This Selected Issues paper examines financial linkages and the correlation between Australian and U.S. output. It shows that the financial linkages have played an important role in conveying shocks from the United States to the Australian economy, and that these have become increasingly important in the 1990s. The paper examines income and output convergence across Australian states. It also examines the role of the terms of trade and different commodity prices in explaining the real exchange rate.

Abstract

This Selected Issues paper examines financial linkages and the correlation between Australian and U.S. output. It shows that the financial linkages have played an important role in conveying shocks from the United States to the Australian economy, and that these have become increasingly important in the 1990s. The paper examines income and output convergence across Australian states. It also examines the role of the terms of trade and different commodity prices in explaining the real exchange rate.

II. Income and Output Convergence Across Australian States9

1. During the 1990s, there was no significant convergence in real per capita income and output across the Australian states.10 Moreover, a pattern of regional unemployment has persisted, with unemployment rates being consistently higher than the national average in those states with lower than average real per capita income and output.

2. The lack of convergence and persistence of high unemployment in the lower real per capita income/output states suggests the existence of factors that maybe impeding adjustment. Likely suspects are rigidities in labor markets and government transfers. Among these transfers, payments to persons have mitigated part of the divergence in real per capita incomes across states, but they may also have contributed to part of the gaps in real per capita output and unemployment. Transfers to state governments are based on the principle of trying to “equalize” fiscal resources to ensure that states have the capacity to deliver services at a similar standard across the country. These transfers are thought to have potentially created some perverse incentives that could adversely affect real per capita income/output and unemployment. In addition, there is the possibility that structural reforms to the economy paved the way for increased adoption of new technologies that may have spurred faster growth in real per capita income and output growth in the higher income states.11

3. Empirical analysis suggests that the centralized wage bargaining system has restricted the adjustment of real wages to productivity differentials and contributed to higher unemployment rates in some states. Government transfers to households also appear to have adversely affected work incentives in high unemployment states by limiting participation in the labor force. The results suggest that growth in the relatively low-income and output states to some extent converged during the 1990s toward that in higher income states, but the initial differences in per capita income and output across the states largely remained. Since 1997, however, the catch-up effect on output growth has slowed down, and the impact on real per capita income growth across states of government transfers appears to have increased. The results also suggest that federal grants to the states did not have a significant impact on the relative output growth rates across states in the 1990s, and that the impact of skill-biased technological change on real per capita income and output growth is not clear.

A. Developments During the 1990s

4. Analysis of the disparities in real per capita output and disposable income suggests that Australian states can be divided into two groups: one group in which the real per capita output and income are higher than the national average (including New South Wales, Victoria, Western Australia, the Northern Territory, and the Australian Capital Territory (ACT)), and another group with output and income levels below the national average (comprising Queensland, South Australia, and Tasmania) (Figures 1 and 2 and Table 1),12 The cross-state dispersion in output and income variables has increased from about the mid-1990s, with the dispersion in output far higher than in disposable income (Figure 3).13 However, a closer examination of the disposable per capita income data indicates that the income dispersion is largely influenced by the income pattern in the ACT. Without the ACT, income dispersion across states is lower and more or less unchanged during the 1990s. At the same time, the gap between output and income dispersion has increased, suggesting that transfers have had a large impact in narrowing income inequalities. Output dispersion across states is only slightly lower excluding the ACT, and it rises after 1997. The output dispersion seems to have risen during a period when significant economic reforms in Australia began to take hold. Since the early 1990s, a number of microeconomic reforms in transport infrastructure, utilities industry, and telecommunications were implemented across Australia. Significant labor market reforms were also initiated in 1996.

Figure 1.
Figure 1.

Australia: Ratio of Real Per Captia State Disposable Income to the Average for Australia

(Percent)

Citation: IMF Staff Country Reports 2002, 215; 10.5089/9781451802047.002.A002

Figure 2.
Figure 2.

Australia: Ratio of Real Per Captia State Output to the Average for Australia

(Percent)

Citation: IMF Staff Country Reports 2002, 215; 10.5089/9781451802047.002.A002

Figure 3.
Figure 3.

Australia Dispersion of per Captia State Output and Household Income

Citation: IMF Staff Country Reports 2002, 215; 10.5089/9781451802047.002.A002

Table 1.

Australia: Summary Indicators on State Output and Income, 1990-2001

(In Percent)

article image
Sources: Australian Bureau of Statistics; and IMF staff estimates.

Real disposable income is estimated by applying gross state product deflators to income in current prices.

5. Large disparities also exist in regional unemployment rates and the median duration of unemployment across states (Figure 4 and (Table 2)). The unemployment rate and the median duration of unemployment are higher in the states with the higher income and output gaps. In 2001, South Australia and Tasmania had a median duration of about 25 weeks to 35 weeks, compared with the Australian average of 18 weeks (Figure 5). Additionally, long-term unemployment rates in the lower income states are significantly higher than the Australian average (Figure 6). The variability among states of the ratio of employment to population also follows a pattern similar to those for income and output for the two groups of states (Figure 7).

Figure 4.
Figure 4.

Australia Dispersion of Unemployment Indicators Across States

Citation: IMF Staff Country Reports 2002, 215; 10.5089/9781451802047.002.A002

Figure 5.
Figure 5.

Australia: Median Duration of Unemployment by State

(Weeks)

Citation: IMF Staff Country Reports 2002, 215; 10.5089/9781451802047.002.A002

Figure 6.
Figure 6.

Australia Long-Term Unemployment Rate by State

(Percent)

Citation: IMF Staff Country Reports 2002, 215; 10.5089/9781451802047.002.A002

Figure 7.
Figure 7.

Australia Employment to Population Ratio in Australian States

(Percent)

Citation: IMF Staff Country Reports 2002, 215; 10.5089/9781451802047.002.A002

Table 2.

Australia: Unemployment Indicators by State, 1990-2001

article image
Source: Australian Bureau of Statistics.

B. Factors Contributing to Regional Disparities

6. Centralized wage bargaining could be an important factor contributing to labor market rigidities in Australia. The Workplace Relations Act of 1996 has facilitated the transition of the industrial relations system from a centrally determined wage awards system to enterprise bargaining. In the current system, the role of awards has been restricted to setting a safety net of minimum wages and conditions. Award coverage has fallen dramatically from 68 percent in May 1990 to 23 percent in May 2000. Yet, award-based wage setting is still substantial—more so in the lower income states, where the share of employees subject to minimum wages is higher (Table 3). With a substantial proportion of wages, particularly at the low end of the pay scale, still based only on awards, employers continue to have less room to adjust for productivity differentials.14 Under centralized wage bargaining, like the award system, wages across states cannot reflect regional productivity differences, causing unemployment to be above average in those areas with below average productivity.

Table 3.

Australia: Method of Setting Pay, May 2000

(Percent of Employees)

article image
Source: Australian Bureau of Statistics Catalog No. 6306.

7. Government transfers per household are higher for states that have the largest output and unemployment gaps with the rest of Australia (Table 4).15 However, the figures for New South Wales and Victoria do not fit this pattern because government transfers to households in these states contain a relatively high proportion of pension payments. Government programs to provide income support may have unintended adverse economic consequences, particularly on incentives for lower-skilled workers. The current evidence is mixed in this regard. The Industry Commission (1993) notes that income support payments can undermine work incentives for those already prone to long-term unemployment. Also, uniformity in benefits across states suggests that people may have some incentive to migrate from low to high unemployment regions with lower costs of living.16 Studies by Debelle and Vickery (1999) and Lawson and Dwyer (2002), however, find that migration is an important channel for adjustment to labor market shocks, and that out migration has been characteristic of states with higher unemployment rates.

Table 4.

Australia: Ratio of Real Government Transfers Per Household to the Average for Australia

(Percent)

article image
Sources: Australian Bureau of Statistics; and IMF staff calculations.

8. The “equalization” aspect of the current arrangements for federal grants to the states can be viewed as potentially creating some disincentives for state governments to introduce changes designed to enhance growth prospects (Craig, 1997).17 A portion of the federal government grants are allocated to the states on the basis of each state’s ability to generate its own-source revenue in an attempt to at least partially equalize revenue across states, so that each state has sufficient resources to provide a minimum standard of public services. Hence, if a policy change would improve a state’s revenue-generating capacity, federal grants would decline, partially offsetting the revenue gain. Consequently, this grants arrangement may impede actions which would boost the real per capita income and output levels of the lower income states.18

C. Empirical Evidence

Labor market

9. In explaining the dispersion of unemployment across states, a pooled regression (with fixed effects) was estimated to assess the importance of the wage determination system and government transfers to households:

URi,t = βoi + β1 (PWGAP)i,(t-1) + β2 (TRANSFER)i,t + β3 URi,(t-1) + εi,t

UR is the unemployment rate in state i relative to the Australian average, PWGAP is the differential between productivity and real wages in state i relative to that in Australia, and TRANSFER is real government transfers per labor force participant in state i relative to the Australian average.19 While contemporaneous information on the government transfer variable is usually available to labor force participants in making their work-leisure decisions, the labor productivity-wage gap variable is likely to impact the unemployment rate with a lag; the lagged dependent variable controls for any persistence or hysteresis component in unemployment.

10. Results in (Table 5) show that the unemployment rate is higher when real wages in a state are high relative to productivity, which is indicative of the influence of wage-setting arrangements. However, the government transfers variable does not have the expected sign—a priori, one expects that as relative transfers to a state increase, the incentives to work in the state decline and cause higher unemployment. The counter-intuitive result in the equation estimated could be because transfers tend to lower work incentives and efforts to seek jobs, thus keeping people out of the labor force, which in turn serves to keep the unemployment rate down.

Table 5.

Australia: Estimation Results: Labor Market Equation

article image
Note: Results for pooled estimation with fixed effects; fixed effects are small and not presented here.

P-values for the null hypothesis of a coefficient equal to zero.

Estimation excludes Australian Capital Territory for which skill bias data are not available.

11. To allow for this possibility, an alternative equation was estimated regressing the productivity-wage gap and government transfers on the employment to population ratio. This equation shows that both the labor market and government transfer variables have the expected signs and are highly significant (see (Table 6)). The results suggest that states with above average government transfers tend to have lower employment to population ratios, implying that these transfers constrain work incentives and participation in the labor market.

Table 6.

Australia: Estimation Results: Income Growth Equation 1/ Dependent Variable: Per Household Disposable Income Growth

article image
Note: Results for pooled estimation with fixed effects; fixed effects are small and not presented here.

Estimation excludes Australian Capital Territory for which skill bias data are not available.

P-values for the null hypothesis of a coefficient equal to zero are within parentheses.

12. The employment-population ratio could also be higher in states with a higher concentration of skilled labor. The states that adopted the technological advancements of the 1990s more rapidly than others are likely to have experienced a higher demand for skilled labor. Adding an index for skill levels by state relative to the average for Australia to the equation confirms that the employment-population ratio rises as the relative skill level of a state increases relative to the national average.

Income and Output Growth

13. Empirical analysis of the possible causes for the lack of convergence in regional incomes and outputs is based on the following equations:

Incomegrowthequation:(DIG)i,t=β0i+β1(DI)i,(t1)+β2(TRANSFER)i,t+β3(SKILL)i.t+ϵi,t
Outputgrowthequation:(GSPG)i,t=β0i+β1(GSP)i,(t1)+β2(LG)i,t+β3(KG)i.t+β4TRANSFERi,t+β5(GPP)i,t+β6(SKILL)i,t+ϵi,t

where DIG is growth in real per capita disposable income in state i relative to the average for Australia. TRANSFER is as defined above. SKILL is a proxy for skill biased growth—as human capital levels rise in certain regions, those regions tend to grow more rapidly over the long run. GSPG is growth in real per capita gross state product in state i relative to the average for Australia. LG is growth in labor force. KG is growth in real per capita capital (or real per capita investment). GPP is the ratio of real per capita general purpose payment grants to state i to the Australian average.20 The lagged variables DT and GSP represent the “catch-up” effect in real per capita income and output growth of state i to the Australian average. The Commonwealth’s general purpose grants to states and household transfer variables are introduced to empirically examine their impact on output growth.21

14. Results for the income and output equations in Tables 6 and 7 show that there is convergence in growth rates although most of the initial level differences between states are preserved (as indicated by the dispersion of these two variables across states). States with lower income and output tend to grow faster as suggested by the negative estimate of the lagged income and output variables. For the output equation, the catch-up coefficient in a regression over the sub-sample period 1997-2001 is about half the size of the coefficient for the estimation over the sub-sample period 1991-1996, which is consistent with the increase in output dispersion recorded since 1997.22

Table 7.

Australia: Estimation Results: Output Growth Equation 1/ Dependent Variable: Per Capita Gross State Product Growth

article image
Note: Results for pooled estimation with fixed effects; fixed effects are small and not presented here.

Estimation excludes Australian Capital Territory for which skill bias data are not available.

P-values for the null hypothesis of a coefficient equal to zero are within parentheses.

15. The results also suggest that government transfers have helped to reduce income disparities across states. However, it appears that these transfers may have also constrained or delayed convergence in output growth across states, which could partly be reflecting its adverse effect on incentives to work.

16. Relative labor force growth and capital accumulation explain most of the states’ output growth. However, labor turns out to be insignificant and having the wrong sign in the regression for the sub-sample period 1997-2001. This could be because technological advances in the second half of the 1990s, which have contributed to substantial increases in productivity, are not accounted for.23 However, the relative skill bias coefficient in the regression over the 1997-2001 sub-sample is positive, although insignificant, suggesting that a higher skill bias (which is also indicative of technological advancement in that state) implies higher output growth. Finally, the results also suggest that the current federal-state funding arrangement may not be significantly contributing to creating perverse incentives among the states for output growth. The general purpose payments coefficient is negative suggesting that it adversely affected output growth, although the coefficients are statistically insignificant in two out of the three regressions.24

References

  • Cashin, Paul, 1995, “Economic Growth and Convergence Across Seven Colonies of Australasia: 1861-1991”, The Economic Record, vol.71, no. 213, 13244.

    • Search Google Scholar
    • Export Citation
  • Cashin, Paul and Loris Strappazzon, 1998, “Disparities in Australian Regional Incomes: Are They Widening or Narrowing?”, The Australian Economic Review, vol.31, no. 1, pp. 326

    • Search Google Scholar
    • Export Citation
  • Craig, Jon, 1997, “Australia”, in Fiscal Federalism in Theory and Practice, (ed) T. Ter-Minassian International Monetary Fund, Washington, D.C.

    • Search Google Scholar
    • Export Citation
  • Debelle, Guy and J. Vickery, 1999, “Labor Market Adjustment: Evidence on Interstate Labor Mobility”, The Australian Economic Review, vol.32, no. 3, pp. 24963.

    • Search Google Scholar
    • Export Citation
  • Garnaut, Ross and Vince FitzGerald, 2001, Background Paper, Review of Commonwealth-State Funding, Melbourne.

  • Garnaut, Ross and Vince FitzGerald, 2002, Interim Report, Review of Commonwealth-State Funding, Melbourne.

  • Harris, Percy and David Harris, 1992, “Interstate Differences in Economic Growth Rates in Australia, 1953-54 to 1990-91”, Economic Analysis and Policy, vol.22, no. 2, pp. 12948.

    • Search Google Scholar
    • Export Citation
  • “Impediments to Regional Industry”, 1993, Industry Commission Report No. 35, Government Publishing Service, Canberra.

  • Lawson, Jeremy and Jacqueline Dwyer, 2002, “Labor Market Adjustment in Regional Australia”, Reserve Bank of Australia Research Discussion Paper 2002-04, Sydney.

    • Search Google Scholar
    • Export Citation
  • Neri, Frank, 1998, “The Economic Performance of the States and Territories of Australia: 1861-1992”, The Economic Record, vol.74, no. 225, pp. 10220.

    • Search Google Scholar
    • Export Citation

ANNEX: Data Definitions

Data set: Pooled data for all Australian states and territories for the period 1990-2001. For regressions with the skill bias variable, the Australian Capital Territory has been excluded due to the absence of skill data. All nominal variables for states have been converted to real basis using each state’s GSP deflator.

Data Sources: Australian Bureau of Statistics, Department of Employment and Workplace Relations, and the Australian Treasury.

UR = (unemployment rate in state i - unemployment rate in Australia).

PWGAP = In [(productivity in state i - real wage in state i) divided by (productivity in Australia - real wage in Australia], where productivity is output per hour and real wages are average weekly earnings deflated by gross state product deflators.

TRANSFER = In [(real per capita government transfers to state i) divided by (the national average for real per capita government transfer)], where government transfer is the sum of social assistance and workers compensation from the ABS gross household income data. For the unemployment equation, transfers are normalized per labor force participant.

SKILL = In [(ratio of skill vacancies in state i to total vacancies in state i) divided by (ratio of skill vacancies in Australia to total vacancies in Australia)].

DI = In [(real per capita disposable income in state i) divided by (real per capita disposable income in Australia)].

DIG = [(1 + real per capita disposable income growth in state i) divided by (1 + real per capita disposable income growth in Australia)]-1.

GSP = In [(real per capita gross state product in state i) divided by (real per capita GDP in Australia)].

GSPG = [(1 + real per capita GSP growth in state i) divided by (1 + real per capita GDP growth in Australia)]-1.

LG = [(1 + labor force growth in state i) divided by (1 + labor force growth in Australia)]-1.

KG = real per capita investment in state i divided by real per capita investment in Australia.

GPP = In [(real per capita general purpose payments to state i) divided by (average real per capita general purpose payments in Australia)].

9

Prepared by Uma Ramakrishnan (ext. 35413) and Martin Cerisola (ext. 38314), who are available to answer questions.

10

In the paper, the term “states” is used to refer to the six Australian states and two territories.

11

Other factors not considered here may also have contributed to the regional disparities. For example, industry composition and diversification of a state relative to the national economy, the proximity of a region to product and factor markets, and physical and cultural amenities of a specific region could influence regional employment and output (Lawson and Dwyer, 2002).

12

While there are many studies pointing to this issue (Harris and Harris (1992), Cashin (1995), Cashin and Strappazzon (1998), and Neri (1998)), only a few studies (including Debelle and Vickery (1999) and Lawson and Dwyer (2002)) consider possible reasons for these disparities, and they focus mostly on labor market outcomes.

13

Dispersion is measured as the coefficient of variation, which is the standard deviation across the states divided by the mean.

14

The actual proportion of employees subject to only awards in setting wages varies by sector, industry, and enterprise size. For instance, 65 percent of the recreational industry is covered by awards only, compared with less than 6 percent of mining and finance and insurance sectors. Forty-two percent of clerical, sales, and service workers are covered by awards, compared with 3 percent of managers and administrators.

15

The data on government transfers include various social assistance payments, such as unemployment assistance, old age pension, and health allowances.

16

According to the Commission’s report, the Social Security Act provides for discouraging migration to areas with lower employment prospects, but enforcement is characterized as being lax. While the law was intended to apply to all persons who reduced their employment prospects by migrating, the penalty is applied only to those already receiving benefits.

17

A study has also been commissioned by the governments of New South Wales, Victoria, and Western Australia to review the methods of allocating Commonwealth grants to the states and territories, and the appropriateness of the outcomes. See Garnaut and FitzGerald (2001 and 2002).

18

This situation is roughly comparable to that faced by welfare recipients who confront high effective marginal income tax rates owing to the withdrawal of benefits when they return to work.

19

The Annex provides detailed data definitions.

20

Note that for the income equation, the income and government transfer variables are normalized per household because the disposable income data are on a per household basis.

21

Government general purpose grants to states are not added to the income equation since these grants are inter-governmental budgetary transfers and not direct payments to households.

22

The sample split in 1997 partly reflects the intention of assessing the impact of the labor market reforms introduced around that time.

23

For the impact of technology on labor productivity, see “Is Australia a New Economy?” in Australia: Selected Issues, IMF Staff Country Report No. 01/55, April 2001.

24

The impact of grants on growth could be narrowed down by estimating the impact of real per capita National Competition Policy Payments (NCPP) to the states. The results of the estimation of the output equation with NCPP (for the sub-sample from 1997-2001) does not alter the above story—the NCPP remains an insignificant determinant of regional growth differentials.

Australia: Selected Issues
Author: International Monetary Fund
  • View in gallery

    Australia: Ratio of Real Per Captia State Disposable Income to the Average for Australia

    (Percent)

  • View in gallery

    Australia: Ratio of Real Per Captia State Output to the Average for Australia

    (Percent)

  • View in gallery

    Australia Dispersion of per Captia State Output and Household Income

  • View in gallery

    Australia Dispersion of Unemployment Indicators Across States

  • View in gallery

    Australia: Median Duration of Unemployment by State

    (Weeks)

  • View in gallery

    Australia Long-Term Unemployment Rate by State

    (Percent)

  • View in gallery

    Australia Employment to Population Ratio in Australian States

    (Percent)