How does big-box retail entry affect labor productivity in durable goods retailing? A synthetic control approach

Abstract

Using data from 2001 to 2012, the effects of IKEA entry in four Swedish municipalities, 2004–2007, on labor productivity in durable goods retailing is investigated using synthetic control methods. We contribute to the literature on synthetic control methods by considering parametric specifications of the intervention effect, which improves the likelihood of identifying the effect of IKEA entry on labor productivity. Our results indicate that in three out of four entry municipalities, labor productivity increased more than in their synthetic counterparts after IKEA entry, and that larger positive effects are found in rural municipalities where the new IKEA was large relative to the existing durable goods retail market.

Appendices

Appendix 1: Placebo tests

Appendix 2: Estimations of how IKEA entry affects sales and employment

The main purpose of this research has been to investigate how entry by IKEA affects labor productivity in four Swedish municipalities subject to IKEA entry during the period 2004 – 2007. However, in this appendix we also investigate the effects of IKEA entry on CPI-adjusted sales and employment using the methods described in the main text.

As mentioned above, we measure labor productivity using the index

$$ {\text{Prod}}_{{{ }it}} = \frac{{{\text{Sales}}_{{{ }it}} /{\text{ Emp}}_{{{ }it}} }}{{{\text{National}}\_{\text{Sales}}_{{{ }t}} /{\text{ National}}\_{\text{Emp}}_{{{ }t}} }}{ } \times { }100 $$

where ${\text{Sales}}_{{{ }it}}$ is CPI-adjusted sales in durable goods retailing in municipality $i$ at time $t$, ${\mathrm{Emp}}_{ it}$ is employment in durable goods retailing in municipality $i$ at time $t$, while ${\text{National}}\_{\text{Sales}}_{{{ }t}}$ and ${\text{National}}\_{\text{Emp}}_{{{ }t}}$ are the cumulative national average sales and employees in durable goods retailing, respectively. To be consistent with the definition of labor productivity in our main estimations, we define our dependent variables in the estimations for sales and employment as follows:

$$ {\text{Sales index}}_{{{\text{it}}}} = 100{ } \times {\text{ Sales}}_{{{ }it}} /{\text{National}}\_{\text{Sales}}_{{{ }t}} $$

$$ {\text{Employment index}}_{{{\text{it}}}} = 100{ } \times {\text{ Emp}}_{{{ }it}} /{\text{ National}}\_{\text{Emp}}_{{{ }t}} $$

The results are presented in Table

Table 6 Point and interval estimation for Sales index, Employment index and Productivity

6 for the new estimations, as well as for the estimation of the productivity index (see also Table 5 in the main text). The reason for including the productivity results is that it will help understand how to interpret the results from the SCM estimations of sales and employment. While the productivity index was constructed to give the estimate of the productivity gap between the actual municipality and the synthetic control unit as a close approximation of the increase in productivity in percent, this is not the case for our new indexes. As such, we must calculate the exact increase in percent for sales and employment as given by the SCM estimations using the following formula:

$$ {\text{Increase in \% }} = 100{ } \times \left[ {\frac{{{\text{Gap}}}}{{\text{Index value synthetic control unit}}}} \right] $$

We start by discussing the result for Haparanda regarding the productivity index, and then, we turn to the new estimations. We know that the estimate of the gap, giving an approximate measure of the increase in productivity in percent, indicated an 79.62% increase in productivity due to IKEA entry. Making the exact calculation using the above equation instead results in a 77.75% = 100 × [(79.62 /102.40)] increase. Making the percentage change calculations for the Sales index and the Employment index, we find that CPI-adjusted sales increase by 490.70% and employment by 364.60% due to IKEA entry in Haparanda and that both the increase in sales and the increase in employment are statistically significant according to the interval estimations. Making similar calculations for Kalmar, Karlstad and Gothenburg, we find that sales increase by 47.00, 21.10 and 13.70%, respectively, all statistically significant according to the interval estimations. Regarding sales we thus find that IKEA entry into the municipality has a positive impact on durable goods retailing, even in larger cities such as Gothenburg.

Turning to the results related to employment in durable goods retailing, we find statistically significant increases in employment in all municipalities affected by IKEA entry. The size of the effect was 364.60% for Haparanda, while employment in durable goods retailing increased by 37.20, 18.30 and 21.10%, respectively, in Kalmar, Karlstad and Gothenburg, again confirming positive effects also in large municipalities.