Abstract
This paper analyzes the implications of global rebalancing in the post-crisis period for Korea and how high leverage in the household and SMEs sectors could affect this process. The first section of the paper discusses implications of global rebalancing for Korea using simulations from the Global Integrated Monetary and Fiscal Model (GIMF). The second section focuses on how rebalancing growth in Korea is different from the rest of the region, and discusses the challenges of highly leveraged households and SMEs for the rebalancing process. The last section concludes with policy recommendations.
II. Are Korean Households Saving Too Little?1
The combination of the declining saving rates and rising indebtedness of Korean households raises concerns that leveraged consumption growth cannot be sustained and could eventually lead to a vicious cycle of deleveraging, lower economic growth, and risks to financial stability. This paper sheds light on the factors behind declining household saving rate in Korea, utilizing cross-country panel data. The results point to rapid aging and labor market, and retirement system characteristics as fundamental factors driving the decline in the saving rate, while a prolonged period of cheap and easy credit, coupled with a deteriorating terms of trade appear to have magnified these trends. While demographic trends will continue pushing household savings lower, labor market and pension system reforms could help arrest the decline in saving rates. Increasing productivity in the nontradable sector would also help lift terms of trade and household incomes and facilitate deleveraging.
A. Introduction
1. Korean households’ savings rate dropped from 27 percent to 7½ percent of disposable household income since 1998 and remains one of the lowest in the OECD and in Asia. This has coincided with a steady increase in household debt, which reached 143 percent of reported disposable income.2 Furthermore, Koreans spend 7 percent of their income to service debt, more than U.S. households.
Household Saving Rate
Citation: IMF Staff Country Reports 2010, 271; 10.5089/9781455204311.002.A002
Source: OECD.1/ includes the OECD countries in the sample.2. The combination of declining saving rates and increasing household debt has raised concerns about sustainability of consumption and financial stability. Higher debt and lower savings enabled household consumption to grow faster than disposable income, providing a boost to economic growth. In the long-run, however, consumption cannot grow faster than income because there is an upper limit to how much debt households can service, based on their incomes. To achieve a sustainable level of debt relative to income, households may need to undergo a prolonged period of deleveraging, whereby debt is reduced and saving is increased. This paper sheds light to the factors behind the sharp decline in saving rate, understanding of which would be essential in devising policies that would help reverse the decline.
Household Consumption and Credit
(2005=100)
(In percent of disposable income)
Citation: IMF Staff Country Reports 2010, 271; 10.5089/9781455204311.002.A002
Source: CEIC Data Company Ltd.3. The next section of the paper presents a cross-country panel regression and discusses the key factors behind the declining saving rate in Korea. The last section concludes with policy recommendations.
B. What Explains Korean Households’ Declining Saving Rate?
4. The literature suggests four broad motives for household savings: to provide resources for retirement and bequests, to finance large life-time expenditures (e.g., housing and education), to finance unexpected losses of income (precautionary savings) and to smooth consumption as incomes vary over time. These motives point to a large number of variables that can explain saving behavior over time and across countries.
5. A cross-country panel regression was used to explain the dynamics of the household saving rate in Korea (Table 1). 3 In line with past empirical literature, the variables considered were public and corporate savings (both as a share of GDP) to capture Ricardian effects and households’ ability to pierce the “corporate veil,” the level of per capita GDP, real household disposable income growth (to capture the level of development), the old-age and young age-dependency ratios (to capture the impact of demographics), real interest rates (to capture the opportunity cost of current consumption), CPI inflation (to capture uncertainty about future income growth), the unemployment rate (to capture precautionary saving motive to insure consumption against adverse shocks to income), real stock prices (to capture wealth effects),4 the share of household credit in GDP (to capture financial deepening), and the terms of trade and the effective retirement age. The terms of trade is intended to capture divergence between gross national income and output, as a fall in the relative value of exports corresponds to a fall in income which is not matched by a drop in consumption given the less-than-unit marginal propensity to consume and save and may lead to lower saving rates. Retirement tends to be the most important saving motive, making a change in pensions and expected length of the retirement period important demographic variables in saving decisions. Increases in the expected length of the retirement period, either through a higher life expectancy or through a decline in the mandatory retirement age, raise the need for more saving in younger ages, putting upward pressure on the aggregate savings. However, since cross-country data on these variables are not available, instead effective retirement age was used as a proxy.5 This variable captures the decision of the retirees to stay in the labor force and hence takes into account their response to incentives embedded in the pension system. For instance, a higher effective retirement age could be associated with the decisions of retirees to stay in the labor force due to low replacement or low coverage rates of the pension system, which in turn could be associated with lower saving motives as it would reduce actual retirement periods.
Determinants of Household Savings
(In percent of household disposable income)
In thousands of U.S. dollars (ppp)
The youth-dependency ratio is the ratio of the pre-working age population (age category 0 to 19 years) to the working-age population (aged 20 to 64). The elderly-dependency ratio is represented by the ratio of the population in the retirement phase (aged 65 and over) to the working-age population.
Source: OECD.
Figures in parantheses are p-values. *, ***,**** denote significance at 10, 5, and 1 percent, respectively.
Determinants of Household Savings
(In percent of household disposable income)
| Coefficients | 4/ | |
|---|---|---|
| GDP per capita 1/ | 0.000 | [0.008]** |
| Old age dependency ratio 2/ | -0.257 | [0.010]** |
| Young age dependency ratio 2/ | 0.214 | [0.033]** |
| Effective retirement age 3/ | -0.534 | [0.000]*** |
| Terms-of-trade | 0.052 | [0.000]*** |
| Real household disposable income growth | 0.310 | [0.000]*** |
| Corporate saving/GDP | -0.472 | [0.000]*** |
| Public saving/GDP | -0.464 | [0.000]*** |
| Real interest rate | 0.332 | [0.001]*** |
| Real stock prices | 0.001 | [0.003]*** |
| Inflation rate | 0.529 | [0.000]*** |
| Unemployment rate | 0.177 | [0.078]* |
| Household credit/GDP | -0.045 | [0.000]*** |
In thousands of U.S. dollars (ppp)
The youth-dependency ratio is the ratio of the pre-working age population (age category 0 to 19 years) to the working-age population (aged 20 to 64). The elderly-dependency ratio is represented by the ratio of the population in the retirement phase (aged 65 and over) to the working-age population.
Source: OECD.
Figures in parantheses are p-values. *, ***,**** denote significance at 10, 5, and 1 percent, respectively.
Determinants of Household Savings
(In percent of household disposable income)
| Coefficients | 4/ | |
|---|---|---|
| GDP per capita 1/ | 0.000 | [0.008]** |
| Old age dependency ratio 2/ | -0.257 | [0.010]** |
| Young age dependency ratio 2/ | 0.214 | [0.033]** |
| Effective retirement age 3/ | -0.534 | [0.000]*** |
| Terms-of-trade | 0.052 | [0.000]*** |
| Real household disposable income growth | 0.310 | [0.000]*** |
| Corporate saving/GDP | -0.472 | [0.000]*** |
| Public saving/GDP | -0.464 | [0.000]*** |
| Real interest rate | 0.332 | [0.001]*** |
| Real stock prices | 0.001 | [0.003]*** |
| Inflation rate | 0.529 | [0.000]*** |
| Unemployment rate | 0.177 | [0.078]* |
| Household credit/GDP | -0.045 | [0.000]*** |
In thousands of U.S. dollars (ppp)
The youth-dependency ratio is the ratio of the pre-working age population (age category 0 to 19 years) to the working-age population (aged 20 to 64). The elderly-dependency ratio is represented by the ratio of the population in the retirement phase (aged 65 and over) to the working-age population.
Source: OECD.
Figures in parantheses are p-values. *, ***,**** denote significance at 10, 5, and 1 percent, respectively.
6. Regression results suggest that the decline in the Korean households’ saving rate is consistent with fundamentals. Without even accounting for Korea specific factors, beyond those captured by the various regressors, the decline in Korean household savings rate can be reasonably predicted by the model. This suggests that rather than country specific factors, economic trends in Korea, consistent with other country experiences, tend to explain the declining trend in savings. Decomposing the estimates into their respective contribution to the household savings, few key conclusions emerge; demographic trends and retirement were the key structural factors behind the decline in the saving rate until the Asian crisis and they continue to contribute significantly to the negative trend since 1997. However, since the Asian crisis increased access to finance along with low interest rates and deteriorating terms of trade have been the main reasons behind the sharp fall in savings rates. Corporate deleveraging since the Asian crisis has also led to lower household savings, while Ricardian effects were more dominant prior to 1997 reflecting the larger increase in public savings during that period.
Demographics. An aging population would reduce household savings, a trend observed in most of the OECD countries. Similarly, a decline in the young-age dependency ratio would reduce the saving motives of the parents. Although the old-age dependency ratio in Korea is still below the OECD average, the fast drop in Korea’s fertility rate led to a significant drop in the young-age dependency ratio and appears to stands out as a key reason for the declining saving rate. Going forward, with old-age dependency increasing rapidly with an aging population, the saving rate is likely to decline further.
The introduction of the public pension system and the unique features of the labor market reduced incentives for saving by altering labor supply decisions. The fully funded and mandatory pension system in Korea has one of the lowest replacement and contribution rates in the OECD.6 Nonetheless, its introduction in 1988 was a break-through for households that relied only on personal savings until then for retirement income reducing incentives to save. On the other hand, the seniority system in the labor market translates into low mandatory company retirement age, well below the official retirement age, which in turn is already lower than the OECD average. As a result, high life expectancy pushes most retirees to second jobs, increasing the effective retirement age beyond the official retirement age. However, most post-retirement employment tends to concentrate in the services, where mom and pop shop proliferate with retirement lump-sums used for start-up capital, or in irregular employment with lower wages.7 These features of the pension system and the labor market contribute to high old-age poverty rates and the higher indebtedness of older cohorts in Korea (see Karasulu, 2007).
Corporate savings have substituted for some of the household savings. Corporate and household savings are intimately connected through the household ownership of corporates. Absent any tax distortions and liquidity constraints, households should be indifferent between holding their savings directly or indirectly via the savings of the firms that they own. However, in the presence of credit constraints or a weak corporate management culture piercing, this corporate veil becomes difficult, breaking this neutrality. While for most of Asia this neutrality does not hold, in Korea, the deleveraging of the corporate sector following the Asian crisis along with substantial corporate governance reforms appears to have provided incentives to reduce savings for households.
Low interest rates and competition for new market segments in the financial system after the Asian crisis fueled credit to the household sector and reduced incentives to save. With less diversification of household balance sheets to financial assets, the low interest rate environment reduced incentives to save and increased household debt. Since 1998, credit to households has increased at an average annual rate of 13 percent, reaching about 70 percent of GDP at e nd-2009 from about 38 percent of GDP prior to the crisis. More than two-thirds of this increase can be attributed to lending by depository money banks. This coincides with retrenchment of credit from the large corporates following the financial crisis, when the banking sector increasingly shifted toward retail lending and since 2000 through a rapid expansion of credit card use. The competition to lend to household sector also appears to have contributed to the rapid rise in household debt. Since 2000, lending rates to household sector declined faster than those charged to corporate sector, despite the expectations of higher risk from such lending.
A deteriorating terms of trade may have supported exports, but undermined real household income growth and reduced savings. The price of a country’s exports relative to its imports is important in the GDP deflator, particularly in an economy with a large trade sector, such as Korea. The deterioration of the terms of trade since 1998 has contributed to almost one-half lower increase in GDP deflator than the CPI or the deflator for private household consumption constraining real income growth and reducing savings in line with the Harberger-Laursen-Metzler effect.
Model Predictions: Korean Household Saving Rate
Citation: IMF Staff Country Reports 2010, 271; 10.5089/9781455204311.002.A002
Sources: OECD; and IMF staff estimate.Korea: Change in Household Saving Rate: Contributions
(Percentage points)
Citation: IMF Staff Country Reports 2010, 271; 10.5089/9781455204311.002.A002
Sources: CEIC Data Company; and IMF staff calculations.Demographic Trends
Citation: IMF Staff Country Reports 2010, 271; 10.5089/9781455204311.002.A002
Source: World Bank, World Development Indicators database.Pension Replacement Rates and Effective Retirement Age, 2000-07
Citation: IMF Staff Country Reports 2010, 271; 10.5089/9781455204311.002.A002
Source: OECD.Old Age Poverty
Citation: IMF Staff Country Reports 2010, 271; 10.5089/9781455204311.002.A002
Sources: Rein et al. (2004); and for Korea, Kim & Cho (2005).Household and Corporate Savings
(In percent of GDP)
Citation: IMF Staff Country Reports 2010, 271; 10.5089/9781455204311.002.A002
Household* and Corporate Loans
(In percent of GDP)
Citation: IMF Staff Country Reports 2010, 271; 10.5089/9781455204311.002.A002
Source: CEIC Data Company Ltd.* Include general loans and loans for housing. ** Major nonbanks Include KDB (until 01), Merchant Banks, Trust account of banks, and life insurance companiesInterest Rates and Credit Risk
(1999Q4=100)
Citation: IMF Staff Country Reports 2010, 271; 10.5089/9781455204311.002.A002
Sources : BOK and CEIC Data Company Ltd.Terms of Trade, GDP and Household Income
(1988=100)
Citation: IMF Staff Country Reports 2010, 271; 10.5089/9781455204311.002.A002
Sources: CEIC Data Company; and IMF staff calculations.C. Conclusions and Policy Recommendations
7. The decline in the Korean household savings rate is driven by a combination of structural and cyclical factors. While the rapidly aging population and the introduction of the pension system are the key structural reasons for the declining saving rate, the availability of cheap credit also contributed to this trend, especially at a time when real household income growth was constrained by a deteriorating terms of trade.
8. Going forward demographic factors are likely to push the saving rate even lower. One of the lowest fertility rates in the world and an increasing life-expectancy will translate into an extraordinarily large increase in the old-age dependency ratio from 14% percent in 2008 to 65 percent in 2050. As a result, more people who are currently in the work force and are accumulating assets would reduce their saving in the medium term.
9. However, cyclical factor should help support savings. With the Great Moderation over, and tighter financial regulations expected to temper credit growth, a deleveraging of the household sector is likely. Considering the high share of borrowing at floating rates, the deleveraging could be rather rapid depending on the speed of interest rate normalization.
10. Addressing the fundamental factors behind the fall in the saving rate would require wide-ranging structural policies. Besides immigration and child- friendly policies to reduce long-term aging pressures, the design of the pension system as well as labor market policies may need to be recalibrated. The 2008 pension reform will gradually reduce the pension benefit replacement rate from the current 60 percent of wages to 40 percent by 2028, but does not raise the contribution rate from its current 9 percent, reflecting a social choice of low contribution and low benefits in Korea. However, the seniority system in the labor market coupled with low coverage and low benefits from the pension system are increasing the poverty risk of an aging population as working years do not provide sufficient accumulation of assets for the longer life expectancy. Reversing this trend would require eliminating the seniority wage system and increasing regular employment of old-age workers.
References
Bérubé, G., and D. Côté, 2000, “Long-Term Determinants of the Personal Savings Rate: Literature Review and Some Empirical Results for Canada,” Bank of Canada Working Paper 2000-3.
Deaton, A.S., 1977, “Involuntary Savings Through Unanticipated Inflation,” American Economic Review, 67: 899–910.
Feyzioğlu, T, and M. Skaarup, and M. Syed, 2008, “Addressing Korea’s Long-Term Fiscal Challenges,” IMF Working Paper No. 08/27 (Washington: International Monetary Fund).
Hüfner, F., and I. Koske, 2010 “Explaining Household Saving Rates in G7 Countries: Implications for Germany,” OECD Economics Department Working Papers No. 754.
Kwack, S.Y., and Y.S. Lee, 2005, “What Determines Saving Rates in Korea? The Role of Demography,” Journal of Asian Economics, 16, 861–73.
Muellbauer, J., 1994, The Assessment: Consumer Expenditure, Oxford Review of Economic Policy, 10, Issue 2, p. 1–41, http://econpapers.repec.org/RePEc:oup:oxford:v:10:y:1994:i:2:p:1-41.
Sandmo, A., 1970, “The Effect of Uncertainty on Saving Decisions.” Review of Economic Studies 37: 353–60.
Prepared by Meral Karasulu. Janice Lee provided assistance with data.
Survey data suggest that underreporting of household incomes may overstate the debt ratio by about 22 percentage points, consistent with the estimated size of the informal economy.
An unbalanced panel regression comprising 20 countries was estimated using generalized method of moments (GMM) with country dummies and lagged values of the variables as instruments. The countries are Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Japan, Korea, New Zealand, Netherlands, Norway, Sweden, Switzerland, United Kingdom and United States. The longest series in the unbalanced sample starts in 1970 and ends in 2008.
Real house prices were also included, but were not statistically significant.
The average effective retirement age is defined as the average age of exit from the labor force during a five-year period. Labor force (net) exits are estimated by taking the difference in the participation rate for each five-year age group (40 and over) at the beginning of the period and the rate for the corresponding age group aged 5 years older at the end of the period.
Although the former helps raise the sustainability of the pension system and reduce the associated fiscal burden, it also affects labor supply and saving decisions by households.
Reflecting the large number of self-employed and irregular workers, the Korean pension system has a low coverage (60 percent compared to around 85 percent for the OECD), although notional coverage was made universal in 1999.
