurrent one! In other words, it takes time before sentiment affect inventory – again business takes time to adjust.
In summary, there are 2 implications we would like to emphasize. First, despite a consequential size, inventory could not be ignored during GDP forecast process. We also attempted to determine factors which might help forecasting the inventory. The second implication is on inventory management. Our analysis suggested that, on macro level, inventory respond to changing economic environment too slow. Admittedly, a lot more works should be done before any conclusive result could be reached. But one thing is certain, inventory moves GDP and our economy more than most people realize. And now you know. ¾
(Published in Bangkok Post on February 5, 2016)
TMB Analytics is the economic analysis unit of TMB Bank. Behind the Numbers is co-authored by Peerawat Samranchit and Benjarong Suwankiri. They can be reached at firstname.lastname@example.org