Distribution of household assets in Croatia

MARINA KUNOVAC, dott. mag. des*

Article** JEL: D1, D31, C35 https://doi.org/10.3326/pse.44.3.1

* The author wants to thank the two anonymous referees, as well as Maja Bukovšak, Vedran Šošić, Krunoslav Zauder, Mate Rosan, Ervin Duraković, Lana Ivičić and Ivica Rubil for their useful comments and suggestions. I would like to express my gratitude to Mrs Frédérique Savignac from the HFCN, the co-author in the paper by Arrondel, Roger and Savignac (2014) for providing a part of the Stata module to integrate generalized ordered probit models with multiple imputed and weighted data. This paper presents the views of the author and should not be interpreted as reflecting the views or opinions of the Croatian National Bank. ** Received: October 21, 2019 Accepted: February 24, 2020

Marina KUNOVAC Croatian National Bank, Trg hrvatskih velikana 3, 10000 Zagreb, Croatia e-mail: [email protected] ORCiD: 0000-0002-3200-5928 1 that the Croatian National Croatian the that The Household Finance and Consumption Survey (HFCS) was carried out in coordination with the Euro- 1 1 INTRODUCTION householdsamong assets of types different of distribution in inequality Potential macropru- monetary, as such policies, public designing when considered be should social and regional policy). Indential and other public policies (tax, demographic, on households depending effect that sense, monetary policy may have a reallocation and liabilities among households and theiron initial distribution of income, assets 2019). Macro- Tzamourani, 2017; exposure to the interest rates channel (Auclert, prudential policy contributing to maintaining stability of financial Oli- system distribution of assets among households (Carpantier, same time may also affect at the of monetary policy Kerm, 2017). For these reasons, implementation Van vera and and design of macroprudential measures to maintain financial stabilityshould be in distribution of household assets.accompanied by the analysis of inequality has previously been analysed on the basis Inequality among households in Croatia of income data only (e.g. Nestić, 2005; Rubil, 2013; Rubil, Stubbs and Zrinščak, in on inequality literature paper builds upon the existing The following 2018). of distribution of analysis an provides and income household of distribution in the collected The analysis is based on the data household assets in Croatia. (HFCS) Survey Consumption and Finance Household Bank (CNB) first carried out on a sample of households in Croatia in mid 2017. liabilities, assets, financial and real household on data detailed covered survey The of households. characteristics income, consumption and other socio-demographic Abstract assets net household of distribution and components main the analyses paper This Consumption and Finance Household the from data the of basis the on Croatia in Survey (HFCS) by taking into account different socio-demographic - characteris distributed widely are assets real that indicate results main The households. of tics among households, whereby 85% of households own the household main - resi dence. Financial assets and liabilities account for among share larger wealthier an of position the establishing determinants main the of analysis The households. individual household in distribution of assets has additionally highlighted importance of the the household main residence (HMR). Households with inherited HMR less are likely to be positioned in the lowest net asset quintile. In addition, households with HMR in the city level the of that found also Zagreb has or survey The on groups. quintile the asset higher Adriatic in be Coast to likely are more the of age and status market labour attainment, educational income, household of person affect of the positioning household probability a reference household in a certain net asset quintile. Keywords: survey data, Household Finance and Consumption Survey (HFCS), transfers, Croatia intergenerational net household assets, household inequality, pean Central Bank. The European Central Bank has already coordinated two previous HFCS waves, the first coordinated Bank has already The European Central pean Central Bank. one in the 2008-2010 period and the second in 2013. Since Croatia entered the EU in July 2013, the CNB joined the survey in the third wave.

public sector marina kunovac: economics distribution of household assets in croatia 44 (3) 265-297 (2020) 266 public sector marina kunovac: economics distribution of household assets in croatia

267 44 (3) 265-297 (2020) - - - The median value of the main residence residence of the main value median The 2 . To the author’s author’s the To generalized probit model. ordered By comparing the data with those collected in the EU in the second survey wave (2013), one can observe By comparing the data with those collected in the EU in the second survey wave (2013), one can observe higher participation rate of the household main residence in Croatia compared to other EU countries (where higher participation euros 165 thousand at stood of which value median HMR the the owned households of 62% of average an (ECB, 2016)). 2 This paper brings new insights into distribution of different components of house - of components of different distribution into new insights paper brings This an of position relative establish that determinants main the assets and net hold the survey, the assets. Prior to of net distribution of terms in household individual data on aggregate depended completely assets in Croatia of household analysis sources such as financial accounts thatcontained the data on total financial assets and liabil of assets on distribution information include but did not and liabilities, Apart from detailed analysis of distribution of household assets, the paper analy paper assets, the of household distribution of analysis detailed from Apart among categories asset certain of distribution in found inequality has analysis The share of household total for a large account assets households. Real Croatian given that 85% of households own assets, much more than in other EU countries, (HMR). residence main household the ities and inequality among households. In addition, researchers did not have ade- did not have researchers In addition, among households. ities and inequality analysis of household real assets. quate data source for of a household in the distribu- the position establishing ses the main determinants using the assets by of net tion in the position of a household affecting of determinants knowledge, the analysis Croatia. On the other assets has not yet been carried out in the distribution of net for the euro area counties (e.g. a subject of extensive literature hand, this has been Peichl, and Kontbay-Busun 2013; Medgyesi, and Sierminska 2016; Du Caju, Arrondel, Roger and Savignac, 2015; Fessler and Schürz, 2015; 2015; Leitner, the two previous survey waves. 2014) on the basis of the HFCS data from With regard to the determinants establishing the position of a household in the the position of establishing regard to the determinants With indicate model probit ordered generalized assets, results of the of net distribution and assets, of income position in distribution between a household a correlation but its statistical significance andintensity varies depending on the position of a household in the income distribution. This result is robust for different specifica amounts to 66 thousand euros and it makes up the bulk of total net household net total of bulk the up makes it and euros 66 thousand to amounts assets. Significant inequality was observed in terms of financial assets because only a certain portion of households own The substantial median financial assets. value of household financial assets stands at The 500 resultseuros. of descriptive in distribution of total net household assets, analysis have pointed to variation asset real income, characteristics, socio-demographic different on depending ownership and geographic location. the helps to explain way the HMR was acquired The tions of household income. in case of of having a certain household in a certain asset quintile probability lower asset levels, but it is not significant for determining the probability of posi- tioning a household in higher net asset quintiles. - - - The first stage included stratification of segments ). In the second stage of sample stratification, the 3 2 and over 120m over and 2 According to the HFCS methodology prescribed by the ECB, the survey did not cover institutional households.According to the according to occupied dwellings in spatial units of the country (belonging to the units of the country (belonging in spatial to occupied dwellings according in of Zagreb, or neighbourhood in the case of the City (city) same municipality The Census). Population 2011 the from areas enumeration the with accordance dwelling the of size the on groups depending two into divided then were segments 120m (up to 3 2 IMPLEMENTATION OF THE SURVEY OF 2 IMPLEMENTATION Household Finance and Consumption The Croatian National Bank ordered the research agency in Ipsos market by the Survey (HFCS) and it was implemented survey question- The (CBS). Statistics Bureau of Croatian with the cooperation was harmonized network and it research ECB the was designed within naire Croatia in out 2016, was carried covering survey, The states. across EU member from March to June 2017. The stratification of private households from the population to the gross sample was carried out in two stages. The analysis has also emphasised the importance of geographical location of the location of geographical the importance also emphasised has The analysis in the main residence with households being equal, factors all other whereby, HMR, net likely to be in higher are much more Adriatic Coast Zagreb or on the the City of Croa- main residence in Eastern households with the compared with asset quintiles very much status and age labour market educational attainment, tia. In addition, thatIn asset quintile. certain net a in household a positioning of probability the affect householdsreference persons and educated and older with more respect, households reference person are more likely to be wealthier. with self-employed basic infor includes chapter of the paper is as follows: the second The structure The third and survey design. of the survey implementation on the technical mation of net assets: real components about the main information covers detailed chapter and financial assets and liabilities, their distribution among households and val- taking into account in distribution of net assets among households, ues. Inequality is discussed in the of households, characteristics socio-demographic different estab- to model econometric the includes chapter fifth the whereas chapter, fourth a household of having the probability affect characteristics lish which household gives an overview of the the sixth chapter Finally, net asset quintile. in a certain main survey conclusions. segments were divided according to the geographical location as follows: Adriatic as follows: location to the geographical according segments were divided of admin to the types and according Croatia and Central Croatian Coast, Eastern istrative units (city or municipality). Cities of Zagreb, Split and Rijeka formed and Rijeka Split of Zagreb, Cities or municipality). units (city istrative After obtaining the strata were obtained. 16 different separate strata. In this way, within were selected size to their proportional of segments number certain a strata, irrespec of selection, households had equal probability each stratum. In this way, 800 seg- contained 16 strata all, in All to. assigned were they stratum of the tive ments from 552 settlements. Finally, five occupied dwellings were randomly

public sector marina kunovac: economics distribution of household assets in croatia 44 (3) 265-297 (2020) 268 public sector marina kunovac: economics distribution of household assets in croatia

269 44 (3) 265-297 (2020) - - 4 CAPI method was employed in the – were thus represented in higher numbers, higher in thus represented were 2 . The calculation of weight takes into account the proba- The calculation of weight takes into . 5 . This methodology allows for several different final survey ver 6 Computer Assisted Personal Interview To take into account unequal probability of participation in the sample among households. in the sample among households. take into account unequal probability of participation To the HFCN. Models for estimation of missing values for particular variables are pre-defined within Recent studies (e.g. Blanchet, Flores and Morgan, 2018) indicate that the combination of survey and tax data Recent studies (e.g. Blanchet, Flores and Morgan, represent the most detailed adjustment in case of underrepresentation of richer households in wealth surveys. wealth households in of richer of underrepresentation case in adjustment detailed most the represent and tax data on the num- procedure that combines survey income microdata The authors propose a statistical ber of taxpayers by income brackets, which gives a new adjusted dataset with new weights and observations At the same time, all the other survey data remain consistent. with adjusted income values. 5 6 4 The survey. A total of 1357 households from the gross sample took part in the survey so A survey. the relatively high unit non response andthe response rate amounted to 33%. Given 1), the Table segments (see population heterogeneity of response rate in different net sample was weighted Vermeulen (2014; 2016) has shown (2014; 2016) has status may due to wealth that non-response Vermeulen - the prob reduce To the survey. by estimated assets of total to underestimation lead (2016) suggests Vermulen households, wealthier among non-response of lem of house- of measurement kind by using some households of richer oversampling pur for the HFCS for Croatia, of the In case criterion. selection as a wealth hold selected from each segment and this corresponds to the sample of 4000 occupied occupied of 4000 sample the to corresponds this and segment each from selected were included that Census) the 2011 to (according households of 4070 dwellings gross sample. in the survey Households with dwellings over 120m over dwellings Households with poses of the survey sampling process, the size of the dwelling (in square meters) sampling process, the size of the dwelling poses of the survey of household assets. was used as a measurement Stochastic multiple imputation was used to compensate for the item non responseStochastic multiple imputation was used rec- ECB to the According of households. that may be registered in the net sample the net sample were imputed by using theommendations, the missing data within are missing at random so the miss- €MIR methodology that implies that responses values obtained by the estimates from the ing data are replaced by several different model stochastic bility of selection of households and heterogeneity of responses among different of responses among different bility of selection of households and heterogeneity all weights were additionally weighted tosegments of population. Furthermore, Census. according to the 2011 reflect age and sex distribution of population accounting for 25% of the gross sample. At the same time, they make up 10% of they make At the same time, gross sample. for 25% of the accounting Split and households from the City of Zagreb, In addition, population. the overall of the gross sample (35% in the gross sample were also overrepresented Rijeka to results from previous according since, population) with 25% of the compared were methods These rates. response low their known for were cities surveys, these richer households in the sample. used to minimize underrepresentation of sions, the only difference being imputed values of the missing data. The sampling, being imputed values of the missing data. sions, the only difference outweighting and imputing processes as described in the text above were carried Household Finance and by the CBS in accordance with the guidelines of the ECB’s Consumption Network (HFCN). A total of five imputed survey versions with the 8 6 19 15 9 16 9 7 2 1 1 2 1 1 % share Population 111,113 124,958 96,573 279,420 230,580 140,397 232,817 128,137 26,420 19,602 20,557 33,039 18,635 17,641 occupied population Number of Number dwellings in dwellings in 5 4 10 17 9 12 5 8 5 4 4 9 4 2 % share 74 59 133 229 128 158 64 103 63 52 60 125 50 28 Net sample in HFCS Number of Number households households respondent respondent 6 4 21 12 6 12 6 10 4 3 5 6 2 3 % share 115 220 150 830 485 225 480 230 380 155 100 195 225 95 Gross sample Gross selected occupied randomly the strata* Number of Number dwellings in 1 able Central Croatia, city, Central Croatia, city, HMR < 120 square meters Strata Central Croatia, HMR municipality, < 120 square meters City of Zagreb, HMR < 120 square meters Eastern Croatia, HMR < 120 city, square meters Eastern Croatia, HMR municipality, < 120 square meters Adriatic Coast, city, Adriatic Coast, city, HMR < 120 square meters Adriatic Coast, HMR municipality, < 120 square meters Cities of Split and Rijeka, HMR < 120 square meters Central Croatia, city, Central Croatia, city, HMR > 120 square meters Central Croatia, HMR municipality, > 120 square meters City of Zagreb, HMR > 120 square meters Eastern Croatia, HMR > 120 city, square meters Eastern Croatia, HMR municipality, > 120 square meters Adriatic Coast, city, Adriatic Coast, city, HMR >120 square meters T population in the overall and net sample and in gross households of Structure

public sector marina kunovac: economics distribution of household assets in croatia 44 (3) 265-297 (2020) 270 public sector marina kunovac: economics distribution of household assets in croatia

271 44 (3) 265-297 (2020) 1 % 0,2 100 share 8 Population 14,064 2,605 occupied 1,496,558 population Number of Number dwellings in dwellings in 2 1 % 100 share

7 21 10 Net sample 1357 in HFCS Number of Number households households respondent respondent 2 1 % 100 share 85 30 4,000 Gross sample Gross selected occupied randomly the strata* Number of Number dwellings in dwellings in Adriatic Coast, Adriatic Coast, HMR municipality, > 120 square meters Strata Cities of Split and Rijeka, HMR > 120 square meters TOTAL The selected parts of the Chapter were presented in CNB (2019). A similar approach was applied in Arrondel, Roger and Savignac (2014) and Georgokoponus (2019). Weights Weights (2019). Arrondel, Roger and Savignac (2014) and Georgokoponus similar approach was applied in A are based on the data on geographic distribution of households and respondents’ age and sex. Controlling for age and sex. Controlling distribution of households and respondents’ on geographic are based on the data these variables in the model is achieved by their direct inclusion in the regression. 7 8 3 MAIN COMPONENTS OF HOUSEHOLD ASSETS AND LIABILITIES ASSETS HOUSEHOLD 3 MAIN COMPONENTS OF The Table 2 shows the main components of household real and financial assets under consideration, the liabilities of assets and For each category and liabilities. the in assets/liabilities of category certain a with households of share the on data In addition, values. median and mean as their as well given, are population overall in the total value component rate of each particular 2 shows participation Table the of assets/liabilities, reflecting the relative importance of different categories of of components of all values presented The for households. assets and liabilities assets and liabilities reflect subjective households’ estimates that may not match real market values. The results presented in the paper underwent a statistical process as described in process as described a statistical The results presented in the paper underwent In addi- Stata. in data imputed of multiple for processing Boes (2006), designed were not but they analysis, descriptive weights were used in the the estimated tion, model. applied in the estimation of the econometric accompanying weights were submitted to the CNB. Detailed information about theaccompanying weights were submitted questionnaire design and imputation andsample selection, survey implementation, weighting of the results will be available in Jemrić and Vrbanc 2019. Additional A4. Table Appendix in in the information on the net sample is presented Note: Note: Geographical location “Adriatic Coast” includes the following counties: Primorje- Šibenik-Knin, Split-Dalmatia, Istria and Lika-Senj, Dubrovnik-Neretva. Zadar, Gorski Kotar, Geographical location “Eastern includes Croatia” the following counties: Sisak-Moslavina, Požega-Slavonia, Osijek- Brod-Posavina, Karlovac, Bjelovar-Bilogora, Virovitica-Podravina, Geographical location Baranja “Central and includes Croatia” Vukovar-Srijem. the following Koprivnica-Križevci and Međimurje. Krapina-Zagorje, Varaždin, counties: Zagreb, calculations. ECB and author’s Source:

- - The figures shown inTable the 2 indicate that98% of Croatian households own euros. thousand 67 of value median the with financial) or (real assets of kind some for assets financial and value asset total the of 97% for account assets real this, Of impor is it figures, these interpreting when hand, other the On 3%. remaining the tant to note that the survey strongly underestimates the value of financial assets, because, according to financial accounts, the value of household-owned financial survey by the collected data other Yet, higher. times seven assets is approximately the socio-demographic with other data sources (e.g. that allow for comparison and the share of household value income the total of households, characteristics data alternative in recorded figures the with line in are ownership) residence main 2019). Vrbanc, sources (Jemrić and estate, real household-owned of types different includes portfolio assets Real detailed A etc.). artwork, antiques jewellery, (valuable and other valuables vehicles HMR the of value, in terms shows that, assets real of of components analysis assets. In general real of total value of the 75% share, i.e. largest for the accounts The median value own the household main residence. terms, 85 % of households col- those with data the euros. By comparing 66 thousand HMR amounts to of the in the second survey wave (2013), one can observe higher par in the EU lected EU other to compared Croatia in residence household main of the rate ticipation households owned the HMR the median countries (where an average of 62% of high similarly hand, other On the euros). thousand 165 at stood which of value privatization share of HMR can be observed in other countries that implemented in the 1990s owned housing stock of socially – (Estonia, Slovenia, Slovakia ECB, 2016). data for Croatia shows More detailed analysis of HMR ownership that, of 85% 77% own the whole residence and the of households with the main residence, for 6% of households, account Also, renters of it. part 8% own some remaining differ- owners. Substantial not real are HMR though they whereas 9% use the HMR was acquired. of how the terms households in among observed were ences it or received it as a gift, it, 34% inherited In that respect, 36% of households built of through a combination it 2% acquired remaining and the 28% purchased it these options.

public sector marina kunovac: economics distribution of household assets in croatia 44 (3) 265-297 (2020) 272 public sector marina kunovac: economics distribution of household assets in croatia

273 44 (3) 265-297 (2020) . - - - 4 0 9 3 1 2 0 0 0 0 0 0 3 9 1 11 97 73 66 63 34 24 liabilities Share in total Share value of assets/ % EUR thousands 6 6 5 9 6 4 6 4 4 1 8 94 54 30 13 69 10 30 26 0,2 0,8 111 114 209 107 Mean 4 2 5 5 3 2 2 0 2 2 1 5 70 66 20 20 25 20 16 67 61 0,5 0,2 0,1 0,4 Median in EUR thousands 9 4 5 6 3 5 9 6 85 23 94 69 82 80 14 41 36 27 13 98 1,4 0,4 0,7 0,4 100 in % Share of Share households 2 for main residence for other real estate Credit lines/overdrafts Credit card debt Other non-mortgage loans able Components of net assets Components Main residence (85%) Other real estate (1) Real assets Mortgage debt Vehicles Other valuables Self-employment business assets* (2) Financial assets Sight accounts Savings accounts funds/ pension Voluntary whole life insurance Mutual funds Money owed to household Shares Bonds Other types of financial assets (3) Liabilities Non-mortgage debt (1+2) Gross assets (1+2) Gross ((1+2)-3) Net assets In respect of other components of household real assets, the survey results show survey results show assets, the of household real of other components In respect value was property whose median that 23% of households own other real estate Also, euros. thousand 20 at stood and HMR the with compared lower significantly to 4 thousand amounted value whose median 69% of households owned vehicles 5% of house- business assets, euros per household. In terms of self-employment holds reported it. Its median value was 25 thousand euros, compared to the sub- self- that means euros, which of 209 thousand value mean higher stantially employment business assets significantly contribute to the overall inequality Note: *Self-employment business assets means any household-owned component of assets real in running a self-employment business. estates, vehicles or valuables) used (real of amount the equal assets Net assets. financial and real of sum the as calculated are assets Gross mem- EU across harmonised been has survey the Since liabilities. household of net assets gross households the for calculated are means and Median euros. in expressed are values its states, ber that own a certain category of assets. calculations. ECB and author’s Source: T Liabilities Assets and Household of Components

- 9 Non- 10 Privatization of the socially owned housing stock was implemented in accordance with the Act on the Sale with the in accordance of the socially owned housing stock was implemented Privatization - Zaud and Rosan and (2019) CNB the in described HFCS are the in collected liabilities household on Details mortgage debt was reported by 36% of households. This type of debt mostly cov- This type of debt households. mortgage debt was reported by 36% of 27% of households) and other non-mort- ers credit lines/overdrafts (reported by thousand euros. gage loans, with the median value of 2 According liabilities. household of net gross assets of amount the assets equal Net of household net assets stands at 61 thou- to the survey results, the median value shows Figure 1A The 107 thousand euros. stands at value mean The sand euros. 5% of the poorest that indicating of net assets in percentiles, the distribution increases The value of net assets gradually households have almost no assets. Above the 75th percentile, 75th percentile. up to the above the 5th percentile at the distribution tail above the the increase becomes more rapid, especially of types important of most topology shows the 1B Figure The percentile. 90th er (2019). Household debt distribution in Croatia had previously been analysed by Herceg and Šošić (2011) and Herceg and Nestić these (2014). analyses However, were based on the Household Consumption Survey of house characteristics socio-demographic main the and data debt micro included survey The CBS. of the holds. On the other hand, the data on assets (real or financial) were not available. of Apartments with Tenancy Rights (Zakon o prodaji stanova nad kojima postoji stanarsko pravo, NN 27/91). Tenancy Apartments with of 9 10 Household Household financial assets are veryhomogeneous, in linewith the results of the previous HFCS waves that found lack of diversification- Portu Slovenia, Slovakia, such as Estonia, average, euro area below the countries offinancial assets in survey the According to 4). 2016: Room, and (Merikull and Greece Malta gal, results for Croatia, the median value of financialassets is500 euros per- house of households) with (reported by 81% share largest account for the hold. Deposits for rates were observed Highest participation value of 300 euros. the median deposits those for savings account with deposits (80%), compared sight account compo- from deposits, other most important Apart own it). (14 % of households nents of household-owned financial assets includeshares in voluntary pension Very shares (5%). traded publicly and (6%) insurance life whole funds and owed to money funds, bonds, mutual were observed for rates low participation the household and other types of financial assets. Yet, theseinterpreted with caution, resultssince the aggregated values should of financial assets from be the finan- in registered those with compared values lower times seven to point HFCS cial accounts. households. Croatian of liabilities on the data detailed collected also survey The In that sense, 41% of households are in some way indebted. Mortgage debt (66% Mortgage debt are in some way indebted. sense, 41% of households In that of the total debt) is the most significant component of household debt, compared debt the mortgage value, total debt (34%). Despite its high with non-mortgage was not significantly distributed among households since only 9% of households 20 thousand euros. stood at debt mortgage of the value median The it. reported to Low share of mortgage debt and high HMR ownership rates may be attributed 1990s. Back in through went economy Croatian the that process transition the the household of today) acquired (the elderly then, the vast majority of population stock. owned housing of socially through privatisation main residence

public sector marina kunovac: economics distribution of household assets in croatia 44 (3) 265-297 (2020) 274 public sector marina kunovac: economics distribution of household assets in croatia

275 44 (3) 265-297 (2020)

- -

) t

l e s % t a s i 3 s e

) s r 2 a

( o ) r

% e p e e 6 e % e t ( h r

4 a t d h t

s 1 t s e O e

( l e p l m y i t A T

) n i a % 5

m 8

( b d

l e o c 1 h n e e s d u i s o e H r igure F of household assets Topology components), (most important of households share s 9 t 9 e s s 5 a 9

t e 0 9 N 5 8

0 s 8 e i t 5 i l 7 i b 0 a 7 i l

l 5 a 6 t o 0 T s 6 e l 5 ti 5 n

s e 0 t c 5 e r s s 5 a 4 Pe

l a 0 i 4 c n 5 a 3 n i 0 F 3 5 2 s 0

t 2 e s a s 5 a 1

l 1 a 0 e 1 R 5 0 0 0 0 0 0 0 0 0 0 0 igure 0 0 0 0 0 0 0 0 0 0 8 7 6 5 4 3 2 1 -1 -2 The Figure 2 shows more pronounced inequality in the distribution of financial assets (Gini coefficient of 0.88) compared with inequality in the distribution of real assets (Gini coefficient of 0.59). This is typical of countries with a high share 4 INEQUALITY IN THE DISTRIBUTION OF HOUSEHOLD NET ASSETS NET HOUSEHOLD DISTRIBUTION OF THE IN 4 INEQUALITY by using the presented assets is net of household the distribution in Inequality Lorenz curve and Gini coefficient. In addition, since the existing income, due to household on inequalities is based in Croatia household inequality literature on col- data the of basis the on income of curve Lorenz the presents also paper the of inequality The Lorenz curve is a graphical representation lected by the HFCS. verti and households of share cumulative represents axis horizontal the whereby The Gini coefficient for total household net assets stands at 0.61, which indicates households com- Croatian of net assets among in distribution lower inequality pared with the euro average since the Gini coefficient for net household assets in HFCS second the from results the to according 0.69, to amounts area euro the wave from 2013 (ECB, 2016). Source: ECB and author’s calculations. calculations. ECB and author’s Source: of of household assets. In case of equal distribution share cal axis cumulative diagonal of the square (the so-called line assets, the Lorenz curve would match the curve Lorenz the closer the of inequality, lower the level The equality). of perfect and vice line, diagonal is to the away the farther the level, the higher versa, the curve is from the baseline. The Gini coefficient is a ratio of the area between the below the area triangle and the total of the square curve and the diagonal Lorenz diagonal line. to 85% according of Croatia, ownership (in case residence of household main survey results) that adds to significant prevalence of real assets among house­ F thousands of net assets, in EUR Distribution household assets. It indicates that some households own several types of assets, types of several own some households that It indicates assets. household However, deposits. time and estate real other residence, main household the e.g. (6%). relatively small fraction represent a these households - Even though the concept of net income is of net income Even though the concept 12 , in which the Lorenz curve shows savings of savings shows curve Lorenz the which in (2016) , CNB Evident inequality in the distribution of financial assets is consistent with with consistent is assets financial of distribution the in inequality Evident 11 The data from the CBS Statistics on Income and Living Conditions (SILC) for 2016 point to somewhat The data from the CBS Statistics on Income and Living Conditions (SILC) for 2016 point to somewhat Austria is is an interesting example in that sense where inequality in the distribution of financial assets is is assets of financial distribution in the inequality where sense that in example an interesting is is Austria the results presented in presented results the of Croatia in 2014. in the Republic natural persons The Gini coefficient for individual subcomponents of real and financial assetsis A5. Table in Appendix in the presented of assets and gross of the distribution comparison In terms of income, income of gross distribution in the inequality suggests less pronounced households among income compared with inequality in the distribution of real or (Gini coefficient of 0.51 for financial income). assets lower than inequality in the distribution of real assets, given the low share residence ownership of main (45%). Linder and Schurz (2019). For more details, see Fessler, lower income inequality in Croatia, but in case of the CBS SILC, the Gini coefficient is significantly lower is significantly lower lower income inequality in Croatia, but in case of the CBS SILC, the Gini coefficient from because the Gini coefficient comparable The results are not completely and stands at 0.3 (CBS, 2017). to gross amounts, i.e. includ in the HFCS relate the data collected whereas all CBS SILC covers net income, ing taxes and social insurance contributions. 12 11 holds. mostly used in analyses on inequality and welfare, the HFCS on inequality and welfare, the HFCS mostly used in analyses collects the data on so the contributions, insurance social taxes and i.e. including gross income only, respect, In that of the text. is analysed in the remainder in gross income inequality income of gross in the distribution inequality overestimates the HFCS probably because a significant number of households reported no income and their annual possessed some of them time, same the at (7%), whereas, was zero gross income income, employment includes gross income annual total the assets. Since valuable rent, income from financial assets, pensions, social transfers or any other sources the in presented of data value actual the that indicates result this income, of responses was deliberately omitted. For this reason, the Gini coefficient for gross income was for the households whose also estimated annual gross income exceeds EUR 1,300 (the selected amount reflects the fact that a single-person household of HRK min- 800, i.e. the amount of the guaranteed a monthly minimum received 0.44. at stands coefficient The 152/14). NN skrbi, socijalnoj o (Zakon benefit imum

public sector marina kunovac: economics distribution of household assets in croatia 44 (3) 265-297 (2020) 276 public sector marina kunovac: economics distribution of household assets in croatia

277 44 (3) 265-297 (2020) - - 1.0 s t e 0.9 s s a

t e N 0.8 e 0.7 m o c n I 0.6 s 0.5 t e s s a

l a i c n 0.4 a n i F 0.3 , such as sex, educational attainment, age or labour or age attainment, educational sex, as such , 13 Cumulative share of households s t e 0.2 s s a

l a e R 0.1 0.0 2

1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 igure assets/income of types different of share Cumulative The reference person is the one chosen by household members as the person most informed about house- The reference person hold finances. market status, presented in the Figure 3, shows that the educational attainment can attainment educational the shows that 3, Figure the in presented status, market ref- households with highly educated assets and that of net value to the be related share (30%) of persons with net asset value in the erence persons have the largest highest, fifth quintile. The share of persons in the highest asset quintile increases once the refer age and slightly decreases in proportion with the reference person’s ence person retires. In terms of labour market status, self-employed persons stand status, self-employed market In terms of labour person retires. ence (over 50% of the self- highest asset quintile share in the out given their largest employed classified in the fifth net asset quintile). On the other hand, relatively poor households make up the majority among the households with non-active 13 nection between inequality in the distribution of net assets and different household of net assets and different in the distribution inequality between nection of the characteristics socio-demographic main the of analysis The characteristics. person reference household The availability of detailed survey data provides additional evaluation of intercon evaluation additional survey data provides of detailed The availability Source: ECB and author’s calculations. ECB and author’s Source: F and income net assets and financial for real, curve Lorenz

- Non-active

Unemployed Employed

households

Retired

Labour market status

employed - Self

K5

65+ K4 45-65

K3

35-44 Age

K2 16-34

K1 Higher

high value (17% of households are both in the lowest (17% of households are high value

Secondary Primary Education

Net asset quintiles

Women Men Sex 3 0 igure 90 80 70 60 50 40 30 20 10 100 income quintile and the highest asset quintile). Even though the literature offers offers the literature though Even quintile). asset highest the and quintile income quintiles of why some households are in the lowest income several explanations first the in pensioners of proportion high a as (such quintiles asset highest the and a having low current incomes, have accumulated who, despite quintile, income transfers of intergenerational impact potential assets, or a of amount considerable decomposition of data has level), a detailed that are not related to the income As previously households. to Croatian do not apply explanations these shown that of number large a quite Croatia, in out carried survey the in explained, Source: ECB and author’s calculations. ECB and author’s Source: with also connected are and age status labour market attainment, educational The of the value of net a determinant being this income of household income, the level accumu income by the savings from current assets, which can be approximated transfers and gifts (for a by intergenerational and increased through time lated between income The interconnection discussion, see Du detailed Caju, 2016). in the distribution of net assets among households is shown in level and inequality also are quintile) income highest the households (in earning Top 4A. Figure the quintile). asset net in highest positioned are (40% of them wealthiest the among assets, yet net own low-value usually quintile lowest income Households in the have assets of some of them with line in since, (7%), and, kind of any income have not did they that responded reference persons. In that sense, over 50% of these households are in the lowest the lowest are in households of these over 50% sense, In that persons. reference quintile. net asset F % in assets, net of quintiles and households of characteristics Socio-demographic

public sector marina kunovac: economics distribution of household assets in croatia 44 (3) 265-297 (2020) 278 public sector marina kunovac: economics distribution of household assets in croatia

279 44 (3) 265-297 (2020)

- 5 Q 5 Q 4 Q 4 s Q e l i t n i 3 u 3 q

Q Q e m o c 2 n I Q 2 Q 1 Q above EUR 1,300 1 The Figure 4B therefore shows therefore 4B Figure The Q 14 B) annual gross household income B) annual gross

0 0 0 0 0 0 0 0 0 0 0

0 9 8 7 6 5 4 3 2 1 s e l i t n i u q t e s s a t e 1 N 5 Q 5 Q 4 Q 4 s Q e l i t n i 3 u 3 q

Q Q e m o c 2 n I Q 2 Q 1 Q 1 Q 4 A) total 0

90 80 70 60 50 40 30 20 10

100 s e l i t n i u q t e s s a t e igure N Income components included in the definition of the annual gross income and components of all the other the of all components and gross income annual of the definition the in included components Income the distribution of assets and incomes for the households whose annual gross households whose incomes for the of assets and the distribution income exceeds EUR 1,300 (the amount of the guaranteed than EUR 1,300 lower gross income annual households with even when minimum However, benefit). low very with households observe still can one sample, the from excluded are ine- that may affect This is why other factors asset values. incomes and high net of net assets are also examined. quality in the distribution variables whose values are collected in the survey are available in CNB (2019a). 14 minant. In the first decile of net assets, only 3% of households inherited the HMR. the inherited households of 3% only assets, net of decile first the In minant. Source: ECB and author’s calculations. ECB and author’s Source: irrespec- 2015) suggests that, Piketty, and Zucman 2011; (Piketty, research Recent in inequality transfers may substantially affect tive of income, intergenerational may household residence of net assets. In that sense, inherited the distribution status in terms of the tenure the between interconnection The role. a pivotal play net of component important most the as residence, main way the the HMR and asset value, is shown in the household net total the including assets, was acquired, shows rent or freely use the that households that The Figure 5A i 5B. Figures 5A of households share the hand, other On the poorest. the among are residence main net that own the HMR increases from 5% among households with the lowest of In terms highest. the assets are assets to 95% among households whose net among Figure 5B shows that, the residence, household main the ways of acquiring proportion does not have the largest of net assets, households in the lowest deciles asset deter most valuable the HMR, which comes as no surprise since this is the Joint distribution of income and net assets Joint distribution of F the survey definition, annual gross income includesemployment income, rent, income from financial assets, pensions, social transfers or any other sources of responses the in presented data of value actual the that indicates result this income, interview. the during omitted was deliberately 10 Other 9 8 CNB, 2019). 7 6 5 Net asset deciles Acquired by purchase / built 4 3

b 5 2 1 No real estate / as a gift Acquired through inheritance 0 90 80 70 60 50 40 30 20 10 igure 100 F household main of acquiring the Way in % and deciles of net assets, residence 10 9 Free use 8 Rent 7 also reflect regional disparities, so the remainder of the 6 15 5 Net asset deciles Own in part 4 3

a 2 5 Own whole 1 0 90 80 70 60 50 40 30 20 10 100 igure For more details see Christiaensen et al. (2019). 15 Other economic trends Other economic Source: ECB and author’s calculations. calculations. ECB and author’s Source: value of net assets, apart from the of the HMR value in total Given the importance of main location the HMR, the geographic status and the way of acquiring tenure residence also has a significant effect on the value of total net household assets. heterogeneity pronounced regional is known for market real estate The Croatian and significant pricedifferences depending on the geographic locationestate (for more information see and Tkalec, Vizek Žilić, of real 2018 and In other deciles, the share of these households remains relatively the same and the same relatively remains households of these the share deciles, In other on average. to 30% amounts F tenure – main residence Household in % deciles of net assets, status and paper provides an analysis of household net asset value depending on the geo- paper provides an analysis of household graphic location of a particular household.

public sector marina kunovac: economics distribution of household assets in croatia 44 (3) 265-297 (2020) 280 public sector marina kunovac: economics distribution of household assets in croatia

281 44 (3) 265-297 (2020) - 100 90 80 y t y i t d in the region l i l l a 70 o a p h i p i e c i c s i n u n 60 u o u m m –

– r a h

50 a t o i t f s

a a s o t o r e C 40 s C

s c n i a

t r t a e e i t 30 r s n

d a f o E A

s 20 e l i

t n b e 10 rc 6 Pe 0

0

90 80 70 60 50 40 30 20 10

100 Croatia in

igure

d d l o h e s u o h a r o f s t e s s a t e n f o s e l i t n e c r e F percentiles net asset of Comparison and at in the region for a household Croatia the level of P pality Eastern munici- Croatia – 5 Q Eastern Croatia – city 4 Q pality Central munici- Croatia – 3 Q – city Central Croatia 2 Q Zagreb City of 1 Q – city Coast

Adriatic a 6 pality Coast – munici- Adriatic 0

90 80 70 60 50 40 30 20 10

100 s e l i t n i u q t e s s a t e N igure The Figure 6A shows that, on the Adriatic Coast and in the City of Zagreb, over Zagreb, of City the in and Coast Adriatic on the shows that, 6A Figure The while the the 40% of the wealthiest, 50% of households can be grouped among sense, the In that 20%. than lower is Croatia Eastern in such households of share have These municipalities out. stand in Eastern Croatia poorest municipalities over 60% of households classified among 40% of those with the lowest value of among inequality of breakdown detailed More Croatia. of level the assets at net locations in Croatia is shown various geographical the in the Figure 5B, whereby of assets have net observed percentile areas below the slope of 45 degrees in each a For instance, country. of the entire asset values lower than those in the sample municipali in the value asset net of the terms in percentile the 50th in household Note: Note: Geographical location “Adriatic Coast” includes the following counties: Primorje- Šibenik-Knin, Split-Dalmatia, Istria and Lika-Senj, Dubrovnik-Neretva. Zadar, Gorski Kotar, Geographical location “Eastern includes Croatia” the following counties: Sisak-Moslavina, Požega-Slavonia, Osijek- Brod-Posavina, Karlovac, Bjelovar-Bilogora, Virovitica-Podravina, Geographical location Baranja “Central and includes Croatia” Vukovar-Srijem. the following Koprivnica-Križevci and Međimurje. Krapina-Zagorje, Varaždin, counties: Zagreb, calculations. ECB and author’s Source: value asset net the of terms in percentile 30th the in also is Croatia Eastern of ties of a municipality in household average words, an In other of Croatia. level the at Croatian average. On the other hand, a Eastern Croatia is much poorer than the geographical a in value asset net of the terms in percentile the 50th in household in Adriatic Coast is in the 65th percentile on the area comprising municipalities that conclusion draw a can One of Croatia. level the at value asset net of the terms wealthier is much Coast Adriatic on the municipality in a household an average in heterogeneity regional These results point to marked average. than the Croatian terms of the net asset value among households. F households of heterogeneity Regional assets to the value of net with regard

- - - -

HMR value among differ In line with the descriptive analysis car analysis descriptive the with line In 16 ANALYSIS OF THE MAIN DETERMINANTS OF INEQUALITIES THE MAIN DETERMINANTS OF OF ANALYSIS IN DISTRIBUTION OF NET ASSETS NET OF IN DISTRIBUTION Selection of dependent variables and methodological approach is similar to those presented in Arrondel, Arrondel, Selection of dependent variables and methodological approach is similar to those presented in 16 The purpose of this chapter is to establish the basic determinants of inequality in of inequality determinants the basic is to establish chapter The purpose of this assets. household net of distribution the 5 Descriptive statistics used in this chapter show that the value of total net house - net of total value the show that chapter used in this statistics Descriptive socio-demo the on depending households, among varies significantly assets hold and geo- real estate ownership and household income, graphic characteristics alone analysis a descriptive However, household. of a particular graphic location - house different of significance relative into insight detailed more a us give cannot house- assets among of net on distribution impact their and characteristics hold in order to the next chapter is used in model why an econometric This is holds. - ine on characteristics household different of impact the detail more in examine of net household assets. qualities in the distribution ried out in the previous chapter, the dependent variable used for measuring ine- used for measuring variable dependent the chapter, previous the in out ried for net assets that the group of net assets is a quintile distribution in the quality to with a value of 1 to 5. household is assigned several main categories: Explanatory variables are divided into distribution on inequal in the income position relative of a household The impact ity in the distribution of net assets is taken into account by in a certain 1 if a household is positioned on value take that variables dummy using a set of five took part in the second that countries of By using the sample quintile. income was no there (2014) have shown that Arrondel, Roger and Savignac HFCS wave, - in some coun respect, In that and asset distribution. income link between unique net assets, whereas in others, this tries a rise in income implies a rise in household household on the depending changes distribution asset and income between link relative position in the asset distribution. the total net asset value, as discussed inGiven the importance of HMR value for the remainder of the analysis includes more detailed informa- the previous chapter, tion on household main (2015),Piketty Zucman and (2011), Piketty . residence householdof establishing value crucial for among others, think that inheritance is in the two previous HFCS waves (ECB,net assets. In addition, the data collected role of inheritance in establishing value2013; 2016) also emphasise the important This is why the analysis also includes a dummy variable that takes on of net assets. value 1 if a household reported that they inherited the main residence or received it as a gift. In addition, given the heterogeneity in terms of - ent geographic locations in Croatia, a set of four dummy variables is created, des ignating the geographic location of household main and the household residence Mathä,Zagreb). of City the and Croatia Eastern Croatia, Central Coast, (Adriatic Roger and Savignac (2014), Martinez and Uribe (2018) and Georgokoponus (2019). Roger and Savignac (2014), Martinez and Uribe (2018) and Georgokoponus

public sector marina kunovac: economics distribution of household assets in croatia 44 (3) 265-297 (2020) 282 public sector marina kunovac: economics distribution of household assets in croatia

283 44 (3) 265-297 (2020) - - - - Another ele- 17 number of household members and according to Modigliani and Brumberg (1954). according to Modigliani and Brumberg household members and the net asset value and negative cor household members and the net asset value life-cycle hypothesis household as total , such structure - (2015) has ana and over 65). Hammer 16-34, 35-45, 45-64 age (age groups: In line with the 17 Porpiglia and Ziegelmeyer (2014) use the Oaxaca – Blinder decomposition andBlinder decomposition – use the Oaxaca (2014) and Ziegelmeyer Porpiglia the most are area countries euro prices across property in that differences show values. Socio- household net asset in differences factor for explaining important demographic characteristics of reference on the data on the households are based about house- most informed members as the person by household person chosen hold finances. The analysis includes a set of dummy variables for reference per son’s son’s lysed age-specific householdbalance sheet depending on the reference person’s referencethe with proportion in increase assets household that shown has and age reference person retires. age and slightly decreases once the person’s Socio-demographic characteristics of households include dummy variables that variables dummy of households include characteristics Socio-demographic describe . A dummy variable that takes dummy variable that A sex. is the reference person’s ment used in the analysis research (Sier person is male was thus created. Previous on value 1 if the reference per reference whose households that shown has 2017) Grabka, and Frick minska, person’s reference the of effect The value. asset net higher have may man a is son referenceusing three dummy variables classifying education was examined by or with no education, persons with primary education persons into three groups: of the The effect education and those with higher education. those with secondary labour reference person’s market status on inequality in the distribution of net variables assigning reference personsassets is measured by using a set of dummy employed, retired and unem- into one of the following categories: self-employed, points to the fact that inequality in distribution of ployed or non-active. Lise (2011) with unemployed and non-active by the labour market status, assets is affected position.individuals in a significantly disadvantaged number of dependent between , because we expect positive correlation children the total number of In addition, the analysis includes other household characteristics, such as the such characteristics, household other includes analysis the addition, In dummy variable measuring household willingness to finan- take significant or substantial take to willingness reported takes on risks (that household a if 1 value variable and the dummy decisions) savings or investment risks when making cial equals 1 if a household receives that some kind of social . Fessler and benefits the state (1995) show that social services provided by Schürz (2015) and Jappelli some receive that households and accumulation wealth for private substitutes are kind of social benefits have substantially lower net asset values compared with other households. relation in terms of number of dependent children in the household. Fessler, children in the household. Fessler, relation in terms of number of dependent analysis of net assets should appropriate Linder and Segalla (2014) show that structure. include variables to control for household - - - - - J} (2) ..., 18 J} indexes ..., household income

j ϵ {1,2, J – 1} defined as: * (1) i i I + ε j j = {2,..., β i = X for

* i I

j ≤ K * i

1 j–1 ≤ I ≤ K ≥ K j–1 can assume the values defined within the set {1,2, * * i i i I K I I if = j if = J if = 1 i i i I I I represents estimated cut-off values that position each household in the household in each position that values cut-off represents estimated j = and sample the in households the 1, ..., n indexes K i For the discussion on the effect of debt and different types of social benefits on inequality in the distribu- types of social benefits on inequality of debt and different For the discussion on the effect where

tion of assets, see Maestri, Bogliacino and Salverda (2014). 18 where 5.1 METHODOLOGY – THE GENERALIZED ORDERED PROBIT MODEL THE GENERALIZED ORDERED PROBIT – 5.1 METHODOLOGY The estimation. in the econometric ordered probit model use the generalized We model is based on a latent dependent variable Finally, two dummy variables were used to examine the role of debt in accumula in of debt role the examine to used were variables two dummy Finally, the with collateralised loans mortgage with for households assets of net tion with consumer debt. and households main residence (HMR) household A8 A7 and A6, Tables Appendix variables, the the above-mentioned Apart from of variables. selection alternative analysis with results of the robustness shows the 3) EUR 1,300 (Chapter not exceed did income whose annual Since households esti- the presents paper the sample, the in share high relatively a for accounted EUR 1,300. By using below income with households without model of the mates in to position each household values were estimated cut-off the new sample, the robustness anal The additional assets quintiles. and net the corresponding income ysis was carried out by using an alternative specification of equiva OECD the to according members of household number for the adjusted are in line with The results of the robustness analysis 2011). lence scale (OECD, of the construction the explains below text The model. main results of the the var of explanatory set above-mentioned of the effects the estimate used to model position in distribution of net assets. iables on a household egories reflecting net asset quintile is defined as follows: corresponding net asset quintile. that a household i is positioned in one of the J cat In that respect, the probability . The observable variable is defined as follows: The observable where J = 5. the categories of the probit model. variable Observable

public sector marina kunovac: economics distribution of household assets in croatia 44 (3) 265-297 (2020) 284 public sector marina kunovac: economics distribution of household assets in croatia

285 44 (3) 265-297 (2020) - - - - (3) categories of the j categories J – 1} vary between j j = {2,..., β for ) values of the dependent variable catego variable dependent of the j values j–1 β i ) j–1 β i ) 1 ) – F(X j β β i i equal for each of for each equal

j β = 1) = F(X = 1) = j) = F(X F(X = J) = 1 – i i i Pr (I Pr (I Pr (I ries, the end result would be the ordered probit model. The ordered probit model probit ordered The model. probit the ordered would be result end the ries, on the dependent variable. variables independent of all effect would assume linear of posi- probability the to establish would be equal effect the income For instance, Wald the Since quintile. asset net fifth and second the both in household a tioning of the test parallel line the for assumption homogeneity the rejects assumption probit ordered generalized to use the appropriate is more it parameters, estimated (test results shown analysis in the model probit ordered the as opposed to model Also, the same tables presents the results A8). A7 and A6, Tables Appendix in in the dependent (log) as the value asset net with estimate squares method least of the regression also assumes linear effect variable. On the other hand, the use of linear variable, but the previous univariate of all independent variables on the dependent sub- analysis has shown household characteristics that the structure of relevant of net on the household position in the distribution changes depending stantially as the probit model was selected ordered assets. For these reasons, the generalized independ of effect for non-linear allows it because tool methodological primary categories of the dependent variable. ent variable depending on different dependent variable (net asset quintiles for households). If we want to have esti- to have households). If we want for asset quintiles (net variable dependent mated coefficients The generalized ordered probit model is selected as opposed to the ordered probit probit ordered the to opposed as selected is model probit ordered generalized The model as a preferred model specification because categories subject to different of the independent variables on the dependent effect it allows for heterogeneous why in is This 2010). Hensher, and Green 2006; (Williams, variable dependent of parameters the estimated model the generalized equals a normal cumulative distribution function. function. cumulative distribution where F equals a normal tory variables, the marginal effect in accordance with the mean value of explana with the mean value in accordance effect the marginal tory variables, tory variables is thus presented. The text below presents the main results of the estimate of the generalized ordered generalized of the estimate of the results main the presents below text The probit model as defined in the equation (3). Since the regression coefficients can- 3 presents Table the model, probit the within effects as marginal interpreted be not of the unit change of the explanatory that show the effect effects the marginal asset net particular a in household a of positioning probability on the variables of explana values for different differ may effect marginal the Given that quintile. 5.2 MAIN RESULTS 5.2 MAIN RESULTS - 5 0.19*** 0.18*** 0.08*** 0.02 0.09*** 0.00 0.24*** 0.12* 0.20*** 0.05 0.11** 0.12** 0.02* 0.10** -0.20*** -0.28*** -0.22*** -0.08** -0.02 -0.01 -0.05 -0.03 -0.08* 4 0.00 0.01 0.07* 0.09*** 0.03 0.00 0.08 0.06 0.08 0.05 0.01 0.05 0.03** 0.03 0.04 0.02 -0.05 -0.06 -0.03 -0.02 -0.01 -0.05*** -0.02 3 0.05 0.01 0.05* 0.02 0.03 0.05 0.04 0.07 0.04 0.00 0.00 0.04 -0.04 -0.05 -0.05 -0.06** -0.08*** -0.03 -0.06 -0.14 -0.03 -0.05 -0.06 Net asset quntiles 2 0.14*** 0.13*** 0.11*** 0.05 0.07*** 0.04* 0.02 0.06 0.13** 0.00 0.07 0.02 0.11** 0.02 -0.21*** -0.24*** -0.03 -0.09** -0.06 -0.01 -0.02* -0.03 -0.08 1 0.00 0.02 0.00 0.05 0.05*** 0.07*** 0.07** 0.01 0.14*** 0.16*** 0.10*** 0.11*** -0.11*** -0.02 -0.01 -0.24*** -0.14*** -0.19*** -0.13*** -0.18*** -0.27*** -0.03*** -0.08** -

1 2 3 4 Inheritance_HMR HMR_City of Zagreb HMR_Primorje HMR_Central Croatia Sex (male) Retirement Unemployed/ non-active Self-employed Secondary education Higher education 35-45 age group 45-64 age group 65+ age group Number of chil dren in household Number of house­hold members Mortgage for HMR Consumer debt Social benefits recipients Willingness to Willingness take risks

- - - - 3 able Income per Income per quintiles HMR charac- teristics (inheritance and location) graphic char Socio-demo acteristics of reference person Household characteris tics Indicators of debt burden Other charac teristics The main results are presented in the text below. The main results are presented in the text below. the in position household a between link the confirmed has analysis empirical The for income effects marginal estimated The assets. and net of income distribution * Generalized ordered probit model, marginal effects model, marginal probit * Generalized ordered respectively. 90% and 95% 99%, of significance statistical indicate * and ** ***, Symbols Note: loca _HMR – location HMR the for group Income_5quintile – income for categories: Reference tion Eastern Croatia; for labour market status – Employed; for educational attainment – Primary – attainment educational for Employed; – status market labour for Croatia; Eastern tion education or no education, for age – Up to 34 years of age. calculations. author’s ECB and Source: T quintile* net asset in a certain household a of positioning Probability

public sector marina kunovac: economics distribution of household assets in croatia 44 (3) 265-297 (2020) 286 public sector marina kunovac: economics distribution of household assets in croatia

287 44 (3) 265-297 (2020) - - - - households in the lowest net asset quintile asset net lowest the households in

19 Other research has also backed this result (e.g. Arrondel, Roger and Savignac, 2014; Leitner, 2015). Arrondel, Roger and Savignac, 2014; Leitner, Other research has also backed this result (e.g. 19 mostly have the expected sign. Low income households are more likely to be in be to likely more are households income sign. Low expected the have mostly household a For instance, quintiles. asset net higher in be to likely less and lower asset net first the in positioned be to likely more 14% is quintile income first the in in quintile, asset net (highest) fifth the in be to likely less 20% and well as quintile comparison with a household in the fifthincome quintile. Regardless of a certain connection between income and net asset value, statistical significance and link of of the distribution position in terms on a household depending vary, intensity This leads us only in distribution of assets can to conclude that inequality assets. some other that among households and levels to the income be attributed partially in position on a household impact a relevant have may from income, apart factors, assets. the distribution of net The location of HMR is extremely important to establish a household position in position a household to establish important of HMR is extremely location The households with the All other factors being constant, of net assets. the distribution are much Coast or in the City of Zagreb Adriatic on the residence household main more and quintiles) third and (second quintiles asset net lower in be to less likely with compared quintiles) fifth and (fourth quintiles asset net higher in be to likely should be It category. reference up the make that Croatia Eastern households in of location of geographic effect important the of irrespective that, however, noted, distribution of net assets, a location does the HMR on a household position in the not have a significant effect on the likelihood of positioning a household in anal descriptive previous the the with line in is This quintile. assets net (first) poorest par only status market labour the characteristics, socio-demographic of terms In hood of positioning the household in the lowest net asset quintile (by 11%) and quintile (by 11%) the household in the lowest net asset hood of positioning quintiles asset net higher somewhat in it positioning of likelihood the increases main resi- the other hand, the inherited (second and third net asset quintile). On quin- in the wealthiest of having a household the likelihood does not affect dence tiles because the results for the fourth and fifth net asset quintile are not statisti cally significant for this variable. of majority the that has shown ysis that do not even own the HMR. net asset quin- in a certain of positioning a household the probability affects tially to be are 24% less likely persons households with self-employed For instance, tile. net asset quin- to be in the wealthiest and 23% more likely in the poorest quintile This is in line person. reference an employed with households with compared tile with the findings of the descriptive analysis that among households. of its distribution business assets and inequality employment point to high values of self- status, the results have the expected of labour market In terms of other categories The results of the estimated model show that information on the household main household the on information show that model estimated the of results The residence (HMR) remains significant when establishing a household position in HMR assets. In that respect, inherited the distribution of net decreases the likeli - - . vice versa 6 CONCLUSION to paper this in used was (HFCS) survey consumption and finance household The The household net assets and its main components. of the distribution analyse Croatian assets among of net distribution in the inequality moderate reveal results households. Inequality in possession of financial assets measuredcoefficient is more pronounced than by incase of realthe Gini assets because only a certain portion of households own substantial financial households’ financial assets. assets stands The at EUR 500. median Real assets value account for of a large household main assets and 85% of households own the total share of households’ 66 assets with net total of households’ bulk up the makes (HMR) that residence assets consider net household The value of total value. thousand euros in median and demo- social different on their ably varies among households depending graphic characteristics, income, real asset ownership and geographic location. of numerous factors interaction that model indicate results of the econometric The sense, the that assets. In of net distribution the in position household a affects importance of HMR stands out. The main residence represents the most signifi- it was in terms of the way cant component of value of net assets, especially Households with inherited HMR are less likely acquired and geographic location. households with HMR In addition, in the lowest net asset quintile. to be positioned sign but they are not statistically significant. The only exception arehouseholds net the highest to be in more likely are 9% that person reference a retired with person. an employed reference households with in comparison with asset quintile age of the reference and older attainment with higher educational Households For example, in higher asset quintile. to be positioned more likely persons are (19% are education those with secondary persons and reference highly educated 12% and (20% and quintile poorest the in be to likely less 14% respectively) and with compared net asset quintile to be in the highest more likely respectively) older with Households education. no or education primary persons with reference net asset quintile are much less likely to be in the poorest reference persons (+65) with house- in comparison net asset quintile in the wealthiest to be and more likely though somewhat effect, similar A is below 35. person holds whose reference reference aged middle for households with observed was also moderate, more of the importance The analysis has also emphasised group). person (45-64 age of net assets with for its position in the distribution household characteristics more likely to be in children and fewer household members households with more poorer net asset quintiles and willing reported perceived the results have shown that households that Finally, ness to take financial risks are more likely to be in the highest net asset quintile more are households indebted and benefits social receive that households whereas debt consumer with households (whereby quintiles asset net poorer in be to likely quintile). are more likely to be in the lowest net asset

public sector marina kunovac: economics distribution of household assets in croatia 44 (3) 265-297 (2020) 288 public sector marina kunovac: economics distribution of household assets in croatia

289 44 (3) 265-297 (2020) - having a

Households with higher income and educational attainment and older age of the attainment and older and educational with higher income Households The quintiles. in highest net asset to be positioned persons are more likely reference same effect was found for financial risktaking and self-businesses(self-employ - of to the likelihood were linked these characteristics ment) because in the City of Zagreb or on the Adriatic Coast are more likely to be in higher asset be in higher likely to are more Coast Adriatic the or on City of Zagreb in the groups. quintile In that sense, this paper represents the first attemptto estimate inequality in the As a it. factors affecting in Croatia and the accompanying distribution of assets other survey and admin with data the collected to compare crucial is follow-up, it statement Disclosure by the author. No potential conflict of interest was reported household in highest net asset quintiles. On the other hand, households with more net asset quintiles. On the other hand, household in highest children, those that receive social benefits and those with net asset quintiles. are most likely to be in the poorest younger reference person poorly educated and - col CBS) and continue with systematic (tax administration, data sources istrative distribution the in inequality of terms in research further and data assets of lection type in Croatia. there are currently few analyses of this of assets given that - - NBER ECB Working Working ECB , pp. 27-43. , Q4(18), pp. 36-66. ECB Working ECB Paper, Working , No. 2. Zagreb: Croatian Bureau of Bureau Zagreb: Croatian . Eurosystem Household Finance Household Eurosystem Monetary Policy & the Economy, Monetary Policy & the Economy, , No. 18. . December 2016. , WP 3-2019. , WP ECB Statistic Paper Series . ECB Statistic Paper The Household Finance and Consumption Survey, Wave 2, Core 2, Core Wave The Household Finance and Consumption Survey, Statistics on Income and Living Conditions (SILC), 2016 Indica- Arrondel, L., Roger, M. and Savignac, F., 2014. Wealth and income in the euro in income and Wealth 2014. F., Savignac, M. and L., Roger, Arrondel, , No. 1709. Paper ECB Working behaviour. in households’ area Heterogenity channel. redistributional the and policy 2017. Monetary A., Auclert, policy 2017. Macroprudential Kerm, P., Van Olivera, J. and J. F., Carpantier, Society for inequality. and household wealth the Study of Economic Inequal- 2017 – 442. WP Paper, ECINEQ Working ity, CBS, 2017. Nationalbank (Austrian Central Bank) Oesterreichische Area Euro Across the Wealth 2014. Net E., Segalla, and P. Linder, P., Fessler, It. for Control How to and Matters Structure Household Why – Paper, No. 1663. Countries: across European Wealth and Schürz, M., 2015. Private P. Fessler, State. Welfare and the The role of income, Inheritance No. 1847. from Micro in Malta: Evidence Wealth I., 2019. Income and Georgakopoulos, Data. Central Bank of Malta tors of Poverty and Social Exclusion in 2016. No. 23451. https://doi.org/10.3386/w23451 No. 23451. Paper, Working Improv- Rich: the of Weight The 2018. M., Morgan, and I. Flores, T., Blanchet, . 2018/12 Paper, Working Data. WID.world Tax Using ing Surveys gener effects random estimate to module Stata Boes, S., 2006. REGOPROB: - Components S456604. Bos Software Statistical models. probit ordered alized Department of Economics ton: Boston College Statistics Sub- A – Croatia in Slavonia, Christiaensen, L. [et al.], 2019. Jobs Challenges Bank Paper Group, Jobs World Assessment. Working Labor Market national No. 35. https://doi.org/10.1596/32300 , No. 16. Zagreb: Croatian National Bank. CNB, 2016. Financijska stabilnost , No. 20. Zagreb: Croatian National Bank. CNB, 2019. Financijska stabilnost National Croatian Zagreb: , 9. dijagnostika 2019a Makroprudencijalna CNB, Bank. initial Belgium: in wealth of household distribution The 2016. P., Du Caju, Sur Consumption and Finance Household the of wave second the of findings Economic Review vey (HFCS). National Bank of Belgium, The Household Finance and Consumption Survey: results from ECB, 2016. the second wave M., 2019. Schurz and P. Lindner P., Fessler Austria. Consumption Survey 2017 for and ECB, 2013. The eurosystem household financeResults from the first wave. and consumption survey, ECB, 2016a and derived variables catalog LITERATURE 1. 2. 5. 6. 16. 17. 18. 3. 4. 7. 8. 9. 10. 11. 13. 15. 12. 14.

public sector marina kunovac: economics distribution of household assets in croatia 44 (3) 265-297 (2020) 290 public sector marina kunovac: economics distribution of household assets in croatia

291 44 (3) 265-297 (2020) - Emerg , 49, pp. https://doi.org/10.1093/

Ageing Europe, Working , WP 122. , WP Ricerche Ricerche Economiche Working Working Papers Central Bank of . Determinants of Household Position within Position Household of Determinants Anketa o financiranju i kućanstava. potrošnji https://doi.org/10.1017/CBO9780511845062 ECB Working Paper, 1690. https://doi.org/10.2139/ Working ECB IZA Discussion Paper, No. 9139. https://doi.org/10.2139/ssrn.2562315 Discussion IZA , No. 827. Modeling Modeling D., 2010. and Hensher, W. Ordered Green, . Choices New York: Press. University Cambridge Hammer, Hammer, B., 2015. The ownership of assets and the role of Agespecific age: countries. area euro for sheets balance household . No. 9 Paper, Vulnerability Financial Cluster-Based New A 2014. D., Nestić, and I. Herceg, in Croatia. Testing to Household Stress Application Its and Indicator ing markets finance and https://doi.org/10.2753/ trade, 50(5), pp. 60-77. REE1540-496X500504 Enlisting Croatia: in Household of Anatomy The 2011. V., Šošić, and I. Herceg, ComparativeHouseholds or Relaxing Lending Standards? More Creditworthy https://doi.org/10.1057/ces.2011.14 Economic Studies, 53, pp. 199-221. of private reduce the accumulation 1995. Does social security T., Jappelli, survey data. from Italian Evidence wealth? acprof:oso/9780199687435.001.0001 2018. F., Uribe, and F. Martinez, Distribution Household’s Wealth Chilean Chile in M., 2014. Household wealth A. and Ziegelmeyer, J., Porpiglia, T. Mathä, transfers, home ownership of intergenerational The importance the euro area: dynamics. house price and ssrn.2487119 of Estonian and wealth The assets, liabilities 2016. T., Merikull, J. and Room, households: Results of the Household Finance and Consumption Survey. Paper, 3-2016. https://doi.org/10.23656/25045520/32 Occasional Pank, Eesti 016/0007 R., 1954. Utility analysis and the Consumption and Brumberg, Modigliani, F. Function: an Interpretation of Cross-section Data. In: K. Kurihara. Post-Keynes- ian Economics. New Brunswick: Rutgers University Press, pp. 388-436. Lise, J., 2011. On-the-Job Search and Precautionary Savings: Theory and Theory Savings: Precautionary and Search On-the-Job J., 2011. Lise, Institute for Inequality. Fiscal Studies, WP Wealth and of Earnings Empirics https://doi.org/10.1920/wp.ifs.2011.1116 11-2016. and the Inequality Wealth 2014. W., and Salverda, F. Bogliacino, V., Maestri, Rich in Inequalities Salverda [et al.]. Changing W. of Debt. In: Accumulation Perspectives. Comparative and Analytical Countries: 1-31. https://doi.org/10.1016/0035-5054(95)90008-X Jemrić, I., I. 2019. and Vrbanc, In preparation. Kontbay-Busun, S. and Peichl, A., Cross-Country Comparison Using the A in the Eurozone: 2015. Wealth Income and Multidimensional Affluence in HFCS. Area Countries. The in Euro Inequality Wealth S., 2015. Drivers of Leitner, Institute for International Economic Studies Vienna

19. 20. 21. 22. 23. 29. 30. 31. 32. 28. 27. 24. 25. 26. ECB ECB Radni . Privredna Privredna , 11. The Stata Journal, 6, pp. 58-82. Pregled tržišta nekretnina Republike nekretnina tržišta Pregled https://doi.org/ 59-116. , 26(2) (141), pp. , No. 1. Handbook of Income Distribution. Volume 2B. North Holland, Volume Distribution. Income of Handbook The Quarterly Journal of https:// pp. 1071-1131. Economics , 3(126), Nestić, Nestić, D., 2005. Income Distribution in Croatia: What Do pp. 39-53. , 29(1), and practice theory Us? Financial the Tell Data Survey Budget Household France 1820 – Evolution of Inheritance: On the Long-Run 2011. T., Piketty, 2050. doi.org/10.1093/qje/qjr020 House- of Croatian Analysis Micro-level A and Zauder K., 2019. Rosan, M. In preparation. Debt Participation. holds’ u Hrvatskoj: uloga raz- u siromaštvu na razini regija Rubil, I., 2013. Razlike lika u prosječnim dohocima i nejednakostima distribucije dohodaka. materijali EIZ-a kretanja i ekonomska politika 10.15179/pkiep.26.2.2 between of wealth The distribution Medgyesi M., 2013. Sierminska, E. and Note Commission Research households. European gender the Examining M. M., 2010. Grabka, and J. R. Frick E., Sierminska, wealth gap. Oxford Economic , 62(4), pp. 669-690. https://doi.org/ Papers 10.1093/oep/gpq007 2018. I., Žilić, and M. Vizek, M., Tkalec, Paper, No. 1692. Working of the wealth distribution the top tail 2016. Estimating P., Vermeulen, Paper, No. 1907. Working Odds Proportional Logit/partial Ordered Generalized R., 2006. Williams, Variables. Models for Ordinal Dependent https://doi.org/10.1177/1536867X0600600104 postoji stanarsko pravo, NN Zakon o prodaji stanova nad kojima 27/91. Zagreb: Narodne novine. Zagreb: Narodne novine. 152/14. Zakon o socijalnoj skrbi, NN in the Long Run. and Inheritance Wealth 2015. T., G. and Piketty, Zucman, 15. in: Chapter Rubil, I., Stubbs, P. and Zrinščak, S., 2018. Child Poverty Study. Quantitative-Qualitative A Croatia: in Coping Strategies and Household pp. 1303-1368. Hrvatske 2012.-2017. Zagreb: Institute of Economics. households. area euro of exposure rate interest The 2019. P., Tzamourani, Paper, No. 1-2019. https://doi.org/10.2139/ Discussion Bundesbank Deutsche ssrn.3209226 distribution. of the wealth is the top tail 2014. How fat P., Vermeulen, 33. 34. 35. 36. 38. 39. 40. 43. 44. 45. 46. 47. 37. 41. 42.

public sector marina kunovac: economics distribution of household assets in croatia 44 (3) 265-297 (2020) 292 public sector marina kunovac: economics distribution of household assets in croatia

293 44 (3) 265-297 (2020) 26 57 17 26 41 14 19 38 4 31 27 90 10 2.8 EU SILC 24 59 16 25 37 17 21 45 3 42 9 85 15 2.8 HFCS Secondary Higher Under 24 25-54 55-64 65+ Employed Self-employed Retirement Non-active/unemployed Owners Renters Primary A4

able Age Labour market status size (no. of individuals) household Average residence Information on main Education Information on the reference person on the reference Information * Comparison of the Household Finance and Consumption Survey (HFCS) and the EU Statistics EU the and (HFCS) Survey Consumption and Finance Household the of Comparison * 2016. on Income and Living Conditions (EU SILC) for Note: Information person on for the the reference HFCS was calculated by using the estimated weights and all five versions of data imputations. Information persons on for the the reference data. the Eurostat EU-SILC was taken from (EU SILC). ECB and Eurostat Source: Comparison of socio-demographic characteristics of households* characteristics of socio-demographic Comparison APPENDIX T 0.59 0.56 0.93 0.69 0.99 0.99 0.88 0.83 0.96 0.99 0.99 0.99 0.98 0.99 0.96 0.90 0.96 0.89 0.51 0.59 0.61 Gini coefficient A5 able Real assets Main residence Other real estate Vehicles Other valuables assets Self-employment business Financial assets Current accounts Savings accounts insurance pension/life Voluntary Mutual funds Money owed to household Shares Bonds assets Other types of financial Liabilities Mortgage debt Non-mortgage debt annual gross income Total assets Gross Net assets Components of net assets Components Source: ECB and author’s calculations. ECB and author’s Source: T assets financial real and of for subcomponents coefficient Gini

public sector marina kunovac: economics distribution of household assets in croatia 44 (3) 265-297 (2020) 294 public sector marina kunovac: economics distribution of household assets in croatia

295 44 (3) 265-297 (2020) 5 0.21*** 0.18*** 0.08*** 0.02 0.08** 0.23*** 0.10 0.18*** 0.06 0.12** 0.13** 0.05*** 0.09* -0.18*** -0.22*** -0.18*** -0.10*** -0.02 -0.01 -0.02 -0.06 -0.03 -0.07* 4 0.01 0.06* 0.09*** 0.02 0.01 0.08 0.06 0.07 0.05 0.01 0.05 0.03*** 0.03 0.03 0.01 -0.06 -0.07* -0.05 -0.05 -0.02 -0.01 -0.05*** -0.03 3 0.01 0.03 0.03 0.04* 0.02 0.03 0.03 0.04 0.06 0.04 0.00 0.05 -0.01 -0.04 -0.07*** -0.08** -0.03 -0.05 -0.11 -0.01 -0.04* -0.05 -0.05 Net asset quintiles 2 0.13*** 0.12*** 0.10** 0.05 0.08*** 0.04* 0.03 0.06 0.06 0.03 0.10** 0.02 -0.22*** -0.24*** -0.03 -0.08* -0.06 -0.11** -0.02 -0.01 -0.03*** -0.02 -0.07 1 0.13*** 0.16*** 0.10*** 0.06 0.00 0.03 0.00 0.04 0.05*** 0.07*** 0.07** 0.02 -0.11*** -0.01 -0.02 -0.26*** -0.14*** -0.17*** -0.13*** -0.18*** -0.27*** -0.04*** -0.08* 1 2 3 4 Inheritance_ HMR HMR_City of Zagreb HMR_Primorje HMR_Central Croatia Sex (male) Retirement Unemployed/ non-active Self-employed Secondary education Higher education 35-45 age group 45-64 age group 65+ age group Number of children in household Number of household members Mortgage for HMR Consumer debt Social benefits recipients Willingness to Willingness take risks - - - - A6 able Income per Income per quintiles (OECD) HMR charac- teristics (inheritance and location) son Socio-demo graphic char acteristics of reference per Household characteristics Indicators of debt burden teristics Other charac * Generalized ordered probit model, marginal effects, income quintiles adjusted for the number model, marginal probit * Generalized ordered scale to the OECD equivalence of household members according respectively. 90% and 95% 99%, of significance statistical indicate * and ** ***, Symbols Note: loca- _HMR – location HMR the for group Income_5quntile – income for categories: Reference Primary – attainment educational for Employed; – status market labour for Croatia; Eastern tion education or no education, for age – Up to 34 years of age. calculations. author’s ECB and Source: T quintile* net asset in a certain household a of positioning Probability 5 0.20*** 0.18*** 0.06** 0.02 0.08** 0.25*** 0.15 0.22* 0.06 0.12** 0.13** 0.02* 0.08 -0.26*** -0.25*** -0.18*** -0.09* -0.02 -0.01 -0.02 -0.06 -0.03 -0.07 4 0.00 0.06* 0.11*** 0.06* 0.00 0.04 0.01 0.02 0.04 0.03 0.05 0.03** 0.02 0.05 0.06 -0.08 -0.03 -0.04 -0.02 -0.03 -0.01 -0.05** -0.04 3 0.04 0.05 0.08 0.00 0.04 0.03 0.00 0.03 -0.01 -0.03 -0.01 -0.05 -0.06 -0.06* -0.08** -0.03 -0.01 -0.01 -0.11 -0.01 -0.03 -0.08* -0.03 Net asset quintiles 2 0.07 0.06 0.09** 0.02 0.12*** 0.04 0.02 0.79 0.06 0.02 0.11* 0.04* 0.14*** 0.03 -0.25*** -0.27*** -0.04 -0.03 -0.08 -0.20*** -0.04** -0.04 -0.07 1 0.28*** 0.18*** 0.14*** 0.13*** 0.05 0.04 0.00 0.01 0.04** 0.08*** 0.07** -0.15*** -0.01 -0.02 -0.97 -0.13*** -0.12** -0.16*** -0.20*** -0.32*** -0.02 -0.07* -0.04 1 2 3 4 Inheritance_ HMR HMR_City of Zagreb HMR_Primorje HMR_Central Croatia Sex (male) Retirement Unemployed/ non-active Self-employed Secondary education Higher education 35-45 age group 45-64 age group 65+ age group Number of children in household Number of household members Mortgage for HMR Consumer debt Social benefits recipients Willingness to Willingness take risks - - - -

A7 able Income per Income per quintiles > EUR 1,300 HMR charac- teristics (inher itance and location) graphic char Socio-demo acteristics of reference person Household characteristics Indicators of debt burden Other charac teristics - report that households the excludes sample effects, marginal model, probit ordered Generalized * quin- asset net and income accompanying the and 1,300 EUR than lower income gross annual ed tiles modified accordingly respectively. 90% and 95% 99%, of significance statistical indicate * and ** ***, Symbols Note: loca- _HMR – location HMR the for group Income_5quntile – income for categories: Reference Primary – attainment educational for Employed; – status market labour for Croatia; Eastern tion education or no education, for age – Up to 34 years of age. calculations. author’s ECB and Source: T quintile* net asset in a certain household a of positioning Probability

public sector marina kunovac: economics distribution of household assets in croatia 44 (3) 265-297 (2020) 296 public sector marina kunovac: economics distribution of household assets in croatia

297 44 (3) 265-297 (2020) - 0.22*** 0.02 0.16 0.000 0.15** 0.61*** 0.53*** 0.15* 0.03 0.29*** 0.97*** 0.50*** 0.75*** 0.48*** 0.69*** 0.81*** -0.18*** -0.17** -0.23** -0.12 -0.71*** -0.88*** -0.64*** -0.39*** Ordered (Model 3) 219.84 probit model probit OLS 1.21*** 0.26*** 0.24 0.18 0.54* 3.96 0.04*** 0.55*** 0.54*** 0.40** 0.24 0.07 0.23 1.02*** 0.77*** 1.03*** -0.25** -0.52** -0.39*** -0.42* (Model 2) OLS 0.33*** 0.24 0.15 0.37 8.44 0.55*** 0.50** 0.38** 0.23 0.06 0.48** 1.03*** 0.59** 0.81** 0.75*** 0.97*** 1.06*** -0.27** -0.57** -0.45** -0.78*** -1.16*** -0.69*** -0.41** (Model 1) Number of children in household Number of household members Mortgage for HMR Consumer debt Social benefits recipients to take risks Willingness Constant chi2 Prob>chi2 Inheritance_HMR HMR_City of Zagreb HMR_Primorje HMR_Central Croatia Retirement Unemployed/non-active Self-employed Secondary education Higher education 35-45 age group Sex (male) 45-64 age group 65+ age group log (age) 1 2 3 4 log (annual gross income) log (annual gross income)^2 A8 able Household characteristics Indicators of debt burden Other characteristics thest for Wald homogeneity of coefficients HMR characteristics (inheritance and location) Socio- demographic characteristics of reference person Income per Income per quintiles (OECD) ent categories of the dependent variable). The of results the Model 3 show the estimated model coefficients rather also than effects. shows marginal The test the of Table that results the Wald assumption (parallel line assumption test). the homogeneity rejects calculations. author’s ECB and Source: Note: The Model 2 of the least squares estimator includes the income square due to the non-line- the to due square income the includes estimator squares least the of 2 Model The Note: differ for coefficient (homogeneous model probit ordered an is 3 Model The income. of effect ar T probit ordered 2) and 1 and (Models estimator squares Least