The Treasury

Global Navigation

Personal tools

You are here: Home > Publications > Media Statements, Speeches and Guest Lectures > Guest Lectures by Visiting Academics > Professor Erwin Diewert - On the Role of Government


Guest Lecture: Professor Erwin DiewertOn the Role of Government

Page updated 20 Sep 2007

Lecture notes, abstract and outline from Professor Erwin Diewert's Guest Lecture presented at the Treasury on 27 July 2004.

Professor Erwin Diewert

University of British Colombia


This paper was stimulated by a report by Winton Bates (2001), who examined the possible roles of government for the New Zealand economy. In particular, Bates examined the role of government in facilitating growth as well as in providing services. He then finished his review by asking how big should the optimal government be?

Following Bates, the discussion of the optimal size of government could be organized around 5 topics:

  • The recent historical record on the size of governments.
  • The determinants of economic growth.
  • The role of government in facilitating growth.
  • The role of government (in addition to facilitating growth).
  • Given a consensus on the possible roles of government, how best can the government achieve society's goals; i.e., how can the role of government be optimized?


        In sections 2 and 3, we discuss the various determinants of growth. It turns out that one of the most important determinants of economic growth is the growth of total factor productivity ; i.e., the growth of outputs that cannot be readily explained by the growth in inputs. We devote section 3 below to this somewhat esoteric topic. Section 4 provides a summary of the various factors that determine productivity growth. In the light of the growth theories discussed in sections 2-4, section 5 then examines the role of government in facilitating growth.

        In section 6, we turn our attention to the role of government as a provider of services . We follow Bates (2001) and divide government services into two categories: core and noncore services . Noncore government services are services that could be provided by the private sector but instead are provided by the government. An interesting question is: why does the government provide these services instead of the private sector? In section 7, we consider two alternative answers to this question: (i) market failures and (ii) risk reduction .

        Finally, in sections 8 and 9, we consider methods by which the government might be able to deliver noncore government services in a less costly fashion. The main new idea that we suggest in this section is for the government to provide each New Zealander at birth cash endowments that would be invested in the stock market and grow into funds that would be used to provide pensions, funding for higher education and to provide medical services.

        The level 1 core functions of a government

        These functions include:

        • Rule of law and security of property rights (internal security including the courts and the police).
        • Defence (external security).
        • Production of national statistical information.
        • Foreign relations.
        • Immigration policy.
        • Product and workplace safety.
        • Maintaining macroeconomic stability (monetary policy).
        The level 2 core functions of a government

        These functions include:

        • Provision of elementary and secondary education.
        • Infrastructure spending.
        • Support of scientific research.
        • Environmental protection.
        • Regulation of natural monopolies.
        • Finally, the noncore functions of government might include:
        • Provision of higher education.
        • Provision of health services or health insurance.
        • Provision of pensions.
        • Provision of income support to the poor.
        • Provision of unemployment insurance .

        The main puzzle about the noncore functions of government is: why do governments provide so many noncore services when there are substantial costs of doing so?

        The first justification for the government to provide the 5 types of noncore services listed above is that some form of market failure is involved in the private provision of these services.

        However, rather than using market failures to justify the government provision of the above 5 types of services, one could appeal to an entirely different type of argument that is based on the work of the philosopher John Rawls. Rawls (1971) suggests the following thought experiment: imagine that we do not know who we are but we want to consider what kind of a society that we would like to be born into. Thus we are situated behind what Rawls calls a ‘veil of ignorance'; we do not know our class, skills, age, gender, sexuality, or religious views. Under these conditions, what roles would we like the government to take?

        Given that most people appear to be risk averse, it could be argued that most taxpayers in the Rawlsian ‘original position’ behind the veil of ignorance would opt for a relatively large dose of noncore government programs in order to insure themselves against risks of being born into a body with poor skills or into a family with extremely low income. Thus the world is a big bad risky place and Rawlsian taxpayers would probably want the government to provide subsidized higher education, some level of health insurance, some minimal level of income support for the poor, pensions for the retired and perhaps a minimal unemployment insurance scheme .

        Thus both market failure and Rawlsian social insurance arguments provide strong justifications for governments to engage in the provision of noncore services to the less well off segments of society.

        However, if these services are provided to the middle and upper classes, we run into the “leaky bucket” problem flagged by Bates (2000; 33-35): the churning of taxes extracted and benefits paid to the middle class wastes a large amount of resources . An apparent solution to this churning problem is to restrict the benefits of noncore government services to the needy but this solution runs into the incentives problem .

        Bates describes this problem as follows:

        “It seems likely that the social safety net that could be provided within a government spending ceiling of 14-15 percent of GDP would provide generous help to people in need, provided fairly stringent eligibility tests could be applied. Stringent income tests (with high abatement rates) can have disincentive effects similar to those of high marginal tax rates and can lead to poverty traps. However, low abatement rates raise the budgetary cost of programmes and are not likely to reduce the number of beneficiaries or the length of time they receive benefits. The problem of poverty traps should be addressed through improved programme design to avoid perverse incentives associated with eligibility conditions and forms of assistance, and by requiring beneficiaries to meet appropriate obligations to help themselves.” Winton Bates (2001; 59-60).

        High effective marginal tax rates

        Thus attempting to restrict the benefits of government services to the poor runs into the problem of high marginal effective tax rates as benefits are withdrawn. This will tend to encourage some percentage of the poor to remain poor in order to avoid these punitive tax rates. However, the above quotation by Bates points toward a solution to this problem: namely, attempt to design programs that will somehow avoid these high rates of taxation as benefits are withdrawn .

        Provision of pensions.

        Our key suggestion here is for the government to make use of the power of compound interest .

        Suppose that as each New Zealander was born, the government gave the child an endowment of $10,000 , which would be invested in the stock market (under rules to be determined). At age 65 or age 70, this investment would be converted into pension income.

        Table 1: Pension Fund Values at Ages 65 and 70: $10,000 initial endowment ( in thousands of dollars)

        Annual rate of return (%)

        Retire at


















        Compound Returns

        Thus if the investments could achieve a real return of 7% per annum, then by age 65, the initial $10,000 endowment would have grown to $813,000. This should provide a satisfactory pension for most people. If the initial investment happens to make a 9% annual rate of return and retirement is postponed until age 70, then the $10,000 endowment would have grown to $4,167,000. This would provide a very satisfactory pension!

        What is the evidence on real rates of return in the stock market? Fortunately, a comprehensive study of investment returns on the stock markets of 12 countries for the past 100 years has just appeared: namely Dimson, Marsh and Staunton (2000). In Table 2 below, we list the annual (geometric) average real rate of return that the stock market of each of these 12 countries has achieved over the 50 years from 1950 to 2000. We also list the annual (geometric) average that bonds earned in each country over the 50 years as well as the annual average consumer price inflation rate.

        Table 2: Geometric Annual Average Percentage Real Rates of Return for Stocks and Bonds and Annual Consumer Price Inflation Rates: 1950-2000

        CountryStock Market RateBond RateInflation Rate
        United States9.01.64.0
        United Kingdom8.91.66.2
        Arithmetic Average8.22.34.8

        Source: Dimson, Marsh and Staunton (2000)


        The stock market returns in Table 2 include both capital gains and dividends while the bond returns include both yields and capital movements. Stock market indexes are weighted by their market capitalization. Thus from Table 2, it can be seen that the 12 countries had an average stock market annual rate of return in the neighborhood of 8.2 %. In particular, the very important U.S. stock market had an average real rate of return around 9 % per year over the past half century. Also from Table 2, it can be seen that the 12 countries had an average bond annual rate of return in the neighborhood of 2.3 %.

        Thus on average, stocks returned about 6% more per annum than bonds!

        Costs of the Endowment Plan

        What about the cost aspects of the above endowment fund suggestion? If per capita income is $20,000 and population growth is 2% per year, then the provision of the $10,000 endowment will cost only 1% of national income, which is very affordable.

        Advantages of an Endowment

        What is the advantage of the above endowment idea over traditional pay as you go government pension schemes? After all, the same idea could be implemented in a pay as you go scheme. However, if a government attempted to set aside 1% of GNP each year in a pension fund that will begin to pay out money in 65 years, that government will end up with an enormous pension fund asset and the demands to spend those accumulated funds on urgent current needs will be overwhelming.

        What are some of the problems with the plan?

        Here is a partial list:

        • There will be transition problems; how do we integrate the existing pension plan with the new one?
        • What should be done with immigrants? Should they get an endowment of $10,000 (or even more) as well?
        • What should be done with emigrants? Should they be allowed to take their accumulated pension funds with them when they leave the country?
        • Under the above plan, cohorts born at different times will earn different rates of return and hence at retirement will receive different pensions. Is this “fair”?
        • Will governments be willing to introduce a plan where the benefits will become evident only after 65 years?
        • Should the plan be restricted to the domestic market or should it be global in scope? If it is global, there could be repercussions on the exchange rate as capital flowed out in the initial stages of the plan and then flowed in during the later stages.
        • What are the precise institutional arrangements for the fund? What is the class of investments that would be allowed? Should only approved mutual funds be permitted investments? Should some fraction be held in market type portfolios? Should individuals have any say on how their endowments should be invested?
        Provision of higher education

        Again make use of the power of compound interest.

        As each New Zealander is born, $10,000 would be set aside as a higher education endowment for that baby. At ages 18 to say 25, this endowment could be accessed by the individual to pay educational fees

        Table 3: Higher Education Fund:Values at Ages 18 and 25 ($10,000 initial endowment) in thousands of dollars

        Annual rate of return (%)

        Commence studies at


















        Thus if the educational fund earned an average annual rate of return of 7% per year, at age 18, the student's educational endowment would have grown to $34,000. If the student were to postpone studies until age 25 and the fund earned 9%, then the endowment would have grown to $86,000.

        Is an Education Fund Affordable?

        If every baby born decided to go to university or college, then the cost to the economy (assuming again a per capita income of $20,000 and a population growth of 2%) would again be 1% of GDP , which again seems affordable.

        Of course, not every child would choose to enroll in a university or college in which case, the endowment could perhaps be used to provide other forms of training. Any unused funds would revert to the state. If the higher educational participation rate were 50%, then the cost of the plan to the nation would only be 0.5% of GDP.

        Provision of health services or health insurance

        I suggest that:

        1. the government could provide some sort of a minimal health plan coverage for every resident and
        2. beyond this minimal level, each individual would be obligated to purchase health insurance or pay directly for health services.

        However, again the endowment idea could be used to good advantage. At birth, each New Zealander could be given a $10,000 health endowment. As usual, this money would be allowed to accumulate in a health account and each individual could use his or her health account to purchase additional health insurance or to pay directly for noninsured services. Any unused balances in the health account would revert to the state at death.

        To get an idea of how large the annual expenditures out of such a health account could be, let us assume that the individual decided to spend nothing on extra health insurance until age 25. At that point, the initial $10,000 health endowment would have cumulated to $34,000 if the average rate of return was 5%, $43,000 if the average rate of return was 6%,…, $86,000 if the average rate of return was 9% (see the last line in Table 2 above).

        Now assume the individual draws down the health account by $2,000, $3,000 or $4,000 per year until he or she reaches age 75. Table 4 below lists how much money would be in the health account under alternative assumptions on the average annual rate of return.

        Table 4: Residual Balance of Health Fund after Annual Health Expenditures from age 25 to 75 (in thousands of dollars) and Assuming an Initial $10,000 Endowment at Birth

        Annual rate of return (%)

        Annual spending
























        If the rate of return were only 5% per annum and annual health expenditures were $2,000 on average from age 25 to 75, then the health account of the average individual would end up at the deficit level of $30,000 at age 75.

        However, if the rate of return increased to 7%, then at the same annual levels of health expenditures, the average health account would show a surplus of $789,000 at age 75!

        Note that if per capita income in the economy were $20,000, health expenditures were $2000 per year per person of age 25 to 75 and 0 for ages 1 to 25 and life expectancy was 75 years, then with an even population distribution, these health care expenditures would be about 6.7 % of GDP. However, if the rate of return increased to 7% per annum and annual health expenditures were $4,000 on average from age 25 to 75, then the health account of the average individual would end up at the deficit level of $27,000 at age 75.

        Note that if per capita income in the economy were $20,000, health expenditures were $4000 per year per person of age 25 to 75 and 0 for ages 1 to 25 and life expectancy was 75 years, then with an even population distribution, these health care expenditures would be about 13.3 % of GDP.

        Again, this plan illustrates the power of compound interest: if the annual rate of return is 5%, a one percent endowment investment in health would finance annual health expenditures in the neighborhood of 7% of GDP. If the annual rate of return is 7%, then a one percent endowment investment in health would finance annual health expenditures in the neighbourhood of 13% of GDP.

        Macroeconomic Effects

        What would be the likely macroeconomic effects of our targeted endowment funds plans? It would almost surely increase the savings rate in the economy (the endowments would crowd out some private savings but overall savings should rise) thus tending to lower the overall rate of return in the world economy.

        In addition, there would be a tendency to decrease the debt-equity ratio in the economy as additional funds pour into the stock market. There would no doubt be other effects of our suggested plan but I do not see any adverse effects that would be substantial enough to outweigh the very large benefits that are implicit in Tables 1, 3 and 4 above.

      Page top