Life in the fast lane: Origins of competitive interaction in new vs, established markets, страница 10

Within the categories of R&D and market moves, we selected six competitive moves (three R&D and three market moves) to study further. Based on our analyses of the data described above, these six moves were considered the most important and captured the most crucial aspects of firm strategy. We confirmed these choices by analyzing several other moves, such as pricing and advertising. These did not significantly influence the results.

Measures

The dependent variables are frequencies of competitive moves. Consistent with competitive dynamics research measuring move frequency (Young et al., 1996; Ferrier et al., 1999), we use a count of competitive moves. Our first dependent variable is the frequency of market moves. We began by separating market moves into three types: market probe, market entry, and market exit. We used specific market moves in each market and in each round to code these moves. We then counted their number. We summed these moves for each firm in each round to compute market moves. Our second dependent variable is frequency of R&D moves, also measured as a count of moves made by each firm. Similar to market moves, we separated R&D moves into three types: R&D probe, R&D product, and R&D process. We used R&D moves in each market and in each round to code them, and then we counted their number. We summed these moves for each firm in each round to compute R&D moves. See the Appendix for technical details.

We also created two related dependent variable measures: diversity of market moves and diversity of R&D moves, because prior work has shown that both frequency and diversity of moves may boost competitive advantage. We measured diversity of moves as the diversification of the firm's market and R&D moves across customer segments in each round (number of Sonite segments ranged from zero to five and Vodite segments from zero to three). We calculated diversity using a Herfindahl index: 1 − ∑(N_a/N_T)², where N_a equals the number of market or R&D moves made in segment a, and N_T equals the total number of (market or R&D) moves made. The measure ranged from 0 to 1, where higher values indicate greater move diversity.

The main independent variable is firm performance. We measure firm performance as the firm's market share in each round. Depending on the competitive move of interest, Sonite market share or Vodite market share was used. As a widely used assessment of performance relative to competitors (Armstrong and Collopy, 1996), market share measures the relative success of firms by providing explicit comparison to rivals. It is particularly appropriate for our study because it allows us to compare across simulation runs by controlling for industry size and other extraneous differences, such as pricing. Consistent with these arguments, studies of competitive moves have frequently used market share as the firm performance measure (Chen and MacMillan, 1992; Makadok, 1998; Ferrier et al., 1999; Ferrier, 2001). Our qualitative data confirm our choice as most interviewees discussed market share (and related revenues) at length in describing their competitive move strategies. We also used alternative measures of performance, including revenue, stock price, and total profit with qualitatively similar results.

There are several control variables. We control for firm resources to ensure that the availability of resources (rather than firm performance) did not explain frequency of moves. We measured resources as a firm's total revenue in each round (in millions of dollars, standardized). This is an appropriate measure because firms with more revenue are likely to have more resources than firms with less (Greve, 1998). Like other studies using Markstrat (Ross, 1987; Glazer et al., 1987), we also used starting position as a measure of firm resources. Markstrat assigns different starting positions in the Sonite market to each of the five firms such that some teams begin with more resources than others (e.g., superior product portfolios). We used a dummy variable for discrete starting positions to control for this difference. The results, available from the authors, were similar to those for the original resources measure.