What factors make some districts rich and well-developing while others are poor and underdeveloped?

1. Introduction

One of the objectives that we set ourselves in the analysis of the socio-economic condition and development of Bulgarian districts was to look for an explanation of what makes some districts richer and more prosperous while others get poorer and underdeveloped. Having collected voluminous statistical data at the district level covering the period from 2000 to 2012, we decided that in the second edition of „Regional Profiles: Indicators of Development” we can attempt to complement the descriptive analysis with an analysis of the correlations between different indicators.

For this purpose, we constructed an econometric model based on the traditional linear production function: Q = aX₁+ bX₂ + cX₃ + dX₄ + … + zX_n, where a, b, c, d, and z are coefficients to be determined empirically by the specific model, Q is the aggregate production, and X_1, X₂, X₃, X₄ and X_n are the factors of production. Usually the factors of production include, as a must, indicators of labour (human capital) and physical capital, which can be supplemented by optional factors such as entrepreneurship, land, natural resources, and technology.

The choice of specific indicators to represent statistical approximations for input factors of production (labour, capital, land, resources, etc.) is made individually by each researcher according to the available data, their quality, and theoretical considerations about the weight of one factor or another. For example, when analysing a territory that is resource-poor, this factor could be excluded. On the contrary, when examining an area part or all of which is rich in any particular natural resource (e.g., fossil fuels), this factor should be included in the model.

2. Panel model for the socio-economic development of Bulgarian districts

2.1 Model

Given the available data, spatial and temporal scope of the study, the selected technique is panel analysis with a random-effects assessment. The period covered by the model is 2004-2010 because it is for this period that we have a complete time series available on both the dependent variable side and the independent variable side. All data included is annual.

2.2 Dependent variable and time period

The gross domestic product per capita has been selected as an aggregate indicator of the degree of development and welfare of districts in Bulgaria. It is the most appropriate measure of aggregate production, the datasets for which are available at the district level. It should be borne in mind that this indicator also has some drawbacks affecting the possibilities of use and further analysis. Firstly, there is a relatively large lag for published GDP data at the district level. At the time of preparing the model (August 2013) the latest data on GDP at national level referred to 2012 (albeit as preliminary estimates) while at the district level the data referred to 2010.

This two-year lag in the publication of district-level data determines the time period for which the model has been constructed, namely the period ending 2010. Therefore, when interpreting the results of the model, it should be considered that any conclusions refer only to this period of time and do not cover the last two years.

Another major shortcoming of district GDP data is that unlike GDP statistics at national level, the district series are calculated and published only in current prices, with no other available options. This feature is due to the absence of district price indices/deflators that could be used to deflate current-price data into constant prices referring to a selected base period. This drawback does not allow for the calculation of the real economic growth rate of separate districts.

2.3 Independent variables

On the side of independent variables, several indicators were tested for statistical significance with a view to using these indicators for the approximation of the main groups of factors of production:

2.3.1 Human capital

a) education - the assumption here is that a more highly educated workforce is more productive (ceteris paribus).

The following variables were tested for statistical significance: share of the population with tertiary education; population density; share of dropouts from primary and secondary education; number of teachers at primary and secondary education per 1,000 students.

b) healthcare - the assumption is that better healthcare and the availability of health professionals and hospitals have a positive effect on the working capacity of the population.

In this category, the following variables were tested for potential explanatory power: number of people per general practitioner; percentage of health insured persons; infant mortality rate.

c) demographics - in terms of demographic variables, we have built on the assumption that the large number of population and in particular the population in working age, is a factor in attracting investment and the development of local economy.

The following demographic variables were tested: rate of natural increase; net migration rate; age dependency ratio (65 to 15-64 yrs.).

2.3.2 Physical capital

Physical capital is generally understood in two ways. On the one hand, in the narrower sense of this term, "physical capital" means investment is the purchase of land, buildings, machinery, technology, etc. In a broader sense physical capital also includes factors that are external to the specific economic unit, such as infrastructure and networks - roads, airports, seaports, energy, water and sewage networks, the Internet, etc. It is reasonable to expect that investors would prefer to invest in areas that are characterized by better infrastructure availability (ceteris paribus). Based on these two aspects of physical capital, two groups of variables were tested in the model to describe the state of the physical capital: those related to investments and those related to infrastructure.

a) investments: expenditures for acquisition of fixed tangible assets (FTAs) per 1,000 population, foreign direct investment in non-financial enterprises (per 1,000 population)

b) infrastructure: road network density (in km per sq. km.); railway network density.

2.3.3 Entrepreneurship

The presence of entrepreneurial initiative undoubtedly contributes to investment and the development of districts. If we look at the entrepreneur as an inventor or entrepreneurship generally as a state of being "on the alert" for profitable opportunities, then it inevitably is a fundamental factor for growth, preceding labour and capital[1].

As a quantitative proxy of the concept of entrepreneurship, we used the number of non-financial enterprises per 1,000 of the population.

It is important to note that all listed indicators tested as potential explanatory variables in the model of production described above are part of the database of indicators of the socio-economic development of districts collected for the purposes of the entire study "Regional Profiles: Indicators of Development". We believe that the database of 58 different indicators in eight categories collected for the purpose of the study is comprehensive enough and actually includes most of the important data on the socio-economic development of Bulgarian districts. Each of the indicators included in the study was selected after a careful review of the available statistics at the district level, and a series of consultations with professionals in regional development within the project’s Advisory Board and Expert Roundtable.

However, several indicators collected for the purposes of the study, which could, a priori, have been included in the model as potential independent variables, were finally dropped due to the short statistical series or doubts about the quality of statistical information at district level. For example, data on EU funds absorbed by municipalities under operational programmes has been published since 2011, that is, although these funds are extremely important for the development of infrastructure in districts, the short time series does not allow for their use as explanatory variables.

3. Results

The model shows several important and statistically stable results which are in line with international research on the key factors affecting economic development. Based on the indicators examined, in the case of Bulgarian districts the statistically significant variables at a significance level of 5% include indicators of human capital (labour), physical capital (investment and infrastructure), and entrepreneurship. That is, the model confirms the validity of the production function used, namely that the development of any area is affected by the quality of human capital and physical capital available, and the activity of entrepreneurs as leading figures and drivers of progress.

In the constructed model of the development of districts, the dependent variable is the gross domestic product per capita (at current prices). It is important to mention that "income per household member" was also tested as a dependent variable, but the results were not statistically significant and sustainable. A possible explanation could be the fact that the total household income also includes income from pensions and social benefits, i.e. government transfers which do not depend on the performance of the local economy, but only on the government's political decisions.

Finally, the dependent variables for which statistical significance was proven at a significance level of 5% are as follows:

1. Relative share of the population aged 25-64 with tertiary education (percentage).
2. Expenditures for acquisition of fixed tangible assets (FTAs) per 1,000 population (in BGN).
3. Railway network density (km per sq. km. of the area).
4. Number of non-financial companies per 1,000 population.

The first indicator (the share of university graduates among the population aged 25-64 years) is beyond doubt a measure of the quality of human capital. Expectations are that a higher number of people with higher (tertiary) education will be directly proportional to the development of the districts. The main argument in support of this link is that a highly educated workforce as a whole is characterised by a relatively higher productivity. The model results for Bulgarian districts confirm these expectations. In case of an increase by one percentage point of the share of university graduates, the GDP per capita increases by about BGN 21.

It is worth noting that the positive relationship between the proportion of people with tertiary education and economic prosperity does not necessarily mean that, in order to develop, the districts need as many university graduates as possible. On the contrary, in many districts (especially those where industry is more important for the local economy) business representatives have complained of a shortage of well-trained professionals with specialised secondary education. The result of the model rather shows that, ceteris paribus, the greater percentage of people with tertiary education implies better quality and hence a higher productivity of human capital. It is a different matter altogether that, although unemployment among university graduates has traditionally been lowest[2], some of the employed individuals do not work in their specialty and/or occupy positions for which a lower level of education would be sufficient.

The expenditures for acquisition of fixed tangible assets per 1,000 people are a measure of the intensity of investments in the districts, i.e. the physical capital invested in local production. Not surprisingly, in districts with relatively larger investments, higher levels of development and income are also reported. The specific results of our model show that for an increase of BGN 1 in acquisition costs for fixed tangible assets (1,000 population) per year, the GDP per capita increases by BGN 28 for the same period.

Interestingly, out of the two measures of the degree of development of local infrastructure - the density of the road network and the density of the railway network - only the railway network density, which is more than four times lower than that of the national road network, proved to be statistically significant for the development of Bulgarian districts. One possible explanation for this somewhat surprising result lies in the characteristics of the indicators used, and in particular in the definition and scope of the "Road network density" indicator. This indicator includes national roads from the first, second and third class, and highways. There are no such roads in Sofia (capital city), that is, although the capital is the most prosperous of all districts in the country, it ranks at the bottom of the list on this indicator. Indeed, all roads of the national road network, including highways, reach the border of Sofia (capital city), but, in theory, do not cross it. Thus, even though the two highways whose construction was finalised in recent years ("Trakia", "Lyulin") and the "Hemus" highway, which is still under construction, start from Sofia or end in Sofia, from the administrative-territorial perspective they are not part of the capital city's road network.

What the econometric model showed is that a railway network density that is one linear kilometre per square kilometre higher causes an increase in GDP per capita of about BGN 270. The direct proportional relationship between the density of the railway network and economic prosperity comes as no surprise. The better connected a district is with the surrounding districts, the more attractive it becomes for investment, while job opportunities, including outside the district's territory, also increase.

Not least, the model confirms the key role of entrepreneurship for the local economy. The more companies are set up in any district, the more its chances for higher incomes and prosperity increase. The number of non-financial enterprises per 1,000 of the population has been used as an approximate indicator of entrepreneurship. The numbers (rounded to integer values) by district for 2011 ranged from 29 to 78 companies per 1,000 people. The results show that an increase in the number of non-financial companies by only one company per 1,000 people causes the per-capita GDP to increase by BGN 1,640. However, when interpreting this result we should take into account the scale of the explanatory variable, namely the number of non-financial enterprises per 1,000 of the population. With this variable, changes occur smoothly and at relatively small increments over the years. For example, the national average number of non-financial enterprises per 1,000 of the population increased from 27 to 50 for the period 2000-2011, while for individual districts the increase ranged from 9 to 46 for the same period.

As a whole, the results of the econometric model confirmed the importance of human, physical, and entrepreneurial capital for the economic development of any territory. The share of highly educated population, the level of investment, entrepreneurship and infrastructural connectivity of the territory stand out as important preconditions for prosperity: their variation explains close to three-fourths of the variations in welfare measured by means of the per capita GDP. Given that the presence of these factors provides a competitive advantage for districts and enhances their faster development, local policies should be geared, as a matter of priority, to a longer and better education of the workforce, retention of the highly educated population, simplified procedures and activities to attract investment, better infrastructure connectivity, and promoting entrepreneurship. This simple recipe for development and progress can be easily applied to all Bulgarian districts. The only condition is the will of the central and local governments.