Journal Issue
Share
Article

Republic of Latvia: Selected Issues

Author(s):
International Monetary Fund. European Dept.
Published Date:
September 2018
Share
  • ShareShare
Show Summary Details

Demographic Headwinds to Convergence1,2

Latvia’s rapidper-capita income convergence rates with Western Europe have on average more than halved in the years after the Global Financial Crisis (GFC) compared to pre-crisis levels. At the same time, Latvia’s population and labor force have been shrinking and aging rapidly. Although the challenges of the ongoing demographic transition will be felt gradually over many years, postponing policy action now could result in much larger required policy adjustments in the future.

A. Background

1. Latvia’s per capita income convergence with Western Europe has slowed, as demographic trends have turned unfavorable. While the income gap with the EU-15 decreased by 7.7 percent annually in 2000–07, it has been closing at an average rate of 3.7 percent since 2011. Since the GFC outburst, Latvia’s population has been shrinking and aging rapidly as fertility rates dropped and life expectancy increased. Persistent emigration has exacerbated this trend, as younger cohorts are more likely to emigrate than older ones. As Latvia is still some distance away from the per capita income levels of advanced Europe, demographic headwinds pose the risk that the country may grow old before becoming rich.

Figure 1.Snapshot of Recent Convergence and Demographic Developments

2. Demographic headwinds can affect economic outcomes through various channels. Fewer workers produce less output, which, if not offset by more capital or higher productivity, will lead to lower potential output. If shrinking and aging coincide, a higher dependency ratio can translate into lower per capita GDP.3 Aging may also affect labor productivity due to depreciating skills and physical capabilities as workers become older (Aiyar et al, 2016). In addition, population aging will create challenges for government finances as fewer workers will provide social security contributions for old-age pensions, health, and long-term care. Demographics also affect external balances and capital accumulation through aggregate investment-saving balances as, for example, a relative shift in the population structure toward older people leads to higher share of consumers relative to savers.

3. Against this backdrop, this SIP examines the implications of projected demographic changes for Latvia and presents policy options to address some of the resulting challenges. Using population projections from the UN’s World Population Prospects (WPP) and labor market statistics from the International Labour Organization (ILO), the SIP examines how projected changes in the size and structure of the labor force could impact productivity and growth, and ultimately income levels. The results underscore the key role of policies aimed at boosting the productivity of workers, increasing labor force participation rates (LFPRs), opening the market for immigrant labor, and spurring innovation.

B. Demographic Trends and Implications for Convergence

4. The total population has been on a declining path since the beginning of the 1990s, a trend which is projected to continue. Since Latvia regained independence in 1990, the country has lost about ¼ of its population, of which about 60 percent was attributable to emigration. Going forward, emigration is projected to continue, albeit at a diminishing rate until about 2027. Coupled with a negative contribution from natural population change (births and deaths), Latvia’s population is projected to decline to about 1.5 million in 2050 from currently just below 2 million under the UN WPP’s medium fertility scenario, which is equivalent to an average annual decline of ¾ percent.

Contributions to total population growth

(thousand)

Sources: United Nations World Population Prospects 2017; Staff Calculations

5. At the same time, the population has aged, which is reflected in a rapidly increasing old-age dependency ratio. While the median age in 1990 was about 35 years, it was 43 in 2015 and is projected to increase to 46 years by 2050. Similarly, the old-age dependency ratio, i.e. the ratio of the population aged 65+ to the population aged 15–64, was 18 percent in 1990, 29 percent in 2015, and is projected to increase to 48 percent by 2050.

Population Distribution by Age Cohort

Source: UN WPP.

6. Similarly, the labor force has been shrinking and getting older. While the labor force in 1990 was about 1.4 million, it decreased to about one million in 2015 and is projected to decline to about 800 thousand in 2050. At the same time, a significant aging of the labor force took place. The share of old-age workers, defined as workers aged 55 years or older, was about 14 percent in 1990, increased to about 20 percent in 2015, and is projected to reach 30 percent in 2050. The share of prime-age workers (45–54 years), which is generally considered as being the most productive, increased by about 1.3 percentage points over the same horizon. The share of the youngest cohorts has been declining from about 14 percent in 1990 to 9 percent in 2015—a reflection of the population bulge passing through to older cohorts.

Size of the labor force and age composition

(Thousand)

Sources: Eurostat, United Nations World Population Prospects 2017, and IMF staff calculations.

Size of the labor force and age composition

(Thousand)

Sources: Eurostat, United Nations World Population Prospects 2017, and IMF staff calculations.

7. Unfavorable demographics are particularly pronounced for Latvia, which may result in a stronger slowing of income convergence compared to other countries. In a cross-country comparison with Central, Eastern, and South-Eastern Europe (CESEE) and Western Europe (WE), Latvia’s projected working-age population has the least favorable dynamics until 2030. In particular, by 2030 the projected decline in Latvia’s working-age population is about 6 percentage points greater than the CESEE average, and about 13 percentage points greater than the WE average. By 2050, these differences are projected to increase to 8 and 26 percentage points respectively. The accelerated reduction in employable labor will, all else equal, translate into a significantly reduced productive capacity of the economy. Consequently, absent any policy measures, closure of the income gap with Western Europe may become progressively elusive.

Changes in Working-Age Population

(Cumulative, percentage points)

Sources: UN WPP; IMF Staff Calculations

C. Growth Implications of Demographic Trends

8. Using a production function approach can help understand the strength of the transmission channels through which demographics affect growth. We employ a growth accounting framework based on Amaglobeli and Shi (2016), which decomposes aggregate labor into age and gender-specific cohorts, with cohort-specific labor force participation and employment rates.4 Using this framework, we estimate the impact of different demographic scenarios on growth, and compare the outcome of a hypothetical “no-aging” scenario, in which the population size and age structure remain constant, to that of (i) an aging scenario, where labor changes according to demographic projections from the UN WPP’s medium-fertility variant, excluding the impact of net migration; and (ii) a net migration scenario, which includes the impact of both aging and net migration. To isolate the impact of demographics from possible policy-induced outcomes, the net-migration scenario is calculated under the assumptions of (i) unchanged total factor productivity (TFP) growth (based on historical average growth); (ii) unchanged age- and gender-specific labor force participation and employment rates; and (iii) a constant capital-to-effective-labor ratio.

9. Demographic shifts could decrease average long-run growth by 1–2 percentage points in 2020–50. Under the assumptions outlined above, the hypothetical no-aging scenario projects average long-term growth at 3 percent in 2020–50, of which capital contributes about 1.7 percentage points.5 Allowing for the projected natural population change (i.e. births and deaths), but abstracting from migration, lowers long-term growth by about 0.7 percentage points as the projected natural decline in labor imposes a drag on growth. The net-migration projection, which includes full population dynamics, projects average growth at 2 percent in 2020–50 due to continued, but diminishing, outward migration.6 Consequently, absent any policy changes, demographic shifts can potentially turn into strong headwinds for Latvia in the coming decades.

Average Contributions to Growth 2020–2050

(Annualized, percent)

Sources: UN WPP; ILO; Statistics Latvia; and IMF staff calculations.

10. An aging workforce can impose an additional drag on growth through lower productivity. An age-related deterioration in physical and mental capabilities and depreciation of knowledge may adversely impact aggregate labor productivity if the share of older workers in the workforce increases. Older workers may also find it more difficult to adapt if job requirements change over time (OECD, 1998). For example, the increased use of information technologies might place older workers at a disadvantage (Dixon, 2003). However, older workers may have more work experience, with potentially positive effects on productivity (Disney, 1996). The combination of these factors has been found to result in a strong increase in productivity until workers are in their 40s and a decline toward the end of their working lives (Feyrer, 2007). The impact of aging may also differ across occupations, as the productivity of manual workers (manufacturing, construction) might decline faster with age, while for doctors, lawyers and managers, it might increase with experience (Veen, 2008). Applying this taxonomy, Latvia has a higher share than the EU-15 average of employed persons—especially male—in professions associated with declining productivity of older workers.

Share of Workforce Whose Productivity Rises or Falls with Aging

(In percent of total employment, 2016)

Note: Category productivity “increases” with age comprises managers and professionals; category “neutral” comprises clerical support workers and services and sales workers; category “decreases” comprises technicians, skilled agricultural workers, forestry and fishery workers, craft and related trades workers, plant and machine operators and assemblers, elementary occupations, and armed forces occupations.

Sources: International Labour Organization; Veen 2008; and IMF staff calculations.

11. We empirically estimate the impact of the growing share of older workers on labor productivity and TFP growth. Using a two-step approach described in Aiyar et al. (2016) and Adler et al. (2017), we first estimate the effect of workforce aging on labor productivity and then decompose the impact on labor productivity into factor accumulation and TFP by employing the panel regression specification:

Where TFPit denotes TFP, αi is a country fixed effect, γi is a time fixed effect, oadr is the old-age dependency ratio, and yadr is the young age dependency ratio. Old-age workers are defined as workers 55 years or older. The above equation is initially estimated using fixed-effects ordinary least squares. To address the potential endogeneity of the share of older workers, we also use a panel-fixed-effects two stage least squares estimation (2SLS) with the 10-year lagged share of the population aged 45–54 as an instrument.7

12. The results show that an aging labor force can significantly slow TFP and GDP growth. We estimate the above 2SLS specification using a panel dataset of 167 countries over the period 1990–2015. A 1 percentage point increase in the share of older workers in the labor force is associated with a statistically significant decrease in TFP growth of about 0.8 percentage points per year (Table 1).8 Using the average projected increase in the share of older workers under the net-migration scenario, and relaxing the assumption of constant TFP growth in the production function growth projections, this translates into lower GDP growth of about 1.1 percentage points on average in 2020–50, which is almost half as high as under the net-migration scenario.9

Table 1.Latvia: The Effect of Workforce Composition on Output per Worker and its Channels
Dependent Variables(1) Δln(YL)(2) α1αΔln(KY)(3) Δln(HC)(4) Δln(TFP)
Workforce share 55+-1.006***

(-4.297)
0.333***

(3.514)
-0.0152

(-0.470)
-0.779***

(-3.312)
Old-age dependency ratio0.413

(1.248)
-0.258**

(-1.993)
-0.0196

(-0.526)
0.236

(0.744)
Young-age dependency ratio0.115**

(2.432)
-0.0468**

(-2.400)
-0.00620

(-0.503)
0.0597

(0.958)
Country fixed effects
Year fixed effects
Number of observations4,1504,1523,5852,883
Number of countries167167144116
Source: IMF staff calculation.
Source: IMF staff calculation.

13. The results do not provide strong support for an offsetting effect through higher capital accumulation. An aging workforce may encourage firms to shift production toward more capital-intensive technologies to complement the gradually less-productive workers. The first step in the empirical strategy outlined above allows to estimate the impact of aging on factor accumulation. Overall, the results show that physical capital does have a statistically significant impact on the real growth of output per worker, but its economic significance is small compared to the role of TFP. In particular, allowing the capital-output ratio to grow as a function of the older worker share according to the regression coefficient in Table 1 would have an impact on the factor-share-adjusted capital output ratio of about ¼ percentage point on average in 2020–50. This translates into higher GDP growth of about 0.1 percentage points compared to the TFP scenario. In addition, the dependency ratios have a statistically significant impact in the opposite direction making the magnitude of the overall impact even less strong.

14. Demographic shifts can impact growth through several additional channels. Demographic changes affect aggregate savings and investment decisions and therefore current account (CA) balances (IMF, 2016a). For example, theory predicts a higher old-age dependency ratio (DR) to have a negative effect on the CA, reflecting a higher share of consumers relative to savers. Similarly, population growth, if mainly driven by birth rates, should exert a negative impact on the CA by increasing the share of the non-saving youth. If driven by growth in the working-age population, higher population growth may also affect the CA negatively through increased investment to stabilize the capital-labor ratio. However, higher-old age dependency ratios may trigger disinvestment by increasing the capital-labor ratio making its overall impact ambiguous. Aging speed (the expected change in old-age dependency in 20 years as a proxy for the future old age dependency ratio) may exert a positive effect on saving and the current account, driven by a higher life expectancy and the resulting need for higher lifetime savings. As many effects work in opposite directions and interact with each other, net effects are not clear cut in open economies. The savings-investment balance also determines interest rates. However, having an open capital account and being a euro area member, Latvia’s interest rates are largely determined by regional and global savings and investment dynamics.10

Potential Impact Channels of Demographic Shifts
SavingInvestmentCurrent AccountInterest Rates
Population GrowthLargely Exogenous
Old-Age DependencyAmbiguousLargely Exogenous
Aging SpeedLargely Exogenous
Source: International Monetary Fund (IMF), 2016. Methodological Note on EBA-Lite, EBS/16/8, February (Washington)
Source: International Monetary Fund (IMF), 2016. Methodological Note on EBA-Lite, EBS/16/8, February (Washington)

15. The projected population rebalancing will impose fiscal costs, which could become a drag on growth. Spending on health and long-term care is set to rise as the old-age dependency ratio is projected to increase from about 29 percent in 2015 to 48 percent in 2050, and life expectancy is projected to increase from about 74 to 80 years. Spending on old-age pensions on the other hand is projected to decrease over time, resulting in falling replacement rates, which is a direct consequence of the mechanics of Latvia’s notional defined contribution system assuming unchanged policies.11 Using long-term pension and health spending projections from Clements et al. (2014; 2015), Figure 2 shows the difference in the fiscal balance between the net-migration and no-aging scenarios. The fiscal deficit is projected to be about 2.5 percentage points higher under the net-migration scenario than under the no-aging scenario by 2050. Assuming that public investment decreases by an equal amount to offset the additional deficit, would result in lower average growth of 0.2 percentage points in 2020–50. This estimate does not include possible second-round effects, which could potentially arise from lower productivity-supporting investment.

Figure 2.Fiscal Implications of Demographic Change

D. Labor Market Policies and Per-Capita Income Implications

16. We gauge the effectiveness of policies aimed at increasing LFPRs to offset the negative effects of demographic changes on workforce size and age structure. We examine 4 policy scenarios, which are calibrated based on historical experiences of other EU countries and compare the outcome to the net-migration scenario based on the UN WPP’s medium fertility scenario:

  • i. Policy scenario 1 assumes the LFPR for women aged 25–45 years to increase to the maximum level of the corresponding age-gender cohort observed in the EU at a constant annual increase of 1.3 percentage points.12

  • ii. Policy scenario 2 assumes an increase in the LFPR of men and women in age cohorts 55–69 of 0.9 percentage points per year until the EU maximum for countries with a retirement age of 65 is reached.13

  • iii. Policy scenario 3 assumes that the statutory retirement age for men and women increases to 67 by 2030, which triggers a coincident increase in the LFPR for age cohorts 55–69 to the average of EU countries that already have a legislated retirement age of 67. Subsequently, LFPRs are projected to increase with ILO projections.14

  • iv. Policy scenario 4 is a hybrid scenario combining scenarios 1–3. It assumes (i) an average annual increase in the LFPR for women aged 25–45 years of 1.3 percentage points until the EU maximum level is reached, (ii) the LFPR for age cohorts 55–69 and older to reach the maximum level of EU countries that already have a legislated retirement age of 67 by 2030, after which LFPRs are projected to increase with ILO projections.15

17. A combination of policies yields the strongest effect on the workforce size offsetting about one third of the negative growth effect induced by demographic shifts. Increasing female labor force participation of younger cohorts has a limited effect under scenario 1, as female LFP is already comparatively high, and thus the maximum level observed in the EU is reached quickly (Figure 3). Scenario 3 outperforms scenario 2 by 2 percentage points in the long run. Overall, the combined policies under scenario 4 are most effective in slowing the labor force decline by about 6 percentage points by 2030 and about 7 percentage points by 2050 compared to the net-migration scenario. Under the combined scenario, long-run growth is projected to be 2.3 percent in 2020–50 and thus about 0.3 percentage points higher than under the net-migration scenario.

Figure 3.Labor Force Participation Rates Across Countries and Policy Implications

Cumulative Labor Force Decline(In percent, 2015=100)
20302050
Baseline-15-28
Scenario 1-13-27
Scenario 2-13-27
Scenario 3-13-25
Scenario 4-9-21

18. Latvia is currently on a path to income convergence with Western Europe, but per-capita GDP levels may turn out lower as demographic headwinds are taking hold. Under the no-aging scenario, per-capita income is projected to reach almost 60 percent of the EA-19 average by 2050. Equivalently, this means that, under the no-aging scenario, Latvia’s real per-capita income level is projected to be about 2.8 times higher in 2050 than in 2015. As Latvia’s population declines and ages faster than Western Europe’s, this likely presents an upper bound absent any policies. As such, adding the projected natural population change as well as net migration to the underlying population dynamics would result in an income, which is about 2.5 times higher by 2050. Including the productivity-restraining effect of workforce aging would yield a per-capita income, which is about 1.8 times higher by 2050. Consequently, swift implementation of policies, which aim at arresting the labor force decline and increasing productivity will be paramount to counter these effects.

E. Policy Options

19. Policies to curb shrinkage of the labor force should aim at slowing the population decline and increasing participation rates. Better utilizing the remaining workforce through an increase in labor force participation among certain groups could be achieved by increasing the retirement age, for example by linking it to life expectancy. Providing more flexible work arrangements, including increased part-time work, both for workers transitioning into retirement and parents of young children can enable longer working lives for an aging workforce. ALMPs and training should be aimed at reducing skills mismatches and increasing the share of workers whose productivity increases with age. Slowing or reversing emigration could be achieved by creating a more attractive environment (e.g. through improving governance) that encourages potential emigrants to stay, as could engaging with the diaspora and promoting return migration (Atoyan et al, 2016).

Income Convergence of Latvia and EA-19

(EUR and percent)

Sources: UN WPP; ILO; WEO; IMF Staff Calculations

GDP per Capita in 2050

(Index, 2015=100)

Sources: UN WPP; ILO; IMF Staff Calculations

20. Structural reforms can support TFP growth. Addressing structural and institutional obstacles that prevent the efficient use of available technologies, or lead to inefficient allocation of resources, will be key to reaching this goal. The largest efficiency gains are likely to come from improving the quality of institutions (such as protection of property rights, upgrading legal systems including insolvency and judicial reforms), and increasing access to financial services (especially for small, but productive firms) (IMF, 2016b; IMF, 2016c). Furthermore, reducing the regulatory burden and red tape (OECD, 2017) for businesses and further improving corporate governance of state-owned enterprises would foster competition and efficient resource allocation, as would greater technology diffusion. Fiscal structural reforms, aimed at improving efficiency in the tax system, can also boost firm-level productivity by reducing resource misallocation (IMF 2017b). Structural reforms in the areas of R&D and education would also help boost productivity and reduce costs.

References

Prepared by Andreas Tudyka.

This SIP is based on the cross-country project “Demographic Headwinds to Convergence in Eastern Europe.”

Several studies document a negative impact of a higher dependency ratio on per capita GDP growth in different parts of the world, e.g. Persson (2002) for the US; Bloom, Canning and Malaney (2000) for East Asia; Aiyar and Mody (2013) for India.

We follow a standard Cobb-Douglas production function specification with constant returns to scale, where output per worker (yit) is given by: yit=kitα(Aithit)1α, where kit is capital per worker, Ait is total factor productivity, hit is human capital per worker, and α is the share of capital in output. All series are taken from the Penn World Table 9.0. Moreover, we decompose aggregate employment as Lt=Σj1jNtjLFPtjEtjwtj, where Ntj is the population in age-gender cohort j in year t, LFPtj is cohort-specific labor force participation, Etj is the cohort-specific employment rate, and wtj is a weight factor to adjust for the difference between the number of employees and the effective units of labor supplied. 1 – α is obtained as the share of the wage bill in GDP and the capital stock is estimated using the perpetual inventory method. Abstracting from policy changes, TFP is assumed to grow at its post-recession trend, and capital accumulation is assumed to follow the balanced-growth condition 1+gtK=(1+gtTFP)1/1α*(1+gtL).

The projections assume a balanced growth path in that capital adjusts to maintain a constant capital-output ratio.

The difference between the net-migration and aging scenarios is only 0.3 percentage points as emigration in projected to slow significantly after 2027.

Experienced individuals may supply more labor in response to wage augmenting technological innovations. At the same time, higher income arising from faster aggregate productivity growth may induce older workers to leave the labor force. Hence, the direction of the possible endogeneity bias is unclear (Adler et al 2017).

This coefficient reflects the historical relationship between workforce aging and TFP growth and does therefore preclude any behavioral shifts that may occur in the future. For example, one may consider that the definition of “older workers” has changed over time and will continue to do so. As such, calculating the impact on GDP growth using the current definition of “older workers” may lead to inflated estimates as longer and healthier life spans lead to workers continuing to be productive even at higher ages. Accounting for this upward shift in the definition of ‘older workers” is very difficult however, as the speed of the shift, as well as the coefficient quantifying the strength of the relationship are difficult to predict.

This exercise can also be conducted in a backward-looking fashion. Accordingly, workforce aging has reduced TFP growth by about 0.2 percentage points per year from 1990 to 2017 on average.

In the extreme case of perfect capital mobility, arbitrage in financial markets should equalize interest rates across borders, and demographic factors of each country should not have an impact on domestic interest rates.

In an NDC system, contributions over an unchanged length of working period will have to cover consumption over longer retirement period as life expectancy increases. At the time of retirement, an annuity is calculated by dividing the individual’s account value by a divisor reflecting life expectancy at the date of retirement. An increase in life expectancy therefore reduces the annual benefit such that the net present value of total expected pension benefits is nearly invariant to changes in the cohort’s remaining life expectancy and the individual’s retirement age. Overall, this would lead to continuously falling replacement rates, which seems unrealistic given the resulting impact on old-age poverty.

The calibration is based on the average annual increase of Spain, which was the best performer in the EU in 1995–2016.

The calibration is based on the average speed of increase in senior participation rates for EU countries which did not change the statutory retirement age in 2000–16.

For age cohorts 55–59 and 65–69 no increase is projected, as the current LFPR is already above the EU average of countries with a retirement age of 67.

For female age cohorts 55–59 and 65–69 no increase is projected as the current LFPR is already above the EU average of countries with a retirement age of 67. speed of increase in senior participation rates for a subsample of countries with unchanged statutory retirement age in 2000–16. The average rate of increase is 0.9.

Other Resources Citing This Publication