2. Literature Review
5.1 The Data
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
22
5 The Estimation and Result
5.1 The Data
Our main goal is to identify the optimal capital requirement strategies in each country under the theme of European debt crisis. Therefore, we focus our analysis on the 17 country in the Euro Zone, including Greek, Spain, Ireland, Portugal, Italy, Belgium, Germany, France, Finland, Netherlands, Austria, Cyprus, Slovak, Slovenia, Luxembourg, Malta, and Estonia. We use seasonally adjusted quarterly data obtained from European Central Bank (ECB) and DATASTREAM from 2005Q1 to 2011Q3.
The data obtained from two databases uses different methodology to calculate the data, but in general the data is matched.
We define tax (T) as total tax of the government, which is the sum of direct, indirect tax, social contributions, government sales, and capital taxes. As for government total expenditure (G2), it is the sum of current expenditure, which includes transfer, subsidy etc., and capital expenditure.11
When we observe missing data in taxes and other current account items, we do not calculate Interest- Rate-Change-to-Income Effect for the quarter. However, if the missing data are only observed in interest rate, which occurs in some country, we still calculate Interest- Rate-Change-to-Income Effect for the quarter. The total
interest-rate-change-to-income effects calculated are 398.
5.2 Simple Regression
To calculate Interest- Rate-Change-to-Income Effect, we must first identify each value
11 More detailed data definition will be described in Appendix A.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
23
of marginal effect in equation (4.6).We use simple regression to obtain them. In the next section we discuss the common factors that influence Interest-
Rate-Change-to-Income Effect by constructing panel data.
To calculate Interest- Rate-Change-to-Income Effect 12, we first calculate marginal propensity to consumption (Cyd), the effect on investment and money demand from a unit change in interest rate (Ir, Lr), and the effect on money demand and import result from a unit change in income (Ly, My).We use simple regression to obtain these values. We regress consumption on disposable income to obtain marginal propensity to consumption (Cyd); Gross fixed Capital Formation on interbank interest rate, which is the risk-free rate, to obtain effects on investment from a unit change in interest rate (Ir); Real money supply on GDP and interbank interest rate to obtain effect on money demand result from a unit change in income (Ly) as well as a unit change in interest rate (Lr); import on GDP to obtain the effect on import result from a unit change in income (My).According to macroeconomic theory, we expect marginal propensity to consumption (Cyd), the effect on money demand and import result from a unit change in income (Ly, My) to be positive. We expect he effect on investment and money demand from a unit change in interest rate (Ir, Lr) to be negative values.
The result is presented in table 1.
There are few observations worthy of mention here. First of all, almost all sighs are compatible with what we have expected, except effects on investment from a unit change in interest rate (Ir). For all 17 countries effects on investment from a unit change in interest rate (Ir) are all positive, which means when the interest rate
increases, the investment will increases. We contribute this result to the large flow of foreign investment in to Euro Zone countries. When interest rate increases, it will attract more foreign investment, thus making the effects on investment from a unit
12 Refer to equation (4.6) to see how to calculate Interest-Rate-Change-to-Income Effect, or h’.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
24
change in interest rate (Ir) to be positive. The values of the effects on investment from a unit change in interest rate (Ir) are especially large in Spain, Italy, France, Germany and Ireland, indicating that there are large flows of foreign investment injecting into these countries. Besides, is especially large in Germany, Belgium, Netherlands, Luxemburg and Malta, indicating that these countries may have high trade flows with foreign countries. This is also consistent with what we have seen in the figures 4 and 5 in previous section. Third, marginal propensity to consumption (Cyd) is
comparatively large in Greece, Cyprus, Estonia, Slovakia and Slovenia. It may show that these country exhibit strong consumption propensities once its GDP increases.
Table1 Simple Regressions Results on
.Cyd Lr Ly Ir My
Greece
0.760621 -45.6756 0.060462910.745
0.375369P-VALUE
6.27E-05 0.000294 2.31E-12 1.26E-08 0.000186Italy
0.045136 -357.08 0.0446142310.32
0.677315P-VALUE
0.56401 9.14E-10 4.02E-10 2.34E-11 5.1E-07Spain
0.333202 -333.203 0.07125590.04
0.292788P-VALUE
0.000228 1.04E-07 1.65E-14 1.41E-14 0.00452Portugal
0.19139 -12.6566 0.03738292.673
0.603843P-VALUE
0.128069 0.12221 2.5E-05 4.77E-07 2.51E-05Ireland
0.22941 -65.4263 0.0334211714.03
0.287126P-VALUE
0.020841 0.23356 0.195125 1.87E-06 0.017306France
0.245478 -210.535 0.0513492007.71
0.412321P-VALUE
0.000319 4.32E-07 2.31E-18 0.013741 3.12E-07Germany
0.326226 -366.776 0.0383151931.21 0.938013
P-VALUE
0.00537 3.45E-05 1.4E-09 0.038941 8.25E-13Belgium
0.111423 18.4409 0.020069192.652 1.148747
P-VALUE
0.020284 0.086139 1.06E-06 0.239066 9.04E-08Austria
-0.05439 -43.2671 0.078566295.36
0.687598P-VALUE
0.482612 0.031854 6.15E-06 0.003305 1.77E-06Finland
0.514274 -38.7561 0.033283330.179
0.574516P-VALUE
0.002149 3.7E-08 5.78E-12 0.000202 3.36E-08‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
25
Cyd Lr Ly Ir My
Netherlands
0.20708 -64.5517 0.063625654.99 1.293333 P-VALUE
0.000742 0.02457 1.08E-10 0.011595 1.71E-11Cyprus
0.760621 -45.6756 0.060462910.745
0.375369P-VALUE
4.93E-05 0.022857 2.91E-10 0.402154 3.91E-05Estonia
0.756925 -0.36829 0.01366781.4458
0.751692P-VALUE
2.91E-11 0.58406 2.68E-05 0.011023 1.12E-05Luxemburg
0.336994 139.983 0.00541471.3167 1.573355 P-VALUE
1.87E-07 1.03E-11 0.55291 0.336454 2.56E-12Melta
0.499736 0.51036 0.0266159.41366 1.087049 P-VALUE
6.28E-05 0.332697 1.9E-05 0.030197 3.08E-08Slovakia
0.773039 -8.90026 0.012423145.663
0.880222P-VALUE
4.48E-11 0.000101 4.23E-05 0.027833 1.11E-07Slovenia
0.982028 -2.39844 0.009719135.124
0.867106P-VALUE
2.15E-05 7.57E-09 2.37E-14 0.000178 1.54E-075.3 The Result
In this section, we will give descriptive analysis of Interest- Rate-Change-to-Income Effect, identifying the trend and variation across the 17 countries. In the following section we will we discuss the common factors that influence Interest-
Rate-Change-to-Income Effect by constructing panel data.
We first discuss the sign of Interest- Rate-Change-to-Income Effect. According to equation (4.6), if Interest- Rate-Change-to-Income Effect is positive, it means the pro-cyclical capital requirement is more appropriate while if the effect is negative, counter-cyclical capital requirement strategies would be more suitable.13 There are few arguments we want to make here. First of all, during 2005Q1 to 2011Q3 Interest- Rate-Change-to-Income Effect exhibits both positive and negative value, though most countries are dominated by positive sign. This indicates that capital requirements should adjust according to each country’s general economic situation at different point
13 More detailed results are presented in Appendix.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
26
of time. Second, dominated positive Interest- Rate-Change-to-Income Effect for most countries, except Belgium and Malta, means that pro-cyclical capital requirement policy should be adopted. When GDP increases, the government should tighten its capital regulation while in economic recessions, the government is supposed to loosen capital requirement. Third, we can see that all countries, except the small country such as Cyprus and Malta, in 2011Q3 have positive Interest- Rate-Change-to-Income Effect. It indicates that during the time of ongoing European debt crisis, the austerity policy might not be optimal in the times of economic downturn. Deeply in-debt countries, especially Greece, Spain, Ireland, which have all positive Interest- Rate-Change-to-Income Effect during our research period, need stimulus expansionary policy to help the economy recover.
Next we turn to the trend and variation of Interest- Rate-Change-to-Income Effect.
Figure 1 and Figure 2 shows the trend and variation of interest rate change to income effect across 11 main Euro Zone countries. We focus on major countries and compare the movement. Among the major 11 Euro Zone countries, Italy, Portugal, France and Germany exhibit very large variation of Interest- Rate-Change-to-Income Effect. In some point, it may indicate a unit change in interest rate, which results from fiscal policy or monetary policy, will cause income to shift by a large amount in these countries. It also indicate that capital regulation policy in these countries should be more flexible, switching between pro-cyclical and counter-cyclical.
We also observe from the figures that in the time near 2008, most countries have larger positive value of Interest- Rate-Change-to-Income Effect relative to other period. When Interest- Rate-Change-to-Income Effect has higher values, it means effects on capital requirement from a unit change in income will be smaller.14 In other
14 We can refer this from equation (4.4). Since h’ is in the denominator,
will decrease.
‧
words, if GDP decreases that quarter, which happens in 2008, the government should adopt loose capital regulation policies but has an upper limit.15
Overall, Interest- Rate-Change-to-Income Effect exhibits large variation within each country as well as across different countries.
Figure1 Trend of Interest- Rate-Change-to-Income Effect for PIIGS
Figure2 Trend of Interest- Rate-Change-to-Income Effect for Relatively Small Variation
15 Under the assumption of counter-cyclical capital requirement, different values of
Interest-Rate-Change-to-Income-Effect will have opposite effects on the change of capital requirement.
Assume GDP all change the same unit, as Interest-Rate-Change-to-Income-Effect increases,
will decreases, and capital requirement will have lower adjustment compared with the situation when Interest-Rate-Change-to-Income-Effect decreases. In other words, as
Interest-Rate-Change-to-Income-Effect increases, one unit increase in GDP requires lower decrease in capital requirement compared with the situation when Interest-Rate-Change-to-Income-Effect increases.
2005 2006 2007 2008 2009 2010 2011
H4G_SPAIN
2005 2006 2007 2008 2009 2010 2011
H4T_GREECE
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
28
Figure3 Trend of Interest- Rate-Change-to-Income Effect for Relatively large Variation
-80000 -40000 0 40000 80000 120000
2005 2006 2007 2008 2009 2010 2011
H4T_ITALY H4T_PORTUGAL
H4T_FRANCE H4T_GERMANY
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
29
6 Panel Data Estimation
6.1 The Unit Root Test
To find whether there are common factors in each countries influencing Interest- Rate-Change-to-Income Effect, we construct panel data estimation. The estimation equation goes as follows:
Interest- Rate-Change-to-Income Effect= Intercept+ Monetary Policy Effectiveness+ Fiscal Policy Effectiveness+ Trade Openness+ Fiscal Policy Weight + Debt Flow (6.1)
All the variables are stated before, except for debt flow. Instead of cumulative debt level, we use debt increase in every quarter as variable to see whether the debt increase level across countries have common influence on Interest-
Rate-Change-to-Income Effect.
Before constructing the panel data, we should perform unit root test. Table 1 shows the statistic and p-value of Argumented Dickey Fuller test, Phillip-Pearson Test, and Breitung-T Test. The result shows that all variables, except monetary policy
effectiveness and fiscal policy effectiveness, are stationary. After first differencing Monetary Policy Effectiveness and Fiscal Policy Effectiveness, the two variables become stationary. In the following sections, our result will be presented in both estimation before first differencing and estimation after first differencing.
‧
Table2 Unit Root Test Result for variables in (6.1)
The Unit Root TestSample: from 2005Q2 to 2011Q3
ADF PP Breitung t-stat
Statistic P-Value Statistic P-Value Statistic P-Value
Interest-rate-change-to
-income effect16 240.901 0.0000* 269.874 0.0000* -5.18864 0.0000*
Monetary Policy
Effectiveness17 22.6953 0.9303 12.7478 0.9997 -2.37752 0.0087*
Fiscal Policy
Effectiveness18 24.2809 0.8909 12.764 0.9996 -2.62222 0.0044*
Trade openness19 65.3137 0.0010* 47.274 0.0647 -3.29405 0.0005*
Fiscal policy weight20 231.007 0.0000* 278.715 0.0000* -8.23047 0.0000*
Debt flow21 212.311 0.0000* 278.154 0.0000* -6.76618 0.0000*
6.2 Panel Data Analysis
Table 2 is the original panel data result of (6.1) before we take care of the unit root issues. Because we have identified Fiscal and Monetary Policy Effectiveness might not be stationary variables, we present first differencing result in table 3. Therefore, our interpretations below follow the first difference adjustment result in Table 3.
We construct our estimation by splitting countries into three groups. First, we pool all 17 countries together, with 398 samples. According to our previous results, most countries have positive Interest- Rate-Change-to-Income Effect. Under the
w16 As stated in (4.6), Interest-Rate-Change-to-Income Effect is defined as how much unit of income (Y) will be affected by a unit change in interest rate (Rf).
17 Monetary policy effectiveness is the elasticity of IS, calculated as
19 Trade openness is the proportion of the sum of import and export to GDP.
20 Fiscal policy weight is calculated as
.
21 Debt flow is the proportion of additional increase of debt in the calculated quarter to GDP.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
31
assumption of procyclicality capital requirement, the significance of trade flow indicates that countries with high trade flow relative to its GDP are tend to require larger adjustment to its capital requirement rule. To think intuitively, high trade-flow countries have higher dependency on other countries, thus the economic situation is more relative to others, especially in crisis. In order to prevent the spillover effects, more flexible capital requirement is needed22. In the second column, we have excluded six small countries which we consider extreme values, including Cyprus, Estonia, Luxemburg, Malta, Slovakia, and Slovenia. The results are similar with the previous panel, indicating that extreme values in the 6 small countries do not affect the overall trend of our data. Finally, we pool only the 5 in-debt countries, including Greece, Ireland, Spain, Portugal and Italy. The significance of expansionary policy weight and effectiveness shows under pro-cyclical capital requirement, the more weight and more effective fiscal policy is, the less capital requirement adjustment needed. In some point, this can be linked to the fact that countries with more
government expenditure should loosen capital requirements during recession but has an upper limit, i.e. the degree of loosen capital regulation should not be out of control, preventing these countries to go into further break down. Besides, though not obvious in other countries, debt flow is significant when we pool the 5 countries together. The result indicates the higher the debt level, the higher level of capital requirement adjustment is needed. Last but not the least, the trade openness is not significant in these 5 countries. We interpret it as though trade openness play a part in determining capital requirement policy, other indicators, such as debt increase level in PIIGS countries have dominated trade openness. Therefore, for PIIGS, the capital requirement policy should be adjusted depending on the fiscal policy weight and
22 For more detailed explanation to the relation between Interest-Rate-Change-to-Income-Effect, see footnote 15.
‧
effectiveness, debt increase level in each quarter.
Table3 Panel Estimation Result for Variables in (6.1) Before First differencing
Dependant Variable:Interest-Rate-Change- to-Income Effect Sample: from 2005Q2 to
2011Q3
All 17 countries excluding 6 small countries
5 in-debt countries
Coefficient P-Value Coefficient P-Value Coefficient P-Value
Monetary Policy Effectiveness
-235984.5 0.1572 -84754.8 0.0091* -98499.8 0.0000*
Fiscal Policy Effectiveness 1902.619 0.0094* 90630.28 0.0060* 120275.7 0.0000*
Trade Openness -52.16518 0.0001* -61.3473 0.0028* -51.743 0.0031*
Fiscal Policy Weight -2637.831 0.2653 -2267.12 0.4789 5292.276 0.0353*
Debt Flow -7194.699 0.2856 -7623.28 0.3542 -8538.43 0.1958
Samples 398 276 128
R-square 0.046948 0.061236 0.253382
Adjusted R-square 0.037293 0.043851 0.22271
F-Statistic Probabilities 0.001281 0.004214 0.000001
Table4 Panel Estimation Result for Variables in (6.1) After First differencing
Dependant Variable:Interest-Rate-Change- to-Income Effect Sample: from 2005Q2 to
2011Q3
All 17 countries excluding 6 small countries
5 in-debt countries
Coefficient P-Value Coefficient P-Value Coefficient P-Value
Monetary Policy Effectiveness
24143.14 0.6974 -154187.3 0.2980 -215478.2 0.0932
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
33
Dependant Variable:
Interest-Rate-Change- to-Income Effect Sample: from 2005Q2 to
2011Q3
All 17 countries excluding 6 small countries
5 in-debt countries
Coefficient P-Value Coefficient P-Value Coefficient P-Value
Fiscal Policy Effectiveness 2458.784 0.4181 163814.3 0.1781 223308.8 0.0620*
Trade Openness -18.2393 0.0550* -43.06878 0.0129* -17.19854 0.2970 Debt Flow -6341.946 0.3060 -7000.504 0.3278 -15316.68 0.0308*
Samples 398 276 128
R-square 0.046948 0.061236 0.126240
Adjusted R-square 0.037293 0.043851 0.0889
F-Statistic Probabilities 0.001281 0.004214 0.000001
6.3 Forecasting and Future Policy Suggestion
In this section, we want to use the estimation result in previous section to forecast Interest- Rate-Change-to-Income Effect in 2012. We present the forecast result before first differencing in figure 4 and after first differencing in figure 5. Our result shows that most countries exhibit positive Interest- Rate-Change-to-Income Effect. After dealing with unit root by first differencing, Interest- Rate-Change-to-Income Effects in all the countries are positive. This indicates that in 2012, where the European Sovereign debt crisis deepens, counter-cyclical capital requirement should be adopted.
Countries with decreasing GDP need loose capital requirement to inject revival to their economy. This is especially true in PIIGS countries. According to our previous analysis, the higher the value of Interest- Rate-Change-to-Income Effects, the less capital requirement adjustment needed to make. Observing from figure 4 and figure 5,
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
34
we find that among PIIGS, Greece, Italy, Spain and Portugal all have relative high value of Interest- Rate-Change-to-Income Effect. We conclude that instead of austerity measure, the authority should adopt counter-cyclical capital requirement.
When recessions occur, the authority should loosen capital requirement but still holds an upper-limit to prevent further break down. This also applies to other major Euro Zone countries, such as Germany and France. However, for Ireland the result is different. It still shows Ireland need to loosen capital requirement in 2012 since it has encountered economic recession, and the lower value of Interest-Rate-to-Income Effect means Ireland should further decrease capital requirement since its GDP has declined in 2012.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
35
Figure4 Forecast of 2012 Interest-Rate-Change-to-Income-Effect Before First differencing
Figure5 Forecast of 2012 Interest-Rate-Change-to-Income-Effect After First differencing
Ireland Slovakia Malta Estonia Luxembo
urg Slovenia Belgium Netherla
nds Austria Germany Finland Spain Greece Italy Portugal France h' -3773 -1066.5 -729.57 -2.9036 158.961 197.317 740.242 976.898 3013.54 3976.04 4445.31 4528.73 4779.16 5629.5 5906.27 6582.82 -6000
-4000 -2000 0 2000 4000 6000 8000
Forecast 2012 Interest-Rate-to-Income-Effect(1)
Malta Slovakia Ireland Estonia Netherla
nds Belgium Slovenia Austria Germany Finland Spain Italy France Greece Portugal Luxembo urg H' 1528.17 1880 1988.25 2088.41 2384.83 2505.36 2614.67 3037.4 3515.5 3678.23 4061.09 4070.53 4355.53 4378.56 4789.91 5721.76 0
1000 2000 3000 4000 5000 6000 7000
Forecast 2012 Interest-Rate-to-Income-Effect(2)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
36
7 Conclusion
In this paper, we followed the framework of Len-Kuo Hu (2012) to derive capital requirement strategies. Depending on interaction with liquidity shock, the loan’s project’s return, and the moral hazard problem, we derive the optimal capital requirement function, which determines on general economic condition in each country. Our model suggests that counter-cyclical capital requirement should be adopted when a country’s GDP is negatively related with its interest rate. On the contrary, the government should implement a pro-cyclical capital requirement policy when a country’s GDP is positively related with its interest rate. With Keynesian's IS-LM framework, we are able to further specify how to identify whether capital requirement should increase or decrease as GDP changes.
We use the result of our model to test the optimal capital requirement strategies under the theme of ongoing European debt crisis. Our results can be summarized as following. First, among the 17 countries across Euro Zone, most countries exhibit positive relation between GDP and interest rate during 2005Q2 to 2011Q3, indicating counter-cyclical capital requirement strategies should be adopted. Besides, in 2011, GDP in Greece, Ireland, Italy, Spain and Portugal are declining, meaning that these countries need loose capital requirement to foster its economic demands. The austerity measure will only worsen the already weak economic condition. Third, by
constructing panel data, the result indicates that under the counter-cyclical capital requirement strategies, countries involved with higher trade are required to adopt more flexible capital requirement policy. When we only estimate PIIGS countries, the result shows that debt increase, fiscal policy weight and effectiveness dominate Interest- Rate-Change-to-Income Effects. Finally, when we use our data and equation
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
37
estimation result to forecast Interest- Rate-Change-to-Income Effects in 2012, we find all 17 Euro Zone countries should adopt counter-cyclical capital requirement policies.
In addition, Greece, Italy, Portugal, France, and Spain are the countries needed to have an upper-limit when loosen capital requirement is adopted.
‧
Appendix A: Detail Definition of Variables
Variables Definition
C Household final consumption
expenditure
Consists of expenditure incurred by residents household on individual consumption of goods and services, household payment to the product provided by the government. It also includes various kinds of imputed expenditure of which the imputed rent for services of owner-occupied housing (imputed rents). Household final consumption includes household expenditure made on the domestic territory by residents and inbound tourists, but excludes residents' expenditure made abroad.
I Gross Capital Formation
(GFCF)
Gross Capital Formation consists of outlays on addition to the fixed assets of an economy and net changes in
inventories. Fixed asset includes land improvements, plant, machinery, equipment purchase, construction of the roads, railways, schools, offices, hospitals, private residential dwellings, and commercial and industrial buildings.
Inventories are stocks of goods held by firms to meet
temporary or unexpected fluctuations in production or sales.
Net acquisitions of valuables are also considered capital formation.
G1 Government final consumption
expenditure (GFCE)
Government expenditure on goods and services that are used for the direct satisfaction of individual needs
(individual consumption) or collective needs of members of the community (collective consumption).
It mainly consists of the current expenditure, which is the sum of service and goods purchased by government, social payments and compensation to government employees.
G2 Total expenditure The sum of current expenditure and capital expenditure.
1. Current expenditure
The sum of compensation to government employees, the service and goods purchased by government, interests, social payments (social benefits and pensions paid in
‧
money, and service funded by government that are
produced and delivered to household by market units), and subsidies.
2. Capital expenditure
The sum of government investment which deals with the acquisition of fixed capital asset, capital transfers, acquisition of stocks, valuable lands less the disposal of such asset.
T Tax Sum of current revenue and capital revenue.
1. Current revenue
The sum of direct taxes payable by household and cooperation, indirect taxes which received by EU institutions, social contributions of employers and
The sum of direct taxes payable by household and cooperation, indirect taxes which received by EU institutions, social contributions of employers and