Data on capacity utilization ratio (CUR) announced for the last couple of months are quite surprising. It goes like this:  CUR for a certain month is announced a month before the industrial production figures for that month is announced. As the declaration made in November suggests; CUR for October 2009 dropped by 4.9 points compared to the same period the year before. However, October industrial production for October, as announced last week, suggests that industrial production rose by 6.5 percent compared to the same period the year before.

The CUR pertains to capacity utilization in industrial sector. Therefore, if CUR drops as significantly as by 4.9 points, you expect a similar drop in industrial production as well. Of course CUR is an ambiguous concept for certain sectors. Nonetheless, note that capacity utilization rate fell by 4.9 points while industrial production increased by 6.5 percent. If the difference was small, for instance if the former dropped by 0.5 points while the latter rose by 1 percent, we could have ignored this considering the mentioned ambiguity.

The question is why is there such a large difference between CUR and industrial sector growth? A friend of mine came up with a possible explanation for this inconsistency: To derive a single capacity utilization rate for one sector, capacity utilization rates derived from annual data declared by individual firms are weighted. If you take a short tour at TURKSTAT's website, you see that production levels for the year before are used to weight firms.

This is a problematic method. I prepared a simple example to ensure better understanding of the phenomena. Assume that there are three factories in one certain industry. Table 1 gives production levels of these hypothetical factories for 2008. The next column gives their share in production which indicates the weights to be used to calculate CUR for that firm. Other three columns show hypothetical capacity utilization rates for different months in 2009 for each firm. The last raw is quite important for us as it gives the overall CUR for the industry composed of three factories.

The point here is that the first factory did not hold production in 2008 and thus is weighted as zero when calculating the CUR for 2009 regardless from the level of its production in 2009. This way, CUR of the industry for January is calculated as 45.7. However, in the coming months economy recovers, production level increases and thus the factory one starts production. CUR of this factory reaches 50 percent in March and 100 percent in October. Nonetheless, CUR of the industry does not change at all as the other two firms produce with the same CUR throughout the year. What is more, factory one has no significance for the index as its weight is considered zero.

Is this possible? Assume that the first factory only produces for exports and thus suspends production as export demand came to a halt with the crisis. But then export demand recovers slowly and the factory starts producing. However, under the current method of calculation, that production and the resultant rise in capacity utilization rate are not taken into account. So, can we employ another method? Of course we can. Table 2 suggests an alternative calculation method: When calculating the weight of each firm, the share of the firm in total industrial production is calculated as if the firm operates under full-capacity.

This is a more precise method. In such case, CUR of the industry proves much different than that given in Table 1. There are two main differences. First, in January where industrial production is low, CUR in the alternative method is lower than that in the current method. Second, as the first factory steps in, the rise in production and capacity utilization in March and October is easily reflected in the index under the alternative method.

A warning: Colleagues at TEPAV called TURKSTAT and asked the weighting method they employed. They channeled us to TURKSTAT website. What I refer to as the 'current method' is what I 'understood' from the explanations in the website. I might have been mistaken; I would be glad to correct it one I receive a comment.

Table 1: CUR (%) under the current method

Table 2: CUR (%) under the second method

This commentary was published in Radikal daily on 14.12.2009