We present a theoretically robust and empirically tractable representation of the aggregate labor supply curve at the extensive (employment) margin. First, we introduce the simple and basic concept of the reservation wedge: the hypothetical percent shift in an individual's potential earnings required to render her indifferent between employment and nonemployment. This concept generalizes reservation wages to the context of heterogeneity in earnings. For any given specific model, the reservation wedge serves as the sole scalar sufficient statistic for employment preferences. The CDF of the reservation wedges is the aggregate labor supply curve at the extensive margin. Second, we directly measure the wedge distribution in a representative household survey -- thereby nonparametrically mapping out the global labor supply curve of the U.S. population. For small deviations, the empirical curve exhibits large Frisch elasticities above 3, hence locally consistent with business cycle evidence. Rather than constant, the empirical arc elasticities shrink towards 0.5 for larger, upward shifts, thereby potentially also reconciling large local elasticities with the small arc elasticities implied by recent quasi-experimental evidence from tax holidays. Third, in a model meta-analysis, existing models would fail to match the global shape of this empirical curve. Fourth, we engineer one model to fit the empirical curve. A business cycle accounting exercise reveals that this fitted model (under the assumption of efficient rationing) would help reconcile cyclical employment fluctuations with labor supply.