048 048 This paper is a product of the Poverty and Equity Global Practice Group and the Macroeconomics and Fiscal Management Global Practice Group. It is part of a larger effort by the World Bank to provide open access to its research and contribute to development policy discussions around the world. The authors may be contacted at mmatytsin@worldbank.org, lmoorty@worldbank.org and kaspar.richter@ec.europa.eu. The Poverty & Equity Global Practice Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. ‒ Poverty & Equity Global Practice Knowledge Management & Learning Team This paper is co-published with the World Bank Policy Research Working Papers. From Demographic Dividend to Demographic Burden? Regional Trends of Population Aging in Russia Mikhail Matytsin, Lalita Moorty and Kaspar Richter1 Key words: Demographic Dividend, Population Aging, Regional Convergence, Russian Regions JEL Classification Numbers: J11, O47, O15, O52, 1 This paper was written by Mikhail Matytsin (World Bank), Lalita Moorty (World Bank) and Kaspar Richter (European Commission) as a background paper for the 2015 World Bank report "Russia: Economic Impact of Aging", launched while Kaspar Richter was working at the World Bank. The views expressed in this paper are those of the authors and should not be attributed to the World Bank or the European Commission.  Summary Russia benefited from favorable demographic changes until recently. Since the early 1990s, as in many East European countries, Russia experienced three transitions – political, economic and demographic. On the demographic front, Russia entered the transition period with a predominantly young population and a low dependency ratio, and benefited from large cohorts of people entering the labor market. However, with declining fertility rates and longer life expectancy, this has changed in recent years and the population in Russia is projected to shrink and age in the coming decades. Will this affect Russia’s longer-term growth prospects?2 This paper looks at the impact of demographic trends on regional economic convergence. It asks the question: do regions with a higher share of working-age population grow faster? The paper is organized as follows: first, we contrast demographic trends in Russia with those in Europe, and explore the link of demographic change and growth for Russia's regions from 1996 to 2011. Second, we investigate econometrically whether the demographic dividend3 supported regional convergence over this period. Third, drawing on Rosstat4 population projections, we simulate the demographic dividend. The fourth and final section draws policy conclusions. Section I: Demographic and economic trends We look at demographic changes in Russia to inform the subsequent analysis. First, we compare demographic trends in Russia with Eastern and Western Europe drawing on UN population statistics and projections.5 Second, we look at the interplay of growth and demography across Russian regions based on data from Rosstat using the national definition of age groups. Russia and Europe We compare Russia with the 15 high-income economies of Western Europe and 21 transition economies of Eastern Europe. The UN data track populations in 5-year intervals. We find four interesting trends related to population, working age population, dependency ratios and composition of dependency ratios: 2 Existing evidence supports the link between growth and demographic changes. For example, the East Asian and Irish economic miracles are often attributed in part to changes in age structure. However, Bloom and Canning (2004) argue that countries benefit from favorable demographic trends only if their economies are open. 3 What is a demographic dividend? It refers to an opportunity that opens up when the labor force in a country grows faster than the population dependent(children and elderly) on it – and therefore leads to increases in per capita income (all else equal). As countries develop, they undergo a demographic transition where fertility rates fall. With fewer children due to declining fertility and fewer elderly people because of lower life expectancy, the largest segment of the population is of working age. This lowers the dependency ratios leading to a demographic dividend, where countries experience increases in per capita income if effective public policies are in place. Eventually, this dividend disappears as lower birth rates lead to lower growth in labor force and better health practices lead to an expanding elderly population. This saps additional income and puts to an end the demographic dividend. 4 Russia’s official statistical agency. 5 We use UN definitions and data to facilitate cross country comparisons. But we switch to using Rosstat data and definitions when analyzing Russian regions. Although the levels may differ slightly, the trends and structures are similar in both sources. 2  Russia’s population declined since 1995, roughly at the same pace as in Eastern Europe (Figure1). The decline is set to continue over the next 40 years. This is in contrast to the rise of the population in Western Europe. Globally, Russia is projected to move from being the ninth most populous nation in 2010 to the 14th most populous nation in 2050.  In contrast to trends in the total population, the working age population (ages 15-64) in Russia increased from 1990-2010, but is projected to drop sharply in the coming decades at a rate similar to Eastern Europe (Figure1). According to UN projections, the working age population in Russia is set to decline from 103 million in 2010 to 76 million in 2050. Western Europe experienced stronger increases from 1990-2010, and is set to experience more moderate declines in the coming decades.  The trends in working-age population translate into a v-shaped pattern for dependency ratios, defined as the share of the population aged less than 15 years or older than 65 as a ratio of population aged 15 to 64 (Figure 2). In Russia, the dependency ratio declined about 12 percentage points from 1995 to 2010 but is projected to increase about 15 percentage points from 2010 to 2030 because the decline in the working-age population is faster than in the total population. In Eastern Europe, the V-shape is flatter than in Russia. In Western Europe, the dependency ratio was constant in the last two decades, and is set to increase about 15 percentage points by 2030, albeit from a higher level. 3 Figure1: Trends in population and working age population in Russia compared to Western and Eastern Europe, 1990-2050 110 100 Population Aged 15 to 64 (2010=100) 105 95 Population (2010=100) 100 85 90 95 80 90 75 85 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Year Year Western Europe Eastern Europe Russia Western Europe Eastern Europe Russia Source: UN World Population Prospects: The 2010 Revision. Projections after 2010, median variant. Source: UN World Population Prospects: The 2010 Revision. Projections after 2010, median variant. Figure 2: Dependency ratios in Russia compared to Eastern and Western Europe, 1990-2030 65 60 Dependency Ratio 50 45 40 55 1990 1995 2000 2005 2010 2015 2020 2025 2030 Year Western Europe Eastern Europe Russia Source: UN World Population Prospects: The 2010 Revision. Projections after 2010, median variant.  When we separate the two components of the dependency ratio, it becomes clear that the changes in the dependency ratio prior and after 2010 are driven by different factors (Figure3). The decline in the dependency ratio prior to 2010 is linked to the drop in the share of the population aged 0 to 14 relative to the population aged 15 to 64 (young-age dependency ratio). By contrast, the rise in the dependency ratio after 2010 is driven by the rise in the share of the population aged 65 or older relative to the population aged 15 to 64 (old-age dependency ratio). The same is true for Eastern Europe. In Western Europe, in the last two decades, the moderate rise in the old-age dependency ratio was compensated by small decreases in the young-age dependency ratio, whereas the future rise in the dependency ratio is driven, as in Eastern Europe and Russia, by the rise in the old-age dependency ratio. 4 Figure3: Old and young age dependency ratios in Russia compared to Eastern and Western Europe, 1990-2060 50 50 Young-Age Dependency Ratio Old-Age Dependency Ratio 40 40 30 30 20 20 10 10 1990 2000 2010 2020 2030 2040 2050 2060 1990 2000 2010 2020 2030 2040 2050 2060 Year Year Western Europe Eastern Europe Russia Western Europe Eastern Europe Russia Source: UN World Population Prospects: The 2010 Revision. Projections after 2010, median variant. Source: UN World Population Prospects: The 2010 Revision. Projections after 2010, median variant. Three factors account for the rapid aging of Russia’s population in the coming decades (Figure 4). First, fewer people are entering the labor force today due to the sharp decline in fertility rates in the early 1990s. Second, life expectancy, which declined between around 1985 to the 2005 before increasing in recent years, is projected to continue to rise in the future. Third, this decade, the large cohorts born in the mid- and late 1950s are set to retire. In the late 2030s and 2040s, the large cohorts of children born in the 1980s in response to Soviet family policy will retire. Figure 4: Total fertility rate and life expectancy in Russia compared to Eastern and Western Europe, 1990-2010 80 2 Total Fertility Rate (Birth Per Woman) Life Expectancy at Birth (Years) 1.8 75 1.4 1.6 70 1.2 65 1 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Year Year Western Europe Eastern Europe Russia Western Europe Eastern Europe Russia Source: World Development Indicators Source: World Development Indicators Regional trends in Russia We explore the link between demographic trends and economic growth at the regional level in Russia.6 This regional focus is motivated by three factors. First, cross-country analyses of growth are often unable to factor in country-specific circumstances. Therefore, we investigate differences within a country, while at the same time exploiting the heterogeneity of demographic trends among regions. For example, the share 6 For this, we use Rosstat data and Russian definitions – where working age population is from 16-59 years for men and 16-54 for women. 5 of the population aged 65 or older ranges from 2 percent in Yamalo-Nenetsky AO to over 16 percent in Tulskaya oblast. Second, Russia’s shift in demographic changes will affect most regions differently. Third, recent research suggests that the evidence of convergence7 of output in Russian regions since the mid-1990s is mixed (Guriev and Vakulenko 2012). This evidence contrasts with trends for other countries, including the states in the United States or countries in the European Union. In view of large changes in demographic composition across regions, it is important to investigate how demographic factors influence economic convergence. Data trends show evidence of economic convergence. Since we are interested in the link of demographic change with economic convergence, we split the regional sample into two using the median real GDP per capita income level for 1996, the first year in our sample. The two groups are labeled “poor” and “rich” regions. Data show some evidence of unconditional convergence. Econometric estimations of the next section confirm the weaker form – conditional convergence. Using control variables for differences in economic, geographic or institutional factors, poor regions grew faster than rich regions since 1999 (Figure 5). Figure 5: Real GDP Per Capita in Poorer Regions Grew Faster than Richer Regions, 1996-2011 .1 200 average annual GDP growth (percent) Real GDP Per Capita, 1996=100 180 .08 160 .06 140 .04 120 100 .02 1996 1999 2002 2005 2008 2011 3 4 5 6 Year GDP per capita in 1996 (log format) Rich 1996 Poor 1996 poor 1996 rich 1996 What were the underlying demographic trends in the regions? The decline in total population was similar for poor and rich regions. However, poor regions experienced a stronger increase in working-age population up to 2007 compared to rich regions (Figure 6). Conversely, since the early 2000s, poor regions experienced a faster drop in dependency ratios than richer regions (Figure 7). 7 Β-convergence can be unconditional and conditional. Unconditional convergence assumes that all countries (or regions) converge to the same steady state point. So based on their initial level of development, countries (or regions) would grow at different rates to reach this steady state point – higher the GDP per capita, the lower its growth rate and vice versa. The implicit assumption here is that countries (or regions) have similar structures. Unconditional convergence, on the other hand, assumes that countries (or regions) converge to different steady state points because they have different structures. Convergence is therefore said to be conditional on the structural factors (for example, geography, culture, natural resources, demographics etc.). 6 Figure 6: Broad Regional Demographic Trends in Russia, 1996-2011 100 106 Working-Age Population (million), 1996=100 Population (million), 1996=100 104 98 102 96 100 94 92 98 1996 1999 2002 2005 2008 2011 1996 1999 2002 2005 2008 2011 Year Year Rich 1996 Poor 1996 Rich 1996 Poor 1996 Figure 7: Dependency ratios fell faster in poorer regions compared to richer regions, 1996-2011 A. Total dependency ratio B. Old-age dependency ratio 100 115 Dependency Ratio, 1996=100 95 110 Dependency Ratio, 1996=100 90 105 85 100 80 95 75 90 1996 1999 2002 2005 2008 2011 1996 1999 2002 2005 2008 2011 Year Year Rich 1996 Poor 1996 Rich 1996 Poor 1996 Looking at projected trends, the dependency ratio is expected to be larger in the poorer regions in the future. This bivariate analysis suggests that faster growth in poor regions in the past might be in part related to the more favorable demographic trends. The flipside of this is that the worse demographic trends in poor regions in the future could dampen economic convergence. Section II: Econometric estimation Theoretical model Following Aiyar and Mody (2011), we use a conditional β-convergence equation to study the effect of demographic trends on regional GDP growth per worker8 in logarithmic form: ∗ 1 , z – Logarithm of regional GDP per worker, 8 Here, regional GDP per worker refers to regional GDP per capita of economically active population – employed and unemployed. We considered looking at GDP per employed worker only, but given that conventional definition is per capita of economically active population, we decided to use the former in our analysis. We also checked our results using GDP per employed worker and do not find much of a difference. 7 - Growth in regional GDP per worker, ∗ - Long-run (steady-state) level of regional GDP per worker, which depends on social, economic, geographical factors, – Initial level of regional GDP per worker, - Speed of convergence. The equation shows that the gap between the steady state level of GDP per worker and the level at the beginning of the period determines the growth in GDP per worker over time. Therefore, the equation posits that growth in labor productivity (GDP per worker) over time in less developed regions will be higher than in more developed regions if they are to reach the long-run conditions ∗ . These long-run conditions could depend on a number of factors – time invariant or not ( ∗ . Examples of time invariant factors include geographical location, climate, or natural resource endowments, and these are captured by fixed individual effects ( . Examples of time variant factors include quality of labor (level of education), income inequality, hard and soft infrastructure, etc. These factors (or available proxies) further would be included in the estimated econometric equation as control variables in the panel specification ( . Rewriting equation (6) from Annex 4 for each region (i) and time period (t) and relaxing the relation of coefficients we get the model that could be estimated using various econometric techniques. Description of data used in this study is in Annex 3. 7 , , , , , , , ,, , – Level of GDP per capita in the region in period t, , – Working age ratio in the region in period t, , – Participation in the region in period t, . – Growth rate (difference in logarithms) of the variable, , – Matrix of control variables, which are time-varying factors affecting steady-state GDP growth of the region, – Individual effects, including all time-invariant characteristics of region, – Time effects to control for macroeconomic changes in the Russian economy, , – Shocks (independent and identically distributed with zero mean). The coefficients in Equation (7) are expected to have the following signs. We posit that initial level of GDP per capita has a negative effect on the growth rate (reflecting regional convergence). Both initial levels and growth rates of the working age ratio and participation rate should have positive signs. The signs of other factors cannot be defined a priori and would depend on the set of variables used for controlling the long- run level of real GDP. We use a panel specification, which is important given the high heterogeneity among Russian regions.9 This model allows us to use individual fixed effects for each region to capture features that are invariant over time. We also introduced time effects to capture the macroeconomic changes that have taken place at the regional level. A panel model with individual fixed effects and time effects is not flexible enough to 9 A cross-section approach requires a wide range of control variables to capture all possible factors that influence regional heterogeneity – but we do not have the relevant data. 8 test for regional conditional convergence. As mentioned before, the idea of conditional convergence is that the steady state point could be different for each region depending on its characteristics. The model with individual effects captures this idea only partially because the individual effects are fixed in time and the only time variation could be through time dummies, which are common for all regions. Therefore, they could reflect only general macroeconomic shocks that affect the whole country, but cannot distinguish between regions. However, it is very likely to be the case that some regions could change their steady stage regime during the period under consideration. For example, a region could accumulate higher stock of social or physical capital and become more productive. Alternatively, a region could start to derive benefits from its geographical location. In this case, the long run level of GDP per worker could change and the model with only time and individual effects cannot deal with this. This effect is partly compensated by a relatively short sample of time-series used in our research. In this situation, not many regions could change fundamental factors that affect the steady state regime. We would like our model to be relevant for forecasting – so we needed the model to be more flexible. To deal with this, we use control variables that could increase the quality of estimation and allow greater flexibility in the long run. In theory, these variables should reflect social and physical capital stock, some policy implications and other relevant factors. In practice, we include as many relevant factors as we were able to collect for Russian regions in the period of 1997-2011. There are five control variables in our model. Two of them are indicators of social situation in a region – the poverty rate and the Gini coefficient, which is the measure of income inequality. One more control variable used in our estimation is the share of urban population – it is a proxy for industrial development of the region what is important given Russia’s economic structure. In the 5-year periods model presented later, we control for human capital (school enrollment ratio) and gender (sex ratio). We estimate the model in logarithmic form. There could be different approaches to estimating equation (7) for Russian data. In this paper, we present two approaches that supplement each other. The first is direct estimation for the whole sample of regions and time periods. In the second, we aggregate data into 5-year periods. To estimate direct panel model we have to revise equation (7) slightly: 8 , , , , , , , , , The interpretation of Equation (8) is as follows. It assumes the changes between GDP and working age ratio and participation rate between t and initial period t-1. In this specification, the initial period is rolling and lagged from current period only for one year. This assumption is debatable and we return to this discussion later when we use the second approach based on aggregating data into 5-year periods. However, for the moment it is important to stay in this specification when we use Arellano-Bond estimator. The results of direct panel estimation are shown in the first column of Table 1. All the coefficients have expected signs and are significant. Coefficients (α1 and α2) are of interest and they have the expected positive signs. It means that regions with higher level of working age ratio have higher growth of GDP per capita (ceteris paribus). The elasticity is high – for one percent of working age ratio, real GDP growth 9 would increase by almost 1.5 percent.10 The growth rate of the working age population is insignificant in this specification, which could be caused by possible endogeneity. Higher initial level of participation rate as well as higher growth in it also positively contributes to GDP per capita growth. One percentage point of each could add 0.4 percent to GDP growth. Direct estimation could lead to biased results because of endogeneity. The problem is that working age ratio and participation ratio are contemporaneous with the dependent variable. And in this case they may be affected by the previous level of GDP per capita and as a result by current level of GDP per capita growth (what is simple difference of current and previous level). The main channel of such reverse causality is migration. Internal migration is not very high in Russia (see Guriev and Vakulenko 2012), but could still be significant. There are no strong administrative barriers for migration in Russia and more developed and faster growing regions could attract more immigrants from other regions. And the majority of migrants would likely be working-age and this would increase the working-age ratio. To control for this, we use two methods. Table 1: Estimation Results (equation 8) (1) (2) (3) (4) Panel Estimation Panel Estimation  Arellano‐Bond model Arellano‐Bond model Growth in GDP per capita Growth in GDP per capita GDP per capita GDP per capita VARIABLES (in logs) (Adjusted for migration) (Adjusted for migration) initial GDP per capita  ‐0.234*** ‐0.218*** 0.712*** 0.739*** (0.028) (0.028) (0.009) (0.011) initial working‐age ratio 1.497*** 1.453*** 1.788*** 1.640*** (0.137) (0.130) (0.044) (0.028) working‐age ratio growth 0.745*** 0.537*** (0.264) (0.118) adjusted working‐age ratio  growth 1.201*** 0.977*** (0.242) (0.084) initial participation rate  0.443*** 0.429*** 0.636*** 0.592*** (0.061) (0.057) (0.025) (0.027) participation rate growth 0.449*** 0.425*** 0.586*** 0.547*** (0.050) (0.049) (0.019) (0.021) Observations 1,162 1,162 1,083 1,083 R‐squared 0.262 0.273 Number of reg 78 78 78 78 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 First, we use growth of the migration-adjusted working age ratio. By this we mean the estimated working age ratio (and its growth) which could be without any migration flows (both internal and external). This variable in theory is free of endogeneity and that it why could be used in the estimation. To estimate this, we would need to subtract the net inflow of migrants from the total population of each region by age groups and recalculate the growth of the working age ratio. However, we did not have good information for migration flows in the Russian regions.11 Instead, we had good demographic information for estimating the 10 The high elasticity could be driven by annual dynamics. The approach that follows is less sensitive to short-term fluctuations. 11 The information about the migration flows is collected directly but the quality of these data is poor. In our approach, we do not distinguish between the interregional and external migration. 10 natural increase of population (the difference between birth and mortality rates). We used these data to estimate the adjusted working age population as working age population minus the net migration inflow, which in turn was estimated as natural increase subtracted from total population increase. This method assumes that all migration inflow is of working age, which is a conservative approach.12 But using both adjusted and non-adjusted measures of working age ratio allows us to estimate an upper bound for adjustment. The results of estimating the contribution of adjusted working are ration growth to GDP per capita growth are shown in Column (2) of Table 1. We do not find significant differences compared to the unadjusted model except the growth of the working age ratio. The contribution of the growth rate of this factor is significant in the adjusted model. It means that the bias caused by possible migration into faster growing regions could be significant. To verify this result, we used another approach estimating dynamic panel developed by Arellano and Bond (1991). This method is based on GMM estimation, where current values of contemporaneous or predetermined values of regressors are instrumented by their lagged values.13 In theory, this method should provide consistent estimations without any systematic bias. The results of the estimation is presented in column (3) and (4) of the Table 1. We would like to pay attention that coefficient for the growth rate of real GDP per capita ρ+1. To compare the results with first column one should subtract 1 from the value of the coefficient. Column (3) represents the results for not adjusted working age ratio, while column (4) is for adjusted. As we mentioned before, in theory the Arellano-Bond approach should lead to consistent estimates and there is no need for additional adjustment. In practice, we estimated both models – with and without the adjustment of working age ratio growth to be on the safe side because we reduced the number of instruments significantly. The quantitative results of column (3) and (4) are very close to each other and to the general panel regression ((1) and (2)). We conclude that there is negative significant contribution of the former level of GDP per capita showing convergence. The positive effect of working age growth and level in previous period ( and ) is still significant for adjusted for migration model that confirms tested hypothesis that demographics have a signification impact on growth. Following Aiyar and Mody (2011), we split our sample into time periods. In principle, a decade might seem appropriate for demographic changes. However, in view of the small time period, we use 5-year periods. This approach can be justified for the case of Russia, as our sample period coincided with fast demographic changes in Russia. For growth rate within the 5-year period, we use the average of annual growth rates and for initial level – the value in the first year of the period. For control variable, we also calculated the time means. In this case, the equation for estimation is modified slightly. The only difference lies in the initial 12 This adjustment is relevant only for working age ratio growth. While the initial (previous) level of working age ratio could also be affected by migrants who came earlier, we cannot distinguish between them and natural population of the region. However, this indicator is not contemporaneous with current level of GDP and that is why it bias of the coefficients should be smaller. 13 The method is based on the property that past value of dependent variable and regressors are not contemporaneous with the error term and can therefore be used as instruments. They correlate with the current levels of regressors and that is why could be strong enough instruments. The idea is to use as many instruments as possible. In our case the contemporaneousness of regressors limit the amount that could be used. We could use only lagged levels of dependent variable as well as regressors, while in standard Arellano-Bond approach all regressors (including future) are used as instruments. 11 levels of GDP per capita, working age ratio and participation rate. They are noted with the sub-index 0 and relate to the first year of the 5-year period – 1997, 2002 and 2007, respectively. The results are presented in Table 2. Column (1) is for not adjusted working age ratio, column (2) is adjusted. 9 , , , , , , , , , Table 2: Results Using 5-Year Averages (1) (2) Panel Estimation Panel Estimation  Growth in GDP per capita Growth in GDP per capita VARIABLES (in logs) (Adjusted for migration) 1 initial GDP per capita  ‐0.056*** ‐0.048** (0.019) (0.019) 2 initial working‐age ratio 0.190 0.244* (0.150) (0.139) 3 working‐age ratio growth 0.993 (0.892) 4 adjusted working‐age ratio growth 1.627*** (0.601) 5 initial participation rate  0.102 0.099 (0.065) (0.064) 6 participation rate growth 0.700*** 0.620*** (0.194) (0.171) Observations 234 234 R‐squared 0.481 0.494 Number of reg 78 78 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 This model confirms the conclusions from the dynamic panel approach. The coefficients ρ is negative and significant – higher level of GDP per capita in the beginning of the 5-year period leads to lower average growth rate in the period. The elasticity is much smaller than in the dynamic panel model. The impact of working age ratio on to GDP per capita growth is positive. In the specification without the adjusted working age ratio (column (1) of Table 2), the coefficients and are insignificant. However, they are insignificant in specification with adjustment for migration (column (2) of Table 2). Again, considering longer periods (five years instead of one) the initial levels of working age ratio is less important for average GDP per capita growth, while the average growth rate of this ratio has coefficient significantly higher than one in both cases – adjusted and not adjusted. It means that one percent of working age ratio growth on average translates to more than two percent of regional GDP per capita real growth. We also note that participation rates also matter for growth. Robustness check To be confident that that our results are stable, we conducted robustness checks. We used several modifications of the model to see if our results were stable. As shown above, the results are stable. The main conclusion is that the working age ratio matters for economic growth in Russia at the regional and national levels. Other robustness checks include changing the definition of working age ratio. We tried to use a more commonly used retirement age of 64 for both genders. In this case, the econometric results did 12 not change dramatically. Most of the coefficients have almost the same values and levels of significance. The most important change is that level of working age ratio has higher elasticity – it matters more, while the growth rate of this modified ratio is insignificant in some models. We interpret this result as evidence of the demographic structure effect. From another point of view the Russian definition of working age looks more relevant, reflecting that people mostly retire close to the official age. We also made several other modifications. Among them, we changed the length of the period for the second model. We consider a shorter period (3 years instead of 5) and two longer periods (one 7 years and the other 8 years). We also dropped the observations for Moscow and Saint Petersburg, where the labor market and general structure of the economy could be different from other regions. In all cases, the results were stable. Some coefficients became insignificant, but the results do not contradict the baseline specification. Section III. Simulating the demographic dividend Sample period We use the estimated results to calculate the demographic dividend. When dependency ratios are low (either because of falling fertility rates or due to older generations having shorter life expectancy), it produces a demographic dividend (DD). If effective policies are in place to employ the burgeoning working age population, this could produce more rapid economic growth. There are at least three channels that lead to a DD (Bloom et al 2003) – labor force supply, savings and human capital accumulation. First effect is related to the higher share of people in working age translated to lower ratio of dependent categories of people (young-aged and old-aged). As the number of workers increases and the number of dependents declines (relatively), the output per capita of total population goes up. Another effect takes place through savings. Usually people save during the working period when they have the highest income. As the amount of people in the labor force increases, the savings also become bigger. These savings could translate into investment to support economic growth. In addition, the third effect is related to improving people’s quality of life and life expectancy. In these conditions, people could invest more in their education and health and become more productive and stay longer in the labor market. In this paper, we focus only on the first type of demographic dividend – the direct effect of increasing the working age ratio. Since we estimated two models using a dynamic panel approach and 5-year period panel approach, we could use either of them to calculate the demographic dividend. As in Aiyar and Mody (2011), we chose to use 5-year averages to forecast DD because it can smooth out year-to-year variations. Based on equation (7) we have the relation of average annual growth rate and the demographic factors (working- age ratio and its growth rate). To calculate the demographic dividend, we assume that the aging structure is constant for the sample period and equaled the aging structure of 1997 – the first year in our sample. Under this assumption, the working-age ratio for the all years in the sample period was constant and equaled the level of 1997. The growth rate of the working age population in this situation would be zero for all considered years. We assume that other factors are not fixed and take on their actual values in each year. Then equation (7) would be: 10 , , , , , , , 13 Subtracting equation (10) from equation (7) yields the formula for calculating the demographic dividend. The formula is quite intuitive. First term reflects the net increase in working age ratio compared to base (1997) year multiplied by the estimated elasticity. The second terms is the growth rate of working age ratio multiplied by its elasticity. We take the total growth rate since under the assumption of constant working age ration at the level of 1997 year the growth rate would be zero. 11 , , , We applied formula (11) to calculate the DD for the sample period of our model from 1997 until 2011. The results are presented in the table below. The results are both at the national and Federal Okrugs (district) levels. Federal Okrugs contain 7-12 regions, divided by geographical location. We also reported information for two groups of regions: rich/poor and leaders/laggards. Rich/poor is the division based on the level of GDP per capita in 1996 (the year before sample period). Leaders/laggards – based on average growth rates of first 5-year period. Leaders are 10 regions with highest average annual growth rate in the period of 1997-2001, laggards – with the lowest. Table 3: Demographic Dividend by Groups Demographic dividend (DD), p.p. GDP per capita annual  growth, % Net of DD GDP per capita annual  growth, % 1997‐2001 2002‐06 2007‐11 1997‐2011 1997‐2001 2002‐06 2007‐11 1997‐2011 1997‐2001 2002‐06 2007‐11 1997‐2011 Russian Federation 1.1 2.1 1.4 1.5 3.8 7.8 3.3 4.9 2.7 5.7 2.0 3.4 Central  Federal  Okrug 1.1 2.1 1.4 1.5 5.2 9.1 2.1 5.5 4.1 7.1 0.7 3.9 North‐ Western Federal  Okrug 1.0 1.9 1.0 1.3 3.9 8.5 3.8 5.4 2.9 6.7 2.9 4.1 Southern Federal  Okrug 1.0 2.2 1.5 1.6 4.2 7.4 4.6 5.4 3.2 5.2 3.1 3.8 North‐ Caucas Federal  Okrug 1.2 2.8 2.5 2.2 3.3 8.1 5.1 5.5 2.1 5.3 2.6 3.3 Volga Federal  Okrug 1.0 2.3 1.5 1.6 3.3 6.2 4.2 4.6 2.3 3.9 2.7 3.0 Ural  Federal  Okrug 1.1 2.2 1.1 1.5 3.5 7.8 2.7 4.6 2.4 5.6 1.6 3.2 Siberia Federal  Okrug 1.2 2.2 1.2 1.5 1.8 6.9 4.0 4.2 0.7 4.7 2.7 2.7 Far East Federal  Okrug 0.8 1.3 0.3 0.8 2.5 6.0 6.0 4.8 1.7 4.8 5.6 4.0 10 Leaders 1.0 2.2 1.5 1.6 5.9 10.0 6.3 7.4 4.8 7.8 4.8 5.8 10 Laggards 1.0 1.9 0.9 1.2 3.1 3.4 2.4 3.0 2.2 1.6 1.5 1.8 Rich 1.0 2.0 1.0 1.3 3.2 6.7 3.9 4.6 2.2 4.7 2.9 3.2 Poor 1.1 2.3 1.6 1.7 3.9 7.0 4.6 5.1 2.8 4.7 3.0 3.5 Table 3 shows that DD explains a large part of growth in regional GDP per capita. On average, at the national level, it explained 30 percent of total GDP growth. This contribution is in line with other countries. For example in India the DD explained up to 40% of total GDP per growth (Aiyar and Mody 2011). During 2002-06 the absolute contribution of DD (2.1 percentage points) was highest supported by both channels – fast growth and high initial level of working age ratio, but the relative contribution was the lowest (explained only 27% of total GDP growth) because of high oil prices and other factors driving the growth in Russia. In the first and third five-year periods, the DD was much smaller – 1.1 and 1.4 percentage points, respectively. In 1997-2001, DD was explained by the fast growth of working age ratio, while in the third period by high level of working age ratio. As other factors of economic growth were on average weak in late 1990s (including 1998 crisis) and after the 2008 crisis relative contribution of DD in first and third five- year periods was higher than in the second. On a regional level, the results are similar. Highest share of DD contribution to the total GDP per capita growth is in North Caucasus Federal Okrug where on average up to 40 percent GDP growth was explained by DD and in Siberian and Volga Federal Okrugs, where share of DD is higher than 35 percent. The smallest share of DD contribution is in Far East Federal Okrug – less than one-sixth of GDP growth was driven by 14 DD. The relative effect of DD was much higher for regions with low initial growth rates (laggards) comparing to leaders – 41 and 21 percent respectively. The difference in contribution of DD between initially rich and poor regions is 23 percent and 33 percent of total GDP growth. Forecasting of demographic dividend We use the estimated model to forecast the demographic dividend. To do this, we used the official demographic forecast provided by Rosstat. It compares the population by regions and divides it into three age groups: before working age, working age and older than working age. Also there are three scenarios – named high, low and baseline – but we renamed the scenarios. According to Rosstat, the high scenario assumes higher amount of total population and the higher fertility rate what translates to higher young- dependency ratio and therefore corresponds to a lower working is ratio. The low is vice-a-versa. Instead, we chose to call them high/low fertility rate scenarios as it reflects they structure better (Figure 8 in the Annex 1). We simulated the DD by three scenarios for all regions (results reported only for the macro regions in table 4) and for the national level assuming that aging structure in Russia would be the same as it was in 2011 – the last year for which we have actual data. Table 4: Forecasts of Demographic Dividends Simulation of demograpfic dividend based on official  population by age  groups forecast Scenario 1 (low fertility) Scenario 2 (baseline) Scenario 2 (high fertility) 2012‐16 2017‐21 2022‐30 2012‐30 2012‐16 2017‐21 2022‐30 2012‐30 2012‐16 2017‐21 2022‐30 2012‐30 Russian Federation ‐1.2 ‐1.8 ‐1.5 ‐1.5 ‐1.5 ‐2.3 ‐2.2 ‐2.0 ‐1.6 ‐2.6 ‐2.5 ‐2.3 Central  Federal  Okrug ‐1.3 ‐1.9 ‐1.7 ‐1.7 ‐1.6 ‐2.5 ‐2.4 ‐2.2 ‐1.7 ‐2.7 ‐2.6 ‐2.4 North‐Western Federal  Okrug ‐1.3 ‐1.7 ‐1.3 ‐1.4 ‐1.6 ‐2.3 ‐2.1 ‐2.0 ‐1.7 ‐2.4 ‐2.2 ‐2.1 Southern Federal  Okrug ‐1.2 ‐1.8 ‐1.5 ‐1.5 ‐1.4 ‐2.2 ‐2.1 ‐2.0 ‐1.6 ‐2.6 ‐2.6 ‐2.3 North‐Caucas Federal  Okrug ‐0.6 ‐1.5 ‐1.5 ‐1.3 ‐0.8 ‐2.0 ‐2.1 ‐1.8 ‐1.2 ‐2.6 ‐2.8 ‐2.3 Volga Federal  Okrug ‐1.2 ‐1.9 ‐1.6 ‐1.6 ‐1.5 ‐2.4 ‐2.3 ‐2.1 ‐1.6 ‐2.7 ‐2.7 ‐2.4 Ural  Federal  Okrug ‐1.3 ‐1.7 ‐1.3 ‐1.4 ‐1.6 ‐2.3 ‐1.9 ‐1.9 ‐1.7 ‐2.5 ‐2.2 ‐2.2 Siberia Federal  Okrug ‐1.3 ‐1.8 ‐1.3 ‐1.4 ‐1.6 ‐2.3 ‐2.0 ‐2.0 ‐1.8 ‐2.6 ‐2.4 ‐2.3 Far East Federal  Okrug ‐1.2 ‐1.6 ‐1.2 ‐1.3 ‐1.5 ‐2.1 ‐1.9 ‐1.8 ‐1.7 ‐2.4 ‐2.2 ‐2.1 10 Leaders ‐1.1 ‐1.7 ‐1.5 ‐1.5 ‐1.3 ‐2.2 ‐2.1 ‐1.9 ‐1.5 ‐2.6 ‐2.5 ‐2.3 10 Laggards ‐1.2 ‐1.7 ‐1.3 ‐1.4 ‐1.4 ‐2.2 ‐1.9 ‐1.9 ‐1.6 ‐2.5 ‐2.3 ‐2.1 Rich ‐1.2 ‐1.7 ‐1.3 ‐1.4 ‐1.5 ‐2.2 ‐2.0 ‐1.9 ‐1.6 ‐2.5 ‐2.3 ‐2.2 Poor ‐1.2 ‐1.8 ‐1.5 ‐1.5 ‐1.4 ‐2.3 ‐2.2 ‐2.0 ‐1.6 ‐2.7 ‐2.6 ‐2.4 We see that in all scenarios for all macro regions (and almost all regions) the contribution of the DD would be negative for regional GDP per capita growth. Negative contribution increases after 2017 when the decline of working age ratio becomes faster and reaches low levels. The distribution of DD effect across Federal Okrugs is quite uniform. As we discussed before, the dependency ratio in the scenario with a high fertility rate is higher. Therefore, the negative contribution of DD in the scenario with a high fertility rate would be 1 percentage point higher than in the scenario with a lower fertility rate. It means that government is faced with a tradeoff. Supporting fertility and trying to increase the total population, the policy makers would put additional pressure on regional GDP growth at least in the medium term (until 2030). We are now in position to simulate the impact of increasing the retirement age. For this, we would need information on the “effective” retirement age – the real age at which people retire which could differ from the official retirement age in practice. Some people are allowed to and prefer to retire earlier because of specific labor conditions or problems with health, while others may choose to work longer. It means that on average people retire “gradually.” It is hard to estimate exact retirement age, so we assume that official retirement is a good proxy for this “effective” retirement age. We note in Table 5 that a gradual increase in 15 the retirement age to 64 for both men and women has a significant impact on the demographic dividend possibly because the share of population ages 60-64 for men and 55-64 for women is fairly large. Table 5: Demographic Dividend with Higher Retirement Age Simulation of demograpfic dividend based asuumtion of gradual  increase  of retirment age  and official  population by age  groups forecast Scenario 1 (low fertility) Scenario 2 (baseline) Scenario 2 (high fertility) 2012‐16 2017‐21 2022‐30 2012‐30 2012‐16 2017‐21 2022‐30 2012‐30 2012‐16 2017‐21 2022‐30 2012‐30 Russian Federation ‐0.2 1.0 1.8 1.0 ‐0.5 0.5 1.2 0.6 ‐0.6 0.3 0.9 0.3 Central  Federal  Okrug ‐0.2 1.1 1.8 1.0 ‐0.5 0.6 1.2 0.6 ‐0.6 0.4 1.0 0.4 North‐Western Federal  Okrug ‐0.2 1.2 2.0 1.2 ‐0.5 0.7 1.4 0.7 ‐0.6 0.5 1.2 0.6 Southern Federal  Okrug ‐0.1 1.1 1.9 1.1 ‐0.4 0.7 1.3 0.7 ‐0.5 0.4 1.0 0.4 North‐Caucas Federal  Okrug 0.0 0.6 1.0 0.7 ‐0.2 0.1 0.4 0.2 ‐0.5 ‐0.5 ‐0.1 ‐0.3 Volga Federal  Okrug ‐0.2 1.0 1.7 1.0 ‐0.4 0.5 1.2 0.6 ‐0.6 0.2 0.8 0.3 Ural  Federal  Okrug ‐0.3 1.0 1.9 1.1 ‐0.6 0.5 1.3 0.6 ‐0.7 0.3 1.0 0.4 Siberia Federal  Okrug ‐0.4 0.9 1.8 1.0 ‐0.6 0.5 1.2 0.5 ‐0.8 0.2 0.9 0.3 Far East Federal  Okrug ‐0.3 1.2 2.0 1.2 ‐0.5 0.7 1.4 0.7 ‐0.7 0.4 1.1 0.5 10 Leaders ‐0.1 1.0 1.7 1.1 ‐0.3 0.6 1.2 0.6 ‐0.5 0.3 0.8 0.3 10 Laggards ‐0.3 1.0 1.8 1.1 ‐0.5 0.5 1.3 0.6 ‐0.6 0.3 1.0 0.4 Rich ‐0.2 1.2 2.0 1.2 ‐0.5 0.7 1.4 0.7 ‐0.6 0.4 1.1 0.5 Poor ‐0.2 1.0 1.7 1.0 ‐0.4 0.5 1.1 0.6 ‐0.6 0.2 0.8 0.3 Section IV. Conclusions Absent policy changes, the aging of the Russian population is likely to have an adverse impact on growth and convergence. Russia's economy benefited from favorable demographic trends from the mid-1990s until the last few years. In the coming decades, however, Russia’s population is expected to age, which could dampen prospects for growth and convergence among Russia’s regions. This challenge underlines the importance of wide-ranging policy changes to help families, workers, and businesses to adjust to population aging. The paper discusses one crucial reform, the increase in the official retirement age. This policy would need to be complemented by measures to ensure that this translates into a rise of the effective retirement age. Other policies include, among others, the following: increased investment in preventative care to improve health status and thus enable people to work longer; lifelong learning programs and other steps to refresh the skills of an aging workforce; taking legal measures against age discrimination and reducing other constraints on mobility (e.g. making pensions portable); and the provision of quality day care and long term care to facilitate greater labor force participation (which would particularly benefit women, who provide the bulk of such care). By contrast, simply focusing on increasing fertility rates will only address the aging challenge partially. Without supporting policies, mothers might drop out of the labor market to take care of their children, which would further reduce the number of workers. All of these efforts would not only bolster growth and convergence, but also make social security systems more sustainable. 16 References 1. Aiyar, Shekhar, and Mody, Ashoka. 2011. The Demographic Dividend: Evidence from the Indian States. IMF Working Paper 11/38, International Monetary Fund, Washington, DC. 2. Arellano, Manuel, and Stephen Bond. 1991. “Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations.” Review of Economic Studies, 58 (2. 3. Barro, Robert, and Xavier Sala-i-Martin. 1995. Economic Growth. New York: McGraw-Hill. 4. Bloom, David E., and David Canning. 2004. Global Demographic Change: Dimensions and Economic Significance.” NIBR Paper No. w10817, National Bureau of Economic Research, Cambridge, MA. 5. Bloom, David E., David Canning, and Jaypee Sevilla. 2003. “The Demographic Dividend: A New Perspective on the Economic Consequences of Population Change. Rand, Vol. 5. No. 1274. 6. Bloom, David E., and Jocelyn E. Finlay. 2009. “Demographic Change and Economic Growth in Asia.” Asian Economic Policy Review, 4 (1). 7. Bloom, David E., and Jeffrey G. Williamson. 1998. “Demographic Transitions and Economic Miracles in Emerging Asia.” The World Bank Economic Review, Vo. 12, No. 3, 1998. 8. Brainerd, Elizabeth, and David Cutler. “Autopsy of an Empire: Understanding Mortality in Russia and the Former Soviet Union. Journal of Economic Perspectives, 19(1). 9. Denisova, Irina. 2010. Adult Mortality in Russia. Economics of Transition, 18. 10. Eich, Frank, Mauricio Soto, and Charleen Gust. 2012. “Reforming the Public Pension System in the Russian Federation. IMF Working Paper 12/201, International Monetary Fund, Washington, DC. 11. Guriev, Sergei, and Vakulenko, Elena. 2013. :Internal Migration and Interregional Convergence in Russia.” Social ScienceResearch Network. Available at SSRN 2200539. 12. Queiroz, Bernardo L., and Cassio M. Turra. 2010. Window of Opportunity: Socioeconomic Consequences of Demographic Changes in Brazil. Washington, DC: NTA. 13. Valulenko, Elena. 2014. “Does Migration lead to Regional Convergence in Russia?” Higher School of Economics Working Paper Basic Research Program 53/EC/2014. 14. Wang, F., and Andrew Mason. 2008. “The Demographic Factor in China’s Transition.” In China’s Great Economic Transformation, edited by L. Brandt and T. Rawski (pp. 136–66). Cambridge, UK: Cambridge University Press. 15. World Bank. 1993. The East Asian Miracle: Economic Growth and Public Policy. New York: Oxford University Press, 1993. 17 Annexes Annex 1. Charts Figure 8. Dynamics of dependency ratios and their forecast in different sources of data. 30% 25% 20% 15% 10% 1990 1995 2000 2005 2010 2015 2020 2025 2030 young_dep_ratio forecast (low population scenario) forecast (base scenario) forecast (high population scenario) 60+ dep ratio forecast (low population scenario) forecast (base scenario) forecast (high population scenario) 64+ dep ratio forecast (low population scenario) forecast (base scenario) forecast (high population scenario) UN young dep rat UN 60+ dep ratio UN 65+ dep ratio Figure 9. Dynamics of working age ratio and forecast 75% 75% 70% 70% 65% 65% 60% 60% 55% 55% 50% 50% 45% 45% 40% 40% 35% 35% 30% 30% 1990 1995 2000 2005 2010 2015 2020 2025 2030 wa_ratio (Russian) forecast (low population scenario) forecast (base scenario) forecast (high population scenario) wa_ratio (16‐64) forecast (low population scenario) forecast (base scenario) forecast (high population scenario) wa_ratio (World Bank) forecast (low population scenario) forecast (base scenario) forecast (high population scenario) share of 60(55)‐64 in population older w/a share of 60(55)‐64 forecast 18 Annex 2. Data 1. Russia consists of 83 regions (republics, oblasts, krais and so on). Because of regional conflicts, data for some regions is not available and are excluded from this analysis. Using data for 15 years and 78 oblasts means that we have 1170 observations. 2. Estimating the impact of demographic trends on regional growth requires availability of relevant data.  First, data on all relevant indicators has to be available at the regional level. Regional data – especially demographic data – is available in Russia and are posted on the Rosstat website. Besides, a number of variables are also available at the regional level – some of which were used as control variables in this analysis. There is, however, data on migration is of poor quality at the regional level. As we will see later, this information is necessary for controlling endogeniety in the demographic structure. Therefore, we used birth and mortality rates to calculate natural increases in population and then subtracted it from the regional population to get an approximation of migration statistics.  In addition, we need long time series because demographic changes are largely inert. Unfortunately, the time series in Russia is not long enough both at the regional and national level because of the relatively short period since the transition. Data are at best available for 15-20 years on a comparable basis. So this study is based on data for 15 years (1997-2011) – which is not long enough for demographic analysis. However, given that this period coincided with important demographic changes, we think that the analysis would still be relevant. 19 Annex 3. Descriptive Statistics GDP per capita (thsd 2003 constant rub) GDP per capita annual  growth, % 1997‐ 2001 2002‐ 06 2007‐11 1997‐ 2011 1997‐2001 2002‐06 2007‐ 11 1997‐2011 Russian Federation 55 69 102 75 3.5 7.5 3.1 4.7 Central  Federal  Okrug 67 86 136 96 4.9 8.7 1.8 5.1 North‐ Western Federal  Okrug 54 72 110 79 3.7 8.2 3.7 5.2 Southern Federal  Okrug 32 42 63 46 3.9 7.1 4.3 5.1 North‐ Caucas Federal  Okrug 18 23 35 26 3.0 7.7 5.0 5.2 Volga Federal  Okrug 46 54 78 59 3.1 6.0 4.0 4.3 Ural  Federal  Okrug 101 123 181 135 3.3 7.5 2.5 4.4 Siberia Federal  Okrug 47 56 80 61 1.6 6.7 3.8 4.0 Far East Federal  Okrug 66 79 110 85 2.3 5.9 5.7 4.6 10 Leaders 38 60 84 61 6.8 6.2 3.9 5.6 10 Laggards 49 51 67 56 ‐ 0.2 5.5 4.7 3.3 Rich 64 78 110 84 2.9 6.4 3.7 4.3 Poor 31 41 59 43 3.4 6.6 4.3 4.8 working age  ratio, % working age  ratio annual  growth, % 1997‐ 2001 2002‐ 06 2007‐11 1997‐ 2011 1997‐2001 2002‐06 2007‐ 11 1997‐2011 Russian Federation 57 61 63 60 0.6 0.6 ‐ 0.4 0.3 Central  Federal  Okrug 57 60 63 60 0.7 0.6 ‐ 0.3 0.3 North‐ Western Federal  Okrug 59 62 64 62 0.6 0.5 ‐ 0.4 0.2 Southern Federal  Okrug 56 59 62 59 0.6 0.6 ‐ 0.3 0.3 North‐ Caucas Federal  Okrug 54 58 62 58 0.7 0.8 ‐ 0.1 0.5 Volga Federal  Okrug 57 60 63 60 0.6 0.7 ‐ 0.3 0.3 Ural  Federal  Okrug 59 62 65 62 0.7 0.6 ‐ 0.5 0.3 Siberia Federal  Okrug 58 62 64 61 0.7 0.6 ‐ 0.5 0.3 Far East Federal  Okrug 62 65 66 64 0.5 0.4 ‐ 0.5 0.1 10 Leaders 57 61 64 61 0.6 0.7 ‐ 0.4 0.3 10 Laggards 59 62 65 62 0.6 0.6 ‐ 0.5 0.2 Rich 59 62 64 62 0.6 0.6 ‐ 0.5 0.2 Poor 56 60 63 60 0.6 0.7 ‐ 0.3 0.3 partcipation rate, % partcipation rate  annual  growth, % 1997‐ 2001 2002‐ 06 2007‐11 1997‐ 2011 1997‐2001 2002‐06 2007‐ 11 1997‐2011 Russian Federation 80 82 83 82 ‐ 0.2 0.2 0.8 0.3 Central  Federal  Okrug 80 85 86 84 0.4 0.0 0.2 0.2 North‐ Western Federal  Okrug 84 84 88 86 ‐ 0.5 0.5 0.6 0.2 Southern Federal  Okrug 77 81 82 80 0.2 0.1 0.6 0.3 North‐ Caucas Federal  Okrug 59 65 75 66 1.2 1.5 0.8 1.2 Volga Federal  Okrug 83 83 83 83 ‐ 0.3 ‐0.2 1.1 0.2 Ural  Federal  Okrug 83 81 82 82 ‐ 1.0 0.3 1.4 0.2 Siberia Federal  Okrug 81 80 80 80 ‐ 0.9 0.2 1.3 0.2 Far East Federal  Okrug 83 80 82 82 ‐ 0.9 0.2 1.6 0.3 10 Leaders 80 82 83 82 ‐ 0.3 0.5 0.4 0.2 10 Laggards 82 81 82 82 ‐ 0.5 ‐0.2 1.8 0.4 Rich 83 83 84 83 ‐ 0.6 0.1 1.2 0.2 Poor 80 81 81 81 ‐ 0.3 0.1 0.9 0.2 20 Annex 4. Model 3. We are mostly interested in estimating the effect of demographic changes especially the working age ratio onto the regional GDP growth. To include demographic variables as factors that determine the long-run level of GDP per worker, we considered the following identity. 2 ∙ ∙ , Y – Real regional GDP (in 2003 rubles), N – Regional population, L – Labor force (number of employed and unemployed workers), W – Working age population. 4. According to equation (2), GDP per capita is the product of GDP per worker times the participation rate times the working age ratio . We are interested in the effect of last term onto the GDP per capita growth to study the demographic dividend. If we denote equation (2) in logarithm with the following notations, we get: 3 , ln , ln , ln , ln . 5. Equation (3) should be relevant to each period t: 4 . 6. If we subtract the previous period (t-1) from equation (4), we get the relation for growth rates, considering, that ln ln ≡ . 5 . 7. Combining (5) with (1) and considering that , we get: 6 . As in Aiyar and Mody (2011), we relax restrictions on the coefficients of demographic variables because in its current form, equation (6) is restrictive. It assumes, for example, that growth rate of participation rate and working-age ratio should be directly translated into growth rate of regional GDP with unit elasticity, which may not be the case. So no restrictions are imposed on the coefficients of demographic variables. 21 Poverty & Equity Global Practice Working Papers (Since July 2014) The Poverty & Equity Global Practice Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. This series is co‐published with the World Bank Policy Research Working Papers (DECOS). It is part of a larger effort by the World Bank to provide open access to its research and contribute to development policy discussions around the world. For the latest paper, visit our GP’s intranet at http://POVERTY. 1 Estimating poverty in the absence of consumption data: the case of Liberia Dabalen, A. L., Graham, E., Himelein, K., Mungai, R., September 2014 2 Female labor participation in the Arab world: some evidence from panel data in Morocco Barry, A. G., Guennouni, J., Verme, P., September 2014 3 Should income inequality be reduced and who should benefit? redistributive preferences in Europe and Central Asia Cojocaru, A., Diagne, M. F., November 2014 4 Rent imputation for welfare measurement: a review of methodologies and empirical findings Balcazar Salazar, C. F., Ceriani, L., Olivieri, S., Ranzani, M., November 2014 5 Can agricultural households farm their way out of poverty? Oseni, G., McGee, K., Dabalen, A., November 2014 6 Durable goods and poverty measurement Amendola, N., Vecchi, G., November 2014 7 Inequality stagnation in Latin America in the aftermath of the global financial crisis Cord, L., Barriga Cabanillas, O., Lucchetti, L., Rodriguez‐Castelan, C., Sousa, L. D., Valderrama, D. December 2014 8 Born with a silver spoon: inequality in educational achievement across the world Balcazar Salazar, C. F., Narayan, A., Tiwari, S., January 2015 Updated on December 2015 by POV GP KL Team | 1 9 Long‐run effects of democracy on income inequality: evidence from repeated cross‐sections Balcazar Salazar,C. F., January 2015 10 Living on the edge: vulnerability to poverty and public transfers in Mexico Ortiz‐Juarez, E., Rodriguez‐Castelan, C., De La Fuente, A., January 2015 11 Moldova: a story of upward economic mobility Davalos, M. E., Meyer, M., January 2015 12 Broken gears: the value added of higher education on teachers' academic achievement Balcazar Salazar, C. F., Nopo, H., January 2015 13 Can we measure resilience? a proposed method and evidence from countries in the Sahel Alfani, F., Dabalen, A. L., Fisker, P., Molini, V., January 2015 14 Vulnerability to malnutrition in the West African Sahel Alfani, F., Dabalen, A. L., Fisker, P., Molini, V., January 2015 15 Economic mobility in Europe and Central Asia: exploring patterns and uncovering puzzles Cancho, C., Davalos, M. E., Demarchi, G., Meyer, M., Sanchez Paramo, C., January 2015 16 Managing risk with insurance and savings: experimental evidence for male and female farm managers in the Sahel Delavallade, C., Dizon, F., Hill, R., Petraud, J. P., el., January 2015 17 Gone with the storm: rainfall shocks and household well‐being in Guatemala Baez, J. E., Lucchetti, L., Genoni, M. E., Salazar, M., January 2015 18 Handling the weather: insurance, savings, and credit in West Africa De Nicola, F., February 2015 19 The distributional impact of fiscal policy in South Africa Inchauste Comboni, M. G., Lustig, N., Maboshe, M., Purfield, C., Woolard, I., March 2015 20 Interviewer effects in subjective survey questions: evidence from Timor‐Leste Himelein, K., March 2015 21 No condition is permanent: middle class in Nigeria in the last decade Corral Rodas, P. A., Molini, V., Oseni, G. O., March 2015 22 An evaluation of the 2014 subsidy reforms in Morocco and a simulation of further reforms Verme, P., El Massnaoui, K., March 2015 Updated on December 2015 by POV GP KL Team | 2 23 The quest for subsidy reforms in Libya Araar, A., Choueiri, N., Verme, P., March 2015 24 The (non‐) effect of violence on education: evidence from the "war on drugs" in Mexico Márquez‐Padilla, F., Pérez‐Arce, F., Rodriguez Castelan, C., April 2015 25 “Missing girls” in the south Caucasus countries: trends, possible causes, and policy options Das Gupta, M., April 2015 26 Measuring inequality from top to bottom Diaz Bazan, T. V., April 2015 27 Are we confusing poverty with preferences? Van Den Boom, B., Halsema, A., Molini, V., April 2015 28 Socioeconomic impact of the crisis in north Mali on displaced people (Available in French) Etang Ndip, A., Hoogeveen, J. G., Lendorfer, J., June 2015 29 Data deprivation: another deprivation to end Serajuddin, U., Uematsu, H., Wieser, C., Yoshida, N., Dabalen, A., April 2015 30 The local socioeconomic effects of gold mining: evidence from Ghana Chuhan-Pole, P., Dabalen, A., Kotsadam, A., Sanoh, A., Tolonen, A.K., April 2015 31 Inequality of outcomes and inequality of opportunity in Tanzania Belghith, N. B. H., Zeufack, A. G., May 2015 32 How unfair is the inequality of wage earnings in Russia? estimates from panel data Tiwari, S., Lara Ibarra, G., Narayan, A., June 2015 33 Fertility transition in Turkey—who is most at risk of deciding against child arrival? Greulich, A., Dasre, A., Inan, C., June 2015 34 The socioeconomic impacts of energy reform in Tunisia: a simulation approach Cuesta Leiva, J. A., El Lahga, A., Lara Ibarra, G., June 2015 35 Energy subsidies reform in Jordan: welfare implications of different scenarios Atamanov, A., Jellema, J. R., Serajuddin, U., June 2015 36 How costly are labor gender gaps? estimates for the Balkans and Turkey Cuberes, D., Teignier, M., June 2015 37 Subjective well‐being across the lifespan in Europe and Central Asia Bauer, J. M., Munoz Boudet, A. M., Levin, V., Nie, P., Sousa‐Poza, A., July 2015 Updated on December 2015 by POV GP KL Team | 3 38 Lower bounds on inequality of opportunity and measurement error Balcazar Salazar, C. F., July 2015 39 A decade of declining earnings inequality in the Russian Federation Posadas, J., Calvo, P. A., Lopez‐Calva, L.‐F., August 2015 40 Gender gap in pay in the Russian Federation: twenty years later, still a concern Atencio, A., Posadas, J., August 2015 41 Job opportunities along the rural‐urban gradation and female labor force participation in India Chatterjee, U., Rama, M. G., Murgai, R., September 2015 42 Multidimensional poverty in Ethiopia: changes in overlapping deprivations Yigezu, B., Ambel, A. A., Mehta, P. A., September 2015 43 Are public libraries improving quality of education? when the provision of public goods is not enough Rodriguez Lesmes, P. A., Valderrama Gonzalez, D., Trujillo, J. D., September 2015 44 Understanding poverty reduction in Sri Lanka: evidence from 2002 to 2012/13 Inchauste Comboni, M. G., Ceriani, L., Olivieri, S. D., October 2015 45 A global count of the extreme poor in 2012: data issues, methodology and initial results Ferreira, F.H.G., Chen, S., Dabalen, A. L., Dikhanov, Y. M., Hamadeh, N., Jolliffe, D. M., Narayan, A., Prydz, E. B., Revenga, A. L., Sangraula, P., Serajuddin, U., Yoshida, N., October 2015 46 Exploring the sources of downward bias in measuring inequality of opportunity Lara Ibarra, G., Martinez Cruz, A. L., October 2015 47 Women’s police stations and domestic violence: evidence from Brazil Perova, E., Reynolds, S., November 2015 48 From demographic dividend to demographic burden? regional trends of population aging in Russia Matytsin, M., Moorty, L. M., Richter, K., November 2015 49 Hub‐periphery development pattern and inclusive growth: case study of Guangdong province Luo, X., Zhu, N., December 2015 50 Unpacking the MPI: a decomposition approach of changes in multidimensional poverty headcounts Rodriguez Castelan, C., Trujillo, J. D., Pérez Pérez, J. E., Valderrama, D., December 2015 51 The poverty effects of market concentration Rodriguez Castelan, C., December 2015 52 Can a small social pension promote labor force participation? evidence from the Colombia Mayor program Pfutze, T., Rodriguez Castelan, C., December 2015 Updated on December 2015 by POV GP KL Team | 4 53 Why so gloomy? perceptions of economic mobility in Europe and Central Asia Davalos, M. E., Cancho, C. A., Sanchez, C., December 2015 54 Tenure security premium in informal housing markets: a spatial hedonic analysis Nakamura, S., December 2015 For the latest and sortable directory, available on the Poverty & Equity GP intranet site. http://POVERTY WWW.WORLDBANK.ORG/POVERTY Updated on December 2015 by POV GP KL Team | 5