The fiscal multiplier during the recession of 2008 in the Netherlands

(1)

The fiscal multiplier during the recession of 2008 in

the Netherlands.

Bachelor Thesis

Faculty of Economics and Business

By: Marinda de Vries

Student number: 11029285

Thesis supervisor: Péter Foldvari

University of Amsterdam, 26-06-2018

(2)

2

Statement of originality

This document is written by Marinda de Vries, who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Abstract

In this research it is examined what the effect of additional government purchases is on consumption and output during a recession. For this reason the fiscal multiplier is estimated in the Netherlands for the period 1999 until 2017. The long term effect of government purchases on consumption is close to zero, and thus are not effective in stimulating consumption. The long term effect of government purchases on output is close to zero and probably even negative. Moreover this effect is not significantly changed by the recession. The multiplier is not significantly different from zero and is not significantly changed by the recession. Thus fiscal policy is not effective in stimulating consumption or output, and does not become more effective during a crisis.

I. Introduction

The Netherlands has never endured such a vehement crisis as the great recession recently. There is much disagreement among economists about the best fiscal policy to increase gross domestic product and decrease unemployment to get the economy out of the recession (Parker, 2011). The variety of fiscal policies adopted during the crisis by different countries suggest uncertainty about which approaches are most effective (Auerbach, Gale, & Harris, 2010). Some say no additional fiscal stimulus is needed since this additional government spending leads to an equal amount of decrease in other components of Gross Domestic Product, such as consumption and investment (Barro, 2009). Others pursue the vision of Keynes with his countercyclical policy and plead for more stimulus. They argue that each additional dollar of government purchases would raise Gross Domestic product by 1,5 dollars (Krugman, 2009). This division among economists proves it is unclear to what extent this additional government spending is responsible for the growth in Gross Domestic Product. Besides additional government spending also leads to higher government deficits which needs to be repaid in the future. According to Ricardian Equivalence an increase in government spending leads to a short term stimulus of output and employment (Seater, 1993). This additional stimulus in its turn raises interest rates because private debt has to compete with public debt and in this way crowds out private investment. Besides rational economic agents know this increase in debt will be financed by an increase in taxes in the future. In this reasoning additional government spending will have no to little effect on economic activity.

One way economists use as an approach of assessing the effectiveness of fiscal policy is to look at the fiscal multiplier. According to Hall (2010) the multiplier is measured by the increase in Gross Domestic Product caused by each dollar of government spending. Since output is added by the amount of additional government spending times the fiscal multiplier, a higher multiplier would

(5)

5 mean higher effectiveness of fiscal policy. The multiplier became an important point of interest again during the Great Recession when nominal interest rates were brought down to its zero lower bound and monetary policy was no longer as effective as always thought (Ramey, 2011). Before the Great Recession most economists were more interested in the effects of monetary policy than those of fiscal policy on the economy. They believed that the lag in the implementation was too big to be useful to overcome a recession. But the huge amount of additional fiscal stimulus in the United states during the recession would indicate economists and policy makers still believe that fiscal policy is an effective method to overcome a recession (Parker, 2011). One of the main concerns of economists was if fiscal policy is still as effective during a crisis as in times of economic growth in stimulating output? Therefore this research tries to answer this question by examining the multiplier before, during and after the recession that started in 2008. There has been done research in estimating the multiplier during recessions, but mostly this research has been focussed on Amerika and there is a lot of variations in these estimates (Auerbach, Gale, & Harris, 2010). Depending on the models used to estimate the multiplier, it varies from 0,5 to far above one. So this research will focus on the multiplier in The Netherlands for almost the last two decades and question whether the multiplier has changed and if this change was due to the recession. First literature is discussed in section II and based on these previous findings a hypothesis is formed for this research. Hereafter the research method and data collection is explained in section III followed by the results of this research in section IV. At last a discussion in section V is followed by the conclusion in section VI.

(6)

6

II. Literature review

There are many different ideas about the effect of government spending on consumption and output. One of the first to write about this subject was John Maynard Keynes. Keynes (1936) explains the idea of countercyclical policy to overcome a recession. This means that the government should stimulate the economy through additional spending in times of economic downfall, and exert contractionary fiscal policy when the economy is in a boom. According to Auerbach and Gorodnichenko (2012) most of the recent literature is in line with the Keynesian idea that the expansionary effect is likely larger during recessions than during booms. Keynes (1936) states that consumption increases through the increase in government spending. This increase in output is measured by the fiscal multiplier. He uses the marginal propensity to consume to calculate the fiscal multiplier. The higher the marginal propensity to consume the higher the fiscal multiplier is. The intuition behind this is, if people spend a greater part of their income on consumption instead of saving it, the greater the effect of additional government spending. In his model Keynes assumes that real wages are rigid. Which means that firms equate the marginal product of labour to the real fixed wage and in this way determine output. When output is unresponsive, private investment is crowded out with increased government spending. According to Hall (2009) this will lead to a multiplier of zero. In his paper Parker(2011) discusses that the countercyclical fiscal policy that Keynes (1936) designed could be effective, but only if the recourses in the economy are slack. Besides Auerbach, Gale and Harris (2010) claim that the traditional Keynesian multiplier could be large when there is much unemployment. Beetsma and Giuliodori (2011) agree with Keynes that an increase in government spending leads to an increase in output, consumption and investment. Moreover they discuss that the effect of additional government purchases would be larger in a closed economy than in an open economy due to the fact that these benefits leak to other countries in an open economy. However there is much criticism on the model of Keynes. Woodford (2011) claims that the model of Keynes does not asses the expectations of aggregate economic activity and the role of intertemporal optimization as much as they should.

As a reaction to the criticism on the Keynesian model, the ideas were revised and transformed into the New Keynesian model. In the New Keynesian model sticky prices and wages are assumed. Besides that economic agents are forward looking and are assumed to have rational expectations. Furthermore monopolistic competition is assumed instead of a free market. In his paper Woodford (2011) explains the key factors that determine the fiscal multiplier of government spending in New Keynesian models. First fiscal policy will not be very effective if the central bank will tighten monetary policy in response to the increase in government spending. Since the short term nominal interest rate was at its lower zero bound during the great recession, this was not the case

(7)

7 and it is expected that fiscal policy would be more effective. In fact the European Central Bank also tried to stimulate the economy by conventional methods, such as open market operations, and also by some unconventional methods such as quantitative easing. According to Woodford (2011) and Hall (2009) the multiplier should be above one when the interest rate is at the zero lower bound. So that would mean that the multiplier would be higher during a recession when the interest rate is at its zero lower bound as was the case in the recession of 2008.

Further the level of unemployment could be a determinant of the effectiveness of fiscal policy. According to Parker (2011) fiscal stimulus is only effective when unemployment is high and capacity is not fully utilized. Since unemployment rate rose quite severe during the Great recession in the Netherlands, this could lead to a bigger multiplier (Graaf-Zijl, Horst, Vuuren, Erken, & Luginbuhl, 2015). In other countries of the European union the wages absorbed a larger part of the shock than in the Netherlands, which indicates more wage rigidity in the Netherlands. Since the wages could not decrease in response to a decrease in the demand of labour, employers were forced to fire people in times of financial distress. This higher wage rigidity led to an increase in the unemployment rate in the Netherlands during the recession, and could contribute to a higher multiplier. Besides that Ramey (2011) argues that the total output can only increase in the short run when the hours worked increase. Hall (2009) concludes that the multiplier would be fairly higher if employment would be elastic to an increase in demand.

According to Ramey (2011) also the way these additional government spending is implemented plays a crucial role in the effectiveness and the size of the multiplier. A temporary increase in transfers to economic agents should have little effect on output according to the Ricardian equivalence. If economic agents expect an increase in government spending today to be financed with extra taxes in the future, they anticipate this and will not see this as a permanent increase in their income. Therefore they will not increase their consumption.

Not all key factors and models discussed above will be used to formulate a hypothesis that will be researched. Previous literature suggests that the fiscal multiplier would be higher if the resources are slack in an economy. So when unemployment is high, the effect of government spending would be larger and the additional stimulus would have a greater effect (Barro, & Redlick, 2011). Since there was an important rise in unemployment in the Netherlands caused by the shock of the recession it is expected that the multiplier would increase (Graaf-Zijl et al., 2011). Besides Beetsma and Giuliodori (2011) found that additional government purchases increases output, consumption and investment. Therefore this research expects that the multiplier will increase due to higher government purchases during the Great Recession.

(8)

8

III. Research method

III.I Data

In some of the research above (Hall, 2009; Ramey, 2011; Barro, & Redlick, 2011) military spending was used as a proxy of government spending in the United States to estimate the multiplier. Military spending would be the optimal example according to Hall (2009). If consumers would receive goods and services from the government which they would have bought anyway, the multiplier would be much lower. Besides military spending, government purchases is often considered as the most important indicator for government spending. Since military spending is only a small amount of total government expenditure in the Netherlands, and it is not likely to increase during the recession, this research will use government purchases as an indicator of government spending to estimate the multiplier.

For this research a time period is chosen of the first quarter of 1999 until the last quarter of 2017. Because of this relatively large sample the Central limit theorem applies. According to the central limit theorem the sampling distribution of the mean become closer to the real value of the population when the sample size increases and is approximately normally distributed (Keller, 2012). The sample must be a random draw of the population. When the central limit theorem applies and the variables are approximately normally distributed a t-test can be performed for which normality is required. Besides it is useful when the sample is relatively large to examine whether the effect of fiscal policy has changed over time, since you have a lot observations before and after the crisis. Quarterly data about the variable Gt, government purchases of the Netherlands, is obtained from the Central Bureau of Statistics (CBS) of the Netherlands for the time period 1999 until 2017. For the variable consumption Ct, quarterly data is obtained from Eurostat of final consumption of expenditure of households for the same time period. For the variable output Yt, quarterly data is obtained from the Central Bureau of Statistics (CBS) of Gross Domestic product for the time period. For the variable tax revenue of the government Tt, quarterly data is obtained from the Central Bureau of Statistics (CBS) of total tax income for the time period. All variables are not seasonally adjusted and at current prices. For all variables a lag is used to create a dynamic model which can account for a lag in the effect of variables as is suggested by Ramey (2011).

III.II Empirical testing

First this research wants to check whether government spending indeed has a positive effect on household consumption as literature suggests. To test if this relationship holds the following regression equation is estimated using an OLS regression which uses the variables as explained above.

(9)

9 𝑙𝑛𝐶𝑡 = 𝛼1 + 𝛼2 ∗ 𝑙𝑛𝐶𝑡 − 1 + 𝛼3 ∗ 𝑙𝑛𝐺𝑡 + 𝛼4 ∗ 𝑙𝑛𝐺𝑡 − 1 + 𝜀𝑡

(1) The natural logarithm of each variable is taken to get elasticities, which as a result can be interpreted as percentages rather than the actual value (Stock, & Watson, 2011). Where Ct is consumption at time t and Gt is government purchases at time t. Ct-1 and Gt-1 are the lags of the variables Ct and Gt and εt is the error term

.

The coefficients of this OLS regression will be used to test with a t-test whether they are significantly different from zero, and thus if government purchases have an effect on consumption. The alpha’s are estimated using heteroskedastic robust standard errors to account for any heteroskedasticity in the error terms (Stock, & Watson, 2011).

Thereupon a regression will be executed to estimate the effect of government purchases on output with an OLS regression using the following regression equation, with taxes as a control variable.

𝐿𝑛𝑌𝑡 = 𝛼1+ 𝛼2 ∗ 𝐿𝑛𝑌𝑡 − 1 + 𝛼3 ∗ 𝐿𝑛𝐺𝑡 + 𝛼4 ∗ 𝐿𝑛𝐺𝑡 − 1 + 𝛼5 ∗ 𝐿𝑛𝑇𝑡 + 𝛼6 ∗ 𝐿𝑛𝑇𝑡 − 1 + 𝜀𝑡 (2) Where Yt is Gross Domestic product at time t, Yt-1 is the lag of Gross Domestic product, Gt is government purchases at time t, Gt-1 is the lag of government purchases, Tt is total tax revenue at time t, Tt-1 is the lag of total tax revenue, Ct is consumption at time t, Ct-1 is the lag of consumption and

𝜀𝑡

is the error term. The coefficients of government purchases and government purchases of the previous period will be used to test with a t-test whether there is a significant effect of government purchases on output with a 5% level.

Further there will be tested if there is a structural breakpoint in the data to check whether output responds differently to government purchases during the crisis than before. This will be tested using the Quandt-Andrews unknown breakpoint test (Quandt, 1960; Andrews, 1993). This test is a generalized form of the Chow test. The Chow test determines if all observations belong to the same regression (Chow, 1960). In other words this test regards whether an economic relationship is stable over time or if there is a change in this relationship at a known point. This model assumes that the residuals are independent and identically distributed from a normal distribution with unknown variance. First a regression is executed of the data before the date that you expect there is a break. From this regression you receive the sum of squared residuals. Then a regression is carried out of the data after the date where you expect the break point and collect the sum of squared residuals. Thereafter a regression is executed of the total data and collect the sum of squared residuals again. Once al these sum of squares residuals are collected a Chow test can be done using the following equation.

(10)

10

𝐹 =

(𝑆𝑆𝑅3−(𝑆𝑆𝑅1+𝑆𝑆𝑅2))/𝑝

(𝑆𝑆𝑅1+𝑆𝑆𝑅2)/(𝑛+𝑚−𝑝) (3)

Where SSR1 is the sum of squared residuals of the data before the breakpoint, SSR2 is the sum of squared residuals of the data after the breakpoint, SSR3 is the sum of squared residuals from the combined data, p is the number of parameters in the regression, n is the number of observations in the first regression and m is the number of observations in the second regression. The Quandt-Andrews unknown breakpoint test (Quandt, 1960; Andrews, 1993) works similar, but the difference is that for this test the time of the break does not need to be known. The Quandt-Andrews unknown breakpoint test determines in a trimmed period in the sample whether there is a structural break and when this structural break is. For every period in the trimmed period it executes a Chow test, and the breakpoint is chosen where the test statistic is the highest. Further the multiplier will be generated with a similar method as Hall (2009).

𝑌𝑡−𝑌𝑡−1

𝑌𝑡−1

=

𝛼1 +

𝑚 ∗

𝐺𝑡−𝐺𝑡−1

𝑌𝑡−1

+ 𝜀𝑡

(4)

Where Yt is Gross Domestic Product at time t, Yt-1 is the lag of Gross Domestic product, Gt is government purchases at time t, Gt-1 is the lag of government purchases, m is the multiplier and

𝜀𝑡

is the error term. First a variable is generated which represents the change in output divided by the output of the previous period. Another variable is generated which represents the change in government purchases divided by the output of the previous period. Then the coefficient of the multiplier for the period of 1999 until 2017 of the Netherlands is estimated using an OLS regression and the variables explained above.

Once the multiplier is estimated there will be tested if there is a structural break in the multiplier using the Quandt-Andrews unknown breakpoint test (Quandt, 1960; Andrews, 1993). If there is a structural break in the multiplier this means that government spending is less or more effective after the breakpoint in stimulating output. If the structural break is significant, an OLS regression will be performed on the data with a dummy break included which is one if the period is at the breakpoint or later and zero otherwise. The result of this regression will be used to determine whether the multiplier has changed during the crisis and if fiscal policy is more of less effective in stimulating output during a crisis. If the dummy break is negative and significant this means that government purchases is less effective in stimulating output during a recession. If the dummy break is positive and significant this means that government purchases are more effective in stimulating output during a recession.

(11)

11

IV. Empirical analyse

First is tested whether government spending has a positive relationship on consumption using equation one. As can be seen in table one all three variables are significant at a 1% level. Which means that consumption at time t can be explained by the lag in consumption, government purchases and the lag in government purchases. Government purchases at time t indeed have a positive effect on consumption as literature suggests. However the lag of government purchases have a negative effect on consumption. This negative coefficient can be due to the fact that economic agents expect these government purchases of the past to be financed in the nearby future and thus will not change their consumption according to the Ricardian equivalence. This result indicates that government purchases only have an effect on consumption in the same period, but this effect is almost completely reversed again in the next period. This means that in the long term the effect of government purchases on consumption will be close to zero and probably even have a negative effect due to crowding out.

Table 1. Ordinary Least Squares estimates of government purchases on consumption.

(1)

VARIABLES

lnc

L.lnc

0.960***

(0.0381)

lng

0.147***

(0.0196)

L.lng

-0.134***

(0.0224)

Constant

0.342 (0.260)

Observations

75 R-squared

0.984 Robust standard errors in parentheses

* p<0.01, p<0.05, * p<0.1

Next the regression which tests whether government purchases has a significant effect on Gross Domestic Product was constructed. These results are summarized in table two below. Striking is the fact that both government purchases at time t as well as its lag are negative. An explanation for this could be the Ricardian Equivalence. The rational economic agents expect that the increase in government purchases will be financed by an increase in taxation in the future and therefore start saving instead of consuming. Which decreases output in its turn. Another explanation for these

(12)

12 negative coefficients could be the effect of crowding out. As explained before additional government purchases tend to increase the interest rates which causes private investment to be crowded out and thus decreases. This decrease in private investment in its turn leads to lower output. This could also lead to higher unemployment for instance since investments are postponed which makes companies less efficient and not all capacity is utilized in the economy. Besides it is surprising that government purchases at time t are not significant at any level. This could indicate that the multiplier that is investigated later in this research will be close to zero. However government purchases at time t-1 are significant. This could illustrate the lag that is observed in previous literature in the effect of the implementation of government spending on output. An explanation could be that economic agents do not always observe the change in government purchases immediately but it takes some time in finding out. Further total tax revenue at time t as well as its lag are significant. Which indicates economic agents respond more rapidly to a change in taxation than to a change in government purchases. However the lag of total tax revenue is still significant, so this also indicates there is a lag in the effect on output.

Table 2. Ordinary Least Squared estimates of government purchases on Gross Domestic Product.

(1)

VARIABLES

lny

L.lny

0.979***

(0.0295)

lng

-0.00836

(0.0124)

L.lng

-0.0284***

(0.00978)

lnt

0.0167*

(0.00886)

L.lnt

0.0255**

(0.0118)

Constant

0.155*

(0.0850)

Observations

75 R-squared

0.998 Robust standard errors in parentheses

* p<0.01, p<0.05, * p<0.1

To test whether there is a structural break in the effect of government purchases on output a Quandt-Andrews unknown breakpoint test is performed on this regression. The trimmed period of the sample consists of the second quarter of 2002 until the first quarter of 2015. The highest test statistic is shown at the third quarter of 2009, F(5, 75) = 30.06, p < .01. This result implies that the

(13)

13 structural break on Gross Domestic Product took place in the third quarter of 2009, roughly a year later than the beginning of the recession. Now the date of the structural break is known, a dummy variable is created which is one if the period is at the third quarter of 2009 or later and zero otherwise. Another regression is run with the variable break included using the following equation. 𝐿𝑛𝑌𝑡 = 𝛼1+ 𝛼2 ∗ 𝐿𝑛𝑌𝑡 − 1 + 𝛼3 ∗ 𝐿𝑛𝐺𝑡 + 𝛼4 ∗ 𝐿𝑛𝐺𝑡 − 1 + 𝛼5 ∗ 𝐿𝑛𝑇𝑡 + 𝛼6 ∗ 𝐿𝑛𝑇𝑡 − 1 + (5) 𝛼7 ∗ 𝑏𝑟𝑒𝑎𝑘 + 𝜀𝑡

Table 3. Ordinary least squared estimates with dummy break.

(1)

VARIABLES

lny

L.lny

0.965***

(0.0239)

lng

-0.00584

(0.0110)

L.lng

-0.0300***

(0.0101)

lnt

0.0185**

(0.00919)

L.lnt

0.0285***

(0.0106)

break

0.00350

(0.00362)

Constant

0.265***

(0.0996)

Observations

75 R-squared

0.998 Robust standard errors in parentheses

* p<0.01, p<0.05, * p<0.1

As can be seen from table three the dummy variable break is positive and not significant. This means that output is slightly increased when government purchases is increased after the break date. Since this increase is not significant at any level and very close to zero, government purchases are not effective in stimulating output when the economy is in a recession. Since the variable Lny and the dummy variable break have a log-linear relationship, this means that the coefficient is multiplied by one hundred percent (Stock, & Watson, 2011). So output is increased by 0.35 percent when government purchases is increased by one percent at the breakpoint and thereafter.

(14)

14 Next the multiplier of government purchases is estimated using equation four. The estimated output multiplier using an OLS regression was found to be 0.125 for the time period 1999-2017 and not significant. Since the estimate of the multiplier is not significant the hypothesis that the multiplier is zero cannot be rejected. So these results imply that the government purchases have very little effect in stimulating output and the effect of additional government purchases will be close to zero in the long term.

Table 4. Ordinary least squared estimates of the multiplier.

(1)

VARIABLES

D.lny

gpery

0.125 (0.180)

Constant

0.00810***

(0.000968)

Observations

75 R-squared

0.005 Robust standard errors in parentheses

* p<0.01, p<0.05, * p<0.1

Once the output multiplier was estimated a Quandt-Andrews test was performed to test whether the multiplier has significantly changed during the during the recession. The trimmed period in which the test is executed is the same as for the Quandt-Andrews test above. From the first quarter of 2002 until the first quarter of 2015. The highest test statistic is shown at the second quarter of 2008, F(1,75) = 27.22, p < .01. This result implies that there is a structural break in the output multiplier in the second quarter of 2008. Now the date of the structural break is known a new dummy variable is created which is one if the period is at the second quarter of 2008 or later and zero otherwise. Another regression is run with the variable break included to check whether the multiplier has increased during the recession. An interaction variable is included which test the effect of government spending on output when the period is at the break point or thereafter. The following formula is used to estimate the effect of the break on the multiplier.

𝑌𝑡−𝑌𝑡−1 𝑌𝑡−1

=

𝛼1 +

𝑚 ∗

𝐺𝑡−𝐺𝑡−1 𝑌𝑡−1

+ 𝛼2 ∗ 𝐵𝑟𝑒𝑎𝑘 + 𝛼3 ∗

𝐺𝑡−𝐺𝑡−1 𝑌𝑡−1

∗ 𝐵𝑟𝑒𝑎𝑘 + 𝜀𝑡

(6)

This formula can be rewritten to the formula seven when the dummy break is equal to one.

𝑌𝑡−1

= 𝛼1 + 𝛼2 + (𝑚 + 𝛼3) ∗

(15)

15 The results from this regression can be found in table five. Before the recession the dummy break is zero and only the multiplier and the constant remain in the equation and the rest becomes zero. The multiplier is still not significant and is even negative in the period before the recession. This means that if the government increased its purchases output actually decreased. The constant is positive and significant which means that output grows on its self, regardless of government purchases. When the dummy break is one the following equation is obtained. The overall effect of government purchases on output is positive but not significant.

𝑌𝑡−1

= 0.0126 − 0.00862 + (−0.0184 + 0.0998) ∗

𝑌𝑡−1 (8)

Table 5. Ordinary least squared estimates of multiplier with dummy break included.

(1)

VARIABLES

D.lny

gpery

-0.0184

(0.238)

break

-0.00862***

(0.00164)

gperybreak

0.0998

(0.296)

Constant

0.0126***

(0.000983)

Observations

75 R-squared

0.279 Robust standard errors in parentheses

* p<0.01, p<0.05, * p<0.1

A reason for this low multiplier could be the openness of an economy of a country as suggested by Beetsma and Giuliodori (2011). The multiplier tends to be much lower in open economies because fiscal stimulus leaks abroad. Since the Netherlands is a very open economy with high net exports this would suggest that the multiplier for the Netherlands would be very low, as is observed by this research.

Another explanation for the low and insignificant multiplier can be found in the results of Acemoglu and Scott (1994). They show that an increase in labour income does not predict consumption growth which is expected according to the Rational Expectations Permanent Income Hypothesis of Hall (1978). Consumer confidence is an important indicator and predictor for the growth in consumption. This result implies that economic agents will not increase consumption just because their income is increased by additional government purchases. Thus if economic agents lack

(16)

16 confidence in the economy they will not increase their consumption, regardless of the fiscal stimulus. As can be seen in figure one consumer confidence in the Netherlands collapsed halfway 2007, at the beginning of the financial crisis in the United States. It has remained negative until the beginning of 2015. So this lack of consumer confidence could explain the insignificant effect of government purchases on output and thus the low multiplier.

Figure 1. Consumer confidence in the Netherlands for the period 1986-2018.

Source: Central bureau of Statistics of the Netherlands.

However Konstantinou and Tagkalakis (2011) argue that fiscal stimulus can boost consumer confidence and in this way can stimulate private spending and thus economic activity. A decrease in the tax rate of direct taxes would boost consumer confidence. However the Dutch government did the exact opposite and increased the ratio on value added tax from nineteen to twenty-one percent in October 2012. This increase of the value added tax had a considerable effect on the purchasing power of households. Barro and Redlick (2011) suggest the multiplier is likely to be very small when additional government spending is financed with higher tax rates. So this increase in the tax rate could have attributed to the low multiplier that is observed in the Netherlands.

(17)

17

V. Discussion

Most of the research that has been done in estimating the multiplier during the Great Recession that started in 2008, has been focussed on the United States. In a reaction to the recession, the government of the United States set up many stimulus packages in the form of tax cuts and additional government spending (Parker, 2011). The Netherlands on the other hand was a bit more cautious to implement such policies. This result in much lower additional government spending in the Netherlands than in the United States in a reaction to the recession. This makes it more difficult to asses the effect of additional government spending on output in the Netherlands and in that way can give a biased view on the estimated multiplier after the crisis.

Another difficulty in estimating the multiplier is the lack of data on the effect of fiscal policy on output during a recession. There have not been many recessions in the recent history from which data can be collected and a multiplier can be estimated. In the Netherlands the only recession of which data is publicly available, is the recession of 2008. So this makes it extremely difficult to estimate the multiplier since this is a single occurrence in the data and there cannot be made a comparison to the multiplier of other recessions.

Besides estimates of the multiplier in most previous research rely on the assumption that government purchases have the same effect of output, consumption and other economic variables during economic booms as during a recession (Parker, 2011). Since economic agents have asymmetric reactions to shocks in the economy, this assumption needs to be relaxed to give a more accurate estimate of the multiplier during the crisis. Further research could think of a way to account for these asymmetric reactions of economic agents in the models. In previous research it is suggested that the state of the economy is relevant in estimating the multiplier. Especially the level of unemployment could be a considerable determinant on the size of the multiplier and thus the effectiveness of fiscal policy. When unemployment is high, the multiplier would be higher. Since unemployment rose quite severely in the Netherlands during the crisis according to Graaf-Zijl et al (2015), it would be desirable to include this variable in the model. Furthermore the government of the Netherlands introduces a new tax law in October 2012, as mentioned above, which resulted in lower purchasing power and thus lower consumption of households. Due to this new law the multiplier could have a downward bias for the periods after the implementation. Further research is needed on the effects of this new tax law on consumption and output.

(18)

18

VI. Conclusion

In this research evidence has been found for the positive relationship between government spending and consumption. However there is a negative relationship between consumption and government spending of the previous period. Rational agents will expect these additional purchases to be financed by increased tax burdens in the future and therefore will not increase their consumption. Besides consumption is crowded out by additional government purchases. The total effect of government spending on consumption on the long term is close to zero and maybe even slightly negative. This means that increasing government spending to try to stimulate the economy in a recession through consumption is not effective.

From this research it becomes clear that models that take Keynes’s ideas as their base are inaccurate. An important assumption of Keynesian model, namely the rationality of economic agents, is violated. Besides it is not as simplistic as Keynes stated, and there are many other factors that influence the multiplier except for the marginal propensity to consume. As is seen in the result above, government purchases do not increase consumption in the long run as is assumed by Keynes. Besides economic agents know that additional government purchases are never free and have to be repaid in the future, following the Ricardian equivalence.

These results also indicates evidence for the lag in implementation of fiscal policy as was suggested by Ramey (2011). There have been found significant coefficients for the lagged variables of government purchases as well as tax. So this suggests that there is a lag in the implementation of government purchases as well as in the implementation of a new tax law. Moreover the break point in the third quarter of 2009 suggests that fiscal policy indeed has a different effect on output during recessions than in times of economic booms. However when the dummy break is included in the regression, it is found not to be significant.

The multiplier is very small and not significant in the Netherlands which indicates that fiscal policy is not very effective as a tool to stimulate output. Possible reasons for the small multiplier could be the collapse of consumer confident during the crisis. If economic agents lack confidence in the economy they will not consume and companies will delay their investments, regardless of government purchases. Besides the economic agents expect the additional government purchases to be financed with an increase in their tax burden in the future. This increase in the value added tax rate occurred in the Netherlands in October 2012, probably contributed to the low multiplier. Besides the Netherlands is a very open economy, so it is likely that the benefits of the fiscal stimulus leak abroad.

The Quandt-Andrews test shows there is a structural break in the multiplier, which suggests that the multiplier is different during a crisis. However when the dummy break is added to the

(19)

19 regression it is not significant, which implies that the multiplier of before the crisis has not significantly changed with the multiplier of during and after the crisis. The effect of government purchases on output has not changed by the recession. So this indicates that there is not a significant difference in the effect of fiscal policy during recessions.

To conclude the results of this research, fiscal policy is not effective as a tool to stimulate output, regardless of the state of the economy. A recession does not change the size of the multiplier significantly and in the long run the multiplier is close to zero. Which means that additional government purchases actually decrease other components of output, such as consumption and investment. Instead of trying to stimulate the economy through direct fiscal stimulus, perhaps it would be better if the government would try to stimulate economic agents to work, invest and produce through incentives during a crisis. For instance a reduction in the marginal income tax rate would increase households disposable income and in that way increase consumption and private investment which in its turn increase output.

(20)

20

References

Acemoglu, D., & Scott, A. (1994). Consumer Confidence and Rational Expectations: Are Agents' Beliefs Consistent with the Theory? The Economic Journal: The Quarterly Journal of the Royal Economic Society, 104(422), 1.

Andrews, D. W. K. (1993). Tests for Parameter Instability and Structural Change With Unknown Change Point. Econometrica,61(4), 821-856.

Auerbach, A., Gale, W., & Harris, B. (2010). Activist Fiscal Policy. Journal of Economic Perspectives, 24(4), 141-164

Auerbach, A., & Gorodnichenko, Y. (2012). Measuring the Output Responses to Fiscal Policy. American Economic Journal: Economic Policy, 4(2), 1-27.

Barro, Robert J. (2009, 22 January). “Government Spending Is No Free Lunch.” Wall Street Journal, Retrieved June 19, 2018, from

https://terpconnect.umd.edu/~jneri/Econ305/files/Barro_WSJ_Jan22.pdf

Barro, R. J., & Redlick, C. J. (2011). Macroeconomic effects from government purchases and taxes.(Report). Quarterly Journal of Economics, 126(1), 51-102.

Beetsma, R., & Giuliodori, M. (2011). The effects of government purchases shocks review and estimates for the EU. The Economic Journal : The Journal of the Royal Economic Society, 121(550), 4-32.

Chow, G. (1960). Tests of Equality Between Sets of Coefficients in Two Linear Regressions. Econometrica, 28(3), 591-605.

Graaf-Zijl, M., Horst, A., Vuuren, D., Erken, H., & Luginbuhl, R. (2015). Long-Term Unemployment and the Great Recession in the Netherlands: Economic Mechanisms and Policy Implications. De

Economist, 163(4), 415-434.

Hall, R. (1978). Stochastic Implications of the Life Cycle-Permanent Income Hypothesis: Theory and Evidence. Journal of Political Economy, 86(6), 971-987.

Hall, R. (2009). By How Much Does GDP Rise If the Government Buys More Output? Brookings Papers on Economic Activity, 2009, 183-231.

(21)

21 Keller, G. (2012). Managerial statistics / (9th ed., International ed.). Australia: South-Western

Cengage Learning.

Keynes, J. (1936). The general theory of employment, interest and money. London: Macmillan. Konstantinou, & Tagkalakis. (2011). Boosting confidence: Is there a role for fiscal policy? Economic

Modelling, 28(4), 1629-1641.

Krugman, P. (2009, January 09). The Obama Gap. New York Times (1923-Current File), p. A27. Parker, J. (2011). On Measuring the Effects of Fiscal Policy in Recessions. Journal of Economic

Literature, 49(3), 703-718

Quandt, R. (1960). Tests of the Hypothesis That a Linear Regression System Obeys Two Separate Regimes. Journal of the American Statistical Association, 55(290), 324-330.

Ramey, V. (2011). Can Government Purchases Stimulate the Economy? Journal of Economic Literature, 49(3), 673-685.

Seater, J. (1993). Ricardian Equivalence. Journal of Economic Literature, 31(1), 142.

Stock, J., & Watson, Mark W. (2011). Introduction to econometrics / (3rd ed., Addison-Wesley series in economics). Boston: Addison-Wesley.

Woodford, M. (2011). Simple Analytics of the Government Expenditure Multiplier. American Economic Journal: Macroeconomics, 3(1), 1-35.

The fiscal multiplier during the recession of 2008 in the Netherlands