Link between Household Debt and Savings

Many analysts and business executives are becoming apprehensive with the recent rises in the consumer debt burden, defined as the level of consumer debt relative to ability to repay which may predict an economic growth slowdown.

A higher debt reduces the credit worthiness of households who would then experience financial anguish caused by unfavourable economic shock, such as the loss of a job or large uninsured medical expenses. In the event of this situation, they would be less disposed to spend on consumer goods, particularly big ticket items such as automobiles and home computers. Consequently, the reduction in consumer spending would hurt economic growth as firms cut back on the production of consumer goods and laid off workers.

Households have spent in excess of income, in part because increased house prices have led to increased household wealth. The rise in house prices reflects an adjustment to sustained low inflation and interest rates, among other factors. However, activity in the housing market cannot be sustained at the pace seen in recent years. As the housing market cools, growth in consumer spending should ease and household saving rise, resulting in a tendency for the current account deficit to fall, everything else equal. The increase in household debt also partly reflects the removal of government controls of the financial system over the past two decades.

Based on the results of empirical works of many authors, most studies favour the hypothesis that the causality is from economic growth rate to growth rate of savings. Based on the empirical results, the main conclusion of this study is that income class of a country does play an important role in determining the direction of causality.

A rising consumer debt burden also might predict future activities in broad methods of economic activity, such as real gross domestic product. A decline in consumer spending on durable goods would lower real GDP growth because such spending is a large constituent of real GDP.

1.2 Objectives and Organisation of the dissertation

Many tests have been carried out by many authors throughout the world to see if there is a link between household debt, household savings and economic growth and hence, analyse its impact on the discussed variables. A panel cross country analysis has been carried out on 25 countries to determine how household savings and debt may act as a deterrent for economic growth. Chapter 2 reviews the literature and empirical evidence pertaining to the works of various authors concerning economic growth, household debt and household savings. The next chapter deals with the review of variables of interest to us, which will be used in the empirical testing part, hence, the household savings as a proportion of disposable income, household debt as a proportion of GDP per capita, growth rate of Real GDP per capita, consumption share of GDP per capita, price level of GDP, investment share of GDP per capita, interest on savings will be scrutinized in the chapter. In Chapter 4, the Haussman tests have been mostly used to predict the impact of these independent and exogenous variables on the dependent variable of economic growth. Finally in Chapter 5, we conclude on the subject and make some policy recommendation and alongside cite some limitations of the work carried out.

2.1 THEORETICAL LITERATURE

When there is a positive change in the level of production of a country’s goods and services over a certain point in time, it is referred to as economic growth. It is also influenced by many factors but one of the pinnacles of economic history is the impact household saving and debt has on economic growth. Most working papers and journal articles on cross countries studies assume a positive relationship between household saving and economic growth and an adverse relationship between consumer debt and economic growth.

The difference between a household’s disposable incomes (primarily wages obtained, proceeds of the self-employed and net property returns) and its consumption (spending on products) is known as household saving. When the household saving is divided by household disposable income, the household savings rate is computed. When a household uses more than it obtains as expected income and funds some of the spending through credit (growing debt), through returns coming from the sale of resources, or by making cash and deposits, there is usually a negative savings rate.

These discrepancies are fairly due to institutional distinctions between countries. These include the degree to which old-age pensions are financed by government rather than through personal savings, and the level to which governments offer insurance against sickness and unemployment. The age composition of the population is also significant, as the elderly tend to run down financial assets obtained during their working life. This implies that a country with an ageing population will generally have a low household saving rate.

The conformist view is that savings contribute to higher investment and hence higher GDP growth in the short run (Bacha, 1990; DeGregorio, 1992; Jappelli and Pagano,1994). The central idea of Lewis’s (1955) traditional development theory was that increasing savings would accelerate growth. Kaldor (1956) and Samuelson and Modigliani (1966) studied how different savings behaviors induced growth. On the other hand, many recent studies have concluded that economic growth contributes to savings (Sinha and Sinha, 1998; Salz, 1999;Anoruo and Ahmad, 2001).

Over the last 10-15 years, household saving rates have increased in Austria, Germany and Sweden and remained stable in Belgium, France and Switzerland. A downward trend over the same period has occurred in Canada, Italy, Japan, Korea, Poland and the United States. (OECD (2010), National Accounts of OECD Countries, OECD, Paris)

The main factors contributing to differences among countries are listed below:

The income effect: in general higher income leads to a higher saving rate;

The wealth effect: profits or losses on financial and non-financial assets and liabilities affect built up wealth, and thus probably expenditure, but not on income. Higher wealth may then lower the saving rate;

Credit facilities: in countries (e.g. UK and US) where consumption credit was easier to finance, saving rates may be comparatively lower;

Institutional factors such as differences in social security schemes, especially pension schemes and the tax system;

The proportion of own-account entrepreneurs and small unincorporated enterprises, within the household sector, because producers may have a different saving behaviour;

Households’ expectations as regards the future economic situation;

Cultural and social factors.

Hondroyiannis (2004) analyses the long term and short term causal factors of aggregate private savings in Greece using data for the time frame of 1961-2000. By considering the financial and demographic advances during this phase, the long run savings utility which is susceptible to real interest rate, public funds, liquidity, old dependency ratio and fertility changes, is approximated on the foundation of an absolute life-cycle hypothesis. The significance of short-run divergences is obtained using vector error-correction model estimation. The empirical evidence proposes the continuation of a stable long-run savings function in Greece both in the long- and short-run periods and the policy inferences of such an association are accessible.

According to Barba and Pivetti (2008), rising household debt in USA made low wages and increasing aggregate demand to arise simultaneously. In the USA, according to the figures of the Federal Reserve Board, consumer credit outstanding reached 25% of disposable personal income (DPI) in 2006. This was the peak of an upward trend that has characterised the period since the first half of the 1980s, following 15 years during which the consumer credit-income ratio averaged around 18%. Increasing household debt in developed countries like USA has been mostly due to the noticeable fall in household savings and this had an adverse effect on economic growth.

Salotti (2009) claims that the current account is inclined by changes in US private savings which aid to generate and maintain world imbalances. A panel of 18 developed countries for the time dimension of 1980-2005 is used to check this claim by examining the components of total household savings. They merge two lines of literature: the first line from consumer theory, bearing in mind particularly the `wealth effect’, the second line from aggregate private savings theory. Unit root and “cointegration” tests are performed to evaluate the most suited method for estimation of the long run savings function and to derive the “cointegrating” relationship. The group means FMOLS is exercised to approximate the model. The empirical evidence goes in line with the theory where a rise in wealth should adversely affect the household savings. In addition, when significant descriptive variables, such as national savings and populace dependence ratios, are incorporated in the model, material wealth becomes the only type of wealth to (inadequately and negatively) control household savings in developed countries.

Howitt, Agnion, Comin and Tecu (2009) wanted to test if a country can grow more rapidly by saving further as they believe that household saving is of deep concern as it allows entrepreneurs to undertake their business and also reducing the agency cost that usually acts a hindrance for foreign investors. Since domestic saving counts for improvement, and consequently growth, it thus allows the home industrialist to put equity into this joint enterprise, which reduces an organization setback that would else discourage the foreign shareholder from contributing. In rich countries, domestic entrepreneurs are already known with limit know-how and consequently do not need to draw foreign outlay for investment, so domestic saving is not important for growth. The higher the household savings and the lower the household debt a country has, the more economic growth it can at least forecast to make. The finding is based on a cross-country non-overlapping panel over the period from 1960 to 2000. They use a sample of 118 countries, all those for which there exists data on per-worker GDP and on the saving rate. The cross-country regression shows that lagged savings is positively related with productivity growth in poor countries but not in rich countries.

2.0 EMPIRICAL EVIDENCE

Empirical evidence deals mainly with the previous works of various authors all around the world. There have been many works carried out by different authors and they reached certain conclusions which may be further developed and their results vary among the countries. The first case considered is on the United States of America (USA) and then they further scrutinise what happened in the developed and emerging countries.

2.1 STUDIES ON THE USA

As noted in Thomas and Towe (1996), research into household saving/consumption behaviour in recent years has inclined to centre on probing for long-run relationships between saving (or consumption) and selected macroeconomic variables. In large part, this shows the fact that the data involved have been found to be non-stationary. This implies that conventional statistical methods cannot be used to test relationships between movements in the savings rate and other (non stationary) macro variables. This approach also implies that short-run movements in the savings rate may be driven by deviations from the long-run relationship between saving and its fundamental determinants.

Callen and Thimann (1997) studied the empirical determinants of household saving in USA using cross sectional and panel data from 21 OECD countries for 1975-95.) They find that household saving fell from 13% during 1975-81 to only 11% in 1982-89 but it has then stayed stable in general. Variables that capture the structure of the tax system and the financing and generosity of the social security and welfare system are added to the set of potential explanatory variables. The results indicate that there is an central role for public and corporate saving, growth, and demographics in controlling household saving, while some role is also established for inflation, unemployment, the real interest rate, and financial deregulation. The results also propose that the tax and the social security and welfare systems have an important impact on household saving.

Bérubé and Côté (2000) examine the structural factors of the household savings rate in Canada over the previous 30 years, using co integration techniques. The main result is that the real interest rate, expected in¬‚ation, the ratio of the all-government ¬scal balances to nominal GDP, and the ratio of household net worth to personal disposable income are the most significant causal factors of the trend in the personal savings rate, as calculated in the National Income and Expenditure Accounts (NIEA). The outcomes also recommend that the fast fall in the NIEA personal savings rate in current years mainly shows a change in the trend constituent of the savings rate, rather than a temporary different approach from the trend.

Tipett (2010) uses many methodological approaches and draws on “longitudinal data from the National Longitudinal Survey of Youth 1979 and also uses multilevel logistic regressions to investigate the relationship between the hypothesized mechanisms and the probability of holding non-collateralized debt. Analysis of Survey of Consumer Finance data shows that the amount of household debt increased faster than household asset increases (see also Bucks, Kennickell, Moore, Fries, and Neal 2006; Kennickell 2009), and Keister (2000) shows that overall wealth has been growing at the same time that the percentage of households with zero or negative net worth has also been rising.

2.2 STUDIES ON DEVELOPED ECONOMIES

Carroll and Weil (1994) present “Granger”-“causality” tests for 38 countries for which they have fine data, and show that increases in growth radically head increases in saving. Dekle (1993) presents comparable “Granger” “causality” regressions for a group of fast-growing countries and finds that growth positively “Granger”-causes saving in every country in his sample.

Edwards (1995) looked at data from a panel of 36 countries over the period 1970-92. Using lagged population growth, openness, political instability, and other lagged variables as instruments, he concludes that the rate of output growth has an important, positive effect on saving.

Andersson (1999) believes that the worldly interdependence between saving and output has been measured in recent empirical studies which obliged some authors to question the conventional idea of a “causal” chain where saving precedes growth via capital accumulation. As divergent to the previous studies, which have mostly used panel-estimation processes, the tests of “causal” chains are performed in time-series sets. Saving and GDP are approximated in” bivariate vector autoregressive or vector error-correction models” for Sweden, UK, and USA, and tests of “Granger non-causality” are executed within the estimated systems. The core results shows that the “causal” chains linking saving and output vary across countries, and also that “causality” linked with amendments to long-run dealings might go in diverse directions than “causality” associated with short-term instabilities.

Jappelli and Padula (2007) reconsidered savings inclinations in Italy, summarizing existing empirical evidence on Italians’ motives to save, relying on macroeconomic indicators as well as on data drawn from the Bank of Italy’s Survey of Household Income and Wealth from 1984 to 2004. The macroeconomic data indicate that households’ saving has fallen considerably, although Italy continues to class above most other countries in terms of saving. The microeconomic data show a strong correlation between the propensity to save and the level of current income, as well as a strong correlation between income and indebtedness. International panel data put forward that saving is robustly linked with the growth rate of income, and that saving changes parallel growth change, as shown by Attanasio, Picci and Scorcu (2000) using the 150 countries of the World Bank Saving Database.

2.3 STUDIES ON EMERGING MARKETS

Emerging markets are economies which are currently in the process of fast growth and industrialisation. There are at present 28 emerging markets in the world with the economies of China and India being considered certainly as the two largest. New conditions were surfaced in recent years to portray the largest developing countries such as BRIC standing for Brazil, Russia, India, and China.

The relationship between savings and economic growth has received increased notice in recent years especially in developed and emerging economies [see Bacha (1990), DeGregorio (1992), Levine and Renelt (1992), and Jappelli and Pagano (1994)]. This might not be distinct to the central foundation of Lewis’s (1955) traditional development theory that increasing savings would accelerate economic growth. Research efforts by Kaldor (1956) and Samuelson and Modigliani (1966) examined how different savings behaviours would induce economic growth.

Caroll and Weil (1994) used five year averages of the economic growth rate and savings for OECD countries and found that economic growth “Granger” caused savings. However, the reverse was obtained when dummies were included in their estimation. Using “Granger” “causality” tests, findings by Sinha and Sinha (1998) and Sinha (1999) found that economic growth rate “Granger” caused the savings growth rate for Mexico and Sri Lanka respectively.

Jappelli, Tullio and Marco Pagano (1994) test whether the measures of liquidity

constraints help to explain the international differences in national saving rates, as forecasted by their model. They also test an outcome of that model, namely that the effect of growth on saving is greater where liquidity constraints are more determined. The data cover a panel of 19 countries (all the main OECD countries are included) and are drawn from Modigliani [1990]. Observations are averages of annual data for three periods: 1960-1970, 1971-1980, and 1981-1987). Findings show that the two variables are negatively linked (the correlation coefficient for the entire sample is -0.55). They have empirically measured the soundness of three propositions, namely that liquidity constraints on households raise the saving rate, strengthen the effect of growth on saving, and promote productivity growth in models in which growth is endogenous.

Using cross section data between 1960 and 1997 and “Granger” “causality” methodology, Anoruo and Ahmadi (2001) observed the causal relationships between the growth rate of domestic savings and economic growth for seven African countries -namely Congo, Cote d’Ivoire, Ghana, Kenya, Nigeria, South Africa and Zambia. Their studies established that savings are co-integrated in all of the countries except for Nigeria and that economic growth “Granger”-causes the growth rate of domestic savings for all the countries considered except Congo where reverse “causality” was obtained.

Matos (2002) used among other parameters, the ratio of residents’ funds deposited in the financial system to aggregate monetary asset M2 (1947-2000) as a proxy of financial development, empirical tests support the view that it is vital to maintain the public’s confidence in domestic financial assets to improve GDP growth prospects. This ratio may reflect an intangible asset of the financial intermediaries, i.e. the general public’s confidence that contracts between customers.

Kwack and Lee (2005) investigate the extent to which income growth and uncertainty and demographic factors affect the domestic real saving rate in Korea. They test an extended life cycle hypothesis and demography hypothesis with Korean time series data from 1975 to 2002. The results of the tests show that the aggregate saving rate is positively affected by the moving average of the growth rate of income and the variance of the income growth. The positive effect of the income growth differs from the negative effect found household survey data were used.

Adebiyi (2005) employed quarterly data spanning between 1970 and 1998 to examine savings and growth relationships in Nigeria using “Granger” “causality” tests and impulse response analysis and concluded that growth, using per capital income, is sensitive to, and has an inverse effect on savings.

Mohan (2008) believes that household savings in India has contributed significantly to its economic growth which recorded a steady rise over the last decades. Mohan found some empirical relations whereby in the argument that high levels of debt-GDP lead to high interest payments relative to GDP, which crowd out government capital expenditure and reduce the overall saving rate, two relationships are of critical importance: the responsiveness of changes in the saving ratio with respect to changes in the fiscal deficit levels; and the responsiveness of government capital expenditure to changes in the level of interest payments. Mohan (2006) experienced the path of “causality” between economic growth and savings in different economic income classes. The ADF test indicates that both log GDP and log GDS have unit roots in the level data. In the presence of unit roots, the variables need to be differenced in order for the series to be stationary. Without differencing the data, a “causality” test would lead to misspecification.

To examine the “direction of causality” between saving and economic growth in Nigeria during the time frame 1970-2007, Oladipo ( 2009) used the “Toda and Yamamoto (1995) and Dolado and Lutkepohl (1996) TYDL” methodology. The variables of interest for savings and economic growth are positively “co-integrated” indicating that there exists a steady long run equilibrium relationship. Furthermore, the findings also revealed a “unidirectional causality” between savings and economic growth and thus the corresponding role of FDI in growth.

In order to establish the link between economic growth and saving in Nigeria during the time frame of 1970-2007, Abu (2010) used the “Granger-causality and co-integration techniques”. There exists “co-integration” and “long-run equilibrium” between the variables savings and economic growth according to the “Johansen co-integration test”. There is also the “causality” runs from economic growth to saving, implying that growth triggers and “Granger” produces saving. Hence, the “Solow’s hypothesis” that saving leads to economic growth, and recognize the Keynesian theory that it is economic growth that leads to higher saving, is discarded.

CHAPTER 3-DATA ANALYSIS

3.1 Sources of data

The economic growth rate, household debt and household saving rate, price level are available on the Global Finance website. The interest on savings, consumption and investment are available on the Nationsmasters website, the World Bank website and the Penn World Table website.

3.2 The Econometric Model

In this section, a model is developed to measure the impact of household debt and household saving among other factors, on economic growth. The model for growth for country i in time t is as follows:

EGit= α +β1 HDit + β2 HSit + β3 Rit + β4 Pit + β5 Cit+ β6 Iit + Uit

Where

EGit= Growth Rate of Real GDP per capita at constant prices

HDit = Household Debt as a % of Gross Domestic Product (GDP)

HSit= Household Savings as a % of Disposable Income

Rit = Interest on Savings

Pit= Price Level of Gross Domestic Product (GDP)

Cit= Consumption Share of CGPD (GDP PER CAPITA)

Iit= Investment Share of CGDP (GDP PER CAPITA)

Uit = the disturbance term

3.3 Economic Growth

When per capita GDP or any other means of calculating total income rises, economic growth arises and this is usually registered as the yearly rate of change in GDP. Economic growth results from advances in productivity in terms of more production of goods and services with the same factors of production.

The dependent variable economic growth is measured by real GDP per capita. At times, total GDP figures are not reflective of the actual performance in the economy. Hence, GDP per capita is a better measure as it is liable to fewer errors and some errors tend to affect population estimates and thus they have offsetting impacts. Furthermore, the natural log of real GDP will be taken into account to avoid any large outliers.

Screen-shot-2009-09-01-at-14

3.4 Household Saving

Household saving can be defined as a percentage of household disposable income which is not consumed and household savings rate can be calculated on gross or net basis. Depreciation is considered in the net savings rate which is more commonly used compared to the gross savings rate.

Comparisons of savings rate among countries become hard by these two different measures of gross and net savings rate due to distinct social security and pension programmes, variable tax schemes which have an impact on disposable income. The household savings rate of a country can be affected by age of the economy’s population, the accessibility of credit, general wealth issues, cultural and social factors. Nevertheless, household savings rates are still a good a measure of an economy’s income in relation to consumption over time.

A country can finance its debt domestically if it has a relatively high level of household savings. High debts levels funded mostly by foreign creditors are less persistent than high debts levels financed by internal savings.

Consumption allows GDP to grow and this is a significant factor in economic expansion. With the existence of financial crisis, the whole economy could be dampened with lower consumption due to higher debt and lower savings level. A larger portion of GDP growth should then come from FDI, exports and government expenditure.

Household saving is the most essential domestic source of funds to back capital outlay and this is a substantial boost for economic growth on the long term basis. Household savings rate vary greatly among countries as shown in the chart. This is partially due to the level pensions schemes are financed by government rather through personal saving and also to the extent governments offer insurance against sickness and unemployment.

savings01

Considering the time dimension in the table above, the savings rate were relatively steady or somehow rising mildly in France, Austria, Italy, Norway and Portugal but have been decreasing in United States, Canada, Japan and Australia. If the social security and insurance payments of USA are considered, its savings rate would be striking.

3.5 Household Debt

When a country has a substantial degree of household debt, it increases its inclination to financial crisis and this acts as a hindrance for economic growth. There have been forecasts about house bubbles which were caused and thus creating the countries to be overheated. A large portion of the economic growth was centred on household consumption which was backed by loans from banks.

When banks noticed the lack of credit worthiness from consumers who even lost their confidence in the financial system, there had been strict controls over the lending conditions for loans. As a result, the ongoing vicious circle preceded a major decline in economic growth following the fall in consumption and repayments of debts.

Analysing the graph results with the conclusion that USA is not the only main country having experienced the worst GDP slowdown but many other countries like Iceland and Portugal are following suit with the level of household debt actually rising substantially. It would not be logical for a country burdened by a large level of household debt to expect its economic performance to flourish in the coming years.

HouseholdDebtSelectedCountries

household-debt-vs-savings

Source:  Lew Rockwell

3.6 Rates of interest

The rate of interest has a great influence on the given level of aggregate disposable income which is divided between consumption and saving. However, it cannot be predicted with conviction that a lower interest rate would imply more disposable income will be dedicated to consumption and less to saving or vice versa.

As a matter of fact, there can be a rise or fall in the total amount saved following a change in interest rate and this depends on the income and substitution effects and their strengths of their net effects. A higher level of future consumption arises at the detriment of present consumption with substitution effects due to higher interest rates and thus resulting in more savings in the present period.

On the other hand, a consumer’s future income compared to his present income can be increased following higher interest rate and this leads to higher consumption by borrowing from future income and hence, less is saved. However, this may not be necessarily the case for lower income earners who would save only a small part of their incomes even when interest rates are high. The substitution effect will then outweigh the income effect and there will be a direct link between income and rate of interest. For some people who prefer to save a greater portion of their incomes, the income effect may offset the substitution effect and thus higher interest rates would result in lower present savings level

real-interest-rates

3.7 Price level/Inflation

One of the theoretical concepts of economics says that when there is a change in the price level, this may affect consumption and savings positively or negatively. It is usually believed that household’s confidence in money erodes when there is inflation and hence, they have the tendency to save more since inflation actually raises the variance of expected real income. The fact that consumers have greater preference for unplanned increases in savings compared to withdrawals, it usually incites consumers to save more when inflation is high.

There is also an indirect effect of inflation whereby the real value of nominal asset is diminished and thus the real value of liquid assets decreases the net household wealth. Real consumption is often reduced and savings rate increases.

080625_global_inflation (1)

3.8 Consumption

The total value of goods and services purchased by people aggregated over time is called consumption and it is usually the greatest GDP component. A country’s economic performance is often assessed on its consumption levels. Different income earners would be consuming differently depending on their standard of living and purchasing power. Consumption is usually determined by current income, accumulated savings and expectations on future income.

Consumption and consumer debt trends

3.9 Investment

When an owner usually acquired property for the purpose of generating income like plants and equipments, this is called investment as it is spending on income-generating assets.

If a country wants to achieve long term sustainable economic growth, it should be able to the rates of accumulation of capital – be it human or physical so that it can result in more efficient assets and so that the whole population can have access to those assets.

With the help of financial instruments, markets, and institutions, the extent to which information, enforcement and transactions costs can have their impact on savings rates, investment decisions, technological innovations and steady-state growth rates can be improved.

Average annual investment growth in the first six quarters of recovery

Source: National Bureau of Economic Research; National Income and Product Accounts (NIPA) from the Bureau of Economic Analysis.

.

Investment as a share of GDP

Personal debt and disposable income

Chapter 4:

Methodology and

Empirical Analysis

4.1: Introduction

In this chapter, the empirical results will be presented and analyzed. However, before doing so, the objective of this dissertation will be re-emphasized. The impact of household debt and household savings on economic growth would be analysed. Section 4.2 deals with the sample of 25 countries selected for the period of 1995 to 2004. There is also the model specification in the Section 4.3. The section 4.4 is dedicated to panel data explanations. Finally, Section 4.5 and 4.6 deal with the interpretation of the empirical results.

4.2 Sample of countries selected

Table 4.2.1 below summarizes the list of countries selected for the testing purpose.

Data availability has put severe restrictions on the number of countries. Ideally many more countries should be taken into consideration to avoid bias; however restrictions on figures reduced the sample size to 25. Furthermore, once again following restriction on data, the testing period under review has been for only ten years starting year 1995 to year 2004.

Table 4.2.1: List of Countries:

Australia

6. Denmark

11. Greece

16. Netherlands

21. Spain

Austria

7. Estonia

12. Hungary

17. Norway

22. Sweden

Belgium

8. Finland

13. Italy

18. Poland

23. Switzerland

Canada

9. France

14. Japan

19. Portugal

24. United Kingdom

Chile

10. Germany

15. Korea

20. Slovak Republic

25. United States of America

Source: Author’s computation

4.3 MODEL SPECIFICATION

In this section, a model is developed to measure the impact of household debt and household saving among other factors, on economic growth. The model for growth for country i in time t is as follows:

EGit= α +β1 HDit + β2 HSit + β3 Rit + β4 Pit + β5 Cit+ β6 Iit + Uit

Where

EGit= Growth Rate of Real GDP per capita at constant prices

HDit = Household Debt as a % of Gross Domestic Product (GDP)

HSit= Household Savings as a % of Disposable Income

Rit = Interest on Savings

Pit= Price Level of Gross Domestic Product (GDP)

Cit= Consumption Share of CGPD (GDP PER CAPITA)

Iit= Investment Share of CGDP (GDP PER CAPITA)

Uit = the disturbance term

4.4: Panel Data

Panel data, also called longitudinal data or cross-sectional time series data, are data where multiple cases (people, firms, countries etc) were observed at two or more time periods.

There are two kinds of information in cross-sectional time-series data: the cross-sectional information reflected in the differences between subjects, and the time-series or within subject information reflected in the changes within subjects over time. Panel data regression techniques allow us to take advantage of these different types of information.

While it is possible to use ordinary multiple regression techniques on panel data, they may not be optimal. The estimates of coefficients derived from regression may be subject to omitted variable bias – a problem that arises when there is some unknown variable or variables that cannot be controlled for that affect the dependent variable. With panel data, it is possible to control for some types of omitted variables even without observing them, by observing changes in the dependent variable over time. This controls for omitted variables that differ between cases but are constant over time. It is also possible to use panel data to control for omitted variables that vary over time but are constant between cases.

4.4.1: Fixed v/s Random Effects model

Fixed-effects (FE) explore the relationship between predictor and outcome variables within an entity (country, person, company, etc.). Each entity has its own individual characteristics that may or may not influence the predictor variables. When using Fixed Effect, we assume that something within the individual may impact or bias the predictor or outcome variables and we need to control for this. This is the rationale behind the assumption of the correlation between entity’s error term and predictor variables. FE removes the effect of those time-invariant characteristics from the predictor variables so we can assess the predictors’ net effect

The general form of a panel regression is as follows:

Yit = αi + λXit + ε t (i)

Where Yit is the dependent variable of a particular country I at time t

Xit is a matrix of explanatory variables of country I at time t, αi is the intercept term of country i

Because countries are likely to vary in several respects, individual specific effects have to be allowed. This is handled by the fixed effect model, which is one method of dealing with panel data set. Since countries are likely to vary, country specific effects imply that

αi = α + μi (ii)

Where μi is the cross-sectional component of the error term and a is a constant

Replacing (ii) in (i), we obtain

Yi = α + λXit + μi + ε it (iii)

The fixed effect model assumes that μi, the error term, will be fixed in repeated sampling. In other words, each country will have a specific μi.

On the other hand, the random effect model is an alternative way for dealing with panel data sets. In the random effects model, there are no specific effects. This implies that the error term μi and the explanatory variable Xit will not be correlated, that is Xit and μi are independent.

4.4.2: Hausman Specification Test

In order to determine whether the fixed or random effects model is appropriate for our data set, a Hausman test is to be carried out. The test evaluates the significance of an estimator versus an alternative estimator. It helps one evaluate if a statistical model corresponds to the data and testing for correlation between the regressors and the error term.

If the linear model y = bX + e, where y is univariate and X is vector of regressors, b is a vector of coefficients and e is the error term. We have two estimators for b: b0 and b1.

Under the null hypothesis, both of these estimators are consistent, but b1 is more efficient

(has smaller asymptotic variance) than b0. Under the alternative hypothesis, one or both of these estimators is inconsistent.

We can derive the statistic: where T is the number of observations. This statistic has chi-square distribution with k

(Length of b) degrees of freedom.

If we reject the null hypothesis, one or both of the estimators is inconsistent.

The hypothesis under test is

H0 : λ = 0(RE specification is acceptable)

H1: λ ≠0 (RE specification is invalid: FE should be used)

If an insignificant P-value is obtained (P> χ2 greater than 0.05) random effects must be used. However, if a significant P-value is obtained, fixed effect must be used.

Nevertheless, the Hausman test often leads to negative test statistics caused by estimated parameter variance differences that are not positive semi-definite (not PSD). In such cases, the absolute value of the statistic must be used which leaves the test statistic asymptotically unchanged under H0. Moreover, finding a non-PSD parameter variance difference with a negative test statistic should not be interpreted as evidence in favor of H0.

4.5 Empirical Estimates

In this section, the empirical results will be presented for the four regressions. Resultswill be that of the fixed effect, random effect and the Hausman test.

4.5.1 Summary Statistic

Table 4.5.1.1 represents the summary statistics for the variables of 25 selected countries for the period 1995 to 2004.

Table 4.5.1.1

Variable

Mean

Std Deviation

Min

Max

GDP

2.63052

2.100717

-9.02

9.87

HS

7.198

5.845731

-11.7

23.2

HD

55.056

20.84261

15

151

C

57.22436

6.406815

42.05

73.7

I

23.3944

4.152051

13.09

42.71

P

94.79224

32.73583

33.2

193.8

R

4.669957

4.274243

0.0355833

26.78333

Source: Author’s Computation

4.5.2: Results for regression (1)

Table 4.5.2.1 presents the results of the fixed and random effect models while Table

4.5.2.2 reports the Hausman test results. The tables are presented below.

Table 4.5.1.1: Fixed and Random effect results for regression (1)

Dependant variable: GDP

Fixed Effects

Random Effects

HS

-0.1860329

(-3.57)

-0.1251016

(-5.19)

HD

-0.336063

(-3.09)

-0.0276192

(-3.94)

C

-0.2195534

(-2.45)

-0.0184041

(-0.82)

I

0.3645943

(5.94)

0.117618

(3.49)

P

0.0078807

(0.86)

-0.0136453

(-2.74)

R

0.0667397

(1.41)

0.0721238

(2.03)

Constant

8.795451

(1.58)

4.309822

(2.55)

R2 (Within)

0.2250

0.1253

Source: Author’s computation

Values in parenthesis represent t-values for fixed effects and z-values for random

Hausman test

GDP

Coefficients-Fixed Effects

Coefficients-Random Effects

Difference

HS

-0.1860329

-0.1251016

-0.0609312

HD

-0.336063

-0.0276192

-0.0059871

C

-0.2195534

-0.0184041

-0.2011493

I

0.3645943

0.117618

0.02469762

P

0.0078807

-0.0136453

0.021526

R

0.0667397

0.0721238

-0.0053841

Source: Author’s Computation  Ï‡2 (6)= 47.62 P- χ2 (6)=0.0000

Based on the Hausman test, H0 is rejected. Hence, for the regression, the fixed effect model will be most appropriate. Therefore, the fixed effect results are discussed below.

4.6 Interpretation of empirical results and findings

It can be observed that the coefficient of savings, debt, consumption, investment are significant while inflation, interest rate and the constant term are all insignificant at the 10% level.

The coefficient of savings is consistent with the theoretical literature which postulates that savings has a significant role to play in determining economic growth in host countries. The empirical result show that a 1% increases in savings negatively affects economic growth by 0.186 % in the selected countries. If the household savings ratio is analysed over the previous 15 years, the results show that it was unstable varying from 13% of disposable income in 1995 to only 3% in 2004. Households have decided to save less after tax income compared to the other periods due to the rise in consumer loans and mortgage removal from the market for housing.

Household debt is defined as consumer debt is consumer credit which is outstanding and as per the empirical results, there is a negative coefficient for debt which suggests that a 1% increase in debt will decrease economic growth by 0.336%. with a decline in household savings, unemployment and interest rates, people started to borrow more and save less. Consumer credit has gone beyond 10% from 1994 till mid of 2005. When households usually borrow money, it allows him to spend more than he has as current income and this creates dissavings and this is risky with all problems linked with lack of credit worthiness which decreases economic growth.

Consumption is assumed to be having a positive relationship with economic growth. Households can actually consume more goods and services and hence, increasing their standard of living. This increases total tax revenues for government which can now invest more on infrastructure for the benefits of everyone. The exports of goods can thus become possible and due to demand pull tendencies, there can be appreciation of the home currencies. As a result, exports will fall and economic growth will decline as well. The coefficient of consumption is negative and this finding is significant at the 1% level. An increase in consumption of 1% will cause economic growth to decrease by 0.219%.

The finding that the coefficient of investment is both consistent with theory and statistically significant is not surprising at all. Investment is usually a major determinant of economic growth and this has been confirmed by the results. Hence a 1% increase in investment increases economic growth by 0.364%.

The coefficient of inflation is positive which implies that a 1% increase in inflation would increase economic growth by 0.00788%. Though inflation can lead to uncertainty about the future profitability of investment projects (especially when high inflation is also associated with increased price variability), this leads to more conservative investment strategies than would otherwise be the case, ultimately leading to lower levels of investment and economic growth.

The constant term is also significant at 1% level.

The coefficient of interest rate is positive implying that a 1% increase in interest rate would cause economic growth to rise by 0.067%. There is thus a positive relationship between interest rates and economic growth. When interest rates rise, economic growth is assumed to be increasing as well since there would be incentives to produce more goods and services with the increase in investment following the rise in interest rates.