Technological innovation and the demand for labor by firms in expansion and recession

ABSTRACT

With firm data from the Spanish Community Innovation Survey (CIS) for the period 2003–2014, we find a positive and significant effect of innovation in the demand for labor when firms introduce product and process innovations in the same time period. The effect of innovation on the demand for labor is countercyclical, higher in the recession, after 2008, than in the expansion, before 2008, but the probability that firms innovate in product and process is counter-cyclical, i.e. lower in the recession. Altogether, the elasticity of the demand for labor to the probability that firms introduce product and process innovations remains stable throughout the sample period, at around 0.035. Innovation contributes to stabilize average employment during the cycle, more so when the innovation is in product, alone or together with process, than when it is only in process. These results are broadly consistent with product and process innovations shifting firms’ demand and production functions upwards, but differentially in expansions (less product market competition) than in contractions (more competition).

KEYWORDS:

• Product and process innovations
• demand for labor
• business cycle
• Spanish CIS panel data

JEL CLASSIFICATION:

• O31
• O32
• O33

Acknowledgments

The authors thank the Referees and Journal Editor for their comments and suggestions in the revision of the paper.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Javier Ortiz http://orcid.org/0000-0003-2114-5033

Vicente Salas Fumás http://orcid.org/0000-0001-8405-1056

Notes

1 The structural model of Crepon, Duguet, and Mairesse (1998) for the analysis of the economic effects of innovation at the firm level has been labeled CDM, for the initials of the three authors of the paper. See Hall (2011) for a survey of applications of the CDM, and Peters et al. (2013), Doraszelski and Jaumandreu (2013), Conte and Vivarelli (2014), Garcia-Quevedo, Pellegrino, and Vivarelli (2014) and Baumann and Kritikos (2016) for more recent applications. Most of the applications of the CDM approach use productivity and profitability as performance variables of the firm; in this paper, the performance variable is the demand for labor.

2 Earlier papers (Entorf and Pohlmeier 1990; Brouwer et al. (1993); König, Licht, and Buscher 1995; Klette and Førre 1998) estimate static labor demand equations and find a positive effect of product innovations on employment, while in the case of process innovations the effects on employment are non-significant or positive.

3 Papers differ between those that examine the influence of the business cycle in the volume of resources that firms dedicate to innovation, i.e. R&D expenditures (Ouyang 2011; López-Garcia, Montero, and Moral-Benito 2012), and those that examine the effect of the cycle in the timing of introducing innovation outcomes (Hingley and Park 2017; Berchicci, Tucci, and Zazzara 2013; Geroski and Walters 1995).

4 Other papers that have used the data base, alone or combined with data from other countries, to study the relationship between innovation and employment growth include Harrison et al. (2014), Segarra and Teruel (2014), Dachs et al. (2017), Calvino (2019). Only Calvino (2019) models and estimates a dynamic growth model with panel data, although no explanation is given on the way to deal with the time overlap in the responses to the innovation questions.

5 With the perpetual inventory model, the stock of R&D capital in period t, RDK_t s given by RDK_t= RDK_t-1(1-δ)+RDI_t, where δ is the depreciation rate and RDI_t is the flow of R&D investment in year t (internal and external). The depreciation rate δ  = 0.15 is the same as in Crepon, Duguet, and Mairesse (1998) and Hall (2007).

6 The PITEC database gives information on the annual flow of investment in tangible capital, but not on the stock of tangible assets of firms. We estimate the stock of capital with the perpetual inventory method, as we did with the stock of R&D capital, but now with a depreciation rate of 10% (from companies’ annual accounting reports data).

7 Martínez-Ros and Labeaga (2009) present a different analysis of the determinants that Spanish firms introduce product and process innovations in the same time period, with a different data source. D' Este, Amara, and Olmos-Pañuela (2015) model the probability that firms introduce product innovations, with Spanish CIS data, but only with cross-sectional data.

8 In the prediction, the estimated coefficient of the time dummy variable is the same for each of the three years of the time interval. The results are robust to estimations of the probit models in Table 3 and predicted latent values, with GDP growth rates rather than time dummies. They are also robust to including the perception of firms about the obstacles to innovation as explanatory control variables of the probit model (none of the estimated coefficients of the obstacles variables is statistically significant).

9 The simple correlations between the latent values of innovation output are: −0.27 between Î_D and Î_DP; 0.29 between Î_P and Î_DP, and –0.06 between Î_D and Î_P.

10 The ratio of tangible capital to output controls for differences in capital-intensive technologies across firms; the share of exports accounts for the possibility that firms increase sales abroad when internal demand falls; the human capital of employees and being part of a business group may condition the job adjustment decisions in response to external shocks From (1), the reduced demand for labor depends inversely on prices of production inputs, capital, and labor but no information on these variables for each firm and year is available in the database. The sector dummy variables, the firm-specific effects, and the time dummy variables control, among other things, for sector, firm, and time differences in input prices. Lachenmaier and Rottmann use sector-level wages as substitute for firm-level wages but when controlling for sector fixed effects, the estimated coefficient of sector wages is not statistically significant.

11 The coincidence in time of joint introduction of product and process innovation among those firms that report introducing a product or process innovation could occur in research that models employment growth rather than innovation levels. Herstad and Sandven (2015), with Norwegian CIS data, find a positive effect of the joint introduction of product and process innovation on employment growth. Calvino (2019), with Spanish CIS data from manufacturing firms, does not find robust evidence of a positive effect of joint product and process innovation on employment growth (with innovations in products new to the market), but consistent evidence that the introduction of only product innovations new to the market contributes positively to employment growth.

12 From the first-order conditions of profit maximization, we have $\frac{\mathrm{\Delta }R}{\mathrm{\Delta }Q}\frac{\mathrm{\Delta }Q}{\mathrm{\Delta }K}K=\alpha \mu R=cK$ and $\frac{\mathrm{\Delta }R}{\mathrm{\Delta }Q}\frac{\mathrm{\Delta }Q}{\mathrm{\Delta }L}L=\text{β}\mu R=wL$, where R=BQ^μ.

Additional information

Funding

The funding for this work was supported by Government of Aragón (Group Reference CREVALOR: S42_17R) and co-financed with FEDER 2014-2020 “Construyendo Europa desde Aragón”, and the Spanish Ministry of Economy and Competitiveness – FEDER (ECO2017-86305-C4-3-R).