The myth of intertemporal labour supply substitution: Evidence from tax holidays

Isabel Z. Martínez, Michael Siegenthaler, Emmanuel Saez 22 August 2018



Do people work more when their wages are temporarily higher? Macroeconomists believe that they do and that this is a key driver of employment fluctuations (e.g. King and Rebelo 1999). While working more when your wages are higher makes sense theoretically, can you do it in practice? Many people have jobs with fixed hours and may not be able to switch jobs easily. People have careers where harder work does not immediately translate into higher earnings. Those out of the labour force might not be able to find jobs quickly. Empirically, estimating such intertemporal labour supply responses is challenging because wages rarely change on a temporary basis for purely exogenous reasons – especially not across the whole economy (see Chetty et al. 2013 for a recent overview). 

The Swiss income tax holiday 

In a recent study (Martínez et al. 2018), we break new ground on this important question by documenting the labour supply decisions of the entire Swiss population during an exceptional tax experiment: a massive, temporary increase in workers’ after-tax wages caused by one- to two-year long income tax holidays. During the tax holidays, earnings were fully exempted from income tax, with the average tax rate falling from around 11% to zero. The marginaltax rate fell from more than 20% to zero. The holidays thus created large incentives to work more in the tax-free years. 

The tax holidays were a peculiar side effect of Switzerland’s change from a past-oriented tax system to a modern, pay-as-you-earn tax system in the late 1990s and early 2000s. To avoid double taxation, income earned during the transition had to be exempt from taxation so that people would not pay taxes both on past incomes (as in the old system) and on current income (in the new system).  Interestingly, not all Swiss cantons transitioned at the same point in time, so that the tax holidays were staggered across cantons. Two cantons had a tax holiday for years 1997 and 1998; 16 cantons had a tax holiday for the years 1999-2000; four cantons had a tax holiday in 2000 only; and three cantons had a tax holiday for the years 2001-2002.

Obviously, responses to the tax holiday would happen only if the public are well informed about the reform and understand that it generates a tax holiday. The timing of the tax holiday was discussed at length in the press well before the transition took place, making it salient to the public. In fact, various press articles discussed how working and earning more during the tax holiday was fiscally advantageous. We conclude that the tax holidays were salient, particularly in the cantons that transitioned last.

Surprisingly small labour supply effects despite large incentives 

Did the zero tax burden during the holidays affect workers’ decisions of whether or not to work? This question is studied in Figure 1 using population-wide social security earnings records matched to 2010 Census data. The figure displays the share of employed men (top panel) and women (bottom panel) in the total population aged 20-60 from 1990 to 2010. For the purpose of this column, these employment rates are only shown for three groups of cantons: 16 cantons with a tax holiday in 1999 and 2000 (in light green); four cantons with a tax holiday for 2000 (in darker green); and three cantons with tax holiday in 2001-02 (in brown). The employment rate is the fraction of individuals in the sample with positive labour earnings during the year. 

Figure 1 Effects of the tax holiday on employment rates for men (top) and women (bottom)

Notes: This figure displays the employment rate by year and groups of cantons from 1990 to 2010. The top panel is for men and the bottom panel for women. The sample in a given year is all individuals aged 20-60. The employment rate is computed as the fraction of individuals in the sample with positive earnings (either from wages or self-employment) during the year. The three groups of cantons are: 16 cantons with a tax holiday in 1999 and 2000 (in light green), 4 cantons with a tax holiday for 2000 (in darker green), 3 cantons with tax holiday in 2001 and 2002 (in brown). For each of the three groups, we represent the corresponding tax holiday periods using the vertical shading and the same colour code. In the series, the dots corresponding to tax holidays are bigger and are blanked out. The figure shows no evidence of employment effects due to the tax holiday.

Two findings are worth noting. 

  • First, all three groups of cantons follow remarkably similar trends for employment over the full period. This implies that for each group of cantons, the two other groups constitute good control groups. 
  • Second, there is no evidence of any relative increase in employment rates during the tax holidays represented by the shaded areas. This implies that a temporary tax holiday does not affect labour supply along the extensive margin, in sharp contrast to the prediction of modern macro business cycle models. 

In our paper, we also show that the tax holiday had very small effects on average earnings in the overall population – we find limited evidence of workers increasing their hours of work in response to the tax holiday.

Evidence of income shifting among high earners and the self-employed

However, we do find evidence of a response among workers who can adjust their reported earnings more easily through tax avoidance. High wage earners are able to do so by negotiating with management (or they are sometimes management itself), retiming bonus and overtime payments, and so on. The self-employed may find it easier to increase their real labour supply and they may be able to accelerate or defer payments and receipts to maximise tax holiday incomes. 

Figure 2 illustrates such responses by zooming in on high earners, defined as workers with average annual labour earnings (wages plus self-employment) above CHF200,000 in 1994-1996 (approximately $150,000 at the time), a couple years before the reform started. The figure follows these high earners and depicts their average wage earnings (top panel) and average self-employment earnings (bottom panel) by year and groups of cantons from 1990 to 2010. The top panel shows a clear and significantly larger response of wage earnings for this high-income group of around 5% excess earnings during the tax holidays. The bottom panel also shows a large spike in self-employment income for the 2001 and 2002 tax holiday of about 7-10% excess self-employment income in these years.

Figure 2 Effects of the tax holiday on high earners: Wages (top) and self-employment earnings (bottom)

Notes: This figure displays average wage earnings (top) and average self-employment earnings (bottom) by year and groups of cantons from 1990 to 2010. The sample is all individuals aged 20-60 who had average annual labour earnings (wages plus self-employment) above 200,000 CHF in 1994-1996. Earnings are expressed in 1000s of 2010 CHF (adjusted for inflation). The three groups of cantons are the same as in Figure 1. For each of the three groups, we represent the corresponding tax holiday periods using the vertical shading and the same colour code. In the series, the dots corresponding to tax holidays are bigger and are blanked out. The figure shows spikes in earnings during the tax holiday, particularly so for cantons who transitioned last.

Overall, we find surprisingly little evidence that the large drop in taxes affected the labour supply of the average worker in the population, despite workers having one or even two years to adjust. We find no evidence of a response in employment, not even for subgroups usually considered to react more strongly along this margin, such as married women. Our evidence suggests that people cannot easily adjust their labour supply to take advantage of the tax holidays. Therefore, the intertemporal labour supply responses appear to be very small on average at the macroeconomic level. However, some groups – such as high earners or the self-employed – can take advantage of the tax holidays by artificially shifting their income into the tax holiday. This evidence of small real labour supply responses overall with large tax avoidance responses for specific groups who can avoid taxes is very consistent with the large literature on behavioural responses to income taxation (Saez et al. 2012).

Implications for macroeconomics

Although this may not seem obvious at first, these results have important implications for macro models used to understand employment fluctuations during the business cycle. The reason is that our results identify a key parameter in these models – the so-called Frisch elasticity, which measures workers’ willingness to work more if wages increase temporarily. The parameter is important in business cycle models because it governs the extent to which these models can match an important fact about business cycles – namely, that recessions lead to large falls in employment but to only limited reductions in wages. If the Frisch elasticity is large, workers’ labour supply decisions can explain this fact, with even a relatively small wage decrease in a recession inducing workers to leave the labour force voluntarily. Comparing the prevailing wage with the gains from enjoying leisure time, workers decide to abstain from participating in the labour market.  Our results suggest that voluntary withdrawals do not play an important role in explaining employment reductions during recessions. Indeed, the Frisch elasticities that we estimate are orders of magnitude smaller than the Frisch elasticities that many macro business cycle models require to match the employment reductions during recessions. 


Chetty, R, A Guren, D  S Manoli, and A Weber (2013), “Does Indivisible Labor Explain the Difference between Micro and Macro Elasticities? A Meta-Analysis of Extensive Margin Elasticities”, NBER Macroeconomics Annual 27(1): 1–56.

King, R G and S T Rebelo (1999), “Resuscitating Real Business Cycles”, in J B Taylor and M Woodford (eds), Handbook of Macroeconomics, Vol. 1: 927–1007.

Martìnez, I, M Siegenthaler, and E Saez (2018), “Intertemporal labor supply substitution? Evidence from the Swiss Income Tax Holiday”, NBER Working Paper No. 24634.

Saez, E, J Slemrod, and S Giertz (2012), “The Elasticity of Taxable Income with Respect to Marginal Tax Rates: A Critical Review", Journal of Economic Literature 50(1): 3–50.



Topics:  Labour markets Taxation

Tags:  tax holidays, labour supply, wages

Postdoctoral Researcher, University of St Gallen

Labour market specialist, KOF Swiss Economic Institute, ETH Zurich

Professor of Economics and Director of the Center for Equitable Growth, University of California Berkeley


CEPR Policy Research