Diversification-Performance Analysis (Rumelt Categories)
- Introduced by
- Richard P. Rumelt; Krishna Palepu
- Year
- 1974
- Type
- Classification-and-comparison pipeline relating diversification type to firm performance
- Family
- Strategic management / corporate strategy
What does it do?
Diversification-performance analysis asks whether the kind of diversification a firm pursues — staying focused, expanding into related businesses, or building an unrelated conglomerate — is systematically associated with how well the firm performs.
The categorical version originates with Rumelt's 1974 Strategy, Structure, and Economic Performance, which classified diversified firms by specialization and relatedness ratios into single-business, dominant-business, related, and unrelated types and found that related diversifiers tended to outperform unrelated ones. Palepu's 1985 study reframed diversification with the continuous Jacquemin-Berry entropy measure, again finding that related diversification was associated with superior profit growth, and showed how the index approach and Rumelt's categorical method can be combined to gain both objectivity and conceptual richness.
The intuition behind the method
Firms grow in different directions. Some stick to one core business; others add lines that share technology, customers, or know-how with what they already do; still others assemble a portfolio of unrelated businesses held together only by finance and control. The strategic question is whether these patterns pay off differently. The case for related diversification is that shared resources and capabilities create real synergies — economies of scope — that a focused or unrelated firm cannot capture. The case against unrelated diversification is that without operating linkages, the corporate parent adds little beyond an internal capital market and may destroy value through complexity and weak oversight. Diversification-performance analysis turns this debate into measurement: sort firms by how and how much they diversify, then compare their performance, controlling for industry and risk, to see whether relatedness actually matters.
History
Richard Rumelt's 1974 Strategy, Structure, and Economic Performance, extending Wrigley's earlier typology, established the categorical study of diversification by classifying large U.S. firms into single, dominant, related, and unrelated business types via specialization and relatedness ratios, and reporting that related-constrained diversifiers achieved the best risk-adjusted performance. The finding ignited the corporate-strategy literature on the diversification-performance relationship. In 1985 Palepu addressed the subjectivity of categorical coding by applying the Jacquemin-Berry entropy measure, decomposing diversification into related and unrelated components and confirming that related diversification was associated with superior profit growth, while advocating that the index and categorical approaches be combined. Subsequent decades produced a large, sometimes conflicting body of work, debating the diversification discount, curvilinear relationships, and the endogeneity of diversification choices.
When to use it
Use diversification-performance analysis when your question concerns corporate-level strategy — the scope of the firm's portfolio and whether its diversification creates or destroys value — and you can characterize each firm's business composition and performance. The Rumelt categorical approach is especially useful when the strategic distinction between related and unrelated diversification is central and you want interpretable, theory-laden categories. It is less appropriate when business-segment data are too coarse to judge relatedness, when sample sizes are small (categories thin out), or when you need a continuous, fully replicable measure for econometric modeling — in which case the entropy index, possibly alongside the categories, is preferable. Throughout, the analysis must control for industry and risk and confront endogeneity, since firms self-select into diversification strategies.
Limitations and considerations
- Rumelt's categorical coding requires subjective judgments about what is 'related,' raising replicability and reliability concerns.
- Results are sensitive to the thresholds chosen and to the granularity of available business-segment data.
- Diversification is endogenous: firms choose their scope, so performance differences may reflect selection rather than a causal diversification effect.
- Industry and risk confounds are severe, and inadequate controls can manufacture or mask the related-versus-unrelated performance gap.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Captures the strategically crucial distinction between related and unrelated diversification, not just the amount of diversification.
- Grounded in Rumelt's careful, operations-based classification, giving categories real conceptual meaning rather than mechanical codes.
- Produced one of strategy's most influential findings — that related diversification tends to outperform unrelated — shaping corporate-strategy theory.
- Can be cross-validated with Palepu's objective entropy measure, combining interpretability with replicability.
Applications
- Classifying firms' corporate strategies into single, dominant, related, and unrelated diversification types for comparative study.
- Testing whether related diversifiers outperform unrelated diversifiers on risk-adjusted accounting and market measures.
- Studying changes in firms' diversification posture over time, such as refocusing waves and conglomerate breakups.
- Linking corporate scope to organizational structure, control systems, and the corporate parent's value-adding role.
- Providing the strategy categorization that complements continuous entropy-based diversification measures in econometric models.
How it works
How it works
The classification starts by measuring how concentrated a firm's revenues are. Rumelt defines the specialization ratio as the share of total revenue from the firm's largest single business, and the related ratio as the share from its largest group of related businesses. These ratios summarize, in two numbers, how focused the firm is and how much of its activity hangs together around a common thread. Computing them requires judgment about what counts as one business and which businesses are 'related,' which is why Rumelt's scheme is grounded in careful reading of each firm's actual operations rather than mechanical segment codes.
Formula
SR=\dfrac{\text{revenue of largest business}}{\text{total revenue}},\quad RR=\dfrac{\text{revenue of largest related group}}{\text{total revenue}}Using thresholds on these ratios, firms are sorted into Rumelt's categories: single business (very high specialization), dominant business (one large core with modest diversification), related diversified (substantial diversification but with businesses linked by shared skills or resources), and unrelated diversified (substantial diversification with little linkage). This typology is the conceptual heart of the method: it distinguishes diversification that exploits common capabilities from diversification that merely spreads across industries, encoding the synergy logic that the rest of the analysis tests.
Formula
\text{category}=\begin{cases}\text{Single} & SR\ge 0.95\\ \text{Dominant} & 0.70\le SR<0.95\\ \text{Related} & SR<0.70,\ RR\ \text{high}\\ \text{Unrelated} & SR<0.70,\ RR\ \text{low}\end{cases}Performance is then measured with accounting and market indicators — return on assets, return on equity, sales and earnings growth, and sometimes market-based measures like Tobin's q — adjusted for industry effects and risk so that differences across categories are not just artifacts of which industries firms happen to operate in. Rumelt emphasized risk-adjusted, industry-comparable performance because diversification types are unevenly distributed across sectors, and failing to control for the industry mix would confound the diversification effect with sector profitability.
Formula
P_i\in\{\,\text{ROA},\ \text{ROE},\ \text{growth},\ q\,\},\ \text{adjusted for industry and risk}The core test compares performance across categories, classically asking whether related diversifiers outperform unrelated ones. This is done with analysis of variance across the Rumelt categories and with regressions that include category dummy variables alongside controls for size, leverage, risk, and industry. Rumelt's finding that related-constrained diversifiers earned the highest risk-adjusted returns, while unrelated diversifiers lagged, became one of strategy's most influential empirical claims and the template for decades of corporate-strategy research linking the type of diversification to value creation.
Formula
H_0:\ \bar P_{\text{related}}=\bar P_{\text{unrelated}};\quad P_i=\beta_0+\textstyle\sum_c \beta_c D_{ic}+\boldsymbol{\gamma}'\mathbf{x}_i+\varepsilon_iBecause the categorical scheme involves subjective judgments, Palepu introduced a complementary continuous approach using the Jacquemin-Berry entropy measure, which quantifies total diversification and decomposes it into related and unrelated components. Validating the categorical results against this objective, replicable index — and finding that related diversification again predicts better profit growth — strengthens the conclusion and addresses the criticism that Rumelt's categories rest on coder judgment. Palepu argued the two approaches are best used together: the entropy index supplies objectivity while Rumelt's categories preserve the strategic meaning of relatedness.
Formula
DT=\sum_j P_j\ln(1/P_j)\ \Rightarrow\ \text{cross-check categorical vs continuous diversification}Common pitfalls
- Treating relatedness as obvious and coding it inconsistently, undermining the reliability of the Rumelt categories.
- Comparing raw performance across categories without adjusting for industry mix and risk, confounding strategy with sector effects.
- Reading a cross-sectional association as causal evidence that diversifying relatedly raises performance, ignoring self-selection.
- Relying solely on coarse segment data to judge relatedness, or solely on a continuous index while losing the strategic meaning of relatedness.
Frequently asked questions
What are Rumelt's diversification categories?
Rumelt sorted firms using specialization and relatedness ratios into four main types: single business (almost all revenue from one business), dominant business (one large core with limited diversification), related business (substantial diversification but with businesses linked by shared technologies, markets, or skills), and unrelated business (substantial diversification with little operating linkage, the classic conglomerate). The categories encode the strategic distinction between diversification that exploits common capabilities and diversification that merely spreads across industries, which is the relationship the performance analysis tests.
Why do related diversifiers tend to outperform unrelated ones?
The theoretical rationale is economies of scope: when businesses share technologies, customers, brands, or know-how, the firm can spread these resources across activities and create synergies a focused or unrelated firm cannot. Unrelated diversification lacks these operating linkages, so the corporate parent adds value mainly through an internal capital market and may instead destroy value via complexity and weaker oversight. Both Rumelt's categorical evidence and Palepu's entropy-based analysis found related diversification associated with better performance, though the relationship is contested and contingent on context and proper controls.
How does the entropy measure relate to Rumelt's categories?
Rumelt's categories are judgment-based and rich in strategic meaning but hard to replicate. Palepu's entropy measure, borrowed from Jacquemin and Berry, quantifies total diversification continuously and decomposes it into related and unrelated components objectively from segment data. Palepu showed the two are complementary: the entropy index supplies objectivity, replicability, and a continuous variable for regression, while Rumelt's categories preserve the conceptual distinction of relatedness. Using them together lets researchers cross-validate findings and guards against the weaknesses of either approach alone.
Related methods
Sources
Rumelt, R. P. (1974). Strategy, Structure, and Economic Performance. Division of Research, Graduate School of Business Administration, Harvard University.
ISBN: 9780875841090Copied.Palepu, K. (1985). Diversification strategy, profit performance and the entropy measure. Strategic Management Journal, 6(3), 239-255.
Open DOI referenceCopied.