The Baltic Journal of Economics
Jorgen Drud Hansen Editor in Chief University of Southern Denmark and EuroFaculty, University of Vilnius
Raul Eamets Editor University of Tartu
Mihails Hazans Editor University of Latvia
Daunis Auers Managing Editor EuroFaculty, University of Latvia
The Baltic Journal of Economics (ISSN-140X-099X) is published twice a year in December and July by EuroFaculty, Raina Blvd. 19, Riga LV-1586, Latvia.
Information for subscribers: Copy requests, orders and other enquiries should be addressed to the BJE at the address above, or by email to [email protected].
The Baltic Journal of Economics is published by EuroFaculty and refereed by internationally recognized scientific standards. The journal is intended to provide a publication medium for original research in economics for scholars working in the Baltic states or those who are working on topics relevant to the Baltic states. Papers may be theoretical, empirical or political economy in emphasis. Papers with policy relevance or which combine economic theory with empirical findings are particularly welcome. The Journal aims to stimulate a dialogue between scientists in social science, policy makers as well as other decision makers involved with economic development in the Baltic States. In order to make the journal relevant to a wide audience of academics trained in social science the articles should be presented in a form where explanations and the intuition behind the conclusions should be given priority above technical derivations.
For more information about the journal see the website: www.eurofaculty.lv/bje
E Copyright EuroFaculty 2002 ISSN-140X-099X Contents
1 Foreword Gustav Kristensen
Articles
3 Some Unpleasant (and then Some Pleasant) Transition Arithmetic Robert Elder
8 Financing Constraints as Determinants of the Investment Behaviour of Estonian Firms Jaan Masso
31 Gender Wage Differences in Soviet and Transitional Estonia Charles Kroncke and Kenneth Smith
50 The Outward Expansion of the Largest Baltic Corporations – Survey Results Kari Liuhto & Jari Jumpponen
Book Reviews
76 Bertola G, Boeri T and Nicoletti G (eds) Welfare and Unemployment in a United Europe: A Study for the Fondazione Rodolfo Debenedetti (MIT 2001) Alf Vanags
79 Keuschnigg, M Comparative Advantage in International Trade: Theory and Evidence (Physica-Verlag 1999) Alf Vanags
FOREWORD The first volume of the Baltic Journal of Economics (BJE) was published by EuroFaculty in 1997, the second volume following in winter 1999. There then followed a lull in activity. However, the growing need for a prestigious Baltic-based refereed economics journal called for the renewal of the organization of the Baltic Journal of Economics. A high technical standard, stable financing, and stable professional leadership has been secured through EuroFaculty Baltic resources and the generous support of our sponsors.
The Baltic scientific capacity in the field of economics is currently quite small. This is the strongest argument for cooperation around one common scientific journal of economics for all three Baltic states. It is the goal of the BJE to raise the Baltic states scientific capacity in economics, over a number of years, to the level of comparably sized European countries e.g. Denmark, Sweden and the Netherlands
Thus the function of the Baltic Journal of Economics is threefold: (i) to encourage Baltic scientists in their economics research by giving them a potential medium for refereed publication; (ii) to create a network of national and international referees who, through interaction with Baltic researchers, will increase the scientific level in the Baltic states; and (iii) to disseminate economics research on the Baltic States through the distribution of free copies of the BJE to academic and research institutions globally and the creation of an open web-site featuring all past and future editions of the journal.
Gustav Kristensen Director EuroFaculty-Tartu-Riga-Vilnius
1
Some Unpleasant (and then Some Pleasant) Transition Arithmetic
Some Unpleasant (and then Some Pleasant) Transition Arithmetic Robert Elder1
Twenty-one years ago, Thomas Sargent and Neil Wallace wrote an article entitled “Some Unpleasant Monetarist Arithmetic” (1981). I am no Monetarist, but I do pay attention to what Monetarists say, and I cheerfully acknowledge that these two Monetarists thought up a clever title for the paper they wrote for the Minneapolis Fed’s Quarterly Review back in 1981. Here in 2002, the subject I address in the paragraphs below is not Monetarism, but instead the experience of transition economies. In particular, I call attention to some arithmetic that can help organize our thoughts with regard to some of the hardships a country endures and some of the successes a country achieves during the transition from central planning to free markets. I focus on equations involving important stocks and flows of people in such a country in transition, subsequently citing data on those stocks and flows between the years of 1992 and 2000 here in Latvia.
When analyzing an economy in transition, a good place to start is with an equation describing the division of employment (E) between the state sector (ES) and the private sector (EP). E = ES + EP Each of the three magnitudes in this equation is a stock. At any moment in time, for example, we can observe the quantity of people with jobs (the stock of employment E), the quantity of people working for the state (the stock of state sector employment ES), and the quantity of people working for private firms (the stock of private sector employment EP). This equation can also be termed static, since it describes the state of employment at any given moment in time. If we move forward from one moment in time to some subsequent moment in time, we can make the equation dynamic, observing how these stocks change as time elapses.
ΔE = ΔES + ΔEP
Each of the three magnitudes in this equation is a flow. Positive flows add to stocks, while negative flows subtract from stocks. (For example, the Daugava River that runs through Riga is a flow that adds to the Baltic Sea, a stock.) Equipped with the
1 Professor of Economics, Beloit College ([email protected]), and for the 2001-2002 academic year, Fulbright Scholar in the Eurofaculty program at the University of Latvia ([email protected]). This essay benefits from valuable suggestions that Mihails Hazans, my colleague at the University of Latvia, made on earlier drafts, and I am grateful to him for his helpful comments.
3 Baltic Journal of Economics Autumn/Winter 2002
flow equation shown above, we can start to highlight a key element of transition. During the transition from central planning to free markets, there is a flow of employment out of the state sector (ΔES < 0) and into the private sector (ΔEP > 0). If each worker leaving the state sector found work in the private sector, note that
ΔEP = – ΔES, so from the flow equation above ΔE = 0, and there would be no change in aggregate employment. In this event, the sectoral reallocation of labor associated with transition might appear relatively painless. But transition is not that easy, and indeed can be painful, because the outflow from state sector employment does not necessarily result in an equal inflow into private sector employment. Overall, if ΔEP < – ΔES, then ΔE < 0, and aggregate employment can fall during the transition. Thus, the flow out of state sector employment might not lead entirely to a flow into private sector employment but also to a flow into unemployment. To bring unemployment into the arithmetic, we begin once more with an appropriate stock equation. At any given moment in time, the economically active segment of the population (EA) consists of those who have jobs (in the state sector, ES, and in the private sector, EP) and those who seek jobs (the unemployed, U):
EA = ES + EP + U.
Setting this equation into motion by allowing time to elapse, the associated dynamic expression would be
ΔEA = ΔES + ΔEP + ΔU.
From this flow equation, we can verify the fact that for an economically active population of constant size (ΔEA = 0), a flow out of state sector employment (ΔES < 0) can imply a flow into private sector employment (ΔEP > 0) as well as a flow into unemployment (ΔU > 0). But unfortunately, the unpleasantness of state sector shrinkage does not stop with rising unemployment. To see why, observe first that the assumption of no change in the size of the economically active population (ΔEA = 0) is unrealistic, and we can relax this assumption in our final pair of stock and flow equations. Starting once more with the pertinent stocks, note that we can dichotomize the entire population (POP) between the economically active (job- holders, ES and EP, and job-seekers, U) and the economically inactive (EI):
POP = ES + EP + U + EI.
The associated flow equation follows as:
ΔPOP = ΔES + ΔEP + ΔU + ΔEI.
Here, for a given population (ΔPOP = 0), note that a flow out of state sector em- ployment (ΔES < 0) can imply (1) a flow into private sector employment (ΔEP > 0), (2) a flow into unemployment (ΔU > 0), and (3) a flow into economic inactivity (ΔEI > 0). Unpleasant dimensions of a flow into economic inactivity are apparent when any such flow involves involuntary retirement. And finally, when we
4 Some Unpleasant (and then Some Pleasant) Transition Arithmetic
acknowledge that the assumption of a constant population (ΔPOP = 0) is inappropriate to transition scenarios, we should add that a flow out of state sector employment can imply (4) a flow out of the population (ΔPOP < 0). From the somewhat unpleasant to the definitely unpleasant, such flows out of the population can range from the temporary life disruptions of emigration to the permanent life terminations of death. As shown by the data below, emigration in excess of immigration and deaths in excess of births have been a steady feature of the transition years from 1992 through 2000 in Latvia2. As it chronicles these nine consecutive years of population decrease, notice that the table reveals net emigration as the larger source of population decrease during the first three years and net deaths as the larger source of population decrease during the final six years.
Table 1: Sources of Population Change in Latvia, 1992-2000 (in thousands)
The next table allows us to monitor the behavior of our final stock equation in Latvia during these same years.
Table 2: POP = ES + EP + U + EI, Latvia, 1992-2000 (in thousands)
Here in Table 2, the first thing that we see is the monotonically decreasing
2 Observe that these years do not span the entire transition, which could be dated from the redeclaration of Latvia’s independence in May 1990 through the present. I focus on 1992 through 2000 because these are the years for which there exists available data for each of the variables that I have introduced in the preceding discussion.
5 Baltic Journal of Economics Autumn/Winter 2002
population implied by Table 1. Also readily apparent from this table of stock magnitudes are two distinguishing features of transition: monotonically decreasing state sector employment and monotonically increasing private sector employment. In contrast, as shown by the final two columns of data, the stocks of the unemployed and the economically inactive display non-monotonic behavior, decreasing during some years and increasing in others. To organize our thoughts about the behavior of the five stocks recorded in Table 2 and to examine the transition from a different angle, I use the stock data provided by Latvia’s Central Statistical Bureau to construct the table pertinent to our final flow equation. The results appear in Table 3. Since each stock in Table 2 is an annual average, each flow in Table 3 is the difference between each pair of consecutive annual averages, as indicated below.
Table 3: ΔPOP = ΔES + ΔEP + ΔU + ΔEI, Latvia, 1992-2000 (in thousands)
To think about what’s going on in the first four rows of the table, observe first that we can re-express our final flow equation as follows:
– ΔES – ΔEI + ΔPOP = ΔEP + ΔU.
For the period from 1992 to 1996, this re-expression of our final flow equation groups outflows on the left-hand side and inflows on the right-hand side. During these years, this equation says that the sum of outflows from state sector employment, the economically inactive, and the population were equal to the sum of inflows into private sector employment and unemployment. For example, from
1992 to 1993 the observation ΔES = – 170 implies – ΔES = 170, an outflow of 170,000 people from state sector employment available as a source of inflows into private sector employment and unemployment. Similarly, the 1992 to 1993 observation ΔEI = – 25 implies – ΔEI = 25, an outflow of 25,000 people coming from the ranks of the economically inactive to provide a second source of inflows into private sector employment and unemployment. In contrast, the 1992 to 1993 observation ΔPOP = – 52 remains negative on the left-hand side of this equation, since an outflow of 52,000 people from the population diminished what was available to facilitate inflows into private sector employment and unemployment between these years. To highlight the leading destinations of workers leaving the state sector, notice that the while private sector employment absorbed the largest
6 Some Unpleasant (and then Some Pleasant) Transition Arithmetic
share of shrinking state sector employment from 1992 to 1993, unemployment absorbed the largest share of shrinking state sector employment from 1993 to 1994, and the population absorbed the largest share of shrinking state sector employment in 1994 to 1996.
Following 1996 to 1997, which was unique for its flow into economic inactivity, we move to the final three years from 1997-2000, during which the outflows-equals- inflows equation becomes
– ΔES – ΔEI + ΔPOP – ΔU = ΔEP.
For each of these last three years, this equation features four outflow terms on the left and one inflow term on the right. Happily, note that the only stock absorbing inflows during these last three years is private sector employment. Although outflows from the population reduce what could have been additional inflows into private sector employment, positive contributions to these recent years of inflows into private sector employment are made by outflows from state sector employment as well as outflows from the ranks of the economically inactive and the unemployed.
This is a positive point with which to conclude. Overall, the subject of economic transition is sprawling, and there are many aspects of it left unaddressed in this brief note. Further, there are many countries involved in economic transition, and a comparison between their experiences and those of Latvia would provide additional perspective. As we have seen here in Latvia, the early years of transition can feature some unpleasant dynamics, but as we have also seen more recently, more pleasant dynamics can follow.
References
Central Statistical Bureau of Latvia, Demographic Yearbook of Latvia 2001.
Central Statistical Bureau of Latvia, Statistical Yearbook of Latvia 2001.
Sargent, T.J., and Wallace, N (1981), “Some Unpleasant Monetarist Arithmetic,” Federal Reserve Bank of Minneapolis Quarterly Review, Volume 5, Number 3, pages 1-17.
7 Baltic Journal of Economics Autumn/Winter 2002
FINANCING CONSTRAINTS AS DETERMINANTS OF THE INVESTMENT BEHAVIOUR OF ESTONIAN FIRMS1 Jaan Masso2
Abstract Lack of financing is arguably the main obstacle for making profitable investments in transition economies. In this paper we investigate whether there is underinvestment due to financing constraints in Estonian manufacturing firms. Firm level panel data from 1995 through 1998 with several items from financial statements were used. The unique data includes many very small firms with assets less than 1 million USD that are often not explored in empirical studies. The existence of liquidity constraints was tested with estimating regression coefficients of inside - firm financing from reduced form investment regressions and using the investment Euler equation. Results show that internal finance played a bigger role for investments made by small firms and firms owned by domestic (non-foreign) capital. The only exception is that in the Euler equation, cash flow influenced investments more for small than for large foreign firms. The results imply that foreign direct investments and lower corporate income taxes can promote investments through the relaxation of liquidity constraints.
JEL classification: G31, E22 Keywords: Financing constraints; Investment; Cash flow; Estonia
1 Introduction Since the seminal article by Fazzari et al. (1988) numerous papers discuss the effects of financing conditions on the investment decisions of private firms. In several papers, it has been found that investment is more sensitive to the availability of internal funds among certain groups of firms that are more subject to the presence of information and agency problems in financial markets. These include, among others, small and young firms, firms without credit ratings, and firms without affiliation to an industrial or banking group. The presence of financing or
1The author is grateful for many helpful comments and remarks from Riku Kinnunen, three anonymous referees, the editor and participants of seminars held in Stockholm, Tartu and Tallinn that have significantly contributed to the quality of the paper. I am also obliged to Urmas Varblane for generously providing the data that was used in this study. All remaining errors are of course my sole responsibility. 2Faculty of Economics and Business Administration, University of Tartu, Estonia. E-mail address: [email protected]
8 Financing constraints as determinants of the investment behaviour of Estonian Firms
liquidity constraints (in the paper both notions are used interchangeably) has several implications for tax policy, corporate take-overs and the channels of macroeconomic policy (Hubbard, 1998).
There is a widespread view among economists that capital market imperfections are especially severe in Central and Eastern European transition economies (Coricelli, 1996). This is because many firms in the new market economies are newly established without credit history, track record, collateral etc. Also, the weakness of the banking sector creates problems due to the banks’ inexperience in monitoring and gathering information about loan applicants. Economic uncertainty has lead to an unwillingness or inability among the banks to lend long-term (Pissarides, 1998). On the demand side, firms need to invest heavily in order to modernize obsolete capital stock and increase competitiveness in world markets. Thus, the lack of financing probably constitutes one of the main obstacles to growth.
There have not been many studies on this topic in the transition economies, mainly due to a lack of enterprise-level data. Perotti and Gelfer (1998) have shown that investment in firms belonging to financial-industrial groups in Russia is less sensitive to cash flow than investment in independent firms. On the other hand, Lizal and Svejnar (1998) did not find evidence of a positive link between internal finance and gross investment, although in their later study (2000) of net investment retained profits were shown to have positive effect. Anderson and Kegels (1997) also found evidence of the influence of financial variables like cash flow, beginning-of- period bank debt and trade credit on the fixed investment of Czech enterprises. Bratkowski et al. (1999) argue that imperfections in capital markets in Central European economies do not seem to affect the growth of new private firms. For Bulgaria, Budina et al. (2000) found liquidity constraints to be important for small firms but not for large. This finding was explained by the inefficiency of the financial sector because of loans granted to large unprofitable firms. One weakness of these studies is that their inferences have been based on reduced form investment regressions, rather than explicit conditions of optimal capital accumulation.
In the present paper, we try to see whether firm size and ownership are important in determining whether Estonian industrial firms can finance profitable investment projects. We succeeded in getting access to enterprise level panel data that allows an examination of how the severity of financing constraints varies across different types of firms. Also, in this way the aggregation bias can be avoided. Thus we focus only on fixed investments; however financing constraints could influence also investments in inventories, research and development, market share etc. (Hubbard, 1998). The existence of liquidity constraints was tested in two ways. First, we estimated simple reduced form investment regressions in order to observe
9 Baltic Journal of Economics Autumn/Winter 2002
whether internal finance affects investment positively. Second, we used the investment Euler equation derived from the objective of maximizing a firm’s value under convex adjustment costs and constant returns to scale. The results from both approaches show that the availability of internal finance plays a significantly bigger role for investments of small and domestically owned private companies. The policy implication of the results is that recent changes in taxation law in Estonia that have made taxation more favourable for retained earnings than for dividends, promote investment of the aforementioned firms. Also, the range of benefits for attracting foreign direct investments into the country can be seen more widely - alongside other positive effects foreign direct investments could also loosen liquidity constraints.
The remainder of the paper is organized as follows. Section 2 provides some stylized evidence of investment and financing behaviour of Estonian firms. Section 3 describes the dataset of Estonian manufacturing firms. Section 4 describes the method and results of the reduced form investment equations and section 5 those of the more elaborate Euler equation. The final section concludes with discussion of the findings and policy implications.
2 Stylized facts of the investment and financing problems of Estonian firms Before a formal statistical analyses we will present some stylized evidence about the investment and financing behaviour of Estonian firms. First, the Estonian Institute of Economic Research has conducted inquiries among Estonian firms about their investment decisions (Konjunktuur, 1999). Table 1 summarizes the relevance of different factors limiting investment.
As we can see, in all years the biggest obstacle for investment has been a small profit. On the one hand this could simply reflect a low internal rate of return in comparison to the cost of capital, as noted by Raudsepp and Leoshko (1999). But on the other hand this could also show a tendency to mostly rely on internal funds when carrying through investments. Many studies in developed economies show internal finance or cash flow to be the primary source of funds, e.g. Fazzari and Petersen (1993) found that cash flow constitutes 71 % of net sources of finance for US public firms paying dividends less than 10 % of earnings. For Estonia it has been argued that internal financing constitutes a smaller part of funds than in developed countries because of a lack of internal funds and unstable economic development. For instance, Kangur et al. (1999) estimated that about 49 % of total investment was internally financed.
10 Financing constraints as determinants of the investment behaviour of Estonian Firms
Table 1 Factors limiting the investments of manufacturing firms in Estonia (percent of enterprises surveyed)
The high cost of capital is often considered an obstacle (up to 32 % noted this as a problem). For example, in a survey among firms in the United Kingdom only 6-10 % viewed the cost of finance as a limitation to capital expenditures (Bond and Jenkinson, 1996). This finding for Estonian firms could result from the cost disadvantage of external funds due to high transaction costs, agency costs and asymmetric information. Earlier papers on the financial problems of Estonian firms have found the cost of capital to be too high, especially for small and medium size enterprises (Raudsepp and Leoshko, 1999). Difficulties in obtaining credit (19-31 % of respondents) could possibly reflect credit rationing, i.e. some applicants are denied loans in spite of their readiness to carry all the price and non-price components of the loan contract (see Stiglitz and Weiss, 1981) or difficulties with necessary collateral, both of which are consistent with an imperfect capital market story. Similarly, in the aforementioned study of UK firms only 2-3 % of firms reported an inability to raise external finance as a problem (Bond and Jenkinson, 1996).
We could infer that financial factors seem to constrain capital investments more in transition economies than in developed western economies. In particular the availability and price of external (new debt and equity) versus internal financing (internally generated cash flow) is an issue. According to the financing hierarchy hypothesis firms prefer to use internal financing due to asymmetric information between managers and potential new equity investors or creditors; external funds are only used after internal sources are exhausted (Fazzari et al., 1988). One survey among Estonian non-financial firms listed in the Tallinn Stock Exchange showed the presence of a financing hierarchy – internal equity was ranked as the most preferred source of financing (Raudsepp et al., 2000). Low dividend payout rates (on average 10 %) confirm this finding (Ibid.) because the cost disadvantage of external funds forces firms to retain profits inside the firm (Fazzari et al., 1988a). The amount of internal financing may be constrained by relatively small rates of
11 Baltic Journal of Economics Autumn/Winter 2002
depreciation – in some sectors (machinery, chemical industry) these are 2 to 4 times smaller than in European firms (Pärn and Lumiste, 2000). This may have some negative effect on the amount of internal finance available due to a smaller tax shield of depreciation.
The Bank of Estonia carried out a study on the financing of the Estonian entrepreneurial sector during 1994-1998 as well as connections between financing and investment (see Kangur et al., 1999). The authors argued that aggregate data shows co-movement between external financing and total investment (investments in fixed assets plus), so that (external) financing can be an important factor affecting the level of investments.
3 Data and summary statistics In the present study we used firm-level financial statements panel data collected and compiled by the Statistical Office of Estonia. The original dataset includes 373 industrial enterprises for the period 1995-1998. Only firms in manufacturing industries were considered. Here manufacturing firms are those with 2-digit EMTAK codes between 15 and 39 that correspond to section "D" of European Union NACE classification; these codes also correspond to SIC codes between 20 and 39. The firms in the sample represent about 70 % of the total sales of manufacturing industry. The total number of firms in the industry in this period was about 4500 (Statistical Yearbook of Estonia 2000), so the sample is biased towards large rather than small firms. When we consider that financing constraints can prevent business from starting (see Evans and Jovanovic, 1987), so that some survivorship bias is introduced, it can be suggested that the present study will tend to underestimate rather than overestimate the importance of financing constraints.
Several firms were deleted from the sample. First, all firms with negative or zero fixed tangible assets were deleted. Second, the possible effect of outliers on regression estimates was controlled by excluding firms with observations of sales growth, investment to capital ratio or cash flow to capital ratios below or above 5% upper and lower tails of distribution. The number of firms left is 195. The justification for excluding firms with extreme growth rates in sales or investment is that if both investment and cash flow grow at a rate similar to growth rate of sales, then part of the co-movement could be due to the scale factor. This effect would bias the estimates of investment-cash flow sensitivities towards one, particularly in firms with higher annual growth rates (Kaplan and Zingales, 1997).
Table 2 presents summary statistics for some of the regression variables as well as the relative importance of different sources of finance for different sub samples of firms (the three last rows of the table). First, the total sample was split into three equally sized groups by the average value of real assets. As we can see
12 Financing constraints as determinants of the investment behaviour of Estonian Firms
from table 2, small and medium sized enterprises grow faster and invest more, so the need for extra financing is greater. As expected, cash flow plays a bigger role as a source of financing for smaller firms. Both cash flow and investments are more volatile for smaller firms. In earlier studies other researchers have found the same evidence for classes of a-priori constrained firms (Fazzari et al., 1987). Only for the 3rd group is new equity an important source of funds. In total (last column of table 2) firms have been investing quite actively (average investment to capital ratio 0.40). This has been largely financed by cash flow. Still the relative importance of cash flow is somewhat smaller than in studies made with developed countries’ data, e.g. Fazzari and Petersen (1993) estimated the average cash flow to the net sources ratio to be 0.715.
We can also see that firms belonging to foreign capital are on average much bigger in terms of total assets and capital, and grow faster. The first evidence can be explained by the fact that Estonian residents do not have enough capital (neither could they borrow the funds) to privatize large state-owned firms. Both investments and cash flow are more volatile for domestic firms. Foreign firms also got remarkably more new equity capital, which indicates their better access to external financing. Here the firm is defined as belonging to foreign or Estonian capital if in all years (1995-98) more than 50 % of the share capital belonged to foreign or Estonian residents.
Table 2 Means of selected variables: sample of 195 manufacturing firms, period 1995-1998 (sample split by average value of real assets and form of ownership)
13 Baltic Journal of Economics Autumn/Winter 2002
4 Evidence of the existence of liquidity constraints from reduced form investment equations The existence of liquidity constraints is usually tested by regressing the investment on variables that measure the availability of financing generated inside the firm and some proxy for the investment demand (affected by productivity of capital, expectations, required rates of return). Often in the part of the latter is Tobin’s q that theoretically should capture all relevant information and is basically the ratio of market value of firm’s equity and debt to replacement value of assets.3 Unfortunately the firms in the current sample are not listed on the stock market, so we are unable to calculate such a measure. Instead we use employment growth to control for the existence of investment opportunities, as with Bratkowski et al. (2000). Second, in place of liquidity variables cash flow and cash stock are used. The liquidity variables proxy for internal net worth (liquid assets plus the collateralizeble value of illiquid assets), while they also convey information about what proportion of investment spending can be internally financed (Schiantarelli, 1996). Firms with a higher level of liquidity can better collateralize debt issues and receive loans at lower interest rates as well as exploit more relatively cheap internal funds. It means that we are testing whether internal and external financing are perfect substitutes or not. The expected impact of cash flow and cash stock on investment is positive. The intuition for including the leverage variable is that agency costs occurring due to diverging interest of lenders and borrowers (e.g. monitoring and bankruptcy costs) are assumed to increase in the amount of debt used. A higher level of debt in the beginning of the period makes it more difficult to finance new investment projects, if there is a limit to the debt a firm can have. So we estimated the following equation:
(4.1) where I denotes gross investment, LGROWTH employment growth measured in logarithms, CF cash flow, CS cash stock, K capital stock and DEBT/A is the ratio of 4 γ γ short- and long-term debt to total assets. The intercept coefficients, i and t allow γ for firm specific and year intercepts; uit is random error term. Firm dummies i control for the effect of variables that are constant over time but are excluded from the model (e.g. industry classification of firm). Hereby investment is measured as change in fixed tangible assets plus depreciation; cash flow is the sum of net income and depreciation. All variables (except debt and employment growth) are normalized by the initial size of capital in order to reduce possible
3 See e.g. studies by Fazzari et al. (1988) and Hoshi et al. (1991). 4 We also tried to proxy for investment demand with change in output and sales growth, as suggested by accelerator models of investment. Still the coefficients were almost always statistically insignificant. This result should not be surprising, as other studies that were made by using data from transition countries have observed similar results, see e.g. Anderson and Kegels (1997), Prasnikar and Svejnar (1998). Accelerator model of investment means hereby that if the desired capital stock is proportional to output, then the investment in capital will be proportional to changes in output (see e.g. Bond et al., 1997).
14 Financing constraints as determinants of the investment behaviour of Estonian Firms
heteroscedasticity arising from varying size of firms. Capital stock (K) is measured as the net value of fixed tangible assets. The stock variables are measured at the end of the year; for instance, Kit is the value of capital of firm i at the end of year t.
A standard criticism to interpreting positive cash flow coefficients as evidence of financing constraints is that cash flow might actually proxy for the profitability of new investment projects. Since Fazzari et al. (1988) the strategy has been to split the sample by some criteria associated with problems of raising funds on the credit and capital markets and compare the relevance of inside firm liquidity between different sub-groups. Plausible criteria include inter alia firm size, firm age, the existence of close relationships with industrial or financial groups, the presence of credit rating or commercial paper programs, dividend policy etc. If for the class that is a-priori classified as financially constrained, the cash-flow sensitivity is significantly bigger and statistically more significant, then this is interpreted as evidence of the presence of financing constraints, assuming that profits have the same relevance as measure of profitability of new investment for different firms.
We split the sample along two lines. First we use firm size as a proxy for the ability to raise funds through external financing. The rationale is that firm size could be a proxy for firm age and other unobservable firm attributes that affect the degree to which public information about the firms’ investment projects is available. Small firms probably include many newly created de-novo firms, which lack credit history and collateral. It is also plausible that the transaction costs of obtaining funds contain a significant fixed cost component. The presence of such increasing returns suggests that the cost of obtaining external funds are higher for small than for large firms.5 It has also been emphasised in earlier studies that in transition economies the financing of small and medium sized firms is an important obstacle to growth (Pissarides, 1998). The sample is divided into three equally sized groups (noted as "small", "medium" and "large") according to the average size of real assets over the sample period. Real assets were calculated with GDP deflator.
One possible criticism to the usage of firm size as a criterion of whether particular firm is liquidity constrained or not, is that the costs of financing could decline with size due to a lower risk for the bank, not necessarily due to smaller information problems.6 Smaller firms in particular usually have a lower survival probability than large firms (Audretsch et al., 1999) and banks’ loan losses are found to be much higher for loans made to small firms in comparison to large firms (Churchill and Lewis, 1985). We offer two arguments against this criticism. First, the aggregate risk for banks is smaller in a portfolio consisting of several small loans than just a few big loans, because in the former case, due to the law of large numbers, the total return is more stable and the overall risk is smaller. Similarly, in
5 Oliner and Rudebusch (1989) found that transaction costs account for up to 25 % of the gross proceeds of small stock issues and one-seventh of the proceeds of small debt issues. 6 We thank one of the anonymous referees for drawing our attention to that issue.
15 Baltic Journal of Economics Autumn/Winter 2002
the insurance industry smaller risks are considered to be more insurable than large ones due to a better spread of claims over time. Secondly, if firms’ owners and banks had exactly the same information about project risk, then the required rate of return from the risky project is probably higher anyway, so the owners are less willing to finance these projects. The source of liquidity constraints (or that firms internal funds and profits are correlated) is the asymmetric information concerning projects returns, not just the possibility of the failure of the project.
After investigating the effect of firm size on investment-cash flow sensitivity, we tried to see whether there are any differences in the investment behaviour of firms owned by foreign capital versus those belonging to private domestic capital. The enterprises of the first group are at least partly subsidiaries of foreign parent companies (as argued by Kangur et al., 1999). So they could receive funds from the internal capital market of the international corporation, (as first studied by Hoshi et al., 1991), as well as receive cheaper and longer-term credits from foreign credit markets. We defined firms as belonging to foreign or Estonian private capital if in all years of the sample period (1995-98) more than 50 % of the share capital belonged to foreign owners or Estonian private capital.
In both classifications firms are not allowed to change their group affiliation, although in a rapidly developing economy this may be inadequate: small firms grow, their net worth increases, and more information on them becomes available, so firms’ financial constraint status may change.
Next we report results of estimating equations (4.1) for different sample splits. As stated, ‘fixed-effects’ or ‘within-groups’ estimators were used. This means that the deviations of variables from their firm-means were used in regressions. As the regression equation was not derived explicitly from any structural model, the parameters should be interpreted as partial correlation coefficients rather than estimates of structural coefficients. First, the results for different size groups (see table 3 below) indicate that the coefficients of both measures of internal liquidity (cash flow and cash stock) decrease monotonically with firm size. The same applies to the statistical significance of the parameters. This is evidence in favour of the hypothesis that large firms can more easily finance their investments and face less severe financing constraints. It is important to emphasize that because cash flow may actually proxy for the firms' investment demand, it is the difference in the estimated values of parameters that matters rather than just the size of the individual parameters. The t-statistic under the null hypothesis that small and medium size firms have the same cash flow coefficient is 2.54. The t-statistic under the null hypothesis that large and medium sized firms have the same cash flow coefficient is 2.34. This means that the difference is also statistically significant. Coefficients of leverage variable are negative for small and medium sized enterprises, but insignificant for large firms. It suggests that strength of balance
16 Financing constraints as determinants of the investment behaviour of Estonian Firms
sheet is more important for smaller firms. Parameters of the employment growth variable are significant in two out of three regressions, so hopefully we have been able to control for the existence of investment opportunities at least partially. Adjusted R2-s in investment equations are similar to the ones observed in other studies.
Table 3 Effects of employment growth, cash flow, cash stock and leverage on investments. Estonian manufacturing firms, sample split by firm size, 1996-1998.
Next the results for firms belonging to Estonian vs. foreign capital are presented. Let us first note that foreign firms tend to be much larger than domestic in terms of average value of assets (see table 2). In order to control for the firm-size effect we split the sample of domestic corporations ordered by the period's average real assets into 3 groups (48 firms each): small, medium and large enterprises. Similarly, the sample of foreign corporations was split into two groups (15 firms each). The foreign firms were divided into two groups due to the much smaller number of foreign firms in our database. As we can see from table 4, both cash flow and stock have a strong positive effect on investment for different groups of Estonian firms (except the cash stock for medium sized firms). In comparison to domestic firms, the coefficients are much smaller and less significant for both large and small foreign corporations. This finding was also robust to other specifications of the model not reported here (for example, when investment was regressed only on cash flow and cash stock etc.). It is interesting that the cash flow parameter for small foreign firms is smaller than that of large Estonian firms although the firms in the second group are about four times larger in terms of total assets (respectively 1.09 and 5.13 millions of USD). If only firm size affected cash flow – investments relationship, then the cash-flow parameter would be bigger among large Estonian firms, not among small foreign firms. The medium Estonian firms are almost of equal average size (1.05 million USD) to small foreign firms, but the cash flow parameter is about 60 % bigger in that group. We can conclude that affiliation to foreign capital significantly loosens financing constraints, increases investment and thereby supports firm growth. On the other hand the results should be treated with
17 Baltic Journal of Economics Autumn/Winter 2002
caution since the sample of foreign firms is quite small. Table 4 Effects of employment growth, cash flow, cash stock and leverage on investments: sample split by both firm size and ownership (Estonian/foreign owners)
Fazzari and Petersen (1993) argue that estimating equations like (4.1) underestimate the full long-run effect of financing constraints on fixed capital investments since firms smooth investment with working capital to maintain desired investment levels. So we also estimated the investment regressions that were augmented with the working capital investment variable. In order to account for the endogeneity of working capital investment, two-stage least squares estimation was used whereby the working capital investment was instrumented with cash flow, employment growth, beginning of period stock of working capital, firm- and year dummies. The results are not reported here due to lack of space (these are available upon request). In general the cash-flow coefficients increased significantly in size but the pattern across size and ownership classes remained the same. The sign of the working-capital investment variable after inclusion in the left side of regression (4.1) turned out to be negative. According to Fazzari and Petersen (1993) the last outcome should address the criticism that positive correlation between investment and cash flow arises because cash flow proxies for investment demand. The intuition is that if it is less costly to decrease working capital investments than fixed investments, liquidity constrained firms should in the periods of temporary cash flow shortfall decrease rather investments in working capital (up to drawing these to negative levels) than in fixed assets that generates the negative relationship between the two kinds of investments. The other possible way to modify the model concerns how far the variation of parameters is tested. Instead of dividing firms into sub-groups and then estimating the same equation separately for each group one could also use the expansion method defined by Casetti (1986).7 Let us have the initial model of the
7 We thank the anonymous referee for suggesting the usage of expansion method. Schiantarelli (1996) has also discussed and suggested the usage of interaction terms in the single investment equation when testing for liquidity constraints instead of grouping firms into sub-samples and then estimating the equation separately for each of them.
18 Financing constraints as determinants of the investment behaviour of Estonian Firms
form (4.2) and the expansion equation for parameters of the form
(4.2)
where FOR is the dummy variable indicating whether particular firm belongs to the foreign capital and SIZE is a measure of firm size defined as the natural log of the average value of firm’s assets. Then the terminal model becomes
(4.3)
Only the financial variables are expanded here in respect to ownership and size, as it is the variation in these variables that is of interest here. The advantage of model (4.3) is that it saves degrees of freedom, keeps the data together and explains the differences due to size and due to ownership in one model. Alternatively, one may argue that in the first model some variables are omitted, which we expect to be of importance (size, type of owners), and hence we would expect biased estimates. The estimation results are presented hereby in the table 5.
As the reader may see, the qualitative results still hold: both cash flow and cash stock have significant positive effect on investments (as shown by the positive values of parameters δ21 and δ31). For the domestic firm with average size (FOR=0 and SIZE=log(16 000 EEK)=9.68) 1 kroon increase in cash flow increases investments by 0.522 kroons (i.e. the value of parameter δ21 plus 9.68 times the value of δ22). The positive effect of liquidity declines both with firm size (due to the negative value for parameter δ22) and is smaller for foreign owned firms (negative δ23). The impact of the leverage or indebtedness variable on investments is still negative, but diminishes with the firm size (negative δ41 and positive δ42). Finally it seems not to matter much for the results whether the effect of liquidity is assumed to change with firm size continuously (like here) or discretely (results in tables 3 and 4).
19 Baltic Journal of Economics Autumn/Winter 2002
Table 5 Effects of employment growth, cash flow, cash stock and leverage on investments: the parameters of financial variables expanded with firm size and ownership
5 Test of liquidity constraints with Euler equation In order to strengthen the robustness of the above results we used the so-called Euler equation approach. The Euler equation is a relation between investment in successive periods derived from a dynamic maximization problem. Thereby the impact of expectations and profitability on investments are considered in a more plausible manner than simply estimating ad-hoc regressions. We used the model that has been derived and estimated by Bond and Meghir (1994). The later modifications of it have been used inter alia by Bond et al. (1997) for Belgium, France, Germany and United Kingdom data and by Lizal (1998) for Czech data.
Bond and Meghir (1994) have derived the model that will be presented here. The derivation of the model starts with the assumption that the firm’s objective is to maximize its net present value. In the absence of taxes it is given at the start of period t as
(5.1)
In the equation above Π stands for firms net revenues or profits, for capital, τ( ) Kt for investment, for variable factors and βt for discount factor between periods It Lt τ+1 t and t+1 that derives from nominal required rate of return as βτ 1 -1 . rt τ+1 =( +rt) Differently, value of firm can be expressed as the sum of discounted future profits:
(5.2)
The motion of capital stock over time is described with the equation δ , Kt= (1- )Kt-1+It where δ is the rate of economic depreciation. As usual, quadratic linearly homogenous function is assumed for adjustment costs (i.e. constant returns to scale
20 Financing constraints as determinants of the investment behaviour of Estonian Firms with respect to capital and investment):
(5.3)
Parameter b measures the size of adjustment costs; parameter c indicates the optimal investment/capital ratio based on the adjustment costs. The judgement for the quadratic form is based on the fact that higher deviations from the equilibrium are more costly than just an oscillation around the optimal level (Lizal, 1998). Differently, due to convex adjustment costs firms should try to divide the desired investment over several consecutive periods instead of making all the investment in one period. Firm’s net revenues Πτ will be then as follows:
(5.4)
The expression F(Kt,Lt) is the constant returns to scale production function, I wtis the vector of prices for the variable inputs Lt and pt denotes the price of investment goods. In order to allow for imperfect competition, pt is let to depend on output, with constant price elasticity of demand (ε >1). The optimal investment path can be described in terms of an Euler equation that under the aforementioned assumptions derives as follows (the complete derivation can be found at Bond and Meghir 1994):
(5.5) where Jit denotes user cost of capital. So, in the Euler equation (that relates marginal adjustment costs in adjacent periods), current investment is positively related to expected investment and to the current average profits (reflecting the marginal profitability of capital under constant returns), and negatively related to the user cost of capital.8 An attractive feature of the Euler equation model is that all relevant expectational influences are captured by one-step-ahead investment forecast (Bond et al., 1997). So, it should control for the usual criticism to other types of models, that financial variables do not capture the effect of liquidity constraints, but rather expectations of future profitability.
When expectations are replaced with realizations and parameters in the resulting regression equation are assumed to be constant over time, then the empirical specification of the Euler equation will be as follows:
8 Jorgenson (1965) introduced the notion of the user cost of capital. Absent taxes, the user cost is calculated as follows. First, the sum of opportunity cost of funds and depreciation minus expected appreciation of capital goods is calculated. Then the result is multiplied with the relative price of capital goods. See also equation (5.7).
21 Baltic Journal of Economics Autumn/Winter 2002
(5.6)
β φ β φ β φ β φ ε β φ α β β where 1=c(1- t+1), 0=(1+c) t+1, 2=- t+1, 3=- t+1/b( -1), 4=- t+1/b , 5=- 4 , φ δ α ε > t+1=(1+r t+1)(pt/pt+1)/(1- ) and =1-(1/ ) 0. The term Cit = ptYit - wtLit is cash flow and the term Yit =Fit - Gitis output net of adjustment costs. The regression variables are calculated from empirical data as follows. First, like previously investment (Iit) is measured as a change in fixed tangible assets plus depreciation. Output Yit is the sum of sales revenues plus change in finished goods inventories. Cash flow Cit is here defined as operating profits before taxes and interest plus depreciation of fixed 9 assets . Finally, the user cost of capital Jit is derived from the model as:
(5.7)
When we calculated the user cost of capital, two-digit producer price index I was used for pt and the deflator of gross capital formation expenditures for pt . The Statistical Office of Estonia publishes the data on both price indices. Required rate of returnrt was proxied with long-term interest rate on bank loans nominated in Estonian kroons in the beginning of the period. The Bank of Estonia publishes the data about the interest rates. Notice that quite often the user cost of capital term is eliminated from the model altogether and replaced with fixed time and firm effects (see e.g. Bond, Meghir, 1994). Following Whited (1992) the rate of economic δ δ depreciation i is calculated as = 2/L, where L denotes the estimated average life of capital goods, and Lt is calculated as Lt=(GKt-1+It)/DEPRt, where GKt-1 is the reported value of gross fixed tangible assets and DEPRt is reported depreciation.
In the equation (5.13) the term (Yit/Kit-1) is non-zero in the case of imperfect competition or non-constant returns to scale. Lagged investment terms consider the effect of adjustment costs. By estimating the equation we are controlling for the relation between current profits and expected future profitability (Gaston and Gelos, 1999). Under the null hypothesis of no financial constraints, the parameters β ≥ β ≤ β should satisfy conditions 1 1, 2 1 and 4<0. Under the alternative, the equation is misspecified and then one would expect a positive sign for the coefficient of the profit in the equation due to liquidity constraints.
Consistent estimates for the parameters of the described dynamic panel data model can be obtained with an appropriate method of moments (Bond and Mehgir, 1994). Ordinary least squares estimates of dynamic panel data models may lead to over- or underestimation of autoregressive coefficients (Bond et al., 1997). The estimator of Arellano and Bond is used here (Arellano and Bond, 1991). 9 Here the cash flow is calculated before taxes as the model disregarded taxes. Actually one may argue that for firms the relevant figure is rather the cash flow available after corporate income taxes. Section 6 discusses briefly the impact of changes in taxation on liquidity constraints.
22 Financing constraints as determinants of the investment behaviour of Estonian Firms
Instruments used are lagged one or more periods, because efficient GMM estimators will typically exploit a different number of instruments in each time period. (Doornik et al., 1999) The estimator proposed by Arellano and Bond for estimating the linear dynamic panel data models (as the one here) is a two-step estimator. In the first step, some known matrix is used as a weighting matrix. Here, it means using an identity matrix (as variables are not transformed, for example, into levels or orthogonal deviations). In the second step, the residuals of the first step will be used to produce an optimal new weighting matrix. Both estimates are consistent, but the two-step estimator is more efficient if the residuals from the first step are heteroscedastic. On the other hand, in small samples the estimated standard errors of the second-step estimates are usually biased downwards, so (as usual) results of both first- and second step estimation are reported. Finally, the Sargan test of overidentifying restrictions is reported to check for the validity of instruments, as well as tests of serial correlation in residuals. The model was estimated with DPD (Dynamic Panel Data) program written in Ox programming language by Doornik, Bond and Arellano (Doornik et al., 1999). In order to measure the different effect of liquidity constraints on investment through different size classes of firms, interaction dummies are used for small, medium and large firms
(denoted as Dsmall,Dmedium,Dlarge) as well as domestically incorporated vs. foreign firms ( Dest,Dfor"est" is for "Estonian" and "for" for "foreign").
Results in table 6 indicate that cash flow affects investment positively, so the null hypothesis of the absence of liquidity constraints is rejected. Also, as expected, internal funds are more important for investments among small and medium sized firms. So smaller firms have more liquidity constraints than the big ones, but as the sign of cash flow is positive even for the latter, they may also have some problems in raising funds. The parameters of the lagged investment terms are correctly signed, but smaller in absolute value than suggested by the model in the absence of liquidity constraints. Output term is non-zero, which can be due to either imperfect competition or nonconstant returns to scale. Also the parameter of the user cost of capital has the expected sign and its two-step estimate is significant at 10 per cent. Sargan test (the test of overidentifying restrictions) shows that the instruments are not correlated with the residuals, which confirms the validity of the chosen instruments. This is also shown by the lack of autocorrelation in residuals. The total number of moment restrictions generated by instruments was 43 (7 for the first, 14 for the second and 21 for the last period; the one extra instrument is the constant term).
23 Baltic Journal of Economics Autumn/Winter 2002
Table 6 Results of estimating Euler equation: effects of firm size on cash- flow investment relationship
Table 7 in columns (i) and (ii) presents the results of the estimation of equation (5.6) when firm ownership dummies are interacted with cash flow variables. As we can see, cash flow coefficient is significant for firms owned by Estonian capital but not for the group owned by foreign capital. This is in accordance with the previous results from reduced-form equations. The coefficients for the other right-hand–side variables are correctly signed and their statistical significance has increased in comparison with the previous table. In order to control for the different average size of Estonian and foreign firms the two sub-samples were further divided according to size. That was done with interaction dummies Dsmall_est, Dmedium_est, Dlarge_est, Dsmall_for, Dlarge_for. So we have five groups of firms as in table 4 of section 4. As we can see from table 7, the cash-flow sensitivity of Estonian firms decreases with size, but that of foreign firms actually increases (coefficient is negative for small, but positive for large firms)!
The latter finding is not consistent with the hypothesis about presence of liquidity constraints. The finding of a larger cash flow parameter in the group of large firms is sometimes interpreted as an evidence of over-investment among large firms due to managers’ incentives to cause firms to grow beyond optimal size (Vogt, 1994). On the other hand the over-investment hypothesis would rather apply to firms with dispersed than concentrated ownership structure, because then the shareholders’ control over managers actions is weaker (Schaller, 1993). Although we do not have more detailed information regarding firms’ ownership structure that is in general not the case of foreign owned firms in Estonia. Another argument could be that the over-investment may occur not only when ownership is dispersed but also when it is remote, and this is the case for foreign firms in Estonia.10 Still, as
10 We thank the editor for suggesting this argument in favour of the over-investment hypothesis.
24 Financing constraints as determinants of the investment behaviour of Estonian Firms
most foreign investments have come into Estonia from geographically close countries such as Sweden and Finland, it may not be too hard for foreign owners to constantly control the actions of local managers. Also, the scatter plot of investment-capital ratio versus firm size (log of assets) seemed to provide no evidence whatsoever for over-investment by large foreign firms. The counterintuitive results here could be due to the small sample size – there are only 15 firms in the group of large foreign owned firms that is well below the usual size of firm group in similar studies. It is important to mention that the various simpler regressions in this paper did not show significant effects of cash on large foreign firms’ investments. Thus we could rather remain with the conclusion that affiliation to foreign owners is associated with better possibilities to finance profitable investments than see liquidity constraints affecting foreign owned firms.
Table 7 Results of estimating Euler equation: effects of firm size and ownership on cash-flow investment relationship
6 Conclusions and implications This paper discussed the relationship between financing conditions and investments in Estonian manufacturing firms. We argued that financing constraints should be quite material determinants of investment levels for many firms, in particularly small firms and domestically incorporated firms (as compared to firms owned by foreign capital). We found from both simple OLS regressions and the more elaborate Euler equation that small (and Estonian) firms are more dependent on their own cash flow and cash stock than larger (and non-Estonian) firms. We
25 Baltic Journal of Economics Autumn/Winter 2002
interpreted these results to be the evidence of the presence of the liquidity constraints because inside firm funds influenced investments significantly more exactly in those enterprises that we assumed to be more financially constrained. A second argument in favour of our interpretation is that cash flow affected investments even when investment opportunities (or the profitability of investments) were controlled.
This study confirms the claims of Raudsepp and Leoshko (1998) that financial problems are more acute for small and medium-sized enterprises in Estonia. Secondly, most previous studies of liquidity constraints have not included the lower tail of the size distribution of firms (i.e. firms whose size is well below the average firm size) in the sample because they were made with databases on firms listed in stock exchanges. For example, in a thorough study of cash flow-investment relationship and firm size made by Kadapakkam et al. (1998) small firms had average assets of $ 57 million in the United States, £ 36 million in Great Britain and DM 355 million in Germany. So the results here may have importance for investment literature in general because they show that there is a negative relationship between investment-cash flow sensitivity and firm size among enterprises with assets ranging from less then 0.5 million US dollars to the maximum of about 50 million USD.
Another issue concerns government subsidies and whether they should be more actively used to solve the financing problems of small and medium sized enterprises by reducing the cost of capital and improving access to long-term sources of external funding. In particular, Raudsepp et al. (2000) are concerned about the lack of subsidizing of small firms in Estonia resulting in too high cost of capital exceeding the internal rate of return of investment projects. So the results of this study can be viewed to confirm these claims. This study showed that small firms might indeed have problems in financing their investments, however it is not sure whether subsidies would solve the problem. For instance, Demirguc and Maksimovic (1996), for the thirty-country sample, found no evidence that government subsidies were associated with the ability of firms to grow faster than when relying only on the their internal resources. Rather one can see here a role for venture capital funds targeting at providing finance for growing small and medium- sized businesses – as we saw, small and medium sized firms receive substantially less new equity capital.
One other possible implication of the results is related to the recent changes in taxation of corporate income. In Estonia retained earnings are exempt from taxation since 2000. Thereafter, corporate income tax in Estonia can be viewed like the "cash flow tax" suggested by Fazzari et al. (1988b). The notion means that only the share of profits that exceed investments are taxed. On the assumption that firms
26 Financing constraints as determinants of the investment behaviour of Estonian Firms
follow the financing hierarchy (investments are first made from retained earnings till these are exhausted) and pay dividends only when they do not have profitable investment projects, the new Estonian corporate income tax is similar to a "cash flow tax". This kind of change in taxation lowers average tax rates for growing firms that invest all of their internal finance and thereby increases their capital expenditures. On the other hand, for mature firms with internal cash flow excess of investments marginal incentives to invest are preserved, so the overall level of investments will increase. Based on a model from the Tobins q theory, Funke (2001) predicted that capital stock would increase 6.1% in the long run due to the year 2000 income tax law. According to the previous arguments financing constraints may amplify this positive effect.
The other main finding of the analysis was that cash flow is not an important determinant of investment for foreign owned firms. Indeed, this is not a new result for transition economies; Lizal and Svejnar (2000) reached the same conclusion when studying the investments of Czech enterprises. However, this finding might well serve for further discussion and interpretation. One way in which liquidity constraints could be relaxed is through the development of the banking sector: if banks become more capable of monitoring loan applicants then the asymmetric information problems will be reduced and profitable investments are more likely to receive outside funds. But the development of banking may take quite some time. On the other hand, the inside flow of foreign direct investment may relax binding liquidity constraints much faster by enabling firms to obtain funds from foreign debt and capital markets.
References
Anderson, R., Kegels, C., 1997. "Finance and Investment in Transition: Czech Enterprises, 1993-1994." William Davidson Institute Working Paper, No. 164, 18 pp.
Arellano, M., Bond, S., 1991. "Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application of Employment Equations". Review of Economic Studies 68, 277-97.
Audretsch, D.B., van Leeuwen, G., Menkveld, B., Thurik, A.R., 1999. Are Small Firms Really Sub-Optimal? Compensating Factor Differentials in Small Dutch Manufacturing Firms. EIM, Resarch Report 9902/E, 47pp.
Bond, S., Elston, J., Mairesse, J., Mulkay, B., 1997. "Financial factors and Investment in Belgium, France, Germany and the UK: A Comparison Using Company Panel Data." NBER Working Paper, No. 5900.
27 Baltic Journal of Economics Autumn/Winter 2002
Bond, S., Jenkinson, T., 1996. "The Assessment: Investment Performance and Policy." Oxford Review of Economic Policy, 12, 1-29.
Bond, S., Meghir, C., 1994. "Dynamic Investment Models and Firm’s Financial Policy." Review of Economic Studies 61, 197-222.
Bratkowski, A. Grosfeld, I., Rostowski, J., 2000. "Investment and Financing in de novo Private Firms: Empirical Results from the Czech Republic, Hungary and Poland." Economics of Transition 8, 101-16.
Budina, N., Garretsen, H., de Jong, E., 2000. "Liquidity Constraints and Investment in Transition Economies: The Case of Bulgaria". The Economics of Transition 8, 453-65.
Casetti, E., 1986. The Dual Expansion Method: An Application for Evaluating the Effects of Population Growth on Development. IEEC Transactions on systems, Man, and Cybernetics, Vol. SMC-16, 29-39.
Churchill, N. C., Lewis, V. L., 1985. "Profitability of small business lending." Journal of Bank Research, 16, 63-71.
Coricelli, F., 1996. "Finance and Growth in Economies in Transition". European Economic Review, 40, 645-53.
Demirgüc-Kunt, A., Maksimovic, V., 1996. "Financial Constraints, Uses of Funds and Firm Growth: an International Comparison." World Bank Working Paper No. 1671, 25 pp.
Doornik, J. A., Arellano, M., Bond, S. 1999. Panel Data Estimation using DPD for Ox. Mimeo Evans, David S. and B. Jovanovic. "An Estimated Model of Entrepreneurial Choice under Liquidity Constraints," Journal of Political Economy, IIIC (1989), 808-27
Fazzari, S., Hubbard, R. G., Petersen, B., 1998a. "Financing Constraints and Corporate Investment." Brookings Papers on Economic Activity 1, 141-95.
Fazzari, S., Hubbard, R. G., Petersen, B., 1998b. "Investment, Financing Decisions, and Tax Policy." American Economic Review, 78, 200-5.
Fazzari, S., Petersen B. C., 1993. "Working Capital and Fixed Investment: New Evidence on Financing Constraints." RAND Journal of Economics, 24, 328-42.
28 Financing constraints as determinants of the investment behaviour of Estonian Firms
Funke, M., 2001. Taxation and investment: the impact of the year 2000 income tax law in Estonia. Unpublished paper, 12 pp.
Hubbard, R. G., 1998. "Capital-Market Imperfections and Investment." Journal of Economic Literature, Vol. 36, 193-225.
Jorgenson, D. W., 1963. "Capital Theory and Investment Behavior." American Economic Review, 53, 247-59.
Kangur, A., Rajasalu, T., Randveer, M., 1999. Kapitali liikumine ja ettevõtluse rahastamine. Eesti Panga Toimetised, No. 4.
Kaplan, S., N., Zingales, L., 1997. "Do Financing Constraints Explain Why Investment is Correlated with Cash Flows?" NBER Working Paper, No. 5267.
Konjunktuur, 1999. Eesti Konjunktuuriinstituut, No. 131.
Lizal, L. Svejnar, J., 1998. "Enterprise investment during the Transition: Evidence from Czech Panel Data", CEPR Discussion Paper, No. 1835, 1998.
Lizal, L., 1998. "Does a Soft Macroeconomic Environment Induce Restructuring on the Microeconomic Level during the Transition Period? Evidence from Investment Behavior of Czech Enterprises." The William Davidson Institute Working Paper, No. 235.
Lizal, L., Svejnar, J., 1999. "Financial Conditions and Investment during the Transition: Evidence from Czech Firms." CERGE-EI Working Paper, No. 153.
Oliner, S. D., Rudebusch, G. D., 1992. "Sources of Financing Hierarchy for Business Investment." The Review of Economics and Statistics, 74, 643-54.
Pärn, H., Lumiste, R., 2000. Tööstus. In: Rääts, E. (Ed.)., Eesti uue aastatuhande lävel. TEA Kirjastus, Tallinn, 115-21.
Perotti, E. S., Gelfer, S., 1998. "Investment Financing in Russian Financial- Industrial Groups." Russian European Centre for Economic Policy Working Paper, No. 3.
Pissarides, F., 1998. "Is lack of funds the main obstacle to growth? The EBRD’s experience with small and medium-sized businesses in central and Eastern Europe." EBRD Working Paper, No. 33, 1998.
29 Baltic Journal of Economics Autumn/Winter 2002
Prasnikar, J., Svejnar, J., 1998. "Investment and Wages during the Transition: Evidence from Slovene Firms." The William Davidson Institute Working Paper, No. 184.
Raudsepp, V., Leoshko, T., 1998. Financing Problems in Small and Medium-Sized Enterprises in Estonia. Unpublished Working Paper, University of Tartu, 1988.
Raudsepp, V., Sander, P., Kask, K., 2000. Patterns of Financing Small and Medium-Sized Enterprises in Estonia. Unpublished Working Paper, University of Tartu, 2000.
Schaller, H., 1993. "Asymmetric Information, liquidity constraints, and Canadian Investment." Canadian Journal of Economics 26, 552-74.
Schiantarelli, F., 1996. "Financial Constraints and Investment: Methodological Issues and International Evidence." Oxford Review of Economic Policy, 12, 70-89.
Statistical Yearbook of Estonia 2000, 2000. CD-ROM Issue, Statistical Office of Estonia.
Vogt, S. C., 1994. "The Cash Flow Investment Relationship: Evidence from U. S. Manufacturing Firms." Financial Management, 23, 3-21.
Whited, T. M., 1992. "Debt, Liquidity Constraints and Corporate Investment: Evidence from Panel Data." The Journal of Finance, 48, 1425-60.
30 Gender Wage Differences in Soviet and Transitional Estonia
Gender Wage Differences in Soviet and Transitional Estonia1 Charles Kroncke2 and Kenneth Smith3
Abstract We use the retrospective, covering the years 1989 - 1994, Estonian Labor Force Survey to examine potential wage discrimination against women. We look at full- time workers of Estonian and Russian ethnicity in the years 1989, the Soviet period, and 1994, the last full year of the survey and three years after Estonian independence and the beginning of the transition to a market economy. We find substantial evidence of wage discrimination against women in both years. In fact, despite the official rhetoric of gender equality in the Soviet Union, our results indicate the relative level of wage discrimination against female workers in Estonia changed very little between 1989 and 1994 when occupational dummies are excluded from the wage equations.
Keywords: Gender, wage discrimination, wage decomposition JEL-Code: J71, P23
I. Introduction In the former Soviet Union gender related wage and employment differences were not widely analyzed due to the official declaration of equality between men and women. The breakup of the Soviet Union has opened this topic up to discussion and analysis. In 1995, a detailed retrospective labor force survey was released in Estonia. The survey covers the 1989-1994 period. Thus it contains empirical data representing both the end of the Soviet period and the beginning of the transitional period to a market economy.
We use the Estonian Labor Force Survey (ELFS) to measure potential wage discrimination against women. The years 1989 and 1994 are analyzed as 1989 represents the Soviet period (to our knowledge, detailed analyses of gender wage differentials in the Soviet Union do not exist), and 1994 represents the last full year
1 We would like to the thank James Long, Dorothe Bonjour, Alf Vanags, Bruce Smith, Jens Larsen, participants of the European Association of Labour Economists 1998 meeting in Blankenberge, Belgium, and an anonymous referee for useful comments on earlier drafts. Dmitri Kulikov provided excellent research assistance. Any remaining errors are the sole responsibility of the authors. 2 Gordon College. E-mail: [email protected] 3 Millersville University. E-mail: [email protected]
31 Baltic Journal of Economics Autumn/Winter 2002
of the survey and a year two years past the start of substantive economic reform in Estonia.4 The data provide substantial evidence of significant wage differentials between men and women in both 1989 and 1994 that cannot be accounted for by productivity relevant characteristics available in the data set. The relative level of the wage differentials explained by the data are quite close in 1989 and 1994 when occupational dummies are not used in the wage regressions. This would imply approximately the same amount of wage discrimination against women in the Soviet period and the more recent transitional period. When the wage regressions include occupational dummies, a significantly greater portion of the wage differential is explained in 1989 than in 1994. Throughout this study we focus on full-time workers.
A caveat is in order when examining earnings inequality in the Soviet and transition periods. It is likely that inequality in material standard of living is closely related to income inequality in a market or transition economy. However, in the Soviet Union, due to goods shortages and the provision of services by the state, it is likely that material well-being was more evenly distributed than was income.
This paper has five sections. Section II presents a brief literature review and an overview of models of wage discrimination. Section III gives a more detailed description of our data and methodology. Empirical results are presented in Section IV, and Section V presents concluding remarks.
II. A Model of Wage Discrimination Our method of estimating gender wage discrimination in Estonia is based on the standard Becker (1971) model that was formalized for empirical testing by Oaxaca (1973). We assume that, in the absence of discrimination, an individual’s actual wage (W) is equal to his or her marginal productivity (W*). In the absence of discrimination the following equality should hold:
(1)
Where bar notation indicates mean values (the M subscript denotes male and the F subscript denotes female). If this equality does not hold, we take it as evidence of gender-based wage discrimination. The larger the deviation from equality, the greater the evidence of wage discrimination. Oaxaca used this model to measure gender-based wage discrimination in the U.S. using wage decomposition methodology. The first step in the wage decomposition procedure is to use OLS to estimate the returns to certain individual characteristics (education, experience, etc.) and other work-related factors (industry, region, etc.) that affect
4 Recently, the results of the 1997 ELFS were released. This data covers 1995-1997. However, this data does not use the same sample of individuals as the 1995 ELFS. Thus, a comparison of 1997 (or 1996 or 1995) and 1989 would be more problematic than a 1989-1994 comparison.
32 Gender Wage Differences in Soviet and Transitional Estonia
productivity. Once these estimates are obtained they may be used to decompose the male-female wage differential into two components; one component is the result of differences in productivity relevant characteristics, and the second unexplained component is generally attributed to discrimination.
β β The following is a brief overview of wage decompositions. If M and F represent the OLS coefficient estimates for vectors of personal characteristics and work-related factors XM and XF, then the mean preserving nature of OLS regression implies that:
(2)
If there is no wage discrimination in the labor market, then we should be able to make the following substitutions:
(3)
Equations (2) and (3) imply that we should be able to substitute the returns to individual characteristics for males and females without affecting wages if no discrimination is present. In other words, in the absence of discrimination, differences in personal characteristics and job related factors (that may themselves be a result of discrimination) should cause wage differentials not differences in the returns to those characteristics and factors.
Employing algebraic manipulation and taking the natural logs of individuals’ wages, Oaxaca decomposed the natural log wage (ln(wage)) differential as follows:
(4) (5) (6)
The unexplained portion of the wage differential (equation (6)) is then attributed to potential discrimination. Using natural logs implies in equations (4) - (6) and throughout the paper that:
(7) where n is sample size. The way the decomposition is presented above implies that eliminating discrimination would tend to pull up female wages toward the male level. Conversely, one could perform an analogous decomposition applying the
33 Baltic Journal of Economics Autumn/Winter 2002
female coefficient estimates to male characteristics. This would imply that eliminating discrimination would tend to pull male wages down toward the female level. Oaxaca recognized this "index problem". Since the two decompositions yield very different results, the standard practice is to do both decompositions and recognize that the actual level of discrimination is somewhere between the two indicated levels5.
The imprecision of this method is certainly unsatisfactory. The decomposition methodology has been refined by Cotton (1988). Cotton’s primary, and quite logical, criticism of the standard decomposition method, was that, were discrimination suddenly eliminated, we should expect the wages of the majority group to fall and those of the minority group to increase. This fact was implicitly accepted by Oaxaca - hence the two decompositions. Cotton thus reasoned that the returns to personal characteristics and other factors would, in the absence of discrimination, be a weighted average of the two groups with the weights being equal to the respective groups’ proportion of the population or sample within the data set. In our case this implies using the following in our decompositions:
(8) where α represents the weight placed on the male coefficients.
Employing Cotton’s method, we can decompose the total ln(wage) differential, ln(WM) - ln(WF), as follows: (9) (10) (11) Equation (9) can be rewritten as:
(12)
Equation (12), in turn, implies:
(13)
5 The decomposition shown tends to attribute more of the wage differential to discrimination than would applying the female coefficient estimates to male personal characteristics. For a detailed explanation of this phenomenon, see Cotton (1988).
34 Gender Wage Differences in Soviet and Transitional Estonia
Thus, the decomposition methodology yields a direct means of comparing the ratios presented in equation (1).
The implication of this decomposition is that discrimination is manifested in two ways: the first through the overvaluation of the male group’s characteristics and the second through the undervaluation of the female group’s characteristics. Cotton’s methodology is used in this paper and recently has been used to study ethnic wage discrimination in the Netherlands (Kee (1995)).
This study is not the first analysis of male-female wage differentials in Estonia. Kandolin (1996), using 1993 survey data, found that women’s wages were 82% of men’s wages in her sample.6 In this study, the author estimates that if Estonian women were to receive the same returns as men for their education, job tenure, and family responsibilities, their wages would be 108% of men’s wages.
Using the more recent, more detailed, and much larger ELFS, our findings for adjusted wages in 1994 are rather similar to Kandolin’s 1993 result. As mentioned, the ELFS also allows for comparisons across economically divergent years, and allows us to compare how women fared relative to men in 1989 as well as in the early transition. The ELFS data indicate that women’s relative wage position (in financial terms and in terms of the part of the financial differential that can be explained by the data) in a transitional economy might not have changed by as much as many would think.
Further our results indicate that gender inequality in the Soviet Union may have changed relatively little in the wake of Gorbachev’s economic reforms. The levels of gender inequality we find in 1989 are consistent with what little evidence exists for earlier Soviet periods. For example, Vinokur and Ofer (1987) find evidence of substantial gender earnings inequality in the mid-late 1970s.
III. Methodology and Data Here, the wage decomposition used is as demonstrated in equations (8) - (11) where β* is a weighted average of the OLS coefficient estimates for males and females. The weights used are the groups’ relative proportions in the data samples (as in Kee). Four wage decompositions are constructed: two each for 1989 and 1994. In each year, one decomposition is constructed without occupational dummies and one is constructed using occupational dummies.
The data are from the retrospective 1995 ELFS. In all, 9608 people between the ages of 15 and 74 were interviewed. The interviewees were selected at random and distributed throughout Estonia. The results of the interviews were compiled by
6 Only ethnic Estonians are included in the Kandolin study.
35 Baltic Journal of Economics Autumn/Winter 2002
the Commission for Population and Social Statistics Working Group for Labor Force Surveys. The Working Group was initially formed by the Estonian Labor Ministry. The interviews took place in early 1995. Interviewees were asked questions concerning their work history (that is their entire employment record) during the 1989 - 1994 period. The retrospective nature of the survey also makes the choice of comparing 1989 and 1994 logical. Unemployment in 1989 was essentially nonexistent in Estonia and the stability of the ruble and of incomes in general make the wage data relatively reliable. For the 1994 portion of the survey, the questions pertained to a period only a few months prior to the interviews.
While, in general, the retrospective nature of the survey may tend to cast doubt on data reliability, this survey represents the only detailed labor data available for Estonia during the late Soviet - early transition period. Further, the retrospective nature of the survey does largely eliminate the possibility of selection bias when comparing 1989 and 1994. We do not need to worry about the possibility of losing people to emigration, death, or failure to report between our 1989 and 1994 samples as all respondents were interviewed in early 1995 about their work histories. Thus, most of the people in our 1989 sample are also present in our 1994 sample. There are normal changes due to the retirement of older workers and young people entering the labor force between 1989 and 1994. There was also a significant decline in labor force participation rates between 1989 and 1994. This decline was greater for women than for men (Eamets et al. (1997)). This change in labor force participation is also reflected in our samples.
Table 2 presents mean values of characteristics for the two groups in both 1989 and 1994. Definitions of the variables are in Table 1. Wages and natural logs of wages for 1989 are in Soviet rubles (RB). The ruble was stable throughout 1989 at an approximate value of 1 RB = 1.67 USD (this was the official, if artificial, exchange rate (Buckley and Ghauri (1994)). Wages and natural logs of wages for 1994 are in Estonian Kroons (EEK). In December 1994, the EEK/USD exchange rate was 12.57 (Estonian Institute of Economic Research (1997)). The EEK was pegged to the German Mark at a rate of 8 EEK = 1 DEM. The mean values for the dummy variables can be interpreted as percentages. For example, the values for ENGLISH in 1989 imply that 17.82 percent of males and 21.64 percent of females in our respective samples indicated an ability to speak English at that time. In our regressions, ln(WAGE) is the dependent variable. All other variables in Tables 1 and 2 are used as regressors except WAGE.
In many respects, our choice of independent variables is comparable to other studies of wage discrimination (Oaxaca, Cotton, Kandolin, and Kee all serve as examples). Naturally, the unique circumstances of Estonia and the limitations of the data make our choice of regressors somewhat unique.
36 Gender Wage Differences in Soviet and Transitional Estonia
English language ability is used as English now serves, and has served, as the primary language of business between Estonia and its primary new trade partners in the Nordic countries, the EU, and the U.S. English language ability tends to be in high demand in Estonia though it is far from a universally spoken language. ETHNIC is included to account for wage differences that may be the result of differences between Estonians and ethnic Russians7. TALLINN and NARVA are our two choices for regional dummies. Tallinn is the capital and the largest city of Estonia. In fact, Tallinn is the only "large" city in Estonia with a population of about 450,000. The next largest city, Tartu, has less than one-fourth the population of Tallinn. Narva, in northeast Estonia, was where Estonia’s heavy industry was concentrated during the Soviet period. This heavy industry is now largely idle and has been since the breakup of the Soviet Union. The Narva region was in 1994, and still remains, a depressed region in Estonia.
The other variables in Tables 1 and 2 are standard measures of returns to experience and job tenure and a series of industry dummies. There is a debate as to whether occupational dummies should be used in wage discrimination regressions as wage differentials may be the result of discrimination against women (or racial or other minorities) wishing to enter certain high-paying occupations. Dolton and Kidd (1994) found little evidence of occupational segregation amongst the sexes leadng to unexplained wage differentials in the United Kingdom. However, Gill (1994) found evidence that differential access to high-paying occupations played a significant role in racial wage differentials in the U.S. The implication here is that there may be (or may have been in Soviet times) barriers in place preventing women from entering some high-paying occupations in Estonia. If so, women might become concentrated in low-wage occupations. Thus the inclusion of occupational dummies could be inappropriate leading to a severe underestimation of actual wage discrimination.
Bearing this in mind, we do examine male-female wage differentials including occupational dummies. Including occupation has a significant effect on the explained differential in 1989 but very little effect in 1994. As is discussed below, it is quite plausible that the lack of occupational effects in 1994 is due to labor market restructuring that occurred during the early transition period. Table 3 presents employment by occupational category.
IV. Empirical Results Since empirical estimations of wage discrimination are derived from profit maximization models where the wage is equal to the marginal product of labor, it is legitimate to ask whether this type of model is appropriate for looking at wage
7 Only ethnic Estonians and Russians are included in this study. Ukrainians, Belarussians, non- Estonian Balts, Finns, etc., make up a total of about seven percent of the Estonian population. (UNDP (1996)).
37 Baltic Journal of Economics Autumn/Winter 2002
differentials in the former Soviet Union or any other planned economy. While it is unlikely that profit maximization is a good hypothesis for Soviet enterprises, the empirical estimations are truly only estimating returns to certain factors relevant to one’s work. Thus, one would assume, that in a socialist economy where equality was in principle a key objective of the state, it should also hold true that individuals are compensated equally for like characteristics. In fact, one should expect that the main difference between wages in a socialist economy and a market economy would be that wage differentials for disparate characteristics are minimal in a socialist economy. Thus, while the initial discussion of wages and marginal productivity may not be relevant for Soviet Estonia, we still believe the empirical wage decomposition model is a legitimate and interesting means of exploring wage differentials in a socialist economy. The results bear this belief out.
Table 4 presents the OLS regression results for 1989 when occupation is excluded, and Table 5 presents the OLS regression results when occupation is included.
To perform the wage decomposition for 1989, we substitute our regression coefficients into equations (9) - (11). β* is computed as in equation (8) with α = 0.492. This yields the following:
Occupation NOT included (14) (15) (16)
The observed female-male wage ratio and the adjusted (productivity) female-male wage ratio from equation (14), respectively, are:
(17)
Occupation included (18) (19) (20)
38 Gender Wage Differences in Soviet and Transitional Estonia
Equation (18) then yields the following adjusted wage ratio:
(21)
The mean ln(wage) differential in the 1989 samples (male-female) is 0.4221 (without occupation) and .4220 (with occupation - this difference reflects minor rounding errors when using the coefficient estimates and mean values to perform the decompositions).
Table 8 presents a specific breakdown of the 1989 ln(wage) differential by individual characteristics, region, and industry. As equation (17) indicates, slightly less than 16 percent of the female-male wage differential can be explained by the data when occupation is excluded. While women made 34.5 percent less than men did in 1989, their estimated productivity was only 5.4 percent less. When occupational dummies are included (see Table 9 for the decomposition results) in the wage equations, almost 38 percent of the female-male wage differential is explained by the data. Thus, even with occupation, there is still substantial evidence of wage discrimination against women in Soviet Estonia. In both decompositions, the unexplained wage differential is, roughly, split evenly between an undervaluation of female characteristics and an overvaluation of male characteristics.
Table 6 presents the regression results for 1994 with occupation excluded, and Table 7 presents the regression results including occupational dummies.
Again, for 1994, we substitute our regression coefficients into equations (9) - (11) with β* computed as in equation (8) (α = 0.508). This yields the following results: Occupation NOT included
(22) (23) (24)
In 1994, the observed and adjusted wage ratios are:
(25)
39 Baltic Journal of Economics Autumn/Winter 2002
Occupation included
(26) (27) (28)
With occupational dummies included, the adjusted wage ratio becomes:
(29)
The mean ln(wage) differential in 1994 is 0.3094 (without occupation) and 0.3085 (with occupation).
In this case, with occupation excluded, slightly more than the entire actual ln(wage) differential is unexplained implying women should have earned more than men given the characteristics we have examined and the data we have. Again the unexplained differential is almost evenly split between an overvaluation of male characteristics and an undervaluation of female characteristics, and again Table 8 presents a specific breakdown of explained wage differentials by variable. When occupation is included (see Table 9), equation (29) indicates women are slightly less productive than men. However, the inclusion of occupation has remarkably little effect in 1994. With or without occupation included, the data indicate men and women should have had virtually identical labor earnings even though, in reality, women only earned about 69 percent of what men did in our samples.
Table 8 shows that, taken together, the industry dummies are the most important factor explaining the 1989 male-female wage differential in the absence of occupational effects. Personal characteristics, as might be expected, played a relatively small role during the Soviet period. However, they along with region favored higher female wages overall. However, differentials due to the sector one worked in were quite important determinants of individual wages (as indicated by Table 4) as well as male-female differentials. Two sectors, agriculture and fishing (AGFISH) and mining, manufacturing, electricity, and construction (MMEC), paid both men and women particularly well in 1989. As Table 2 indicates, these two sectors employed over 70 percent of all male full-time workers. However, the same sectors employed only 46.5 percent of female full-time workers. Conversely, the educational (EDUC) and health care (HEALTH) sectors were two of the poorer paying areas even in 1989. While these two sectors accounted for only 4.1 percent of male full-time employment, 19.4 percent of women working full-time were employed in EDUC or HEALTH. Clearly there was a high degree of industrial
40 Gender Wage Differences in Soviet and Transitional Estonia
segregation by gender in Soviet Estonia with men dominating employment in the high-paying sectors and women dominating employment in the low-paying sectors. If this fact itself is a result of gender discrimination, it is possible our results in equations (14) - (16) significantly underestimate overall gender wage discrimination in 1989.
Table 9 illustrates the importance of occupational effects in 1989. Occupational effects dominated the explained 1989 wage differential and, together with sectoral effects, accounted for slightly more than the entire explained differential. Two occupational categories had positive regression coefficients for both men and women in 1989: management and skilled labor. As Table 3 indicates, these two occupational categories accounted for about 78.6 percent of male employment. However, the same two occupations employed only around 32.5 percent of women. Conversely, women dominated employment in low-wage occupations. Four occupational categories had negative OLS coefficients for both genders - technical, clerical, service, and unskilled labor. These occupations employed over 47.5 percent of all women working full-time in 1989 but just 9.1 percent of men. While the data do not allow for testing of occupational segregation in 1989 (at least not by conventional methodology employed by Gill and Dolton- Kidd), it is difficult to believe that occupational segregation in Soviet Estonia was not due, at least in part, to discrimination. It is likely that this was also true of sectoral segregation.
Personal characteristics were far more significant in 1994, and the overall significance of sector diminished somewhat (see Table 8). Table 6 shows employment in AGFISH continued to significantly influence wages for both men and women in 1994. However, in 1994 working in the AGFISH sector tended to depress wages - hurting far more men than women. Men continued to dominate employment in the still fairly lucrative MMEC sector. Women were still hurt by their relatively high concentration in the low-paying health care and education sectors. As in 1989, differences in personal characteristics tended to favor women - particularly education (EDU) and English language ability. Women on average had approximately one-half more year of education than men. Further, as market forces (marginal productivity) tended to drive wages more, the returns to education increased sharply8. The fact that women were considerably more likely to speak English and work in Tallinn also had significant positive effects on their wages. In 1994, these personal characteristics dominated sectoral wage effects with the end result that, according to our estimations, women should have earned slightly more than men.
8 Increasing returns to education during the economic transition to a market economy have been found elsewhere in Central and Eastern Europe. For a detailed example, see Rutkowski (1996).
41 Baltic Journal of Economics Autumn/Winter 2002
Table 7 presents the effects of individual variables on wages in 1994 with occupational dummies included in the wage regressions. The net effect of occupation is extremely small and the overall explained wage differential changes only slightly with the inclusion of occupation. In fact the direction of the overall occupational effect is unique to our knowledge. As mentioned, many economists argue that including occupation is problematic as it may lead to an underestimation of wage discrimination against women (or other minority groups) who likely face discrimination when trying to enter high-wage occupations. However, in transitional Estonia, the net effect of occupational dummies actually favors women - if only slightly. While the high level of occupational segregation may well have been the result of discrimination against women in Soviet Estonia, the nature of the economic transition seems to have worked against men as far as occupation is concerned. In 1994, three occupational categories had positive coefficients for both sexes - management, professional, and technical. It is not surprising in a market economy that the three "white-collar" occupations requiring specialized training would receive wage premiums. While men were considerably more likely to be in managerial positions in 1994 (15.04 percent of men as opposed to 9.57 percent of women held such positions), women dominated employment in professional and technical positions (35.17 percent of female full-time employment as opposed to 14.86 percent of male full-time employment). Not surprisingly, managerial positions paid quite well in 1989 while professional and technical positions generally paid men (as well as women) more poorly than skilled labor. Women were hurt by the fact that they also dominated the one occupational category that had a large and significant negative impact on wages for both men and women - unskilled labor. Just over eight percent of women worked as unskilled laborers in 1994 while less than four percent of men were in unskilled labor positions.
V. Conclusions Despite official Soviet proclamation, our analysis indicates a large gender wage differential in Estonia at the end of the Soviet period - in fact larger in relative terms than the 1994 wage gap. Little of this differential can be explained by differences in personal characteristics and other job related factors present in the ELFS. Even including occupational dummies leaves the majority of the differential unexplained. This large unexplained differential remained throughout the early transitional period. Though, in fact, in relative terms, the unexplained male-female wage differential grew. However, the relative growth in the unexplained portion of the differential might be less than many would expect.
Potential gender wage discrimination (as well as other types of potential discrimination) provides a particularly acute policy problem for transitional economies in general and for Estonia specifically. While established market economies have wrestled with the problem of wage discrimination for years and
42 Gender Wage Differences in Soviet and Transitional Estonia
erected elaborate (if not always effective) legal means of dealing with discrimination problems, discrimination was ignored (as it officially did not exist) in socialist economies. As Estonia gets deeper into negotiations for EU membership, it will be forced to deal with the question of discrimination. As of yet, Estonian policy makers and legislators have not addressed the issue in any substantial way. As mentioned above, the job segregation that occurred in Soviet Estonia might actually be aiding women during the transition. Women in Soviet Estonia were relatively concentrated in jobs requiring specialized training despite the fact they did not pay particularly well. While the market has pushed up the wages for these occupations (particularly professional and technical occupations) individuals had not, as of 1994, had time to adjust to these market forces. It is plausible that more men will be attracted to now relatively high-paying professional and technical positions. It is also plausible that women may be crowded out of these occupations. It is quite possible the relative position of women in Estonia (and perhaps transitional economies in general) will get worse before it improves. Examination of the 1997 ELFS and subsequent labor force surveys may indicate how the problem of gender wage discrimination is addressing itself.
Table 1: Variable Definitions
43 Baltic Journal of Economics Autumn/Winter 2002
Table 2: Mean Characteristics
Table 3: Employment by occupation (percentages)
44 Gender Wage Differences in Soviet and Transitional Estonia
Table 4: Regression Results, 1989 - Occupation not included dependent variable: ln(wage)
Table 5: Regression Results, 1989 - occupation included dependent variable: ln(wage)
45 Baltic Journal of Economics Autumn/Winter 2002
Table 6: Regression Results, 1994 - occupation not included dependent variable: ln(wage)
Table 7: Regression Results, 1994 - occupation included dependent variable: ln(wage)
46 Gender Wage Differences in Soviet and Transitional Estonia
Table 8: The wage effects of variables (Occupation not included)
Table 9: The wage effects of variables Occupation included
47 Baltic Journal of Economics Autumn/Winter 2002
References
Becker, Gary S. (1971), The Economics of Discrimination, 2nd edition. Chicago: University of Chicago Press.
Buckley, Peter J. and Ghauri, Pervez N. (1994), "Statement of the Issues." The Economics of Change in East and Central Europe: Its Impact on International Business. Academic Press,: 1-32.
Commission for Population and Social Statistics Working Group for Labor Force Surveys, (1995) Estonian Labor Force Survey 1989-1995.
Cotton, Jeremiah (1988), "On the Decomposition of Wage Differentials." The Review of Economics and Statistics 70: 236-43.
Dolton, Peter J. and Kidd, Michael P. (1994), "Occupational Access and Wage Discrimination." Oxford Bulletin of Economics and Statistics. 56, 4: 457-74.
Estonian Institute of Economic Research (1997), Economic Indicators of Estonia, August 1997.
Gill, Andrew M. (1994), "Incorporating the Causes of Occupational Differences in Studies of Racial Wage Differentials." The Journal of Human Resources 29, 1:20- 41.
Kandolin, Irja (1996), "Pay Differentials Between Men and Women in Estonia and Finland." The Finnish Review of East European Studies 3, 4: 56-78.
Kee, Peter (1995), "Native-Immigrant Wage Differentials in the Netherlands: Discrimination?" Oxford Economic Papers 47: 302-17.
Oaxaca, Ronald (1973), "Male-Female Wage Differentials in Urban Labor Markets." International Economic Review 14, 3: 693-709.
Rutkowski, Jan (1996), "High Skills Pay Off: The Changing Wage Structure During Economic Transition in Poland." Economics of Transition 4, 1: 89-112.
48 Gender Wage Differences in Soviet and Transitional Estonia
United Nations Development Programme (1996), Estonian Human Development Report 1996.
Vinokur, Aaron and Ofer, Gur (1987), "Inequality of Earnings, Household Income, and Wealth in the Soviet Union in the 1970s, in Politics, Work, and Daily Life in the USSR: A Survey of Former Soviet Citizens, ed. James R. Millar, Cambridge University Press, Cambridge, UK.
49 Baltic Journal of Economics Autumn/Winter 2002
The Outward Expansion of the Largest Baltic Corporations - Survey Results Kari Liuhto1 & Jari Jumpponen2
Abstract A relatively high percentage of Baltic corporations have already started their operations abroad, over 40% of the companies studied. It is surprising that the approaching EU membership does not seem to be the driving force of the Baltic corporations’ internationalization, though the EU is clearly the major export destination. The empirical evidence shows that the operations of the Baltic companies in foreign markets, have concentrated on the ex-CMEA countries, especially on the former USSR. The empirical data indicates that most of the operations abroad are related to marketing, such as the foundation of their own representative office or their own sales unit in a foreign market.
Key Words: Baltic States, international business, internationalization, Estonia, Latvia, Lithuania
1. INTRODUCTION The Baltic States are very small. Their population, even combined, is only 7.5 million, which is less than the population of Austria. The small size of the Baltic economies is emphasized, when their GDP is analyzed. In 2001, the GDP of all three Baltic States, measured at purchasing power parity (PPP) was some USD 60 billion. Even the GDP of Ireland (USD 105 billion), which is among the least wealthy of EU members, is higher than total Baltic GDP. Finland's GDP was double that of the Baltics (CIA, 2002).
The small size of their economy obviously pushes Baltic companies abroad. Clear evidence of Baltic firms' internationalization at the macroeconomic level is the high exports-GDP ratio. In 2000, the exports of goods and services were some 45- 91% compared to GDP, depending on the Baltic State in question (EU, 2002)3.
1Kari Liuhto, The Research Group for Russian and East European Business, Lappeenranta University of Technology, P.O. Box 20, FIN-53851 Lappeenranta, Finland. 2Jari Jumpponen, The Research Group for Russian and East European Business, Lappeenranta University of Technology, P.O. Box 20, FIN-53851 Lappeenranta, Finland.
50 The Outward Expansion of the Largest Baltic Corporations - Survey Results
In 1990, the overwhelming majority of Baltic States' foreign trade was directed to other socialist countries. Then, the CMEA covered over 90% of the Baltic States' exports (EBRD, 2000). Ten years later, the direction of the foreign trade has almost completely reversed. In 2001, the EU was the main trading partner of the Baltic States. Exports to the EU covered some 69% of the Estonian exports. The respective share in Latvia was close to that of Estonia, but in Lithuania the EU share was remarkably lower, less than 50%. Also the imports from the EU are significant. The EU represented roughly 50% of the Baltic countries' imports, Estonia being the most dependent and Lithuania the least dependent on imports from the EU. (Foreign Trade, 2002)
Whilst the EU's importance in Baltic foreign trade has grown rapidly, the dependence on Russian trade has declined. In 2001, Russia covered only 3-10% of Baltic exports, Estonia being the least Russia-oriented and Lithuania the most Russia-oriented. Russia's proportion of Baltic imports is considerably higher than that of their exports. In Lithuania, the dependence on imports from Russia is by far the highest, over 25%. In Estonia and Latvia, Russia formed just some 10% of the total imports (Foreign Trade, 2002).
Besides foreign trade flows, foreign direct investment (FDI) inflows verify that the Baltic countries are open economies. In 2000, the FDI-inflow represented 3-6% of Baltic GDP. The Baltic States have attracted much more FDI per capita than other ex-Soviet republics. The cumulative FDI inflow per capita during 1989-2000 in the Baltic States was on average over USD 1000, while in other former Soviet republics it was less than USD 170 (EBRD, 2001).
Finland and Sweden are the most important investor countries in Estonia, where they together form 65% of Estonian FDI stock in 2001. Denmark, in turn, is the biggest investor in Lithuania. The Baltic States covered only a modest part of the FDI in another Baltic State. Only Estonia managed to climb among the top 10 investor countries with a 6.5% FDI stake in both Latvia and Lithuania (see Table 12).
FDI has supported the recovery of the Baltic States from the transition slump and has enhanced the improvement of enterprise competitiveness both directly (foreign owner impact) and indirectly (via competition or copying competitiveness). Along with the development of their competitiveness, the Baltic companies have not only intensified their export activities, but they have also begun to invest outside their home market. In fact, a Latvian company was ranked the third most international company among the Central and East European firms in 2000 (see Table 1).
3According to the Economist Intelligence Unit, the exports/GDP ratio was in Estonia 90.6%, while in Latvia it was 44.9%, and in Lithuania 50.4%, respectively (EIU, 2002).
51 Baltic Journal of Economics Autumn/Winter 2002
Table 1. Top 25 Non-Financial Transnational Corporation based in Central Eastern Europe (ranked by foreign assets, USD mn, 2000)
Though only one Baltic company reached the top 25 list, the importance of monitoring the Baltic companies' expansion abroad is emphasized, due to the accelerating globalization of business. Baltic corporations cannot simply afford to underestimate the pressures created by globalization.
2.METHODOLOGY The authors searched for answers to the following research questions: * To what extent have the largest Baltic companies already moved their operations abroad? * What are the main driving forces behind internationalization? * What are the main target environments of internationalization? * What are the main operation modes used?
Due to limited research funds, the researchers were forced to limit the sample size, and thus, they focused the study on the 100 largest companies in each Baltic State. These 300 companies were selected on the basis of their net turnover/sales.
The authors deliberately decided to focus the study on the largest corporations for three main reasons. First, should the researchers have aimed at
52 The Outward Expansion of the Largest Baltic Corporations - Survey Results
random sampling, the outcome of the study would most probably have been less successful, because a large proportion of the registered enterprises do not operate. This would inevitably have caused an enormous non-response. Second, smaller companies have usually less need, resources or skills for their internationalization. This would most probably have resulted in a great percentage of those answers indicating that the firm has not yet started its internationalization. Third, the investigation was focused on the largest companies, due to their economic importance. The success of these companies' internationalization is crucial for the economic development of the Baltic States, since they form a significant part of Baltic GDP and industrial production.
The questionnaire designed for the research deals with the reason, environment, and mode of the internationalization. The authors considered that the questionnaire should be linguistically as clear as possible, to avoid the possibility of misunderstanding. It was also decided that the questionnaire should not exceed two pages and should not include overly sensitive issues, such as exact performance indicators or ownership arrangements, since both a lengthy questionnaire and overly sensitive questions would have reduced the Baltic managers' willingness to respond to the questionnaire (see Appendix 1).
The above methodological decisions proved to be correct, since the response rate was rather satisfactory, over one-third, especially taking into consideration that the mail survey was conducted among post-socialist companies, which are usually reluctant to reveal any information to researchers (see Michailova & Liuhto, 2000). In this context, it should be mentioned that due to the researchers' persistent efforts, two reminders, the response rate increased from 20% to 38% (see Table 2).
Estonian companies were more active in participating in the survey than Latvian and Lithuanian firms. Even if Latvian and Lithuanian corporations were less enthusiastic about taking part in the research, the response is not so much unbalanced by their lesser enthusiasm that the over- or under-representation of any country would distort the analysis on the internationalization of the largest Baltic corporations. The participating companies also represent various business fields in each of the countries in question, so there is also no distortion in this issue (see Appendix 2).
Analysis of returned questionnaires indicates that those questionnaires received by the researchers were accurately answered; though deficiencies could be detected from questions concerning the geographical division of the exports. On the basis of the response analysis, it can be assumed that using English in the questionnaire did not result in an incorrect interpretation of the questions, and
53 Baltic Journal of Economics Autumn/Winter 2002
thus, the received answers are believed to be valid and credible. Most probably, the research language did not cause the non-response as much as managers' hectic timetables or a fear of the data getting into the 'wrong' hands.
Table 2. The Response to the Survey
The questionnaires were first sent out on the 12th of January and the last questionnaire to be included in the analysis was received two months later, on the 11th of March, 2001. Because firms from transition economies expand their activities abroad at an ever-increasing speed, the empirical data will become outdated relatively fast, and therefore, it is extremely important to conduct follow- up studies frequently.
3. EMPIRICAL RESULTS
3.1. Exports of the Baltic Companies Almost two thirds of the respondents (64%) indicated that their company has exports. The export frequency among the Latvian corporations was considerably lower. Only one half of the studied Latvian companies have exports. When one remembers that Latvian firms were more active in their activities abroad than the Estonian and Lithuanian ones, their lower export activity is rather puzzling.
The companies that have exports were asked to indicate the share of exports out of their total sales. The data reveals that over one-third of the companies have no exports, a fifth of the companies export less than one-fifth, and for the rest, exports compose 50% or more of the total sales (see Figure 1).
54 The Outward Expansion of the Largest Baltic Corporations - Survey Results
Figure 1. The Share of Exports of Total Sales in the Baltic Companies (N=113)
The EU and other Baltic State(s) are the most common destinations for the exports. Of those companies that have exports, more than two-thirds export to the EU and/or to another Baltic State(s). The EU is especially favored among export- oriented companies i.e. if the proportion of the exports from the total sales is high, the company is likely to export to the EU. To put it differently, if a Baltic company exports to the EU, it seldom has any other significant destinations for exports. Respectively, if a company exports elsewhere, the exports are divided between many countries (see Figure 2).
Figure 2. The Division of Baltic Companies’ Exports to Other Baltic State(s) and to the EU
55 Baltic Journal of Economics Autumn/Winter 2002
Russia is the third most favored export destination, after the EU and another Baltic State(s). The share of Eastern Europe was not high. In fact, the Baltic companies export more to the USA than to Eastern Europe.
3.2. The Baltic Corporations' Operations Abroad It is not exceptional to find a Baltic enterprise, which has already started its operations abroad. Some 42% of the studied companies have begun their operations in a foreign market (see Table 3).
Table 3. Studied Baltic Companies Abroad
The table above shows that operations abroad are more common among Latvian corporations than Lithuanian and Estonian ones. The empirical data cannot reveal any apparent explanation, as to why Latvian companies are more active in starting operations abroad than their Estonian and Lithuanian counterparts.
The majority of Baltic companies stated that the driving force for their internationalization was getting a foothold in a larger economy. The option "internationalization is a necessity" was in second position. The aim of getting a better price was the third most frequently selected alternative. Surprisingly, "preparation for EU accession" was selected by only 13% of those companies that have operations abroad. All in all, it can be concluded that the domestic factors pushing Baltic companies abroad seem to be behind their internationalization rather than the attractions of foreign markets per se (see Table 4).
56 The Outward Expansion of the Largest Baltic Corporations - Survey Results
Table 4. Reasons for Baltic Companies to Operate Abroad (N=48)4
Though the EU has an important role as a destination for Baltic companies’ exports, companies do not self-evidently seem to turn to the West in their operations. In fact, operations in other Baltic State(s) and in Russia are more common than operations in the EU (see Table 5).
Table 5. The Operations of the Baltic Companies Abroad (N=48)5
Starting operations in other Baltic State(s) is natural, as the Baltic States form a relatively familiar market place. Their geographical proximity can be another explanatory factor. Estonian and Latvian companies, in particular, seem to have chosen to expand their operations in another Baltic State, while Lithuanian firms have penetrated into other regions.
It is noteworthy to mention that also distant regions, like the United States and Asia, are represented among the environments where operations have been started. Latvian companies, in particular, have discovered these 'remote' environments.
If one analyzes the reasons for internationalization and environment selection together, an extremely interesting finding can be discovered. The EU is not
4 As a company may have several reasons for operating abroad, the sum of percentages exceeds 100%. 5 As a company may have operations in many regions, the sum exceeds 100%.
57 Baltic Journal of Economics Autumn/Winter 2002
regarded as "a larger economy", but Russia and the CIS are. In other words, Russia and the CIS are selected if the Baltic corporation's goal is to search for a larger market. The EU is chosen on a different basis.
The data clearly indicates that the largest Baltic companies do not prefer to start production abroad. Similarly, joint ventures are not a very widely used operation mode. Instead, almost half of the companies with operations abroad indicated that they have their own representative offices (see Table 6).
Table 6. Baltic Companies’ Operating Modes Abroad (N=48)6
All in all, 32 of the studied companies indicated that they had activities abroad. 27 out of 32 corporations indicated that they have employees abroad. However, not more than two firms stated that they have the majority of their staff abroad. 28 companies announced they have assets abroad, but not more than six companies have moved over 50% of their assets outside their country. Estonian companies have been more active than their counterparts in Latvia and Lithuania in shifting their assets abroad (see Table 7).
As indicated in Table 1, only one Baltic company made the list of the 25 most international companies in Central and Eastern Europe. The empirical evidence of this study indicates that several other transnational Baltic companies exist. The data also makes reference to the fact that the field of operation is not the main explanatory factor for moving assets and employees abroad. Rather, several different fields of operations can be detected behind these Baltic companies.
6 As a company may use many operation modes simultaneously, the sum exceeds 100%.
58 The Outward Expansion of the Largest Baltic Corporations - Survey Results
Table 7.The Transnationality Analysis of the Studied Baltic Corporations
3.3. Future Operations Abroad Table 10 shows that only 28% of the companies studied planned to start operations abroad. The data does not reveal a significant difference between Baltic companies' interest in beginning operations abroad in the future. Moreover, the answers indicate that the company's existing operations abroad do not seem to reflect whether a company plans to start further operations abroad i.e. firms with no experience in foreign operations are planning to start operations abroad (15%) as frequently as those enterprises with experience (13%).
7 The transnationality index is calculated as the average of three ratios: foreign assets to total assets, foreign sales to total sales and foreign employment to total employment.
59 Baltic Journal of Economics Autumn/Winter 2002
Table 8. The Companies’ Intentions to Operate Abroad
While eleven companies indicated that they have plans to expand their operations in another Baltic State, only seven companies mentioned the EU as the target environment. In fact, Russia was more popular than the EU. Nine companies planned to start their operations in the EU. Keeping in mind the small number of the response, the empirical evidence tentatively indicates that the largest companies in the Baltic States perceive the EU as a trading partner rather than a destination for their expansion.
4. CONCLUSION Over 60% of the studied enterprises claimed to have exports. A relatively high percentage of firms (40%) indicated that they have started operations abroad. These high percentages do not come as surprise, since the Baltic States are small markets, which automatically push most of the largest Baltic corporations abroad.
Some 60% of the companies indicated that a foothold in a larger economy was one reason for starting operations abroad. The second most frequently given answer (over 50%) was "internationalization is a necessity to survive in future business". Third, Baltic corporations expand their activities in foreign markets to receive a better price for their commodity.
These responses could be easily anticipated, but it is very surprising that preparation for EU accession did not rank higher among the reasons for starting internationalization. The response of the Baltic managers indicate that approaching EU membership is not the driving force for Baltic corporations' internationalization, even though the EU is clearly the major export destination.
Baltic corporations' management may think that they are able to maintain sales to the EU even without starting-up their own operations inside the EU. In a way, maintaining production inside the Baltic States can be a rational decision since it allows Baltic corporations to take advantage of lower production costs while enjoying the benefits of the European Single market. On the other hand, EU
60 The Outward Expansion of the Largest Baltic Corporations - Survey Results
membership may attract more EU and even non-EU companies to the Baltic States, and thus increase competition inside the Baltics.
Consequently, increasing competition will force Baltic companies to improve their effectiveness, either via increasing their size or by sharpening their focus. If the Baltic corporations do not manage to improve their competitiveness, we could witness an increase in bankruptcies, mergers and takeovers in the Baltic States over the next decade (see Table 9).
Table 9. Summary of the Empirical Findings REASON FOR INTERNATIONALIZATION: WHY INTERNATIONALIZE? 10 main reasons behind the Baltic corporations’ internationalization:
1) Small domestic market forces the Baltic companies abroad (small economy-related driving force). 2) Survival in future business requires internationalization (a global trend in business). 3) Baltic firms expect to receive a better price for their commodity abroad (relatively low buying power in the post-socialist countries). 4) Securing a resource supply (the Baltic States are relatively poor in natural resources). 5) Baltic companies are searching for less competitive markets, especially in other former Soviet republics (inter-enterprise competition seems to be fiercer in the Baltic States than in other ex-Soviet republics). 6) Baltic firms are searching for more stable markets in the West (to increase predictability in their enterprise development). 7) Foreign ownership in the company influences their internationalization decision (internal driving force) 8) Domestic clients have expanded their operations abroad (following the own client principle). 9) Preparation for EU accession (a need for Pan-European "internationalization"). 10) Logistical reasons have attracted Baltic companies abroad (a goal to improve efficiency).
ENVIRONMENT SELECTION: WHERE TO INTERNATIONALIZE? 5 main environments, where Baltic firms have started their operations: 1) The Baltic market is the key foreign environment (a familiar and close foreign market). 2) Russia’s potential is attractive (earlier business relationships and experience). 3) The EU has attracted surprisingly few Baltic companies to start their operations there, though the EU is the main export direction (a fear of competition or EU regulations?). 4) Other ex-Soviet republics (earlier experience and less-fierce competition). 5) Eastern Europe (Baltic products’ price-quality ratio suit both East European demand and buying power).
MODAL CHOICE: HOW TO INTERNATIONALIZE? * Various marketing operations dominate (to increase sales, while keeping financial investment low). * Subcontracting, licensing, franchising (minimizing risks, while penetrating a foreign market). * Joint venturing is a mode, allowing partners to join their resources and knowledge. * Their own production unit abroad is still a relatively rarely used operation mode. * Acquisition of a foreign company is still a rare option, mainly due to financial constraints.
61 Baltic Journal of Economics Autumn/Winter 2002
PECULARITIES CONCERNING THE BALTIC FIRMS’ INTERNATIONALIZATION: * Despite the EU dominance in exports and the approaching EU membership of the Baltic States, surprisingly few Baltic firms have started their operations within the current EU. * The ex-socialist bloc clearly dominates as an environment, where foreign operations have been started. * The majority of Baltic firms are not planning to start operations abroad in the near future.
The empirical evidence shows that the operations of the Baltic companies in foreign markets have concentrated on the ex-CMEA countries, especially in the former USSR. The The empirical evidence shows that the operations of Baltic companies in foreign markets have concentrated on the ex-CMEA countries, especially in the former USSR. The explanation for focusing on the ex-CMEA market may stem from the fact that the Baltic commodities' price-quality ratio better fits these markets than those of the developed West. Also, their earlier business relations and experience in these markets may have offered a competitive advantage to the Baltic corporations, compared to their Western rivals.
The empirical evidence supports the presumption that most of the operations abroad are related to marketing, such as establishing their own representative office or their own sales unit in a foreign market. These sales- increasing activities are a logical modal choice since they do not require heavy financial investment. It can be assumed that operational modes, which require more investment and risk taking, will increase along with the improvement of the Baltic firms' financial position.
In closing, it can be argued that internationalization is a necessary condition, though not a sufficient condition by itself, for securing the Baltic corporations' survival in future business. Therefore, Baltic corporations must build strategic alliances with each other or foreign companies in order to be able to cope with the competitive pressures arriving both from the EU and from the East, as it can be predicted that Russian companies will intensify their investment activities in the Baltic States in years to come.
Until now, Russian investments in the Baltics have remained relatively modest (see Table 10). In Latvia, Russia formed some 5% of the FDI stock in 2001. Both in Estonia and Lithuania Russian investments represented only some 1-2% of the FDI stock. However, it would not be a surprise if Russian companies decided to use the Baltic States as a familiar foothold to the EU single market, and would thus decide to increase their investments in the Baltic States before the Baltic States receive their EU membership (Liuhto & Jumpponen, 2002).
62 The Outward Expansion of the Largest Baltic Corporations - Survey Results
Table 10. Foreign Direct Investment Stock in the Baltic States by Investing Countries
63 Baltic Journal of Economics Autumn/Winter 2002
REFERENCES
Bank of Estonia (2002) Direct investment stock by countries, available at http://grizli.ml.ee/itp1/itp_report.jsp [referred 6.11.2002]
CIA (2002) http://www.odci.gov/cia/publications/factbook/fields/2001.html [referred 6.11.2002]
EBRD (2000) Transition Report 2000, European Bank for Reconstruction and Development, London.
EBRD (2001) Transition Report 2001, European Bank for Reconstruction and Development, London.
EU (2002) Regular Report, The European Union, Brussels.
Foreign Trade (2002), Estonia, Latvia, Lithuania Foreign Trade 2001, Statistical office of Estonia, Tallinn.
LCB (2002) Latvian Central Bank, Balance of Payments database, available at: http://www.bank.lv/izdevumi/Latvian/maksbil/2002-02/LMB5.xls [referred 6.11.2002]
Liuhto Kari & Jumpponen Jari (2002) ‘International Activities of Russian Corporations - Where Does Russian Business Expansion Lead?’, Russian Economic Trends.
Michailova Snejina & Liuhto Kari (2000) ‘Organisation and Management Research in Transition Economies: Towards Improved Research Methodologies’, Journal of East-West Business 6/3.
Statistics Lithuania (2001) Statistical Yearbook of Lithuania 2001.
UNCTAD (2002) World Investment Report 2002, United Nations, New York.
64 The Outward Expansion of the Largest Baltic Corporations - Survey Results
APPENDIX 1. The Questionnaire
1. Does your company have exports (mark appropriate alternative with X)? ( )Yes ( ) No - if no, move to question 4.
2. The share of exports of total sales? ( ) 1-5% ( ) 6-10% ( ) 11-20% ( ) 21-30% ( ) 31-40% ( ) 41-49% ( ) 50-60% ( ) 61-70% ( ) 71-80% ( ) 81-90% ( ) 91-99% ( ) 100%
3. What is the share of the following markets of your company's exports? EU ____ % Another Baltic State ____ % Russia ____ % Other ex-Soviet republic/s ____ % Eastern Europe ____ % USA ____ % Asia ____ % Other, what ______%
4. Does your company operate abroad (not taking into account exports)? ( )Yes ( ) No - if no, move to question 7.
5. Which operation mode/s is your company using abroad (many answers possible)? ( ) Marketing co-operation with a foreign firm/s ( ) Own representative office/s ( ) Own sales unit/s ( ) Joint venture with another firm ( ) Completely owned production unit/s ( ) Equity ownership in a foreign company/ies ( ) Own investment / holding company abroad ( ) Subcontracting / licensing / franchising agreement with a foreign company ( )Other,what ______
65 Baltic Journal of Economics Autumn/Winter 2002
6. In which regions your company has started business operations (not exports)? EU ( )Yes ( ) No Another Baltic State ( )Yes ( ) No Russia ( )Yes ( ) No Other ex-Soviet republic/s ( )Yes ( ) No Eastern Europe ( )Yes ( ) No USA ( )Yes ( ) No Asia ( )Yes ( ) No Other, what -______( )Yes
7. What is the share of the following activities of your company's performance? Home market Abroad Assets ______% ______% Sales ______% ______% Employees ______% ______% Profits ______% ______%
8. What are the reasons why your company has started operations abroad? ( ) To get a foothold in a larger economy ( ) To get a better price ( ) Production costs are lower abroad ( ) To decrease transportation costs ( ) To secure availability of raw materials or skilful labor ( ) To avoid / to reduce custom duties or other tariffs ( ) To reduce tax burden ( ) Due to investment incentives offered by host or home government ( ) Due to more stable business environment ( ) Due to better business infrastructure ( ) Domestic clients have started their operations abroad ( ) Influence of foreign owner in your company's management ( ) Competition is not so hard abroad as in the home market ( ) Preparation for the accession of your country in the EU ( ) Internationalization is a necessity to survive in the future business ( ) Other, what______
9. Are you planning to start operations abroad (not exports)? ( )Yes , when______( ) No - if no, move to the end of the questionnaire.
66 The Outward Expansion of the Largest Baltic Corporations - Survey Results
10. In which regions you are planning to start operations (not exports)? EU ( )Yes ( ) No Another Baltic State ( )Yes ( ) No Russia ( )Yes ( ) No Other ex-Soviet republic/s ( )Yes ( ) No Eastern Europe ( )Yes ( ) No USA ( )Yes ( ) No Asia ( )Yes ( ) No Other, what ______( )Yes
Thank you for your valuable contribution! If you wish to receive the research report on the internationalization of the 300 largest Baltic companies, please write your company's address below or enclose your business card in the reply letter. ______
67 Baltic Journal of Economics Autumn/Winter 2002
APPENDIX 2. The Sample8 ESTONIA Company Field Net sales(1000 EEK)9
1. Eesti Energija AS Energy 3 923 722 2. Eesti Telefon AS Telecommunication 2 404 577 3. Hansatee Grupp AS Transport 1 825 772 4. Eesti Mobiiltelefon AS Telecommunication 1 469 597 5. Eesti Põlevkivi AS Mining 1 455 408 6. Eesti Raudtee AS Transport 1 401 398 7. Sylvester Grupp AS Wood processing 1 070 939 8. Tallinna Kaubamaja AS Trade 996 079 9. Kreenholmi Grupp Textiles 984 846 10. Pakterminal AS Transit 978 644 11. ETK Hulgi AS Trade 878 205 12. Neste Eesti AS Oil - petroleum trade 874 918 13. Tallinna Soojus AS Energy 822 504 14. Kaupmees & Ko AS Trade 812 140 15. Tallinna Sadam AS Port services 803 732 16. Stockmann AS Trade 793 736 17. Eesti Gaas AS Energy 762 646 18. Balti Laevaremonditehase AS Shipbuilding 761 264 19. Merko Ehitus AS Construction 733 657 20. Estonian Air AS Transportation 720 981 21. Saku Ílletehase AS Beverages 715 040 22. EMV AS Construction 686 735 23. Bankend Eesti AS Trade 652 237 24. Onako Eesti AS Oil - petroleum trade 646 990 25. Premium Oil AS Oil - petroleum trade 627 320 26. E.O.S. AS Transit 621 627 27. Eesti Statoil AS Oil - petroleum trade 592 466 28. Estline AS Transport 590 729 29. EE Grupp AS Construction 590 727 30. Stora Enso Mets AS Wood processing 586 865 31. Fanaal AS Building materials 582 393 32. NT Marine AS Services for ships 578 321 33. Eriõli Kaubanduse AS Oil - petroleum trade 553 534 34. Viru Keemia grupp AS Chemicals 552 538 35. Rakvere Lihakombinaat AS Foodstuffs 539 555 36. Eesti Metallieksport AS Metal trade 531 159
8 The companies marked with bold returned a usable reply and companies marked in the brackets were not included in the research either because the researchers were not able to find the company's mail address. Also corporations operating in banking or in insurance business were dropped out of the survey. Moreover, the Latvian Privatization Agency was no approached. 9 In February 2001, one US dollar equaled 16,9 Estonian kroons (EEK).
68 The Outward Expansion of the Largest Baltic Corporations - Survey Results
37. FKSM AS Port services 524 326 38. Norma AS Automobile seat belts 516 400 39. Eesti Post AS Post services 507 602 40. G.S.G. AS Oil - petroleum trade 507 316 41. ETK Maksimarketi AS Trade 499 084 42. Saurix Petroleum AS Oil - petroleum trade 496 409 43. Falck Baltics (ESS AS) Security services 480 547 44. Tallinna Külmhoone AS Foodstuffs 454 988 45. Eesti Mereagentuur AS Stevedoring 438 830 46. Jungent OÜ Trade 430 151 47. Tallinna Vesi AS Utilities 426 803 48. S-Marten AS Trade 408 116 49. Elcoteq Tallinn AS Electronics 403 663 50. Rannila Profiil AS Building materials 400 150 51.(AVR Trans AS Transit 390 718)10 52. Tamro Eesti Pharmaceuticals 388 930 53. Estravel AS Travel services 388 500 54. Eesti Coca-Cola Joogid AS Foodstuffs 382 195 55. JOT Eesti OÜ Electronics 380 144 56. Horizon Tselluloosi ja Paberi AS Paper production 379 042 57. Kunda Nordic Tsement AS Building materials 373 967 58. Tallinna Piimatööstuse AS Dairy 366 748 59. Maseko AS Foodstuffs 366 923 60. SI-Kaubabaasi Trade 350 959 61. Kalev AS Confectionery production 338 511 62. ES Sadolin AS Building materials 332 417 63. Famar-Desl AS Trade 329 876 64. Toyota Baltic AS Trade 326 277 65. Repo Vabrikud AS Wood processing 325 793 66. Kesko Eesti AS Trade 325 460 67. Veho Eesti AS Trade 323 167 68. Hiiu Kalur AS Foodstuffs 314 721 69. HTM Sport Eesti OÜ Sport equipment 309 213 70. Siemens AS Electronics 308 168 71. Silmet Grupp AS Chemicals 305 213 72. Tallegg AS Foodstuffs 303 772 73. Metsind AS Timber products 292 247 74. Silberauto AS Trade 291 018 75. Baltika AS Beverages 288 926 76. Ericsson Eesti AS Trade 283 362 77. Tech Data Eesti AS Information technology 282 024 78. Mets & Puu AS Forestry 281 405 79. Marat AS Textiles 279 964
10 The companies marked in the brackets were not included in the research either because the researchers were not able to find the company's mail address. Also corporations operating in banking or in insurance business were dropped out of the survey. Moreover, the Latvian Privatization Agency was no approached.
69 Baltic Journal of Economics Autumn/Winter 2002
80. Rapla Dairy Dairy 279 157 81. ABB AS Energetics 274 881 82. Liviko AS Alcohol products 258 813 83. Radiolinja Eesti AS Telecommunication 253 419 84. Nitrofert AS Chemicals 253 151 85. (NB Oil Group OÜ 248 427) 86. Nordic Jetline AS Travel services 248 406 87. AbeStock AS Wholesale trade 243 285 88. TVMK AS Wood processing 242 551 89. Forestex AS Wood trade 239 161 90. Valga Liha- ja Konservitööstus AS Foodstuffs 236 662 91. Saksa Auto AS Vehicle trade 235 706 92. Microlink Arvutite AS Information technology 235 370 93. Tartu Ílletehas AS Beverages 235 132 94. Skanska Ehituse AS Construction 232 651 95. Holmen Mets AS Trade 227 005 96. Tarmeko AS Furniture manufacturing 226 283 97. Baltex 2000 AS Textiles 224 963 98. EVR Koehne AS Construction 224 843 99. Teede REV-2 AS Construction 224 061 100. Amisco AS Shipping services 220 049 101. Kommest Auto AS Trade 219 648 102. Södra Eesti AS Paper products trade 216 447
APPENDIX 2. Continued LATVIA Company Field Net Turnover(LVL million)11
1.Latvenergo PVAS Energy 167,56 2.Lattelkom SIA Telecommunication 129,30 3.Latvijas kugnieciba PVAS Shipping 111,79 4.Latvijas dzelzcels VAS Transport 110,72 5.Turiba CS Trade, catering 84,00 6.Latvijas Gaze AS Energy 83,08 7.(Latvijas Privatizacijas agentura VAS Privatization 70,04) 8.Kurzemes degviela AS Oil products 67,77 9.Latvijas Mobilais telefons SIA Telecommunication 64,60 10.Rigas siltums AS Heating 63,95 11.Liepajas metalurgs AS Metal industry 56,64 12. Latvija Statoil SIA Oil products 53,00 13.Latvijas finieris AS Woodworking 50,10 14.Ventpils nafta AS Transit services 45,86 15.Ventpils tranzita serviss SIA Oil transit 44,92 16.Procter & Gamble Marketing Latvia SIA Trade 43,47 17.Alianse-2 SIA Trade, foodstuffs 38,73
11 In February 2001, one US dollar equaled 0,62 Latvian lats (LVL).
70 The Outward Expansion of the Largest Baltic Corporations - Survey Results
18.Dinaz SIA KU Oil products 31,14 19.(Latvijas unibanka AS Finance 30,98) 20.Nelss SIA Woodworking, trade 29,37 21.(Parekss banka AS Finance 28,16) 22.Bravo SIA Trade, beverages 27,91 23.Severstallat AS Steel trade 27,57 24.Neste Latvija SIA Oil products 23,36 25.Interpegro Latvija SIA KU Trade 23,30 26.(Latvijas Banka Finance 22,96) 27.Air Baltic Corporation AS Transport 22,74 28.Rimi - Baltija SIA Retail trade 22,20 29.LUKoil Baltija R. SIA Oil products 21,92 30.LatRos Trans SIA KU Oil transit 21,06 31.Lex-U SIA Trade, foodstuffs 20,86 32.Lindeks AS Wood trade 20,01 33.Greis SIA Trade 19,99 34.Skonto buve SIA Construction 19,78 35.Latvijas pasts VAS Post service 19,49 36.Aldaris AS Beverages 19,35 37.BMGS AS Construction 19,10 38.Tamro SIA Trade, medicines 18,01 39.Rigas udens PU Municipal services 17,70 40.Linda SIA Wood trade 17,65 41.(Hansabanka AS Finance 17,45) 42.Ogre AS Textiles 17,25 43.Venceb AS Construction 17,00 44.Krasainie lejumi AS Metal working 16,86 45.Skonto metals SIA Metals 16,49 46.Tramvaju un trolejbusuparvalde PU Transport 16,38 47.Elko Riga SIA KU Computers, trade 16,03 48.Mono SIA Trade, insurance 15,84 49.(Balta AAS Insurance 15,77) 50.Ventbunkers AS Transit services 15,77 51.Rigas piena kombinats AS Dairy 15,33 52.Riga kugu buvetava AS Mechanical engineering 15,19 53.Viada SIA Oil products 15,10 54.Baltkom GSM SIA KU Telecommunications 15,08 55.Laima AS Foodstuffs 15,00 56.Shell Latvia SIA Oil products 14,80 57.Unilever Baltic LLC SIA Trade 14,62 58.Latvijas Balzams AS Beverages 14,60 59.Augstceltne SIA Oil products 14,51 60.Preses apvieniba AS Trade 14,02 61.Lauma AS Textiles 13,65 62.Weeluk (Baltic) Ltd. SIA Wood export 13,64 63.Ventspils ekspedicija SIA Transit services 13,60 64.Cido partikas grupa SIA Foodstuffs 13,37
71 Baltic Journal of Economics Autumn/Winter 2002
65.Valmieras stikla skiedra AS Chemical industry 13,33 66.(Latvijas krajbanka Finance 13,11) 67.Alkolats SIA Beverages, trade 13,00 68.Karsten Latvian SIA Trade 12,82 69.Rigas vini AS Beverages 12,80 70.Lido nafta SIA Oil products 12,54 71.Nelss TT Trade 12,17 72.Silva SIA Forestry 12,06 73.Unifex SIA Trade 11,87 74.Skanska konstrukcija SIA Construction 11,80 75.Latvijas nafta PVAS Oil products 11,78 76.Dobeles dzirnavnieks AS Foodstuffs 11,70 77.Baltimar VT SIA Oil products 11,69 78.Fortech SIA Computers 11,59 79.SEL Ð II SIA Beverages, trade 11,59 80.Hanzas maiznicas AS Foodstuffs 11,16 81.Baltfor SIA Wood export 11,10 82.Klangu kals SIA Fuel trade 11,08 83.Rigas piensaimnieks AS Dairy 10,91 84.Motors Latvia SIA Cars 10,85 85.CHS Riga SIA Computers, trade 10,80 86.Kurekss SIA Wood export 10,73 87.Cido logistika SIA Foodstuffs, trade 10,60 88.Siemens SIA Electronic equipment 10,50 89.Ventamonjaks AS Transit services 10,28 90.(Austrumu alianse AAS Insurance 10,22) 91.Tolaram Fibers AS Chemical industry 9,88 92.Grindeks AS Pharmaceuticals 9,85 93.Diena AS Publishing 9,84 94.Lattransrail SIA Construction, transport 9,80 95.Oilands SIA Fuel trade 9,80 96.Nokia Latvija SIA Telecommunications 9,72 97.Ventspils tirdzniecibas osta AS Stevedores 9,70 98.Rigas transporta flote AS Shipping 9,54 99.Juraslicis AS Fish industry 9,50 100.Nelda SIA Trade 9,50 101.Rimako AS Textiles 9,50 102.Ziemelu nafta SIA Oil products 9,50 103.(Stalkers AS Trade 9,48) 104.Latvijas Gaisa satiksme VAS Air navigation 9,43 105.Philips Latvija SIA Trade 9,40 106.Bolderaja Woodworking 9,32 107.Kalija parks AS Port services 9,32 108.Baltijas transporta apdrosinasana AASInsurance 9,31 109.Jelgavas cukurfabrika AS Sugar producer 9,20 110.(SBV SIA Construction 9,18) 111.Rezeknes piena konservu kombinats AS Dairy 9,03
72 The Outward Expansion of the Largest Baltic Corporations - Survey Results
APPENDIX 2. Continued LITHUANIA Company Field Sales(LTL)12
1. Mazeikiu nafta Oil - petroleum products 2.283.923.797 2. Lietuvos Energija Electric utilities 1.468.362.109 3. Lietuvos Telekomas Telecom 969.493.511 4. Lietuvos Dujos Natural gas utilities 555.796.474 5. Lietuvos Gelezinkeliai Transport 555.036.257 6. Lifosa Chemicals 494.171.030 7. Achema Chemicals 339.954.125 8. Ekranas Electrical engineering 279.361.166 9. Lietuvos juru laivininkyste (Lisco) Shipping 227.876.164 10. Rokiskio Suris Dairy 224.573.000 11. Kraft Foods Lietuva Food 214.940.933 12. Lietuvos Avialinijos Transport 196.718.834 13. Kauno Energija Heating 190.423.314 14. Alytaus Tekstile Textile 170.637.187 15. Lietuvos Kuras Oil - petroleum products 164.145.681 16. Snaige Electrical engineering 149.903.872 17. Pieno Zvaigzdes Dairy 147.098.796 18. Birzu Akcine Pieno Bendrove Dairy 145.811.210 19. Zemaitijos Pienas Dairy 138.682.501 20.Dirbtinis Pluostas Chemicals 121.059.952 21. Akmenes Cementas Building materials 112.559.709 22. Utenos Trikotazas Clothing 109.000.973 23. Kauno Tiltai Construction 106.210.619 24. Klaipedos Nafta Shipbuilding 105.044.716 25. Klaipedos Maistas Oil - petroleum products 102.128.317 26. Klaipedos Juru Kroviniu Kompanija Stevedoring 98.318.852 27. Kalnapilis Brewery 96.789.184 28. Kausta Construction 96.191.725 29. Vilniaus Duona Confectionery & bread 95.779.952 30. Alkesta Construction 94.627.691 31.Klaipedos Energija Heating 92.372.773 32.Klaipedos Mediena Wood products 91.013.546 33.Panevezio Keliai Construction 87.793.625 34.Panevezio Silumos Tinklai Heating 85.781.631 35.Baltik Vairas Vehicles 85.543.158 36.Baltijos Laivu Statykla Shipbuilding 83.609.478 37.Marijampoles Pieno Konservai Dairy 81.336.825 38. Vilniaus Vingis Electrical engineering 81.225.034 39. Alita Drinks 79.095.636 40. Alytaus Silumos Tinklai Heating 78.794.470 41. Drobe Textiles 76.211.074
12 In January 2001, one US dollar equaled 4,00 Lithuanian litas (LTL).
73 Baltic Journal of Economics Autumn/Winter 2002
42. Linas Textiles 76.112.350 43. Anyksciu Vynas Drinks 75.550.468 44. Stumbras Drinks 74.315.800 45. Svyturys Brewery 73.714.249 46. Panevezio Statybos Trestas Construction 72.743.822 47. Panevezio Pienas Dairy 72.107.309 48. Kretingos Grudai Cereals 72.089.326 49.Mesa Meat products 70.956.604 50.Klaipedos Transporto Laivynas Shipping 67.924.971 51.Ventus-Nafta Oil – petroleum products 66.633.657 52.Vilniaus Paukstynas Meat products 65.722.547 53.Kauno Grudai Cereals 65.274.339 54.Lytagra Trade 62.286.712 55.Vievio Paukstynas Meat products 61.690.456 56.Viti Construction 59.391.886 57.Utenos Pienas Dairy 57.551.366 58.Klaipedos Pienas Dairy 57.056.903 59.Siauliu Energija Heating 55.715.994 60.Grigiskes Paper and printing 55.032.095 61.Montuotojas Construction 53.707.841 62.Hidrostatyba Construction 52.531.905 63.Siauliu Plentas Construction 52.208.124 64.Alna Computer technologies 50.439.330 65.Apranga Trade 50.140.484 66.Medienos Plausas Paper and printing 49.090.874 67.Plasta Plastics 46.334.624 68.Klaipedos Baldai Furniture 46.259.300 69.Vilniaus Pergale Confectionery and bread 43.654.170 70.Krekenavos Agrofirma Meat products 43.438.295 71.Malsena Cereals 43.212.918 72.Siauliu Stumbras Leather, leather products 42.436.382 73.Kaisiadoriu Paukstynas Meat products 40.888.938 74.Vilniaus Mesos Kombinatas Meat products 40.825.482 75.Dvarcioniu Keramika Building materials 40.609.237 76.Klaipedos Duona Confectionery and food 38.554.060 77.Levuo Trade 37.451.682 78.Siauliu Pienas Dairy 36.314.344 79.Lietuvos Tara Packaging 35.726.233 80.Vernitas Chemicals 34.722.023 81.Kauno Pienas Dairy 34.168.133 82.Nemunas Building materials 33.952.173 83.Ragutis Brewery 32.841.357 84.Vilniaus Tauras Brewery 32.667.879 85.Metalu Komercija Trade 32.320.756 86.Vilniaus Baldu Kombinatas Furniture 31.846.515 87.Audejas Textiles 31.759.833 88.Liteksas Textiles 31.551.686
74 The Outward Expansion of the Largest Baltic Corporations - Survey Results
89.Skaites Electrical engineering 30.758.276 90.Kauno Ketaus Liejykla Building materials 30.080.704 91.Aliejus Oil production 29.722.948 92.Silutes baldai Furniture 29.432.643 93.Vilniaus Degtine Drinks 29.394.465 94.Ekinsta Construction 29.054.622 95.Kuro Aparatura Electrical engineering 28.356.018 96.Kelmes Pienine Dairy 28.129.653 97.Satrija Clothing 28.124.632 98.Siulas Textiles 27.910.378 99.Kedainiu grudai Cereals 26.362.957 100.Naujoji ruta Confectionery and bread 25.971.625
75 Baltic Journal of Economics Autumn/Winter 2002
BERTOLA G, BOERI T and NICOLETTI G (eds) Welfare and Unemployment in a United Europe: A Study for the Fondazione Rodolfo Debenedetti MIT 2001
Tito Boeri (general editor of the volume) writes in his introduction ‘This book aims at putting the current debate on the future of (social) welfare and employment in a united Europe on a new footing. We need fewer apocalyptic statements, fewer Cassandras, more facts, and deeper theoretical perspectives’ (p 1) It has to be said that the book does indeed live up to such a billing. Largely this is because, in contrast to many edited volumes that feature numerous disparate contributions, often of variable quality, here we are offered two extensive and solid pieces of work by top-flight researchers.
The overall theme of the book is the impact on social welfare and unemployment of more than twenty years of unprecedented integration in Europe. The first study, by Guiseppe Bertola, Juan Francisco Jimeno, Ramon Marimon and Christopher Pissarides (Bertola et al), deals with the past, present and future of EU welfare systems. The second is authored by Guiseppe Nicoletti, Robert Haffner, Steven Nickell, Stefano Scarpetta, and Gylfi Zoega (Nicoletti et al) and aims to analyse the interrelationships between product and labour market liberalization or deregulation. The quality extends to the commentary and discussion. Thus, in addition to his interesting introduction Tito Boeri gives a brief summing-up and there are comments on the first study by Charles Bean and Goesta Esping-Andersen and on the second by Olivier Blanchard and Andre Sapir.
Bertola et al show that the social welfare systems of the EU remain remarkably heterogeneous despite the overall progress of integration in EU goods and factor markets. They identify four broad welfare systems within the EU corresponding to distinct groups of countries. The Nordic group, which includes the Netherlands as well as Denmark, Finland and Sweden, has a tradition of universal welfare provision. At 30% or more of GDP these countries have the highest levels of social protection expenditures in the EU, they have rather generous unemployment benefits together with an important role for active labour market policies. The Nordic group also has a rather low degree of income inequality both in terms of earnings and when adjusted for taxation and transfers.
Another group, described as the Continental group, comprises Austria, Belgium France and Germany. At 27% to 30% of GDP total social expenditures are marginally lower than in the Nordic countries but income inequality is somewhat higher. These are countries regarded as being in the Bismarkian tradition where benefits are linked to employment, wage determination is centralised, and employment protection legislation is rather stringent.
76 The Outward Expansion of the Largest Baltic Corporations - Survey Results
Next come the Anglo-Saxon countries. Effectively, this means the UK and Ireland. These are regarded as being in the Beveridgian tradition with targeted benefits, a high degree of income inequality, low unemployment benefits, weaker employment protection legislation, decentralised wage-setting, and minimal active labour market policies.
Finally, the Southern European countries have more recently developed welfare states and, at 20% plus or minus a few points, have the lowest levels of social protection expenditures (arguably, Ireland with only 18.1% according to Table 1.1 should also be in this group), and high income inequality (not much influenced it appears by transfers. A notable feature of the Southern countries, that is mentioned but not really researched, is the role played by the family in their social protection arrangements. The family would also very likely be rather important in the context of, say, the Baltic states.
Having categorized the various national systems Bertola et al discuss developments and future options. They note that fears of a so-called ‘race to the bottom’ in terms of national social protection in the face of increased mobility of capital and labour have not so far materialized. They examine the relationship between the welfare systems and economic performance, concluding that ‘the welfare state appears to achieve its intended poverty-reduction purpose with no obvious adverse effects on income levels or growth rates’ (p 67). Charles Bean in his comment regards this view as ‘rather sanguine’ and points out that Bertola et al may have underestimated the growth effects of alternative systems of provision by focusing only on the labour market. He notes that ‘the most obvious omission here is the effect of unfunded pension schemes on national savings and investment rates’ (p 124).
Bertola et al conclude that while the diversity of European welfare states has hitherto proved rather durable, it may not be fully sustainable in the long run. Accordingly, they propose a basic safety net that is provided at the EU level, with a specific budget-line in the EU budget. They regard the associated problems of cost-of- living differentials as difficult but ‘technical’. As for contingent insurance provision, Bertola et al believe that this should address real market failures, participation should be mandatory and minimum contribution levels should be coordinated at the EU level. They also regard it as important that the link between contributions and benefits should be emphasised.
Finally they note a whole category of ‘local’ social provisions that should properly remain organised at the appropriate local level.
Nicoletti et al first explore the degree and dynamics of price convergence in the EU in the wake of deeper integration. For this they construct similarity indices, based
77 Baltic Journal of Economics Autumn/Winter 2002
on the Grubel-Lloyd index of intra-industry trade. It appears that price similarity in the EU for all products increased by 2.6 percentage points between 1985 (when the index was 81.9) and 1996 (when it was 84.6). Not all sectors experienced price convergence over the period. Most notably, in the energy sector prices diverged by 5.0 points over the period. However, it is interesting that the prices of EU non-tradables did converge and at a faster rate than the prices of tradables. Other interesting findings are that prices are more similar in the EU than they are in the other OECD countries and they have tended to converge faster. However, curiously, the rate of EU price convergence slowed down after 1992.
The remainder of Nicoletti et al is devoted to examining the impact of product market deregulation on labour markets. Partial theoretical analysis suggests that an increase in product market competition shifts the demand for labour outwards and thereby has a positive effect on employment, but at the same time if labour markets are unionized wages would tend to fall. On the other hand, general equilibrium effects are more complicated – some arguments suggest that an economy-wide increase in product market competition would increase both wages and employment while other arguments suggest that both wages and employment might fall as competition intensifies in product markets.
The empirical evidence is complex, not least because measurement of the degree of both product and labour market regulation is difficult. Nicoletti et al construct regulatory indicators that are then used to examine the interrelationship between regulation and labour market performance. Boeri sums up their results as showing ‘that restrictive product-and labour-market are closely interlinked. Environments more prone to competition in the product markets also tend to offer more protection to labor- market insiders and vice versa’ (p 252). Simultaneous regulatory reform in both sets of markets can deliver improved economic performance. Hence Nicoletti et al suggest that such reforms should be coordinated. Boeri points out that this implies a case for extending EU level policy making to labour markets ‘at least to match [that] prevailing in product markets’ (p 253).
This book will repay study by researchers in the Baltic states and other European transition economies. Transition together with the EU accession process can be represented in terms of very much the kind of developments analyzed here. Accordingly, although not dealing explicitly with Eastern Europe or enlargement the book is a rich source of ideas for labour market and social welfare research in the transition economies. Definitely something to have on the reference shelf.
ALF VANAGS BICEPS1
1 Director, Baltic International Centre for Economic Policy Studies. E-mail: [email protected]
78 The Outward Expansion of the Largest Baltic Corporations - Survey Results
KEUSCHNIGG, M Comparative Advantage in International Trade: Theory and Evidence, Physica-Verlag, 1999
Paul Samuelson once remarked that the principle of comparative advantage was one of the few propositions in economics that was both non-trivial and true. What Samuelson meant in this instance was that the theory of comparative advantage is logically correct. Demonstrating ‘empirical truth’ is another matter and pursuit of the empirical validity of the theory of comparative advantage has generated a whole industry of activity by economists and econometricians.
This is especially true of the Heckscher-Ohlin (HO) version of the theory in which it is asserted that comparative advantage is determined by relative factor endowments. Leontief started the ball rolling in 1953 when he unveiled his famous ‘paradox’ in which it was proposed that either the United States was abundant in labour relative to the rest of the world or the HO theory was empirically false.
Leontief’s challenge to received theory has subsequently been taken up by many economists and this volume, which is one of a series devoted to empirical economics, offers a useful summary of both empirical and theoretical responses to the Leontief paradox.
Although Leontief’s original study was conducted within the framework of the two-good, two-factor model familiar to students of international trade (the two- country aspect of the standard HO model was implicit in Leontief) it is rather obvious that serious empirical investigation of the determinants of international trade patterns needs to address the reality that the world is populated by many goods and many factors. In a many-factor, many-commodity world, the neatness of the 2x2x2 model evaporates and ‘when the number of goods exceeds that of factors, the precise commodity pattern of production and trade is indeterminate’ (p 13).
This is Mirela Keuschnigg’s starting point. She outlines a compact fashion the HO model as developed by Vanek (known as the HOV version of the model) in which the spirit of the original is retained in the form of a proposition that relates the factor content of a country’s net trade to its relative factor endowments. Thus the HOV model postulates that international trade in goods implicitly trades factors of production and that a country will ‘export’ factors in which its world endowment share exceeds it consumption share, and ‘import’ those in which it consumes more than its endowment.
The quantity version of this proposition requires the strong assumption that factor prices are internationally equalized. Accordingly, Keuschnigg proposes a value version combined with Cobb-Douglas technology. This approach has the
79 Baltic Journal of Economics Autumn/Winter 2002
property that factor cost shares can be treated as parameters. Further, Keuschnigg generalises the basic value version of the HOV model to incorporate non-neutral technological differences, increasing returns (external and internal), mobile capital and Armington demands. The relationship between the factor content of trade and factor endowments survives the generalisation of the HOV model in this manner and the remainder of the book is devoted to developing and implementing tests of the more general model.
The empirical work reported in the book reflects a data base of 46 countries and 116 (108) manufacturing industries in 1989 (1979). Three factors of production are postulated – capital, low-skilled labour and high skilled labour. The countries include both developed and developing countries. Keuschnigg conducts a variety of both direct and indirect tests and concludes that ‘the analysis provides a reasonably good explanation of trade data in terms of a very short list of production factors …. international trade, factor intensities, and factor endowments are related as predicted by the HOV theory.’ (p 139). Thus, developing countries tend to ‘export’ low-skilled labour and developed ones high-skilled labour. The results for capital are not as clear cut and Keuschnigg reports that generally the theory fares better for developing country trade than it does for developed countries.
Interesting subsidiary results include: i) when factor intensities are defined as relative cost shares no factor intensity reversals are observed; and ii) the perfect competition version of the HOV model cannot be rejected in favour of one which incorporates scale economies and imperfect competition.
Sometimes in the transition economies questions of the following form have been asked ‘in what does (say Latvia) have a comparative advantage?’ This volume comprehensively demonstrates at least two things. Firstly, even to begin to address such a question needs an approach that goes well beyond the usual textbook version of HO and Keuschnigg provides a solid account of the necessary theoretical framework. So the book would be rather useful for a graduate student working in this area. Secondly, the book suggests that the answer to such a question is likely to be rather unsurprising.
Finally, a personal comment on popular misconceptions of the meaning and usefulness of the principle of comparative advantage. From time to time it is suggested that comparative advantage has a normative content, that comparative advantage should be identified so as to identify the sectors a country should specialise in. This is quite the wrong way of interpreting the idea. Specialisation according to comparative advantage is simply the outcome of market forces when technology, markets etc satisfy certain restrictions. The merit of this book is to inform us a little more about the exact scope of those restrictions.
ALF VANAGS BICEPS
80
Call for Papers
TAX POLICY IN EU CANDIDATE COUNTRIES
ON THE EVE OF ENLARGEMENT
12-14 September 2003 - Riga, Latvia
Keynote Speakers: Gibert Metcalf, Tufts University – "Indirect Taxation" Helmuth Cremer, University of Toulouse – "Optimal Tax Policy" Peter Lambert, York University – "Taxation and Equity"
Topics EU tax harmonisation Tax competition Tax administration Local taxation Efficiency and equity of taxation Tax evasion EU Accession Tax abatement Tax policy in the era of globalisation Taxation in cyberspace
Schedule Submitted papers and abstracts (up to 200 words) must include title, keywords, JEL classification, full name, affiliation, address, email, fax and telephone of the author/s. Only email submissions will be accepted. Abstracts should be submitted by March 31st 2003. Acceptance or rejection will be notified by April 30th 2003. Accepted final papers should be received before June 30th 2003.
Participation fees (EUR 100) include conference papers, welcome reception and refreshments.
Papers and abstracts should be sent to: Dr. Mark Chandler EuroFaculty, University of Vilnius Sauletekio al. 9 Bldg. 2, Room 702 LT-2040 Vilnius, Lithuania [email protected] URL: www.eurofaculty.lv/taxconference