Privatisation of socially owned enterprises, the methods used and the impact on economic growth: empirical evidence from Kosovo

Abstract Privatisation together with the related social consequences and impact on the economy represent key challenges facing the former communist countries. This paper aims to assess how the privatisation of socially owned enterprises (SOEs) affects economic growth, entailing an empirical test using a panel effects regression analysis on a sample of 571 SOEs (or 1,600 assets) over a 16-year period (2003–2018). We find that privatisation at the aggregate level does not boost economic growth; in particular, the methods used to privatise SOEs or parts of them are not a determining factor. We also show that the quality of institutions is fragile, confirming a negative associations with economic growth. We also show that the effects of privatisation vary according to the method used, although we note that the sale of SOEs or parts thereof in the first decade of privatisation has been quite selective, devoid of development effects and faced with serious impediments to privatisation funds being directly invested in the economy.


Introduction
The late 1980s signalled the inevitable collapse of the communist system. Its downfall forced most former communist countries to introduce substantial reforms, including economic ones, build state institutions and a host of other governance and social reforms. Moreover, socially owned enterprises (SOEs) also underwent substantial changes in order to boost economic growth (Dolenc, 2010;Jefferson et al., 2000). Measuring the impact of privatisation on economic growth is challenging due to various economic and social factors and the improper implementation of the privatisation model. So far, few empirical studies have measured the direct impact of privatization on economic growth, and Kosovo is among those countries that require further investigation. The privatisation in Kosovo does not seem to be the best, but as other countries apply certain models to transform socially-owned property, the comparison becomes even more difficult. Therefore, the purpose of this research is to provide an initial analysis of the transformation of SOEs; respectively, it assesses whether the model used to privatise the SOEs portfolio is appropriate and ensures a positive association with the economic growth of Kosovo.
Theoretically, economic growth is driven by a series of inputs, labour, capital, total productivity, and several auxiliary variables (e.g. institutional reforms) which, in turn, corresponds to the production function (Bajra et al., 2020;Buterin et al., 2017). While growth is most likely to be nourished by different source channels such as trade, investment, social progress, human development, on the other hand, privatisation is also expected to have significant effects on economic growth (Estrin & Pelletier, 2018;Maw, 2002;Skufli c et al., 2013;Vuk si c, 2016). Accordingly, critics have often pointed to the dilemma that while several countries encounters allegations of unfair privatisation of social assets (Megginson & Sutter, 2006), in Kosovo counterintuitive processes are underway, calling for in-depth research on what has been achieved over a longer period, namely whether privatisation is linked to economic growth. Also the efficiency of institutions (i.e. KTA/UNMIK and PAK/Kosovo) plays a key role in ensuring a strict and orderly privatisation, therefore, the period of asset management also serves as a proxy of whether institutions have acted responsibly with the socially owned property, and thus whether such an institutional approach is associated with growth (M endez & Sep ulveda, 2006).
The research aims to contribute to literature in several ways. Initially, using the panel regression analysis, the study not only assesses the transformation of SOEs but reveals the effect of the extent of privatisation in Kosovo. It also measures the effectiveness of the methods used, which is newer than what is most commonly used in other countries. Five privatisation methods are used, yet it is still a common concern to determine which method is more suitable and allows better organisation of the ownership transformation. The extent of privatisation has never been directly measured, thus the net effect remains insufficiently addressed. One reason is that the focus has been given only to the evocative aspects, almost entirely bypassing empirical analysis. Since different regimes have administered SOEs in Kosovo, it is expected that application of the methodology will vary. In addition to methodology, this study affirms the assessment that the administration of SOEs is straightforward, effective in adopting the procedures and rules via which privatisation is to occur (sales regulations, asset-selection procedures, offers, prices, purchase approval, etc.). The paper also is particularly valuable by considering the quality of institutions within the economic cycle. On the other hand, credible institutions do not necessarily mean that appropriate models of privatisation are introduced and therefore our research reveals whether the government is on the side of increasing growth, fighting corruption and remaining accountable to its citizens for ensuring economic development.
The rest of the paper is structured as follows. Section 2 summarises the literature review and the concept of SOEs. Section 3 shows the development of the hypothesis, while Section 4 presents the research methodology, including selection of the sample and details. Section 5 presents the results and Section 6 concludes.

Theoretical background
The socially owned enterprise concept largely arises from the former Soviet Union, but is otherwise identified as a legal entity operating wholly within the hands of the state, with all of its governing bodies, finances and other strategic matters being under state control, whereas only operational matters are entrusted to the enterprise (rights and obligations vis-a-vis suppliers and other parties, etc.). Apart from the transfer of ownership, privatisation implies the redistribution of social assets to private entities. The outcome impacts current owners and employees in many ways, allowing some to gain something of value while others almost nothing (Upchurch & Marinkovi c, 2011).
There is little empirical evidence to support that privatisation is significantly linked to economic growth, but on the other hand, there are many investigations criticizing the way in which the privatisation of SOEs has been pursued (Altshuler, 2001;Estrin et al., 2009;Iwasaki & Ko cenda, 2017;Lou zek, 2005). Alongside the theoretical aspect, very controversial practices may be observed in particular Central and Southeast European countries where the privatisation process has been criticised for irregularities, for being accompanied by corruption and non-transparency and having little impact on economic growth. In short, most scholars state that privatisation was ineffective and failed to drive growth, creating some economic and social consequences ( Cu ckovi c, 1993; Goldstein, 1997;Gray & Gray, 1996;Pra snikar et al., 2012;Shukarov, 2012;Simoneti et al., 2005;Uvalic, 2008;Vuja ci c & Vuja ci c, 2011).
Likewise, the literature does not show whether the reform of Kosovo's economy, namely the transfer of SOEs or parts thereof, is linked with economic growth. The privatisation is incomparable with other developing countries due to specific political and economic factors. The development of enterprises in Kosovo (mainly socially owned) has gone through several periods which we will not elaborate on here, but starting in the 1970s and in the late 1980s the entire economic system was centralised by the former communist regime (former Yugoslavia) 1 . With the collapse of communism, many SOEs established for specific markets were left without any need to produce, making their restructuring and privatisation inevitable (Grasten & Uberti, 2017).
Several methods have been used for the privatisation of SOEs or parts thereof. Commercial leasing (CL) is the first method used by UNMIK, which selectively enables certain private entrepreneurs to rent SOEs for a period of 10 years. Moreover, application of this method gives the tenant a pre-purchase right, which later proved to be an ineffective method of transfer of ownership since in almost no case was the ultimate goal achieved, raising suspicions of individual abuses and benefits (Korovilas, 2006). With respect to an ordinary spin-off (OSO), it is the most preferred method for transferring SOEs considered to have sustainable business activities. It operates by establishing a new company (New Co), and then transferring the main SOE asset to the New Co, which is subsequently sold as a new company with the assets of the old enterprise (i.e. the SOE) 2 . The next method is a special spin-off (SSO), which is more complex than an OSO, imposing several additional commitments while owning SOEs (e.g. investment in equipment, employment and holding current business activities). Liquidation is the last method and is used to sell only assets or parts of SOEs not sold entirely with the above methods. In addition, leasing and sale via direct negotiations are two complementary methods that have been applied in the last decade. The leasing method primarily includes assets that have not been sold and for which the parties have an interest in temporary use. In contrast, the direct negotiation method is applied to those assets that for objective reasons cannot be sold using any of the above methods (e.g. non-access to roads, construction of private housing on SOE land, etc.).
As the purpose of privatisation is a broad concept, our study is limited to the privatisation of Kosovo's SOEs, primarily focussing on how this has impacted economic growth

Hypothesis development
Often referred to as the economic cycle, economic growth is an abstract concept constantly influenced by several factors like purchasing power parity, namely income level, return on equity, capital stock, etc. Although many scholars argue various factors that monotonically impact economic growth, including foreign investment, trade (i.e. export), level of education, new technology, institutional reforms, and so on (Bajra et al., 2020;Bermejo Carbonell & Werner, 2018;Buterin et al., 2017;Hayat, 2019;Krueger & Lindahl, 2001), on the other hand, there are rare investigations that have proven whether privatisation is linked to economic growth. The need to covering privatisation-related corruption activities is another reason that prior research does not focus sufficiently on the empirical aspects of privatisation. Therefore, it is today widely discussed that privatisation methods are not that effective and are associated with many irregularities, increasing the perception of corruption and bringing the systems into question (Mencinger, 1996;Zhang, 2006;Z ıdek, 2016). For example, Gr€ undler and Potrafke (2019) and Silajdzic and Mehic (2016) claim that corruption activities reduce buyers' demand (i.e. direct investment), whose consequences are also related to economic growth. Sharma and Mitra (2019) argue that weak state institutions adversely affect certain economic segments and phenomena. Also, it stimulates unfair competition and favours policy-related individuals who expect to take a portion of the social and public property (Nguyen et al., 2018).
So, while the economic cycle represents the economy's natural fluctuations between periods, such as GDP growth, it is still important to discover the direct effects of privatisation on economic growth (Goldstein, 1997;Gray & Gray, 1996;Zhang, 2006). Western countries do not have much experience with the transfer of social property (Hamm et al., 2012;Starr, 1988), although quite different situations are encountered by most post-communist countries. In developed countries, the debate has concentrated on privatising public sectors such as health, education, electricity (distribution), mobile telephony, etc. (Bone, 2013;Levine, 2001;Marcelin & Mathur, 2015), while former communist countries are still faced with the need to restructure SOEs to make them more attractive to investors ( Cu ckovi c, 1993). Kosovo is such a story and, despite it relying on a diversified privatisation methodology, the effects of its privatisation on the economy have not been sufficiently measured. Therefore, at this early stage we claim that the transfer of property rights is an inexplicable process.
Kosovo's economy transition was opaque and meaningless, with many factories and enterprises having closed, lacking investment in equipment and technology, being faced with lost markets, etc. (Z ıdek, 2016). This led to a decline in SOE productivity and low economic output, the consequences of which are still evident today.
Total funds from the sale of SOEs and their assets amounted to almost 10% of Kosovo's GDP in 2018, although some of these funds remain frozen and are inactive. Despite this, there is no empirical evidence showing the extent of privatisation's impact on Kosovo's economic development and the level of success privatisation may be said to have accomplished from the current perspective. When it comes to increasing social and economic well-being, the evidence gives rise to conflicting arguments. For example, Knudsen (2008) claimed that the economic situation did not improve after several years of internationally managed privatisation. To assess this, the study begins by considering the production function to help explain the patterns of transfer of SOEs and their impact on the Kosovo economy (Faro, 2013;Yang et al., 2015). Using this rationale, we posit the following comprehensive hypothesis: H1: The privatisation of SOEs is positively associated with economic growth.
In contrast to the above hypothesis, the methods used to transfer SOEs also require separate treatment from each other. The focus of the SSO method is to transfer those SOEs which were larger and employed huge numbers of workers. In addition, extra conditions were imposed by first-time buyers (e.g. investing in property, plant and equipment), which are thought to have had a significant effect on the economic cycle. Therefore, to understand the effect of each method, the testing of separate hypotheses is inevitable. The volume of sales implemented by SSO is not large (around 20 SOEs) and must thus refer to a small number of SOEs, but the proceeds from using this method are relatively high. The SSO was applied until 2008, respectively up to 2012, facilitating the sale of the most vital SOEs (15 were sold under the UNMIK mandate). With this reasoning, we posit the following hypothesis: H1a: The privatisation of SOEs or parts thereof through the special spin-off method is positively linked with economic growth.
In addition to SSO, the second major method used to transfer SOEs is an ordinary spin off. With to this method, the new company (New Co) is initially established and then the assets of the SOE are transferred to the New Co., but not the SOE itself. Thus, only the New Co. is sold along with the assets transferred from the SOE. The transfer in terms of physical sales of assets using OSO is high as is the revenue from the transfer. The method has been in use for over 14 years in which time over 465 SOEs have been privatised, with about a 13% share of the total proceeds over the years. Consequently, the impact on economic growth is expected to be significant. Therefore, we posit our hypothesis that: H1b: The privatisation of SOEs or parts thereof through the ordinary spin-off method is positively linked with economic growth.
No less important is the LIQ method whereby only a few items of SOEs are sold, but not the SOE as a whole. Application of this method has resulted in large-scale sales, although the liquidation proceeds are not very big compared to the two methods mentioned above. Currently, 76% of property sales is made through this method, averaging 18% of total privatisation proceeds over the 16-year period. This liquidation method has been used more often in the last 8 years, but the effect on the economic cycle is expected to be less significant. Despite this, the next hypothesis raises the question of whether: H1c: The privatisation of SOEs through the liquation method is positively linked with economic growth.
Moreover, the management of the SOE is under the mandate of two authorities. The first refers to the time of UNMIK (1999UNMIK ( -2008. As part of this, the general management and transfer of rights in relation to SOEs or parts of them were the sole responsibility of international staff (2003)(2004)(2005)(2006)(2007)(2008). While the second refers to the period from 2008 onwards when the SOE was administered by the Kosovo authorities and the International Civilian Representative for Kosovo (ICO) as a monitoring body. So, implementation of the methodology by these institutions differs in principle according to the executive power they held, but also to the efficiency in treating SOEs and their assets. With this reasoning, we present the pre-2008 privatisation model (KTA/UNMIK) and the post-2008 privatisation model (PAK/Kosovo) which refers to the transfer of competencies to the Kosovo authorities with the aim to evaluate the efficiency of the two regimes in economic growth. Therefore, we present the following hypothesis.
H2: Economic growth in the period following the SOEs privatisation by local management (PAK/Kosovo) is higher than it is in the period following the SOEs privatisation by international management (KTA/UNMIK).
Privatisation is a sensitive issue and has sparked debate on whether it is achieving the intended effect. Building credible institutions plays a critical role not only in influencing the effectiveness of privatization, but also in economic growth. A key factor that will serve as a proxy for addressing this issue is the quality of institutions (IQ), which has a multidimensional impact and indicates whether resource use is appropriate at the country level. The lack of a friendly environment is likely to increase uncertainty for potential investors and make comparable economic growth unlikely. Moreover, quality indicates a state's inability to address issues related to governance, accountability, corruption, civil society participation in decision-making, rule of law, quality of regulators, etc. But what is most associated with the quality of institutions is corruption itself which undermines fair competition and clearly disrupts the economy in the long run (Baum€ ohl et al., 2019). Since the quality of institutions is not directly measured, we used World Governance Indicators (WGI) as a proxy for it. Collectively, the WGI provide a sufficient basis for assessing the institutional quality as well as its relevance to economic growth in Kosovo. Therefore, in addition to the effects on the privatisation process, we assumed that institutional quality positively supports economic growth.
H3: Economic growth is positively affected by the quality of state institutions (IQ).

Data and sample
To the best of our knowledge, this is the first empirical investigation to be conducted on the privatisation data and process in Kosovo. The data collection occurred on the basis of a special permit (for non-public registries) for all financial and non-financial transactions made in the respective period. The study covers the long period of 2003 to 2018, creating a 16-year time span of data. The initial sample includes data on 590 SOEs. In an effort to better analyse privatisation, we first selected all SOEs under the UNMIK administration until Kosovo declared its independence in 2008 and those SOEs under the administration of the Kosovo authorities from 2008 onwards. From this population, we removed all SOEs that were transferred to municipalities or government authorities (e.g. forest economy, urban transport bus station, and others). Further, we ignored all SOEs for which no information had been provided for at least 3 years. The final sample includes 571 SOEs, including their assets of different types and sizes.
Data (financial and nonfinancial information) mainly account for sales proceeds, costs associated with the sale and liquidation of assets, creditors 'claims, physical assets sold annually, bidders' waves, etc. Thus, the data source on privatisation activities is collected by the Privatisation Agency of Kosovo, while regarding macroeconomic indicators like gross domestic product, foreign direct investment and other financial control variables the sources are the Kosovo Agency of Statistics, the World Bank, the CIA factsheet, UNCTAD, the Central Bank of Kosovo, etc.

Dependent variable
Various institutions have established criteria on which economic growth is calculated, but gross domestic product (real GDP or GDP at current prices) is the main measure used everywhere to determine the level of an economy's output. Therefore, real GDP is considered the most reliable measure of economic growth. In this sense, the study relied on real GDP as a proxy for economic growth, measured as the value of the economy's output accounting for inflation and represents the dependent variable in the study. Moreover, we extract real GDP according to the five main regions (i) of Kosovo, calculating each region's share of GDP per capita at given point in time (t). Namely, real GDP is a given number and calculated at the country level by the Kosovo Agency of Statistics.

Independent variables
The first test variable is total privatisation proceeds (TPP). Since privatisation as a process is unable to be directly measured, as a proxy we used total privatisation proceeds (i.e. sub-variables SSO, OSO, LIQ, LE and DSN) for each region (i) at time (t). Transportation SOEs (bus station) (-) (6) 5 Forestry economics (-) The methods used in our study are presented as sub-variables which measure the total proceeds generated from the sale of an SOE or assets through any of the methods. On this basis, TPP is measured as the summarised amount of cash inflows coming from each method used to privatise SOEs or any assets.
Apart from the methods used, the model employs KTA/PAK which is measured as a dummy variable and represents the period before/after the UNMIK regime.
Further, institutional quality (IQ) is a proxy for the Worldwide Governance Indicators (WGI), which summarises the quality of state institutions, and is measured on the basis of six broad governance dimensions: (1) voice and accountability; (2) political stability and lack of violence/terrorism; (3) government effectiveness; (4) regulatory quality; (5) rule of law; and (6) control of corruption. The six general indicators are based on over 30 sources of basic data reporting perceptions of governance of a large number of survey respondents and expert evaluations worldwide (WB, 2018). It uses a scale of 0 (-2.5 index, very corrupt) through to 100 (þ2.5 index very clean). Assuming a good institutional environment, institutional quality is expected to have a positive relationship with growth.

Control variables
Several control variables are included in the comprehensive model, driven by logical reasoning that suggests they may have a positive impact on economic growth. The values of the control variables are refined and there is no difficulty in measuring them.
1. FDI is measured as the value of inward direct investment made by non-resident investors in the reporting economy. FDI stands for Foreign Direct Investment (FDI) per region (i) and measured as the total level of direct investment at a given point in time (t). 2. Population growth is measured as the difference between births and deaths. Thus, an increase in the net population (np) is calculated as: np ¼ (births-deaths)/ population size for region (i) at time (t). 3. GINI Index is a given index which represents the income or wealth distribution of a nation's residents and is the most commonly used measurement of inequality. It measures inequality among values of a frequency distribution (e.g. levels of income). The coefficient ranges from 0 (or 0%) to 1 (or 100%), with 0 representing perfect equality and 1 perfect inequality for region (i) at time (t); 4. ExpEdu stands for education. While it is hard to directly measure education, many researchers measure the level of education as a proxy for government spending on education, expressed as a share of GDP. Accordingly, we measured education as a proxy for government spending on education per capita for region (i) at time (t). 5. Export is measured as the value of all goods and services produced in one country that are sold in another country for region (i) at time t. 6. Import is measured as the value of all goods and services produced in one country that are bought in another country for region (i) in a given period in time (t).

Research model
With panel data, as a starting point, the study used the growth model introduced by Cobb Douglas and Solow, the so-called production function (Biddle, 2012). An increase in input factors will lead to an increase in output. On this basis, transforming the core model into a comprehensive model helps better understand the effect of social property transfer along with other ancillary variables. Therefore, the research seeks to test hypotheses 1-5 using the following model.
where EG stands as a proxy for real GDP for region (i) at a given point in time (t).
Total privatisation proceeds (TPP) include the sum of the proceeds from the SSO, OSO, LIQ LE and DSN for each region (i) at time (t). In addition, several control variables are introduced into the model since the literature showed they have an impact on economic growth (Biddle, 2012;Hamm et al., 2012;Okten & Arin, 2006). Population growth (PG) it is a very important indicator on which GDP growth is supposed to be contingent. Preliminary evidence supports many growth theories but, as suggested by (Headey & Hodge, 2009), the decline in population growth significantly slows economic growth. It is therefore expected that PG is positively associated with the economy's output (Peterson, 2017). Next, we control also for inequality (INEQ) to ensure whether privatisation has widened income inequality. While this is possible, our expectations are that the relationship between inequality and economic growth will be less distinct (Adams & Mengistu, 2008). Moreover, much of the nation's capital has accumulated in the hands of oligarchs or individuals who support corrupt public officials, and it seems that much of the wealth created also flows to these individuals (d' Agostino et al., 2016;Mo, 2001). Foreign direct investments (FDI) such as inward FDI, transfer of technology and knowledge, etc. are often considered a crucial factor for economic growth. The import of capital, in particular the inflow of FDI into the economy, has a multiplying impact. The higher the volume of inflows, the more the economy benefits directly, healthy competition increases, production capacities increase and new trade lines are created. With this rationale, we expect FDI to have a positive relationship with economic growth (Jenkins, 2006). Spending on education (EDU), namely, on the training/educating of human capital, is an important causal determinant of income for individuals within countries. In this sense, the increase in income is linearly related to the education level of individuals: the more one is educated, the higher the income (Saller, 2012). Finally, we include in the model export (EXP) and import (IMP) as control variables. Both of these variables are expected to affect the volume of the balance of payments, which in turn affects economic activity. Compared to imports, EXP is expected to have a positive impact on economic growth and vice versa (Pietrucha & _ Zelazny, 2020). Next, the dummy fixed effects (uit) and error term (eit) are included in the model specification.

Summary statistics
Summary statistics are presented in Table 1. To date, about 540 SOEs have undergone transfer procedures (i.e. privatisation and liquidation).
As Table 1 shows, EG (real GDP) over the period of 16 years multiplied from a minimum value of EUR 2,912.50 million (2003) to EUR 6,715.35 million (2018. Thus, average GDP growth is 5% per year, ranging between À0.03 (-3%) to 0.54 (5.4%) a year. Regarding privatisation proceeds (TPP), Table 1 shows that approximately EUR 9.25 million on average (at the regional level) was collected/collected (due to assets sold) in the period 2003-2018. Specifically, most proceeds from the transfer of social assets (privatisation) have come via the spin-off method (with an average of EUR 5.32 million per year at the regional level), followed by liquidation (EUR 1.1 million per year at the regional level) and finally by the special spin-off method (around EUR 1.48 million per year at the regional level). Moreover, on average, DSN proceeds are EUR 0.3 million at the regional level. With respect to the initial asset (extent of the SOEs), we originally identified about 3,900 assets of 571 SOEs during the period 2003-2018, while on average about 608 assets were scattered across the regional offices. Moreover, on average about 27% (1017[2003], 611[2018]) of them were concentrated in the Prishtina region, followed by Pej€ e with 980 assets (2003), Gjilan with 690 assets (2003) etc. Asset sales were realised through waves (public tenders) that on average amounted to around 6.37 waves per year. This shows that during a single year it has been possible to prepare assets for sale in 6.37 bidding waves (whether the privatisation, liquidation or special spin-off methods).

Empirical results
This study employed fixed panel regression analysis to consider the impacts of variables that may change in time. Prior to interpreting the results of the regression analysis, we tested for several potential problems that might threaten our data and regression analysis in general (i.e. collinearity diagnostics, endogeneity and heteroskedasticity). With respect to collinearity, we tested whether some independent variables are highly correlated and ran Pearson's correlation analysis. Table 2 shows that most correlations between the explanatory variables and the dependent variable are significant, as expected. In addition, in our case the collinearity diagnostics, most importantly the variance inflation factor (i.e. 1/VIF ¼ 7.15), reveal that reasonable collinearity indeed exists. However, the literature suggests that multicollinearity is a potential problem when the VIF exceeds 10, and we thus dismiss multicollinearity as a serious threat to the validity of the estimated parameters (Lewis-Beck et al., 2012). Since endogeneity and heteroskedasticity were a concern, we used the Hausman and Breusch-Pagan tests. In both cases, it was suggested that the problem of endogeneity and heteroskedasticity is not serious, allowing us to ignore it as a threat to further treatment. We also conducted another test which provides a flexible functional form that is useful for identifying nearly any pattern of heteroskedasticity. Thus, heteroskedasticity is robust also using the White test since it enables the independent variable to have a nonlinear and interactive effect on the error variance.
Tables 3 and 4 introduce hypothesis testing at two levels: (1) an aggregated model; and (2) segregate regression analysis (i.e. as to the method used to privatise SOEs or part thereof). At the aggregated level, Model (1) in Table 3 shows that TPP is positively significant related with economic growth. It suggests that privatisation funds have had a positive impact on economic growth, but not as high as expected. As for the next hypothesis on KTA/PAK, we found a significant positive relationship with growth, but not as expected. Further, it appears the post-UNMIK period is negatively associated with economic growth. Concerning the third hypothesis on the quality of institutions (IQ index), Model (1) in Table 3 shows a significant relationship with economic growth, but in the opposite direction to what was expected.
In addition to the aggregate level, when testing at the level of each method applied we find that only Model (4), Model (5) and Model (6) in Table 4 are associated with economic growth. In contrast, Model (2) in Table 4 shows that sales through the SSO method do not have a significant relationship with economic growth. Similarly, we find that application of the OSO method also does not have a significant impact on economic growth. Contrasting situations are shown with regard to the LIQ method (Model 4), where B coefficients are significantly related to GDP growth. Further,  Table 4 also shows significant links between CL and economic growth, while finally Model (6) reveals that negotiated sales are significantly related to growth. Further, the results shown in Table 4 are robust to those shown in Table 3, thus confirming that neither KTA/PAK nor IQ index are positively related to GDP growth as hypothesised, despite having a significant relationship with it. The outcomes suggest that overall privatisation is not well-performed and the link between privatisation (methods used) and growth is not satisfactory. Further, the results of the models presented in Tables 3 and 4 show the impact of particular control variables on GDP growth. In this context, EXP has a significant relationship with GDP growth in almost all models in Tables 3 and 4, yet Model 2 shows no relation with it. In addition to Models (1) and (2), IMP is significantly negatively related to GDP. Next, PG at all stages of the regression procedure is significantly associated with GDP growth. Moreover, the analysis of the regressions in all models employed shows that FDI, INEQ and EDU are also significant for economic growth. Finally, year effects included in the models are reported to count for any variation in the outcome that happens over time and is not attributed to other explanatory variables.

Discussion
This section discusses the results of the study presented according to the research hypotheses. Therefore, the paper deals with privatisation in Kosovo, the methods used and empirically reveals the effect it brings for economic growth. Concerning the first hypothesis, despite a weak association, we find that TPP is positively related to economic growth. The main reason for the small effect on the economic growth is precisely the inability to include TPPs in economic activities. Thus, the proceeds generated by privatisation were blocked for years and not used for development purposes.
Further, we also tested whether the methods used to privatise SOEs were effective and appropriate. As for SSO method, we find no association with growth and therefore describe it as non-effective. According to the SSO requirements, a potential investor is required to invest in local hire new workers or continue with the same economic activity. However, we find that this is inappropriate because most privatised SOEs are left behind and declare bankruptcy as soon as they are transferred to the new owners. Those that survive bankruptcy are likely to be followed by other accrual irregularities that further reduce the impact on the economy (e.g. failure/withdrawal of the first bidder [e.g. for EUR30 million] and the sale of the SOE to a second bidder at a fraction of the price [e.g. for EUR 15 million], disregard of the commitment to invest and recognition of suitability/commitment, etc.). With respect to OSO method, we find no relation to economic growth. Surprisingly, although the SOE was transferred to the New Co., and then the New Co. (core asset) was sold, many parts of the SOE (non-core assets) were not sold together with the New Co., thereby remaining unsold even many years after the New Co. was sold. Only a few cases of purchased SOEs were used for the same economic activity as they were established. With this reasoning, the OSO method remains ineffective and has not affected economic growth. Liquidation (LIQ) is the third method of transferring ownership. By testing the hypothesis, we find that use of the LIQ procedure is significantly associated with economic growth yet, although large amounts of assets have been transferred through this method, it has only had a modest impact on economic growth. The reasons for this relationship do not suggest this method is the best possible one, except that the cash flows are continuously uninterrupted more than the two methods mentioned above.
Leasing (LE) and Direct Negotiation Sales (DNS) are two complementary methods to the above. Asset leasing does not confer any property rights and the administration of the SOE's assets is tied to their lease until they are privatised. Our analysis shows that active rents relate to economic activities and, in another form, generate proceeds or add value to the economy. We use leasing as a privatisation method because all rental proceeds are held in trust in the same SOE account, which is then to be distributed to potential creditors in the SOE liquidation phase. With regard to direct negotiations, we find significant links with economic growth, but they are not strong enough to ensure that the effect will continue in the future.
Two very important issues that were hypothesised to have an increasing effect are the pre-and post-UNMIK periods (KTA/PAK) and the quality of state institutions (IQs) (IQ index). Yet, in contrast, we find a negative association with economic growth for these two test variables. First, the PAK period brings no changes of past practices (in the KTA/UNMIK era), meaning the methodology introduced by the UNMIK is almost the same and does not provide the proper effects for the economy. With respect to the third hypothesis, as mentioned we find that IQ index (a proxy for WGI) does not encourage economic growth. Economies denoted by poor governance are unable to function properly because it prevents law enforcement and professional competencies from functioning freely. In such environments, fair competition is impeded as business owners who can use their connections or money to deceive government officials can manipulate policies and mechanisms to ensure they are the sole provider of goods or services in the market. Furthermore, the inequality doesn't show a high disproportion in income distribution among the population in Kosovo, and since the number of unsold assets is extremely large (about 60% of total assets), this may overturn our future findings. Otherwise, we find that the developmental imbalance among the regions deeply widens the gap and does not guarantee sustainable economic growth in all 5 regions of Kosovo.
We also find that population growth (PG) and education strongly influence economic growth. Spending on education (accounting for about 4% of Kosovo's GDP), with the aim of preparing students for the labour market and in line with the current trend, is strongly linked to economic growth. This trend could be badly damaged by the migration of skilled people, further supported by the negative (in some years) population growth in the last decade. Further, we make the strange finding that foreign investment does not stimulate favourable economic growth. FDI inflows are influenced by many factors, including the poor quality of the institutions potentially related to corruption, the quality of the regulators, political stability, investment protection, etc., that in turn reduce foreign capital inflows (Vuk si c, 2016).
On top of enriching the literature, the study has some limitations that for objective reasons were not addressed in this study. Although privatisation data are largely internal and difficult to access publicly, in our case, we managed to own the privatisation data but handled it responsibly and discreetly. As a result, a number of sensitive issues which may be raised by this paper remain unexplained at this stage (i.e., purchase price, board decisions on sales approval, sales contracts, budget, no. of claims and claimants of SOEs, etc.). In contrast, since the paper is based on macroeconomic variables the privatisation data are also aggregated and highly informative for the study in question.

Conclusion
In the last 20 years, the term privatisation has been heard frequently, not so much with respect to the cash generated but to how the money was managed and who benefits from it. The lack of evidence showing which method is more effective than the other, and vice versa, has created space for prejudice and corruption. The period 2003-2008 is characterised by the sale of Kosovo's most vital SOEs, which also provided a large amount of proceeds that was, however, blocked and could not be used for development purposes i.e. the ban on the use by UNMIK and ICR of privatisation funds. Despite this, less important SOEs and less attractive assets remained unsold, meaning the privatisation could not be completed in time and higher operating costs. Although this transition has been underway for some three decades in some Central and Southeast European countries, the privatisation process in Kosovo is almost 19 years old.
Moreover, the privatisation model in Kosovo does not seem to be the best, but we conclude that privatisation in the country has been accompanied by selective approaches that have not had a significant impact on economic growth. Further, the fragile institutions emerging in the countries of the former socialist bloc, the slow transition taking place, and the lack of preliminary evidence linking to economic growth are complemented by this empirical study, which to our knowledge is the first study to address the effects of privatisation on Kosovo's economic growth. Notes 1. The 1990s were the worst years for the Kosovo economy. Due to circumstances at the time, a large number of factories closed down, markets were lost, a high level of inflation, etc. led to economic collapse, which later degenerated into war. In the 2000s, Kosovo was under an international protectorate where the administration of socially owned assets was the responsibility of the United Nations Mission in Kosovo (UNMIK). 2. Using an OSO, only New Co.'s are sold. Thus, the main asset of the SOE is the transfer to New Co. and then the New Co. is sold. The other smaller assets (secondary assets) of the SOE are sold/expected to be sold through liquidation method.

Disclosure statement
No potential conflict of interest was reported by the authors.