Full article: Bent Hansen’s theory of fiscal policy

https://www.tandfonline.com/doi/full/10.1080/09672567.2023.2172196

Bent Hansen's theory of fiscal policy

Abstract

Bent Hansen's The Economic Theory of Fiscal Policy contains a macroeconomic model, based on optimising agents, to analyse how fiscal policy can be used to secure full employment and a constant value of money. Focus is on the coordination of economic policy. In addition to giving an account of Hansen's analysis, the choice of model (equilibrium model, inflation model or a Keynesian unemployment model) is discussed. The role of and the modelling of monetary policy is treated, and the possibility of implementing the policy.

1. Introduction

Up to the 1970s Keynesian macroeconomics was the dominating intellectual basis for fiscal policy aiming at stabilising the economy.¹ Besides lacking a generally acknowledged competing macroeconomic theory², an important reason for this (abstracting from a few turbulent years in some countries immediately after the war) was the generally low rate of inflation during the first two decades after WWII. However, there were two periods of higher inflation. The first one occurred during the Korea war 1950–53, the second one occurred during the late 1960s and the early 1970s in connection to the war in Vietnam.

The second of these inflation occurrences started a revision of macroeconomic theory initiated by Milton Friedman and Edmund Phelps and their analyses of the Phillips curve.³ The first one was of shorter duration and did not result in any notable interruption of the development of macroeconomics based on the Keynesian income-expenditure theory. There was a notable exception, Bent Hansen's analysis of how to use fiscal policy to secure both full employment and a stable price level. Hansen (1951) analysed open and repressed inflation starting from an aggregate model with excess demand both in the goods market and the labour market, see Siven (2021). When a few years later in The Economic Theory of Fiscal Policy analysing how to combine full employment with a stable value of money, he instead used a macroeconomic model with choice-theoretic foundations where markets were in non-rationed equilibrium. Markets were however not cleared through decentralised price movements as in the Walrasian model (a neoclassical general equilibrium model where a tâtonnement process guarantees equilibrium. The price system clears all markets, there is no rationing).⁴ Equilibrium was instead attained by fiscal and monetary policy.

Bent Hansen (1920–2002) was one of the most influential economists in Sweden in the 1950s and 1960s. Erik Lindahl was Bent Hansen's tutor in the work with the dissertation A Study in the Theory of Inflation finished in 1951. This work is methodologically influenced by the Stockholm School. After finishing The Economic Theory of Fiscal Policy in 1955 he succeeded Erik Lundberg as head of the Konjunkturinstitutet (National Institute of Economic Research), responsible for making business cycle forecasts for Sweden. Eventually, he made an international career, working for the OECD, in Egypt and moving to the United States, serving at Berkeley where he was chairman for the department of economics for about eight years.

The paper gives a background to Hansen's work The Economic Theory of Fiscal Policy, presents the essence of his analysis, and discusses various aspects of his analysis. I first give a background and an overview of Hansen's analysis. Hansen's analysis of fiscal policy is then put into perspective by a discussion of Hansen's choice of model and the kernel of his analysis, the coordination of economic policy.

2. Background

Sweden has a long tradition of thoroughly worked out official reports of the Swedish Government. Some of these concerned stabilisation policies. The work on the Arbetslöshetsutredningen (The Official Government Report of the Employment Issue) started in 1927 but it is best known for its reports from the 1930s. Here Myrdal (1934) analysed how fiscal policy could be used to expand employment during a depression and Ohlin (1934) how monetary policy could be used for the same purpose.

The treatises by Myrdal and by Ohlin belong to a Swedish tradition of analysing macroeconomic phenomena which Ohlin named the Stockholm School. Ohlin (1937a, 1937b) characterised the Stockholm School by inter alia: (i) aggregate analysis, (ii) distinction between ex ante and ex post, (iii) consistent use of period analysis, (iv) interest in the plans and expectations of individual economic agents, but less so in the development of the quantity of money. Ohlin could have added that the Stockholm School was mainly concerned with problems of unemployment and not by changes of the price level but that was rather self-evident given that the research by the members of the school was essentially completed during the 1930s.⁵ The only notable exception of a work in the vein of the Stockholm School after WWII is Bent Hansen's doctoral dissertation A Study in the Theory of Inflation from 1951. As indicated by the title, Hansen analysed a different situation than the depression economy analysed by the Stockholm School during the 1930s but with similar methods. Hansen was in particular influenced by the period analysis of his tutor Erik Lindahl, see Siven (2021, 4–5).

In 1951 the work on a new official government report concerning stabilisation policy was started. This time the world-wide inflation of the Korea crises had shifted focus from employment to inflation issues, which was reflected in the name of the committee, the Penningvärdesundersökningen (Swedish Government Official Report on the Value of Money). The committee published three reports, Metelius (1955), Hansen (1955)⁶ and Lindberger (1956). Hansen (1955) discussed how fiscal policy could be used to combine full employment and a stable price level. In contrast to the Arbetslöshetsutredningen some twenty years before there was no published official government report on the corresponding role of monetary policy. One possible reason was that at the time monetary policy was not considered as a potent instrument for affecting aggregate demand, another that monetary policy had been fixed to a low-interest goal in Sweden during WWII and the following ten years, see Lindbeck (1975, chapter 7) and Jonung (1993).⁷

The same year as the work on the government report started Hansen had completed and defended his doctoral dissertation on inflation, Hansen (1951). Besides being of high quality and taking up a problem relevant for the government report, the dissertation also contains a discussion of how to combine different means in order to reach certain goals. This was precisely the aim of his work for the government report on the value of money.⁸ Contrary to Keynes (1940) who concentrated his analysis to excess demand in the aggregate goods market Hansen (1951) analysed the interplay between wage and price increases by analysing both the aggregate goods and the aggregate labour market. Depending on the policy used to affect excess demand in the respective market the effect on real wages could be different. This means that there was at least one additional goal⁹ in addition to that of curbing inflation. Hansen was consequently already working on problems of goals and means when he started his work for the committee.

The following discussion will be concentrated to Bent Hansen's report to the Penningvärdeundersökningen, The Economic Theory of Fiscal Policy. While the dissertation analysed an economy with excess demand in the aggregate goods and the aggregate labour market (the two markets explicitly analysed in the book) the main concern of the report is an economy with equilibrium in the aggregate goods and the aggregate labour market.¹⁰ The reason is not an assumption of the price mechanism efficiently equilibrating the markets of the economy. Fiscal policy is instead assumed to be so efficient that not only the goals of economic policy are attained (constant consumer goods prices and full employment) but as a side effect the markets are equilibrated as well. Different worlds are consequently depicted in the dissertation and in the report to the committee, respectively. However, the two works are connected, and not only by the fact that they have the same author, and that only four years separate the publication of them. They were produced in an environment that had similarities, not least by the fact that Erik Lindahl was Hansen's tutor for the dissertation and that Lindahl also (together with inter alia Ragnar Bentzel, Erik Lundberg, Gösta Rehn and Herman Wold) frequently discussed the manuscript to The Economic Theory of Fiscal Policy. There will therefore be some comments below on the interrelation between the two works.

The Economic Theory of Fiscal Policy is divided into three parts, general theory of fiscal policy, micro-economics of fiscal policy and macro-economics of fiscal policy. The three parts will be discussed below.

3. General theory of fiscal policy

Musgrave (1959) discussed three branches of fiscal policy, the allocation of resources, the distribution of income and the stabilisation of the overall economy. Of these branches Hansen's discussion was concentrated to the third.¹¹ The question at issue was given by the committee, how fiscal policy could be used to attain full employment and a stable price level. The first part of Hansen ([1955] 1958) contains an analysis of principle of this question.

Lacking a "true" quantitative model of the Swedish economy, the aim of the government report was to explain analytically how fiscal policy could be used to attain the different goals. Hansen thought that the economy could be described by an extensive equation system where the goals of fiscal policy could be formulated as restrictions on some of the endogenous variables.¹² If the values of the goal variables are specified the next step is to choose a number of fiscal policy parameters (for example tax rates or expenditure decisions) and solve out the values that give the required values of the goal variables. Hansen also discussed monetary policy parameters where monetary policy in contrast to fiscal policy is defined as measures that are directly connected to transactions in the credit market.¹³

Certain goals can be fulfilled automatically, for example full employment in a Walrasian model. The task is then to find a parameter constellation (tax rates and/or expenditure norms) that makes the other goals attain their desired values. A rule of thumb is that the number of means should equal the number of goals. But it is neither a necessary nor a sufficient condition that the number of goals should equal the number of parameters at hand. It is generally not possible to connect a certain parameter to a certain goal, but it may be possible if the model is not fully simultaneous. If all endogenous variables are interdependent (all equations in the system are functions of all endogenous variables) a changed value of a parameter will change the values of all endogenous variables (see pp.14 and 27–28). In the special case that a certain goal variable only occurs in one of the equations (and is the only endogenous variable in the equation), it will be a function of the parameters of that equation. The model is then (partially) recursive.¹⁴

The values of the policy parameters affect the endogenous variables. But it is not possible to talk about the effect of an endogenous variable, for example the budget surplus. A given budget surplus may for example have different effects depending on the volume of government expenditure and income. Here Hansen on p. 41n refers to the works on the multiplier effects of the balanced budget by Gelting (1941) and Haavelmo (1945). The same value of a certain endogenous variable can be brought about by different parameter constellations which in their turn have different effects on the other endogenous variables. This does not contradict the statement that (as noted by Hansen p. 61) the deficit will build up the public net debt and increase future net interest payments. The budget balance will affect the economy in future periods. Myrdal (1934) stressed the importance of budget balance over the cycle. Active fiscal policy however meant underbalancing in the slump and overbalancing in the boom. The old belief in "sound finances" was based on the thought that what is wise for an individual of for a family would also be a wise policy for steering a whole economy. Myrdal translated the principle to hold from a fiscal year to a number of years comprising a business cycle. For further discussion, see Lindbeck (1975, 85–87).¹⁵

Assume that the goals of economic policy are fulfilled to begin with but that the economy is disturbed and that consequently the values of the goal variables are changed. A possibility is that a number of fiscal policy variables are changed so that the goals are fulfilled again. But even if no fiscal policy variables are changed it might be possible that (some of) the goal-variables will not deviate very much from their original values. One example is the Walrasian economy mentioned above. Another example is that the tax scales are constructed so that the system reacts less to shocks.¹⁶ The automatic stabilising effect may be different for different shocks and for different goals. Different parameter constellations may furthermore be more or less effective from a stabilising point of view. Hansen stressed that in order to talk about the automatic stabilising effect a norm of comparison was required. One such norm could be that the different budget items were kept constant through parameter variations.¹⁷ However, there may be different sets of parameter variations that keep the budget items constant. This means that it is impossible to unambiguously talk about the automatic stabilisation properties of the budget.¹⁸

The above reasoning builds on the assumption that economic policy goals are formulated as fixed values of different endogenous variables, not through a goal index. This means that Hansen's treatise did not aim at discussing how to balance different goals against each other but rather how fiscal policy could be used to attain predetermined goals.

4. Micro-economics of fiscal policy

In a number of chapters Hansen treated the effects of income and consumption taxes on household intertemporal planning. The Hicks-Allen consumption theory¹⁹ was the theoretical starting-point for the analysis. This is in contrast to Myrdal (1927, 1934) who in the tradition of Gustav Cassel dissociated himself from utility theory.²⁰

Samuelson (1974) wrote about the "Hicks-Allen revolution in demand theory." Marginal utility theory was abandoned for ordinal preferences with the corresponding indifference curves. Hansen's discrimination between "utility theory" and "modern preference theory" on pp. 116–117 reflects this theoretical change. For choice under risk cardinal measurability of utility was still required. For early discussions of the relationships between the "utility" in the von Neumann-Morgenstern and the Hicks-Allen sense see Ellsberg (1954) and Baumol (1958).

The papers of Hicks and Allen from 1934 were not referred to directly, but Hansen referred to Hicks (1939) and Wold (1952). Chapter 4 of Wold (1952) contains a mathematical presentation of the theory of consumer demand built on ordinal utility. Both Hicks and Allen (1934a, 1934b) and Hicks (1939) are included in the literature list. See also the discussion on ordinal and cardinal utility by Wold, Shackle, and Savage (1952). Wold had read and criticised chapters VII and VIII of Hansen's work.

Even if the analysis of consumer behaviour in chapters VII–IX is based on the assumption of ordinal utilities, there is an exemption. In the derivation of a consumption function on pp. 162–163, Hansen assumes a utility function of a Cobb-Douglas type, which implies cardinal measurability.

In chapter VII Hansen presents an intertemporal theory of household choice. The main argument for this was that fiscal policy will affect household saving. The principles are outlined in Lindahl (1939, 40–51) but Hansen presents an analytical and geometrical²¹ analysis of the question in chapter VII. In the next chapter the effects of direct and indirect taxation on household consumption and saving plans are analysed.

A concrete result of the analysis of household behaviour is that that the difference between the effects of an income and a consumption tax can be formulated as the effect of the rate of interest after tax. Hansen's analysis furthermore gives a good structure for analysing the effects of taxation of household income. But to get clear-cut results, it is necessary to start from more specific assumptions. On p. 162 Hansen presents a Cobb-Douglas type preference function (which assumes cardinal measurability). The function is written as

$$U=q_1^{{\alpha }_1}{q}_2^{{\alpha }_2}$$

where

$q_1$

and

$q_2$

denote consumption in period 1 and 2 and

${\alpha }_1$

and

${\alpha }_2$

are positive constants. Maximisation of the function subject to the budget equation and a savings goal generates a consumption function which resembles the Keynesian one, see pp. 162–163.

Chapter IX contains a number of suggestions of extending the analysis of consumer behaviour in order to take other aspects into considerations. For example, what are the effects on consumer behaviour of taxes on capital, death duties or tax progression? The possibility of extending consumption theory in various directions of relevance for fiscal policy is also discussed. How is behaviour affected by risk or uncertainty? Another example discussed by Hansen.is to take the utility of stocks into considerations, for example the utility of real balances, see Patinkin (1950–51; 1952–53). Individual household behaviour might furthermore be affected by what other households do, see Duesenberry (1962).

The three chapters on the microeconomics of consumer behaviour analyse a large number of aspects of relevance for fiscal policy. Given the level of abstraction, few concrete results were achieved. Less general assumptions, as for example specifying the form of the utility function would be an alternative but would lead to specific results of less general value. Empirical knowledge would in this case be of great value but as Hansen (p. 185) stressed, in 1955 empirical research was practically absent in the area.

The discussion of the effects of taxation on consumer behaviour follows in Chapter X by a corresponding analysis of the effects on firm planning of various taxes under different market forms. Hansen choose not to start from a general intertemporal theory of firm behaviour (analysing for example financial and real investments) as that outlined by Svennilson (1938) and Hicks (1939). Instead he studied the effects of taxes on firm behaviour under different market forms. The chapter also contains a study of the effects of investment duties on real investments of firms. Here a simple form of intertemporal production theory is used. The problem of price stickiness is treated both as a possibility in connection with oligopoly and as a consequent of pricing built on rules of thumb, see Hall and Hitch (1939).

In a chapter on organisations and fiscal policy Hansen discussed the question whether organisations have a preference function and consequently could be analysed analogously to that of households and firms. He also treated the question that organisations may be so large that their behaviour may affect the market solution in macro (for example the wage level). It is then not sufficient to in addition to external facts only start from the preferences of the organisations in order to describe their behaviour. Organisations may namely have a theory of their own of the economic relationships.²²

The microeconomic part of the book only contains analysis of the behaviour of individual households, firms and organisations. No market solutions are analysed. This means for example that the incidence effects of fiscal policy is not treated (but is referred to in the first part of the book) and that there is no discussion of the stability problems of individual markets.²³ The absence of a discussion of market solutions in this part of the book means that its main function is to offer a more extended analysis of individual behaviour than the simplified functions used in the macroeconomic part of the book.

Hansen (1951, 126–129) discussed the aggregation problem when firms are rationed in a labour market with excess demand (firms with a low marginal cost may have greater difficulties in recruiting labour than firms with a higher marginal cost). This is an example of complications arising due to disequilibrium. Hansen (1951, 221–228) also discussed price indexes when relative prices are not constant. Why Hansen ([1955] 1958) did not discuss the aggregation problem may be due to different causes. One possibility is that he did not consider it as serious when working with equilibrium models. Another possibility is that the problem had not been generally formulated at the time (Hansen 1951, 13). The absence of a systematic discussion of the relationship between the microeconomics and the macroeconomics part of the book did not stand out. Keynesian macroeconomics which dominated during the first decades after WWII had no clear microeconomic foundation (this is in contrast to the Stockholm School where Myrdal 1927 and Svennilson 1938 made important contributions, see Siven 1991).

5. Macro-economics of fiscal policy

The third part of Hansen's book deals with macro-economics of fiscal policy. What are the goals of fiscal (and monetary) policy, what fiscal policy means are at hand and in what way will variations of individual means affect the goal variables?

Chapter XII discusses how the goal of a constant value of money and full employment should be formulated. The first of these goals was subject to a long discussion by Knut Wicksell, recommending a constant price level for consumer goods, and David Davidson, who recommended that the price level should vary inversely to total productivity in society. Erik Lindahl held a position close to, but not identical to Davidson's (see Lindahl [1924] 1929 and Hammarskjöld 1955). The goals were formulated before the 1930s, that is before Keynesianism, and concerned monetary policy. When Sweden left the gold standard in September 1931, there was even an official declaration from the Riksbank (the Swedish central bank) that the aim was to stabilise the consumer price level, see Lundberg (1957, 98).

Lindahl (1930, 69–105; 1939, 199–244) contains an analysis of monetary policy in two cases, the first that the goal is a constant price level and the second that the goal is that the price level should vary in inverse proportion to productivity. There is thus only one goal at a time in Lindahl's analysis, but at least two means, the short term loan rate of interest and the bond rate, respectively. Lindahl studied how the means at hand could be used to counteract the effects on the goal of a changed demand for consumer goods or a changed productivity in the production of consumer or capital goods production. The extent to which Lindahl's analysis was a fore-runner to that of Hansen will be commented on later in the paper.

Since fiscal policy might imply changed direct or indirect taxes, the question arises how these taxes should be taken into consideration when constructing the price index. Should fiscal policy aim at constant prices including or excluding indirect taxes? The following discussion is based on a constant consumer price index. The price index chosen is calculated at market price that is including indirect taxes, and excluding direct taxes.

The next question is how to define the second goal of fiscal policy, namely full employment. Hansen choose to define full employment not as a certain amount of employment or allocation of the labour force but as equilibrium in the labour market. The definition implies productive employment, both in the private and in the public sector.

6. Fiscal policy in a closed economy

Hansen's analysis of fiscal policy for a closed economy starts from a simple macro model with a consumption goods, an investment goods market, and a public sector. Given an exogenous wage level and perfect competition in the consumption and investment goods markets the first order conditions for the demand for labour by the firms can be written²⁴ as:

$$w=p_C(1-t_C)\cdot f\prime (N_C)$$

(1)

$$w=p_I(1-t_I)\cdot \phi \prime (N_I)$$

(2) where

$w$

is the market wage level,

$p_C$

the market price of consumption goods,

$t_C$

the indirect (production) tax rate on consumption goods²⁵,

$f\prime$

marginal productivity of labour in the production of consumption goods and

$N_C$

the number of employees in the consumption goods industry. In the case of monopoly

$p_C$

is substituted for by the marginal revenue

$g=p_C(e-1)/e$

where

$e$

equals the price elasticity. The corresponding equation with self-evident notation holds for the investment goods industry.

The disposable incomes of wage earners can be written as

$$L_d=(w(N_C+N_I)+w_O{N}_O+T_L)\cdot (1-t_L)$$

(3) where

$w_O$

is the wage rate in the public sector,

$N_O$

the number of employees in the public sector,

$T_L$

total transfers to wage earners and

$t_L$

the income tax rate for wage earners. The corresponding equation for disposable incomes of entrepreneurs²⁶ is given by

$$V_d=(p_C(1-t_C)\cdot f(N_C)-w{N}_C+p_I(1-t_I)\cdot \phi (N_I)-w{N}_I+T_V)\cdot (1-t_V)$$

(4) where the notation should be evident.²⁷

The demand for consumption goods is a function of disposable incomes of wage earner and entrepreneur households. The equilibrium condition for the consumption goods market is then:

$$(q_C=d_C=)f(N_C)=F(\frac{L_d}{p_C},\frac{V_d}{p_C})+q_C^O$$

(5) where

$f(N_C)$

signifies production of consumer goods,

$F(\frac{L_d}{p_C},\frac{V_d}{p_C})$

is the demand for consumer goods by wage earner and entrepreneur households as a function of their respective disposable real incomes and

$q_C^O$

is the demand for consumption goods by the public sector.

The supply of investment goods

$\phi (N_I)$

is a function of the number of employees in the investment goods industry,

$N_I.$

The demand for investment goods,

$\Phi (q_C,p_I,r)+q_I^O$

depends on the production of consumption goods, the price of investment goods and the rate of interest.²⁸ The demand by the public sector should furthermore be added. The equilibrium condition can then be written as:

$$(q_I=d_I=)\phi (N_I)=\Phi (q_C,p_I,r)+q_I^O$$

(6)

Total employment²⁹ is the sum of employees in the consumption goods and investment goods industry and in the public sector:

$$N=N_C+N_I+N_O$$

(7)

The budget surplus can finally be written as

$$B=t_L(L+T_L)+t_V(V+T_V)+t_C{p}_C{q}_C+t_I{p}_I{q}_I-N_O{w}_O-q_C^O{p}_C-q_I^O{p}_I-(T_L+T_V)$$

(8) where

$L$

denotes total factor incomes of wage-earners households,

$L=w(N_C+N_I)+w_O{N}_O$

and V denotes total factor incomes of entrepreneur households,

$V=p_C(1-t_C)q_C+p_I((1-t_I)q_I-w{N}_C-w{N}_I.$

Note that interest payments on the public debt may be included in the term

$T_V.$

The budget surplus equals the decrease of the net debt of the public sector (outstanding interest carrying debt + quantity of money – claims on the private sector).³⁰ It corresponds to the difference between investments and saving in the private sector. The demand for money vis-à-vis bonds by the public depends on the rate of interest.

The macroeconomic model for a closed economy consists consequently of eight equations which can be used to solve out the eight endogenous variables market price of consumer goods,

$p_C,$

market price of investment goods, pI, disposable incomes of wage earners, $L_d,$ disposable incomes of entrepreneurs, $V_d,$ employment in the consumer goods industry, $N_C,$ employment in the investment goods industry, $N_I,$ total employment, $N$ and the budget surplus, $B.$ Two additional variables can be obtained via the production functions, the production of consumer goods $q_C=f(N_C)$ and the production of investment goods $q_I=\phi (N_I).$

When it comes to exogenous variables the indirect (production) tax on consumer goods, $t_C,$ can be used to influence the market price for consumer goods at given wage rate and marginal productivity of labour in the same industry.³¹ The corresponding holds for the indirect tax on investment goods, $t_I.$ The income taxes for wage earners and entrepreneurs, $t_L$ and $t_V,$ and the respective transfers $T_L$ and $T_V$ can be used to affect the demand for consumer goods whereas the rate of interest, $r,$ in addition to influencing the portfolio decisions of households affects the demand for investment goods. Employment in the public sector, $N_O,$ directly affects total employment, $N.$

The model can now be used to analyse how the government could use parameter variations to preserve full employment and a constant consumer price level in the face of different shocks. One of Hansen's examples of shocks emanates from government itself, namely a mobilisation. The draft implies that a part of the labour force will be withdrawn from production. Assume that the men will be taken from the production of consumer goods. The marginal productivity of labour in this sector will increase so that the indirect tax has to be increased in order to keep a constant market price level of consumer goods. Demand for investment goods falls when production in the consumption goods sector falls. This could be counteracted by a decreased rate of interest or that the government buys more investment goods. The demand for consumer goods will decrease since incomes of both wage earner and entrepreneur households are lower after the draft. The income in military service is normally lower than from civil employment. The profits of the entrepreneurs in the consumption goods industry has moreover fallen since the price before tax as well as the production volume is lower than before. If the demand for consumption goods has fallen more than the production, the income tax has to be decreased in order to preserve equilibrium. We see that three means are required to guarantee that in this example the goals of a constant price level and full employment are preserved after the shock. We could in this example however talk about a third goal, a certain allocation of labour between the consumer and the investment goods sector.

Another example of a shock could be an increased marginal or average productivity of labour in the consumer goods industry. In the first case the indirect tax has to increase in order to keep a constant market price. The wage level will not be affected but the incomes of entrepreneurs in the sector will fall so that the income tax rate has to decrease. If instead only average productivity increases, marginal productivity and consequently the wage rate will be constant. The increased production of consumer goods increases entrepreneurial income, but not enough to let the demand for consumer goods to increase as much as production (a result of the income tax and of a marginal propensity to consume less than unity). The fulfilment of the two goals therefore requires a decreased income tax. The increased production of consumer goods may increase the demand for investment goods. This can be countered by an increased interest rate.³²

The examples together with the fact that some goals can be automatically fulfilled (for example full employment in a Walrasian economy) show that the number of required means are not always equal to the number of goals. Some goals may not even be possible to realise at all with the battery of means at hand.

Up to now the discussion of how to preserve full employment and a constant price level has taken place within the framework of a one-period model. In his treatment of intertemporal problems in Chapter XVI Hansen discussed capital accumulation which could cause both changes of the marginal and average productivity of labour and of private wealth. The budget surplus/deficit will also period by period change government debt and consequently household financial assets. The short-term model is thus a special case of an intertemporal model where the development in one period may result in new shocks in the next period; these shocks require new parameter changes period by period to secure the goals. A possibility discussed by Hansen in Chapter XV (especially pp. 311–317) is that parameter changes within an early period may facilitate economic policy in future periods.

Assume that capital accumulation leads to increased marginal and average productivity of labour by the same percentage in the consumer and the investment goods sectors. If wages for all employees increased by the same percentage the two first order conditions for the demand for labour by the consumption and investment production industries would be left unchanged. The incomes before taxes and transfers of wage earners and entrepreneurs increase proportionally. Disposable incomes may however not increase by the same proportion. The reason may be that transfers are constant or a progressive tax scale. Even if disposable incomes increased proportionally, consumption might depend on an income elasticity of saving different from unity (see p. 334). Wage increases have therefore to be combined with tax changes. The government has furthermore to increase its purchases of goods by the same per cent and change the rate of interest in order to preserve full employment, a constant price level and an unchanged allocation of labour between the sectors.

Wage changes may consequently be a part of a policy battery to secure the different goals. But wage changes may also act as shocks that have to be counteracted by fiscal policy. Wages in Sweden were according to Hansen only partly determined by market forces. They were mainly determined by the powerful labour market organisations (primarily by LO (The Swedish Trade Union Confederation) and SAF (The Swedish Employers' Confederation) and by the public sector. In addition to the wage setting by collective agreement, wage drift³³ might represent the direct effect of the market situation on wages. Naturally this does not hinder that the determination of wages via collective agreement may to a large extent be influenced by the market situation as perceived by the parts behind the agreement.

If economic policy can be used to create a situation where market forces result in wage changes compatible with the goals all is well. But if wage changes to some extent are autonomous, fiscal and monetary policy must be used to counteract the shocks. The direct effect of excessive wage increases on the price level of consumer goods can be counteracted by decreases of the indirect tax $t_C.$ The direct tax has moreover to increase. The policy will however as stressed by Hansen in the long run lead to unreasonable results, see the discussion on pp. 341–342. The direct tax will namely tend towards 100 per cent and the indirect tax will be converted into indirect subsidies where the sky is the limit.

Up to now a homogenous labour market has been discussed. The equilibrium concept is then rather simple. Assume instead that the labour market is composed of several sub-markets with excess demand in some of them and excess supply in some of them. The result is that some firms will have difficulties in recruiting labour. But a "number of factors, geographical, occupational, personal, etc., make it difficult for labour to be transferred from one part of the labour market to another. Thus, we get frictional unemployment, structural unemployment, etc., and difficulties in the definition of the concept of full employment." (p. 345) Can we speak about a macroeconomic equilibrium when the excess demands and excess supplies of the different sub-markets approximately even out even if there then will be some unemployment?³⁴

In a perfect competitive economy, the (short run) equilibrium wage rate could differ between different sub markets depending for example on the geographical situation and the type of production. Some of these wage differences may lead to flows of workers between different locations and different production sectors. There is a parallel structural change among the firms. Various movement costs and costs of change imply that the processes will not be instantaneous. Relative wage rates will change during the process so that some of them will increase and others decrease.

The process will be different in an economy with powerful labour market organisations. Wages may then be fixed by collective agreements. The result is that wages may be flexible upwards via wage drift, but rigid downwards (due to asymmetric wage drift). There might then be a stable wage level if there is excess supply in almost all sub-markets. Successive increases of the demand for labour in in a heterogeneous market may as Hansen describes on pp. 349–350 lead to lower unemployment and a higher wage level.³⁵ Full employment can then only be reached if wage increases are accepted – wage increases that may be incompatible with a stable price level (note however the discussion above about the role of indirect taxes). Policies to alleviate structural problems may consequently be important for creating the necessary conditions for macroeconomic stability.

On pp. 358–367 Hansen discusses the possibilities for the government to influence the policy of the labour market organisations so that it would be in their own interest to demand wage increases which would make the extensive changes of direct and indirect taxes discussed above unnecessary. The government could declare to implement a different tax structure for each outcome of the wage negotiations such as the real disposable income of the workers would be maximised if the wage level preferred by the government was chosen. Hansen's suggestion was more of a thought experiment than a realistic suggestion. It was apparently based on the assumption that the government declaration was trustworthy and politically possible to implement. It also required that the employers' side was weak in comparison to that of the workers.

Hansen's thought experiment can be seen as an alternative to the situation where a strongly centralised trade union movement has the initiative and forms a certain wage policy. How should the government policy in this case be constructed so that full employment and a constant value of money is attained? The matter was extensively discussed in Sweden during the 1950s starting from a proposal (the Rehn-Meidner policy) presented at the LO congress in 1951, Fackföreningsrörelsen och den fulla sysselsättningen (Trade Unions and Full Employment), see Turvey (1952). The aim of the proposal was to avoid cost inflation in a situation of full employment and strong labour unions. The proposal also aimed at improving the movability of labour and thereby resource allocation, cf. the above discussion of the heterogeneity of the labour market.

The Rehn-Meidner policy built on four pillars, see Erixon (2010). Restrictive fiscal policy should decrease aggregate demand and thereby profits. This would increase the resistance against inflationary wage claims from the labour movement. LO should secondly pursue a solidarity wage policy, that is wages should not depend on the particular market situation for the firm or for the branch, the same wage should be paid for the same kind of labour. Labour market policy should thirdly be active with a public employment service that could help firms to fill vacancies and job-searchers to find a job as well as help people to move from geographic areas with few new jobs to areas with a high demand for labour. The policy fourthly implied that the government should actively participate in the reallocation of capital and labour. This should preferably be done by marginal employment subsidies to expanding firms as well as to firms expanding their capital stock.

Hansen concentrated his discussion of the Rehn-Meidner policy to the aspect that high profits might make it difficult for the union movement to avoid making wage claims incompatible with a stable value of money. Hansen suggested three types of policy, see pp. 372–375. A high direct tax on net profits of the firms would be a simple and logical solution. The two other suggestions involved increased indirect taxes (note that this may be inconsistent with fulfilment of the goal of a constant price level at market price). One of these implied decreased production of consumption goods and an increased employment in the public sector. In order to keep employment in the production of consumer goods the other suggestion implied that increased taxes on consumer goods should be combined with marginal employment subsidies.

On pp. 374–375 Hansen discussed similarities and differences between his thought experiment and the Rehn-Meidner policy. An important difference pointed out by Hansen on p. 375 is that "the Rehn-Meidner policy assumes that the State accepts the income distribution, and the size of profits, considered reasonable by the workers. If, on the other hand, the State has any special desires in these matters, such as that profits are to be large in order to secure large savings by private firms and consequently a high degree of self-financing, then this policy cannot be used."

Hansen's question relates to the discussion in Turvey (1952) by Erik Lundberg and Gösta Rehn. Lundberg (1952, 67) asked "where will be the incentives to re-orientation of production when the structure of demand alters and technical progress is uneven? Will there be enough risk taking? Will firms desire to carry out investment projects if they can only finance them by borrowing, for example, from the new State banks which will be formed to look after the saving forced from private income-recipients in the form of indirect taxation? It is not enough to answer by simply saying that big enough subsidies will solve these problems. The greater the proportion of costs of production covered by subsidies, the wider is the degree of detailed State interference with private industry and the less does the price system fulfil its function of allocating resources."

In answer to Lundberg's argument Rehn (1952, 73–74) maintained that "there must be a more deflationary (or less inflationary) fiscal policy than we have had up to now, but, instead, a seemingly more inflationary wages policy. Thus there should be a larger volume of collective saving and consequently low business profits. Specific State measures will then be necessary to prevent reductions in employment wherever they threaten to appear. To achieve this, rather than to persuade trade unions to be restrictive in their wage claims should be the task of economic policy."

The Rehn-Meidner policy has some similarity with Hansen's method of coordinating fiscal policy parameters in order to reach certain goals since it builds on a combination of different policy means to combine full employment with a constant price level. The difference is that Hansen ([1955] 1958) contains a general theory of fiscal policy which is applicable to a great number of different situations whereas the Rehn-Meidner policy is constructed for a special case – Sweden in the 1950s and 60 s with a strong Social Democratic government in close cooperation with a powerful LO.

7. The open economy

Hansen extended his model to cover the case of an extremely open economy where he assumed that prices in foreign currency were given by the world market. To simplify he assumed that consumer goods were imported and investment goods exported. The domestic price of consumption goods is therefore given by³⁶ $$p_C=(1+t_C^M)\cdot \upsilon \cdot p_C^M$$(9) where $t_C^M$ denotes the rate of customs duty, $\upsilon$ the exchange rate (number of units of domestic currency per unit of foreign currency) and $p_C^M$ the world market price in foreign currency for consumer goods. The price of investment goods is analogously given by $$p_I=(1+t_I^X)\cdot \upsilon \cdot p_I^X$$(10) where $t_I^X$ denotes the export bounty (an export subsidy which in case of a negative value functions as an export tax) and $p_I^X$ the world market price in foreign currency of investment goods. Imports are given by $$M_C=d_C-q_C$$(11) and exports by $$X_I=q_I-d_I.$$(12)

The balance of trade surplus reckoned in foreign currency is finally given by $$S=p_I^X\cdot X_I-p_C^M\cdot M_C.$$(13)

The government has two more parameters at its disposal, the rate of customs duty and the export bounty. It can moreover influence the exchange rate by selling or buying foreign currency, which presupposes a currency reserve. Hansen assumed that there was no direct influence of the rate of interest on the exchange rate (the domestic demand for investment goods and thereby exports are however affected by the rate of interest). The reason was presumably the regulations of foreign payments (including capital movements) of the time, see Wihlborg (1993).

The extension of the model to cover an extremely open economy implies that the domestic prices of consumer and investment goods are no longer determined simultaneously with the other endogenous variables. Equations (9) and (10) contain only exogenous variables on the right-hand side. As soon as the two domestic prices are determined, employment in the consumer goods and investment goods industries will at given values of the exogenous variables be determined as well, see (1) and (2). The rest of the endogenous variables (total employment, production of consumer and investment goods, disposable incomes of wage earner and entrepreneur households, and finally the balance of trade surplus) can likewise successively be determined if we know the values of endogenous variables of a lower order. In contrast to the model of a closed economy where the endogenous variables are determined simultaneously, the model for an extremely open economy is consequently recursive.

Contrary to the case of a closed economy, the price level of consumer goods does not depend on the development in the country. The price level is given by the world market price level, the exchange rate and the rate of customs duty.³⁷ Fiscal policy measures that affect supply and demand conditions are no longer necessary to achieve a stable price level. They influence instead the balance of trade surplus.

An equally large increase of the average and marginal productivity in both the consumer and investment industries can (if we want to keep a constant allocation of labour between the industries) now be countered by an increase of the indirect tax rate by the same percentage in both industries. This will lead to a surplus in the balance of trade which may be countered by a decrease of the income tax rate or by the rate of interest. The increased production of consumer goods will increase the demand for investment goods. This effect may require further policy changes.

A proportional increase of world market prices can be met by a reduction of the exchange rate in the same proportion in order to keep constant internal prices (appreciation of the currency). The country may however be bound by international agreement to keep a constant exchange rate and may therefore have to accept changes of the price level.³⁸ There is an additional possibility, however. If a sales tax (turnover tax) is levied in the country this tax could (contrary to the production tax) be used to counter changes in the international price level. The combination of an increased turnover tax and a decreased production tax would for example have the same effect as increased customs duties or increased export bounties, see pp. 416–419.

Hansen also discussed the case that import and/or export prices are not exogenous as in the model of an extremely open economy. World market prices are likely to be influenced by imports and exports of a "big" economy such as that of the U.S. This may also be the case for a "small" economy such as that of Sweden. The country may for example be a major producer of some highly technical goods. Some home prices may furthermore to some extent be independent of international prices because of high transport costs. Services and housing can be taken as examples. As soon as the internal prices are to some extent independent of the international price system, the model will again be interdependent, not recursive as in the case of the extremely open economy.

8. Fiscal policy in an uncertain environment

The last chapter of the book was entitled "Full Employment and a Stable Value of Money in a World of Uncertainty." Here Hansen discussed some problems connected to the incomplete knowledge of the economy.

One possibility is that those deciding on economic policy know the general structure of the economy, but do not know its exact numerical properties. One example of this is that the "true" model is stochastic so that only the multivariate probability density function of the stochastic parameters is known. Lower variances around the different goals are naturally to be desired: "If, on the other hand, neither of the two sets of means gives a smaller standard deviation for both end-variables than the other, one cannot say which set of means is best, unless there is a 'welfare-function' or 'end-index' established by the policy-maker." (pp. 434–435).³⁹ The fact that there may be several goals of fiscal policy means furthermore that stabilisation of the economy is a very complex area: "In contemporary discussion it is often taken for granted that the degree of 'stability' can be measured by changes in one single entity, e.g., the national income measured in money terms, or the gross national product in real terms; what complicates the matter here is that we have several ends, and consequently several measures of stability, and these measures need not point in the same direction." (p. 442).

A more serious case is that the "true" model is not known, even approximately. This may lead to an economic policy that in extreme cases brings the economy further away from set up goals. One example is according to Hansen p. 431 the application of the policy of "sound" public finances (a balanced Central Government budget) to improve the situation during the 1930s. The result was probably instead even higher unemployment.

Even if one has a fair knowledge of the economy in a given point of time, various parameter changes may occur. In order to fulfil the goals of economic policy, the effects of these parameter changes have to be countered by economic policy. However, there may be various lags that complicate this task. It will for example take some time between the shock occurs and a certain endogenous variable is affected (shock time-lag). The policy time-lag should preferably not be longer than the shock time-lag. The policy time-lag consists of an observation, an administrative and an effect time-lag. The observation time-lag is probably the same for fiscal and monetary policy but the administrative time-lag may be shorter for monetary policy whereas the effect time-lag is often shorter for fiscal policy. One way of shortening the policy time-lag would be to make the policy response to shocks more or less automatic.⁴⁰

9. Hansen's choice of model

Hansen's analysis was constructed to answer a question posed by an official Swedish Government report: How could fiscal policy be used to attain full employment and a stable price level? In order to answer the question, it was necessary to have knowledge of the economy. Insufficient knowledge made it according to Hansen necessary instead to starting from a simple model discuss how policy in principle could be shaped. But how should the model look like?

Hansen (1951) had earlier discussed the goal-means problem starting from models for repressed and open inflation respectively. In both cases Hansen worked with aggregated models with one labour and one goods market. The price system is fixed under repressed inflation. An increased income tax may only eliminate the inflationary gap in the commodity market whereas the inflationary gap in the labour market remains unaffected. To eliminate both excess demands two means are necessary. In the case of open inflation both excess demands can be eliminated by an increased income tax. The resulting change of the real wage rate has in this case to be accepted. If there is goal for the real wage rate two parameters must be changed.

Keynes (1940) used a model with an aggregate goods market to analyse open inflation. In this case the rate of inflation can be affected via aggregate demand and the problem of means and goals does not appear explicitly as in Hansen's (1951) more disaggregate model with one labour and one goods market. This may be one reason why Hansen became interested in the goal-means problem.

Another possibility is to start from a Keynesian multiplier model. This was the starting point for Hansen and Snyder (1969) in an analysis of the government budget effects (both discretionary changes and automatic) during the period 1955–1965 in seven OECD countries.⁴¹ It was consequently not a question of analysing how fiscal policy could be used to fulfil certain goals. But the analysis implicitly implies this since the results (together with estimations of certain parameters, for example MPC) could be used to evaluate how aggregate demand could be affected if one wanted to achieve a full employment equilibrium. An argument for using a Keynesian model was that that there was no inflation gap in the seven countries during the period 1955–1965. However, Hansen and Snyder (1969, 38–39) argued that "for years at or near full employment our estimates of the budget effects can only be conceived of as an expression of the potential expansionary or contractionary impact of the budget changes." Hansen (1966) contains a discussion of how to analyse the effects of the budget under inflationary conditions (both demand and cost inflation). The model was not used since it according to Hansen and Snyder (1969, 39) required information which was not available for any of the studied countries.

Musgrave (1959) gives another example of a Keynesian analysis of how fiscal policy can be used to affect aggregate demand in the economy. In contrast to Hansen ([1955] 1958) Musgrave did not analyse the labour market separately, but in Chapter 22 instead offered a detailed analysis of liquidity aspects of fiscal policy.⁴²

Hansen thought that he in The Economic Theory of Fiscal Policy had used a Keynesian analysis in the model for a closed economy and a neoclassical one in the analysis of an open economy:

The closed model is definitely Keynesian; it does not entail any automatic tendency towards full employment; the "effective demand" for commodities occupies a central position in the model; it could be said that employment and effective demand mutually determine each other; the model is a purely interdependent, static model. The extremely open model is in certain respects extremely neo-classical; employment is determined by prices (given from without) and money wages; unemployment or shortage of labour is due to too high or too low (monopolistic) money-wages; the "effective demand" is of no importance whatsoever for employment but only to the balance of payments; deficit or surplus in the balance of payments may be said to be due to too slack or too tight a monetary policy and, possibly, fiscal policy too; the model is a purely recursive, static one. (p. 412)⁴³

Hansen consequently argued that the model for a closed economy is of a Keynesian character since the price mechanism is not assumed to create equilibrium in the markets. Fiscal policy parameters and the rate of interest are however used to create equilibrium in the markets, as well as to reach the goals of economic policy. This implies that the rationing mechanisms that work both under Keynesian unemployment and demand inflation are lacking in Hansen's analysis.⁴⁴ The economy analysed by Hansen can therefore be characterised as a neoclassical equilibrium model without a corresponding dynamic model.

Alternatively, it could be described as a macroeconomic theory with choice-theoretic foundations. The equations of the macroeconomic model are simplifications and aggregations of the behaviour functions (and in some cases marginal conditions). The equilibrating process after a shock is (at given prices) consequently created by political intervention, not by the price mechanism.⁴⁵

The choice of a macroeconomic theory with choice-theoretic foundations with an adjustment mechanism via fiscal policy was advantageous from a pedagogical perspective. Policy operations after a shock both secured the fulfilment of the goals and equilibrium in the markets. No total differentiation of the system was necessary for the analysis. But there were disadvantages as well.

There are at least two reasons why it may be problematic to concentrate the analysis to a macroeconomic equilibrium model. The first one is that if the economy is initially not in equilibrium as described by Hansen's model for a closed and an open economy, respectively, it may take several periods to reach this situation. The second reason is that great non-expected shocks may hit the economy and that fiscal policy is unable to affect the economy fastly enough, see Hansen pp. 436–440 and Musgrave (1959, 501–505).

The result of the lags will in both cases be disequilibrium in different markets. Some deals will nevertheless be agreed upon since real markets are generally described by non-tâtonnement processes, i.e., that transactions also take place out of equilibrium (see Negishi 1962).⁴⁶ The deals made in disequilibrium will naturally not be the deals that would be the result in equilibrium; some rationing mechanisms will be in operation.⁴⁷ In excess supply the sellers will be rationed, in excess demand the buyers. In both cases the notional demand and supply functions are substituted for by effective supply and demand functions. In a macro model with one goods market and one labour market (note that in Hansen's model the goods market is divided into one market for consumption goods and one for investment goods) and implicitly a money market (money, not commodities, serve as the medium of exchange) four types of disequilibrium may exist: Keynesian unemployment (excess supply both in the goods and in the labour market), a pressure of inflation (excess demand both in the goods and in the labour market), classical unemployment (excess demand in goods market and excess supply in the labour market) and finally shortage of labour (excess supply in the goods market and excess demand in the labour market). For further discussion, see Barro and Grossman (1971, 1976) and Malinvaud (1977).

The above discussion of different regimes builds on a high rate of aggregation. In the discussion of a heterogenous labour market Bent Hansen brought up that there might be excess demand and positive wage drift in some parts of the market, excess supply and possibly (wages might be downwards rigid) negative wage drift in some other parts. The same heterogeneity may of course hold for the aggregate goods market. This means that the coordinating role of the price system must hold for the sub-markets even if the government succeeds in setting the total monetary excess demand (the sum of the values of the excess demands of all sub-markets) equal to zero in the labour market and the goods market.

The choice of model does not only concern assumptions regarding the coordination of individual choices but also the level of aggregation and how markets are grouped together. That the number of markets included in the macroeconomic model is of importance for the formulation of the goal-means problem has already been discussed above in connection with the comparison of the models of Hansen (1951) and Keynes (1940).

The grouping of markets at a given level of aggregation depends inter alia on what questions the model is supposed to give an answer to. That Hansen choose to divide the goods market into a consumer goods and an investment goods market may be related to that this was the way it was done in the Keynesian model. Here consumption was assumed to mainly be dependent of aggregate income and investments to depend both on aggregate income and on the rate of interest. But Hansen's choice was probably also affected by the question given by the government report. It was important to discriminate between the market price of consumer goods (which was to be kept constant) and the correspondent price of investment goods.

For questions other that those given be the report of the Swedish Government, other groupings may as discussed by Hansen on pp. 422–428 be more attractive. The problem with two sectors, one sheltered production sector and one subject to international competition was analysed by Edgren et al. (1969); Edgren, Faxén, and Odhner (1973). The division of the aggregate goods market was consequently different from that of Hansen. The so called EFO-model (named after the three authors) assumed a greater productivity growth in the sector open to international competition and equivalent wages (solidarity wage policy). Prices would then increase faster in the sheltered sector and inflation (a weighted average of price increases in the two sectors) would be a function of wage increases. The EFO-model could also be used to calculate the "room for wage increases" at a constant exchange rate, the sum of productivity growth in the sector open to international competition and international price increases in this sector, see Edgren et al. (1969, 138) and Holmlund and Ohlsson (1992, 4).

The EFO-model was constructed in cooperation between the leaders Edgren, Faxén and Odhner of the research departments of the three labour market organisations TCO (The Swedish Confederation of Professional Employees), SAF and LO. As such the cooperation preceding the publishing of the EFO-report had the effect of coordinating their comprehension of the economic situation as well as of the result of various agreements on wages. If this coordination of apprehension of the economic realities also involved the management of the organisations, it would probably lead to less risks of labour market conflicts.

The EFO-model aimed at describing the situation of a small open economy under the Bretton Woods system. It was consequently constructed to give an answer to a different question to that of Hansen's. At an approximately constant exchange rate to the dollar Sweden had small chances of eliminating the effects of price increases in the international markets. But this also means acceptance that the goal of a constant market price level for consumption goods had to be given up.

10. The modelling of monetary policy

Penningvärdesundersökningen which started in 1951 did not publish any separate report on monetary policy. However, Lindberger (1956, 149–152) contains a discussion of how to limit investments of the private firms by credit rationing and Metelius (1955, 152–171) on how the exchange rate, custom duties, subsidies and indirect taxes could be used to isolate the economy from external price increases. Hansen ([1955] 1958) has some analysis of monetary policy where both the interest rate and the exchange rate can be used; the interest rate to affect the demand for investment goods (but not the demand for foreign currency) and changes in the exchange rate to cut off the internal price level from international inflation.

Monetary policy was consequently discussed in the three reports of the Swedish Government, but only to a minor extent. One could ask why this was the case, especially considering that Hansen (pp. 14–18) stressed the importance of coordination between fiscal and monetary policy in order to simultaneously reach the different goals.

One answer is that Keynesianism after the Second World War emerged as the dominant intellectual system behind stabilisation policy. A consequence of this was that monetary policy was degraded in comparison to fiscal policy as an instrument to affect aggregate demand.⁴⁸ One reason for this is the alleged low interest elasticity of the LM-curve.⁴⁹ Monetary policy was not efficient in comparison to fiscal policy when it comes to affect aggregate demand, especially in a recession. Lundberg (1957, 195) formulated it in the following way:

It should be observed that it has been easy for Keynes's successors, starting from Keynes's own theoretical scheme, to show the ineffectiveness of monetary policy. If in an inflationary situation investment is only slightly sensitive to changes in interest rates (though this was not Keynes's own view), there is little point in limiting the supply of money and stiffening credit policy, if the effect of these measures only occurs by means of a corresponding rise in interest rates. In a depression it is in some ways even more difficult to pursue an effective monetary policy: in view of the character of the liquidity function the interest rate cannot be forced down below a certain limit, while investment is extremely insensitive to changes in interest. In Keynes's last theoretical model, there was no place for a direct influence on expenditure, whether for investment or consumption, arising from changes in liquidity.

The matter was already discussed in the Arbetslöshetsutredningen. Hammarskjöld (1935, 104)⁵⁰ discussed both the problem of using expansionary monetary policy in a deep depression and restraining the economy in a boom. Regarding the latter it is pointed out that:

If credit is strongly restrained in a late stage of a boom and is effective it may easily cause a crisis; but there is reason to assume that a crisis caused in this way will be less serious than the break-down that will otherwise happen in a later point of time. The reason for the ineffectiveness of isolated monetary policy is evidently the exaggerated expectations of the future hold by businessmen.

A second answer to the question why monetary policy was not thoroughly discussed in the Penningvärdesundersökningen is the rigid political adherence to the policy of low interest rates in Sweden during 1945–1955. One reason was according to Ohlin (1975, 174–175) that the social democratic government wanted to preserve the value of government bonds emitted during the war.

The thought that the rate of interest could not play any role concerns primarily the effect on investments, that is on aggregate demand. However, it was also believed that the rate of interest was most important when it comes to costs, above all rents, see Lundberg (1957, Chapter VI). The thought that prices were cost determined was also an important reason for the appreciation of the Swedish krona in July 1946 and the removal of the general sales tax in January 1947.

The degradation of monetary policy the first ten years after WWII was in contrast to the case during the 1930s. Then monetary policy had played a more active role than that of fiscal policy. It was according to Lundberg (1957, 54) the fact that the Swedish krona was undervalued by the devaluation in September 1931⁵¹ that helped Sweden to emerge from the depression, expansionary fiscal policy did not appear until 1934–1935. At that time the revival was already on its way.

One indication of the supremacy of monetary policy during the 1930s was that Lindahl (1930, 63) discussed the possibility that the Ministry of Finance should be subordinated to the Riksbank when preparing the budget policy: "If the views of monetary policy are allowed to dominate those of the fiscal policy, the central bank could give certain directives concerning fiscal policy. Fiscal policy would then − instead of hindering − facilitate the regulation of the value of money." Lindahl and Hansen evidently agreed upon the suitability of coordinating fiscal and monetary policy even if they disagreed on whether the central bank or the Ministry of Finance should have the initiative.

A third answer to the question why monetary policy in form of variations of the rate of interest (open market operations and variations of the discount rate) was so little discussed is that traditional market-conform policy to a large extent was substituted for by regulations of various kinds, see Jonung (1993) and Wihlborg (1993). The regulations were introduced during the war when Sweden was isolated from most of its international trade. Foreign trade, foreign exchange transactions (portfolio investments, trade in securities, borrowing from and lending to foreign countries) and building activity were regulated, there was wide-spread rationing, price and wage controls. Most of the rationing was phased out during the autumn of 1945 but new regulations were instituted during the years 1947–1949. Hansen (1951) devoted consequently an important part of his book to the problem of repressed inflation. The reason why Hansen ([1955] 1958) only assumed that the rate of interest affected the demand for investments, not for example the demand for foreign currency, was probably the existence of regulations regarding foreign exchange transactions.

11. The coordination of economic policy

If each goal was a parameter, not an endogenous variable of the economic system, fiscal policy would be easy since the goals and the policy parameters would be the same. But the goals are likely to be endogenous variables determined simultaneously with the other endogenous variables of the system. This means that the government must take the structure of the economy into consideration in its economic policy. Economic policy can therefore be described as a coordination problem.

Assume that the system initially is in equilibrium and the goal-variables have the required values (for example demand for labour equals the labour force and the price level is stable). If then some exogenous variable (for example the marginal productivity of labour) changes, adjustments of a number of government policy parameters may be necessary in order to deal with the effects of change of the exogenous variable on the goals. However, only the goal variables must return to their original values, not the other endogenous variables of the economy. ⁵² This is also the way Bent Hansen's analysis is constructed.

The essence of Bent Hansen's theory of fiscal policy is not the model of the economy; the theory can be applied to different models. Note for example that Hansen first discussed the coordination problem in connection with situation of repressed or open inflation. The theory focuses on the coordination of the values of the policy parameters to obtain the desired values of the different goal variables. But how well the coordination problem is presented depends on the model, its structure, and its realism. In this sense the presentation of the coordination problem and the chosen model are of equal theoretical importance.

Lindahl (1930, 1939) contains, as mentioned in connection with Hansen's macroeconomic discussion, an analysis of how monetary policy could be used to eliminate the effects on a price-goal (a constant price level or a price level that changes in reverse proportion to productivity) of a changed demand for consumer goods or changed productivity in the consumer or capital goods industry: Lindahl's analysis can be seen as a fore-runner to that of Hansen. But it is important to stress that Lindahl did not treat Hansen's coordination problem. The reason is that even if Lindahl discussed various monetary policy means, he only worked with one goal-variable at a time, a price level measure. The problem of coordinating the policy variables in order to simultaneously attain different goals does not occur in this simplified situation.

In his review of The Economic Theory of Fiscal Policy Lowell Harriss (1959, 588–589) pointed out that "This volume is not intended as a handbook for government officials. It is designed for the advanced student, 'to show how problems … of fiscal policy … ought to be dealt with in economic analysis.' It achieves a substantial measure of success." and Kimmel (1959, 128) considered the work "to be of major interest to economic and fiscal theorists."

A possible reason why the two reviewers did not stress the work as a practical guide for stabilisation policy was Hansen's analysis of a small open economy with a centralised system of wage negotiations. On the other hand, Hansen also analysed fiscal policy in a closed economy, which would better suit the U.S. situation.

The reason why the reviewers did not regard The Economic Theory of Fiscal Policy as a practical guide to fiscal policy was probably its generality. It is true that the analysis of economic policy both for a closed economy and an open economy was based on models which at the chosen level of aggregation could form a realistic description of an economy. But to base practical policy on these models required much additional information. This information concerns both the numerical properties of the "real" economy and of various disturbances affecting the economy. It is also necessary that the government has time to react to the information of the disturbances. Hansen consequently wrote that "the main purpose of this book is […] to expound and demonstrate an analytical method." (p. xv)

In the last chapter of The Economic Theory of Fiscal Policy Hansen stressed the incomplete knowledge of the structure of the economy at hand for the policy body. Shocks hitting the economy may moreover not be observed early enough for warding off their effects on the goal variables. Giving these limitations, the deciding body is still assumed to make the best out of the situation, namely to as closely as possible fulfil the goals.

Hansen's account was however not meant to give concrete instructions to those in power regarding fiscal policy. It was rather meant to introduce an analytical method of analysing fiscal policy. How should the means at hand be combined in order to get as close as possible to the specified goals? Those in charge should be trained to think systematically on how to combine different measures in order to fulfil the goals. This systematic thinking also implies that erroneous thoughts on fiscal policy should be weeded out. Hansen's Economic Theory of Fiscal Policy was primarily a contribution to the theory of economic policy. The aspect that his work was a part of a Swedish Government official report and as such meant to be useful for economic policy was of secondary importance.

12. Summary and conclusion

Bent Hansen's analysis of fiscal policy is divided into three parts, general theory of fiscal policy, microeconomics of fiscal policy and macroeconomics of fiscal policy. The first part contains a general discussion of the ends – goals problem. It contains an extensive discussion of principle of the coordination of means in order to attain the specified set of goals. Even if the problem has been discussed by Frisch and Tinbergen, it contains an original discussion and contains the central message of the work.

Part two, microeconomics, gives detailed analysis of how intertemporal economic planning of individual households and firms are affected by changes of the government's fiscal policy parameters (for example tax changes and changes of government outlays).

Part three contains the macroeconomics of fiscal policy, both for a closed and an open economy. Here it is assumed that the aggregate markets are cleared through fiscal policy. That Hansen did not choose to start the macroeconomic analysis from a Keynesian model (which was natural at the time) was due to the assumption that the two goals of fiscal policy (full employment and a constant price level) were attained.

The macroeconomic models (both for a closed and for an open economy) were at the given level of aggregation chosen in order to give a fair representation of the economy at the time. The choice of model was also affected by the formulation of the two goals. This means that neither the specific question, nor the specific circumstances of the time, constitutes the kernel of Bent Hansen's theory of fiscal policy. The kernel is instead how to coordinate the policy variables on order to reach the goals.

Acknowledgments

The translations from Hammarskjöld (1935) and Lindahl (1930) are my own. I am very grateful for comments from Lars Jonung and two anonymous referees.

Disclosure statement

No potential conflict of interest was reported by the author(s).

This article has been corrected with minor changes. These changes do not impact the academic content of the article.