Tax Simplification and Business Formalization in Colombian Caribbean Cities

2. Theoretical or Conceptual Framework

Compliance costs and simplified taxation

Tax compliance involves substantially more than the payment of a statutory tax liability. Firms generally need to identify applicable obligations, maintain records, prepare declarations, update accounting systems, monitor legal changes, and allocate managerial resources to regulatory tasks. Many of these activities generate costs independently of the amount of tax ultimately paid. When a portion of these costs is fixed, the proportional burden can become particularly relevant for smaller firms. Slemrod and Venkatesh [6] document the importance of business tax-compliance costs and illustrate why administrative obligations may impose uneven burdens across firm sizes. Bird and Zolt [7] similarly argue that taxation in emerging economies should be evaluated not only through statutory rates but also through administrative capacity, implementation requirements, and the characteristics of taxpayers.

This reasoning provides a potential link between simplified taxation and formalization. A firm considering formal participation compares expected benefits from formal operation with fiscal obligations, administrative requirements, and other regulatory costs. If simplified taxation reduces part of the fixed cost of compliance, the relative net benefit of formal operation may improve. The relevant mechanism is therefore broader than a lower nominal tax rate. Simplification can reduce uncertainty about obligations, improve the predictability of payments, facilitate record keeping, and reduce the number of separate interactions required with tax authorities. Wen [2] argues that presumptive and simplified tax systems may be useful where conventional tax structures impose relatively high compliance demands on micro and small enterprises. OECD et al. [8] likewise treat administrative simplification as one element within a broader SME-policy framework involving digitalization, business development, finance, and institutional capacity.

Simplification can also interact with monitoring and information. Greater reliance on electronic invoicing, digital payments, standardized reporting, and integrated tax records may make compliance more manageable while increasing the traceability of transactions. Pomeranz [9] shows that information structures and third-party reporting can affect tax enforcement and compliance incentives. Alm and McClellan [1] additionally suggest that tax compliance reflects perceptions of institutions and tax morale. These perspectives imply that simplified taxation may influence firms through several overlapping channels: reduced transaction costs, improved predictability, greater information traceability, and a potentially more accessible relationship with tax institutions. The relative importance of each channel is likely to vary across firms and territories.

Formalization as a heterogeneous decision

A substantial literature cautions against assuming that all informal firms would prefer formal status if registration became less costly. McKenzie and Sakho [4] find heterogeneous profitability effects associated with tax registration. Some businesses may obtain greater legitimacy, market access, or commercial opportunities from registration, while others may receive more limited benefits relative to the fiscal and administrative obligations they assume. Formalization can therefore depend on where a firm lies relative to an economic threshold at which formal participation becomes worthwhile.

Experimental research provides additional evidence of this heterogeneity. de Mel et al. [10] find that providing information and reimbursing registration expenses generated limited formalization among informal firms in Sri Lanka, whereas larger financial incentives produced stronger registration responses. Even among firms that formalized, subsequent gains were not uniform. These findings suggest that administrative barriers can matter without necessarily constituting the primary constraint for every enterprise. Simplified tax regimes may therefore have stronger effects among businesses whose productivity, scale, or market orientation already place them relatively close to formal operation.

La Porta and Shleifer [5] develop a related structural argument. Their interpretation of informality emphasizes differences in productivity, human capital, business scale, and managerial characteristics. Some informal enterprises may therefore remain economically distinct from typical formal firms even when registration procedures become easier. Ulyssea [11] similarly concludes that reductions in the costs of formality can contribute to formalization but may be insufficient when other structural constraints remain important. This literature suggests that the relationship between tax simplification and formalization should be expected to vary across businesses and should therefore be tested as a conditional empirical relationship rather than assumed in advance.

Latin American evidence on simplified taxation

Latin American tax reforms provide relevant evidence on the possible relationship between simplification and business formalization. Fajnzylber et al. [12] examine Brazil's SIMPLES program and report evidence consistent with improvements in several dimensions of formality among eligible microenterprises. Their findings also suggest changes in revenues, employment, and capital intensity among some newly formal businesses. This evidence indicates that lowering tax and administrative burdens can potentially influence business organization in addition to registration behavior.

The results from Brazil nevertheless appear heterogeneous. Monteiro and Assunção [13] find stronger formalization effects among eligible retail establishments than in several other economic activities. This suggests that the consequences of simplified taxation can depend on sector characteristics. Businesses differ in profit margins, cost structures, labor intensity, seasonality, and their ability to benefit from formal market access. These differences are relevant for Colombia's RST because rates and potential advantages vary across economic activities and income brackets.

Rocha et al. [14] provide additional evidence that lower taxes can encourage some informal firms to formalize even after administrative registration costs have been reduced. However, the magnitude of these responses appears sufficiently moderate to make the cost-effectiveness of tax incentives an important consideration. Ulyssea [15] further distinguishes between extensive and intensive dimensions of informality. A business can possess formal registration while retaining informal workers or incomplete reporting practices. This distinction provides an important reason not to use RST registration itself as the dependent measure of formalization. A multidimensional indicator offers a more demanding assessment of whether tax-regime participation is associated with broader organizational changes.

Theoretical mechanism and research hypothesis

The literature supports a possible sequence through which simplified taxation may contribute to formalization. First, the regime may reduce some administrative and fiscal frictions associated with remaining formally registered. Second, a lower compliance burden may modestly improve the expected net benefits of formal participation. Third, formal operation can facilitate electronic invoicing, banking relationships, contracting, and improved business records. If these mechanisms are sufficiently relevant for a meaningful share of firms, increases in simplified-tax participation could eventually be accompanied by reductions in broader measures of business informality.

However, the same literature identifies several reasons why this association may be limited or delayed. Firms with very low productivity may continue to find informal operation economically preferable. Already-formal firms can shift from the ordinary regime into the RST without generating additional formalization. Labor practices may remain informal after tax registration, while technological requirements may create new difficulties for some microenterprises. OECD [3] emphasizes persistent territorial and productivity differences in Colombia, and OECD et al. [8] highlight the relevance of complementary SME capabilities. These considerations suggest that the relationship may become more informative when changes are observed within cities over time rather than through simple cross-sectional comparisons.

Accordingly, the study evaluates one hypothesis: H1 proposes that greater RST intensity is associated with a subsequent reduction in multidimensional business informality. Because higher IMIE values denote greater informality, evidence consistent with H1 would generally be reflected in a negative estimated relationship. This wording is deliberately moderate. The hypothesis does not assume that RST participation necessarily produces formalization; it examines whether the observed direction and timing of the association are compatible with the proposed mechanism. The three statistical approaches examine different forms of variation. A negative association that remains reasonably stable across lagged correlation, fixed effects, and first differences would provide stronger support than a relationship appearing under only one specification.

3. Methodology

Research design

The empirical framework uses a balanced quarterly panel covering Barranquilla, Cartagena, Santa Marta, and Valledupar from 2021Q1 to 2024Q4. These cities belong to Colombia's Caribbean region but differ in economic size, sector composition, labor-market characteristics, and institutional environment. Holding national taxation broadly constant while examining different regional economies provides a useful setting for studying whether local movements in RST participation correspond to movements in business informality.

Each city contributes sixteen quarterly observations, yielding a total of 64 city-period observations. The design is longitudinal because the principal question concerns changes within cities and temporal ordering. The quarterly values are synthetically generated but calibrated against verified annual benchmarks. Consequently, the study should be interpreted as a calibrated empirical exercise rather than as an analysis of official quarterly administrative microdata. The design is useful for evaluating the statistical framework and the stability of the hypothesized relationship, while stronger substantive conclusions would require direct administrative observations.

Variable construction

The principal explanatory measure is RST intensity per 1,000 formal firms. Within the analysis, it is calculated as RSTIntensity_it = (RSTRegistrations_it / FormalFirms_it) × 1,000. Normalizing the number of registrations is important because raw enrollment is strongly influenced by the overall size of each city's business population. Barranquilla, for example, would be expected to have more registered taxpayers than a smaller urban economy even if relative RST penetration were similar.

Regression specifications use the natural logarithm of RST intensity. The dependent variable is multidimensional business informality, represented by IMIE, where higher values indicate greater informality. For interpretation, a complementary formalization measure is constructed as 100 − IMIE. Unemployment and the share of service-sector activity are included as time-varying controls. These controls should not be interpreted as eliminating all sources of confounding; they provide a limited adjustment for plausible changes in local economic conditions that may be related to both formalization and RST participation.

Data verification procedures

The first verification examines how closely the calibrated quarterly series correspond to the official annual anchors. For RST, fourth-quarter observations are compared directly with the annual DIAN benchmark. The discrepancy is calculated as the proportional difference between the calibrated Q4 value and the corresponding annual value. IMIE is treated somewhat differently because quarterly variation is intentionally introduced around the annual DANE benchmark. The mean of the four quarterly values is therefore compared with the annual reference level.

The second verification procedure applies the median absolute deviation to changes in logged RST intensity. For each quarter-to-quarter change, a robust standardized score is calculated using 0.6745 × (x − median) / MAD. Values with an absolute score greater than 3.5 are flagged for review. These flags are not automatically treated as errors. Rapid changes can reflect the interpolation needed to reach substantially different annual anchors or, in observed data, potentially correspond to policy changes or registration campaigns. The procedure is therefore primarily intended to identify influential or unusual transitions requiring additional examination.

Statistical analysis

The first test of H1 uses Spearman's rank correlation between previous-quarter RST intensity and the subsequent quarterly change in IMIE. The lag introduces temporal ordering and reduces dependence on contemporaneous correlation. Because Spearman's rho uses ranks, it is less sensitive to nonnormal distributions and extreme values than Pearson correlation. A negative coefficient is directionally compatible with H1.

The second procedure estimates two-way fixed-effects models of IMIE on logged RST intensity, city effects, and period effects. The model can be represented within the text as IMIE_it = βln(RSTIntensity_it) + γX_it + α_i + λ_t + ε_it. The third procedure uses first differences, represented as ΔIMIE_it = βΔln(RSTIntensity_it) + γΔX_it + λ_t + ε_it. Fixed effects account for relatively stable differences between cities, while first differences focus directly on quarter-to-quarter changes within the same city. Because the data are synthetically calibrated and the number of cities is small, estimated coefficients and confidence intervals are interpreted cautiously.

4. Results

Calibration and statistical verification

The calibration procedure indicates that the quarterly synthetic structure remains close to the annual empirical benchmarks. RST fourth-quarter observations reproduce their corresponding annual anchor by construction, producing no closing-point discrepancy. For IMIE, the mean absolute discrepancy between annual calibrated averages and the annual benchmarks is approximately 0.268 percentage points. The maximum observed discrepancy is below half a percentage point.

Table 1. Calibration and robust verification results

Note. RST discrepancy compares calibrated fourth-quarter values with the corresponding annual anchors. IMIE discrepancy compares the average of quarterly calibrated observations with the annual benchmark. MAD flags identify changes in ln(RST intensity) for which |robust z| > 3.5.

Source: Authors' own work.

MAD screening identifies four comparatively unusual changes in RST intensity. Two occur in Barranquilla, while Santa Marta and Valledupar contain one flagged transition each. Cartagena contains no flagged observation under the selected threshold. These values remain in the analysis because the verification procedure does not establish that they are erroneous.

The result suggests that the synthetic series broadly retains its intended relationship with the annual empirical anchors while preserving meaningful short-run variation. Nevertheless, calibration accuracy should not be interpreted as evidence that the simulated quarterly trajectories reproduce actual quarterly administrative behavior. The procedure instead provides a consistency check for the analytical dataset.

Cross-city patterns

Mean RST intensity differs across the four cities. Barranquilla records the highest average value at 75.780 registrations per 1,000 formal firms, followed closely by Cartagena at 74.815. Santa Marta records 62.355 and Valledupar 55.756. The overall panel average is 67.177.

Table 2. Descriptive statistics by city

Note. RST intensity is measured as registrations per 1,000 calibrated formal firms. The formalization index equals 100 − IMIE. Quarterly observations are synthetic calibrated values.

Source: Authors' own work.

The descriptive ordering illustrates why cross-sectional comparisons alone would provide limited evidence about H1. Valledupar has the lowest average RST intensity but also the lowest average IMIE. Conversely, Santa Marta records comparatively high business informality despite stronger average RST intensity than Valledupar. These differences may reflect structural characteristics rather than the effects of tax-regime participation.

The longitudinal methods are therefore more informative for the study hypothesis. Fixed-effects models evaluate whether changes around each city's own average are associated with changes in informality after accounting for common time effects. This approach reduces, although does not eliminate, the possibility that persistent territorial differences are driving the results.

Lagged Spearman analysis

The pooled Spearman correlation between previous-quarter RST intensity and the subsequent quarterly change in IMIE is −0.154. The associated p value is .241. The coefficient is negative, which is directionally consistent with H1, but its magnitude is relatively small and the statistical uncertainty remains substantial.

Table 3. Lagged Spearman correlations

Note. Predictor = previous-quarter RST intensity. Outcome = subsequent quarterly change in IMIE. Negative coefficients are directionally compatible with H1.

Source: Authors' own work.

Three of the four cities produce negative estimates. Santa Marta has the largest negative coefficient, followed by Cartagena and Barranquilla. Valledupar shows a small positive association. None of the city-specific estimates reaches conventional levels of statistical significance.

These results may be described as modest directional evidence rather than confirmation of H1. One possible interpretation is that the relationship, if present, operates only for certain firms or becomes visible over longer periods than one quarter. Another possibility is that aggregate RST intensity is an imperfect proxy for genuinely new formalization because some participants may have been formal before changing tax regimes.

Fixed-effects and first-difference models

The two-way fixed-effects models produce coefficients that differ from the direction observed in the lagged correlation. Without additional controls, the estimated coefficient on ln(RST intensity) is 4.044 with a robust standard error of 2.158 and p = .061. With unemployment and service-sector composition included, the coefficient is 4.238, with SE = 2.212 and p = .055.

Table 4. Panel estimates of multidimensional business informality

Note. Fixed-effects models use ln(RST intensity) and include city and period effects. First-difference models use Δln(RST intensity). Positive coefficients correspond to higher IMIE values and are therefore inconsistent with the expected direction of H1.

Source: Authors' own work.

The positive coefficients in the fixed-effects models should be interpreted cautiously. They do not necessarily imply that greater RST participation increases informality. Several alternative mechanisms may generate this pattern. RST enrollment may increase more rapidly during periods in which policymakers place greater emphasis on difficult-to-formalize territories. Already-formal firms may switch into the RST without producing a corresponding movement in the broader informality indicator. It is also possible that changes in organizational formalization take longer to emerge than changes in tax registration.

The first-difference estimates return to negative coefficients but with wide confidence intervals and high p values. The controlled first-difference model produces a coefficient of −0.220. This result is compatible with the expected direction but provides little statistical precision. The variation in coefficient sign across specifications suggests that the estimated relationship is sensitive to the source of identifying variation. Accordingly, the current evidence does not provide a sufficiently stable basis for asserting that greater RST intensity is systematically associated with lower multidimensional informality.