Economic Consequences of Universal Old Age Social Pensions

In my last article I calculated the long term costs of various means-tested old age social security support programs proposed by Mr. Leung Chung-Ying, Mr. Henry Tang, and Professor Chow Wing-Sun. I compared their proposals against the present arrangement that combines two programs: the Comprehensive Social Security Old Age Assistance Scheme (CSSA) and the Social Security Old Age Allowance (SSA). The three proposed programs are means tested rather than universal social pension schemes. In this article I provide estimates of what it would cost in the long term to have a universal social pension scheme that gave each individual over the age of 65 a fixed monthly payment.

 

In making these calculations I follow the same approach as I did in last week’s article and make the following assumptions:

 

First, labor productivity will continue to rise at an annual constant rate in the future. Specifically, I assume that real GDP per working population will rise by 4% per year.

 

Second, I base my calculations on two sets of population forecasts. The Hong Kong Census and Statistics Department (C&SD) forecasts future population up to 2039. The United Nations (UN) forecasts future population up to 2100. Both sets of forecasts project the elderly population to peak in the mid-21st century. The UN forecasts a lower rate of population growth than the C&SD.

 

Third, I assume that the value of old age monthly social security payments will grow at the rate of GDP per capita, which implies they will rise with the average rate of improvement in the standard of living.

 

Using the first two assumptions I calculate the implied real GDP and real GDP per capita growth rates for the two forecasted periods. The results are given in Table 1. I then use all three assumptions to estimate the percentage shares of old age social security payments in GDP over time under the various actual and proposed schemes.

 

Table 1: Population, Working Population, and Real GDP (Annual Percentage Growth)

	2011-2039	2010-2100
Total population	0.8	0.4
Working population	0.1	0.05
Real GDP per working population	4.0	4.0
Real GDP	4.2	4.0
Real GDP per capita	3.3	3.6

 

Alarmingly Low Population Growth Rate

 

In Figure 1 and 2, I plot these percentage shares over time using the projected population figures from the C&SD and UN, respectively. From Figure 1 we observe that the old age social welfare expenditures under each of these schemes rise rapidly from 2011 until 2040. The existing CSSA and SSA schemes are the least expensive. Mr. Leung’s proposal is more expensive, Mr. Tang’s scheme is even more expensive, and Professor Chow’s scheme, which drops the “Bad Son Statement” requirement, is the most expensive. Between 2011 and 2039 the expenditure required for each of these schemes as a percentage of real GDP will approximately double.

 

 

Figure 2, using UN projections, shows the percentage shares begin to level off after 2040 and start to decrease a little after 2060. Between 2010 and 2040 the expenditure of all these schemes will increase approximately 1.5 times as a percentage of real GDP due to the growing proportion of elderly people in the population as a whole. Indeed higher old age social security expenditure will persist until 2100 because of the very low rate of population growth. Between 2010 and 2100 the population is projected to grow by only 0.4% per year. The working population growth rate is even worse at 0.05% per year. Our population growth is reaching alarmingly low levels.

 

 

I next calculate expenditure based on a number of hypothetical universal non-means tested pension schemes. These are schemes where any person over the age of 65 will be paid a fixed monthly payment until death. I assume that the fixed monthly payment will increase at the rate of growth of real GDP per capita so that improvements in the standard of living are maintained. I assume five different schemes where the monthly payments are $1350, $2400, $3600, $4800, and $6000. A monthly payment of $6000 would be approximately equal to the monthly income of a person remunerated at $35 per hour and working 40 hours per week.

 

Table 2 outlines the estimated old age social security expenditure  as a percentage of real GDP for the various schemes. There are four means tested schemes and five non-means tested universal schemes. The estimates show that for the year 2039 or 2040, depending on the two population projections, even the cheapest option will be much more costly than today. The CSSA plus original SSA Scheme will consume an estimated 1.41% to 1.72% of real GDP, compared with 0.67% of real GDP at present.

 

Table 2: Comparison of Old Age Social Security Expenditures as a Percentage of Real GDP under Different Schemes

	Means Tested Schemes				Non-Means Tested Universal Schemes
	CSSA + Original SSA	CSSA + Leung’s SS	CSSA + Tang’s SSA	Chow’s SSA  + Drop Bad Son Statement	$1,350	$2,400	$3,600	$4,800	$6,000
Hong Kong Census and Statistics Department  Population Forecasts
2011	0.67	0.87	1.05	1.54	1.06	1.64	2.30	3.06	3.83
2039	1.41	1.83	2.20	3.19	2.19	3.39	4.76	6.34	7.93
United Nations Population Forecasts
2010	0.66	0.86	1.04	1.52	1.04	1.61	2.26	3.02	3.77
2040	1.72	2.23	2.68	3.90	2.68	4.15	5.82	7.76	9.70
2070	1.75	2.25	2.70	3.88	2.67	4.13	5.79	7.72	9.65
2100	1.63	2.10	2.53	3.64	2.51	3.87	5.43	7.24	9.05

Note: The CSSA figures include only the expenditures for recipients aged 65 or above

 

The other, more costly means tested schemes proposed by Mr. Leung, Mr. Tang and Professor Chow will consume, respectively, 1.83% to 2.23%, 2.20% to 2.68%, and 3.19% to 3.90% of real GDP by 2039 or 2040. But the non-means tested universal pension schemes are far more expensive. Ordered in terms of progressive cost, by 2039 or 2040 they will consume 2.19% to 2.68% of real GDP for a $1350 payout,, 3.39% to 4.15% for a $2400 payout, 4.76% to 5.82% for a $3600 payout, 6.34% to 7.76% for a $4800 payout, and 7.93% to 9.70% for a $6000 payout. Even the most inexpensive of these non-means tested schemes will be more expensive then the schemes proposed by Mr. Leung and Mr. Tang.

 

100% Increase in Tax Revenues

 

It is worth noting that universal schemes are particularly costly not only because everyone is eligible, but also becauseHong Kong’s population is rapidly ageing and the growth in the total population is slowing. This means more recipients will need to be supported by fewer contributors to the schemes.

 

There are two other ways to compare the various schemes. The first is to determine how much tax revenues would need to increase to finance the larger old age social security expenditures. At presentHong Kongsalaries and profits tax revenues are about 9% of GDP. Using the Census and Statistics Department’s population forecasts, we know that expenditure on the existing CSSA and original SSA scheme would have to increase from 0.67% to 1.41% of real GDP from 2011 to 2039. Therefore it would be necessary to raise salaries and profits tax revenues by about 8% (= [1.14%-0.67%] ÷ 9%) if the schemes were to be financed wholly from these sources.

 

Table 3 below shows the percentage increases in salaries and profits tax revenues needed to finance the other old age social security schemes. The estimates are somewhat higher using the United Nations population projections because its projected growth rate is lower than that of the Hong Kong Census and Statistics Department. Of particular interest is Mr. Leung’s proposal. The implied increase in tax revenues would be the lowest of all schemes described here, amounting to about 13% to 17% more than today.

 

Table 3:The Impact of Old Age Social Security Expenditures to Tax Revenue and Fiscal Reserve Under Different Schemes

	Means Tested Schemes				Universal Non-Means Tested Schemes
	CSSA + Original SSA	CSSA + Leung’s SSA	CSSA + Tang’s SSA	Chow’s SSA  + Drop Bad Son Statement	$1,350	$2,400	$3,600	$4,800	$6,000
Hong Kong Census and Statistics Department  Population Forecasts
Increase in Tax Revenues in 2039	8%	13%	17%	28%	17%	30%	45%	63%	81%
United Nations Population Forecasts
Increase in Tax Revenues in 2040	12%	17%	22%	36%	22%	39%	57%	79%	100%
Years it would take to deplete $600 Billion in Fiscal Reserves
Number of years	26	22	19	14	19	14	10	8	7
Year when fully depleted	2037	2033	2030	2025	2030	2025	2021	2019	2018

 

The universal non-means tested schemes are very expensive. Paying everyone above the age of 65 only $1350 per month would require an increase in tax revenues of 17% to 22%. Paying them $3600 per month would require an increase in tax revenues of 45% to 57%. And if we paid out $6000 per month, or the equivalent of $35 per hour for a 40-hour work week (which is at the maximum of the range of recently proposed minimum wage rates), then tax revenues would have to increase by 81% to 100%.

 

Modest Payout Under a Means Tested Scheme

 

The second method for comparing the schemes is to see how long they would take to deplete the accumulated fiscal reserves of $600 billion. Since the reserves are invested and earn a return, we shall assume that the long term rate of return in real terms (or inflation adjusted terms) will be equal to the calculated growth rate of real GDP per capita. From Table 1 we know that the growth rate is 3.3% for 2011-39 and 3.6% for 2010-2100. Using these numbers in the calculations we show in Table 3 the number of years it takes to deplete Hong Kong’s entire stock of fiscal reserves and the year when the current reserves is expected to be fully depleted.

 

The existing CSSA and original SSA schemes would take 26 years to deplete the reserves, Mr. Leung’s scheme would take 22 years, Mr. Tang’s scheme 19 years, and Professor Chow’s scheme 14 years. The non-means tested universal schemes would deplete the fiscal reserves at an even faster rate, depending on the amount paid out: within 19 years if we paid $1350 per month, 10 years at $3600 per month, and 7 years at $6000 per month. Once the fiscal reserves were depleted, raising tax revenues to finance old age social security payments would become an unavoidable decision. It is worth noting that all the schemes considered here would cause the fiscal reserves to be depleted in less than one generation. The fiscal reserves would be exhausted before mid-century when the share of old age social security payments in real GDP will be reaching its peak level.

 

A non-means tested universal old age social security scheme would be an economic disaster for Hong Kong. A modest payout under the means tested scheme to help the poor in their old age, who are in poverty through no fault of their own, can be the only sensible policy choice for Hong Kong. In my article last week I also suggested that if an enhanced means tested old age social security scheme were to be implemented in the near future then it would be desirable to permanently freeze the current non-means tested old age Social Security Allowance at its current level of $1350 per month. This would allow tax revenues to be better targeted at the elderly poor. In the longer run, non-means tested expenditures will then gradually become a declining proportion of total old age social security payments.