Consumption Analysis Based On Distributional National Accounts1
We construct a distributional national accounts dataset for Slovakia, which contains distributional information fully consistent with the national accounts. An analysis using the data finds that to maintain robust consumption growth amid downward pressure from unfavorable demographic dynamics, an increase in consumption per person, primarily driven by income, will be necessary. This underscores the importance of policy measures to raise income growth, such as increasing productivity and encouraging labor force participation, especially among the elderly. The analysis also shows that inequality, as well as the gender gap, has declined owing to a labor shift towards high-skill jobs and fewer unemployed household heads. To maintain these favorable trends, policies to encourage a shift towards higher skilled jobs and labor force participation, such as strengthening active labor market policies and addressing skill mismatches, will be important.
A. Introduction
1. The aim of this paper is to gain distributional insights into household consumption in line with the national accounts by constructing a distributional national accounts dataset. National accounts (NA) provide comprehensive information about economic activities, including GDP, consumption, income, and wealth. However, NA can only describe the behavior of aggregated households and lacks insights into household heterogeneity. On the other hand, microdata such as household surveys contain various household characteristics but are not necessarily consistent with NA data. The distributional national accounts (DNA) combine the advantages of both NA and microdata, enabling us to conduct distributional analysis of household consumption as well as income and wealth in a manner fully consistent with NA. This means that DNA totals (the sum of individual households) are fully aligned with NA data.
2. There is increasing interest in DNA datasets. Research to construct DNA has been undertaken by various bodies. Eurostat (2024) released experimental DNA data for household consumption and income, including data for Slovakia. The ECB also publishes household distributional wealth data called the Distributional Wealth Accounts (DWA) that are consistent with the aggregates of Quarterly Sector Accounts (QSA).2 However, the Eurostat experimental data is missing several important elements important for conducting macroeconomic analysis, such as real values of the variables in the dataset and a breakdown among categories of households. To address this gap, this paper estimated the DNA of household consumption, income, and wealth in Slovakia by employing microdata from the Household Budget Survey (HBS) and the Household Finance and Consumption Survey (HFCS). Given the availability of microdata, DNA data are estimated for 2010, 2015, and 2020. Technical details regarding construction of the DNA can be found in the Appendix.
B. Demographic Analysis Using DNA
3. This paper uses the DNA dataset for Slovakia to analyze the impact of demographic change on consumption. NA data show that Slovakia’s real consumption stagnated during 2010-15 (with only a 0.2 percent compounded annual growth rate), but consumption, as well as income and wealth, grew steadily during 2015-20 (with a 3.2 percent compounded annual growth rate) (Figure 1). Using the DNA data, these growth rates can be decomposed into the following four elements (contributions).
Consumption, Income, and Wealth Growth
(Compound Annual Growth Rate, in percent, percentage points)
Citation: IMF Staff Country Reports 2025, 073; 10.5089/9798229005852.002.A001
Note: All deflated by Household Consumption DeflatorSources: Eurostat and IMF staff estimation.Composition: Contributions due to changes in household share (structure). For example, aggregate consumption will be lower if the share of household categories with lower consumption levels increases. The DNA data allows us to differentiate households according to household type and age of the household head.3
Number of households: Contributions due to changes in the total number of households. Aggregate consumption levels will be higher if a country has more households.
Household size: Contributions due to changes in the number of persons per household. The level of household consumption will be higher if the average household has more persons.4
Real consumption (income/wealth) per person: Contributions due to changes in real consumption (income/wealth) per person (equivalence scale).
In other words, the DNA data allow us to evaluate how total consumption changed as a result of changes to these four factors changes while keeping all other factors fixed.
4. Household composition changes due to aging exerted downward pressure on consumption growth. Figure 2 (left) illustrates the changes in household shares with respect to household type and age of the household head from 2010 to 2020. The chart indicates that over this time period the share of single adult households (aged 60+) increased the most (by 5 percentage points (ppt)), while that of two adults with dependent children (aged 30-59) decreased the most (by 5.4 ppt). Figure 2 (right) compares the changes in household share and the consumption level per household (note that the orange dots, representing changes in household share, are the same across both the right and left charts). Figure 2 (right) suggest that changes in household share and the consumption level per household are negatively correlated.5 For example, the share of single adults (aged 60+) has increased, and the consumption level among this group is the lowest among all household categories. Similarly, the share of two adults with dependent children (aged 30-59) has decreased, and the consumption level among this category of households is relatively higher. This implies that the changes in household composition contributed negatively to consumption growth between 2010-20 and will continue to do so unless this trend changes.6
Household Share and Consumption Level (by Household Type and Age)
Citation: IMF Staff Country Reports 2025, 073; 10.5089/9798229005852.002.A001
Note: w/dep. = with dependent children. Age of household head = Y: less than 30, P: 30-44, M: 45-59, S:60 and overSources: HBS and IMF staff estimation.5. Consumption in Slovakia benefited from an increase in the total number of households, although this is no longer the case after 2022. Until 2022, the total number of households in Slovakia had been increasing, along with the population (Figure 3, left).7 Other things equal, simply having more households helps raise aggregate consumption growth. However, the number of households started to decline in 2022, consistent with the decline in the overall population, implying that this demographic factor now contributes negatively to total consumption growth.
6. The average household size has been shrinking over the past decade. Figure 3 (right) shows the changes in household size (equivalent scale) from 2010 to 2020. The figure indicates that the average household size has been shrinking for all household types, except for households with 3 or more adults with dependent children (aged less than 44). On average, the number of persons per household decreased from 2.9 in 2010 to 2.6 in 2020. Other things equal, this decline in the household size will reduce the level of consumption per household and aggregate consumption.
Number of Households and Household Size
Citation: IMF Staff Country Reports 2025, 073; 10.5089/9798229005852.002.A001
Sources: Population and Housing Census, Eurostat, and IMF staff estimates7. An increase in consumption per person will be needed to offset the negative impact of demographic change moving forward. Based on the estimated DNA, the contributions of the aforementioned four elements are calculated (Figure 4). Note that the total annual growth rates in Figure 4 (from DNA) are the same as in Figure 1 (from NA). The DNA data indicate that both composition and household size have put downward pressure on real consumption growth (-0.3 ppt and -0.2 ppt, respectively), while the number of households contributed positively (0.9 ppt). If trends related to the composition and size of the average household continue, current and future annual consumption growth rate will, other things equal, be around 0.5 ppt (the sum of average contributions from composition and household size) lower than would otherwise have been the case. The anticipated decline in the number of households moving forward implies that the overall impact of demographic change on aggregate consumption will likely be even higher. Similar analysis can be applied to real income and net wealth, where demographic changes could reduce annual growth by at least 0.6 ppt and 0.4 ppt, respectively.8 To mitigate the negative impact from demographic change and achieve robust consumption growth, an increase in consumption per person, which is the last of the four elements discussed above, will be required.
8. Raising consumption per person requires higher income growth. DNA data indicate that real consumption per person was almost constant (zero growth) during 2010-15 but grew strongly during 2015-20 (Figure 4). To examine the drivers behind these movements a simple regression analysis is conducted which attempts to explain real consumption by real income and net wealth. Figure 5 presents the results of this analysis: the left figure shows the compounded annual growth rate of real consumption, real income, and net wealth for 2010-15, and the right figure for 2015-20. This simple regression analysis suggests that while both real income and net wealth growth contributed to real consumption growth, income growth tends to play a more important role. This implies that policy measures to increase income growth per person are particularly important for sustaining consumption growth moving forward.
Decomposition Analysis
Citation: IMF Staff Country Reports 2025, 073; 10.5089/9798229005852.002.A001
Note: All deflated by Household Consumption DeflatorSources: Eurostat, HBS, NFCS, and IMF staff estimation.Drivers of Real Consumption per Person
Citation: IMF Staff Country Reports 2025, 073; 10.5089/9798229005852.002.A001
Note 1: w/dep. = with dependent children. 1A+D = 1 adult with dependent children.Note 2: The charts show the fitted values for a simple regression analysis by setting real consumption per person growth as the dependent variable and growth in real income per person and net wealth per person as independent variables. Some of the household categories with smaller sample sizes are aggregated to achieve more stable numbers.Sources: Eurostat, HBS, NFCS, and IMF staff estimation.C. Inequality Analysis using DNA
9. The second part of this paper uses the DNA dataset to analyze changes in inequality. Inequality measures, such as the Gini coefficient, are generally estimated using income or consumption per person from microdata, but these measures do not necessarily align with the NA numbers. By employing the DNA, which adjusts microdata to be fully consistent with the NA, inequality can also be measured in line with the NA.
10. The DNA dataset suggests that consumption and income inequality in Slovakia has declined. Figure 6 indicates the real consumption growth (left figure) and real income growth (right figure) by income quintile over 2010-15 and 2015-20.9 For both periods, annual growth rates of real consumption tend to be higher for households in lower quintiles, as these households experienced higher income growth than those in the higher income quintiles. From 2010 to 2020, this helped reduce the Gini coefficient (measured by income) from 0.3 to 0.26 and the coefficient (measured by consumption) from 0.28 to 0.23, implying that inequality has declined.
Consumption and Income Growth by Income Quintile
Citation: IMF Staff Country Reports 2025, 073; 10.5089/9798229005852.002.A001
11. The decline in inequality can be attributed to a shift towards high skilled jobs and lower unemployment. Figure 7 (left) shows the changes in household share by labor status of the household head from 2010 to 2020. The figure shows that over this period there was a decrease in the share of unemployed households (by 5.2 ppt) and an increase in the share of households with employment in high skill jobs (by 12 ppt). Figure 7 (right) compares the changes in household share by labor status and the real income level per person. Note that the orange dots, representing the changes in household share, are the same across the right and left figures. Figure 7 (right) demonstrates that there is a positive correlation between the changes in household share that occurred between 2010 and 2020 and the level of real income.10 For example, the share of households with high skill jobs has increased, and the level of income per person among this category of households are the highest, while the opposite holds for unemployed households. This suggests that the labor shift towards high-skill jobs and away from unemployment contributed to lowering income inequality.
12. The DNA data suggests that the gender gap (especially in income) has declined, as relatively more females have moved toward higher skilled jobs. Figure 8 (left) presents the results of a regression analysis estimating the percentage difference in income and consumption between male- and female-headed households. The analysis suggests that there is a statistically significant difference between male- and female-headed households, especially in terms of income, even after controlling for household type, age, and size. However, the chart also suggests that the income difference between genders has narrowed over the years. One possible reason for this narrowing is a shift in the occupation of female workers. Figure 8 (right) shows the changes in household share by the household head's gender and occupation. The figure suggests that more female heads of households are moving toward higher skilled jobs. For example, the increase in the share of managers and professionals is higher for female heads compared with male heads.
Household Share and Income Level (by Labor Status)
Citation: IMF Staff Country Reports 2025, 073; 10.5089/9798229005852.002.A001
Note 1: Gender of household head = M: male, F: female.Note 2: Labor status of household head = emp & low: low skill job, emp & med: medium skill job, emp & high: high skill job. Skill classification is based on the defintion by the ILO, which is ISCO major groups 1–3: high skill, major groups 4–8: medium skill, and major group 9: Low skill.Sources: Eurostat, HBS and IMF staff estimation.D. Conclusions and Policy Considerations
13. Efforts to improve the quality and usability of DNA data should continue. Despite its comprehensiveness, the interpretation of NA data can be challenging due to its lack of distributional insights. As seen in this note, DNA data have the potential to help uncover hidden dynamics behind NA. The new G20 Data Gaps Initiative (IMF, 2023) also includes specific recommendations for household distributional results. To enhance the quality and usability of DNA data, the authorities need to recognize the importance of distributional insights and invest resources for further data improvements. For example, increasing the frequency of micro datasets and improving micro-macro linkages to reduce discrepancies between the micro and macro results would be beneficial.11
14. In Slovakia, measures to increase income are essential to achieve robust consumption growth amid the challenges from demographic changes. Unfavorable population dynamics characterized by changing household composition, a decreasing total number of households, and shrinking household size will exert downward pressure on consumption growth. To mitigate these negative trends and maintain robust aggregate consumption growth, an increase in consumption per person, primarily driven by income, will be necessary. This underscores the importance of policy measures to raise income growth, such as increasing productivity and encouraging labor force participation.12 In particular, the old-age labor force participation rate in Slovakia has room to increase, given that the current rate remains one of the lowest in Europe.
15. Policies that encourage higher skilled jobs and labor force participation are important to maintain the favorable trend in inequality. Inequality measured by both income and consumption has declined from 2010 to 2020 due to a shift towards high skill jobs and decline in unemployed household heads. While the gender gap (especially in terms of income) still exists, it has narrowed as relatively more females moved toward higher skilled jobs. To sustain these positive trends, the authorities should continue to encourage labor force participation among females, and enhance the education system and options for reskilling and upskilling, including strengthening active labor market policies and addressing skill mismatches.
Appendix I. Methodology of Estimating the Distributional National Accounts
The basic strategy of estimating DNA can be found in OECD (2024), Eurostat (2024), ECB (2024), and Coli et al. (2022), although ad-hoc adjustments are made to serve this paper’s purposes. Given that consumption is our primary variable of interest, HBS is treated as our main dataset. Note that the HBS microdata for Slovakia is available for 2010, 2015, and 2020.13
Weight Adjustment
Eurostat experimental data adjusted NA data proportionally to deduct consumption/income from the NA that corresponds to the population not covered in EU-SILC (statistics on income and living conditions) and HBS. However, given this paper’s aim is to explain trends fully consistent with NA, this analysis adjusted the sample weight of HBS as follows.
First, the total number of households is estimated. Although the EU-LFS (labor force survey) provides estimates of the number of households, the data series has breaks in 2011 and 2021. Hence, the paper calculated the number of persons per household based on the Population and Housing Census 2011 and 2021 from the Statistical Office of the Slovak Republic (SOSR). The numbers of persons per household for other years (other than 2011 and 2021) are extended based on the growth rate from EU-LFS.14 Using the total population (break adjusted) from Eurostat, the total number of households can be calculated as the total population divided by the number of persons per household. Since both population and number of households are as of 1st January in each year, the average is taken to calculate the mid-year estimates. The estimated results can be found in Figure 3.
Next, based on the estimated number of households and population distributions, the weight of HBS is adjusted. The right chart shows the population distribution of gender-age groups for 2010, 2015, and 2020 from Eurostat. The original HBS weight is adjusted by employing entropy reweighting to be consistent with these population distributions.15 Finally, the weight scale is adjusted to match with the total number of households.
Matching HBS and HFCS
Based on the discussion by Balestra and Oehler (2023), HBS microdata from 2010, 2015, and 2020 are matched with HFCS microdata from 2010, 2014, and 2021, respectively. In this analysis, HBS is the recipient dataset and HFCS is the donor dataset (providing wealth data to HBS). The integration of the two datasets is conducted through statistical matching. In this paper, this is performed using K-nearest neighbor (KNN) matching based on the 13 common variables between HBS and HFCS.16
As distance is a critical factor for the KNN algorithm, three datasets with different distances are prepared: one emphasizes household structure, another emphasizes income and consumption, and the third is somewhere in between.17 Using these three datasets, households between HBS and HFCS are matched three times. A simple average of the wealth data from these three results is used as the outcome of this matching exercise.
Gap adjustment
Consumption adjustment
Following Eurostat (2024) and Coli et al. (2022), a proportional allocation method is used for gap allocation. This method distributes the entire micro-macro gap proportionally across households. The coefficients used for upscaling or downscaling the microdata are calculated as the weighted sum of the microdata divided by the corresponding NA item. In our exercise, the coefficients are calculated to match the main 12 ECOICOP (European Classification of Individual Consumption according to Purpose) divisions and four durability categories (durable, semi-durable, non-durable, and services) from NA.
Income adjustment
Given the limited availability of income information from HBS, the variable for net income from HBS — defined as total income from all sources, including non-monetary components, minus income taxes — is used. This variable is adjusted to match the gross disposable income from NA. The gap between microdata and NA is adjusted based on the ascending gap shares by decile (consistent with the Eurostat (2024) method). The method assumes under-coverage or under-reporting of higher income households and allocates the higher gap shares to higher income decile groups. Then, each household within the respective decile is adjusted by an equal amount. In our exercise, the gap share per decile is slightly adjusted (see the table below) so that the estimated Gini coefficient using HBS data is close to the Gini coefficient from Eurostat experimental data, which is estimated from EU-SILC.
|
M3.1 Ascending gap shares by decile |
decile |
D1 |
D2 |
D3 |
D4 |
D5 |
D6 |
D7 |
D8 |
D9 |
D10 |
|
gap share,% |
0 |
0 |
4 |
6 |
8 |
12 |
14 |
16 |
18 |
22 | |
|
adjusted |
0 |
2 |
4 |
6 |
8 |
12 |
14 |
16 |
18 |
20 | |
|
Each household within the respective decile has been adjusted by an equal amount rather than proportionally to their initial relative contribution. | |||||||||||
|
M3.1 Ascending gap shares by decile |
decile |
D1 |
D2 |
D3 |
D4 |
D5 |
D6 |
D7 |
D8 |
D9 |
D10 |
|
gap share,% |
0 |
0 |
4 |
6 |
8 |
12 |
14 |
16 |
18 |
22 | |
|
adjusted |
0 |
2 |
4 |
6 |
8 |
12 |
14 |
16 |
18 |
20 | |
|
Each household within the respective decile has been adjusted by an equal amount rather than proportionally to their initial relative contribution. | |||||||||||
Wealth adjustment
Following the study by Ahnert et al. (2020), the following HFCS items are matched with NA financial assets/liabilities (see table below). For fixed assets, the sum of AN.112 (Buildings other than dwellings and other structures) and AN.211 (Land) from NA is matched with HFCS codes HB0900 and HB280$x.
|
ESA 2010 code |
NA instrument name |
HFCS variable(s) |
HFCS variable label and/or codes |
|---|---|---|---|
|
F.22 |
Transferable deposits |
HD1110 |
Sight accounts |
|
F.29 |
Other deposits |
HD1210 |
Savings accounts |
|
F.3 |
Debt securities |
HD1420 |
Bonds |
|
F.4 |
Loans |
HD1710 |
Amount owed to household |
|
F.511 |
Listed shares |
HD1510 |
Publicly traded shares |
|
F.521 |
Money market fund (MMF) shares or units |
HD1320c |
Investments in mutual funds c |
|
F.522 |
Non-MMF investment fund shares/units |
HD1320a+HD1320b+HD1320d+HD1320e+HD1320f |
Investments in mutual funds a, b, d, e, f |
|
F.62 |
Life insurance and annuity entitlements |
DA2109 = sum of (PF0920) over household members |
Voluntary pension/whole life insurance |
|
F.4 |
Loans |
DL1000 |
Total outstanding balance of household’s liabilities |
|
ESA 2010 code |
NA instrument name |
HFCS variable(s) |
HFCS variable label and/or codes |
|---|---|---|---|
|
F.22 |
Transferable deposits |
HD1110 |
Sight accounts |
|
F.29 |
Other deposits |
HD1210 |
Savings accounts |
|
F.3 |
Debt securities |
HD1420 |
Bonds |
|
F.4 |
Loans |
HD1710 |
Amount owed to household |
|
F.511 |
Listed shares |
HD1510 |
Publicly traded shares |
|
F.521 |
Money market fund (MMF) shares or units |
HD1320c |
Investments in mutual funds c |
|
F.522 |
Non-MMF investment fund shares/units |
HD1320a+HD1320b+HD1320d+HD1320e+HD1320f |
Investments in mutual funds a, b, d, e, f |
|
F.62 |
Life insurance and annuity entitlements |
DA2109 = sum of (PF0920) over household members |
Voluntary pension/whole life insurance |
|
F.4 |
Loans |
DL1000 |
Total outstanding balance of household’s liabilities |
The net wealth in this paper is defined as “deposits + other assets (debt securities + listed shares + investment fund shares + life insurance) + fixed assets (dwellings + land) – financial liabilities (mortgage + non-mortgage).” As our focus is on consumption analysis, this paper excludes business wealth (financial and non-financial business wealth) to simplify the estimation process, which is approximately 10 percent of total assets based on DWA by ECB.
As discussed by ECB (2024), the HFCS/QSA coverage ratios of deposits are large in many countries, including Slovakia (about 35 percent), implying the under-reporting of deposit numbers. Based on the suggestions by ECB (2024), some of the possible outliers of deposit numbers in HFCS (extremely small levels of deposits given the income level) are corrected.18
The gap allocations between macro and micro data are conducted based on the following four items: deposits, other assets, fixed assets, and financial liabilities. A proportional allocation (see consumption adjustment) is used with information from DWA to fill the gap. The DWA data provide the share of household wealth items by the level of households’ net wealth (the bottom 50 percentiles and deciles 6-10 of the, or six household groups in total), as illustrated by the chart below. Although the net wealth of DWA includes business wealth, the differences are assumed to be marginal. Based on this share data from DWA, the corresponding NA items (deposits, other assets, fixed assets, and financial liabilities) are divided into six groups (bottom 50 percentile and deciles 6-10). Similarly, the matched household microdata (HBS-HFCS microdata) are also divided into the six groups based on the net wealth concept. Then, the proportional allocation is applied to the corresponding four wealth items and six net wealth groups between NA and HBS-HFCS microdata.
Deflator adjustment
The deflator for each household is constructed based on the Harmonized Index of Consumer Prices (HICP). At the time of writing, given that the latest deflator of NA is set to 2020 = 100, all HICP deflators are converted to 2020 = 100 (instead of 2015 = 100). Then, the weighted average of the HICP deflator for each household is constructed at the ECOICOP level 3 in 2010 and level 4 in 2015 and 2020. For imputed rent, the deflator from NA is used, as HICP does not cover this price index. Additionally, some items that HICP does not cover (e.g., narcotics, games of chance, life insurance, etc.) are disregarded, given their potentially small impacts on the total index. The weighted average of the constructed HICP-based deflator for all households and the total deflator from NA is compared, and a proportional allocation is applied (the HICP-based deflator is adjusted) to match the two numbers.
References
Ahnert, H., Kavonius, I. K., Honkkila, J., & Sola, P. (2020). “Understanding household wealth: linking macro and micro data to produce distributional financial accounts”. ECB Statistics Paper Series No 37
Balestra, C., & Oehler, F. (2023). “Measuring the joint distribution of household income, consumption and wealth at the micro level”. OECD Papers on Well-being and Inequalities, No. 11, OECD Publishing, Paris
Blatnik, N., Bobasu, A. G., Krustev, G., & Tujula, M. (2024). “Introducing the Distributional Wealth Accounts for euro area households”. ECB Economic Bulletin.
Coli, A., Istatkov, R., Jayyousi, H., Oehler, F., & Tsigkas, O. (2022). “Distributional national account estimates for household income and consumption: methodological issues and experimental results”, Publications Office of the European Union.
Eurostat. (2024), “Distribution of income and consumption for the Household sector Eurostat centralized exercise – methodological note”
European Central Bank. (2024) “Experimental Distributional Wealth Accounts (DWA) for the household sector Methodological note”
Hainmueller, J., & Xu, Y. (2013). “ebalance: A Stata Package for Entropy Balancing”. Journal of Statistical Software, 54(7), 1–18.
IMF (2023), G20 Data Gaps Initiative, https://www.imf.org/en/News/Seminars/Conferences/g20-data-gaps-initiative
OECD. (2024). “OECD Handbook on the Compilation of Household Distributional Results on Income, Consumption and Saving in Line with National Accounts Totals”, OECD Publishing, Paris
Prepared by Shinya Kotera (EUR).
See Blatnik et al. (2024) for a detailed analysis of DWA. QSA data are constructed by the ECB following international standards for NA and provide comprehensive wealth information for all resident institutional sectors.
Household type contains the following six categories: one adult, one adult with dependent children, two adults, two adults with dependent children, three or more adults, and three or more adults with dependent children. The age of the household head contains the following four categories: aged less than 30, aged 30-44, aged 45-59, and aged 60 and over. In this paper, the household head refers to the household’s reference person in the HBS to simplify the argument, although they are not necessarily the same. Generally, the reference person in the HBS is the adult with the highest income.
The household size here refers to equivalent size (modified OECD scale). The modified OECD scale attributes a weight to all household members: a weight of 1 to the first adult, a weight of 0.5 to each additional adult, and a weight of 0.3 to each child. The equivalent size is the sum of the weights of all the members.
They have a correlation of -0.4.
Similar trends (negative correlation) can also be observed between the changes in household share and the level of real income (or net wealth), implying that changes in household composition also put downward pressure on the growth of real income and net wealth.
The number of households is estimated based on the Population and Housing Census and EU-LFS (labor force survey). See appendix.
Demographic change could also impact savings rate. From 2010 to 2020, the composition effect contributed negatively (on average, -0.1 ppt annually) to the change in gross savings rate.
Quintiles based on equivalized income.
They have a correlation of 0.6.
See OECD (2024) for more discussion about this.
See “A Micro-Meso-Macro view of labor productivity growth in the Slovak Republic” in this Selected Issue Paper for more discussion about productivity.
The fieldwork period for the HBS 2020 wave in Slovakia is 2019-2021.
Since extending the 2011 census data by using the growth rates of EU-LFS does not reach the 2021 census level, the growth rates of EU-LFS are adjusted by minimizing the sum of squared residuals between the growth rates (from EU-LFS) and adjusted growth rates, subject to the constraint in two census years (2011 and 2021).
Stata packaged prepared by Hainmueller and Xu (2013) is used.
These variables are age, education, labor status, household size, number of children, number of elderly members, number of household members in employment, household type, income ventiles, food consumption ventiles, utility consumption ventiles, food and utility consumption ventiles, and goods and services consumption ventiles. Due to data availability, for the 2010 KNN matching only, the categories of household type differ from those in other matching years, and the goods and services consumption ventiles is missing.
Following three datasets are prepared. 1) greater distances for age and household type, 2) greater distances for income and food and utility ventiles and less distances for number of children and elderly members, and 3) distances that are around the average between 1 and 2 datasets.
Deposit numbers are corrected to 10 percent of households’ monthly income if their annual gross incomes are above €10,000 and their reported deposit numbers are less than 10 percent of the monthly income. Deposit numbers are corrected to 5 percent of households’ monthly income if their annual gross income is above €5,000 but below €10,000, they do not have any financial liabilities, and their reported deposit numbers are less than 5 percent of the monthly income.
