Dalarna University's logo and link to the university's website

This paper presents the first results from a newly developed BI (business intelligence) model for the Swedish horse industry. Compared to previous studies of the impact from the horse industry we are able to present both figures for the national level as well as a decomposition to regional levels.

The size of the horse industry in Sweden for 2016, is measured departing from the expenditure approach, i.e. summing the final use of horse related goods and services. One implication of the approach is that results are comparable with overall GDP figures for a country and with other subsectors of an economy, e.g. the tourism industry or the car producing industry. The model has two main inputs. Firstly, estimates of the geographical position of all Sweden’s 355.500 horses of different type and use, based on JBVs statistics and postal codes from horse associations. Secondly, estimates of the horse owner’s consumption pattern related to their leisure or professional use. Other horse related activities like riding schools, education, race tracks, betting etc. is treated separately, measured and added to the overall calculation.

The preliminary results indicate that the horse industry in Sweden amounts to somewhere in the interval of 26-32 Billion SEK corresponding to approximately 0,5-0,6 percentage of Swedish GDP. Looking at regional variations, the region of Skåne has most horses and consistently also the region with largest share of the horse industry.

Denna rapport presenterar skattningar av den svenska hästnäringens samhällsekonomiska effekter på nationell och regional nivå. År 2016 uppskattades hästnäringen i Sverige generera en direkt omsättning på 31,3 miljarder kronor motsvarande en sysselsättning om ca 16 900 helårsarbetskrafter. Skattningarna bygger på en modell för hästnäringens samhällsekonomi som har sin utgångspunkt i summering av total konsumtion av hästrelaterade varor och tjänster i Sverige under ett år. Rapporten presenterar även den Business Intelligence (BI) lösning som ligger till grund samt hur bakgrundsdata har samlats in, lagrats och paketerats för att slutligen presenteras med hjälp av ett BI-verktyg (https://hastnaringen-i-siffror.se).

This thesis consists of a summary and five self-contained papers addressing dynamics of firms in the Swedish wholesale trade sector.

Paper [1] focuses upon determinants of new firm formation in the Swedish wholesale trade sector, using two definitions of firms’ relevant markets, markets defined as administrative areas, and markets based on a cost minimizing behavior of retailers. The paper shows that new entering firms tend to avoid regions with already high concentration of other firms in the same branch of wholesaling, while right-of-the-center local government and quality of the infrastructure have positive impacts upon entry of new firms. The signs of the estimated coefficients remain the same regardless which definition of relevant market is used, while the size of the coefficients is generally higher once relevant markets delineated on the cost-minimizing assumption of retailers are used.

Paper [2] analyses determinant of firm relocation, distinguishing between the role of the factors in in-migration municipalities and out-migration municipalities. The results of the analysis indicate that firm-specific factors, such as profits, age and size of the firm are negatively related to the firm’s decision to relocate. Furthermore, firms seems to be avoiding municipalities with already high concentration of firms operating in the same industrial branch of wholesaling and also to be more reluctant to leave municipalities governed by right-of-the- center parties. Lastly, firms seem to avoid moving to municipalities characterized with high population density.

Paper [3] addresses determinants of firm growth, adopting OLS and a quantile regression technique. The results of this paper indicate that very little of the firm growth can be explained by the firm-, industry- and region-specific factors, controlled for in the estimated models. Instead, the firm growth seems to be driven by internal characteristics of firms, factors difficult to capture in conventional statistics. This result supports Penrose’s (1959) suggestion that internal resources such as firm culture, brand loyalty, entrepreneurial skills, and so on, are important determinants of firm growth rates.

Paper [4] formulates a forecasting model for firm entry into local markets and tests this model using data from the Swedish wholesale industry. The empirical analysis is based on directly estimating the profit function of wholesale firms and identification of low- and high-return local markets. The results indicate that 19 of 30 estimated models have more net entry in high-return municipalities, but the estimated parameters is only statistically significant at conventional level in one of our estimated models, and then with unexpected negative sign.

Paper [5] studies effects of firm relocation on firm profits of relocating firms, employing a difference-in-difference propensity score matching. Using propensity score matching, the pre-relocalization differences between relocating and non-relocating firms are balanced, while the difference-in-difference estimator controls for all time-invariant unobserved heterogeneity among firms. The results suggest that firms that relocate increase their profits significantly, in comparison to what the profits would be had the firms not relocated. This effect is estimated to vary between 3 to 11 percentage points, depending on the length of the analyzed period. 

This study analyses the effects of firm relocation on firm profits, using longitudinal data on Swedish limtied liability firms and employing a difference-in-differnce propensity score method in the empirical analysis. Using propensity score matching, the pre-relocalization differneces between relocating and non-relocating firms are balanced. In addition to that, a difference-in-difference estimator is employed in order to control for all time-invariant unobserved heterogeneity among firms. For matching, nearest neighbour matching, using the one-, two- and three nearest neighbours is employed. The balanacing results indicate that matching achieves a good balance, and that similar relocating and non-relocating firms are being compared. The estimated average treatment on the treatment effects indicate thats relocations has a significant effect on the profits of the relocating firms. In other words, firms taht relocate increase their profits significantly, in comparison to what the profits would be had the firms not relocated. This effect is estimated to vary between 3 to 11 percentage points, depending on the lenght of the analysed period after relocation. 

Firm entry into local markets has often been studied using administrative areas such as municipalities as the assumed relevant markets. However, administrative areas and the actual relevant markets based on local demand for firms’ products often do not coincide, which could bias the results of studies treating administrative areas as the relevant markets. Based on a behavioral assumption regarding how retailers act when purchasing products from wholesale trade firms, we create alternative markets using Voronoi diagrams. We then compare the empirical results of investigating the determinants of firm entry using municipalities as the relevant markets with the results obtained using Voronoi markets. The results indicate that, in both cases, the same variables are statistically significant in affecting entry, though the estimated effects differ in size.

Applying microeconomic theory, we develop a forecasting model for firm entry into local markets and test this model using data from the Swedish wholesale industry. The empirical analysis is based on directly estimating the profit function of wholesale firms. Profits are assumed to depend on firm- and location-specific factors, and the profit equation is estimated using panel data econometric techniques. Using the residuals from the profit equation estimations, we identify local markets in Sweden where firm profits are abnormally high given the level of all independent variables included in the profit function. From microeconomic theory, we then know that these local markets should have higher net entry than other markets, all else being equal, and we investigate this in a second step, also using a panel data econometric model. The results of estimating the net-entry equation indicate that 19 out of 30 estimated models have more net entry in high-return municipalities, but the estimated parameter is only statistically significant at conventional levels in one of our estimated models, and then with an unexpected negative sign.