One who wants to conduct a research on Turkey must definitely be aware of TURKSTAT (Turkish Statistical Institute). If you need data, that is the databank. The TURKSTAT was formerly called State Statistical Institute (SSI). But the European Union (EU) said that it was not state's duty to collect data and that there is need for a separate institution with relative autonomy, the TURKSTAT was established. Back then, the EU accession process was not crippled by the way. It was thought that the problem was resolved as the SSI was replaced with the TURKSTAT. The statistical institute was never a beloved one, neither as the SSI nor as the TURKSTAT. I currently am not contented with the TURKSTAT nowadays. They continuously change the definition of the data; assumptions change or coefficients are modified. Yes, they have an explanation. But it does not prevent the confusion; you get unable to carry out long term analysis. Today let me tell you why the labor force statistics upsets me.

Have you ever clicked on the "Labor Force Statistics" icon on TURKSTAT's website? If you do, comes the explanation that upset me. The first input that you receive before you see the data is a pop-up message. If you blocked pop-up messages, you do not see the explanation. For instance, I did not see the explanation at first. The explanation in question reads, do not even think of conducting time-series analysis because we started to revise the series and for now only the data for 2007, 2008 and 2009 have been revised. What about the previous years? The data for the previous years is not reliable for long-term analysis and they needed to be revised as well. When? We do not know. In fact at the heart of the problem lies the Address Based Population Registration System (ABPRS). The said system gave a population size 3.7 million people lower than the actual population. Thus, the TURKSTAT had to alter some coefficients used in the Household Budgetary Survey. The surveys are still carried out with approximately 13.000 households. But the weighed coefficients related with the size of population represented by one survey are revised on the basis of the ABPRS data. Therefore, you cannot compare the data for 2005 and 2006 with those for 2007 and the proceeding years. Thr TURKSTAT had changed the data analysis methodology also in 2005. So what is the issue here?

First let me stress what is not the issue here: What I am trying to say here is not arguing that the TURKSTAT has been manipulating the data. If it is possible to improve the data compilation methodology, not doing so would be wrong. Methodology can be changed with highly logical and scientific motivations. The critical point here is that it should first be announced why the change was done so that those analyzing Turkey's economy will not compare apples with oranges. This is the first point to stress. Second, after the change is made and announced to everyone, the revision of the data must not be extended over a long period of time. In fact, the methodological change had better be carried out together with long-term data revision so that the change will not be a surprise for anyone and that it is possible to conduct long-term analysis on Turkey's economy. The third point is about how the change will be made. It is necessary to tie methodological changes to a rule. When methods and coefficients shall be changed and the preceding and proceeding announcements with this regard must be written. The last point must go as follows:  it should be noted that changing methodology in the face of a crisis only confuses people more. In a milieu where everyone seeks to analyze the recovery process of the economy, what happens if you say "Oh, sorry. The assumptions of the dataset were outdated, so we have revised them. Meanwhile, you cannot compare the new data with the old data." What happens is, everyone tries to assess the impacts of the crisis on Turkey's economy on the basis of a dataset they are not completely accustomed to. So, people get confused further.

Let me conclude with an example. Female labor force participation rate has been increasing since 2008. Why? First, the ABPRS might have a direct effect on this, but I do not know why. Second, the share of agricultural employment in total employment has tended upwards in the recent period. Participation of women into the labor force rate in Turkey is higher in agricultural sector; as unpaid family worker. The constant drop in female labor force participation since 1988 was mainly brought by migration from rural to urban areas. Then either people are migrating back to the rural or the coefficient change in the ABPRS has a role in this. The third point would be the following: with the effect of the crisis, real wage per unit has been decreasing since 2008 in Turkey. In this context, to secure a certain level of income flow, one extra family member must work in each household. This might have triggered the participation of women into the labor force.

Which one of the following do you think is correct? A, B or C, all of the above (D) or none of the above (E)?

Is it good if a statistical institute raises more questions than it answers? It certainly is bad.

This commentary was published in Referans daily on 27.07.2010