INTRODUCTION

Retirement planning is a challenging issue for people around the world. In developed countries, demographic changes and the looming insolvency of pension systems have been widespread, and these countries are moving forward to reform their pension systems away from defined benefit and toward defined contribution pensions. This shifts to individuals the risks of retirement income volatility and sustainability, as well as responsibility for managing retirement portfolios. In less-developed and emerging market countries, the trend toward defined contribution pension systems is also becoming more apparent as well. But the trend is less important, as emerging market pension systems cover only a relatively small fraction of the population, and often the amount of pension promises is small and inadequate. Annuity markets are also of limited availability in many of these emerging market countries. These issues underscore the increasing importance of individual responsibility and reliance for retirement planning.

Much retirement planning research at the individual level is about the United States. Importantly, a main focus of this literature is exploring methods and guidelines to help retirees manage their retirement wealth by choosing a sustainable withdrawal rate (for example, see Bengen;1, 2 Guyton and Klinger;3 Spitzer et al;4 Tezel5). These studies provide various decision rules about safe withdrawals that retirees can use to make withdrawal decisions. An example of these guidelines is the popular 4 per cent withdrawal rule, which suggests that retirees can safely sustain their retirement withdrawals without outliving their wealth, if they initially withdraw 4 per cent of their savings and adjust this amount for inflation in subsequent years. However, the recommendations from these studies do not capture the whole retirement planning process. This becomes apparent in a situation where individuals unknowingly over-consume during their working lives and end up with small accumulated savings at retirement. In such a case, they will be forced to dramatically reduce their living standard after retirement. Saving too much or not withdrawing enough are two possibilities that could also lead to an undesired reduction in living standards. The retirement planning process and guidelines are more important and meaningful if they involve smoothing lifetime consumption rather than retirement consumption.

Ibbotson et al6 create savings guidelines for individuals who want to sustain their pre-retirement lifestyle to smooth their lifetime consumption into their retirement. They use a Monte Carlo simulation approach to project how much savings are necessary for individuals to maintain their pre-retirement lifestyle with 80 per cent of pre-retirement net income. Ervin et al7 also use a Monte Carlo simulation approach to examine the savings rates required for individuals to sustain their pre-retirement income levels between 80 and 100 per cent replacement rates. They assume a specified income level for the individual at retirement (US$100 000 in their study) and investigate how much savings rates need to be adjusted when varying the saving and retirement periods, varying investment portfolio weights and excluding Social Security income. These studies are based on past US data, which cover a fortuitous period for any country in world history with high stock returns and low volatility. Such a record is not likely to be obtainable in other countries or even in the future United States (see, for example, Bogle;8 Dimson et al;9 Krugman;10 Pfau11).

It is possible that retirement planning in pursuit of lifetime consumption smoothing can be accomplished with safe withdrawal rules by deriving a wealth accumulation target for retirement (obtained by dividing the desired income replacement rate by the withdrawal rate that a retiree feels comfortable using), and then calculating the savings rate needed during the working years to obtain the wealth accumulation target. For example, a person who follows the 4 per cent safe withdrawal rule and desires an 80 per cent replacement of their final pre-retirement income for their retirement expenditures will need 20 times their final income as accumulated wealth at retirement if no other pensions or income sources are available. Knowing this target, they can plan their savings during their working period to reach it. However, such an approach that considers the accumulation phase and decumulation phase separately misses out on incorporating mean reversion.12 A worst-case accumulation period will rarely be followed by a worst-case retirement period. This means saving more than was necessary in many cases. Instead, both the accumulation and decumulation phases should be combined into one integrated whole, providing guidelines in terms of safe savings rates for individuals to follow, which are sufficient to finance their planned retirement expenditures.

Moreover, this research on providing savings guidelines is very important to induce more awareness and knowledge for people to prepare for their retirement planning, as surveys conducted across OECD (Organisation for Economic Co-operation and Development) countries and worldwide consistently show disturbingly low and insufficient knowledge of individuals to prepare savings for their retirement.13 Such trends could also be expected for individuals in emerging market countries. Our objective is to consider the retirement planning process by integrating the working and retirement phases and to provide savings guidelines for individuals in 25 emerging market countries.

DATA AND METHODOLOGY

This study uses data from a variety of sources available through the end of 2009. First, returns on domestic stocks for the 25 countries are obtained from the MSCI Stock Indices. They are calculated as the annual percentage change at year-end for the MSCI Standard Core Gross Indices. We use domestic currency deposit rates, taken from the International Monetary Fund's International Financial Statistics (IFS), to represent the local fixed-income returns. Some exceptions are that we use the central bank discount rate for India and Jordan in 1988–1989 and the call money rate for Pakistan. In addition, for Poland, there was a change in the methodology for reporting deposit rates in 2002, and we make adjustments so that rates remain consistent after this. Inflation rates are also taken from the IFS. We exclude the period of hyperinflation occurring in Argentina and Brazil, which were extraordinary and would severely affect the realism of our simulation results. The analysis will be conducted in real terms. Even though we would also like to consider short-term and long-term government debt instruments, such data are not available for many of the emerging countries.

Our approach starts from the point where an individual chooses a fixed proportion of their income to save each year for retirement. These savings are invested in a portfolio with domestic stocks and fixed income. Given the accumulated savings at retirement, the individual starts to take inflation-adjusted withdrawals that are fixed in real terms each year from the savings portfolio for a certain number of retirement years. A particular savings rate is considered successful if it provides enough wealth at retirement to sustain the desired withdrawals for the retirement duration without exhausting wealth.

In order to determine the savings rates, we need to define initial parameters during the individual's working and retirement life, including initial wealth, career length, income path, replacement rate, retirement duration and portfolio asset allocation. We consider a hypothetical individual who starts with an arbitrary value of salary. The salary grows at 1 per cent in real terms every year for a 30-year career length. The individual contributes to the retirement savings portfolio at the end of each year with a constant proportion of their salary. Estimating an appropriate savings rate is the objective. Our analysis does not incorporate other pension or income sources, which would reduce the necessary replacement rate from savings, as well as the necessary savings rate accordingly. Even though we do not consider social security benefits, our results will still provide a useful guide for those who are covered by it, as necessary withdrawal amounts from savings can be reduced by the amount of other income sources.

After saving for 30 years, the individual begins retirement and takes withdrawals from their accumulated wealth. We consider fixed withdrawals amounting to 60, 80 and 100 per cent replacement rates from pre-retirement net income, which is defined as gross income less savings for retirement. An 80 per cent replacement rate is considered because there is widespread agreement that the retirement income needed to maintain a person's living standard is less after retirement (for example, see Munnell14) because of the elimination of some taxes and costs such as mortgage payments and children's schooling. We consider a 60 per cent replacement rate for some individuals who might receive social security benefits or income from other sources, and a 100 per cent replacement rate for those wishing to maintain the same expenditures. As mentioned by Ibbotson et al,6 using net income is a realistic approach and helps retirees avoid dramatic changes in lifestyle. Basing the retirement income target on gross income could confuse an individual into saving more unnecessarily. We assume a 20-year retirement duration in which fixed withdrawal amounts (in real purchasing power) are made at the beginning of each year as an amount equal to the defined replacement rate of pre-retirement net income.

We consider the case where individuals self-manage their savings and retirement portfolio. They invest in domestic assets and remain responsible for all risks related to investment returns and longevity. This self-management of the retirement portfolio, in particular, should reflect the current and future trend in emerging market economies where pension reforms from defined benefit toward defined contribution pension plans have been extensive, although the annuity markets are still underdeveloped or non-existent.

With the data, we use a bootstrapping approach, in which the annual in-sample returns are randomly selected with replacement by year to form a variety of hypothetical asset paths over the life cycle. We simulate 10 000 hypothetical return sequence paths for each country. For each return sequence in each country, we optimize across the two domestic assets, finding the fixed asset allocation that allows for the lowest savings rate to meet the retirement spending goals. In so doing, we assume that individuals have perfect foresight to choose the fixed asset allocation that would achieve the lowest savings rates for sustaining 20 years of withdrawals in retirement. We will relax this assumption as well by showing how the savings rates change for various asset allocations. We consider 21 possibilities for fixed asset allocations, ranging in 5 percentage point increments from 0 to 100 per cent stocks, with the remainder of assets in fixed income. We assume that the investment portfolio will be rebalanced without considering tax implications and transaction costs at the end of the year to maintain the targeted asset allocation.

RESULTS

Table 1 provides summary statistics for 25 emerging market countries. The statistics are provided for the available period in real terms after removing the effect of inflation. Stocks provide double-digit average returns for most countries, except China, Israel, Jordan, Morocco and Poland. However, their volatility, as measured by standard deviation, is also very high. The standard deviation of stock returns for most countries is more than twice as high as the mean, and even more than three times as high for some countries. This demonstrates the high risks prevailing in the stock markets of emerging market countries. On the other hand, fixed-income assets, in general, provide relatively low average returns in these countries, compared with stocks. Fixed-income assets provide less than 5 per cent average real returns, except in Brazil, and even negative returns for some countries. However, the lower returns for most countries are also accompanied by less volatility and risk, compared with stocks. Inflation rates are also high with an average of more than 6 per cent for many countries. Brazil, Columbia, Hungary, Indonesia, Mexico, Russia, Sri Lanka and Turkey experienced an average double-digit inflation rate. Table 1 also provides the correlations between stocks and fixed-income assets. As seen, the correlation coefficients are small for most countries and negative for some countries, implying potential diversification benefits.

Table 1 Summary statistics
Full size table

Table 2 shows the minimum necessary savings rates calculated from the integrated approach, which jointly considers both the working and retirement phases, and the minimum necessary savings rates calculated from the isolated approach, which are the savings rates needed to achieve the retirement wealth target that was calculated to provide the necessary wealth for the estimated safe withdrawal rate at the same percentile. The table is based on a 30-year work period, 20-year retirement period, 1 per cent annual real wage growth, a 30/70 per cent asset allocation to stocks and fixed-income assets during the entire 50-year period, and an 80 per cent replacement of pre-retirement net income. For the 90, 95 and 99 per cent success rates (in other words, 90th, 95th and 99th percentiles of the simulations), the minimum necessary savings rates obtained from the isolated approach are higher than those obtained from the integrated approach. The difference between the savings rates from both approaches varies across countries, ranging from 2 percentage points to more than 10 percentage points. This difference is larger for higher percentiles as well. The reason for these differences is because the isolated approach is doubly pessimistic. First, a safe withdrawal rate is chosen with a particular success rate. Separately, the savings rate needed to provide enough wealth to sustain this withdrawal is calculated for the same percentile. This is too pessimistic because the chances of experiencing a bad luck scenario both before and after retirement are not related in this way. In fact, for US data, Pfau12 shows the opposite to be the case because of mean reversion in stock returns and market valuations. The integrated approach here provides a lower savings rate for a particular level of confidence because it allows for independence between the pre- and post-retirement periods. This is why it is more appropriate to consider the accumulation and retirement phases together.

Table 2 Minimum necessary savings rates under integrated and isolated approaches
Full size table

Table 3 provides results for the minimum necessary savings rates over 30 years of work and 20 years of retirement required to attain 60, 80 and 100 per cent replacement rates of pre-retirement net income with 90, 95 and 99 per cent success rates. This table incorporates the perfect foresight assumption, using the asset allocation for each simulation that provides the lowest savings rate. Figure 1 will show more about the results for varying asset allocations. By looking at the minimum necessary savings rates for the 80 per cent replacement rate, the results show that individuals in most of the countries need high savings rates to smooth their lifetime consumption. If we allow for a 90 per cent success rate, only individuals in 3 out of 25 countries (Chile, Brazil and Peru) experience sustainable savings rates of less than 10 per cent. Individuals in 17 countries would need more than 20 per cent savings rates, and 7 countries (Czech Republic, Pakistan, Jordan, Sri Lanka, China and particularly Russia) need 30 per cent or higher. The impossibly high savings rates needed in Russia result from the very high volatility of stock returns and the extraordinarily high inflation rate. If we focus on improved chances for success, higher savings rates are necessary. For example, with a 99 per cent chance for success, individuals in 13 countries need more than a 30 per cent savings rate. The table also considers a 60 per cent replacement rate and a 100 per cent replacement rate. Naturally, savings rates and replacement rates are positively related, and this additional information provides more perspective about their relationship. The results show that necessary savings rates increase by between 20 and 25 per cent of the current savings rates for many countries when the desired replacement rate (of pre-retirement net income) increases from 60 to 80 per cent, and between 15 and 20 per cent when the desired replacement rate increases from 80 to 100 per cent.

Table 3 Minimum necessary savings rates with perfect foresight assumption (30 working and 20 retirement years)
Full size table
Figure 1
figure 1

Optimal stock allocation for savings rates at various success rates.Note: Assumptions include perfect foresight, 80 per cent replacement of pre-retirement net income, a 30-year saving duration, 20-year retirement duration, 1 per cent annual real wage growth, no administrative fees and annual rebalancing.

Full size image

Table 4 shows the minimum necessary savings rates to attain an 80 per cent replacement of pre-retirement net income for varying career lengths and success rates. Longer savings periods produce a remarkable effect in lowering safe savings rates. At a 90 per cent success rate, individuals who save for only 20 years will require in some countries a more than doubling of the 30-year savings rate. Individuals in only two countries, Brazil and Chile, can afford an 80 per cent replacement rate with a less than 20 per cent savings rate, whereas 18 countries need 30 per cent or higher savings rates. On the other hand, with 40 years of saving, individuals in 16 countries may save less than 20 per cent. Among them, seven countries need lower than 10 per cent savings rates. As the replacement rate is calculated as a percentage of pre-retirement net income, lower necessary savings rates also mean higher absolute retirement income. Looking at higher probabilities for success, the results are more terrifying for the individuals who do not start to save early enough. Overall, given the high savings rates needed for most emerging markets because of the high volatility of saving instruments and inflation, the results suggest that individuals should start saving as early as possible.

Table 4 Minimum necessary savings rates to attain 80 per cent of pre-retirement net income
Full size table

Figure 1 shows the optimal asset allocations to achieve the minimum necessary savings rates shown in Table 3 for various success rates. Interestingly, the results from Figure 1 reveal that for most countries, the optimums occur with a low proportion of stocks. This is in contrast to many previous studies about the United States, a country that is characterized by quite favorable market conditions and a well-developed financial system. Between the 90 per cent and 99 per cent success rates, the optimums occur at points with a less than 40 per cent stock allocation for most countries except Chile, Peru and Mexico at the 90, 95 and 99 per cent success rates, Egypt and Czech Republic at the 90 and 95 per cent success rates, and Colombia, South Africa and Hungary at the 90 per cent success rate.

Finally, Figure 2 shows the distribution of the minimum necessary savings rates with a 95 per cent chance of success across stock allocations for each country. The smallest savings rate is labeled with the country's name code. The figure shows that the optimal stock allocation is low in many countries, and that savings rates often slope upward above about 30 per cent stocks. Allocating a high proportion to stocks tends to increase necessary savings rates in these emerging markets.

Figure 2
figure 2

Savings rate by various stock allocation at 95 per cent success rate.Note: Assumptions include 80 per cent replacement of pre-retirement net income, a 30-year saving duration, 20-year retirement duration, 1 per cent annual real wage growth, no administrative fees and annual rebalancing.

Full size image

CONCLUSION

Although much retirement planning literature focuses on finding ways for retirees to safely withdraw their accumulated wealth so that they smooth their consumption during retirement, these studies capture only a part of retirement planning. The full retirement planning process involves both the accumulation and decumulation phases. It is possible that from the concept of safe withdrawal rates, individuals can work to achieve a wealth target. However, such an approach is doubly pessimistic, and to capture a necessary savings rate with a particular chance for success we recommend connecting the entire life cycle and determining the savings rate required to finance retirement expenditures without regard for the actual implied withdrawal rate. Savings rates calculated in this manner are lower and more reflective of the intended risk level.

Even though the savings rates necessary to attain a desired replacement rate vary across the 25 emerging market countries, our results show that the situation in most of these countries would require high savings rates to safely provide for planned retirement expenditures. The savings rates, of course, are higher for individuals who desire higher success probabilities or higher income replacement rates. Individuals in emerging market countries should start saving early, as longer savings periods can dramatically reduce the required savings rate, given the high volatility of saving instruments and the inflationary environment in these countries. With regard to asset allocation for the saving and retirement portfolios, our study indicates that stocks do not play a large role in the optimal allocation for most emerging market countries.