PMA WORKING PAPER ANALYSIS OF THE DEMAND SIDE IN THE PALESTINIAN ECONOMY. Michel Dombrecht Mohammed Aref Saed Khalil WP/12/03.

Transcription

1 PMA WORKING PAPER WP/12/03 ANALYSIS OF THE DEMAND SIDE IN THE PALESTINIAN ECONOMY Prepared by: Michel Dombrecht Mohammed Aref Saed Khalil

2 PMA Working Paper Research and Monetary Policy Department ANALYSIS OF THE DEMAND SIDE IN THE PALESTINIAN ECONOMY Prepared by: Michel Dombrecht 1, Mohammed Aref and Saed Khalil February 2012 February 2012 All Rights Reserved. Suggested Citation: Palestine Monetary Authority (PMA), ANALYSIS OF THE DEMAND SIDE IN THE PALESTINIAN ECONOMY, Ramallah Palestine All Correspondence should be directed to: Palestine Monetary Authority (PMA) P. O. Box 452, Ramallah, Palestine. Tel.: Fax: JEL Classification Numbers: C53; E21; E22 Keywords: Econometric forecasting; Aggregate demand; Consumption; Saving; Investment. Authors addresses: 1 University of Antwerp, Belgium and Hogeschool Universiteit Brussels, Belgium. 2

3 CONTENTS I. INTRODUCTION 6 II. RECENT ECONOMIC DEVELOPMENTS IN THE PALESTINIAN TERRITORY 6 III. SUMMARY ACCOUNT OF HOUSEHOLD 8 IV. HOUSEHOLD CONSUMPTION FROM ATHEORETICAL PERSPECTIVE 10 V. EMPRICAL ANALYSIS OF HOUSEHOLD FINAL CONSUMPTION EXPENDITURES IN PT 12 VI. THE PRIVATE FINAL CONSUMPTION DEFLATOR 18 VII. CLASSIFICATION OF INVESTMENT EXPENDITURES 20 VIII. HOUSING INVESTMENT FROM A THEORETICAL PERSPECTIVE 20 IX. CORPORATE FIXED CAPITAL FORMATION FROM A THEORETICAL PERSPECTIVE 22 X. EMPIRICAL ANALYSIS OF GROSS CAPITAL FORMATION 22 XI. CONCLUSIONS 27 3

4 LIST OF TABLES TABLE 1: SUMMARY ACCOUNT OF HOUSEHOLD AT CURRENT PRICES 9 TABLE 2: FORMATION OF FINANCIAL ASSETS AND NEW FINANCIAL LIABILITIES OF HOUSEHOLDS 10 TABLE 3: LONG RUN DETERMINANTS OF THE AVERAGE PROPENSITY TO CONSUME IN PT 17 TABLE 4: COINTEGRATION TEST 17 TABLE 5: THE STATIONARITY TESTS FOR VARIABLES USED IN THE PRIVATE CONSUMPTION EQUATION 18 TABLE 6: THE PRIVATE CONSUMPTION PRICE DEFLATOR 19 TABLE 7: THE STATIONARITY TESTS FOR THE VARIABLES USED IN THE PRIVATE CONSUMPTION PRICE DEFLATOR EQUATION 19 TABLE 8: COINTEGRATION TEST FOR THE PRIVATE CONSUMPTION PRICE DEFLATOR 19 TABLE 9: BREAKDOWN OF FINAL INVESTMENT EXPENDITURES 20 TABLE 10: AN AVERAGE YEARLY DECLINE IN THE CAPITAL/ OUTPUT RATIO 26 LIST OF FIGURES FIGURE 1: ANNUAL GDP GROWTH RATES DURING ( ) 7 FIGURE 2: THE CONTRIBUTION OF ECONOMIC ACTIVITIES TO REAL GDP DURING ( ) 7 FIGURE 3: THE DISTRIBUTION OF INVESTMENT AND CONSUMPTION BETWEEN THE PUBLIC AND THE PRIVATE SECTORS DURING ( ) 8 FIGURE 4: PRIVATE CONSUMPTION PRICE DEFLATOR AND CONSUMER PRICE INDEX IN PT 14 FIGURE 5: THE REAL AVERAGE PROPENSITY TO CONSUME IN PT (APC) 15 FIGURE 6: THE AVERAGE PROPENSITY TO CONSUME AND THE NIS/USD EXCHANGE RATE 16 FIGURE 7: THE AVERAGE PROPENSITY TO CONSUME AND THE RISK FREE INTEREST RATE 16 FIGURE 8: AVERAGE GROSS PROPENSITY TO INVEST PUBLIC SECTOR 23 FIGURE 9: AVERAGE GROSS PROPENSITY TO INVEST PRIVATE SECTOR 24 FIGURE 10: AVERAGE GROSS PROPENSITY TO INVEST BUILDING 24 FIGURE 11: AVERAGE GROSS PROPENSITY TO INVEST NON- BUILDING 24 FIGURE 12: PRIVATE SECTOR CAPITAL/OUTPUT RATIO AND USER COST OF CAPITAL 26 4

5 ABBREVIATIONS CNFI DDAWPR$ DWBPU$ DWBPR$ DWBIS$ DGE DPRE DISE DDAW$ DDAWIS$ DAMWD DAMWIDIS GREMIT GDP GNDI GNI GS JD MoF NCTAPR NCDT NIS PMA PT PA PCBS PCPIP$ RWB RWBG RVAPU T USD Net Factor Income from Abroad Daily Average Wage in the Private Sector Total Wage Bill in the Government Sector in USD Total Wage Bill in the Domestic Private Sector in USD Compensation of Employees in Israel and Settlements Government Employment Number of Persons Employed in the Domestic Private Sector Number of Persons Employed in Israel and Settlements Daily Average Wage in the Government Sector Daily Average Wage in Israel and Settlements Average Monthly Works Days in PT Average Monthly Works Days in Israel and Settlements Gross Remittances Gross Domestic Product Gross National Disposable Income Gross National Income Gaza Strip Jordanian Dinar Ministry of Finance Net Current Transfers from Abroad in USD received by the Private sector Number of Closure Days for Trade New Israeli Shekel Palestine Monetary Authority Palestinian Territory Palestinian Authority Palestinian Central Bureau of Statistics Consumer Price Index Remaining West Bank Remaining West Bank and Gaza Real Value Added in the Public Sector Trends US dollar 5

6 I. INTRODUCTION ANALYSIS OF THE DEMAND SIDE IN THE PALESTINIAN ECONOMY In the supply side of the economy, firms set prices so as to maximise profits. Firms are therefore price setters, implying that output is basically demand determined. Of course demand itself is driven by real income growth, which is itself sourced in the labour productivity growth which itself originates in the supply environment of the economy. In a world of perfect competition, prices would be solely determined by deviations from demand and supply and the mark up would be zero. In most countries markets are basically characterised by imperfect competition, more in particular by so called 'monopolistic competition'. Therefore the demand for goods and services is mainly driven by the real income growth of all sectors in the economy and by the price setting behaviour of profit maximizing firms. The main demand components are consumption by households, private investment expenditures, government expenditure, exports and imports. In this paper we will mainly focus on private final consumption and investment expenditures. II. RESENT ECONOMIC DEVELOPMENTS IN THE PALESTINIAN TERRITORY The annual real GDP growth rate showed obvious fluctuations during due to the political conditions that PT lived under. Real GDP growth rate was the highest (12 percent) in 2004 due to the relative political stability during that year 2 ; 2006 witnessed the lowest growth rate, in which GDP declined by 5.2 percent as a result of the new political and economic conditions after the legislative elections, which resulted in a cut in foreign financial support and the refusal of Israel to transfer tax returns to PA. In the years , the growth rate increased due to the relatively stable political conditions despite of the Palestinian division resumption of the international financial support to PA, and the release of tax revenues by Israel. Thus 2010 was better than its predecessor, as GDP grew by 9.3 percent, as a result of the development policies adopted by the Palestinian government under the Palestinian Reform and Development Plan, which ended by that year. In general, the average annual real GDP growth rate reached 5.3 percent during ( ), which is acceptable under the conditions that the Palestinian economy experienced; but this rate was not sufficient to enable the Palestinian economy to achieve sustainable development. In terms of the contribution of different economic sectors to real GDP, it is noticed that the contribution of the industrial sector has declined during the period from 17.3 percent in 2004 to 12.3 percent in However the contribution of the agriculture sector at 6.3 percent improved in 2010 compared to the previous years, but it was still lower than its levels in On the other hand, the contribution of the construction sector improved during the period ( ) as a result of building new cities and different constructions in WB. The 2 Therefore, PCBS adopted it as a new base year instead of

7 services sector 3 continued to be the highest contributor to real GDP with 60.3 on average during the period. Figure 1: Annual GDP growth rates during ( ) 14% 12% 10% 8% 6% 4% 2% 0% -2% -4% -6% -8% 12.0% 8.6% 9.3% 7.1% 7.4% 5.4% % The weak contribution of productive activities compared to the services and trade activities is linked to the weak investment in these activities and the increase in consumption patterns of the Palestinian society as a result of the political and social conditions. Thus the Palestinian economy is characterized by a high volume of final consumption, with about percent of GDP on average during ( ); private consumption accounted for percent of GDP, while gross investment accounted for 24.7 percent of GDP on average during the same period, out of which the private sector accounted for 16.9 percent while the public sector accounted for 7.8 percent. Figure 2: The contribution of economic activities to real GDP during ( ) 70% 60% 50% 40% 30% 20% 10% 0% Agriculture Industry Construction Services 3 Includes productive services (retail and wholesale trade, transport, storage and communications, financial intermediaries, and hotels and restaurants), and social services (housing services, social services, education, health and social work, public administration and defense, and home services). 7

8 This structural defect in the economy comes as a result of the weak investment environment in the PT and as a result of the high correlation between the economic activities and the volatile and unstable political and security conditions beside the dependence of a large segment of the population on aid and foreign transfers. The average annual net factor income reached million USD (in current prices). While average annual net foreign transfers reached million USD in the same period. This makes the difference between GNDI and GDP high, with the gap between them reaching million USD (in current prices) or 39.8 percent of average GDP during the period. This also leads to an increase in the share of consumption expenditure in total expenditure. Figure 3: The distribution of investment and consumption between the public and the private sectors during ( ) 120% 100% 80% 60% 40% 20% 0% Private consumption Public consumption Private investment Public investment On the other hand, the low in government investment is attributed to the financial problems which the government budget faces and the lack of commitment by donor countries to disburse aid on a regular basis. This makes the priority for government go to cover the wages and the current expenses necessary for the continued operation of ministries and public sector institutions. III. SUMMARY ACCOUNT OF HOUSEHOLDS According to the national accounts methodology, a household is defined as a small group of people who share the same living accommodation, pool some of their income, and consume certain types of goods and services collectively, mainly housing and food. The sector includes resident household units and also all their economic activities that are unincorporated. Unincorporated enterprises which are owned by households but have complete business accounts are classified as quasi-corporations in the corporations sector. The household sector may be subdivided into: Employers; Own-account workers; Employees; a. Recipients of property and transfer incomes. 4 A valid starting point to analyze households behaviour is to look at the households accounts which relate their incomes, expenditures and net savings. The national accounts provide such information which may usefully be summarized and presented as in table United Nations, National Accounts: A Practical Introduction, New York,

9 Table 1: Summary account of households at current prices 6 (In Millions of USD) ITEM Value 1. Gross primary income - Wages and salaries 7 - Property incomes (net) - Gross mixed income 8 - Gross operation surplus - 2. Current transfers 9 - Transfers received - Transfers paid - 3. Gross disposable income (1+2) - 4. Change in net claims of households on institutions for occupational - retirement provision 5. Final consumption expenditure - 6. Gross saving (3+4-5) - 7. Capital transfers Gross capital formation - 9. Overall balance (6+7-8) - In fact the overall net financing capacity of households (line 9 in table 1) should in principle correspond to the difference between the formation of financial assets and new financial liabilities of the households. Table 2 provides an example of a useful classification of these flows 11. The accumulated values of these assets net of liabilities together with the households accumulated capital stock (mainly in housing) constitute overall net household wealth. As will become obvious in the next section, this net household wealth evaluated at market prices is considered to be an important explanatory factor in household expenditure, saving and asset allocation decisions. Most countries calculate the different items in the overall stocks of assets and debts of the household sector 5 National Bank of Belgium, Annual reports. 6 The data in this table are calculated in gross terms before deduction of consumption of fixed capital. 7 Remuneration, excluding that of owner entrepreneurs, including social security contributions and civil service pensions. 8 Mixture of compensation of employees and gross operating surplus of households that carry out business activities. 9 These are net amounts, i.e. the difference between transfers received from other sectors and those paid to other sectors, excluding transfers in kind. 10 These are net amounts, i.e. the difference between transfers received from other sectors and transfers paid to other sectors, including net acquisitions of non-financial non-produced assets and net acquisition of valuables. 11 National bank of Belgium, Annual reports. 9

10 Table 2: Formation of financial assets and new financial liabilities of households (In millions of USD) ITEM Value Formation of financial assets - At up to one year - Notes, coins and sight deposits - Savings deposits - Time deposits - Fixes income securities - Units of monetary UCI s - At over one year - Time deposits - Fixed income securities - Shares and other equity - Units of non-monetary UCI s - Net claims on insurance technical reserves - Other assets - New financial liabilities - Loans at up to one year - Loans at over one year - Mortgage loans - Instalment loans - Other - Other liabilities - Financial balance - In PT most elements of the households income and capital accounts are not available. As will be explained later in this paper, this absence hampers considerably any detailed macroeconomic analysis and will constrain the information set on the basis of which the PMA would be able to conduct its future monetary policy. IV. HOUSEHOLD CONSUMPTION FROM A THEORETICAL PERSPECTIVE Households' final objective is to maximize their expected consumption over their expected life time, given their expected intertemporal budget constraint 12. Instantaneous utility is assumed to depend on actual consumption. Consumption today is preferred to consumption 12 Current analysis of household consumption and investment decisions are based on the original contributions of, among others, R.C. Merton, Lifetime Portfolio Selection under Uncertainty: The Continuous Time Case, The Review of Economics and Statistics, 1969 and P. Samuelson, Lifetime Portfolio Selection by Dynamic Stochastic Programming, The Review of Economics and Statistics, 1969.

11 later, implying that consumers exhibit time preference. The contemporaneous budget constraint links current real wealth to previous period wealth, the real return on that wealth and real consumption (the wealth accumulation rule). Real wealth includes all financial and non-financial assets net of all debts. Households consumption, savings and wealth allocation decisions are then derived from a stochastic intertemporal optimization exercise. Under these general assumptions, the equilibrium consumption path basically depends on: Real total wealth with a unitary elasticity. Total wealth may be subdivided into nonhuman wealth and permanent disposable labour income (the discounted value of expected future net real labour income over the expected life time); The real risk free interest rate with a negative semi-elasticity; The expected real return on total household wealth; Parameters reflecting consumers time preference and their degree of risk aversion. In the short run consumption may deviate from its equilibrium value, but consumers will gradually adjust their consumption pattern to bring it back to its steady state value. More concretely, suppose a negative income shock occurs, consumers will first be reluctant to adjust their standard of living, implying a decline in consumption that is less than proportional when compared to the decline in their income. The average propensity to consume will therefore temporarily increase. Only after time, consumers will realize that they will have to bring their consumption in line with their disposable income. The reverse will happen after the occurrence of a positive income shock. The speed with which consumers adjust their consumption pattern may depend on their overall confidence concerning the future course of their real income. Sometimes this confidence or risk parameter is measured by the unemployment rate. An increase in that rate makes consumers more pessimistic about their future income expectations through increasing the likelihood of getting unemployed. Another phenomenon that is sometimes taken into account in the analysis of short run dynamics in consumption behaviour is the fact that part of the households has a binding budget constraint. They would therefore not base their current consumption decisions on expected life time real labour income, but on their current income, which may even need to be supplemented by bank credit. Applying this theoretical framework to any country and more specifically to the case of PT requires an evaluation and comparison between the data needed (as suggested by the mainstream theory) and those that are actually available. In this respect the following observations can be made: In most countries data on the market value of non-human wealth (but expressed in real terms by the consumer price index) are constructed using available information concerning financial assets and liabilities of households and an estimation of the market value of the housing stock. In PT data concerning financial assets and liabilities of households are only recently calculated. Other wealth items are not yet collected or calculated. In PT national accounts and assets and liabilities are basically expressed in USD. That means that incomes received in other currencies and assets and liabilities held in other currencies (essentially NIS, EUR and JOD which however is linked to the USD) are translated into USD. This implies that changes in the USD 11

12 exchange rate may create income and wealth effects which may affect consumption expenditures 13 ; Human wealth is mostly defined as permanent net disposable labour income which is the discounted value of the expected future net disposable labour income flow. That implies that in terms of logarithms one would expect the paths of current and permanent income to be co-integrated; The real risk free interest rate would be a nominal deposit rate adjusted for inflation. Here too, in PT, deposits are held in several currencies and inflation can be measured in terms of these several currencies. Furthermore average interest rates on deposits and loans are available since 2000 only; The real expected return on total wealth (of which the risk free return is a composite part) is otherwise composed of the return on financial assets and liabilities and on the housing stock and the net capital gains on these assets. Here again exchange rate effects may be important. These observations lead us to expect that, taking account of data availability, real private consumption behaviour in PT may show a long run relationship with one or more of the variables mentioned above: Current real net disposable labour income as a proxy for human wealth; Real wealth effects and expected real returns on wealth represented by the inflation rate and exchange rate movements; Policy interest rates in the US and/or Israel as a proxy for the nominal risk free discount rate; The unemployment rate as a measure of the household uncertainty. V. EMPIRICAL ANALYSIS OF HOUSEHOLDS FINAL CONSUMPTION EXPENDITURES IN PT In fact some of the variables relevant for analyzing private consumption should be present in the income and capital accounts of the household sector of which an example was presented in table 1. More specifically, households disposable labour income has to be derived from the national accounts. Unfortunately, this information is not yet provided by the PCBS. Therefore for the purpose of an empirical analysis of private consumption in PT, we had to approximate the concept of labour income using the limited available information. Furthermore these approximations need to be coherent with the activities in the other sector such as private sector and government sector employment and wage formation. Nominal disposable labour income of households is defined as the sum of their gross wage earnings, which themselves are the product of the volume of employment and the wage rate, and the net income transfers received from other sectors. The latter include net income transfers received from government and from abroad and income taxes and social security contributions paid. In this respect the available information includes (all expressed at current prices): 13 These effects were already documented in S. Sarsour, D. Cobham, Analyzing Transmission Mechanism of Monetary Policy in Absence of a National Currency: Case of Palestinian Territory, PMA, May

13 Number of persons employed by the government (DGE); Number of persons employed in the domestic private sector (DPRE); Number of persons employed in Israel and settlements (DISE); Daily average wage in government sector in USD (DDAW$); Daily average wage in private sector in USD (DDAWPR$); Daily average wage in Israel and settlements in USD (DDAWIS$); Average monthly work days in PT (DAMWD); Average monthly work days in Israel and settlements (DAMWDIS); Net factor income from abroad in USD (CNFI); Net current transfers from abroad in USD received by the private sector (NCTAPR), source National Accounts; Gross remittances (private sector received transfers from abroad (GREMIT)) from the Balance of Payments. This available information allows calculating by approximation the: Total wage bill in the government sector in USD: DWBPU$ = DDAWPU$ x DAMWD x 12 x DGE; Total wage bill in the domestic private sector in USD: DWBPR$ = DDAWPR$ x DAMWD x 12 x DPRE; Compensation of employees in Israel and settlements: DWBIS$ = DDAWIS$ x DAMWDIS x 12 x DISE; Part of current transfers, more specifically transfers received from abroad. These will be used in the empirical analysis of household behavior. It is important to note though that the available data do not allow calculating household gross disposable income, because data are unavailable for the following items: All elements of property (wealth) income of the household sector; Gross operating surplus of self employed and owner entrepreneurs; Transfers and taxes paid and received from the government sector. Some of these are available in the government sector account, but they have to be integrated and made coherent with the (yet non-existent) household account and furthermore it requires that the government account should be extended to the level of general government instead of only central government as is currently the case. 13

14 Given these constraints we have considered two income variables: Total wage bill including the compensation of employees working in the government sector, in the domestic private sector and augmented with wage income received by those living in PT but working in Israel and the settlements: DWBTOTIS$ = DWBPU$ + DWBPR$ + DWBIS$ The remittances received from abroad (Private sector current transfers received according to the balance of payments): GREMIT$. For the purpose of analyzing consumption behavior, we need to explore the relationship between private final consumption at constant prices and real household (disposable income) expressed in terms of purchasing power, i.e. nominal income deflated by a consumer price index. As to the price index, two alternatives are available: The private consumption price deflator taken from the national accounts; The consumer price index, measuring the cost of living and which is the main target for monetary policy. Figure 4 compares both consumer price indexes, both expressed in USD. As can be seen in some periods, both price indexes can deviate substantially. Since the PCBS does not yet publish its methodology as to how it calculates the national accounts and more specifically the deflators, we opt to use the consumer price index in the rest of our analysis. In the empirical analysis household consumption at constant prices will be compared with the calculated real income variable. And as is usual in economic analysis these variables will be expressed in logarithms so that the estimated parameters can be interpreted as elasticities. Figure 4: private consumption price deflator and consumer price index in PT (2004 = 1, expressed in USD) Consumer price index private consumption price deflator Figure 5 shows the average propensity to consume (APC) as the ratio of real private final consumption and real household wage income (measured as the sum of all labour income 14

15 (compensation of employees working within PT and in Israel and the settlements). As to the remittances received from abroad we will treat them as a separate variable (later alternative income measures can be explored). A number of observations can be made regarding this graph: The (APC) is larger than one. This is an atypical result since in most countries it is found to be smaller than one. It may be the result of: o Incomplete income statistics. Some components of household income may be lacking in the available statistics (part of which may be generated in the informal sector); o Real private consumption may be overestimated in the national accounts. This cannot be verified due to absence of information concerning the national accounts methodology; o If the statistics do provide a reasonable estimate of the APC, it would imply that in PT consumption expenditures are financed by running down assets or accumulating ever increasing indebtedness; The APC has been volatile and has been hit by the specific political and economic shocks on the PT economy in the past; The APC shows an upward trend since Figure 5: the real average propensity to consume in PT The main challenge, apart from interpreting the level of the APC (which requires more analysis and more information in the data) is to explore the existence of a long run equilibrium relationship between the movements of the APC and other potential factors suggested by the mainstream theory. As discussed before we can expect a negative relationship between the APC and the exchange rate of the USD in NIS. The national accounts are expressed in USD, but part of labour income is received in NIS and part of assets is held in NIS as well. 15

16 An appreciation of the USD (i.e. a depreciation of the NIS) may therefore cause negative income and wealth effects. Therefore one may expect an inverse relationship between the APC and the NIS/USD exchange rate index. Furthermore theory would predict a negative relationship to hold between the APC and the risk free interest rate. Since the accounts are in USD, the short term US policy rate was considered to be valid indicator for measuring this effect. Figures 6 and 7 confirm these inverse relationships between the log of the APC (LAPC) and the log of the NIS/USD exchange rate index (LMEXGRISINDEX) and the US Federal Funds interest rate (MFFRUS). Figure 6: the average propensity to consume and the NIS/USD Exchange rate LAPC LMEXGRISINDEX Figure 7: the average propensity to consume and the risk free Interest rate LAPC FEDERAL FUNDS INTEREST RATE US

17 Tables 3 and 4 confirm the existence of a long run equilibrium relationship between the APC and the mentioned explanatory variables. The equation was estimated by non-linear OLS, in that the sum of all income components was constrained to add up to one as required from theory. Also the equation accounts for the exceptional surge and busts of consumption in 1997 and 2000 respectively each by about 15% exceptional increase and decrease respectively. The cointegration test confirms stationarity of the residuals of the long run equilibrium equation. A Breusch Godfrey serial correlation LM test showed absence of serial correlation in the residual of the long run equation and therefore this equation can be directly used in the structural and financial programming models. Table 3: long run determinants of the Average Propensity to Consume (LRPC) in PT Estimated equation by OLS ( ) LRPC= C(1)+C(2)* LRDWBTOTIS+(1-C(2))* LRGREMITCPI +C(3) *LMEXGRISINDEX+C(4)* MFFRUS+C(5)*(DUM97-DUM00) Coefficient Std. Error t-statistic Prob. C(1) C(2) C(3) C(4) C(5) R-squared Mean dependent var Adjusted R-squared S.D. dependent var S.E. of regression Akaike info criterion Sum squared resid Schwarz criterion Log likelihood Hannan-Quinn criter F-statistic Durbin-Watson stat Prob(F-statistic) Table 4: Cointegration test Null Hypothesis: RESIDUAL has a unit root Exogenous: Constant Lag Length: 0 (Automatic based on SIC, MAXLAG=3) t-statistic Prob.* Augmented Dickey-Fuller test statistic Test critical values: 1% level % level % level

18 Table 5: the Stationarity tests for the variables used in the private consumption equation* Variables ADF PP l(rpc) Level st difference l(rdwbtotis) Level st difference l(ddawpu$) Level st difference l(rgremitcpi) Level st difference l(mexgrisindex) Level st difference (mffrus) Level st difference * Presented values represent the P-value In table 5: l stands for the logarithm of a variable. rpc = real private consumption rdwbtotis = real total wage bill including the worker in Israel & settlement ddawpu$ = daily average wage in the public sector in US Dollar rgremitcpi = real received transfers from abroad private sector (BOP) mexgrisindex = exchange rate index mffrus = fedral funds interest rate US Apparently variables (expressed in logarithms) are integrated of order one, implying that the first differences are stationary at the 95% level for some variable and 90% level for other variable according to the Augmented Dickey Fuller (ADF) test and the Phillips-Perron (PP) test. VI. THE PRIVATE FINAL CONSUMPTION DEFLATOR Nominal private consumption is the product between private consumption at constant prices (on which theory focuses) and the private consumption price deflator (PCPD$). Normally this deflator should be highly correlated with the consumer price index (PCPIP$). A source off difference between both is that the deflator is implicitly calculated on a changing basket, whereas the price index is calculated on a fixed basket (that only changes infrequently). Therefore shocks like in the cost of imports or world food prices may temporarily make both price indexes deviate. We therefore estimated the deflator in function of the price index, the cost of imports and the world food price index all expressed in USD and in logs. The results are shown in tables 6 and 8 18

19 Table 6: The private consumption price deflator (LPCPD$) Estimated equation by OLS ( ) Variable Coefficient Std. Error t-statistic Prob. C LPCPIP$ LPWFPI LPCIM$ R-squared Mean dependent var Adjusted R-squared S.D. dependent var S.E. of regression Akaike info criterion Sum squared resid Schwarz criterion Log likelihood Hannan-Quinn criter F-statistic Durbin-Watson stat Prob(F-statistic) Table 7: the stationarity tests for the variables used in the private consumption price deflator equation Variables ADF PP Lpcpd$ Level st difference Lpcpip$ Level st difference Lpwfpi Level st difference Lpcim$ Level st difference In table 7: capital L stands for the logarithm of a variable. pcpd$ = private consumption price deflator pcpip$ = palestine consumer price index (Dollar) pwfpi = world food price index pcim$ = cost of import (dollar) Apparently variables (expressed in logarithms) are integrated of order one, implying that the first differences are stationary at the 95% level for some variable and 90% level for other variable according to the Augmented Dickey Fuller test and the Phillips-Perron test. Table 8: Cointegration test for the private consumption price deflator Null Hypothesis: RESID04 has a unit root Exogenous: Constant Lag Length: 0 (Automatic based on SIC, MAXLAG=2) t-statistic Prob.* Augmented Dickey-Fuller test statistic Test critical values: 1% level % level % level

20 As expected the coefficient of the consumer price index is close to unity. The unit root test confirmed the existence of a cointegration vector between these variables and an LM test confirmed absence of serial correlation in the residuals. VII. CLASSIFICATION OF INVESTMENT EXPENDITURES From the point of view of economic analysis and more in particular from the needs to monitor the stance of monetary policy and the transmission mechanisms of monetary policy, a useful breakdown of total investment expenditures both at constant and current prices is presented in table Table 9: Breakdown of final investment expenditures (Millions USD) 2009 At constant prices At current prices Gross fixed capital formation - - Housing - - Enterprises - - General Government - - Change in inventories - - Total - - The breakdown of gross fixed capital formation according to housing, enterprise and general government categories is not available in PT. This absence clearly hampers a detailed macroeconomic analysis and the information set to conduct well informed monetary policy. VIII. HOUSING INVESTMENT FROM A THEORETICAL PERSPECTIVE The consumer s stochastic intertemporal optimization problem does not only provide for the optimal consumption path but also the household s optimal wealth accumulation and wealth allocation decisions. As was already obtained by Merton (1969) the sequential nature of the consumer s behaviour can be summarized as follows. First the share of each wealth item in total wealth is determined on the basis of expected returns, variances and covariances of returns and risk aversion. Then the total expected return on the portfolio can be derived on the basis of expected returns and the optimal portfolio allocation. Given this return, together with previous period wealth, time preference and risk aversion, the consumer decides on desired consumption and saving. This in turn determines current period wealth. The latter together with the optimal wealth allocation determines the optimal level of demand for all individual assets. 14 National Bank of Belgium, Annual Reports. 20

21 Households housing investment has to be analyzed as part of its global asset allocation decision (the allocation of wealth between housing, shares, fixed prices assets and mortgage loans) 15. According to this theory total wealth is the scaling factor for the demand for each class of assets and debts. But in the housing market, permanent income may be the relevant wealth item. The reason is that permanent income can be traded for immediate investment finance through the mortgage market (or other types of longer term loans). Therefore the relevant scaling factor in the demand for houses may rather be permanent income. As mentioned before permanent income can be expected to be cointegrated with current income. The desired share of wealth (income) invested in houses then depends on the risk premia on houses and mortgage loans and on the debt ratio. Given income, this determines the optimal housing stock evaluated at market prices. But because housing prices are much more volatile as compared to housing numbers or volume, this optimal demand for houses rather explains the movement of market prices for existing houses. Within the framework of this theory, housing investment itself is conceived as a supply of new houses which adds to the existing housing stock. This supply of new houses can be modeled as a modified Tobin s-q. Desired investment depends on the ratio of the cost of building a new house versus the prices of existing houses on the secondary market. Modeling housing investment would therefore require the following data: Housing investment at constant and current prices from the national accounts; An estimate of the housing stock, using for example the permanent inventory method; Average market price of existing houses; Household disposable income; Stock of outstanding mortgage debt; Index of the cost of building a new house; The rent earned on the existing housing stock, which should be available in the national accounts and which should allow to calculate the return on housing investment.. In PT an empirical analysis of the housing market is not possible because of the absence of the data mentioned. Even national accounts data for housing investment are not available. Since the housing activity in each country, and certainly also in PT, is an important contributing factor to economic activity and growth and is also an important transmission mechanism of monetary policy, it is of prime importance that these data be collected as soon as possible. 15 See for example, M. Dombrecht, R. Wouters, Interest Rates and Household Absorption in Belgium, Bank for International Settlements 21

22 IX. CORPORATE FIXED CAPITALFORMATION FROM A THEORETICAL PERSPECTIVE The production technology described in the supply side implies that in the long run the capital - output ratio is constant. Real private investment is defined as the increments of the real stock of capital augmented with replacement investment (due to capital depreciation). This implies that in the long run also the ratio of real private investment to output remains constant. Deviations between the real short term interest rate from its equilibrium value may change this ratio. But such deviations are in principle only temporary, since in the monetary sector mechanisms are at work that tend to make the real interest rate converge to its equilibrium value. In the short run, firms do not adjust immediately and completely to changes in the level of output. Investments occur in stages and are planned well in advance. Temporary shocks in output will therefore typically not result in immediate and complete stabilization of the investment - output ratio. Real private investment will therefore adjust with delay to shocks in output. The speed of adjustment may depend on the degree of confidence firms have concerning future prospects and other variables such as changes in relative factor prices or deviations of the real interest rate from its equilibrium level and accessibility to internal and external finance. Modeling companies gross fixed capital formation would require data at the very minimum on: Gross fixed capital formation of companies at constant and current prices; The calculation of a real business capital stock, for example using the permanent inventory method; Private real value added In PT data on companies gross fixed capital formation are not available. X. EMPIRICAL ANALYSIS OF GROSS CAPITAL FORMATION In the national accounts, gross capital formation is part of the expenditure approach to GDP. The PCBS publishes the following breakdown: Gross fixed capital formation: o Buildings o Non-buildings Change in inventories For analysis necessary for monetary policy making, this classification is unsatisfactory. The PMA s analysis would benefit very much if the PCBS would provide for the classification of total investment expenditure as presented in table 7. The current classification in the national accounts seriously hampers the PMA to analyze and monitor the transmission mechanisms of monetary policy, which is one of the fundamental inputs in policy making in a central bank. 22

23 For the time being, the PMA will base its analysis on an unofficial estimate of private and government gross fixed capital formation. This implies that household investment in housing cannot be separated from gross fixed capital formation of companies which considerably limits the usefulness of the analysis. Indeed both categories of final investment expenditures are basically driven by different motives. This implies that the estimated investment function may be unstable over time. As was mentioned, from a theoretical point of view one would expect the capital/output ratio as well as the investment/output ratio to be stationary variables. We can check whether this is actually the case in PT. Figure 8 shows the average gross propensity to invest by the government sector (gross government investment as a proportion of public value added, all expressed at constant prices). Figure 8: Average Gross Propensity to Invest in Public Sector As can be seen the public average propensity to invest does not show any clear trend, which is consistent with a steady state assumption. On the other hand the average propensity to invest in the private sector gives a different picture. As shown by Figure 9, the average propensity to invest in the private sector has been on a declining trend over most of the sample period. From a theoretical perspective this is difficult to explain. Probably politicoeconomic shocks have affected the inclination to engage in risky investment projects negatively. It should also be added that in recent years the building activity has picked up somewhat as demonstrated in Figure 10. On the other hand non-building investment remains relatively weak (see Figure 11). In principle in a stable environment, it is to be expected that the private sector average propensity to invest would return to more normal levels. 23

24 Figure 9: Average Gross Propensity to invest in the Private Sector Figure 10: Average Gross Propensity to Invest in Building Figure 11: Average Gross Propensity to Invest in Non-Building

25 As to the modeling of private investment, given all the constraints and reservations mentioned above, we followed a neo-classical investment theory approach. As was explained in the supply side paper, we expect the private sector capital/output ratio to be affected only by the real user cost of capital. But from a steady state point of view, the real user cost of capital is expected to be a stationary process, just like the capital/output ratio itself. In order to check these propositions for the case of PT, we need to construct a real capital stock and the real user cost of capital. The construction of the real private sector capital stock needs a number of preliminary assumptions, whose validity has not been thoroughly checked. The results should therefore be treated with great caution. The main assumptions relate to the depreciation ratio and the beginning of period capital stock to be employed in a permanent inventory method. Because the available data on real private investment contain both housing and other capital expenditure, we assumed an expected life time of the average capital stock of 30 years, implying a yearly depreciation rate of 3.33%. Furthermore we assumed the starting value of the capital multiple to be about 10. It goes without saying that these parameters can be changed on the basis of more and better information. According to the permanent inventory method and the assumed parameters, the real private sector capital stock can be estimated using the following formula: KPR t KPR (1 ) RPI t 1 Where: KPR is the real private sector capital stock; δ is the depreciation rate; RPI is real private sector investment. From this, we can estimate the private sector capital/output ratio as: CORPR = KPR/RVAPR Where: RVAPR is the real value added in the private sector. The real user cost of capital is traditionally calculated as follows: RUCC PI( R ) PI PVA Where: RUCC is the real user cost of capital; R is the nominal rate of interest; PI is the price of the capital good (the private sector investment price deflator); PVA is the output price (the private sector value added deflator). Theoretically we would expect both the CORPR and the RUCC to be stationary variables (see Figure 12). Unit root tests indicate that: The calculated RUCC is indeed stationary; The calculated CORPR is only trend stationary at the 10% level. 25

26 We conclude from this that the CORPR has been subjected to a negative trend over the sample period, which confirms the observed decline in the average propensity to invest in the private sector. Figure 12: PRIVATE SECTOR CAPITAL/OUTPUT RATIO (CORPR) AND USER COST OF CAPITAL real user cost of capital CORPR This negative trend is confirmed in table 10 that reports an average yearly decline in the capital/output ratio of about 2.8%. Furthermore table 10 shows that: The real user cost of capital does not affect the capital output ratio, which was to be expected since this ratio, contrary to the capital/output ratio, was found to be stationary; The number of closures dates positively affects the capital/output ratio. This is to be explained by the fact that closure days immediately negatively affect output. On the other hand the capital stock is the sum of past net investment expenditures (net additions to the capital stock) and therefore the instantaneous effect of closure dates on capital is much more mitigated. Therefore an increase (decrease) in closure days automatically increases (decreases) the capital/output ratio. Table 10: An average yearly decline in the capital/output ratio (LCORPR) Estimated equation by OLS ( ) Variable Coefficient Std. Error t-statistic Prob. C RUCC LNCDT R-squared Mean dependent var Adjusted R-squared S.D. dependent var S.E. of regression Akaike info criterion Sum squared resid Schwarz criterion Log likelihood Hannan-Quinn criter F-statistic Durbin-Watson stat Prob(F-statistic)

27 In fact these results basically belong to the supply side of the economy. From the demand perspective the private investment expenditures can be obtained from the capital/output equation estimated above, using the following identity: RPI t KPR t 0.33KPR t 1 XI. CONCLUSIONS Analyzing the demand size of the PT economy along internationally accepted approaches is a particularly challenging exercise. The main reason is the absence of the necessary set of complete and coherent national account data. The income and capital account of the household sector is not available and therefore the analysis of household consumption and investment expenditures are impossible both in a descriptive way and in a scientific approach using internationally accepted theories. Furthermore the analysis of investment expenditures using internationally accepted approaches is impossible due to the absence of the appropriate breakdown in the investment statistics. The sectoral breakdown of gross fixed capital formation between households, enterprises and general government is unavailable. This absence of the necessary data very much hampers detailed macro-economic analysis and is bound to constrain the necessary information set for conducting future monetary policy. Using available data for the real private consumption analysis, we had to construct for the main explanatory variables suggested by theory. The most important one is household s disposable income for which no official data are available. Calculating and using a crude proxy based on partial information yielded the result that the average propensity to consume of PT households is substantially larger than one, which is an atypical result compared to what is found in other countries. The movements in the average propensity to consume were found,to be related to exchange rate and interest rate developments. An empirical analysis of the housing market was not possible due to the absence of national accounts data on households housing investment expenditures and of the explanatory variables suggested by mainstream theories. This is particularly harmful since the housing market in most countries is an important variable for economic growth and for business cycle fluctuations. It is also an important transmission mechanism of monetary policy. Another important variable from the point of view of economic growth, business cycle fluctuations and transmission of monetary policy is corporate gross fixed capital formation. Data related to this variable are missing in the national accounts. The investment analysis in this paper makes use of a proxy for total private investment and uses these data to construct a real private sector capital stock using the permanent inventory model. This allowed the construction of a proxy for the private sector capital-output ratio which was explained in terms of a constructed proxy for the user cost of capital according to standard investment theory. 27