1                This working paper describes the assessment of the effect on State expenses of unavoidable differences — to the extent these are due to circumstances outside a State's control — in the price of labour, office accommodation and electricity.  These differences are assessed as input costs disability factors.  They are intended to influence a State's share of GST revenue. 

2                Differences in input costs due to States' policy choices are avoidable and are not intended to influence a State's share of GST revenue. 

3                More detailed and comprehensive discussion on the development of the assessment method is in Volume 7 of the 2004 Review Working Papers and the Commission's various discussion papers[1].  Only a summary of the concepts and methods are presented here. 

General

4                The inputs in paragraph 1 are used in virtually all State services.  Input cost disability factors were applied to 29 expenses categories, including Depreciation, but not to those categories assessed by equal per capita or actual per capita methods.  The factors capture differences in the cost per unit of inputs, and are additional to differences in the demand per person for services. 

5                Input cost factors for labour, accommodation, and electricity were assessed separately.  The factors for labour were of most importance to GST revenue distribution. 

6                Box 1 provides a summary of how the three factors are combined.  Detailed discussion of each factor follows. 

Box 1:  Combining individual input costs For each expense category where input costs are assessed: (i) the wages factors were multiplied by the wages 'weights', being the average share of wages in the total average cost of the service.  The wages weights varied, depending on the importance of wages in service delivery cost for particular services.  They were estimated at 20, 35, 50, 60, 62.5, 70 or 80 per cent of expenses (for urban transit, the wages factor were used without 'weights'); (ii) the accommodation factors were multiplied by the accommodation 'weight', being the average share of accommodation costs in the total average of services.  They were estimated at 2 per cent; (iii) the electricity factors were multiplied by the electricity 'weight', being the average cost share of electricity in the total average cost of services.  They were estimated at 0.5 per cent; and (iv) the overall input cost factors were derived by summing the contributions from (i), (ii) and (iii) above.

 

7                The factors are updated annually using the latest available mean resident populations and other data for each of the reference years.  The cost weights are not updated between reviews. 

Wages and salaries

Why do wages and salaries differ

8                Wages and salaries are a large component of the costs of government services.  As noted, the wages and salaries input cost factor (called the wages factor) applied to between 20 and 100 per cent of average costs in 29 expense categories.  Thus, even when interstate differences in the assessed cost factors were small, their impacts on the distribution of GST revenue distribution were large. 

9                The main issue in the assessment of the wages factor was whether the level of compensation of similar public sector employees (defined as employees with identical labour productivity characteristics) differed according to their location and therefore for a State as a whole (the 'location' effect).  If interstate differences existed and were unavoidable — that is, they were not due to the policy choices of the State governments — they constituted an expense advantage/disadvantage.  A State facing, say, higher costs of labour would incur higher expenses to provide an average level of service. 

10            Actual observed wages in the States did not allow the measurement of cost differences because they reflected the differences in productivity characteristics or working conditions of individuals or groups in the State, as well as the differences in wage rates.  None of the standard data series on aggregate State wages compiled by the ABS and others related to similar employees.

11            Hence, the Commission undertook analysis of wages data for individual employees in the private sector to separate the 'location' effects from labour productivity factors[2], based on a widely used standard econometric model.  By accounting for the productivity effects, location effects could be indirectly estimated[3]. 

12            The use of the private sector as the benchmark for analysing public sector wages is the usual route in economic literature[4].  The main reason for using it in Commission analysis was that, aside from the greater policy contamination of public sector data, the cost base in each State was defined by the greater size of the private sector — accounting for around 75 to 80 per cent of employment — and the opportunities for cross-location trading. 

13            The analysis found different levels of wages across States because of 'location' effects after adjusting for productivity differences.  Differences arose because the conventional cost‑of‑living for individuals differed by location based on:

·                the cost of a representative basket of consumption of goods and services;

·                cost of housing (reflecting an economic cost rather than cash outlays of the value of housing services); and

·                the 'intangible' and subjective value placed on location specific amenity (characteristics such as congestion, climate, isolation, big city/small city and proximity to friends/relatives). 

14            Moreover, the analysis indicated broad conformity in the location effects for wage rates paid in the public and private sectors, including occupations concentrated in the public sector. 

15            But, why do differences in conventional cost-of-living measures arise between locations?  There would, in fact, be no differences if:

·                there were no restrictions on trade in goods and services;

·                all goods and services could be freely traded across locations; and

·                there were no frictional costs (for example, freight costs). 

16            However, not all goods and services can be freely traded between locations, for example, taxi services, hairdressing salons and repair services. 

17            Housing is a major non-tradeable service because mostly employees have to locate close to their job.  Because housing competes with alternative economic uses of land, housing costs could be higher if land were relatively scarce. 

18            In summary, the main drivers of differential location effects are differences in the nature and extent of non-tradeable services and 'intangibles'.  These differences exist, even in an otherwise competitive national labour market[5], because potential employees respond to wage signals in deciding whether there is any benefit in relocating, that is, whether the differences in wages exceeds differences in the broader cost-of-living. 

The 2004 Review approach

19            The assessment was done using an econometric model that reflected the conceptual issues outlined above.  The analysis used data from the ABS's 1997 and 2001 Surveys for Education and Training Experience (SET).  The Confidentialised Unit Record File (CURF) data covered all persons who had a wage or salary job in the 12 months prior to the surveys.  The data included details by sector of wage levels, hours worked, location, occupation, industry, education and other characteristics for sampled individuals. 

20            The model sought to explain interstate differences in wage levels in terms of differences in:

·                characteristics of employees that are usually associated with productivity (such as type and level of education, experience, occupation, industry); and

·                location specific effects, such as cost of living and especially housing costs[6].

21            In other words, the model estimates the interstate differences in wages paid to comparable employees and attributes them to differences in location specific effects such as cost of living. 

22            The regression results indicated that, for both the private and the public sectors, there was consistent evidence of 'location effects' across the States after excluding differences in employee skills, occupations, industry and other productive characteristics. 

23            The Commission subjected the estimated location effects to reality testing.  That testing and detailed analysis of the concepts, results and supporting evidence all indicated the model results were robust.  However, as discussed later, the estimates were modified to recognise special features of labour markets in some smaller States. 

24            Essentially, the model regressed, separately for the 1997 and 2001 data, the logarithm of earnings (w_t (t = 1997, 2001)) on measurable labour market influences (X_x)in the datasets and the State of employment as a location dummy (DS). 

                        ln (w_t)                  

                 =     X_ijt B_it+ ε_t                           ………………...…………………………(A)

                 =     I_t                                         (I: Fixed Intercept)

                        + Σ_iDS_it*I_it                            (DS_i represents dummy variables for each State i;
I_i represents the location effect for each state)

                        + Σ_xX_xtb_xt                                   (X_x: represents a set of measures of individuals' labour market characteristics such as type and level of education (EDU), field of education (FEDU), experience (EXP) and square of experience (EXPSQ), employment history, etc.; b_xrepresents returns to such characteristics)

                        + ε_t                                                  (stochastic error)

25            The regression results indicated that there was consistent evidence of 'locational effects' across the States after adjusting for differences in employee skills, occupations and other productive characteristics, for both the private and the public sectors.  That is, the values of I_it for the States were generally different from each other. 

26            The Commission subjected the location effects derived from the regression model to substantial reality testing in which it:

·                compared the results of its model with results from other studies of interstate wage differentials;

·                examined widely used ABS data on wages (especially average weekly earnings and the wage cost index) for evidence of divergent trends in wage patterns across States;

·                compared estimates  of interstate differences in private and public sector wages based on average weekly earnings data and the raw SET data at both the aggregate level and by occupation; and

·                examined the main causes of the location effect — especially the changes in house prices over time in the State capital cities. 

27            A detailed analysis of the concepts, results, supporting evidence and substantial reality testing all indicated that the model results were robust.  However, the model results required modification to account for special features of labour markets in some smaller States, which were not captured in the modelling. 

·                The location effect for Tasmania appeared understated because of the low migration of labour into Tasmania's private sector and because the pattern of economic activity was broadly comparable to regional areas of other States.  As a matter of judgment, the Commission set the Tasmanian location effect at the level of the second lowest State (Queensland). 

·                The location effect for the ACT underestimated the effect of the Commonwealth sector on wage levels, especially in recent years.  Using judgment, the Commission set the ACT location effect at the average of the calculated effects for the ACT and New South Wales. 

·                The raw wages data from SET were consistently higher than those from average weekly earnings, especially in 2001.  To deal with this and the inevitable uncertainty of model results, the Commission discounted the estimated location effects for 1997 by 5 per cent and those for 2001 by 15 per cent. 

Developments in the 2007 Update

28            In the 2004 review, the Commission committed to re-estimating the location effects when the data from the 2005 survey became available.  Those data were available from the ABS, and were similar in structure and depth to the 2001 survey. 

29            The wages data from the 2005 survey related to May to August of 2005.  For all practical purposes, the same model was applied to the 2005 survey data[7]. 

30            Table 1 summarises the modelled location effects for 1997, 2001 and 2005. 

Table 1                        Modelled location effects for 1997, 2001 and 2005

	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aust
1997
Location effects	0.015	0.010	-0.027	-0.016	-0.015	-0.056	0.040	0.094	0.000
Standard error(b)	0.005	0.006	0.007	0.010	0.011	0.022	0.027	0.037	0.000
t-value	3.32	1.75	-4.02	-1.62	-1.35	-2.59	1.46	2.57	0.000
Level of Significance	0.001	0.080	0.000	0.104	0.177	0.010	0.145	0.010	0.000
2001
Location effects	0.050	-0.005	-0.048	-0.021	-0.046	-0.079	0.019	0.043	0.000
Standard error	0.005	0.006	0.007	0.010	0.011	0.021	0.027	0.042	0.000
t-value	11.03	-0.90	-7.04	-2.18	-4.09	-3.80	0.69	1.02	0.000
Level of Significance	0.000	0.369	0.000	0.030	0.000	0.000	0.490	0.306	0.000
2005
Location effects	0.037	-0.004	-0.035	-0.020	-0.024	-0.069	0.031	0.032	0.000
Standard error	0.005	0.006	0.008	0.011	0.014	0.025	0.031	0.044	0.000
t-value	7.10	-0.57	-4.67	-1.86	-1.77	-2.79	0.98	0.76	0.000
Level of Significance	0.000	0.566	0.000	0.063	0.076	0.005	0.327	0.448	0.000
Per cent point differences between 2001 and 2005 location effects
	-0.013	0.001	0.013	0.001	0.022	0.010	0.012	-0.011	0.000

31            A staff discussion paper 2006/05 (see Appendix A to this paper) was sent to the States for comments. States' responses specific to the update issues, including adjustments if any to the modelled results, follows.  State comments of a more general nature are included in Appendix B to this paper. 

32            New South Wales said that the model results provided an appropriate base for updating the wages factors.  They however argued that there was no reason to discount the modelled location effects, because the results accorded with reality and the data were of reliable quality. 

33            Victoria argued for removing the adjustments for:

·                Tasmania because of marked increase in net interstate migration to Tasmania since 2001-02; and

·                the ACT because the modelled location effects for it over the long term were similar to those for New South Wales, and an adjustment if and when the ACT's modelled location effect was greater would not make sense.  

34            However, they argued for retaining the discount to all States location effects because in their view the interstate variance in the SET data was now similar to that in the AWE data — the basis for the discount — because of higher variance in the AWE data.  

35            Queensland said that there was a high level of uncertainty with the estimated location effects for 2005, suggesting that one off factors in the 2001 location effects had exaggerated them.  They urged the Commission to take a conservative approach, indicating particularly that it would be appropriate for the Commission to recognise the full reduction in the location effects — that is, not apply any discounts to the wage differentials evident from the 2005 SET data. 

36            South Australia indicated that the movements in the modelled results from the 2001 to the 2005 results were likely reflective of a move back to a more plausible position for the State.  They argued for retaining the general discount to the location effects. 

37            They also expected the modelled location effects for the reference years 2001 to 2005 to be interpolated using a straight line, as was done in the 2004 Review between 1997 and 2001, and said that they would consider any variation to be a method change.

38            Western Australiadid not regard the results of the model as plausible for 2005, pointing to the recent above average wages growth in the State.  They proposed that the Commission estimate the location effect by combining the results of the 2005 SET and EEBTUM data, changing Western Australia's location effect from -0.020 to -0.014. 

39            Western Australia also raised a number of technical questions about the wage regression model used by Commission staff.  Specifically, they said that:

·                the wage regression model has not been tested for robustness — cross-effects between variables (except those involving the worker's gender) have been ignored, and there was no analysis of correlation between location and non location variables;

·                the values of some of the coefficients produced by the regression analysis seem implausible (e.g. rate of pay is a decreasing function of work experience for values of work experience greater than about 10 years); and

·                for most States, including Western Australia, the 2005 location effects were not statistically significant. 

40            Tasmania urged the Commission to be conservative.  They noted that while the modelled location effects were consistent with past evidence as to relatively high/low wage states, those effects showed significant shifts for at least some States since 2001. 

41            They also said that there was no reason to believe that the characteristics of Tasmania had materially changed since 2001 and asked the Commission to set the Tasmania's location effect at the level of the second lowest state (which is still Queensland).  They asked the Commission to continue to discount because of the sample based nature of the data. 

42            The ACT supported updating the wages input cost factors based on the 2005 SET survey, indicating that the 'reality checks' were rational.  They said that special feature adjustments from the 2004 Review for Tasmania and the ACT should continue. 

43            The Northern Territory had no comments. 

44            The technical issues raised by Western Australia. The specification of the model used for the regression analysis was very similar to that used in similar studies by other researchers over many years.  Those researchers developed the specification over time on the basis of strong conceptual arguments or other research findings.  The model used was also refereed by an external expert. 

45            Questions similar to those raised by Western Australia were raised and considered during the 2004 Review.  At that time it was considered unnecessary to allow for possible cross-effects between variables, except for gender where known differences in labour force experience for females were thought to be relevant.  Compelling evidence of conceptual or methodological flaws were necessary for consideration of whether the model should be adapted to allow for specific cross-effects.  In this context, Western Australia had not provided any conceptual argument.  Besides, these were matters for the review rather than the update. 

46            Western Australia's point that 'rate of pay is a decreasing function of work experience for values of work experience greater than about 10 years' did not appear factually correct.  It was a more plausible 27 years (or more if a fuller consideration is given). 

47            The decision of the Commission to apply a 15 per cent general discount to the modelled location effects in the 2004 Review and the preliminary decision to continue that discount were in part intended to deal with the issue of standard errors raised by Western Australia. 

48            Comments on the adjustments.  Not all States commented on each adjustment, and the comments were not unexpected.  Continuing the adjustments made in the 2004 Review, including continuing the general 15 per cent, was broadly but not unanimously supported. 

49            The Tasmanian adjustment. In Tasmania, while recent house price increases were well above average, wages growth was not.  The effects of migration on the labour force did not appear dramatic — while there were migration gains to the State, these appeared to be in the older age groups with limited direct impact on wages.  Moreover, the level of migration declined in 2005-06 with increases in house prices.  In these respects the situation was similar to that in 2001 and it was likely that the modelled private sector location effect continued to understate the policy neutral wage levels that might have faced the State government. 

50            However, whether the Tasmanian location effect should continue to be adjusted to the level of the second lowest (Queensland) was less clear.  In this regard, comparisons of average weekly earnings for the public sector in the two States were interesting, even though they did not strictly compare like with like.  At or about the time of the 2001 SET, average public sector wage levels in Queensland and Tasmania were reasonably comparable — about 4 per cent below the national average.  Since then, however, the average public sector wages in the two States diverged.  Public sector wages as a proportion of national public sector wages were mostly higher in Queensland than in Tasmania.  The difference was in the order of 1 to 2 percentage points. 

51            Taken at face value, this suggested that a lower adjustment to the modelled location effect for Tasmania could be appropriate.  In 2001 the modelled location effect for Tasmania was adjusted from -7.9 per cent to -4.8 per cent.  Its modelled effect for 2005 was -6.9 per cent and that for Queensland is -3.5 per cent.  On this basis, retaining Tasmania at the adjusted level for 2001 represented a 2.1 percentage point adjustment to its location effect which appeared reasonable. 

52            The ACT adjustment. At the time of the 2004 Review, the Commission considered that the estimated location effect based on the ACT's private sector understated the effects on the wage levels faced by the ACT Government because of higher wages in the Commonwealth sector.  It adjusted the ACT's location effect to the average of that in New South Wales and the ACT.  That adjustment moved the ACT location effect from 1.9 per cent to 3.4 per cent. 

53            Following the same adjustment process for the 2005 results would have resulted in very little change because the New South Wales location effect was now 3.7 per cent compared to the ACT's 3.1 per cent.  The average would have been 3.4 per cent.  The adjustment did not appear justified on materiality grounds any longer. 

54            The general discount.  Table 2 compares the raw private sector wages from Average Weekly Earnings (AWE) and SET for 2001 and 2005. 

Table 2            Relative full-time adult ordinary time average weekly earnings, private sector, 2001 and 2005

Year	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aus
2001
AWE	1.07300	0.97156	0.92108	1.02448	0.92694	0.87100	1.04440	0.95831	1.00000
SET	1.08977	1.03024	0.87582	0.97238	0.89065	0.86237	0.97976	0.97576	1.00000
Differences between 2001 and 2005	0.017	0.059	-0.045	-0.052	-0.036	-0.009	-0.065	0.017	0.000
2005
AWE	1.06104	0.99111	0.91917	1.04682	0.90280	0.85364	1.06516	1.00033	1.00000
SET	1.05849	1.01710	0.95042	0.95553	0.91137	0.83643	0.98645	1.12000	1.00000
Differences between 2001 and 2005	-0.003	0.026	0.031	-0.091	0.009	-0.017	-0.079	0.120	0.000

55            In 2005, the raw wages relativities between AWE and SET for the larger States in particular appeared more comparable.  Those for some of the smaller States did not. 

56            However, general statistical uncertainties such as with the standard errors remained.  These uncertainties were sufficient to justify continuing to apply the 15 per cent discount to the 2005 model results. 

57            In response to South Australia, the location effects for years between 2001 and 2005 using a straight line interpolation of the results for 2001 and 2005 continued, as was done in the 2004 Review for 1997 to 2001. 

58            The main premise for considering a case for adjusting the modelled results for Western Australia was that:

·                Western Australia was experiencing boom conditions in recent times; and

·                the supporting evidence for changes in Western Australia in 2005-06 — which were largely not captured in the SET data — were markedly different from those in 2004-05. 

59            The Commission re-examined the supporting economic indicators for Western Australia by extending those to 2005-06. 

60            Average weekly earnings. Table 3 shows average weekly earnings for each State by sector.  Two specific cautions apply.  

·                The public sector figures for the ACT include the Commonwealth sector; and

·                Comparisons over time and across States do not control for compositional changes. 

61            Between 2000-01 and 2004-05 the increase in earnings for the private sector of Western Australia was the third highest.  In the public sector, the increase between 2000-01 and 2004-05 was the lowest of all States. 

62            However, the change between 2004-05 and 2005-06 for Western Australia was the second highest for all States for the private sector, but still the lowest of all States for the public sector.  Thus while there were above average increases in Western Australia's private sector average weekly earnings, especially in the most recent year, they had not yet flowed through to the public sector.  

Table 3            Full-time adult ordinary time earnings, original, by sector

Year	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aus
	$	$	$	$	$	$	$	$	$
Private
2000-01	829.6	755.9	725.8	806.1	723.9	693.1	835.0	765.4	779.8
2004-05	1004.1	964.1	879.6	992.0	867.0	827.3	978.5	948.8	956.6
Differences between 2000-01 and 2004-05 (%)	21.0	27.5	21.2	23.1	19.8	19.4	17.2	24.0	22.7
2005-06	1057.5	993.5	921.6	1058.9	916.6	861.3	1105.0	987.6	1002.6
Differences between 2004-05 and 2005-06 (%)	5.3	3.0	4.8	6.7	5.7	4.1	12.9	4.1	4.8
Public
2000-01	917.6	936.4	851.7	876.2	895.3	863.8	1000.8	888.4	905.3
2004-05	1082.6	1111.7	1046.0	1024.5	1056.8	1019.9	1210.1	1066.9	1078.9
Differences between 2000-01 and 2004-05 (%)	18.0	18.7	22.8	16.9	18.0	18.1	20.9	20.1	19.2
2005-06	1163.8	1156.1	1090.6	1062.7	1106.9	1067.3	1282.5	1114.0	1136.2
Differences between 2004-05 and 2005-06 (%)	7.5	4.0	4.3	3.7	4.7	4.6	6.0	4.4	5.3

Source: Average Weekly Earnings, 6302.0, May 2006, ABS. 

63            Labour price index.  Table 4 shows, by State, the labour price index for the private sector.  This index was corrected for changes in the composition of the labour force over time within each State, but not for differences in composition across States.  It was thus in part closer in concept to the modelled location effect than AWE — the modelled location effect sought to compare like with like over time and across States. 

Table 4            Labour Price Index by State, total hourly rates of pay excluding bonuses, private sector, 2000-01 to 2004-05

Year	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aus
2000-01	90.7	90.5	90.9	90.6	89.9	91.1	90.7	91.8	90.6
2004-05	103.4	103.8	103.7	104.3	103.3	103.8	103.4	103.8	103.7
Differences between 2000-01 and 2004-05 (%)	14.0	14.7	14.0	15.0	14.9	13.9	14.0	13.1	14.4
2005-06	107.2	107.9	108.4	109.2	106.9	107.9	107.3	107.9	107.8
Differences between 2004-05 and 2005-06 (%)	3.7	3.9	4.5	4.7	3.5	4.0	3.8	3.9	4.0

Source: Labour Price Index, March 2006.  2003-04 = 100 for all States. 

64            It supported above average underlying wage pressure in Western Australia in 2005-06. 

65            Residential land values. Table 5 shows total residential land value in each State in dollars and changes over time.  It shows that, over four years to 2004-05, the change in total land values in Western Australia was below average.  This was consistent with the modelled location effects.  However, the change between 2004-05 and 2005-06 was the second highest of all States and well above the average. 

Table 5            Total residential land values, 2000-01 to 2005-06

Year	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aus
	$m	$m	$m	$m	$m	$m	$m	$m	$m
Total residential land value
2000-01	386 687	233 647	113 585	78 645	38 818	6 791	11 868	4 789	87 4830
2004-05	688 611	414 201	277 254	137 415	82 006	11 657	27 503	6 029	164 4676
Differences between 2000-01 and 2004-05 (%)	78.1	77.3	144.1	74.7	111.3	71.7	131.7	25.9	88.0
2005-06	730 477	435 728	289 456	172 526	94 369	17 060	27 835	6 357	177 3808
Differences between 2004-05 and 2005-06 (%)	6.1	5.2	4.4	25.6	15.1	46.4	1.2	5.4	7.9

Source: State data returns. 

66            House prices.  Table 6 shows the ABS price index for established homes in the capital cities.  This series tracks changes in the price of the same (notional) representative house in each city.  It was less prone to compositional effects than some other data series. 

67            Between March 2002 (when the series began) and 2004-05, Queensland, South Australia and Tasmania experienced higher changes than Western Australia.  Between 2004-05 and September 2006, the change for Western Australia was much greater. 

Table 6            Price index of established homes in the capital cities (2003-04 = 100)

	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	All  capitals
Mar-2002	75.9	79.3	61.9	75.5	69.7	56.3	66.5	81.6	74.3
2004-05	96.1	101.9	104.2	114.4	106.5	111.8	99.9	115.9	101.2
Change (%)	26.6	28.5	68.4	51.5	52.7	98.5	50.2	42.0	36.2
2005-06	93.2	105.9	108.1	146.1	111.4	120.4	102.7	138.2	104.9
Change from 2004-05 (%)	-3.1	4.0	3.7	27.7	4.6	7.7	2.8	19.2	3.6
Sep-2006	94.0	111.2	112.5	186.3	115.1	125.6	110.7	150.7	111.4
Change from 2005-06 (%)	0.9	5.0	4.0	27.5	3.4	4.3	7.8	9.1	6.2

(a)        ABS 6416.0 June and September Quarters 2006.   The ABS' new house price series started only in Mar 2002. 

68            Table 7 shows changes in the median price of established homes in the capital cities.  This series is not based on the same representative house in each city, and is affected by compositional effects.  Within that limitation, Western Australia experienced between 2001 and 2005 changes similar to other small States (except Hobart), but between June 2005 and June 2006 the change for Western Australia has been much greater than for any other State. 

Table 7            Median house prices in capital cities

Quarter and Year	NSW	Vic	Qld	WA	SA	Tas	ACT	NT
	$ 000	$ 000	$ 000	$ 000	$ 000	$ 000	$ 000	$ 000
June 2001	316.0	291.0	180.0	165.7	148.2	120.3	203.0	187.0
June 2005	528.0	360.0	312.5	295.0	275.0	260.0	352.5	279.8
Changes between June 2001 and June 2005 (%)	67.1	23.7	73.6	78.0	85.6	116.1	73.6	49.6
June 2006	523.0	375.0	326.0	395.0	286.5	277.0	380.0	350.0
Changes between June 2005 and June 2006 (%)	-0.9	4.2	4.3	33.9	4.2	6.5	7.8	25.1

Source:  Real Estate Institute of Australia. 

69            Looking across the indicators, there was in 2005-06 a departure from the previous years.  This was consistent with changes in other indicators for Western Australia — growth in business investment, population, engineering construction activities and commodity prices — all of which showed marked increases in 2005-06. 

70            Recognising that the SET data did not capture changes in relative wages in 2005-06, there was a conceptual case for adjusting Western Australia's location effect for 2005-06. 

71            However, there were some concerns. 

·                The strength in the private sector wage growth had not yet flown through to the public sector.

·                Any adjustment made this year would have been small.  For example, the above average movement in the labour price and the Western Australian suggestion of combining the results of applying the model to SET and EEBTUM data implied an adjustment for 2005-06 that was in the vicinity of half a percentage point. 

·                The 2006 EEBTUM data for August 2006, which would allow modelling similar to that based on the SET data, would become available in the second quarter of 2007.  Using these data together with a longer time span of other data (such as labour price index and average earnings) would provide a better guide to an adjustment if any. 

72            Commission decision.  The Commission decided :

·                to use the modelled location effects for 2005 based on 2005 SET data as the basis for calculating wages input cost factors for the 2007 Update;

·                not to make a specific adjustment to the modelled location effect for Western Australia for the 2007 Update, because the movements in private sector wages in Western Australia in 2005-06 shown in average weekly earnings and labour price index data were not reflected in similar data for the public sector and were not comparable across States;

·                to reconsider the location effects in the 2008 update when EEBTUM data for 2005-06 and other indicators would allow the location effects to be updated more reliably;

·                to set the 2005 Tasmanian location effect at the same level as its 2001 adjusted location effect because recent movements in wages in Tasmania, especially in the public sector, were not as strong as in Queensland;

·                not to adjust the 2005 modelled location effect for the ACT because it was immaterial; and

·                to discount the modelled location effects of all States by 15 per cent as was done in 2001. 

2007 Update factors

73            The final factors, shown in Table 8, were calculated using the same steps as for the 2004 review. 

·         The raw location effects (A) for 1997, 2001 and 2005 were adjusted for the Commission's decisions for Tasmania and rescaled (adjusted location effects (B)). 

·         The adjusted location effects (B) for 1997, 2001 and 2005 were discounted by 5, 15 and 15 per cent respectively (discounted location effects: 1997 (C), 2001 (D) and 2005 (E)). 

·         For 1997-98 to 2001-02 (F), the discounted location effects were calculated using a straight line interpolation, or time weighting (TW), of discounted location effects for 1997 (C) and 2001 (D). 

·         For 2001-02 to 2005-06 (F), the discounted location effects were calculated using a straight line interpolation, or time weighting (TW), of discounted location effects for 2001 (D) and 2005 (E). 

·         The raw factors (F) were calculated by taking the exponential of the calculated discounted location effects (E). 

·         The final factors (G) were calculated by rescaling the raw factors by mean resident population (MRP) to force the Australian average factor to equal one. 

Table 8            Factors based on location effects, private sector, 1997, 2001 and 2005

	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aust
Location effects, Original (A)
1997	0.015	0.010	-0.027	-0.016	-0.015	-0.056	0.040	0.094	0.003
2001	0.050	-0.005	-0.048	-0.021	-0.046	-0.079	0.019	0.043	0.003
2005	0.037	-0.004	-0.035	-0.020	-0.024	-0.069	0.030	0.032	0.000
Location effects, Adjusted For Tasmania and the ACT, and Rescaled (B)
1997	0.015	0.009	-0.028	-0.016	-0.016	-0.028	0.027	0.094	0.000
2001	0.049	-0.006	-0.049	-0.022	-0.047	-0.049	0.033	0.042	0.000
2005	0.037	-0.004	-0.036	-0.021	-0.024	-0.048	0.030	0.031	0.000
Location effects, adjusted For Tasmania and the ACT, and discounted
1997 (C)	0.014	0.009	-0.026	-0.015	-0.015	-0.026	0.026	0.089	0.000
2001 (D)	0.042	-0.005	-0.041	-0.019	-0.040	-0.041	0.028	0.036	0.000
2005 (E)	0.031	-0.003	-0.030	-0.018	-0.021	-0.041	0.026	0.026	0.000
Calculated Location effect For Six Years using Time Weights (TW): (F = C*(1 - TW) + TW*D) for 2000-01 and 2001-02 and (F = D*(1 - TW) + TW*E) for other years
2000-01 (TW = 0.75)	0.035	-0.002	-0.038	-0.018	-0.033	-0.038	0.028	0.049	0.000
2001-02 (TW = 1.00)	0.042	-0.005	-0.041	-0.019	-0.040	-0.041	0.028	0.036	0.000
2002-03 (TW = 0.25)	0.039	-0.005	-0.039	-0.018	-0.035	-0.041	0.028	0.033	0.000
2003-04 (TW = 0.50)	0.037	-0.004	-0.036	-0.018	-0.030	-0.041	0.027	0.031	0.000
2004-05 (TW = 0.75)	0.034	-0.004	-0.033	-0.018	-0.025	-0.041	0.026	0.029	0.000
2005-06 (TW = 1.00)	0.031	-0.003	-0.030	-0.018	-0.021	-0.041	0.026	0.026	0.000
Calculated Raw Factors (G =EXP (F))
2000-2001	1.03541	0.99831	0.96310	0.98229	0.96706	0.96310	1.02801	1.05011	1.00045
2001-2002	1.04262	0.99487	0.95952	0.98144	0.96116	0.95952	1.02870	1.03620	1.00057
2002-2003	1.03991	0.99528	0.96219	0.98168	0.96569	0.95964	1.02803	1.03384	1.00048
2003-2004	1.03722	0.99569	0.96488	0.98192	0.97025	0.95976	1.02735	1.03149	1.00039
2004-2005	1.03453	0.99610	0.96757	0.98215	0.97483	0.95989	1.02668	1.02914	1.00033
2005-2006	1.03185	0.99651	0.97026	0.98239	0.97944	0.96001	1.02601	1.02680	1.00032
Final factors (H)
2000-2001	1.03494	0.99785	0.96266	0.98185	0.96662	0.96266	1.02754	1.04963	1.00000
2001-2002	1.04202	0.99430	0.95897	0.98088	0.96061	0.95897	1.02812	1.03561	1.00000
2002-2003	1.03941	0.99480	0.96173	0.98120	0.96523	0.95918	1.02753	1.03334	1.00000
2003-2004	1.03681	0.99530	0.96450	0.98153	0.96987	0.95939	1.02695	1.03108	1.00000
2004-2005	1.03419	0.99577	0.96724	0.98182	0.97451	0.95957	1.02634	1.02880	1.00000
2005-2006	1.03152	0.99620	0.96996	0.98208	0.97913	0.95971	1.02568	1.02648	1.00000

Estimating wage components of assessed expenses

74            The wage cost disabilities shown in Table 1 were applied to the wages proportion of expenses in each category to derive the wage cost factor.  The formula used was:

75            For each category, the wages proportion of expenses (including grants and subsidies used by the recipients to pay wages and salaries) was estimated using information from the adjusted budget, departmental annual reports and other documents. 

76            The Commission assumed that wages were 80 per cent of the minimum fixed costs of head office in every category.  Table 9 shows the wage proportions applied to the remainder of each category. 

Table 9            Wage cost as proportion of total expenses

Expense category	Wage proportion
	(%)
Pre-school Education	70
Government Primary Education	70
Non Government Primary Education	Not applicable
Government Secondary Education	70
Non Government Primary Education	Not applicable
Vocational Education and Training	70
Inpatient Services	62.5
Non-inpatient and Community Health Services	60
Population and Preventive Health	60
Family and Child Services	70
Aged and Disabled Services	70
Homeless and General Welfare	70
Housing	20
Services to Indigenous Communities	70
Police	80
Administration of Justice	60
Corrective Services	70
Public Safety	80
Culture and Recreation	70
National Parks and Wildlife Services	50
Electricity and Gas	80
Water, Sanitation and Protection of the Environment	80
Non-urban Passenger Transport	80
Roads	60
Urban Transit	100
Primary Industry	60
Mining, Fuel and Energy	60
Tourism	60
Manufacturing & Other Industry	60
General Public Services	80
Depreciation	35

Accommodation

77            The accommodation cost factor recognises differences in the cost of office accommodation, either through renting, leasing or purchasing.  The assessment was based on commercial rental data provided by the Australian Valuation Office (AVO), with an adjustment for the ACT because it has no 'outside capital city' locations. 

Calculation of the accommodation factor

78            The factor used data provided by the AVO on net average commercial rents (expressed in $/m², dollars per square metre), for various types of office locations in the central business districts (CBDs) (primary and secondary) and outside capital cities.  The factor was calculated as follows:

·                Calculate the weighted net average commercial rents for each State by multiplying the net rents by the average proportion of accommodation occupied by State governments in each type of location. 

·                Scale up the result for the ACT because it has no 'outside capital city' locations, by adding to its result the ACT's secondary fringe CBD rate multiplied by the average proportion of outside capital city accommodation. 

·                Calculate the rental ratio for each State by dividing its weighted average rents by the weighted national average rent. 

·                Rescale the ratios by mean resident populations so that the Australian average factor equalled one. 

79            The calculation steps are shown later under Results for 2007 Update — accommodation input costs. 

Estimating the accommodation component of standard expenses

80            The cost factors in Table 13 were applied to the average share of accommodation costs in total expenses (2 per cent) to derive the accommodation cost factors for each category.  The formula used was:

Electricity

81            The electricity input costs factors take account of the effects of unavoidable interstate differences in the cost of electricity on the relative costs of providing services. 

Calculation of electricity cost factors

82            The electricity factors were calculated using the electricity generated by each plant type and State available from the published data of Energy Supply Association of Australia Limited (ESAA). 

83            For each State, electricity generated by each plant type was weighted by the average cost of generation for that plant type and summed to derive a weighted generation cost. 

84            The generation costs were adjusted as follows:

·                For New South Wales, Victoria, Queensland and the ACT the final adjusted costs were set equal to the average of the generation costs in New South Wales, Victoria and Queensland. 

·                For Western Australia, the final adjusted cost was set equal to the estimated generation cost in the State plus 10 per cent of the standard cost of capital (which was estimated to be equal to 50 per cent of the standard generation cost) to allow for the greater reserve capacity required in the State. 

·                For South Australia, the final adjusted cost was set equal to the average of its generation cost and the average generation cost for the three large interconnected States (New South Wales, Victoria and Queensland). 

·                For Tasmania and the Northern Territory, no adjustments were made. 

85            The calculation steps are shown later under Results for 2007 Update — electricity input costs. 

86            The factors are updated annually using published generation cost data from ESAA.  The estimated cost share of electricity however remains unchanged. 

Estimating electricity component of standard expenses

87            The weighted electricity cost factor for each category was calculated by applying the estimated share of electricity costs in total expenses (0.5 per cent) to the calculated electricity cost factor.  The formula used was:

Results for the 2007 Update

Wages input costs

88            Table 10 compares the wages input costs factors calculated for the five years of the 2007 Update as described earlier with those assessed for the 2006 Update. 

Table 10     Wages input costs factors for 2006 Update and 2007 Update

	NSW	Vic	Qld	WA	SA	Tas	ACT	NT
2006 Update
2000‑2001	1.03494	0.99785	0.96266	0.98185	0.96662	0.96266	1.02754	1.04963
2001‑2002	1.04202	0.99430	0.95897	0.98088	0.96061	0.95897	1.02812	1.03561
2002‑2003	1.04216	0.99443	0.95909	0.98101	0.96073	0.95909	1.02825	1.03574
2003‑2004	1.04232	0.99458	0.95924	0.98116	0.96088	0.95924	1.02841	1.03590
2004-2005	1.04245	0.99470	0.95936	0.98128	0.96100	0.95936	1.02853	1.03603
Average	1.04078	0.99517	0.95986	0.98124	0.96197	0.95986	1.02817	1.03858
2007 Update
2001‑2002	1.04202	0.99430	0.95897	0.98088	0.96061	0.95897	1.02812	1.03561
2002‑2003	1.03941	0.99480	0.96173	0.98120	0.96523	0.95918	1.02753	1.03334
2003‑2004	1.03681	0.99530	0.96450	0.98153	0.96987	0.95939	1.02695	1.03108
2004‑2005	1.03419	0.99577	0.96724	0.98182	0.97451	0.95957	1.02634	1.02880
2005-2006	1.03152	0.99620	0.96996	0.98208	0.97913	0.95971	1.02568	1.02648
Average	1.03679	0.99527	0.96448	0.98150	0.96987	0.95936	1.02692	1.03106

Accommodation input costs

89            The accommodation factors are updated annually using rental data from the AVO.  The accommodation expense weight are not updated between updates. 

90            Table 11 shows the net average rent for 2005‑06 by types of office location and the standard weight for each location, the latter reflecting the average proportion of accommodation occupied by State governments in each type of location. 

Table 11          Net average commercial rents for CBDs and areas outside capital cities, 2005-06

Area	Std prop(a)	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aust
	%	$/m2	$/m2	$/m2	$/m2	$/m2	$/m2	$/m2	$/m2	$/m2
Prime CBD	15	510	310	350	325	240	270	380	340	381
Secondary CBD	15	330	230	230	220	195	220	235	270	260
Prime fringe CBD	18	340	270	250	290	210	170	320	230	285
Secondary fringe CBD	27 (52)	270	220	190	210	195	130	225	180	225
Outside capital city(b)	25 (0)	167	174	246	190	135	180	0	175	182

(a)        Standard proportions of commercial areas occupied by State governments. 

(b)        Except for Canberra, referring to the population weighted average of net average commercial rentals in the selected locations outside capital cities in the States.  The selected locations were:

             New South Wales: Broken Hill and Wagga Wagga;

             Victoria: Bendigo and Mildura;

             Queensland: Cairns and Mount Isa;

             Western Australia: Kalgoorlie‑Boulder;

             South Australia: Whyalla;

             Tasmania: Launceston; and

             Northern Territory: Alice Spring. 

Source: Australian Valuation Office. 

91            The factors were calculated as follows:

·                Calculate the weighted net average commercial rents for each location by multiplying the net average commercial rents by the standard weight for that location in Table 11.  Table 12 shows the weighted average rents for 2005‑06. 

Table 12          Weighted average rents, 2005-06

Area	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aust
	$/m2	$/m2	$/m2	$/m2	$/m2	$/m2	$/m2	$/m2	$/m2
Prime CBD	76.5	46.5	52.5	48.8	36.0	40.5	57.0	51.0	57.1
Secondary CBD	49.5	34.5	34.5	33.0	29.3	33.0	35.3	40.5	39.0
Prime fringe CBD	61.2	48.6	45.0	52.2	37.8	30.6	57.6	41.4	51.3
Secondary fringe CBD	72.9	59.4	51.3	56.7	52.7	35.1	60.8	48.6	60.9
Outside capital city(b)	41.8	43.5	61.5	47.5	33.8	45.0		43.8	45.5
Sum of weighted average	301.9	232.5	244.8	238.2	189.5	184.2	210.6	225.3	253.7

·                Scale up the ACT results because it has no 'outside capital city' locations.  The ACT's secondary fringe CBD rate was scaled up by the standard proportion of 'outside capital city' accommodation. 

·                Calculate the ratio by dividing the States weighted average rents divided by weighted average national rent. 

·                These ratios were rescaled by mean resident population to calculate the accommodation cost factors. 

92            Table 13 shows the accommodation cost factors for the 2007Update. 

Table 13          Accommodation factors, 2007 Update

	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aust
2001‑2002	1.17883	1.06579	0.82972	0.76466	0.82876	0.68932	1.07058	0.79313	1.00000
2002‑2003	1.17941	1.06632	0.83013	0.76504	0.82917	0.68966	1.07110	0.79352	1.00000
2003‑2004	1.17230	1.05479	0.85060	0.77435	0.84272	0.70180	1.06058	0.78199	1.00000
2004‑2005	1.15820	1.01043	0.95280	0.78494	0.79433	0.66380	1.02767	0.82773	1.00000
2005-2006	1.18568	0.91327	0.96159	0.93547	0.74417	0.72355	1.04820	0.88479	1.00000

93            Table 14 compares the accommodation input costs factors calculated for the five years of the 2007 Update with those for the 2006 Update. 

Table 14          Accommodation costs factors, comparison of factors, 2006 Update and2007 Update

	NSW	Vic	Qld	WA	SA	Tas	ACT	NT
2006 Update
2000‑2001	1.17868	1.06565	0.82961	0.76456	0.82866	0.68923	1.07044	0.79302
2001‑2002	1.17883	1.06579	0.82972	0.76466	0.82876	0.68932	1.07058	0.79313
2002‑2003	1.17941	1.06632	0.83013	0.76504	0.82917	0.68966	1.07110	0.79352
2003‑2004	1.17230	1.05479	0.85060	0.77435	0.84272	0.70180	1.06058	0.78199
2004‑2005	1.15814	1.01037	0.95275	0.78490	0.79429	0.66376	1.02762	0.82768
Average	1.17347	1.05259	0.85856	0.77070	0.82472	0.68676	1.06006	0.79787
2007 Update
2001‑2002	1.17883	1.06579	0.82972	0.76466	0.82876	0.68932	1.07058	0.79313
2002‑2003	1.17941	1.06632	0.83013	0.76504	0.82917	0.68966	1.07110	0.79352
2003‑2004	1.17230	1.05479	0.85060	0.77435	0.84272	0.70180	1.06058	0.78199
2004‑2005	1.15820	1.01043	0.95280	0.78494	0.79433	0.66380	1.02767	0.82773
2005-2006	1.18568	0.91327	0.96159	0.93547	0.74417	0.72355	1.04820	0.88479
Average	1.17883	1.06579	0.82972	0.76466	0.82876	0.68932	1.07058	0.79313

94            In the 2004 Review, the Commission used commercial rental data from the AVO for 2002‑03 for all assessment years.  Since the 2005 Update, the Commission used new commercial rental data for 2003‑04 and subsequent years. 

95            Over the five years to 2005-06, the average annual rate of growth in Aggregate Gross Domestic Product was 3.7 per cent, while for each of Queensland, Western Australia and the Northern Territory it was over 5 per cent.  Consistent with this, Queensland, Western Australia and the Northern Territory showed above average increases to rental cost for office premises over this period mirroring the rentals in the prime CBD areas (Brisbane: 59 per cent; Perth: 63 per cent; Darwin: 58 per cent).  New South Wales also recorded modest growth for the five years to 2005-06.  This was mainly because of rental growth of over 20 per cent in the last year as a result of limited space availability in the prime CBD and prime fringe sectors. 

Electricity input costs

96            For the 2005 Update, the electricity factors were calculated using published generation data by types of plant for the years 1998‑99 to 2002‑03 from the Energy Suppliers Association of Australia (ESAA) annual publications.  The data for 2002‑03 were also used for 2003‑04. 

97            On that basis, in each update, data for the pre-final year are brought in, but the same data are used for the final year as well.  In this update, the data were updated for 2004-05 using the same publication for 2006.  The same data are used for 2005‑06.  The detailed calculations are in Appendix C. 

98            The weights for each type of plant are not updated between reviews. 

99            Table 15 sets out the final factors calculated for 2005‑06 on these bases. 

Table 15          Electricity factors based on generation costs in $/mwh, 2005‑06

	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aus
	$/mwh	$/mwh	$/mwh	$/mwh	$/mwh	$/mwh	$/mwh	$/mwh	$/mwh
Weighted cost	33.73	33.48	34.33	36.72	36.90	20.92	33.73	68.28	34.36
Final modified weighted cost(a)	33.85	33.85	33.85	38.44	35.38	20.92	33.85	68.28	34.45
Raw factors	0.98264	0.98264	0.98264	1.11582	1.02700	0.60718	0.98264	1.98233	1.00000
Final factors	0.97156	0.97156	0.97156	1.20137	1.02520	0.56689	0.97156	2.11356	1.00000

 (a)       See paragraph 85 for conversion from weighted cost to final modified weighted cost. 

             mwh ‑ million kilowatt hours. 

100         The factors for the three bigger States — which served as the benchmark for the 'national market' — were generally similar in magnitude.  The factor for South Australia was over one, because of its less than full integration with the national market for topographical and historical reasons. 

101         Western Australia and the Northern Territory had factors substantially greater than one mainly because they were not part of the national market.  This meant they incurred greater costs because they needed to maintain a greater reserve capacity (to respond to sudden increases in demand) than other States who could draw upon reserves in the national grid. 

102         Table 16 shows electricity cost factors for the 2007 Update. 

Table 16          Electricity cost factors, 2007 Update

	NSW	Vic	Qld	WA	SA	Tas	ACT	NT
2001‑2002	0.97429	0.97429	0.97429	1.16200	1.04429	0.53055	0.97429	2.21742
2002‑2003	0.97111	0.97111	0.97111	1.21907	1.03958	0.57216	0.97111	1.85801
2003‑2004	0.97249	0.97249	0.97249	1.23142	1.02460	0.62140	0.97249	1.62580
2004‑2005	0.98251	0.98251	0.98251	1.11568	1.02686	0.60710	0.98251	1.98207
2005‑2006	0.98236	0.98236	0.98236	1.11550	1.02671	0.60700	0.98236	1.98177
Average	0.97294	0.97294	0.97294	1.19670	1.03325	0.56398	0.97294	1.99610

103         Table 17 compares the electricity input costs factors calculated for the five years of the 2007 Update with those for the 2006 Update. 

Table 17          Electricity input cost factors, comparison of factors, 2006 Update and 2007 Update

	NSW	Vic	Qld	WA	SA	Tas	ACT	NT
2006 Update
2000‑2001	0.97387	0.97387	0.97387	1.17433	1.02452	0.53182	0.97387	2.28318
2001‑2002	0.97429	0.97429	0.97429	1.16200	1.04429	0.53055	0.97429	2.21742
2002‑2003	0.97111	0.97111	0.97111	1.21907	1.03958	0.57216	0.97111	1.85801
2003‑2004	0.97249	0.97249	0.97249	1.23142	1.02460	0.62140	0.97249	1.62580
2004‑2005	0.97239	0.97239	0.97239	1.23129	1.02450	0.62133	0.97239	1.62564
Average	0.97156	0.97156	0.97156	1.20137	1.02520	0.56689	0.97156	2.11357
2007 Update
2001‑2002	0.97429	0.97429	0.97429	1.16200	1.04429	0.53055	0.97429	2.21742
2002‑2003	0.97111	0.97111	0.97111	1.21907	1.03958	0.57216	0.97111	1.85801
2003‑2004	0.97249	0.97249	0.97249	1.23142	1.02460	0.62140	0.97249	1.62580
2004‑2005	0.98251	0.98251	0.98251	1.11568	1.02686	0.60710	0.98251	1.98207
2005-2006	0.98236	0.98236	0.98236	1.11550	1.02671	0.60700	0.98236	1.98177
Average	0.97294	0.97294	0.97294	1.19670	1.03325	0.56398	0.97294	1.99610

104         The changes to the relativities since the 2006 Update were due to changes in the assessment data and the mean resident population.  The changes were small because of the stable patterns of electricity generation by plant type over time.  The fall in the factor for the Northern Territory until 2003-04 was due, to less reliance on the most expensive form of electricity, the diesel engine.  This had accounted for 16% of generating capacity in 2000-01 to 4.6% in 2003-04.  However in 2004-05 it rose to 10.6%. 

Summary of the input cost factors

105         The overall input costs factors, combining the wages, accommodation and electricity cost sub-components, were calculated by:

·                weighting the extent of the cost advantage/disadvantage for each component by the proportion of costs for that component; and

·                summing and adding 1, to turn the weighted cost advantages/disadvantages into a cost factor relative to the Australian average of 1. 

106         This is shown in the following formula. 

Input costs factor    =             wage proportion * (wage cost factor - 1)                                     + accommodation proportion * (accommodation cost factor - 1)                                     + electricity proportion * (electricity factor - 1)                                     + 1   where                           wage proportion = range from 0.2 to 0.8 (see Table 9);                                     accommodation proportion = 0.02; and                                     electricity proportion = 0.005.    Note: For input costs applied to the 'fixed cost' component, wage proportion = 0.8.

 

107         Table 18 shows the fixed cost input costs factors for all expenses categories to which these apply. 

Table 18          Input cost factors applied to the fixed costs component, 2005‑06

	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aust
Input Costs factor	1.02885	0.99514	0.97511	0.98495	0.97832	0.96027	1.02142	1.02379	1.00000

108         Table 19 shows, for each expense category, the input costs factors applied to all components other than the fixed cost component. 

Changes in GST revenue distribution: 2007 Update compared with 2006 Update

What has changed?

109         The main changes the Commission examines are:

·                revisions to the financial and assessment data that were used in the 2006 Update; and

·                advancing the reference period one year — a new year comes into the reference period and the oldest year drops out. 

110         Figure 1 shows the reference periods for the two inquiries. 

Table 19          Input cost factors by category, 2005-06

Expense category	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aust
Pre-school Education	1.02569	0.99552	0.97811	0.98674	0.98041	0.96430	1.01885	1.02114	1.00000
Government Primary Education	1.02569	0.99552	0.97811	0.98674	0.98041	0.96430	1.01885	1.02114	1.00000
Government Secondary Education	1.02569	0.99552	0.97811	0.98674	0.98041	0.96430	1.01885	1.02114	1.00000
Vocational Education and Training	1.02569	0.99552	0.97811	0.98674	0.98041	0.96430	1.01885	1.02114	1.00000
Inpatient Services	1.02333	0.99580	0.98037	0.98809	0.98197	0.96732	1.01693	1.01915	1.00000
Non-inpatient and Community Health Services	1.02254	0.99590	0.98112	0.98853	0.98249	0.96833	1.01629	1.01849	1.00000
Population and Preventive Health	1.02254	0.99590	0.98112	0.98853	0.98249	0.96833	1.01629	1.01849	1.00000
Family and Child Services	1.02569	0.99552	0.97811	0.98674	0.98041	0.96430	1.01885	1.02114	1.00000
Aged and Disabled Services	1.02569	0.99552	0.97811	0.98674	0.98041	0.96430	1.01885	1.02114	1.00000
Homeless and General Welfare	1.02569	0.99552	0.97811	0.98674	0.98041	0.96430	1.01885	1.02114	1.00000
Housing	1.00993	0.99742	0.99314	0.99570	0.99084	0.98445	1.00601	1.00790	1.00000
Services to Indigenous Communities	1.02569	0.99552	0.97811	0.98674	0.98041	0.96430	1.01885	1.02114	1.00000
Police	1.02885	0.99514	0.97511	0.98495	0.97832	0.96027	1.02142	1.02379	1.00000
Administration of Justice	1.02254	0.99590	0.98112	0.98853	0.98249	0.96833	1.01629	1.01849	1.00000
Corrective Services	1.02569	0.99552	0.97811	0.98674	0.98041	0.96430	1.01885	1.02114	1.00000
Public Safety	1.02885	0.99514	0.97511	0.98495	0.97832	0.96027	1.02142	1.02379	1.00000
Culture and Recreation	1.02569	0.99552	0.97811	0.98674	0.98041	0.96430	1.01885	1.02114	1.00000
National Parks and Wildlife Services	1.01939	0.99628	0.98412	0.99033	0.98458	0.97236	1.01372	1.01584	1.00000
Electricity and Gas	1.02885	0.99514	0.97511	0.98495	0.97832	0.96027	1.02142	1.02379	1.00000
Water Sanitation and Protection of the Environment	1.02885	0.99514	0.97511	0.98495	0.97832	0.96027	1.02142	1.02379	1.00000
Non-urban Passenger Transport	1.02885	0.99514	0.97511	0.98495	0.97832	0.96027	1.02142	1.02379	1.00000
Roads	1.02254	0.99590	0.98112	0.98853	0.98249	0.96833	1.01629	1.01849	1.00000
Urban Transit	1.03152	0.99620	0.96996	0.98208	0.97913	0.95971	1.02568	1.02648	1.00000
Primary Industry	1.02254	0.99590	0.98112	0.98853	0.98249	0.96833	1.01629	1.01849	1.00000
Mining, Fuel and Energy	1.02254	0.99590	0.98112	0.98853	0.98249	0.96833	1.01629	1.01849	1.00000
Tourism	1.02254	0.99590	0.98112	0.98853	0.98249	0.96833	1.01629	1.01849	1.00000
Manufacturing & Other Industry	1.02254	0.99590	0.98112	0.98853	0.98249	0.96833	1.01629	1.01849	1.00000
General Public Services	1.02885	0.99514	0.97511	0.98495	0.97832	0.96027	1.02142	1.02379	1.00000
Depreciation	1.01101	0.99867	0.98951	0.99374	0.99271	0.98592	1.00897	1.00925	1.00000

Figure 1          Advancing the reference period, 2007 Update

2000‑01	2001‑02	2002‑03	2003‑04	2004‑05	2005-06
2006 Update
	2007 Update

111         The effect of revisions is estimated by replacing 2005 Update data with 2006 Update data for the years 1999‑00 to 2003‑04.  The effect of advancing the reference period one year is estimated by comparing the data of the new year entering the reference period (2004‑05) with the financial and assessment data of the year dropping out (1999‑00).  In both cases, the Commission considers the impact of replacing financial data separately from the effect of replacing assessment data. 

Wages input costs

112         Table 20 shows the redistribution of GST revenue due to wages cost factors in the 2005 Update and the 2006 Update. 

Table 20          Wages cost factor, contributions and changes to the GST revenue distribution^(a), 2006 Update and 2007 Update

Contribution	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Total redist'd
	$m	$m	$m	$m	$m	$m	$m	$m	$m
2006 Update	919.4	-64.9	-541.2	-131.1	-196.1	-73.0	33.3	53.6	1006.3
2006 Update	811.7	-63.5	-474.0	-126.3	-150.7	-71.8	31.1	43.6	886.4
Effect of updating	-107.7	1.4	67.2	4.8	45.4	1.1	-2.1	-10.0	119.8

(a)        The distributions were calculated by applying the 2006 Update and 2007 Update relativities to State populations as at 30 December 2006 and the 2006-07 GST and Health Care Grants (HCGs) pool of $47.634 billion, and by removing the influence of wages. 

113         The changes to redistribution were consistent with the direction of changes to the factors (see Table 10).  Although the changes to the factors appear small, they have large redistributive effects because wages factors apply to a large number of expenses categories and have high weights associated with them (see Table 9). 

114         The main change arises from revising the factors for the reference years. 

115         This decreased the GST revenue distribution significantly for New South Wales and the Northern Territory, and to a smaller extent for the ACT.   The GST revenue distribution increased for the other States. 

Electricity and accommodation input costs (only changes since the 2006 Update)

116         Table 21 shows changes to redistribution of GST revenue since the 2006 Update due to the accommodation and electricity cost factors. 

Table 21          Electricity and accommodation input costs, changes to the GST revenue redistribution since 2006 Update

	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Total redist'd
	$m	$m	$m	$m	$m	$m	$m	$m	$m
Effect of updating	3.2	-12.8	8.1	3.9	-3.0	0.1	-0.1	0.7	16.0

(a)        The distributions were calculated by applying the 2006 Update and 2007 Update relativities to State populations as at 30 December 2006 and the 2006-07 GST and Health Care Grants (HCGs) pool of $47.634 billion, and by removing the influence of electricity and accommodation. 

117         The changes to the GST revenue distribution, since the 2006 Update, due to electricity and accommodation cost factors were modest.  The effects were relatively high for Victoria, Queensland and Western Australia. 

118         Over the five years to 2005-06, the average annual rate of growth in Gross Domestic Product was 3.7 per cent, while for each of Queensland, Western Australia and the Northern Territory it was over 5 per cent.  Consistent with this, Queensland, Western Australia and the Northern Territory showed above average increases to rental cost for office premises over this period mirroring the rentals in the prime CBD areas (Brisbane: 59 per cent; Perth: 63 per cent; Darwin: 58 per cent).  New South Wales also recorded modest growth for the five years to 2005-06.  This was mainly because of rental growth of over 20 per cent in the last year as a result of limited space availability in the prime CBD and prime fringe sectors. 

119         There was a significant decrease for Victoria, a decrease for South Australia, and some decrease for the ACT because rents were relatively steady in absolute terms, but fell relative to the average of the other States.  Hence there was redistribution of GST revenue away from these States. 

120         The changes to redistribution from electricity costs were generally small because of the stable patterns of electricity generation by plant type over time.  Over four years to 2004‑2005, the decrease in the disability factors for the Northern Territory and Western Australia was due, according to the data, to less reliance on the most expensive form of electricity, the diesel engine[8].    This was so although its use increased between 2003‑2004 and 2004‑2005 in the Northern Territory. 

This chapter was prepared by the Revenue section of the Commonwealth Grants Commission.  If you have any questions about its content please contact Gautam Biswas on (02) 6229 8889 or Gautam.biswas@cgc.gov.au

 

121             Date:  15 February 2007

WAGES INPUT COST FACTORS FOR
THE 2007 UPDATE

 

 

 

 

 

CGC STAFF INFORMATION PAPER CGC 2006/10-S

 

 

122          

123          

124          

125          

126         NOVEMBER 2006

127          

128          

 

129          

130       INTRODUCTION   1

131       THE 2007 UPDATE RESULTS  1

132       REALITY CHECKING   3

133       Reality checking — differences in location effects measured from the average  3

134       Reality checking — differences in location effects between 2001 and 2005  6

135       Revisiting the judgments  11

136          

137       ATTACHMENT A:  IMPORTANT DIFFERENCES BETWEEN
THE 2001 AND 2005 SET DATA   13

138       ATTACHMENT B:  REGRESSION RESULTS  14

139          

INTRODUCTION

140         Wages input costs factors reflect the effects of unavoidable interstate differences in the cost of labour (wages) on the relative costs of providing State services. 

141         In the 2004 Review, the Commission estimated the differences across States in the wages of comparable employees in the private sector using an econometric model[10] based on data[11]from the ABS's Survey of Education and Training Experience (SET) for 1997 and 2001.  Those differences were called 'location effects'.

142         The Commission committed to re-estimating the location effects when data from the 2005 survey became available.  The data, which are similar in structure and depth to the 2001 survey, are now available.  The wages data in the 2005 survey relate to May to August of 2005.

143         This information paper presents updated location effects for 2005. 

The 2007 UPDATE RESULTS

144         The 2004 Review model sought to explain interstate differences in wage levels in terms of differences in:

·                characteristics of employees that are usually associated with productivity (such as type and level of education, experience, occupation, industry); and

·                location specific effects[12]. 

145         In other words, the model estimates the interstate differences in wages paid to comparable employees and attributes them to differences in location specific effects such as cost of living. 

146         For all practical purposes, the same model has been applied to the 2005 survey data[13]. 

147         Table 22 summarises the modelled location effects for 1997, 2001 and 2005. 

Table 22                      Modelled location effects for 1997, 2001 and 2005

	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aust
1997
Location effects	0.015	0.010	-0.027	-0.016	-0.015	-0.056	0.040	0.094	0.000
Standard error(b)	0.005	0.006	0.007	0.010	0.011	0.022	0.027	0.037	0.000
t-value	3.32	1.75	-4.02	-1.62	-1.35	-2.59	1.46	2.57	0.000
Level of Significance	0.001	0.080	0.000	0.104	0.177	0.010	0.145	0.010	0.000
2001
Location effects	0.050	-0.005	-0.048	-0.021	-0.046	-0.079	0.019	0.043	0.000
Standard error	0.005	0.006	0.007	0.010	0.011	0.021	0.027	0.042	0.000
t-value	11.03	-0.90	-7.04	-2.18	-4.09	-3.80	0.69	1.02	0.000
Level of Significance	0.000	0.369	0.000	0.030	0.000	0.000	0.490	0.306	0.000
2005
Location effects	0.037	-0.004	-0.035	-0.020	-0.024	-0.069	0.031	0.032	0.000
Standard error	0.005	0.006	0.008	0.011	0.014	0.025	0.031	0.044	0.000
t-value	7.10	-0.57	-4.67	-1.86	-1.77	-2.79	0.98	0.76	0.000
Level of Significance	0.000	0.566	0.000	0.063	0.076	0.005	0.327	0.448	0.000
Point differences between 2001 and 2005 location effects
	-0.013	0.001	0.013	0.001	0.022	0.010	0.012	-0.011	0.000

148          

149         Major differences between the results for 2001 and 2005 are a lower location effect for New South Wales and the Northern Territory and higher location effects for the other States, except Victoria and Western Australia where there was little change.  The precision of measurements as indicated by standard errors is similar in 2001 and 2005, although the t values for some States have fallen in 2005 (reflecting the fall in their location effect). 

150         The reduced location effect for New South Wales is consistent with a slowing of its economy and a slower rate of increase in house prices.  The reduced location effect for the Northern Territory accords with greater integration of its labour force with other States (which has been showing as a long term movement in the relative wage level towards the national average), which offset the effects of the recent economic growth and higher house prices. 

151         The changes for the other States are consistent with the growth in house prices.  The change for South Australia is broadly consistent with its house price movements but appears larger than expected.  The small change for Western Australia may appear unexpected when judged against movements in more common wage indicators such as average earnings.  

Reality Checking

152         During the 2004 Review, a range of indicative information was examined to see whether the results of the modelling accorded with reality.  This reality checking was subject to an important caveat that no single readily available data series, particularly wages data series, could directly benchmark the results.  This is because the model attempts to compare wages paid to like employees across all States by removing the effects of differences between States and over time in skills, industry and so on.  The existing wages data series do not do that — they are affected by changes in composition across States and/or time.  The reality checking thus requires balancing different, and sometimes apparently contradictory, information. 

153         The next sections provide information that forms a basis for reality checking the:

·                pattern of the 2005 location effects relative to the average (no locality effect); and

·                movements between the estimated locality effects for 2001 and 2005.

Reality checking — differences in location effects measured from the average

154         Some of the evidence for the 2004 Review related to the existence of generic location effects.  For example, we examined other studies of wage differentials.  They gave clear indications that interstate wage differentials existed, after controlling for many possible influences[14]. 

155         Spatial experimental cost of living index. Table 23 presents an update of the spatial experimental cost of living index constructed by the ABS for the capital cities.  This indicator, although limited to the capital cities, relates directly to the main concept behind the model. 

Table 23          ABS experimental spatial price indices, September 2001 and 2003-04
(Sydney = 100.0)

Item	Sydney	Melbourne	Brisbane	Adelaide	Perth	Hobart	Darwin	Canberra
September 2001(a)
ALL except Housing	100.0	99.4	98.3	95.1	98.1	98.6	96.7	98.2
Housing Cost Index(b)	100.0	85.1	78.0	70.7	71.5	65.7	-	95.5
2003-04(c)
All items	100.0	91.8	85.2	87.7	86.3	88.4	92.8	93.2
ALL except Housing	100.0	97.8	95.1	96.4	96.2	97.9	99.8	98.7
Housing	100.0	80.5	65.4	69.9	65.7	67.7	79.8	83.0

(a)        Source:  Based on experimental estimates for the 32^nd Conference of Economists held in Canberra from 29 September to 1 October, 2003.  This work was yet to be endorsed by the ABS for use in policy formulation.  Housing was not included at that time.  Aggregated indices were compiled using 1998-99 HES weights.  Omitted items were set to 100 for each location. 

(b)        Source:  2004 Review Working Papers, Vol 7.  This was constructed by the Tasmanian Treasury. 

(c)        Source:  Measuring differences in prices across Australian capital cities, (Paper presented at Economic Measurement Group Workshop at the Centre for Applied Economic Research, Sydney, 12 - 13 December 2005), Keith Woolford et al, Prices Research and Development, Australian Bureau of Statistics.

156         This indicator does not cover the same time period as the modelling of the location effects.  However, it shows differences across the capital cities in the measurable components of the cost of living, and housing in particular.  It highlights the high cost of living in Sydney, Canberra and Darwin, and the higher housing costs in Sydney.  This strongly supports the conceptual basis, and the direction, of the results from the model. 

157         Econometric modelling using Employee Earnings, Benefits and Trade Union membership (EEBTUM) data.  In analysis for Heads of Treasuries, the ABS suggested an alternative to the SET data — the EEBTUM survey data, a collection conducted in August each year as a supplement to the monthly labour force survey.  The ABS made the detailed data from this survey available for the first time in 2005. 

158         The survey collected most, though not all, data items of interest.  In particular, the survey did not collect details of education for the respondent, a key determinant of wages.   

159         We applied a model similar to that used for the SET data.  Table 24 compares the modelled location effects for 2005 using the SET and the EEBTUM data. 

Table 24          Modelled location effects, SET and EEBTUM data, 2005

	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aust
SET
Location effects	0.037	-0.004	-0.035	-0.020	-0.024	-0.069	0.031	0.032	0.000
Standard error(b)	0.005	0.006	0.008	0.011	0.014	0.025	0.031	0.044
t-value	7.10	-0.57	-4.67	-1.86	-1.77	-2.79	0.98	0.76
Level of Significance	0.000	0.566	0.000	0.063	0.076	0.005	0.327	0.448
EEBTUM
Location effects	0.030	-0.002	-0.030	-0.009	-0.027	-0.055	0.042	0.046	0.000
Standard error	0.004	0.005	0.006	0.010	0.011	0.021	0.027	0.036
t-value	6.712	-0.414	-4.736	-0.937	-2.514	-2.63	1.573	1.266
Level of Significance	0.000	0.679	0.000	0.349	0.012	0.009	0.116	0.205

160         Despite the data differences, the results and the standard errors were broadly similar.  Both analyses indicate the location effects for New South Wales, the ACT and the Northern Territory were above average, and those for the other States below average. 

161         Comparing the location effects using private and public sector data.  The model was also run using the public sector data from SET (which included Commonwealth public servants).  Table 25 compares the modelled location effects for the two sectors for 2001 and 2005.  It shows the directions of changes between 2001 and 2005 in the two sectors mirror each other for most States. 

Table 25          Comparison of modelled location effects by sector, SET data, 2001 and 2005

	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aust
2001
Private	0.050	-0.005	-0.048	-0.021	-0.046	-0.079	0.019	0.043	0
Public	0.057	-0.024	-0.026	-0.026	-0.063	-0.081	0.076	0.017	0
2005
Private	0.037	-0.004	-0.035	-0.020	-0.024	-0.069	0.031	0.032	0
Public	0.052	-0.041	-0.022	-0.025	-0.021	-0.036	0.096	0.049	0
Point differences between 2001 and 2005
Private	-0.013	0.001	0.013	0.001	0.022	0.010	0.012	-0.011	0
Public	-0.004	-0.016	0.004	0.001	0.042	0.044	0.020	0.032	0

 

162         The results for 2005 are graphed in Figure 2.

Figure 2          Comparison of modelled location effects between sectors, SET data, 2005

 

163         The analysis shows that despite government policy differences, there are many similarities in the location effects for the two sectors — similar results were obtained for 2001.  The large difference between the two sectors for the ACT is because the public sector data include the Commonwealth public service employees. 

Reality checking — differences in location effects between 2001 and 2005

164         The evidence relates to land values, house prices and observed wage levels in States. 

165         Land values.  Table 26 shows total value of residential land in each State in dollars and relative to the Australian average per capita value.  We note the reductions in relative per capita values in New South Wales and the Northern Territory are consistent with the changes in their modelled location effects.  The increases in relative per capita values in Queensland, South Australia and the ACT are also consistent with the changes in their location effects.  The below average increases in land values in Victoria and Western Australia are not consistent with their stable location effects and Tasmania's land values move in the opposite direction to its location effect.

Table 26          Total residential land values, 2000-01 to 2004-05

Year	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aus
	$m	$m	$m	$m	$m	$m	$m	$m	$m
Total land value
2000-01	339840	204432	98231	67299	33531	5709	10597	3438	763077
2001-02	367235	227954	110780	75131	37192	5910	12557	3470	840229
2002-03	434948	248704	135930	89299	45264	6344	18139	3633	982263
2003-04	549317	296170	169646	102434	57938	6917	22562	3724	1208708
2004-05	618553	360089	247493	122352	73095	10321	25516	4135	1461555
Differences between 2000-01 and 2004-05 (%)	82.0	76.1	151.9	81.8	118.0	80.8	140.8	20.3	91.5
Relative per capita land value
2000-01	1.3147	1.0823	0.6886	0.9004	0.5642	0.3078	0.8442	0.4421	1.0000
2001-02	1.2938	1.0969	0.6977	0.9124	0.5724	0.2923	0.9129	0.4082	1.0000
2002-03	1.3163	1.0235	0.7246	0.9257	0.5998	0.2691	1.1372	0.3708	1.0000
2003-04	1.3550	0.9915	0.7293	0.8621	0.6262	0.2382	1.1530	0.3095	1.0000
2004-05	1.2674	0.9969	0.8718	0.8483	0.6575	0.2950	1.0872	0.2844	1.0000
Point differences between 2000-01 and 2004-05	-0.047	-0.085	0.183	-0.052	0.093	-0.013	0.243	-0.158	0.000

Source: State data returns. 

 

166         House Price Index. Differences in housing costs are the major determinant of differences in cost of living between States (see Table 23).  More importantly, changes in the cost of housing can differ between regions and can materially affect the relative cost of living in a short time (say, 5 years).  Changes in prices of many other goods and services tend to follow a national pattern affecting all States more evenly given interstate and international trade. 

167         In that light, we examine to what extent changes in house prices accord with changes in the modelled location effects between 2001 and 2005. 

168         Table 6 shows the ABS price index for established homes[15] in the capital cities.  This series conceptually tracks changes in the price of the same (notional) representative house in each city and is less prone to compositional effects than some other data series. 

Table 27          Price index of established homes (2003-04 = 100)

	Sydney	Melbourne	Brisbane	Perth	Adelaide	Hobart	Canberra	Darwin	All capitals
Mar-2002(a)	75.9	79.3	61.9	75.5	69.7	56.3	66.5	81.6	74.3
2005(b)	95.0	102.8	105.0	119.6	107.4	113.3	100.3	121.0	101.7
Change (%)	25.2	29.6	69.6	58.4	54.1	101.2	50.8	48.3	36.8

(a)        The ABS' new house price series started only in Mar 2002. 

(b)        This represents the average of four quarters, Dec 2004 to Sep 2005.  The series underwent a method change after
                   September 2005.  

 

169         The below average change for Sydney is consistent with the negative change in its location effect.  The direction of change is also consistent for most other capital cities.  For Perth however, the SET data imply the changes in house prices were not translated into higher wages by mid-2005.  For Darwin, the SET data imply the effects of recent house price activity on relative wage levels are more than offset by those of greater mobility. 

170         Median house prices across capital cities in recent years.  Table 7shows the changes in the median house prices of established homes in the capital cities.  This series is not based on the same representative house in each city, and is affected by compositional effects. 

Table 28              Median house prices in capital cities, 2000-01 and 2004-05

Year	Syd	Mel	Bri	Per	Ade	Hob	Can	Dar
	$ 000	$ 000	$ 000	$ 000	$ 000	$ 000	$ 000	$ 000
Jun, 2001	316.0	291.0	180.0	165.7	148.2	120.3	203.0	187.0
Jun, 2005	528.0	360.0	312.5	295.0	275.0	260.0	352.5	279.8
Prices relative to Sydney
Jun, 2001	100.0	92.1	57.0	52.4	46.9	38.1	64.2	59.2
Jun, 2005	100.0	68.2	59.2	55.9	52.1	49.2	66.8	53.0
Per cent point changes		-23.9	2.2	3.4	5.2	11.2	2.5	-6.2

Source:  Real Estate Institute of Australia.  Median prices were averaged over quarters. 

 

171         While the median house prices in different locations do not, strictly speaking, compare like with like they still show rises in house prices — relative to Sydney — in all capital cities except Melbourne and Darwin.  They also show that, up to June 2005, Perth house prices had not increased relative to Sydney to the extent implied by Table 27. 

172         These changes are generally consistent with the changes in modelled lo location effects. 

173         The recent fall in Sydney's house prices relative to other capital cities has also been noted in the Reserve Bank of Australia Bulletin of February 2006^[16]. 

174         In summary, changes to house prices between 2001 and 2005, together with a long term decline in relative wages in the Northern Territory, are generally consistent with changes in the modelled location effects between 2001 and 2005. 

175         Wages series.  Next, we looked for indicative evidence on how changes in average wages data from standard ABS publications compared with the modelled location effects.  The earlier caution that the readily available wages data series are not directly comparable with the modelled results because of compositional differences should be borne in mind.  

176         Average weekly earnings. Table 29 shows wages in dollars and relative to the Australian figure by sector.  Two specific cautions apply to these figures: 

·                the public sector figures for the ACT include the Commonwealth sector; and 

·                comparisons over time and across States do not control for any compositional changes. 

We note that changes in private sector wages in New South Wales, Victoria and the ACT are consistent with changes in the modelled effects, those in other States are not. 

Table 29          Full-time adult ordinary time earnings, trend, by sector, 2001 to 2005

Year	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aus
	$	$	$	$	$	$	$	$	$
Private
2001	860.3	779.0	738.5	821.4	743.2	698.3	837.4	768.3	801.8
2002	900.1	853.7	771.3	853.4	768.2	744.9	871.1	791.4	847.3
2003	950.3	904.2	812.1	899.3	816.9	776.7	954.2	846.1	896.3
2004	975.7	945.2	855.0	951.0	835.2	809.5	937.5	905.1	930.1
2005	1 044.6	975.8	905.0	1030.6	888.8	840.4	1 048.7	984.9	984.5
Differences between 2001 and 2005 (%)	21.4	25.3	22.5	25.5	19.6	20.3	25.2	28.2	22.8
Public
2001	939.1	950.5	884.3	899.1	916.2	890.1	1 019.5	906.8	927.6
2002	977.3	995.9	921.2	943.8	947.5	921.5	1 057.3	946.6	967.1
2003	1 024.1	1039.3	960.2	986.3	999.0	947.4	1 118.4	994.2	1011.7
2004	1 074.4	1079.9	1018.9	1008.9	1041.0	984.1	1 167.6	1038.8	1057.8
2005	1 119.9	1133.4	1064.8	1048.1	1078.4	1044.1	1 251.3	1096.4	1106.4
Differences between 2001 and 2005 (%)	19.3	19.2	20.4	16.6	17.7	17.3	22.7	20.9	19.3
Private
2001	1.07300	0.97156	0.92108	1.02448	0.92694	0.87100	1.04440	0.95831	1.00000
2002	1.06232	1.00752	0.91033	1.00714	0.90664	0.87915	1.02806	0.93397	1.00000
2003	1.06031	1.00881	0.90606	1.00332	0.91141	0.86659	1.06460	0.94405	1.00000
2004	1.04903	1.01624	0.91925	1.02255	0.89804	0.87036	1.00801	0.97312	1.00000
2005	1.06104	0.99111	0.91917	1.04682	0.90280	0.85364	1.06516	1.00033	1.00000
Differences between 2001 and 2005	-0.012	0.020	-0.002	0.022	-0.024	-0.017	0.021	0.042	0.000
Public
2001	1.01240	1.02469	0.95329	0.96930	0.98771	0.95955	1.09913	0.97755	1.00000
2002	1.01057	1.02983	0.95254	0.97596	0.97976	0.95285	1.09327	0.97885	1.00000
2003	1.01228	1.02728	0.94910	0.97492	0.98742	0.93642	1.10544	0.98268	1.00000
2004	1.01572	1.02092	0.96327	0.95377	0.98414	0.93032	1.10388	0.98208	1.00000
2005	1.01222	1.02443	0.96240	0.94733	0.97471	0.94371	1.13099	0.99098	1.00000
Differences between 2001 and 2005	0.000	0.000	0.009	-0.022	-0.013	-0.016	0.032	0.013	0.000

Source: Average Weekly Earnings, 6302.0, May 2006, ABS.  Since the SET data relate to May to August of 2001 and 2005, we have used comparative data for calendar years 2001 and 2005 to approximate that. 

 

177         Mean weekly earnings in all jobs.   Table 30 summarises earnings from another survey — Employee Earnings, Benefits and Trade Union Membership (EEBTUM), Australia, August 2005[17].  Like the average weekly earnings, earnings from this survey are in raw form and are not corrected for composition effects. 

Table 30          Mean weekly earnings in all jobs, by State, August, 2001 to 2005

Year	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aus
	$	$	$	$	$	$	$	$	$
August, 2001	732	683	640	664	643	598	767	734	687
August, 2002	744	695	664	711	663	615	821	777	707
August, 2003	788	723	670	735	675	648	847	762	734
August, 2004	816	758	724	747	700	668	878	843	766
August, 2005	838	806	767	821	763	697	912	858	807
Differences between August, 2001 and 2005 (%)	14.5	18.0	19.8	23.6	18.7	16.6	18.9	16.9	17.5
Mean weekly earnings relative to average
August, 2001	1.0655	0.9942	0.9316	0.9665	0.9360	0.8705	1.1164	1.0684	1.0000
August, 2002	1.0523	0.9830	0.9392	1.0057	0.9378	0.8699	1.1612	1.0990	1.0000
August, 2003	1.0736	0.9850	0.9128	1.0014	0.9196	0.8828	1.1540	1.0381	1.0000
August, 2004	1.0653	0.9896	0.9452	0.9752	0.9138	0.8721	1.1462	1.1005	1.0000
August, 2005	1.0384	0.9988	0.9504	1.0173	0.9455	0.8637	1.1301	1.0632	1.0000
Differences between August, 2001 and 2005	-0.027	0.005	0.019	0.051	0.009	-0.007	0.014	-0.005	0.000

Source: ABS Catalogue number 6310.0 Employee Earnings, Benefits and Trade Union Membership, August 2005, Table 1. 

 

178         The changes in relative mean weekly earnings generally confirm the changes in AWE, particularly those in New South Wales, Western Australia and the ACT. 

179         Labour price index.  Table 31 shows, by State, the labour price index for the private sector.  This index is corrected for changes in the composition of the labour force over time within each State, but not for differences in composition across States.  It is thus closer in concept to the modelled location effect than AWE — the modelled location effect seeks to compare like with like over time and across States.

Table 31          Labour Price Index by State, total hourly rates of pay excluding bonuses, private sector, 2001 to 2005

Year	NSW	Vic	Qld	WA	SA	Tas	ACT	NT	Aus
2001	92.2	92.1	92.4	92.2	91.3	92.4	92.2	93.1	92.2
2002	95.2	95.3	95.1	95.3	94.5	95.2	95.1	95.9	95.2
2003	98.4	98.4	98.3	98.4	98.2	98.4	98.3	98.5	98.4
2004	101.6	101.7	101.8	102.0	101.6	101.8	101.7	101.8	101.7
2005	105.3	105.9	105.9	106.7	105.1	105.9	105.3	105.9	105.7
Differences between 2001 and 2005	14.1	15.0	14.7	15.6	15.1	14.6	14.3	13.8	14.6

Source: Labour Price Index, March 2006.  2003-04 = 100 for all States.    Since the SET data relate to May to August of 2001 and 2005, we have used comparative data for calendar years 2001 and 2005 to approximate that. 

 

180         The labour price index indicates below average increases in New South Wales and above average increases in South Australia and Western Australia. 

181         Raw earnings are often above the average in Western Australia particularly in the private sector (see Table 29).  This is mainly because wage premiums are paid in some regions for isolation and in the mining industry (where the national male average wage has been 50 per cent above the 'all industry' average and steady at that level over the last five years).  However, when the Western Australian observed average wage is adjusted to reflect the national composition of labour, the observed differential falls. 

182         The labour price index changes for Western Australia are not as far above the average as the changes in the AWE (Table 29) because the effects of changes over time in the composition of Western Australia's labour force are removed.  However, there are indications that labour price movements in Western Australia increased more rapidly after the SET survey was taken in mid-2005.  

Revisiting the judgments

183         In the 2004 Review, the model results were modified to account for special features of the data and the labour markets in some smaller States. 

·                The estimated location effect for Tasmania appeared to understate its underlying wage level because of low migration of labour into Tasmania's private sector and because its pattern of economic activity was broadly comparable to regional areas of other States.  The Commission set the Tasmanian location effect at the level of the second lowest State (Queensland). 

·                The estimated location effect for the ACT underestimated the effect on ACT wage levels of the Commonwealth, the largest employer in the ACT and the ACT Government's main competitor for labour.  The Commission set the ACT location effect at the average of the calculated effects for the ACT and New South Wales. 

·                The raw wages data from SET were consistently higher than those from average weekly earnings, especially in 2001.  To deal with this and the inevitable uncertainty of model results, the Commission discounted the estimated location effects for 1997 by 5 per cent and those for 2001 by 15 per cent. 

184         The Commission is still considering whether similar adjustments should be applied to the location effects derived from 2005 data.