A Better Test for Whether Trade Makes Work Precarious

The labour-market response to expanding services trade is best judged by asking several linked questions rather than searching for one verdict. This final chapter answers the apparent paradox at the centre of the subject: more services trade can be associated with shorter current job tenure without a broad rise in measured employment exits, sector switching, or temporary-contract use. The practical value of that answer is conditional. It lets readers distinguish greater dynamism from contractual precarity while still asking whether the movement reaches different workers, places, tasks, and dimensions of job quality unevenly.

The evidence does not support either of the easy conclusions. Shorter tenure is not proof that work has become precarious, but the absence of a detected rise in temporary contracts is not proof that trade is good for workers. The first claim confuses a measure of time with a measure of security. The second confuses one narrow form of security with welfare. A defensible judgment has to keep quantity, movement, access, quality, and evidence strength in view at the same time.

The paradox resolves when the outcomes stay separate

The headline tenure result concerns the time people have accumulated in their current jobs when they are observed. Greater services-trade exposure is generally associated with shorter current job duration in the OECD analysis. That is evidence of a faster-changing set of employment relationships at the level of the groups being studied. It is not a complete history of the workers in those groups, and it is not a direct measure of whether their next job is better or worse.

The companion outcomes change the interpretation. The analysis does not identify a systematic rise in the share of workers entering or leaving employment. It does not identify a systematic rise in switching into or out of broad sectors. It does not find greater use of temporary contracts for services exports or imports; imported services used as inputs are associated with lower temporary-contract use. The four measures together make it difficult to describe the tenure result as a broad shift toward visible unemployment, sectoral displacement, or temporary employment.

That is why the most accurate answer to the title is a conditional one: more services trade can shorten current tenure without making work more precarious in the measured contractual sense because the observed tenure decline is not accompanied by a broad increase in the other measured routes. The phrase “in the measured contractual sense” is essential. The evidence narrows the explanation; it does not certify the experience of every worker or every job change.

A plausible interpretation is that more workers are moving from one job to another within broad sectors, sometimes in search of better matches. A worker may leave an employer while remaining employed, carry an occupation into a new firm, or enter a growing activity after a shorter period in the previous job. The pattern of stable employment-status and sector-switching shares is compatible with those routes. But the studies do not observe the reason for each move, the wage or autonomy of the next job, or whether the worker had a short spell of unemployment between two survey observations. “Better match” is therefore a credible inference, not a measured classification.

The distinction between absence of evidence and evidence of absence does most of the work here. A null relationship means that the study does not detect a systematic change in the measured outcome under its definitions, sample, and controls. It does not mean that no worker lost employment, accepted a temporary contract, or moved across a meaningful task boundary. Annual endpoint comparisons can miss short interruptions. Broad sector categories can hide substantial changes inside the same label. A permanent contract can coexist with pressure, weak prospects, or strained workplace relations.

The causal status is equally important. The OECD analyses of tenure, employment-status transitions, sector switching, and temporary contracts use extensive fixed effects, not a further causal-inference design. Those controls make comparisons more informative by accounting for persistent differences and common changes, but they do not automatically remove every time-varying shock that could affect both trade exposure and labour-market movement. The results should therefore be described as associations. The stronger instrumental-variable designs elsewhere in the material cannot be borrowed to upgrade them: an instrument for non-local demand in a local job-multiplier analysis answers a different question from an association between services-trade exposure and current tenure.

This leaves a useful but bounded conclusion.