In the Easter break it came again – new roads, dark forests and unknown people. I left to Sweden to visit my brother and Swedish-Latvians. This time the homework I took with me was to evaluate the natural (structural) and cyclical unemployment in Latvia. “What is the natural unemployment level and its dynamics in Latvia?” was the question I was asked to answer by using econometric techniques. One of the main sources used was Barlevy (2011) Evaluating the role of labor market mismatch in rising unemployment. In the homework the task was to create (1) the Beveridge Curve, (2) the matching function with Cobb-Douglas specification, and (3) the Phillips Curve. As the last two are really econometrics and most people are not interested in reading mathematical equations and their interpretation but prefer nice graphs instead, I’ll share with you the findings of the Beveridge Curve only.
The long story short – The Beveridge Curve shows how the proportion of unemployment and vacancies as well as the efficiency of matching both changes over time. There can be movement on the curve – the efficiency of employment does not change while changes the proportion of unemployment and vacancies, showing whether economy is on increasing or decreasing pattern. Or the curve itself can move – if the curve shifts towards the origin (0;0) or to the left, the efficiency of matching improves and both unemployment and number of vacancies decrease. This shows that the people needed in the labour market are actually out there and waiting to be employed. If the curve shifts to the right it means that companies are searching for employees with skills, knowledge or qualification different from those unemployed persons have.
In the pictures below two different data sets are used both starting at 2005 until 2014. The starting point of plot is with the unemployment around 11% and vacancies around 0,6 thousand from NVA* data and above 1,0 thousand from CSP** data. In years 2006 – 2007 the lowest unemployment and highest number of vacancies was reached. Afterwards unemployment increased again, but the curve shifted to the left, which means that the efficiency of matching both parameters increased. The highest unemployment and lowest level of vacancies was in year 2009.
However, there is no clear appreciation about the latest situation. From NVA data the efficiency of matching unemployment with vacancies has decreased (curve shifts back to the right) and we are approximately on the level as in 2005 , but from CSP data it seems that we have the same unemployment as in 2005 with more than twice as less vacancies, therefore, the efficiency of matching unemployment with vacancies has improved. Who has the right data? Personally, I do not know. The conclusion of the homework is that you should not believe in what you see when only one measurement is given! Also other methods I used confirmed that it is possible to get different results depending not only on data set used, but also the time period taken, the parameters chosen (econometricians will understand) etc.
For those still interested in the natural unemployment that I mention in the title of the this blog post but did not actually analyse here – it is possible to get whatever results again. I got natural unemployment in interval between 7 and 12% depending on time period, data set and parameters chosen. As well as division between structural (natural) and cyclical components were different in every model. The only result that was consistent for any model used was that the overall unemployment was below the natural unemployment level in years 2006 and 2007 (“Gold times”, economic expansion). Whether it was good or bad (knowing the low productivity of workforce at that time) is again a new question to be researched on. The work on this homework will be continued during April, and most likely updated presentation (that I first presented on April 8) will be published in my SlideShare profile on May. If you still can’t find it there, please Contact me (maybe I have forgotten to keep my promises)!
After having all Saturday by the computer, I went to Uppsala next day to clear my mind. I took many pictures but this particular is one of my favourites. Non-traditional forms of traditional things (chair and table in a park).
And this picture has the same message as homework. It is captured at some particular time and place and it looks like it is autumn there but it was an early spring (April 5). The same works in economics and econometrics – one model can still lead to contrary results depending on your personal point of view (and background knowledge).
Have a nice Spring 2015 and be careful with your judgements – do some sensitivity analysis and check if there are any stable results.
This also reminds me about “Fake it until you Make it“. Funny – we econometricians also are willing to use this approach from time to time.
* NVA – Nodarbinātības Valsts aģentūra – State Employment Agency
**CSP – Centrālās Statistikas pārvalde – Central Statistical Bureau of Latvia
You’re always welcome to leave a comment or send me a private message through the Contacts form!
Have a nice day!
View this post on Instagram
Hello Yellow, Spring is here! And sun is shining! Wanted to share with you my news on agneesze.wordpress.com Be #awesome! Stay #cool! And love #econometricians! The #Easter #break as all my breaks took me abroad, this time to #Sweden, #Uppsala. Again as always, I had great #leisure time but also did some #homework, #econometrics. This time about the unemployment in Latvia. Do you have any opinion about the #natural level of unemployment in Latvia? If you do, please share it with me, would be great to hear!