Hans Hvide, Tom G. Meling 16 December 2019
While a large literature has studied the decision to become an entrepreneur (e.g. Evans and Jovanovic 1989, Levine and Rubinstein 2017), less is known about the factors that make some startups successful and others not. Does a temporary demand shock have long-term effects on startup outcomes?
Our recent paper (Hvide and Meling 2019) combines data on more than 10,000 procurement auctions held by the Norwegian Public Roads Administration with registered data on all Norwegian firms. We focus on auctions where the lowest price is the only winning criterion to better understand the long-term effects of temporary demand shocks on startups. The average work contract takes about one year to complete and accounts for about 25% of the winning startup’s annual sales. Thus, auction winners experience a short-lived but economically significant shock to demand.
Figure 1 Difference in log sales between winners and runners-up
In Figure 1, we plot the difference in log sales between winning startups and runners-up in the years surrounding a procurement auction. Even five years after the contract period has ended, winning startups are about 20% larger than runners-up, despite having indistinguishable levels and trends on key firm characteristics before auction. The effects on other measures of firm size, such as employment and value added, are similar.
It is possible that auction winners would follow a different trend than runners-up after auction, even absent the auction win. To alleviate this concern, we perform two placebo tests. First, we compare the trends of winners and runners-up prior to the auction and show that they are the same. Second, we compare the post-auction performance of runners-up with startups that end up in third place. Differences between these two groups cannot be due to winning an auction, as neither startup won. We find that the startups in these two groups have very similar trends and levels on observables, both before and after auction.
Figure 2 Pre-auction differences between winners and runners-up
Temporary demand shocks may have long-term effects for several reasons. The contract work could have learning-by-doing effects (e.g. Arrow 1962, Thompson 2012). Consistent with learning-by-doing, auction winners expand vertically by becoming more likely than runners-up to subsequently participate in larger auctions. Winners also expand horizontally by increasing their participation in auctions that involve new products. For example, if the initial auction involved the procurement of a bicycle path, the winning firm may later bid for a contract that involves building road fences.
Although winning an auction does not seem to affect firm productivity as conventionally measured (e.g. Syverson 2011), auction winners do tend to enter subsequent auctions with higher-productivity competitors and to hire managers of higher quality, suggesting a latent productivity increase consistent with learning-by-doing. Furthermore, the long-term effects of auction wins appear unique to startups – we find no or minimal long-term effects on mature firm outcomes of winning an auction, consistent with learning-by-doing effects being more important for young firms
We also find evidence that winning startups undertake investments that could lead to lasting firm size differences, or sunk cost effects (e.g. Sutton 1991, Das et al. 2007). For example, using balance sheet data, we find that winners significantly expand their tangible assets (including machinery) compared to runners-up during the contract work, and that these differences persist several years after the contract work ends. Interviews with the managers of winning startups, obtained from newspaper articles describing startups’ auction wins, further support the idea of winners making investments that are later sunk.
We find less support for other mechanisms, such as winning a procurement auction alleviating the startup’s financial constraints (e.g. Evans and Jovanovic 1989), or winning an auction boosting the startup’s demand through the development of customer relationships or the building of a brand name (e.g. Foster et al. 2016). Neither do we find evidence that establishing a relationship with the Norwegian Public Roads Administration can explain the effects. To arrive at this result, we use data on auctions where criteria other than price (such as quality) determine which firm wins.
Thus, a stroke of luck – winning an auction rather than coming in second – gets reinforced via learning-by-doing and investment effects, making winning and runner-up startups look very different even in the long run.
Using data from procurement auctions in Norway, we find that temporary demand shocks have long-term effects for startups: startups that win a procurement auction are 20% larger than runner-up startups, even several years after the contract work has ended. This has potential policy implications. In order to enhance their startup and innovation ecosystems, the US and the UK both employ policies that promote startups’ participation in government procurement auctions. Our findings provide a possible rationale for such policies: in terms of job creation and sales growth, winning a procurement auction seems to have much larger effects for startups than for mature firms.
Recent work by Sedlacek and Sterk (2017) documents that firms born in cohorts where job creation is weaker are persistently smaller, even when the aggregate economy recovers. As both demand-side and supply-side factors vary with the business cycle, it is challenging to establish what drives these patterns from aggregate data alone. With the important caveat that general equilibrium effects make extrapolating from micro to macro difficult, our findings suggest that part of the reason for the persistent cohort effects documented by Sedlacek and Sterk could be the demand component of business cycle variations.
More broadly, our work contributes to empirical research that explores path-dependence of firms in various settings. For example, recent work by Aghion et al. (2016) demonstrates path-dependence in the type of research that R&D-intensive firms pursue.
Aghion, P, A Dechezlepretre, D Hemous, R Marting and J Van Reenen (2016), “Carbon Taxes, Path Dependency, and Directed Technical Change: Evidence from the Auto Industry”, Journal of Political Economy, 124: 1-51.
Arrow, K J (1962), “The Economic Implications of Learning by Doing”, Review of Economic Studies 29(3): 155-173.
Das, S, M J Roberts and J R Tybout (2007), “Market Entry Costs, Producer Heterogeneity, and Export Dynamics”, Econometrica 75(3): 837-873.
Evans, D S and B Jovanovic (1989), “An Estimated Model of Entrepreneurial Choice under Liquidity Constraints”, Journal of Political Economy 97: 808-827.
Foster, L, J Haltiwanger and C Syverson (2016), “The Slow Growth of New Plants: Learning about Demand?”, Economica 83: 91-129.
Hvide, H K and T G Meling (2019), “Do Temporary Demand Shocks Have Long-Term Effects for Startups?”, CEPR discussion paper 1413.
Levine, R and Y Rubinstein (2017), “Smart and Illicit: Who Becomes an Entrepreneur and Do They Earn More?”, Quarterly Journal of Economics 132: 963-1018.
Sedlácek, P and V Sterk (2017), “The Growth Potential of Startups over the Business Cycle”, American Economic Review 107(10): 3182-3210.
Sutton, J (1991), Sunk Costs and Market Structure, Cambridge, MA: MIT Press.
Syverson, C (2011), “What Determines Productivity?”, Journal of Economic Literature 49: 326-365.
Thompson, P (2012), “The Relationship between Unit Cost and Cumulative Quantity and the Evidence for Organizational Learning-by-Doing”, Journal of Economic Perspectives 26(3): 203-224.