PUBLISHED: June 2021
By William Keating, Christopher Smart and Samuel Gow [1].
Document download
Contents
Overview
1. Literature review
2. Data
3. Methodology
4. Empirical results
5. Discussion
6. Conclusion
References
AbstractHow responsive are the remuneration and hiring decisions of Victorian businesses to payroll tax? Using a natural experiment based on the payroll tax rate being reduced in regional Victoria but not metropolitan Melbourne, we show businesses facing lower payroll tax rates increase their total wage bill. Our difference-in-difference estimates demonstrate liable business’s total wages increased by approximately 7 to 9 per cent in response to the payroll tax rate cuts totalling 2.425 percentage points over a 24‑month period. Notably, the effect was largely concentrated in the first year following an initial tax rate cut of 1.2 percentage points. |
Overview
States and Territories in Australia levy payroll tax on the wage bill of liable employers. Payroll taxes are usually imposed at a flat rate, after a tax-free threshold that applies to an employer’s total wage bill (not the individual employee) has been removed.
Payroll tax is one of the States’ and Territories’ most important own-source revenues. In 2019–20, total state and territory payroll tax revenue was $27.0 billion representing 17 per cent of own-source revenue — making this tax the largest collected by states and territories.
State and territory governments have generally reduced the share of revenue raised through payroll taxes over time, in attempts to encourage businesses to increase jobs in their jurisdiction. By evaluating a policy designed to make regional Victoria more attractive to employers, this paper adds to the narrow empirical literature relating to the efficacy of such reforms in Australia.
The payroll tax rate in Victoria has been set at 4.85 per cent since 2014–15. In the 2017–18 financial year, a lower payroll tax rate of 3.65 per cent was introduced for regional businesses. The rate was reduced further to 2.425 per cent from 1 July 2018. This payroll tax cut created a natural experiment which we exploit using difference-in-difference estimators. This paper aims to measure the impact of this cut on business behaviour by comparing the wage bill of businesses eligible for the regional payroll tax cut against the trends among ineligible businesses.
The structure of this article is as follows. Section 1 provides background detail on Victoria’s payroll tax system and reviews relevant literature. Section 2 describes the Victorian State Revenue Office (SRO) data and how it differs from more widely used data sources. Section 3 details our estimation strategy. Section 4 describes and discusses the results, and Section 5 concludes.
1. Literature review
Research on payroll tax in Australia tends to revolve around simulated deadweight loss and efficiency such as KPMG Econtech (2011) and Nassios, et al. (2019). There is little empirical literature which attempts to measure business decisions made due to changes in payroll tax policy in Australia.
Two prominent Australian papers use datasets in the Business Longitudinal Analysis Data Environment (BLADE) to measure effects of changes in payroll tax on business behaviour. Ralston (2020) finds little evidence for behavioural effects from payroll taxes. Majeed & Sinning (2019) did not find any evidence of changes affecting wages, employment, or capital expenditure.
Both studies evaluate the effect of tax-free thresholds (and the effective tax rate these imply), rather than the tax rate. In contrast, the natural experiment created in Victoria by the introduction of the regional employer payroll tax rate, provides an opportunity to investigate the effect of rate changes. Unlike a tax-free threshold, rates have a proportional and correctly identifiable impact for all businesses and therefore are more likely to have an identifiable treatment effect.
Studies based on thresholds likely also suffer from the datasets used not being particularly effective for estimating the actual threshold each business is able to deduct or properly identifying wages subject to the tax. Not all businesses can deduct the full amount of each state’s or territory’s tax-free threshold from their wages when calculating their payroll tax. For example, when a business:
- is part of a group, the threshold is assigned to one business in that group or shared across the group rather than each business being able to deduct the full amount [2]; and
- operates nationally, the amount they can deduct will be determined by multiplying the state’s or territory’s legislated threshold by wages paid in that state divided by their national wages.
Using datasets that do not properly identify payroll tax groups, including those in BLADE data, would result in multiple thresholds being attributed to a group (one for each business), masking the effects of changes to the threshold on these businesses. The BLADE-based papers would have partially mitigated the issues with grouping provisions by discarding businesses with wage bills larger than some threshold value, which are more likely to be part of a business group. However, this issue could still confound the results.
Furthermore, not all forms of wages are subject to payroll tax (for example, wages paid to parents on parental leave are exempt from payroll tax). Identifying the proportion of wages which are taxable is near‑impossible using datasets such as those in BLADE that aggregate business wage data. This paper uses data on actual payroll liable Victorian wages.
Internationally, the literature around payroll tax-driven changes to business behaviour is more comprehensive and likely less prone to data issues. This literature provides evidence that targeted payroll tax cuts do effect individual wages of employees and total employees hired.
Saez, Schoefer & Seim (2019a) exploit an age-based payroll tax concession in Sweden to measure the effect of payroll tax cuts on youth unemployment using both worker and business level data. The study finds concessions positively affected the employment rate of eligible younger workers, but not their after-tax wages. Further research by the same authors — Saez, Schoefer & Seim (2019b) — finds the long-run effects on employment of the Swedish payroll tax cuts for young workers were larger than in the short-run and persistent in that they continued after eligible workers became too old (and therefore ineligible) as well as after the policy was repealed.
Stokke (2016) in Norway observed limited effects of regional payroll tax cuts on employment and suggested increased wages is a more likely result. This paper also notes these wage increases become less strong as worker education increases. Cruces, Galiani & Kidyba (2010) in Argentina, also observed limited effects of regional payroll tax cuts on employment but measurable increases to wages. Similarly, Bennmarker, Mellander & Öckert (2009) found that payroll tax cuts in Sweden had a measurable effect on the average wage bill per employee.
Korkeamäki & Uusitalo (2006) performed similar analysis in Finland but used matching techniques on business pairs in order to control for business and industry effects and found a reduction in payroll taxes led to measurably faster wage growth in the target region.
Interestingly Ku, Schönberg & Schreiner (2018) found businesses responded to the abolition of regional Norwegian payroll tax cuts by firing workers to compensate for the larger wage bill, as they could not pass the tax increase on to workers. This finding demonstrates the response of businesses to payroll tax may not be symmetric. This would imply payroll tax initiatives should not be pursued under an assumption they may be reversed without consequences.
The natural experiment created by the introduction of the regional employer rate in Victoria provides an opportunity to apply the basic concept of exploiting place-based changes in payroll tax policy used in this international research to an Australian context. Though the data available doesn’t allow us to distinguish between the effect of the tax rate on wages and employment, as we only have access to business level wage data, the natural experiment may allow us to better understand the magnitude of the effect of the tax rate on total wage bills.
2. Data
The analysis in this paper uses data from the Victorian SRO payroll tax unit records, for the financial years 2016–17 to 2018–19. The 2016–17 financial year was the last in which regional employers paid the full rate of payroll tax, 4.85 per cent. In the 2017–18 financial year, a lower payroll tax rate of 3.65 per cent was introduced for regional businesses. The rate was reduced further to 2.425 per cent from 1 July 2018.
The SRO payroll tax unit records are administrative data collected from businesses to allow the SRO to effectively manage the calculation and collection of payroll tax liabilities. The SRO payroll tax unit records possess some unique features compared to datasets used in similar studies.
- No identification strategy is needed to remove payroll exempt businesses, as these are not recorded by the SRO or have a tax liability of zero. Other studies have had to either classify businesses manually or exclude data based on self-reported industry codes, which may be unreliable due to widespread inaccuracies in self-reporting of industry codes.
- Victorian wage bills are clearly reported by businesses and do not need to be derived or estimated from Australia-wide wages.
However, the payroll tax unit records do not contain any information about either the number of employees or any form of compensation per employee or hours worked. This means any observable effect of the regional payroll tax rate on wages will be at a business or group level.
Furthermore, though tax liability data are accurately reported (and heavily scrutinised), user entry is often problematic for wage bills. In cases where wages are clearly misstated, such as when the taxable wage bill is entered instead of the total wage bill, wages are imputed from the tax liability of the payer.
Businesses’ regional employer status is recorded in the dataset. This means a business’s eligibility for the regional rates does not need to be inferred from locational data, such as postcodes, which may represent the location of a business’s (or its business group’s) headquarters, not where most workers are located. However, the regional employer status of businesses is only recorded from the first treatment year. So, we cannot differentiate between businesses that would have qualified as regional employers in 2016–17 and those that shifted employees (or increased regional wages) to qualify in 2017–18.
Businesses grouped for payroll tax purposes are separately assessed for eligibility for the lower regional employer rate of payroll tax. For example, a business group could consist of three businesses located in Melbourne, Geelong and Ballarat. The first business would be ineligible for the regional rate due to its location (though this criterion was revoked from 1 July 2019) and would be liable to pay tax at the rate of 4.85 per cent. Eligibility for the Geelong and Ballarat businesses would depend on the location of their employees. If more than 85 per cent of their wage bill was paid to employees performing work in regional Victoria, they would be eligible for the regional employer reduction, even on the small amount of wages paid to employees located in Melbourne.
Where one business in a business group is eligible for the discounted rate but the other two are not, that member is eligible for the lower payroll tax rate of 3.65 per cent (2.425 per cent from 1 July 2018). The other two businesses in the business group will pay payroll tax at the rate of 4.85 per cent.
3. Methodology
We use group level wage panel data to estimate the impact of the regional payroll tax cut using a difference-in-differences (DiD) strategy. This strategy compares the effect of the regional payroll tax rate on groups that include one or more regional employer with a control group of groups comprised entirely of non‑regional employers. DiD has become common in a policy evaluation and micro-econometric context since its usage in several seminal papers (Card & Krueger, 1994; Ashenfelter & Card, 1985). Discussed below in Box 1, DiD exploits variation in the application of a policy change across groups and time to identify the causal effect of the policy.
|
Box 1: Difference-in-differences Difference-in-differences (DiD) is an econometric method that compares outcomes across groups. In its simplest form, the outcome is observed for two groups in two time periods. Neither of the groups are exposed to ‘treatment’ in the first period (in our case, treatment is defined as lowering the payroll tax rate for regional employers). In the second period, however, one group is exposed to the treatment, while the other control group is not (Angrist & Pischke, 2008). Under several identifying assumptions, such as common trends in the absence of treatment, the difference in differences over time between the two groups then represents the causal effect of the treatment. In our example, metropolitan businesses form the control group, and businesses which receive the lower regional tax rate are the treatment group (they have a sudden drop in their tax rate). The advantage of the DiD approach is that broad macroeconomic trends in payrolls and employment are absorbed by the control and treatment groups, and so will not confound the analysis. |
A similar approach has been used in overseas payroll tax assessments (Bennmarker, et al., 2009) and prior studies in Australia such as those described in Section 2 (Majeed & Sinning, 2019). In comparison to the latter work, we benefit from a business-level fixed effects specification in addition to direct tax status and liable wage reporting.
The key identifying assumption of this technique is parallel trends. After accounting for any covariates (in our case only fixed effects), both groups that include one or more regional employer and the control group of non-regional employers would follow the same growth path in the absence of policy intervention. Were this not the case, the estimate does not have a causal interpretation. In our context, a theoretical basis for this assumption is that payrolls are fundamentally driven by the level of economic activity, which is linked across the state. We further attempt to mitigate this issue by restricting the sample to businesses on the border of metropolitan Melbourne and regional Victoria which are likely to share most unobserved shocks not otherwise controlled for.
Our DiD specification is the standard form for panels with several years and units, which includes fixed effects at the group level and for every distinct year (‘two-way’ fixed effects), as opposed to the simpler indicators as discussed in Box 1.
The relevant regression equations are as follows:
(1)
(2)
- wi,t refers to the natural logarithm of Victorian taxable wages (payrolls) for business i in period t
- δi and γt refer to business and time fixed effects for business i and time t respectively
- Di,t is an indicator variable equal to 1 if business i was eligible for the regional payroll tax rate time t, and 0 otherwise
Specifications (1) and (2) are very similar, except (1) identifies the cumulative effects of the two policy changes (the first and second rate reductions) independently as β1 and β2. β3 identifies the pre-post average impact of all regional tax changes to date — it numerically equates to a weighted average of β1 and β2, but provides a useful evaluation of the average effect on payrolls to date.
We calculate standard errors for all estimates clustered at the panel unit (business, in our case) level, as recommended by Bertrand et al. (2004).
While we are confident parallel trends is a reasonable assumption across the data, we apply this model to two subsets of the dataset, being:
- the complete dataset of businesses in the SRO unit records; and
- a subset of businesses in areas near the border of metropolitan Melbourne and regional Victoria. These border areas are shown below in Figure 1 and Table 1.
In both cases, the treatment group is made up of groups that included at least one business that claimed the regional employer rate and the control group is made up of groups comprised entirely of businesses that paid payroll tax at 4.85 per cent.
We have chosen to analyse the effect of the tax policy change on groups rather than individual businesses as responses to the treatment may comprise an income and a substitution effect. Groups can respond to the reduced rate of tax payable on regional employees by either employing more regional employees, as the cost of regional employees has fallen (the substitution effect), or employing more staff in any or all of their businesses, as they have increased purchasing power (the income effect). If we assessed each business individually, then the income effect as it relates to the metropolitan part(s) of grouped businesses would not be captured as an effect of the treatment.
Limiting analysis to the subset of business groups around the border provides an initial robustness check against geographically based confounding factors. For example, the onset of drought or bushfires in regional areas in the same years we measure would confound our estimates as these trends would not apply equally to metropolitan and regional businesses.
Using the subset of border businesses should minimise the potential for unobserved confounding variation, including by removing inner-city-based businesses (which may have substantially different growth trends) and retaining a significant number of both regional and non-regional employers. We include regressions of specifications (1) and (2) on both samples in Section 4.2, with the whole‑of‑state results primarily acting as a validity check.
Figure 1: Border business area
Table 1: Metropolitan-Regional border LGAs
|
Metropolitan |
|
|
Greater Geelong City Council |
Wyndham City Council |
|
Moorabool Shire Council |
Melton City Council |
|
Ballarat City Council |
Hume City Council |
|
Macedon Ranges Shire Council |
Whittlesea City Council |
|
Mitchell Shire Council |
Nillumbik Shire Council |
|
Murrindindi Shire Council |
Cardinia Shire Council |
|
Mansfield Shire Council |
|
|
Baw Baw Shire Council |
|
|
Yarra Ranges Shire Council |
|
To alleviate concerns regarding treatment status changes — groups switching between regional and non-regional status between the treatment years — we also present results with these (switching) groups dropped.
One caveat of the DiD technique applied here is that, due to regional and metropolitan businesses drawing from a broadly shared labour force (i.e. there is minimal distinction between potential workers), increases in regional payrolls may, in part, be a result of reductions in metropolitan wages. These general equilibrium effects are not separately identified under this research design (Ku, et al., 2018, p. 13). However, such effects are still captured in our estimates. This means, in part, our results cannot be interpreted generally for business responses to reductions in the overall payroll tax rate as the increase in wages may be zero sum. Uniform payroll tax rate cuts do not incentivise region-based labour shifting in the same way targeted variation does, so the impacts are likely to be smaller.
4. Empirical results
4.1 Evaluating parallel trends
We plot the trends of the variable of interest (log Victorian wages) among groups with regional employers and groups without regional employers and examine their growth paths. Graphical evidence that the trends are very different would potentially invalidate our identification strategy. The wage trends are plotted on an index against the relevant group average in 2015‑16.
Figure 2: Parallel Trends Plots, 2015–16 to 2018–19
*Groups which recorded as regional payers in either 2017-18 or 2018-19 are carried backwards and assigned as 'regional in earlier years. The mean log-wages show the average of the logarithm of firm wage wills, across the firms in each subset shown.
With regional and without regional average log wages appear to follow shared paths, with a significant increase following the initial regional payroll tax change as demonstrated by the increased slope on the yellow series between 2016–17 and 2017–18. This suggests parallel trends is a reasonable assumption and an effect is present in the data.
4.2 Difference-in-differences estimation results
We estimated both DiD specifications as described in Section 3by ordinary least squares (OLS), with the results shown in Table 2.
We find positive increases to businesses’ total wage bills from the regional tax rate in Victoria as a whole and within border areas. Overall, the effect of rate cuts up to 2018–19 was an increase in the value of the regional-employer payrolls of 6.8 per cent for all businesses in Victoria, and 7.9 per cent for all those within border areas.
Table 2: Fixed Effects Regressions — Border Businesses and All Businesses
| Log Victorian Wages | ||||
| Border Businesses | All Businesses | |||
|
(1) |
(2) |
(3) |
(4) |
|
|
Diff-in-Diff 2017–18 (β1) |
0.070*** |
0.071*** |
||
|
|
(0.014) |
|
(0.009) | |
|
Diff-in-Diff 2018–19 (β2) |
0.088*** |
0.066*** |
||
|
|
(0.017) |
|
(0.010) | |
|
Diff-in-Diff Aggregate (β3) |
0.079*** |
0.068*** |
||
|
|
|
(0.015) |
|
(0.009) |
|
Fixed effects: |
|
|
|
|
|
Yes |
Yes |
Yes |
Yes |
|
Yes |
Yes |
Yes |
Yes |
|
Businesses |
6,661 |
6,661 |
46,185 |
46,185 |
|
Years |
4 |
4 |
4 |
4 |
|
Observations |
21,927 |
21,927 |
149,341 |
149,341 |
Notes:
***Significant at the 1 percent level.
**Significant at the 5 percent level.
*Significant at the 10 percent level.
Under our assumption of conditional parallel trends, these findings indicate that the introduction of the regional payroll tax rate increased regional business payrolls, which is significant statistically and with a noticeable large magnitude.
We conducted hypothesis tests to evaluate whether there was a statistically significant difference in the coefficients for each treatment (β1 and β2). The results of these tests, shown below in Table 3, demonstrate the null hypothesis of equal coefficients cannot be rejected in the data. This means we cannot conclude there was any increased effect on payrolls moving from the initial regional payroll tax cut in 2017-18 to the subsequent rate in 2018-19. This result is somewhat counterintuitive and further work confirming this outcome would be prudent.
Table 3: Testing for coefficient variation — H0:β1=β2
|
Model |
χ2 statistic |
p-value |
|
Marginal businesses (model (1)) |
1.769 |
0.183 |
|
All businesses (model (3)) |
0.473 |
0.491 |
4.3 Robustness check and additional tests
The primary robustness check of our results is the application of the methodology to the subset of near-border businesses as well as the whole-of-state dataset. However, as some businesses switch treatment status, this is a potential area which could confuse the interpretation of the primary results. A business which is not treated in 2017‑18 but is in 2018-19, will be affected differently to those treated in both years. The same issue arises when businesses exit the treated group. This necessitates a robustness check to determine these businesses do not dramatically change the outcome of our analysis. Though the study design consisting of two distinct treatment estimators validates the primary results, we assess the robustness of the results by discarding all businesses which did not have a consistent treatment status across the policy years. This results in 261 businesses (or 3.9 per cent) being dropped from the border business sample.
Shown in Table 4, this approach yields a smaller estimate for each year of the policy, though still statistically significant (within half a percentage point of the 1 per cent level). The same coefficient hypothesis testing as in Table 3 was also conducted for this model, again returning insufficient evidence to conclude the rate reduction had a further marginal effect between 2017-18 and 2018-19.
Table 4: Regression results — border business subset, switching businesses dropped
|
Log Victorian wages |
||
|
(5) |
(6) |
|
|
Diff-in-Diff 2017–18 (β1) |
0.044*** |
|
|
|
(0.016) | |
|
Diff-in-Diff 2018–19 (β2) |
0.037** |
|
|
|
(0.017) | |
|
Diff-in-Diff Aggregate (β3) |
0.041*** |
|
|
|
(0.015) | |
|
Fixed effects: |
|
|
|
Yes |
Yes |
|
Yes |
Yes |
|
Businesses |
6,400 |
6,400 |
|
Years |
4 |
4 |
|
Observations |
20,013 |
20,013 |
Notes:
***Significant at the 1 percent level.
**Significant at the 5 percent level.
*Significant at the 10 percent level.
5. Discussion
The estimates shown in Section 4.2 are useful from a policy analysis standpoint. Our findings demonstrate businesses that received the regional rate of payroll tax increased total wages at a faster rate than businesses which were not eligible. However, for several reasons, the interpretation of this disparity is unclear without further research.
The relative effects of the policy interventions in 2017-18 and 2018-19 highlight this. Though the rate cut in 2018-19 was larger than the one in 2017-18, its marginal impact was statistically insignificant, contrary to the response to the first reduction. The elasticity of business-level aggregate employee remuneration with respect to the payroll tax rate in the first year was -0.287, but the cumulative elasticity over both periods is lower at -0.136. This is unexpected if the benefits were derived purely from the reduced marginal costs induced by payroll tax cuts, as these should be reasonably linear.
We considered several hypothetical explanations for this observed reaction. First, full adjustment in advance of subsequent rate changes — that is, businesses may have reacted to future reductions in the initial year. However, the timing of regional rate announcements means this effect seems unlikely [3]. We are sceptical of this because the regional rates taken up in 2017-18 and 2018-19 were announced in their respective financial years’ Victorian Government budget. It is possible businesses foresaw a future policy agenda of further cuts, but there is little chance this effect would be sufficient to explain the subdued reaction to the second cut we observe. An alternative explanation is increased regional payrolls in the first policy year resulted from a larger workforce due to improved competitiveness with metropolitan businesses or relocating resources within businesses.
Under the first explanation, the primary benefit to regional businesses came from having any significant cost reduction, and so the subsequent change may have minimal benefits. Under the second explanation, businesses benefited from the regional rate reduction by relocating employees to qualify rather than changing remuneration. These explanations are difficult to fully assess at this time but could be assisted by investigating the impact of further legislated regional rate changes.
We find the alternative explanation compelling and, while further research is warranted, it suggests the primary benefit from the regional payroll tax rate may be yielded by a non-negligible tax advantage of businesses employing staff in regional locations compared to metropolitan locations. The diminished response to the secondary rate cut suggests this is a nonlinear effect, possibly resulting from work that is significantly substitutable between regions being captured in the initial change. However, reviewing Figure 2 brings this substitution effect into question — there is no clearly visible reduction in payroll growth for non-regional businesses, as would be expected if regional businesses were ‘crowding out’ metropolitan employers.
It is impossible to say, however, whether the nonlinearity means future rate cuts will be ineffective. This is primarily because there may be a step-functional relationship where cuts elicit reactions as they pass thresholds, as opposed to a diminishing effect in general. As DTF receives periodic updates to this administrative data from the SRO, it would be prudent to continue to monitor the trends among regional and non-regional businesses using the framework set out in this paper.
Lastly, as our dataset does not include information on employee numbers or work hours, we cannot distinguish between employment (hours worked) and wage (increased pay per hour) effects. This is a significant avenue for future inquiry but would require additional data.
6. Conclusion
Our analysis concludes cutting the rate of payroll tax for regional Victorian employers from 4.85 per cent to 3.65 per cent had a statistically significant positive effect on business-level aggregate employee remuneration in the first financial year it was introduced. However, the effect of the further reduction of the regional employer rate from 3.65 per cent to 2.2425 per cent from 1 July 2018 was not statistically differentiable. These results suggest the regional employer rate has been successful in influencing the behaviour of eligible regional businesses in 2017-18.
The second finding suggests subsequent changes in regional payroll tax rates may not motivate further increases in regional wage bills. The significant change in wage bills in 2017‑18 may have resulted primarily from improved competitiveness of employing workers in regional businesses relative to metropolitan businesses. We hypothesise that subsequent changes to the tax rate may have effects on payrolls depending on if they pass some ‘threshold’ levels required for businesses to relocate or increase activity. As further regional payroll tax rate cuts in Victoria have been planned, this will be observable and provides an avenue for future work to identify these ‘threshold’ levels and to confirm that this finding is valid.
The results may also have implications for state and territory policies designed to attract businesses or increase employment by lowering payroll tax rates. If business-level aggregate employee remuneration responds nonlinearly to reductions in payroll tax rates, then policies that lower payroll tax rates (including threshold increases) to attract businesses may only be effective if the reductions exceed ‘threshold’ levels required for businesses to relocate activity [4]. If this is the case, a State or Territory reducing the rate of its payroll tax may do little to further encourage business to relocate if the rate reduction does not exceed some threshold.
The use of SRO payroll tax unit records mitigates many issues that have historically made evaluating payroll tax policy difficult, such as grouping provisions and state wage bill separation. Our significant results demonstrate the benefit of using state level unit records to answer state and territory tax policy questions.
This study raises questions about how we view and discuss payroll tax policy in Australia. This work is important as discussions around payroll tax policy in Australia have seldom included empirical evaluations of the efficacy of payroll tax policy initiatives. Once policy has been implemented, the ability to empirically test is greatly increased when tax record data can be obtained. We believe this is key to discussions of tax reform work as it provides governments with greater information as to what initiatives should be implemented in future and encourage other agencies with access to similar data to evaluate the empirical effects of their own payroll tax policies.
References
Angrist, J. D. & Pischke, J.-S., 2008. Mostly Harmless Econometrics: An Empiricist's Companion. Princeton: Princeton University Press.
Ashenfelter, O. & Card, D., 1985. Using the Longitudinal Structure of Earnings to Estimate the Effect of Training Programs. The Review of Economics and Statistics, 67(4), pp. 648-660.
Bennmarker, H., Mellander, E. & Öckert, B., 2009. Do regional payroll tax reductions boost employment?. Labour Economics, 16(5), pp. 480-489.
Bertrand, M., Duflo, E. & Mullainathan, S., 2004. How much should we trust differences-in-differences estimates?. The Quarterly Journal of Economics, 119(1), pp. 249-275.
Card, D. & Krueger, A., 1994. Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania. American Economic Review, 84(4), pp. 772-793.
Cruces, G., Galiani, S. & Kidyba, S., 2010. Payroll taxes, wages and employment: Identification through policy changes. Labour Economics, 17(4), pp. 743-749.
Korkeamäki, O. & Uusitalo, R., 2006. Employment effects of a payroll-tax cut - evidence from a regional tax exemption experiment. IFAU Working Paper Series, Volume 10.
KPMG Econtech, 2011. Economic Analysis of the Impacts of Using GST to Reform Taxes. [Online]
Available at: https://www.cpaaustralia.com.au/~/media/corporate/allfiles/document/pro…
Ku, H., Schönberg, U. & Schreiner, R. C., 2018. How Do Businesses Respond to Place-Based Tax Incentives?. CReAM Discussion Paper Series, Volume 1811.
Majeed, O. & Sinning, M. G., 2019. Do payroll tax cuts for Australian businesses affect their use of capital and labor?. TTPI Working Papers, Volume 8.
Nassios, J. et al., 2019. The Economic Impact and Efficiency of State and Federal Taxes in Australia. CoPS Working Paper, Volume G-289.
Ralston, B., 2020. Does payroll tax affect business behaviour?. Economic Papers: A journal of applied economics and policy, 39(1), pp. 15-27.
Saez, E., Schoefer, B. & Seim, D., 2019a. Hysteresis from Employer Subsidies. NBER Working Paper, Volume 26391.
Saez, E., Schoefer, B. & Seim, D., 2019b. Payroll Taxes, Business Behavior, and Rent Sharing: Evidence from a Young Workers’ Tax Cut in Sweden. American Economic Review, 109(5), p. 1717–1763.
Stokke, H., 2016. Regional payroll tax cuts and individual wages: Heterogeneous effects across education groups. European Regional Science Association Conference Papers.
Footnotes
[1] The authors would like to thank the following Department of Treasury and Finance (DTF) staff for their comments: James Brugler, Omid Mousavi, Hao Wang, Shenglang Yang, Gillian Thornton, and Georgina Grant. The views expressed in this paper are those of the authors and do not necessarily reflect the views of DTF.
[2] Businesses can be grouped for payroll tax purposes if there is a link between the businesses. A link exists where:
- Corporations are related bodies corporate within the meaning of s50 of the Corporations Act 2001 (this situation is commonly known as a holding and subsidiary relationship).
- Employees are shared between businesses.
- The same person has (or the same persons together have), a controlling interest in at least two businesses.
[3] Change for 2018-19 was read first in May 2018. Technically this could allow pre-emptive behaviour by businesses, but this seems unlikely given the short time frame (remembering that 2018-19 ended on 30 June 2019). The relevant bills and timelines can be found in the State Taxation Acts Amendment Bill 2017 and State Taxation Acts Amendment Bill 2018.
[4] A threshold increase reduces the effective payroll tax rate for a business by reducing the share of its wages subject to tax. For example, the scheduled increase in Victoria’s tax-free threshold from $650,000 to $700,000 will reduce the effective tax rate, or tax as a share of total wages, for a business with wages of $1 million from 1.7 per cent to 1.45 per cent.