Decomposing Wage Stagnation: Employment Reallocation, Wage Structure,and Demographics

Beyond the Rise of Part-Time Work: Unpacking Japan's Wage Stagnation

Why did Japanese wages stop growing significantly after the mid-1990s? While economists have long debated the causes, the answer remains elusive. The forces involved—aging populations, shifting job types, and changing pay scales—are often studied in isolation.

Researchers have traditionally focused on whether specific groups, such as part-time workers, see lower pay. However, this ignores how the entire workforce moves between different sectors. A new study from Kyoto University argues that wage stagnation is not caused by a single factor. Instead, it results from a complex interplay of demographic shifts, employment reallocation (the movement of workers between different types of jobs), and changes in the relative pay structure of different jobs.

The limitations of single-factor explanations

Current economic literature often treats different drivers of wage trends as separate phenomena. Some studies emphasize the expansion of nonstandard employment (contract or part-time work). Others focus on the aging of the workforce. This fragmented view creates a blind spot. It fails to distinguish between who is working and what kind of jobs they are doing.

If you only look at wage growth within specific job categories, you might conclude that wages are stagnant everywhere. But aggregate wages can drop even if individual job wages are rising. This happens simply because the workforce is shifting into lower-paying sectors. As seen in, employment shares have shifted significantly away from prime-age men toward women and older workers.

Previous models that failed to account for these simultaneous movements between "worker types" (demographics) and "job types" (industry, size, and status) likely misattributed the cause of the stagnation.

A multi-stage accounting framework

To resolve this, the author develops a three-step decomposition framework. This framework treats average log real hourly wages (a mathematical way to represent wage changes) as a weighted sum of various components. This approach acts like a financial audit. It traces exactly where every "yen" of change originates.

1. Shift-Share Decomposition: The framework first separates changes caused by demographic composition from changes occurring within specific worker types. This distinguishes "who works" (e.g., a shift from young men to older women) from wage changes within those groups.
2. Extended Olley–Pakes Decomposition: Within each worker type, the study applies an extension of the Olley–Pakes method. This technique is typically used to analyze firm productivity. Here, it separates wage changes within job types from the "allocation term." The allocation term represents how workers are distributed across different jobs.
3. Allocation Refinement: Finally, the framework breaks the allocation term into two distinct parts. These are changes in relative employment shares (moving from one job type to another) and changes in the relative wage structure (how much more one job pays compared to another).

By layering these steps, the paper can isolate four distinct drivers: demographic change, employment reallocation, relative wage structure changes, and wage growth within specific job cells.

Four drivers of a stagnant economy

The study uses nationally representative data from 1980 to 2024 to map these components. The results reveal that Japan's wage stagnation (specifically the 1996–2014 period) was uniquely severe. This is because all four components moved in a negative direction simultaneously.

The paper reports that while wage growth within job types (the "cell-wage" component) contributed positively to wages over the long run, it was heavily offset by demographic and reallocation pressures. As shown in, the cumulative impact of demographic change and employment reallocation has been consistently negative since 1980.

One of the most striking findings is the magnitude of the "reallocation" effect. The author uses an accounting counterfactual to simulate what wages would look like if demographics and employment shifts had not occurred. The paper finds that by 2024, the counterfactual increase in wages would have been 34.7 log points. This compares to the actual increase of only 22.6 log points .

This 12.1 log point gap quantifies the massive downward pressure exerted by the redistribution of the workforce.

Moving beyond the part-time narrative

While much of the existing research blames the rise of part-time work, the author demonstrates that the reality is more nuanced. The negative contribution from employment reallocation is not just a story of "more part-time workers."

The paper finds that reallocation also occurs within full-time employment. It also occurs across different industries and establishment sizes (the number of employees in a business). For example, the study shows that even among full-time workers, employment shares have shifted away from higher-wage sectors and larger establishments toward lower-wage ones [Table 2]. This suggests that the "dualization" of the labor market—the split between protected and unprotected jobs—happens across multiple dimensions. It includes company size and industrial sector, not just employment status.

The verdict

Is this framework ready to guide policy? The answer depends on what you intend to fix. The paper provides a powerful diagnostic tool. It proves wage stagnation is a multi-front problem. It effectively demonstrates that focusing solely on increasing wages within specific sectors will not solve the aggregate problem. This holds true if the underlying demographic and sectoral shifts continue unabated.

However, the study is strictly an accounting exercise. The author explicitly states that the framework identifies what happened, not why it happened. It does not uncover the structural reasons why companies choose certain industries. It also does not explain why workers move between them. For a policymaker, this means the paper provides the "map" of the stagnation. It does not provide the "engine" for growth. To drive wages upward, one would still need to address the deeper structural causes behind the reallocation patterns identified here.

Figures from the paper