Call for applications: Klaus Liebscher ­Economic Research Scholarship

The scholarship program targets Austrian and international experts with a ­proven research record in economics and finance, and postdoctoral research experience. Applicants need to be in active employment and should be interested in broadening their research experience and expanding their personal research networks. Given the OeNB’s strategic research focus on Central, Eastern and South­eastern Europe, the analysis of economic developments in this region will be a key field of research in this context.

The OeNB offers a stimulating and professional research environment in close proximity to the policymaking process. The selected scholarship recipients will be expected to collaborate with the OeNB’s research staff on a prespecified topic and are invited to participate actively in the department’s internal seminars and other research activities. Their research output may be published in one of the department’s publication outlets or as an OeNB Working Paper. As a rule, the consultancy services under the scholarship will be provided over a period of two to three months. As far as possible, adequate accommodation for the stay in Vienna will be provided. ¹

Applicants must provide the following documents and information:

• a letter of motivation, including an indication of the time period envisaged for the consultancy
• a detailed consultancy proposal
• a description of current research topics and activities
• an academic curriculum vitae
• an up-to-date list of publications (or an extract therefrom)
• the names of two references that the OeNB may contact to obtain further information about the applicant
• evidence of basic income during the term of the scholarship (employment contract with the applicant’s home institution)
• written confirmation by the home institution that the provision of consultancy services by the applicant is not in violation of the applicant’s employment ­contract with the home institution

Studies

A sleeping beauty or a dead duck?The state of capital market development in CESEE EU ­Member States

Thomas Reininger, Zoltan Walko ²

For quite some while, there have been high expectations that stronger capital markets could generate a broader range of financing sources and reduce the buildup of vulnerabilities for the corporate sector in European Union Member States in Central, Eastern and Southeastern Europe (CESEE). These expectations prompted various supportive measures by international institutions, national authorities and market participants over the past 10 to 15 years. ­However, despite these efforts, capital market developments in the region have been far from dynamic over the past decade. Capital markets continue to be substantially less developed than in the euro area and the U.S.A., judging from the balances of debt securities, listed shares and investment fund shares outstanding both in relation to GDP and as a share of total financial liabilities. Even taken together, these three types of securities account for a smaller portion of total ­financial liabilities than loans, with loan penetration levels (loans as a percentage of GDP) ­already approaching euro area and U.S. levels. Data on financial flows show narrower gaps with the euro area for funding via debt securities, but issuance is dominated by government entities. Analyzing the financial liabilities of nonfinancial corporations, we find some relevance for listed shares only in Poland and Croatia (but less of a relevance than in the euro area) and a negligible role for debt securities (with the euro being the dominant issuing currency in most non-euro area CESEE EU Member States). The predominantly bank-based nature of the financial systems of the CESEE EU Member States is also confirmed by the fact that the total ­assets of nonbank financial institutions are well below the level of domestic bank credit to the private sector, thus playing a relatively smaller role in financial intermediation than in the euro area and the U.S.A. In view of this evidence, we review the key factors which have so far prevented a more dynamic development, describe major efforts undertaken to overcome these detrimental factors and synthesize proposals by various institutions for future action to deepen local capital markets in the region, including in the context of the European Union’s capital markets union.

JEL classification: D14, D18, D31, D63, E44, G21, G28, H81

Keywords: CESEE, capital markets, financial intermediation, European Union, capital markets union

This descriptive study gives a comparative overview of the status quo of key segments of the local capital markets in CESEE EU Member States and of major developments since 2010. To complete the picture, we often distinguish between euro area members and nonmembers, and we provide comparisons with euro area and U.S. totals. We draw on a variety of data sources, such as quarterly financial balance sheets, securities statistics and securities exchange-trading statistics ­published by the ECB, the World Bank’s Global Financial Development database, the IMF’s Financial Development Index and market statistics provided by – among others – the World Federation of Exchanges and the Federation of European Securities Exchanges. Note that we did not intend to analyze in detail, e.g. econometrically, the determinants of capital market development.

In what follows, section 1 provides the big picture view using the IMF’s Financial Development Index. Section 2 adds insights from financial sector accounts data into the role of securities as a financing instrument in CESEE, for the national economies as a whole and for individual sectors. Section 3 cross-checks this information with World Bank data on the role of the nonbank financial sector in financial intermediation. Section 4 discusses structural issues of capital market development in CESEE EU Member States, covering impediments to development and efforts made so far, and highlighting potential ways forward. Section 5 provides a concluding overview.

1 The big picture view with the IMF’s Financial Development Index

A more detailed analysis of the IMF’s Financial Markets Index does not identify a general pattern in development across the three dimensions (see chart 2 and table 1). While market depth is most pronounced in some countries (Bulgaria, Croatia, ­Estonia, Lithuania and Slovakia), market access ranks first in four other countries (Hungary, Poland, Latvia and Slovenia), and market efficiency is the most advanced of the three areas in two other countries (Czechia and Romania). A more homogeneous picture arises when we compare the data with the euro area: for most CESEE EU Member States, the largest gaps with the euro area were observed for market depth and market access in 2017. Market efficiency levels were closer to euro area levels (with Hungary and Poland even outperforming the euro area), which is partly explained by the fact that this is the least developed dimension in the euro area (also compared to the U.S.A.). ⁷

Table 1: IMF Financial Development Index: Financial Markets Index

	Financial Markets Index	of which:
Depth			Access			Efficiency
1995	2010	2017	1995	2010	2017	1995	2010	2017	1995	2010	2017
	Values range between 0 (worst) and 1 (best)
Non-euro area CESEE EU Member States
Bulgaria	0.04	0.07	0.06	0.03	0.11	0.10	0.01	0.03	0.02	0.08	0.05	0.05
Czechia	0.19	0.17	0.19	0.09	0.14	0.16	0.17	0.03	0.09	0.32	0.35	0.32
Croatia	0.13	0.12	0.10	0.03	0.24	0.23	0.02	0.05	0.03	0.36	0.04	0.02
Hungary	0.24	0.58	0.34	0.04	0.28	0.18	0.51	0.53	0.51	0.19	1.00	0.35
Poland	0.32	0.34	0.34	0.03	0.22	0.23	0.35	0.42	0.45	0.64	0.38	0.36
Romania	0.01	0.05	0.09	0.00	0.04	0.05	0.00	0.01	0.01	0.01	0.12	0.21
Euro area CESEE EU Member States
Estonia	0.39	0.13	0.11	0.14	0.15	0.14	0.09	0.08	0.09	1.00	0.14	0.10
Lithuania	0.05	0.06	0.05	0.01	0.08	0.06	0.05	0.04	0.04	0.11	0.06	0.04
Latvia	0.07	0.08	0.07	0.01	0.06	0.06	0.05	0.15	0.12	0.17	0.02	0.03
Slovenia	0.46	0.32	0.10	0.02	0.24	0.10	0.72	0.66	0.14	0.73	0.05	0.06
Slovakia	0.03	0.06	0.05	0.03	0.05	0.08	0.02	0.06	0.02	0.04	0.08	0.02
EA-19	0.32	0.47	0.44	0.21	0.49	0.50	0.37	0.51	0.47	0.39	0.41	0.34
U.S.A.	0.68	0.91	0.90	0.74	0.99	0.99	0.51	0.71	0.68	0.79	1.00	1.00
Source: IMF.

2 Digging into balance sheet data: the role of securities as a financing instrument

2.1 The role of securities in financing the economy as a whole

These statistics document (1) the generally lower level of financial penetration (i.e. outstanding balances as a percentage of GDP) in the CESEE EU Member States compared to the euro area and the U.S.A. and (2) the fact that financing through loans plays a substantially more important role than financing through ­securities in most CESEE EU Member States – at least at the level of the total economy. As securities play a much smaller role in financing the economy than in the U.S.A., the financing structure in the CESEE EU Member States resembles at best that of the euro area.

2.1.1 Securities and loans outstanding

If we look at the share of these four types of financing instruments in total ­financial liabilities, the dominant role of loans in CESEE EU Member States be­comes even clearer (see chart 4). Loans are a more relevant financing instrument in the CESEE EU Member States than in the euro area and the U.S.A., while ­investment fund shares play a much smaller role. Taken together, the share of listed shares and debt securities in total financial liabilities has been approaching the euro area average in several CESEE EU Member States, while remaining considerably lower in Bulgaria, Estonia and Latvia. From 2010 to 2018, the combined share of debt securities, listed shares and investment fund shares in total financial liabilities tended to rise in the CESEE EU Member States, predominantly on account of debt securities and investment fund shares.

For the sake of completeness, we mention that the remaining part of total financial liabilities consists mainly of (1) nonlisted shares and other equity, (2) other accounts payable including trade credits and (3) deposits taken in by banks. Besides, a substantial part of nonlisted shares and other equity stems from foreign direct investment, particularly in the case of CESEE EU Member States.

Next, we take a closer look at total debt securities and listed shares outstanding. Concerning debt securities, a breakdown of the balances outstanding by issuing sector ⁸ reveals that these balances tend to be heavily skewed toward debt securities issued by the general government (see chart 5 and table 2): In 2018, the government sector accounted for more than 75% of overall issuance, compared to around 50% in the euro area and the U.S.A. Notable exceptions were Estonia (where nonfinancial corporations were the most prominent issuers of debt securities, followed by banks and nonbank financial corporations) and Czechia (where banks were ­important issuers beside the general government).

Table 2: Debt securities and listed shares by issuing sector

		Total economy	Central bank	Banks	Nonbank­ ­financial ­corporations	Nonfinancial corporations	General ­government	Households1
		2010	2018	2010	2018	2010	2018	2010	2018	2010	2018	2010	2018	2010	2018
		Balances outstanding in % of GDP
Non-euro area CESEE EU Member States
Bulgaria	Debt securities	13.0	22.5	0.0	0.0	0.4	0.3	0.7	0.7	1.7	2.8	10.3	18.7	0.0	0.0
	Listed shares	11.3	10.9	0.0	0.0	1.1	0.7	2.1	4.7	8.1	5.5	0.0	0.0	0.0	0.0
Czechia	Debt securities	49.3	54.1	0.0	0.0	6.3	14.9	0.4	0.9	7.1	6.9	35.6	31.4	0.0	0.0
	Listed shares	20.3	11.4	0.0	0.0	4.3	3.7	0.0	0.0	16.1	7.6	0.0	0.0	0.0	0.0
Croatia	Debt securities	36.3	59.5	0.0	0.0	0.1	0.2	0.0	0.0	3.0	3.3	33.2	55.9	0.0	0.0
	Listed shares	43.0	34.7	0.0	0.0	9.5	9.1	0.7	0.3	32.8	25.4	0.0	0.0	0.0	0.0
Hungary	Debt securities	84.7	74.9	11.5	0.0	13.3	5.0	0.4	0.9	2.0	1.1	57.6	67.9	0.0	0.0
	Listed shares	20.4	18.5	0.0	0.0	5.3	7.5	0.2	0.1	14.8	10.9	0.0	0.0	0.0	0.0
Poland	Debt securities	56.8	54.4	5.2	2.5	1.1	3.5	0.2	0.6	3.2	5.2	47.1	42.7	0.0	0.0
	Listed shares	35.1	25.8	0.0	0.0	10.9	8.2	3.5	2.3	20.7	15.2	0.0	0.0	0.0	0.0
Romania	Debt securities	16.7	30.0	0.0	0.0	0.2	0.2	0.0	0.0	0.1	0.1	16.3	29.7	0.0	0.0
	Listed shares	8.0	8.1	0.0	0.0	2.0	1.9	0.0	0.0	6.0	6.2	0.0	0.0	0.0	0.0
Euro area CESEE EU Member States
Estonia	Debt securities	6.8	7.6	0.0	0.0	0.2	1.5	0.3	1.3	4.7	4.1	1.6	0.8	0.0	0.0
	Listed shares	11.3	10.4	0.0	0.0	0.0	0.0	0.0	1.1	11.3	9.3	0.0	0.0	0.0	0.0
Lithuania	Debt securities	35.9	34.5	0.0	0.0	1.4	0.0	0.4	0.1	0.4	2.4	33.8	32.0	0.0	0.0
	Listed shares	15.2	7.4	0.0	0.0	1.5	0.5	0.5	0.1	13.2	6.7	0.0	0.0	0.0	0.0
Latvia	Debt securities	13.2	33.2	0.0	0.0	1.7	0.5	0.0	1.6	0.5	0.6	11.0	30.5	0.0	0.0
	Listed shares	1.6	2.6	0.0	0.0	0.4	0.0	0.0	0.0	1.3	2.6	0.0	0.0	0.0	0.0
Slovenia	Debt securities	52.5	71.3	0.0	0.0	14.0	0.3	0.1	0.2	2.3	2.1	36.1	68.6	0.0	0.0
	Listed shares	19.3	13.9	0.0	0.0	1.7	2.7	2.1	2.7	15.5	8.5	0.0	0.0	0.0	0.0
Slovakia	Debt securities	43.1	66.6	0.0	0.0	5.4	7.6	0.2	4.8	0.1	5.8	37.4	48.3	0.0	0.0
	Listed shares	0.9	1.8	0.0	0.0	0.4	1.0	0.0	0.0	0.6	0.8	0.0	0.0	0.0	0.0
EA-19	Debt securities	164.7	156.9	0.0	0.0	54.4	35.2	30.8	31.9	9.5	11.9	70.1	77.8	0.0	0.0
	Listed shares	48.7	60.7	0.0	0.1	4.8	4.0	6.0	9.3	37.9	47.3	0.0	0.0	0.0	0.0
U.S.A.	Debt securities	212.6	202.7	0.0	0.0	4.5	1.4	88.4	68.1	26.4	30.6	91.5	101.6	1.8	1.1
	Listed shares	111.0	148.0	0.0	0.0	4.2	5.3	22.7	37.1	84.1	105.6	0.0	0.0	0.0	0.0
Source: ECB, Eurostat, OECD, authors‘ calculations.
1 Including nonprofit institutions serving households.

To complete the picture, we perform a currency breakdown of debt securities outstanding, based on the ECB’s securities statistics. However, it should be noted that financial balance sheet data on financial instruments and the ECB’s securities statistics are not strictly comparable, resulting in substantial differences between the published values in a few cases. ⁹ At the end of 2018, the outstanding balances of debt securities were dominated by issues in the national currencies in almost all CESEE EU Member States. The two notable exceptions were Bulgaria and Croatia. In Bulgaria, the overwhelming majority of debt securities issued was denominated in euro (which is not surprising given the euro-based currency board operated in that country), while in Croatia nearly half of the outstanding volume was denominated in euro (possibly encouraged by the tight peg of the kuna to the euro) and about a quarter in other foreign currencies. In other non-euro area CESEE EU Member States, debt securities denominated in euro accounted for up to a third of the outstanding balances, and up to around 10% were denominated in other foreign currencies. Clearly, foreign currency issuance was less widespread among the CESEE EU Member States participating in the euro area, except for Lithuania.

For listed shares, a breakdown by issuing sector also finds evidence of substantial concentration, with nonfinancial corporations accounting for the bulk of the balances outstanding in the CESEE EU Member States, same as in the euro area and the U.S.A. Unlike in the euro area and the U.S.A., a significant portion of total listed shares outstanding was also attributable to banks, partly even (subsidiaries of) foreign parent banks, in 2018 (see table 2). The only exception are the Baltics and Bulgaria, with Bulgaria moreover being the only country in which nonbank financial corporations accounted for a notable portion of the outstanding amount in 2018.

On the currency structure of listed shares outstanding, the ECB’s securities statistics indicate that listed shares are almost exclusively denominated in the national currencies.

More generally, the ECB’s securities statistics on listed shares confirm the picture painted by the financial balance sheet data by showing that equity market ­capitalization ¹⁰ decreased in the majority of CESEE EU Member States ¹¹ from 2010 to 2018 (see table 3). To some extent, this decrease reflected declines in equity prices. However, in half of the CESEE EU Member States we also observe a drop in market capitalization deflated by the respective main equity index. Moreover, the number of listed companies fell in most of the countries in the sample from 2010 to 2018. This may not necessarily have been a significant phenomenon, though, in so far as it reflects merely the delisting of companies which were hardly ever traded or otherwise active in the stock exchange.

Table 3: Equity market size and trading activity

	Equity market capitalization in % of GDP	Number of companies with listed shares	Equity market capitalization in % of GDP	Value traded in % of GDP	Turnover velocity
	2010	2018	2010	2018	2010–2018
Non-euro area CESEE EU Member States
Bulgaria	14.5	24.4	390	274	15.2	1.3	0.1
Czechia	20.4	11.4	27	54	14.9	4.9	0.3
Croatia	42.3	34.9	240	127	38.3	0.9	0.0
Hungary	20.8	18.9	52	43	16.7	9.1	0.5
Poland	39.3	28.2	585	851	33.0	13.8	0.4
Romania	7.8	8.9	69	87	10.0	1.4	0.1
Euro area CESEE EU Member States
Slovenia	19.3	13.9	72	31	14.3	1.0	0.1
Slovakia	5.0	6.2	165	58	6.0	0.1	0.0
Euronext	68.7	86.0	1.135	1.208	80.0	72.9	0.9
U.S.A.	115.3	148.5	..	..	137.3	214.0	1.6
Source: ECB, Eurostat, Federation of European Securities Exchanges, World Bank (WDI), World Federation of Exchanges, authors‘ calculations.

The size of stock markets (measured as equity market capitalization as a percentage of GDP) does not necessarily correlate with trading activity. In terms of size, the Zagreb Stock Exchange (Croatia) led the league among the CESEE EU Member States both in 2018 and on average in the 2010–2018 period. However, equity trading activity in the Zagreb Stock Exchange was comparatively low, being the third-lowest in the region in 2018 and the second-lowest over the 2010–2018 period (see table 3). Trading activity in this sample of countries was highest in the stock exchanges in Warsaw, Budapest, Prague and Bucharest, expressed both as trading volumes in percent of GDP and as turnover velocity (i.e. as a ratio of market capitalization). However, even in the two most liquid equity markets in the CESEE EU Member States, namely Hungary and Poland, trading activity was substantially lower than in Euronext ¹² or the U.S.A.

2.1.2 Volume of transactions in financing instruments

However, net issuance of debt securities eased in almost all CESEE EU countries over the past few years, as the net incurrence of liabilities in debt securities as a percentage of GDP was lower during 2016–2018 than in the longer period of 2010–2018. In several CESEE EU Member States, this development reflects the lower issuance volume of government debt securities as a result of fiscal consolidation. The declining net issuance of debt securities contrasted with a rising net incurrence of loan liabilities in most countries of the region.

Based on financial flows, the CESEE EU economies compare better with the euro area than on the basis of outstanding volumes. ¹³ Financial transactions in debt securities were larger as a percentage of GDP in nearly half of the CESEE countries than in the euro area – both during 2010–2018 and 2016–2018 – yet still reaching only a tenth of the U.S. level. (Similarly, transactions in loan liabilities as a percentage of GDP surpassed the euro area average in five CESEE EU countries in the 2016–2018 period.)

For completeness, we note that in the 2010–2018 period funding from sources other than those discussed here – including unlisted shares, trade credit and ­deposits – was far more important on balance in almost all countries, including the euro area and the U.S.A., reaching a share in total funding substantially above 50%. The only exceptions were Slovakia and Slovenia where the share of these other funding sources was about 45% and 20%, respectively.

Next, we look at the structure of new net funding via debt securities by issuing sector. Like outstanding volumes, the net incurrence of debt securities liabilities was concentrated on debt securities issued by general government entities in the CESEE EU Member States as well as in the euro area and the U.S.A. (see chart 7). The two most notable exemptions were Czechia (where banks and nonfinancial corporations issued above all debt securities to raise funds during the past few years) and Slovakia (where nonbank financial corporations accounted for most of debt securities net issuance during 2016–2018).

2.2 The role of securities in financing individual economic sectors

2.2.1 Securities issued by banks

Table 4: Share of securities in total financial liabilities

	Banks	Nonbank financial ­corporations	Nonfinancial corporations	General government
		2010	2018	2010	2018	2010	2018	2010	2018
		% of total financial liabilities of issuing sector
Non-euro area CESEE EU Member States
Bulgaria	Debt securities	0.4	0.3	2.8	1.5	0.5	0.9	41.1	57.6
	Listed shares	1.1	0.7	8.3	10.6	2.3	1.8	0.0	0.0
	Investment fund shares	0.1	0.0	2.3	3.3	0.0	0.0	0.0	0.0
Czechia	Debt securities	6.1	10.3	1.0	1.4	3.4	3.7	71.8	69.1
	Listed shares	4.1	2.6	0.0	0.0	7.8	4.1	0.0	0.0
	Investment fund shares	1.1	0.1	8.1	13.5	0.0	0.0	0.0	0.0
Croatia	Debt securities	0.1	0.2	0.0	0.0	1.0	1.2	28.5	40.9
	Listed shares	7.7	8.2	1.7	0.5	11.5	9.1	0.0	0.0
	Investment fund shares	1.6	1.6	6.9	7.7	0.0	0.0	0.0	0.0
Hungary	Debt securities	10.4	5.4	0.2	0.8	0.8	0.4	65.3	73.3
	Listed shares	4.2	8.0	0.1	0.1	5.7	4.3	0.0	0.0
	Investment fund shares	4.0	0.9	5.4	13.7	0.0	0.0	0.0	0.0
Poland	Debt securities	1.3	3.9	0.5	1.2	2.0	3.3	72.6	58.9
	Listed shares	12.8	9.3	7.7	4.9	13.3	9.8	0.0	0.0
	Investment fund shares	0.1	0.0	17.8	27.9	0.0	0.0	0.0	0.0
Romania	Debt securities	0.3	0.4	0.1	0.0	0.0	0.1	36.9	59.6
	Listed shares	3.0	4.0	0.2	0.1	3.1	3.9	0.0	0.0
	Investment fund shares	0.9	0.0	13.3	22.0	0.0	0.0	0.0	0.0
Euro area CESEE EU Member States
Estonia	Debt securities	0.2	1.5	0.7	1.7	1.6	1.4	7.7	3.7
	Listed shares	0.0	0.0	0.0	1.4	3.8	3.1	0.0	0.0
	Investment fund shares	0.0	0.0	8.9	5.1	0.0	0.0	0.0	0.0
Lithuania	Debt securities	1.5	0.1	2.1	0.5	0.2	1.3	63.0	76.0
	Listed shares	1.6	0.8	2.8	0.5	7.1	3.8	0.0	0.0
	Investment fund shares	0.1	0.0	3.0	7.2	0.0	0.0	0.0	0.0
Latvia	Debt securities	1.1	0.6	0.0	2.9	0.2	0.3	16.3	54.3
	Listed shares	0.2	0.0	0.0	0.0	0.6	1.4	0.0	0.0
	Investment fund shares	0.4	0.0	3.1	2.2	0.0	0.0	0.0	0.0
Slovenia	Debt securities	10.1	0.4	0.3	0.6	0.9	1.1	62.1	73.8
	Listed shares	1.2	3.2	4.4	6.9	6.3	4.5	0.0	0.0
	Investment fund shares	0.0	0.2	12.7	13.0	0.0	0.0	0.0	0.0
Slovakia	Debt securities	6.7	8.3	0.8	12.4	0.1	3.9	56.6	53.0
	Listed shares	0.4	1.1	0.0	0.0	0.4	0.5	0.0	0.0
	Investment fund shares	3.2	0.0	12.1	19.8	0.0	0.0	0.0	0.0
EA-19	Debt securities	16.9	13.4	11.4	8.7	3.4	3.8	69.6	69.3
	Listed shares	1.5	1.5	2.2	2.5	13.8	14.9	0.0	0.0
	Investment fund shares	3.6	3.8	22.6	25.7	0.0	0.0	0.0	0.0
U.S.A.	Debt securities	3.9	1.2	24.8	19.1	9.2	8.7	70.3	72.7
	Listed shares	3.7	4.5	6.4	10.4	29.4	30.0	0.0	0.0
	Investment fund shares	16.1	12.5	16.9	20.0	0.0	0.0	0.0	0.0
Source: Eurostat, OECD, authors‘ calculations.

The issuance of debt securities in euro (beside the given national currency) was common among banks in non-euro area CESEE EU Member States. By contrast, banks in euro area CESEE EU Member States rather avoided debt securities issuance in foreign currencies, with the exception of Latvia (see chart 9 and table 5).

Table 5: Currency structure of debt securities

	Banks	Nonbank financial corporations	Nonfinancial ­corporations	General ­government
		2010	2018	2010	2018	2010	2018	2010	2018
		% of total debt securities issued by sector
Non-euro area CESEE EU Member States
Bulgaria	National currency	12.1	12.7	12.2	17.8	3.7	9.8	37.5	22.7
	Euro	87.9	87.3	87.8	79.4	96.3	90.1	38.5	77.3
	Other foreign currencies	0.0	0.0	0.0	2.7	0.0	0.1	23.9	0.1
Czechia	National currency	81.2	70.4	90.5	32.5	11.7	33.2	81.5	89.7
	Euro	15.5	29.5	9.5	67.5	79.0	52.8	17.1	9.9
	Other foreign currencies	3.3	0.1	0.0	0.0	9.3	14.0	1.3	0.4
Croatia	National currency	43.1	42.2	100.0	18.5	10.4	23.1	29.9	33.0
	Euro	56.9	57.8	0.0	81.5	66.7	44.1	38.2	44.2
	Other foreign currencies	0.0	0.0	0.0	0.0	23.0	32.9	31.9	22.9
Hungary	National currency	49.0	64.2	100.0	96.7	11.6	32.1	66.7	79.8
	Euro	50.4	20.6	0.0	2.4	83.1	65.5	20.2	6.9
	Other foreign currencies	0.6	15.2	0.0	1.0	5.3	2.4	13.1	13.4
Poland	National currency	85.7	59.6	100.0	96.9	72.8	79.8	76.8	76.6
	Euro	7.4	37.9	0.0	2.0	23.2	19.9	15.1	15.9
	Other foreign currencies	6.9	2.5	0.0	1.1	4.0	0.3	8.1	7.4
Romania	National currency	85.0	93.7	0.0	100.0	0.0	53.0	64.4	54.9
	Euro	15.0	6.3	0.0	0.0	0.0	47.0	35.6	35.1
	Other foreign currencies	0.0	0.0	0.0	0.0	0.0	0.0	0.0	10.0
Euro area CESEE EU Member States
Estonia	National currency	100.0	100.0	100.0	98.0	94.7	99.5	100.0	98.5
	Euro	0.0	..	0.0	..	0.0	..	0.0	..
	Other foreign currencies	0.0	0.0	0.0	2.0	5.3	0.5	0.0	1.5
Lithuania	National currency	62.5	100.0	100.0	100.0	14.0	100.0	10.0	69.4
	Euro	32.1	..	0.0	..	86.0	..	53.2	..
	Other foreign currencies	5.4	0.0	0.0	0.0	0.0	0.0	36.7	30.6
Latvia	National currency	6.7	75.9	0.0	70.4	62.8	100.0	58.6	88.7
	Euro	79.7	..	0.0	..	37.2	..	41.4	..
	Other foreign currencies	13.6	24.1	0.0	29.6	0.0	0.0	0.0	11.3
Slovenia	National currency	100.0	100.0	100.0	100.0	100.0	100.0	100.0	90.4
	Euro	..	..	..	..	..	..	..	..
	Other foreign currencies	0.0	0.0	0.0	0.0	0.0	0.0	0.0	9.6
Slovakia	National currency	97.2	98.7	100.0	96.7	100.0	98.9	100.0	93.9
	Euro	..	..	..	..	..	..	..	..
	Other foreign currencies	2.8	1.3	0.0	3.3	0.0	1.1	0.0	6.1
EA-19	National currency	82.2	74.6	83.7	70.9	81.5	82.9	97.6	97.8
	Euro	..	..	..	..	..	..	..	..
	Other foreign currencies	17.8	25.4	16.3	29.1	18.5	17.1	2.4	2.2
Source: ECB, authors‘ calculations.

2.2.2 Securities issued by nonbank financial corporations

Debt securities issuance in euro (beside the national currency) was very common among nonbank financial corporations in some non-euro area CESEE EU Member States (Bulgaria, Czechia and Croatia) but negligible in others (Hungary, Poland and Romania; see table 5). Like in the case of banks, nonbank financial corporations in euro area CESEE EU Member States rather avoided issuance in foreign currencies, again except for Latvia.

2.2.3 Securities issued by nonfinancial corporations

Nonfinancial corporations in most non-euro area CESEE EU Member States preferred debt securities issuance in euro (in Croatia also in other foreign currencies) to issuing debt in their national currency (see chart 11 and table 5). Only in Romania and Poland was the share of foreign currency-denominated debt securities below 50%, at about 45% and 20%, respectively. In the euro area CESEE EU Member States, almost all debt securities issued by nonfinancial corporations and outstanding at end-2018 were denominated in euro, reflecting the avoidance of issuance in foreign currency like in the case of banks and nonbank financial corporations.

2.2.4 Securities issued by general government entities

The currency structure of general government debt securities (see chart 12 and table 5) strongly resembles that of the total economy, which is not surprising, given that the total outstanding volume of debt securities was dominated by general govern­ment debt instruments.

3 Cross-checks with World Bank data: the role of the nonbank financial sector in financial intermediation

Data collected by the World Bank (Global Financial Development database) indicate that the nonbank financial institution sector – such as mutual funds, insurance companies and pension funds; corresponding roughly to the sector of nonbank financial corporations discussed above – is much less developed in the CESEE EU Member States than in the euro area, which itself looks less developed than the U.S. market. Moreover, in contrast to the euro area and the U.S.A., the total ­assets of nonbank financial institutions are smaller – in many countries markedly smaller – than domestic bank credit to the private sector (see chart 13), corroborating evidence from financial balance sheets about the predominantly bank-based nature of the CESEE EU Member States’ financial systems.

Within the nonbank financial institution sector, mutual funds account for the smallest portion of this sector’s total assets in most CESEE EU Member States. Also, the gap of mutual funds in the assets-to-GDP ratio vis-à-vis the euro area is substantial and larger than in the case of insurance companies or pension funds. The only exception is Hungary, where mutual fund assets as a percentage of GDP are much larger than elsewhere in the region and are also closer to the euro area, despite a still substantial gap. Mutual fund assets as a percentage of GDP expanded between 2010 and 2016 (latest available data) across the region, but due to the substantial rise in the corresponding euro area ratio, the gaps widened further.

Insurance corporations’ assets as a percentage of GDP were also substantially smaller in CESEE EU Member States than in the euro area in 2016, although, as mentioned above, the gaps were much smaller than for mutual funds. The gaps with the euro area narrowed from 2010 to 2016, mostly due to the decrease of the assets-to-GDP ratio in the euro area, while the corresponding ratios roughly stagnated or rose modestly in the CESEE EU Member States.

Pension funds’ assets-to-GDP ratio rose markedly in most CESEE EU Member States (except for Poland and Hungary) from 2010 to 2016, thus coming closer to euro area levels. Differences in the size and development of this sector across the region have reflected policy choices affecting the pension systems. Specifically, pension fund assets expanded most dynamically in those countries which operate mandatory second-pillar funded pension schemes or have recently started operating such schemes. The expansion was less pronounced in countries where second-pillar pension funds are not mandatory (e.g. Czechia, Lithuania, Slovenia or Slovakia). Pension fund assets even decreased substantially as a percentage of GDP in Hungary and Poland, where pension reforms were reversed, and pension fund ­assets were nationalized partially (Poland in 2014) or almost completely (Hungary in 2011) a few years ago.

Alternative financial sources are also substantially less widespread in the ­CESEE EU Member States than elsewhere in the EU. Gross private equity investments reached an aggregated amount of at most 0.2% of GDP in the region over the period from 2007 to 2018, compared to the European average of nearly 0.4%. The biggest annual average inflows (as a percentage of GDP) were registered in Czechia, Estonia, Poland, Serbia and Hungary. Considering also annual divestments by private equity firms, cumulative net private equity investments were highest in Estonia and Serbia (0.14% of GDP), followed by Poland and Latvia (0.11% of GDP), but still falling short of the European average (0.15%). Other forms of early stage financing are even less common in the CESEE EU Member States. For instance, business angel investments reached 0.05% of GDP in the ­region in 2017. The online alternative finance market (such as peer-to-peer lending, crowdfunding, minibonds, profit-sharing, etc.) is also generally rather small (up to 0.03% of GDP in 2017), with the notable exception of the three Baltic countries (0.15%–0.34% of GDP).

4 Structural issues of capital market development in CESEE EU ­Member States

4.1 Impediments to capital market development

First, there are historical reasons and legacy issues. ¹⁴ The first issue in this context has its roots in the period of socialism, when nonfinancial companies that were state-owned and administratively linked to the government had a comparatively strong debtor position, while banks were formed for their benefit and partly even by them. Thus, loan decisions were not the result of banks’ assessment of creditworthiness. Rather, they reflected the arrangement of companies and government authorities using financing entities. These practices contributed to the legacy of a culture of informality and lack of transparency and openness in society with respect to financing decisions. Companies have thus been hesitant to expose themselves to constant disclosure requirements and investor control following a stock exchange listing. Hence, for instance, in some countries, firms were listed initially in line with legal requirements but then delisted as soon as legal requirements al­lowed it. However, one might argue that this legacy would not inhibit the emergence of stronger private equity markets.

Hence, it is important to recall the following second legacy issue that stems from the final stage of the centrally planned system and the years of transformation: a low level of trust, particularly on the side of investors. People’s financial confidence was eroded by very high inflation coupled with (initially suppressed) currency crises and then banking crises. Add to this disappointing experiences with problematic forms of (coupon) privatizations and financial fraud like e.g. ­pyramidal schemes in the early 1990s. In the banking sector, the entrance of foreign-owned banks decisively contributed to stabilization and the stepwise build­up of confidence in many countries, luring “mattress money” in foreign currency into foreign currency deposits and, finally, local currency deposits. Following partly clearly excessive and imprudent foreign currency lending before the Great Financial Crisis, the avoidance of a severe banking crisis, thanks to public support in parent banks’ home countries and ample liquidity provision by the ECB, strength­ened public confidence further, implying a large increase of domestic local currency deposits. By contrast, following the problems in the 1990s, there was no comprehensive “quick fix” for the capital markets like that for the banking sector in most countries. However, in some countries, capital markets got some boost from the establishment of pension funds in the context of systemic pension reforms. Then, during the Great Financial Crisis, stock market investors, mostly foreign and domestic institutional investors, but also some domestic retail investors, were badly hit. Moreover, in some countries, systemic pension reforms were reversed, mostly due to fiscal policy considerations considering the European Union’s fiscal framework, which supported funded pension schemes only partially and temporarily. These reversals implied a substantial downsizing of pension funds and a weakening of capital market liquidity. Above all, it did not encourage trust in capital market investments in the private sector as an alternative to bank deposits.

Second, building capital markets takes time. It is now only around 20–25 years that CESEE countries started to implement the legal framework for effective capital markets. Despite substantial progress, for example, in the area of insolvency regulation, key indicators such as the length of insolvency procedures or recovery rates still lag behind the EU average in the majority of CESEE EU Member States. In the same vein, the protection of minority investors and corporate transparency are comparably worse in some countries than in the EU on average. The same applies to the regulation of securities exchanges, at least according to data from 2017 when the World Economic Forum last included this indicator in its Global Competitiveness database.

Third, levels of capital market development correlate positively with levels of economic development. On the demand side, more advanced economies tend to have more companies that reach the critical size or have sufficiently large growth potential in order to be interesting candidates for local stock markets or bond issuance. On the supply side, higher levels of disposable income allow bigger capacities to save and accumulate a larger proportion of those savings in the form of more sophisticated financial assets. Moreover, in the early phase of transition, which was characterized by high inflation and high exchange rate volatility or episodes of sudden large devaluation/depreciation, economic agents in CESEE countries suffered from the “original sin,” i.e. the lacking opportunity to borrow abroad in their domestic currency, or from the unavailability of long-term borrowing, even domestically. This was detrimental for the development of capital markets and caused delays and setbacks.

Fourth, size matters. Small economies and small populations reduce economies-of-scale effects, meaning higher relative costs for establishing capital market infrastructure and related regulatory and legal system development. Moreover, with euro adoption being a strategic economic policy target for most CESEE EU Member States, ¹⁵ the prospective future participation in the common capital markets of the euro area may have discouraged a proactive development of local markets for only a “transitional period.”

Small markets also tend to suffer from the lack of liquidity (as suggested by the comparably low stock market turnover velocity rates), which remains a significant disincentive both for potential issuers and investors in the CESEE region. There is a vicious circle in that insufficient liquidity levels prevent investors from entering the market, which in turn prevents liquidity from rising. Insufficient liquidity and the lack of a sufficiently large local investor base may drive companies to foreign capital markets, although there is a substantial domestic bias in equity and debt ­issuance (owing to e.g. language barriers, documentation requirements), or more often to bank loans.

Fifth, the demand for capital has been driven by specific characteristics. Thus, the involvement of foreign companies has been crucial in the transformation of the economies in the CESEE region into market economies. Consequently, FDI-related capital plays an important role as a form of financing of (especially larger) nonfinancial corporations in the region. Financing needs can often be met with FDI equity or intercompany lending from foreign parent companies. Moreover, in foreign-owned companies financing decisions are often taken and implemented at the group level by the foreign headquarters, thus making local financing unnecessary. In addition, as pointed out above, the local banking sectors in CESEE are predominantly foreign-owned, so that banks can rely on parent bank financing (both in form of capital injections and parent bank credit) to complement the local deposit base.

By contrast, SMEs, which account for a substantial portion of value added and employment in the CESEE EU Member States (generally a bigger portion than in the EU as a whole), rather shy away from capital markets due to the complexity of information and cost requirements, covering their external financing needs predominantly by bank loans. Obviously, for individual companies, listing equities on stock exchanges or publicly issuing bonds and complying with related legal requirements, involve substantial costs. These costs may be prohibitive especially for smaller companies, keeping them away from capital markets altogether. Correspondingly, according to a great majority of SMEs in the CESEE region, the costs and regulatory constraints of being listed outweigh potential benefits, particularly in an environment where domestic bank loans are available in abundance.

In recent years, the environment of high liquidity and low interest rates may have fueled search-for-yield attitudes among capital investors willing to supply ­capital, but it provided also rather easy access to bank financing for nonfinancial corporations and hence reduced their incentive to venture into the corporate bond market to gain some interest advantage. However, this may change in the wake of the COVID-19 crisis in as far as banks could tighten their lending standards in view of the uncertainties, the economic recession and probably rising provisions for nonperforming loans.

Sixth, capital market development has also been impeded by specific supply-side characteristics. Thus, investor preferences may also play a role in the relatively slow development of capital markets in the CESEE EU Member States. Institutional investors in several countries in the region have a preference for domestic sovereign bond holdings while households hold a larger share of their financial assets in cash and bank deposits than in the EU as a whole. This issue may be in part related to the fact that financial literacy levels tend to be lower in the CESEE EU Member States than in more advanced EU Member States with deeper capital markets. In addition, given higher owner-occupancy rates in the CESEE EU Member States than in the EU average, presumably a bigger part of households’ overall wealth is locked up in real estate (rather than in financial assets) in the CESEE EU Member States than elsewhere in the EU. Finally, foreign portfolio investors may prefer the shares of a foreign parent company to the domestically listed shares of a subsidiary.

To sum it up, let us quote from a member survey conducted by the CFA Institute, an independent global association for investment management professionals, in early 2018 among 263 investment professionals (portfolio managers, risk managers, consultants, analysts, senior executives) from eight CESEE countries. ¹⁶ The survey sought to identify the main factors preventing growth of capital markets and identify possible solutions to foster deeper and more integrated markets in the ­region. Above all, the respondents shone a light on the scarce supply of listed shares and debt securities, low retail investor demand and administrative burdens that discourage companies from seeking public listings. Further down the list ranked low institutional investor demand, the low level of investor protection and uncertainty about the impact of certain EU regulations and directives. Low levels of ­investor confidence were another factor considered to restrain financial market development in CESEE, with only 38% of respondents having trust in their local market. Nevertheless, respondents did note an improvement in investor protection standards and transparency in their local capital markets compared to the situation five years earlier.

4.2 Efforts already taken to alleviate hindrances

4.2.1 EBRD activities

According to an internal assessment of the initiative prepared in 2017, the EBRD has indeed dedicated considerable effort and skill to delivering specific transactions and technical cooperation projects, resulting in discrete accomplishments in numerous cases. For example, the EBRD contributed to legislative and regulatory improvements through policy dialogue and technical assistance, helped develop benchmark indices, invested in corporate bonds, listed equities and stock exchanges in the region, issued bonds denominated in local currencies on local and international markets and sponsored SEE Link, a regional infrastructure platform for trading securities listed on stock exchanges in six CESEE countries. ¹⁷ At the same time, the assessment came to the rather critical conclusion that there “appears to be a significant disconnect between, on the one hand, the Bank’s high but undefined strategic ambitions for the complex task of transforming LCMs and, on the other, its limited actual capacity to accomplish this given choices it has made about resourcing, prioritization, organization and collaboration with other institutions.”

Therefore, drawing on these lessons, in late 2018 the EBRD’s Board of Directors approved an adapted LC2 Strategy for 2019–2024, setting out in detail how the EBRD would support local currency and capital markets development in its countries of operation. It has been envisaged that LC2 will, for one, facilitate the transition from policy dialogue to investments by the EBRD to support the local currency and capital market development process. Furthermore, efforts will be concentrated on increasing the share of EBRD investments in local currencies and on identifying and supporting sequenced reforms in local capital markets based on four defined priority areas (upgrading the capital markets policy framework, enhancing the legal and regulatory environment, improving capital market infrastructure, and expanding the product range and the investor base).

4.2.2 National authorities’ strategies

National governments – alone or in cooperation with stock exchanges – have also been developing financial and investor education programs to raise awareness of the potential benefits of capital markets and increase financial literacy (e.g. in Latvia, Lithuania, Hungary, Croatia, Czechia, Slovakia). Most CESEE countries have also embarked on pension reforms, introducing second and third pillar-funded pension systems with often mandatory participation (at least in the second pillar). Reforms of the insolvency regimes to reduce complexity and time to resolve insolvencies and raise recovery rates have likewise been beneficial to capital market developments. Also, some countries (e.g. Poland, Hungary) have started public schemes to offer financial support for SMEs in preparing their initial public offerings. Some CESEE countries have introduced public support and a favorable regulatory regime to improve the availability of venture capital, which is especially important for financing start-ups and innovative companies. In some cases, tax legislation has been modified to create incentives for issuers and investors. In Hungary, the central bank bought a majority stake in the Budapest stock exchange (BSE) in 2015, with the explicit aim of implementing a variety of strategic innovations and providing issuers and investors proper incentives.

However, beyond the generally supportive activities for capital market development over the past two decades, national authorities have taken also measures that have weakened or may weaken capital market development. In Poland and Hungary, the reversal of previous systemic pension reforms, leading to the partial (Poland) or almost complete (Hungary) nationalization of the assets of mandatory second-pillar pension funds, has weakened the institutional investor base. Also, deliberate measures to make government securities more attractive for retail investors (e.g. by preferential above-market interest rates, tax exemption for interest income, broad marketing channels – like in Hungary since mid-2019) may create undue competition for other, “nonprivileged” forms of investment (like corporate bonds, equity or investment fund shares).

4.2.3 Market participants’ initiatives

Furthermore, in order to improve the supply of tradable instruments for investors and to ease access to the capital market for smaller companies, stock exchanges in several countries of the CESEE region have launched dedicated stock market segments for SMEs and bond markets for smaller issuers. These offer lighter listing requirements and lower compliance costs and listing fees, to improve the access of smaller companies to capital markets.

4.3 Potential ways forward

• Continue to develop national capital market strategies.
• Foster a friendly business environment, e.g. stable legal and judiciary system and efficient administration. Ensure the rule of law for all market participants. Continue to improve insolvency frameworks to give viable companies a second chance to restructure and offer investors more certainty.
• Establish a sound and harmonized regulatory and supervisory environment for financial markets with strong institutions (eventually through the establishment of a common EU regulatory body). The introduction of common standards would reduce investment costs, boost cross-border investments, improve the level of transparency and enhance investor protection. Common rules across markets would also help narrowing differences in market liquidity.
• Support the implementation of EU-wide regulatory requirements. In view of ongoing regulatory changes affecting the banking sector (e.g. modified minimum requirements for own funds and eligible liabilities (MREL)), there are concerns that local bond markets in several CESEE EU Member States are not ­sufficiently deep and liquid to allow for the smooth absorption of required issuance volumes. Moreover, there are concerns that increased bond issuance by banks established in the CESEE region to raise bail-in-able funds may counteract efforts to develop local bond markets for nonfinancial corporations by crowding out nonfinancial corporations’ bond issuance in so far as investor demand does not strengthen sufficiently. As a result, support by international financial institutions in the placement of such bank bonds has been called for. In this vein, the EBRD actively participated in the placement of such bank bonds in Romania in December 2019, by acting as an investor itself and by mobilizing other investors.
• Ease investment regulations of local institutional investors to allow them to invest in a wider range of asset classes (e.g. venture capital, unlisted securities, real estate funds, etc.).
• Strengthen financial literacy among investors, particularly retail investors, so that they better understand capital markets and become aware of potential benefits as well as risks involved by diversifying away from bank deposits and loans. Enhance financial literacy also among (potential) issuers, particularly in the SME segment.
• Develop a local base of issuers. Encourage state-owned enterprises to raise funds through the local capital market by issuing bonds or listing equity. Privatize state-owned enterprises through the local stock exchange.
• Offer targeted assistance to SMEs in their access and participation in the capital market to compensate for excessive costs to prepare initial public offerings and comply with capital market-related administrative and legal requirements (e.g. prospectus, regular public financial reporting, analytical research reports).
• Promote alternative financing for SMEs via instruments like venture capital, private equity, private placements, minibonds or equity crowdfunding.
• Start-ups and high-tech companies, which are essential for economic development beyond the “middle-income trap,” are said to have rather low asset turnover (due to, for instance, long research and development phases) and have quite often had difficulties getting bank loans. Fostering private equity funds and ­venture capital would support these companies and thus strengthen economic development. During the recovery after the corona crisis, the low interest rate environment will probably prevail for longer and policymakers should make use of the tide from safe-haven flows toward a renewed search for yields, thus benefiting these alternative forms of investment as well.
• Meet the growing demand for green finance. In the coming years, policymakers could use the increased demand for green finance to channel funds into ecologically advanced companies through programs that incentivize, screen and rate such companies.
• Adopt supportive tax measures. Simplify the tax system, including with respect to capital gains and withholding tax. Consider tax incentives for investors in ­order to attract savings to capital markets, and for certain capital market vehicles for funding to SMEs (e.g. venture capital, private equity). Implement favorable tax treatment and easier tax procedures for companies going public to offset high costs that listed companies and to-be-listed companies must incur in relation to higher transparency and regulatory requirements.
• Increase economies of scale by enhanced regional cooperation among governments, market participants and market infrastructure providers, e.g. by expanding existing geographical alliances to non-allied markets and deepening their ­integration. Create new stock market segments dedicated to listing and trading foreign equities in order to broaden the scope of instruments available to local investors. Facilitate listing on other EU markets for local issuers. Develop cross-border cooperations of national central security depositories to support cross-border trading by simplified clearing and settlement procedures.
• Tackle the impact of COVID-19 on banks’ balance sheets rather sooner than later. The corona-induced recession will probably leave banks with higher volumes of nonperforming loans (NPLs). In the years of NPL work-out following the Great Financial Crisis, capital investors specialized in NPL investments have already penetrated the CESEE region. Often, these are foreign-based, e.g. London-based, investors that focus on specific industries across countries. In cooperation with these investors, NPL trading platforms could be used as a loan price determination tool that should be more efficient than outright sales to single investors.

5 Conclusions

Three decades after the start of economic transition, capital markets in the CESEE EU Member States continue to be substantially less developed than in the euro area and the U.S.A., judging from the balances of debt securities, listed shares and investment fund shares outstanding. In the case of listed shares outstanding, the volume-to-GDP ratio even declined in most countries of the region over the past few years. Progress has been just as mixed across countries and uneven across time using a longer horizon of 25 years. However, based on financial flows, fund­ing via debt securities in the CESEE EU Member States compares considerably better with the euro area than based on stock data. Nevertheless, this result stems almost exclusively from the issuance of government debt securities.

Looking at the financial liabilities of the total economy as a percentage of GDP, the comparably limited role of debt securities and even more so of listed shares and investment fund shares is apparent. By contrast, loan penetration levels already come closer to the euro area and the U.S.A. Concerning the structure of liabilities, the dominant role of loans in CESEE EU Member States becomes even clearer: loans in the CESEE EU Member States account for a larger portion of total financial liabilities than debt securities, listed shares and investment fund shares taken together, while in the euro area (and even more so in the U.S.A.) loans play a less important role than these securities together. In sharp contrast to the euro area and the U.S.A., debt securities outstanding in the CESEE EU Member States are dominated by general government papers. Concerning listed shares, nonfinancial corporations account for the bulk of issuance in all three regions. Listed shares ­issued by financial corporations are dominated by banks, often subsidiaries of foreign parent banks, in the CESEE EU Member States and by nonbank financial institutions in the euro area and the U.S.A. With respect to listed shares, CESEE EU Member States lag behind more developed countries not only in terms of ­market depth but also market efficiency and liquidity, as suggested by substantially lower turnover velocity.

Looking at the financial liabilities of nonfinancial corporations only, listed shares had some relevance in this sector’s funding structure only in Poland and Croatia, but still falling short of their role in the euro area. The role of debt securities issued by nonfinancial corporations was negligible (like in the euro area), with the euro being the currency of denomination of choice in most non-euro area CESEE EU Member States.

Finally, development gaps are also substantial with respect to the role of nonbank financial institutions, such as mutual funds, insurance companies and pension funds in financial intermediation, while the market size of alternative financing sources (e.g. private equity, business angels, online alternative financing) is negligible. Moreover, in contrast to the euro area and the U.S.A., the total assets of nonbank financial institutions are smaller – in many countries substantially smaller – than domestic bank credit to the private sector, confirming the predominantly bank-based nature of the CESEE EU Member States’ financial systems.

Further to the overall finding of gaps between the CESEE countries and the euro area, we also found substantial heterogeneity across the CESEE region with respect to the overall development of capital markets, the role of securities in the financing of both the total economy and individual institutional sectors and the currency structure of debt securities.

Expecting that the development of capital markets could generate substantial additional financing opportunities for the corporate sector, international institutions, national authorities and market participants have taken various supportive measures over the past 10 to 15 years. These have included the launch of national capital market strategies, measures to promote financial literacy, support for SMEs in going public, new market segments on stock exchanges with lighter regulatory requirements designed specifically for smaller companies, regional cooperation of stock exchanges, etc. We discussed several historical and structural reasons why capital markets in CESEE continue to play a rather marginal role in financing the private sector. Among these figure inter alia the legacy of informality, the disappoint­ment with certain privatization procedures, the small size of most CESEE econ­omies, the dominance of FDI, higher owner-occupancy rates locking a larger part of household wealth in real estate, the stock market bust during the Great Financial Crisis discouraging retail equity investors and also national measures that ­weakened capital markets development, such as for example the dismantling or partial dismantling of private pension funds in Hungary and Poland in line with new fiscal rules and policy considerations. Assessing or estimating the effects of the various measures to promote capital markets in the individual CESEE countries and gauging to what extent this might explain the heterogeneity within the region would go beyond the scope of this overview and remains a challenge for other authors.

The still fairly limited state of development of CESEE capital markets suggests that more remains to be done if one wants to increase the contribution of capital markets to financing domestic investment. In recent years, various institutions have suggested a wide range of measures to prop up capital markets in the region. These proposals include the launch of additional national capital market strategies, improvements in the business environment (e.g. judiciary, government administration, rule of law), enhancements to the regulatory and supervisory frameworks (including at the regional level) and support for the implementation of EU-wide regulatory requirements (e.g. related to MREL). Moreover, the strengthening of financial literacy among (retail) investors, and also among potential (SME) issuers, the development of the local investor and issuer base, including by tax incentives and targeted assistance especially for SMEs, start-ups and high-tech companies, as well as the regional cooperation of national governments and stock exchanges to increase economies of scale rank prominently among the suggestions. Taking the opportunities of green finance proactively and tackling the impact of COVID-19 on banks’ balance sheets via NPL trading platforms linked to specialized capital investors remain current challenges. Such measures, if decided and implemented, could possibly prevent the sleeping beauty from turning into a dead duck.

References:

Arpa M., T. Reininger and Z. Walko. 2005. Can Banking Intermediation in the Central and Eastern European Countries Ever Catch up with the Euro Area? In: Focus on European Economic Integration Q2/05. OeNB. 110–133.

Cambridge Centre for Alternative Finance. 2016. Sustaining Momentum: The 2nd European Alternative Finance Industry Report.

Cambridge Centre for Alternative Finance. 2018. Expanding Horizons: The 3rd European Alternative Finance Industry Report.

Cambridge Centre for Alternative Finance. 2019. Shifting Paradigms: The 4th European ­Alternative Finance Benchmarking Report.

EBAN – European Trade Association for Business Angels, Seed Funds and Early Stage Market Players. 2016 and 2017. Statistics Compendium: European Early Stage Market ­Statistics.

EBRD. 2017. EBRD Support for the Development of Local Capital Markets. EBRD Evaluation ­Department Special Study. July. www.ebrd.com/evaluation

EBRD. 2018. LC2 Strategy 2019–2024. November.

European Commission. 2018a. European Financial Stability and Integration Review 2018.

European Commission. 2018b. The 2018 Pension Adequacy Report: Current and Future ­Income Adequacy in Old Age in the EU. Volume II. Country profiles.

European Commission. 2019. Capital Markets Union. Communication (2019) 136 final.

Invest Europe. 2012 to 2018. Central and Eastern Europe Statistics. Annual private equity statistics.

Invest Europe. 2019. European Private Equity Activity 2018. Statistics on Fundraising, Investments and Divestments. May.

Klapper L., A. Lusardi and P. van Oudheusden. 2015. Financial Literacy Around the World: Insights from the Standard & Poor’s Ratings Services Global Financial Literacy Survey. November. https://gflec.org/wp-content/uploads/2015/11/3313-Finlit_Report_FINAL–5.11.16.pdf

OECD. 2016. OECD/INFE International Survey of Adult Financial Literacy Competencies. www.oecd.org/finance/OECD-INFE-International-Survey-of-Adult-Financial-Literacy-Competencies.pdf

Panagiotis A. and W. Wright. 2019. Unlocking the Growth Potential in European Capital Markets. June. www.newfinancial.org

Silvestri R. 2019. Unlocking the Potential of Capital Markets in Central and Eastern European Countries. January. https://blogs.cfainstitute.org/marketintegrity/2019/01/29/unlocking-the-potential-of-capital-markets-in-central-and-eastern-european-countries/

Vienna Initiative. 2018. Report by the Working Group on Capital Markets Union. March. http://vienna-initiative.com/resources/themes/vienna/wp-content/uploads/2018/03/VI-CMU-Working-Group-Final-Report-March-2018.pdf

Whitehouse E. 2011. Reversals of Systemic Pension Reforms in Central and Eastern Europe: Implications. OECD Social Policy Division.

Wright W., J. Suarez, P. McGhee and L. Bax. 2016. The Benefits of Capital Markets to High Potential EU Economies. November. www.newfinancial.org, www.afme.eu.

How financially literate is CESEE? Insights from the OeNB Euro Survey

Sarah Reiter, Elisabeth Beckmann ¹⁹

Drawing on data from the OeNB Euro Survey, we document financial literacy across ten countries in Central, Eastern and Southeastern Europe (CESEE-10) between 2012 and 2018. We analyze people’s understanding of the “big three” concepts of financial literacy: interest rates, inflation and risk diversification. We show that financial literacy differs across and within countries. On average, just one in five adults can be considered financially literate. Our results show that low financial literacy levels are more common among older and less educated individuals and that self-employment is only weakly related to higher literacy. In line with previous research, females show lower levels of financial literacy than their male counterparts. However, the gender gap observed in the CESEE-10 (countries with a communist legacy) is small compared to the gap in countries that do not have a communist legacy. Individuals who experienced economic turbulence during transition from planned to market economies tend to be more financially literate regarding inflation. While indicators of economic and financial development are correlated with higher financial literacy at the country level, interactions are more complex at the intracountry level.

JEL classification: D14, D83, D91, G53

Keywords: financial literacy, interest rates, inflation, risk diversification, gender gap, CESEE

From previous research, financially literate individuals are known to be (1) more successful at job planning and saving for retirement (Behrman et al., 2012); (2) more likely to participate in the stock market (Almenberg and Dreber, 2015; van Rooij et al., 2011); and (3) more likely to diversify their savings (Hastings et al., 2013). In contrast, individuals who lack financial literacy are more prone to take high-cost loans and become overindebted (Lusardi and Tufano, 2015) and to encounter ­repayment difficulties (Gerardi et al., 2013).

To remedy this situation, many countries have implemented national strategies for financial education seeking to improve “financial literacy with a view to promoting healthier financial behaviors and improving financial well-being” (OECD, 2015). In this context, the OECD argues that “policymakers, educators and ­researchers need high-quality data on levels of financial literacy in order to inform financial education strategies” (OECD, 2013). Indeed, the number of surveys undertaken to gain a better understanding of financial literacy is steadily increasing. However, some of these surveys are not appropriate for comparisons across countries as they vary regarding the sociodemographic groups that are surveyed. For instance, some studies consider only young or only old people whereas others consider adults in general. Moreover, the studies also differ in the way they define and measure financial literacy. ²¹

To deal with this shortcoming, numerous efforts have been made to come up with harmonized definitions and measures of financial literacy, allowing for cross-country comparisons. One of the most prominent of these projects was initiated by the OECD International Network on Financial Education (OECD/INFE), which provides harmonized data on financial literacy for 30 countries (Atkinson and Messy, 2012; OECD, 2016). Furthermore, the World Bank has developed ­financial capability surveys and is actively involved in developing financial education strategies (World Bank, 2014). Another well-known initiative is the so-called Financial Literacy around the World (FLat World) ²² project, which collects answers to three standard financial literacy questions on interest rates, inflation and risk diversification (Lusardi and Mitchell, 2008). In the literature, these questions have come to be known as the “big three” (see, for example, Barboza et al., 2016).

In this paper, we present unique survey evidence on individuals’ understanding of the “big three” concepts of financial literacy for ten Central, Eastern and Southeastern European (CESEE-10) countries for the period from 2012 to 2018. One distinct advantage of our data is that the wording of the financial literacy questions put to respondents (always the adult population) and the survey mode were exactly the same in all ten countries. We contribute to the literature by presenting evidence for countries with a rather short history of developed financial systems and consumer finance. To the best of our knowledge, this is the first paper that analyzes evidence for the three standard financial literacy questions over a period of more than five years using a dataset that has sufficient observations to dig into intracountry heterogeneities. Moreover, we add to the FLat World project by providing comparable statistics on financial literacy for countries that have not yet been covered by the project. We present evidence of how financial literacy varies across sociodemographic groups and, by comparing our results to those from other surveys, we show that in the CESEE-10 (all countries with a communist legacy) the gender gap in financial literacy is smaller than in countries that do not have a communist legacy (similar findings were made by Cupák et al., 2018). For each of the ten CESEE countries under study, the online annex of our paper provides indicators of financial literacy for different sociodemographic groups and regions. This evidence can be used as input for policy work and further research.

The rest of this paper is organized as follows: Section 2 describes the data; section 3 presents details on the “big three” financial literacy questions and discusses the strengths and weaknesses of these questions. Section 4 presents the corresponding results, describing the variation in financial literacy across countries. ²³ We also compare our financial literacy results with those of other surveys that have been conducted in the CESEE-10. Section 5 analyzes intracountry variation of ­financial literacy and its correlates. Section 6 discusses the variation in financial literacy across sociodemographic groups, section 7 describes how the transition experience is related to literacy and section 8 provides concluding remarks.

1 Data: the OeNB Euro Survey

The Euro Survey has been conducted on a regular basis since 2007 as a repeated cross-sectional face-to-face survey. In each country and in each survey wave, a sample (based on multistage random sampling procedures) of around 1,000 individuals is interviewed. Each sample reflects a country’s population characteristics in terms of age, gender, region and ethnicity.

When interpreting the results presented in this paper, the following issues should be taken into account: Nonresponse varies across countries and across survey waves. The gross sample size ranges approximately from 1,500 to 3,000 across countries and waves. The number of interrupted interviews is zero in some countries and up to 200 in other countries. In the absence of information on the number of individuals who refused to participate in the survey, we cannot construct nonresponse weights. Regarding unit nonresponse, we do not impute missing values but assume that nonresponse is random, which is arguably a strong assumption. However, for the central questions of interest – the questions on financial literacy – the share of “no answer” responses is below 3% in all countries and waves.

Weights are calibrated on census population statistics for age, gender, region, and, where available, on education and ethnicity. Weights are calibrated separately for each wave and country. For the majority of countries, population statistics ­relate to the year 2011; more recent census data were available only for some countries. ²⁴

All in all, we use data from six Euro Survey waves between 2012 and 2018, ²⁵ meaning that our dataset covers a total of around 60,000 observations. ²⁶ The central variables of our analyses are derived from the three questions on financial literacy referred to above. Beyond that, the survey questionnaire elicits a rich set of information on socioeconomic characteristics, indicators of wealth and household ­finances, individual beliefs, expectations and trust.

The survey also contains the addresses (at the street level) of the primary sampling units (PSUs), i.e. of the units that are selected in the first stage of the multistage random sampling process, which is ultimately aimed at selecting individual elements. Put simply, these are the points where the interviewer starts walking to select specific addresses, and ultimately individuals, to participate in the survey. Depending on the country, there are between 100 and 300 PSUs per country. The maximum number of interviews conducted around one PSU is 25. We geocode the PSU addresses and calculate the area around the PSU points at different radii (5km, 10km, 20km). For each area, we compute (1) indicators of urbanicity and (2) proxies for local economic activity such as average stable night light following Henderson et al. (2012). We merge the survey data with these indicators at the area level. Both indicators have been shown to be associated with financial literacy at the country level (Klapper and Lusardi, 2019).

3 Measurement of financial literacy

The three financial literacy questions as put to Euro Survey respondents are shown in table 1. The questions on interest rates and inflation use the original wording. With regard to the question on risk diversification, which originally ­referred to stock mutual funds, various surveys have used a different wording, as stock mutual funds are not commonly known in all countries; see for example the S&P Global Finlit Survey (Klapper and Lusardi, 2019). ²⁷ The Euro Survey follows this approach: respondents are asked whether they think that the risk of losing money when spread among different assets increases, decreases or stays the same, rather than whether they consider a single company stock to provide a safer return than a stock mutual fund. ²⁸

Based on the three questions, previous research commonly defines three binary variables where the correct answer is coded as 1, wrong answers and “do not know” responses are coded as 0, and “refuse to answer” responses are coded as “missing.” The three binary variables are then aggregated to a financial literacy score, defined as the number of correct answers (see, for example, Bucher-Koenen and Lusardi, 2011; and Bucher-Koenen and Ziegelmeyer, 2013). In our paper, we follow this approach and define (1) three separate binary variables for each of the three financial literacy questions and (2) a financial literacy score taking on integer values between 0 and 3.

Table 1: The big three financial literacy questions included in the OeNB Euro Survey

Concept	Question
Interest rate	Suppose you had 100 [local currency] in a savings account and the interest rate was 2% per year. Disregarding any bank fees, how much do you think you would have in the account after 5 years if you left the money to grow: (i) More than 102 [local currency]*(ii) Exactly 102 [local currency](iii) Less than 102 [local currency](iv) Do not know(v) No answer
Inflation	Suppose that the interest rate on your savings account was 4% per year and inflation was 5% per year. Again disregarding any bank fees – after 1 year, would you be able to buy more than, exactly the same as, or less than today with the money in this account?(i) More(ii) Exactly the same(iii) Less*(iv) Do not know(v) No answer
Risk diversification	When an investor spreads his money among different assets, does the risk of losing money(i) Increase(ii) Decrease*(iii) Stay the same(iv) Do not know(v) No answer
Source: OeNB Euro Survey.
Note: Correct answers are marked with an asterisk.

As Lusardi and Mitchell (2014) point out, any given set of financial literacy measures can “only proxy for what individuals need to know to optimize their ­behavior in intertemporal models of financial decision making.”

Indeed, for the specific sample of countries we cover, the concept of risk diversification is probably less important than in the U.S.A. and other high-income countries where these questions were initially developed and implemented. For example, stock market capitalization to GDP is 152% in the U.S.A., compared to 32% in Croatia (the Euro Survey country with the highest percentage). Further indicators, such as life insurance premiums to GDP, pension fund assets to GDP and mutual fund assets to GDP provide a similar picture. ²⁹ It is, therefore, not surprising that saving instruments such as life insurance, pension funds, bonds, stocks and mutual funds are not widespread in the countries we analyze. Instead, for the countries covered by the OeNB Euro Survey, for example, understanding of ­exchange rate risk may be much more relevant in terms of optimizing household financial decisions (Beckmann and Stix, 2015). Nevertheless, we will stick to the “big three” and follow the concept of the FLat World project as this is the only way to allow comparison with results from countries not covered by the OeNB Euro Survey.

Compared to the OECD Toolkit for Measuring Financial Literacy (OECD, 2018), the “big three” questions have the advantage that they can be integrated into existing surveys at low cost. At the same time, this limits the measure to three concepts and gives rise to two general concerns that may be particularly relevant for this limited number of questions: (1) Are responses to the questions affected by measurement error? (2) Do the questions provide a comprehensive measure of financial knowledge?

With regard to the first concern, Crossley et al. (2017) provide an in-depth discussion of measurement error related to financial literacy. They point out that, in contrast to many other survey questions, financial literacy questions test respondents’ knowledge (instead of, say, their opinion), and so the interviewers, who presumably know the correct answers, would be able to help respondents. Crossley et al. (2017) indeed find that interviewer effects are larger for financial literacy questions than for other survey questions. Interviewer variation does not seem to drive the “do not know” responses but is more complex. Guessing the correct answer when being asked a financial knowledge question might also lead to measurement error. Analyzing the framing of financial knowledge questions, van Rooij et al. (2011) find that correctly guessed answers might be mistaken for true financial knowledge.

With regard to the second concern, it has been argued that although the “big three” provide a narrow measure of financial literacy, the three concepts covered by the questions are most relevant to saving and investment decisions (Bucher-­Koenen et al., 2017). Furthermore, compared to research using more complex measures of financial literacy, research based on the “big three” finds similar sociodemographic patterns with regard to lack of financial literacy. For example, a growing body of research on gender and financial literacy documents a gender gap irrespective of the survey measure of financial literacy: Cupák et al. (2018) use the OECD surveys; Driva et al. (2016) use the “big three” plus additional questions on financial literacy; Bucher-Koenen et al. (2017) use the “big three”; and Klapper and Lusardi (2019) use the S&P Global Finlit Survey. Irrespective of the financial literacy measure used, robust patterns have been identified also with respect to other sociodemographic characteristics such as age, employment status and education.

In this paper, we address the above-mentioned concerns in the following way: We compare our results to evidence from other surveys conducted in the ten ­CESEE countries covered by the OeNB Euro Survey. ³⁰ As can be seen in the next section, the relative levels in financial knowledge among countries remain more or less the same no matter what measure of financial knowledge is used.

4 Financial literacy: variation across countries

In line with the common approach in the above-mentioned studies, we compute indicators of cross-question consistency. On average, across all countries, 19% of respondents answer all three questions correctly whereas 22% fail to answer any of the three questions correctly. The percentage of respondents who answer both the inflation and interest rate questions correctly is significantly higher at 33%. Overall, the six CESEE EU countries perform better than the four CESEE non-EU countries. In the online annex, we also provide a detailed overview of how financial literacy evolved over time between 2012 and 2018.

In table 3, we compare the financial literacy results for the CESEE-10 countries with those of other countries participating in the FLat World project. ³²

With a share of 38.6% of respondents correctly answering all three financial literacy questions, Czechia ranks among the best-performing countries. However, most of the other CESEE countries under study are at the lower end of the ranking. The countries differ considerably in terms of economic and financial development. Hence, cross-country comparisons between the CESEE-10 countries and the other FLat World countries should be taken with caution. GDP per capita is in general low in the CESEE-10 countries when compared to the other FLat World countries. The only FLat World country with GDP per capita figures in the range of most CESEE-10 countries is Chile, which also exhibits low financial literacy rates. Italy and Czechia are comparable in terms of GDP per capita, but less so in terms of overall financial literacy. In general, table 3 also shows that literacy does not steadily increase with GDP per capita.

Table 3 also suggests that the gender gap in financial literacy increases with overall financial literacy. See section 6 for a discussion of this aspect.

Table 2: Summary statistics on the big three financial literacy questions

	Full sample	CESEE EU	CESEE non-EU
	%
Interest rate
More than 102*	51.7	51.5	52.7
Exactly 102	16.3	16.0	18.2
Less than 102	11.7	12.3	8.1
Do not know	18.1	18.3	16.8
No answer	2.3	1.9	4.2
N	61,564	36,777	24,787
Inflation
More	11.3	10.9	13.2
Exactly the same	17.8	17.2	21.3
Less*	48.5	49.6	42.4
Do not know	19.9	20.2	18.1
No answer	2.5	2.0	5.1
N	61,564	36,777	24,787
Risk diversification
Increase	19.7	18.3	27.5
Decrease*	39.8	41.4	30.6
Stay the same	16.3	16.4	16.2
Do not know	21.7	21.8	20.9
No answer	2.6	2.2	4.8
N	61,564	36,777	24,787
Cross-question consistency
Correct answers for interest rate and inflation	32.6	32.9	31.0
All answers correct	18.6	19.7	12.1
None of the answers correct	21.7	21.6	22.5
“Do not know” selected at least once	33.2	33.6	30.6
“Do not know” selected for all answers	9.5	9.5	9.6
N	58,732	35,573	23,159
Source: OeNB Euro Survey, 2012–2016 and 2018.
Note: The statistics are based on weighted data. Correct answers are marked with an asterisk. N = number of observations. The “cross-question consistency” panel covers only those respondents who gave an answer to all three questions.

To test the robustness of our financial literacy measure, we provide an overview in table 4 comparing our results with those from other studies on financial literacy conducted in the CESEE-10: Klapper et al. (2015) analyze people’s financial knowledge in more than 140 economies including all CESEE-10. With its International Network on Financial Education, the OECD analyzes financial knowledge, financial behavior and financial attitudes in various countries, including four of the ten CESEE countries in their first study (Atkinson and Messy, 2012) and five of them in their second study (OECD, 2016). ³³ Subject to the constraint that the studies differ in their definition of financial literacy and that the financial literacy measures are based on a different range of questions (ranging from three to eight), we find that the relative position of the ten CESEE countries with respect to financial knowledge is robust across the different surveys. Across all surveys, Czechia and Hungary consistently show the highest levels of financial knowledge within the CESEE-10. In contrast, the four CESEE non-EU countries generally show low levels of financial knowledge when compared to the other six CESEE EU countries.

Table 4: Financial knowledge in the CESEE-10: a comparison across different studies

	OeNB Euro Survey	Klapper et al. (2015)	Atkinson and Messy (2012)	OECD/INFE (2016)
Country	3 out of 3	Rank	N	Year	3 out of 4	Rank	N	Year	6 out of 8	Rank	N	Year	5 out of 7	Rank	N	Year
	%				%				%				%
Bulgaria	22.8	3	5,850	x	35	6	n.a.	2014	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
Croatia	19.1	5	5,885	x	44	3	n.a.	2014	n.a.	n.a.	n.a.	n.a.	46	4	1,049	2015
Czechia	38.6	1	6,109	x	58	1	n.a.	2014	57	2	1,005	2010	52	3	1,000	2015
Hungary	25.1	2	5,912	x	54	2	n.a.	2014	69	1	998	2010	60	1	1,000	2015
Poland	19.6	4	5,800	x	42	4	n.a.	2014	49	3	1,008	2010	55	2	1,000	2015
Romania	7.5	10	6,017	x	22	8	n.a.	2014	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
Albania	9.5	9	6,021	x	14	10	n.a.	2014	45	4	1,000	2011	43	5	1,000	2015
Bosnia and Herzegovina	10.2	8	5,706	x	27	7	n.a.	2014	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
North Macedonia	10.9	7	5,835	x	21	9	n.a.	2014	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
Serbia	14.7	6	5,597	x	38	5	n.a.	2014	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
Source: OeNB Euro Survey (2012–2016; 2018), Klapper et al. (2015), Atkinson and Messy (2012) and OECD/INFE (2016).
Note: The definition of “being financially knowledgeable” varies across studies: to qualify as financially knowledgeable, respondents must answer 3 out of 3 questions correctly (OeNB Euro Survey); 3 out of 4 questions (Klapper et al., 2015); 6 out of 8 questions (Atkinson and Messy, 2012); or 5 out of 7 questions OECD/INFE, 2016). x indicates the survey waves from 2012–2016 and 2018; n.a. stands for “not available”; and N refers to the number of observations.

Table 3: Financial literacy in the FLat World and OeNB Euro Survey countries

Country	All correct	Gender
	Rank	Result	Male	Female	Difference	Rank	N	Data collection	Source	GDP per capita
		%	%		pp.
Germany	1	53.2	59.6	47.5	12.1	11	1,059	2009	Bucher-Koenen and Lusardi (2011)	46,988
Switzerland	2	50.1	62.0	39.3	22.7	1	1,500	2011	Brown and Graf (2013)	77,452
Austria	3	48.8	55.0	43.0	12.0	12	1,342	2019	Fessler et al. (2020)	49,190
Netherlands	4	44.8	55.1	35.0	20.1	2	1,665	2010	Alessie et al. (2011)	53,920
Australia	5	42.7	52.0	34.0	18.0	4	1,024	2012	Agnew et al. (2013)	56,229
Canada	6	42.5	51.4	32.9	18.5	3	6,805	2012	Boisclair et al. (2017)	51,126
Czechia	7	38.6	41.2	36.2	5.0	15	6,109	x	Own analysis	22,755
Finland	8	35.6	44.0	27.1	16.9	5	1,477	2014	Kalmi and Ruuskanen (2018)	47,559
France	9	30.9	36.3	26.0	10.3	13	3,616	2011	Arrondel et al. (2013)	43,002
U.S.A.	10	30.2	38.3	22.5	15.8	7	1,488	2009	Lusardi and Mitchell (2011b)	53,356
Japan	11	27.0	34.3	20.6	13.7	9	5,268	2010	Sekita (2011)	48,439
Hungary	12	25.1	25.7	24.6	1.1	24	5,912	x	Own analysis	15,696
Italy	13	24.9	29.5	17.0	12.5	10	3,992	2007	Fornero and Monticone (2011)	35,029
New Zealand	14	24.0	32.0	16.0	16.0	6	850	2009	Crossan et al. (2011)	37,678
Bulgaria	15	22.8	24.5	21.2	3.3	16	5,850	x	Own analysis	8,331
Sweden	16	21.4	29.3	13.6	15.7	8	1,302	2010	Almenberg, Säve-Söderbergh (2011)	56,611
Poland	17	19.6	20.5	18.8	1.7	22	5,800	x	Own analysis	15,826
Croatia	18	19.1	22.0	16.6	5.4	14	5,885	x	Own analysis	15,332
Serbia	19	14.7	16.2	13.2	3.0	17	5,597	x	Own analysis	6,560
North Macedonia	20	10.9	12.0	9.9	2.1	20	5,835	x	Own analysis	5,257
Bosnia and Herzegovina	21	10.2	11.6	8.9	2.7	18	5,706	x	Own analysis	5,828
Albania	22	9.5	10.8	8.2	2.6	19	6,021	x	Own analysis	4,868
Romania	23	7.5	8.6	6.5	2.1	21	6,017	x	Own analysis	11,017
Chile	24	7.7	n.a.	n.a.	n.a.	n.a.	14,463	2009	Garabato Moure (2016)	14,749
Russia	25	3.1	3.8	2.5	1.3	23	1,366	2009	Klapper and Panos (2011)	11,470
Source: FLat World project and OeNB Euro Survey; World Bank (Global Financial Development).
Note: For the ten OeNB Euro Survey countries (“own analysis”), the statistics are based on weighted data; x indicates waves from 2012–2016 and 2018. For Romania, also see Beckmann (2013). For Austria, the statistics are not based on the big three questions, but on a survey that uses the OECD toolkit for measuring financial literacy. n.a. stands for “not available” and N refers to the number of observations. GDP per capita refers to 2017 and is expressed in constant 2005 USD.

5 Financial literacy: regional variation

Klapper and Lusardi (2019) analyze economic and financial factors that drive variation in financial literacy across countries. They find a significant positive correlation between literacy and GDP per capita as well as between literacy and consumer protection laws. Indicators of financial development and financial stability are significantly correlated with literacy levels in developed countries only. As our sample covers only ten countries, we do not analyze what determines differences in literacy levels across these ten countries. However, we take the analysis of Klapper and Lusardi (2019) to the intracountry level, examining which factors might explain the observed regional variation in financial literacy in the CESEE-10.

Table 5 shows correlations at the PSU level between financial literacy (computed as the average literacy score ranging from 0 to 3) and various indicators of local economic and financial development. ³⁴ Following Henderson et al. (2012), we use average stable night lights as an indicator of economic development. Our results show that the positive correlation of financial literacy and economic development only holds in some countries. For others, such as Bulgaria, Hungary, Poland and North Macedonia, we find that more economically developed areas have lower levels of financial literacy. It is important to note, however, that these correlations do not control for other factors and that night light is, for example, highly correlated with indicators of urbanization (see column 2 of table 5, which shows the same negative/positive correlation pattern as column 1).

Unlike Klapper and Lusardi (2019), we find a significant correlation between financial literacy levels and indicators of financial development. ³⁵ Again, bearing in mind that these correlations do not control for other factors, our results indicate that literacy levels increase with bank proximity and density in most countries. While higher bank concentration is associated with lower literacy levels in Croatia, Hungary and Serbia, the opposite is the case in Bulgaria, Czechia, Poland and North Macedonia. To investigate whether the local banking environment is merely a proxy for the development of the local infrastructure, we also look at correlations with local road density (see the final column of table 5). The positive and significant correlation of road density and literacy for Croatia, Czechia and Serbia coincides with the positive and significant correlation of bank proximity and literacy. Taken together, these results suggest that it is worth investigating in depth the determinants of intracountry heterogeneity in financial literacy in future research. In particular, it would be informative for policymakers to investigate to what extent intracountry heterogeneities are related to heterogeneities in economic activity alone or to what extent institutions and factors that may be influenced by concrete policy measures are relevant.

Table 5: Intracountry correlation of financial literacy and indicators of economic and ­financial development

	Night light	Urban fabric	km to next bank branch	Number of banks within 5 km	Bank concentration	Road density
Bulgaria	–0.0520***	–0.0582***	–0.0631***	0.0402***	0.0359***	0.005
Croatia	0.0703***	0.0786***	–0.0610***	0.0808***	–0.0772***	0.0717***
Czechia	0.2017***	0.1905***	–0.3132***	0.2049***	0.1052***	0.1545***
Hungary	–0.0372***	–0.0371***	–0.025	0.0626***	–0.0708***	–0.012
Poland	–0.0344***	–0.0619***	0.008	–0.0508***	0.0621***	–0.0596***
Romania	0.0407***	0.0783***	–0.0801***	0.0452***	0.002	0.030
Albania	0.023	0.031	–0.0358***	0.0396***	–0.001	–0.016
Bosnia and Herzegovina	0.02	0.010	–0.029	0.026	0.029	0.0376***
North Macedonia	–0.0663***	–0.1133***	–0.1296***	0.0877***	0.0372***	–0.017
Serbia	0.1153***	0.1249***	–0.0995***	0.1345***	–0.0852***	0.1411***
Source: OeNB Euro Survey, 2012–2016 and 2018.
Note: The table shows the “Pearson product-moment correlation coefficient” between average financial literacy and different economic and financial development indicators measured at the level of the primary sampling unit (PSU). The OeNB Euro Survey data are combined with non-survey data by collecting the geographic coordinates of the PSUs and computing indicators of economic and financial development for different perimeters around the PSUs. *** indicates significance at the 1% level (not adjusted for sampling design). “Night light” is obtained from the National Oceanic and Atmospheric Administration (NOAA) and represents average stable night lights based on the VIIRS Nighttime light series for a ­radius of 20km around the PSUs. “Urban fabric” is obtained from the CORINE Land Cover database and represents the area covered by continuous and discontinuous urban fabric for a radius of 20km around the PSUs. Indicators of bank proximity (km to next bank branch from the PSUs), density (number of banks within 5km around the PSUs) and concentration (Herfindahl index of bank concentration calculated based on the number of bank branches within 5km around the PSUs) are derived by combining data from the OeNB Euro Survey with bank branch data as described in Beckmann et al. (2018). “Road density” is obtained from the Global Road Inventory Dataset Project and represents road infrastructure for a radius of 5km around the PSUs.

6 Financial literacy: variation across sociodemographic groups with a focus on gender gaps

In this section, we investigate whether these patterns are also prevalent across the ten Euro Survey countries and whether differences in financial literacy among certain sociodemographic groups (e.g. between men and women) are more pronounced in some countries than in others. We do so by aggregating results for the six EU Member States and the four EU candidates and potential candidates. ³⁶ To maintain comparability with the other studies in the FLat World project, we stick to the structure of the tables in the related publications and especially to the definitions of the sociodemographic categories.

We find that financial literacy is lowest among adults aged 65+ both in CESEE EU countries (see table 6) and in CESEE non-EU countries (see table 7). This is the case for all three aspects of financial literacy. This finding is in line with results from previous empirical research (e.g. Klapper and Lusardi, 2019) and with theoretical models, where financial literacy is defined as a choice in the context of life-cycle models (Lusardi et al., 2015). Learning by doing would suggest that ­financial literacy is also lower for the youngest (Frijns et al., 2014). In line with empirical research for developed countries, table 6 shows an inverted U-shaped relationship between literacy and age (with the exception of literacy regarding ­interest rates). Results in table 7, however, are more in line with empirical evidence for emerging markets (Klapper and Lusardi, 2019): While literacy levels are not highest among the youngest, they are higher than among those aged 50+.

As expected – and corroborating previous research (for example, Christelis et al., 2010) – financial literacy is strongly correlated with education. The share of “do not know” responses for financial literacy is lowest among respondents with tertiary education.

Regarding respondents’ labor market status, tables 6 and 7 show that financial literacy is higher for those who are working. In the six EU Member States, it is highest for self-employed respondents. In the CESEE non-EU countries, the literacy score for inflation and interest rates, and also the overall level of financial literacy, is highest for non-self-employed working respondents; risk literacy is highest for self-employed respondents. These results hold also when controlling for age and gender in a multivariate analysis.

Empirical evidence on financial literacy has shown large and persistent gender differences. The “gender gap” in financial literacy is also present in the CESEE-10 – with regard to all three concepts of financial literacy and overall literacy. Furthermore, we confirm a gender gap in the share of “do not know” responses. Previous research has sought to explain these differences by studying, for example, whether differences are due to life experience (Driva et al., 2016) or household decision-­making (Fonseca et al., 2012), or whether they are present only for more complex questions (Bucher-Koenen et al., 2017). More recently, Cupák et al. (2018) provide a cross-country analysis of the gender gap using a measure of financial literacy that goes beyond the “big three.” They argue that while some of the gender gap can be explained by personal characteristics, the rest may be due to an individual’s economic and social environment. Their analysis shows that the gender gap in ­financial literacy is particularly small in Eastern European countries. They hypothesize that the “more equal financial literacy scores may be related to social and economic norms left over from times of communism.”

Table 6: Financial literacy in CESEE EU countries: differences across sociodemographic groups

	Sample	Interest rate	Inflation	Risk diversification	Overall
		Correct	Do not know	Correct	Do not know	Correct	Do not know	All correct	None correct
	%
Age
35 and younger	32.6	56.1	16.3	49.6	18.6	43.5	19.5	19.7	20.0
36–50	25.9	54.8	14.3	52.8	16.7	45.6	18.2	21.2	18.6
51–65	24.2	52.8	17.2	53.9	19.2	42.8	21.8	20.5	20.2
Older than 65	17.4	43.9	29.6	45.7	30.7	36.8	32.2	16.1	30.9
N	35,573
Gender
Male	47.9	54.7	16.1	53.0	18.2	44.7	19.1	20.9	19.4
Female	52.1	51.2	20.3	48.8	22.4	40.8	24.6	18.5	23.5
N	35,573
Education
Low	16.3	43.7	31.5	38.3	33.0	35.3	33.4	13.1	32.8
Medium	65.9	53.4	16.8	52.3	18.7	43.3	20.7	20.0	20.5
High	17.7	59.5	11.7	56.9	14.9	47.6	15.9	24.8	15.1
N	35,488
Employment
Retired	25.5	45.2	27.3	47.2	28.3	35.7	30.4	15.1	28.8
Student	6.4	56.1	18.2	49.5	20.9	43.4	21.2	19.1	20.9
Unemployed/other	10.1	49.4	23.7	45.3	24.9	34.0	26.7	14.0	25.9
Working	50.9	55.9	13.6	53.2	16.2	46.7	17.4	22.0	18.0
Self-employed	7.2	61.1	11.7	57.0	13.6	52.6	15.8	28.6	14.6
N	35,276
Source: OeNB Euro Survey, 2012–2016 and 2018.
Note: The table shows descriptive statistics of the “big three” financial literacy questions for different sociodemographic groups in six CESEE EU Member States. The statistics are based on weighted data. N indicates the number of observations. The sample consists of those respondents who provide answers to all three financial literacy questions and for whom information on sociodemographic characteristics is available.

Chart 2 provides a striking confirmation of this finding, bringing together ­descriptive results from all FLat World studies. ³⁷ The chart illustrates that the gender gap in financial literacy is comparatively lower in transition economies. However, it is important to bear in mind that the overall level of literacy is lower in most of these countries as well, as illustrated in chart 3. The gender gap in financial literacy can be expressed in absolute terms (the share of men with correct answers to all three questions minus the share of women with correct answers to all three questions) or in relative terms (absolute gender gap as a percentage of the overall literacy level). Both measures are of interest: The absolute gender gap illustrates how much “catching up” is necessary so that women’s financial literacy becomes equal to that of men within one country. The relative gender gap lends itself to cross-country comparisons. Such a comparison shows, for example, that Russia is on a similar level with the Netherlands in terms of relative gender differences in literacy. However, these countries are not comparable in terms of GDP per capita, and their overall levels of literacy differ substantially. A comparison of Czechia and ­Italy, which have similar levels of GDP per capita, shows that the overall level of literacy is higher in Czechia and that the gender gap is larger in Italy, both in absolute and relative terms.

Table 7: Financial literacy in CESEE non-EU countries: differences across sociodemographic groups

	Sample	Interest rate	Inflation	Risk diversification	Overall
		Correct	Do not know	Correct	Do not know	Correct	Do not know	All correct	None correct
	%
Age
35 and younger	35.6	57.9	14.5	44.6	16.7	35.1	18.0	12.7	19.9
36–50	26.8	58.4	12.5	47.8	14.0	34.6	16.0	13.8	19.0
51–65	26.0	52.3	19.5	44.2	20.4	30.3	24.9	11.3	25.1
Older than 65	11.6	46.7	28.5	39.4	30.7	25.2	34.1	8.1	32.8
N	23,159
Gender
Male	48.6	58.1	13.8	46.9	15.8	34.7	17.8	13.5	20.1
Female	51.4	52.6	19.8	42.7	21.2	30.5	24.3	10.8	24.8
N	23,159
Education
Low	27.5	46.3	27.3	37.3	28.5	28.0	31.6	8.5	32.0
Medium	57.0	57.7	14.0	46.9	15.8	33.3	18.3	12.7	19.8
High	15.5	62.4	8.9	50.1	11.1	38.2	13.0	16.3	15.9
N	23,151
Employment
Retired	19.9	48.1	24.7	41.3	26.2	25.9	31.9	8.7	29.8
Student	9.8	58.8	16.1	44.4	17.9	36.3	20.4	14.8	20.2
Unemployed/other	27.1	50.7	20.8	42.1	21.9	30.9	22.8	10.0	26.1
Working	36.8	61.5	10.7	49.4	12.7	35.3	15.0	14.8	16.9
Self-employed	6.4	56.2	11.6	43.5	13.2	39.4	15.3	12.7	19.1
N	22,839
Source: OeNB Euro Survey, 2012-2016 and 2018.
Note: The table shows descriptive statistics of the “big three” financial literacy questions for different sociodemographic groups in four CESEE EU candidates or potential candidates. The statistics are based on weighted data. N indicates the number of observations. The sample consists of those respondents who provide answers to all three financial literacy questions and for whom information on sociodemographic characteristics is available.

Hungary, Poland and Chile have roughly similar GDP per capita levels. Unfortunately, the FLat World study for Chile does not contain information on the percentage of men and women that answered all questions correctly (Garabato Moure, 2016). However, the data allow us to compute a gender gap for the individual financial literacy questions and to compare the results with Hungary and Poland (see tables 8 and 9 in the online annex).

This comparison yields the following results: For the interest rate question, a gender gap of 7.8 percentage points for Chile, of 0.8 percentage points for Hungary, and of 3.2 percentage points for Poland. For the inflation question, a gender gap of 4.1 percentage points for Chile, of 2.8 percentage points for Hungary, and of 4.2 percentage points for Poland. For the risk diversification question, a gender gap of 5.4 percentage points for Chile, of 2.7 percentage points for Hungary, and of 4.3 percentage points for Poland. Thus, with the exception of inflation literacy, the gender gap is generally lower in the two countries with a communist legacy than in the country without such a legacy. Providing an in-depth analysis of the low gender gap in Eastern Europe, Beckmann and Reiter (mimeo) argue that the experience of communism alone does not suffice to explain the lower gender gap. The gender gap has decreased even further among cohorts born shortly before or after 1989.

7 Financial literacy and the experience of economic turbulence during transition

By 2012, the first year of our analysis, at least a decade (and mostly a decade and a half or more) had passed since inflation had been in the 20% range or beyond in any of the CESEE-10 countries (see table 8). We could have combined the information on the final high-inflation year with the information on respondents’ age in order to analyze whether older respondents are more literate with regard to inflation than younger respondents. However, the OeNB Euro Survey contains a question whether respondents “remember periods of high inflation during which the value of the local currency dropped sharply,” which provides a more specific measure of personal experience. When comparing the financial literacy measures of those who have memories of high inflation with the literacy of those who do not, we find that the percentage of inflation-literate individuals is higher among those who remember high inflation than among those who do not. For interest and risk diversification literacy, there is no such clear pattern. In fact, the percentage of risk diversification-literate individuals is on average higher among those who do not remember previous economic crises.

Table 8: Transition experience and financial literacy

	Last period of annual average inflation of 20% or higher	Interest rate	Inflation	Risk diversification
		Correct		Correct		Correct
	Year	%
Memory of high inflation*		yes	no	yes	no	yes	no
Bulgaria	1998	51	50	75	69	46	42
Croatia	1994	66	66	64	62	37	39
Czechia	1993	67	65	70	67	61	61
Hungary	1996	59	50	64	54	52	50
Poland	1995	56	55	49	40	50	48
Romania	2002	41	42	52	44	27	29
Albania	1998	41	36	33	28	45	46
Bosnia and Herzegovina	1995	47	46	42	41	34	34
North Macedonia	1994	60	64	46	40	30	35
Serbia	2002	63	65	57	54	27	37
Source: OeNB Euro Survey and wiiw database.
Note: * Memory of high inflation is derived from the survey question “Please tell me on a scale from 1 to 6 whether you agree or disagree with the following statement: “I remember periods of high inflation during which the value of the [LOCAL CURRENCY] dropped sharply.” Answers from 1 to 3 are coded as “yes” while answers from 4 to 6 are coded as “no.”

8 Conclusion

At least one of the “big three” questions is answered correctly by 80% of adults in ­CESEE-10, but correct answers to all three questions are provided by barely one in five adults, and the share of correct answers is generally lowest for risk diversification. These figures vary substantially across and within countries. Our results show that financial literacy varies within countries across sociodemographic groups: In line with previous research, financial literacy is highest among well-­educated and employed individuals, and it is lowest among older and female individuals. Those who experienced economic turbulence during transition from planned to market economies have a higher level of inflation literacy, but overall literacy does not differ.

While previous research has found that higher GDP levels are correlated with higher financial literacy, we show that – within countries – lower economic development is not necessarily associated with lower financial literacy levels. We also find that the development of local financial markets is associated with higher financial literacy levels.

Our findings are relevant both for policymakers and for future research. In this context, it is important to note that explicit mandates to conduct financial education policies are lacking in many countries; and if a mandate has been established, different institutions are in charge in different countries. Therefore, this article does not in any way aim to assess the success of various policy measures that have been carried out in different countries to address – and perhaps already successfully narrow – the financial literacy gaps we describe. Furthermore, we cannot contribute to the discussion to what extent any gaps in financial literacy could perhaps be addressed by financial education efforts in the private sector. Nevertheless, the OECD has shown that in order for financial education to be effective it should target specific segments of the population (OECD, 2019b). Our paper provides some evidence to identify the segments of the population that would benefit most from financial education programs; or put differently, that currently have lower levels of financial literacy. We show that such segments can be identified both in sociodemographic terms and regionally.

The OECD argues that financial literacy can be enhanced by financial education, and that financial education and literacy make a difference with regard to ­financial and economic inclusion (OECD, 2013). With both financial literacy and financial development being low in some regions, our results imply a possible potential for financial education programs to enhance not only literacy but financial ­inclusion as well. However, more research is needed to investigate these indicative results in more depth.

Thus, our paper points out several lines for future research that are relevant for policymakers: Some of the countries we investigate still have a significant share of individuals who do not have a bank account. Do these arguably financially excluded individuals benefit from a developed financial market in terms of improved literacy? Are financially literate individuals more likely to be financially included? More generally, how does financial literacy affect economic outcomes? Finally, there are potential ­research questions that are particularly relevant for the countries covered by the OeNB Euro Survey, which includes a number of countries that will adopt the euro at some point in the future. How does financial literacy affect expectations regarding the development of inflation and the exchange rate? How does financial literacy affect expectations regarding adoption of the euro as legal tender? While some of these questions can only be analyzed based on microdata, for other questions our detailed online annex of descriptive statistics for individual countries will provide a useful reference.

References

Agnew, J. R., H. Bateman and S. Thorp. 2013. Financial Literacy and Retirement Planning in Australia. In: Numeracy 6(2). Article 7.

Alessie, R., M. van Rooij and A. Lusardi. 2011. Financial literacy and retirement preparation in the Netherlands. In: Journal of Pension Economics and Finance 10(4). 527–545.

Almenberg, J. and J. Säve-Söderbergh. 2011. Financial literacy and retirement planning in Sweden. In: Journal of Pension Economics and Finance 10(4). 585–598.

Almenberg, J. and A. Dreber. 2015. Gender, stock market participation and financial literacy. In: Economics Letters 137. 140–142.

Arrondel, L., M. Debbich and F. Savignac. 2013. Financial Literacy and Financial Planning in France. In: Numeracy 6(2). Article 8.

Atkinson, A. and F. Messy. 2012. Measuring Financial Literacy: Results of the OECD/International Network on Financial Education (INFE) Pilot Study. OECD Working Papers on Finance, Insurance and Private Pensions 15.

Barboza, G., C. Smith and J. G. Pesek. 2016. Assessing Financial Literacy, Gender Gap and Cognitive Differences. In: Journal of Financial Education 42(3-4). 205–242.

Barr, N. and P. Diamond. 2006. The Economics of Pensions. In: Oxford Review of Economic Policy 22(1). 15–39.

Beckmann, E. 2013. Financial Literacy and Household Savings in Romania. In: Numeracy 6(2). Article 9.

Beckmann, E. and H. Stix. 2015. Foreign currency borrowing and knowledge of exchange rate risk. In: Journal of Economic Behavior & Organization 112. 1–16.

Beckmann, E., S. Reiter and H. Stix. 2018. A geographic perspective on banking in Central, Eastern and Southeastern Europe. In: Focus on European Economic Integration Q1/18. OeNB. 26–47.

Behrman, J. R., O. S. Mitchell, C. K. Soo and D. Bravo. 2012. How Financial Literacy Affects Household Wealth Accumulation. In: American Economic Review 102(3). 300–304.

Boisclair, D., A. Lusardi and P.-C. Michaud. 2017. Financial literacy and retirement planning in Canada. In: Journal of Pension Economics and Finance 16(3). 277–296.

Bolaji-Adio, A., G. Iarossi, V. Perotti and S. Zottel. 2013. Making sense of financial capability surveys around the World: a review of existing financial capability and literacy measurement instruments. World Bank.

Brown, M. and R. Graf. 2013. Financial Literacy and Retirement Planning in Switzerland. In: ­Numeracy 6(2). Article 6.

Bucher-Koenen, T. and A. Lusardi. 2011. Financial literacy and retirement planning in Germany. In: Journal of Pension Economics and Finance 10(4). 565–584.

Bucher-Koenen, T. and M. Ziegelmeyer. 2013. Once Burned, Twice Shy? Financial Literacy and Wealth Losses during the Financial Crisis. In: Review of Finance 18(6). 2215–2246.

Bucher-Koenen, T., A. Lusardi, R. Alessie and M. van Rooij. 2017. How Financially Literate Are Women? An Overview and New Insights. In: Journal of Consumer Affairs 51(2). 255–283.

Célérier, C. and B. Vallée. 2017. Catering to Investors Through Security Design: Headline Rate and Complexity. In: Quarterly Journal of Economics 132(3). 1469–1508.

Christelis, D., T. Jappelli and M. Padula. 2010. Cognitive abilities and portfolio choice. In: European Economic Review 54(1). 18–38.

Crossan, D., D. Feslier and R. Hurnard. 2011. Financial literacy and retirement planning in New Zealand. In: Journal of Pension Economics and Finance 10(4). 619–635.

Crossley, T. F., T. Schmidt, P. Tzamourani and J. Winter. 2017. Interviewer effects and the measurement of financial literacy. ISER Working Paper Series 2017-06.

Cupák, A., P. Fessler, A. Schneebaum and M. Silgoner. 2018. Decomposing gender gaps in financial literacy: New international evidence. In: Economics Letters 168. 102–106.

Driva, A., M. Lührmann and J. Winter. 2016. Gender differences and stereotypes in financial literacy: Off to an early start. In: Economics Letters 146. 143–146.

Fessler, P., M. Jelovsek and M. Silgoner. 2020. Financial literacy in Austria – focus on millennials. In: Monetary Policy and the Economy Q3/20 (first view). OeNB.

Fonseca, R., K. J. Mullen, G. Zamarro and J. Zissimopoulos. 2012. What Explains the Gender Gap in Financial Literacy? The Role of Household Decision Making. In: Journal of Consumer Affairs 46(1). 90–106.

Fornero, E. and C. Monticone. 2011. Financial literacy and pension plan participation in Italy. In: Journal of Pension Economics and Finance 10(4). 547–564.

Frijns, B., A. Gilbert and A. Tourani-Rad. 2014. Learning by doing: the role of financial ­experience in financial literacy. In: Journal of Public Policy 34(1). 123–154.

Garabato Moure, N. 2016. Financial literacy and retirement planning in Chile. In: Journal of ­Pension Economics and Finance 15(2). 203–223.

Gerardi, K., L. Goette and S. Meier. 2013. Numerical ability predicts mortgage default. In: Proceedings of the National Academy of Sciences 110(28). 11267–11271.

Hastings, J. S., B. C. Madrian and W. L. Skimmyhorn. 2013. Financial Literacy, Financial Education, and Economic Outcomes. In: Annual Review of Economics 5(1). 347–373.

Henderson, J. V., A. Storeygard and D. N Weil. 2012. Measuring Economic Growth from Outer Space. In: American Economic Review 102(2). 994–1028.

Hung, A., A. Parker and J. Yoong. 2009. Defining and Measuring Financial Literacy. RAND Working Paper Series WR-708.

Hung, A., J. Yoong and E. Brown. 2012. Empowering Women Through Financial Awareness and Education. OECD Working Papers on Finance, Insurance and Private Pensions 14.

Huston, S. J. 2010. Measuring Financial Literacy. In: Journal of Consumer Affairs 44. 296–316.

Jump$tart Coalition for Personal Financial Literacy. 2015. National Standards in K-12 Personal Finance Education. www.jumpstart.org/what-we-do/support-financial-education/standards/ .

Kalmi, P. and O.-P. Ruuskanen. 2018. Financial literacy and retirement planning in Finland. In: Journal of Pension Economics and Finance 17(3). 335–362.

Klapper, L. and G. A. Panos. 2011. Financial literacy and retirement planning: the Russian case. In: Journal of Pension Economics and Finance 10(4). 599–618.

Klapper, L., A. Lusardi and P. van Oudheusden. 2015. Financial Literacy Around the World: Insights from the Standard & Poor’s Ratings Services Global Financial Literacy Survey.

Klapper, L. and A. Lusardi. 2019. Financial literacy and financial resilience: Evidence from around the world. In: Financial Management. 1–26.

Lusardi, A. and O. S. Mitchell. 2008. Planning and Financial Literacy: How Do Women Fare? In: The American Economic Review 98(2). 413–417.

Lusardi, A. and O. S. Mitchell. 2011a. Financial Literacy and Planning: Implications for Retirement Well-being. In Lusardi, A. and O. S. Mitchell (eds.). Financial Literacy: Implications for ­Retirement Security and the Financial Marketplace. Oxford University Press.

Lusardi, A. and O. S. Mitchell. 2011b. Financial literacy and retirement planning in the United States. In: Journal of Pension Economics and Finance 10(4). 509–525.

Lusardi, A. and O. S. Mitchell. 2011c. Financial literacy around the world: an overview. In: Journal of Pension Economics and Finance 10(4). 497–508.

Lusardi, A. and O. S. Mitchell. 2014. The Economic Importance of Financial Literacy: Theory and Evidence. In: Journal of Economic Literature 52(1). 5–44.

Lusardi, A. and P. Tufano. 2015. Debt literacy, financial experiences, and overindebtedness. In: Journal of Pension Economics and Finance 14(4). 332–368.

Lusardi, A., P.-C. Michaud and O. S. Mitchell. 2015. Using a Life Cycle Model to Evaluate Financial Literacy Program Effectiveness. CIRANO Working Papers 2015s-14.

Malmendier, U. and S. Nagel. 2011. Depression Babies: Do Macroeconomic Experiences ­Affect Risk Taking? In: Quarterly Journal of Economics 126. 373–416.

Mandell, L. 2008. Financial Literacy of High School Students. In Xiao, J. J. (ed.). Handbook of Consumer Finance Research. Springer. 163–183.

Moore, D. 2003. Survey of Financial Literacy in Washington State: Knowledge, Behavior, Attitudes, and Experiences. Technical Report 03-39, Washington State University.

OECD. 2006. Improving Financial Literacy: Analysis of Issues and Policies. In: Financial Market Trends 2005/2.

OECD. 2013. PISA 2012 Assessment and Analytical Framework: Mathematics, Reading, Science, Problem Solving and Financial Literacy.

OECD. 2015. National Strategies for Financial Education: OECD/INFE Policy Handbook. https://www.oecd.org/daf/fin/financial-education/national-strategies-for-financial-education-policy-handbook.htm.

OECD. 2016. OECD/INFE International Survey of Adult Financial Literacy Competencies. https://www.oecd.org/finance/oecd-infe-survey-adult-financial-literacy-competencies.htm.

OECD. 2018. OECD/INFE Toolkit for Measuring Financial Literacy and Financial Inclusion. https://www.oecd.org/financial/education/2018-INFE-FinLit-Measurement-Toolkit.pdf.

OECD. 2019a. Pensions at a Glance 2019: OECD and G20 Indicators.

OECD 2019b. Financial Education of the Future – Summary Record. OECD-FSCA International Conference. http://www.oecd.org/financial/education/future-fin-edu-2019-summary-record.pdf.

Van Rooij, M., A. Lusardi and R. Alessie. 2011. Financial literacy and stock market participation. In: Journal of Financial Economics 101(2). 449–472.

Sekita, S. 2011. Financial literacy and retirement planning in Japan. In: Journal of Pension Economics and Finance 10(4). 637–656.

Servon, L. J. and R. Kaestner. 2008. Consumer Financial Literacy and the Impact of Online Banking on the Financial Behavior of Lower-Income Bank Customers. In: Journal of Consumer Affairs 42. 271–305.

Vitt, L. A., C. Anderson, J. Kent, D. M. Lyter, J. K. Siegenthaler and J. Ward. 2000. Personal Finance and the Rush to Competence: Financial Literacy Education in the U.S. http://www.isfs.org/documents-pdfs/rep-finliteracy.pdf.

World Bank. 2014. Financial education programs and strategies: approaches and available resources.

Zottel, S., V. Perotti and A. Bolaji-Adio. 2013. Financial capability surveys around the World: why financial capability is important and how surveys can help. World Bank.

Annex

A1 Definitions of financial literacy

1. Definitions that focus on knowledge;

2. Definitions that focus on ability;

3. Definitions that focus both on knowledge and ability, and

4. Definitions that focus both on knowledge and ability and, in addition, specify an outcome to which financial literacy should finally lead.

5. Definitions that cover additional dimensions, including people’s actual financial behavior.

Table A1: Definitions of financial literacy

	Source
Focus on knowledge
1 “Financial literacy is a basic knowledge that people need in order to survive in a modern society.”	Kim (2001) (as cited in Huston, 2010: p. 311)
Focus on ability
2 “Financial literacy is the ability to make informed judgements and to take effective decisions regarding the use and management of money.”	Noctor et al. (1992) (as cited in Huston, 2010: p. 311)
3 Financial literacy is the “ability to evaluate the new and complex financial instruments and make informed judgements in both choice of instruments and extent of use that would be in their own best long-run interests.”	Mandell (2008): pp. 163f
4 “Personal financial literacy is the ability to read, analyze, manage and communicate about the personal ­financial conditions that affect material well-being. It includes the ability to discern financial choices, discuss money and financial issues without (or despite) discomfort, plan for the future and respond competently to life events that affect everyday financial decisions, including events in the general economy.”	Vitt et al. (2000): p. 2
Focus on knowledge and ability to apply knowledge
5 “Financial literacy could be defined as measuring how well an individual can understand and use personal finance-related information.”	Huston (2010): p. 144
6 “Financial literacy refers to a person’s ability to understand and make use of financial concepts.”	Servon and Kaestner (2008): p. 273
7 “Individuals are considered financially literate if they are competent and can demonstrate they have used knowledge they have learned. Financial literacy cannot be measured directly so proxies must be used. Literacy is obtained through practical experience and active integration of knowledge.”	Moore (2003): p. 29
Focus on knowledge, ability to apply knowledge, and outcome
8 “Financial literacy is the ability to use knowledge and skills to manage one’s financial resources effectively for a lifetime of financial security.”	Jump$tart Coalition for Personal Financial Literacy (2015): p. 1
9 Financial literacy is the “knowledge of basic economic and financial concepts, as well as the ability to use that knowledge and other financial skills to manage financial resources effectively for a lifetime of financial well-being.”	Hung et al. (2009)
10 “Financial literacy represents the level of aptitude in understanding personal finance. It often refers to awareness and knowledge of key financial concepts required for managing personal finances and is generally used as a narrower term than financial capability.”	World Bank (2014): p. 1
Definitions that cover additional dimensions, including people’s actual financial behavior
11 Financial literacy is “a combination of awareness, knowledge, skill, attitude and behaviour necessary to make sound financial decisions and ultimately achieve individual financial wellbeing.”	Atkinson and Messy (2012): p. 3
12 “Financial literacy is knowledge and understanding of financial concepts and risks, and the skills, motivation and confidence to apply such knowledge and understanding in order to make effective decisions across a range of financial contexts, to improve the financial well-being of individuals and society, and to enable participation in economic life”	OECD (2013): p. 144
Source: Authors’ compilation.
Note: The table should not be seen as an exhaustive list of all the existing financial literacy definitions. The definitions selected are grouped according to the findings of Huston (2010).

A2 Overview of financial literacy surveys in CESEE-10

Table A2: Overview of financial literacy surveys in CESEE-10

Country	Survey	Year
Bulgaria	Financial Literacy Survey, prepared for the World Bank§	2010
	Standard and Poor’s Global Financial Literacy Survey*	2014
Croatia	Standard and Poor’s Global Financial Literacy Survey*	2014
	OECD/INFE Survey of Financial Literacy Competencies**	2016
Czechia	OECD/INFE Pilot Study***	2012
	Standard and Poor’s Global Financial Literacy Survey*	2014
	OECD/INFE Survey of Financial Literacy Competencies**	2016
Hungary	OECD/INFE Pilot Study***	2012
	Standard and Poor’s Global Financial Literacy Survey*	2014
	OECD/INFE Survey of Financial Literacy Competencies**	2016
Poland	OECD/INFE Pilot Study***	2012
	Standard and Poor’s Global Financial Literacy Survey*	2014
	OECD/INFE Survey of Financial Literacy Competencies**	2016
Romania	Financial Literacy and Financial Services Survey±	2010
	Standard and Poor’s Global Financial Literacy Survey*	2014
Albania	OECD/INFE Pilot Study***	2012
	Standard and Poor’s Global Financial Literacy Survey*	2014
	OECD/INFE Survey of Financial Literacy Competencies**	2016
Bosnia and Herzegovina	Financial Literacy Survey, prepared for World Bank~	2011
	Standard and Poor’s Global Financial Literacy Survey*	2014
North Macedonia	Standard and Poor’s Global Financial Literacy Survey*	2014
	Adult Financial Literacy Competencies in Macedonia^	2018
Serbia	Standard and Poor’s Global Financial Literacy Survey*	2014
Source: * https://gflec.org/wp-content/uploads/2015/11/3313-Finlit_Report_FINAL-5.11.16.pdf?x66755 ** https://www.oecd.org/finance/oecd-infe-survey-adult-financial-literacy-competencies.htm*** https://www.oecd-ilibrary.org/finance-and-investment/measuring-financial-literacy_5k9csfs90fr4-en§ https://microdata.worldbank.org/index.php/catalog/1026/study-description± https://datacatalog.worldbank.org/dataset/romania-financial-literacy-and-financial-services-survey-20 10 ~ http://documents.worldbank.org/curated/en/116701483501101366/Financial-literacy-in-Bosnia-and-Herzegovina ^ https://www.efse.lu/fileadmin/user_upload/user_upload/Financial_Literacy_Report_Final.pdf

Typology of multinationals in Austria: CESEE focus and foreign control as distinct features

Thomas Cernohous, Tomáš Slačík ³⁹

Multinational enterprises (multinationals) play an important role in every economy as they tend to be larger, more capital- and R&D-intensive, more productive and more integrated in global value chains than domestic enterprises. Focusing on multinationals active in Austria, this paper discusses essentially two research questions: Can we categorize Austrian units of multinationals in consistent groups? And can these groups be characterized by meaningful variables? To address these questions, we undertake a microdata-linking exercise to build a comprehensive dataset of multinationals in Austria and use adequate clustering techniques to identify homogeneous and distinct groups without imposing any prior knowledge regarding the number of such groups or their features. This approach enables us to characterize more than 2,500 multinationals in Austria and meaningfully identify eight types of multinationals, the main grouping factors being (1) foreign or Austrian control, (2) special purpose entity versus other form of company, (3) the share of outward investment in Central, Eastern and Southeastern Europe (CESEE) and (4) the degree of trade openness. With this basic research work, we open up a wide range of questions that may serve as the basis for future (applied) analytical work.

JEL classification: C49, F13, F14, F15, F21, F36, F41

Keywords: multinational enterprises, cluster analysis, globalization, partitioning around medoids, typology

“Multinational enterprises (MNEs) are a key channel of globalisation. They serve as the backbone of many global value chains by linking and organizing production across countries and are an important channel for exchanging capital, goods and services, and knowledge across countries. Foreign direct investment (FDI) is necessary for the creation of an MNE.” (OECD, 2015)

Most multinational enterprises are classified in the nonfinancial corporate ­sector, whose fast-growing role is evident from international investment statistics (see chart 1 with data for Austria).

This paper sheds some light on the relevance and role of multinational enterprises active in Austria. After all, a better understanding will enable better ­informed and more targeted decision-making and thus more efficient and effective policymaking on issues like regional development, employment or taxes. Our ­approach is to first identify those Austrian companies that satisfy the OECD ­definition of multinational enterprises with a view to compiling a dataset covering all Austrian companies that are on the giving or receiving end of foreign direct ­investment (FDI). For every company with a direct investment relationship, i.e. with a cross-border equity participation of at least 10%, we then identify further Austrian companies that are part of the same multinational group and assign them to the initial enterprise. Hence, we build a dataset of so-called “truncated enterprise groups” ⁴⁰ of multinationals in Austria, for which we use the term “multinationals” in this paper for the sake of readability. To populate the database further, we link a set of variables, both numerical and categorical, from different statistical areas and sources to the individual enterprises and calculate aggregated values for each multinational.

In a second step, we partition this dataset into homogeneous and distinct groups. Rather than imposing prior knowledge about the number of such groups or their features, we want to “let the data speak” and suggest meaningful subsets of the dataset. We do this by breaking the dataset into clusters using the “partitioning around medoids” algorithm, which is a rather robust and well-designed method for clustering mixed datasets (see annex). ⁴¹

The resulting dataset contains more than 2,500 multinationals consisting of 3.6 enterprises on average. The clustering algorithm allows us to meaningfully identify eight types of multinationals, with the key grouping factors being foreign or Austrian control; special-purpose entity (SPE) versus other company forms; the share of outward investment in Central, Eastern or Southeastern Europe (CESEE); and the degree of trade openness.

The remainder of the paper is structured as follows. In section 1, we describe how we set up the dataset and which data sources we used. In section 2, we explain the clustering algorithm and our choice of key input parameters. Following analysis of the results in section 3, we interpret the results and suggest further avenues of research in section 4. Section 5 concludes.

1 Data sources

1.1 Identifying multinationals in Austria

• any domestic companies with at least one direct investment relationship to ­another economy; or
• any groups of domestic companies controlled by a domestic group head, of which at least one unit maintains an FDI relationship with another economy.

1.2 Using microdata-linking to enrich the database

• Annual FDI survey
• Business register
• Structural business statistics
• Foreign trade statistics
• Services according to the balance of payments
• External statistics compilation system

2 Data clustering with partitioning-around-medoids (PAM) algorithm

2.1 Selection and weighting of input variables

1. The starting point are a set of k random observations in the dataset. These observations, called medoids, represent centers of k clusters which, at this point, consist of single observations.

2. All observations are (re-)assigned to their closest medoid.

3. In each of the clusters thus built, the algorithm finds the observation that would yield the lowest average distance to its cluster members. If this is a different observation than the one in step 1, this observation becomes the new medoid.

4. If at least one out of k medoids has changed, the algorithm goes back to step 2; otherwise the process ends.

Hence, it is crucial to strike a balance between including all major variables that are of interest in clustering the data, and not choosing too many ⁴⁹ and/or highly correlated variables. Table A1 in the annex displays and describes the 20 variables that we carefully selected as input for the PAM algorithm. While the ­focus lay on numerical characteristics, three binary attributes were assessed by ­expert judgment to be crucial for grouping multinationals, namely “SPE” (special purpose entity), “FOREIGN_CONTROL” (multinationals controlled by non­residents) and “BANK” (one of the units is classified as a bank). In other words, we created a mixed-type dataset consisting of numerical and categorical data. In addition, while not being part of the clustering procedure, other nominal scaled attributes were important for the ensuing analysis and interpretation of results. In particular, the variables economic activity ⁵⁰ and controlling region ⁵¹ were of high explanatory value.

In weighting the variables, we basically followed the concept of equal weights, assigning each variable a weight of 100% divided by the number of variables. ­Exceptions were made only for the numerous balance of payments/international investment position variables ⁵² because of their high correlation to each other (see chart 7 in the annex), and the fact that they cover similar aspects (external funding). To avoid the excessive influence issues described above, these attributes were ­assigned only one-quarter of the weight other variables have.

2.2 Choice of key input parameters

Regarding the number of clusters k to be defined, a large number of methods and indices has been proposed in the literature for identifying the optimal number of clusters (Mirkin, 2011). Yet, most of these indices and evaluation methods are not applicable to mixed data. For this reason, we picked one of the few indicators available also for mixed data, the popular silhouette plot, which indicates the ­so-called silhouette width for a given number of clusters. The silhouette width is a normalized ratio between the average dissimilarity within clusters relative to the nearest neighboring cluster. It is normalized to a range between –1 and 1, with values closer to 1 suggesting good clustering. The purpose of the silhouette plot is to find the relative maximum silhouette width for a reasonable range of possible numbers of k. ⁵³ In our case, the silhouette plot shown in chart 2 suggested either eight or – with an even slightly better value – twelve as the optimal number of clusters. ⁵⁴ While we had a close look at both suggestions, we considered eight ­clusters to be the more reasonable choice, since interpreting and comparing four more clusters bears the risk of losing focus.

3 Results

3.1 Clustering multinationals in Austria

On initial eyeballing, we see one major observation that we can take away from chart 3: as the resolution level k increases, the typical pattern is that a cluster splits up into two. The number of elements moving to the newly built groups from other clusters is fairly limited; or put differently, the remaining clusters stay relatively stable as k rises. This implies that the clustering algorithm is rather robust; otherwise we would see a lot more reshuffling among clusters at each k.

While it has to be borne in mind that the PAM algorithm does not ­proceed in the iterative, hierarchical way suggested by the clustering tree in chart 3, this tree may, nonetheless, be interpreted as a dynamic decision tree. At each node, we can identify the variable(s) that cause(s) a cluster to split off. How do we do that? At each splitting node, we compute for each variable the standardized difference between the average values of the respective variable in the two subsequent clusters:

See table 1 for the results and the table rows for the key variables triggering the split, highlighted with the darkest color. For example, when we look at table 1 in combination with chart 3, we see that at clustering level 2 (i.e. k = 2), for the split into the two resulting clusters 1 and 2, it is the variable FOREIGN_CONTROL that makes the biggest difference. At the next level (k = 3), cluster 1 splits up into cluster 1 and 2, with the split essentially being driven by the variable EXP_QUOTA (export of goods as a share of turnover). Next (k = 4), variable IMP_QUOTA prompts the division of cluster 2 into clusters 2 and 4. At k = 5 a subset of enterprises with significant foreign investment activities in CESEE (variable CESEE_SHARE_ODI) spins off from the cluster of Austrian-controlled multinationals. The export share and CESEE investment focus are the main clustering drivers also at the next two levels. Finally, at k = 8, the variable SPE triggers the split of cluster 2 into clusters 2 and 8.

Table 1: Standardized mean value difference between new clusters at subsequent clustering levels

k	Cx	Cy	AW	AX	AZ	BA	BD	O	AU	BC	BB	C	D	G	H	P	Y	Z	AA	AB
∆ SPE	∆ FOREIGN_CONTROL	∆ CESEE_SHARE_ODI	∆ CESEE_SHARE_IDI	∆ BANK	∆ EMP	∆ AGE	∆ IMP_QUOTA	∆ EXP_QUOTA	∆ ODI	∆ IDI	∆ OI_A	∆ OI_L	∆ TURN	∆ FININS_EXP	∆ FININS_IMP	∆ TECH_EXP	∆ TECH_IMP
2	1	2	0.2	2.3	0.7	0.1	0.1	0.2	0.3	0.4	0.0	0.1	0.1	0.1	0.1	0.2	0.1	0.0	0.0	0.1
3	1	2	0.3	0.2	0.1	0.1	0.2	0.1	0.2	1.0	1.9	0.1	0.1	0.0	0.0	0.1	0.0	0.0	0.3	0.2
4	2	4	0.4	0.0	0.0	0.2	0.2	0.0	0.4	1.9	0.3	0.1	0.2	0.1	0.1	0.0	0.1	0.1	0.0	0.1
5	3	5	0.0	0.2	2.7	0.1	0.2	0.2	0.5	0.3	0.1	0.0	0.0	0.3	0.2	0.3	0.0	0.0	0.0	0.0
6	4	5	0.0	0.0	0.5	0.2	0.2	0.3	0.4	1.2	2.0	0.1	0.0	0.0	0.0	0.5	0.2	0.0	0.2	0.1
7	1	7	0.1	0.0	2.9	0.1	0.3	0.4	0.4	0.4	1.5	0.1	0.1	0.0	0.1	0.2	0.0	0.1	0.0	0.0
8	2	8	7.1	0.0	0.1	0.3	0.2	0.1	0.9	0.2	0.1	0.8	0.7	0.1	0.1	0.1	0.1	0.1	0.0	0.1
9	1	3	0.0	0.0	0.1	0.0	0.0	0.0	0.2	0.8	1.2	0.0	0.1	0.0	0.0	0.1	0.0	0.2	0.0	0.1
10	2	9	0.0	0.1	0.1	6.0	0.1	0.1	0.5	0.3	0.2	0.4	0.5	0.0	0.1	0.0	0.1	0.0	0.1	0.1
11	1	4	0.0	0.0	0.2	0.2	0.0	0.1	0.1	1.9	0.2	0.0	0.0	0.0	0.0	0.2	0.0	0.0	0.0	0.1
12	6	10	0.0	0.0	2.1	0.0	0.0	0.1	0.4	0.2	0.4	0.1	0.0	0.0	0.1	0.2	0.1	0.0	0.1	0.0
Source: Authors’ calculations.

Chart 4 summarizes the key driving variables at each node by translating the clustering tree into a decision tree where the “branches” represent crucial dimensions for characterizing the different multinationals groups while the “leaves” visualize the clusters. Of the eight identified clusters, three are controlled by domestic companies while five are dominated by entities outside Austria. Due to their key characteristics, which will become more apparent in the detailed analysis below, we label the eight clusters as follows: (C1) Austrian links in global value chains, (C2) small foreign-controlled services providers, (C3) Austrian ­nonmanufacturers, (C4) Austrian export champions, (C5) retail multinationals, (C6) Austrian (-controlled) CESEE experts in contrast to (C7) (foreign-controlled) CESEE hubs and (C8) SPEs. The decision tree shows that the most relevant ­variables for the partitioning algorithm were SPE, FOREIGN_CONTROL, ­CESEE_SHARE_ODI, IMP_QUOTA and EXP_QUOTA.

By way of example, we can demonstrate the important role that the variables IMP_QUOTA and EXP_QUOTA played for cluster-building, yet from another perspective. The scatter plot (chart 5), which plots import against export shares for the eight clusters, shows a clear concentration of multinationals in the upper half of the quadrant (i.e. high EXP_QUOTA) for the clusters labeled “Austrian links in global value chains” (C1) and “Austrian export champions” (C4). In line with the decision tree, the only definite concentration of high IMPORT_QUOTA values was calculated for “retail multinationals” (C5). The multinational groups “small foreign-controlled service providers” (C2), “Austrian nonmanufacturers” (C3) and “SPEs” (C8) show small or no values for IMP_QUOTA and EXP_QUOTA. The clusters “Austrian CESEE experts” (C6) and “CESEE hubs” (C7) do not exhibit a clear pattern of distribution. The reason for this random distribution is that the multinationals clustered into these groups are to a high degree defined by other variables (especially FOREIGN_CONTROL and CESEE_SHARE_ODI).

To deepen the picture of how the clusters of multinationals differ from each other, we look at the main underlying economic activity ⁵⁶ (chart 6) and find that the identified clusters are rather homogeneous with respect to the industry breakdown (as represented by MNE_NACE, which was no input variable for the cluster analysis). Two clusters stand out with manufacturing as the predominant economic activity: “Austrian links in global value chains” (59%) and “Austrian export champions” (73%). The other economic sectors of these two clusters show a similar distribution, the main distinguishing feature being foreign vs. domestic control. In the “trade multinationals” cluster, 80% of the group members are classified as trade companies. In the “SPEs” cluster, activities like manufacturing and trade are absent almost by definition (“up to five employees” in the IMF’s definition of SPEs), leaving the vast majority of SPEs to be classified in the service sector or as financial companies. The other clusters show no clear indication of a dominating economic activity within a cluster.

3.2 Cluster results in detail

3.2.1 “Austrian links in global value chains”

While the median number of employees (113) is neither exceptionally high nor low compared to other clusters, the median age (23) indicates a comparatively young set of multinationals. The median turnover (EUR 38 million) is also in the mid range of the total population. These figures are well in line with the large share of medium-sized ⁵⁷ enterprises (58%) in this cluster. Other distinguishing features include the comparatively high share of external trade in services (service exports: 75%, service imports: 72%). The dominant economic activity is ­manufacturing (59%), followed by trade (18%) and services (14%). SPEs aside, this is the most globalized group in terms of the location of ultimate investors: 26% of them are located outside Europe, with the Americas contributing strongly with 16%.

The homogeneity of clusters measured by intra-cluster variance is in the mid range compared to other groups, ⁵⁸ with intra-cluster variance being rather high for import share and turnover in particular. At the same time, employment and ­outward FDI are homogenous. Cluster stability is low, as more detailed resolutions of the clustering tree (see chart 3) show two additional upcoming splits of this group until resolution 12 (k = 2).

3.2.2 “Small foreign-controlled service providers”

The cluster’s median turnover is among the lowest (EUR 7 million), as is the median number of employees (17). Only a small fraction of the companies grouped into this cluster trade in services, and the average share of imports (5%) and ­exports (4%) is very low. Essentially, this group of foreign-controlled multinationals in Austria serves the domestic market for services (e.g. hotel industry, catering, car rentals, transport services, financial services), goods (a wide range of industries, e.g. food, office equipment, opticians, petrol stations) and professional business services (holding companies). A close “neighboring” group is the “SPEs” cluster, which exists as a distinct group only from resolution k = 8 downward (see chart 3 for details). The main regions of origin of FDI investors in this cluster are Western Europe (74%), particularly Germany (36% of all cases). This group also contains the highest concentration of CESEE investors, with a small subgroup of the CESEE-­controlled multinationals branching off at resolution 10 (k = 10). Thus, cluster ­stability is neither high nor low.

3.2.3 “Austrian nonmanufacturers”

The multinationals in this cluster are mainly engaged in services (31%) and professional business services (in the form of holding companies; 26%). More than half of all private individuals included in the population were clustered into this group, ⁶⁰ which is one explanation for the low median number of employees (9), low turnover (EUR 4 million) and low degree of foreign trade activity (export share: 4%, import share: 5%). More than half (58%) of the companies in this ­cluster qualify as small multinationals, but there are also some widely known larger enterprises, such as SPAR (HOLDAG Bet. GmbH), FLUGHAFEN WIEN, PORR and RAIFFEISEN-HOLDING NÖ/W. Essentially, this cluster contains Austrian enterprises in the nonmanufacturing business and individuals engaged in outward FDI without a specific CESEE focus.

3.2.4 “Austrian export champions”

Many of the multinationals in this cluster are household names in Austria, e.g. OMV, ANDRITZ, KTM, VERBUND, VOESTALPINE, LENZING and RED BULL. With a comparatively low CESEE outward FDI ratio (20%), investment targets and markets are spread globally. Another explanation for this rather low value could be that large multinationals with a very strong CESEE focus were ­clustered into the group of “Austrian CESEE experts.” While homogeneity is low because of outliers in many variables, stability is high (this cluster remains broadly unchanged until cluster resolution k = 12).

3.2.5 “Retail multinationals”

The main characteristics are a high share of imports (61%) combined with a low share of exports (11%). Most companies are neither small nor large, and ­homogeneity within the group is high. The median number of employees is 52, median turnover is EUR 28 million. Typical representatives of this cluster are companies like H&M, IBM, MAN, EDUSCHO, DEICHMANN, MIELE, NEW YORKER or ZARA. Typically, they do not hold outward FDI; this is the case only for 27 out of 472 companies in this cluster. The group heads of these multinationals are overwhelmingly located in Western Europe (80%; Germany: 42%).

3.2.6 “Austrian CESEE experts”

In this cluster, 94% of all outward FDI is invested in CESEE countries on ­average. The industry mix is highly balanced, led by professional business services (i.e. holding companies; 24%) and followed by trade (19%), services (18%) and manufacturing (18%). This cluster stands out with regard to the variables “BANK” (7%) and “capital market participation” (25%), which are likely to be correlated since all banks in this cluster are active on international financial markets. With a median age of 35 years, this cluster contains the most mature of all companies. Moreover, the cluster is highly stable; the partitioning algorithm would build ­almost identical clusters when forced to build nine, ten or eleven groups. Only at k = 10 would some companies (mainly the manufacturing companies) break off to form a new cluster.

3.2.7 “CESEE hubs”

Household names in this cluster are REWE, TELEKOM AUSTRIA, SIEMENS and ALLIANZ. Homogeneity is also comparatively low given the broad mix of companies (companies of all sizes and industries) as in the “neighboring” cluster 6 above. Stability is high, with no further split occurring at least until k = 12.

3.2.8 “SPEs”

“An SPE resident in an economy, is a formally registered and/or incorporated legal entity ­recognized as an institutional unit, with no or little employment up to maximum of five ­employees, no or little physical presence, and no or little physical production in the host ­economy.” ⁶²

Additional characteristics include foreign control and an exceptionally high ­degree of cross-border assets and liabilities.

As was to be expected, this cluster is highly homogeneous, with 51 out of the 52 multinationals in the dataset marked as SPE having been clustered into this group. The median number of employees in this cluster is zero, as is turnover and foreign trade. The PAM algorithm conducts a “SPE split” at k = 12. The closest “neighboring” group is “small foreign-controlled service providers,” which builds a common cluster with SPEs at k = 7.

4 Interpretation, conclusion and further research

The analysis delivers eight meaningfully interpretable groups of multinationals, three of which are Austrian-controlled, with the other five clusters being in ­foreign hands. There is a significant difference in complexity between these two segments. The Austrian-controlled units are characterized by a high degree of stability from a relatively early partitioning stage (k6), with no further splits occurring until k = 12, except for the branching-off of a CESEE group containing larger, export-­orientated manufacturing companies. As the foreign-controlled units are more heterogeneous, they tend to be subject to splits between k = 7 and k = 11. Looking ahead, further experimentation in input parameters could verify the stability of our results at these cluster resolutions.

Additionally, one must bear in mind that we conducted a one-off exercise based on 2017 data only. An obvious next step could be to perform a similar analysis with 2018 data or time series with historic data to be able to assess the stability of the clusters are over time. Furthermore, we did not investigate all available variables, and there might be additional relevant data sources that could be linked to the ­multinationals database (e.g. R&D expenditure). Another rewarding question could be a more detailed investigation of “neighboring” clusters, e.g. “Austrian ­CESEE experts” and “CESEE hubs,” to be able to establish the effect foreign ­control has had over time compared with domestic control. Likewise, intra-cluster consistency could be the subject of further research.

Finally, future research may look at the impact of various policy measures ranging from taxes to labor market policies for different types of multinationals, which would then allow for more effective and efficient policymaking. Last but not least, once the proposed taxonomy of environmentally sustainable activities has been completed, another avenue of future research might focus on identifying the characteristics and drivers of multinationals’ “green” activities.

References

Budiaji, W. and F. Leisch. 2019. Simple K-medoids partitioning algorithm for mixed variable data. In: Algorithms. 12(9). 177.

Baser, P. and J. Saini. 2013. A comparative analysis of various clustering techniques used for very large datasets. In: International Journal of Computer Science and Communication Networks. 3(5). 271–275.

De Backer, K., S. Miroudot and D. Rigo. 2019. Multinational enterprises in the global ­economy: Heavily discussed, hardly measured. https://voxeu.org/article/multinational-enterprises-­global-economy.

Fraiman, R., A. Justel and M. Svarc. 2009. Selection of variables for cluster analysis and ­classification rules. In: Journal of the American Statistical Association. 1294–1303.

Gower, J. C. 1971. A general coefficient of similarity and some of its properties. In: Biometrics 27. 857–874.

Jain, A. K. and R. C. Dubes. 1988. Algorithms for clustering data. Prentice-Hall.

Jin X. and J. Han. 2017. K-Medoids Clustering. In: Sammut C. and G. I. Webb (eds.). Encyclopedia of Machine Learning and Data Mining. Springer.

Kaufman, L. and P. J. Rousseeuw. 1987. Clustering by means of medoids. In: Dodge, Y. (ed.). Statistical Data Analysis Based on the L1-Norm and Related Methods. 405–416.

Kim, I. S. and H. V. Milner. 2019. Multinational Corporations and their Influence Through ­Lobbying on Foreign Policy. Princeton University Working Paper.

Mirkin, B. 2011. Choosing the number of clusters. In: WIREs Data Mining Knowledge Discovery 1(3). 252–260.

Nguyen, T. T., D. Dinh., S. Sriboonchitta and V. Huynh. 2019. A method for k-means-like clustering of categorical data. In: Journal of Ambient Intelligence and Humanized Computing.

OECD. 2015. Measuring International Investment by Multinational Enterprises. Implementation of the OECD’s Benchmark Definition of Foreign Direct Investment, 4th edition. www.oecd.org/corporate/FDI-BMD4-brochure.pdf

OECD. 2018. Multinational enterprises in the global economy. Heavily debated but hardly ­measured. www.oecd.org/industry/ind/MNEs-in-the-global-economy-policy-note.pdf

Rousseeuw, P. J. 1987. Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. In: Journal of Computational and Applied Mathematics 20. 53–65.

Sambandam, R. 2003. Cluster analysis gets complicated. In: Marketing Research 15(1).

Tryfos, P. 1998. Methods for business analysis and forecasting: Text & Cases, Chapter 15. Wiley.

Annex

Cluster analysis

Similarly, cluster analysis can be performed with a plethora of different ­methods classifiable across multiple dimensions (see e.g. Baser and Saini, 2013). In general, different clustering methods lead to different outcomes. Cluster analysis is an ­explorative technique, meaning that the best approach is highly subjective, and it should be one that is practical to handle and that delivers the “best” – i.e. meaningful, useful and well interpretable – results for the analyst.

Having said that, important decisions need to be made with respect to the ­clustering method and its parameters in the course of the data exploration process. One of the choices to be made is between a hierarchical or a partitional approach. The hierarchical approach is based on a nested sequence, proceeding either ­bottom-up – i.e. starting from as many clusters as there are observations and ­finishing with a single cluster comprising all observations (agglomerative approach) – or vice versa (divisive approach) (see e.g. Jain and Dubes, 1988).

In contrast to hierarchical methods, partitional clustering algorithms generate single partitions of the data into mutually disjoint subsets the number of which – conventionally denoted as k – needs to be specified by the researcher ex ante. ­Essentially, these algorithms first assign each data point to one of k clusters and then reshuffle the observations across clusters until each observation has the smallest distance to the center of the cluster. The methods differ, inter alia, in the way the “center” is defined and in the distance metric.

In general, hierarchical clustering methods are rather useful for smaller datasets. Moreover, different parameter specifications tend to produce very different outcomes, as was the case with our dataset. Having experimented with various ­hierarchical clustering methods, we ultimately opted for a partitional clustering method.

A very popular partitional approach is the so-called k-means method, well known for its efficiency in clustering large datasets. However, one of its key features is the fact that it uses arithmetic data “means” (so-called centroids) as the center of the clusters. The upshot is that this method typically works only on numerical variables. ⁶³ In addition, its results are sensitive to outliers and noise in the data (Budiaji and Leisch, 2019). Since our dataset is a mixture of numerical and categorical variables and contains a number of outliers, the k-means method was not an option. The most common alternative for mixed-variable datasets is the “partitioning around medoids” (PAM) algorithm. It can be considered a more robust and universal ­algorithm than the k-means, not only because it can handle mixed data but also because it is less sensitive to outliers (Jin and Han, 2017). Rather than using ­“centroids,” this method uses “medoids,” which are not computed statistical means but actual data points from the dataset representative of each cluster.

It follows from the description of the PAM algorithm in section 2 that one of the key inputs for the algorithm is some distance metric. If the dataset contains purely numerical variables, different distance measures can be applied directly to the raw dataset just as with k-means. However, in case of mixed data, the distance between observations needs to be computed beforehand and provided as input to the algorithm as a distance matrix. A common option to compute distances for mixed data sets is the Gower-distance matrix (Gower, 1971). It uses an appropriate distance metric for each variable type, i.e. Manhattan for continuous and Dice for categorical datapoints, which is subsequently scaled to fall between 0 and 1. Then, a linear combination using user-specified weights is calculated to create the final distance matrix.

Table A1: Variables and weights for PAM clustering

Variable	Name	Type	Values	Description	Weight for clustering in %
SPE	Special purpose entity	Nominal	1 = “yes”0 = “no”	One or more units are classified as a special ­purpose entity	8
FOREIGN_CONTROL	Foreign control	Nominal	1 = “yes”0 = “no”	One or more units are controlled by a ­nonresident	8
CESEE_SHARE_ODI	CESEE share of outward FDI	Interval	0-100%	Outward FDI in CESEE as a share of total ­outward FDI	8
CESEE_SHARE_IDI	CESEE share of inward FDI	Interval	0-100%	Inward FDI in CESEE as a share of total inward FDI	8
BANK	Banking license	Nominal	1 = “yes”0 = “no”	One unit is classified in ESA sector 1220A	8
EMP	Employees	Interval	ℕ	Total number of employees (all units)	8
AGE	Age	Ordinal	ℕ	Age of the oldest unit	8
IMP_QUOTA	Import share	Interval	0-100%	Import of goods divided by turnover	8
EXP_QUOTA	Export share	Interval	0-100%	Export of goods divided by turnover	8
ODI	Outward FDI	Interval	ℤ	Outward FDI (extended direction principle)	2
IDI	Inward FDI	Interval	ℤ	Inward FDI (extended direction principle)	2
OI_A	Other investment assets	Interval	ℕ	Other investment assets (BOP/IIP concept)	2
OI_L	Other investment liabilities	Interval	ℕ	Other investment liabilities (BOP/IIP concept)	2
TURN	Turnover	Interval	ℕ	Turnover as reported in structural business ­statistics	8
FININS_EXP	Insurance and financial services ­exports	Interval	ℕ	Insurance and financial services exports as ­reported for ITSS	2
FININS_IMP	Insurance and financial services ­imports	Interval	ℕ	Insurance and financial services imports as ­reported for ITSS	2
TECH_EXP	Technical services exports	Interval	ℕ	Technical services exports as reported for ITSS	2
TECH_IMP	Technical services imports	Interval	ℕ	Technical services imports as reported for ITSS	2
OtherS_EXP	Other services exports	Interval	ℕ	Other services exports as reported for ITSS	2
OtherS_IMP	Other services imports	Interval	ℕ	Other services imports as reported for ITSS	2
Source: OeNB, Statistics Austria, Authors’ calculations.
Note: ℕ = positive integers; ℤ = positive or negative integers; BOP/IIP = balance of payments/international investment position; ITSS = international trade in services statistics.

A CESEE conundrum: low trust in government but high hopes for government-led job creation

Markus Eller, Thomas Scheiber ⁶⁴

OeNB Euro Survey results for ten countries in Central, Eastern and Southeastern Europe ­(CESEE) from 2018 indicate that a major share of respondents is disappointed with public governance. Yet, while trust in national governments is lacking, there is still a widespread belief that creating jobs is primarily a state responsibility, even 30 years after the onset of transition. As shown by a series of probit regressions, respondents are more likely to consider job creation to be above all a state responsibility if they belong to a low-income household, have ­comparatively little wealth and comparatively little education, rely on welfare payments, have worked for the public sector or reside outside the capital city. The views of respondents who express a lack of trust in government are also colored strongly by past economic hardship ­experiences. While there is, of course, a limit to how big the welfare state can get, our survey results imply that there is a case for national governments to build up buffers to be able to tide people over when incomes dry up in crisis episodes, and to invest more in developing ­human capital and improving social inclusion to address the concerns of marginalized societal groups.

JEL classification: A13, H11, P35

Keywords: public preferences, trust in government, government-led job creation, survey data, CESEE

This short study provides the relevant stylized facts (section 1) and sheds light on the characteristics of individuals expressing such views (section 2). With that we try to understand better the origins of subjective attitudes toward politics in the CESEE region. On the back of a large body of political science literature that provided evidence for public attitudes impacting public policy (e.g. Burstein, 2003; Hager and Hilbig, 2020; Hobolt and Klemmemsen, 2005; Manza et al., 2002), this knowledge should be useful for the design of targeted policies that enhance people’s trust in, and support for, the respective institutional frameworks – not in the sense that people are manipulated for this purpose, but in the sense that their respective socioeconomic conditions are adequately accounted for. This could ultimately ­allow for more stable political and economic outcomes in the region (EBRD, 2019; IMF, 2016).

1 Distrusting government but counting on government for job creation – stylized facts

Consistency between the scores of the Corruption Perception Index (CPI) published by Transparency International ­in 2018 and the OeNB Euro Survey ­responses on vested interests confirms the validity of our survey data (see chart 2). Countries with lower CPI scores (indicating higher corruption) coincide with higher shares of agreement that politicians of the respective country only serve vested interests. Albania, Hungary and Serbia seem to be an exception; their respondents less frequently report a prevalence of vested interests than countries with a similar CPI score.

In a second step, the survey asked respondents whether job creation should primarily be a state responsibility or a private sector responsibility (or a shared responsibility; for detailed country-specific answers, see also chart A1 in the annex). Despite the underlying concerns about public governance, demand for state intervention in the economy is nonetheless very substantial, with about 45% of respondents in the full sample arguing that job creation should primarily be a responsibility of the state (see lower panel in chart 1). Support for this role is largest in Bosnia and Herzegovina and North Macedonia with shares of about 60%, followed by Albania and Romania with shares of about 53%. The overall share of respondents who do not trust government but primarily look to the state for job ­creation (blue bars in the lower panel of chart 1) is about 20% in the full sample, with Romania, Bosnia and Herzegovina, and North Macedonia showing the largest shares (around 30%). Even though the share of respondents thus holding seemingly contradictory beliefs is, at first glance, not extraordinarily high, it should be ­emphasized that respondents who express distrust in government often make up a significant share of those who expect jobs to be created above all by government (e.g. more than half in the SEE EU Member States and Bosnia and Herzegovina).

Existing literature points to a paradoxical relationship between distrust in ­public institutions and demand for more state regulation (e.g. Aghion et al., 2010), where the main trigger comes from private businesses imposing negative externalities on society in an environment with poor social capital and the society apparently choosing to demand more state regulation and tolerating occasional corruption to reduce these externalities. However, based on our sample we cannot confirm a statistically significant conditional correlation between respondents’ demand for government-led job creation and distrust in government (unlike Denisova et al., 2010, for Russia). In any case, respondents holding both beliefs are characterized by a mixed socioeconomic profile that differs from other groups in the sample (see the next section).

2 Who are the people that do not trust government but look to the state for job creation first?

Selected marginal effects are reported in chart 3 for the full sample, while the complete set of regression results – also emphasizing differences between country groups – is shown in table A2 in the annex. ⁷¹ We find that distrust in government (upper left-hand panel in chart 3) and the perception of vested interests (upper right-hand panel), short-term rent-seeking behavior (middle left-hand panel) and the waste of taxpayer money (middle right-hand panel) are very robustly associated with economic hardship undergone in the recent past: Respondents who had to cut back on basic consumption for some time in the period from 2008 to 2018 are more likely to have very skeptical views of public governance. Conversely, receipt of social transfers has a mitigating effect on all four dependent variables: Respondents who received unemployment benefits for some time in the period from 2008 to 2018 are less likely to voice concerns about politicians’ behavior, and respondents for whom welfare benefits are an important source of income are less likely to voice concerns about vested interests and waste of taxpayer money. Moreover, ownership of assets – in the form of a secondary ­residence or other real estate
in addition to a main residence – is associated with less pronounced distrust
and ­skepticism. Interestingly, other economically important variables such as ­educational attainment, employment status (except for self-employment) or net household income do not appear to matter in this context. ⁷² In addition, and as ­expected, respondents who would vote for a ruling political party, who are ­interested in politics or who report a high degree of life satisfaction are significantly less skeptical about public governance. Furthermore, respondents’ attitudes vary systematically with age, gender and religion (see table A2). The qualitatively similar results do not come as a surprise, given that these four variables capture similar dimensions of people’s attitudes toward public governance.

Our findings for the respondents who see job creation primarily as a state ­responsibility are quite different (lower left-hand panel in chart 3): in this case, responses show more differentiation in line with individuals’ current economic situation – economic “fundamentals” matter much more. Those who are better off or less marginalized (i.e. those who have a higher level of education, belong to the upper income tercile, hold more assets – either in the form of a secondary ­residence or more savings –, have to rely less on public welfare benefits or live in the capital city) are less likely to look to the state for job creation first. The aggregate picture presented in chart 3 conceals some interesting cross-regional heterogeneity. For instance, as can be seen in table A2, being better off in terms of higher household income or owning a secondary residence makes respondents less likely to expect government-led job creation only in the three CEE EU Member States. Moreover, the regressions explaining preferences for looking to the state for job creation first are the only ones where no explanatory variable is statistically significant across all the three country groups, pointing to stronger cross-country differences in this case.

Finally, when focusing on the last dependent variable capturing respondents who hold both seemingly contradictory beliefs, i.e. express distrust in the government and expect governments to take the initiative in providing jobs (lower right-hand panel in chart 3), it becomes evident that they share a mixture of characteristics: in addition to the socioeconomic factors that were already important for respondents who voiced distrust in government, marginalization effects play an important role as well. In particular, respondents are more likely to support both beliefs if they have experienced economic hardship, have not received unemployment benefits, have a low level of household income, do not have a secondary residence or live outside the capital city.

3 Key takeaways: adequate institution-building and social inclusion matter

Social inclusion also plays an important role according to our results. Those who are less marginalized (i.e. those who have comparatively higher levels of ­education, belong to the upper income tercile, possess comparatively more assets, do not rely heavily on public welfare benefits or live in the capital city) do not see job creation primarily as a state responsibility. Rather than expecting the state to intervene strongly in the job market to address the concerns of marginalized societal groups, which could be very expensive and would add to fiscal constraints, these respondents support a more active role for the state in investing in human capital and improving social inclusion by enhancing (access to) education and making ­targeted social transfers to the poor. The funds to pay for that will most likely have to come from longer-term-oriented budget policies.

However, given that populists, of both right- and left-wing varieties, could take advantage of distrust of government and people’s worries about market turmoil fallout, there is a danger that short-term rent-seeking behavior gains further prevalence. One answer to these challenges consists in adequate institution-building. Sound institutions are crucial for providing the analytical capacities for policy ­design and implementation and promoting a conscientious and prudent management of tax revenues. Independent audit offices and fiscal councils that recently have been set up in several CESEE countries (or whose establishment has been debated) could prove to play essential roles in achieving these objectives and thus fostering trust and promoting satisfaction with public governance (OECD, 2017). ⁷³ Moreover, appropriate fiscal rules could be helpful in bringing reelection-minded ­incumbents away from a solely short-run, election-oriented budget policy behavior ⁷⁴ to a longer-run, growth-promoting orientation (Rose, 2006).

References

Aghion, P., Y. Algan, P. Cahuc and A. Shleifer. 2010. Regulation and Distrust. In: The ­Quarterly Journal of Economics 125(3). 1015–1049.

Begovic, S., N. Adnett and G. Pugh. 2016. An investigation into the credibility of currency board arrangements in Bosnia and Herzegovina and Bulgaria. In: Journal of Comparative ­Economics 44(3). 787–799.

Belabed, C. A. and M. Hake. 2018. Income inequality and trust in national governments in Central, Eastern and Southeastern Europe. OeNB Working Paper 222. May.

Brender, A. and A. Drazen. 2005. Political Budget Cycles in New Versus Established Democracies. In: Journal of Monetary Economics 52(5). 1271–1295.

Burstein, P. 2003. The Impact of Public Opinion on Public Policy: A Review and an Agenda.
In: Political Research Quarterly 56(1). 29–44.

Denisova, I., M. Eller, T. Frye and E. Zhuravskaya. 2009. Who Wants To Revise Privatization? The Complementarity of Market Skills and Institutions. In: American Political Science ­Review 103(2). 284–304.

Denisova, I., M. Eller and E. Zhuravskaya. 2010. What do Russians think about transition?
In: Economics of Transition 18(2). 249–280.

EBRD. 2006. Life in Transition I – A survey of people’s experiences and attitudes.

EBRD. 2016. Life in Transition III – A decade of measuring transition.

EBRD. 2019. Better Governance, Better Economies. Transition Report 2019-20.

ECB. 2020. Convergence Report. June.

Hager, A. and H. Hilbig. 2020. Does Public Opinion Affect Political Speech? In: American ­Journal of Political Science. DOI: 10.1111/ajps.12516.

Hayo, B. and F. Neumeier. 2019. Public Preferences for Government Spending Priorities: ­Survey Evidence from Germany. In: German Economic Review. 20(4), e1-e37

Hobolt, S. B. and R. Klemmemsen. 2005. Responsive Government? Public Opinion and ­Government Policy Preferences in Britain and Denmark. In: Political Studies 53. 379–402.

IMF. 2016. Central, Eastern, and Southeastern Europe – Effective Government for Stronger Growth. Regional Economic Issues. November.

Manza, J., F. L. Cook and B. I. Page. 2002. Navigating Public Opinion – Polls, Policy, and the Future of American Democracy. Oxford University Press.

Nordhaus, W. D. 1975. The Political Business Cycle. In: Review of Economic Studies 42. 169–190.

OECD. 2017. Trust and Public Policy: How Better Governance Can Help Rebuild Public Trust. OECD Public Governance Reviews.

Pitlik, H., G. Schwarz, B. Bechter and B. Brandl. 2011. Near Is My Shirt but Nearer Is My Skin: Ideology or Self-Interest as Determinants of Public Opinion on Fiscal Policy Issues.
In: Kyklos 64(2). 271–290.

Rose, S. 2006. Do Fiscal Rules Dampen the Political Business Cycle? In: Public Choice 128(3/4). 407–431.

Stix, H. 2013. Does the Broad Public Want to Consolidate Public Debt? – The Role of Fairness and of Policy Credibility. In: Kyklos 66(1). 102–129.

Annex

Table A1: Variable description

Label	Description
(1) Perceptions of public governance (dependent variables in the regressions)
Distrust in ­government	Dummy equals 1 if respondents reported to somewhat or completely distrust the central government/cabinet of ­ministers; 0 if they indicated complete/some trust or picked neither trust nor distrust (5-point Likert scale).
Vested interests	Dummy equals 1 if respondents rather/strongly agreed with the statement “most politicians in my country serve the ­interests of particular groups”; 0 if they rather/strongly agreed with the statement “most politicians in my country act in line with the ­general public’s interest” or if they indicated no preference or considered both statements to be ­somewhat true.
Short-term ­rent-seeking	Dummy equals 1 if respondents rather/strongly agreed with the statement “most politicians in my country are ­concerned ­about the next elections”; 0 if they rather/strongly agreed with the statement “most politicians in my ­country are concerned about the country’s long-term well-being” or if they indicated no preference or considered both statements to be somewhat true.
Waste of taxpayer money	Dummy equals 1 if respondents rather/strongly agreed with the statement “the state is wasting taxpayer money”; 0 if they ­rather/strongly agreed with the statement “the state manages tax revenues conscientiously” or if they indicated no preference or ­considered both statements to be somewhat true.
Government-led job creation	Dummy equals 1 if respondents considered the state to be primarily responsible for creating jobs; 0 if they considered job ­creation to be primarily a private sector responsibility or a shared public/private responsibility, or said that it did not matter as long as jobs were available.
Distrust in government and expectations of government-led job creation	Dummy equals 1 if the dummy “distrust in government” equals 1 AND the dummy “government-led job creation” equals 1 for a given respondent; 0 otherwise.
(2) Socioeconomic characteristics
Cut back on ­consumption	Dummy equals 1 if respondents had to cut back on basic consumption in the period from 2008 to 2018; 0 otherwise.
Unemployment ­benefits	Dummy equals 1 if respondents received unemployment benefits in the period from 2008 to 2018; 0 otherwise.
Welfare benefits	Dummy equals 1 if respondents considered public welfare benefits to be very/rather important for their individual (or shared) budgets at the time of asking (with welfare benefits including unemployment compensation, public pension, benefits for families and children, etc.); 0 if they indicated that such benefits were not important or not part of ­household income.
Wage cut	Dummy equals 1 if respondents had to accept wage cuts or delayed wage payments in the period from 2008 to 2018; 0 otherwise.
Employment status	Dummy variables: self-employed, unemployed, retired, student. Base category: employed.
Worked for the ­public sector	Dummy equals 1 if respondents worked for the public sector and received a wage income in the period from 2008 to 2018; 0 otherwise.
Secondary residence	Dummy equals 1 if respondents confirmed that they or a household member owned a secondary residence or other real estate (beyond the main residence); 0 if they said that this was not the case, refused to answer or were not able to provide an answer.
Savings	Dummy equals 1 if respondents reported to have savings; 0 otherwise.
House in poorer/ better condition	Dummy variables take value 1 if interviewers considered a given dwelling to be in a better or poorer condition than the ­neighboring dwellings. Base category: similar condition.
(3) Sociodemographic characteristics
Age	Dummy variables for three age groups: 19 to 34 years, 35 to 54 years (base category) and 55 and more years.
Female	Dummy variable that takes the value 1 if respondents were female, 0 otherwise (base category).
Educational ­attainment	Dummy variables; degree of educational attainment (high, medium, low). Base category: low level of education.
Religion	Dummy variables: Muslim, Orthodox Christian, Catholic Christian, Protestant Christian (and other Christian faith), other faith groups (e.g. Jewish, Buddhist), no answer. Base category: atheist/agnostic.
Interest in politics	Dummy equals 1 if respondents rather/strongly agreed with the statement “I am very interested in politics”; ­0 otherwise.
Life satisfaction	Dummy equals 1 if respondents rather/strongly agreed with the statement “all things considered, I am satisfied with my life now”; 0 otherwise.
Party preference	Respondents were asked which party they would vote for if parliamentary elections were held during the coming week. Dummy variables: would vote for a ruling party, would not vote, were undecided or refused to answer. Base ­category: opposition party.
(4) Household characteristics
Income	Dummy variables for the level of total household income after taxes (high, medium, don’t know/no answer). Base ­category: low income.
Household size	Dummy variables for the number of persons living permanently in a given household (two, three and more individuals). Base category: single person.
Capital city resident	Dummy equals 1 if respondents’ main residence is in the capital city; 0 otherwise.
Source: Authors’ compilation.
Note: Unless otherwise stated, respondents with “don’t know” answers or who refused to answer are excluded.

Probit regression results

Table A2 : Probit estimations: distrust in government and public perception of public governance; in CESEE and per country group

	Distrust in government	Most politicians only serve vested interests
CESEE	CZ/HU/ PL	BG/HR/ RO	AL/BA/ MK/RS	CESEE	CZ/HU/ PL	BG/HR/ RO	AL/BA/ MK/RS
1	2	3	4	5	6	7	8
Experienced wage cuts (2008-2018)	0.016	–0.040	0.040	0.025	–0.003	–0.013	–0.024	0.019
	(0.024)	(0.033)	(0.056)	(0.034)	(0.019)	(0.038)	(0.026)	(0.035)
Had to cut back on consumption (2008-2018)	0.079***	0.085***	0.090***	0.066**	0.117***	0.136***	0.095***	0.120***
	(0.016)	(0.027)	(0.025)	(0.026)	(0.015)	(0.027)	(0.019)	(0.031)
Public welfare benefits are currently ­important	–0.019	–0.038	0.011	–0.022	–0.045**	–0.092***	0.015	–0.063**
	(0.020)	(0.027)	(0.034)	(0.029)	(0.019)	(0.027)	(0.029)	(0.028)
Received unemployment benefits (2008-2018)	–0.055***	–0.001	–0.062*	–0.069**	–0.041**	–0.037	0.002	–0.049
	(0.020)	(0.034)	(0.036)	(0.035)	(0.019)	(0.032)	(0.022)	(0.035)
Public sector work experience (2008-2018)	–0.012	0.011	–0.027	–0.022	–0.009	0.031	–0.036	–0.028
	(0.020)	(0.022)	(0.030)	(0.048)	(0.018)	(0.030)	(0.026)	(0.026)
Would vote for a ruling party	–0.287***	–0.298***	–0.286***	–0.272***	–0.136***	–0.152***	–0.084***	–0.169***
	(0.025)	(0.039)	(0.041)	(0.046)	(0.020)	(0.034)	(0.027)	(0.035)
Would not vote	0.007	0.035	0.037	–0.031	0.046**	0.084**	0.047**	0.015
	(0.022)	(0.040)	(0.029)	(0.044)	(0.022)	(0.041)	(0.021)	(0.047)
Don’t know for whom to vote	–0.079***	–0.095*	–0.056	–0.088	0.043	0.047	0.041	0.012
	(0.029)	(0.052)	(0.041)	(0.056)	(0.027)	(0.039)	(0.034)	(0.044)
Refused to reveal voting preference	–0.124***	–0.051	–0.077	–0.176***	0.003	–0.005	0.038	–0.036
	(0.030)	(0.057)	(0.048)	(0.052)	(0.031)	(0.049)	(0.044)	(0.052)
Aged 19 to 34 years	0.025	–0.027	0.042	0.050*	–0.026**	–0.051***	–0.010	–0.025
	(0.016)	(0.023)	(0.028)	(0.028)	(0.013)	(0.018)	(0.023)	(0.024)
Aged 55+ years	0.002	–0.049**	0.029	0.032	0.038**	0.042	–0.005	0.065***
	(0.017)	(0.023)	(0.027)	(0.027)	(0.015)	(0.029)	(0.025)	(0.022)
Female	–0.008	–0.014	–0.006	–0.014	–0.031***	–0.022	–0.030***	–0.038*
	(0.010)	(0.017)	(0.020)	(0.013)	(0.011)	(0.023)	(0.011)	(0.023)
Educational attainment: high	0.019	0.044	0.101*	–0.034	–0.023	0.013	–0.022	–0.052
	(0.029)	(0.045)	(0.056)	(0.05)	(0.027)	(0.047)	(0.039)	(0.045)
Educational attainment: medium	0.023	0.042	0.100*	–0.013	–0.023	0.017	–0.045	–0.028
	(0.023)	(0.039)	(0.054)	(0.034)	(0.019)	(0.039)	(0.041)	(0.027)
Self-employed	–0.051**	–0.005	–0.071**	–0.102***	0.018	0.039	–0.023	0.015
	(0.020)	(0.031)	(0.032)	(0.034)	(0.023)	(0.040)	(0.035)	(0.033)
Unemployed	–0.006	–0.009	–0.015	–0.012	0.009	0.028	–0.072**	0.035
	(0.020)	(0.042)	(0.032)	(0.028)	(0.023)	(0.043)	(0.031)	(0.029)
Retired	–0.015	0.057*	–0.037	–0.057*	–0.004	0.049	–0.026	–0.029
	(0.022)	(0.032)	(0.035)	(0.034)	(0.023)	(0.040)	(0.034)	(0.038)
Student	–0.052**	–0.006	–0.049	–0.090***	0.005	0.014	–0.017	0.021
	(0.023)	(0.041)	(0.044)	(0.032)	(0.027)	(0.049)	(0.040)	(0.049)
Muslim	–0.009	0.000	0.037	–0.350***	0.048	0.000	0.082	0.050
	(0.051)	(.)	(0.058)	(0.073)	(0.052)	(.)	(0.059)	(0.059)
Orthodox Christian	–0.039	–0.014	0.031	–0.374***	0.058	–0.048	0.055	0.040
	(0.042)	(0.079)	(0.058)	(0.075)	(0.066)	(0.083)	(0.061)	(0.063)
Catholic Christian	–0.066**	–0.055**	0.031	–0.500***	0.001	0.038	–0.175**	0.060
	(0.026)	(0.022)	(0.052)	(0.073)	(0.026)	(0.024)	(0.076)	(0.091)
Other Christian faith (e.g. Protestant)	–0.057*	–0.109***	0.130*	–0.294***	–0.019	–0.042	–0.005	0.111
	(0.034)	(0.030)	(0.078)	(0.099)	(0.037)	(0.040)	(0.069)	(0.088)
Source: Authors’ calculations. OeNB Euro Survey 2018.
Note: Average marginal effects from probit estimations with country-fixed effects using data from the OeNB Euro Survey 2018; standard errors are adjusted for clustering at the regional level and reported in parentheses. ***, **, * denote that the average marginal effect is statistically different from zero at the 1%, 5% and 10% level, respectively. For a definition of the variables, see annex table A1. P(DepVar=1) denotes the unconditional sample probability of the respective dependent variable. By construction, a positive (negative) average ­marginal effect implies that a respondent is more (less) likely to agree with the question underlying the dependent variable. Base categories are: preference for the opposition party, 35 to 54 years old, male, low educational attainment, employed, atheist/agnostic, main residence in a similar condition as neighboring homes, no savings, no secondary residence, not living in the capital city, low income, single household, Czech resident. The CESEE group comprises all ten OeNB Euro Survey countries. The second group consists of three Central and Eastern European countries that became EU members in 2004. The third country group consists of three Southeastern European countries that became EU members in 2007 and 2013, respectively. The final group consists of four EU candidates or potential candidates.

Table A2 continued: Probit estimations: distrust in government and public perception of public governance; in CESEE and per country group

	Distrust in government	Most politicians only serve vested interests
CESEE	CZ/HU/ PL	BG/HR/ RO	AL/BA/ MK/RS	CESEE	CZ/HU/ PL	BG/HR/ RO	AL/BA/ MK/RS
1	2	3	4	5	6	7	8
Other faith groups (e.g. Jewish, Buddhist)	–0.012	0.033	–0.158*	–0.064	–0.037	–0.013	–0.105	0.003
	(0.075)	(0.102)	(0.084)	(0.1)	(0.083)	(0.144)	(0.144)	(0.196)
Religion: no answer	–0.023	–0.042	–0.047	–0.161	–0.024	–0.062	0.076*	–0.081
	(0.048)	(0.041)	(0.094)	(0.146)	(0.042)	(0.059)	(0.043)	(0.133)
Very interested in politics	–0.034**	0.011	–0.015	–0.075***	–0.047***	–0.074***	–0.011	–0.059*
	(0.015)	(0.029)	(0.028)	(0.018)	(0.018)	(0.025)	(0.021)	(0.032)
Satisfied with life as it is	–0.136***	–0.130***	–0.131***	–0.137***	–0.014	–0.001	0.011	–0.037
	(0.017)	(0.024)	(0.032)	(0.027)	(0.016)	(0.024)	(0.025)	(0.026)
House in better condition than that of ­neighbors	0.019	0.021	0.013	0.015	–0.021	–0.055**	0.001	–0.015
	(0.013)	(0.026)	(0.027)	(0.016)	(0.015)	(0.022)	(0.025)	(0.025)
House in poorer condition than that of neighbors	0.041*	0.031	0.009	0.056	–0.005	–0.034	–0.006	0.022
	(0.022)	(0.029)	(0.043)	(0.038)	(0.018)	(0.039)	(0.024)	(0.023)
Has secondary residence or other real ­estate	–0.038***	–0.062**	–0.044*	–0.017	–0.034**	–0.087***	–0.022	–0.007
	(0.014)	(0.025)	(0.025)	(0.02)	(0.016)	(0.023)	(0.026)	(0.022)
Respondent reports savings	–0.016	–0.025	–0.032	0.003	–0.005	–0.003	–0.051*	0.043**
	(0.017)	(0.019)	(0.032)	(0.029)	(0.017)	(0.032)	(0.028)	(0.021)
Don’t know/no answer on savings	–0.031	–0.123***	0.000	0.011	–0.051	–0.058	–0.042	–0.044
	(0.037)	(0.042)	(0.058)	(0.089)	(0.039)	(0.060)	(0.061)	(0.053)
Capital city resident	–0.016	–0.165***	0.076**	0.004	–0.045**	–0.107***	0.008	–0.066***
	(0.035)	(0.027)	(0.032)	(0.044)	(0.018)	(0.031)	(0.024)	(0.023)
High income	0.021	–0.011	0.027	0.026	0.014	0.029	–0.039	0.059*
	(0.019)	(0.037)	(0.032)	(0.026)	(0.022)	(0.041)	(0.029)	(0.032)
Medium income	0.025	–0.032	0.039	0.054**	–0.004	–0.016	0.000	0.002
	(0.016)	(0.027)	(0.033)	(0.023)	(0.016)	(0.024)	(0.023)	(0.032)
Don’t know/no answer on income	0.050***	0.073**	0.005	0.061**	–0.007	0.067*	–0.082**	0.027
	(0.019)	(0.034)	(0.032)	–0.03	(0.030)	(0.038)	(0.042)	(0.037)
2-person household	0.005	0.036	0.014	–0.024	0.010	0.017	0.006	–0.001
	(0.022)	(0.035)	(0.037)	(0.036)	(0.019)	(0.037)	(0.029)	(0.034)
3-plus-person household	–0.018	–0.004	0.005	–0.035	0.027	0.005	0.016	0.048
	(0.024)	(0.043)	(0.040)	(0.035)	(0.021)	(0.036)	(0.032)	(0.039)
Country-fixed effects	YES	YES	YES	YES	YES	YES	YES	YES
Constant	YES	YES	YES	YES	YES	YES	YES	YES
Log likelihood	–5,104.0	–1,516.5	–1,580.1	–1,902.1	–4,625.5	–1,670.0	–1,116.7	–1,732.5
Pseudo R squared	0.16	0.13	0.10	0.16	0.10	0.07	0.08	0.11
Prob > Chi squared	1,205.7	. .	. .	. .	2,027.3	. .	. .	. .
Number of observations	8,777	2,667	2,702	3,404	8,411	2,656	2,619	3,132
P(DepVar=1)	0.45	0.36	0.64	0.38	0.70	0.60	0.82	0.69
Source: Authors’ calculations. OeNB Euro Survey 2018.
Note: Average marginal effects from probit estimations with country-fixed effects using data from the OeNB Euro Survey 2018; standard errors are adjusted for clustering at the regional level and reported in parentheses. ***, **, * denote that the average marginal effect is statistically different from zero at the 1%, 5% and 10% level, respectively. For a definition of the variables, see annex table A1. P(DepVar=1) denotes the unconditional sample probability of the respective dependent variable. By construction, a positive (negative) average ­marginal effect implies that a respondent is more (less) likely to agree with the question underlying the dependent variable. Base categories are: preference for the opposition party, 35 to 54 years old, male, low educational attainment, employed, atheist/agnostic, main residence in a similar condition as neighboring homes, no savings, no secondary residence, not living in the capital city, low income, single household, Czech resident. The CESEE group comprises all ten OeNB Euro Survey countries. The second group consists of three Central and Eastern European countries that became EU members in 2004. The third country group consists of three Southeastern European countries that became EU members in 2007 and 2013, respectively. The final group consists of four EU candidates or potential candidates.

Table A2 continued: Probit estimations: public perception of public governance; in CESEE and per country group

	Most politicians are just concerned about the next elections	The state is wasting taxpayer money
CESEE	CZ/HU/ PL	BG/HR/ RO	AL/BA/ MK/RS	CESEE	CZ/HU/ PL	BG/HR/ RO	AL/BA/ MK/RS
9	10	11	12	13	14	15	16
Experienced wage cuts (2008-2018)	–0.010	–0.046	–0.016	0.008	0.044***	0.024	0.014	0.078***
	(0.018)	(0.032)	(0.027)	(0.028)	(0.016)	(0.030)	(0.030)	(0.028)
Had to cut back on consumption (2008-2018)	0.105***	0.132***	0.067**	0.116***	0.080***	0.094***	0.067**	0.077**
	(0.017)	(0.026)	(0.028)	(0.030)	(0.017)	(0.032)	(0.028)	(0.031)
Public welfare benefits are currently ­important	–0.053***	–0.084***	0.021	–0.097***	–0.031*	–0.051*	0.015	–0.061**
	(0.019)	(0.024)	(0.026)	(0.030)	(0.019)	(0.027)	(0.032)	(0.027)
Received unemployment benefits (2008-2018)	–0.019	0.015	0.007	–0.041	–0.048**	–0.028	–0.034	–0.033
	(0.020)	(0.035)	(0.024)	(0.032)	(0.020)	(0.034)	(0.028)	(0.037)
Public sector work experience (2008-2018)	0.000	0.016	–0.009	–0.010	–0.003	0.047*	–0.017	–0.046**
	(0.016)	(0.025)	(0.027)	(0.026)	(0.015)	(0.025)	(0.023)	(0.022)
Would vote for a ruling party	–0.142***	–0.134***	–0.085**	–0.202***	–0.181***	–0.237***	–0.083***	–0.220***
	(0.024)	(0.036)	(0.034)	(0.039)	(0.022)	(0.036)	(0.028)	(0.040)
Would not vote	0.027	–0.009	0.069***	–0.005	0.022	0.052	0.054**	–0.027
	(0.020)	(0.031)	(0.024)	(0.041)	(0.023)	(0.042)	(0.024)	(0.045)
Don’t know for whom to vote	0.047*	–0.007	0.098***	0.002	0.016	–0.023	0.064	–0.039
	(0.024)	(0.032)	(0.031)	(0.045)	(0.028)	(0.040)	(0.041)	(0.041)
Refused to reveal voting preference	–0.009	–0.031	0.073	–0.072*	–0.013	–0.002	0.065	–0.091**
	(0.030)	(0.049)	(0.044)	(0.042)	(0.030)	(0.045)	(0.044)	(0.045)
Aged 19 to 34 years	–0.020	–0.037*	–0.015	–0.007	–0.028**	–0.080***	–0.007	–0.006
	(0.013)	(0.021)	(0.020)	(0.022)	(0.013)	(0.024)	(0.020)	(0.018)
Aged 55+ years	0.018	0.065**	–0.003	–0.003	0.002	–0.007	–0.046*	0.048
	(0.015)	(0.028)	(0.026)	(0.024)	(0.021)	(0.044)	(0.026)	(0.032)
Female	–0.026***	–0.005	–0.037***	–0.037*	–0.013	0.009	–0.011	–0.030
	(0.010)	(0.020)	(0.011)	(0.020)	(0.010)	(0.017)	(0.013)	(0.021)
Education attainment: high	–0.022	0.022	–0.047	–0.051	–0.011	0.060	–0.069*	–0.025
	(0.028)	(0.049)	(0.037)	(0.049)	(0.025)	(0.043)	(0.039)	(0.034)
Education attainment: medium	–0.037*	0.009	–0.072**	–0.050	–0.008	0.063*	–0.054	–0.027
	(0.019)	(0.039)	(0.036)	(0.030)	(0.019)	(0.036)	(0.036)	(0.025)
Self-employed	0.042**	0.094**	–0.010	0.011	0.023	0.023	–0.027	0.069*
	(0.021)	(0.037)	(0.027)	(0.032)	(0.022)	(0.039)	(0.026)	(0.037)
Unemployed	–0.005	–0.012	–0.037	0.008	–0.008	–0.044	–0.060*	0.014
	(0.021)	(0.048)	(0.026)	(0.032)	(0.020)	(0.046)	(0.032)	(0.027)
Retired	–0.002	–0.002	–0.024	0.005	0.013	0.088**	–0.008	–0.035
	(0.021)	(0.040)	(0.025)	(0.034)	(0.024)	(0.044)	(0.036)	(0.037)
Student	–0.006	0.020	–0.030	–0.012	0.009	0.020	–0.008	0.008
	(0.028)	(0.047)	(0.039)	(0.048)	(0.030)	(0.040)	(0.052)	(0.049)
Muslim	–0.012	–0.341*	0.039	–0.101*	–0.002	0.000	–0.021	0.082
	(0.051)	(0.197)	(0.074)	(0.058)	(0.054)	(.)	(0.069)	(0.080)
Orthodox Christian	0.038	–0.053	0.065	–0.066	0.017	–0.103	0.031	0.078
	(0.064)	(0.081)	(0.068)	(0.058)	(0.069)	(0.092)	(0.070)	(0.093)
Catholic Christian	–0.016	0.010	–0.122	–0.066	–0.035	–0.023	–0.172*	0.087
	(0.023)	(0.022)	(0.077)	(0.084)	(0.027)	(0.030)	(0.095)	(0.089)
Other Christian faith (e.g. Protestant)	–0.035	–0.079*	–0.032	0.081	–0.024	–0.069*	–0.066	0.221**
	(0.035)	(0.041)	(0.076)	(0.065)	(0.036)	(0.041)	(0.096)	(0.110)
Source: Authors’ calculations. OeNB Euro Survey 2018.
Note: Average marginal effects from probit estimations with country-fixed effects using data from the OeNB Euro Survey 2018; standard errors are adjusted for clustering at the regional level and reported in parentheses. ***, **, * denote that the average marginal effect is statistically different from zero at the 1%, 5% and 10% level, respectively. For a definition of the variables, see annex table A1. P(DepVar=1) denotes the unconditional sample probability of the respective dependent variable. By construction, a positive (negative) average ­marginal effect implies that a respondent is more (less) likely to agree with the question underlying the dependent variable. Base categories are: preference for the opposition party, 35 to 54 years old, male, low educational attainment, employed, atheist/agnostic, main residence in a similar condition as neighboring homes, no savings, no secondary residence, not living in the capital city, low income, single household, Czech resident. The CESEE group comprises all ten OeNB Euro Survey countries. The second group consists of three Central and Eastern European countries that became EU members in 2004. The third country group consists of three Southeastern European countries that became EU members in 2007 and 2013, respectively. The final group consists of four EU candidates or potential candidates.

Table A2 continued: Probit estimations: expectations of government-led job creation, subgroup of both beliefs; in CESEE and per country group

	Expecting jobs to be created primarily by ­government	Distrusting government but expecting jobs to be created primarily by government
CESEE	CZ/HU/ PL	BG/HR/ RO	AL/BA/ MK/RS	CESEE	CZ/HU/ PL	BG/HR/ RO	AL/BA/ MK/RS
17	18	19	20	21	22	23	24
Experienced wage cuts (2008-2018)	–0.007	–0.034	0.025	–0.011	0.006	–0.044*	0.017	0.011
	(0.024)	(0.042)	(0.041)	(0.042)	(0.019)	(0.025)	(0.037)	(0.034)
Had to cut back on consumption (2008-2018)	–0.014	0.003	–0.021	–0.019	0.037***	0.033**	0.034	0.047*
	(0.019)	(0.027)	(0.036)	(0.026)	(0.014)	(0.016)	(0.027)	(0.026)
Public welfare benefits are currently ­important	0.057***	0.03	0.026	0.093***	–0.012	–0.030*	0.004	–0.001
	(0.02)	(0.034)	(0.025)	(0.033)	(0.015)	(0.017)	(0.026)	(0.025)
Received unemployment benefits ­(2008-2018)	–0.011	–0.008	–0.003	–0.037	–0.036**	–0.024	–0.026	–0.027
	(0.023)	(0.044)	(0.04)	(0.038)	(0.018)	(0.027)	(0.033)	(0.031)
Public sector work experience (2008-2018)	0.050**	0.100***	0.058*	0.004	0.025	0.031**	0.016	0.029
	(0.025)	(0.035)	(0.034)	(0.037)	(0.016)	(0.015)	(0.031)	(0.030)
Would vote for a ruling party	0.012	0.017	–0.014	0.001	–0.134***	–0.131***	–0.098***	–0.162***
	(0.017)	(0.033)	(0.029)	(0.031)	(0.018)	(0.024)	(0.022)	(0.040)
Would not vote	0.028	0.059	0.056**	–0.002	0.017	0.033	0.071***	–0.027
	(0.02)	(0.037)	(0.024)	(0.036)	(0.020)	(0.028)	(0.022)	(0.045)
Don’t know for whom to vote	0.031	–0.047	0.091**	0.024	–0.021	–0.058*	0.041	–0.056
	(0.027)	(0.033)	(0.042)	(0.057)	(0.026)	(0.030)	(0.030)	(0.060)
Refused to reveal voting preference	0.022	–0.094*	0.112***	0.009	–0.051**	–0.051	0.040	–0.109**
	(0.027)	(0.055)	(0.037)	(0.041)	(0.026)	(0.040)	(0.040)	(0.045)
Aged 19 to 34 years	–0.016	–0.057***	0.042**	–0.035	0.015	–0.048***	0.055**	0.033
	(0.013)	(0.019)	(0.018)	(0.021)	(0.012)	(0.017)	(0.024)	(0.020)
Aged 55+ years	0.001	–0.01	–0.018	0.024	–0.004	–0.026	–0.024	0.036
	(0.016)	(0.027)	(0.031)	(0.027)	(0.014)	(0.019)	(0.025)	(0.022)
Female	0.007	–0.003	0.007	0.012	–0.005	–0.019	0.002	–0.000
	(0.012)	(0.02)	(0.022)	(0.023)	(0.010)	(0.015)	(0.019)	(0.019)
Education attainment: high	–0.071**	–0.116***	–0.123*	–0.027	–0.030	–0.025	–0.028	–0.047
	(0.033)	(0.04)	(0.066)	(0.048)	(0.021)	(0.036)	(0.040)	(0.034)
Education attainment: medium	–0.044**	–0.051*	–0.093**	–0.011	–0.011	–0.009	–0.025	0.003
	(0.02)	(0.029)	(0.04)	(0.03)	(0.016)	(0.029)	(0.030)	(0.022)
Self-employed	–0.052*	–0.06	0.033	–0.089**	–0.035*	–0.006	–0.035	–0.065*
	(0.027)	(0.044)	(0.038)	(0.043)	(0.019)	(0.025)	(0.031)	(0.035)
Unemployed	0.028	0.080*	–0.016	0.034	0.005	0.066**	–0.037	0.007
	(0.022)	(0.048)	(0.041)	(0.029)	(0.017)	(0.027)	(0.034)	(0.027)
Retired	0.01	0.028	0.024	–0.007	0.009	0.045	0.007	–0.016
	(0.023)	(0.04)	(0.04)	(0.038)	(0.021)	(0.028)	(0.035)	(0.034)
Student	–0.045*	–0.086	–0.035	–0.02	–0.039	–0.006	–0.038	–0.056
	(0.027)	(0.057)	(0.043)	(0.045)	(0.024)	(0.043)	(0.040)	(0.041)
Muslim	0.064	–0.018	0.005	0.098	0.073*	0.000	0.016	–0.173**
	(0.054)	(0.245)	(0.072)	(0.082)	(0.042)	(.)	(0.049)	(0.078)
Orthodox Christian	–0.018	0.042	–0.008	–0.028	0.023	0.001	0.064*	–0.236***
	(0.047)	(0.089)	(0.063)	(0.081)	(0.031)	(0.042)	(0.036)	(0.088)
Catholic Christian	0.099**	0.078	0.135**	0.159	0.029	0.024	0.135***	–0.321***
	(0.038)	(0.05)	(0.067)	(0.115)	(0.021)	(0.020)	(0.028)	(0.080)
Source: Authors’ calculations. OeNB Euro Survey 2018.
Note: Average marginal effects from probit estimations with country-fixed effects using data from the OeNB Euro Survey 2018; standard errors are adjusted for clustering at the regional level and reported in parentheses. ***, **, * denote that the average marginal effect is statistically different from zero at the 1%, 5% and 10% level, respectively. For a definition of the variables, see annex table A1. P(DepVar=1) denotes the unconditional sample probability of the respective dependent variable. By construction, a positive (negative) average ­marginal effect implies that a respondent is more (less) likely to agree with the question underlying the dependent variable. Base categories are: preference for the opposition party, 35 to 54 years old, male, low educational attainment, employed, atheist/agnostic, main residence in a similar condition as neighboring homes, no savings, no secondary residence, not living in the capital city, low income, single household, Czech resident. The CESEE group comprises all ten OeNB Euro Survey countries. The second group consists of three Central and Eastern European countries that became EU members in 2004. The third country group consists of three Southeastern European countries that became EU members in 2007 and 2013, respectively. The final group consists of four EU candidates or potential candidates.

Table A2 continued: Probit estimations: public perception of public governance; in CESEE and per country group

	Most politicians are just concerned about the next elections	The state is wasting taxpayer money
CESEE	CZ/HU/ PL	BG/HR/ RO	AL/BA/ MK/RS	CESEE	CZ/HU/ PL	BG/HR/ RO	AL/BA/ MK/RS
9	10	11	12	13	14	15	16
Other faith groups (e.g. Jewish, Buddhist)	–0.101	–0.146	–0.006	–0.309**	–0.152	–0.203	–0.163	–0.033
	(0.096)	(0.144)	(0.158)	(0.146)	(0.094)	(0.132)	(0.184)	(0.186)
Religion: no answer	–0.075*	–0.121**	0.021	–0.114	–0.039	–0.119*	–0.011	0.208
	(0.043)	(0.058)	(0.051)	(0.106)	(0.047)	(0.071)	(0.046)	(0.155)
Very interested in politics	–0.054**	–0.074**	–0.020	–0.072*	–0.044**	–0.070**	0.008	–0.071**
	(0.022)	(0.031)	(0.020)	(0.041)	(0.020)	(0.030)	(0.028)	(0.031)
Satisfied with life as it is	–0.022	–0.024	0.021	–0.054	–0.037**	–0.016	–0.004	–0.079***
	(0.020)	(0.031)	(0.029)	(0.034)	(0.018)	(0.024)	(0.031)	(0.030)
House in better condition than that of ­neighbors	–0.022	–0.052**	–0.031	0.004	–0.002	–0.009	–0.008	0.005
	(0.014)	(0.024)	(0.020)	(0.022)	(0.015)	(0.025)	(0.025)	(0.021)
House in poorer condition than that of neighbors	0.002	–0.040	–0.015	0.042	–0.003	0.013	–0.028	0.006
	(0.022)	(0.032)	(0.026)	(0.042)	(0.020)	(0.027)	(0.037)	(0.028)
Has secondary residence or other real ­estate	–0.036**	–0.067**	–0.031	–0.017	–0.033**	–0.059***	–0.032	–0.016
	(0.015)	(0.026)	(0.023)	(0.023)	(0.015)	(0.021)	(0.032)	(0.022)
Respondent reports savings	–0.020	–0.007	–0.042	–0.013	–0.002	0.009	–0.031	0.015
	(0.015)	(0.022)	(0.028)	(0.021)	(0.015)	(0.028)	(0.021)	(0.023)
Don’t know/no answer on savings	–0.049	–0.032	–0.043	–0.102	–0.031	–0.031	–0.030	–0.036
	(0.042)	(0.063)	(0.060)	(0.088)	(0.037)	(0.061)	(0.054)	(0.085)
Capital city resident	–0.047**	–0.151***	0.003	–0.035*	–0.027	–0.094***	0.000	–0.023
	(0.021)	(0.029)	(0.027)	(0.019)	(0.020)	(0.028)	(0.040)	(0.024)
High income	0.014	0.013	–0.020	0.047	0.006	–0.005	–0.042	0.064*
	(0.022)	(0.044)	(0.023)	(0.039)	(0.021)	(0.040)	(0.029)	(0.034)
Medium income	0.009	–0.025	0.001	0.048*	0.014	–0.013	0.007	0.046*
	(0.014)	(0.020)	(0.024)	(0.025)	(0.015)	(0.024)	(0.022)	(0.025)
Don’t know/no answer on income	0.018	0.087*	–0.052	0.048	0.019	0.061	–0.028	0.051
	(0.030)	(0.052)	(0.043)	(0.032)	(0.028)	(0.041)	(0.035)	(0.037)
2-person household	0.022	0.034	0.014	0.017	0.031	0.059*	0.015	0.021
	(0.016)	(0.029)	(0.023)	(0.034)	(0.020)	(0.035)	(0.028)	(0.041)
3-plus-person household	0.035**	0.043	0.018	0.037	0.019	0.026	0.002	0.022
	(0.018)	(0.028)	(0.025)	(0.037)	(0.020)	(0.033)	(0.034)	(0.035)
Country-fixed effects	YES	YES	YES	YES	YES	YES	YES	YES
Constant	YES	YES	YES	YES	YES	YES	YES	YES
Log likelihood	–4,607.0	–1,653.9	–1,119.4	–1,730.3	–4,848.1	–1,673.2	–1,263.6	–1,810.2
Pseudo-R-squared	0.09	0.06	0.08	0.12	0.09	0.07	0.06	0.10
Prob > Chi squared	1,019.2	. .	. .	. .	1,012.4	. .	. .	. .
Number of observations	8,414	2,650	2,617	3,147	8,329	2,637	2,580	3,108
P(DepVar=1)	0.71	0.62	0.83	0.68	0.66	0.57	0.78	0.65
Source: Authors’ calculations. OeNB Euro Survey 2018.
Note: Average marginal effects from probit estimations with country-fixed effects using data from the OeNB Euro Survey 2018; standard errors are adjusted for clustering at the regional level and reported in parentheses. ***, **, * denote that the average marginal effect is statistically different from zero at the 1%, 5% and 10% level, respectively. For a definition of the variables, see annex table A1. P(DepVar=1) denotes the unconditional sample probability of the respective dependent variable. By construction, a positive (negative) average ­marginal effect implies that a respondent is more (less) likely to agree with the question underlying the dependent variable. Base categories are: preference for the opposition party, 35 to 54 years old, male, low educational attainment, employed, atheist/agnostic, main residence in a similar condition as neighboring homes, no savings, no secondary residence, not living in the capital city, low income, single household, Czech resident. The CESEE group comprises all ten OeNB Euro Survey countries. The second group consists of three Central and Eastern European countries that became EU members in 2004. The third country group consists of three Southeastern European countries that became EU members in 2007 and 2013, respectively. The final group consists of four EU candidates or potential candidates.

Table A2 continued: Probit estimations: expectations of government-led job creation, subgroup of both beliefs; in CESEE and per country group

	Expecting jobs to be created primarily by ­government	Distrusting government but expecting jobs to be created primarily by government
CESEE	CZ/HU/ PL	BG/HR/ RO	AL/BA/ MK/RS	CESEE	CZ/HU/ PL	BG/HR/ RO	AL/BA/ MK/RS
17	18	19	20	21	22	23	24
Other Christian faith (e.g. Protestant)	–0.018	0.036	0.12	–0.388***	–0.027	–0.023	0.155	–0.410***
	(0.069)	(0.069)	(0.153)	(0.076)	(0.039)	(0.032)	(0.132)	(0.084)
Other faith groups (e.g. Jewish, Buddhist)	0.124	0.005	–0.03	0.372***	0.119	–0.075	–0.013	0.343***
	(0.092)	(0.104)	(0.114)	(0.097)	(0.103)	(0.060)	(0.074)	(0.072)
Religion: no answer	0.03	0.025	0.001	0.129	0.013	–0.021	–0.012	0.027
	(0.044)	(0.052)	(0.106)	(0.14)	(0.036)	(0.031)	(0.049)	(0.151)
Very interested in politics	–0.044***	–0.05	–0.003	–0.058***	–0.026**	0.012	–0.029	–0.049***
	(0.017)	(0.034)	(0.026)	(0.022)	(0.012)	(0.021)	(0.021)	(0.017)
Satisfied with life as it is	–0.008	–0.009	–0.056**	0.019	–0.082***	–0.065***	–0.088***	–0.088***
	(0.02)	(0.034)	(0.025)	(0.03)	(0.012)	(0.015)	(0.022)	(0.021)
House in better condition than that of neighbors	0.011	–0.035	0.089***	–0.006	0.030**	0.010	0.069***	0.013
	(0.02)	(0.035)	(0.025)	(0.029)	(0.013)	(0.022)	(0.024)	(0.020)
House in poorer condition than that of neighbors	–0.018	–0.009	0.041	–0.049*	0.010	–0.021	0.022	0.022
	(0.022)	(0.03)	(0.052)	(0.029)	(0.016)	(0.018)	(0.038)	(0.024)
Has secondary residence or other real ­estate	–0.039**	–0.055*	–0.026	–0.047	–0.028**	–0.065***	–0.016	–0.020
	(0.019)	(0.031)	(0.023)	(0.035)	(0.012)	(0.024)	(0.020)	(0.018)
Respondent reports savings	–0.059***	–0.062**	–0.017	–0.067**	–0.016	–0.033	–0.011	0.007
	(0.017)	(0.031)	(0.021)	(0.028)	(0.012)	(0.020)	(0.018)	(0.022)
Don’t know/no answer on savings	–0.014	0.092	–0.016	–0.131**	–0.024	–0.001	0.003	–0.106***
	(0.043)	(0.079)	(0.058)	(0.058)	(0.035)	(0.049)	(0.075)	(0.032)
Capital city resident	–0.085*	0.044	–0.04	–0.183***	–0.050**	–0.041**	0.009	–0.097**
	(0.046)	(0.037)	(0.031)	(0.055)	(0.024)	(0.018)	(0.023)	(0.040)
High income	–0.070***	–0.127***	–0.067	–0.026	–0.046***	–0.075***	–0.064*	–0.020
	(0.024)	(0.035)	(0.041)	(0.045)	(0.015)	(0.023)	(0.034)	(0.019)
Medium income	–0.047**	–0.086**	–0.069*	–0.006	–0.023	–0.063***	–0.042	0.016
	(0.022)	(0.036)	(0.04)	(0.037)	(0.018)	(0.022)	(0.035)	(0.027)
Don’t know/no answer on income	0.021	0.070*	0.005	0.018	0.028*	0.055***	0.006	0.024
	(0.024)	(0.042)	(0.042)	(0.04)	(0.016)	(0.021)	(0.032)	(0.024)
2-person household	0.005	0.058*	–0.049**	0.031	0.009	0.035	–0.010	0.013
	(0.02)	(0.031)	(0.024)	(0.042)	(0.018)	(0.023)	(0.029)	(0.037)
3-plus-person household	0.012	0.107***	–0.054*	0.004	0.002	0.029	–0.025	0.013
	(0.024)	(0.029)	(0.029)	(0.046)	(0.018)	(0.020)	(0.031)	(0.033)
Country-fixed effects	YES	YES	YES	YES	YES	YES	YES	YES
Constant	YES	YES	YES	YES	YES	YES	YES	YES
Log likelihood	–5,574.1	–1,669.2	–1,623.0	–2,151.1	–3,949.5	–950.1	–1,348.0	–1,534.7
Pseudo-R-squared	0.08	0.07	0.09	0.08	0.10	0.13	0.09	0.14
Prob > Chi squared	749.7	. .	. .	. .	2,373.3	. .	. .	. .
Number of observations	8,804	2,674	2,703	3,427	8,706	2,636	2,676	3,390
P(DepVar=1)	0.45	0.39	0.37	0.56	0.20	0.15	0.24	0.22
Source: Authors’ calculations. OeNB Euro Survey 2018.
Note: Average marginal effects from probit estimations with country-fixed effects using data from the OeNB Euro Survey 2018; standard errors are adjusted for clustering at the regional level and reported in parentheses. ***, **, * denote that the average marginal effect is statistically different from zero at the 1%, 5% and 10% level, respectively. For a definition of the variables, see annex table A1. P(DepVar=1) denotes the unconditional sample probability of the respective dependent variable. By construction, a positive (negative) average ­marginal effect implies that a respondent is more (less) likely to agree with the question underlying the dependent variable. Base categories are: preference for the opposition party, 35 to 54 years old, male, low educational attainment, employed, atheist/agnostic, main residence in a similar condition as neighboring homes, no savings, no secondary residence, not living in the capital city, low income, single household, Czech resident. The CESEE group comprises all ten OeNB Euro Survey countries. The second group consists of three Central and Eastern European countries that became EU members in 2004. The third country group consists of three Southeastern European countries that became EU members in 2007 and 2013, respectively. The final group consists of four EU candidates or potential candidates.

Detailed country-specific responses to the main survey questions of interest

CESEE-related abstracts from other OeNB publications

Austrian banks’ expansion to Central, Eastern and Southeastern Europe Milestones – review and outlook

Stefan Kavan, Tina Wittenberger

Published in Monetary Policy & the Economy Q1–Q2/20.

Mapping financial vulnerability in CESEE: understanding risk-bearing capacities of households is key in times of crisis

Nicolas Albacete, Pirmin Fessler, Maximilian Propst

Published in Financial Stability Report 39.