Previous Main Table of Content Print PDF Next


Juris Ozols, Dr.phys
School of Business Administration Turība, Latvia

Oskars Onževs,
School of Business Administration Turība, Latvia

Elita Ardava
Stradins University, Latvia

The work aims to find the metrics of the econometric parameters, which could be used for the building of multi-purpose models and also for solving problems of forecasting. The paper explores the choice and usage of a definite set of parameters for the forecasting of the future development of regional pharmacies. Regional pharmacies, situated in distant areas, are significant  representatives of the real economic situation and living standards of the inhabitants. The peculiarity of pharmacies is their location in rather closed areas; clients consist mainly of local people. Consuming of medicines depends on the purchasing power in the appropriate region. Turnover of a pharmacy is affected by the location of the pharmacy and by the economic situation in the region. The main research methods are: the turnover analysis of pharmacies using the database of enterprises, statistical processing of data with pivot methods, the statistical analysis of the dynamic parameters of time series and the regression analysis. The research shows the possibility to perform forecasting of the turnover based on a such parameters as the turnover of a particular regional pharmacy, changes of the pharmaceutical market for the whole country and dynamics of GDP per capita in Purchasing Power Standards in the country as a whole. Common algorithm and approach is used. Research was performed using data from selected pharmacies and Lursoft database of enterprises.

Key words: regional, pharmacies, turnover forecast, Gross Domestic Product.


Small and micro business research is of particular interest under the current economic conditions. The 23 million SMEs in the EU represent 99% of businesses, and are a key driver for economic growth, employment and social integration. They are the true back-bone of the European economy, being primarily responsible for wealth and economic growth According to the analysis, 85% of new jobs in the EU between 2002 and 2010 were created by small and medium sized enterprises. (European Comission, 2012).

Due to their nature (mostly regional companies, important job providers) these enterprises and their survivability and development capabilities are an important factor in the national economy. The turnover and turnover dynamics of small and micro enterprises reflect the turnover trends in the country, as well as the purchasing power of inhabitants. Therefore, it is important to forecast the possible turnover and the consequent profit and future perspectives. Business analysts and forecasters need to focus on the key business indicators that impact turnover, and forecasting is regarded as an important factor for the manager to make decisions on the development of the company (DeLurgio, 1998), (Hoshmand, 2010). However, the forecasting should not be too costly and complicated (Gilliland, 2010).

The relationship between the turnover of a small enterprise (regional pharmacy) and the gross domestic product (GDP) per capita in Purchasing Power Standards was studied during the previous years. Based on a tight correlation of these data, a model was developed for the forecasting of the regional pharmacy turnover for 2010. The potential 2010th annual GDP dynamics scenarios were taken into account and turnover was predicted in accordance with each scenario. The forecasted GDP change level for 2010 compared to 2009 was -0.07%. According to these forecasts, pharmacy expected turnover was estimated at 80 thousand Ls which corresponded very well to the actual turnover of 79 679 Ls (Ardava, Onževs, & Ozols, 2011).

Note that the pharmacy activities are characterized by a fixed profit and turnover clearly determines the pharmacy profits and the necessary production and sales costs. In the enterprise data sometimes the production and sales costs are not identified, and they were not used as initial for the study. From the customer side the turnover is characterized by the purchasing power of vitally necessary goods and services (medicine). The obtained turnover forecasting results for a separate enterprise led initiative to further research.

Previous research showed that models of pharmacy operation should be developed for further analysis (Ardava E., Onževs O., Namatēvs I., Ozoliņa V., 2010);(Ardava E., Onževs O., Vīksne I., Namatēvs I. 2010). Pharmacies are typical representatives of small businesses whose business and operating models are better identifiable in contrast to other sectors of small businesses. The experience acquired will help to develop models for other small businesses.

The research object is approximately 400 individual pharmacies all over the country. To determine the sample size of the research a 95% significance level was assumed. The standard deviation of the rate of turnover changes of individual pharmacies was estimated (s=10%). Evaluating the necessary confidence level of the turnover changes rate less than ±4%, the appropriate sample size 24 was estimated (Bartlett, J. E., Kotrlik, J. W., Higgins, C.C., 2001).

As a representative model of the small and micro-enterprises 30 individual pharmacies in different regions of the country - both urban and remote rural areas - were selected.To obtain a sufficiently long-term data for the turnover investigations, only the pharmacies which operated continuously from 2004 by 2010 were chosen. As the most efficient pharmacy operation characteristics the annual 2005 by 2010 turnover and its dynamics was used (Fig. 1).Pharmacy turnover data, without specifying them individually, were obtained from Lursoft business registry databases.

Other authors' studies have shown that half of the Latvian pharmacies (400) belong to pharmacy chains. The number of separate pharmacies in these chains varies in time (Igaune, 2010). The rapid increase in turnover of some pharmacy chains can be explained by the increase in the number of pharmacies. Pharmacy chains cannot be considered as small enterprises either; therefore, their total turnover data analysis is not included in this study (Fig. 2).

Figure 1: Turnover of the individual pharmacies in the 2005 -2010.

The annual turnover and its dynamics of individual pharmacies was compared with the 2005 turnover, which is considered as the stability and reference year. The turnover increased up to the 2008 and the crisis followed in  2008 - 2010 (Fig. 1).

Figure 2: Turnover of some pharmacy chains in 2005 - 2010.

The primary purpose of the research done was to classify data on the individual pharmacy turnover and its dynamics for further development of the models of the turnover forecast. Direct turnover figures in absolute terms (thousands of Ls) are not suitable for model building since the absolute values of the observed turnover are very different (from 70 thousand Ls up to 1 mln. Ls). Furthermore, absolute changes in turnover are growing increasingly with each subsequent year in relation to the reference year 2005.

Figure 3: Turnover of individual pharmacies and turnover of medicines compared with the 2005.

To perform smoothing and inter-comparison of the absolute turnover data the turnover percentage change curves in comparison to the reference year were obtained. These curves clearly demonstrate the turnover trends independently of each pharmacy absolute turnover. Using State Agency of Medicines of Latvia information (Medicines consumption statistics) in the pharmaceutical market volume percentage trends relation to the reference year were calculated (Fig. 3). It appears that the medicine market trends well coincide with the turnover trends of individual pharmacies. However, such a data form is also badly applicable to develop forecast models. The turnover change curves pass from one point (2005th years) and with each subsequent year the changes become increasingly larger.

The outpilot studies show that the forecasting model of the development fits well  only the sales data of the pharmacies, which changes of the turnover are routine and there are no very rapid irregular changes in the turnover related to the closure of a pharmacy for a period of reconstruction, new outlets opening, or even acquisition of another pharmacy. Such "irregular" pharmacy turnover does not meet the general trend and can be predicted only by individual methods, and their data should be excluded from the output data set.

Turnover trends are best seen as a percentage of an annual turnover compared with the previous year turnover(Fig.4). This representation is suitable for forecasting because the annual increases are quite smoothly for all the years from 2006 by 2010.

Figure 4: Turnover of individual pharmacies compared with previous year.

To find the statistical regularities of distribution of the increase in turnover the average annual turnover growth and the annual standard deviations of the rate of turnover changes were calculated (Fig. 5).

Figure 5: The average changes of the turnover of individual pharmacies, the turnover standard deviation ±s field, the changes of the total turnover of medicines and the changes of the GDP, compared with the previous year

Here the grey field represents the average value of the increase of turnover with its standard deviations ±s, resulting in the so-called s field. The graph shows the annual national medicine market trends, which fits well the selected pharmacy turnover changes. When comparing Fig. 4 and Fig. 5, it can be seen that the majority of pharmacy turnover curves fit into the ±s variations in the grey field. Pharmacy turnover changes were found close to the normal distribution.

A pilot study was conducted to predict the possibility of forecasting the percentage increase in pharmacies turnover for the next year on the basis of the data from the previous years (2005th - 2010). The initial data set for modelling turnover changes was as follows:

When using the 2005 - 2010 data as the data source, the whole set of initial data can be obtained only for the years 2006 - 2009 (T+1of 2009 is equal to T0of 2010) and the resulting forecasting model covers four years of data.

The turnover increase model was created as next year's turnover growth forecast T+1F as a linear combination of T0, T-1, GDP-1, GDP0, GDP+1. In order to actually forecast changes in the turnover T+1F, the predicted value of the possible GDP changes must be  taken into account instead of the GDP+1.

Figure 6: Comparison of the real pharmacy turnover changes T+1 and turnover changes T+1F, calculated in accordance with the model.

When checking the developed model, calculated turnover changes T+1F were compared with the fair previously known turnover changes T+1 (Fig. 6).  Correlation between the estimated T+1F and turnover changes fair value T+1 showed a good congruence match (regression coefficient square value R2 = 0.746).

The chart (Fig. 6). shows three explicit fields expressing the relationship between T+1Fand T+1, corresponding to the three distinct turnover change T+1Fvalues. The drawn trend line is nearly at 450 angle, which also characterizes a good proportionality. It appears that some points "fall out" of the linear relationship; they are turnover figures of the irregular pharmacies data, which must be excluded from the data set.

Discussion and conclusions

The research shows that pharmacy model can be successfully synthesized, taking into account the time series parameters of the turnover, GDP and medicine market. The pharmacy turnover has a pronounced correlation with the GDP changes in the previous year.

It is shown that the metrics of the econometric parameters should contain dimensionless characteristics of the turnover, GDP and medicine market (compared with the previous year) taking into account the current, previous and next values (for example, forecasted values of GDP). Such a metric enables to balance various economic parameters and significantly different parameters of pharmacies. These claims were tested by linear pilot model. The established metrics allows to build a variety of target models using, for example, Group Method of Data Handling (GMDH) algorithms (Madala H.R., Ivakhnenko A.G., 1994). In the future the division of pharmacies into clusters depending on the specific turnover dynamics tendencies can be performed.

The authors express a special gratitude for consultations and aid to Lursoft and Lursoft Information Department Head Ģirts Ķēberis.


Ardava E., Onževs O., Namatēvs I., Ozoliņa V. (2010), The Research of the Main Factors at the Pharmaceutical Care for Latvian Individual Pharmacies. Seventh Annual International Scientific Conference Human Potential Development: Search for Opportunities in the New EU States. [on CD 12 pp.] Vilnius, Lithuania, June 2-3. ISBN 978-9955-19-181-0

Ardava E., Onževs O., Vīksne I., Namatēvs I. (2010), Empowerment of Electronic Prescription for Latvian Individual Pharmacies Entailed with Innovation Initiation. Innovation Driven Entrepreneurship. International Scientific Conference BMRA2010. [on CD] Vilnius, Lithuania, October 14-16th. ISSN 2029-5448

Ardava E., Ozols J., Onževs O. (2011), Regional pharmacy turnover forecasting feasibility  study. XII International Scientific Conference. "New values for tourism and community development". 2-3rd June. Riga. School of Business Administration Turība.

Bartlett, J. E., Kotrlik, J. W., Higgins, C.C. (2001), Organizational Research: Determining AppropriateSample size in Survey Research.Informational Technology, Learning and Performance Journal, Vol. 19, No 1, Spring 2001.

DeLurgio, Stephen A. (1998), Forecasting Principles and Applications. Boston, Massachusets: McGraw Hill, 804 p.

European Comission(2012).Small and medium-sized enterprises (SMEs). (02.02.2012). Retrieved 02.06.2012, from

GDP (purchasing power parity). (n.d.). Retrieved 12 15, 2011, from The World Factbook:

Gilliland, M. (2010). The Business Forecasting Deal. New Jersey: John Wiley & Sons, Inc., Hoboken, 252 p.

Hoshmand, A. (2010). Business Forecasting: a practical approach. New York: Routledge, 372 p.

Igaune, S. (2010, 3 26). Puse no Latvijas aptiekām pieder aptieku ķēdēm. Retrieved 12 14, 2011, from

Madala H.R., Ivakhnenko A.G. (1994), Inductive Learning Algorithms for Complex System Modeling, London, Tokyo: CRC Press.

Zāļu patēriņa statistika. (n.d.). Retrieved 12 14, 2011, rom Zāļu valsts aģentūra:

Previous Ineta Luka, Kristiana Talente, Valerija Malavska, Vaidesvarans Sundars
          Dalia Perkumienė, Danutė Kleinienė