The Buffer Stock Employment Model in a Small Open Economy
William F. Mitchell
Centre of Full Employment and Equity Research
Department of Economics
University of Newcastle
Australia
1. Introduction
High and persistent unemployment has pervaded almost every OECD country since the mid-1970s. Mass unemployment arises because the budget deficit is too small relative to the desires of the private sector to meet its tax obligations, to save and to hold money for transactions purposes. Unemployment is thus a macroeconomic phenomenon and can never be a "real wage" problem The solution to this problem is for government to use deficit spending to introduce a Buffer Stock Employment (BSE) policy (Mitchell 1996, 1998a).
The BSE approach to full employment is counter the current policy direction of governments in the OECD economies. The rising unemployment began with the rapid inflation of the mid-1970s. The inflation left an indelible impression on policy-makers who became captives of the resurgent new labor economics and its macroeconomic counterpart, monetarism. The goal of low inflation led to excessively restrictive fiscal and monetary policy stances by OECD governments driven by what we might call "backward" thinking (Mitchell 1996, 1998a). This has led to GDP growth in OECD countries being generally below that necessary to absorb the growth in the labor force in combination with rising labor productivity.
Backward reasoning reflects a fundamental misunderstanding of the way fiat currency operates. It begins with the fallacious analogy that government spending, taxation, and debt issue is equivalent to the spending and financing decisions of the household. Accordingly, governments are supposed to seek financing prior to spending. The analogy has led orthodox economists to advocate balanced budgets to avoid higher tax rates and interest rates. The reemerging free market ideology has convinced us wrongly that government involvement in the economy imposes costs on us, and we have thus supported governments that have significantly reduced their fiscal involvement in economic activity.
The economies that avoided the plunge into high unemployment maintained a "sector of the economy which effectively functions as an employer of the last resort, which absorbs the shocks which occur from time to time..." (Ormerod, 1994, p.203). The BSE policy fulfills this absorption function. While the BSE approach will cure unemployment it also delivers price stability.
In this paper, the inflation control mechanisms of the BSE model are contrasted with those in an economy subject to a NAIRU. The BSE concept of the Non-Accelerating Inflation Buffer Employment Share (NAIBER) first developed in Mitchell (1998a) is further explained.
There are two broad ways in which government can maintain price stability. First, it can adopt the NAIRU approach by suppressing the budget deficit and generating unemployment. Second, it can conduct a BSE policy whereby the public sector absorbs all the current idle workers into paid employment at a base level wage that it sets and maintains. The relevant price stability concept is the NAIBER. The change in the buffer employment ratio (BER) disciplines the wage-price pressures in the private sector by asserting the buffer stock wage as the numeraire. No rational government, which understood how its own currency works and the role of the budget deficit, would choose the NAIRU approach. The costs from lost output and social alienation are enormous.
The paper also considers the financial implications of the BSE model in the context of a small open economy. In addition to the normal arguments that monetarists and others use to justify their case against fiscal activism (crowding out, inefficient resource usage), it is often argued that increased globalisation imposes further restrictions on the ability of governments to pursue independent fiscal and monetary policy. In Australia’s case, it is alleged that budget deficits only result in growing current account deficits and rising debt levels. Reacting to this, it is alleged that external funds managers can enforce higher interest rates and thus even lower growth and higher unemployment in the domestic economy.
There are several testable hypotheses included in the monetarist case, which are rarely confronted with empirical scrutiny.
Is there evidence of a relationship between budget deficits and short-term and long-term interest rates? If there is no discernable statistical relationship found it is difficult to argue against fiscal activism based on financial crowding out arguments.
Is there evidence of a relationship between long-term interest rates across countries in globalised financial markets? If there is no relationship detected then the view that financial traders in the large markets like Japan and the United States can render domestic monetary policy ineffective is problematic.
Is there any evidence that the relationship between domestic long-term and short-term interest rates is unstable? Stability implies that the cash rate, which is set as a policy instrument, and the longer-term interest rates, which are influenced by market considerations, move together in a proportional manner over the long-run and that therefore the determinant is the officially controlled cash rate.
Is there any evidence to support the twin-deficits hypothesis that imposes causality from the fiscal deficit changes to changes in the current account deficit? A lack of such a direct relationship also provides further support for the use of budget deficits under the BSE policy.
It is found that none of the principal claims used against fiscal activism are empirically sustainable. The evidence is supportive of the conceptual basis of monetary theory that underpins the BSE model (see Mitchell, 1996; Mosler, 1997).
The BSE model is thus be justified on two separate grounds: First, it is appealing from social welfare considerations; and, second, it is the only rational strategy for a government that supplies a fiat currency and wishes to maximize macro benefits and retain price stability. A third justification exists in terms of environmental sustainability but that is addressed elsewhere (see Mitchell, 1999a).
The paper is set out as follows. Section 2 discusses the BSE in relation to the problem of price stability and explains how the NAIBER is preferable as an inflation control mechanism to the NAIRU. Section 3 discusses the role of the budget deficit and explains why we should not be concerned about its size. Section 4 conducts a range of econometric tests, which establish that the monetarist case against budget deficits is not empirically founded. Accordingly it is argued that the budget deficit implications of introducing the BSE policy should be disregarded. Concluding remarks are provided.
2. The Buffer Stock Employment Model
Under the BSE policy, the government continuously absorbs workers displaced from the private sector. The "buffer stock" employees would be paid the minimum wage, which defines a wage floor for the economy. Government employment and spending automatically increases (decreases), as jobs are lost (gained) in the private sector. A full discussion of the cost of such schemes is to be found in Mitchell (1998a).
The BSE model allows currently idle workers to contribute in many socially useful activities including urban renewal projects and other environmental and construction schemes (reforestation, sand dune stabilization, river valley erosion control, and the like), personal assistance to pensioners, and other community schemes. While the existence of the buffer stock would reinforce the automatic stabilization built into the fiscal system, it remains a fluctuating work force. The design of the jobs and functions would have to reflect this. Projects or functions requiring critical mass might face difficulties as the private sector expanded, and it would not be sensible to use only buffer stock employees in functions considered essential. Private sector employers who tender for government sponsored projects could also provide buffer stock employment.
Inflation and the NAIRU
Would the BSE policy violate the NAIRU constraint and generate inflation? The expectations-augmented Phillips curve became the centerpiece of the resurgence of orthodox thinking in the late 1960s as rising inflation rates challenged the credibility of the demand-oriented Keynesian macroeconomics. The conventional notion of an inflation-unemployment trade-off that had driven the conduct of fiscal and monetary policy since the end of World War II was abandoned in favor of a natural rate of unemployment (NRU) that was considered to be insensitive to aggregate policy. The quest for balanced budgets and deregulation replaced fiscal activism. The NAIRU became the target for governments obsessed with inflation.
The OECD experience of the 1990s shows that persistently high unemployment has eventually delivered low inflation -- the Phillips curve is alive and well (Mitchell 1996). Orthodox theory, in denying the existence of involuntary unemployment, had to adopt an esoteric explanation for the observed Phillips curve behavior. Adherents of the natural-rate hypothesis claimed that when a government stimulus pushes the inflation rate up, workers confuse the rise in nominal wages with a rise in real wages and increase their supply. The rise in labor supply lasts until the workers learn the truth and withdraw their labor, and ultimately the economy settles at the competitive equilibrium position -- the NAIRU. Attempting to maintain unemployment below the NAIRU results in accelerating inflation. The most damning piece of evidence against these supply-side explanations of unemployment is that quits are procyclical -- exactly the opposite hypothesized in the natural-rate story.
Drawing from the competing claims literature, a NAIRU relationship can be established without the orthodox theory (Mitchell 1987). Accordingly, inflation results from incompatible claims on available real income, and unemployment acts to discipline the aspirations of labor so that they are compatible with the profit expectations of capital (Kalecki 1971). The depressed product market demand also reduces the ability of firms to pass on prices. The temporary price stability defines what Mitchell (1987) termed a macroequilibrium unemployment rate (MRU). Adding hysteresis, where the MRU is functionally related to the actual unemployment, defines a long-term trade-off between inflation and unemployment (Hargreaves Heap 1980; Mitchell 1987).
If the government pays market prices for everything in a NAIRU world, then it is forced to use unemployment to maintain price stability. How would the introduction of the BSE policy change this? Suppose we characterize an economy with two labor markets: A (primary) and B (secondary) broadly corresponding to the dual labor market depictions. Wage setting in A is contractual and responds in an inverse and lagged fashion to relative wage growth (A/B) and to the wait unemployment level (displaced Sector A workers who think they will be re-employed soon in Sector A).
A government stimulus to this economy increases output and employment in both sectors immediately. Wages are relatively flexible upwards in Sector B and respond immediately. The compression of the A/B relativity stimulates wage growth in Sector A after a time. Wait unemployment falls due to the rising employment in A but also rises due to the increased probability of getting a job in A. The net effect is unclear. The total unemployment rate falls after participation effects are absorbed.
The wage growth in both sectors may force firms to increase prices, although this will be attenuated somewhat by rising productivity as utilization increases. A combination of wage-wage, and wage-price mechanisms in a soft product market can then drive inflation. This is a Phillips curve world. To stop inflation, the government has to repress demand. The higher unemployment brings the real income expectations of workers and firms into line with the available real income and the inflation stabilizes - a typical NAIRU story.
Inflation and the BSE
Introducing the BSE policy into the depressed economy effectively makes Sector B the BSE sector because its wage levels are fixed by the government in accordance with its desire to set the value for its fiat money. This sets a floor in the economy's cost structure for given productivity levels. The dynamics of the economy change significantly. The elimination of all but wait unemployment in Sector A and frictional unemployment does not distort the relative wage structure so that the wage-wage pressures that were prominent previously are now reduced.
But the rising demand may soften the product market because it becomes easier to pass on higher margins and accede to wage demands. The rising demand also stimulates employment in Sector A. There are no new problems faced by employers who wish to hire labor to meet the higher sales levels. They must pay the going rate, which is still preferable to appropriately skilled workers, than the BSE wage level. The rising demand per se does not necessarily invoke inflationary because firms may increase capacity utilization to meet the higher sales volumes.
How do Sector A workers react? Wendell Gordon (1997, 833) said, "If there is a job guarantee program, the employees can simply quit an obnoxious employer with assurance that they can find alternative employment." With the BSE policy, wage bargaining is freed from the general threat of unemployment. However, it is unclear whether this freedom will lead to higher wage demands than otherwise.
In professional occupational markets, it is likely that some wait unemployment will remain. Skilled workers who had previously been laid off are likely to have received payouts that forestall their need to get immediate work. They have a disincentive to immediately take a BSE job, which is a low-wage and possibly stigmatized option. Wait unemployment disciplines wage demands in Sector A but the demand pressures may eventually exhaust this stock and wage-price pressures may develop.
Further, buffer stock employees are more attractive than when they were unemployed, not the least because they will have basic work skills, like punctuality, intact. This reduces the hiring costs for firms in tight labor markets who previously would have lowered their hiring standards and provided on-the-job training and vestibule training. They can thus pay higher wages to attract workers or accept the lower costs that would ease the wage-price pressures. The BSE policy thus reduces the "hysteretic inertia" embodied in the long-term unemployed and allows for a smoother private sector expansion because growth bottlenecks are reduced.
Exchange rate changes may induce cost pressures. With flexible exchange rates, the demand stimulus may increase the price of foreign exchange, which under usual conditions increases the competitiveness of the economy while also adding to the domestic price level. Vickrey (1996) said, "The danger of world speculative gyrations under freely floating conditions would be greatly diminished under a well-established full-employment policy, especially if combined with a third dimension of direct control over the overall domestic price level." The direct control to allow the depreciation to be insulated from the wage-price system could be an incomes policy.
Inflation Control -- the NAIBER
The BSE wage provides a floor that prevents serious deflation from occurring and defines the private sector wage structure. However, if the private labor market is tight, the non-buffer stock wage will rise relative to the BSE wage and the buffer stock pool drains. The smaller this pool, the less influence the BSE wage has on wage patterning. Unless the government stifles demand, the economy will then enter an inflationary episode, depending on the behavior of labor and capital in the bargaining environment.
In the face of wage-price pressures, the BSE approach maintains inflation control in much the same way as monetarism -- by choking aggregate demand using contractionary fiscal and/or monetary policy and inducing slack in the non-buffer stock sector. The slack does not reveal itself as unemployment, and in that sense the BSE may be referred to as a "loose" full employment.
The BSE policy generates inflation stability because the suppression of non-buffer sector output asserts the numeraire price -- the BSE wage. This leads to the definition of a new concept, the Non-Accelerating Inflation Buffer Employment Ratio (NAIBER), which, in the buffer stock economy, replaces the NAIRU/MRU as an inflation control mechanism. The Buffer Employment Ratio (BER) is the ratio of buffer stock employment to total employment.
As the BER rises, due to an increase in interest rates and/or a fiscal tightening, resources are transferred from the inflating non-buffer stock sector into the buffer stock sector at a price set by the government; this price provides the inflation discipline. The disciplinary role of the NAIRU, which forces the inflation adjustment onto the unemployed, is replaced by the compositional shift in sectoral employment, with the major costs of unemployment being avoided. That is a major advantage of the BSE approach.
The NAIBER also relates the private sector employment rate to the inflation rate. In the switch from a NAIRU regime to a NAIBER model, the initial BER would be an understatement of the medium-term, steady state BER. As the government absorbed the unemployment into buffer stock employment the demand levels that were delivered via the NAIRU would rise and disturb the distributional peace. The government would then have to exert pressure on the private sector and be prepared to absorb increased numbers into the buffer stock pool.
However, relying on the NAIBER may introduce other costs. For example, the rising BER may lower overall productivity growth, as resources are transferred out of the higher productivity, non-buffer sector. While this will not have direct implications for competitiveness in the export sector, it is possible that productivity growth in the non-buffer stock sector itself may also fall as scale declines (Kaldor 1978).
The BSE economy thus has some new policy choices to make. Minimizing the BER improves productivity growth but leaves the economy open to inflation. By maximizing the BER, it controls inflation, but may reduce productivity growth overall.
The alternative is to separate the BER from the inflation control via an income policy (Mitchell and Watts 1997). With the BSE economy, the government sets a wage floor and thus the price that it is willing to pay to transfer resources from the non-buffer stock sector to the buffer stock sector. An income policy using this numeraire as the basis for wage adjustment would allow the economy to achieve both full employment and price stability with a lower BER. The design of such a policy is not considered here.
3. The BSE and the Budget Deficit
The critics of the BSE approach point to financial constraints they allege would arise from the higher budget deficits. The willingness of government to allow the budget deficit to increase and decrease as is necessary to maintain full employment is essential to the viability of the BSE policy. In this section, it is argued that the rising budget deficits that are likely to accompany the introduction of the BSE policy are not a cost and should be ignored. In a BSE world the size of the budget deficit necessary to maintain the policy is irrelevant. One of the most damaging analogies in economics is the supposed equivalence between the household budget and the government budget. This immediately leads to what we might call "backward" reasoning. For example, Barro (1993, p.367) says "we can think of the government's saving and dissaving just as we thought of households' saving and dissaving."
The analogy is flawed at the most fundamental level. The household must work out the financing before it can spend. Whatever sources are available, the household cannot spend first. Moreover, by definition a household must spend to survive. The government is totally the opposite. It spends first and does not have to worry about financing. The important difference is that the government spending is desired by the private sector because it brings with it the resources (fiat money) which the private sector requires to fulfill its legal taxation obligations. The household cannot impose any such obligations. The government has to spend to provide the money to the private sector to pay its taxes, to allow the private sector to save, and to maintain transaction balances. Taxation is the method by which the government transfers real resources from the private to the public sector. A budget deficit is necessary if people want to save.
The logic according to those who draw the household analogy follows like this. Debt would have to be issued to finance the deficit. Accordingly, bond sales finance government, which will accumulate as debt. Like a household, the rising debt cannot be sustained indefinitely and so spending must be curbed and brought in line with the financial reality. In the meantime, the demands that the debt places on available savings pushes interest rates up and crowds out "more efficient" sources of private spending.
The backward logicians divide into two camps. The orthodox monetarists who eschew government debt and advocate balanced budgets. Their wrong-minded logic has imposed extremely high macroeconomic costs in terms of lost growth and high unemployment on the western economies since the mid-1970s. The other camp is the group, which includes some Post Keynesians, who while comfortable with using deficit spending to increase economic activity, couch their recommendations in conservative logic bounded by appropriate movements in the debt to GDP ratio. As long as the ratio is stable there is no problem.
For example, The Bank of International Settlements (1994) analysed the level of deficit that would be sustainable in terms of a stable government debt to GDP ratio (hereafter the debt ratio). They concluded that a deficit is sustainable as long as the debt ratio does not increase permanently. A framework for analysing the relation between deficits and the debt ratio is provided by Bispham (1988) and Glyn (1977). Glyn (1997, p.226), an advocate of expansionary fiscal policy to reduce unemployment, points out that this literally means the higher is the debt ratio the higher sustainable deficit as long as the real interest rate is below the GDP growth rate. He also argues that "financial markets, the ultimate arbiters of such matters, may look simply at the size of the deficit." The BIS (1995, p.88) concur that "it is difficult to persuade markets that low inflation is sustainable in the presence of large budget deficits." Glyn (1997, p.227) concludes that "Given the experience of the past twenty years it would be difficult to convince that increased deficits at the beginning of the expansionary programme would be rapidly scaled down as the private sector took up the main thrust of expansion. There seems little alternative to financing through taxation most of an expansionary programme." Further, Glyn (1997, p.224) says "it is misleading to treat them (interest rates) as entirely exogenous. It is likely that beyond a certain level, a higher deficit will lead financial markets to exact a higher real-interest rate."
The two camps however fail to understand the relationship between fiat currency, public debt and taxation in a monetary capitalist economy, a topic, which is examined by Mitchell (1998b) and Mosler (1997). They show the priority of spending and argue that debt issue is not essential for governments to spend beyond tax revenue. Mosler (1997) shows that bond issues are essential only to support the cash rates set by the Central Bank. Deficit spending without Treasury bond sales would generate excess reserves in the banking system, so that government debt helps to maintain a positive overnight interest rate for private banks. The idea of crowding out in this environment is as meaningless as debates about the term maturity of the debt. Deficits add to the net disposable income of households in the economy and the income provides markets for private production. An endogenous credit economy then serves to provide the deposits necessary to make payments, which facilitate production. The higher demand stimulates investment that creates capacity as a legacy to the future. The higher is current demand, the higher is productive capacity in the future. Spending brings forth its own savings. Savings are not required to exist as a prior pool for spending to occur.
William Vickrey (1996, p.10) argued that "the 'deficit' is not an economic sin but an economic necessity. Its most important function is to be the means whereby purchasing power not spent on consumption, nor recycled into income by the private creation of net capital, is recycled into purchasing power by government borrowing and spending. Purchasing power not so recycled becomes non-purchase, non-sales, non-production, and unemployment." In an endogenous money world, there can be no crowding out unless the monetary authority stops lending.
The recent Asian financial troubles and IMF intervention have once again given credence to the view that increasing levels of debt will eventually lead to lenders refusing to take up further public borrowing. Usually this is cast in terms of countries with low levels of capital that have major private debt denominated in a foreign currency which is used to finance imports. Crises occur when the export revenue, which services the debt, falls for one reason or another. But none of these countries would have any trouble issuing debt in its own currency.
The point is clear. When fiat money is used, government spending increases reserves in the banking system. Taxation and borrowing drain the reserves. This gives the clue to the function of borrowing. A deficit generates a net build up in reserves in the banking system. The spending occurs and the private firms and individuals that sell goods and services to the government deposit the proceeds in the commercial banks, which build up reserves. Unless those reserves are drained from the system, they will earn a the official discount rate. The role of the government bond issues is to give these returns a way to earn a return in excess of the discount rate.
To fine-tune this point further, the spending would still have occurred if there were no bond issues. The excess reserves would be held somewhere in the banking system earning zero return. If the Treasury offers too few or too many bonds relative to the holders of reserve balances at the Central Bank, the Central Banks "offsets" those operations to balance the system. In any case, the 'money' is in one account or another at the Central Bank. We then ask the question - why should government care if the holders of the excess balances chose the one that doesn't pay interest as opposed to the ones that do (buying bonds)? The answer is simple - they would be indifferent.
4. Interest Rates, Budget Deficits and Current Account Performance
The BSE policy requires that the government have the ability to implement a largely independent monetary and fiscal policy. In this section we examine the effects of budget deficits on interest rates and current account performance and also seek to establish causality within the term structure of interest rates.
As noted above Glyn (1997, pp.226-227) an advocate of fiscal activism believes that taxation should be used to "finance" the necessary spending. His contention is based on his acceptance of the notion that international financial markets will react to higher budget deficits and "exact a higher real-interest rate." (p. 224)
This contention forms a set of empirically testable hypotheses outlined in the introduction. We examine each in turn in this section. The following ideas are examined:
Is there evidence of a relationship between budget deficits and short-term and long-term interest rates? If there is no discernable statistical relationship found it is difficult to argue against fiscal activism based on financial crowding out arguments.
Is there evidence of a relationship between long-term interest rates across countries in globalised financial markets? If there is no relationship detected then the view that financial traders in the large markets like Japan and the United States can render domestic monetary policy ineffective is problematic.
Is there any evidence that the relationship between domestic long-term and short-term interest rates is unstable? Stability implies that the cash rate, which is set as a policy instrument, and the longer-term interest rates, which are influenced by market considerations, move together in a proportional manner over the long-run.
Is there any evidence to support the twin-deficits hypothesis that imposes causality from the fiscal deficit changes to changes in the current account deficit? A lack of such a direct relationship also provides further support for the use of budget deficits under the BSE policy.
Crowding Out
The crowding out notions of monetarism which was anticipated by Keynes (1937) are well known. Whenever there is an exogenous planned rise in demand there is a concomitant rise in demand for money to meet the extra contractual commitments. If the banking system does not meet the demand for credit the rate of interest will rise before any additional output is sold. Accordingly, a budget deficit which "draws" on scarce savings via debt issue will push interest rates up in the domestic markets.
However, if the extra wage bill that the Government requires for BSE workers are paid out of deposit-balances held by the Treasury at the Central Bank then there are no short-term interest rate effects. The demand for credit is not independent of the level of real activity, irrespective of whether this is a demand for loans from private entrepreneurs from the commercial banks, or whether it is a demand by the Treasury for balances at the Central Bank. Either route to increased money are consistent with an overdraft system which Keynes certainly approved of (Keynes, 1937). Thus, in a credit money economy we should expect to find no relationship between changes interest rates and the changes in the budget deficit.
Table 1 presents evidence from Australian financial markets which strongly rejects the null hypothesis that there is a statistical causation flowing from changes in the budget deficit to changes in long- and short-term interest rates. Granger causality tests were conducted which formulate the problem in the following way:
x is a Granger cause of y (denoted as
), if present y can be predicted with better accuracy by using past values of x rather than by not doing so, other information being identical. (Granger, 1969).
In other words, in a general Autoregressive-Distributed lag model, the rejection of Granger causality amounts to the acceptance of the restriction that all the coefficients of the distributed lag (starting at lag one) are zero. The testing model regressed the change in the interest rate measure on lagged changes of the interest rate measure and lagged changes of the budget deficit to GDP ratio.
For every lag tested, no relationship between changes in the Deficit/GDP ratio and the changes in the real interest rates could be detected. The evolution of real interest rates appears to be independent of the changes in the relative size of the deficit.
Long-term interest rate convergence
While the measurement of real long-term interest rates is open to question there appeared to be a convergence among real long term interest rates in the major economies during the mid-1990s despite what appeared to be different domestic situations in each country. Real long-term interest rates in Europe and Japan seemed to rise in response to American monetary conditions. Christiansen and Pigott (1997, p.5) argue that if "long-term interest rates were responding more to external factors than domestic economic conditions and, if so, might be less free, even under floating exchange rates, to vary independently across the major regions than earlier believed."
The issue bears on the ability of a sovereign government to implement policy, which is likely to be suspected by global financial markets. Christiansen and Pigott (1997) investigate the extent to which external factors constrain the freedom of long-term interest rates to vary with domestic fundamentals and the extent to which globalisation has reduced the ability of monetary authorities to influence long-term interest rates.
With flexible exchange rates we expect that domestic long-term real interest rates will reflect domestic economic conditions. The budget deficit is usually included because it is seen as a major factor determining domestic saving. However, this reflects a view of saving as a finite pool, which can be made available either to finance the budget deficit or to finance private spending. We have argued that spending creates its own saving and within the BSE framework no such influence is expected. In the previous section, we failed to find evidence to support a relationship between the budget deficit and long-term interest rates.
What factors might lead to international conditions dominating domestic influences on a country’s long-term interest rates? First, when portfolio diversification is possible, risk premia may be determined by conditions in world markets. The argument is that large financial traders can impose their view on a nation’s interest rates. If, for example, it is thought that inflation is rising a higher risk premium will be imposed. There is very little evidence in the literature to support this view. Further, the antagonism towards large budget deficits is usually in terms of higher expected inflation rates. Under the BSE model, any relation between budget deficit and expected inflation is negated by broken by the influence of the NAIBER. Second, the work by Summers (1986) on noise trading could provide the linkage. Expectations rather than economic fundamentals drive speculation. Traders who are unable to determine exact equilibrium information will use price information derived from large bond markets (such as the US bond market) to guide their trading behaviour. However, it is hard to argue that these effects which are likely to impact on short-term rates will be influential on long-term rates.
There is thus a time dimension to the degree to which long-term interest rates may converge between countries. It is useful to distinguish between relationships in the:
very-near term (daily or weekly)
medium and longer term.
Relationships between daily US long-term rates and daily Australian long-term rates
Initially we test the relationship between the daily US long-term nominal bond rate and the daily Australian nominal long-term bond rate. Table 2 provides evidence from Granger causality tests relating the US nominal long-term interest rate to Australian nominal long-term rates. The tests for Granger-Causality were performed on daily changes of using 5, 10, and 20 lags. In other words we are assuming that the US rates affect Australian rates on the following day.
We conclude that changes in the United States long-term bond market is statistically significant as a predictor of changes in the Australian long-term bond market over extremely short periods. The bi-lateral causality is also confirmed with Australian changes helping predict US changes the following day (see Remolona, 1991).
Another significant consideration is the magnitude of the response of the rate change in Australia to a change in a large market. The causality merely indicates that the spillover effects between markets occur. To estimate the responsiveness the daily change in Australian nominal long-term interest rates was regressed on the daily change in nominal long-term interest rates a simple regression with a lagged dependent variable added was run over the full sample. The result below is the steady-state solution and the R2 refers to the coefficient of determination in the dynamic model and the Partial R2 refers to the partial effect of the addition of DUS to the dynamic model.
The conclusion is that only a minute fraction of daily changes in the Australian nominal long-term interest rates are attributable to daily changes in the United States nominal long-term interest rates.
Relationships between monthly real US long-term rates and monthly real Australian long-term rates
Table 3 provides evidence from Granger causality tests relating the monthly foreign real long-term interest rate to monthly Australian real long-term rates. The tests for Granger-Causality were performed on monthly changes of real long-term interest rates using 36, 24, 12, 6 and 1 lags. To examine the hypothesis that globalisation has changed the way long-term interest rates are determined in a domestic economy the tests were performed over the entire sample and then for the post-1984 period when Australia essentially experienced financial deregulation and opened it financial markets. If the hypothesis that large markets dominate smaller markets like Australia, even under flexible exchange rates, is valid, then we would expect the foreign rates to Granger-cause Australian rates for both samples.
The results show clearly that while there is some influence from American and Japanese rates for the entire sample (part of which spans a fixed exchange rate period), there is no causality detected in the post-1984 sample. In other words, the move to flexible exchange rates has been associated with a period where changes in monthly foreign long-term interest rates have had no influence on changes in monthly domestic rates. Similar results were found by testing for the influence of other monthly OECD real long-term rates.
These results are consistent with those found in the literature. There is little evidence to support the proposition that there has been a decrease in the sensitivity of long-term rates to domestic short-term rates with the rising incidence of globalisation (see Kasman and Rodrigues, 1991; Skinner and Zettelmeyer, 1996; Christiansen and Pigott, 1997).
Testing the Stationarity of the Yield Gap
It appears that the long-term interest rates in the large markets do not "cause" enduring movements in the long-term rates in Australia. Indeed, the evidence appears to support the idea that globalisation has led to more independence of long-term rates between Australia and the rest of the world after the move to freely determined exchange rates. The only causal relations detected appear to be due to the period of fixed or heavily managed exchange rates or very small near-term effects.
Given that there is very little evidence to support the notion that globalisation has led to Australia’s long-term interest rate being dominated by the US rates or other external long-term rates, it is important to focus on the relationship between the short-term rates and the long-term rates within Australia. There are two issues of interest. First, do long-term interest rates drive short-term interest rates or vice-versa? Second, if monetary authorities are able to set the cash rate and if the medium and long-term rates exhibit a stable relationship over the long run then the monetary authorities retain the ability to influence movements in long-term interest rates in their own financial markets and achieve their policy targets.
Table 4 provides evidence of causality between the components of the Australian term structure. The tests seek to determine if there is any evidence to support the notion that the changes in the cash rate are caused by changes in other rate. The results indicate that causality can be detected in both directions for each of the pairs tested and are thus not supportive of any primary determinacy.
Given that bi-directional causality cannot be eliminated, it is important to examine the notion that monetary authorities can set the cash rate within a stable term structure. This contention is examined by testing whether the yield gaps between the cash rate and various other rates in the term structure are stable. A series of unit root tests were conducted for this purpose. We would reject the notion that the yield gap is stable if we detected a unit root in the series.
For Australia, the results shown in Table 5 clearly support the notion that the yield gap is stationary over a range of time periods. The tests reject the null that the yield gap is a unit root process. In other words, the evidence is consistent with the statement that the difference between domestic long-term interest rates and short-term rates is stable over time. This is also consistent with the view that Australian monetary authorities are able to pursue their policy objectives and are not at the behest of global funds managers as if often alleged by antagonists of the use of activist deficit-based government policy.
Christiansen and Pigott (1997, p.14) perform similar tests for the United States, Japan, Germany, France, Italy, the United Kingdom and Canada and conclude that "the evidence does not suggest that the effects of globalisation have been so great as to prevent monetary authorities from being able to achieve their fundamental objectives…. Most of the studies cited earlier suggest that domestic short-term interest rates have a greater impact on long-term interest rates than do US interest rates, at least for European countries. … Indeed, for most countries, the gap between domestic long-term interest rates and short-term rates is stable in the long run. This indicates that long-term interest rates are ultimately linked to short-term interest rates. Through their influence on the supply of liquidity to markets, authorities remain capable of controlling the evolution of short-term interest rates over these horizons." Other studies reach similar conclusions (see Radecki and Reinhart, 1989; Kasman and Rodrigues, 1991).
To further examine the stability of the term structure, cointegration tests were performed using the Engle-Granger (1987) approach. The cointegration equations confirmed the evidence shown in Table 6. The pairwise relationships between the cash rate and the medium and long-term rates are cointegrated, which indicates that there are no systematic departures over a long period between the rates of interest.
The finding of pairwise cointegration means that the term structure defined in terms of the rates we are using in this study is also cointegrated and the following cointegrating regression establishes the long-term stability of the term structure over the period 1979(1) to 1997(2):
CASHRATE = + 1.049 - 0.1929 TB3M + 1.248 BILL90 - 0.1819 TB10Y
[0.419] [0.102] [0.108] [0.069]
R2 = 0.970139
ADF = -4.3640**
Additional evidence was adduced from a series of error correction models of the form:
The model is consistent with theories of the term structure of interest rates, which imply a long-run relationship between long and short-term rates. If the gap between the long- and short-term rates is large relative to the long-run (equilibrium) relationship, then the yield gap has to close by some amount each period. The models do not tell us exactly how the gap is closed. But the models strongly support the notion that there is a long-run relationship between the short-term and long-term rates and that the short-run dynamics in the short-term rate is influenced by deviations from the long-run relationship. In each case, the error-correction term, g, is strongly defined and of the correct sign. The speed of adjustment varies between 0.20 to 0.73. Results are available from the author.
Twin Deficits
The Twin Deficits Hypothesis (TDH) was used by monetarists to justify restrictive fiscal policy stances in the OECD economies during the 1980s and 1990s. The hypothesis is based on sectoral flow relationships which hold in an accounting sense in the national accounts. The TDH, however, imputes a strict causality where the private sector savings and investment gap is zero or stable, and changes in the budget deficit translate directly into current account deficit. Noting that in these circumstances the current account deficit represents a nation "spending more than it is earning", the budget deficits are then considered to "cause" a rising external debt. Accordingly, the risk of foreign financial market retribution via downgrading by international ratings agencies and the like is related to rising budget deficits. The cure for a chronic current account deficit then is logically to be found in increased domestic savings emanating from budget surpluses.
The problem is that the causality is not guaranteed. The evidence in Australia is that the private savings gap is not stable (see Argy, 1992). Further, the current account position at any point in time can be driven by international factors like imperfect competition, barriers to entry, economies of scale and general conditions of world trade. All these factors may constrain export revenue. A world recession may cause a trading economy with automatic stabilisers to experience a current account deficit, which then drives a rising budget deficit. Further, a rising budget deficit can increase domestic income and reduce the private savings gap.
Table 7 shows that the tests fail to support any notion of causality between changes in the Current Account deficit and changes in the Budget Deficit. Neither direction of causality was detected.
Future directions
Leading proponents of Post Keynesian economic theory develop the open economy model in terms of fixed exchange rate regimes (see Davidson, 1984). The BSE approach to full employment and price stability requires a flexible exchange rate system to allow monetary authorities the scope to pursue independent policies. Under fixed exchange rates, globalisation of financial markets lead to a convergence of both short-term and long-term interest rates across countries within the exchange rate bloc. The rates also tend to move together and are thus determined by shared conditions. Individual economies cannot run independent monetary policy. The BSE approach thus challenges not only the monetarist orthodoxy but also the Post Keynesian orthodoxy.
Post Keynesian orthodoxy is also somewhat ambivalent to the role of the budget deficit. Glyn’s (1997) idea that taxation should finance spending to avoid issuing debt is an example of the Post Keynesians who are worried about the endogeneity of the deficit. The BSE policy requires that the deficit is endogenous and not a policy target in itself. A reading of Keynes (1940) suggests that the BSE approach to full employment, while ostensibly "Keynesian" is in fact not derived from the ideas that he had on functional finance. For example, Keynes (1940, p.23) "The last row of figures leaves us with the incomes out of which the increased war expenditure has to be met either by additional taxes or by borrowing, after allowing for what can be provided out of existing capital." Keynes (1919) was also negative about the use of budget deficits to finance reconstruction.
Further work is required to resolve this conflict. The development of a coherent approach to full employment will always be hampered if economists constrain activity by making the budget deficit a target variable, either in absolute terms (the Balanced Budget School) or in relative terms (stable deficit to GDP School). Neither view is consistent with the dynamics of a government which issues fiat currency.
Conclusion
Unemployment arises because the budget deficit is too low in relation to private saving and the desire to hold money. It is always a macroeconomic problem. Australia’s persistently high unemployment rate is largely the outcome of demand deficiency brought on by successive governments who have failed to understand the implications and logic of their own monetary position. The Buffer Stock Employment model is the only logical way of providing jobs for everyone with guaranteed price stability. Whether it is accompanied by an income policy is a matter of refinement rather than substance.
Once we understand the role of public spending and why there is no financing imperative for the government then it is possible to see why there is no requirement to balance the budget position of the government.
The evidence presented in this paper suggests that the major financial objections raised to fiscal activism do not hold. Under flexible exchange rates, there appears to be no enduring constraints against a government running an independent monetary policy.
Table 1 The relationship between changes in the budget deficit and real interest rates in Australia
Interest Rate Change Lag Sample Period
Test Statistic Cash Rate 11.00 am. 4
1980 (2) to 1997 (3)
F(4,61) = 0.90068 8
1981 (2) to 1997 (3)
F(8,49) = 0.74283 12
1982 (2) to 1997 (3)
F(12,37) = 0.4068 3 Month Treasury 4
1979 (4) to 1997 (3)
F(4,63) = 1.0074 Bill Rate 8
1980 (4) to 1997 (3)
F(8,51) = 0.55517 12
1981 (4) to 1997 (3)
F(12,39) = 0.49747 90-day Bank-accepted 4
1979 (4) to 1997 (3)
F(8,63) = 1.52741 Bill rate. 8
1980 (4) to 1997 (3)
F(8,51) = 0.54597 12
1981 (4) to 1997 (3)
F(12,39) = 0.38844 Federal Government 10 4
1979(4) to 1997 (3)
F(4,63) = 0.62536 Year Bond Yield 8
1980 (4) to 1997 (3)
F(8,51) = 0.56285 12
1981 (4) to 1997 (3)
F(12,39) = 0.5763 Data: Datastream
Quarterly interest rate changes: Cash Rate 11 a.m., 3-month Treasury Bill Rate, 90-day Bank-accepted Bill Rate, and 10-year Commonwealth Government Bond Yield.
The nominal rates were converted to real rates using a smoothed moving-average of the inflation rate.
Table 2 Tests of causality among daily changes in long-term interest rates
January 1993 to March 1998
Change in Australian Rates Adding Change in USA Rates 7 F(7,1325) = 80.658 ** 10 F(10,1316) = 56.755 ** 20 F(20,1286) = 28.267 ** Change in USA Rates Adding Change in Australian Rates 7 F(7,1325) = 2.3614 * 10 F(10,1316) = 1.972 * 20 F(20,1286) = 1.3717 Data: Datastream for the period January 29 1993 to March 31 1998.
Changes in the nominal US Treasury benchmark 10-year bond and changes in the Australian 10-year bond yield (middle rate)
Table 3 Relationship between Australian real long-term interest rates and those of other countries
Country Lag
Sample Period
Test Statistic for adding the Distributed lag
FULL SAMPLE: United States 1
1969(2)-1997(7)
F(1,332) = 1.9595 6
1970(2)-1997(7)
F(6,317) = 1.2358 12
1970(8)-1997(7)
F(12,299) = 1.1161 24
1971(8)-1997(7)
F(24,263) = 1.6139 * + 36
1972(8)-1997(7)
F(36,227) = 1.5035 * + Japan 1
1971(3)-1997 (7)
F(1,314) = 7.0949 ** 6
1971(8)-1997(7)
F(6,299) = 2.5883 * 12
1974(2)-1997(7)
F(12,281) = 1.5385 24
1973(2)-1997(7)
F(24,245) = 1.4956 36
1974(2)-1997(7)
F(36,209) = 1.3458 United Kingdom 12
1970(8)-1997(7)
F(12,299) = 0.61168 24
1971(8)-1997(7)
F(24,263) = 0.55368 36
1972(8)-1997(7)
F(36,227) = 0.79009 GLOBALISATION PERIOD: United States 1
1984 (1)-1997(7)
F(1,160) = 2.4689 6
1984 (1)-1997(7)
F(6,150) = 1.1229 12
1984(1)-1997(7)
F(12,138) = 1.2414 Japan 1
1984(1)-1997(7)
F(1,160) = 3.784 6
1984(1)-1997(7)
F(6,150) = 2.0498 12
1984(1)-1997(7)
F(12,138) = 1.319
Source: OECD Main Economic Indicators, monthly long-term interest rates. Real interest rates were calculated using a smoothed annualized inflation rate.
+ Causality was also detected flowing from changes in Australian long-term interest rates to changes in the USA long-term interest rates. No such causality could be detected flowing from Australia to Japan.
* Denotes the 5 per cent level of significance.
Table 4 Causality within the Australian Term Structure
Lag
Sample Period
Test Statistic for adding the Distributed lag Cash rates on 3-month Treasury Bills 12
1982 (2) to 1997 (3)
F(12,37) = 5.394 ** 8
1981 (2) to 1997 (3)
F(8,49) = 3.2831 ** 4
1980 (2) to 1997 (3)
F(4,61) = 4.0947 ** 3-month Treasury Bills on Cash rates 12
1982 (2) to 1997 (3)
F(12,37) = 2.4243 * 8
1981 (2) to 1997 (3)
F(8,49) = 1.8358 4
1980 (2) to 1997 (3)
F(4,61) = 1.8719 Cash rates on 90-day Bank-accepted Bill Rate 12
1982 (2) to 1997 (3)
F(12,37) = 7.9528 ** 8
1981 (2) to 1997 (3)
F(8,49) = 8.1097 ** 4
1980 (2) to 1997 (3)
F(4,61) = 9.4402 ** 90-day Bank-accepted Bill Rate on Cash rates 12
1982 (2) to 1997 (3)
F(12,37) = 2.9441 ** 8
1981 (2) to 1997 (3)
F(8,49) = 2.839 * 4
1980 (2) to 1997 (3)
F(4,61) = 4.6364 ** Cash rates on 10-year Treasury Bonds 12
1982 (2) to 1997 (3)
F(12,37) = 2.0358 * 8
1981 (2) to 1997 (3)
F(8,49) = 2.2491 * 4
1980 (2) to 1997 (3)
F(4,61) = 2.1809 10-year Treasury Bonds on Cash rates 12
1982 (2) to 1997 (3)
F(12,37) = 2.0394 * 8
1981 (2) to 1997 (3)
F(8,49) = 1.438 4
1980 (2) to 1997 (3)
F(4,61) = 1.497
Data: Datastream
Quarterly interest rate changes: Cash Rate 11 a.m., 3-month Treasury Bill Rate, 90-day Bank-accepted Bill Rate, and 10-year Commonwealth Government Bond Yield.
The nominal rates were converted to real rates using a smoothed moving-average of the inflation rate.
Table 5 Testing the relationship between short-term and long-term interest rates in Australia
Yield Gap 1980(3)-1997(3)
1984(1)-1997(3)
1990(1)-1997(3)
Cash rate – 3-month Treasury Bills -6.0912**
-5.0143**
-2.6924**
Cash rate – 90-day Bills -4.8933**
-3.9441**
-2.6162 *
Cash rate – 10-year Treasury Bonds -2.8827**
-2.1521 *
-2.4720 *
- CASHRATE is the 11.00 am cash rate, TB3M is the 3-monthly Treasury Bill rate, BILL90 is the 90-day commercial bill rate, and TB10Y is the 10-year Treasury Bond rate.
- There was no constant and trend included in the final augmented Dickey-Fuller tests after starting with 5 lags with a trend and constant. All lags generated test statistics that reject the unit root null.
* denotes significance at the 5 per cent level and ** denotes significance at the 1 per cent level.
Table 6 Cointegration Tests, 1979(1)-1997(3), normalised on the Cash Rate.
Variable Coefficient
t-value
ADF
CRDW
TB3M 0.99168
22.583
-6.7092**
1.64
BILL90 0.97421
45.607
-4.7402**
1.17
TB10Y 1.4063
14.556
-3.2832*
0.55
ADF is the Augmented Dickey-Fuller test and the CRDW is the Cointegrating Durbin Watson Test. Both yielded significant results.
* denotes statistical significance at the one per cent level.
** denotes statistical significance at the one per cent level.
Table 7 Twin Deficits Causality
Lag
Sample Period
Test Statistic Does Budget Deficit 12
1981(4)-1997(2)
F(12,38) = 0.5614 Granger-cause the 8
1980(4)-1997(2)
F(8,50) = 0.41082 Current Account Deficit? 4
1979(4)-1997(2)
F(4,62) = 0.80449 Does Current Account 12
1981(4)-1997(2)
F(12,38) = 1.0185 Granger-cause the 8
1980(4)-1997(2)
F(8,50) = 1.2217 Budget Deficit Deficit? 4
1979(4)-1997(2)
F(4,62) = 2.0031
Source: Australian Bureau of Statistics, AUSSTATS Time Series Service.
The Budget Deficit is expressed as a percentage of GDP as is the Current Account Deficit. The data is quarterly and the Granger-causality testing regression used the change in each variable.
References
Argy, Victor (1992). Australian Macroeconomic Policy, (Sydney: Allen and Unwin).
Bank of International Settlements (1994). Annual Report 1993/94, (Basel: BIS).
Barro, Robert J. (1993). Macroeconomics, Fourth Edition, (New York: John Wiley).
Bispham, J. (1988). "Rising Public-Sector Indebtedness: Some More Unpleasant Arithmetic", in M. Boskin et al, (eds), Private Saving and Public Debt, (Oxford: Blackwell).
Blinder, Alan (1987). Hard Heads Soft Hearts, (Reading: Addison-Wesley).
Christiansen, Hans and Pigott, Charles (1997), "Long-term Interest Rates in Globalised Markets", OECD Working Paper No. 175.
Davidson, Paul (1994). Post Keynesian Macroeconomic Theory, (Vermont: Edward Elgar).
Engle, R.F. and Granger, C.W.J. (1987). "Cointegration and error correction: Representation, estimation and testing", Econometrica, 55, pp.251-276.
Glyn, Andrew (1997). "Paying for Job Creation", in Jonathan Michie and John Grieve Smith (eds.), Employment and Economic Performance: Jobs, Inflation and Growth, (Oxford: Oxford University Press), pp. 221-233.
Gordon, Wendell (1997). "Job Assurance - The Job Guarantee Revisited", Journal of Economic Issues, Vol. XXX1, No. 3, September, pp. 817-825.
Granger, C.W.J. (1969). "Investigating causal relations by econometric models and cross-spectral methods", Journal of Econometrics, 39, pp. 199-221.
Hargreave-Heap, S. (1980). "Choosing the wrong 'Natural' Rate: accelerating inflation or decelerating employment and growth", The Economic Journal, Vol. 90.
Johansen, S. (1988). "Statistical analysis of cointegration vectors", Journal of Economic Dynamics and Control, 12, pp.231-254.
Kaldor, N. (1978). Further Essays on Economic Theory, (New York: Holmes and Meirer).
Kalecki, Michal (1971). "Political Aspects of Full Employment", in Kalecki, selected essays on the dynamics of the capitalist economy, (Cambridge: Cambridge University Press), pp. 138-145.
Kasman, B. and Rodrigues, A. (1991). "Financial liberalisation and monetary control in Japan", Federal Reserve Bank of New York, Quarterly Review, Fall.
Keynes, J.M. (1919). The Economic Consequences of the Peace, (London: Macmillan and Co).
Keynes, J.M. (1937). "The Ex-Ante Theory of the Rate of Interest", The Economic Journal, 47.
Keynes, J.M. (1940). How to Pay for the War, (London: Macmillan and Co).
Kitson, M., Michie, J. And Sutherland, H. (1997). "A Price Well Worth Paying? The Benefits of a Full-Employment Strategy", in Jonathan Michie and John Grieve Smith (eds.), Employment and Economic Performance: Jobs, Inflation and Growth, (Oxford: Oxford University Press), pp. 234-253.
Layard, R., Nickell, S. and Jackman, R. (1991). Unemployment, Macroeconomic Performance and the Labour Market, Oxford University Press, Oxford.
Mitchell, William F. (1987). "The NAIRU, Structural Imbalance and the Macro-equilibrium Unemployment Rate", Australian Economic Papers, Vol. 26, No. 48, pp. 101-118.
Mitchell, William F. (1996). "Inflation and Unemployment: A Demand Story", Presented to European Unemployment Conference, sponsored by the European Commission, at the European University Institute, Florence, November 21-22, 1996, forthcoming in published proceedings.
Mitchell, William F. (1997). "Measuring the impact of the Accord on Inflation", Paper presented to the End of the Accord Conference, Victoria University, Melbourne, November 26th., forthcoming in published proceedings.
Mitchell, William F. (1988a). "The Buffer Stock Employment Model and the NAIRU: The Path to Full Employment, Journal of Economic Issues, Vol XXXII, No. 2, June, pp. 1-9.
Mitchell, William F. (1988b). "Why deficits do not need to be financed", mimeo, Department of Economics, University of Newcastle.
Mitchell, William F. (1999a). "Full Employment and Environmental Sustainability", mimeo, Department of Economics, University of Newcastle, forthcoming.
Mitchell, William F. and Burgess, J. (1998a). "Eight Propositions about Unemployment", IIRA World Congress Proceedings, Bologna, Italy.
Mitchell, William F. and Burgess, J. (1998b). "Full Employment as a Human Right", Australian Journal of Human Rights, June.
Mitchell, William F. and Watts, M.J. (1997). The Path to Full Employment, Australian Economic Review, 4th Quarter, 1997.
Mosler, Warren (1996). Full Employment and Price Stability, available at http://www.warrenmosler.com.
Ormerod, Paul (1994). The Death of Economics, (London: Faber and Faber).
Orr, Adrian, Edey, Malcolm and Kennedy, Michael (1995). "The Determination of Real Long-term Interest Rates: 17 Country Pooled-Time-Series Evidence", OECD Working Paper No. 155, (Paris: OECD).
Radecki, L.J. and Reinhart, V.R. (1989). "The globalisation of financial markets and the effectiveness of monetary policy instruments in the United States", International Interest Rate Linkages and Monetary Policy, BIS, March.
Remolona, E.M. (1991). "Do international stock and bond markets reflect macroeconomic fundamentals?", Federal Reserve Bank of New York, Quarterly Review, Autumn.
Summers, L. (1986). "Does the stock market rationally reflect fundamental values?" Journal of Finance, July.
Thurow, Lester (1983). Dangerous Currents, (Sydney: Allen and Unwin).
Thurow, Lester (1996). The Future of Capitalism, (Sydney: Allen and Unwin).
Tool, Marc (1996). "Employment as a Human Right", Conference Paper for the European Association for Evolutionary Political Economy, Antwerp, November 1996.
William Vickrey (1996), " Fifteen Fatal Fallacies of Financial Fundamentalism : A Disquisition on Demand Side Economics", published electronically by Columbia University and available at http://www.columbia.edu/dlc/wp/econ/vickrey.html. This was Vickrey's last academic paper.
