Productivity, Taxes, and Hours Worked in Spain: 1970-2014 Juan Carlos Conesa Stony Brook University Timothy J. Kehoe University of Minnesota and Federal Reserve Bank of Minneapolis
February 2017
Questions: What has driven economic growth and fluctuations in Spain since 1970? Changes in productivity or changes in factor inputs? What has been the impact of changes in taxes on aggregate hours worked and on output?
Framework: Cole-Ohanian (1999) and Kehoe-Prescott (2002): growth accounting + general equilibrium growth model Other researchers role of labor market institutions: Blanchard and Jimeno (1999), Blanchard and Summers (1986), Sargent and Ljungqvist (1995, 1999, 2000); sectoral composition: Marimon and Zilibotti (1998), Rogerson (2004)
2
Growth accounting Yt At Kt L1t 1
1 Kt Yt 1 At Nt Yt
Lt N t
Note: On a balanced growth path, the capital-output ratio, K t / Yt , and hours per working age person, Lt / N t , are constant.
3
Growth accounting data Yt : real GDP factor prices (national income accounts – SNA93, 1954-2014)
X t : real investment (national income accounts) Lt : hours worked (labor surveys, 1959-1980, 1977-2014) Construct capital stocks:
Kt 1 1 Kt X t . Total factor productivity is the residual:
At Yt Kt L1t . 4
Growth accounting for Spain 225
GDP per working age person
200
Index 1970=100
175
TFP factor
150
Capital factor 125
100
Labor factor
75
50 1970
1980
1990
2000
2010
5
Growth Accounting for the US 1970-2005 250 230 210
)190 0 0 1170 = 0 7150 9 1 ( x130 e d n i110
Yt / N t
At1/(1 ) Lt / N t
K t / Yt
/(1 )
90 70 50 1970
1975
1980
1985
1990
1995
200 0
2005
6
Model Stand-in consumer: max
t t 1970 [ log Ct (1 )log( Nt h Lt )]
s.t. (1 tc )Ct Kt 1 Kt (1 t ) wt Lt (1 tk )(rt ) Kt Tt . K1970 K1970 .
Feasibility: Ct Kt 1 (1 ) Kt Gt At Kt L1t .
Government budget constraint: Gt Tt tcCt t wt Lt tk (rt ) Kt . 7
Calibration 0.045 ( t 1970 2014
Kt / 45 0.1385 ) Yt
0.371 Tax data: 1970-2014 using methodology of Mendoza, Razin, and Tesar (1994). Income tax rates: marginal > average (by a factor of 1.8). Note that, because the model includes indirect taxes, national income is GDP at factor prices, not GDP at market prices. Yt At Kt L1t rt Kt wt Lt GDPt rt Kt wt Lt tC Ct .
8
Parameters of stand-in consumer’s utility function and : estimated using 1970-2014 data. First order conditions: (1 tc1 )Ct 1 1 (1 tc )Ct 1 (1 tk )(rt ) (1 tc )Ct . c (1 t )Ct (1 t ) wt ( N t h Lt )
9
Numerical experiments Set K1970 K1970 . Set At Yt K t L1t , t 1970,..., 2014 . 1. Base case – set taxes equal to their actual values 1970-2014. Estimate 0.9736 and 0.2683 for 1970-2014 data. 2. Constant taxes – set taxes tc , t , tk and government consumption equal to average values 1970-2014. Estimate 0.9688 and 0.2456 for 1970-2014 data.
10
Effective marginal tax rates in Spain 0.6
Taxes on capital
0.5
Taxes on labor 0.4
0.3
0.2
Taxes on consumption
0.1
0 1970
1980
1990
2000
2010
11
Government consumption in Spain 25
Percent GDP
20
15
10
5 1970
1980
1990
2000
2010
12
Results The model with taxes performs much better relative to the data than a model with constant taxes. The trend of lower hours worked can be associated to the evolution of taxes and government expenditure. Significant departures at business cycle frequencies. Hours worked more volatile in the data than in the model.
13
Figure 4: GDP per working age person 40000
Constant taxes
35000
2010 Euros
30000
Benchmark model 25000
20000
Data
15000
10000 1970
1980
1990
2000
2010
14
Figure 5: Capital output ratio 4.5
Constant taxes
4.0
Ratio
3.5
Benchmark model 3.0
Data 2.5
2.0 1970
1980
1990
2000
2010
15
Figure 6: Hours worked in the benchmark model 26
Constant taxes
24
Hours per week
22
20
Benchmark model
18
16
Data 14
12 1970
1980
1990
2000
2010
16
We learn Unlike in Prescott (2002) taxes are not everything in explaining hours worked Key difference: the study of transitional dynamics rather than focusing on steady state comparisons Key aspect: what does the government do with taxes?
17