A simple model for examining global influences

The analytical model

The basic analytical model presented below is constructed with three basic design elements in mind: (1) a simultaneous system of equations; (2) variables are expressed in terms of shares; and (3) time is explicitly introduced.

URBSHR_rt = URBSHR_rt (GLBLSHR_rt-1, LCLSHR_rt-1) (1)

GLBLSHR_rt = GLBLSHR_rt (URBSHR_rt, LCLSHR_rt, STRCSHR_rt) (2)

LCLSHR_rt = LCLSHR_rt (URBSHR_rt, GLBLSHR_rt, STRCSHR_rt) (3)

where

URBSHR_rt - share of total urban population by region r in time t

GLBLSHR_rt - share of total global factor influence present in region r at time t

LCLSHR_rt - share of total local influence present in region r at time t

STRCSHR_rt - share of total structural facilities in region r at time t

Equation 1 shows how urbanization may be influenced by both global and local factors. It will be through this equation that the question of whether or not global influences exacerbate urban primacy is discussed. Equation 2 shows how the application of global influence on the urban system may be influenced by existing urbanization patterns, local factors, and structural facilities. Equation 3 likewise shows how the distribution of local influences is affected by existing urban patterns, global factors, and structural facilities.

The model is constructed as a simultaneous system of equations in order to address three basic issues: (1) the nature of the influence of both global and local factors; (2) the influence, in turn, of existing urban patterns on the way these factors exert their influence; and (3) the interaction between global and local factors.

Since the focus of the analysis is on the influence of global factors on the pattern of urbanization in the Philippines rather than on the urban growth of individual provinces or regions, the model is cast to discern relative rather than absolute changes and influences. A more meaningful examination of the question concerning globalization and urban primacy, for example, can be made only in these terms.

The variables used in the model are, therefore, measured as shares. For example, urbanization is measured as the urban population in the region over the total urban population in the country. Hence, an increase in the share of urbanization of one region necessarily reduces that of others. The influences on urban primacy become more interesting to study under this formulation.

Take, for example, the relationship between foreign direct investment and urbanization. If the variables in the model were to be measured in absolute terms, an observed positive relationship would not be sufficient to show that foreign direct investment promotes urban primacy. Cast in relative terms, on the other hand, the model should be able to provide a sufficient test of this hypothesis.

To demonstrate this, consider the following simple linear regression model: URBSHR_r = b₀ + b₁FINVSHRr + u, where FINVSHR_r is the share of total foreign investment going to region r and u is the error term. If b₁ were estimated to be negative, this would mean that, as the foreign investment share of the average region increases, the share of the total urban population of the average region declines. This happens only when a substantial share of foreign investment is applied to only one or two regions. As the urbanization levels in these favoured regions increase in response, the shares of the other regions decline and then the average region will experience a marginal decline. Hence, b₁ < 0 will indicate a positive relationship between foreign investment and urban primacy.

The other element built into the simple model is time. This is deemed necessary considering that, whereas global and local factors have substantial short-run effects, urbanization patterns tend to change only in the long term. Simple dynamics are introduced in the model to reconcile the simultaneous examination of both short-run and long-run variables.

Main hypotheses

The design parameters of the simple model presented earlier should allow for the analysis of the following hypotheses:

(1) Global factors and urban primacy. As discussed earlier, the use of relative values would allow us to determine whether global factors such as direct foreign investment and exports help strengthen the dominance of the national capital city. If this is the case, we should expect the partial effect of GLBLSHR_rt-1 on URBSHR_rt in Equation 1 to be negative.

(2) Local factors and balanced growth. The introduction of local factors as an argument in Equation 1 should allow us to determine if local factors such as local investment and government spending are more responsive to regional development objectives. If this were so, we can expect the partial effect of LCLSHRrt-1 on URBSHRrt to be positive.

(3) Existing urban patterns and the distribution of global and local factors. In both Equations 2 and 3, the relative urbanization level of a region is presented as a determinant of the way global and local factors are geographically distributed. The interest here is to determine the ability of regional centres to attract these influences. If the regions, through their city centres, had such abilities, the partial effects of URBSHRrt on the right-hand-side variables of Equations 2 and 3 should be positive.

(4) Crowding between global and local factors. LCLSHR_rt and GLBLSHR_rt are introduced as arguments in Equations 2 and 3, respectively, to test whether global and local factors crowd-in or crowd-out one another. If both sets of factors crowd-in one another, the partial effects should be positive. Otherwise, we should expect negative partial effects.

Empirical specification and data

The model represented by Equations 1-3 is specified as a system of five linear equations using regional-level data for 1980, 1985, and 1990. Urbanization is presented in terms of the regional share of the total urban population (URBSHR) as mentioned earlier. The influence of global factors is to be examined using regional shares of total foreign direct investment (FINVSHR) and of total exports (EXPSHR). On the other hand, the influence of local factors will be examined in terms of the regional shares of total local investment (LINVSHR) and of government spending (GSPDSHR).

Other variables introduced in the model as exogenous determinants are regional shares of: total banking offices (BNKSHR), total electrical connections (ELCSHR), total telephone exchanges (TELSHR), and total infant deaths (INFDSHR). The first three are used as measures of structural facilities, while the fourth is introduced as a determinant of government spending representing broad social objectives.

The basic empirical model is presented in the following system of equations:

URBSHR_rt = b₁ + b₂FINVSHR_rt-1 + b₃EXPSHR_rt-1 + b₄LINVSHR_rt-1 + b₅GSPDSHR_rt-1 + u₁ (4)

FINVSHR_rt = b₆ + b₇URBSHR_rt + b₈LINVSHR_rt-1 + b₉ELCSHR_rt + b₁₀TELSHR_rt + u₂ (5)

EXPSHR_rt = b₁₁ + b₁₂URBSHR_rt + b₁₃FINVSHR_rt-1 + b₁₄BNKSHR_rt + b₁₅TELSHR_rt + u₃ (6)

LINVSHR_rt = b₁₆ + b₁₇URBSHR_rt + b₁₈FINVSHR_rt-1 + b₁₉BNKSHR_rt + b₂₀ELCSHR_rt + u₄ (7)

GSPDSHR_rt = b₂₁ + b₂₂URBSHR_rt + b₂₃INFDSHR_rt-1 + u₅ (8)

The system that consists of Equations 4-8 is taken to be the basic empirical specification of the simple analytical model represented by Equations 1-3. It must be noted that the way the empirical model is specified was primarily driven by data restrictions rather than by theoretical soundness. First of all, data to complete a three-year, thirteen-region panel are available only for the variables specified here. Secondly, degrees of freedom limitations allow us to introduce truly exogenous variables as well as lagged endogenous variables only sparingly.

In order to get around the second limitation, an alternative specification of the basic empirical model is also presented. In the new system, more lagged variables are introduced to test dynamic effects underlying the model. In particular, previous period values of the dependent variables are introduced in each equation. The alternative specification is as follows:

URBSHR_rt = b₁ + b₂URBSHR_rt-1 + b₃FINVSHR_rt-1 + b₄EXPSHR_rt-1 + b₅LINVSHR_rt-1 + b6GSPDSHR_rt-1 + u₁ (9)

FINVSHR_rt = b₇ + b₈URBSHR_rt + b₉FINVSHR_rt-1 + b₁₀ELCSHR_rt + b₁₁BNKSHR_rt + u₂ (10)

EXPSHR_rt = b₁₂ + b₁₃URBSHR_rt + b₁₄FINVSHR_rt-1 + b₁₅BNKSHR_rt + b₁₆EXPSHR_rt-1 + u₃ (11)

LINVSHR_rt = b₁₇ + b₁₈URBSHR_rt + b₁₉FINVSHR_rt-1 + b₂₀LINVSHR_rt-1 + u₄ (12)

GSPDSHR_rt = b₂₁ + b₂₂URBSHR_rt + b₂₃INFDSHR_rt-1 +b₂₄GSPDSHR_rt-1 + u₅ (13)